Citation
Development of Radio Frequency Identification (RFID) Temperature Tracking Systems for Food Supply Chains

Material Information

Title:
Development of Radio Frequency Identification (RFID) Temperature Tracking Systems for Food Supply Chains
Creator:
Amador, Cecilia
Place of Publication:
[Gainesville, Fla.]
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (223 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Agricultural and Biological Engineering
Committee Chair:
Bucklin, Ray A.
Committee Members:
Geunes, Joseph P.
Leary, James D.
Emond, Jean-Pierre
Nunes, Cecilia N.
Wysocki, Allen F.
Brecht, Jeffrey K.
Engels, Daniel
Graduation Date:
8/7/2010

Subjects

Subjects / Keywords:
Ambient temperature ( jstor )
Antennas ( jstor )
Boxes ( jstor )
Cooling ( jstor )
Pineapples ( jstor )
Radio frequency identification ( jstor )
Sensors ( jstor )
Supply chain management ( jstor )
Temperature profiles ( jstor )
Temperature sensors ( jstor )
Agricultural and Biological Engineering -- Dissertations, Academic -- UF
chain, cold, environmental, food, frequency, identification, logistics, monitoring, perishables, process, radio, rfid, supply, temperature, traceability, tracking, transportation
City of Gainesville ( local )
Genre:
Electronic Thesis or Dissertation
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
Agricultural and Biological Engineering thesis, Ph.D.

Notes

Abstract:
DEVELOPMENT OF RADIO FREQUENCY IDENTIFICATION (RFID) TEMPERATURE TRACKING SYSTEMS FOR FOOD SUPPLY CHAINS Food items require temperature controlled supply chains since exposure to certain temperature conditions can diminish product quality and create safety threats. Temperature tracking systems should then be in place in order to monitor the temperature management along their supply chains. Radio frequency identification (RFID) has been suggested as having the potential to become an enhanced temperature tracking method; yet, very few studies have explored the subject in depth. Furthermore, practical details about its application, such as the proper use of RFID tags with probe and without them, how to surpass environmental interactions taking place along the supply chain or how to achieve relevant monitoring keeping the costs down, and the economic benefit it will bring to the food industry still remain unclear. The following work aims to offer insight on these matters and to create viable applications for the technology in real-life food supply chains. Four objectives were established: 1) To compare the performance of RFID temperature tags versus conventional temperature tracking methods in a food supply chain; 2) To compare the utilization of RFID temperature tags with probe and without them along a food supply chain; 3) To determine the level of instrumentation (amount of sensors and the best locations for their placement) of an efficient temperature tracking system in three different scenarios for food supply chains (For products prone to low and high temperature abuse, for products susceptible to high temperature abuse, and for shelf-stable products); and 4) To create the business case for a RFID temperature tracking system when combined with shelf-life prediction software by performing an economic analysis in one of the systems previously designed. In order to achieve the first three objectives, a shipping trial was performed with crownless pineapples; while spherical water bottles mimicking produce and First Strike Rations (FSRs) were subjected to thermal relevance and readability studies for the third one. Additionally, software material allowing temperature estimations inside the pallet of FSRs and shelf-life prediction were developed as support material for objectives three and four. Finally, a return on investment (ROI) study was performed for a load management system based on the final monitoring system developed for FSRs. Results indicate that, although analogous with respect to accuracy in the temperature measurements, RFID systems are superior as temperature tracking method to conventional methods. In addition, RFID tags with probe are important to monitor the critical points of the load, which are the areas of the load where temperature abuse is most likely to occur inside the product; while RFID tags without probes are relevant during monitoring of ambient conditions during storage and transportation. Also, a monitoring system for crownless pineapples was designed, which in some cases involved the use of more than one tag per pallet. Moreover, RFID monitoring systems were also designed for loads of certain varieties of produce such as apples, oranges, pomegranates, passion fruit and tangerines, and for FSRs; but these allowed the use of only one tag per pallet. Lastly, the Return on Investment (ROI) analysis of the RFID-based load management system designed for FSRs was calculated to be 719.49%; which proved that this technology can be an important tool for value generation in food supply chains. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2010.
Local:
Adviser: Bucklin, Ray A.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-08-31
Statement of Responsibility:
by Cecilia Amador.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
8/31/2011
Resource Identifier:
004979706 ( ALEPH )
705932736 ( OCLC )
Classification:
LD1780 2010 ( lcc )

Downloads

This item has the following downloads:


Full Text






APPENDIX B
RECOMMENDED LEVEL OF RFID INSTRUMENTATION FOR THE CROWNLESS
PINEAPPLE SUPPLY CHAIN WHEN THIS USES THE REFRIGERATED TRUCK/OPEN
HOLD TRANSPORTATION METHOD



Proper monitoring is achieved recording two or three temperatures per pallet, according to

the location of the pallet during forced air cooling. In order to monitor HTA, ambient probeless

tags are recommended for the top layers of the pallets and tags with probe are advised for their

center. Furthermore, tags with probe are also suggested to observe LTA in the bottom layers of

the pallets. In this case, no areas are specified as specific tag locations at the cargo level.









In the case of the fitting curves for 26.670C (80.00F), the lack of experimental data points

could be promoting the simplification of a more complex polynomial relationship into only a

first order polynomial equation. So, the model estimations for temperatures within the 26.670C

(80.00F) and 48.890C (120.00F) range could have a certain level of inaccuracy. Yet, once the

experimental data is available and the resulting curve is subjected to regression, this software

could be easily updated for model improvement.

The best way to determine whether the current or future versions of this software are

indeed accurate is by validating the model. This would allow a comparison of the calculated

acceptability scores versus the ones obtained by sensory evaluations in product subjected to a

particular temperature profile during a real-life supply chain. Anderson (2010) used this

methodology as a tool for concluding that empirical models are better suited for shelf-life

prediction than theoretical models in sweet corn cold chains. Likewise, this process could assist

in the determination of errors and minimization of them in this software; and thus it is necessary

as future work.

In addition, a feature indicating the amount of shelf-life in weeks left in the pallet given the

latest temperature pattern could be particularly useful in the field; and should also be pursued in

the future.

Finally, Figures 6-1 and 6-2 present the two recommendation scenarios in the final

software product.

Maximum Amount of Weeks of Shelf-Life at the Deployment Areas

The results indicate that for all sensor positions, all the meals in the pallet of FSR have less

than three weeks of shelf-life at storage conditions in the deployment area (Table 6-2). Thus, in

order to avoid any risks, it would be recommended to distribute and consume the entire product

present in the load around the two week period, when all products are still over the acceptability









Table 7-13. Total cost of a lost pallet according to its final destination.
Item Cost
Pallet losses when shipped to Iraq
FSRs $4,968.00
Cost of shipping to Iraq/pallet $134.95
Total loss per pallet when shipped to Iraq $5,102.95


Pallet losses when shipped to Pakistan (en route to
Afghanistan)
FSRs
Cost of shipping to Pakistan (en route to Afghanistan)/pallet
Total loss per pallet when shipped to Pakistan (en route to
Afghanistan)


$4,968.00
$101.26

$5,069.26


Table 7-14. Amount of product sent as yearly emergency shipments.
Item Afghanistan Iraq
Number of mishandled pallets/year (18.75%) 1860.00 930.00
FSR shipments sent for safety (pallets/year) 1280.00 640.00
Emergency shipments (pallets/year) 580.00 290.00
Emergency shipments (containers/year) 14.50 7.25
Actual emergency shipments (containers/year) 15.00 8.00


Total
2790.00
1920.00
870.00
21.75
23.00


Table 7-15. Amount of money destined to yearly emergency shipments.
Item Unit price Number of units
Emergency shipment-pallets sent to Afghanistan $5,069.00 600.00
Emergency shipment-pallets sent to Iraq $5,103.00 320.00


Total


Total
$3,041,555.00
$1,632,944.00

$4,674,499.00


Table 7-16. Yearly savings created by the proposed system.
Pallet loss avoided/year Savings per unit Number of units saved Total Savings
From pallets sent to Afghanistan $5,069.00 930.00 $4,714,409.00
From pallets sent to Iraq $5,103.00 465.00 $2,372,872.00
From emergency shipments $4,674,499.00

Total savings/year $11,761,779.72









Table 7-2. Yearly costs of RFID tags.
Tag Number of units/Year
Caen RFID A927Z
In transit 14880.00
Safety stock 2232.00
Subtotal Caen RFID -A927Z 17112.00


Avery AD-233
Rollsx 1000 tags
Subtotal Avery AD-233


50.00


Unit price


25.00
25.00
25.00


200.00


Total


Total price

$372,000.00
$55,800.00
$427,800.00


$10,000.00
$10,000.00

$437,800.00


Table 7-3. Cost of the project during the first year
Item Number of units


Work station hardware +
software
Backup hardware + software
Maintenance + training
Tags
Integration


50.00


of operation.
Unit price

$14,065.00


Total


Total

$703,268.00
$70,327.00
$175,817.00
$437,800.00
$2,500,000.00

$3,887,212.00


Table 7-4. Cost of the project during the second and fourth year of operation.
Item Total
System replacement $175,817.00
Maintenance + training $175,817.00
Tags $437,800.00


$789,434.00


Total









has a temperature (n) that will be the middle point of an interval that gathers the temperature

readings of 85.00% of the measured locations. The proposed range for the time periods used is

[n-6.00, n+6.00]; a total of 12.000C of temperature differential. The likelihood of gathering

85.00% of the temperature readings in this interval varies according to the duration of the

heating/cooling episode, decreasing as the heating/cooling episode increases in time:

* For 0.50/0.50 hours: 100.00%
* For 1.00/1.00 hours: 96.00%
* For 2.00/2.00 hours: 64.00%
* For 4.00/4.00 hours: 42.00%

Comparisons were made between the temperature profile of this hypothetical point and the

ones obtained in the trials. Three locations in the pallet were then considered Points of Relevance

(PoR) and suitable for sensor placement (where temperature "n" will be recorded):

* Point A: Layer 9 (Row 3), X= 14.00, Y=9.00.
* Point B: Layer 13 (Row 5), X=8.00, Y=1.00.
* Point C: Layer 18 (Row 6), X=14.00, Y=1.00.

Where the layers move along the vertical axis (Z) and are labelled, along rows, in an

ascending way from the top to the bottom. Figure 4-3 illustrates the places where these three

points are located. It is necessary to indicate that, because of thermal symmetry, the

recommended thermally relevant points will be located in its equivalent position on the other

side of the pallet.

Analysis of the likelihood of gathering 85.00% of the temperature readings of the pallet

using these points as middle points for the interval was conducted (Table 4-4). Comparing these

estimates, point B would be the best for temperature monitoring because it has the highest values

in the first three exposure times, which are more likely to occur along the supply chain than the

last one. Point A would follow in preference and then point C. The selection of one of the









LIST OF REFERENCES


Abada, E., Zampolli, S., Marcoc, S., Scorzoni, A., Mazzolai, B. et al., 2007. Flexible tag
microlab development: Gas sensors integration in RFID flexible tags for food logistic.
Sensors and Actuators B 127, 2-7.

Abdullah, H., Rohaya, M.A., Rosli, H., Mohamed Selamat, M., 2000. Handling and
transportation trial of pineapple by sea shipment from Malaysia to the United Kingdom.
Acta Horticulturae 529, 317-327.

Acedo, A., Akinaga, T., Tanabez, T., 2004. Inhibition of chilling injury and quality changes in
pineapple fruit with prestorage heat treatment. Journal of Food, Agriculture and
Environment 2, 81-86.

Akamine, E.K., Goo, T., Steepy, T., Greidanus, T., Iwaoka, N., 1975. Control of endogenous
brown spot of fresh pineapple in postharvest handling. Journal of the American Society for
Horticultural Science 100, 60-65.

Amador, C., Emond, J.P., Nunes, M.C.N., 2009. Application of RFID Technologies in the
Temperature Mapping of the Pineapple Supply Chain. Journal of Sensing and
Instrumentation for Food Quality and Safety 3 (1), 26-33.

Amador, C., Emond, J.P., 2010. Evaluation of sensor readability and thermal relevance for RFID
temperature tracking. Computers and Electronics in Agriculture 73 (1), 84-90.

Ames, H., 2006. Authentication from a Cold Chain Perspective. Pharmaceutical Commerce.
July/August 2006.

Amos, N.D., 2001. Factors affecting fruit temperature maintenance within refrigerated
containers. Proceedings of the IRHACE Annual Conf. Vol I, 7-10.

Anderson, K. E., 2010. Development of a simulator for sweet corn cold chain distribution.
M.Sc. Thesis. University of Florida, Agricultural and Biological Engineering.

Angeles, R., 2005. RFID Technologies: Supply-Chain Applications and Implementation Issues.
Information Systems Management, Vol. 22 (1), 51-65.

Aroor, S. R., Deavours, D. D., 2007. Evaluation of the state of passive UHF RFID: An
experimental approach. IEEE Systems Journal 1 (2), 168-176.

Asimakopoulos, G., Spiros Louvros, G., Triantafillou, V., 2007. METATRO: A Real Time RFID
enabled haulage monitoring system for perishable comestibles. Paper Presented at the
IEEE Conference on Emerging Technologies and Factory Automation (ETFA), Patras,
Greece.

Banerjee S.R., Jesme, R., Sainati, R.A., 2007. Performance Analysis of Short Range UHF
Propagation as Applicable to Passive RFID. Paper Presented at 2007 IEEE International
Conference on RFID, Grapevine, TX.


210









BIOGRAPHICAL SKETCH

Cecilia Amador was born in 1980 in Lima, Peru. She attended the Universidad Nacional

Agraria La Molina (UNALM), where she received her bachelor's degree in food industries in

2003. In 2005 she began her Master of Science program at the Agricultural and Biological

Engineering Department at the University of Florida, and completed it in December 2006, under

the direction of Dr. Jean-Pierre Emond. In order to continue her work with Dr. Emond, she was

admitted into the doctoral program at UF in the winter of 2007. As a graduate student in the

UF/IFAS Center for Food Distribution and Retailing (CFDR), she has been involved in projects

related to cold chain, RFID technologies and postharvest management. In addition, during her

years of graduate school at UF, she has also served as the vice-president of the Graduate Student

Council (GSC), as a senator in UF's Student Government, and as the president of the Hispanic

Graduate and Professional Student Association (HGSA).


223









Room Cooling

Although not a precooling method, room cooling is still used in some places whenever the

aforementioned precooling methods are not available. According to Meana (2005), room cooling

is the simplest method for precooling, since it only needs a refrigerated room with proper cooling

capacity. This is used mostly in products with relatively long shelf-life (such as potatoes and

onions), as a step prior to storage. These are inside loosely stacked packaging in the cooling

room, allowing for ventilation in the side of the containers (Thompson et al., 1998; Thompson et

al., 2002). Other relevant factors are the presence of proper package venting and good air flow in

the room so the cold air can past near and through each package. When these are accomplished,

most products will cool in less than 24 hours. According to Thompson et al. (1998), poor room

air flow, tightly stacked product, and poor box venting will extend cooling to many days.

Most of the internal heat load of the package needs to be transferred by conduction to the

surface so that the cold air can remove it by convection. As a result, the cooling rate of this

method is very slow when compared to other precooling methods (Meana, 2005).

According to Cortbaoui (2005), the advantages of room cooling include its low labor and

equipment cost. Yet, as mentioned before, since its cooling rates are low, this method is

appropriate for products with a low respiration rate and those who are not affected by slow

cooling such as onion or potatoes (Sargent et al., 1988). Dincer (2003), states that, if not

controlled properly, room cooling can end up creating moisture loss issues in the product.

Short Term Storage

Once cooled, produce will gain temperature rapidly when exposed to warmer

environments. Therefore, it is necessary to place it in temperature controlled environments along

its supply chain, so optimum conditions are kept and product quality is preserved as much as

possible.









Temperatures, 2010. Temperature Sensor Types. Temperatures.com, Inc. Available at:
http://www.temperatures.com/sensors.html. Accessed June 28th, 2010.

Thompson, A.K., 2003. Fruit and Vegetables: Harvesting, Handling and Storage. Blackwell
Publishing Ltd., Oxford.

Thompson, J., 2004. Pre-cooling and storage facilities. The Commercial Storage of Fruits,
Vegetables, and Florist and Nursery Stocks. In: Agriculture Handbook 66. USDA,
Washington, D.C.

Thompson, J., Mitchell, F., Rumsey, T., Kasmire, R., Crisosto, C., 2002. Commercial Cooling of
Fruits, Vegetables, and Flowers, Revised Edition. University of California-Division of
Agriculture and Natural Resources, Oakland.

Thompson, J.F., Mitchell, F.G., Rumsey, T.R., Kasmire, R.F., Crisosto, C.C., 1998. Commercial
Cooling of Fruits, Vegetables, and Flowers. University of California-Division of
Agriculture and Natural Resources, Oakland.

Thompson, M., Sylvia, G., Morrissey, M.T., 2005. Seafood Traceability in the United States:
Current Trends, System Design, and Potential Applications. Comprehensive Reviews in
Food Science and Food Safety. 4 (1), 1-7.

Thyagaraja, K., 2007. GatorPacker: A worker productivity monitoring system using RFID
(Radio Frequency Identification) technology. M.Sc. Thesis, University of Florida,
Agricultural and Biological Engineering Department.

Tutar, M., Erdogdu, F., Toka, B., 2009. Computational modeling of airflow patterns and heat
transfer prediction through stacked layers' products in a vented box during cooling.
International Journal of Refrigeration 32 (2), 295-306.

United States Department of Agriculture/Economic Research Service, 2010. Food availability-
Spreadsheets: Fresh fruits. Available at
http://www.ers.usda.gov/Data/FoodConsumption/FoodAvailSpreadsheets.htm, Accessed
on January 16th, 2010.

Uysal, I., 2010. Development of models for a temperature profile estimator. Unpublished report.
Agricultural and Biological Engineering Department, University of Florida, Gainesville,
FL.

van Boekel, M.A.J.S., 2008. Kinetic Modeling of Food Quality: A Critical Review.
Comprehensive Reviews in Food Science and Food Safety 7, 144-158.

Verboven, P., Tijskens, E., Ramon, H., Nicolai, B.M., 2005. Virtual filling and airflow
simulation of boxes with horticultural products. Acta Horticulturae 687, 47-54.

Vergara, A., Llobet, E., Ramirez, J. L., Ivanova, P. et al., 2007. An RFID reader with onboard
sensing capability for monitoring fruit quality. Sensors and Actuators B 127, 143-149.









identifier number in a machine-readable form that can be accessed quickly and reliably, with no

human translation. However, unlike bar codes, RFID does not require line-of-sight.

A RFID system consists of four elements: RFID tags, RFID readers, reader antennas, and

information processing software (Thyagaraja, 2007). The reader sends radio frequency (RF)

waves through its antenna and collects the RF waves emitted or reflected from the tag. The

information, such as a unique serial number for every tag, is carried with these RF waves.

Industrial-Scientific-Medical (ISM) Bands

Radio waves are part of the electromagnetic spectrum which use is regulated by

governments around the world. Governments have then assigned different uses for the various

parts of this spectrum (Angeles, 2005). Industrial-Scientific-Medical (ISM) bands are special

license-free bands that have been set aside by regulatory bodies across the world and are

available for use in all countries. Anyone can use unlicensed frequencies as long as they follow

the rules of transmission and broadcast (Sweeney, 2005). By using ISM bands RFID system

operators can avoid licensing processes; but, will still have to follow closely the ISM rules on the

use of the band, limits on radiated power, and tolerance of interference (Hunt et al., 2007).

Carrier Frequency

Carrier frequency is the center of a particular RF bandwidth. For instance, a 915 MHz

carrier frequency with a 10 MHz bandwidth encompasses bands from 910 to 920 MHz (Reed,

2009). Common standard frequencies include 135 kHz, 13.56 MHz, 868-870 MHz and 2.45

GHz. The frequency determines read range, the ability of the wave to penetrate objects and data

transfer rate. Higher frequencies have longer read range but less penetrating ability. They can

also accommodate higher data rates, thus using less power per unit of data (Angeles, 2005; Reed,

2009). Yet, these systems need more power to transmit the same distance as systems using lower









Identification

According to Panos and Freed (2007), the majority of the produce facilities in the US have

minimal information systems capabilities, being most of their logistic operations still

documented on paper. Nonetheless, some members of this industry use automatic data collection

through bar code systems (pallet tags) in order to identify the products being moved along the

supply chain. This feature facilitates the inventory management of all parties involved, increases

their inventory accuracy and enhances the visibility of the products throughout the chain.

RFID provides a faster alternative to bar code systems because it does not require line-of-

sight between the tag and the reader and can collect the information of many tags at the same

time. It also reduces labor costs and allows the storage of bigger amounts of information than the

ones currently contained in the bar code. Taking these benefits into account, in 2003 the U.S.

Department of Defense mandated all of its suppliers to comply with RFID tagging at both the

pallet and case levels. In a similar fashion and almost at the same time, Wal-Mart also demanded

its main suppliers to execute the same command; having by 2008 incorporated all of their 15,000

suppliers into it. In the supermarket arena, Metro AG, in Germany, has also introduced the

pallet/case RFID tagging into its daily business.

RFID also provides protection against imitation products. Using non-reprogrammable

RFID tags, information about where that product was harvested or processed can be stored and

remain intact until the end of the supply chain, which could be important for products with

denomination of origin, certifications or an established branding that the consumer perceives as

an added value.

Trace back

Opara (2003) explains that the traceability in an agricultural and food supply chains

contributes to the demonstration of the transparency of the chain, adding value to the overall









Table C-4.
Time
1400
1402
1405
1407
1410
1412
1415
1417
1420
1422
1425
1427
1430
1432
1435
1437
1440
1442
1445
1447
1450
1452
1455
1457
1500
1502
1505
1507
1510
1512
1515
1517
1520
1522
1525
1527
1530
1532
1535
1537
1540
1542
1545


Continued.
Ambient temperatures (oC)
11.58
11.66
11.74
11.82
11.96
12.01
12.09
12.13
12.19
12.27
12.33
12.41
12.44
12.53
12.65
12.63
12.63
12.25
11.61
11.06
10.82
10.69
10.53
10.32
10.30
10.66
10.57
10.58
10.49
10.41
10.55
10.73
10.81
10.86
10.91
10.90
10.90
10.85
10.87
10.86
10.84
10.84
10.85


206









overnight in a cold room at 4.000C. The next morning, the pallet was moved from the cold

storage to the outside of the building, where it was exposed to ambient temperatures for a

specific amount of time (either 4.00, 2.00, 1.00 or 0.50 hours). It was stored once again in the

cold room and recordings were taken for the same amount of time it was exposed to ambient

temperatures (4.00, 2.00, 1.00 or 0.50 hours, respectively). Quantitative analysis was performed

using quintiles in Microsoft Excel (Microsoft Corporation, Redmont, WA).

Readability Study

The readability of Class 3 battery assisted passive tags (BAP) was tested in two sides of the

pallet using the IntelleflexTM DK900 (Intelleflex Coorporation, Santa Clara, CA) RFID system

operating at 915 MHz. The pallet was located in an environment mimicking the unloading area

of a distribution center, in open air surroundings but with a sea container and metallic doors

nearby. The placement of the tags was determined by the results of the relevance study. Tags had

to be placed in one of the 1.22 m sides (Configuration a-I) and in both 1.02 m sides

(Configuration 3), but since these had a similar layout only one of them was considered for this

testing. Each side was studied independently. The antennas were positioned in the middle point

of the front of the side being read. Readings were obtained using four different antenna distances

(0.50 m, 1.00 m, 2.50 m and 5.00 m) with respect to the pallet.

Tag placement

The placement of the tag inside the load could be a determinant factor in the readability of

the system and its resulting reading range. In this trial, the depth of tag placement varied with the

side configuration. For the area studied in configuration 3 it was 0.12 m, equivalent to the width

of two bottles. For configuration a-I it was set to 0.14 m, the height of one bottle. Figure 4-2









for all products (4.00), a window displays "Pass" or "Discard" indicating the required action

regarding the pallet.

Application of the Shelf-Life Software

Ambient temperature profiles were developed using the information collected in previous

trials performed by the U.S. Department of Defense (DoD) on shipments from the U.S. to

Kuwait. A temperature profile was created based on worst-case scenario ambient conditions

recorded during those trials. This was generated by taking the maximum temperatures registered

in 17 non-refrigerated sea containers during 39.00 days where environmental monitoring of the

shipment and storage periods was performed.

Additionally, two other temperature profiles were developed also for ambient

temperatures. All of them had as initial input the temperature profile described before plus one or

two extra weeks of data of hypothetical temperature information corresponding to the storage

phase in the Middle East. These data points were created by repeating in each week the

temperatures obtained originally in the trials under this storage stage.

The following is the description of the ambient temperature profiles created:

* Profile 1: Original shipment information (storage in the US + marine shipment + 1.00
week of storage in the Middle East).

* Profile 2: Original shipment information + 1.00 week of storage in the Middle East.

* Profile 3: Original shipment information + 2.00 weeks of storage in the Middle East.

All ambient temperature profiles were assumed to be recorded in the outside wall of the

pallet; in the first location considered for the temperature sensor. The second and third sensor

positions corresponded to the locations of the Points of Relevance (PoR) A and B from the

research work described in the previous chapter.









Kader (2002) recommends using a stowage pattern that forces refrigerated air to flow

through and around the packages and do not allow air to bypass around the pallet units. Along

the same lines, Vigneault et al. (2009) suggests employing pallets, boxes and inner packaging

with enough venting and airspaces to allow vertical airflow through the pallet load.

Main Transportation Modes

According to Peleg (1985), some of the many factors involved in choosing an optimal

transportation chain are as follows: distances to markets; cost per ton-kilometer; types and

varieties of produce; and climatic conditions en route requiring refrigeration, ventilated cooling,

or heating for prevention of freezing. Additional constraints are types of packaging used, types of

available handling techniques, and unitizing methods (pallets, slip sheets, intermodal containers,

etc.).

Overland transport by trucks

This is by far the most popular mode of overland fresh produce transportation. There are

two types of vehicles used in highway transport: refrigerated semitrailers and intermodal

containers. Refrigerated semitrailers can be classified as intermodal transport vehicles. Detached

from the tractors the semi-trailers can be transported on railroad flatcars, driven right into sea

vessels, or simply hauled by a tractor-trailer on the highway. They are available in lengths of 12

m, 13.7 m, 14.6 m, or 16.2 m; and most of them use mechanical refrigeration (Hui et al., 2003).

In addition, new units can provide heat when the trailer is operated in ambient conditions colder

than the set point temperature (Kader, 2002).

Most of the heat that the refrigeration unit removes comes from heat conducted across the

walls and from air leaking into the trailer. Therefore, if the product is not center-loaded it will be

warmed when in contact with the walls. Controlled atmospheres are not applied in highway









Configuration 3 only point b was read, and very weakly (0.09% in percentage of readings). Thus,

only point d in Configuration a will be taken into account in the comparison of the RFID

systems.

As can be seen in Table 5-7 the lowest readability results at all distances were obtained by

point d; which shows the strong impact that the metallic components of the packaging and the

metal surfaces of the sea container had on the signal. This phenomenon was expected, given the

fact that this model of Caen tag was not designed for applications in metallic environments; and,

as aforementioned, this element can affect greatly the operation of the system. Differences in tag

design between the Caen system and the Intelleflex system; as well as other technical

considerations, such as reader power output and antenna size and pattern might have played an

important role in their performances. Equalizing some of these variables might present a more

accurate panorama and allow proper comparisons.

The tag placed on the surface of the wooden structure of the pallet had the highest level of

readability at every single reading distance. There are two possible causes for this: First, these

tags were in contact with wood, which does not reflect RF waves, and not with the FSR boxes

full of packages with metal in their structure. And second, the angle of the reader antenna and the

location of the tag might have determined reflection patterns different than those in the other tag

positions, which created a less detrimental multipath effect.

The results corresponding to the tags placed on the surface of the FSR load are relatively

similar, though Configuration 3 seems to provide more robust readings than a. This is most

likely also product of different multipath effects.

Even though conclusions cannot be made with respect to the robustness of the systems

with respect to antenna distances, the use of those higher than 1.50 m is not recommended. This,









sensors was better than dealing with the bigger and thicker ALB-2484; problem quite evident at

the time of taping them to the exterior of the pallet of corrugated boxes. Even less convenient

were the big ILR i-Q 32T sensors, which could not be taped and had to be placed inside the

primary packages at all times. During the setup ofRPCs, the sensors that were easier to place on

the package (because they fit in the slot included in the package design) were the ThermaProbe

RF and ThermAssure RF tags; while the other two were secured using the same method as in the

previous package.

When comparing the results obtained with the ThermaProbe RF RFID tag and the HOBO

sensor placed in the pineapple of the core of the pallet the statistical analysis (Table 3-2) showed

significant differences between the sensors' readings (P = 5.66e-03). The ThermaProbe RF RFID

tag consistently produced higher values in all pallets, being more evident in the case of the ones

using corrugated boxes as packaging (Table 3-3). This phenomenon could be explained by the

difference in accuracies of the sensors used ( 0.500C HOBO, +0.100C ThermaProbe RF). As far

as RFID sensor performance, only two of the ambient ones (ThermAssure RF and ILR i-Q 32T)

were able to provide readings point by point. ALB-2484 RFID tags only supplied a final graphic

with the temperature profile, but did not allowed access to the data for further comparison. The

three temperature profiles had the same shape in each pallet (Figure 3-4). However, for both

kinds of packaging, comparison of the curves obtained with ThermAssure RF and ILR i-Q 32T

sensors showed that the former had higher values most of the time (Figures 3-4, 3-5 and 3-6).

The statistical analysis (Table 3-4) later on showed that there were significant differences

between the readings of both sensors (P < 1.55e-07). The maximum delta obtained between these

in each pallet reached up to 2.700C in the pallets with corrugated boxes. This temperature

difference was most likely due to the fact that the bulky ILR i-Q 32T sensor was placed on the









Readability Study

Fixed system

Figure 5-7 shows the placement of the tags for Configuration a (1.22 m). These were placed

both in the PoRs found, as in their thermally equivalent positions inside the pallet. For

Configuration 3, the pallet was rotated 900 clockwise. Configuration a provided the highest

number of tags read: A minimum of five for each distance (a, b, c, d, e), and six during two of

the repetitions held at one meter. With Configuration 3, four tags were read during the first five

distances tested (b, c, d, e), but only three were recorded at three meters (b, c, e). The location

with the highest readability ofPoR A was position "e"; whereas for PoR B, the highest reading

percentages were distributed in locations "b" and "c" in Configurations a and respectively.

Tables 5-4 and 5-5 summarize the highest readability levels for each PoR in each one of the

configurations studied at different reading ranges.

In Configuration a~ the best reading distance for PoR A is 1 m, and for PoR B is either 2.00

m or 2.50 m. In configuration 3, this distance is 2.00 m for both PoRs. These results are quite

positive when considering the metallic environment that surrounded the pallet; and the fact that

the quad-laminate packaging of the components of the FSRs in direct contact with the tags

contained a layer of foil. Furthermore, the high density of the FSR packets inside the boxes

restricted the existence of air pockets able to propagate the RF signal and favor tag/reader

communication. Since metallic objects reflect RF waves, problems arise whenever the tags need

to harvest energy in order to backscatter its information (Sydanheimo et al., 2006). Mo et al.,

(2007) also reported that whenever there is a metallic object near an antenna, this can change the

antenna's radiation pattern, input impedance, radiation efficiency, and resonant frequency; which

can decrease the performance of the RFID system. Additionally, Lazaro et al. (2009) accounts









package locations and kinds of food products. Also necessary is the development of more

convenient sensors and portable readers in regards to their size and shape.

Finally, recommendations were given with respect to the instrumentation level required for

a RFID temperature tracking system for this supply chain.











Environmental temperature vs, temperature at point 36-prime


-; ,'
a I I
o


I s *
CL i


i-i





0 1000 2000 300 4000 5000 6000 7000 8000
Time samples, one per five minutes


Figure 5-9. Temperature profiles showing how the temperature in PoR B (36-prime) changed
with the temperature outside the pallet in all eight trials.


Pallet temperature vs. estimated temperature


-Environmental lemC:ralihje
.. Temperature at F-:,ni 20 alplra
--Estimated temprierai ,re


Time samples, one per five minutes


Figure 5-10. Comparison of ambient temperatures, temperatures measured in PoR A and the
estimated pallet temperatures for PoR A (point 20-alpha).










Table A-4. Percentage (%) of transportation time during which pineapples were exposed to different high temperature abuse (HTA)
scenarios when using the refrigerated truck trailer/open hold combination.
Front/lA Middle/2A Back/2B
Top Center Bottom Top Center Bottom Top Center Bottom
Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC


HTA
No.1 (T> 19.9 20.3 3.5
120C)


4.3 7.9


24.1 26.3 100.0 6.2


DS 20.6 28.2 22.1 11.8 -


HTA No.2
7.9
(T > 150C)

HTA No.3
(T17 4.5
(T > 17C)


3.3 8.0 26.8


0.6 10.7


DS: Defective sensor




Table A-5. Risk of fruit exposure to different high temperature abuse (HTA) scenarios according to the position of the fruit in the
pallet when using the refrigerated truck trailer/open hold combination.
Overall Top Center Bottom

HTANo.1 (T > 120C) H H H H

HTANo.2 (T > 150C) M H L N

HTANo.3 (T> 170C) L M N N

H = High, M = Medium, L = Low, and N = None.


5.2 4.2









Table 2-4. EPCTM tag classification.
Class Power Range Memory
0 None < 3 m 1 to 96 bits, Read
Only
1 None < 3 m 1 to 96 bits,
Read/Write Once
2 None < 3 m 1 to 96 bits,
Read/Write
3 Battery assisted < 100m < 100 Kilobytes,
Read/Write
4 Battery assisted < 300m < 100 Kilobytes
Read/Write
5 Battery assisted, Unlimited Unlimited,
AC/DC Read/Write
connection
Source: Banks et al., 2007. Table 3-1, pp. 69.


Communication
Backscatter

Backscatter

Backscatter

Backscatter

Active
Transmission
Active
Transmission


Peripherals
None

None

Security

Security, Sensors

Security, Sensor

Security, sensors, can
communicate with other
tags


Cost
Low

Low

Medium

High

High

Very High









or quadruple (HTA No. 3: > 17 C) (Dull et al., 1967). HTANo. 1 was considered the least

intense exposure, while HTA No.3 was considered the most intense exposure. The values of

HTA given represent the percentage of time when the fruit were exposed to these temperatures

with respect to the total duration of the storage and transport operations analyzed.

Low temperature abuse (LTA) analysis

Only one level of exposure to low temperature abuse (LTA) were defined. This LTA level

was created by adding the 0.5 C error margin of the HOBO sensors to the chilling injury

threshold temperature for pineapple (i.e., 7.23 C). As in the HTA analysis, the values of LTA

given represent the amount of time when the fruit were exposed to LTA temperatures with

respect to the total duration of the storage and transport operations as a percentage.

Results and Discussion

Temperature Profiles of Pineapples using Different Combinations of Packaging,
Transportation Methods and Locations in the Cargo Environment (Sea Container and
Truck Trailer).

Significant differences were found in the majority of the temperature profiles with regards

to transportation method, position within the cargo environment, packaging, and the resulting

combinations. However, at the bottom of the pallets no significant differences were obtained

with regards to the transportation method employed, and the placement inside the container

(Tables A-i and A-2). In addition, no significant differences were obtained due to packaging at

the top of the pallets loaded in the open holds (Table A-3). Overall, the transportation method

(truck trailer or sea container), the position inside the cargo environment analyzed (back, middle

and front) and the type of packaging (corrugated or RPC) had a significant effect on the

temperature profiles measured during handling and shipping of crownless pineapple from the

Pacific coast of Costa Rica to the U.S..









-t
= ( V-Vo \
In vc_ ) (5-2)

Hence, if the ambient and the initial temperature in the PoR are known, and also the temperature

in the PoR a given time "t"; it becomes trivial to calculate the time constant. Unlike the

electronic circuit described above, it was possible to have a different time constant for the

heating and cooling cycles due to the way the experiments were designed.

Calculations and graphics were performed in Matlab (The MathWorks, Inc.; Natick,

MA).

Results and Discussion

Relevance Study

No thermally relevant position representing 85.00% of the temperatures of the pallet was

detected. A great level of temperature variability was present in the load. This was mostly

because of the existence of high temperature differentials due to heat accumulation in the middle

of the boxes after the warming episodes. Even though the temperature differentials within boxes

and along the pallet were quite significant in all trials, at the end of the 3.00 d/3.00 d and 4.00

d/4.00 d ones these were reduced considerably because the pallet was reaching thermal

equilibrium.

An approach already employed in the previous chapter of using a location that has a

temperature (n) that represents the middle point of an interval that gathers the temperature

readings of 85.00% of the measured locations was proposed. The suggested range was [n+8.00,

n-8.00]; a total of 16.000C of temperature differential. Table 5-2 displays the likelihood of

gathering 85.00% of the temperature readings in this interval for each heating/cooling episode.

No particular trend was found between this particular likelihood and the duration of the episodes.









intense exposure to LTA, since lower temperature values were recorded compared to those

registered in the sea container.

Refrigerated truck trailer/open holds

Overall, there was a high risk of exposure to LTA in the refrigerated truck trailer/open hold

combination (Tables A-12 and A-13). This exposure was mostly concentrated in the bottom and

central layers of the pallets, also high risk areas, while the top layers only presented medium risk

of LTA.

When analyzing the effect of packaging on LTA incidence and duration in the Front/ A

pallets, there was not much of a difference between the use of corrugated cartons and RPCs.

However, an examination of the other two positions showed that the pineapples packed in RPCs

were subjected to LTA for longer periods of time, especially in the center of the pallet. Thus, it

seems that the number of vents in RPCs increased the level of penetration of low temperature air

in the pallets (Table A-14).

The uniformity of the incidence and duration of LTA in the Frontl/1A pallets was due to

the combined effect of short-circuiting in the truck trailer, by which the majority of top-delivered

cold air went through the pallets placed in the front of the trailer with high speeds and the bottom

air delivery system in the vessel, which affected mostly the bottom layers. Furthermore, the

exposure to LTA of the pallets placed in the center and back of the trailer took only place when

these pallets were inside the refrigerated holds. Therefore, the fruit located in the top and central

layers of the front pallets in the trailer were more prone to exposure to LTA during land

transport, while similar pallets in the other positions and all the packages placed in the bottom

layers of pallets were subject to LTA during sea transportation. Overall, exposure to LTA was

correlated with the position of the cargo in the truck trailer and the airflow pattern in it. Yet, LTA


















Top View


Front View


Probe of ThermaProbe RF
Environmental Readings: ThermAssure RF, ALB-2482, ILR i-Q32T
Core Reading: Body of ThermaProbe RF


Figure 3-3. Placement of the RFID temperature sensors in the experimental pallets.













TemperatureI History forJ.I00.'l Iu I gilll Im


Figure 3-4. Ambient temperature recordings obtained by the ALB-2484RFID tag placed in a
pallet with corrugated boxes.


Figure 3-5. Ambient temperature recordings obtained by two RFID tags placed in a pallet with
corrugated boxes: ThermAssure RF and ILR i-Q 32T.


-TherrAssure
RF
-I LR -Q 32T


5YM07 s5a7 50110?0 11o7 S12J07 131?07 5114107 ,15fI 7 1fiT07
Tme









produce, and in the cargo areas that harbored them along their supply chains. For example, Punt

and Huysamer (2005) discovered temperature differentials of about 4.5C in air temperature at

the pallet level in plum shipments.

Research performed by Billing et al (1993, 1995) and Smale (2004) also reported the

existence of temperature gradients within loads of produce during transportation in refrigerated

sea containers and vessels. Tanner and Amos (2003), found delivery air temperatures that varied

across the width of a container severely enough to result in air temperatures capable of freezing

the cargo. They also discovered significant time-based temperature variability during the

shipment, particularly between defrosts and in the time period corresponding to travel over the

equator.

In addition, studies in refrigerated trucks (Meffert and Van Nieuwenhuizen, 1973; Gogfis

and Yavuzturk, 1974; Lenker et al., 1985; Bennhamias, 1993; Le Blanc et al., 1994; Moureh et

al., 2002; Finn and Brennan, 2003; Tapsoba et al., 2006; and Moureh et al., 2009) have

discovered the existence of "hot spots" in the rear part of these vehicles created by poor

ventilation because of uneven air distribution and air short-circuiting around the front of the

truck.

Temperature Measurement Devices

There are many kinds of temperature sensors in the market. The easiest way to classify

them is either as contact or non-contact temperature measurement devices. Contact sensors

measure their own temperature; but they obtain the temperature of the obj ect they are in contact

with by assuming that the two are in thermal equilibrium and there is no heat flow between them.

Noncontact temperature sensors generally measure the thermal radiant power of the infrared or

optical radiation that they receive from a known or calculated area on its surface, or a known or









Cold rooms then become a suitable storage area before refrigerated transport operations. It

is recommended to precool the product and avoid loading it warm into refrigerated transport

systems. These systems do not possess the refrigeration capacity needed to bring the temperature

down to optimum transport conditions, and so, the product can be exposed to high temperature

abuse for long amounts of time (Thompson et al., 1998).

Storage facilities should remain around +1C of the desired temperature for the produce

stored. According to Kader (2002), temperatures below the optimal range for a given commodity

can cause freezing or chilling injury; while temperatures above it can shorten its shelf-life.

Transportation Systems

Temperature Control

One of the most important features for transportation systems used in a cold chain

management is their capacity to control temperature. There are many methods available to

maintain optimal temperatures in transport vehicles. Mechanical refrigeration, ice cooling, and

cryogenic cooling are the most common. Yet, the most widespread method worldwide is

mechanical refrigeration. This is used in road, rail, marine, and intermodal transport.

In mechanical refrigeration systems, it is very important to control frost accumulation on

the evaporator coils, for it reduces the cooling capacity of refrigeration units. Defrosting these

coils requires using electric heaters or hot refrigerant gas. During the defrost cycle, airflow is

stopped to prevent the movement of the heat produced in the coils towards the produce (Hui et

al., 2003).

Heat Loads

The refrigeration system used in a transport vehicle must remove all heat entering the

vehicle from the outside, all heat generated within the vehicle, and any heat contained in the

vehicle itself (Hui et al., 2003). For cooling capacity calculations, it is recommended using









dissertation, with regards to cold chain monitoring, RFID sensors posses ample advantages

compared to traditional methods of temperature tracking which will translate into cost reductions

and increased profits throughout the supply chain. Their fast and simple instrumentation and data

recovery allow the use of less manpower for these duties, reducing labor costs and delays in

packing lines and warehouse management procedures. In addition, because RFID temperature

sensors supply vital information about the product's temperature abuse, they will become a

powerful decision-making tool for suppliers and retailers, assisting in the application of the

concept of "First Expires First Out" (FEFO) in supply chain and logistic operations (Emond and

Nicometo, 2006).

This breakthrough idea will be put into practice because of the incorporation of shelf-life

modeling into RFID data processing software. Distribution centers all around the world will be

able to automatically retrieve the thermal history of each pallet entering their facilities, while at

the same time obtaining an accurate estimate of the remaining shelf-life of the products in those

pallets based on the temperatures they experienced until that point in the supply chain (Emond

and Nicometo, 2006). As a consequence, informed decisions will be made with regards to the

products future: The produce with longer shelf-life will be sent to distant locations or remain in

storage while the one with shorter shelf-life will either be destined to near-by stores or discarded,

according to the amount of shelf-life it still possess.

Furthermore, RFID temperature tags also provide something impossible to attain with

conventional temperature tracking methods: the reading and reprogramming of the sensors when

they are already into place. This is a point of high importance when considering the need for

real-time remote monitoring in food supply chains, which is becoming possible with the

combination of RFID and other wireless technologies. Once this is fully achieved and combined









with. Thank you to Pedro, Sharon and Lacey for understanding and loving me at home. To the

members of my Peruvian family in Gainesville, thank you for your love and encouragement. My

deepest thank you to Ceci and Guillermo for behaving as my "big siblings" and always having

the best advice available for me. Thank you to Mari, Dane, Betsi, Karol, Juanca, Isa, Marce,

Raul, Kathy, Milton, Jose, Daniel, Alonso, Wendy-Maria, Nadia, Chris, Alexis and Ingrid; for

always being by my side even though we are thousands of miles apart.

I would also like to thank Dr. Arthur Teixeira, for changing my life for the best ten years

ago; to him and his wife Marjorie, my most sincere gratitude for having opened your hearts to

me when I truly mostly needed it.

And last, but never least, I would like to thank God for his love and for all the many

blessings he has granted me. Thank you Lord for this unique opportunity and for giving me the

strength to complete it.


















Front View


Figure 3-1. Placement of the HOBO temperature sensors in the experimental pallets.


Figure 3-2. RFID tags used during the trial. From left to right: ThermaProbe RF, ALB-2484,
ThermAssure RF and ILR i-Q 32T.


Top View












Table A-6. Risk of fruit exposure to different high temperature abuse (HTA) scenarios according to the packaging used and the
position of the fruit in the pallet when using the refrigerated truck trailer/open hold combination.
Overall Top Center Bottom
Corr RPC Corr RPC Corr RPC Corr RPC


At HTA No. 1 (T> 120C)

AtHTANo.2 (T> 150C)

AtHTANo.3 (T> 170C)


H H

L H

L L


H H

H L

M N


M M

N N

N N


H = High, M = Medium, L = Low, and N = None.


Table A-7. Temperatures recorded in the experimental pallets at different stages of the supply chain when using the refrigerated truck
trailer/open hold combination.
Front/lA Middle/2A Back/2B
Top Center Bottom Top Center Bottom Top Center Bottom
Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC
Temperature at
beginning of
beginning of 27.6 27.91 27.06 28.7 27.06 28.9 27.91 27.6 26.8 26.52 DS 26.88 27.91 26.52 26.34 26.7 26.73 28.33
forced air
cooling (C)
Temperature at
the end of
14.15* 10.21 14.99* 8.58 13.14* 9.77 14.09* 10.11 14.8* 10.28 DS 9.6 18.66* 10.62 15.23* 9.6 12.93* 9.43
forced air
cooling (C)
Temperature
difference after
differenceafter 13.45 17.7 12.07 20.12 13.92 19.1 13.82 17.49 12 16.24 DS 17.28 9.25 15.9 11.11 17.1 13.8 18.9
forced air
cooling (C)
Max.temperatur
during 20.4 22.86 12.13 13.31 14.15 11.6 16 18.02 16.4 12.47 DS 13.48 16 18.87 12.16 11.5 12.55 12.81
transportation
(C)
Exposure to HTA. DS: Defective sensor









Estimation of the economic loss avoided by the load management system

The economic loss avoided by the proposed system will be the amount of money saved by

the DoD once this system is deployed. The calculations were done by multiplying the amount of

pallets that the system will allocate before expiration and prevent from becoming a loss with the

economic loss per pallet given its country of destination. Additional savings will derive from

eliminating the emergency shipments, since the resulting pallet loss levels after the

implementation of the system can easily be covered by the yearly FSR safety stock. Table 7-16

summarizes the results obtained. The aggregate for both items totals $11,761,779.72 per year;

and, consequently, the amount saved during the five years of its initial operation will reach

$58,808,898.63.

Return on Investment (ROI) Estimation

The ROI was determined by replacing the corresponding data in Equation 7-1. The current

discount rate given by the U.S. Federal Reserve (0.75%) was used for the calculations (Wall

Street Journal, 2010).



n Vt

ROI -=0 1 (7-1)
Vi
to (1+ D)t



Where Vi is the investment required by the project at the end of the time period t, Vt is the

monetary yield of the project at the end of time period t, n is the number of periods in the

analysis (5), t is the iterative time period; and D is the discount rate for the time value of money

(0.75%).









and inside one of the holds (2), on two decks (A and B). The temperature set point of the holds

was also established at 7.5 C. A schematic of the loading pattern can be seen in Figure A-3. The

sea container remained sealed and after the holds were closed, it was placed on the deck of the

vessel. Once fully loaded, the vessel was sent to Port Manatee (FL), where the sensors were

retrieved 3 days later.

Statistical Analysis

A Generalized Linear Model (GLM) procedure with a Gamma family was used in order to

determine the effects of time, packaging, location within the cargo, and transportation method

employed on the temperature within the pineapple loads (Freund and Wilson, 2003). The same

statistical procedure was also used to determine the equality of means of the temperatures

obtained with the different combinations of transportation method, location in the cargo area, and

packaging. Significance was accepted at level a=0.05. Statistical analyses were computed using

R 2.7.2 (The R Foundation for Statistical Computing., Wiedner Hauptstrabe 8-10/1071 1040

Vienna, Austria).

High Temperature Abuse (HTA) and Low Temperature Abuse (LTA) Risk Estimation

Risk levels were defined as the following percentages of positives from the total number of

temperature readings considered for the specific thermal abuse studied: "High risk" at values

equal to or more than 70% positives; "Medium risk" at values from 40%to less than 70%

positives; and "Low risk" at values of less than 40% positives.

High temperature abuse (HTA) analysis

Three intensity levels of exposure to high temperature abuse (HTA) were defined. These

were established based on the temperatures at which the respiration rate of a pineapple stored at

7 C (the optimal storage temperature for this variety of pineapple is 7.23 C or 45 F; Saenz and

D'Alolio, 2007) would approximately double (HTA No. 1: >12 C), triple (HTANo. 2: > 15 C),









6-3 Means and SDs of the combination of the 3 FSR meals at all temperature profiles. ...... 147

6-4 GLM of the acceptability scores for Profile 1 and its related profiles...................... 147

6-5 GLM of the acceptability scores for Profile 2 and its related profiles....................... 147

6-6 GLM of the acceptability scores for Profile 3 and its related profiles....................... 148

6-7 Means and SDs for each meal when combining 3 temperature profiles ........................ 148

7-1 Hardware and software costs for each work station. ................................................ 158

7-2 Y early costs of RFID tags .............. .... ..... .................................... ............... 159

7-3 Cost of the project during the first year of operation..................... ...................... 159

7-4 Cost of the project during the second and fourth year of operation................................. 159

7-5 Cost of the project during the third and fifth year of operation. ................................ 160

7-6 Y early co sts of th e project. ............................................ ................................................ 160

7-7 Distribution of U.S. troops in the zone of conflict studied.............................................. 160

7-8 Estimated number ofFSR eaten daily in the zone of conflict studied............................ 160

7-9 Estimated demand for FSRs in the zone of conflict studied. ............ ...... ............... 161

7-10 Actual demand for FSRs in the zone of conflict studied. .............................................. 161

7-11 E stim nations of yearly pallet losses. ..................................................................................... 16 1

7-12 Maritime shipping costs for FSR loads according to its destination............................... 161

7-13 Total cost of a lost pallet according to its final destination ............. ....... ............... 162

7-14 Amount of product sent as yearly emergency shipments........ ............ ................ 162

7-15 Amount of money destined to yearly emergency shipments..................... ................ 162

7-16 Yearly savings created by the proposed system ........................................ 162

A-1 GLM comparing the refrigerated truck trailer/open holds with a sea container ............ 186

A-2 GLM comparing three positions inside the sea container. .............. .............. 187

A-3 GLM comparing refrigerated holds inside a vessel. .................................................... 187

A-4 Percentage of time of HTA exposure in refrigerated truck trailer/open holds ................. 188










Table 5-7. Readability values for four different tag positions using the handheld reader.
Percentage of readings
Antenna Configuration ao Configuration 3
distance Point d Surface of wooden
Pot B) Surface of FSR load Surface of FSR load a
(PoR B) pallet
0.50 m 11.73% 39.50% 32.67% 41.66%
1.00 m 1.74% 19.32% 45.81% 63.52%
1.50 m 0.53% 1.33% 3.55% 58.48%
2.00 m 0.00% 25.64% 24.17% 94.85%
2.50 m 0.00% 0.00% 22.67% 90.73%
3.00 m 0.00% 0.00% 0.00% 98.19%


Environmental temperature vs. temperature at point 20-alpha
-En.,rjrimantal Lamc rature
-- T m' rl[., iur 1 al poirl 2 alpha


1000 2000 3000 4000 5000 6000 7000 8000
Time samples, one per five minutes

Figure 5-8. Temperature profiles showing how the temperature in PoR A (20-alpha) changed
with respect to the temperature outside the pallet in all eight trials.









tracking work either in the frequencies that can not work near products containing high amounts

of water (such as produce) or in frequencies that have short reading ranges, which makes them

unsuitable for portal applications in warehouses and distribution centers.

Another technical challenge is the use of multiple frequencies around the globe. Even

though organizations such as the International Association for Standardization (ISO) and EPC

Global are working towards creating standards promoting global parity for specific RFID

applications; the different regulations with respect to frequency ranges allocated for RFID

commercial purposes, vary according to the country and complicate global RFID ventures. For

example, UHF tags sent in a shipment from the U.S. (with a UHF frequency allocation between

902-928 MHz) will be able to be read in Europe (frequency allocation of 865-868 MHz) if used

EPC Global Gen-2 standards (which call for readers to be able to read along the entire UHF

spectrum) but will not be able to be reprogrammed and sent back with a shipment.

Cost is also one of the restrictive factors for RFID temperature tracking implementation.

The use of semi-passive, semi-active and active tags able to perform data collection and the

infrastructure necessary to perform the readings throughout the cold chain considerably surpass

the cost of conventional temperature monitoring methods in the market. Yet, the implementation

of protocols able to reduce the level of instrumentation, along with the imminent reduction in the

manufacturing costs of data-logging RFID tags, could increase the competitiveness of these

systems in the near future.

