• TABLE OF CONTENTS
HIDE
 Front Cover
 Disclaimer
 Metric conversion chart
 Technical report documentation...
 Executive summary
 Table of Contents
 List of Figures
 List of Tables
 Introduction
 Data collection
 Analysis of field data
 Simulation of the I-95 site in...
 Comparisons of field-observed capacities...
 Summary, conclusions and recom...
 Reference






Title: Field data collection and analysis for freeway work zone capacity estimation
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00087019/00001
 Material Information
Title: Field data collection and analysis for freeway work zone capacity estimation
Physical Description: Book
Language: English
Creator: Transportation Research Center
Publisher: University of Florida Transportation Research Center
Place of Publication: Gainesville, Fla.
Publication Date: 2008
 Subjects
Subject: University of Florida.   ( lcsh )
Spatial Coverage: North America -- United States of America -- Florida
 Record Information
Bibliographic ID: UF00087019
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved, Board of Trustees of the University of Florida

Downloads

This item has the following downloads:

FDOT%20BD545%2082%20Final%20Report ( PDF )


Table of Contents
    Front Cover
        Page i
    Disclaimer
        Page ii
    Metric conversion chart
        Page iii
    Technical report documentation page
        Page iv
    Executive summary
        Page v
        Page vi
        Page vii
    Table of Contents
        Page viii
        Page ix
    List of Figures
        Page x
    List of Tables
        Page xi
    Introduction
        Page 1
        Page 2
        Page 3
    Data collection
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
    Analysis of field data
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
    Simulation of the I-95 site in Jacksonville, Florida
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
    Comparisons of field-observed capacities to other techniques
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
    Summary, conclusions and recommendations
        Page 45
        Page 46
    Reference
        Page 47
Full Text




FINAL REPORT


to


THE FLORIDA DEPARTMENT OF TRANSPORTATION
ROADWAY DESIGN OFFICE


on Project


"Field Data Collection and Analysis
for Freeway Work Zone Capacity Estimation"


FDOT Contract BD-545, RPWO #82 (UF Project 67207)


June 25, 2008


The University of Florida









DISCLAIMER

The opinions, findings, and conclusions expressed in this publication are those of the authors and

not necessarily those of the State of Florida Department of Transportation.










METRIC CONVERSION CHART


U.S. UNITS TO METRIC (SI) UNITS


LENGTH
WHEN YOU
SYMBOL W U MULTIPLY BY TO FIND SYMBOL
KNOW

in inches 25.4 millimeters mm

ft feet 0.305 meters m

yd yards 0.914 meters m

mi miles 1.61 kilometers km




METRIC (SI) UNITS TO U.S. UNITS


LENGTH
WHEN YOU
SYMBOL WHENYOU MULTIPLY BY TO FIND SYMBOL
KNOW

mm millimeters 0.039 inches in

m meters 3.28 feet ft

m meters 1.09 yards yd


km kilometers 0.621 miles mi











Technical Report Documentation Page
1. Report No. 2. Government Accession No. 3. Recipient's Catalog No.

4. Title and Subtitle 5. Report Date
June 25, 2008
Field Data Collection and Analysis for Freeway Work Zone
Capacity Estimation 6. Performing Organization Code

7. Author(s) 8. Performing Organization Report No.
Lily Elefteriadou and Kevin Heaslip
9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)


Transportation Research Center 11. Contract or Grant No.
University of Florida FDOT Contract BD-545, RPWO #82
512 Weil Hall, PO Box 116580
Gainesville, FL 32611-6580

12. Sponsoring Agency Name and Address 13. Type of Report and Period Covered
Final
Florida Department of Transportation 14. Sponsoring Agency Code
605 Suwannee Street, MS 30
Tallahassee, FL 32399
15. Supplementary Notes