Passive RFID systems are mostly in use in the produce industry for product identification

and trace back due to the U.S. Department of Defense (DoD) and Wal-Mart mandates. These

systems are not as expensive as the ones employed for temperature tracking and are more likely

to be purchased by companies. However, the allocation of funding by the DoD to projects









APPENDIX D
TEMPERATURE PROFILE USED AS WORST CASE SCENARIO FOR FSR SHIPMENTS

Table D-1. Maximum daily temperatures obtained in a combat feeding shipment from the U.S. to
Kuwait.
Days Temperatures (OF)
1 104
2 104
3 91.4
4 107.6
5 82.4
6 73.4
7 73.4
8 96.8
9 89.6
10 91.4
11 91.4
12 91.4
13 87.8
14 86
15 86
16 86
17 87.8
18 89.6
19 95
20 96.8
21 102.2
22 102.2
23 102.2
24 102.2
25 100.4
26 96.8
27 100.4
28 100.4
29 118.4
30 114.8
31 120.2
32 129.2
33 122
34 125.6
35 131
36 122
37 122
38 125.6
39 129.2


209









DEVELOPMENT OF RADIO FREQUENCY IDENTIFICATION (RFID) TEMPERATURE
TRACKING SYSTEMS FOR FOOD SUPPLY CHAINS



















By

CECILIA R. AMADOR


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2010









Nunes, M.C.N., Emond, J.P., Rauth, M., Dea, S., Chau, K.V., 2009. Environmental conditions
encountered during typical consumer retail display affect fruit and vegetable quality and
waste. Postharvest Biology and Technology 51, 232-241.

Nunes, M.C.N., Proulx, E., Emond, J.P., 2003b. Quality characteristics of "Horn of Plenty" and
"Medallion" yellow summer squash as a function of storage temperature. Acta
Horticulturae 628, 607-614.

Omega, 2010. Technical Reference. Omega Engineering, Inc. Available at:
http://www.omega.com/techref/. Accessed June 28th, 2010.

Opara, L., 2003. Traceability in agriculture and food supply chain: a review of basic concepts,
technological implications, and future prospects. Journal of Food, Agriculture and
Environment, 1 (1), 101- 106.

Paull, R.E, 1993. Pineapple and papaya. In: Seymour, G., Taylor, J., Tucker, G. (Eds.),
Biochemistry of Fruit Ripening. Chapman & Hall, London, pp. 291-323.

Panos, R. C., Freed, T., 2007. The Benefits of Automatic Data Collection in the Fresh Produce
Supply Chain. Paper Presented at the 3rd Annual IEEE Conference on Automation Science
and Engineering, Scottsdale, AZ.

Peleg, K., 1985. Produce Handling, Packaging and Distribution. Avi Publishing Company, Inc.,
Westport.

Perez-Aloe, R., Valverde, J.M., Lara, A., Carrillo, J.M., Roa, I., Gonzalez, J., 2007. Application
of RFID tags for the overall traceability of products in cheese industries. Paper Presented
at 2007 RFID Eurasia, Istanbul, Turkey.

Potdar, V., Hayati, P., Chang, E., 2007. Improving RFID Read Rate Reliability by a Systematic
Error Detection Approach. Paper Presented at 2007 RFID Eurasia, Istanbul, Turkey.

Potter, N.N., Hotchkiss, J.H., 1998. Food Science. Springer-Verlag, New York.

Produce Marketing Association., 2008. Traceability Frequent Asked Questions. Available at
www.pma.com/viewdocument.cfm?doclD=221. Accessed on October 20th, 2008.

Proulx, E., Nunes, M.C.N., Emond, J.P., 2005. Quality attributes limiting papaya postharvest life
at chilling and non-chilling temperatures. Proceedings of the Florida State Hort. Soc. 118,
389-395.

Punt, H., Huysamer, M., 2005. Temperature variances in a 12 m integral reefer container
carrying plums under a dual temperature shipping regime. Acta Horticulturae 687, 289-
296.

Raza, N., Bradshaw, V., Hague, M., 1999. Applications of RFID Technology. Paper Presented at
the IEEE Colloquium on RFID Technology, London, UK.


218









spoilage and pathogenic microorganisms in their meals (Ross et al., 1985; Ross et al., 1987;

Banwart, 1989; Narayan et al., 1997; Ross et al., 1997; Shaw et al., 1997; Ng et al., 2002;

Natress et al., 2009). Consequently, high temperature abuse will diminish the shelf-life of FSRs

and, in some cases, significantly.

Since the majority of current U.S. military operations takes place in desert areas where

high temperatures are reached and the loads of FSRs do not move along environmentally

controlled supply chains, there are many opportunities for high temperature abuse. In the

previous chapter the use of RFID temperature tracking systems as a way of monitoring the

conditions of these loads and detecting thermal abuse was proposed. However, during fast-paced

military deployments, the data collected in these could be difficult to decipher; since the

temperature profiles recorded are by no means a stand alone indicative of the remaining shelf-life

of the product. Therefore, a load management system that uses a software tool able to decode

such data, providing intelligible information about the shelf-life status of the product and giving

recommendations on the actions required with respect to the load is needed.

The main step in the creation of such software is the selection of an appropriate model for

the prediction of the FSR's shelf-life. There have been a number of studies on modeling the

shelf-life of food products. Many of them are based in variations of the Arrhenius equation,

which describes the kinetics of simple chemical reactions as dependant on temperature

conditions (Hertog et al., 1999; Brody, 2003; Giannakourou and Taoukis, 2003; Hough et al.,

2006). Also popular is the use of predictive microbiology models, such as the Gompertz

equation, as a mean of determining the deterioration and safety thresholds limiting shelf-life

(Labuza and Fu, 1993; Zwietering et al., 1993; McMeekin and Ross, 1996; Riva et al., 2001;

Corbo et al., 2006). Less common are empirical models, which are built upon experimental data









In addition, there was significant variability in the temperature profiles of crownless

pineapples shipped from the Pacific coast of Costa Rica to the U.S.. The lack of consistency in

the environmental and physical conditions during handling could result in the delivery of loads

of pineapples with variable quality, which could not only limit the usage of the fruit, but also the

shelf-life and the final quality of the fresh-cut products.

Prior research performed by Billing et al (1993, 1995), Smale (2004), and Punt and

Huysamer (2005) already reported the existence of temperature gradients within loads of produce

during transportation in refrigerated sea containers and vessels. While these studies aimed to

quantify the magnitude of the gradients, they did not provide an overall statistical comparison of

the temperature management throughout the whole supply chain.

The lack of significant differences in temperature profiles at the bottom layer of the pallets

with regards to the transportation method employed could be explained by the fact that both sea

containers and open holds use bottom-air delivery systems. However, it would also indicate that

the effect of the temperature management during the land transportation period, when truck

trailers with top-air delivery systems were used, was minimal. Furthermore, it suggests that the

thermal effect of the exposure to ambient conditions of the same set of pallets during the process

of loading and unloading the vessel was not a significant factor in determining the temperature

profiles.

The lack of significant differences in temperature profiles at the bottom of the pallets with

respect to placement inside the container could suggest the existence of uniform airflow along

the length of the container. Even though this observation argues against the occurrence of any air

short-circuiting phenomenon inside the container;_in order to obtain a conclusion on this matter,









High Temperature Abuse (HTA) and Low Temperature Abuse (LTA) Risk
E stim atio n ................... ....................................................... ................ 1 7 3
High temperature abuse (HTA) analysis .................................... ............... 173
Refrigerated truck trailer/open holds ............... ................................................. 174
S ea co n ta in er ...................................................................................... 17 8
Low temperature abuse (LTA) analysis............................. ............... 180
Refrigerated truck trailer/open holds ............... ................................................. 181
S e a c o n ta in e rs ............................................................................................................ 1 8 2
C o n clu sio n s .... ............... ............................................... .......... ...... 183
F u rth er w ork .... ................. ........................................... ........... ...... 184

B RECOMMENDED LEVEL OF RFID INSTRUMENTATION FOR THE
CROWNLESS PINEAPPLE SUPPLY CHAIN WHEN THIS USES THE
REFRIGERATED TRUCK/OPEN HOLD TRANSPORTATION METHOD .................... 197

C AMBIENT TEMPERATURE PROFILES FOR THE THERMAL RELEVANCE
STUDY OF A PALLET OF SPHERICAL BOTTLES OF WATER.................................. 198

D TEMPERATURE PROFILE USED AS WORST CASE SCENARIO FOR FSR
SHIPMENT S .......... ..... ......................... ........ 209

L IST O F R E FE R E N C E S ...................................................................................... .....................2 10

BIOGRAPHICAL SKETCH ........... .... ....... ..................... ........................ 223





























10









Table 5-5. Maximum readability values for both Points of Relevance (PoR) in Configuration
f using the fixed system.
Antenna distance Position PoR read Percentage of readings
0.50 m e A 69.50%
0.50 m c B 47.80%
1.00 m e A 52.40%
1.00 m c B 69.40%
1.50m e A 45.00%
1.50m c B 78.40%
2.00 m e A 70.50%
2.00 m c B 88.10%
2.50 m e A 28.50%
2.50 m c B 68.80%
3.00 m e A 25.60%
3.00 m c B 74.90%


Table 5-6. Readability values using the fixed system of positions "b" and "c" throughout the
pallet.
Configuration Position Percentage of readings
a b 82.00%
a c 62.60%
P b 32.60%
3 c 74.90%























Configuration a-I


Figure 4-2. Top view of the depth of tag placement for each configuration.


Table 4-1. RFID reader configuration.
Power Q TX RX Forward
Protocol
(dBm) value Ant Ant link
C3 30.00 4.00 1.00 1.00 8.00 kbps


Reverse
lin e Inventory BS type
8.00 kbps portal fsk
8.00 kbps portal fsk


Table 4-2. Average temperature differentials present at the pallet level.
Exposure time Average temperature Standard deviation
(hours) differential
0.50/0.50 11.95 2.90
1.00/1.00 13.01 3.30
2.00/2.00 16.68 4.10
4.00/4.00 20.07 5.03




Table 4-3. Average temperature differentials present at the RPC level.
Exposure time Average temperature Standard deviation
(hours) differential


0.50/0.50
1.00/1.00
2.00/2.00
4.00/4.00


7.15
7.54
8.85
10.71


0.98
1.30
2.01
2.60


fC X X 3C
-3 T CS-X~r '-r)rCr~


Configuration p









Table C-4.
Time
1547
1550
1552
1555
1557
1600
1602
1605
1607
1610
1612
1615
1617
1620
1622
1625
1627
1630
1632
1635
1637
1640
1642
1645
1647
1650
1652
1655
1657
1700
1702
1705
1707
1710
1712
1715
1717
1720
1722
1725
1727
1730
1732


Continued.
Ambient temperatures (oC)
10.83
10.81
10.78
10.78
10.73
10.74
10.71
10.68
10.66
10.64
10.61
10.56
10.49
10.46
10.40
10.33
10.30
10.21
10.17
10.10
10.05
9.98
9.91
9.84
9.76
9.73
9.61
9.54
9.47
9.39
9.34
9.26
9.19
9.14
9.08
9.00
8.94
8.85
8.81
8.77
8.79
8.83
8.84


207










CHAPTER 4
EVALUATION OF SENSOR READABILITY AND THERMAL RELEVANCE FOR RFID
TEMPERATURE TRACKING

Introduction

Technological advances have enabled precise traceability and visibility mechanisms within

temperature controlled supply chains. The existence of these technologies is driving a

requirement for precise traceability in perishable goods chains, such as the food and

pharmaceutical supply chains (Bollen, 2005; Ames, 2006; Metzger et al., 2007; Panos and Freed,

2007; Zhen-hua et al., 2007; Jedermann et al., 2008). RFID technology has been suggested as a

solution to address this need since it can store both pedigree and temperature/time information

and transmit it even as the product is in transit (Forcinio, 2004; Smith, 2005; Wyld, 2006; Leake,

2007). As discussed in the previous chapter, as a temperature tracking method, RFID has shown

superior performance when compared to traditional monitoring devices since it can surpass the

simplicity problems that these encounter during real-life fast-paced supply chains, while

maintaining their levels of accuracy. In spite of the promises it holds, the implementation of

RFID in this particular area is still in its infancy and, therefore, still presents some challenges.

A major hurdle for RFID's application in food supply chains is that, based upon the

frequency used, certain environmental factors, such as water and the presence of metal, affect the

RF signal (Dobkin and Weigand, 2005; Redemske and Fletcher, 2005; Gaukler and Seifert,

2007; Hartvanyi and Marek, 2007; Sivakumar and Deavours, 2008). Current RFID systems for

temperature tracking present in the market operate either in the frequencies with reduced

readability near loads of perishable products with high water content such as produce (915 MHz,

868 MHz), or in frequencies that have short reading ranges (13.56 MHz), which makes them

unsuitable for portal applications in warehouses and distribution centers.





















35 -


30






20

a) Forced Air Cooling
b) Refrigeration
c) Loading
d) Land Transportation
10 e) Warehouse/Port
f) Sea Transportation

5
5o l --------------------------------


0 0 0 0 0
o o 0
o o 0 0 0
0) 0) 0) 0) 0)
U) 0 (0 ^- C
o ^- CMi 0
o o 0 0 0
o o 0 0 -
U) ) ) U U
o o 0


0 0 0 0
0 0 0 0

I 6) (0
--00






TI ME


0 0 0 0 0 0 0 0 0 0 0


0 0 0- 0- 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0
N Co r 0 )- (N Coo 0 0o
^- Ci 0 C'i 0 0 0
o o o o o ooo o o o
U) UM Un UN UN U) U N U) U) U) U)
0 0 0 0 0 0


Figure A-4. Ambient air temperature profile inside the sea container displaying the three heating

episodes along the supply chain: 1) During land transportation 2) During the process

of loading the containers onto the vessel 3) During sea transportation.


a b c d


e f









temperatures below their threshold, which damages membranes and promotes cellular

breakdown (Nunes et al., 2003b). CI creates a wide variety of symptoms, though some of them

might not manifest themselves until the product is exposed to higher temperatures (Shewfelt and

Prussia, 1993; Thompson, 2003). Another physiological disorder, freezing injury takes place

when the commodity is overcooled and ice crystals form inside its cells (Hui et al., 2003;

Thompson, 2003). According to Thompson (2003), in order to avoid these disorders, produce

handling and storage should take place at temperatures just over their threshold temperatures for

chilling and freezing injury. Lastly, high-temperature injury can be the product of either long

exposure to temperatures within the 30-40C range, or to short exposure to even higher

temperatures (Shewfelt and Prussia, 1993).

In conclusion, proper temperature management along food and produce supply chains will

delay the deterioration of the product, extending its shelf-life and maintaining its economic value

along the chain.

The Pineapple Supply Chain

Pineapple is the second most important tropical fruit in the world. Around 21 million tons

of pineapples were produced worldwide in 2007. Twelve percent of these were devoted

exclusively to the international fresh market, a business of more than 2 USD billion dollars

(FAO, 2009). Central America (Costa Rica, Panama) and the Philippines are the biggest sources

of fruit; while the US is the top importer, having reached its levels of consumption to 2.16 kg of

fresh pineapple per capital in 2007 (USDA, 2010).

International commerce of fresh pineapples requires highly efficient temperature controlled

supply chains. Due to the sensitivity of this product to chilling injury, temperatures must remain

within a specific range at all times (70C to 120C), avoiding exposures to temperatures lower than

the products' threshold (Abdullah et al., 2000; Acedo et al., 2004). However, exposure to higher









wavelength (Dagdelen, 2007; Reed, 2009). Circularly polarized antennas can have a clockwise

rotation or a counter clockwise one. The main benefit of these antennas is that they are largely

unaffected by tag orientation. Circular polarized antennas can read tags in a wider area compared

to linear polarized antennas because they have a wider radiation beam (Lahiri, 2006).

Information Processing Software

In a RFID system software is critical for filtering and analyzing data as well as integrating

RFID technology into backed databases (Thyagaraja, 2007). According to Banks et al. (2007),

middleware is the software component that transforms low level RFID hardware information

coming from the readers to usable event information. Hunt et al. (2007) states that middleware

plays an important role in the quality and usability of the information produced by RFID systems

because it allows the transmission of data between the RFID networks and the IT systems within

an organization. According to Clarke et al. (2006), legacy systems can be used to make

instantaneous decisions based on the incoming data from the RF readers; for example, triggering

replenishment when an order causes inventory to drop to the reorder point.

RFID Applications

Some of the current applications of RFID are: Supply chain automation, inventory

management, parcel and postal tracking, access control, airport luggage, self check-out, and

medical ID bracelets, automatic toll collection, retail parking access, retail stock management,

library book tracking, vehicle immobilization, theft prevention systems, livestock tracking,

environmental monitoring, and tracking exact timing in sports events (Griffin et al., 2005; Potdar

et al., 2007).

Applications of RFID in the Produce Industry

RFID offers three major applications suited for the produce packing industry:

Identification, Trace Back and Process System Traceability.












Table A-2. GLM results for the effects of position, time step and packaging when comparing the
fruit temperatures (C) recorded at the bottom of the pallets during transportation in
the three positions (back, middle and front) inside the sea container.
Estimate Std. error t Value P
(Intercept) 8.765e-02 1.157e-03 75.782 <2e-16


Position

Time step

Packaging

Position:
Time step


-1.705e-05


5.466e-04


2.366e-05 1.086e-06

5.395e-03 4.133e-04


-1.019e-06


5.555e-07


-0.031

21.789

13.054

-1.834


0.9751

<2e-16

<2e-16

0.0667


Table A-3. GLM results for the effects of the hold, time step and packaging when comparing the
fruit temperatures (C) recorded at the top of the pallets during transportation in the
holds of the refrigerated vessel.
Estimate Std. error t Value P
(Intercept) 1.104e-01 1.781e-03 61.958 <2e-16


Hold


Time step

Packaging

Hold: Time
step


-1.057e-02 6.990e-04

-2.289e-05 1.417e-06

-9.425e-05 5.735e-04

1.193e-05 6.571e-07


-15.121

-16.156

-0.164

18.159


<2e-16

<2e-16

0.87

<2e-16









typical modern middle size reefer ship will have a capacity of approximately 12,000 m3, divided

into around 4 holds, 8 temperature zones and 12 to 16 cargo chambers (Stera, 1999). The main

difference between using containers or refrigerated ships is their cargo-carrying capacity: A

container can carry about 1,000 to 1,500 packages, while a refrigerated ship has capacity of

about 350,000 packages (Kader, 2002). Travel times, however, are generally longer, often

around 1 to 4 weeks; but produce being shipped overseas may spend on board even 5 to 6 weeks.

Thus, good temperature and humidity control is essential.

Thompson (2003) states that produce for exportation is generally transported in

temperature-controlled cargo space be it in "break bulk" (conventional refrigerated ships) or in

containers or "reefers" (container ships).

Conventional ships

These are completely insulated vessels and have a series of holds; each one divided into

three to five cargo areas. Generally, each cargo compartment will have its own refrigeration coil

and fresh air ventilation with independent temperature control (Heap et al., 1998). These ships

can carry refrigerated goods, frozen foods, and nonperishable items in different holds or rooms at

the same time. They mostly use the break bulk system (Hui et al., 2003).

Break bulk refers to a system of transport where individual boxes or pallets of produce are

stacked directly in the hold of the ship. Yet, most produce has been palletized before shipping

and remains palletized along the supply chain. Thompson (2003) mentions that the main benefits

of palletization are the reduction of handling which then reduces labor costs and damage to the

product.

The method of stacking cargo in the hold is critical for the maintenance of optimal

temperatures. Spaces must be left between rows, stacks, or layers, so that the upward circulating

air can reach one or more surfaces of each shipping container (Peleg, 1985; Hui et al., 2003).









configuration the tag placed in PoR B was never detected and only the tag in C3 (PoR C in this

side) was read at antenna distances of 1.00 m, 2.50 m and 5.00 m.

The highest level of readability found in configuration a-I was 7.90% ( 0.01, standard

deviation) and belonged to PoR A at a reading distance of 0.50 m. Its counterpart in side 3 was

point C3 at a reading distance of 5.00 m with 95.59% ( 0.02, standard deviation) of readability.

Tables 4-5 and 4-6 summarize the average readability values for the PoRs detected in each

configuration at different reading ranges.

The results obtained can be explained in part by the findings of Fletcher et al. (2005) who

analyzed the signal propagation of 915 MHz waves through water. Their research indicates that

the strength of the RF signal depends not only on the amount of water it has to go through, but

also on the period of the sinusoidal wave and the point on it inherent to each water depth. Under

their analysis the period of this wave lasts approximately 0.04 m; and therefore, the lowest signal

strength will be located in amounts of depth of water multiple of four. In the current study, point

B, which is located in the middle of the 3 side, had no readability and point A, on the a-I side,

obtained some readability at two antenna distances. In the first case the RF wave had to

propagate through 0.12 m of water, where the signal will decrease at its maximum; while in point

A the signal had to go through 0.14 m of water, the middle point of the period between 0.12 m

and 0.16 m, at the peak of the RF wave. Therefore, it is logical to expect higher readability

during this particular point than at 0.12 m of depth. It can be then said that the bottle orientation

and the side configuration it determined influenced the level of readability of the tags.

Side configuration seemed to have played a less relevant role in the case of point C3,

where according to the previous discussion readability should have been hindered. A possible

explanation for this behaviour lies with the fact that there were no bottles of water on one of the









than the fruit in corrugated cartons. This fact is in agreement with previous research on the

influence of packaging in cooling operations for pineapples, papayas and mangoes

(Chonhenchob and Singh, 2003a; Chonhenchob and Singh, 2003b; Chonhenchob et al., 2008).

A good example of the importance of adequate pre-cooling was the fruit in corrugated

cartons in the center of the pallet that were shipped in the Middle/2A combination (Table A-7).

In this case, the pineapples were exposed to temperatures that would presumably have, on

average, doubled the respiratory metabolism during the entirety of the voyage. Furthermore, the

level of heat accumulation was such that during more than 25% of the shipping time, the fruit

reached temperatures that would presumably have tripled the respiration rate, accelerating the

pineapple fruit metabolism and its quality deterioration (Mohamed and Wickham, 1995).

The effect of position within the cargo environment in HTA was not entirely clear when

using the current quantitative analysis as a basis. A detailed review of the temperature profiles of

the pallets allowed the determination of the thermal behavior of the load in the two stages studied

(data not shown). In the first stage, during land transportation by refrigerated truck, the

temperatures obtained along the length of the cargo area were strongly correlated with the

expected trend given the top air delivery system airflow pattern. Regardless of the packaging

used, the pallets at the front of the trailer experienced colder temperatures than the pallets placed

in the middle and in the back of the trailer. Even though, in the case ofRPCs, the temperatures

recorded in middle and back positions inside the refrigerated trailer had similar values, for

corrugated cartons the temperature differential between positions was approximately 2 C higher

in the back pallet than in the middle pallet, at almost all times. Earlier studies (Meffert and Van

Nieuwenhuizen, 1973; Goguis and Yavuzturk, 1974; Lenker et al., 1985; Bennahmias and

Labonne, 1993; Le Blanc et al., 1994; Finn and Brennan, 2003; and Moureh et al., 2009) have









pomegranate, passion fruit or tangerines. Further research is recommended in order to determine

if the thermally relevant locations identified in this work could also be used for products with

storage temperatures different than 4.000C (tomatoes, limes, etc).

Readability Study

The three locations found as thermally relevant in the previous study ("PoR") were tested

for readability. Special considerations were taken since a sensor placed in any of the sides has

the risk of not being read by the system if it is used as the entry port for the forklift. This will

occur because of the strong interaction of the RF waves emitted by the tag and the metal of the

forklift, which might hinder the communication process with the antennas. In addition, the

forklift could also shield the signal from the antenna to the tags, thereby, preventing

communication with them. Even though it would be feasible to mark the specific side where the

tag is present, this would complicate the handling of the pallet at the receiving end, where the

RFID system will be in place. Since point C is located in the bottom corner of the lowest row of

product on both of the 3 sides (varying from left to right corner according to the side chosen), it

was decided that both these locations and the nearest ones in the a-I side (where the corner is

completed) needed to be tested for RFID readability in case reading the sides of the pallet was

unsuccessful because of forklift interaction. Tags were then placed in these positions, aiming

towards the corresponding face being read, in the first internal layer of all sides (Figures 4-4, 4-5

and 4-6).

In configuration a-I two out of the three tags tested were read, while in configuration 3

only one of the two tags set up was detected. In the first configuration, the tag placed in the PoR

A was read at antenna distances of 0.50 m and 1.00 m. In addition, the tag corresponding to PoR

C placed on the left of this side (Cl) was only found with a 2.50 m reading distance. Finally, in









CHAPTER 6
DEVELOPMENT OF A LOAD MANAGEMENT SYSTEM FOR COMBAT FEEDING
LOGISTICS BASED ON SHELF-LIFE PREDICTION SOFTWARE

Introduction

Combat feeding during military campaigns has represented a challenge for armies

throughout history. The lack of refrigeration and the possibility of limited food supply in the

field have driven some of the biggest developments in food conservation and processing; the

most remarkable being canning. During the last fifty years, technological efforts have been

aimed towards developing safe and nutritious foods able to withstand the rigorous aspects of

military supply chains. And so, in 1993 the U.S. Department of Defense (DoD) created a

successful set of meal-based shelf-stable products named Meals, Ready-to-Eat (MRE) able to

defy most of the time and temperature limitations existent in food distribution and storage, and

suitable for their logistics.

Sometime after their introduction, feedback from the field indicated that soldiers were

opening the MRE packs, and only taking with them and eating certain elements of these in order

to reduce the amount of weight carried. This not only generated big amounts of food waste, but

also endangered the nutritional status of the troops. A new shelf-stable product called First Strike

Rations (FSR) was then designed to take on the place of MREs in assault operations. With it,

Warfighters could bring a lightweight load with enough food to fulfill their nutritional

requirements for a whole day during fast-paced maneuvers (DoD Live, 2009).

A First Strike Ration (FSR), although shelf-stable, will still deteriorate faster if its storage

temperature of 26.67C has been exceeded (DoD Live, 2009). Exposure to high temperatures

accelerates negative changes in the meals' sensory attributes, which influence its level of

acceptability. In addition, storage of FSRs under these environmental conditions could affect

adversely the health status of the troops, by promoting nutritional changes, and the growth of









Vigneault, C., Goyette, B., Markarian, N.R., Hui, C.K.P., C6te, S., Charles, M.T., Emond, J.P.,
2004. Plastic container opening area for optimum hydrocooling. Canadian Biosystems
Engineering 46, 3.41.

Vigneault, C., Wu, S., Thompson, J.F., Hui, K.P.C., Leblanc, D.L., 2009. Transportation of
Fresh Horticultural Produce. In: Benkeblia, N., Shiomi, N. (Eds.), Postharvest
Technologies for Horticultural Crops, Vol. 2. Research Signpost, Kerala, pp. 1-24.

Wall Street Journal, 2010. Money Rates. Dow Jones & Company, Inc. Available at:
http://online.wsj.com/mdc/public/page/2_3020-moneyrate.html. Accessed July 7th, 2010.

Wang, J.K., Tunpun, K., 1968. Forced-air cooling of tomatoes in cartons. Transactions of the
American Society of Agricultural Engineers 12, 804-806.

Wang, N., Zhang, N., Wang, M., 2006. Wireless sensors in agriculture and food industry-Recent
development and future perspective. Computers and Electronics in Agriculture 50 (1), 1-
14.

Want, R., 2006. An introduction to RFID technology. Pervasive Computing (January/March).

Watts, A., Miller, P., Godwin, D., 2002. Applications of RFID (Radio Frequency Identification)
in Agriculture. Pesticide Outlook (December).

Wikipedia, 2010. Bimetallic Strip. Wikimedia Foundation, Inc. Available at:
http://en.wikipedia.org/wiki/Bi-metallic strip#Applications. Accessed June 28th, 2010.

Wilcke, C., 1992. Model for predicting storability and internal quality of apples. Acta
Horticulturae 313, 115-124.

Wyld, D., 2006. RFID: The Right Frequency for Government. The IBM Center for the Business
of Government. Washington, D.C.

Zhen-hua, D., Jin-tao, L., Bo, F., 2007. Radio Frequency Identification in Food Supervision.
Paper Presented at the 9th International Conference on Advanced Communication
Technology, Gangwon-Do, South Korea.

Zwietering, M. H., Wijtzes, T., De Wit, J. C., Van't Riet, K., 1993. A decision support system
for prediction of microbial spoilage in foods. Journal of Industrial Microbiology and
Biotechnology 12 (3-5), 324-329.


222









Systems operating in the high UHF range (915 MHz in the US, 868 MHz in Europe) are highly

affected by the presence of metals and liquids in their surroundings (Leong et al., 2006);

nevertheless, low UHF frequencies such as 315 MHz and 433 MHz perform better under the

same conditions. The UHF range is not accepted worldwide; each country has restrictions with

respect to which UHF frequencies are allowed within their territory.

Microwave frequency

Microwave frequency ranges upward from 1 GHz. RFID systems operate either at 2.45

GHz or 5.8 GHz. They also use electromagnetic coupling as communication method. They

possess the fastest data-transfer rates between the tag and the reader (reaching up to the Mbit/s

range); and are the most affected with respect to the existence of metals or liquids in their

surroundings. The 2.4 GHz frequency band belongs to the ISM band, and therefore it is accepted

worldwide (Lahiri, 2006).

Materials Hindering the Signal

According to Lahiri (2006), materials can be grouped within three classes with respect to

the way they affect the radio waves propagating through them (Table 2-3). RF-lucent materials

to a certain frequency allow radio waves at this frequency pass through it without any substantial

loss of energy. RF-opaque materials block, reflect, and scatter RF waves. And finally, RF-

absorbent materials allow the radio waves to propagate through them but with significant loss of

energy.

The RF-absorbent or RF-opaque property of a material depends on the operating

frequency. So, a material that is RF-absorbent at a certain frequency could be RF-lucent at a

different frequency. For example, water and metal are RF-lucent at low frequencies while at the

highest ones (UHF, Microwave) they turn into RF-absorbent and RF-opaque materials,

respectively.









differentials could be greater, because the temperatures inside this will be considerably lower

than the ones registered by the weather services. Likewise, these temperature differentials could

decrease if the load is being transported by sea, and protected from solar radiation. From a heat

transfer perspective, the worst case scenario for a temperature tracking system similar to the one

being designed is when the temperatures in the pallets are quite variable, due to a high

temperature differential between the pallet and the environment. The temperature differential of

95.00C used in these trials represents then, an extreme situation very rare to find under normal

circumstances; but that gives a safety factor even for the outlier observations encountered in real-

life situations. Consequently, it is expected that under normal supply chain conditions, the

likelihood of gathering 85.00% of the temperature readings present in the pallet within the

16.000C range proposed when using the PoRs found as middle points of this interval will

increase.

The use of a 16.000C temperature range as a reference for quality and shelf-life

assessments along this supply chain will not be a major concern, because FSRs are shelf-stable

products. However, if a component of the ration has a characteristic particularly sensitive to heat

or cold (degradation of vitamins, color and texture changes, etc) that might happen within that

range under ordinary storage and transport conditions; then, a higher level of quality variability

in the pallet is expected, and the application of this system could derive in the incorrect

acceptance of a faulty load. As well, a stronger level of sampling and quality control of the load

before acceptance should be established when exposure to high temperatures (over 26.670C) is

recorded. Mainly, because the 15.00% of temperatures inside the pallet that are not represented

by this temperature tracking system might promote pathogen growth, and create food safety

hazards for the troops.









Thermistors (Thermal Resistors or Bulk Semiconductor Sensors)

According to Michalski et al. (2001) thermistors are non-linear, temperature dependent

resistors with a high resistance temperature coefficient. Only thermistors with a negative

temperature coefficient are used for temperature measurements. Unlike an RTD's metal probe

where the resistance increases with temperature, the thermistor uses ceramic semiconducting

materials which respond inversely with temperature (Efunda, 2010). The resistance is measured

by passing a small, measured direct current (dc) through it and measuring the voltage drop

produced (Temperatures, 2010).

Typical thermistor sensors can measure temperatures across the range of-40 150

+0.350C (-40 ~ 302 0.630F). A thermistor probe can take the shape of a bead, washer, disk, or

rod. Typical operation resistances are in the kM range, although the actual resistance may range

from several MQ to several Q (Efunda, 2010).

Radiation Thermometers (Pyrometers)

A Pyrometer, is a non-contact temperature sensor that detects an object's surface

temperature by measuring the temperature of the electromagnetic radiation (infrared or visible)

emitted from the object (Efunda, 2010). This group of sensors includes both spot measuring

devices in addition to line measuring radiation thermometers, which produce one-dimensional

and can sometimes produce two-dimensional temperature distributions, and thermal imaging, or

area measuring, thermometers which measure over an area from which the resulting image can

be displayed as a two-dimensional temperature map of the region viewed (Temperatures, 2010).

According to Efunda (2010) pyrometers manipulate the fact that all objects above absolute

zero temperature 0K (-273.150C; -459.670F) radiate and absorb thermal energy. Then, if the

relationship between the radiation intensity and wavelength and the temperature can be









which one of its items deteriorated faster. These would then be considered the shelf-life limiting

items (SLLI) for the pallet. The following were the results of this preliminary testing (Nunes et

al., 2010, unpublished data): In Menu 1, the SLLI was the Bacon Cheddar Sandwich; in Menu 2,

the Italian Style Sandwich; and in Menu 3, the Zapplesauce.

These products were then subjected to controlled storage at two different temperatures

(48.890C or 120.000F, and 60.000C or 140.000F). Since the easiest way to determine the shelf-

life of the product during a real-life military deployment is the level of acceptability for

consumption, this attribute would then be considered in the model as the limiting quality factor

in the shelf-life of FSRs. Thus, sensory analysis was performed on each product/temperature

combination at different storage times. The scale for the acceptability evaluations was from 1 to

9; being 1 the worst acceptability score and 9 the best acceptability score. A score of 4 was

considered the limit value of acceptability fit for consumption.

Curves were then created when plotting the acceptability for consumption versus the weeks

of storage. For each temperature, the resulting curve was fitted into a third order polynomial

equation.

A third curve was built for all products in order to describe the deterioration behavior of

these at their ideal storage temperature of 26.67C (80.000F). Due to the fact that similar storage

tests under this temperature were not finished by the time of the completion of this study, it was

assumed that each product started with an initial quality equal to the average initial quality they

presented for the first two temperatures considered. It was also assumed that the limit of

acceptability of consumption for all products (a score equal to 4.00) was reached at 104.00

weeks (2.00 years) of storage under 26.670C (80.000F). Given that only the initial and final

acceptability values were given, the resulting curves were only first order polynomial curves.










Table 3-4. Results of the Generalized Linear Model for the Effects of the Sensor, Time Step and
Packaging when comparing the temperatures (C) obtained with the ThermAssure RF
and the ILR i-Q 32T RFID tags recording ambient temperatures on pallets with
corrugated boxes or reusable plastic containers in the sea container.


(Intercept)

Sensor


Estimate
6.07e-02


Std. Error
9.20e-04


2.75e-03 5.23e-04


Time Step 2.07e-05

Packaging 7.75e-03


Sensor:
Time Step


8.53e-07

2.93e-04


2.75e-06 5.48e-07


t Value
65.96

5.25

24.20

26.41

5.01


P
< 2e-16

1.55e-07

< 2e-16

< 2e-16

5.47e-07


25


20
S
15 -ThernAssure RF
S--TherraProbe RF







0
5W7 597 5110 07 51107 5f12/07 5R1307 5114/07 51S07 5/11607
Tmie


Figure 3-6. Temperature profiles obtained by RFID tags with probe and without probe in the
central layer of a pallet with RPCs.









(Verboven et al., 2005) and there is a broad variance in the speed of temperature changes,

depending on the transport conditions (Jedermann and Lang, 2007)

Once an optimal set-point temperature for transportation has been established, achieving

proper temperature management depends on several factors. Packaging design and materials, for

example, can hasten or hinder the heat transfer process, and dramatically affect the temperature

distribution inside the pallets and throughout the cargo environments (Smale, 2004).

With produce, respiratory heat generation can pose problems, causing the cartons in the

centre of a pallet to be significantly warmer than those on the outside of the pallet (Tanner and

Amos, 2003). Such temperature variability can lead to variable quality at out-turn, which in turn

can lead to marketing problems. But packaging design provides a relatively simple means by

which these processes can be controlled, so packaging becomes a major factor in determining the

rate of heat and mass transfer during refrigerated storage and transport of horticultural products.

The traditional approach to dealing with respiratory heat has been to design cartons with open

vents; however the vents are usually designed to facilitate horizontal airflow despite vertical

airflow being the standard configuration in transport systems (Smale, 2004).

In the majority of food refrigeration systems, heat is transferred primarily by convection;

therefore, the temperature and its homogeneity are directly governed by the patterns of airflow.

Most of the time, non-uniform airflow is one of the major causes of temperature variability. For

sensitive products, this level of temperature variability may have significant food quality and

safety implications (Smale et al., 2006).

Previous research (Billing et al., 1993; Billing et al., 1995; Amos, 2001; Tanner and Amos,

2003; Moureh and Flick, 2004; Punt and Huysamer, 2005; Rodriguez-Bermejo et al., 2007) has

described the existence of spatial and time-based temperature variability both inside pallets of









CHAPTER 7
RETURN ON INVESTMENT (ROI) DETERMINATION FOR THE DEPLOYMENT OF A
RFID-BASED LOAD MANAGEMENT SYSTEM IN COMBAT FEEDING LOGISTICS

Introduction

The particularities of dealing with perishable inventory have been studied for many years

(Nahmias, 1982; Goyal and Giri, 2001; Bogataj et al., 2005; Ferguson and Ketzenberg, 2006).

RFID has been recently suggested as a new tool for obtaining operational efficiencies in the

supply chains of perishables as a mean of obtaining stock level transparency and improving the

application of inventory retrieval policies such as "First In First Out" (FIFO) and "Last In First

Out" (LIFO) (Karkkainen, 2003; Chande et al., 2005; Broekmeulen and van Donselaar, 2007).

However, the need for combining this information with the thermal history of the load due to

safety and quality concerns (Giannakourou and Taoukis, 2003; Sahin et al., 2007) opened the

door for the application of sensor-enabled RFID tags and the possibility of sorting the products

based on the amount of thermal abuse suffered during the supply chain and its impact in product

quality (Ilic et al., 2009).

Consequently, it has been suggested empowering RFID temperature tracking systems with

shelf-life prediction models in order to improve sorting at the distribution center and store level

(Emond and Nicometo, 2006; Jedermann et al., 2008; Ilic et al., 2009). Based on the shelf-life

information collected for each pallet or load, FIFO and LIFO could be replaced by a "First

Expired First Out" (FEFO) retrieval policy, that reduces the economic loss and quality and safety

concerns generated by handling expired product in the supply chain.

In the previous chapters, a relevant and feasible instrumentation protocol for RFID

environmental monitoring of First Strike Rations (FSR) logistics was created; and the resulting

RFID temperature tracking system has been combined with shelf-life models in order to build a

load management system. The aim of this system is achieving a FEFO management of the FSR









threshold score of four. An optimization could be made in order to determine the particular

amount of days of shelf-life (within 14.00 and 21.00), but it was considered more appropriate, to

facilitate the load handling in-situ, giving the recommendations to the Warfighters in weeks.

It is important to remark that the temperature profile employed was the worst case scenario

in the set of shipments studied previously by the DoD. Within these shipments there was a high

level of variability in the temperature profiles recorded depending on the position of the

container on the ship and the position of the sensor inside the container. So, even under similar

environmental conditions, the level of variability in temperature profiles was wide and thus,

possibly the acceptability scores of the load and its expected shelf-life.

In addition, these shelf-life results could differ considerably according to the weather

patterns. For example, if in real-life the temperatures increase more than the ones used in this

simulation; then, the product will probably have less acceptability scores than the ones obtained,

and could also reach its acceptability threshold after just one or two weeks of storage at the

deployment area. Since the temperature profiles provided by the DoD correspond to operations

in the months of September and October, it is expected that the shelf-life of the load gets reduced

during the months of July and August, at the peak of the summer. Likewise, if the ambient

temperatures are lower, the amount of shelf-life of the product could be extended, particularly

during the winter and fall months.

Therefore, given the high level of variability in temperature profiles and environmental

conditions, it is important to perform this kind of shelf-life analysis for every single shipment

received throughout the year.

Differences between Acceptability Estimates for FSR Meals

Table 6-3 shows the averages and standard deviations of the three meals combined for each

one of the ambient temperature profiles and its related profiles. As can be seen, standard










Table 3-1. Comparison of the features of the RFID temperature tracking systems analyzed and a traditional temperature monitoring
system (HOBO Sensor).
ILR ThermAssure ThermaProbe ALB- HOBO
Features
Features i-Q RF RF 2484 Sensor
32T
Programming time 19.09 21.50 s + 21.50 s + 47.68 s 53.71 s +
S5 2.43 2.43 + 6.36 3.92
(Mean SD) 3.15

Detects all sensors at
Yes No No Yes No
once

User-friendly software Yes Yes Yes No Yes

Allows distant
interaction between the
Yes No No Yes No
reader or computer and
the sensor

Requires placing a No No Yes No Yes
No No Yes No Yes
probe in the load

Easy to handle and affix
No Yes Yes No No
to the packaging

Software facilitates data
management and Yes Yes Yes No Yes
analysis









packaging must be designed to facilitate both horizontal air flow (for forced air cooling and

refrigerated trailers) and vertical air flow (for sea containers and refrigerated cargo vessel holds).

Stowage patterns must also accommodate the air flow design of the transport equipment being

used. The existence of spatial and time-based temperature variability, both inside pallets of

produce and in the cargo areas that harbored them along their supply chains, has been described

in previous research reports (Billing et al., 1993; Billing et al., 1995; Amos, 2001; Tanner and

Amos, 2003; Moureh and Flick, 2004; Punt and Huysamer, 2005; Rodriguez-Bermejo et al.,

2007). Since deviations like these could potentially increase the risk of temperature abuse on the

load, any attempt to optimize a produce cold chain will require that these factors and their

possible interactions be taken into account.

The objectives of this research were to determine if there are significant differences in the

temperature profiles of pineapples handled using different combinations of packaging and

transportation methods, or among different locations in the cargo; and to estimate the risk of

exposure to high temperature abuse (HTA) and the development of chilling injury due to

exposure to low temperature abuse (LTA) in the fruit among these different parameters, based on

duration of exposure to deleterious temperatures and thermal history. It is expected that, as a

result of this study, the produce industry will be able to recognize which one of the

package/transportation method combinations synergize when trying to design an efficient

crownless pineapple supply chain that will secure high-quality produce.

Materials and Methods

Experimental Design

At a pineapple packinghouse in the Pacific coast region of Costa Rica, two samples of

primary packages of crownless pineapples were instrumented with temperature sensors before

palletizing. Both samples consisted of 9 primary packages containing six 'MD-2' pineapples









frequencies, a design trade-off. Lower frequency devices have less read range, and slower

communication speed, but good penetrating ability (Watts et al., 2002).

Following is the description of the main characteristics of different RFID systems

associated to specific frequency bands:

Low frequency (LF)

LF ranges from 30 KHz and 300 KHz. A typical LF RFID system operates at 125 KHz or

134.2 KHz. These systems use inductive coupling as communication method. LF waves perform

very well in environments containing metals, liquids, dirt, snow, or mud (Lahiri, 2006; Roussos,

2006). However, according to Dagdelen (2007) and Leong et al. (2006) they have longer read

times (with data rates, on the order of Kbits/s), higher costs, and larger tag sizes. Nonetheless,

one of its advantages is that this range is accepted worldwide.

High frequency (HF)

These systems use frequencies ranging from 3 MHz to 30 MHz. The ISM band commonly

used in HF RFID systems is 13.56 MHz. HF also uses inductive coupling as communication

method. HF waves can penetrate through water, yet metallic objects are still problematic

(Dagdelen, 2007). In addition, they present longer read times compared to other systems. The HF

frequency range is also accepted worldwide (Lahiri, 2006).

Ultra high frequency (UHF)

The UHF band ranges from 300 MHz to 1 GHz. According to Dagdelen (2007), in this

frequency range RFID systems communicate through electromagnetic coupling backscatteringg,

for example). The UHF frequency band has a number of advantages. This frequency range

provides fast reading rates and offers longer reading distances than other RFID frequencies

(Sivakumar and Deavours, 2008). In addition, greater transmitter power is permitted at UHF

frequencies than at microwave frequencies, such as 2.4 GHz (Redemske and Fletcher, 2005).















A


Y


Figure 5-6. Locations of the Points of Relevance (PoR) detected in the pallet.


Table 5-3. Likelihood of gathering 85.00% of the temperature readings in the 16.000C range
proposed when using the Points of Relevance (PoR) found as the interval's middle
points.
Episode Point A Point B


6.00 h, 6.00 h
8.00 h, 8.00 h
9.00 h, 15.00 h
18.00 h, 18.00 h
1.00 d, 1.00 d
2.00 d, 2.00 d
3.00 d, 3.00 d
4.00 d, 4.00 d


72.58%
41.24%
63.41%
32.15%
23.87%
49.89%
42.32%
58.07%


85.23%
35.00%
74.78%
48.58%
32.45%
44.76%
32.27%
43.11%


.. "f









extreme environmental circumstances encountered along the supply chain. According to

Vigneault et al. (2009) the estimations of total cooling capacity can be simplified by adding all

heat loads and multiplying the result by a safety factor. In addition, transportation of produce in

cold weathers requires the use of heating systems to prevent chilling and freezing injuries.

Internal heat loads

These include respiratory heat generated by the produce and any field heat that remains

within the produce at the beginning of the transportation process. If produce is not adequately

precooled or if it has gained heat from loading areas, then its internal heat load is larger.

External heat loads

The external heat loads include all the heat that enters the vehicle from the outside. This

can do so through conduction, convection, air infiltration, as well as by radiation (Vigneault et

al., 2009). Heat is conducted through the floor, walls, and ceiling of the vehicle; while warm air

infiltrates into it through small holes, cracks, drainage holes, broken door seals, and when the

doors are opened unnecessarily. Infiltration is by far the most copious source of external heat

load in refrigerated trailers. Finally, solar radiation also has an effect on internal temperatures.

For example, studies have found that the cooling requirements of stationary vehicles increased

by 20% after exposure to sunlight for several hours (Hui et al., 1993).

Residual heat loads

These include any heat initially contained in the transport vehicle or any heat load not

included as internal or external loads. The most common source is the heat present in the air and

surfaces inside the transport vehicle; as well as any remaining heat in boxes, pallets and devices

used to secure the load (Vigneault et al., 2009).









Table A-10. Risk of fruit exposure to different high temperature abuse (HTA) scenarios according to the type of packaging used and
the position of the fruit in the pallet when using sea containers.
Overall Top Center Bottom
Corr RPC Corr RPC Corr RPC Corr RPC
At HTA No. 1
T 120C H M H H H M N N
(T > 12'C)
At HTA No.2
At 15N2 M M H H N N N N
(T > 15'C)
At HTA No.3
At HT L M M H N N N N
(T > 17HC)
H = High, M = Medium, L = Low, and N = None.









Metal is an electromagnetic reflector and radio signals cannot penetrate it (Roussos, 2006).

This situation prevents the tag from absorbing enough energy from the reader, since the metallic

material holds most of it, therefore detuning the operating resonant frequency at which the tag

was supposed to operate (Banks et al., 2007; Reed, 2009). As a result, metal will not only block

communication if placed between a tag and an interrogator; but it can also affect the operation

and characteristics of the antennas located nearby (Hunt et al., 2007). A proposed solution for

tagging metallic items is to either use antennas with precise tuning or to employ tags mounted on

a foam platform, which avoids the direct contact with the metal (Banks et al., 2007).

Liquid materials such as soap, water, or salty solutions are RF-absorbent in the UHF and

Microwave frequency ranges, and diminish the functionality of the RFID systems (Leong et al.,

2006; Roussos, 2006; Aroor and Deavours, 2007; Banks et al., 2007; Sivakumar and Deavours,

2008). These materials absorb considerable amounts of the energy of the radio waves arriving to

the transponder, thus reducing the available energy necessary to operate the tag. If there is severe

absorption, the transponder will not be able to capture the necessary energy to operate and the

reader will not detect its signal (Banks et al., 2007; Reed, 2009).

Banks et al. (2007) also states that the differences in the impact of RF-absorbent or RF-

opaque objects on HF and UHF derive from the ratio of the wavelength to the object size. The

author also proposes a classification of the object based on its dimensions relative to the

wavelength:

Rayleigh range. The wavelength is much larger than the object dimensions. So, for objects

smaller than half the wavelength, the reflection is minimal and metal would not interfere in the

normal operation of a tag.


































2010 Cecilia R. Amador









Antenna polarization

Propagation is the direction and path that a radio wave travels, and polarization describes

the electric field vector along the propagation path (Reed, 2009). Dagdelen (2007) defines it as

the orientation of the electric field of the wave. Antenna polarization and the angle at which the

tag is presented to the reader greatly impact the readability of a tag, its reading distance and the

reading robustness of the system (Lahiri, 2006). Most antennas radiate via either linear or

circular polarization.

According to Reed (2009), antennas are designed to have a particular polarization. Using

the analogy of a thread on a nut and bolt, the author explains how antennas and RF waves must

match to work together in wireless communications. For example, an antenna with right-hand

circular polarization would have difficulty receiving a reverse-polarized signal. Consequently,

optimized antenna orientation promotes the best wireless communication performance.

Linear polarized antenna

Linear polarization takes place when the electric field is aligned in a single plane (Reed,

2009). These waves have only one energy field and the antenna's radiation pattern remains in the

same plane at all times. Dagdelen (2007) explains that linear polarization can be vertically or

horizontally oriented depending on the propagation direction with respect to the earth surface.

Thus, a linear polarized antenna will be sensitive to tag orientation with respect to its polarization

direction. Finally, when compared to circular polarized antennas, linear polarized ones present a

narrower radiation beam with a longer read range (Lahiri, 2006).

Circular polarized antenna

A circularly polarized antenna radiates an electric field which is the sum of two fields that

have equal amplitude and magnitude with a 900 phase difference (Dagdelen, 2007). These two

components form a spiraling field vector, which rotates and completes a full 3600 within one














Instrumented
packet





Figure 5-1. Three-dimensional view of the FSR packets instrumented in each box.


Figure 5-2. Pallet and antenna position during the readability study at an antenna distance of 2 m.
Figure 5-2. Pallet and antenna position during the readability study at an antenna distance of 2 m.


Table 5-1. RFID reader configuration.
Power


Forward link
8.00 kbps


Reverse link
8.00 kbps


Protocol
C3


(dBm)
30.00


Q value
4.00


TX Ant
1.00


RX Ant
1.00









Refrigerated ships are loaded in open docks and expose produce to the elements (heat,

freezing temperatures, or rain). This is especially problematic in ships that use a common

refrigeration system for two compartment levels, because the refrigeration system is not operated

until both compartments are loaded (Kader, 2002).

Smale (2004) explains that there are two common systems of air distribution in reefer

holds: The longitudinal air delivery system and the Robson system. Yet, with respect to their

interaction with the load, their effect is similar since both of them provide bottom-air delivery

systems in the cargo area.

Due to their exposure to the elements during loading, it is expected that produce

transported in cargo holds regain some of the temperature lost during precooling operations.

Hence, the ship must have sufficient refrigeration capacity and air circulation to lower the

temperature of the cargo (Hui et al., 2003).

Reefer holds are less affected by infiltration heat and to the frosting of their evaporator

coils when compared to reefers. However, both of these systems share the need for renewal of

refrigerated air as a mean to avoid the build-up of respiratory products in produce shipments

(Smale, 2004).

Container ships

Container ships have specially designed holds that have vertical guides to stack containers

below the deck. These can be stacked six to nine high below deck, and three to four high above

deck. Protection from the sun to those carried on deck may be given with an upper layer of non-

refrigerated containers. One of the main benefits of this type of ship is that they require minimal

handling of the container's cargo; and as such, give fewer opportunities to break cold chain when

compared to the break-bulk system. Therefore, the quality of the produce transported in reefers at









LIST OF ABBREVIATIONS

DoD United States of America Department of Defense

EPC Electronic Product Code

FSR First Strike Rations

HF high frequency

HTA high temperature abuse

ISO International Standards Organization

LF low frequency

LTA low temperature abuse

MRE Meal, Ready-to-eat

PoR point of relevance

RF radio frequency

RFID radio frequency identification

RPC reusable plastic container

SLLI shelf-life limiting item

UHF ultra-high frequency









illustrates the end point of the internal layers in the configurations studied. The RFID tags were

set along the red lines.

Equipment configuration

In order to maximize the read range of the Class 3 tags inside the pallet, reader power was

set at its maximum and air interface data rates were set at their minimum. Since our research was

targeted towards portal applications, the reader's inventory configuration was set in "portal".

Table 4-1 summarizes the reader's configuration.

Once set up, the system was run in three occasions; in each one, about 100 attempts of

communicating with each one of the tags were made. The level of readability of the tags was

determined by averaging the percentage of times a tag was read with respect to the attempts

made of contacting it.

Results and Discussion

Relevance Study

Temperature distribution

The temperature differentials were quite significant at the pallet level and sometimes also

at the RPC level. In addition, the magnitude of these differentials increased with the duration of

the heating/cooling episode. For example, for the longest one (4.00 hours/4.00 hours), the

temperature deltas were an average of 20.000C ( 5.000C, standard deviation) at the pallet level

and an average of 10.700C ( 2.600C, standard deviation) at the RPC level. Tables 4-2 and 4-3

summarize the average temperature differentials present at the pallet and RPC level, respectively.

As expected, the hottest temperatures were obtained near the top layer and the external

walls of the pallet; where exposure to solar radiation and environmental conditions was direct.

The coldest temperatures were recorded near the center of the pallet and at the bottom. Similar

trends in pallet temperature distribution had been previously reported by Jedermann and Lang














Table A-i 1. Temperatures recorded in the experimental pallets at different stages of the supply chain when using sea containers.

Front Middle Back
Top Center Bottom Top Center Bottom Top Center Bottom
Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC

Temperature at
beginning of forced 28.15 26.7 25.95 26.9 25.95 26 DS 26.34 26.34 26.7 26.34 26.3 27.78 25.56 26.34 25.56 25.95 25.6
air cooling ( C)

Temperature at the
end of forced air 11.29 9.94 12.55* 9.6 10.6 9.94 DS 10.11 13.32* 10.79 10.99 8.75 12.47* 9.03 13.7* 10.21 10.21 8.63
cooling ( C)

Temperature
difference after
difference after 16.86 16.76 13.4 17.3 15.35 16 DS 16.23 13.02 15.91 15.35 17.6 15.31 16.53 12.64 15.35 15.74 16.9
forced air cooling
(C)

Max.temperature
during 16.5 17.35 12.16 11.8 10.99 11 DS 18.36 12.16 12.13 10.6 10.5 17.01 17.14 12.55 12.93 10.99 11.4
transportation ( C)

Exposure to HTA. DS: Defective sensor









is due to the fact that it was observed that the speed of reading of the systems was inversely

proportional to the reading distance, and that obtaining readings at 2.00 m and more took a

considerable amount of time. Since the use of a handheld reader during unloading operations will

require its operator to move constantly towards the pallets and back, in order to avoid the

upcoming forklifts, high speed readings are desirable. Consequently, antenna distances of 1.50 m

are recommended as a way of combining the tag read velocity with the minimization of the

distance the operator will have to walk to read each pallet.

The temperatures obtained by the tag in this handheld system will later become the input

data for the temperature profile estimator; and after being processed, will give origin the

temperatures obtained in PoR B.

Temperature Profile Estimator

The temperatures in the Points of Relevance A and B (labeled also as 20-alpha and 36-

prime in the figures) showed a capacitor effect to rapidly changing temperature by slow

heating/cooling cycles (Figures 5-8 and 5-9). Therefore, the temperatures rose and decayed with

a time constant that can be determined from the eight trials from the relevance study, and can

subsequently be modeled to estimate the temperature in the PoRs given the ambient temperature.

The model was the applied by first estimating the time constants for the heating and

cooling processes for each trial. As an example, the calculations corresponding to the constant

for the 4.00 days heating period were performed for PoR A (Trising).

In this experiment, the average ambient temperature sat at 60.500C. Ift=0.00 is defined as

the time PoR A's temperature started to increase from 24.600C, then a second temperature point-

time pair can be chosen to calculate the time constant. For this example, at t=76.00 hours the

temperature of PoR A reached 60.00C, and this point was then used in the calculations below

where T represents temperature whereas t represents time.










APPENDIX A
ANALYSIS OF THE VARIABILITY AND COLD CHAIN PERFORMANCE FOR
CROWNLESS PINEAPPLE WITH RESPECT TO TRANSPORTATION METHODS,
LOCATION WITHIN THE CARGO, AND PACKAGING

Introduction

Pineapple is the second most important tropical fruit in the world. Around 21 million tons

of pineapples were produced worldwide in 2007 (FAO, 2009); 12% of these were devoted

exclusively to the international fresh market, a business of more than 2 billion USD. Central

America (i.e., Costa Rica and Panama) and the Philippines are the biggest sources of exported

fruit; while the US is the top importer, having reached a level of consumption in 2007 of 2.16 kg

of fresh pineapple per capital (USDA, 2010).

International commerce of fresh pineapples requires highly efficient temperature controlled

supply chains. Due to the sensitivity of this product to chilling injury, temperatures must remain

within a specific range at all times, avoiding exposures to temperatures lower than the products'

threshold (Abdullah et al., 2000; Acedo et al., 2004). However, exposure to higher temperatures

than those recommended for pineapple storage (70C to 120C; Paull, 1993) will accelerate the

senescence and decay rates of the load (Mohammed and Wickham, 1995). Poor temperature

management during pineapple shipments will then result in postharvest losses and in poor

product quality; which generates lower customer satisfaction and impacts the produce companies

with economic losses and lack of public credibility (Nunes, 2008; Machado et al., 2009; Nunes et

al., 2009).

During recent years, the fresh pineapple trade has been positively affected by the use of the

fruit by the fresh-cut industry, which can import either pineapples with crown (i.e., bracts on the

fruit) or crownless (Gonzalez-Aguilar et al., 2004). Crownless pineapples facilitate fresh-cut

operations; however, their postharvest management becomes more challenging, since the









trailers because these are not airtight enough; yet, modified atmospheres can be achieved when

using semi-permeable films around the packages or pallet (Kader, 2002).

Overland transport by railcars

The main advantages of overland rail transport are fast service to distant points and better

efficiency in terms of diesel fuel per ton-kilometer; also, the lack of traffic (Peleg, 1985). They

are mostly used to transport potatoes, citrus fruits, onions, carrots, and other lesser perishable

commodities (Kader, 2002). According to Hui et al. (2003), four different vehicles can be used

for rail transport: The ice-refrigerated railcar, the mechanically refrigerated railcar, the

refrigerated semitrailer on flatcar ("piggy-back"), and the intermodal container on flatcar.

In mechanically refrigerated railcars the cold air is distributed vertically downward through

the load from the ceiling or from wall flues on side walls and the far end of the car. In both

systems, warm air returns to the refrigeration unit through the floor ducts (Huit et al., 2003).

Their refrigeration systems use an electric motor that is powered by a diesel generator which can

be detached and allow the connection of the railcar to electricity (Kader, 2002).

Apart from refrigeration, heating equipment may also be required for winter transport of

fresh produce in cold climates. Insulated standard boxcars and artificial heating can be used to

protect from chilling and freezing injuries (Peleg, 1985).

These rail cars are suitable for modified atmosphere transportation. However, since they

are very airtight, unintended atmospheric modification can take place in them if drain vents are

clogged or water in them freezes in the winter (Kader, 2002).

Marine transport

This method is generally selected for intercontinental shipments since it is the most

economical mode of transportation over long distances and the most energy efficient one too,

especially when large volume is involved (Peleg, 1985; Heap et al., 1998; Hui et al., 2003). A









TpR A(t=0) = 24.60C

Tambient = 60.50C

TpoR A(t=76.00 hours) = 60.00C

Thus,

-76
Trising = (60.S = 17.8 hours
n60.5 -24.6

Similarly, to find Tfalling, is only needed to define two time-temperature points where the

temperature ofPoR A slowly approaches the ambient temperature.

TpoR A(t=0) = 60.50C

Tambient = -35.00C

TpoR A(t=67.50 hours)= -33.00C

-67.5
Tfaling = (33) = 17.5 hours
In(35 --(-33)
-35 60.5)

The fact that the falling time constant was higher shows that the pallet cooled down faster

than it heated up in this trial. Based on these time constants, estimating the

temperature inside the pallet can be modeled by the following two equations:

If TAmbient > (T-1)PoR, then


TpoR = ([(T l)]PoR + (TAmbient (T 1)PoR ) (1-e( s (5-3)

If TAmbient < (T-1)PoR, then


TPoR = ([(T 1)]PoR + (TAmbient (T 1)PoR) (1-e(t lng))(5-4)

Where TpoR is the current estimated temperature in the PoR and (T-1)poR denotes the

previously estimated temperature sample.









even raise the internal temperature of the load. Yet in most cases, the first two factors are the

most relevant, particularly when the product exposes a large surface to the air flow. In addition,

the temperature difference between the product and the cooling air, as well as the velocity of the

air passing through the products are the main elements influencing the heat transfer rate from the

product to the air stream (Dincer, 2003).

The efficiency of forced-air cooling is determined by process time and product temperature

uniformity. Ventilated packaging favors rapid and uniform cooling. The airflow inside the

packaging is a strong factor in the heat transfer taking place. Effective venting is then necessary

to maximize cooling efficiency. Tutar et al., 2009 mentions that a compromise between container

structure and venting areas should be reached. Thus, packaging integrity has to exist even though

openings large enough to avoid hampering airflow are distributed along the bottom and the walls

of the container. According to Wang and Tunpun (1968) and Mitchell et al. (1971) boxes should

have about 5% sidewall vent area; nonetheless, Dincer (2003) recommends a minimum of 6% of

the total face area of a carton on the incoming air side as acceptable.

The main advantages of using forced-air cooling are its simplicity, economy, sanitation,

and the fact that it is relatively noncorrosive to the equipment (Dincer, 2003). Its main

disadvantage is that it causes some moisture loss during cooling which could be significant for

produce with a low transpiration coefficient. This moisture loss is correlated to the difference

between initial and final product temperatures and can be reduced at the expense of longer

cooling times by wrapping product in plastic or packing it in bags (Thompson et al., 2002).

Hydrocooling

Cooling is accomplished by moving cold water around produce with a shower system or

by immersing produce directly in cold water (Thompson et al., 1998). These can be flow-through

or batch systems (Dincer, 2003). The produce can be cooled in bins or be in bulk before packing









the RTD element with temperature. Most RTD elements consist of a length of fine coiled wire

wrapped around a ceramic or glass core. The element is usually quite fragile, so it is often placed

inside a sheathed probe to protect it. The RTD element is made from a pure material whose

resistance at various temperatures has been documented and has a predictable change in

resistance as the temperature changes (Omega, 2010). They are active devices requiring an

electrical current to produce a voltage drop across the sensor that can be then measured by a

calibrated read-out device (Temperatures, 2010).

The resistance of commercially available RTDs ranges from 10 to 25,000 Q. The most

common are 100, 200, and 1000 Q strain-free platinum (>99.999%) probes and 10 Q copper

probes. On a general basis, the higher the resistance, the less affected the RTD will be due to

small resistance/voltage fluctuations in the lead wires and circuit. Some of the common metals

used in RTDs include platinum, copper, nickel, BalcoTM (70% Ni-30% Fe), and tungsten

(Efunda, 2010). Table 2-2 lists their temperature ranges.

Platinum RTDs (also called PRTs and PRT100s) are the most popular RTD type. These are

nearly linear over a wide range of temperatures and some are small enough to have response

times of a fraction of a second. PRTs are among the most precise temperature sensors available

with resolution and measurement uncertainties or +0.1 C or better (Temperatures, 2010).

RTDs are usually encapsulated in probes for temperature sensing and measurement with an

external indicator, controller or transmitter, or enclosed inside other devices where they measure

temperature as a part of the device's function, such as a temperature controllers or precision

thermostats (Temperatures, 2010).









Smale, 2004). Some reefer containers are suitable for modified atmospheres and controlled

atmospheres usage during produce transport (Thompson, 2003).

Intermodal transport

Intermodal freight transport allows the movement of goods in a cargo unit by successive modes

of transport with no handling of the goods themselves during changes in transport modes.

According to Ruiz-Garcia (2008), intermodality is characterized by the transferability of the

transported items between modes and a unique system of administration and billing.

Intermodality is important because it promotes continuity in the cold chain. Additionally, it

allows handling the product without breaking up the cargo, which results in reduced shipping

costs and less damage to the produce (Hui et al., 2003).

Sea containers are the most common type of equipment used in intermodal transport. There

are two general size classes: 20 ft (6.1 m) and 40 ft (12.2 m). Specifications can be found in the

ISO standard 668 (Smale, 2004).

Airfreight transport

This method is generally used to carry, over long distances, produce that has a short shelf-

life and therefore cannot withstand the trip through road, rail, or sea. Air transport is more

expensive than other transportation methods; however, it gives a quicker return on the capital

that is tied up during transportation (Hui et al., 2003). According to Kader (2002) air

transportation provides poor temperature control when compared to the other transportation

methods. In addition, great concern is paid to the high environmental costs and the carbon

footprint related to transporting produce via air cargo (Thompson, 2003).

Cargo is generally stacked in unit load devices (ULDs) before the time of loading the

aircraft. The two main categories of ULDs are aircraft pallets and aircraft containers. Intermodal









Although present, the effect of pallet position inside the cargo area on pineapple thermal

handling was mostly relevant in the case of exposure to LTA in sea containers. It was also

determined that whenever inappropriate temperature control existed during transportation

corrugated cartons offered more protection from temperature abuse than RPCs due to slower heat

transfer with the cartons. Moreover, the refrigerated truck trailer/open hold combination

increased the intensity (peaks and dips) of exposure to HTA and LTA, with a large volume of

product being affected, while transporting the fruit in the sea container increased the duration of

the exposures to HTA and LTA. Thus, the use of corrugated cartons and refrigerated sea

containers for handling and transportation of crownless pineapples from Costa Rica to U.S. can

be recommended as this combination minimized the amount of product exposed to HTA and

LTA.

Further work

A comprehensive quality study should be completed in order to determine the extent of the

effect of the exposure to HTA and LTA documented in this study on pineapple shelf life and

overall quality, and to determine whether there is an intermittent warming protection effect with

regard to chilling injury when using RPCs in refrigerated truck trailer/open hold transportation.

In addition, corrugated carton should be redesigned in order to reduce the HTA exposure due to

inefficient pre-cooling that was documented here.









Table C-2.
Time
1127
1130
1132
1135
1137
1140
1142


Continued.
Ambient temperatures (oC)
5.73
5.72
5.34
5.37
5.53
5.30
5.35


200









APPENDIX C
AMBIENT TEMPERATURE PROFILES FOR THE THERMAL RELEVANCE STUDY OF A
PALLET OF SPHERICAL BOTTLES OF WATER



Table C-1. Ambient temperature profiles registered in the 0.5 h/0.5 h heating and cooling
episodes.
Time Ambient temperatures (C)
942 26.16
945 27.15
947 27.13
950 26.82
952 28.25
955 27.64
957 27.54
1000 30.08
1002 28.84
1005 29.39
1007 31.02
1010 30.64
1012 32.92
1015 22.90
1017 18.64
1020 9.57
1022 8.30
1025 9.51
1027 11.17
1030 11.24
1032 10.58
1035 11.14
1037 9.91
1040 8.17









Table C-4. Ambient temperature profiles registered in the 4 h/4 h heating and cooling episodes.
Time Ambient temperatures (C)
1022 7.20
1025 7.50
1027 7.61
1030 7.68
1032 7.67
1035 7.75
1037 7.82
1040 7.80
1042 7.77
1045 7.74
1047 7.65
1050 7.60
1052 7.58
1055 7.60
1057 7.67
1100 7.69
1102 7.69
1105 7.61
1107 7.59
1110 7.55
1112 7.58
1115 7.65
1117 7.68
1120 7.74
1122 7.69
1125 7.66
1127 7.61
1130 7.54
1132 7.53
1135 7.56
1137 7.61
1140 7.69
1142 7.70
1145 7.76
1147 7.80
1150 7.83
1152 7.91
1155 8.10
1157 8.17
1200 8.22
1202 8.23
1205 8.28
1207 8.33
1210 8.35


204









involving temperature tracking, such as those described in this dissertation, indicate a tendency

leaning towards the combination of ID applications and environmental monitoring along the

supply chain, which could possibly lead to new mandates for its suppliers (including the produce

industry) involving the use of this type of technology in the near future.











Table 6-6. GLM analysis of the acceptability scores for Profile 3 and its related profiles.
Estimate Std. error t Value P
(Intercept) 3.46 0.36 9.43 1.31e-05

Sensor location 0.02 0.17 0.10 0.92

Meal 0.08 0.13 0.57 0.58

Sensor location:
-0.00 0.06 -0.04 0.97
Meal







Table 6-7. Averages and standard deviations for each meal when combining the three
temperature profiles analyzed.
Profile 1 Profile 2 Profile 3
Meal 1 Meal 2 Meal 3 Meal 1 Meal 2 Meal 3 Meal 1 Meal 2 Meal 3
Average 5.52 5.58 5.35 4.55 4.71 4.37 3.39 3.88 3.75


Standard
deviation 0.02
deviation


0.01 0.02 0.01 0.01


0.02


0.02









Results and Discussion

RFID Sensor Performance versus Conventional Methods

The major difference noticed within both kinds of sensors was the time required for the

instrumentation of the load (Table 3-1). The RFID systems allowed a faster setup because of the

little time taken while programming the sensor and their ease of placement in the load.

The main factors in the speed of programming of the RFID systems were the amount of

tags read at once by the readers and the user-friendliness of the software employed. The fastest

was the ILR i-Q 32T system (19.09 s + 3.15; mean SD) which, because of its active nature,

allowed the distant reading of all the tags at the same time, facilitating a remote activation

process. This tag was followed by the ThermAssure RF and ThermaProbe RF sensors (21.50 s +

2.43; mean SD), semi-passive RFID tags using the same software and hardware, which

required a closer interaction with the readers and could only be read one by one. The user-

friendliness of the software employed was decisive in the programming time of the semi-passive

ALB-2484 (47.68 s 6.36; mean SD) tags, which, unlike the others, had management software

with poor interface and took the longest to be activated even though all the tags were able to be

read at once.

On the other hand, HOBO sensors required a lengthier period of time during programming

(53.71 s + 3.92; mean SD) and had to be wired to the computer which made impossible any

sensor re-programming or reading once the pallets were instrumented.

As far as the ease of setup in the load, ThermaProbe RF tags were as difficult to handle as

HOBO sensors because their probe also needed to be placed in a specific location inside the

pallet. All other RFID sensors were quite easy to place; however, amongst all, some of them

were more convenient than others. For example, due to their small size and thickness (0.05 m x

0.05 m x 0.04 m) and light weight (6 g), handling the ThermAssure RF and ThermaProbe RF









the port of destination should be better than when this is handled by the break-bulk system in

conventional refrigerated ships (Hui et al., 2003; Smale, 2004).

Mechanically refrigerated containers or reefers are insulated containers usually equipped

with a recessed refrigeration unit (Thompson, 2003; Hui et al., 2003). Many of these run on a

diesel engine during highway or rail travel and then are plugged on shipboard to operate with an

electric motor. Reefers are generally carried on deck to ensure that there is sufficient air to

remove the heat rejected by the refrigeration unit's condenser (Hui et al., 2003).

Air circulations rates both in vessels and in reefers are high right after loading the cargo as

part of the initial cool-down. Once the product is cooled to the optimal transport temperature and

the refrigerated air has been properly distributed along the cargo space, the system switches to

lower air circulation rates in order to reduce refrigeration load and power consumption (Smale,

2004).

The refrigeration capacity of a reefer depends on ambient temperature, insulation quality,

respiration rate of the cargo, initial temperature of the cargo, the rate of fresh air exchange and

leakage (Smale, 2004). Irreversible damage and loss of product quality can be caused by small

periods of equipment malfunction; thus continuous monitoring of the temperature and the

functioning of the entire cooling system should take place all along the chain (Ruiz-Garcia,

2008).

Heat leakage in a insulated container will be about 22 W K-1 for newer 6.1 m (20 ft) ones,

but some containers specially made for fresh fruit will have thinner insulation with a heat

leakage of around 35 W K-1. Values for 12.2 m (40 ft) containers are approximately double

these figures. Insulation efficiency reduces with time by about 3-5% per year (Heap, 1989;









years. Table 7-6 summarizes the costs of the RFID deployment year by year and gives the total

amount required for the project: $7,076,628.00. Note that the costs for the last two years are

assumed to be the same of the second and third year, respectively.

Estimation of the Benefits resulting from the RFID and Load Management System
Implementation

In order to estimate the economic benefits of the RFID system proposed, it is necessary to

first determine the amount of FSRs being shipped to Afghanistan and Iraq and then, estimate the

economic loss avoided by this system.

Estimation of FSR demand in Afghanistan and Iraq

According to recent reports (BBC, 2010a), the amount of U.S. soldiers deployed in Iraq

will be decreasing from around 98,000 to 50,000 by the end of 2010. Consequently, this will be

the amount of soldiers present in Iraq for this analysis. Likewise, these reports also mention that

the U.S. troops in Afghanistan will be close to 100,000 by the summer of 2010. Therefore, this

will be the number of soldiers assumed as stationed in this country for FSR demand calculations.

Table 7-7 summarizes the distribution of U.S. soldiers in these regions considered for this study.

Assuming only 10.00% of the 150,000 Warfighters in these areas consume FSRs in any

given day, then there would be a demand for 15,000 FSR meals per day (Table 7-8). The number

of pallets per day necessary to satisfy this demand can then be calculated and, taking into

account the number of pallets transported in a 40 foot sea container is 40, so is the number of

containers these would require per month and year (Table 7-9). Since the shipments to the zone

of conflict take place every month and the containers have to be fully loaded, an approximation

is made by rounding the monthly numbers obtained above and new yearly estimations were

found (Table 7-10).















Top View


Figure A-1. Placement of the HOBO temperature sensors in the experimental pallets.


Figure A-2. Position of the experimental pallets in the cargo areas (14-m refrigerated truck trailer
and 12-m sea container). Pallets depicted in dark color correspond to those containing
corrugated cartons; while the lighter ones represent pallets containing reusable plastic
containers (RPCs).


Refrigerated Truck


Door
f f f-- ^---

Back Middle Front


Sea Container



Door


Back Middle Front


Front View









Jedermann, R., Lang, W., 2007. Semi-passive RFID and beyond: steps towards automated
quality tracing in the food chain. International Journal of Radio Frequency Identification
Technology and Applications 1 (3), 247-259.

Jedermann, R., Stein, K., Becker, M., Lang, W., 2008. UHF-RFID in the Food Chain From
Identification to Smart Labels. Paper Presented at Cold Chain Manangement: 3rd
International Workshop, Bonn, Germany.

Jones, P., 2006. Networked RFID for use in the Food Chain. Paper Presented at the IEEE
Conference on Emerging Technologies and Factory Automation (ETFA). Prague, Czech
Republic.

Kader, A.A., 2002. Postharvest Technology of Horticultural Crops. University of California-
Division of Agriculture and Natural Resources, Oakland.

Kapoor, G., 2008. Secure Ownership Transfer and Authentication Protocols for Radio Frequency
Identification (RFID). Ph.D. Thesis, University of Florida, Department of Business
Administration.

Karkkainen, M., 2003. Increasing efficiency in the supply chain for short shelf life goods using
RFID tagging. International Journal of Retail and Distribution Management, 31 (10), 529-
536.

Labuza, T.P., Fu, B., 1993. Growth kinetics for shelf-life prediction: theory and practice. Journal
of Industrial Microbiology 12, 309-23.

Lahiri, S., 2006. RFID Sourcebook. IBM Press, Upper Saddle River.

Laniel, M., Emond, J.P., Altunbas, A.E., 2009. RFID behavior study in enclosed marine
container for real time temperature tracking. Journal of Sensing and Instrumentation for
Food Quality and Safety 3 (1), 34-40.

Lazaro, A., Girbau, D. Salinas, D., 2009. Radio link budgets for UHF RFID on multipath
environments. IEEE Transactions on Antennas and Propagation 57 (4), 1241-1251.

Leake, L., 2007. The Search for Shelf Life Solutions. Food Technology 61 (11), 66-70.

LeBlanc, D., Beaulieu, C., Lawrence, R., Stark, R., 1994. Evaluation of temperature variation of
frozen foods during transportation. The Refrigeration Research Foundation Information
Bulletin, Bethesda, MD.

Lenker, D.H., Wooddruff, D.W., Kindya, W.G., Carson, E.A., Kasmire, R.F., Hinsch, R.T.,
1985. Design criteria for the air distribution systems of refrigerated vans. ASAE Papers 28
(6), 2089-2097.

Leong, K. S., Ng, M. L., Cole, P. H., 2006. Dual-frequency antenna design for RFID application.
Paper Presented at the 21st International Technical Conference on Circuits/Systems,
Computers and Communications (ITC-CSCC 2006), Chiang Mai, Thailand.


215




















F I _


+ +


I I I I A) L_ IL = B)

Figure 5-3. Tag placement in the outside of the pallet for: A) Configuration a. B) Configuration
P.


Figure 5-4. Tag placement in the surface of the wooden pallet in Configuration
Figure 5-4. Tag placement in the surface of the wooden pallet in Configuration P.




Full Text

PAGE 1

1 DEVELOPMENT OF RADIO FREQUENCY IDENTIFICATION (RFID) TEMPERATURE TRACKING SYSTEMS FOR FOOD SUPPLY CHAINS By CECILIA R. AMADOR A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLM ENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Cecilia R. Amador

PAGE 3

3 To Patty Amador (My mentor, my hero, my guardian angel)

PAGE 4

4 ACKNOWLEDGMENTS I would like to express my deepest gratitude to my f ormer advisor, Dr. Jean -Pierre Emond, for his guidance, support, example and trust; and for giving me the great opportunity of pursuing my research in an exciting new area. I would also like to thank my current advisor, Dr. Ray Bucklin, for all his guidanc e and advice; your assistance has been invaluable in these last months. In addition, I would like to express my sincere gratitude to my committee members Dr. Cecilia do Nascimento Nunes, Dr. Jeffrey Brecht, Dr. Allen Wysocki, Dr. Daniel Engels, Dr. Joseph Geunes, and Dr. James Leary for their support, advice and assistance. I would like to thank the United States Department of Defense and Motorola, Inc. for funding this research; as well as the members of Franwell, Inc., who provided me part of the equipme nt used. I would also like to thank Dr. Ismail Uysal for his extremely valuable advice and assistance with the DoD project. Additionally, I would like to thank all the wonderful faculty and staff at the Agricultural and Biological Engineering Department a t UF, particularly to my good friend Billy Duckworth, whose help and support was essential in the completion of the experiments involved in this research. Thank you to my wonderful parents and siblings for believing in me since the very beginning; I woul d be nothing without your unconditional love, support, encouragement and example. Patrica, thank you for watching over me from above and for guiding me every step of the way; I know Ill never be alone. Thank you to Tere and Felix, you have been a source o f support, love, happiness and motivation beyond words. Thank you to Magalie for being my partner in crime and for being a truly great friend. Thank you to Erdem for always having the right words; to Elliot for his tremendous help during these years; to Alan for always making me smile; and to all my current and former lab mates, who have been a joy to be around and to work

PAGE 5

5 with. Thank you to Pedro, Sharon and Lacey for understanding and loving me at home. To the members of my Peruvian family in Gainesvill e, thank you for your love and encouragement. My deepest thank you to Ceci and Guillermo for behaving as my big siblings and always having the best advice available for me. Thank you to Mari, Dane, Betsi, Karol, Juanca, Isa, Marce, Raul, Kathy, Milton, J ose, Daniel, Alonso, Wendy -Maria, Nadia, Chris, Alexis and Ingrid; for always being by my side even though we are thousands of miles apart. I would also like to thank Dr. Arthur Teixeira, for changing my life for the best ten years ago; to him and his wif e Marjorie, my most sincere gratitude for having opened your hearts to me when I truly mostly needed it. And last, but never least, I would like to thank God for his love and for all the many blessings he has granted me. Thank you Lord for this unique opp ortunity and for giving me the strength to complete it.

PAGE 6

6 TABLE OF CONTENTS ACKNOWLEDGMENTS .................................................................................................................... 4 page LIST OF TABLES .............................................................................................................................. 11 LIST OF FIGURES ............................................................................................................................ 14 LIST OF ABBREVIATIONS ............................................................................................................ 16 ABSTRACT ........................................................................................................................................ 17 CHAPTER 1 INTRODUCTION ....................................................................................................................... 19 2 LITERATURE REVIEW ........................................................................................................... 22 Temperature Management in Food and Produce Supply Chains ............................................. 22 The Pineapple Supply Chain ...................................................................................................... 23 Cold Chain ................................................................................................................................... 24 Precooling .................................................................................................................................... 25 Forced -Air Cooling .............................................................................................................. 26 Hydrocooling ....................................................................................................................... 27 Vacuum Cooling .................................................................................................................. 29 Ice Cooling ........................................................................................................................... 29 Room Cooling ...................................................................................................................... 31 Short Term Storage ..................................................................................................................... 31 Transportation Systems ............................................................................................................... 32 Temperature Control ............................................................................................................ 32 Heat Loads ........................................................................................................................... 32 Internal heat loads ........................................................................................................ 33 External heat loads ....................................................................................................... 33 Residual heat loads ....................................................................................................... 33 Air Circulation Systems ...................................................................................................... 34 Top air delivery system ............................................................................................... 34 Bottom air delivery system ......................................................................................... 34 Main Transportation Modes ................................................................................................ 35 Overland transport by trucks ....................................................................................... 35 Overla nd transport by railcars ..................................................................................... 36 Marine transport ........................................................................................................... 36 Conventional ships ....................................................................................................... 37 Container ships ............................................................................................................. 38 Intermodal transport ..................................................................................................... 40 Airfreight transport ....................................................................................................... 40 Temperature Variations during Transport .......................................................................... 41

PAGE 7

7 Temperature Measurement Devices ........................................................................................... 43 Thermocouples ..................................................................................................................... 44 Resistance Temperature Detectors (RTD) ......................................................................... 44 Thermistors (Thermal Resistors or Bulk Semiconductor Sensors) .................................. 46 Radiation Thermometers (Pyrometers) .............................................................................. 46 Non -Electric Thermometers ................................................................................................ 47 Radio Frequency Identification (RFID) ..................................................................................... 48 Industrial Scientific Medical (ISM) Bands ........................................................................ 49 Carrier Frequency ................................................................................................................ 49 Low frequency (LF) ..................................................................................................... 50 High frequency (HF) .................................................................................................... 50 Ultra high frequency (UHF) ........................................................................................ 50 Microwave frequency .................................................................................................. 51 Materials Hindering the Signal ........................................................................................... 51 Multipath Effect ................................................................................................................... 53 RFID System Components .................................................................................................. 54 RFID tags ...................................................................................................................... 54 Classes of RFID tags .................................................................................................... 54 EPCTM tag classifications ............................................................................................. 55 RFID Reader ........................................................................................................................ 56 Stationary reader .......................................................................................................... 56 Handheld reader ........................................................................................................... 56 Reader Antenna .................................................................................................................... 57 Antenna footprint ......................................................................................................... 57 Antenna polarization .................................................................................................... 58 Linear polarized antenna .............................................................................................. 58 Circular polarized antenna ........................................................................................... 58 Information Processing Software ....................................................................................... 59 RFID Applications ............................................................................................................... 59 Applications of RFID in the Produc e Industry .................................................................. 59 Identification ................................................................................................................. 60 Trace back ..................................................................................................................... 60 Process system tra ceability .......................................................................................... 62 Opportunities of RFID usage in the produce industry ............................................... 62 Challenges for the application of RFID in the produce indust ry .............................. 64 3 APPLICATION OF RFID TECHNOLOGIES IN THE TEMPERATURE MAPPING OF THE PINEAPPLE SUPPLY CHAIN .................................................................................. 70 Introduction ................................................................................................................................. 70 Materials and Methods ................................................................................................................ 71 Experimental Design ........................................................................................................... 71 Statistical Analysis ............................................................................................................... 73 Results and Discussion ............................................................................................................... 74 RFID Sensor Performance versus Conventional Methods ................................................ 74 RFID Temperature Tags with Probe versus RFID Temperature Tags without Probe and Their Utilization along the Supply Chain ................................................................ 76

PAGE 8

8 Level of Instrumentation ..................................................................................................... 78 Conclusions ................................................................................................................................. 79 4 EVALUATION OF SENSOR READABILITY AND THERMAL RELEVANCE FOR RFID TEMPERATURE TRACKING ....................................................................................... 88 Introduction ................................................................................................................................. 88 Materials and Methods ................................................................................................................ 91 Relevance Study .................................................................................................................. 91 Readability Study ................................................................................................................. 92 Tag placement .............................................................................................................. 92 Equipment configuration ............................................................................................. 93 Results and Discussion ............................................................................................................... 93 Relevance Study .................................................................................................................. 93 Temperature distribution .............................................................................................. 93 Eighty five percent location ......................................................................................... 94 Readability Study ................................................................................................................. 97 Conclusions ............................................................................................................................... 100 5 DEVELOPMENT OF RFID TEMPERATURE TRACKING SYSTEMS FOR COMBAT FEEDING LOGISTICS ......................................................................................... 106 Introduction ............................................................................................................................... 106 M aterials and Methods .............................................................................................................. 109 Relevance Study ................................................................................................................ 109 Readability Study............................................................................................................... 110 Fixed system ............................................................................................................... 110 Handheld system ........................................................................................................ 110 Temperature Profile Estimator .......................................................................................... 111 Results and Discussion ............................................................................................................. 113 Relevance Study ................................................................................................................ 113 Readability Study ............................................................................................................... 116 Fixed system ............................................................................................................... 116 Handheld system ........................................................................................................ 118 Temperature Profile Estimator .......................................................................................... 120 Conclusion ................................................................................................................................. 122 6 DEVELOPMENT OF A LOAD MANAGEMENT SYSTEM FOR COMBAT FEEDING LOGISTICS BASED ON SHELF -LIFE PREDICTION SOFTWARE ............. 133 Introduction ............................................................................................................................... 133 Materials and Methods .............................................................................................................. 135 Development of the Software ........................................................................................... 135 Application of the Shelf -Life Software ............................................................................ 138 Statistical Analysis ............................................................................................................. 139 Results and Discussion ............................................................................................................. 139 Development of the Software ........................................................................................... 139

PAGE 9

9 Maximum Amount of Weeks of Shelf -Life at the Deployment Areas .......................... 140 Differences between Acceptability Estimates for FSR Meals ........................................ 141 Differences between Acceptability Estimates at Different Sensor Locations ............... 142 Conclusions ............................................................................................................................... 143 7 RETURN ON INVESTMENT (ROI) DETERMINATION FOR THE DEPLOYMENT OF A RFID -BASED LOAD MANAGEMENT SYSTEM IN COMBAT FEEDING LOGISTICS ............................................................................................................................... 149 Introduction ............................................................................................................................... 149 Return on Investment (ROI) Analysis ..................................................................................... 150 Es timation of RFID Deployment Costs ............................................................................ 150 Estimation of the Benefits resulting from the RFID and Load Management System Implementation .............................................................................................................. 153 Estimation of FSR demand in Afghanistan and Iraq ............................................... 153 Estimation of FSR pallets lost to thermal abuse ...................................................... 154 Estimation of th e economic loss of the FSR pallets lost to thermal abuse ............. 155 Estimation of FSR pallets sent in emergency shipments ......................................... 155 Estimatio n of the economic loss avoided by the load management system ........... 156 Return on Investment (ROI) Estimation .......................................................................... 156 Conclusion ................................................................................................................................. 157 8 CONCLUSIONS ....................................................................................................................... 163 For Products Prone to Low and High Temperature Abuse (Case 1) .............................. 164 For Products Susceptible to High Temperature Abuse (Case 2) .................................... 164 For Shelf Stable Products (Case 3) ................................................................................... 165 APPENDIX A ANALYSIS OF THE VARIABILITY AND COLD CHAIN PERFORMANCE FOR CROWNLESS PINEAPPLE WITH RESPECT TO TRANSPORTATION METHODS, LOCATION WITHIN THE CARGO, AND PACKAGING ................................................. 166 Introduction ............................................................................................................................... 166 Materials and Methods .............................................................................................................. 168 Experimental Design ......................................................................................................... 168 Statistical Analysis ............................................................................................................. 170 High Temperature Abuse (HTA) and Low Temperature Abuse (LTA) Risk Estimation ....................................................................................................................... 170 High temperature abuse (HTA) analysis .................................................................. 170 Low temperature abuse (LTA) analysis .................................................................... 171 Results and Discussion ............................................................................................................. 171 Temperature Profiles of Pineapples using Different Combinations of Packaging, Transportation Methods and Locations in the Cargo Environment (Sea Container and Truck Trailer). ......................................................................................................... 171

PAGE 10

10 High Temperat ure Abuse (HTA) and Low Temperature Abuse (LTA) Risk Estimation ....................................................................................................................... 173 High temperature abuse (HTA) analysis .................................................................. 173 Refrigerated truck trailer/open holds ....................................................................... 174 Sea container .............................................................................................................. 1 78 Low temperature abuse (LTA) analysis .................................................................... 180 Refrigerated truck trailer/open holds ....................................................................... 181 Sea containers ............................................................................................................ 182 Conclusions ............................................................................................................................... 183 Further work .............................................................................................................................. 184 B RECOMMENDED LEVEL OF RFID INSTRUMENTATION FOR THE CROWNLESS PINEAPPLE SUPPLY CHAIN WHEN THIS USES THE REFRIGERATED TRUCK/OPEN HOLD TRANSPORTA TION METHOD .................... 197 C AMBIENT TEMPERATURE PROFILES FOR THE THERMAL RELEVANCE STUDY OF A PALLET OF SPHERICAL BOTTLES OF WATER .................................... 198 D T EMPERATURE PROFILE USED AS WORST CASE SCENARIO FOR FSR SHIPMENTS ............................................................................................................................. 209 LIST OF REFERENCES ................................................................................................................. 210 BIOGRAPHICAL SKETCH ........................................................................................................... 223

PAGE 11

11 LIST OF TABLES Table page 2 1 Thermocouple properties according to their ISA classification .......................................... 67 2 2 Temperature ranges for common metals used in RTDs. ..................................................... 68 2 3 List of RF properties of some materials. ............................................................................... 68 2 4 EPCTM tag classific ation. ....................................................................................................... 69 3 1 Comparison of features between RFID tags and the HOBO sensor ............................... 83 3 2 GLM results when comparing data from HOBO a nd ThermaProbe tags ....................... 85 3 3 Comparison of HOBO and ThermaProbe RF RFID tags temperatures ......................... 85 3 4 GLM results when comparing data from ThermAssure RF and ILR i Q 32T. .................. 86 4 1 RFID reader configuration. .................................................................................................. 102 4 2 Average temperature differentials present at the pallet level. ........................................... 102 4 3 Average temperature differentials present at the RPC level. ............................................. 102 4 4 Likelihood of gathering 85% of temper atures in the intervals calculated. ....................... 103 4 5 Readability in the points of relevance (PoR) for Configuration ............................... 105 4 6 Readability in the points of relevance (PoR) for Configuration ................................... 105 5 1 RFID reader configuration. .................................................................................................. 124 5 2 Likelihood of gathering 85% of the temperatures in the 16C range proposed .............. 126 5 3 Likelihood of gathering 85% of the temperatures wh en using the PoRs. ....................... 127 5 4 Maximum readability of both PoRs for Configuration with the fixed system ............ 128 5 5 Maximum readability of both PoRs in Configuration with the fixed system ............... 129 5 6 Readability f or the fixed system in positions b and c. ................................................ 129 5 7 Readability of four tag positions using the handheld reader ............................................ 130 6 1 Fitting curves for the acceptability of the shelf -life limiting items ................................... 146 6 2 Final acceptability score predicted by the software. .......................................................... 146

PAGE 12

12 6 3 Means and SDs of the combination of the 3 FSR meals at all temperature profiles. ...... 147 6 4 GLM of the acceptability scores for Profile 1 and its related profiles. ............................. 147 6 5 GLM of the acceptability scores for Profile 2 and its related profiles .............................. 147 6 6 GLM of the acceptability scores for Profile 3 and its related prof iles .............................. 148 6 7 Means and SDs for each meal when combining 3 temperature profiles .......................... 148 7 1 Hardware and software costs for each work station. ......................................................... 158 7 2 Yearly costs of RFID tags. ................................................................................................... 159 7 3 Cost of the project during the first year of operation. ........................................................ 159 7 4 Cost of the project during the second and fourth year of operation. ................................. 159 7 5 Cost of the project during the third and fifth year of operation. ....................................... 160 7 6 Yearly costs of the project. .................................................................................................. 160 7 7 Distribution of U.S. troops in the zone of conflict studied. ............................................... 160 7 8 Estimated number of FSR eaten daily in the zone of conflict studied. ............................. 160 7 9 Estimated demand for FSRs in the zone of conflict studied. ............................................ 161 7 10 Actual demand for FSRs in the zone of conflict studied. .................................................. 161 7 11 Estimations of yearly pallet losses. ..................................................................................... 161 7 12 Maritime shipping costs for FSR loads according to its destination. ................................ 161 7 13 Total cost of a lost pallet according to its final destination .............................................. 162 7 14 Amount of product sent as yearly emergency shipments. ................................................. 162 7 15 Amount of money destined to yearly emergency shipments ............................................ 162 7 16 Yearly savings created by the proposed system. ................................................................ 162 A 1 GLM comparing the refrigerated truck trailer/open holds with a sea container. ............. 186 A 2 GLM comparing three positions inside the sea container. ................................................ 187 A 3 GLM comparing refrigerated holds inside a vessel .......................................................... 187 A 4 Percentage of time of HTA exposure in refrigerated truck trailer/open hold s ................. 188

PAGE 13

13 A 5 Risk o f HTA exposure in th e refrigerated truck trailer/open hold combination. ............. 188 A 6 Risk of HTA exposure according to packaging in refrigerated truck /open hold s ........... 189 A 7 Temperatures along the supply chain in refrigerated truc k trailer/open hold s ................. 189 A 8 Percentage of time of HTA exposure in sea containers. .................................................... 190 A 9 Risk of HTA exposure in sea containers. ............................................................................ 190 A 10 Risk of HTA exposure according to packaging in the sea containers. .............................. 191 A 11 Temperatures along the supply chain in using sea containers. .......................................... 193 A 12 Percentage of time of LTA exposure in refrigerated truck trailer/open holds ................. 195 A 13 Risk of LTA exposure in the refrigerated truck trailer/open hold combination. .............. 195 A 14 Risk of LTA exposure according to packaging in refrigerated truck /open hold s ............ 195 A 15 Percentage of time of LTA exposure in sea containers. .................................................... 196 A 16 Risk of LTA exposure in s ea containers. ............................................................................ 196 A 17 Risk of LTA exposure according to packaging in sea containers ..................................... 196 C1 Ambient temper ature profiles in th e 0.5 h/0.5 h heating and cooling episodes. .............. 198 C2 Ambient temperature profiles in the 1 h/1 h heating and cooling episodes. .................... 199 C3 Ambient temperature profiles in the 2 h/2 h heating and cooling episodes. .................... 201 C4 Ambient temperature profiles in the 4 h/4 h heating and cooling episodes. .................... 204 D 1 Maximum temperatures from a combat feeding shipment from the U.S. to Kuwait. ...... 209

PAGE 14

14 LIST OF FIGURES Figure page 3 1 Placement of the HOBO temperature sensors in the experimental pallets. ........................ 81 3 2 RFID tags used during the trial. ............................................................................................ 81 3 3 Pl acement of the RFID temperature sensors in the experimental pallets. .......................... 82 3 4 Ambient temperatures obtained by the ALB 2484RFID tag ............................................. 84 3 5 Ambient temperatures obtained by ThermAssure RF and ILR i Q 32T ............................ 84 3 6 Temperature profiles from RFID tags with probe and without probe in RPCs ................ 86 3 7 Temperature profiles from RFID tags with probe and without probe in boxes ................ 87 3 8 Temperature profiles from HOBO sensors in 3 areas of a pallet with boxes ..................... 87 4 1 Area of pallet instrumented ................................................................................................. 101 4 2 Top view of the depth of tag placement for each configuration. ...................................... 102 4 3 Location of the points of relevance in one half of the pallet. ............................................ 103 4 4 Top view of the placement of the RFID tags in the pallet. ................................................ 104 4 5 Vertical view of the location of the RFID tags in Configuration I ............................... 104 4 6 Vertical view of the location of the RFID tags in Configuration ................................. 105 5 1 Three dimensional view of the FSR packe ts instrumented in each box. .......................... 124 5 2 Pallet and antenna position at an antenna distance of 2 m ................................................ 124 5 3 Tag placement in the outs ide of the pallet for both configurations ................................... 125 5 4 Tag placement in the surface of the wooden pallet in Configuration ........................... 125 5 5 A typical resistor capacitor circuit. ..................................................................................... 126 5 6 Locations of the Points of Relevance (PoR) detected in the pallet. .................................. 127 5 7 Locations of the RFID tags during the readability study for Configuration ................. 128 5 8 Temperatures in PoR A changing with respect to ambient temperatures ......................... 130 5 9 Temperatures in PoR B changing with respect to ambient temperatures ......................... 131

PAGE 15

15 5 10 Ambient tempera tures and measured and estimated temperatures in PoR A ................... 131 5 11 Ambient temperatures and measured and estimated temperatures in PoR B ................... 132 6 1 Screenshot of the final Shelf -Life Prediction Software passing the load ......................... 145 6 2 Screenshot of the final Shelf -Life Prediction Software rejecting the load ....................... 145 A 1 Placement of the HOBO temperature sensors in the experimental pallets. ...................... 185 A 2 Position of the experimental pallets in the cargo areas ...................................................... 185 A 3 Schematic of the loading pattern inside the refrigerated cargo vessel. ............................. 186 A 4 Ambient air temperature profile inside the sea con tainer .................................................. 192 A 5 Temp erature profiles presenting an air short -circuiting pattern ........................................ 194

PAGE 16

16 LIST OF ABBREVIATION S DoD United States of America Department of Defense EPC Electronic P roduct Code FSR First S trike Rations HF h igh f requency HTA h igh t emperature a buse ISO International Standards Organization LF l ow f requency LTA l ow temperature abuse MRE Meal, Ready -to -eat PoR p oint of r elevance RF radio f requency RFID radio f requency i dentification RPC reusable plastic c ontainer SLLI s helf l ife l imiting i tem UHF u ltra -h igh f requency

PAGE 17

17 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirem ents for the Degree of Doctor of Philosophy DEVELOPMENT OF RADIO FREQUENCY IDENTIFICATION (RFID) TEMPERATURE TRACKING SYSTEMS FOR FOOD SUPPLY CHAINS By Cecilia R. Amador August 2010 Chair: Ray Bucklin Major: Agricultural and Biological Engineering Food items require temperature controlled supply chains since exposure to certain temperature conditions can diminish product quality and create safety threats. Temperature tracking systems should then be in place in order to monitor the temperature management along their supply chains. Radio frequency identification (RFID) has been suggested as having the potential to become an enhanced temperature tracking method; yet, very few studies have explored the subject in depth. Furthermor e, practical details about i ts application, such as the proper use of RFID tags with probe and without them, how to surpass environmental interactions taking place along the supply chain or how to achieve relevant monitoring keeping the costs down, and the economic benefit it will br ing to the food industry still remain unclear. The following work aims to offer insi ght on these matters and to create viable applications for the technology in real life food supply chains. Four objectives were established: 1) To c ompar e the performance o f RFID temperature tags versus conventional temperature tracking methods in a food supply chain; 2) To c ompar e the utilization of RFID temperature tags with probe and without them along a food supply chain; 3) To determine the level of instrumentation (amo unt of sensors and the best locations for their placement) of an efficient temperature tracking system in three different scenarios for food

PAGE 18

18 supply chains (For products prone to low and high temperature abuse, for products susceptible to high temperature a buse, and for shelf -stabl e products); and 4) To create the business case for a RFID temperature tracking system when combined with shelf life prediction software by performing an economic analysis in one of the systems previously designed. In order to ach ieve the first three objectives, a shipping trial was performed with crownless pineapples; while spherical water bottles mimicking produce and First Strike Rations (FSRs) were subjected to thermal relevance and readability studies for the third one. A dditi onally, software material allowing temperature estimations inside the pallet of FSRs and shelf -life prediction were developed as support material for objectives three and four. Finally, a return on investment (ROI) study was performed for a load management system based on the final monitoring system developed for FSRs. Results indicate that, although analogous with respect to accuracy in the temperature measurements, RFID systems are superior as temperature tracking method t o conventional methods. In addit ion, RFID tags with probe are important to monitor the critical points of the load, which are the areas of the load where temperature abuse is most likely to occur inside the product; while RFID tags without probes are relevant during monitoring of ambient conditions during storage and transportation. Also, a monitoring system for crownless pineapples w as designed, which in some cases involved the use of more than one tag per pallet. Moreover, RFID monitoring systems were also designed for loads of certain varieties of produce such as apples, oranges, pomegranates, passion fruit and tangerines, and for FSRs; but these allowed the use of only one tag per pallet. Last ly, the R eturn on Investment (ROI ) analysis of the RFID based load management system designed for FSRs was calculated to be 719.49% ; which proved that this technology can be a n important tool for value generation in food supply chains

PAGE 19

19 CHAPTER 1 INTRODUCTION Globalization has promoted the expansion of the trade of food and produce worldwide. Many o f these products now have to travel longer distances than before in order to reach their final markets. C onsumers currently have the opportunity to eat year round products that use to be considered seasonal, or that were not present in their area before (K ader, 2002). As new flavors are introduced into their diet, offering a high -quality product becomes one m ost effective ways to gain customer loyalty and guard the economic interests of the companies involved in the business. Achieving this expected quali t y, however, requires meeting the challenge of controlling the environmental conditions surrounding the load all along handling, transport, distribution, and retail operations (Moureh et al., 2002). Managing the temperature of the product, in particular, is the most important factor for maintaining the products quality and extending its sh elf life (Jedermann et al., 2008). Bad temperature management can accelerate a wide array of deterioration processes and microbial growth; which in some cases can lead to food safety threats for the public (Potter and Hotchkiss, 199 8 ; James, 2006). Consequently, it is imperative for the industry to count er with mechanisms that actively monitor the temperature conditions food products are exposed to during their supply chain s. Radio Frequency Identification (RFID) is a wireless auto identification technology that has been mentioned as having the potential to become an enhanced method for temperature tracking (Mermelstein, 2002; Karkkainen, 2003; Gaukler and Seifert, 2007). C omparisons amongst different kinds of RFID temperature sensors and between this and other sensing methods have only taken place recently (Ruiz -Garcia, 2008; Jedermann et al., 2007). However, further performance analysis and evaluation are required in order to determine their superiority as a

PAGE 20

20 sensing method and the suitability of systems with and without probes for particular food supply chain applications. The use of RFID for temperature tracking purposes in food supply chains has two main challenges. First the lack of robustness of the most common RFID systems for supply chain applications around products containing high amounts of water, such as produce, or in environments containing metal, which are frequently encountered at different stages of food supp ly chains (Dobkin and Weigand, 2005; Redemske and Fletcher, 2005; Gaukler and Seifert, 2007; Hartvanyi and Marek, 2007; Sivakumar and Deavours, 2008). And second, the possibility of increasing the costs along the supply chain by using this type of sensors (Edwards, 2007), which drives the need for efficient instrumentation. Therefore, real -life implementations of RFID temperature tracking systems in food supply chains must surpass any negative interaction with the load and its environment and keep sensing c osts to a minimum while still providing relevant temperature information. Current RFID systems offer fast collection of data and the software processing capabilities necessary for real time decision making in food supply chains. Emond and Nicometo (2006) p roposed their use in the application of the concept of First Expires First Out (FEFO) in supply chain and logistic operations, which can be achieved by using e stimates o f the remaining shelf life of the product. L oa d management system s able to provide recommendations given the status of the loads quality can be then produced by combining the te mperature data collected by RFID system s with shelf -life prediction software and create efficiencies all along the supply chain s A RFID temperature tracking sy stem empowered with shelf life prediction software can become a powerful tool for the food industry. Yet, the level of profit generated by its

PAGE 21

21 implementation will vary with the particularities of the food supply chain. Economic studies are then important i n order to determine whether the financial benefits obtained by applying these systems in real life surpass the investment and operating costs involved in them (Banks et al., 2007). As can be seen, even though RFID temperature tracking seems to offer grea t possibilities for the food industry, the details involved in its application and the rea l benefits it will bring to food supply chains still remain unclear. The fol lowing work aims to offer insight on these matters and to create viable applications for t he technology in real life food supply chains. The o bjectives of the research presented in this dissertation were: 1 To study the use of RFID in temperature monitoring by comparing the performance of RFID temperature tags versus conventional temperature tra cking methods in a food supply chain. 2 To compare the utilization of RFID temperature tags with probe versus RFID temperature tags without probes along a food supply chain. 3 To determine the level of instrumentation (amount of sensors and the best locations for their placement) of an efficient RFID temperature tracking system at the pallet and cargo level in three different scenarios for food supply chains: For products prone to low and high temperature abuse (Case 1). For products susceptible to high tempera ture abuse (Case 2). For shelf stable products (Case 3). 4 To create the business case for a RFID temperature tracking system when combined with shelf life prediction software by performing an e conomic analysis in one of the systems previously designed.