16. Abstract
A previous FDOT project (FDOT BD 545-51) developed analytical models for estimating the capacity of freeway work zones based entirely on simulated
data. Simulation was used because there were no freeway work zone field data available at the time. As that project was nearing completion, it was determined
that field data could be obtained from a recently installed freeway work zone along 1-95 in Jacksonville, Florida, using the Jacksonville Traffic Management
Center (TMC) cameras. Therefore this project was initiated to collect field data and based on these re-calibrate the models previously developed. To achieve the
objectives of the project, field data were collected at the freeway work zone identified, for a total of 15 days. Various capacity-related measures were obtained
from the field data, including the maximum pre-breakdown flow, the breakdown flow, and the maximum and average discharge flow during congested
conditions. The following were concluded from the research:
The average discharge flow had the smallest standard deviation of all the capacity measures. The other three measures showed higher variability.
The average per lane discharge flow (i.e., capacity during congested conditions) at the study site under good weather conditions was found to be
4,013 vehicles per hour (vph), or 2,007 vehicles per hour per lane (vphpl).
Rainy conditions reduced the average discharge flow in the work zone by 10-29 percent; heavy rain had a much greater impact on capacity than
moderate rain.
It was concluded that the operations model developed under FDOT BD 545-51 can reasonably predict the capacity of the work zone (difference less
than 1 percent) while the planning model under-predicts it by 8.8 percent. The effect of rain is not captured well by either of these models. The
existing FDOT lane analysis method underestimated the discharge flow for that site by approximately 10 percent.
The following are recommended:
The operations model from FDOT BD 545-51 provided good estimates of the discharge flow for the study site over several days of observation. Since
the data were collected only at one site, the model should be evaluated in future freeway construction sites, and adjusted if necessary based on
additional field data, before it is officially adopted by FDOT.
The planning model from FDOT BD 545-51 may also be further tested in future construction sites, however it does not seem to provide results as
close as those of the operations model.

17. Key Word 18. Distribution Statement
Freeway Work Zones, Capacity No restrictions.


19. Security Classif. 20. Security Classif. (of this page) 21. No. of 22. Price
(of this report) Pages
Unclassified Unclassified 58

Form DOT F 1700.7 (8-72) Reproduction of completed page authorized









EXECUTIVE SUMMARY


A previous FDOT project (FDOT BD 545-51), titled "Impact of Trucks on Arterial LOS
and Freeway Work Zone Capacity," developed analytical models for estimating the capacity of
freeway work zones based entirely on simulated data. Simulation was used because there were
no freeway work zone field data available at the time. The models developed estimate the
capacity of various freeway work zone configurations as a function of prevailing traffic, design,
environmental, and work zone characteristics. As that project was nearing completion, it was
determined that field data could be obtained from a recently installed freeway work zone along I-
95 in Jacksonville, Florida, using the Jacksonville Traffic Management Center (TMC) cameras.
Therefore this project was initiated to collect field data to evaluate, and if necessary re-calibrate
the models formulated using simulation.
To fulfill the objectives of the project, the research team collected field data at a freeway
work zone along 1-95 in Jacksonville, Florida, and extracted capacity values for a total of 15
days. Existing TMC cameras were used in conjunction with AUTOSCOPETM devices, which
were installed at the TMC and used to automatically obtain field data. The data collection
consisted of 10 days of lane closures with the left-most lane closed, and 5 days with the right-
most lane closed. Various capacity-related measures were obtained from the field data, including
the maximum pre-breakdown flow, the breakdown flow, and the maximum and average
discharge flow during congested conditions. The field site operations were next simulated using
CORSIM to compare the field results to those obtained by CORSIM, which was the simulator
used in the FDOT BD 545-51 project. The main objective of this analysis was to evaluate the
CORSIM simulator with respect to its ability to replicate work zone operations and assess its
flexibility to evaluate the impacts of various factors that affect work zones. The capacity
estimates from the field data and the CORSIM simulation were compared to the models
proposed in the previous project, as well as to the HCM2000 and the existing FDOT lane closure
methodology. The objective of these comparisons was to determine which models) best
estimated the capacity of the work zone.
The following were concluded from the research:
The average discharge flow had the smallest standard deviation of all the capacity
measures. The other three measures showed higher variability.