PAGE 22

22 CHAPTER 2 LITERATURE REVIEW Temperature Management in Food and Produce Supply Chains Temperature is one of the most important factors in food deterioration for it promotes biochemical, chemical and physical changes and enzymatic reactions. The growth of pa thogens and decay organisms is highly temperature dependent, thus temperature management is imperative in food safety and quality maintenance. In addition, inadequate temperature conditions during storage could also contribute to the products degradation by hastening other enzymatic and nonenzymatic reactions such as lipid oxidation and emulsion breaking. Fu rthermore, physical changes such as moisture loss can have economic implications by promoting the loss of saleable weight while at the same time reduci ng the products shelf -life (Potter and Hotchkiss, 1998 ; James, 200 6 ). In the case of produce, keeping optimal temperatures will also generate physiological changes, such as reducing the commoditys respiration rate and thus, delaying its senescence (Shew felt and Prussia, 1993). The Q10 value is an indicator of the effect of temperature rise in produce metabolism. Values of 2 to 3 are common for most produce, and denote a two to three fold increase in the produce respiration rate once the products tempera ture has in creased 10C (Nunes et al., 1995; Shewfelt and Bruckner, 2000; Thompson, 2003). Storage at lower temperatures also implies a reduction in produce transpiration and water loss (Shewfelt and Bruckner, 2000; DeEll et al., 2003; Proulx et al., 2005) Finally, Kader (2002) asserts that temperature also impacts the effect of ethylene, reduced oxygen and elevated carbon dioxide in produce. When temperature is not properly managed, it can generate physiological disorders in produce. For exa mple, chillin g injury (CI) develops after exposing chilling-sensitive produce to

PAGE 23

23 temperatures below their threshold, which damages membranes and promotes cellular breakdown (Nunes et al., 2003b ). CI creates a wide variety of symptoms, though some of them might not mani fest themselves until the product is exposed to higher temperatures (Shewfelt and Prussia, 1993; Thompson, 2003). Another physiological disorder, freezing injury takes place when the commodity is overcooled and ice crystals form inside its cells (Hui et al ., 2003; Thompson, 2003). According to Thompson (2003), in order to avoid these disorders, produce handling and storage should take place at temperatures just over their threshold temperatures for chilling and freezing injury. Lastly, high temperature inju ry can be the product of either long exposure to temperatures within the 3040C range, or to short exposure to even higher temperatures (Shewfelt and Prussia, 1993). In conclusion, proper temperature management along food and produce supply chains will d elay the deterioration of the product, extending its shelf life and maintaining its economic value along the chain. The Pineapple Supply Chain Pineapple is the second most important tropical fruit in the world. Around 21 million tons of pineapples were pro duced worldwide in 2007. Twelve percent of these were devoted exclusively to the international fresh market, a business of more than 2 USD billion dollars (FAO, 2009). Central America (Costa Rica, Panama) and the Philippines are the biggest sources of frui t; while the US is the top importer, having reached its levels of consumption to 2.16 kg of fresh pineapple per capita in 2007 (USDA, 2010). International commerce of fresh pineapples requires highly efficient temperature controlled supply chains. Due to the sensitivity of this product to chilling injury, temperatures must remain within a specific range at all times (7 C to 12 C) avoiding exposures to temperatures lower than the products threshold (Abdullah et al., 2000; Acedo et al., 2004). However, exp osure to higher

PAGE 24

24 temperatures than the ones recommended for its storage will accelerate the senescence and decay rates of the load (Mohammed et al., 1995). Poor temperature management during pineapple shipments will then result in postharvest losses and in poor product quality; which generates lower customer satisfaction and impacts the produce companies with economic losses and lack of public credibility (Nunes, 2008; Machado et al., 2009; Nunes et al., 2009). During the last years, the fresh pineapple tra de has been positively affected by the use of the fruit in the fresh-cut industry, which can import either pineapples with crown or crownless (Gonzalez -Aguilar et al., 2004). Crownless pineapples facilitate fresh -cut operations; however, their postharvest management becomes more challenging, since the wounding left by crown removal increases metabolic activities and promotes senescence and decay in the fruit. Given the direct relationship between temperature and these processes, adequate temperature managem ent in crownless pineapples is then of the uttermost importance. Cold Chain Perishables such as produce, pharmaceuticals, meat, and dairy products require temperature control along their supply chains in order to avoid safety hazards and market loss. Thei r temperature requirements will vary according to the biology and nature of the product; for example, certain pharmaceutical items require storage temperatures of 2 to 8 C, while certain tropical crops require minimum temperatures of 7.2 C. Cold chain i s a term that describes a supply chain that provides the best temperature conditions required to maintain the initial quality of the product and prolong its shelf life, at all times. As mentioned before, proper temperature control slows down the rate of quality loss. According to Sargent (1988), the cold chain should not be broken after the initial cooling of the product; while Nunes et al. (2003a ) explains that short interruptions in the cold chain can generate a fast decline in product quality. Thompson e t al. (1998) affirms that in the case of

PAGE 25

25 produce, the product condition at the market stage will be the result of the sum of all the quality losses that took place in prior stages of the cold chain. Precooling In produce supply chains, the cold chain sta rts with the rapid removal of field heat after harvest and before storage and/or transport. This is performed in a process called precooling. By dropping the temperature of the produce to the approximately optimum storage and/or transportation temperature, the metabolism and deterioration rates of the product diminish as well as the moisture loss, reducing postharvest losses (Miller et al., 2001; Brosnan and Sun, 2001; Dincer, 2003). Delays before precooling should be avoided since, as mentioned before, tem perature can have a significant impact on fresh produce quality and mar ket life. For example, Thompson et al. (1998) reports that a delay of only 2 hours at 30C/86F before precooling was enough to cause decay and severe bruising symptoms which generated a loss in strawberry quality. Anderson (2010) affirms that rapid cooling is particularly important for products with high respiration rates such as sweet corn where deterioration and market loss could be considerably avoided if the product is subjected to proper temperature management. According to Sargent (1988), the rate of heat transfer or the cooling rate, is critical for the efficient removal of field heat; and is dependent upon three factors: time, temperature and contact. Therefore, in order to ac hieve maximum cooling, the product has to remain inside the precooler for enough time. In addition, the cooling medium (air, water, etc.) must be maintained at constant temperature all along the cooling process. And lastly, the design of the primary and se condary packaging of the product should allow intimate contact between the cooling medium and the surfaces of the individual pieces of produce.

PAGE 26

26 There are different methods used for precooling. The factors that play into the selection process are: the rate of cooling required, compatibility of the method with the product to be cooled, subsequent storage and shipping conditions, and equipment and operating cost (Talbot and Chau, 1991). Forced air cooling is one of the most common methods; being used for many fruits, fruit type vegetables and cut flowers. However, when products tolerate water contact, hydrocooling is a viable option. This method uses water as the cooling medium and requires the use of water resistant packaging. It is commonly used for root, s tem and flower type vegetables, melons and some tree fruits. Crops such as leafy vegetables, can use vacuum and water spray vacuum cooling. And finally, package icing cools and maintains product temperature by means of crushed ice and is used for a small a mount of produce, amongst them broccoli (Thompson, 2004). Forced -Air Cooling In this method, the pallets are placed in a cold room, and cold air is forced to flow through the inside of each container. According to Meana (2005), this allows the air to remove the heat directly from the surface of the product by forced-convective contact. This process creates a pressure differential across the containers producing a driving force through container openings and individual pieces of commodity (Tutar et al., 2009). The cold air then go es though the path of least resistance, which should be present inside the primary packaging of the product and not around it (Talbot et al., 1992). The cooling rate of this precooling method is determined by the available refrige ration capacity, the heat transfer capacity of cooling air, and the heat transfer parameters. For example, cooling rates could be hindered when the product is packed in large bulk bins or tight cartons because of the apparent conduction resistance existent And, in such cases, respiratory heat might

PAGE 27

27 even raise the i nternal temperature of the load Yet in most cases, the first two factors are the most relevant, particularly when the product exposes a large surface to the air flow. In addition, the temperatur e difference between the product and the cooling air, as well as the velocity of the air passing through th e products are the main elements influencing the heat transfer rate from the product to the air stream (Dincer, 2003). The efficiency of forced air c ooling is determined by process time and product temperature uniformity. Ventilated packaging favors rapid and uniform cooling. The airflow inside the packaging is a strong factor in the heat transfer taking place. Effective venting is then necessary to ma ximize cooling efficiency. Tutar et al., 2009 mentions that a compromise between container structure and venting areas should be reached Thus, packaging integrity has to exist even though openings large enough to avoid hampering airflow are distributed al ong the bottom a nd the walls of the container. According to Wang and Tunpun (1968) and Mitchell et al. (1971) boxes should have about 5% sidewall vent area; nonetheless, Dincer (2003) recommends a minimum of 6% of the total face area of a carton on the inc oming air side as acceptable. The main advantages of using forced air cooling are its simplicity, economy, sanitation, and the fact that it is relatively noncorrosive to the equipment (Dincer, 2003). Its main disadvantage is that it causes some moisture lo ss during cooling which could be significant for produce with a low transpiration coefficient. This moisture loss is correlated to the difference between initial and final product temperatures and can be reduced at the expense of longer cooling times by wr apping product in plastic or packing it in bags (Thompson et al. 2002). Hydrocooling Cooling is accomplished by moving cold water around produce with a shower system or by immersing produce directly in cold water (Thompson et al. 1998). These can be flow through or batch systems (Dincer, 2003). The produce can be cooled in bins or be in bulk before packing

PAGE 28

28 or already be in their primary containers after the packing process. According to Vigneault et al. (2004) this system involves less capital cost than the other cooling methods; generates faster cooling rates than forced air and liquid -ice; uniformly distributes the electric -power demand by creating a heat sink in ice; avoids water loss in produce, and may also increase the water content in some commodit ies. Efficient hydrocooling depends upon the adequate flow of water over the produce surface, which is between 10 to 17 L s1 m2 (Thompson et al. 2002), and on the uniformity of the water distribution. Efficient hydrocooling will then highly depend on th e design of the container and the stacking arrangement of the produce. In addition, cooling efficiency will also be affected by other factors, such as the water distribution inside the containers and the amount of water leaving the container by flowing thr ough the side -walls (Vigneault et al., 2004). Dincer (2003) reports that flow -through cooling systems are reasonably effective but leave hot spots throughout the load, especially in loads of sweet corn and celery that are packed in wire -bound crates; whi le batch systems have slower cooling rates. Packages for hydro-cooled produce must allow vertical water flow and must resi st water contact. Thompson (2004) indicates that the best packaging for hydrocooling are plastic or wood containers; however, corruga ted boxes can also be used if they have been previously wax dipped. According to Sargent (1988), lots of attention should be paid to the sanitation of the hydrocooling water, since this is reused for cooling all along the shift. If proper care is not take n, decay -free produce can be inoculated by the decay organisms from previous loads. Additionally, the same author also mentions that two of the main features of hydrocooled produce are that these are resistant to contact with water borne pathogens and mana ge to withstand the force of

PAGE 29

29 the water drench. Thompson et al. (1998) suggests minimizing the levels of decay organisms by obtaining the water from a clean source and treating it (usually with hypochlorous acid from sodium hypochlorite or gaseous chlorine) Vacuum Cooling This is the most rapid method of precooling and it is extremely effective for products that posses both a large surface to mass ratio (leafy vegetables), and an ability to release internal water readily (Dincer, 2003). This method is based on the principle that the boiling point of water lowers as atmospheric pressure is reduced (Sargent, 1988). It achieves cooling by causing water to rapidly evaporate from a product; generating a water loss of 1%, close to 6C (11F) of product cooling. Va cuum cooling units reduce the atmospheric pressure of 101 kPa to 0.6 kPa (Thompson et al. 1998). Spraying the produce with water before vacuum cooling minimizes the product moisture loss, which can range from 2 to 4% of its weight (Thompson, 2004). Prewe tting is especially useful in products with high initial temperatures or that can absorb substantial amounts of the added water on their surfaces before the vacuum is applied (Dincer, 2003). Vacuum coolers are very energy efficient; so, one of their main advantages is their speed and economy. In addition, they also reduce the cost of labor, packaging and the amount of product damaged (Thompson et al. 1998; Dincer, 2003). Their main disadvantage is the need for a high capital investment (Sargent, 1988). Ic e Cooling This method can be used for precooling and also for temperature maintenance during transport. Ice requires a substantial amount of heat to change phase from solid to liquid, hence it has a higher level of heat capacity when compared to water (Sa rgent, 1988). When the product contact s the ice, the heat it has accumulated i s absorbed by the ice, which consequently melts.

PAGE 30

30 This also allows the maintenance of high levels of relative humidity (Sargent, 1988). According to Anderson (2010), there are dif ferent variations of ice cooling, such as topicing and package icing. Top icing involves the placement of finely crushed ice over the top of a package of produce before this one is closed (Sargent, 1988). It is commonly used now as a complement to other c ooling methods (Brosnan and Sun, 2001). Topicing is relatively cheap with respect to other cooling methods; however, the fact that the ice is not uniformly distributed throughout the container can end in slow cooling rates (Sargent, 1988). Cortbaoui (2005) adds that topicing is less efficient in the lower layers of the product, where the ice source is farther away. Package icing allows a more uniform distribution of crushed ice throughout the package, and thus allows a faster and more uniform cooling (Cor tbaoui, 2005). Dincer (2003) explains than, this methods advantage is its simplicity and effectiveness if applied properly, its disadvantage is that it is labor intensive. Dincer (2003) also mentions that, ice -slush, a modification of topicing containin g a mixture of cold water and ice, is also simple and effective if applied properly. Talbot et al. (1991) add that slush-ice allows a wider distribution of the ice due to its liquid nature; facilitating conductive heat transfer. The main advantage of this method is the high relative humidity environment, which hinders moisture loss in the product. Its disadvantages are that it has high capital and operating costs, it requires a package that can withstand constant water contact, it increases considerably the weight on the package, and that the melt water can damage nei ghboring produce (Thompson, 2004).

PAGE 31

31 Room Cooling Although not a precooling method, room cooling is still used in some places whenever the aforementioned precooling methods are not available. According to Meana (2005), room cooling is the simplest method for precooling, since it only needs a refrigerated room with proper cooling capacity. This is used mostly in products with relatively long shelf life ( such as potatoes and onions) as a step prior to storage. These are inside loosely stacked packaging in the cooling room, allowing for ventilation in the side of the containers (Thompson et al., 1998; Thompson et al., 2002). Other relevant factors are the presence of proper package venting and good ai r flow in the room so the cold air can past near and through each package. When these are accomplished, most products will cool in less than 24 hours. According to Thompson et al. (1998), poor room air flow, tightly stacked product, and poor box venting wi ll extend cooling to many days. Most of the internal heat load of the package needs to be transferred by conduction to the surface so that the cold air can remove it by convection. As a result, the cooling rate of this method is very slow when compared to other precooling methods (Meana, 2005). According to Cortbaoui (2005), the advantages of room cooling include its low labor and equipment cost. Yet, as mentioned before, since its cooling rates are low, this method is appropriate for products with a low r espiration rate and those who are not affected by slow cooling such as onion or potatoes (Sargent et al., 1988). Dincer (2003), states that, if not controlled properly, room cooling can end up creating moisture loss issues in the product. Short Term Storag e Once cooled, produce will gain temperature rapidly when exposed to warmer environments Therefore, it is necessary to place it in temperature controlled environments along its supply chain, so optimum conditions are kept and product quality is preserved as much as possible.

PAGE 32

32 Cold rooms then become a suitable storage area before refrigerated transport operations. It is recommended to precool the product and avoid loading it warm into refrigerated transport systems. These systems do not possess the refriger ation capacity needed to bring the temperature down to optimum transport conditions, and so, the product can be exposed to high temperature abuse for long amounts of time (Thompson et al. 1998). Storage facilities should remain around 1C of the desired temperature for the produce stored. According to Kader (2002), temperatures below the optimal range for a given commodity can cause freezing or chilling injury; while temperatures above it can shorten its shelf life Transportation Systems Temperature Con trol One of the most important features for transportation systems used in a cold chain management is their capacity to control temperature. There are many methods available to maintain optimal temperatures in transport vehicles. Mechanical refrigeration, ice cooling, and cryogenic cooling are the most common. Yet, the most widespread method worldwide is mechanical refrigeration. This is used in road, rail, marine, and intermodal transport. In mechanical refrigeration systems it is very important to contr ol frost accumulation on the evaporator coils, for it reduces the cooling capacity of refrigeration units. Defrosting these coils requires using electric heaters or hot refrigerant gas. During the defrost cycle, airflow is stopped to prevent the movement o f the heat produced in the coils towards the produce (Hui et al., 2003). Heat Loads The refrigeration system used in a transport vehicle must remove all heat entering the vehicle from the outside, all heat generated within the vehicle, and any heat contai ned in the vehicle itself (Hui et al., 2003). For cooling capacity calculations, it is recommended using

PAGE 33

33 extreme environmental circumstances encountered along the supply chain. According to Vigneault et al. (2009) the estimations of total cooling capacity can be simplified by adding all heat loads and multiplying the result by a safety factor. In addition, transportation of produce in cold weathers requires the use of heating systems to prevent chilling and freezing injuries. Internal heat loads These incl ude respiratory heat generated by the produce and any field heat that remains within the produce at the beginning of the transportation process If produce is not adequately precooled or if it has gained heat from loading areas, then its internal heat load is larger. External heat loads The external heat loads include all the heat that enters the vehicle from the outside. This can do so through conduction, convection, air infiltration, as well as by radiation (Vigneault et al., 2009). Heat is conducted thr ough the floor, walls, and ceiling of the vehicle; while warm air infiltrates into it through small holes, cracks, drainage holes, broken door seals, and when the doors are opened unnecessarily. Infiltration is by far the most copious source of external he at load in refrigerated trailers. Finally, solar radiation also has an effect on internal temperatures. For example, studies have found that the cooling requirements of stationary vehicles increased by 20% after exposure to sunlight for several hours (Hui et al., 1993). Residual heat loads These include any heat initially contained in the transport vehicle or any heat load not included as internal or external loads. The most common source is the heat present in the air and surfaces inside the transport vehicle; as well as any remaining heat in boxes, pallets and devices used to secure the load (Vigneault et al., 2009).

PAGE 34

34 Air Circulation Systems The air circulation system plays a major role in ensuring temperature control, for it is used to distribute cool ai r around and through the cargo. Vigneault (2009) explains that good air circulation allows the transfer of heat from the cargo and the environment surrounding it, to the refrigeration unit. Moreover, if this air does not get evenly distributed along the lo ad, overheating or overcooling of the product can take place. There are two different types of air delivery systems commonly used in mechanically refrigerated semitrailers, intermodal containers, and railcars: Top air and bottom air. Top -air delivery syst em This is the air circulation method commonly encountered in refrigerated semitrailers and railcars. In this system the refrigeration unit blows cold air all along the ceiling from the front to the end of the trailer. While the cold air is flowing above t he load some of it also moves downward along the side walls. Once it reaches the back end of the trailer, the cold air goes downward and comes back along t he floor, underneath the cargo. When it reaches the front, the air flows upward and returns to the re frigeration unit (Heap et al., 1998; Hui et al., 2003; Vigneault et al., 2009). Bottom -air delivery system The bottom air delivery system is used in intermodal and sea containers. In it, most air movement is vertical, as the refrigeration system blows col d air through the T beam floor of the container. The air flows from the front to the rear of the vehicle and is forced upward through the cargo. When air reaches the rear of the container, it flows up between the load and the rear doors to the ceiling and then returns to the refrigeration unit, in the front through the bulkhead openings (Heap et al., 1998; Hui et al., 2003; Vigneault et al., 2009).

PAGE 35

35 Kader (2002) recommends us ing a stowage pattern that force s refrigerated air to flow through and around th e p ackages and do not allow air to bypass around the pallet units. Along the same lines, Vigneault et al. (2009) suggests employing pallets, boxes and inner packaging with enough venting and airspaces to allow vertical airflow through the pallet load. Main Tr ansportation Modes According to Peleg (1985), some of the many factors involved in choosing an optimal transportation chain are as follows: distances to markets; cost per ton kilometer; types and varieties of produce; and climatic conditions en route requi ring refrigeration, ventilated cooling, or heating for prevention of freezing. Additional constraints are types of packaging used, types of available handling techniques, and unitizing methods (pallets, slip sheets, intermodal containers, etc.). Overland t ransport by trucks This is by far the most popular mode of overland fresh produce transportation. There are two types of vehicles used in highway transport: refrigerated semitrailers and intermodal containers. Refrigerated semitrailers can be classified as intermodal transport vehicles. Detached from the tractors the semi -trailers can be transported on railroad flatcars, driven right into sea vessels, or simply hauled by a tractor -trailer on the highway. They are available in lengths of 12 m, 13.7 m, 14.6 m or 16.2 m; and most of them use mechanical refrigeration (Hui et al., 2003). In addition, new units can provide heat when the trailer is operated in ambient conditions colder than the set point temperature (Kader, 2002). Most of the heat that the refrig eration unit removes comes from heat conducted across the walls and from air leaking into the trailer. Therefore, if the product is not center loaded it will be warmed when in contact with the walls. Controlled atmospheres are not applied in highway

PAGE 36

36 traile rs because these are not airtight enough; yet, modified atmospheres can be achieved when using semi -permeable films around the packages or pallet (Kader, 2002). Overland transport by railcars The main advantages of overland rail transport are fast service to distant points and better efficiency in terms of diesel fuel per ton kilometer; also, the lack of traffic (Peleg, 1985). They are mostly used to transport potatoes, citrus fruits, onions, carrots, and other lesser perishable commodities (Kader, 2002). A ccording to Hui et al. (2003), four different vehicles can be used for rail transport: The ice refrigerated railcar, the mechanically refrigerated railcar, the refrigerated semitrailer on flatcar (piggy -back), and the intermodal container on flatcar. In mechanically refrigerated railcars the cold air is distributed vertically downward through the load from the ceiling or from wall flues on side walls and the far end of the car. In both systems, warm air returns to the refrigeration unit through the floor ducts (Huit et al., 2003). Their refrigeration systems use an electric motor that is powered by a diesel generator which can be detached and allow the connection of the railcar to electricity (Kader, 2002). Apart from refrigeration, heating equipment may also be required for winter transport of fresh produce in cold climates. Insulated standard boxcars and artificial heating can be used to protect from chilling and freezing injuries (Peleg, 1985). These rail cars are suitable for modified atmosphere trans portation. However, since they are very airtight, unintended atmospheric modification can take place in them if drain vents are clogged or water in them freezes in the winter (Kader, 2002). Marine transport This method is generally selected for intercontin ental shipments since it is the most economical mode of transportation over long distances and the most energy efficient one too, especially when large volume is involved (Peleg, 1985; Heap et al., 1998; Hui et al., 2003). A

PAGE 37

37 typical modern middle size reef er ship will have a capacity of approximately 12,000 m3, divided into around 4 holds, 8 temperature zones and 12 to 16 cargo chambers (Stera, 1999). The main difference between using containers or refrigerated ships is their cargo -carrying capacity: A cont ainer can carry about 1,000 to 1,500 packages, while a refrigerated ship has capacity of about 350,000 packages (Kader, 2002). Travel times, however, are generally l onger often around 1 to 4 weeks; but produce being shipped overseas may spend on board eve n 5 to 6 weeks. Thus, good temperature and humidity control is essential. Thompson (2003) states that produce for exportation is generally transported in temperature-controlled cargo space be it in break bulk (conventional refrigerated ships) or in cont ainers or reefers (container ships). Conventional ships These are completely insulated vessels and have a series of holds; each one divided into three to five cargo areas. Generally, each cargo compartment will have its own refrigeration coil and fresh air ventilation with independent temperature control (Heap et al., 1998). These ships can carry refrigerated goods, frozen foods, and nonperishable items in different holds or rooms at the same time. They mostly use the break bulk system (Hui et al., 2003) Break bulk refers to a system of transport where individual boxes or pallets of produce are stacked directly in the hold of the ship. Yet, most produce has been pall etized before shipping and remain s palletized along the supply chain. Thompson (2003) me ntions that the main benefits of palletization are the reduction of handling which then reduces labor costs and damage to the product. The method of stacking cargo in the hold is critical for the maintenance of optimal temperatures. Spaces must be left be tween rows, stacks, or layers, so that the upward circulating air can reach one or more surfaces of each shipping container (Peleg, 1985; Hui et al., 2003).

PAGE 38

38 Refrigerated ships are loaded in open docks and expose produce to the elements (heat, freezing tem peratures, or rain). This is especially problematic in ships that use a common refrigeration system for two compartment levels, because the refrigeration system is not operated until both compartments are loaded (Kader, 2002). Smale (2004 ) explains that th ere are two common systems of air distribution in reefer holds: The longitudinal air delivery system and the Robson system. Yet, with respect to their interaction with the load, their effect is similar since both of them provide bottom air delivery systems in the cargo area. Due to their exposure to the elements during loading, it is expected that produce transported in cargo holds regain some of the temperature lost during precooling operations. Hence, the ship must have sufficient refrigeration capacity and air circulation to lower the temperature of the cargo (Hui et al., 2003). Reefer holds are less affected by infiltration heat and to the frosting of their evaporator coils when compared to reefers. However, both of these systems share the need for ren ewal of refrigerated air as a mean to avoid the build up of respiratory products i n produce shipments (Smale, 2004). Container ships Container ships have specially designed holds that have vertical guides to stack containers below the deck. These can be st acked six to nine high below deck, and three to four high above deck. Protection from the sun to those carried on deck may be given with an upper layer of nonrefrigerated containers. One o f the main benefits of this type of ship is that they require minim al handling of the containers cargo; and as such, give fewer opportunities to break cold chain when compared to the break -bulk system. Therefore, the quality of the produce transported in reefers at

PAGE 39

39 the port of destination should be better than when this is handled by the break bulk system in conventional refrigerated ships (Hui et al., 2003; Smale, 2004 ). Mechanica lly refrigerated containers or reefers are insulated containers usually equipped with a recessed refrigeration unit (Thompson, 2003; Hui et al ., 2003). Many of these run on a diesel engine during highway or rail travel and then are plugged on shipboard to operate with an electric motor. Reefers are generally carried on deck to ensure that there is sufficient air to remove the heat rejected by th e refrigeration units condenser (Hui et al., 2003). Air circulations rates both in vessels and in reefers are high right after loading the cargo as part of the initial cool -down. Once the product is cooled to the optimal transport temperature and the ref rigerated air has been properly distributed along the cargo space, the system switches to lower air circulation rates in order to reduce refrigeration load and power consumption (Smale, 2004 ). The refrigeration capacity of a reefer depends on ambient tempe rature, insulation quality, respiration rate of the cargo, initial temperature of the cargo, the rate of fresh air exchange and leakage (Smale, 200 4 ). Irreversible damage and loss of product quality can be caused by small periods of equipment malfunction; thus continuous monitoring of the temperature and the functioning of the entire cooling system should take place all along the chain (Ruiz Garcia, 2008). Heat leakage in a insulated container will be about 22 W K1 for newer 6.1 m ( 20 ft ) ones, but some c ontainers specially made for fresh fruit will have thinner insulation with a heat leakage of around 35 W K1. Values for 12. 2 m ( 40 ft) containers are approximately double these figures. Insulation efficiency reduces with time by about 3 5% per year (Heap, 1989;

PAGE 40

40 Smale, 2004). Some reefer containers are suitable for modified atmospheres and controlled atmospheres usage during produce transport (Thompson, 2003). Intermodal transport Intermodal freight transport allows the movement of goods in a cargo unit by successive modes of transport with no handling of the goods themselves during changes in transport modes. According to Ruiz Garcia (2008), intermodality is characterized by the transferability of the transported items between modes and a unique system of administration and billing. Intermodality is important because it promotes continuity in the cold chain. Additionally, it allows handling the product without breaking up the cargo, which results in reduced shipping costs and less damage to the produce (Hu i et al., 2003). Sea containers are the most common type of equipment used in intermodal transport. There are two general size classes: 20 ft (6.1 m) and 40 ft (12. 2 m). Specifications can be found in the ISO standard 668 (Smale, 2004 ). Airfreight transpo rt This method is generally used to carry, over long distances, produce that has a short shelf life and therefore cannot withstand the trip through road, rail, or sea. Air transport is more expensive than other transportation methods; however, it gives a q uicker return on the capital that is tied up during transportation (Hui et al., 2003). According to Kader (2002) air transportation provides poor temperature control when compared to the other transportation methods. In addition, great concern is paid to t he high environmental costs and the carbon footprint related to transporting produce via air cargo (Thompson, 2003). Cargo is generally stacked in unit load devices (ULDs) before the time of loading the aircraft. The two main categories of ULDs are aircraf t pallets and aircraft containers. Intermodal

PAGE 41

41 containers, traveling with their refrigeration units off can also fit in certain freight aircrafts (Heap et al., 1998; Hui et al., 2003). Aircraft cargo holds are also pressurized, just like passenger cabins at less than sea level pressure At high altitudes, outside air temperatures can be as low as 55C or 60C, but the inside cargo ambient temperature ranges from 5C 25C (Heap et al.,1998). According to Hui et al. (2003), certain wide -bodied aircraft no w have temperature controlled holds during the flight (i.e. at normal cruising altitude up to 7C on a hot day). However, Kader (2002) asserts that most air transport containers are not refrigerated and provide minimal air circulation. Furthermore, Nunes e t al. (2006) states that the temperatures inside the containers are around 18 C because the cargo holds are generally fitted to accommodate warm -blooded animals. Since mechanical refrigeration systems are not generally used on aircraft containers (Peleg, 1985); flexible insulating material is used as cover for the ULDs as a mean to delay the warming of the produce (Hui et al., 2003). One of the main problems with air transport of produce is that at high altitudes, air humidity in planes is extremely low a nd can cause product dehydration if the product is not packaged correctly or placed in a fairly airtight box (Kader, 2002). Staging at departure and destination airports can represent additional breaks in the cold chain due to the fact that these areas are generally not refrigerated; thus, endangering the final quality of the product (Kader, 2002). Temperature Variations during Transport Recommendations for food transports allow deviations of 0.5C from the set point (Jedermann and Lang, 2007). However, environmental temperature can differ from each other depending in the location of the logger, packing material, or heat dissipation of the product

PAGE 42

42 (Verboven et al., 2005) and there is a broad variance in the speed of temperature changes, depending on the tra nsport conditions (Jedermann and Lang, 2007) Once an optimal set point temperature for transportation has been established, achieving proper temperature management depends on several factors. Packaging design and materials, for example, can hasten or hinde r the heat transfer process, and dramatically affect the temperature distribution inside the pallets and throughout the cargo environments (Smale, 2004). With produce, respiratory heat generation can pose problems, causing the cartons in the centre of a p allet to be significantly warmer than those on the outside of the pallet (Tanner and Amos, 2003). Such temperature variability can lead to variable quality at out turn, which in turn can lead to marketing problems. But packaging design provides a relativel y simple means by which these processes can be controlled, so packaging becomes a major factor in determining the rate of heat and mass transfer during refrigerated storage and transport of horticultural products. The traditional approach to dealing with r espiratory heat has been to design cartons with open vents; however the vents are usually designed to facilitate horizontal airflow despite vertical airflow being the standard configuration in transport systems (Smale, 2004 ). In the majority of food refrig eration systems, heat is transferred primarily by convection; therefore, the temperature and its homogeneity are directly governed by the patterns of airflow. Most of the time, nonuniform airflow is one of the major causes of temperature variability. For sensitive products, this level of temperature variability may have significant food quality and safety implications (Smale et al., 2006). Previous research (Billing et al., 1993; Billing et al., 1995; Amos, 2001; Tanner and Amos, 2003; Moureh and Flick, 2004; Punt and Huysamer, 2005; Rodriguez Bermejo et al., 2007) has described the existence of spatial and timebased temperature variability both inside pallets of

PAGE 43

43 produce, and in the cargo areas that harbored them along their supply chains. For example, Punt and Huysamer (2005) discovered temperature differentials of about 4.5C in air temperature at the pallet level in plum shipments. Research performe d by Billing et al (1993, 1995) and Smale (2004) also reported the existence of temperature gradients with in loads of produce during transportation in refrigerated sea containers and vessels. Tanner and Amos (2003), found delivery air temperatures that varied across the width of a container severely enough to result in air temperatures capable of freezing the cargo. They also discovered significant time -based temperature variability during the shipment, particularly between defrosts and in the time period corresponding to travel over the equator. In addition, studies in refrigerated trucks (Meffert and Van Nieu wenhuizen 1973; Gogs and Yavuzturk 1974; Lenker et al., 1985; Bennhamias, 1993; Le Blanc et al., 1994; Moureh et al., 2002; Finn and Brennan, 2003; Tapsoba et al., 2006; and Moureh et al., 2009) have discovered the existence of hot spots in the rear part of these vehicles created by poor ventilation because of uneven air distribution and air short -circuiting around the front of the truck. Temperature Measurement Devices There are many kinds of temperature sensors in the market. The easiest way to class ify them is either as contact or non -contact temperature measurement devices. Contact sensors measure their own temperature; but they obtain the temperature of the object they are in contact with by assuming that the two are in thermal equilibrium and ther e is no heat flow between them. Noncontact temperature sensors generally measure the thermal radiant power of the infrared or optical radiation that they receive from a known or calculated area on its surface, or a known or

PAGE 44

44 calculated volume within it (Tem peratures, 2010). Following are described the most popular temperature measurement devices in the market. Thermocouples A thermocouple is a thermoelectric temperature sensor that consists on two different kind of wires joined together at the probe tip (mea surement junction) and extended to the reference junction where a known temperature is recorded. According to Efunda (2010), the temperature difference between the probe tip and the reference junction is detected by measuring the change in voltage at the reference junction. Then, the absolute temperature reading is obtained by combining the information of the known reference temperature and the difference of temperature between probe tip and the reference. There are different types of thermocouples based o n the different combinations of metals or calibrations. The four most common calibrations are J, K, T and E, which are base -metal thermocouples and can be used up to about 1000C (1832F). There are high temperature calibrations R, S and B, which are noble -metal thermocouples and can be used up to about 2000C (3632F). Each calibration has a different temperature range and environment, although the maximum temperature varies with the diameter of the wire used in the thermocouple. Even though the thermocoup le calibration determines the temperature range, the maximum range is also limited by the diameter of the thermocouple wire, which means that a very thin thermocouple may not reach the full temperature range (Efund a, 2010; Omega, 2010). Table 2 1 provides a summary of basic thermocouple properties. Resistance Temperature Detectors (RTD) Resistance temperature detectors or resistance thermometers are devices that measure temperature based on the premise that the resistance of metals increases with temperatur e (Michalski et al., 2001). Therefore, measurements are obtained by correlating the resistance of

PAGE 45

45 the RTD element with temperature. Most RTD elements consist of a length of fine coiled wire wrapped around a ceramic or glass core. The element is usually qui te fragile, so it is often placed inside a sheathed probe to protect it. The RTD element is made from a pure material whose resistance at various temperatures has been documented and has a predictable change in resistance as the temperature changes (Omega, 2010). They are active devices requiring an electrical current to produce a voltage drop across the sensor that can be then measured by a calibrated read -out device (Temperatures, 2010). The resistance of commercially available RTDs ranges from 10 to 25,000 The most common are 100, 200, and 1000 strain -free platinum (>99.999%) probes and 10 copper probes. On a general basis, the higher the resistance, the less affected the RTD will be due to small resistance/voltage fluctuations in the lead wires and c ircuit. Some of the common metals used in RTDs include platinum, copper, nickel, BalcoTM (70% Ni 30% Fe), and tungsten (Efunda, 2010). Table 2 2 l ists their temperature ranges. Platinum RTDs (also called PRTs and PRT100s) are the most popular RTD type. The se are nearly linear over a wide range of temperatures and some are small enough to have response times of a fraction of a second. PRTs are among the most precise temperature sensors available with resolution and measurement uncertainties or 0.1 C or bet ter (Temperatures, 2010). RTDs are usually encapsulated in probes for temperature sensing and measurement with an external indicator, controller or transmitter, or enclosed inside other devices where they measure temperature as a part of the device's funct ion, such as a temperature controllers or precision thermostats (Temperatures, 2010).

PAGE 46

46 Thermistors (Thermal Resistors or Bulk Semiconductor Sensors) According to Michalski et al. (2001) thermistors are non-linear, temperature dependent resistors with a high resistance temperature coefficient. Only thermistors with a negative temperature coefficient are used for temperature measurements. Unlike an RTD's metal probe where the resistance increases with temperature, the thermistor uses ceramic semiconducting mat erials which respond inversely with temperature (Efunda, 2010). The resistance is measured by passing a small, measured direct current (dc) through it and measuring the voltage drop produced (Temperatures, 2010). Typical thermistor sensors can measure temp eratures acro ss the range of 40 ~ 150 0.35C ( 40 ~ 302 0.63F). A thermistor probe can take the shape of a bead, washer, disk, or rod. Typical operation resistances are in the k range, although the actual resistance may range from several M to severa l (Efunda, 2010). Radiation Thermometers (Pyrometers) A Pyrometer, is a noncontact temperature sensor that detects an object's surface temperature by measuring the temperature of the electromagnetic radiation (infrared or visible) emitted from the object (Efunda, 2010). This group of sensors includes both spot measuring devices in addition to line measuring radiation thermometers, which produce one -dimensional and can sometimes produce two dimensional temperature distributions, and thermal imaging, or ar ea measuring, thermometers which measure over an area from which the resulting image can be displayed as a two -dimensional temperature map of the region viewed (Temperatures, 2010). According to Efunda (2010) pyrometers manipulate the fact that all objects above absolute zero temperature 0K ( 273.15C; 459.67F) radiate and absorb thermal energy. Then, if the relationship between the radiation intensity and wavelength and the temperature can be

PAGE 47

47 established, the temperature can be found from the radiation. Pyrometers are then essentially photodetectors which are capable of absorbing energy, or measuring the electromagnetic wave intensity, at a particular wavelength or within a certain range of wavelengths. Common pyrometers include: Optical Pyrometers, whi ch are designed for thermal radiation in the visible spectrum; and Infrared Pyrometers, which are designed to detect thermal radiation in the infrared region (0.75 ~ 1000 m; 30 in ~ 0.04 in) usually 2 ~ 14 m (80 ~ 550 in) (Efunda, 2010). According to O mega (2010), the most basic design of an infrared pyrometer consists of a lens to focus the infrared (IR) energy on to a detector, which converts the energy to an electrical signal that can be displayed in units of temperature after being compensated for a mbient temperature variation. Pyrometers allow the measurement of temperatures in applications where conventional sensors cannot be employed, such as in cases dealing with moving objects or where non -contact measurements are required because of contaminat ion or hazardous reasons, where distances are too great, or where the temperatures to be measured are too high for thermocouples or other contact sensors (Omega, 2010). Non -Electric Thermometers These temperature sensors are the ones most recognized by the general public. There are three main kinds: Liquidin glass thermometers, bimetallic thermometers and manometric thermometers. Liquid in -glass thermometers are based upon the temperature dependant variation of the volume of the liquid which is used. The t hermometer consists of a liquid filled bulb connected to a thin capillary where the temperature scale is shown. Common liquids employed include mercury, ethanol, toluene and pentane (Michalski et al., 2001).

PAGE 48

48 Bimetallic thermometers are also called bimetallic strips. They are used to convert temperature change into mechanical displacement. This displacement may be coupled to a switch for simple on off function, to a needle of an indicator, or to a position detector for electronic output (Capgo, 2010). The st rip consists of two strips of different metals which expand at different rates as they are heated. These metals are joined together throughout their length by riveting, brazing or welding. The different expansions force the flat strip to bend one way if he ated, and in the opposite direction if cooled below its initial temperature. The metal with the higher coefficient of thermal expansion is on the outer side of the curve when the strip is heated and on the inner side when cooled (Wikipedia, 2010). In some applications the bimetal strip is used in the flat form, but in others it may be coiled to make it more compact and sensitive, with temperature changes causing the coil to tighten or unwind. The most common application of bimetallic strips is being the on/ off switch of thermostats; they are also used in strip chart recorders (Capgo, 2010). According to Michalski et al. (2001), manometric thermometers can be divided into volume or variable pressure type. Variable volume thermometers are liquid filled units t hat generate the movement of the pointer indicating the temperature measurement due to the liquids expansion when the temperature increases, while the variable pressure type generates measurements depending upon the thermometric behavior of the vapors and gases they enclose. Radio Frequency Identification (RFID) Radio Frequency Identification (RFID) is an automatic identification (Auto -ID) technology that allows the identification of physical objects by employing tags attached to these that communicate inf ormation about them wirelessly through radio waves to a reader. According to Banks (2007), the basic principle is not much different than that of a barcode: Encoding an

PAGE 49

49 identifier number in a machine readable form that can be accessed quickly and reliably, with no human translation. However, unlike bar codes, RFID does not require line of -sight. A RFID system consists of four elements: RFID tags, RFID readers, reader antennas, and information processing software (Thyagaraja, 2007). The reader sends radio fr equency (RF) waves through its antenna and collects the RF waves emitted or reflected from the tag. The information, such as a unique serial number for every tag, is carried with these RF waves. Industrial -Scientific -Medical (ISM) Bands Radio waves are pa rt of the electromagnetic spectrum which use is regulated by governments around the world. Governments have then assigned different uses for the various parts of this spectrum (Angeles, 2005). Industrial -Scientific -Medical (ISM) bands are special license -f ree bands that have been set aside by regulatory bodies across the world and are available for use in all countries. Anyone can use unlicensed frequencies as long as they follow the rules of transmission and broadcast (Sweeney, 2005). By using ISM bands RF ID system operators can avoid licensing processes; but, will still have to follow closely the ISM rules on the use of the band, limits on radiated power, and tolerance of interference (Hunt et al., 2007). Carrier Frequency Carrier frequency is the center o f a particular RF bandwidth. For instance, a 915 MHz carrier frequency with a 10 MHz bandwidth encompasses bands from 910 to 920 MHz (Reed, 2009). Common standard frequencies include 135 kHz, 13.56 MHz, 868 870 MHz and 2.45 GHz. The frequency determines read range, the ability of the wave to penetrate objects and data transfer rate. Higher frequencies have longer read range but less penetrating ability. They can also accommodate higher data rates, thus using less power per unit of data (Angeles, 2005; Reed, 2009). Yet, these systems need more power to transmit the same distance as systems using lower

PAGE 50

50 frequencies, a design trade -off. Lower frequency devices have less read range, and slower communication speed, but good penetrating ability (Watts et al. 2002) Following is the description of the main characteristics of different RFID systems associated to specific frequency bands: Low frequency (LF) LF ranges from 30 KHz and 300 KHz. A typical LF RFID system operates at 125 KHz or 134.2 KHz. These systems use inductive coupling as communication method. LF waves perform very well in environments containing metals, liquids, dirt, snow, or mud (Lahiri, 2006; Roussos, 2006). However, according to Dagdelen (2007) and Leong et al. (2006 ) they have longer read times ( with data rates, on the order of Kbits/s), higher costs, and larger tag sizes. Nonetheless, one of its advantages is that this range is accepted worldwide. High frequency (HF) These systems use frequencies ranging from 3 MHz to 30 MHz. The ISM band commonl y used in HF RFID systems is 13.56 MHz. HF also uses inductive coupling as communication method. HF waves can penetrate through water, yet metallic objects are still problematic (Dagdelen 2007). In addition, they present longer read times compared to othe r systems. The HF frequency range is also accepted worldwide (Lahiri, 2006). Ultra high frequency (UHF) The UHF band ranges from 300 MHz to 1 GHz. According to Dagdelen (2007), in this frequency range RFID systems communicate through electromagnetic coupl ing (backscattering, for example). The UHF frequency band has a number of advantages. This frequency range provides fast reading rates and offers longer reading distances than other RFID frequencies (Sivakumar and Deavours, 2008). In addition, greater tran smitter power is permitted at UHF frequencies than at microwave frequencies, such as 2.4 GHz (Redemske and Fletcher 2005).

PAGE 51

51 Systems operating in the high UHF range (915 MHz in the US, 868 MHz in Europe) are highly affected by the presence of metals and li quids in their surroundings (Leong et al., 2006 ); nevertheless, low UHF frequencies such as 315 MHz and 433 MHz perform better under the same conditions. The UHF range is not accepted worldwide; each country has restrictions with respect to which UHF frequ encies are allowed within their territory. Microwave frequency Microwave frequency ranges upward from 1 GHz. RFID systems operate either at 2.45 GHz or 5.8 GHz. They also use electromagnetic coupling as communication method. They possess the fastest data -transfer rates between the tag and the reader (reaching up to the Mbit/s range); and are the most affected with respect to the existence of metals or liquids in their surroundings. The 2.4 GHz frequency band belongs to the ISM band, and therefore it is acc epted worldwide (Lahiri, 2006). Materials Hindering the Signal According to Lahiri (2006), materials can be grouped within three classes with respect to the way they affect the radio waves propagating through them (Table 2 3) RF lucent materials to a cert ain frequency allow radio waves at this frequency pass through it without any substantial loss of energy. RF -opaque materials block, reflect, and scatter RF waves. And finally, RF absorbent materials allow the radio waves to propagate through them but with significant loss of energy. The RF absorbent or RF -opaque property of a material depends on the operating frequency. So, a material that is RF absorbent at a certain frequency could be RF lucent at a different frequency. For example, water and metal are RF lucent at low frequencies while at the highest ones (UHF, Microwave) they turn into RF absorbent and RF opaque materials, respectively.

PAGE 52

52 Metal is an electromagnetic reflector and radio signals cannot penetrate it (Roussos, 2006). This situation prevents the tag from absorbing enough energy from the reader, since the metallic material holds most of it, therefore detuning the operating resonant frequency at which the tag was supposed to operate (Banks et al., 2007; Reed, 2009). As a result, metal will not o nly block communication if placed between a tag and an interrogator; but it can also affect the operation and characteristics of the antennas located nearby (Hunt et al., 2007). A proposed solution for tagging metallic items is to either use antennas with precise tuning or to employ tags mounted on a foam platform, which avoids the direct contact with the metal (Banks et al., 2007). Liquid materials such as soap, water, or salty solutions are RF absorbent in the UHF and Microwave frequency ranges, and dimin ish the functionality of the RFID systems ( Leong et al., 2006; Roussos, 2006; Aroor and Deavours, 2007; Banks et al., 2007; Sivakumar and Deavours, 2008). Thes e materials absorb considerable amounts of the energy of the radio waves arriving to the transpon der, thus reducing the available energy necessary to operate the tag. If there is severe absorption, the transponder will not be able to capture the necessary energy to operate and the reader will not detect its signal (Banks et al., 2007; Reed, 2009). Ba nks et al. (2007) also states that the differences in the impact of RF absorbent or RF opaque objects on HF and UHF derive from the ratio of the wavelength to the object size. The author also proposes a classification of the object based on its dimensions relative to the wavelength: Rayleigh range The wavelength is much larger than the object dimensions. So, for objects smaller than half the wavelength, the reflection is minimal and metal would not interfere in the normal operation of a tag.

PAGE 53

53 Optical range The wavelength is smaller than the object size. In this case a metal would appear fully "visible'' to an RF signal. Environmental factors also play a very important role in RFID system performance. Wind, rain, electric motors, machinery etc. are some fa ctors that need to be considered when trying to achieve higher readability. For example, Ultra High Frequency (UHF) radio waves are impacted much more by wind, snow, rain and sun than High Frequency (HF) tags. In addition, direct sunlight on the chip also alters the read rates significantly. Furthermore, the presence of other radio sources such as cell phones and mobile radios, and the existence of electrical motors in the surroundings can also have a dramatic effect on the systems performance. Finally, RF waves can also receive interference generated by electrostatic discharge (ESD), which is a sudden flow of electrical current through a material that is an insulator under normal circumstances (Lahiri, 2006; Potdar et al., 2007). Multipath Effect A "multip ath effect'' is created by the reflection of the RF signal on RF reflecting objects when these are present in large amounts in the operating environment (Banerjee et al., 2007; Bankst et al., 2007). It happens because the reflected RF waves get scattered and take different propagation paths and phases after the original RF waves hit the metal. These different phases will then create interference amongst signals (Banks et al., 2007). Reed (2009) adds that in addition to phase changes, polarization alteration s in the wave will also take place. Lahiri (2006) explains that the reflected waves can arrive at the reader antenna at different times through different paths. Sometimes these waves are in phase, matching precisely with the original antenna signals wave pattern; and thus, enhancing the original antenna signal. This event is called constructive interference and is responsible for the protrusions in the antenna patterns On the other hand, when the waves arrive out of phase, as the exact opposite of the

PAGE 54

54 or iginal antenna wave pattern, destructive interference arises as these waves cancel each other out. This results in the creation of nulls in the antenna pattern. RFID System Components As aforementioned, a RFID system consists of RFID tags, RFID readers, re ader antennas, and information processing software. RFID tags A RFID tag is a device that can store and transmit data to a reader in a contactless manner using radio waves (Lahiri, 2006). RFID tags have both a microchip and an antenna. The microchip is use d to store object information such as a unique ID; while the antenna enables the microchip to transmit information to th e reader, which transforms information to a digital format compatible with computers (Angeles, 2005). According to Banks et al. (2007) RFID tags can also contain memory that provides the ability to re cord information sent by the reader. In addition, some tags also contain batteries, and their existence and purpose give origin to a way of classification. Classes of RFID tags Based on their powering mechanisms for transmission and their emission behavior, there are four kinds of RFID tags: passive, semi -passive, active and semi active tags. Passive tags are unable to emit signal unless energized by a RFID reader, have no internal power supp ly and are inexpensive ( Mndlein et al., 2004). They also have a range of operation (in terms of the distance from the reader) limited to only a few meters (Banks, 2007). Semi passive tags contain a battery that only powers thei r logic parts (such as sen sors) It is necessary the stimulation of the RFID reader for them to transmit a signal (Amador et al., 2009). They have a longer range of readability when compared to a passive tag (Angeles, 2005).

PAGE 55

55 Active tags need a power source (battery) to power on t heir own (Clarke et al., 200 6 ; Kapoor, 2008). These tags constantly emit Their main advantage is the fact that they have larger read ranges compared to passive tags and can carry sensors (Chamberlain, 2006). However, they have higher costs and big sizes c ompared to the passive and semi -passive systems. Semi active tags which are quite similar to active tags, but do not continuously emit; instead, they are often activated manually (Dagdelen, 2007). They have an extended battery life compared to active tag s and have a longer read range than any passive or semi -passive system. Their drawback is having costs and sizes similar to the active tags. EPCTM tag classifications Tags can also be classified by the standards created by EPCglobal, Inc. (Table 2 4) Thi s institution has defined six classes for RFID tags (0 to 5). According to Banks et al. (2007), classes have greater capability as their number increase. Class 0 to 3 generally include passive tags, while class 4 describes active tags, and class 5 consists of tag readers and active tags that can read data from other tags Following is this authors description of the classes: Class 0 / Class 1 These classes provide the basic RF passive capability. Class 0 is factory programmed. Beyond class 0, including class 1, the tags are user programmable. Class 2 Additional functionality is added, such as encryption and read-write RF memory. Class 3 Batteries that will power logic component of the tag are found on board. Class 3 provides longer range and broadband comm unications Class 4 Active tags are part of the definition of class 4 tags. Peer to -peer communications and additional sensing are also included. Class 5 Class 5 tags contain enough power to activate other tags and could be effectively classified as a rea der.

PAGE 56

56 RFID Reader Readers (or Interrogators) are the electronic components that transmit and/or receive the radio frequency waves used to communicate with the tags (Banks et al., 2007). They generate signals that provide power for passive and semi -passive t ags as well as interrogate the tags. A tag captures the energy it receives from a reader to supply its own power (when possible) and then carries out commands sent by the reader, the easiest one being sending its ID. Upon receiving information from a tag, the reader decodes the information in its decoding software and then transmits it to the information management system (Want, 2006; Thyagaraja, 2007). Angeles (2005) explains that the technical characteristics of the reader are highly important because the read range of the tag depends in part on both the readers power and the frequency used to communicate. Lahiri (2006) proposes the following classification of readers based on its mobility: Stationary Handheld Stationary reader A stationary or fixed rea der is what its name implies. Fixed -position interrogators can be mounted in dock doors, along conveyor belts, and in doorways to track the movement of objects through any facility (Hunt et al., 2007). Handheld reader A handheld reader is a mobile reader t hat a user can operate as a handheld unit. It generally has built in antenna(s) (Lahiri, 2006). Hand-held readers allow the deployment of RFID systems in locations where it's not feasible to install fixed -position interrogators (Hunt et al., 2007).

PAGE 57

57 Reader Antenna The antenna is a separate device that is physically attached to a reader. Its main goal is to provide a communication channel between the reader and the tags, sending and receiving RF signals from them. According to Reed (2009), transmitting antennas must radiate so that the receiving antenna in the tag can accept the signal. RF signals lose energy with distance, so enough power must be directed to the area where a receiver is located. Transmission power is applied to a devices antenna in order to generate electromagnetic energy in the form of RF waves. For a given transmitter power, a low -frequency signal is able to travel farther than a high -frequency signal. Following is a description of two of the main technical characteristics for antennas: A ntenna Footprint and Antenna Polarization. Antenna footprint An antenna footprint or antenna pattern is a three -dimensional region shaped like an ellipsoid or a balloon projecting out of the front of the antenna where the antennas energy is most effective The footprints of the readers antennas govern the read zone of a reader; and within this range tag readings are easily obtained. The footprint of an antenna generally has protrusions surrounded by dead zones (or nulls), so it is never a perfect ellipsoi d. Since readings near these areas lack robustness, it is recommended to keep the tags inside the main ellipsoid shaped region (Lahiri, 2006). Null spots can also occur from the detuning of tags, which occurs when two tags are placed in close proximity to one another or in close proximity to liquids, metals, and other materials with a high dielectric permittivity (Hunt et al., 2007).

PAGE 58

58 Antenna polarization Propagation is the direction and path that a radio wave travels, and polarization describes the electric field vector along the propagation path (Reed, 2009). Dagdelen (2007) defines it as the orientation of the electric field of the wave. Antenna polarization and the angle at which the tag is presented to the reader greatly impact the readability of a tag, its reading distance and the reading robustness of the system (Lahiri, 2006). Most antennas radiate via either linear or circular polarization. According to Reed (2009), antennas are designed to have a particular polarization. Using the analogy of a thread on a nut and bolt, the author explains how antennas and RF waves must match to work together in wireless communications. For example, an antenna with right hand circular polarization would have difficulty receiving a reverse -polarized signal. Consequently, optimized antenna orientation promotes the best wireless communication performance. Linear polarized antenna Linear polarization takes place when the electric field is aligned in a single plane (Reed, 2009). These waves have only one energy field and the antennas radiation pattern remains in the same plane at all times. Dagdelen (2007) explains that linear polarization can be vertically or horizontally oriented depending on the propagation direction with respect to the earth surface. Thus, a linear polar ized antenna will be sensitive to tag orientation with respect to its polarization direction. Finally, when compared to circular polarized antennas, linear polarized ones present a narrower radiation beam with a longer read range (Lahiri, 2006). Circular polarized antenna A circularly polarized antenna radiates an electric field which is the sum of two fields that have equal amplitude and magnitude with a 90 phase difference ( Dagdelen 2007). These two components form a spiraling field vector, which rotat es and completes a full 360 within one

PAGE 59

59 wavelength ( Dagdelen 2007; Reed, 2009). Circularly polarized antennas can have a clockwise rotation or a counter clockwise one. The main benefit of these antennas is that they are largely unaffected by tag orientati on. Circular polarized antennas can read tags in a wider area compared to linear polarized antennas because they have a wider radiation beam (Lahiri, 2006). Information Processing Software In a RFID system software is critical for filtering and analyzing data as well as integrating RFID technology into backend databases (Thyagaraja, 2007). According to Banks et al. (2007), middleware is the software component that transforms low level RFID hardware information coming from the readers to usable event inform ation. Hunt et al. (2007) states that middleware plays an important role in the quality and usability of the information produced by RFID systems because it allows the transmission of data between the RFID networks and the IT systems within an organization According to Clarke et al. (2006), legacy systems can be used to make instantaneous decisions based on the incoming data from the RF readers; for example, triggering replenishment when an order causes inventory to drop to the reorder point. RFID Applicat ions Some of the current applications of RFID are: Supply chain automation, inventory management, parcel and postal tracking, access control, airport luggage, self check -out, and medical ID bracelets, automatic toll collection, retail parking access, retai l stock management, library book tracking, vehicle immobilization, theft prevention systems, livestock tracking, environmental monitoring, and tracking exact timing in sports events (Griffin et al., 200 5 ; Potdar et al., 2007). Applications of RFID in the Produce Industry RFID offers three major applications suited for the produce packing industry: Identification, Trace Back and Process System Traceability.

PAGE 60

60 Identification According to Panos and Freed (2007), the majority of the produce facilities in the US have minimal information systems capabilities, being most of their logistic operations still documented on paper. Nonetheless, some members of this industry use automatic data collection through bar code systems (pallet tags) in order to identify the produ cts being moved along the supply chain. This feature facilitates the inventory management of all parties involved, increases their inventory accuracy and enhances the visibility of the products throughout the chain. RFID provides a faster alternative to ba r code systems because it does not require line -of sight between the tag and the reader and can collect the information of many tags at the same time. It also reduces labor costs and allows the storage of bigger amounts of information than the ones current ly contained in the bar code. Taking these benefits into account, in 2003 the U.S. Department of Defense mandated all of its suppliers to comply with RFID tagging at both the pallet and case levels. In a similar fashion and almost at the same time, Wal -Mar t also demanded its main suppliers to execute the same command; having by 2008 incorporated all of their 15,000 suppliers into it. In the supermarket arena, Metro AG, in Germany, has also introduced the pallet/case RFID tagging into its daily business. RF ID also provides protection against imitation products. Using non reprogrammable RFID tags, information about where that product was harvested or processed can be stored and remain intact until the end of the supply chain, which could be important for products with denomination of origin, certifications or an established branding that the consumer perceives as an added value. Trace back Opara (2003) explains that the traceability in an agricultural and food supply chain s contributes to the demonstration of the transparency of the chain, adding value to the overall

PAGE 61

61 quality management system by providing the communication linkage for identifying, verifying and isolating sources of noncompliance to agreed standards and customer expectations. In the last decade the produce industry has seen an increased need to perform quick and effective trace back processes. T race back directives in this industry were put into action due to the U. S. Bioterrorism Act of 2002, which requires that every handler of food products establishes and maintains records to document the movement of its products both one step forward and one step back through the supply chain (Produce Marketing Association, 2008). In recent years, some of the food safety recalls experienced have tested thi s system and showed the imminent need for better trace back mechanisms along the supply chain, able to determine the origin of the product in a small period of time and in an even more specific way. Additionally, in October of 2008, the Country of Origin Labeling (COOL) regulations were set into place S uch directives require that information about the country where the product was harvested accurately reaches the retailer and the consumer as a mean of providing more information to the customer during the purchase process. When combined with proper software applications, RFID offers a quick and effective way of storing, maintaining and retrieving the information needed during trace back operations across all the stages of the supply chain; being able to sat isfy all the regulations and needs attached to the produce business. Furthermore, it facilitates and improves the recall process by providing tools to all the parties involved to aim at only affected products for disposal, leaving acceptable and safe ones in their inventories. And finally, as in the previous analysis, the use of nonreprogrammable tags (write once, read many times) provides the opportunity of securing the information of the origin of the product, which protects the customer by preventing i ntentional or unintentional mislabeling.

PAGE 62

62 Process system traceability According to Bollen and Riden (2006), process system traceability involves the ability to verify any processing and postharvest activities and treatments, such as washing, drenches and dips, storage conditions, additives or chemicals used and temperature conditions within the supply chain. Keeping track of some of this information will not only facilitate monitoring the compliance to current certifications (GAP, EUREPGAP, HACCP, Organic, etc), but will also provide a tool for tracing back quality issues that might arise during the supply chain. Although suggested, very little research has been done with regards to the application of RFID in process traceability, especially on the monitor ing of the ambient conditions surrounding food products along the supply chain B eing perishable, produces final quality and shelf life is determined mostly due to their interaction with these conditions (relative humidity, temperature, etc) and therefore there is a valid economic rationale behind monitoring them. As mentioned before, p roper temperature management of produce shipments also reduces the risk associated with pathogenic bacteria survival and growth, which is highly important in order to avoid any food safety concern and possible economically devastating recall scenarios. Semi passive, semi active and active tags possess logic compounds within their structure. These logic compounds can be sensors, and with them RFID systems are enabled to reco rd ambient measurements in conjunction with identification product codes and origin information. RFID technology currently offers tags with temperature, moisture and vibration sensors (Montalbano Technologies, 2008) and the incorporation of gas sensors is under development (Abada et al., 2007; Vergara et al., 2007). Opportunities of RFID usage in the produce industry There are many opportunities for the produce industry when adding RFID to their operations. First of all, a s will be demonstrated by the re search results present ed in this

PAGE 63

63 dissertation with regards to cold chain monitoring, RFID sensors p osses ample advantages compared to traditional methods of temperature tracking which will translate into cost reductions and increased profits throughout th e supply chain. Their fast and simple instrumentation and data recovery allow the use of less manpower for these duties, reducing labor costs and delays in packing lines and warehouse management procedures In addition, because RFID temperature sensors sup ply vital information about the products temperature abuse, they will become a powerful decision-making tool for suppliers and retailers, assisting in the application of the concept of First Expires First Out (FEFO) in supply chain and logistic operatio ns (Emond and Nicometo, 2006). This breakthrough idea will be put into practice because of the incorporation of shelf -life modeling into RFID data processing software. Distribution centers all around the world will be able to automatically retrieve the t hermal history of each pallet entering their facilities, while at the same time obtaining an accurate estimate of the remaining shelf -life of the products in those pallets based on the temperatures they experienced until that point in the supply chain (Emo nd and Nicometo, 2006). As a consequence, informed decisions will be made with regards to the products future: The produce with longer shelf life will be sent to distant locations or remain in storage while the one with shorter shelf life will either be d estined to near by stores or discarded, according to the amount of shelf -life it still possess. Furthermore, RFID temperature tags also provide something impossible to attain with conventional temperature tracking methods: the reading and reprogramming of the sensors when they are already into place This is a point of high importance when considering the need for real time remote monitoring in food supply chains, which is becoming possible with the combination of RFID and other wireless te chnologies Onc e this is fully achieved and combined

PAGE 64

64 with shelf -life prediction software, as described before, companies will be able to make logistic decisions during the products transport process; redirecting product to their customers based on the amount of temperat ure abuse suffered by the load, its remaining shelf -life and the distance between the vehicle containing the load (truck, ship, plane, train) and their customers. The increased visibility obtained with the RFID systems during cold chain operations will a id in process control by detecting failures in the links of the chain, and will provide a rationale behind process improvement actions. All agents involved ( retailers, producers, distributors) will then be able to set up better standards and improve their temperature management. The resulting enhanced cold chain will reduce the product waste in the shape of rejections at the distribution center (DC) level and losses at the DC and the retailer store level. From a sustainability point of view, better cold ch ains imply a reduction in waste: from packaging (from the tossed product), to fuel (used to operate the packing line during their packaging, to force air cool it and to keep the products subjected to temperature abuse somewhat cold during its failed cold c hain), to the product itself and all the factors involved in its generation (pesticides, land, water, etc). All of these reductions plus the ones in manpower will generate a decrease in costs for the different areas of the produce supply chain. In additio n, profit will be generated along this by adding value to the product offered with improved quality, safety and consistency. Better produce will then increase customer satisfaction and, in consequence, brand loyalty to both retailers and producers (if appl icable). Challenges for the application of RFID in the produce industry The implementation of RFID technologies in the produce industry is in its infancy and, therefore, still presents some challenges. From a technical perspective, the most important one for cold chain applications is the fact that the current RFID systems designed for temperature

PAGE 65

65 tracking work either in the frequencies that can not work near products containing high amounts of water (such as produce) or in frequencies that have short reading ranges, which makes them unsuitable for portal applications in warehouses and distribution centers. Another technical challenge is the use of multiple frequencies around the globe. Even though organizations such as the International Association for S tandardization (ISO) and EPC Global are working towards creating standards promoting global parity for specific RFID applications; the different regulations with respect to frequency ranges allocated for RFID commercial purposes, vary according to the country and complicate global RFID ventures. For example, UHF tags sent in a shipment from the U.S. (with a UHF frequency allocation between 902 928 MHz) will be able to be read in Europe (frequency allocation of 865 868 MHz) if used EPC Global Gen 2 standards (which call for readers to be able to read along the entire UHF spectrum) but will not be able to be reprogrammed and sent back with a shipment. Cost is also one of the restrictive factors for RFID temperature tracking implementation. The use of semi -pas sive, semi active and active tags able to perform data collection and the infrastructure necessary to perform the readings throughout the cold chain considerably surpass the cost of conventional temperature monitoring methods in the market. Yet the implem entation of protocols able to reduce the level of instrumentation, along with the imminent reduction in the manufacturing costs of data -logging RFID tags, could increase the competitivene ss of these systems in the near future. Passive RFID systems are mostly in use in the produce industry for product identification and trace back due to the U.S. Depar tment of Defense (DoD) and Wal M art mandates. These systems are not as expensive as the ones employed for temperature tracking and are more likely to be purchased by companies. However, the allo cation of funding by the DoD to projects

PAGE 66

66 involving temperature tracking such as those described in this d issertation indicate a tendency leaning towards the combination of ID applications and environmental monitoring a long the supply chain, which could possibly lead to new mandates for its suppliers (including the produce industry) involving the use of this type of technology in the near future.

PAGE 67

67 Table 2 1. Thermocouple properties according to their ISA (Instrument S ociety of America) classification. ISA Class Material (+ & -) Temperature range C (F) Sensitivity at 25C (77F) V/C (V/F) Error A pp lic ati on ** E Chromel & Constantan (Ni Cr & Cu Ni) 270~1000 (450~1800) 60.9 (38.3) LT:1.67C(3F) HT:0.5% I, O J Iron & Constantan (Fe & Cu Ni) 210~1200 (350~2200) 51.7 (28.7) LT:2.2~1.1C(4~2F) HT:0.375~0.75% I, O, R, V K Chromel & Alumel (Ni Cr & Ni -Al) 270~1350 (450~2500) 40.6 (22.6) LT:2.2~1.1C(4~2F) HT:0.375~0.75% I, O T Copper & Constantan (Cu & Cu Ni) 270~400 (450~750) 40.6 (22.6) LT:1~2% HT:1.5% or 0.42C(0.75F) I, O, R, V R Platinum & 87% Platinum/ 13% Rhodium (Pt & Pt Rh) 50~1750 (60~3200) 6 (3.3) LT:2.8C(5F) HT:0.5% I, O S Platinum & 90% Platinum/ 10% Rhodium (Pt & Pt -Rh) 50~1750 (60~3200) 6 (3.3) LT:2.8C(5F) HT:0.5% I, O B 70% Platinum/ 30% Rhodium & 94% Platinum/ 6% Rhodium (Pt Rh & Pt Rh) 50~1750 (60~3200) 6 (3.3) LT:2.8C(5F) HT:0.5% I, O *: LT = Low temperature range, HT = High temperature range **: I = Inert medi a, O = Oxidizing media, R = Reducing media, V = Vacuum Constantan, Alumel, and Chromel are trade names of their respective owners. Source: Efunda.com 2010.

PAGE 68

68 Table 2 2. Temperature ranges for common metals used in RTDs. Material Temperatur e r ange Note Platinum (Pt) 260~1000 C ( 440~1800 F) < 550 C (1022 F) in most applications Copper (Cu) 200~260 C ( 330~500 F) Nickel (Ni) 200~430 C (330~800 F) Linearity is not good Balco (70% Ni 30% Fe) 100~230 C ( 150~450 F) Linearity is not good; cheap to fabricate; high resistance Tungsten (W) 100~1200 C ( 150~2200 F) Source: Efunda.com 2010. Table 2 3 List of RF properties of some materials. Material LF HF UHF Microwave Clothing RF lucent RF lucent RF lucent RF lucen t Dry wood RF lucent RF lucent RF lucent RF absorbent Graphite RF lucent RF lucent RF -opaque RF -opaque Liquids (some types) RF lucent RF lucent RF absorbent RF absorbent Metals RF lucent RF lucent RF -opaque RF -opaque Motor oil RF lucent RF l ucent RF lucent RF lucent Paper products RF lucent RF lucent RF lucent RF lucent Plastics (some types) RF lucent RF lucent RF lucent RF lucent Shampoo RF lucent RF lucent RF absorbent RF absorbent Water RF lucent RF lucent RF absorbent RF abso rbent Wet wood RF lucent RF lucent RF absorbent RF absorbent Source: Lahiri, 2006. Table 1 2, pp. 6.

PAGE 69

69 Table 2 4 EPCTM tag classification Class Power Range Memory Communication Peripherals Cost 0 None < 3 m 1 to 96 bits, Read Only Backscatter None Low 1 None < 3 m 1 to 96 bits, Read/Write Once Backscatter None Low 2 None < 3 m 1 to 96 bits, Read/Write Backscatter Security Medium 3 Battery a ssisted < 100m < 100 Kilobytes, Read/Write Backscatter Security, Sensors High 4 Battery a ssisted < 300m < 100 Kilobytes Read/Write Active Transmission Security, Sensor High 5 Battery a ssisted, AC/DC connection Unlimited Unlimited, Read/Write Active Transmission Security, sensors, can communicate with other tags Very High Source: Banks et al., 2007. Tabl e 3 1, p p 69.

PAGE 70

70 CHAPTER 3 APPLICATION OF RFID TECHNOLOGIES IN THE TEMPERATURE MAPPING OF THE PINEAPPLE SUPPLY CHA IN Introduction RFID applications within the food supply chain have been mostly proposed in order to satisfy the need for identification and tr ace back mechanisms (Jones, 2006; Wang et al., 2006; Asimakopoulos et al., 2007; Montrucchio et al., 2007; Panos and Freed, 2007). Although suggested, very little research has been done with regards to the application of RFID in process traceability, espec ially on the monitoring of the ambient conditions surrounding food products along the supply chain (Mermelstein, 2002; Karkkainen, 2003; Gaukler and Seifert, 2007). RFID systems present many advantages for temperature monitoring and tracking. One of them i s the reduced amount of instrumentation labor required. RFID does not entail manual scanning and hundreds of RFID tags can be downloaded into a computer at one time without ever requiring line of sight between the tags and the reader (Karkkainen, 2003). T his saves considerable time and man-hours and decreases the number of errors that can occur with manual systems (Thompson et al., 2005). RFID tags can be read from significant distances (especially the active variety) and at fast speed, which favors its application in supply chain operations without generating delays (Raza et al., 1999). According to the kind of tag, they can also record more than 13,000 data points at different time intervals (up to 4.5 0 years of continuous logging) (Identec Solutions Inc. 2007); ensuring the performance of the sensor in longduration supply chains. Ruiz Garcia et al. (2008) explain that these wireless sensors can be embedded within the load which brings their readings closer to the true in situ properties of perishable p roducts. However, ambient temperature recordings throughout the load will differ from the ones measured in the core of the product and within themselves according to the ambient conditions

PAGE 71

71 of the surrounding area. Results obtained by this research show evi dence to support that, compared to temperature sensors with probe, the limitation of probeless RFID systems lies on their failure to provide accurate temperature readings in some of the critical points of the load, where temperature abuse is likely to occu r. Still, their use could be appropriate if the corresponding relationship between their output and the temperature of the product located in the critical point is established. Therefore, the different scenarios for the utilization of RFID tags with and wi thout probes in the temperature mapping of supply chains have to be determined. Cost being one of the current restrictive factors for RFID temperature tracking implementation in real life supply chains, it has been proposed that tagging should be done at the transport unit level (Karkkainen, 2003). Nonetheless, the cost of placing these tags in transport cartons might surpass the level of investment a food company is willing to put into temperature tracking; therefore, it is important to minimize the amoun t of RFID equipment used and the resulting economic investment per load. The objectives of this work were: 1) To study the use of RFID in the temperature monitoring of a load of crownless pineapples packed in two kinds of primary packages by comparing the performance of RFID temperature tags versus conventional temperature tracking methods, as well as RFID temperature tags with probe versus RFID temperature tags without probes and their utilization along the supply chain; and 2) To determine the level of i nstrumentation necessary for proper temperature monitoring in this supply chain. Materials and Methods Experimental Design The initial part of the experiment took place at a pineapple packing house in the Pacific coast of Costa Rica. Eighteen primary packa ges containing six crownless pineapples each were instrumented with temperature sensors befor e the palletizing process. Nine of these primary

PAGE 72

72 packages were corrugated boxes (trays; 0.5 8 m x 0.39 m x 0.14 m) and nine were reusable plastic containers (RPC; 0 .59 m x 0.39 m x 0.20 m). In each package, one fruit was probed with a probe attached to a HOBO temperature sensor (HOBO Series H8; TMC6 HA and TMC6 HD probes, Onset Computer Corporation, MA) so as to obtain their mass average temperature at a sampling ra te of 5 .00 minutes. The probes were installed at 0.04 m of depth at the center of the fruit. The packages were then placed in three different areas of their corresponding pallet: The upper corner, the core at the central layer (seventh from the bottom up) and the lower corner (Figure 3 1). The position of the probed pineapples depended on the location of the primary package in the pallet. For the top and the bottom packages, the pineapple corresponding to the corner was probed; while for the box placed at t he core, it was the fifth pineapple from left to right. Four kinds of RFID Temperature Sensors were also placed in each one of the six pallets studied (Figure 3 2). These RFID tags were: ThermAssure RF (Evidencia, TN), ALB 2484 (Alien Technology, CA), Ther maProbe RF (Evidencia, TN), and ILR i Q 32T (Identec Solutions, TX). Both ThermAssure RF and ThermaProbe RF tags worked at a frequency of 13.56 MHz, while the ALB 2484 worked at 2.4 GHz and the ILR i -Q 32T at 915 MHz. The three RFID sensors initially mentioned are semi passive tags (also known as battery assisted passive tags), which are RFID tags that contain a battery that powers their logic parts but are unable to emit a signal unless energized by a RFID reader. On the other hand, ILR i Q 32T, is an acti ve tag. Active tags are able to make use of their battery not only to power their logic parts but also to actively broadcast a signal. One RFID sensor of each kind was placed at the exterior of the central layer of the pallet; but only the first three men tioned record ed ambient temperatures (Figure 3 -3). The ThermaProbe RF (Evidencia, TN) sensor was recording the mass average temperature of the probed pineapple

PAGE 73

73 in the package placed in the core of the pallet. Based on results obtained in preliminary trials conducted for temperature mapping, the sampling rate for all RFID temperature tags was set in 5 .00 minutes. After forced air cooling and nine hours in refrigerated storage, the two sets of pallets were pla ced inside one refrigerated 12.2 0 m container in three specific locations: Front, Middle and Back. The container had one row of pineapples packed in corrugated boxes and one row with a mixed load of fruit packed in RPCs (the three pallets being studied) and corrugated boxes. The set point of the conta iner was 7.50 C. Once filled, the container was sent to the Port of Limon in the Atlantic coast of Costa Rica through land transportation. Two days after its arrival it was shipped to Florida (US) through a carrier ship. The Temperature/Time readings from both kinds of sensors were obtained until the shipment reached the companys distribution center in the coasts of Florida three days later. Statistical Analysis A Generalized Linear Model (GLM) with a Gamma family was used in order to determine the effect s of the time, the sensor and packaging employed on the temperatures recordings obtained (Freund and Wilson, 2003). It was also used to determine the equality of means of the temperatures obtained with the different kinds of sensors. In addition, a Studen ts t Test was performed in order to compare the means of the temperature deltas between the sensors. Significance was accepte Statistical analyses were computed using R 2.7.2 (The R Foundation for Statistical Computing., Wiedner Hauptstrabe 8 10/1071 1040 Vienna, Austria).

PAGE 74

74 Results and Discussion RFID Sensor Performance versus Conventional Methods The major difference noticed within both kinds of sensors was the time required for the instrumentation of the load (Table 3 1). The RFID systems allowed a faster setup because of the little time taken while programming the sensor and their ease of placeme nt in the load. The main factors in the speed of programming of the RFID systems were the amount of tags read at once by the readers and the user -friendliness of the software employed. The fastest was the ILR i -Q 32T system (19.09 s 3.15; mean SD) whi ch, because of its active nature, allowed the distant reading of all the tags at the same time, facilitating a remote activation process. This tag was followed by the ThermAssure RF and ThermaProbe RF sensors (21.50 s 2.43; mean SD), semi -passive RFID tags using the same software and hardware, which required a closer interaction with the readers and could only be read one by one. The user friendliness of the software employed was decisive in the programming time of the semi -passive ALB 2484 (47.68 s 6 .36; mean SD) tags, which, unlike the others, had management software with poor interface and took the longest to be activated even though all the tags were able to be read at once. On the other hand, HOBO sensors required a lengthier period of time duri ng programming (53.71 s 3.92; mean SD) and had to be wired to the computer which made impossible any sensor re -programming or reading once the pallets were instrumented. As far as the ease of setup in the load, ThermaProbe RF tags were as difficult to handle as HOBO sensors because their probe also needed to be placed in a specific location inside the pallet. All other RFID sensors were quite easy to place; however, amongst all, some of them were more convenient than others. For example, due to their s mall size and thickness (0.05 m x 0.05 m x 0.04 m) and light weight (6 g), handling the ThermAssure RF and ThermaProbe RF

PAGE 75

75 sensors was better than dealing with the bigger and thicker ALB 2484; problem quite evident at the time of taping them to the exterior of the pallet of corrugated boxes. Even less convenient were the big ILR i Q 32T sensors, which could not be taped and had to be placed inside the primary packages at all times. During the setup of RPCs, the sensors that were easier to place on the packag e (because they fit in the slot included in the package design) were the ThermaProbe RF and ThermAssure RF tags; while the other two were secured using the same method as in the previous package. When comparing the results obtained with the ThermaProbe RF RFID tag and the HOBO sensor placed in the pineapple of the core of the pallet the statistical analysis (Table 3 2) showed significant differences between the sensors readings (P = 5.66e 03). The ThermaProbe RF RFID tag consistently produced higher values in all pallets, being more evident in the case of the ones using corrugated boxes as packaging (Table 3 3). This phenomenon could be explained by the difference in accuracies of the sensors used ( 0.50 C HOBO, 0.1 0 C ThermaProbe RF). As far as RFID sens or performance, only two of the ambient ones (ThermAssure RF and ILR i Q 32T) were able to provide readings point by point. ALB 2484 RFID tags only supplied a final graphic with the temperature profile, but did not allowed access to the data for further co mparison. The three temperature profiles had the same shape in each pallet (Figure 3 4). However, for both kinds of packaging, comparison of the curves obtained with ThermAssure RF and ILR i Q 32T sensors showed that the former had higher values most of th e time (Figures 3 4, 3 5 and 3 6). The statistical analysis (Table 3 4) later on showed that there were significant differences between the readings of both sensors (P < 1.55e 07). The maximum delta obtained between these in each pallet reached up to 2.7 0 C in the pallets with corrugated boxes. This temperature difference was most likely due to the fact that the bulky ILR i Q 32T sensor was placed on the

PAGE 76

76 vent inside the pineapple box and not on the surface of the box like the ThermAssure RF and therefore, w as subjected to different airflow exposure. In the pallet with RPCs, even though still present, the temperature differentials between the ThermAssure RF and the ILR i Q 32T tags were reduced considerably with respect to the values obtained with corrugated boxes due to the increased ventilation characteristic of RPCs. Measured against the conventional methods for temperature tracking, such as handheld thermometers and datalogging systems like the HOBO and iButton sensors, RFID sensors provide ample advantag es. Their fast and simple instrumentation and data recovery allow the use of less manpower for these duties, reducing labor costs and delays in packing lines and warehouse management procedures. In addition, because RFID temperature sensors supply vital in formation about the products temperature abuse, they will become a powerful decisionmaking tool for suppliers and retailers, assisting in the application of the concept of First Expires First Out (FEFO) in supply chain and logistic operations (Emond an d Nicometto, 2006). Furthermore, RFID temperature tags also provide something impossible to attain with conventional methods: The reading and reprogramming of the sensors when they are already into place. This last point is of high importance when conside ring the need for real time remote monitoring in food supply chains, which in the near future will become possible with the combination of RFID and other wireless technologies. RFID Temperature Tags with Probe versus RFID Temperature Tags without Probe an d Their Utilization along the Supply Chain Ambient readings with RFID sensors without probes present a wide difference with those obtained by the ThermaProbe RF sensor in the pineapple of the core of the pallets. The biggest deltas happened mostly during t he forced air cooling period, reaching absolute values of up to 16.60 C (6 .00 2.75; mean, SD) for corrugated boxes and 16.50 C (4.96 0.63; mean, SD) for

PAGE 77

77 RPCs. During refrigerated storage and transportation, the maximum absolute values of the deltas we re 6.8 0 C (0.32 0.30 ; mean, SD) and 5.2 0 C (0.39 0.16; mean, SD), respectively. Even though, statistically there was no significance between the absolute values of the temperature difference between the ThermAssure RF and the ThermaProbe RF and the IL R i Q 32T and the ThermaProbe RF in both packages (p -value = 4.41e 01); numerically the first delta (1.03 1.73; mean, SD) presented higher values than the second one (1.01 1.56; mean, SD); which can be explained by the position of the sensors in the loa d. When analyzing the temperatures profiles of these sensors after precooling, we obtain two different situations according to the type of package. For corrugated boxes, both core sensors for the frontal and middle pallets present very similar curves to t he ones of the ambient sensors, where the biggest dissimilarity happens during and after heating episodes and a lag effect takes place. However, in the corresponding back pallet, the core temperature remains hotter than the environment most of the time. In RPCs the profiles from the fronta l and middle pallets behave as in the previous case (Fig. 3 6); but in the back pallet, even though the lag effect is also present, core temperatures always stay lower than ambient ones. Ambient temperature profiles in the two back pallets did correlate to the profiles obtained in the pineapple monitored in the top layer of the pallet, where the highest temperatures of the fruit in the container were recorded. Pulp temperature is important during rapid change of tempera ture such as during precooling. If the temperature at the core of the pallet has not been reduced during precooling, this region becomes one of the critical points of the pallet, an elevated risk area for high temperature abuse throughout the supply chain. Therefore, the use of RFID sensors with probe is quite relevant because they allow pulp temperature monitoring at this point, which could be exposed to insufficient cooling that will not be detected by the sensors without probes (Fig. 37).

PAGE 78

78 Also worthy of monitoring are certain regions where the pineapple would be at risk of low temperature abuse and its manifestation, chilling injury, during transportation and refrigerated storage. Results of the temperature mapping performed in this trial suggest that the fruit in the lowest layers of the pallets will be more at risk to such conditions (Fig. 3 8). However, no specific correlation could be deducted between the readings of the pulp temperatures (measured with HOBO sensors) in these areas and the ambie nt temperatures recorded. Consequently, in order to determine the temperature in these critical points, employing RFID temperature tags with probe is suggested. RFID temperature tags with probe are then necessary during precooling operations and could als o be required during transportation and refrigerated storage if there is an interest in monitoring and preventing chilling injury occurrences. As mentioned previously, during transportation and refrigerated storage, ambient RFID tags will prove useful in s howing the behavior of the fruit subjected to the highest temperatures in the load. Therefore, an efficient temperature tracking system for this supply chain could make use of the potential of combining these two RFID systems for a comprehensive temperatur e abuse analysis. The development of a RFID temperature tag with a probe, able to record both ambient and probed temperatures simultaneously is proposed as a way of ensuring proper temperature tracking along precooling, storage and transportation operations. Its use will reduce the amounts of instrumentation and investment required if two RFID sensors with probes, one for the core (or for the bottom, if monitored) and one for a fruit located in the upper layer of the pallet were employed. Level of Instrum entation Proper monitoring is achieved with one sensor per pallet, if necessary, according to the location of the pallet during forced air cooling and inside the container. Probeless tags are

PAGE 79

79 recommended for monitoring High Temperature Abuse (HTA) in the t op of the pallets placed in the back of the container. Tags with probe are suggested for detecting HTA in the center of the pallets and Low Temperature Abuse (LTA) in the bottom of the pallets of RPCs in the front of the container. Conclusions The applicat ion of RFID temperature monitoring was studied in a commercial shipment of crownless pineapples packed in corrugated boxes and reusable plastic containers from Costa Rica to the U S A comparison between the performance of a conventional method (HOBO sens ors) and RFID temperature tags showed that both methods are analogous with regards to accuracy, but RFID systems have a superior performance because they allow quick instrumentation and data recovery, and the possibility of accessing the sensor program and data at any point of the supply chain without line of sight. Although observed differences between the pulp temperatures taken by the RFID sensor with probe and the HOBO sensors, and between the external RFID tags can be disregarded due to factors such a s the accuracy and the placement of the sensors. In addition, the use of RFID tags with probe was justified by its role in determining the efficiency of the precooling operations and in tracking areas prone to chilling injury during transportation and refr igerated storage; while the RFID tags without probe proved most useful during high temperature abuse monitoring along transportation and refrigerated storage. The creation of a RFID sensor with a probe, able to record both ambient and probed temperatures s imultaneously is suggested as a way of gathering the benefits of both systems and ensuring proper temperature tracking along precooling, storage and transportation operations. Future work in this area is needed in order to determine the readability of RFI D temperature sensors that use different frequencies along a variety of distances, pallet and primary

PAGE 80

80 package locations and kinds of food products. Also necessary is the development of more convenient sensors and portable readers in regards to their size a nd shape. Finally, recommendations we re given with respect to the instrumentation level required for a RFID temperature tracking system for this supply chain.

PAGE 81

81 Figure 3 1. Placement of the HOBO temperature sensors in the experimental pallets. Figu re 3 2. RFID tags used during the trial. From left to right: ThermaProbe RF, ALB 2484, ThermAssure RF and ILR i Q 32T.

PAGE 82

82 Figure 3 3. Placement of the RFID temperature sensors in the experimental pallets.

PAGE 83

83 Table 3 1. Comparison of the features of the RFID temperature tracking systems analyzed and a traditional temperature monitoring system (HOBO Sensor). Features ILR i Q 32T ThermAssure RF ThermaProbe RF ALB 2484 HOBO Sensor Programming time (Mean SD) 19.09 s 3.15 21.5 0 s 2.43 21.5 0 s 2.43 47. 68 s 6.36 53.71 s 3.92 Detects all sensors at once Yes No No Yes No User friendly software Yes Yes Yes No Yes Allows distant interaction between the reader or computer and the sensor Yes No No Yes No Requires placing a probe in the load No No Yes N o Yes Easy to handle and affix to the packaging No Yes Yes No No Software facilitates data management and analysis Yes Yes Yes No Yes

PAGE 84

84 Figure 3 4. Ambient temperature recordings obtained by the ALB 2484RFID tag placed in a pallet with corrugated boxes Figure 3 5. Ambient temperature recordings obtained by two RFID tags placed in a pallet with corrugated boxes: ThermAssure RF and ILR i Q 32T.

PAGE 85

85 Table 3 2. Results of the Generalized Linear Model for the Effects of the Sensor, Time Step and Packaging w hen comparing the temperatures (C) obtained with the HOBO sensor and the ThermaProbe RF RFID tag placed at the core of pallets with corrugated boxes or reusable plastic containers in the sea container. Estimate Std. Error t Value P (Intercept) 5.74 e 02 9.03 e 04 63.50 < 2e 16 Sensor 1.40e 03 5.05e 04 2.76 5 66e 03 Time Step 3.01e 05 8.57e 07 35.11 < 2e 16 Packaging 1.29e 02 2.88e 04 44.89 < 2e 16 Sensor: Time Step 2.23e 06 5.36e 07 4.1 5 3.35e 05 Table 3 3. Comparison between the temperature s (C) obtained with the HOBO sensor and the ThermaProbe RF RFID tag placed at the core of pallets with corrugated boxes according to their position in the sea container. Container Position Front Middle Back HOBO Sensor ThermaProbe RF HOBO Sensor T hermaProbe RF HOBO Sensor ThermaProbe RF Mean 10.39 10.89 10.91 11.33 12.51 12.71 Standard Deviation 4.11 4.10 3.97 3.95 3.57 3.55

PAGE 86

86 Table 3 4. Results of the Generalized Linear Model for the Effects of the Sensor, Time Step and Packaging when comparing the temperatures (C) obtained with the ThermAssure RF and the ILR i Q 32T RFID tags recording ambient temperatures on pallets with corrugated boxes or reusable plastic containers in the sea container. Estimate Std. Error t Value P (Inter cept) 6.07 e 02 9.20 e 04 65.9 6 < 2e 16 Sensor 2.75e 03 5.23e 04 5.2 5 1.55e 07 Time Step 2.07e 05 8.53e 07 24.20 < 2e 16 Packaging 7. 75e 03 2.93e 04 26.41 < 2e 16 Sensor: Time Step 2.75e 06 5.48e 07 5.01 5.47e 07 Figure 3 6. Temperature profiles obtained by RFID tags with probe and without probe in the central layer of a pallet with RPCs.

PAGE 87

87 Figure 3 7. Temperature profiles obtained by RFID Tags with probe (ThermaProbe RF) and without probe (ThermAssure RF, ILR i Q 32T) in the central layer of a p allet with corrugated boxes during forced air cooling. Figure 3 8. Temperature profiles obtained by HOBO sensors in three different areas of a pallet with corrugated boxes.

PAGE 88

88 CHAPTER 4 EVALUATION OF SENSOR READABILITY AND THER MAL RELEVANCE FOR RF ID TEM PERATURE TRACKING Introduction Technological advances have enabled precise traceability and visibility mechanisms within temperature controlled supply chains. The existence of these technologies is driving a requirement for precise traceability in perishab le goods chains, such as the food and pharmaceutical supply chains (Bollen, 2005; Ames, 2006; Metzger et al., 2007; Panos and Freed, 2007; Zhenhua et al., 2007; Jedermann et al., 2008). RFID technology has been suggested as a solution to address this need since it can store both pedigree and temperature/time information and transmit it even as the product is in transit (Forcinio, 2004; Smith, 2005; Wyld, 2006; Leake, 2007). As discussed in the previous chapter, as a temperature tracking method, RFID has sh own superior performance when compared to traditional monitoring devices since it can surpass the simplicity problems that these encounter during real -life fast -paced supply chains, while maintaining their levels of accuracy. In spite of the promises it holds, the implementation of RFID in this particular area is still in its infancy and, therefore, still presents some challenges. A major hurdle for RFIDs application in food supply chains is that, based upon the frequency used, certain environmental factor s, such as water and the presence of metal, affect the RF signal (Dobkin and Weigand, 2005; Redemske and Fletcher, 2005; Gaukler and Seifert, 2007; Hartvanyi and Marek, 2007; Sivakumar and Deavours, 2008). Current RFID systems for temperature tracking pres ent in the market operate either in the frequencies with reduced readability near loads of perishable products with high water content such as produce (915 MHz, 868 MHz), or in frequencies that have short reading ranges (13.56 MHz), which makes them unsuit able for portal applications in warehouses and distribution centers.

PAGE 89

89 Cost is also an important barrier for industry implementations (Edwards, 2007). Battery assisted passive and active tags used in temperature monitoring are considerably more expensive th an passive solutions used for asset tracking and represent a higher level of investment for companies. Therefore, based on the discussions from the previous chapter, any feasible real life application has to utilize efficient instrumentation within the dif ferent levels of aggregation of the load, allowing the minimization of the number of tags used without compromising the accuracy of the system. Within the different frequencies, there are two kinds of RFID temperature tags available in the market: Tags with a probe and probeless tags. Tags with a probe are more suitable for temperature monitoring inside the pallet and the product. This is mostly because these tags can be read on the outside of the pallet; while reading a probeless tag implies surpassing the possible interactions existent inside the pallet between the packaging/product and the RF waves, which could turn into a very difficult matter. In addition, using a probeless tag for monitoring inside the product complicates the instrumentation process ( cut, open and insert; instead of only inserting a probe); and could also pose a safety and quality threat if there is considerable discharge of biological material into the load. As discussed in the previous chapter, in produce industry applications the u se of RFID tags with a probe played an important role in monitoring the efficiency of precooling operations; while RFID tags without a probe proved useful during transportation and refrigerated storage. This study also demonstrated that the limitation of p robeless RFID systems placed on the outer surface of the pallet lies on their failure to consistently provide temperature readings close to many of those in the critical points of the load, which are the areas of the load where temperature abuse is most li kely to occur inside the product. An example of this disparity is given when these

PAGE 90

90 probeless tags register rapid temperature changes, which, if taken as reference, may mislead on the actual product temperature in these locations. Some perishable products ( such as certain pharmaceuticals and chilling sensitive produce) require the monitoring of the critical points of their loads because they can not afford being subjected to any temperature abuse, since reaching threshold temperatures would seriously impair their safety and integrity; consequently, RFID systems using tags with a probe would be the most adequate solution for their temperature tracking needs. Nonetheless, measuring the temperature in the critical points of the load offers disadvantages when trying to determine the total amount of temperature abuse suffered by the entire product placed in a pallet. Since these points are the areas more prone to it, the values they provide are the extremes and will not reflect the average thermal history of the pa llet. For many perishables operations, where certain temperature excursions are allowed without posing a major threat to the product, it is the level of thermal abuse in this other temperature profile what matters at the receiving end. Placing a minimum amount of temperature data loggers in relevant locations able to describe the majority of the temperatures present in the entire pallet seems like a reasonable option in order to obtain a broader perspective of the thermal reality of the load throughout t he supply chain, while taking into account real -life cost constraints. If the acquisition of internal product temperature is not required, probeless RFID temperature tags could be used for this kind of monitoring as long as, once placed in the thermally r elevant areas during packing and palletizing, they surpass any negative interaction with the loads content and its environment and get to be read either in transit or through portal applications during the end stages of the distribution process.

PAGE 91

91 The obje ctives of the research presented in this chapter were: To find a relevant location able to provide a temperature reading representing eighty five percent of the temperatures present in a pallet of bottles of water mimicking high-water content produce subje cted to heating and cooling episodes commonly encountered during loading/unloading operations in produce supply chains. And second, to assess the readability and read range of a commercially available RFID system when a battery assisted passive tag is plac ed in this thermally relevant position so as to determine whether it would be read at the required distance for the supply chain. Materials and Methods Two studies were p erformed. A relevance study d esigned to find the thermally relevant positions of the load was initially done. Then, a readability study able to detect the level of readability of battery assisted passive tags and the reading ranges of the RFID system obtained when placing these tags in the relevant positions found in the previous study w as completed. Relevance Study The thermal relevance of different locations inside a pallet of bottles of water was studied for conditions similar to the ones present during produce loading/unloading operations. Half a pallet with 2160 bottles of water packed in 30 reusable plastic containers (RPC) was instrumented with thermocouples previously calibrated in an ice bath (Figure 4 1) Spherical bottles (Aquapod, Zephyrhills, 0.3 3 L) were used, mimicking circular shaped high -water content produce such as ora nges. The pallet consisted of six layers with five reusable plastic containers (SmartCrate GP 6419; 0.60 m x 0.40 m x 0.20 m, PP; IPL Inc., Saint Damien, Quebec, Canada) each; and each RPC contained a total of three layers of 24 bottles in a four by six c onfiguration. The 126 instrumented bottles provided a three -dimensional perspective for the examination of the heat transfer process along the reusable plastic containers (RPC) and the half pallet analyzed (Figure 4 1). The temperature of the pallet was in itially stabilized by keeping it

PAGE 92

92 overnight in a cold room at 4 .00 C. The next morning, the pallet was moved from the cold storage to the outside of the building, where it was exposed to ambient temperatures for a specific amount of time (either 4 .00 2 .00 1 .00 or 0.50 hours). It was stored once again in the cold room and recordings were taken for the same amount of time it was exposed to ambient temperatures (4 .00 2 .00 1 .00 or 0.50 hours, respectively). Quantitative analysis was performed using quintile s in Microsoft Excel (Microsoft Corporation, Redmont, WA). Readability Study The readability of Class 3 battery assisted passive tags (B AP) was tested in two sides of the pallet using the Intelleflex DK900 (Intelleflex Coorporation, Santa Clara, CA) RFID system operating at 915 MHz. The pallet was located in an environment mimicking the unloading area of a distribution center, in open air surroundings but with a sea container and metallic doors nearby. The placement of the tags was determined by the results of the relevance study. Tags had to be placed in one of the 1.22 m sides (Configuration I) and in both 1.02 m sides (Configuration ), but since these had a similar layout only one of them was considered for this testing. Each side was studied independently. The antennas were positioned in the middle point of the front of the side being read. Readings were obtained using four different antenna distances (0.50 m, 1.00 m, 2.50 m and 5.00 m) with respect to the pallet. Tag placement The placement of the tag inside the load could be a determinant factor in the readability of the system and its resulting reading range. In this trial, the dep th of tag placement varied with the side configuration. For the area studied in configuration it was 0.12 m, equivalent to the width of two bottles. For configuration -I it was set to 0.14 m, the height of one bottle. Figure 4 2

PAGE 93

93 illustrates the end poin t of the internal layers in the configurations studied. The RFID tags were set along the red lines. Equipment configuration In order to maximize the read range of the Class 3 tags inside the pallet, reader power was set at its maximum and air interface da ta rates were set at their minimum. Since our research was targeted towards portal applications, the readers inventory configuration was set in portal. Table 4 1 summarizes the readers configuration. Once set up, the system was run in three occasions; in each one, about 100 attempts of communicating with each one of the tags were made. The level of readability of the tags was determined by averaging the percentage of times a tag was read with respect to the attempts made of contacting it. Results and Discussion Relevance Study Temperature distribution The temperature differentials were quite significant at the pallet level and sometimes also at the RPC level. In addition, the magnitude of these differentials increased with the duration of the heating/ cooling episode. For example, for the longest one (4 .00 hours/4.00 hours), the temperature deltas were an average of 20 .00 C ( 5 .00 C, standard deviation) at the pallet level and an average of 10.70C ( 2.60C, standard deviation) at the RPC level. Table s 4 2 and 4 3 summarize the average temperature differentials present at the pallet and RPC level, respectively. As expected, the hottest temperatures were obtained near the top layer and the external walls of the pallet; where exposure to solar radiation and environmental conditions was direct. The coldest temperatures were recorded near the center of the pallet and at the bottom. Similar trends in pallet temperature distribution had been previously reported by Jedermann and Lang

PAGE 94

94 (2007). The core of the pa llet was insulated by the external layers surrounding it; and its temperature change was slow because heat transfer happened mostly due to conduction (poor air circulation generated low convection). The bottom of the pallet remained cold also due to the p oor air circulation happening underneath the wooden pallet and reduced exposure to solar radiation. It is important to mention that this temperature distribution might change according to the shape, weight, size and composition of the product being handle d; as the heat transfer varies according to them. With regards to this last issue, the products water content will be decisive, because it is directly related to the products capacity to experience faster temperature changes. Additionally, the shape and the size of the product will also determine the amount of air flow present inside the pallet and could facilitate the insulation of the inner layers. Moreover, the environmental factors in existence will contribute to these changes in temperature too (i.e .: an increase in wind speed and/or air moisture content will facilitate the heat exchange and raise temperature faster). Nevertheless, the current experiment was set up with bottles full of water and conducted under the North Central Florida summer precis ely because it represents the worst case scenario for a perishable products cold chain, where most of the conditions existent will encourage the quickest temperature gain possible. Hence, it is expected that in most real life cases the temperature differe ntials in loads of produce packed in the same density and similar in weight, shape, and size to the bottles of water will be less than the ones reported in this research when subjected to comparable circumstances. Eighty five percent location The high var iability of the temperature readings indicate that the initial target set for an optimal location is difficult to determine. The thermal variability existent in the pallet increased with the duration of the heating or cooling episode. It was then suggested to use the location that

PAGE 95

95 has a temperature (n) that will be the middle point of an interval that gathers the temperature readings of 85.00 % of the measured locations. The proposed range for the time periods used is [n 6 .00 n+6 .00 ]; a total of 12 .00 C of temperature differential. The likelihood of gathering 85.00 % of the temperature readings in this interval varies according to the duration of the heating/cooling episode, decreasing as the heating/cooling episode increases in time: For 0.50/0.50 hours: 100.00 % For 1 .00 /1 .00 hours: 96 .00 % For 2 .00 /2 .00 hours: 64 .00 % For 4 .00 /4 .00 hours: 42 .00 % Comparisons were made between the temperature profile of this hypothetical point and the ones obtained in the trials. Three locations in the pallet were then consider ed Points of Relevance (PoR) and suitable for sensor placement (where temperature n will be recorded): Point A: Layer 9 (Row 3), X= 14.00 Y=9 .00 Point B: Layer 13 (Row 5), X=8.00 Y=1 .00 Point C: Layer 18 (Row 6), X=14.00 Y=1 .00 Where the layers mo ve along the vertical axis (Z) and are labelled, along rows, in an ascending way from the top to the bottom. Figure 4 3 illustrates the places where these three points are located. It is necessary to indicate that, because of thermal symmetry, the recommen ded thermally relevant points will be located in its equivalent position on the other side of the pallet. Analysis of the likelihood of gathering 85.00 % of the temperature readings of the pallet using the s e points as middle points for the interval was conducted (Table 4 -4). Comparing these estimates, point B would be the best for temperature monitoring because it has the highest values in the first three exposure times, which are more likely to occur along the supply chain than the last one. Point A would f ollow in preference and then point C. The selection of one of the

PAGE 96

96 proposed locations for temperature monitoring will depend not only in this estimate but also on the readability of the RF signal. The produce industry works with two main packages: RPCs (us ed in this trial) and corrugated boxes. Considering the fact that corrugated boxes generally allow less air flow and delay the heat transfer process when compared to RPCs, it is expected that the results obtained in this study would also be applicable to p roducts packed in them as long as the packing density (amount of units per box) is maintained. The intrinsic variability of having a range of 12 .00 C that includes 85.00 % of the temperatures of the pallet could be more or less detrimental on determining t he condition of the produce depending on how its respiration rate reacts to the rise of temperature. Overall, the more sensitive the produce is to temperature increase, the more quality problems are expected in it. In consequence, it would be advised to a void using this methodology for produce with high temperature quotient (Q10) values (Saltveit, 2003). Transition times will vary widely with the specifics of a particular supply chain. On average, they would not surpass the one hour limit during land or container loading/unloading operations. However, for marine hold loading/unloading operations the situation changes, having on average a one and a half hour span. Given that the level of relevance of the suggested locations decreases with the amount of tim e of the heating/cooling episode, it is recommended that this temperature monitoring system is used mostly for produce transported either by refrigerated truck, or in a sea container, or in a combination of both methods. It is important to remember that t his system is designed solely for products of similar characteristics to the spherical bottles of water employed and which storage temperature is around 4 .00 C; therefore, it could prove useful for produce such as apples, oranges,

PAGE 97

97 pomegranate, passion frui t or tangerines. Further research is recommended in order to determine if the thermally relevant locations identified in this work could also be used for products with storage temperatures different than 4 .00 C (tomatoes, limes, etc). Readability Study The three locations found as thermally relevant in the previous study (PoR) were tested for readability. Special considerations were taken since a sensor placed in any of the sides has the risk of not being read by the system if it is used as the entry port for the forklift. This will occur because of the strong interaction of the RF waves emitted by the tag and the metal of the forklift, which might hinder the communication process with the antennas. In addition, the forklift could also shield the signal fr om the antenna to the tags, thereby, preventing communication with them. Even though it would be feasible to mark the specific side where the tag is present, this would complicate the handling of the pallet at the receiving end, where the RFID system will be in place. Since point C is located in the bottom corner of the lowest row of product on both of the sides (varying from left to right corner according to the side chosen), it was decided that both these locations and the nearest ones in the -I side ( where the corner is completed) needed to be tested for RFID readability in case reading the unsuccessful because of forklift interaction. Tags were then placed in these positions, aiming towards the corresponding face being read, in the first internal layer of all sides (Figures 4 4, 4 5 and 4 6). In configuration I two out of the three tags tested were read, while in configuration only one of the two tags set up was detected. In the first configuration, the tag placed in the PoR A was read at antenna distances of 0.50 m and 1.00 m. In addition, the tag corresponding to PoR C placed on the left of this side (C1) was only found with a 2.50 m reading distance. Finally, in

PAGE 98

98 configuration ed and only the tag in C3 (PoR C in this side) was read at antenna distances of 1 .00 m, 2.50 m and 5.00 m. The highest level of readability found in configuration I was 7.90% ( 0.01, standard deviation) and belonged to PoR A at a reading distance of 0.5 0 m. Its counterpart in side was point C3 at a reading distance of 5.00 m with 95.59% ( 0.02, standard deviation) of readability. Tables 4 5 and 4 6 summarize the average readability values for the PoRs detected in each configuration at different reading ranges. The results obtained can be explained in part by the findings of Fletcher et al. (2005) who analyzed the signal propagation of 915 MHz waves through water. Their research indicates that the strength of the RF signal depends not only on the amount of water it has to go through, but also on the period of the sinusoidal wave and the point on it inherent to each water depth. Under their analysis the period of this wave lasts approximately 0.04 m; and therefore, the lowest signal strength will be locat ed in amounts of depth of water multiple of four. In the current study, point B, which is located in the middle of the side, had no readability and point A, on the -I side, obtained some readability at two antenna distances. In the first case the RF wave had to propagate through 0.12 m of water, where the signal will decrease at its maximum; while in point A the signal had to go through 0.14 m of water, the middle point of the period between 0.12 m and 0.16 m, at the peak of the RF wave. Therefore, it is logical to expect higher readability during this particular point than at 0.12 m of depth. It can be then said that the bottle orientation and the side configuration it determined influenced the level of readability of the tags. Side configuration seemed to have played a less relevant role in the case of point C3, where according to the previous discussion readability should have been hindered. A possible explanation for this behaviour lies with the fact that there were no bottles of water on one of the

PAGE 99

99 s ides of the C3 tag. Compared to the center of that side, where point B was located, C3 had more contact with open air, which facilitated the propagation of the signal and allowed larger reading ranges. In contrast, for point C2, it is believed that any rea dability advantage generated by the side configuration and the presence of air was cancelled by the reflection of the RF waves over the surface of the sea container located near its right side. These readability results indicate that the temperature monit oring of the load could take place in PoRs placed either at the I side of the pallet (point A) or at any of the sides (point C3). However, in order to maximize the consistency of readings and reading range flexibility, the use of this last point in configuration is highly recommended. The success of this prefer red system depends on the protocol used for forklift loading. It requires that all pallets are loaded on the forklift on the side, so as to avoid marking the side where the tag is placed, putting special attention during the loading of the forklifts (so the side with the sensor is not in contact with it), or placing the antennas on an angle on top of the portal (which might reduce the readability obtained with this setup). In this case, placing the sensor on point C3 would only require installing antennas on both sides of the portal, so that the tag could be read in any of the lateral sides of the pallet on the forklift. It is also worth mentioning that since this system can work with antenna distances of 1 .00 m, 2.50 m and 5.00 m, it can accommodate rel atively small and large entry areas with portal dimensions from 3.20 m up to 11.20 m of width. Taking into account that for certain exposure times the likelihood of gathering 85 .00 % of the pallet temperatures in the interval corresponding to point A is bi gger than its analogue for point C3 (Table 4 2); a compromise between the combination re a dability/reading range and system accuracy is necessary.

PAGE 100

100 Even though the current study was designed for portal applications, the use of forklift readers could also be adequate for data retrieval during loading/unloading operations. Further research determining the readability of the PoRs found in this study using forklift reading systems is recommended. Conclusions It is feasible to perform relevant temperature monitor ing using a single RFID tag per pallet for loads of certain varieties of produce such as apples, oranges, pomegranates, passion fruit and tangerines.

PAGE 101

101 A B Figure 4 1. Area of pallet instrumented. A) Top view. B) Three -dimensional view of the bottles ins trumented in each row.

PAGE 102

102 Figure 4 2. Top view of the depth of tag placement for each configuration. Table 4 1. RFID reader configuration. Protocol Power (dBm) Q va lue TX Ant RX Ant Forward l ink Reverse l ink Inventory BS t ype C3 30 .00 4 .00 1 .00 1 .00 8 .00 kbps 8 .00 kbps portal fsk Table 4 2. Average temperature differentials present at the pallet level. Exposure t ime (h ours) Average temperature d ifferential Standard d eviation 0.5 0 /0.5 0 11.95 2.90 1 .00 /1 .00 13.01 3.30 2 .00 /2 .00 16.68 4.10 4 .00 / 4 .00 20.07 5.03 Table 4 3. Average temperature differentials present at the RPC level. Exposure time (h ours) Average temperature d ifferential Standard d eviation 0.5 0 / 0.5 0 7.15 0.98 1 .00 / 1 .00 7.54 1.30 2 .00 / 2 .00 8.85 2.01 4 .00 / 4 .00 10.71 2.60

PAGE 103

103 Fi gure 4 3. Location of the points of relevance in one half of the pallet. Table 4 4. Likelihood of gathering 85.00 % of the pallet temperatures inside the intervals calculated using these locations. Exposure t ime ( h ours) Location Point A Point B Point C 0.5 0 /0.5 0 100 .00 % 100 .00 % 100 .00 % 1 .00 /1 .00 82.35% 92.15% 74.5 0 % 2 .00 /2 .00 63.26% 63.26% 62.24% 4 .00 /4 .00 41.14% 39.06% 42.18%

PAGE 104

104 Figure 4 4. Top view of the placement of the RFID tags in the pallet. Figure 4 5. Vertical view of the locati on of the RFID tags inside the pallet in Configuration I.