The average per lane discharge flow (i.e., capacity during congested conditions) at the
study site under good weather conditions was found to be 4,013 vehicles per hour (vph),
or 2,007 vehicles per hour per lane (vphpl).
Rainy conditions reduced the average discharge flow in the work zone by 10-29 percent;
heavy rain had a much greater impact on capacity than moderate rain.
There was no significant difference in the work zone capacity measures between a left-
and a right-lane closure.
CORSIM appears to over-predict the pre-breakdown flow, while the differences for the
remaining three capacity measures are relatively low. With respect to lane utilization,
CORSIM does not predict very accurately the percent of traffic on each lane, and it does
not have the flexibility to allow the user to enter these as inputs into the simulation for a
particular site.
An operations and a planning model were developed under FDOT BD 545-51. That
project defined capacity as the average discharge flow based on the CORSIM output.
Therefore, that capacity measure from the field data was compared to the results provided
by the two models. It was concluded that the operations model can reasonably predict the
capacity of the work zone (difference less than 1 percent), while the planning model
under-predicts it by 8.8 percent. The effect of rain is not captured well by either of these
models. The existing FDOT lane analysis method underestimated the discharge flow for
that site by approximately 10 percent.


The following are recommended:
The operations model from FDOT BD 545-51 provided good estimates of the
discharge flow for this site over several days of observation. Since the data were
collected only at one site, the model should be evaluated in future freeway
construction sites, and adjusted if necessary based on additional field data, before it is
officially adopted.
The planning model from FDOT BD 545-51 may also be further tested in future
construction sites; however it does not seem to provide results as close as those of the
operations model.
As additional data are gathered, it may be possible to develop and apply a "Rain









Factor" in traffic operations applications around work zones.


The following recommendations are provided regarding possible improvements to
CORSIM with respect to freeway work zone simulation:
The software should consider developing algorithms specifically applying to work
zones, and replicating the use of taper sections.
Guidance should be provided regarding the use of the "rubbernecking" factor and its
relationship to worker and equipment presence in the work zone.
Various geometric elements (such as lane width and shoulder width) are currently not
considered within CORSIM. Its algorithms should be modified to consider such
factors generally, as well as with respect to work zones.










TABLE OF CONTENTS


E X E C U T IV E SU M M A R Y ................................................................................... ..................... v


LIST OF FIGURES .................................. .. .. .... .... ................. .x


LIST OF TA BLES .......... ........... .................................................. .................. xi


1. IN TRODU CTION ............... ................. ........... ................... ...............

1.1 Research Objectives .............................. .. .. .... ... ................ ....
1.2 M methodology .................................................................................. .2
1.2 .1 F field D ata C collection ..................................................... .. ........ .......... .. ....
1.2 .2 D ata A naly sis .......................................................................................... .................... 2
1.2.3 Re-creation of the Field Work Zone in CORSIM ..................................................2
1.2.4 Comparison and Recalibration of Original Models ................................................3

2. D A TA C O L L E C T IO N ............................................................................. ........................ 4

2.1 Study Site D description ....................... ........................... .. ................. .4
2.2 Left Lane Closure D ata Collection ............................................................................ 8
2.3 R ight L ane C losure D ata C collection ......................................................................... ...... 9

3. A N A LY SIS O F FIELD D ATA .............................................................................. ........ 10

3.1 C capacity A analysis M ethodology ........................................................................... ...... 10
3.2 Left Lane Closure Field D ata ............................................ ........... ..... ............... 12
3.2.1 Left Lane Closure Data Collection Summary .....................................................18
3.3 Right Lane Closure Field D ata .................................................. .............................. 20
3.3.1 Right Lane Closure Data Collection Summary.....................................................24
3.4 Summary, Comparisons, and Conclusions from the Data Collection .............................27