PAGE 105

105 Figure 4 6. Vertical view of the location of the RFID tags inside the pallet in Configuration Table 4 5. Average readability values for the points of relevance (PoR) detected in Configuration at different reading ranges. Antenna d istan ce PoR r ead Average percentage of r eadings Standard d eviation 0.5 0 m A 7.90% 0.01 1 .00 m A 5.61% 0.02 2.5 0 m C1 0.62% 0.01 Table 4 6. Average readability values for the points of relevance (PoR) detected in Configuration at different reading ran ges. Antenna d istance PoR read Average percentage of r eadings Standard d eviation 1 .00 m C3 51.89% 0.02 2.5 0 m C3 74.11% 0.04 5 .00 m C3 95.59% 0.02

PAGE 106

106 CHAPTER 5 DEVELOPMENT OF RFID TEMPERATURE TRACKING SYSTEMS FOR COMBAT FEEDING LOGISTICS Introduction Cu rrent U S Military operations require the deployment of millions of men and women around the world. Daily, the Department of Defense (DoD) feeds more than two million people, becoming one of the largest foodservice providers in the globe. In the field, during warfare operations, the use of shelf -stable Meals, Ready to Eat (MRE) has provided for many years a viable option for combat feeding without the time and temperature constraints imposed by most perishable food supply chains. First Strike Rations (FSR) were first introduced into the mix of shelf -stable military feeding by the DoD in 2008 as a way of reducing the waste of food existent with MREs and the weight Warfighters had to carry, while still fulfilling their nutritional requirements. Unlike MREs, t hat provide sustainment on a meal -basis, FSRs are able to supply the soldier the energy requirements of the entirety of the day (DoD Live, 2009). Being shelf -stable, FSRs are able to withstand a wide range of temperatures along their supply chain. Nonethel ess, once their temperature thresholds have been surpassed (26.67C), the deterioration rate of the product is accelerated, reducing considerably its expected shelf -life (DoD Live, 2009). High temperature abuse could not only generate nutritional changes i n it, such as the loss of certain vitamins, and sensory changes in color, texture and flavour, which impact the acceptability of the product (Ross et al., 1985; Ross et al., 1987; Narayan et al., 1997; Ross et al., 1997; Shaw et al., 1997; Natress et al., 2009); but could also favor the development of spoilage microorganisms and dangerous pathogenic growth (Banwart, 1989; Ng et al., 2002). Taking into account the fact that shelf -stable meals are transported without refrigeration, and the wide array of envir onmental conditions encountered along combat feeding logistics, it is then imperative to monitor the temperature conditions of FSR loads along their supply chains. If

PAGE 107

107 inadequate temperature management of FSRs is not discovered in a timely manner, their con sumption could diminish the nutritional status of the military forces deployed, affect their morale, and, moreover, pose a food safety thread for them. Therefore, a reliable temperature tracking system must be in place during combat feeding logistic operat ions in order to detect the thermal abuse of FSR loads. Radio Frequency Identification (RFID) is a wireless auto id technology that has recently gained popularity as an asset tracking tool, mostly due to the DoD and Wal Marts initiative to use it for thi s purpose in their supply chains since 2005. In food supply chains, RFIDs potential lies in the possibility of a conjoint collection of pedigree data and time/temperature information able to satisfy the current needs of the industry for trace -back and pro cess traceability (Karkkainen, 2003; Gaukler and Seifert, 2007; Montrucchio et al., 2007; Perez -Aloe et al., 2007). Furthermore, results presented in C hapter 3 indicate that RFID temperature tracking performs better when compared to conventional temperatur e tracking systems; providing similar accuracy while increasing the simplicity of the sensing system. However the promises RFID holds in food supply chains; in each implementation, there are two major hurdles that need to be surpassed: Applying the technology in a cost -efficient way, and obtaining consistent readings. In order to achieve the first one, and reduce the cost of the implementation; the use of sensor -enabled battery assisted passive tags (BAP) and active tags has to be minimized across the load without compromising the accuracy of the system. Moreover, as presented in the previous chapter, if the thermal and perishability characteristics of the product being handled permit it, this can be easily obtained by monitoring only the relevant locations able to provide a general view of the thermal reality of the load or its Points of Relevance (PoR). Attaining consistent readings, nonetheless, is a more complicated matter; for it

PAGE 108

108 depends considerably on surpassing the interactions between the systems r adio frequency (RF) waves and the product and its environment. Previous studies by Dobkin and Weigand (2005), Redemske and Fletcher (2005), and Hartvanyi and Marek (2007) have detailed how, depending on its frequency, the radio frequency (RF) signal i s affected by the presence of water or metal in the environment. In the ultra high frequency (UHF) spectrum, for example, where most supply chain management applications take place, water absorbs the energy of the RF wave, decreasing communication between the a ntennas and the tags; while metal acts as a reflector, which sometimes can facilitate or hinder this process (Sydnheimo et al., 2006; Lazaro et al., 2009). Solutions for this problem will then depend on many factors particular to each implementation case, such as the class and configuration of the equipment being used, the reading range available, the components, packaging, and level of aggregation of the product handled, and the presence of any element in the surroundings that might interact with the syst em (metallic forklifts, sea containers, etc). If a RFID temperature tracking system is deployed in the military, monitoring the immediate ambient temperature surrounding the pallets of FSRs would simplify the instrumentation process and reduce the amount of labor involved in this task. Furthermore, negative interactions between the RF waves and the metallic components of the meals packaging could be avoided, increasing the robustness of the system. Yet, data collected from the surface of the pallet could be quite distant from the temperatures obtained in its Points of Relevance (PoR), which are the best indicative of the thermal reality and the status of the load. Therefore, a software tool able to estimate the temperature profiles in the PoRs of the pall et based upon ambient temperature readings could be needed in order to facilitate the implementation and accuracy of a RFID temperature tracking system for FSRs.

PAGE 109

109 The objective of this research was to develop two RFID temperature tracking systems for FSR lo gistics by: 1) Finding a thermally relevant location able to represent 85% of the temperatures present inside a pallet of FSRs when subjected to heating and cooling conditions similar to the ones faced along its supply chains; 2) Determining the level of r eadability of two commercially available RFID systems using a battery assisted passive tag in this thermally relevant position; and 3) To develop a software tool able to estimate the temperature profile of the thermally relevant position using as input ambient temperature readings. Materials and Methods Relevance Study A pallet of 432 FSR packets (0.20 m x 0.14 m x 0.09 m) packed in 48 boxes was instrumented with thermocouples previously calibrated in an ice bath. The pallet consisted of four layers of tw elve corrugated boxes (0.42 m x 0.26 m x 0.23 m); and each box contained three layers of FSR packets (four in the bottom, three in the middle, and two in the top), with a total of nine per box. The 142 packets instrumented provided a three -dimensional pers pective for the examination of the heat transfer process along the boxes (Figure 5 1) and the pallet. After instrumentation, the pallet was stored in a temperature controlled chamber. A series of heating episodes followed by cooling ones were staged inside it, with room temperatures ranging from 60.00 C to 35.00 C, respectively. Before each heating/cooling episode pallet temperature was stabilized at 25 .00 C. There were eight heating/cooling episodes with different durations: six hours of heating/six hours of cooling, eight hours of heating/eight hours of cooling, nine hours of heating/15 .00 hours of cooling, 18.00 hours of heating/18.00 hours of cooling, 1.00 day of heating/1 .00 day of cooling, 2 .00 days of heating/2.00 days of cooling, 3.00 days of heating/3 .00 days of cooling, and 4.00 days of heating/4.00 days of cooling. Quantitative analysis was performed using quintiles in Microsoft Excel (Microsoft Corporation, Redmont, WA).

PAGE 110

110 Readability Study Fixed system Two sides of the pallet were tested for rea dability of class 3 battery assisted passive tags (BAP) using the Intelleflex DK900 (Intelleflex Coorporation, Santa Clara, CA) RFID system operating at 915 MHz. The pallet was located inside a sea container, near the door, in the middle of it wide -wise ( Figure 5 2). The container had open air surroundings and a metallic door near by. The placement of the tags was determined by the results of the relevance study. Since the four sides of the pallet could be represented by one of the 1.22 m sides (Configurat ion ) and one of the 1.02 m sides (Configuration ), only these two were tested. Each side was studied independently. The antennas were positioned in the middle point of the front of the side being read. Readings were obtained using six different antenna dist ances (0.50 m, 1.00 m, 1.50 m, 2 .00 m, 2.50 m, and 3.00 m) with respect to the pallet. Equipment configuration. In order to maximize the read range of the Class 3 tags inside the pallet, reader output power was set at its maximum and air interface data rat es were set at their minimum. Table 5 1 summarizes the readers configuration. Once set up, the system was run in five occasions; in each one, 200 attempts of communicating with each one of the tags were made. The level of readability of the tags was deter mined by averaging the percentage of times a tag was read with respect to the attempts made of contacting it. Handheld system As in the previous case, two sides of the pallet were tested for readability. The tags used were Caen EASY2LOG Mod. A927Z (Caen RFID Srl., Viareggio, Italy), also class 3 battery assisted passive tags (BAP) working at 915 MHz. The pallet placement, the sides analyzed and the surrounding environment were the same as in the past trial. Readings were taken independently for each side of the pallet with the Motorola MC3090Z handheld reader

PAGE 111

111 (Motorola, Inc., Holtsville, NY). The equipment was configured to read EPC Gen 2 tags at 27 .00 dBm. It was not possible to perform other changes in the configuration of the reader. Two kinds of readings were taken: Inside and outside the pallet. For readings inside the pallet, the tags were placed in the same locations as in the trial with the fixed system For readings outside the pallet, for both side configurations, one tag was placed on the side wall of the top layer of product, facing the opening for the sea containers door (Figure 5 3). In addition, for configuration a different tag was also placed in the middle of the surface of the wooden pallet facing the back of the container (Figure 5 4 ). These readings were also taken independently. The operator holding the reader was then standing in the middle point of the front of the side being read. The movement of the operators body and the arm holding the reader while collecting data were differ ent for each kind of test. For example, in order to obtain readings in the interior of the pallet, the operator did the following: He bent his arm at the height of his chest, then extended it and pointed it upwards and downwards. He then held the reader at the height of his chest; and bent his knees pointing his arm towards the pallet. Finally, he returned to the initial position. For external readings, the methodology was more direct: The operator simply extended his arm and pointed upwards or downwards towards the tag. As in the experiment with the fixed system six different reading ranges were tested with respect to the pallet (0.50 m, 1 .00 m, 1.50 m, 2 .00 m, 2.50 m, and 3.00 m). Temperature Profile Estimator After analysis of the data obtained in the re levance study it was decided that a model using a capacitive effect should be applied for the estimation of the temperature profiles in the

PAGE 112

112 thermally relevant point existent inside the pallet. The software used to analyze the data was created based upon th e concept created by Uysal (2010, unpublished data ) In order to better explain this methodology, an analogy was created with the use of electrical engineering concepts. In this, the ambient temperature is modeled by the potential difference between the terminals of a voltage source and the temperature of the thermally relevant point (Point of Relevance or PoR) is modeled by the potential difference between the t erminals of a capacitor as depicted in the electronic circuit shown in Figure 5 5. In this figur e, V represents the ambient temperature and Vc represents the temperature of the PoR inside the pallet (Both in degrees Celsius). The relation between the two temperatures can be explained by the equation 5 1 : (5 1) Where Vc initial is the initial temperature of Point of Relevance, R is the resistance of the resistor and C is the capacitance of the capacitor. Thus, the temperature of the PoR rises or falls with a speed determined by the time constant (RC) and the difference between the temperature of the PoR and the ambient temperature. The bigger the difference the faster the temperature changes in the PoR. In order to find the time constant empirically, the potential V was changed and an observation on how the potential Vc changed with time was made. The temperature profiles resulting from the relevance study provided enough information on how to find the time constant for both rising and falling temperatures. After completing the rearra ngement of the terms in the above equation to find the C) factor equation 5 2 was derived:

PAGE 113

113 (5 2) Hence, if the ambient and the initial temperature in the PoR are known, and also the temperature in the PoR a given time t; it becomes trivial to calculate the time constant. Unlike the e lectronic circuit described above, it was possible to have a different time constant for the heating and cooling cycles due to the way the experiments were designed. Calculations and graphics were performed in Matlab (The MathWorks, Inc.; Natick, MA). Results and Discussion Relevance Study No thermally relevant position representing 85 .00 % of the temperatures of the pallet was detected. A great level of temperature variability was present in the load. This was mostly because of the existence of high tempe rature differentials due to heat accumulation in the middle of the boxes after the warming episodes. Even though the temperature differentials within boxes and along the pallet were quite significant in all trials, at the end of the 3 .00 d/3 .00 d and 4.00 d/4 .00 d ones these were reduced considerably because the pallet was reaching thermal equilibrium. An approach already employed in the previous chapter of using a location that has a temperature (n) that represents the middle point of an interval that gath ers the temperature readings of 85.00 % of the measured locations was proposed. The suggested range was [n+8.00 n 8 .00 ]; a total of 16 .00 C of temperature differential. Table 5 2 displays the likelihood of gathering 85.00 % of the temperature readings in this interval for each heating/cooling episode. No particular trend was found between this particular likelihood and the duration of the episodes.

PAGE 114

114 The temperature profiles obtained in the trials were then compared to the one corresponding to this hypothetica l location; resulting in the determination of two Points of Relevance (PoR), where the temperature n was recorded and where the sensors of the projected system would be placed. Point A: Layer 2 (Row 1), X= 11.00 Y= 5 .00. Point B: Layer 4 (Row 2), X= 4.00 Y= 1 .00. Where the layers move along the vertical axis (Z) and are labe led, along rows, in an ascending way from the top to the bottom. The locations of these points are represented in Figure 5 6. Although not shown in the graphic, it is important to remark that, because of thermal symmetry, these thermally relevant locations will also be present in their equivalent positions on the other sides of the pallet (along the X and Y axis). The likelihood of gathering 85.00 % of the measured temperatures of t he pallet within the 16.00 C interval that uses these points as middle point was then calculated (Table 5 3). As can be seen, Point B gives a higher likelihood during most of the short and middle duration episodes (up to 1 .00 d/1 .00 d), while Point A performs better during the longest trials. Still, the selection of the best PoR for temperature monitoring will not only depend in this estimate but also on the level of readability of the RF signal obtained in these locations. Current U S military operations have a focal interest the Middle East. Therefore, the results obtained in this relevance study have to be applicable to the varied conditions of such environment. A review of the year round weather conditions of Baghdad and Kabul (BBC, 2010b ) show the exis tence of considerable temperature di fferentials, which average 34.5 0 C, and can reach up to 44.00 C during certain fall and winter months. These differentials will also depend on where the FSRs are stored, and in which phase of the transportation process i s the load. For example, if the pallets are transported by land to be placed inside a cave, the

PAGE 115

115 differentials could be greater, because the temperatures inside this will be considerably lower than the ones registered by the weather services. Likewise, thes e temperature differentials could decrease if the load is being transported by sea, and protected from solar radiation. From a heat transfer perspective, the worst case scenario for a temperature tracking system similar to the one being designed is when th e temperatures in the pallets are quite variable, due to a high temperature differential between the pallet and the environment. The temperature differential of 95.00 C used in these trials represents then, an extreme situation very rare to find under norm al circumstances; but that gives a safety factor even for the outlier observations encountered in real life situations. Consequently, it is expected that under normal supply chain conditions, the likelihood of gathering 85.00 % of the temperature readings present in the pallet within the 16.00 C range proposed when using the PoRs found as middle points of this interval will increase. The use of a 16 .00 C temperature range as a reference for quality and shelf life assessments along this supply chain will not be a major concern, because FSRs are shelf -stable products. However, if a component of the ration has a characteristic particularly sensitive to heat or cold (degradation of vitamins, color and texture changes, etc) that might happen within that range unde r ordinary storage and transport conditions; then, a higher level of quality variability in the pallet is expected, and the application of this system could derive in the incorrect acceptance of a faulty load. As well, a stronger level of sampling and qual ity control of the load before acceptance should be established when exposure to high temperatures (over 26.67C) is recorded. Mainly, because the 15 .00 % of temperatures inside the pallet that are not represented by this temperature tracking system might p romote pathogen growth, and create food safety hazards for the troops.

PAGE 116

116 Readability Study Fixed system Figure 5 7 shows the placement of the tags for Configuration (1.22 m). These were placed both in the PoRs found, as in their thermally equivalent po sitions inside the pallet. For Configuration the pallet was rotated 90 clockwise. Configuration provided the highest number of tags read: A minimum of five for each distance (a, b, c, d, e), and six during two of the repetitions held at one meter. With Configuration four tags were read during the first five distances tested (b, c, d, e), but only three were recorded at three meters (b, c, e). The location with the highest readability of PoR A was position e; whereas for PoR B, the highest reading percentages were distributed in locations b and c in Configurations and Tables 5 4 and 5 5 summarize the highest readability levels for each PoR in each one of the configurations studied at different reading ranges. In Configura tion .00 m or 2.50 m. In configuration this distance is 2 .00 m for both PoRs. These results are quite positive when considering the metallic environment that surrounded the pallet; and the fact that the quadlaminate packaging of the components of the FSRs in direct contact with the tags contained a layer of foil. Furthermore, the high density of the FSR packets inside the boxes restricted the existence of air pockets able to propaga te the RF signal and favor tag/reader communication. Since metallic objects reflect RF waves, problems arise whenever the tags need to harvest energy in order to backscatter its information (Sydnheimo et al., 2006). Mo et al., (2007) also reported that wh enever there is a metallic object near an antenna, this can change the antennas radiation pattern, input impedance, radiation efficiency, and resonant frequency; which can decrease the performance of the RFID system. Additionally, Lzaro et al. (2009) accounts

PAGE 117

117 that, although challenging, the presence of metallic surfaces near UHF RFID applications can also generate constructive wave interference, and aid the RF system. Nonetheless, a study performed by Laniel et al. (2009) described the existence of signal attenuation at 915 MHz inside a metallic sea container due to path loss, such as multi -path and scattering. Taking into account that the tags utilized in the trial were not designed to work in metallic environments, the existence of favourable results in this readability study could be explained by a combination of factors such as the use of maximum reader output power, and the existence of overall constructive interference. The system being designed will be used during container unloading operations; and the readings will be taken moments before the pallet is removed from the container by the forklift. Consequently, the space needed between the doors of the container and the antennas must allow the transit of a forklift loaded with a pallet. So, 3 .00 m, t he maximum antenna distance tested, will be chosen. At this distance, the readability of PoR A is less than PoR B; thus, the latter will be the thermally relevant point monitored. Given that the orientation of the pallet inside the container might vary acc ording to the stowage pattern used at the time of loading, the RFID tag in this location must be read from any of the two configurations studied. Locations b and c are the only two points of PoR B read at 3 .00 m, which are equivalent to each other depe nding on the rotation of the pallet with respect to the sea container. As a result, instrumenting the pallet in any of them will achieve the systems goal. Table 5 6 shows the readability values of these positions. As can be seen, the maximum amount of re adability existent in the system i s 82 .00 % while its minimum is 32.60 %. Since both tags were detected in all cases, this last value does not imply that the tag will not be read; it only indicates that the reader will take a longer period of ti me

PAGE 118

118 detecting it. Obtaining 32.6 0 % of the 200 reader communication attempts at that particular antenna distance took less than two minutes; so, reading the tag only once will take less than that. However, even if detecting the tag takes two full minutes; this amount of time fits perfectly into this phase of the FSR supply chain, as it uses the dead time when the forklifts are transporting the pallets to their destination after unloading them from the container. Besides, if the readings are done when the forklifts are far away, there are fewer opportunities for negative interactions between its metallic surface and the RF waves. Handheld system Before reporting and discussing the results, it is important to mention that the system did not allow the estimation of tag reada bility as percentage of reads from the total number of attempts to communicate with that tag. Therefore, readability had to be estimated using the percentage of readings of a particular tag from the total number of tag readings. So, the presence of other C lass 1 Gen 2 tags detectable by the reader nearby influenced the result greatly. Since there were some tags in the surroundings these were read along with the CAEN tags. Yet in all the experiments performed with the handheld reader, the impact of these ta gs on the results diminished with reading distances. Even so, these results do not give a complete depiction of the robustness of the communication existent between tag and reader at different distances; but they do, however, allow a comparison between the readability obtained between the five trials performed: Two attempting to read tags placed in all the PoRs, two trying to read the tag on the surface of the FSR load, and one attempting to read the tag on the surface of the wooden pallet structure. Two Po R locations were initially read in Configuration at 0.5 0 m of reading distance. These were points b and d, having a percentage of readings of 3.27% and 11.73%, respectively. Nevertheless, at distances higher than 0.50 m the system could only detect point d. In

PAGE 119

119 Configuration only point b was read, and very weakly (0.09% in percentage of readings). Thus, only point d in Configuration will be taken into account in the comparison of the RFID systems. As can be seen in Table 5 7 the lowest readability resu lts at all distances were obtained by point d; which shows the strong impact that the metallic components of the packaging and the metal surfaces of the sea container had on the signal. This phenomenon was expected, given the fact that this model of Caen t ag was not designed for applications in metallic environments; and, as aforementioned, this element can affect greatly the operation of the system. Differences in tag design between the Caen system and the Intelleflex system; as well as other technical con siderations, such as reader power output and antenna size and pattern might have played an important role in their performances. Equalizing some of these variables might present a more accurate panorama and allow proper comparisons. The tag placed on the surface of the wooden structure of the pallet had the highest level of readability at every single reading distance. There are two possible causes for this: First, these tags were in contact with wood, which does not reflect RF waves, and not with the FSR boxes full of packages with metal in their structure. And second, the angle of the reader antenna and the location of the tag might have determined reflection patterns different than those in the other tag positions, which created a less detrimental multip ath effect. The results corresponding to the tags placed on the surface of the FSR load are relatively similar, though Configuration seems to provide more robust readings than This is most likely also product of different multipath effects. Even though conclusions cannot be made with respect to the robustness of the systems with respect to antenna distances, the use of those higher than 1.5 0 m is not recommended. This,

PAGE 120

120 is due to the fact that it was observed that the speed of reading of the systems was inversely proportional to the reading distance, and that obtaining readings at 2 .00 m and more took a considerable amount of time. S ince the use of a handheld reader during unloading operations will require its operator to move constantly towards the pallets and back, in order to avoid the upcoming forklifts, high speed readings are desirable. Consequently, antenna distances of 1.5 0 m are recommended as a way of combining the tag read velocity with the minimization of the distance the operator will have to walk to read each pallet. The temperatures obtained by the tag in this handheld system will later become the input data for the tem perature profile estimator; and after being processed, will give origin the temperatures obtained in PoR B. Temperature Profile Estimator The temperatures in the Points of Relevance A and B (labeled also as 20 alpha and 36 prime in the figures) showed a ca pacitor effect to rapidly changing temperature by slow heating/cooling cycles (Figures 5 8 and 5 9). Therefore, the temperatures rose and decayed with a time constant that can be determined from the eight trials from the relevance study, and can subsequen tly be modeled to estimate the temperature in the PoRs given the ambient temperature. The model was the applied by first estimating the time constants for the heating and cooling processes for each trial. As an example, the calculations corresponding to the constant for the 4 .00 rising). In this experiment, the average ambient temperature sat at 60.5 0 C. If t=0 .00 is defined as the time PoR As temperature started to increase from 24.6 0 C, then a second tem perature point time pair can be chosen to calculate the time constant. For this example, at t=76 .00 hours the temperature of PoR A reached 60 .00 C, and this point was then used in the calculations below where T represents temperature whereas t represents time.

PAGE 121

121 TPoR_A(t=0) = 24.60 C Tambient = 60.50 C TPoR_A(t=76 .00 hours) = 60.00 C Thus Similarly falling, is only needed to define two time -temperature points where the temperature of PoR A slowly approaches the ambient temperature. TPoR_A(t=0) = 60.50 C Tambient = 35.00 C TPoR_A(t=67.50 hours) = 33.00 C The fact that the falling time constant was higher shows that the pallet cooled down faster than it heated up in this trial. Based on these time constants, estimating the temperature inside the pallet can be modeled by the following two equations: If TAmbient > (T 1)PoR then g ri sin (-t/e (1 PoR Ambient PoR PoR ) 1) (T (T 1)] ([(T T (5 3) If TAmbient < (T 1)PoR then falling(-t/e (1 PoR Ambient PoR PoR ) 1) (T (T 1)] ([(T T (5 4) Where TPoR is the current estimated temperature in the PoR and (T 1)PoR denotes the previously estimated temperature sample.

PAGE 122

122 In order to gauge the performance of this mo del, the model output was compared to the actual measured temperatures of PoR A in the relevance study. Figure 5 10 shows the estimated temperatures of PoR A against the actual temperatures measured by the sensor in that position. As can be seen from this figure, the estimated temperature is much closer to the temperature in PoR A than the ambient temperature, and thus would be a much better candidate to be the temperature profile used for temperature monitoring purposes if only ambient temperatures can be collected. In terms of numerical evaluation, the absolute mean error and standard deviation between the ambient temperature and the PoR A temperature are as follows: eambient PoR_A = 17.05C ambient PoR_A = 23.60 C In contrast, the mean error and standa rd deviation between the estimated temperature for PoR A and the actual PoR A temperature are much lower: ePoR_A PoR_A_estimated = 3.4 0 C PoR_A PoR_A_estimated = 5.4 0 C Even though these calculations were performed for PoR A (point 20alpha), the results are similar to those obtained for PoR B (point 36 prime). Figure 5 11 shows the estimated results for PoR B. Conclusion Two RFID temperature tracking systems, able to monitor 85.00 % of the temperatures of a pallet of First Strike Rations (FSR) with only one RFID tag have been successfully designed for combat feeding logistic operations. The fixed system uses the temperatures collected by a tag placed in a Point of Relevance (PoR) inside the pallet. The handheld system uses as initial input ambient temper atures surrounding the pallet coming from a tag placed on the surface of its

PAGE 123

123 wooden structure, but with the help of a software tool is able to estimate the temperatures in the PoR monitored by the fixed system.

PAGE 124

124 Figure 5 1. Th ree -dimensional view of the FSR packets instrumented in each box. Figure 5 2. Pallet and antenna position during the readability study at an antenna distance of 2 m. Table 5 1. RFID reader configuration. Protocol Power (dBm) Q v alue TX A nt RX Ant Forward l ink Reverse l ink C3 30 .00 4 .00 1 .00 1 .00 8 .00 kbps 8 .00 kbps Instrumented packet

PAGE 125

125 Figure 5 3. Tag placement in the outside of the pallet for: A) Configuration B) Configuration Figure 5 4. Tag pla cement in the surface of the wooden pallet in Configuration A) B)

PAGE 126

126 Figure 5 5. A typical resistor -capacitor circuit to simulate the behavior of the pallet temperature in the presence of changing ambient temperature. Where: V= Ambient temperature ( C), R= R esistance of the resistor, and C= Capacitance of the capacitor. Table 5 2. Likelihood of gathering 85.00 % of the temperature readings in the 16.00 C range proposed. Episode Likelihood 6 .00 h, 6 .00 h 96.55% 8 .00 h, 8 .00 h 84.40% 9 .00 h, 15 .00 h 60.2 0% 18 .00 h, 18 .00 h 60.27% 24 .00 h, 24 .00 h 52.16% 2 .00 d 2 .00 d 56.28% 3 .00 d, 3 .00 d 69 .00 % 4 .00 d, 4 .00 d 76.90%

PAGE 127

127 Figure 5 6. Locations of the Points of Relevance (PoR) detected in the pallet Table 5 3. Likelihood of gathering 85.00 % of the temperature readings in the 16.00 C range proposed when using the Points of Relevance (PoR) found as the intervals middle points. Episode Point A Point B 6 .00 h, 6 .00 h 72.58% 85.23% 8 .00 h, 8 .00 h 41.24% 35 .00 % 9 00 h, 15 .00 h 63.41% 74.78% 18 .00 h, 18 .00 h 32.15% 48.58% 1 .00 d, 1 .00 d 23.87% 32.45% 2 .00 d, 2 .00 d 49.89% 44.76% 3 .00 d, 3 .00 d 42.32% 32.27% 4 .00 d, 4 .00 d 58.07% 43.11% B A X Y Z

PAGE 128

128 Figure 5 7. Locat ions of the RFID tags during the readability study for Configuration Where Table 5 4. Maximum readability values for both Points of Relevance (PoR) in Configuration using the fixed system. Antenna distan ce Position PoR read Percentage of readings 0.50 m e A 17.80% 0.50 m b B 80.50% 1 .00 m e A 32.20% 1 .00 m b B 75.90% 1.50 m e A 15.60% 1.50 m b B 82.30% 2 .00 m e A 18.40% 2 .00 m b B 82.60% 2.50 m e A 22.40% 2.50 m b B 82.60% 3 .00 m e A 6.60% 3 .0 0 m b B 82 .00 % f e Door of Container Back of Container Level 2 b d a c Door of Container Back of Container Level 4

PAGE 129

129 Table 5 5. Maximum readability values for both Points of Relevance (PoR) in Configuration Antenna distance Position PoR read Percentage of readings 0.50 m e A 69.50% 0.50 m c B 47.80% 1 .00 m e A 52.40% 1 .00 m c B 69.40% 1.50 m e A 45 .00 % 1.50 m c B 78.40% 2 .00 m e A 70.50% 2 .00 m c B 88.10% 2.50 m e A 28.50% 2.50 m c B 68.80% 3 .00 m e A 25.60% 3 .00 m c B 74.90% Table 5 6. Readability values using the fixed system of positions b and c throughout the pallet. Configuration Position Percentage of readings b 82 .00 % c 62.60% b 32.60% c 74.90%

PAGE 130

130 Table 5 7. Readability values for four different tag positions using the handheld reader. Antenna d istance Percentage of r eadings Configuration Configuration Point d (PoR B) Surface of FSR l oad Surface of FSR l oad Surface of wooden p allet 0.5 0 m 11.73% 39.50% 32.67% 41.66% 1 .00 m 1.74% 19.32% 45.81% 63.52% 1.5 0 m 0.53% 1.33% 3.55% 58.48% 2 .00 m 0 .00 % 25.64% 24.17% 94.85% 2.5 0 m 0 .00 % 0 .00 % 22.67% 90.73% 3 .00 m 0 .00 % 0 .00 % 0 .00 % 98.19% Figure 5 8. Temperature profiles showing how the temperature in PoR A (20 alpha) changed with respect to the temperature outside the pallet in all eight trials.

PAGE 131

131 Figure 5 9. Temperature profiles showing how the temperature in PoR B (36prime) changed with the temperature outside the pallet in all eight trials Figure 5 10. Comparison of ambient temperatures, temperatures measured in PoR A and the estimated pallet temperatures for PoR A (point 20 alpha).

PAGE 132

132 Figure 5 11. Comparison of ambient temperatures, temperatures measured in PoR B and the estimated temperatures for PoR B (point 36 -prime).

PAGE 133

133 CHAPTER 6 DEVELOPMENT OF A LOA D MANAGEMENT SYSTEM FOR COMBAT FEEDING LOGISTICS BASED ON S HELF LIFE PREDICTION SOFT WARE Introduction Combat feeding during military campaigns has represented a challen ge for armies throughout history. The lack of refrigeration and the possibility of limited food supply in the field have driven some of the biggest developments in food conservation and processing; the most remarkable being canning. During the last fifty y ears, technological efforts have been aimed towards developing safe and nutritious foods able to withstand the rigorous aspects of military supply chains. And so, in 1993 the U S Department of Defense (DoD) created a successful set of meal based shelf -sta ble products named Meals, Ready -to -Eat (MRE) able to defy most of the time and temperature limitations existent in food distribution and storage, and suitable for their logistics. Sometime after their introduction, feedback from the field indicated that s oldiers were opening the MRE packs, and only taking with them and eating certain elements of these in order to reduce the amount of weight carried. This not only generated big amounts of food waste, but also endangered the nutritional status of the troops. A new shelf -stable product called First Strike Rations (FSR) was then designed to take on the place of MREs in assault operations. With it, Warfighters could bring a lightweight load with enough food to fulfill their nutritional requirements for a whole day during fast -paced maneuvers (DoD Live, 2009). A First Strike Ration (FSR), although shelf -stable, will still deteriorate faster if its storage temperature of 26.67C has been exceeded (DoD Live, 2009). Exposure to high temperatures accelerates negative changes in the meals sensory attributes, which influence its level of acceptability. In addition, storage of FSRs under these environmental conditions could affect adversely the health status of the troops, by promoting nutritional changes, and the growth of

PAGE 134

134 spoilage and pathogenic microorganisms in their meals (Ross et al., 1985; Ross et al., 1987; Banwart, 1989; Narayan et al., 1997; Ross et al., 1997; Shaw et al., 1997; Ng et al., 2002; Natress et al., 2009). Consequently, high temperature abuse will d iminish the shelf life of FSRs and, in some cases, significantly. Since the majority of current U S military operations take s place in desert areas where high temperatures are reached and the loads of FSRs do not move along environmentally controlled supp ly chains, there are many opportunities for high temperature abuse. In the previous chapter the use of RFID temperature tracking systems as a way of monitoring the conditions of these loads and detecting thermal abuse was proposed. However, during fast -pac ed military deployments, the data collected in these could be difficult to decipher; since the temperature profiles recorded are by no means a stand alone indicative of the remaining shelf -life of the product. Therefore, a load management system that uses a software tool able to decode such data, providing intelligible information about the shelf -life status of the product and giving recommendations on the actions required with respect to the load is needed. The main step in the creation of such software i s the selection of an appropriate model for the prediction of the FSRs shelf life. There have been a number of studies on modeling the shelf life of food products. Many of them are based in variations of the Arrhenius equation, which describes the kinetic s of simple chemical reactions as dependant on temperature conditions (Hertog et al., 1999; Brody, 2003; Giannakourou and Taoukis, 2003; Hough et al., 2006). Also popular is the use of predictive microbiology models, such as the Gompertz equation, as a mea n of determining the deterioration and safety thresholds limiting shelf -life (Labuza and Fu, 1993; Zwietering et al., 1993; McMeekin and Ross, 1996; Riva et al., 2001; Corbo et al., 2006). Less common are empirical models, which are built upon experimental data

PAGE 135

135 obtained in the laboratory setting (Wilcke, 1992; Jacxsens et al., 2002; Nunes et al., 2004). A ccording to van Boekel (2008), these models are able to describe closely complicated temperature-dependent reactions existing in food within the parameter region on which they are based. Given the variety of the nature of products inside FSRs and the complexity of the reactions existent within them, an empirical model would then be the best option for shelf life prediction. The application of the load manag ement system proposed also relies in the quality of the temperature data collected. Taking into account the sampling methodologies previously developed for this particular supply chain in the last chapter, it is necessary to determine whether variations in the sensor locations proposed in these will alter the shelf -life estimations. Likewise, it is also important for the system to determine whether the pallet is changing its shelf life uniformly or if some of its contents (meals) are doing so at a higher ra te than others. The objectives of this research were: 1) Developing software able to predict the shelf life of pallets of FSRs under current supply chain environmental conditions. 2) To establish the maximum amount of weeks of shelf -life possible for FSRs during storage in the deployment areas in a real life case. 3) To determine whether there are significant differences in the predicted acceptability scores when comparing three kinds of FSR meals. 4) To determine whether there are significant differences in the predicted acceptability scores of FSRs when comparing temperature sampling locations in the pallet. Materials and Methods Development of the Software The first step in the development of the software was collecting the information necessary in orde r to create the shelf life prediction models. Since all boxes present in the pallet of FSR studied contained the same three menu options, each menu was analyzed in order to determine

PAGE 136

136 which one of its items deteriorated faster. These would then be considere d the shelf -life limiting items (SLLI) for the pallet. The following were the results of this preliminary testing (Nunes et al., 2010, unpublished data): In Menu 1, the SLLI was the Bacon Cheddar Sandwich; in Menu 2, the Italian Style Sandwich; and in Menu 3, the Zapplesauce. These products were then subjected to controlled storage at two different temperatures (48.89C or 120.00 F, and 60.00 C or 140.00 F). Since the easiest way to determine the shelf life of the product during a real -life military deploym ent is the level of acceptability for consumption, this attribute would then be considered in the model as the limiting quality factor in the shelf life of FSRs. Thus, sensory analysis was performed on each product/temperature combination at different stor age times. The scale for the acceptability evaluations was from 1 to 9; being 1 the worst acceptability score and 9 the best acceptability score. A score of 4 was considered the limit value of acceptability fit for consumption. Curves were then created when plotting the acceptability for consumption versus the weeks of storage. For each temperature, the resulting curve was fitted into a third order polynomial equation. A third curve was built for all products in order to describe the deterioration behavior of these at their ideal storage temperature of 26.67C (80.00 F). Due to the fact that similar storage tests under this temperature were not finished by the time of the completion of this study, it was assumed that each product started with an initial qua lity equal to the average initial quality they presented for the first two temperatures considered. It was also assumed that the limit of acceptability of consumption for all products (a score equal to 4 .00 ) was reached at 104 .00 weeks (2 .00 years) of stor age under 26.67C (80.00 F). Given that only the initial and final acceptability values were given, the resulting curves were only first order polynomial curves.

PAGE 137

137 The software was then designed by integrating the resulting relationships into a task automat ing system called macro within the Microsoft Visual Basic (Microsoft Corporation, Redmond, Washington, 19752010) platform. The inputs for the program were the initial acceptability scores of each one of the menus (also referred to as initial quality scor es), the number of temperature data points collected along the shipment, and their corresponding temperatures in degrees Fahrenheit. Since the temperature profiles for combat feeding logistics provided by the U.S. Department of Defense contained only one d ata point per day, the number of data points in the software equaled the number of days when temperature information was collected. The rationale behind the model is as follows. For each product, the system estimates the time value corresponding to the in itial quality score along each one of the set curves (26.70 C, 48.89C and 60.00 C). It then adds up the amount of time passed between the data points (in this example, one day) and estimates the acceptability score related to that new time in the set curv es. The resulting quality score after subjecting the product to a certain period of time at the initial temperature will be calculated by interpolating between the acceptability values given at 26.7 0 C (80 .00 F) and 48.89C (120.00 F), if the initial tempe rature is between them; or between 48.89C (120.00 F) and 60.00 C (140 .00 F) if the temperature falls within this range. This is the final acceptability score for that data point, and it will be used as the initial quality for the next iteration. Once all iterations are completed and the last acceptability value has been estimated fo r each product; the system provide s these as outputs. In addition, it also average s these final quality scores and provide s an average acceptability for the pallet. Finally, dep ending on the status of this particular value with respect to the minimum acceptability level for consumption

PAGE 138

138 for all products (4 .00 ), a window display s Pass or Discard indicating the required action regarding the pallet. Application of the Shelf -Life Software Ambient temperature profiles were developed using the information collected in previous trials performed by the U.S. Department of Defense (DoD) on shipments from the U S to Kuwait. A temperature profile was created based on worst -case scenario ambient conditions recorded during those trials. This was generated by taking the maximum temperatures registered in 17 nonrefrigerated sea containers during 39.00 days where environmental monitoring of the shipment and storage periods was performed. Ad ditionally, two other temperature profiles were developed also for ambient temperatures. All of them had as initial input the temperature profile described before plus one or two extra weeks of data of hypothetical temperature information corresponding to the storage phase in the Middle East. These data points were created by repeating in each week the temperatures obtained originally in the trials under this storage stage. The following is the description of the ambient temperature profiles created: Profi le 1: Original shipment information (storage in the US + marine shipment + 1.00 week of storage in the Middle East). Profile 2: Original shipment information + 1 .00 week of storage in the Middle East. Profile 3: Original shipment information + 2 .00 weeks o f storage in the Middle East. All ambient temperature profiles were assumed to be recorded in the outside wall of the pallet; in the first location considered for the temperature sensor. The second and third sensor positions corresponded to the locations of the Points of Relevance (PoR) A and B from the research work described in the previous chapter.

PAGE 139

139 Temperature profiles for PoR A and B were then obtained by entering the three temperature profiles corresponding to the ambient conditions into the temperatur e profile estimator presented in the aforementioned study. Once all temperature profiles were estimated, these were entered into the software individually. For all cases, the system was set up using as initial quality values the initial acceptability sco res obtained by the Shelf -Life Limiting Items in the sensory analysis employed for the development of the model (7.85 for Menu 1; 7.90 for Menu 2; and 8 .00 for Menu 3). Once run, the resulting information was collected. Statistical Analysis The normality of the data was verified with the Anderson -Darling test. A Generalized Linear Model (GLM) with a Gaussian family was used in order to determine the effects of the position of the temperature sensor and the kind of meal studied (Freund and Wilson, 2003). I t was also used to determine the equality of means of the acceptability score predictions obtained with the different combinations of sensor positions and meals. Significance was accepted at level he R Foundation for Statistical Computing., Wiedner Hauptstrabe 8 10/1071 1040 Vienna, Austria). Results and Discussion Development of the Software The model used to construct the software was developed based in a set of equations relating acceptability sc ores and time at different temperatures for each one of the SLLIs. Table 6 1 summarizes these relationships and their coefficients of determination (R2). As can be seen, most of the equations presented had coefficients of determination close to one; which indicates that the results obtained with these relationships were very close to the original input data coming from sensory analysis.

PAGE 140

140 In the case of the fitting curves for 26.67C (80.00 F), the lack of experimental data points could be promoting the sim plification of a more complex polynomial relationship into only a first order polynomial equation. So, the model estimations for temperatures within the 26.67C (80 .00 F) and 48.89C (120.00 F) range could have a certain level of inaccuracy. Yet, once the experimental data is available and the resulting curve is subjected to regression, this software could be easily updated for model improvement. The best way to determine whether the current or future versions of this software are indeed accurate is by val idating the model. This would allow a comparison of the calculated acceptability scores versus the ones obtained by sensory evaluations in product subjected to a particular temperature profile during a real life supply chain. Anderson (2010) used this meth odology as a tool for concluding that empirical models are better suited for shelf -life prediction than theoretical models in sweet corn cold chains. Likewise, this process could assist in the determination of errors and minimization of them in this softwa re; and thus it is necessary as future work. In addition, a feature indicating the amount of shelf life in weeks left in the pallet given the latest temperature pattern could be particularly useful in the field; and should also be pursued in the future. Finally, Figures 6 1 and 62 present the two recommendation scenarios in the final software product. Maximum Amount of Weeks of Shelf -Life at the Deployment Areas The results indicate that for all sensor positions, all the meals in the pallet of FSR have l ess than three weeks of shelf -life at storage conditions in the deployment area (Table 6 2) Thus, in order to avoid any risks, it would be recommended to distribute and consume the entire product present in the load around the two week period, when all pr oducts are still over the acceptability

PAGE 141

141 threshold score of four. An optimization could be made in order to determine the particular amount of days of shelf life (within 14 .00 and 21.00 ), but it was considered more appropriate, to facilitate the load handli ng in -situ, giving the recommendations to the Warfighters in weeks. It is important to remark that the temperature profile employed was the worst case scenario in the set of shipments studied previously by the DoD. Within these shipments there was a high level of variability in the temperature profiles recorded depending on the position of the container on the ship and the position of the sensor inside the container. So, even under similar environmental conditions, the level of variability in temperature p rofiles was wide and thus, possibly the acceptability scores of the load and its expected shelf -life. In addition, these shelf -life results could differ considerably according to the weather patterns. For example, if in real -life the temperatures increase more than the ones used in this simulation; then, the product will probably have less acceptability scores than the ones obtained, and could also reach its acceptability threshold after just one or two weeks of storage at the deployment area. Since the te mperature profiles provided by the DoD correspond to operations in the months of September and October, it is expected that the shelf -life of the load gets reduced during the months of July and August, at the peak of the summer. Likewise, if the ambient te mperatures are lower, the amount of shelf -life of the product could be extended, particularly during the winter and fall months. Therefore, given the high level of variability in temperature profiles and environmental conditions, it is important to perfor m this kind of shelf life analysis for every single shipment received throughout the year. Differences between Acceptability Estimates for FSR Meals Table 6 3 shows the averages and standard deviations of the three meals combined for each one of the ambie nt temperature profiles and its related profiles. As can be seen, standard

PAGE 142

142 deviation values are between the range of 0.11 and 0.26. These values increase with respect to the amount of time spent in storage at the deployment area, which is probably relate d to the acceleration of the products degradation rates at increased amounts of high temperature abuse. Statistical analysis of all the profiles shows that there are no significant differences between the acceptability scores for FSR Meals for all sensor positions (P = 0.65 ; P=0.7 6 ; P=0.5 8 ). Tables 6 4, 65 and 6 6 summarize the GLM Analysis performed when using as input for the model Profiles 1, 2 and 3 and their related profiles for PoRs A and B. These results imply that the degradation of the acceptabi lity in the load takes place uniformly. Thus, using an average of the final acceptability scores of the three meals is a viable option to indicate the overall state of the pallet. Differences between Acceptability Estimates a t Different Sensor Locations Wh en analyzing the resulting data of combining the acceptability scores of all the temperature profiles obtained at different sensor positions given a particular ambient temperature profile (Table 6 7), it is very clear that the levels of variability of thei r original scores were very low. The standard deviations in this case range from 0.01 to 0.02. Further statistical analysis (Tables 6 4, 65, and 6 6) using a Generalized Linear Model shows that there are no significant differences between the acceptabi lity scores obtained when placing the temperature sensor in the three different locations studied. This means that with regards to shelf -life estimations, placing the sensor outside the pallet recording ambient temperature is the same as recording product temperature inside the pallet at the points of relevance A and B. As was found in the previous chapter, working with these two locations allowed a thermally relevant instrumentation, where most of the temperatures in the pallet were represented.

PAGE 143

143 This find ing creates a strong practical implication for the military, since it would reduce considerably their instrumentation times and labor costs at the warehouse stage given that instead of setting up the sensors inside the loads, they can just place these on t he surface of them. Moreover, when using RFID sensors, installing them on the outside of the pallet will increase the readability of the tag and reduce the negative effects of the interactions between the radiofrequency waves and the metallic packaging of the FSRs; facilitating their initial programming and the data retrieval process at the receiving end. Yet, before implementing a sampling system using ambient temperatures, these results should be repeated using a full set of data points with smaller data collection intervals. Since the temperature profiles available for this study had only one daily reading, the temperature profile estimator used to develop the curves corresponding to PoR A and B used this as the timeframe for the heat transfer process in side the pallet. However, data with smaller collection intervals could provide a more dynamic idea of the temperature change process and take into account other temperature variations along the day that might modify the initial resulting temperature profil es for both relevant points. Consequently, it is necessary further research to determine if this phenomenon would change the final acceptability scores of the load and if this will have a negative or positive effect on the current estimates. Conclusions A load management system for First Strike Rations that uses a software tool able to provide recommendations on the actions required with respect to the load and acceptability scores based on shelf -life prediction models and the temperature history of the s hipment was developed. This was applied given a real -life temperature profile and it was found that the pallets corresponding to that shipment had less than three weeks of shelf life during their storage at the deployment area. In addition, it was determin ed that using an average of the final acceptability

PAGE 144

144 scores of the three meals as an indication of the overall state of the pallet is viable. Finally, it was established that relevant temperature sampling for shelf -life prediction purposes can take place ou tside the pallet.

PAGE 145

145 Figure 6 1. Screenshot of the final shelf -life prediction software recommending passing the load. Figure 6 2. Screenshot of the final shelf -life prediction software recommending rejecting the load.

PAGE 146

146 Table 6 1. Fitting curves desc ribing the acceptability of the shelf -life limiting items as a function of time at three different temperatures. Where y is the acceptability score and x is the amount of time spent at a particular temperature. Menu Shelf life limiting i tem (SLLI) Temp erature (C) Equation R 2 1 Bacon Cheddar Sandwich 26.67 y = 0.037x + 7.85 1.000 2 Italian Style Sandwich 26.67 y = 0.037x + 7.85 1.000 3 Zapplesauce 26.67 y = 0.037x + 7.85 1.000 1 Bacon Cheddar Sandwich 48.89 y = 0.0345x3 + 0.3248x2 1.3912x + 7. 9255 1.000 2 Italian Style Sandwich 48.89 y = 0.0174x3 + 0.1936x2 1.2079x + 7.9353 1.000 3 Zapplesauce 48.89 y = 0.0055x3 + 0.1374x2 1.3228x + 7.95 1.000 1 Bacon Cheddar Sandwich 60 .00 y = 0.104x3 + 1.0545x2 3.7997x + 7.7235 0.989 2 Italian St yle Sandwich 60 .00 y = 0.0681x3 + 0.7681x2 3.2935x + 7.8207 0.999 3 Zapplesauce 60.00 y = 0.1197x3 + 1.2223x2 4.1828x + 7.8832 0.990 Table 6 2. Final acceptability score predicted by the software. Sensor l ocation Profile 1 Profile 2 Profile 3 Meal 1 Meal 2 Meal 3 Pallet Meal 1 Meal 2 Meal 3 Pallet Meal 1 Meal 2 Meal 3 Pallet Environment 5.49 5.56 5.34 5.46 4.52 4.70 4.36 4.53 3.37 3.87 3.75 3.66 Point A 5.53 5.59 5.36 5.49 4.56 4.72 4.38 4.55 3.40 3.89 3.76 3.68 Point B 5.53 5.59 5.36 5.49 4.56 4.72 4.38 4.55 3.40 3.89 3.76 3.68 Action Pass Pass Pass Pass Pass Pass Pass Pass Discard Discard Discard Discard

PAGE 147

147 Table 6 3. Averages and standard deviations of the combination of the three FSR meals analyzed when subjected to all temperature profi les. Sensor l ocation Profile 1 Profile 2 Profile 3 Average Standard d eviation Average Standard d eviation Average Standard d eviation Ambient 5.46 0.11 4.53 0.17 3.66 0.26 Point A 5.49 0.12 4.55 0.17 3.68 0.25 Point B 5.49 0.12 4.55 0.17 3.68 0.25 Table 6 4. GLM analysis of the acceptability scores for Profile 1 and its related profiles. Estimate Std. e rror t Value P (Intercept) 5.53 0.17 32.42 8.91e 10 Sensor l ocation 0.02 0.08 0.25 0.81 Meal 0.03 0.06 0.4 7 0.65 Sensor l ocation: Meal 0.00 0.02 0.06 0.95 Table 6 5. GLM analysis of the acceptability scores for Profile 2 and its related profiles. Estimate Std. e rror t Value P (Intercept) 4.59 0.26 17.63 1.09e 07 Sensor l ocation 0.02 0.12 0.16 0.88 Meal 0.03 0.10 0.32 0.7 6 Sensor l ocation: Meal 0.00 0.04 0.05 0.96

PAGE 148

148 Table 6 6. GLM analysis of the acceptability scores for Profile 3 and its related profiles. Estimate Std. e rror t Value P (Intercept) 3.46 0.36 9.43 1.31e 05 Sensor l ocation 0.02 0.17 0.10 0.92 Meal 0.08 0.13 0.57 0.58 Sensor l ocation: Meal 0.00 0.06 0.04 0.97 Table 6 7. Averages and standard deviations for each meal when combini ng the three temperature profiles analyzed. Profile 1 Profile 2 Profile 3 Meal 1 Meal 2 Meal 3 Meal 1 Meal 2 Meal 3 Meal 1 Meal 2 Meal 3 Average 5.52 5.58 5.35 4.55 4.71 4.37 3.39 3.88 3.75 Standard d eviation 0.02 0.02 0.01 0.02 0.01 0.01 0.02 0.01 0 .01

PAGE 149

149 CHAPTER 7 RETURN ON INVESTMENT (ROI) DETERMINATION FOR THE DEPLOYMENT OF A RFID BASED LOAD MANAGEMEN T SYSTEM IN COMBAT F EEDING LOGISTICS Introduction The particularities of dealing with perishable inventory have been studied for many years (Nahmia s, 1982; Goyal and Giri, 2001; Bogataj et al., 2005; Ferguson and Ketzenberg, 2006). RFID has been recently suggested as a new tool for obtaining operational efficiencies in the supply cha ins of perishables as a mean of obtaining stock level transparency a nd improving the application of inventory retri eval policies such as First In First Out (FIFO) and Last In First Out (LIFO) (Krkkinen, 2003; Chande et al., 2005; Broekmeulen and van Donselaar, 2007). However, the need for combining this information w ith the thermal history of the load due to safety and quality concerns (Giannakourou and Taoukis, 2003; Sahin et al., 2007) opened the door for the application of sensor -enabled RFID tags and the possibility of sorting the products based on the amount of t hermal abuse suffered during the supply chain and its impact in product quality (Ilic et al., 2009). Consequently, it has been suggested empowering RFID temperature tracking systems with shelf life prediction models in order to improve sorting at the dist ribution center and store level (Emond and Nicometo, 2006; Jedermann et al., 2008; Ilic et al., 2009). Based on the shelf -life information collected for each pallet or load, FIFO and LIFO could be replaced by a First Expired First Out (FEFO) retrieval po licy, that reduces the economic loss and quality and safety concerns generated by handling expired product in the supply chain. In the previous chapters, a relevant and feasible instrumentation protocol for RFID environmental monitoring of First Strike Ra tions (FSR) logistics was created; and the resulting RFID temperature tracking system has been combined with shelf -life models in order to build a load management system The aim of this system is achieving a FEFO management of the FSR

PAGE 150

150 inventory based upon the products acceptability values in two combat zones: Iraq and Afghanistan. It is believed that if applied, the FSR loss generated by thermal abuse during pallet transport will be reduced and great savings will be achieved. However, the costs of the deployment of such system are considerable; hence, a proper economic analysis of the benefits of this implementation must be performed in order to make the final decision with regards to its application. The objective of the research presented in this chapte r was to develop the business case for the deployment of a load management system for FSRs, resulting from the combination of RFID temperature tracking and shelf -life prediction software, applied during combat feeding logistics operations by determining it s return on investment (ROI). Return on Investment (ROI) Analysis This analysis works within a five year time frame due to the fact that by that time, improved RFID systems might be available and thus, an update of the whole proposed system could be in order. The suggested deployment implies the installation of 50 RFID stations along Afghanistan and Iraq. Based on the current and future distribution of Warfighters in the field, it was determined that thirty four of the work stations will be located in Af ghanistan and sixteen in Iraq. Estimation of RFID Deployment Costs In order to determine the cost of the deployment, yearly cost analyses were performed. During the initial year, the cost of numerous factors were included: The hardware and software for ea ch work station, the RFID tags, the backup hardware and software for all work stations, maintenance and training, and integration costs. Most of the costs included in this study were obtained from quotes from real companies; while some, such as the cost of the shelf -life software, were estimated based on similar applications encountered in real life situations.

PAGE 151

151 The core of each work station included a hand -held RFID reader, a laptop computer and a wireless Smart Label printer. The reader would then be used to read the tags placed in each pallet of the container, collecting inventory information as well as providing shelf -life estimations for those with temperature sensing capabilities. The information collected by the reader would later on be broadcasted eit her through Bluetooth or Wi -Fi 802.11 a/b/g to the laptop computer, which would gather it and add this information to the back -end system of the U.S. Department of Defense (DoD) through a secure internet connection. The wireless Smart Label printer would be used as a back up system, in case any passive RFID tag used for inventory management traveling with the shipment needs to be replaced due to damage or failure. Additionally, the printer would create Smart Labels containing the approximate acceptability values of a pallet at the time of arrival using the average acceptability scores of the load, whenever the pallets temperature RFID tag suffers damage in transit. This action, although does not assure the proper use of the information gathered and cannot aid in further data collection with respect to environmental monitoring, provides more resources to the people in charge of distributing the meals, allowing them to make a more informed decision. Different kinds of software will be necessary in order to operate the work stations. Some of those used in typical inventory management practices using wireless RFID readers have been included, as well as the shelf -life prediction software developed in the previous chapter, the drivers for the printer, and the program necessary for the creation of labels. Other software applications necessary to link each work station and the data gathered in them with the back -end system of the DoD are included in the price of the integration. In addition, a five -year service pla n for each reader offered by Motorola, that extends normal wear and tear repair coverage to include accidental breakage of cracked plastics, broken

PAGE 152

152 screens, keypads and more has also been added into the initial cost. Table 7 -1 offers a detailed description of the hardware and software related items needed for each work station and their costs. The cost of tags was estimated by adding the number of temperature tags needed for temperature tracking plus a safety stock of 15.00 % which would replace all the tags that malfunction or are damaged before being deployed. Also included was the cost of the passive tags used by the wireless printers in each work station (Table 7 2). It is important to mention that the number of temperature tags required for temperature tracking was based on results that will be reported later on in this document on the yearly demand for FSRs in Afghanistan and Iraq. Also included in the initial investment was the purchase of backup hardware and software which, as its name states, is avai lable in case the ones deployed are inoperable or being subjected to repairs. This cost was estimated to be 10 .00 % of the total initial hardware and software sales price. System maintenance and employee training were calculated as 25 .00 % of this particular price too; while the cost of integration was assumed to be 2.5 0 million dollars. Table 7 3 presents the costs estimated for the first year of the project; which total around 3. 9 million dollars. There were three recurring costs present in every year of t he analysis: The cost of tags, the cost of maintenance and training and the cost of system replacement. This last factor gathered all the expenses generated when changing damaged elements of the work stations permanently and has been taken to be 25 .00 % of the initial cost for hardware and software for all work stations. Two additional costs were taken into account for the beginning of years three and five: The replacement of both reader and printer batteries. These need to be changed because of the end of their lifespan after two years of activity. T able 7 4 present s the costs incurred at the beginning of the second and fourth year s of operation ; while Table 7 5 shows the costs for the third and fifth

PAGE 153

153 years Table 7 6 summarizes the costs of the RFID deploy ment year by year and gives the total amount required for the project: $7,076,628. 00. Note that the costs for the last two years are assumed to be the same of the second and third year, respectively. Estimation of the Benefits resulting from the RFID and Load Management System Implementation In order to estimate the economic benefits of the RFID system proposed, it is necessary to first determine the amount of FSRs being shipped to Afghanistan and Iraq and then, estimate the economic loss avoided by this s ystem. Estimation of FSR d emand in Afghanistan and Iraq According to recent reports (BBC, 2010 a ), the amount of U.S. soldiers deployed in Iraq will be decreasing from around 98,000 to 50,000 by the end of 2010. Consequently, this will be the amount of sol diers present in Iraq for this analysis. Likewise, these reports also mention that the U.S. troops in Afghanistan will be close to 100,000 by the summer of 2010. Therefore, this will be the number of soldiers assumed as stationed in this country for F SR de mand calculations. Table 7 7 summarizes the distribution of U.S. soldiers in these regions considered for this study. Assuming only 10 .00 % of the 150,000 Warfighters in these areas consume FSRs in any given day, then there would be a demand for 15 ,000 FSR meals per day (Table 7 8). The number of pallets per day necessary to satisfy this demand can then be calculated and, taking into account the number of pallets transported in a 40 foot sea container is 40, so is the number of containers these would require per month and year (Table 7 9). Since the shipments to the zone of conflict take place every month and the containers have to be fully loaded, an approximation is made by rounding the monthly numbers obtained above and new yearly estimations were found (T able 7 10).

PAGE 154

154 Since military deployments of this magnitude require the use of safety factors for their combat feeding operations in case there are any pro blems with their shipments, a 15.00 % of the total number of yearly containers was added in order to obt ain the grand tota l of FSR pallet demand (Table 7 10). The safety stock will be stored in areas of the field less prone to temperature a buse, where conditions are favorable for the maintenance of product stability and deterioration rates are reduced (i.e. in caves), and deployed whenever necessary to cover for damaged product Estimation of FSR pallets l ost to thermal abuse The results obtained in the previous chapter indicate that, in one of the worst -case scenarios of environmental conditions studied by t he military (the profile used as the basis for the shelf life studies in Chapter 6) the handling of FSR containers allows a shelf life of between two and three weeks in these combat zones. Since transport between the port of Karachi (Pakistan) towards Afg hanistan and transport amongst the port of Umm Qasr in Iraq towards other strategic areas in this country can take between three to eight days, an average of five days will be taken as land transportation time in both cases. Therefore, based upon these ass umptions, the remaining shelf -life of the pallets once they have arrived to their destin ation will be of approximately 1 .5 weeks. Yet, this particular case is an extreme and does not represent a common situation; so, an extra week of shelf life will be con sidered as a way of accounting for this factor. Consequently, the remaining shelf life after the pallets have arrived to their final destination will be considered of about 2.5 weeks. Since the shipment must last for a whole month, some FSR pallets will ha ve to be stored after this deadline, in order to be eaten during the third and fourth week of the month. Yet, if the product is expired and is not suitable for consumption, the load must be discarded, generating economic loss. So, if the pallets that would be consumed during the last 1.5 weeks of the month have been thermally abused, about 37.5% of

PAGE 155

155 the total monthly shipment could be classified as loss. However, since there is no certainty all of them would be subjected to extreme environmental conditions, only half of this percentage (18.75% of the total monthly pallet shipment) was taken into account as a conservative estimate for pallet losses in the upcoming calculations. Given the fact that even with the implementation of the load management system sug gested some pallets might still be deemed as los t due to exposure to extreme circumstances during transit, the efficiency of the system has be en considered only 50.00 %. Table 7 11 summarizes the estimations of yearly pallet losses (pallets mishandled) and the resulting pallet losses avoided once the proposed system is into place. Estimation of the e conomic l oss of the FSR p allets lost to thermal abuse Two factors were taken into account in order to calculate this estimate. First, the cost of the pallet of FSRs; and second, the cost of the maritime shipment from Philadelphia, PA to the two receiving ports: Karachi in Pakistan (en route to Afghanistan) and Umm Qasr in Iraq. The cost of land transportation and handling from these ports up to the final destina tions was not included due to the high variability existent in their locations given the nature of combat operations. Table 7 12 presents the estimation of the total costs of maritime shipment per pallet and container according to the country of fi nal rece ption. Finally, Table 7 13 shows the calculated economic loss of a pallet sent to each one of the combat zones: $ 5,102.95 for Iraq and $ 5,069.26 for Afghanistan. Estimation of FSR pallets sent in emergency shipments Since the number of pallets lost to t hermal abuse surpass that of the safety stock, emergency shipments with more FSRs are necessary in order to feed the troops once the initial inventory has been depleted. Table 7 14 presents the amount of yearly containers devoted to this purpose and Table 7 15 the amount of m oney allocated to yearly emergency shipments

PAGE 156

156 Estimation of the e conomic l oss avoided by the load management system The economic loss avoided by the proposed system will be the amount of money saved by the DoD once this system is deployed. The calculations were done by multiplying the amount of pallets that the system will allocate before expiration and prevent from becoming a loss with the economic loss per pallet given its country of destination. Additional savings will derive from el iminating the emergency shipments since the resulting pallet loss levels after the implementation of the system can easily be covered by the yearly FSR safety stock Table 7 1 6 summarizes the results obtained. The aggreg ate for both items totals $11,761,779.72 per year; and, consequently, the amount saved during the five years of its i nitial operation will reach $58,808,898.63. Return on I nvestment (ROI) Estimation The ROI was determined by replacing the corresponding data in Equation 7 -1. The current disc ount rate given by the U.S. Federal Reserve (0.75%) was used for the calculations (Wall Street Journal, 2010). 1 ) 1 ( ) 1 (0 0 n t t n t tD Vi D Vt ROI (7 1) Where Vi is the investment required by the project at the end of the time period t Vt is the monetary yield of the project at the end of time period t n is the number of periods in the analysis (5) t is the iterative time period ; and D is the discount rate for the time value of money (0.75%)

PAGE 157

157 After performing the calculations the retu rn on investment of the project was estimated to be 719.49%; which provided a solid financial reason for the projects implementation. Furthermore, the savings offered by the first year of operation of the system could pay for more than all the costs of th e deployment; allowing a net gain for the DoD during the next four years of operation. Conclusion Based upon the business case presented, the deployment of a load management system for combat feeding logistics founded on the combination of a RFID temperature tracking system with shelf -life prediction software allows a return on investment of 719.49% during the initial fi ve years of operation; making its deployment a sound economic decision cap able of generat ing considerable amount of value in the supply cha in

PAGE 158

158 Table 7 1 Hardware and software costs for each work station. Item Number of units Unit price Total price Reader Mc3090 Z reader 1 .00 2647.1 0 $2,647 .00 Lithium baterry pack (x2) 2 .00 78.4 0 $157 .00 4 Slot battery charger 1 .00 140 00 $140 .00 AC Line cord for power supp l ies 1 .00 15.8 0 $16 .00 Universal power supply 1 .00 43.5 0 $44 .00 Protective boot 3 .00 6 .00 $18 .00 Holster 1 .00 47.63 $48 .00 Belt for holster 1 .00 21 .00 $21 .00 Screen protectors 5 .00 9 .00 $45 .00 Subtotal reader $3,135 .00 Laptop computer HP Pavilion dv7t Select Edition series 1 .00 949.99 $950 .00 Subtotal laptop computer $950 .00 Smart Label printer RP4T mobile printer 1 .00 2566 .00 $2,566 .00 Vehicle charger 1 .00 89.1 0 $89 .00 Li Ion quad ch arger for RP4T batteries 1 .00 375.24 $375 .00 Spare battery 1 .00 160 .00 $160 .00 USB A to PC cable 1 .00 23 .00 $23 .00 Softcase 1 .00 85 .00 $85 .00 Pinhead cleaning pens 3 .00 40 .00 $120 .00 Subtotal Smart Label printer $3,418 .00 Software Inventor y management software 1 .00 1590 .00 $1,590 .00 I/C Host interface 1 .00 1495 .00 $1,495 .00 MJD nexus 1 .00 995 .00 $995 .00 Shelf life software 1 .00 1600 .00 $1,600 .00 Label Vista software 1 .00 69 .00 $69 .00 Zebra Designer Pro software 1 .00 200 .00 $200 .00 Sub total software $5,949 .00 Other Service Pack for 5 years for MC3090 Z reader (Motorola) 1 .00 613.2 0 $613 .00 Subtotal other $613 .00 Total $14,065 .00

PAGE 159

159 Table 7 2 Yearly costs of RFID tags. Tag Number of units/Year Unit price Total pr ice Caen RFID A927Z In t ransit 14880 .00 25 .00 $372,000 .00 Safety s tock 2232 .00 25 .00 $55,800 .00 Subt otal Caen RFID A927Z 17112 .00 25 .00 $427,800 .00 Avery AD 233 Rolls x 1000 t ags 50 .00 200 .00 $10,000 .00 Subt otal Avery AD 233 $10,000 00 T otal $437,800 .00 Table 7 3 Cost of the project during the first year of operation. Item Number of units Unit price Total Work station hardware + s oftware 50 .00 $14,065 .00 $703,268 .00 Backup hardware + s oftware $70,327 .00 Maintenance + t raining $175,817 .00 Tags $437,800 .00 Integration $2,500,000 .00 T otal $3,887,212 .00 Table 7 4 Cost of the project during the second and fourth year of operation. Item Total System replacement $175,817 .00 Maintenance + t raining $175, 817 .00 Tags $437,800 .00 Total $789,434 .00

PAGE 160

160 Table 7 5 Cost of the project during the third and fifth year of operation. Item Number of units Unit price Total System replacement $175,817 .00 Main tenance + t raining $175,817 .00 Replace ment of reader batteries 100 .00 78.4 0 $7,840 .00 Replacement of printer batteries 50 .00 160 .00 $8,000 .00 Tags $437,800 .00 Total $805,274 .00 Table 7 6 Yearly costs of the project. Year Cost Year 1 $3,887,212 .00 Year 2 $789,434 .00 Year 3 $ 805,274 .00 Year 4 $789,434 .00 Year 5 $805,274 .00 T otal $7,076,628 .00 Table 7 7 Distribution of U.S. troops in the zone of conflict studied. Country Number of U.S. troops Afghanistan 100,000 .00 Iraq 50,000 .00 Total 150,000 .00 Table 7 8 Estimated number of FSR eaten daily in the zone of conflict studied. Country Number of FSR eaten daily Afghanistan 10,000 .00 Iraq 5,000.00 Total 15,000 .00

PAGE 161

161 Table 7 9 Estimated demand for FSRs in the zone of conflict studied. Item Afghanis tan Iraq Total Number of FSR pallets eaten/day 23.26 11.46 34.72 Number of FSR containers eaten/day 0.58 0.29 0.87 Number of FSR pallets eaten/month 707.69 348.56 1,056.25 Number of FSR containers eaten/month 17.69 8.71 26.41 Number of FSR pallets eat en/year 8,491.32 4,182.29 12,673.61 Number of FSR containers eaten/year 212.28 104.56 316.84 FSR containers sent as safety factor/year 31.84 15.68 47.53 Table 7 10. Actual demand for FSRs in the zone of conflict studied. Item Afghanistan Iraq Total Number of FSR containers shipped/month 18 .00 9 .00 27 .00 Number of FSR containers shipped/year 216 .00 108 .00 324 .00 FSR s hipments sent for safety (c ontainers/year) 32 .00 16 .00 48 .00 Total number of FSR containers shipped/year + safety factor of 10% 248 .00 124 .00 372 .00 Table 7 11. Estimations of yearly pallet losses. Item Afghanistan Iraq Total Number of pallets/year 9920.00 4960.00 14880.00 Number of mishandled pallets/year (18.75% of total pallets) 1860 .00 930 .00 2790 .00 Number of mishandle d pallets/year avoided by system (50% of 18.75% of total pallets) 930 .00 465 .00 1395.00 Table 7 12. Maritime shipping costs for FSR loads according to its destination. Item Cost Cost of s hipping to Iraq/c ontainer $5,398.00 Cost of s hipping to Pakista n (en route to Afghanistan )/c ontainer $4,050.30 Cost of s hipping to Iraq/p allet $134.95 Cost of s hipping to Pakistan (en route to Afghanistan )/p allet $101.26

PAGE 162

162 Table 7 13. Total cost of a lost pallet according to its final destination. Item Cost P allet l osses when s hipped to Iraq FSRs $4,968.00 Cost of s hipping to Iraq/p allet $134.95 Total l oss per pallet when s hipped to Iraq $5,102.95 Pallet losses when s hipped to Pakistan (en route to Afghanistan) FSRs $4,968.00 Cost of s hipping to Pak i stan (en route to Afghanistan)/p allet $101.26 Total loss per pallet when shipped to Pakistan (en route to Afghanistan) $5,069.26 Table 7 14. Amount of product sent as yearly emergency shipments. Item Afghanistan Iraq Total Number of mishandled palle ts/year (18.75%) 1860 .00 930 .00 2790 .00 FSR shipments sent for safety (pallets/year) 1280 .00 640 .00 1920 .00 Emergency shipments (pallets/year) 580 .00 290 .00 870 .00 Emergency shipments (containers/year) 14.5 0 7.25 21.75 Actual e mergency shipments (conta iners/year) 15 .00 8 .00 23 .00 Table 7 15. Amount of money destined to yearly emergency shipments. Item Unit price Number of units Total Emergency shipment pallets sent to Afghanistan $5,069 .00 600 .00 $3,041,555 .00 Emergency shipment pallets sent to Ir aq $5,103 .00 320 .00 $1,632,944 .00 Total $4,674,499 .00 Table 7 16. Yearly savings created by the proposed system. Pallet loss avoided /year Savings per unit Number of units saved Total Savings From p allets sent to Afghanistan $5,069 .00 930 .00 $ 4,714,409 .00 From p allets sent to Iraq $5,103 .00 465 .00 $2,372,872 .00 From emergency shipments $4,674,499 .00 Total s avings /year $ 11 761 779.72

PAGE 163

163 CHAPTER 8 CONCLUSIONS The first objective in this research was to study the use of RFID in tempera ture monitoring by comparing the performance of RFID temperature tags versus conventional temperature tracking methods in a food supply chain. The supply chain chosen for such purpose was that of crownless pineapples traveling from Costa Rica to the US. Th e results of the study indicate that, although analogous with respect to accuracy in the temperature measurements, RFID systems are superior as temperature tracking method to conventional methods such as HOBO sensors. The main reasons behind this are the f act that RFID systems allow quick instrumentation and data recovery, and offer the possibility of accessing the sensor program and data at any point of the supply chain without line of sight. The second objective of this work was to compare the utilization of RFID temperature tags with probe versus RFID temperature tags without probes along a food supply chain. The same supply chain studied for the first objective was chosen for this comparison. After completing the trial, it was established that RFID tags with probe were important for determining the efficiency of precooling operations and for tracking low temperature abuse (LTA) during transportation in sea containers and refrigerated storage. RFID tags without probe were found most useful during high tem perature abuse (HTA) monitoring in transportation in sea containers and refrigerated storage. In other words, RFID tags with probe are important to monitor the critical points of the load, which are the areas of the load where temperature abuse is most lik ely to occur inside the product; while RFID tags without probes are relevant during monitoring of ambient conditions during storage and transportation. Finally, recommendations were given with respect to the creation of a RFID sensor with a probe, able to record both ambient and probed temperatures simultaneously.

PAGE 164

164 The third objective of this research was to determine the level of instrumentation (amount of sensors and the best locations for their placement) of an efficient RFID temperature tracking system a t the pallet and cargo level in three different scenarios for food supply chains. These were: For products prone to low and high temperature abuse (Case 1); for products susceptible to high temperature abuse (Case 2); and for shelf stable products (Case 3) Following are the conclusions for each one of the cases studied. For Products Prone to Low and High Temperature Abuse (Case 1) In this case, the supply chain of crownless pineapples from Costa Rica to the US transported by sea containers was also used. Proper monitoring was obtained using one sensor per pallet, if necessary, according to the location of the pallet during forced air cooling and inside the container. It is suggested to place ambient probeless tags in the top of the pallets placed in the ba ck of the container as a way to monitor High Temperature Abuse (HTA). For additional HTA watch, tags with probe were recommended for the center of the pallets prone to suffer insufficient precooling. Finally, tags with probe were also proposed for monitori ng Low Temperature Abuse (LTA) in the bottom of the pallets of RPCs in the front of the container. For Products Susceptible to High Temperature Abuse (Case 2) Relevant temperature monitoring using a single RFID tag per pallet for loads of certain varieties of produce such as apples, oranges, pomegranates, passion fruit and tangerines was achieved. In order to determine the total amount of HTA suffered by the entirety of the pallet, a probeless RFID sensor had to be placed in a thermally relevant position or Point of Relevance (PoR) able to describe the temperatures of 85.00 % of the temperatures of the pallet. The sensor was placed after the first layer of product in one of the bottom corners of one of the 1.02 m smaller pallet sides, although the system can work in any of these sides.

PAGE 165

165 For Shelf Stable Products (Case 3) Two RFID temperature tracking systems for First Strike Rations (FSR) were developed for combat feeding logistics. Relevant temperatures were obtained by monitoring 85 .00 % of the temperatures o f the pallet with only one probeless RFID tag placed in a PoR. In the fixed RFID system the tag was placed near to the surface of the second row (from top to bottom) of boxes in one of the 1.22 m sides of the pallet (although the system can also work in th e other 1.22 m side). In the handheld system the tag was placed on the surface of the pallets wooden structure; yet its initial readings were transformed into the temperatures obtained at the PoR monitored by the fixed system with the help of a software t ool. The final objective of this research was to create the business case for a RFID temperature tracking system when combined with shelf life prediction software by performing an economic analysis in one of the systems previously designed In order to do so, a load management system for First Strike Rations that uses a software tool able to provide recommendations on the actions required with respect to the load and acceptability scores based on shelf life prediction models and the temperature history of the shipment was developed. The return on investment (ROI) of a scenario created from real life conditions was estimated to be 719.49% This proves that for this particular deployment the potential economic benefits of th is technology surpass greatly its costs ; mak i ng it a worthy investment. And most importantly, it also shows the fact that RFID temperature tracking systems can be the foundation of an important tool for value generation in food supply chains

PAGE 166

166 APPENDIX A ANALYSIS OF THE VARI ABILIT Y AND COLD CHAIN PER FORMANCE FOR CROWNLESS PINEAPPLE WITH RESPECT TO TRAN SPORTATION METHODS, LOCATION WITHIN THE CARGO, AND PACKAGING Introduction Pineapple is the second most important tropical fruit in the world. Around 21 million tons of pineapples were produced worldwide in 2007 (FAO, 2009) ; 12% of these were devoted exclusively to the international fresh market, a business of more than 2 billion USD Central America (i.e., Costa Rica and Panama) and the Philippines are the biggest sources of exported f ruit; while the US is the top importer, having reached a level of consumption in 2007 of 2.16 kg of fresh pineapple per capita (USDA, 2010). International commerce of fresh pineapples requires highly efficient temperature controlled supply chains. Due to the sensitivity of this product to chilling injury, temperatures must remain within a specific range at all times, avoiding exposures to temperatures lower than the products threshold (Abdullah et al., 2000; Acedo et al., 2004). However, exposure to highe r temperatures than those recommended for pineapple storage (7C to 12C ; Paull, 1993) will accelerate the senescence and decay rates of the load (Mohammed and Wickham, 1995). Poor temperature management during pineapple shipments will then result in postharvest losses and in poor product quality; which generates lower customer satisfaction and impacts the produce companies with economic losses and lack of public credibility (Nunes, 2008; Machado et al., 2009; Nunes et al., 2009). During recent years, the fresh pineapple trade has been positively affected by the use of the fruit by the fresh -cut industry, which can import either pineapples with crown (i.e., bracts on the fruit) or crownless (Gonzalez -Aguilar et al., 2004). Crownless pineapples facilitate fr esh -cut operations; however, their postharvest management becomes more challenging, since the

PAGE 167

167 wounding left by crown removal increases metabolic activities and promotes senescence and decay in the fruit. Given the direct relationship between temperature an d these processes, adequate temperature management in crownless pineapples is of the uttermost importance. Proper temperature management for pineapples begins with prompt transport of harvested fruit from the field to the packing and cooling facility. Pin eapples are usually packed into cartons or other containers and the containers stacked on pallets at ambient temperature conditions. The pallets of pineapples are then promptly cooled using forced air cooling to near the established temperature for storage and transportation. For forced air cooling to be efficient, the pineapple containers must be properly designed to facilitate the flow of refrigerated air through vent holes from one side of a pallet to the opposite side. Cooled pallets of pineapples are s ubsequently transferred to refrigerated rooms to complete the cooling process to reach a final temperature for transport and for temporary storage prior to transport. Once an optimal set point temperature for transportation has been established, achieving proper temperature management depends on several factors. The air flow pattern and refrigeration capacity of sea containers and in truck trailers is very different. Sea containers have greater refrigeration capacity than trailers and utilize bottom air delivery, which entails delivery of refrigerated air under the floor of the container, from which it moves in a vertical direction through the cargo. Refrigerated trailers utilize top air delivery, in which air moves from the front, top part of the trailer al ong the top of the cargo to the back, door area. The air then returns to the reefer unit through and underneath the cargo. Furthermore, the airflow pattern in the holds of refrigerated cargo vessels is from bottom to top. Packaging design and materials can hasten or hinder the heat transfer process, and dramatically affect the temperature distribution inside the pallets and throughout the cargo environments (Smale, 2004). For pineapples, this means that

PAGE 168

168 packaging must be designed to facilitate both horizont al air flow (for forced air cooling and refrigerated trailers) and vertical air flow (for sea containers and refrigerated cargo vessel holds). Stowage patterns must also accommodate the air flow design of the transport equipment being used. T he existence o f spatial and time-based temperature variability both inside pallets of produce and in the cargo areas that harbored them along their supply chains has been described in previous research reports (Billing et al., 1993; Billing et al., 1995; Amos, 2001; T anner and Amos, 2003; Moureh and Flick, 2004; Punt and Huysamer, 2005; Rodriguez Bermejo et al., 2007) Since deviations like these could potentially increase the risk of temperature abuse on the load, any attempt to optimiz e a produce cold chain will req uire that these factors and their possible interactions be taken into account. The objectives of this research were to determine if there are significant differences in the temperature profiles of pineapples handled using different combinations of packagin g and transportation methods or among different locations in the cargo; and to estimate the risk of exposure to high temperature abuse (HTA) and the development of chilling injury due to exposure to low temperature abuse (LTA) in the fruit among these dif ferent parameters, based on duration of exposure to deleterious temperatures and thermal history. It is expected that, as a result of this study, the produce industry will be able to recognize which one of the package/transportation method combinations synergize when trying to design an efficient crownless pineapple supply chain that will secure high -quality produce. Materials and Methods Experimental D esign At a pineapple packinghouse in the Pacific coast region of Costa Rica, two samples of primary packag es of crownless pineapples were instrumented with temperature sensors before palletizing. Both samples consisted of 9 primary packages containing six MD 2 pineapples

PAGE 169

169 each. These primary packages were corrugated cartons (trays; 0.58 m x 0.39 m x 0.14 m) a nd reusable plastic containers (RPC; 0.59 m x 0.39 m x 0.20 m). In each package, one fruit was instrumented with a probe attached to a HOBO temperature sensor and datalogger (HOBO Series H8; TMC6 HA and TMC6 HD probes, Onset Computer Corporation, MA) in o rder to obtain their mass average temperature at a sampling rate of one measurement per 5 min. All the probes were installed in the center of the fruit, at a depth of 0.04 m. The packages were then placed in three different regions of their corresponding pallet: the upper corner, the core at the central layer (seventh from the bottom up) and the lower corner (Figure A 1). The location of the instrumented pineapples depended on the position of the primary package within the pallet. For the top and the bottom packages, the pineapple corresponding to the corner was instrumented ; while for the package placed at the core, it was the fifth pineapple from left to right. The pallets were forced air cooled using 6.5C air to an average pulp temperature of 10.5C. The pallets were kept for 9 hours in refrigerated storage at 1 1.5 C after forced air cooling was completed. Then, three pallets of each set were loaded into a 12.19 m refrigerated sea container or into a 16.15 m refrigerated truck trailer The pallets were pl aced in three specific locations (front, middle and back) in each one of the cargo areas (Figure A -2). Both of these environments were arranged with one row of pineapples packed in corrugated cartons and one row with a mixed load of fruit packed in RPCs ( the pallets being studied) and corrugated cartons The sea container and truck trailer temperature set points were programmed to 7.5C. Once filled, the sea container and the trailer were taken to the Port of Limon on the Atlantic coast of Costa Rica by la nd. Two days after their arrival at the port, the sea container and the pallets from the truck trailer were loaded onto a cargo vessel The pallets transported by truck were placed in pairs containing one of each set (corrugated cartons and RPCs), in two d ifferent holds (1 and 2);

PAGE 170

170 and inside one of the holds (2), on two decks (A and B). The temperature set point of the holds was also established at 7.5 C. A schematic of the loading pattern can be seen in Figure A 3. The sea container remained sealed and af ter the holds were closed it was placed on the deck of the vessel Once fully loaded, the vessel was sent to Port Manatee (FL), where the sensors were retrieved 3 days later. Statistical A nalysis A Generalized Linear Model (GLM) procedure with a Gamma fa mily was used in order to determine the effects of time, packaging, location within the cargo, and transportation method employed on the temperature within the pineapple loads (Freund and Wilson, 2003). The same statistical procedure was also used to deter mine the equality of means of the temperatures obtained with the different combinations of transportation method, location in the cargo area, and R 2.7.2 (The R Foundation for Statistical Computing., Wiedner Hauptstrabe 810/1071 1040 Vienna, Austria). High Temperature Abuse (HTA) and Low Temperature Abuse (LTA) Risk Estimation Risk levels were defined as the following percentages of positives from the total numbe r of temperature readings considered for the specific thermal abuse studied: High risk at values equal to or more than 70% positives; Medium risk at values from 40% to less than 70% positives; and Low risk at values of less than 40% positives. High temperature a buse (HTA) a nalysis Three intensity levels of exposure to high temperature abuse (HTA) were defined. These were established based on the temperatures at which the respiration rate of a pineapple stored at 7 C (the optimal storage temperature for this variety of pineapple is 7.23 C or 45 F; Saenz and DAlolio, 2007) would approximately double (HTA No. 1: >12 C), triple (HTA No. 2: > 15 C),

PAGE 171

171 or quadruple (HTA No. 3: > 17 C) (Dull et al., 1967). HTA No. 1 was considered the least intense exp osure while HTA No.3 was considered the most intense exposure The values of HTA given represent the percentage of time when the fruit were exposed to these temperatures with respect to the total duration of the storage and transport operations analyzed. Low t emperature a buse (LTA) a nalysis Only one level of exposure to low temperature abuse (LTA) were defined This LTA level was created by adding the 0.5 C error margin of the HOBO sensors to the chilling injury threshold temperature for pineapple ( i.e ., 7.23 C). As in the HTA analysis, the values of LTA given represent the amount of time when the fruit were exposed to LTA temperatures with respect to the total duration of the storage and transport operations as a percentage. Results and Discussion Te mperature P rofiles of P ineapples using D ifferent C ombinations of P ackaging, Transportation M ethods and Locations in the C argo Environment ( S ea C ontainer and Truck Trailer). Significant differences were found in the majority of the temperature profiles wit h regards to transportation method, position within the cargo environment, packaging, and the resulting combinations. However, at the bottom of the pallets no significant differences were obtained with regards to the transportation method employed, and the placement inside the container (Tables A 1 and A 2). In addition, no significant differences were obtained due to packaging at the top of the pallets loaded in the open holds (Table A 3). Overall, the transportation method (truck trailer or sea container) the position inside the cargo environment analyzed (back, middle and front) and the type of packaging (corrugated or RPC) had a significant effect on the temperature profiles measured during handling and shipping of crownless pineapple from the Pacific coast of Costa Rica to the U.S..

PAGE 172

172 In addition, there was significant variability in the temperature profiles of crownless pineapples shipped from the Pacific coast of Costa Rica to the U.S.. The lack of consistency in the environmental and physical conditions during handling could result in the delivery of loads of pineapple s with variable quality, which could not only limit the usage of the fruit, but also the shelf life and the final quality of the fresh -cut product s Prior research performed by Billing et al (1993, 1995), Smale (2004), and Punt and Huysamer (2005) already reported the existence of temperature gradients within loads of produce during transportation in refrigerated sea containers and vessels. While these studies aimed to quantify the magnitu de of the gradients, they did not provide an overall statistical comparison of the temperature management throughout the whole supply chain. The lack of significant differences in temperature profiles at the bottom layer of the pallets with regards to the transportation method employed could be explained by the fact that both sea containers and open holds use bottom air delivery systems. However, it would also indicate that the effect of the temperature management during the land transportation period, whe n truck trailer s with top air delivery systems were used, was minimal. Furthermore, it suggests that the thermal effect of the exposure to ambient conditions of the same set of pallets during the process of loading and unloading the vessel was not a signif icant factor in determining the temperature profiles The lack of significant differences in temperature profiles at the bottom of the pallets with respect to placement inside the container could suggest the existence of uniform airflow along the length o f the container Even though this observation argues against the occurrence of any air short -circuiting phenomenon inside the container; in order to obtain a conclusion on this matter,

PAGE 173

173 additional analysis of the temperature profiles of the fruit located in the bottom layer of the pallets transported in sea containers is discussed later. The lack of significant differences in temperature profiles at the top of the pallets transported in open holds when comparing type of packaging could be explained by the fa ct that these cartons and RPCs are heavily affected by solar radiation when they are loaded o nto the vessel Therefore, the bulk of the heat transfer during this period of time could have taken place through the top surface of the cartons and RPCs not thr ough the horizontal vents; the shape, size and amount of these being one of the key dissimilarities between the two types of packages. However, further analysis of thermal abuse incidence, duration, and intensity is required in order to determine whether t his was indeed occurring. Similar statistical analysis procedures ha ve been successfully used by Amador et al. (2009) to compare the performance of sensors during temperature tracking of pineapple shipments. Even though the procedure used constitutes an ef fective tool to determine the statistical variability of the temperature profiles of pineapples along this supply chain, it fails to provide the information needed to detect the occurrence and timing of temperature abuses that could lead to quality deterio ration of the fruit. And so, this analysis can not solely lead to process improvement in the cold chain. High Temperature Abuse (HTA) and Low Temperature Abuse (LTA) Risk Estimation Following are the results of the estimations of thermal abuse risk corresp onding to the trial. High t emperature a buse (HTA) a nalysis The results indicate that exposure to HTA occurred more often in the load in the truck trailer /open hold combination; but lasted longer when using the sea container (Tables A 4 to A 11).

PAGE 174

174 Refriger ated t ruck t railer / o pen h olds Table A 4 displays the percentages of time that the instrumented fruit spent over the limits of the different levels of HTA during the pineapple cold chain, while Tables A 5 and A 6 present a summary of the degree of HTA risks existing in the locations of the instrumented pallets, and according to the packaging type employed. Overall there was a high risk of exposure to HTA No. 1 (T is quite understandable, taking into account that the top area of the pallets is exposed to direct solar radiation during the processes of loading and unloading the vessel Furthermore, since the refrigeration units in the holds did not start until the decks surface area had been completely filled; the pineapples were exposed to direct sunlight without any forced convection cooling for 1.5 to 2.5 hours. Considering that the loading/unloading of the vessel took place all day long, this allowed the top layers of fruit to reach even higher temperatures when the sun was at its peak. The top layers of the pallets placed on Deck 1A constitute a good example of this situation. That is, the maximum temperatures for these pallets were recorded when loaded onto the vessel around noon and surpassed any of the maximum temperatures registered in the other pallets, which were exposed to am bient conditions at sunset and night time (Table A 7). The exposure to midday temperatures and the resulting heat accumulation in the fruit transported in this particular deck most likely generated the higher HTA No. 3 incidence in this layer (Table A 4). It is also worthy of mention that there was a high risk of HTA No. 2 and medium risk of HTA No. 3 in the top layers of the pallets, making this particular area most likely to be responsible for deterioration of fruit quality (Table A 4). In addition, ther e was also high risk of HTA No. 1 in the center and the bottom of the pallet s which may be related to inefficient forced air cooling operations and exposure to ambient temperatures, respectively.

PAGE 175

175 When comparing the effect of packaging type on the incidence, duration and intensity of HTA and, according to the position of the fruit inside the pallet (Tables A 4, A 5 and A 7), there were two main phenomena taking place. The first one was on the top layer, where the incidences of HTA No. 2 and No. 3 were con siderably higher in RPCs; even though the amount of HTA No. 1 was comparable for both kinds of packaging. Compared to the corrugated cartons RPCs allow ed greater air exchange due to their vast surface area devoted to vents; which is quite desirable for si tuations such as forced air cooling, cold storage or transportation under refrigeration. However, these results indicate that this feature may have been a liability during the loading/unloading process, because it favored the intrusion of warmer air into t he pallet s So, the fruit packed in RPCs in the top layer were exposed to higher temperatures than the fruit packed in corrugated cartons located in the same area because the heat transfer process was more extensive in the RPCs taking place not only throu gh the surface of the primary packages, but also through its vents. The second phenomenon took place in the core of the pallet, where HTA No.1 was more predominant in corrugated cartons than in RPCs. T his pattern could be explained by the fact that corruga ted cartons restrict more airflow than RPCs and hinder pre -cooling operations. This is mostly due to the differences between them in the shape, size, and quantity of the vents for horizontal air flow Hence, since the refrigeration equipment used during sh ipping was designed to maintain temperature instead of reducing it, there was heat accumulation in the load throughout the whole journey. Table A 7 shows the temperatures obtained after pre -cooling and the amount of heat removed from each one of the positi ons within the analyzed pallets. The target final forced air cooling temperature was 10.5 C. Therefore, in all cases, the fruit packed in RPCs achieved lower final pre -cooling temperatures and were closer to the target temperature

PAGE 176

176 than the fruit in corrug ated cartons This fact is in agreement with previous research on the influence of packaging in cooling operations for pineapples, papaya s and mango e s (Chonhenchob and Singh, 2003a; Chonhenchob and Singh, 2003b; Chonhenchob et al., 2008). A good example o f the importance of adequate pre -cooling was the fruit in corrugated cartons in the center of the pallet that were shipped in the Middle/2A combination (Table A 7). In this case, the pineapple s were exposed to temperatures that would presumably have on av erage, doubled the respiratory metabolism during the entirety of the voyage. Furthermore, the level of heat accumulation was such that during more than 25% of the shipping time, the fruit reached temperatures that would presumably have tripled the respirat ion rate, accelerating the pineapple fruit metabolism and its quality deterioration (Mohamed and Wickham, 1995). The effect of position within the cargo environment in HTA was not entirely clear when using the current quantitative analysis as a basis. A de tailed review of the temperature profiles of the pallets allowed the determination of the thermal behavior of the load in the two stages studied (data not shown). In the first stage during land transportation by refrigerated truck, the temperatures obtain ed along the length of the cargo area were strongly correlated with the expected trend given the top air delivery system airflow pattern. Regardless of the packaging used, the pallets at the front of the trailer experienced colder temperatures than the pal lets placed in the middle and in the back of the trailer Even though, in the case of RPCs the temperatures recorded in middle and back positions inside the refrigerated trailer had similar values, for corrugated cartons the temperature differential betwe en positions was approximately 2 C higher in the back pallet than in the middle pallet, at almost all times. Earlier studies (Meffert and Van Nieuwenhuizen, 1973; Gogs and Yavuzturk, 1974; Lenker et al., 1985; Bennahmias and Labonne, 1993; Le Blanc et a l., 1994; Finn and Brennan, 2003; and Moureh et al., 2009) have

PAGE 177

177 reported the existence of hot spots in the rear of refrigerated trailers created by f uneven air distribution. In addition, in the present study the temperature profiles for the bottom laye rs for both kinds of packaging showed that the product placed in the front pallet was colder than that of product placed in the middle and back positions. However, because of its proximity to the air return area, the front pallet location was expected to have the highest temperatures under uniform airflow conditions. So, this phenomenon could only indicate the existence of air short -circuiting around the front of the trailer which favored the formation of the hot spots found at the rear (Moureh et al., 2 002; Finn and Brennan, 2003; Tapsoba et al., 2006).This was more evident for RPCs, for which the air encounters less resistance to its flow than for corrugated cartons In this particular case, temperature differentials between the bottom layers of the pal let in front with respect to the temperature differentials in the middle and the back of the refrigerated truck reached more than 5 C. Finally, even though there was uneven air and temperature distribution along the length of the trailer their influence on the incidence of HTA No. 1 in the load was small, since only the fruit in the bottom of the middle pallet packed in RPCs were exposed to HTA No. 1 during the land stage of this transportation process (data not shown). As previously discussed, the effec t of carrying the pineapples in the hold of the cargo vessel on the exposure of the pineapple s to HTA was more related to the time of loading onto the vessel and to the amount of heat accumulated by the produce during this process, than to the position in the hold itself. Once the cargo had lost the heat that had accumulated during loading, the temperatures in all the pallets were stable for the remainder of the voyage In the case of H old 2, the pallets on D ecks A and B displayed the pattern expected given the set up of the equipment (Smale, 2004). In this pattern the product in the lower deck (B) always presented lower temperatures than its counterpart in the upper deck (A) (data not shown) s ince the same

PAGE 178

178 refrigeration unit was used for the entire hold, a long with a bottom air delivery system that connected both decks. During this period of time, temperatures in H old 1A were within the temperature ranges of both D eck A and Deck B in H old 2. Unlike the first part of the marine transportation process, this s tage did not have a major impact on the exposure of the load to HTA. Sea c ontainer In general, there was a medium risk of HTA No. 1 in the fruit transported by sea container (Tables A 8, A 9 and A 10). However, this risk varied greatly depending on the location of the produce analyzed. Thus, the top and central layers showed a high HTA risk, while the bottom layers showed no risk (Table A 9). With respect to HTA No. 2, there was an overall low HTA risk in the cargo, but a high HTA risk in the top layers Finally, there was a low risk of HTA No. 3 in the load as a whole along with a high HTA preponderance in the top layers. The existence of a propensity toward fruit HTA exposure was not only related to the airflow pattern inside the container, which wil l be discussed later, but also due to the fact that the pineapple fruit endured three major episodes during which the container was deprived of the energy necessary to run its refrigeration unit during the transportation process (Figure A 4). Due to loss o f the feed of cold air, the container was more vulnerable to heat accumulation due to the respiratory heat of the cargo, exposure to solar radiation, leaks in the walls and doors of the container, and to convection heat transfer from the environment. The first of these episodes lasted 7 hours and occurred when the container was being transported to the port, and was crossing a mountainous area with low ambient temperatures. This was most likely and intentional act performed by the transportation company in order to reduce their fuel costs. As a consequence, pineapples packed in both kinds of packaging suffered from HTA No. 1 in the top and the central layers of the pallets (Table A 10). Moreover, these top

PAGE 179

179 layers were also exposed to HTA No. 2 and in some c ases to HTA No. 3. The second episode took place for 6 hours, when the container was in the process of being loaded onto the vessel Then, the top layers of the pallets with both type s of packaging were subjected to HTA No. 1. Lastly, the third event occurred in the middle of the voyage of the vessel and lasted about 20 hours. Once again, regardless of the primary package (corrugated or RPC), the product placed in the top layer faced HTA No. 1. Overall, the top layers were exposed to thermal abuse between 24.4% and 42.7% of the total duration of the journey while the center was exposed to thermal abuse for a maximum of 29.7% of the total journey (Table A 8). Finally, also worthy of attention was the fact that regardless of the kind of packaging used, no HT A was registered in the bottom layer of the pallets. With respect to the effect of packaging on HTA, there were also two situations that took place based on the area of the pallet being studied. Similar to the analysis of loads transported in truck trailer s and cargo holds, in the top layers, the use of RPCs favored an increase of fruit temperature, subjecting it to HTA No. 3. Nevertheless, and unlike the previous case, the heating of the pineapples took place when they were protected from the elements by t he container, during the periods of time when the refrigeration equipment was turned off. Under these conditions, the forced convection cooling stopped, and the warm air rose to the top of the container due buoyancy forces. Then, that warm air came in cont act with some of the fruit in the top layer through the vents of the RPCs while it remained isolated from the product in corrugated cartons precisely because of their lack of vents. In addition, the effect of heat accumulation inside corrugated cartons in the central layers of the pallets was also present and it was seen even more clearly than for the truck trailer / open hold combination. Once again, a review of the temperatures obtained after pre -cooling (Table A 11) confirmed that the pineapples packed in

PAGE 180

180 RPCs were properly pre -cooled, while the pineapples packed in corrugated cartons were not. Furthermore, the center of both kinds of pallets seemed to be the area where the temperature differences between pallets reached its highest values. As a result the temperature differential inside the corrugated cartons was between 2 and 3 C higher than the target temperature of the pre cooling operation. This situation was then magnified by the lack of constant ventilation inside the container, exposing the core pr oduct of the pallet located at its rear to HTA No. 1 for over 29% of the time of travel (Table A 8). In order to determine the effect of the position within the container on temperature management the temperature profiles of the pallets were examined. Re gardless of the kind of packaging, the lowest temperatures in the top layer of the pallets when the refrigeration unit was running were obtained for the pallets in the front of the container. Given that this area was the closest to the air -return for the b ottom air delivery system, it was expected to be the warmest spot when adequate airflow exists. Other temperature profiles were then reviewed in order to detect the possibility of air short -circuiting. A typical case of this anomaly would also involve occu rrence of higher temperatures in the back of the container, which was the case in almost every one of the locations measured (Figure A 5). After further analysis, it was determined that this phenomenon did not really affect the incidence of HTA No. 1 in the load but, once this HTA had originated due to the disconnection of the refrigeration unit, the air short -circuiting prolonged the HTA duration in the rear of the container once the unit was turned on. Low t emperature a buse (LTA) a nalysis Exposure to LT A was found to occur more often in the truck trailer /open hold transportation but it lasted longer in the sea container, surpassing in some locations more than 60% of the traveling time. Meanwhile, the truck trailer /open hold combination had a more

PAGE 181

181 inten se exposure to LTA, since lower temperature values were recorded compared to those registered in the sea container. Refrigerated t ruck t railer / o pen h olds Overall, there was a high risk of exposure to LTA in the refrigerated truck trailer / open hold combinat ion (Tables A 12 and A 13). This exposure was mostly concentrated in the bottom and central layers of the pallet s also high risk areas, while the top layers only presented medium risk of LTA. When analyzing the effect of packaging on LTA incidence and duration in the Front/1A pallets, there was not much of a difference between the use of corrugated cartons and RPCs. However, an examination of the other two positions showed that the pineapples packed in RPCs were subjected to LTA for longer periods of time especially in the center of the pallet. Thus, it seems that the number of vents in RPCs increased the level of penetration of low temperature air in the pallets (Table A 14). The uniformity of the incidence and duration of LTA in the Frontl/1A pallets w as due to the combined effect of short -circuiting in the truck trailer by which the majority of top -delivered cold air went through the pallets placed in the front of the trailer with high speeds and the bottom air delivery system in the vessel, which aff ected mostly the bottom layers. Furthermore, the exposure to LTA of the pallets placed in the center and back of the trailer took only place when these pallets were inside the refrigerated holds. Therefore, the fruit located in the top and central layers o f the front pallets in the trailer were more prone to exposure to LTA during land transport while similar pallets in the other positions and all the packages placed in the bottom layers of pallets were subject to LTA during sea transportation. Overall, ex posure to LTA was correlated with the position of the cargo in the truck trailer and the airflow pattern in it. Yet, LTA

PAGE 182

182 also had a strong relationship with the location of the fruit in the pallet, especially when the fruit were subjected to the bottom air delivery system in the vessel. The magnitude of the development of chilling injury (CI) symptoms in pineapples is strongly dependent on the combination of the temperature (i.e., the lower the temperature the more severe the symptoms) and the duration of the exposure (i.e., the longer the exposure the more severe the symptoms). In addition, these symptoms often and sometimes only become apparent after transferring the fruit from the chilling temperatures to warmer temperatures (Saltveit and Morris, 1990) However, if the fruit are subjected to intermittent warming (removed from chilling temperatures before irreversible harmful changes have started, and then exposed to warmer temperatures), its resistance to CI can actually increase when later returned to chilling temperatures (Akamine et al., 1975; Saltveit and Morris, 1990). In the present study, some of the pineapple fruit transported through the refrigerated truck trailer /open hold combination were first exposed to chilling temperatures (LTA) while in the trailer and then to ambient temperatures when loaded onto the vessel. They were later on exposed again to chilling temperatures when inside the open hold of the vessel Further research is recommended in order to determine the possibility of the exist ence of a positive intermittent warming effect on the fruit when using this transportation method, in case the exposure to chilling temperatures at the truck trailer level does not generate irreversible CI. If found, this intermittent warming effect would make the use of refrigerated truck trailer s/open holds and RPCs quite adequate for this particular supply chain. Sea c ontainers Table A 15 shows that the central and bottom layers of the pallets with RPCs were exposed to LTA for prolonged periods of time (approximately 58 and 73% of the total shipping time for the central and bottom layers, respectively) However, while there was no risk of LTA in the top layer of the pallets for the pineapples transported in the container, and their central and

PAGE 183

183 bottom lay ers showed low risk of LTA (Table A 16). In addition, the use of corrugated cartons seemed to exert a shielding effect against LTA whereas the use of RPCs seemed to favor the risk of LTA in the bottom and central layers of the pallets (Table A 17). The e ffect of position within the container also influenced the exposure of pineapple fruit packed in RPCs to LTA. Since the sea container had a bottom air delivery system and there was also air short -circuiting within it, the bottom and central layers of the c ent er and especially the front pallets with RPCs received the majority of the cold airflow. The air traveled more easily through the RPCs compared with the corrugated cartons because the RPCs offered the path of least resistance. Thus, stagnant areas (i.e. no air flow) occurred in the corrugated cartons and at the back of the container. Consequently, in the pallets with corrugated cartons no effect with respect to the position in the cargo area could be detected. So, exposure to LTA in the sea container w as strongly related to the position of the fruit within the pallet, to the position of the pallet in the cargo environment, and also to the airflow pattern characteristic of a mixed load of RPCs and corrugated cartons which promoted air short -circuiting. Unlike in the refrigerated truck trailer /open hold case, transportation in sea containers does not offer the possibility of an intermittent warming effect at any point in the shipping process. Consequently, the continuous exposure of the fruit to LTA coul d possibly cause the development of chilling injury in pineapple s Conclusions Overall, there were significant differences in the temperature profiles obtained from the different transportation method/location in cargo/packaging combinations in the majori ty of cases. Irr egardless of the transportation system used, the areas of the pallet that showed the highest risk of high temperature abuse (HTA) were the top layers followed by the center layers whereas the bottom and cent er layers displayed the highest risk of low temperature abuse (LTA).

PAGE 184

184 Although present, the effect of pallet position inside the cargo area on pineapple thermal handling was mostly relevant in the case of exposure to LTA in sea containers. It was also determined that whenever inappropriat e temperature control existed during transportation corrugated cartons offered more protection from temperature abuse than RPCs due to slower heat transfer with the cartons Moreover, the refrigerated truck trailer /open hold combination increased the inten sity (peaks and dips) of exposure to HTA and LTA with a large volume of product being affected, while transporting the fruit in the sea container increased the duration of the exposure s to HTA and LTA. Thus, the use of corrugated cartons and refrigerated sea containers for handling and transportation of crownless pineapples from Costa Rica to U.S. can be recommended as this combination minimized the amount of product exposed to HTA and LTA. Further work A comprehensive quality study should be completed in order to determine the extent of the effect of the exposure to HTA and LTA documented in this study on pineapple shelf life and overall quality, and to determine whether there is an intermittent warming protection effect with regard to chilling injury when using RPCs in refrigerated truck trailer /open hold transportation. In addition, corrugated carton should be redesigned in order to reduce the HTA exposure due to inefficient pre -cooling that was documented here

PAGE 185

185 Top View Front View Figure A 1. Placement of the HOBO temperature sensors in the experimental pallets. Figure A 2 Position of the experimental pallets in the cargo areas ( 14-m refrigerated truck trailer and 12 -m sea container). Pallets depicted in dark color correspond to th ose containing corrugated cartons ; while the lighter ones represent pallets containing reusable plastic containers (RPCs). Sea Container Door Front Middle Back Refrigerat ed Truck Door Front Middle Back

PAGE 186

186 Figure A 3 Schematic of the loading pattern inside the refrigerated cargo vessel. Table A 1. GL M results for the effects of the transportation method, time step, and packaging when comparing the fruit temperatures (C) recorded at the bottom of the pallets during transportation using refrigerated truck trailer /open hold combination with the temperat ures obtained using a sea container. Estimate Std. e rror t Value P (Intercept) 8.787e 02 1.379e 03 63.742 <2e 16 Method 2.812e 04 8.133e 04 0.346 0.73 Time s tep 3.265e 05 1.245e 06 26.229 <2e 16 Packaging 5.417e 03 4.304e 04 12.587 <2e 16 Method: T ime s tep 1.086e 05 8.156e 07 13.315 <2e 16 Hold 1 Hold 2 Deck A Deck B Ships Deck Bow Center

PAGE 187

187 Table A 2 GLM results for the effects of position, time step and packaging when comparing the fruit temperatures ( C) recorded at the bottom of the pallets during transportation in the three positions (bac k, middle and front) inside the sea container. Estimate Std. e rror t Value P (Intercept) 8.765e 02 1.157e 03 75.782 <2e 16 Position 1.705e 05 5.466e 04 0.031 0.9751 Time s tep 2.366e 05 1.086e 06 21.789 <2e 16 Packaging 5.395e 03 4.133e 04 13.054 <2e 16 Position: Time s tep 1.019e 06 5.555e 07 1.834 0.0667 Table A 3. GLM results for the effects of the hold, time step and packaging when comparing the fruit temperatures (C) recorded at the top of the pallet s during transportation in the holds of the refrigerated vessel. Estimate Std. e rror t Value P (Intercept) 1.104e 01 1.781e 03 61.958 <2e 16 Hold 1.057e 02 6.990e 04 15.121 <2e 16 Time s tep 2.289e 05 1.417e 06 16.156 <2e 16 Packag ing 9.425e 05 5.735e 04 0.164 0.87 Hold: Time s tep 1.193e 05 6.571e 07 18.159 <2e 16

PAGE 188

188 Table A 4. Percentage (%) of transportation time during which pineapple s were exposed to different high temperature abuse (HTA) scenarios whe n using the refrigerated truck trailer / open hold combination. Front/1A Middle/2A Back/2B Top Center Bottom Top Center Bottom Top Center Bottom Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC HTA No.1 (T 12C) 19.9 20.3 3.5 4.3 7.9 24.1 26.3 100.0 6.2 DS 20.6 28.2 22.1 11.8 5.2 4.2 HTA No.2 (T 7.9 9.5 3.3 8.0 26.8 DS 0.6 10.7 HTA No.3 (T 4.5 6.4 2.3 DS DS: Defecti ve sensor Table A 5. Risk of fruit exposure to different high temperature abuse (HTA) scenarios according to the position of the fruit in the pallet when using the refrigerated truck trailer / open hold combination. O verall T op C enter B ottom HTA No.1 (T H H H H HTA No.2 (T M H L N HTA No.3 (T L M N N H = High, M = Medium, L = Low, and N = None.