4. SIMULATION OF THE 1-95 SITE IN JACKSONVILLE, FLORIDA.............................31

4.1 M odeling of the 1-95 Field Site w ith COR SIM 5.1 ........................................ ..................31
4.2 Simulation Results ........................................ ............. .................. 35
4.3 Comparison of CORSIM to the Field Data ........................................ ...... ............... 38

5. COMPARISONS OF FIELD-OBSERVED CAPACITIES TO OTHER TECHNIQUES .....40

5.1 Comparison of Field Data to FDOT BD 545-51 Models ............................... ...............40
5.2 Comparisons to CORSIM, HCM 2000, and the Existing FDOT Lane Closure









M eth o d o lo g y ...................................... ................................................... 4 3

6. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS................ .............. ....45


R E F E R E N C E S ........................................................ ...................................47












LIST OF FIGURES

Figure 1: Map of Jacksonville Indicating the Work Zone Location ............................................. 4
Figure 2: Data Collection Site Diagram for the Left Lane Closure............................................... 5
Figure 3: Layout of the Study Site for the Left Lane Closure ...................................................... 6
Figure 4: View from the FDOT Camera at the Beginning of the Work Zone............................. 7
Figure 5: Layout of the Study Site for the Right Lane Closure.................................................... 9
Figure 6: Illustration of the Three Parameters on Time Series Plot of Flow and Speed............... 11
Figure 7: Plot of Flow and Speed for 3/27/07....................... .... ............................ 12
Figure 8: Plot of Flow and Speed for 4/5/07....................................................... ... ................. 13
Figure 9: Plot of Flow and Speed for 4/13/07..................................................... ... ................. 13
Figure 10: Plot of Flow and Speed for 4/19/07 ......... .......................................... .............. 14
Figure 11: Plot of Flow and Speed for 4/26/07.......... ........................................ .............. 14
Figure 12: Plot of Flow and Speed for 5/14/07............................ ..... .... .. ............... .. 15
Figure 13: Plot of Flow and Speed for 5/22/07............................ ..... .... .. ............... .. 15
Figure 14: Plot of Flow and Speed for 5/31/07 ........... ........................................ .............. 16
Figure 15: Plot of Flow and Speed for 6/4/07 ........................... ......... ................... 16
Figure 16: Plot of Flow and Speed for 6/14/07.................... ............. ......... ...... ...... ...... .. 17
Figure 17: Histogram of Capacity Measures Observed for the Left-Lane Closure ................. 19
Figure 18: Plot of Flow and Speed for 9/19/07 ......... ........................................ .............. 20
Figure 19: Plot of Flow and Speed for 9/20/07 ............................................................ ..... 21
Figure 20: Plot of Flow and Speed for 9/28/07 ............................................................ ..... 22
Figure 21: Plot of Flow and Speed for 10/2/07 ......... ........................................ .............. 23
Figure 22: Plot of Flow and Speed for 10/3/07........ ............. ................. ................ 24
Figure 23: Histogram of Capacity Measures Observed for the Right-Lane Closure (in vph)...... 26
Figure 24: Histogram of Capacity Measures Observed for Both Work Zone Configurations...... 30
Figure 25: Aerial Photograph of Field Site with Overlay of the Simulated Freeway Network.... 32
Figure 26: The Simulated W ork Zone in CORSIM .............. ................. ................................ 34
Figure 27: Time Series Diagram of Simulated Breakdown in CORSIM.................................. 37