PAGE 189

189 Table A 6. Risk of fruit exposure to different high temperature abuse ( HTA ) scenarios according to the packaging used an d the position of the fruit in the pallet when using the refrigerated truck trailer / open hold combination. Overall Top Center Bottom Corr RPC Corr RPC Corr RPC Corr RPC At HTA No.1 (T H H H H H M M M At HTA No.2 (T M L H H L N N N At HTA No.3 (T L L L M N N N N H = High, M = Medium, L = Low, and N = None. Table A 7. T emperatures recorded in the experimental pallets at different stages of the supply chain when using the refrigerated truck trailer / open hold combin ation. Front/1A Middle/2A Back/2B Top Center Bottom Top Center Bottom Top Center Bottom Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Temperature at beginning of f orced a ir c ooling (C) 27.6 27.91 27.06 28.7 27.06 2 8.9 27.91 27.6 26.8 26.52 D S 26.88 27.91 26.52 26.34 26.7 26.73 28.33 Temperature at the end of f orced a ir c ooling (C) 14.15* 10.21 14.99* 8.58 13.14* 9.77 14.09* 10.11 14.8* 10.28 D S 9.6 18.66* 10.62 15.23* 9.6 12.93* 9.43 Temperature d ifference after f orced a ir c ooling (C) 13.45 17.7 12.07 20.12 13.92 19.1 13.82 17.49 12 16.24 D S 17.28 9.25 15.9 11.11 17.1 13.8 18.9 Max. t emperatur e during t ransportation (C) 20.4 22.86 12.13 13.31 14.15 11.6 16 18.02 16.4 12.47 D S 13.48 16 18.87 12.16 11.5 12.55 12.81 Exposure to HTA. DS: Defective sensor

PAGE 190

190 Table A 8 Percentage (%) of transportation time during which pineapples were exposed to different high temperature abuse (HTA) scenarios when using sea containers. Front Middle Back Top Center B ottom Top Center Bottom Top Center Bottom Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC At HTA No.1 (T 12C) 24.4 34.6 11.4 DS 42.7 4.5 0.8 32.1 29.9 29.7 5.7 At HTA No.2 (T 15C) 4.5 4.5 DS 9.4 4.9 4.9 At HTA No.3 (T 17C) 1.6 DS 4.1 0.6 1.4 DS: Defective sensor Table A 9. Risk of fruit exposure to different high temperature abuse (HTA) scenarios according to the position of the fruit in the pallet when using sea containers. Overall Top Center Bottom At HTA No.1 (T M H H N At HTA No.2 (T L H N N At HTA No.3 (T L H N N H = High, M = Medium, L = Low, and N = None

PAGE 191

191 Tab le A 10. Risk of fruit exposure to different high temperature abuse (HTA) scenarios according to the type of packaging used and the position of the fruit in the pallet when using sea containers. Overall Top Center Bottom Corr RPC Corr RPC Corr RPC Corr RPC At HTA No.1 (T H M H H H M N N At HTA No.2 (T M M H H N N N N At HTA No.3 (T L M M H N N N N H = High, M = Medium, L = Low, and N = None.

PAGE 192

192 0 5 10 15 20 25 30 35 05/08/07 14:55:00.0 05/08/07 20:25:00.0 05/09/07 01:55:00.0 05/09/07 07:25:00.0 05/09/07 12:55:00.0 05/09/07 18:25:00.0 05/09/07 23:55:00.0 05/10/07 05:25:00.0 05/10/07 10:55:00.0 05/10/07 16:25:00.0 05/10/07 21:55:00.0 05/11/07 03:25:00.0 05/11/07 08:55:00.0 05/11/07 14:25:00.0 05/11/07 19:55:00.0 05/12/07 01:25:00.0 05/12/07 06:55:00.0 05/12/07 12:25:00.0 05/12/07 17:55:00.0 05/12/07 23:25:00.0 05/13/07 04:55:00.0 05/13/07 10:25:00.0 05/13/07 15:55:00.0 05/13/07 21:25:00.0 05/14/07 02:55:00.0 05/14/07 08:25:00.0 05/14/07 13:55:00.0 05/14/07 19:25:00.0 05/15/07 00:55:00.0 TIME TEMPERATURE (C) a a) Forced Air Cooling b) Refrigeration c) Loading d) Land Transportation e) Warehouse/Port f) Sea Transportation b c d e f 1 2 3 Figure A 4. Ambient air temperature profile inside the sea conta iner displaying the three heating episodes along the supply chain: 1) During land transportation 2) During the process of loading the containers onto the vessel 3) During sea transportation.

PAGE 193

193 Table A 11. Temperatures recorded in the experimental pallets at different stages of the supply chain when using sea containers. Exposure to HTA. DS: Defective sensor Front Middle Back Top Center Bottom Top Center Bottom Top Center Bottom Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Temperature at begi nning of f orc ed a ir c ooling (C) 28.15 26.7 25.95 26.9 25.95 26 D S 26.34 26.34 26.7 26.34 26.3 27.78 25.56 26.34 25.56 25.95 25.6 Temperature at the end of f orced a ir c ooling (C) 11.29 9.94 12.55* 9.6 10.6 9.94 D S 10.11 13.32* 10.79 10.99 8.75 12.47* 9.03 13.7* 10.21 10.21 8.63 Temperature difference after f orced a ir c ooling (C) 16.86 16.76 13.4 17.3 15.35 16 D S 16.23 13.02 15.91 15.35 17.6 15.31 16.53 12.64 15.35 15.74 16.9 Max. t emperature during t ransportation (C) 16.5 17.35 12.16 11.8 10.99 11 D S 18.36 12.16 12.13 10.6 10.5 17.01 17.14 12.55 12.93 10.99 11.4

PAGE 194

194 0 2 4 6 8 10 12 14 16 18 5/8/07 5/9/07 5/10/07 5/11/07 5/12/07 5/13/07 5/14/07 5/15/07 5/16/07 Time Temperature (C) Front Back Figure A 5. Temperature profiles presenting the characteristic pattern of air short circuiting for pineapples packed in corrugated cartons placed on the t op layer of the experimental pallets inside the sea container. The temperature profile for the experimental pallet located in the middle of the container is not shown due to a malfunction of the sensor.

PAGE 195

195 Table A 12. Percentage (%) of transportation time du ring which pineapples were exposed to low temperature abuse (LTA) when using the refrigerated truck trailer / open hold combination. Front/1A Middle/2A Back/2B Top Center Bottom Top Center Bottom Top Center Bottom Corr RPC Corr RPC Corr RPC Corr RPC Cor r RPC Corr RPC Corr RPC Corr RPC Corr RPC At CI (T 7.73C) 16.7 16.3 21.0 20.2 25.4 25.7 14.9 BS 30.2 7.2 6.4 27.8 27.9 38.6 Table A 13. Risk of fruit exposure to low temperature abuse (LTA) according to the position of the fruit in the pallet when using the refrigerated truck t railer / open hold combination. Overall Top Center Bottom At CI (T H M H H H = High, M = Medium, L = Low, and N = None. Table A 14. Risk of fruit exposure to low temperature abuse ( LTA ) according to the packaging used and the position of the fruit in the pallet when using the refrigerated truck trailer / open hold combination. Where H= High, M= Medium, L=Low, and N=None. Overall Top Center Bottom Corr RPC Corr RPC Corr RPC Corr RPC At CI (T H H M L M H H H H= High M = Medium, L=Low, and N=None.

PAGE 196

196 Table A 15. Percentage (%) of transportation time during which the pineapples were exposed to low temperature abuse ( LTA ) when using the sea containers. Front Middle Back Top Center Bottom Top Center Bottom Top Center Bott om Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC Corr RPC At CI (T 7.73C) 58.1 72.7 29.7 62.6 Table A 16. Risk of fruit exposure to low temperature abuse (LTA ) according to the position of the fruit in the pallet when using sea containers. Overall Top Center Bot tom At CI (T L N L L H= High, M= Medium, L=Low, and N=None. Table A 17. Risk of fruit exposure to low temperature abuse (LTA) according to the packaging used and the position of the fruit in the pallet when using sea containers. Ove rall Top Center Bottom Corr RPC Corr RPC Corr RPC Corr RPC At CI (T N M N N N M N M H= High, M= Medium, L=Low, and N=None.

PAGE 197

197 APPENDIX B RECOMMENDED LEVEL OF RFID INSTRUMENTATION FOR THE CROWNLESS PINEAPPLE SUPPLY CHA IN WHEN THIS USES THE REFRIGERATED TRU CK/OPEN HOLD TRANSPORTATION METHOD Proper monitoring i s achieved recording two or three temperatures per pallet, according to the location of the pallet during forced air cooling. In order to monitor HTA, ambient probeless tags are rec ommended for the top layers of t he pallets and tags with probe a re advised for their center. Furthermore, tags with probe a re also suggested to observe LTA in the bottom layers of the pa llets. In this case, no areas a re specified as specific tag locations at the cargo level.

PAGE 198

198 APPENDIX C AMBIENT TEMPERATURE PROFILES FOR THE THE RMAL RELEVANCE STUDY OF A PALLET OF SPHERICAL BOTTLES OF WATER Table C 1. Ambient temperature profiles registered in the 0.5 h/0.5 h heating and cooling episodes. Time Ambient t emper atures (C) 942 26.16 945 27.15 947 27.13 950 26.82 952 28.25 955 27.64 957 27.54 1000 30.08 1002 28.84 1005 29.39 1007 31.02 1010 30.64 1012 32.92 1015 22.90 1017 18.64 1020 9.57 1022 8.30 1025 9.51 1027 11.17 1030 11.24 1032 10.58 1 035 11.14 1037 9.91 1040 8.17

PAGE 199

199 Table C 2. Ambient temperature profiles registered in the 1 h/1 h heating and cooling episodes. Time Ambient t emperatures (C) 937 16.94 940 30.06 942 34.25 945 28.14 947 17.68 950 18.47 952 19.46 955 18.88 957 18.85 1000 28.05 1002 18.64 1005 21.25 1007 18.91 1010 19.07 1012 19.35 1015 19.35 1017 21.35 1020 21.15 1022 20.34 1025 23.36 1027 19.75 1030 21.02 1032 18.86 1035 20.93 1037 21.65 1040 28.76 1042 23.41 1045 15.16 1047 11.03 1050 9.73 1052 9.14 1055 8.41 1057 8.74 1100 8.28 1102 7.88 1105 7.54 1107 7.24 1110 6.93 1112 6.84 1115 6.41 1117 6.27 1120 6.08 1122 5.94 1125 5.71

PAGE 200

200 Table C 2. Cont inued. Time Ambient t emperatures (C) 1127 5.73 1130 5.72 1132 5.34 1135 5.37 1137 5.53 1140 5.30 1142 5.35

PAGE 201

201 Table C 3. Ambient temperature profiles registered in the 2 h/2 h heating and cooling episodes. Time Ambient t emperatures (C) 1317 34.47 1320 32.54 1322 33.15 1325 34.34 1327 31.88 1330 32.98 133 2 34.51 1335 32.10 1337 36.04 1340 35.38 1342 36.68 1345 35.26 1347 30.79 1350 30.06 1352 29.27 1355 30.77 1357 30.28 1400 31.74 1402 35.78 1405 37.86 1407 39.14 1410 37.80 1412 38.47 1415 38.71 1417 37.13 1420 37.33 1422 38.06 1425 38 .92 1427 39.24 1430 40.78 1432 40.49 1435 39.28 1437 39.67 1440 39.11 1442 39.53 1445 39.14 1447 40.76 1450 39.35 1452 40.92 1455 40.49 1457 40.33 1500 40.63 1502 40.39 1505 41.32

PAGE 202

202 Table C 3. Cont inued. Time Ambient t emperatures (C) 1507 39.14 1510 40.22 1512 39.89 1515 42.14 1517 39.65 1520 19.08 1522 13.73 1525 14.13 1527 14.55 1530 14.46 1532 16.26 1535 11.47 1537 11.16 1540 10.63 1542 10.06 1545 9.66 1547 9.42 1550 8.83 1552 8.58 1555 8.21 1557 7.90 1600 7.56 1602 7.39 1605 7.09 1607 6.91 1610 6.80 1612 6.57 1615 6.55 1617 6.31 1620 6.29 1622 6.14 1625 6.03 1627 5.96 1630 5.99 1632 5.81 1635 5.68 1637 5.66 1640 5.55 1642 5.57 1645 5.46 1647 5.34 1650 5.34 1652 5.31

PAGE 203

203 Table C 3. Cont inued. Time Ambient t emperatures (C) 1655 5.22 1657 5.18 1700 5.15 1702 5.05 1705 5.03 1707 4.99 1710 4.92 1712 4.92 1715 4.76 1717 4.75

PAGE 204

204 Table C 4 Ambient temperature profiles registered in the 4 h/4 h heating and cooling episodes. Time Ambient t empera tures (C) 1022 7.20 1025 7.50 1027 7.61 1030 7.68 1032 7.67 1035 7.75 1037 7.82 1040 7.80 1042 7.77 1045 7.74 1047 7.65 1050 7.60 1052 7.58 1055 7.60 1057 7.67 1100 7.69 1102 7.69 1105 7.61 1107 7.59 1110 7.55 1112 7.58 1115 7.65 11 17 7.68 1120 7.74 1122 7.69 1125 7.66 1127 7.61 1130 7.54 1132 7.53 1135 7.56 1137 7.61 1140 7.69 1142 7.70 1145 7.76 1147 7.80 1150 7.83 1152 7.91 1155 8.10 1157 8.17 1200 8.22 1202 8.23 1205 8.28 1207 8.33 1210 8.35

PAGE 205

205 Table C 4. Cont i nued. Time Ambient t emperatures (C) 1212 8.46 1215 8.57 1217 8.69 1220 8.83 1222 8.90 1225 8.94 1227 9.01 1230 9.08 1232 9.19 1235 9.28 1237 9.41 1240 9.52 1242 9.70 1245 9.80 1247 9.91 1250 10.01 1252 10.07 1255 10.18 1257 10.30 1300 10.33 1302 10.34 1305 10.39 1307 10.46 1310 10.55 1312 10.64 1315 10.65 1317 10.71 1320 10.75 1322 10.73 1325 10.81 1327 10.86 1330 10.92 1332 10.96 1335 11.02 1337 11.03 1340 11.06 1342 11.11 1345 11.11 1347 11.15 1350 11.22 1352 11.3 5 1355 11.44 1357 11.49

PAGE 206

206 Table C 4. Cont inued. Time Ambient t emperatures (C) 1400 11.58 1402 11.66 1405 11.74 1407 11.82 1410 11.96 1412 12.01 1415 12.09 1417 12.13 1420 12.19 1422 12.27 1425 12.33 1427 12.41 1430 12.44 1432 12.53 1435 12.65 1437 12.63 1440 12.63 1442 12.25 1445 11.61 1447 11.06 1450 10.82 1452 10.69 1455 10.53 1457 10.32 1500 10.30 1502 10.66 1505 10.57 1507 10.58 1510 10.49 1512 10.41 1515 10.55 1517 10.73 1520 10.81 1522 10.86 1525 10.91 1527 10.9 0 1530 10.90 1532 10.85 1535 10.87 1537 10.86 1540 10.84 1542 10.84 1545 10.85

PAGE 207

207 Table C 4. Cont inued. Time Ambient t emperatures (C) 1547 10.83 1550 10.81 1552 10.78 1555 10.78 1557 10.73 1600 10.74 1602 10.71 1605 10.68 1607 10.66 1610 10.64 1612 10.61 1615 10.56 1617 10.49 1620 10.46 1622 10.40 1625 10.33 1627 10.30 1630 10.21 1632 10.17 1635 10.10 1637 10.05 1640 9.98 1642 9.91 1645 9.84 1647 9.76 1650 9.73 1652 9.61 1655 9.54 1657 9.47 1700 9.39 1702 9.34 1705 9. 26 1707 9.19 1710 9.14 1712 9.08 1715 9.00 1717 8.94 1720 8.85 1722 8.81 1725 8.77 1727 8.79 1730 8.83 1732 8.84

PAGE 208

208 Table C 4. Cont inued. Time Ambient t emperatures (C) 1735 8.83 1737 8.83 1740 8.83 1742 8.78 1745 8.73 1747 8.69 1750 8.63 1752 8.57 1755 8.52 1757 8.48 1800 8.41 1802 8.37 1805 8.31 1807 8.25 1810 8.23 1812 8.14 1815 8.13 1817 8.04 1820 8.03 1822 7.95 1825 7.93

PAGE 209

209 APPENDIX D TEMPERATURE PROFILE USED AS WORST CASE SCENARIO FOR FSR SHIP MENTS Table D 1. Maximum daily temperatures obtained in a combat feeding shipment from the U.S. to Kuwait. Days Temperatures (F) 1 104 2 104 3 91.4 4 107.6 5 82.4 6 73.4 7 73.4 8 96.8 9 89.6 10 91.4 11 91.4 12 91.4 13 87.8 14 86 15 86 16 86 17 87.8 18 8 9.6 19 95 20 96.8 21 102.2 22 102.2 23 102.2 24 102.2 25 100.4 26 96.8 27 100.4 28 100.4 29 118.4 30 114.8 31 120.2 32 129.2 33 122 34 125.6 35 131 36 122 37 122 38 125.6 39 129.2

PAGE 210

210 LIST OF REFERENCES Abada, E., Zampolli, S., Marcoc, S., Scorzoni, A., Mazzolai, B. et al., 2007. Flexible tag microlab development: Gas sensors integration in RFID flexible tags for food lo gistic. Sensors and Actuators B 127, 27. Abdullah, H., Rohaya, M.A., Rosli, H., Mohamed Selamat, M., 2000. Handling an d transportation trial of pineapple by sea shipment from Malaysia to the United Kingdom. Acta Horticulturae 529, 317327. Acedo, A., Akinaga, T., Tanabez, T., 2004. Inhibition of chilling injury and quality changes in pineapple fruit with prestorage heat t reatment. Journal of Food, Agriculture and Environment 2, 8186. Akamine, E.K., Goo, T. Steepy, T., Greidanus, T., Iwaoka, N., 1975. Control of endogenous brown spot of fresh pineapple in postharvest handling. J ournal of the Am erican Soc iety for Hort icult ural Sci ence 100, 60 65. Amador, C., Emond, J.P., Nunes, M.C.N., 2009. Application of RFID Technologies in the Temperature Mapping of the Pineapple Supply Chain. Journal of Sensing and Instrumentation for Food Quality and Safety 3 (1), 26 33. Amador, C., E mond, J.P., 2010. Evaluation of sensor readability and thermal relevance for RFID temperature tracking. Computers and Electronics in Agriculture 73 (1), 8490. Ames, H., 2006. Authentication from a Cold Chain Perspective. Pharmaceutical Commerce. July/Augu st 2006. Amos, N.D., 2001. Factors affecting fruit temperature maintenance within refrigerated containers. Proceedings of the IRHACE Annual Conf. Vol I, 7 10. Anderson, K. E., 2010. Development of a simulator for sweet corn cold chain distribution. M.Sc. Thesis. University of Florida, Agricultural and Biological Engineering. Angeles, R., 2005. RFID Technologies: Supply-Chain Applications and Implementation Issues. Information Systems Management, Vol. 22 (1), 51 65. Aroor, S. R., Deavours, D. D., 2007. Eval uation of the state of passive UHF RFID: An experimental approach. IEEE Systems Journal 1 (2), 168 176. Asimakopoulos, G., Spiros Louvros, G., Triantafillou, V., 2007. METATRO: A Real Time RFID enabled haulage monitoring system for perishable comestibles. Paper Presented at the IEEE Conference on Emerging Technologies and Factory Automation (ETFA), Patras, Greece. Banerjee S.R., Jesme, R., Sainati, R.A., 2007. Performance Analysis of Short Range UHF Propagation as Applicable to Passive RFID. Paper Presented at 2007 IEEE International Conference on RFID, Grapevine, TX.

PAGE 211

211 Banks, J., Pachano, M., Thompson, L., Hanny, D., 2007. RFID Applied. John Wiley & Sons, Inc., Hoboken. Banwart, GJ., 1989. Basic food microbiology, 2. Van Nostrand Reinhold, New York. BBC, 2010 a. BBC News: US troops in Afghanistan outnumber those in Iraq. British Broadcasting Corporation. Available at: http://news.bbc.co.uk/2/hi/world/us_and_canada/10151175.stm Accessed July 6th, 2010. BBC, 2010b. World Weather: Average Conditions. British Broadcasting Corporation Available at: http://www.bbc.co.uk/weather/world/city_guides Accessed February 5th 2010. Benna hmias, R., Labonne, G., 1993. Etude de la distribution de l'air et de la dispersion des temperatures dans une semi -remorque frigorifique. Proceedings of the I.I.R Commissions C2, D1 et D2/3, Fez, Morocco, 241256. Billing, D.P., Hayes, A.J., McDonald, B., 1993. Performance of refrigerated ships carrying New Zealand produce. Refrigeration Science and Technology 3, 546554. Billing, D.P., Hayes, A.J., McDonald, B., 1995. Temperature and airflow characteristics in reefer vessels carrying New Zealand produce. Proceedings of the 19th International Congress of Refrigeration, La Haye, Netherlands, Vol II, 485492. Bogataj, M., Bogataj, L., Vodopivec, R., 2005. Stability of perishable goods in cold logistic chains. International Journal of Production Economics 939 4 (1), 345356. Bollen, F., 2005. Traceability in Fresh Produce Supply Chains. Acta Horticulturae 687, 279288. Bollen, F., Riden, C., 2006. Traceability in Postharvest Quality Management. International Journal of Postharvest Technology and Innovation 1 (1), 93 96. Brody, A.L., 2003. Predicting packaged food shelf life. Food Technology 57(4), 100102. Broekmeulen, R.A.C.M., Van Donselaar, K.H., 2007. A Replenishment policy for a perishable inventory system based on estimated aging and retrieval behavior. Wo rking Paper, Department of Technology Management, University of Eindhoven, Eindhoven, Netherlands. Brosnan, T., Sun, D., 2001. Precooling techniques and applications for horticultural products A review. International Journal of Refrigeration 24, 154 170. Brown, D., 2007. RFID Implementation. Mc Graw Hill, New York. Capgo, 2010. Introduction to Bimetallic Temperature Sensors. Capgo Pty Ltd. Available at: http://www.capgo.com/Re sources/Temperature/BiMet/BiMetallic.html Accessed June 28th, 2010. Chamberlain, J., 2006. IBM WebSphere RFID Handbook: A Solution Guide. IBM Redbooks.

PAGE 212

212 Chande, A., Dhekane, S., Hemachandra, N., Rangaraj, N., 2005. Perishable inventory management and dyna mic pricing using RFID technology. Sadhana 30 (2), 445462. Chonhenchob, V., Kamhangwong, D., Singh, S.P. 2008. Comparison of reusable and single use plastic and paper shipping containers for distribution of fresh pineapples. Packaging Technology and Scie nce 21, 7383. Chonhenchob, V., Singh, S.P. 2003a. A comparison of corrugated boxes and reusable plastic containers for mango distribution. Packaging Technology and Science 16, 231237. Chonhenchob, V., Singh, S.P. 2003b. Packaging performance comparison for distribution and export of papaya fruit. Packaging Technology and Science 18, 125131. Clarke, R.H., Twede, D., Tazelaar, J.R., Boyer, K.K., 2006. Radio Frequency Identification (RFID) performance: The effect of tag orientation and package contents. Packaging Technology and Science 19, 4554. Corbo, M. R., Del Nobile, M.A., Sinigaglia, M., 2006. A novel approach for calculating shelf life of minimally processed vegetables. International Journal of Food Microbiology 106, 6973. Cortbaoui, P., 2005. Ass essment of Precooling Technologies for Sweetcorn. M. Sc. Thesis. McGill University, Bioresource Engineering. Dagdelen, D., 2007. Improving the Readability of RFID tagged Pharmaceutical Products and Food Supplements inside a Modified Sealed Tote. M.Sc. Thes is, University of Florida, Food Science and Human Nutrition Department. DeEll, J.R., Prange, R.K., Peppelenbos, H.W., 2003. Postharvest Physiology of Fresh Fruits and Vegetables. In: Chakraverty, A., Mujumdra, A.S., Vijaya Raghavan, G.S., Ramaswamy, H.S. ( Eds.), Handbook of Postharvest Technology: Cereals, Fruits, Vegetables, Tea, and Spices. Marcel Dekker, Inc., New York, pp. 455484. Diner, I., 2003. Heat Transfer in Food Cooling Applications. Taylor & Francis, Washington, D.C. Dobkin, D., Weigand, S., 2 005. Environmental effects on RFID tag antennas. Paper Presented at the 2005 IEEE MTT -S International Microwave Symposium, Long Beach, CA. DoD Live., 2009. Armed with science: First Strike Rations offer fast food on the go. US Department of Defense. Availa ble at: http://www.dodlive.mil/index.php/2009/11/first strike rations -offer -fast -food -onthe -go Accessed 2nd February 2010. Dull, G., Young, R., Biale, J. 1967. Respiratory patterns in fruit of pineapple, Ananas comosus, detached at different stages of development. Physiologia Plantarum 20, 10591065. Edwards, J., 2007. Cold Chain Heats Up RFID Adoption. RFID Journal. March/April 2007.

PAGE 213

213 Efunda, 2010. Tem perature Sensors. eFunda, Inc. Available at: http://www.efunda.com/designstandards/sensors/thermocouples/thmcple_intro.cfm Accessed June 28th, 2010. Emond, J.P. Nicometo, M., 2006. Shelf -life prediction and FEFO inventory management with RFID. Presentation in the Cool Chain Association Workshop: Temperature Measurements When, Where and How? Knivsta, Sweden. Available at: www.coolchain.org/uploads/MikeNicometo CCAStockholmNov06.pdf Accessed January 15th 2008. Ferguson, M. E., Ketzenberg M.E., 2006. Information sharing to improve retail product freshness of perishables. Producti on and Operations Management 15 (1), 5773. Finn, D.P., Brennan, S.L., 2003. Sensitivity analysis using CFD of air distribution in transport refrigeration containers. Paper Presented at the 21st International Congress of Refrigeration, Washington, DC. Flet cher, R., Marti, U., Redemske, R., 2005. Study of UHF RFID Signal Propagation through Complex Media. Paper Presented at the 2005 IEEE Antennas and Propagation Society International Symposium, Washington, DC. Food and Agriculture Organization, 2009. FAOSTAT Available at http://faostat.fao.org/ Accessed November 30th, 2009. Forcinio, H., 2004. Packaging Helps Protect Temperature -Sensitive Products. Pharmaceutical Technology 28 (3), 3642. Freund, R., Wilson, W., 2003. Statistical Methods, 2. Academic Press, San Diego. Gaukler, G., Seifert, R.W., 2007. Applications of RFID in Supply Chains. In: Jung, H., Chen, F., Jeong, B. (Eds.), Trends in Supply Chain Design and Management: Technologies and Methodologies. Springer -Verlag, London, pp. 29 49. Giannakourou, M. C., Taoukis, P.S., 2003. Application of a TTI based distribution management system for quality optimization of frozen vegetables at the consumer end. Journal of Food Science 68, 201209. Gogs, A.Y., Yavuzturk, S .,1974. Temperature pull -down and distribution in refrigerated trailers. Paper Presented at I.I.R commissions D2, Wageningen, Netherlands. Gonzalez-Aguilar, G.A., Ruiz Cruz, S., Cruz -Valenzuela, R., Rodriguez -Felix, A., Wang, CY., 2004. Physiological and q uality changes of fresh -cut pineapple treated with anti -browning agents. Lebensmittel Wissenschaft undTechnologie 37, 369376. Goyal, S.K., Giri, B.C., 2001. Recent trends in modeling of deteriorating inventory. European Journal of Operational Research 134 (1), 1 16.

PAGE 214

214 Griffin, J.D., Durgin, G.D., Haldi, A., Kippelen, B., 2005. Radio link budgets for 915 MHz RFID antennas placed on various objects. Paper Presented at 2005 Texas Wireless Symposium, Austin, TX. Hartvanyi, T., Marek, J., 2007. RFID Applications in the Supply Chain. Paper Presented at 2007 RFID Eurasia, Istanbul, Turkey. Heap, R.D., Kierstan, M., Ford, G., 1998. Food Transportation. Blackie Academic and Professional, London. Heap, R.D., 1989. Design and performance of insulated and refrigerated I SO intermodal containers. International Journal of Refrigeration 12, 137145. Hertog, M.L.A.T.M., Boerrigter, H.A.M., Boogaard, G.J.P.M., van den Tijskens, L.M.M., van Schaik, A.C.R., 1999. Predicting keeping quality of strawberries (cv. Elsanta) packed under modified atmospheres: an integrated model approach. Postharvest Biology and Technology 15, 1 12. Hough, G., Garittaa, L., Gmez, G., 2006. Sensory shelf -life predictions by survival analysis accelerated storage models. Journal of Food Quality and Pre ference 17 (6), 468473. Hui, C.K.P., LeBlanc, D.I., Vigneault, C., DeEll, J.R., Sotocinal, S.A., 2003. Transportation and Handling of Fresh Fruits and Vegetables. In: Chakraverty, A., Mujumdra, A.S., Vijaya Raghavan, G.S., Ramaswamy, H.S. (Eds.), Handbook of Postharvest Technology: Cereals, Fruits, Vegetables, Tea, and Spices. Marcel Dekker, Inc., New York, pp. 555 584. Hunt, V.D., Puglia, A., Puglia, M., 2007. RFID: A Guide to Radio Frequency Identification. Wiley -Interscience, Hoboken. Identec Solutions Inc., 2007. Active UHF Tag i Q32T. Version 1.2. Identec Solutions Inc. Addison. Ilic, A., Schuster, E. W., Staake, T., Fleisch, E. 2009. Sensors in Retailing: Reducing the Quality Uncertainty of Perishable Goods. Working Paper, Bits to Energy Lab, Informa tion Management Department, ETH Zurich, Zurich, Switzerland. Jacksens, L., Devlieghere, F., Debevere, J., 2002. Temperature dependence of shelf life as affected by microbial proliferation and sensory quality of equilibrium modified atmosphere package fresh produce. Postharvest Biology and Technology 26, 59 73. James, S.J., 2006. Principles of Food Refrigeration and Freezing. In: Hui, Y.H. (Ed.), Handbook of Food Science, Technology, and Engineering, Vol. 3. CRC Press, Boston. Jedermann, R., Emond, J.P., Lan g, W., 2008. Shelf life prediction by intelligent RFID. In: Haasis, H.D., Kreowski, H.J., Scholz -Reiter, B. (Eds.), Dynamics in Logistics: Proceedings of the First International Conference LDIC 2007. Springer -Verlag, Berlin Heidelberg, pp. 231240.

PAGE 215

215 Jederma nn, R., Lang, W., 2007. Semi passive RFID and beyond: steps towards automated quality tracing in the food chain. International Journal of Radio Frequency Identification Technology and Applications 1 (3), 247259. Jedermann, R., Stein, K., Becker, M., Lang, W., 2008. UHF RFID in the Food Chain From Identification to Smart Labels. Paper Presented at Cold Chain Manangement: 3rd International Workshop, Bonn, Germany. Jones, P., 2006. Networked RFID for use in the Food Chain. Paper Presented at the IEEE Confer ence on Emerging Technologies and Factory Automation (ETFA). Prague, Czech Republic. Kader, A.A., 2002. Postharvest Technology of Horticultural Crops. University of California Division of Agriculture and Natural Resources, Oakland. Kapoor, G., 2008. Secure Ownership Transfer and Authentication Protocols for Radio Frequency Identification (RFID). Ph.D. Thesis, University of Florida, Department of Business Administration. Karkkainen, M., 2003. Increasing efficiency in the supply chain for short shelf life goods using RFID tagging. International Journal of Retail and Distribution Management, 31 (10), 529536. Labuza, T.P., Fu, B., 1993. Growth kinetics for shelf -life prediction: theory and practice. Journal of Industrial Microbiology 12, 30923. Lahiri, S., 200 6. RFID Sourcebook. IBM Press, Upper Saddle River. Laniel, M., Emond, J.P., Altunbas, A.E., 2009. RFID behavior study in enclosed marine container for real time temperature tracking. Journal of Sensing and Instrumentation for Food Quality and Safety 3 (1), 34 40. Lzaro, A., Girbau, D. Salinas, D., 2009. Radio link budgets for UHF RFID on multipath e nvironments. IEEE Transactions on Antennas a nd Propagation 57 (4), 12411251. Leake, L., 2007. The Search for Shelf Life Solutions. Food Technology 61 (11), 6670. LeBlanc, D., Beaulieu, C., Lawrence, R., Stark, R., 1994. Evaluation of temperature variation of frozen foods during tranportation. The Refrigeration Research Foundation Information Bulletin, Bethesda, MD. Lenker, D.H., Wooddruff, D.W., Kindya, W.G., C arson, E.A., Kasmire, R.F., Hinsch, R.T., 1985. Design criteria for the air distribution systems of refrigerated vans. ASAE Papers 28 (6), 20892097. Leong, K. S., Ng, M. L., Cole, P. H., 2006. Dual -frequency antenna design for RFID application. Paper Presented at the 21st International Technical Conference on Circuits/Systems, Computers and Communications (ITC CSCC 2006), Chiang Mai, Thailand.

PAGE 216

216 Machado, F.L.C., Alves, R.E., Figueiredo, R.W., Teixeira, A.S., 2009. Quality maintenance of ripe pineapple as affected by application of wax associated to 1 -Methylcyclopropene. Acta Horticulturae 822, 261268. McMeekin, T.A., Ross, T., 1996. Shelf life prediction: status and future possibilities. International Journal of Food Microbiology 33, 65 83. Meana, M.B., 2005. Forced-Air Cooling Of Strawberries in Reusable Plastic Containers. M.E. Thesis, University of Florida, Agricultural and Biological Engineering Depar tment. Meffert, H.F.T., Van Nieuwenhuizen, G., 1973. Temperature distribution in refrigerated vehicles. Proceedings of the I.I.R Commissions D1, D2 and D3, Barcelona, Spain. Mermelstein, N. H., 2002. A look into the future of food science and technology. F ood Technology 56 (1), 4855. Metzger, C., Michahelles, F., Fleisch, E., 2007. Ambient Energy Scavenging for Sensor Equipped RFID Tags in the Cold Chain. Lecture Notes in Computer Science 4793, 255269. Michalski, L., Eckersdorf, K., Kucharski, J., McGhee, J., 2001. Temperature Measurement. John Wiley and Sons, Inc., Chichester. Miller, W.M., Dou, H., Talbot, M., 2001. Evaluation of refrigerated conditions and fruit quality in cooling fresh Florida citrus. Proceedings of the Florida State Hort. Soc. 114, 181 185. Mitchell, F. G., Parsons, R. A., Mayer, G., 1971. Cooling trials with plastic tray pack nectarines in various containers. California Agriculture. 25(9):1315. Mo, L.F., Zhang, H.J. 2007. RFID antenna near the surface of metal. Paper Presented at the International Symposium on Microwave, Antenna, Propagation and EMC Technologies for Wireless Communications, Hangzhou, China. Mohammed, M., Wickham, L., 1995. Biochemical changes and sensory evaluations in pineapple during storage at refrigerated and non -refrigerated temperatures. Acta Horticulturae 425, 571580. Montalbano Technology., 2008. MT Product Family. Available at http://www.montalbanotechnology.com/index.htm Accessed on October 15th, 2008. Montrucchio, G.F., Rebaudengo, M., Sanchez, E.R., 2007. Analysis of a RFID -based information system for tracking and tracing in an agri -food chain. Paper Presented at 2007 RFID Eurasia, Istanbul, Turkey. Moureh, J., Flick, D., 2004. Airflow pattern and temperature distribution in a typical refrigerated truck configuration loaded with pallets. International Journal of Refrigeration 27 (5), 464474.

PAGE 217

217 Moureh, J., Menia, N., Flick, D., 2002. Numerical and experimental study of airflow in a typical refrige rated truck configuration loaded with pallets. Computers and Electronics in Agriculture 34, 2542. Moureh, J., Tapsoba, S., Derens, E., Flick, D., 2009. Air velocity characteristics within vented pallets loaded in a refrigerated vehicle with and without ai r ducts. International Journal of Refrigeration 32, 220234. Mn dlein M., Nicolics J., Brandl M., 2004. Packaging Concept for a Miniaturized Wirelessly Interrogable Temperature Logger. Proceedings of the 2004 International Spring Seminar on Electronics Te chnology (ISSE) Vol. 1, 68 73. Nahmias, S., 1982. Perishable inventory theory: A review. Operations Research 30 (4), 680708. Narayan, K., Porcella, C., Shaw, C.P., 1997. The Effect of hightemperature storage on the digestibility of carbohydrates in pouch bread. Natick, Mass: US Army Soldier Systems Command, Natick Research, Develoment and Engineering Center. Nattress, D., Bennett, A., Canniff, M., Aylward, J., Moody, S., LaBrode, A., 2009. Improved storage stability of meal ready to eat cheese spread under heat -stressing conditions. US Army Soldier Systems Command, Natick Research, Develoment and Engineering Center, Natick, MA. Ng, T.N., Viard, E., Caipo, M.L., Duffy, S., Schaffner, D.W., 2002. Expansion and validation of a predictive model for the growth of Bacillus Stearothermophilus in military rations. Journal of Food Science 67 (5), 1872 1878. Nunes, M.C.N., 2008. Color Atlas of Postharvest Quality of Fruits and Vegetables. Blackwell Publishing Ltd., Ames. Nunes, M.C.N., 2010. Quality curves for Firs t Strike Rations (FSR) as a function of the storage temperature. Unpublished report. Food Science and Human Nutrition Department, University of Florida, Gainesville, FL. Nunes, M.C.N., Brecht, J.K., Sargent, S.A., Morais, A.M.M.B., 1995. Effects of delays to cooling and wrapping on strawberry quality (cv. Sweet Charlie). Food Control 6 (6), 323328. Nunes, M.C.N., Emond, J. P., Brecht J.K., 2003a. Quality of strawberries as affected by temperature abuse during ground, in -flight and retail handling operation s. Acta Horticulturae 604, 239246. Nunes, M.C.N., Emond, J. P., Brecht, J.K., 2004. Quality curves for highbush blueberries as a function of the storage temperature. Small Fruits Review 3, 423440. Nunes, M.C.N., Emond, J.P., Brecht, J.K., 2006. Brief dev iations from set point temperatures during normal airport handling operations negatively affect the quality of papaya (Carica papaya) fruit. Postharvest Biology and Technology 41(3), 328340.

PAGE 218

218 Nunes, M.C.N., Emond, J.P., Rauth, M., Dea, S., Chau, K.V., 2009 Environmental conditions encountered during typical consumer retail display affect fruit and vegetable quality and waste. Postharvest Biology and Technology 51, 232241. Nunes, M.C.N., Proulx, E., Emond, J.P., 2003b. Quality characteristics of Horn of P lenty and Medallion yellow summer squash as a function of storage temperature. Acta Horticulturae 628, 607614. Omega, 2010. Technical Reference. Omega Engineering, Inc. Available at: http://www.omega.com/tec href/ Accessed June 28th, 2010. Opara, L., 2003. Traceability in agriculture and food supply chain: a review of basic concepts, technological implications, and future prospects. Journal of Food, Agriculture and Environment, 1 (1), 101106. Paull, R.E, 1 993. Pineapple and papaya. In: Seymour, G., Taylor J., Tucker G. ( E ds ), Biochemistry of Fruit Ripening. Chapman & Hall, London, pp. 291323. Panos, R. C., Freed, T., 2007. The Benefits of Automatic Data Collection in the Fresh Produce Supply Chain. Pape r Presented at the 3rd Annual IEEE Conference on Automation Science and Engineering, Scottsdale, AZ. Peleg, K., 1985. Produce Handling, Packaging and Distribution. Avi Publishing Company, Inc., Westport. Perez -Aloe, R., Valverde, J.M., Lara, A., Carrillo, J.M., Roa, I., Gonzalez, J., 2007. Application of RFID tags for the overall traceability of products in cheese industries. Paper Presented at 2007 RFID Eurasia, Istanbul, Turkey. Potdar, V., Hayati, P., Chang, E., 2007. Improving RFID Read Rate Reliability by a Systematic Error Detection Approach. Paper Presented at 2007 RFID Eurasia, Istanbul, Turkey. Pot ter, N.N., Hotchkiss, J.H., 1998 Food Science. Springer -Verlag, New York. Produce Marketing Association., 2008. Traceability Frequent Asked Questions. Av ailable at www.pma.com/view_document.cfm?docID=221. Accessed on October 20th, 2008. Proulx, E., Nunes, M.C.N., Emond, J.P., 2005. Quality attributes limiting papaya postharvest life at chilling and non -chilling temperatures. Proceedings of the Florida State Hort. Soc. 118, 389395. Punt, H., Huysamer, M., 2005. Temperature variances in a 12 m integral reefer container carrying plums under a dual temperature shipping regime. Acta Horticulturae 687, 289296. Raza, N., Bradshaw, V., Hague, M., 1999. Applications of RFID Technology. Paper Presented at the IEEE Colloquium on RFID Technology, London, UK.

PAGE 219

219 Redemske, R., Fletcher, R., 2005. Design of UHF RFID Emulators with Applications to RFID Testing an d Data Transport. Paper Presented at the 4th IEEE Workshop on Automatic Identification Advanced Technologies, Buffalo, NY. Reed, S. 2009. Physical Aspects of Applying Wireless Sensors: Short range, UHF radio waves and antennas in natural environments. Resource (January/February). Riva, M., Pranzetti, L., Galli, A., 2001. Effect of storage temperature on microbiological quality and shelf -life of ready to use salads. Annals of Microbiology 51, 39 52. Rodriguez Bermejo, J., Barreiro, P., Robla, J.I., Ruiz Garc ia, L., 2007. Thermal study of a transport container. Journal of Food Engineering 80, 517527. Ross, E.W., Klicka, M., Kalick, J., Branagan, M.T., 1985. Acceptance of a military ration after 24month storage. Journal of Food Science 50 (1), 178181. Ross, E.W., Klicka, M., Kalick, J., Branagan, M.T., 1987. A time temperature model for sensory acceptance of a military ration. Journal of Food Science 52 (6), 1712 1717. Ross, E.W., Shaw, C.P., Friel, M., 1997. Color measurement as predictor of consumer ratin gs of military ration items. Journal of Food Quality 20 (5), 427439. Roussos, G., 2006. Enabling RFID in retail. IEEE Computer 39 (3), 25 30. Ruiz Garcia, L., 2008. Development of Monitoring Applications for Refrigerated Perishable Goods Transportation. P h.D. Thesis, Universidad Politecnica de Madrid, Escuela Tecnica Superior de Ingenieros Agronomos. Ruiz Garcia, L., Barreiro, P., Robla, J.I., 2008. Performance of ZigBee Based wireless sensor nodes for real -time monitoring of fruit logistics. Journal of Fo od Engineering 87 (3), 405415. Saenz, A., DAlolio, O., 2007. Referencias tecnicas de manejo pre y postcosecha en p ia cultivar MD 2. San Jose, Costa Rica: Consejo Nacional d e Produccion del Gobierno de Costa Rica Direccion d e Calidad Agricola Sahin, E ., Baba, M.Z., Dallery, Y., Vaillant, R., 2007. Ensuring supply chain safety through time temperature integrators. The International Journal of Logistics Management 18 (1), 102124. Saltveit, M. E., Morris, L.L. 1990. Overview on chilling injury of horti cultural crops. In : Wang, C. (Ed.), Chilling Injury of Horticultural Crops. CRC Press, Boca Raton pp. 3 16. Saltveit, M., 2003. Temperature Extremes. In: Brecht, J., Weichmann, J. (Eds.), Postharvest Physiology and Pathology of Vegetables. CRC Press, Boca Raton, pp. 457484. Sargent, S.A., Talbot, M.T., Brecht, J.K., 1988. Evaluating precooling methods for vegetable packinghouse operations. Proceedings of the Florida State Hort. Soc. 101, 175182.

PAGE 220

220 Shaw, C.P., Kluter, R. A., Morse, M., Engell, D., 1997. The effects of fluctuating temperature storage on the acceptibility of meal, ready to -eat components. Natick, Mass: US Army Soldier Systems Command, Natick Research, Develoment and Engineering Center. Shewfelt, R. L., Bruckner, B., 2000. Fruit and Vegetable Q uality: An Integrated View. Technomic Publishing, Lancaster. Shewfelt, R.L., Prussia, S.E., 1993. Postharvest Handling: A Systems Approach. Academic Press, San Diego. Sivakumar, M., Deavours, D., 2008. A Dual Resonant Microstrip Antenna for Paperboard in t he Cold Chain. Paper Presented at the 2008 IEEE Sarnoff Symposium, Princeton, NJ. Smale, N. J., 2004. Mathematical Modelling of Airflow in Shipping Systems: Model Development and Testing. Ph.D. Thesis, Massey University, Food Technology Department. Smale, N.J., Moureh, J., Cortella, G., 2006. A review of numerical models of airflow in refrigerated food applications. International Journal of Refrigeration, 29 (6), 911930. Smith, J., 2005. Specialized Logistics for a Longer Perishable Supply Chain. World Tra de Magazine 18 (11), 4651. Stera, A.C., 1999. Long distance refrigerated transport into the third millennium. Proceedings of the 20th International Congress of Refrigeration Vol 4, 1 18. Sweeney, P.J., 2005. RFID for Dummies. Wiley Publishing, Inc., Hobok en. Sydnheimo, L., Ukkonen, L., Kivikoski, M., 2006. Effects of size and shape of metallic objects on performance of passive radio frequency identification. International Journal of Advanced Manufacturing Technology 30, 897 905. Talbot T.M., Sargent, S.A. Brecht, J.K., 1991. Cooling Florida Sweet Corn. In: Circular 941. Extension Service Institute of Food and Agricultural Sciences, University of Florida, Gainesville, pp. 1 21. Talbot, M.T., Baird, C.D., Sargent, S.A., Brecht, J.K., 1992. Improving Forced -Air Cooler Performance. In: Circular AE 108. Extension Service Institute of Food and Agricultural Sciences, University of Florida, Gainesville. Talbot, M.T., Chau, K.V., 1991. Precooling of Strawberries. In: Circular 942. Extension Service Institute of Foo d and Agricultural Sciences, University of Florida, Gainesville. Tanner, D. J., Amos, N. D., 2003. Temperature variability during shipments of fresh produce. Acta Horticulturae 599, 193204. Tapsoba, M., Moureh, J., Flick, D., 2006. Airflow patterns in an enclosure loaded with slotted pallets. International Journal of Refrigeration 29, 899910.

PAGE 221

221 Temperatures, 2010. Temperature Sensor Types. Temperatures.com, Inc. Available at: http://www.temperatures.c om/sensors.html Accessed June 28th, 2010. Thompson, A.K., 2003. Fruit and Vegetables: Harvesting, Handling and Storage. Blackwell Publishing Ltd., Oxford. Thompson, J., 2004. Pre -cooling and storage facilities. The Commercial Storage of Fruits, Vegetables, and Florist and Nursery Stocks. In: Agriculture Handbook 66. USDA, Washington, D C. Thompson, J., Mitchell, F., Rumsey, T., Kasmire, R., Crisosto, C., 2002. Commercial Cooling of Fruits, Vegetables, and Flowers, Revised Edition. University of California -Division of Agriculture and Natural Resources, Oakland. Thompson, J.F., Mitchell, F.G., Rumsey, T.R., Kasmire, R.F., Crisosto, C.C., 1998. Commercial Cooling of Fruits, Vegetables, and Flowers. University of California Division of Agriculture and Natural Resources, Oakland. Thompson, M., Sylvia, G., Morrissey, M.T., 2005. Seafood Traceability in the United States: Current Trends, System Design, and Potential Applications. Comprehensive Reviews in Food Science and Food Safety. 4 (1), 1 7. Thyagaraja, K., 2 007. GatorPacker: A worker productivity monitoring system using RFID (Radio Frequency Identification) technology. M.Sc. Thesis, University of Florida, Agricultural and Biological Engineering Department. Tutar, M., Erdogdu, F., Toka, B., 2009. Computational modeling of airflow patterns and heat transfer prediction through stacked layers' products in a vented box during cooling. International Journal of Refrigeration 32 (2), 295306. United States Department of Agriculture/Economic Research Service, 2010. Fo od availability Spreadsheets: Fresh fruits. Available at http://www.ers.usda.gov/Data/FoodConsumption/FoodAvailSpreadsheets.htm Accessed on January 16th, 2010. Uysal, I ., 2010. Development of models for a temperature profile estimator. Unpublished report. Agricultural and Biological Engineering Department University of Florida, Gainesville, FL. van Boekel, M.A.J.S., 2008. Kinetic Modeling of Food Quality: A Critical Rev iew. Comprehensive Reviews in Food Science and Food Safety 7, 144158. Verboven, P., Tijskens, E., Ramon, H., Nicola, B.M., 2005. Virtual filling and airflow simulation of boxes with horticultural products. Acta Horticulturae 687, 47-54. Vergara, A., Llob et, E., Ramirez, J. L., Ivanova, P. et al., 2007. An RFID reader with onboard sensing capability for monitoring fruit quality. Sensors and Actuators B 127, 143149.

PAGE 222

222 Vigneault, C., Goyette, B., Markarian, N.R., Hui, C.K.P., Ct, S., Charles, M.T., mond, J .P., 2004. Plastic container opening area for optimum hydrocooling. Canadian Biosystems Engineering 46, 3.41. Vigneault, C., Wu, S., Thompson, J.F., Hui, K.P.C., Leblanc, D.L., 2009. Transportation of Fresh Horticultural Produce. In: Benkeblia, N., Shiomi, N. (Eds.), Postharvest Technologies for Horticultural Crops, Vol. 2. Research Signpost, Kerala, pp. 1 24. Wall Street Journal, 2010. Money Rates. Dow Jones & Company, Inc. Available at: http://online.wsj.com/mdc/public/page/2_3020moneyrate.html Accessed July 7th, 2010. Wang, J.K., Tunpun, K., 1968. Forcedair cooling of tomatoes in cartons. Transactions of the American Society of Agricultural Engineers 12, 804 806. Wang, N., Zhan g, N., Wang, M., 2006. Wireless sensors in agriculture and food industryRecent development and future perspective. Computers and Electronics in Agriculture 50 (1), 1 14. Want, R., 2006. An introduction to RFID technology. Pervasive Computing (January/Marc h). Watts, A., Miller, P., Godwin, D., 2002. Applications of RFID (Radio Frequency Identification) in Agriculture. Pesticide Outlook (December). Wikipedia, 2010. Bimetallic Strip. Wikimedia Foundation, Inc. Available at: http://en.wikipedia.org/wiki/Bi -metallic_strip#Applications Accessed June 28th, 2010. Wilcke, C., 1992. Model for predicting storability and internal quality of apples. Acta Horticulturae 313, 115124. Wyld, D., 20 06. RFID: The Right Frequency for Government. The IBM Center for the Business of Government. Washington, D C. Zhen -hua, D., Jin -tao, L., Bo, F., 2007. Radio Frequency Identification in Food Supervision. Paper Presented at the 9th International Conference o n Advanced Communication Technology, Gangwon-Do, South Korea. Zwietering, M. H., Wijtzes, T., De Wit, J. C., Vant Riet, K., 1993. A decision support system for prediction of microbial spoilage in foods. Journal of Industrial Microbiology and Biotechnology 12 (35), 324329.

PAGE 223

223 BIOGRAPHICAL SKETCH Cecilia Amador was born in 1980 in Lima, Peru. She attended the Universidad Nacional Agraria La Molina (UNALM), where s h e received her b achelors degree in food i ndustries in 2003. In 2005 she began her M aster o f S cience program at the Agricultural and Biological Engineering Department at the University of Florida, and completed it in December 2006, under the direction of Dr. Jean -Pierre Emond. In order to continue her work with Dr. Emond, she was admitted into t he doctoral program at UF in the winter of 2007. As a graduate student in the UF/IFAS Center for Food Distribution and Retailing (CFDR), she has been involved in projects related to cold chain, RFID technologies and postharvest management. In addition, dur ing her years of graduate school at UF, she has also served as the v ice -p resident of the Gradua te Student Council (GSC), as a s enator in UFs Student Government, and as the president of the Hispanic Graduate and Professional Student Association (HGSA)