LIST OF TABLES

Table 1: Capacity-Related Measures for Each Breakdown (Left-Lane Closure) in VPH ............ 17
Table 2: Aggregate Capacity-Related Measures for Left-Lane Closure (in VPH) ...................... 18
Table 3: Lane U tilization at D ata Points 2 and 3 ................................................................ 18
Table 4: Capacity-Related Measures for Each Breakdown Observed (Right Lane Closure) in
V P H ............... ... ..... ......... ....... ...... .......... ............. ... ............................ ......... 2 4
Table 5: Aggregate Capacity-Related Measures for Right-Lane Closure, With Rain Events in
V P H ............... ... ..... .......... ..... .... ..... .............................. ... ... .. .............. 2 5
Table 6: Aggregate Capacity-Related Measures for Right-Lane Closure, Without Rain Events in
V P H ........... .. ..... ... .......... ..... ...... ..................... 2 5
Table 7: Capacity-Related Measures for Both Work Zone Configurations ............................. 27
Table 8: Aggregate Capacity-Related Measures for Both Work Zone Configurations, With Rain
E vents (in V PH ) ........... .............. .... ..... .... .......... ........... .......... ........ ............... 28
Table 9: Aggregate Capacity-Related Measures for Both Work Zone Configurations, Without
R ain E vents .... .......................... .... ... ...... ............... .. .......... ......... .......... 28
Table 10: Comparison of Capacity Measures for Rain and Non-Rain Events............................ 29
Table 11: Results of the 1-95 CORSIM Simulation ................................................................... .. 36
Table 12: Summary of the Capacity Measures Obtained From CORSIM (in VPH).................... 38
Table 13: Comparison of Capacity Measures in CORSIM to Field Data (in VPH) ..................... 39
Table 14: Lane Utilization in the Field and in CORSIM .................................. .................. .. 39
Table 15: Comparison of Field Data to FDOT BD 545-51 Models (in VPH)............................ 42
Table 16: Differences Between Operations Model and Field Data ................................. 43
Table 17: Differences Between Field Data, CORSIM, HCM 2000, and the Existing FDOT Lane
C lo su re M eth o d o lo g y .............................................................................................. .............. 4 5









1. INTRODUCTION
Chapter 10 of the FDOT Plans Preparation Manual (PPM) titled "Work Zone
Traffic Control" contains a lane closure analysis procedure (pp. 10-30 10-43) that
calculates the restricted capacity for roadway segments with a lane closure. This
procedure applies capacity reduction and other factors to the basic capacity flow rate to
determine the capacity of the work zone. Comparing this estimated capacity to the
expected hourly traffic demand, restrictions may then be placed on the time of day/night
that a lane can be closed. This capacity estimation procedure was developed
approximately 10 years ago and it is the desire of the Department to evaluate and update
it against more current publications including the Highway Capacity Manual 2000
(HCM2000) and other pertinent recent research.
A previous FDOT project (FDOT BD 545-51), titled "Impact of Trucks on
Arterial LOS and Freeway Work Zone Capacity," developed analytical models for
estimating the capacity of freeway work zones based entirely on simulated data.
Simulation was used because there were no freeway work zone field data available at the
time. The models developed estimate the capacity of various freeway work zone
configurations as a function of prevailing traffic, design, environmental, and work zone
characteristics. As that project was nearing completion, it was determined that field data
could be obtained from a recently installed freeway work zone along 1-95 in Jacksonville,
Florida, using the Jacksonville Traffic Management Center (TMC) cameras. Therefore
this project was initiated to collect field data and based on these re-calibrate the models
formulated using simulation.


1.1 Research Objectives
The objectives of this project are to:
a) Collect field data at a freeway work zone along 1-95 in Jacksonville, Florida,
using the infrastructure of the Jacksonville TMC;
b) Compare the field-measured capacity of the freeway work zone to the
respective capacity estimates from project FDOT BD 545-51; and
c) Compare the field-measured capacity to that obtained by the HCM2000 and the
existing FDOT procedure, and d) Re-calibrate the previously developed models as









necessary, and provide a recommended freeway work zone analysis method.


1.2 Methodology
To fulfill the objectives of the project, four tasks were performed, which are
briefly described in this section.


1.2.1 Field Data Collection
The research team collected field data at a freeway work zone along 1-95 in
Jacksonville, Florida, and extracted capacity values for a total of 15 days. There are
currently two lanes open and one lane closed through the work zone, which is 2.5 miles
long, and will result in the addition of one lane in each direction of the freeway.
Existing TMC cameras were used in conjunction with AUTOSCOPETM devices,
which were installed at the TMC and used to automatically obtain field data. The data
collection consisted of 10 days of lane closures with the left-most lane closed, and 5 days
with the right-most lane closed. Data collection was conducted during non-congested and
congested conditions, and includes a total of 20 transition periods (i.e., breakdowns).
Approximately 45 hours of field data were obtained. This task is described in detail in
Chapter 2.


1.2.2 Data Analysis
Various capacity-related measures were obtained from the field data, including the
maximum pre-breakdown flow, the breakdown flow, and the maximum and average
discharge flow during congested conditions. Statistical analysis verified that the number
of samples obtained were adequate for estimating the mean capacity flows to within 100
vehicles per hour (vph), with a confidence interval of 95 percent. The data analysis is
presented in Chapter 3.


1.2.3 Re-creation of the Field Work Zone in CORSIM
The field site operations were next simulated using CORSIM to compare the field
results to those obtained by CORSIM, which was the simulator used in the previous
FDOT project. The main objective of this analysis was to evaluate the CORSIM









simulator with respect to its ability to replicate work zone operations and assess its
flexibility to evaluate the impacts of various factors that affect work zones (such as left
vs. right-lane closures). The simulation effort is summarized in Chapter 4.


1.2.4 Comparison and Recalibration of Original Models
The capacity estimates from the field data and the CORSIM simulation were
compared to the models proposed in the previous project, as well as to the HCM2000 and
the existing FDOT lane closure methodology. The objective of these comparisons was to
determine which models) best estimated the capacity of the work zone. Based on these
findings, a final lane closure methodology for freeways is proposed. Chapter 5 presents
the comparisons and discusses the results, while Chapter 6 presents the project reports
and recommendations.










2. DATA COLLECTION

This section describes the study site selected for data collection and summarizes

the data collection activities.



2.1 Study Site Description

The criteria for selecting a suitable freeway work zone were as follows:

1) A work zone with a lane closure;

2) Demands that exceed capacity for a significant amount of time each day, so that
capacity can be measured; and

3) Existing cameras in the vicinity of the work zone that would allow for data
collection at the approach to the work zone and within the bottleneck.


The selected site is located on the north side of Jacksonville, along 1-95

Northbound, approaching the Trout River Bridge widening/bridge reconstruction project.

Figure 1 shows the location of the work zone in the Jacksonville area.


Eiid of \oik ZonIe
IPP
:Si

Beuiiniinu of \\oik Zone

Jacksonville


ig



134 -irma~~


~~1"


IA..A hl

(DA


bEm ic


r. "i


a ,- -
^ ; *',,-. '
, *: *


Figure 1: Map of Jacksonville Indicating the Work Zone Location



Figure 2 provides a diagram of the site indicating the data collection points, while


[


c


i' __ .


I


CD1:









Figure 3 shows a more detailed sketch of the geometry of the site and data collection
points (squares in the travel lanes) during the left lane closure. Data collection point 1 is
located at the beginning of the work zone and was used to provide discharge flows (i.e.
capacity flows) from the work zone.




Data Collection Poi II i
(AUTOSCOPE '' \
Oatewway Shopping Oenter

End of Taper


Beginning of Ta pe i 4




Data Collection Point #2 i
(FDOT Station) S ,
9thSt 5
W2,

BrenpUwood Park
W 24th St
I W 239h S1
2bndSt
W 21sA S4


W l8thSt
Data Collection Point #3
(FDOT Station) 1 4. W iSth
r S ite Di m for Ct

Figure 2: Data Collection Site Diagram for the Left Lane Closure











Interstate 95 Northbound Jacksonville, FL


14 Mile I V4 Mile 3/4 Mile I

Figure 3: Layout of the Study Site for the Left-Lane Closure


The data at point 1 were collected via the SUNGUIDE ITS camera system in
Jacksonville's TMC. Data were collected using existing camera equipment along the 1-95
corridor. Cameras are located approximately every 3A mile, and they have the ability to
pan, tilt and zoom. To automate the data collection and reduction, an AUTOSCOPETM
data collection device was installed at the Jacksonville TMC and used to process video
coming in from the TMC cameras. Figure 4 shows a snapshot of the camera view used to
collect speed and volume data. The other two data collection points are FDOT permanent
count stations, which provide speed and flow upstream of the work zone. Data collection
point 2 is located at the first orange warning sign indicating a work zone ahead (1 mile
upstream of the beginning of the taper to the work zone). Data collection point 3 is
located after the first work zone notification sign (variable message sign), approximately
34 of a mile from point 2. Data were collected at these two locations because the speed
and volume data provide lane utilizations and information regarding queue presence,
which are necessary in simulating this site and comparing the performance in the field to
simulation.
The posted speed limit on the freeway in advance of the work zone is 60 miles per
hour (mph). Free-flow speeds during off-peak time periods were found to range between
65 mph and 75 mph. In addition, the area where the work zone is located regularly
experienced congestion even prior to the work zone installation. During the study period,
the work zone experienced heavy congestion during weekday afternoon peak periods and
it was not uncommon for congested conditions to persist for several hours.

































Figure 4: View from the FDOT Camera at the Beginning of the Work Zone


The work zone had the following geometric and operational characteristics during
the observation periods:

There was an eight-foot shoulder on the right and a three-foot shoulder on the left
of the freeway mainline;
There was no presence of police other than routine patrols of the site;
The lane widths were at least 12 ft throughout the segment;
Truck percentages averaged about 5 percent;
The data were collected during weekdays, when commuter traffic is expected to
be present at the site.

Data were collected in the spring and fall of 2007, for a total of 15 weekdays and
for several hours each day. Of these, 10 days of data collection were completed when the
work zone had a left lane closure and 5 weekdays when the work zone was reconfigured
as a right lane closure. For this later one, the taper section approaching the work zone was
shorter than that of the left lane closure. During the data collection effort, the two









permanent count stations (Data collection points 2 and 3) were functional only during
two days; therefore adequate data could not be collected, and no relationship could be
established between capacity measures and lane distributions. However, the data obtained
for those two days provided some information on the lane distribution upstream of the
work zone. To supplement the information provided by data points 2 and 3 regarding
queue lengths upstream of the work zone, the area was visually inspected during the data
collection using the TMC cameras upstream of the work zone. The data collection is
discussed in the following sections.


2.2 Left-Lane Closure Data Collection
Data were collected over several hours during the PM peak period over 10 days
(from 3/27/2007 to 6/14/2007). The data collection periods ranged from four to six hours
per day. The data were obtained using the AUTOSCOPETM device and closely examined
for erroneous detector readings. The AUTOSCOPETM provided volumes and speeds at
data collection point 1. A test was conducted to compare the results from the
AUTOSCOPETM to manual data collection. Researchers conducted manual counts from
recorded data for two hours to determine the accuracy of the cameras. It was found that
the accuracy of the AUTOSCOPETM was within 2 percent of the hourly flows. The field
data were entered into a spreadsheet, which includes the speed, volumes, and flow rates
for each five minute period of data collection.
Most data collection periods began around 3:00 p.m. and lasted until the
congestion subsided. Typically breakdowns (i.e., the beginning of the congested period)
occurred sometime between 3:00 and 6:00 p.m. There were three data collection periods
where congestion started before the data collection period began. The data collected
includes 14 congested time periods, for which breakdown and the transition to congested
flow was observed 11 times. For most data collection periods, congestion started at the
bottleneck, i.e., immediately upstream of the two lane section. Two occurrences were
notable in the data collection: a) On one day, congestion started within the work zone and
extended upstream; b) On another day there was heavy rain that caused congestion. The
other 12 breakdowns were caused by demand exceeding the capacity of the work zone. A
summary of the field data collected and the capacities measured is provided in Chapter 3.










2.3 Right-Lane Closure Data Collection
Data collection continued at the same location in the fall of 2007 when the right
most lane of the same site was closed. Figure 5 illustrates the geometry of the site along
with the data collection points indicated by squares in the travel lanes. The collection
methods for these data were identical to the methods used for the left lane closure and
described above. Data were again collected during the PM peak period for 5 days (from
9/19/2007 to 10/3/2007) and for time periods ranged from four to six hours per day.
Typically, data collection started around 3 PM, since congestion occurred between
3:00 PM and 6:00 PM. There were three data collection periods where the congestion
occurred before the data collection period began. The data collected include six
breakdowns. During one data collection period, there was very heavy rain, with 4.55
inches (recorded at the National Weather Service at Jacksonville Naval Air Station;
Source: www.weatherunderground.com) during that day. The heaviest rain occurred
during the data collection and precipitated the breakdown. The remaining five
breakdowns were caused by demand exceeding capacity at the work zone. A summary of
the field data collected and the capacities measured is provided in Chapter 3.




Interstate 95 Northbound Jacksonville, FL






4-- 1/4 Mile -1/2 Mile -~/2Ml 3/4 Mile


Figure 5: Layout of the Study Site for the Right-Lane Closure









3. ANALYSIS OF FIELD DATA
This section describes the methodology used to obtain capacity values from the
field data, and presents the data obtained for each day of the data collection.


3.1 Capacity Analysis Methodology
Five-minute intervals were used in the data analysis. For each day of data
obtained, time series plots of flow and speed were generated to identify the breakdown
(i.e., the transition from non-congested to congested conditions), and to obtain the
maximum flows measured at various times throughout the data collection. The data
analysis examined the flows collected before, during, and after the transition from non-
congested to congested flow, because previous research has shown that the maximum
flow at a particular bottleneck may occur during any one of these three time periods.
Furthermore, the actual numbers of the maximum flows vary from day-to-day, therefore
it was important to obtain field data over several days.
In this project the analysis was conducted in accordance to a procedure detailed
partly by Elefteriadou and Lertworawanich (2002). The following process was
undertaken:

1. Identify and quantify each transition interval from non-congested to congested
flow, i.e., breakdown event, and document the corresponding breakdown flow.
2. Identify and document the maximum pre-breakdown flow.
3. Identify and document the maximum discharge flow. This flow is the maximum
observed at the site after the occurrence of breakdown, and prior to recovery to
non-congested conditions.
4. Identify and document the average discharge flow. This flow is the average
observed at the site between the start and end of congestion.


Figure 6 presents a time-series plot of speed and flow during a breakdown event
at a ramp merge junction, and illustrates the first three variables defined above. The
purple line (scale on the right of the graph) shows speed dropping sharply around 19:15
PM, indicating breakdown, and the beginning of congestion. The blue line (scale on the
left of the graph) shows the corresponding average flows for each time interval. Similar
plots were developed for the data obtained in this project to obtain the four flows defined
above.














10000 (5'540 120
vc-h h,
,_o_ l110
9000 110
100
8000

7000
980
5 6ooo 70

40
5000-4P 60

o 4000 laxini F)r. P Erea6 jowCi,5
Flcm R/ ie40
3000 7 i''P72 wnh nM

2000 20

1000 10



S.& ,3' "' A .,* ^ ,,.^. Pd "> ,t^ "' ,' "r (& "b .r 'r "> .,'' i^ f "o '

Time

Figure 6: Illustration of the Three Parameters on Time Series Plot of Flow and Speed
(Source: Elefteriadou and Lertworawanich, 2003)


For this project breakdown was defined as 15 minutes (3 five-minute

observations) of speeds below 55 mph. This value was chosen because it was observed

that when the speed was below 55 mph queuing was present. This also represented a 10-

15 mph drop in speed from the free flow speeds of 65-70 mph. Return to non-congested

conditions was defined to occur when the average speed reached 55 mph and remained

there for at least 15 minutes. To obtain capacity for the study site, each of these four flow

rates (the maximum pre-breakdown, the breakdown flow, the maximum discharge flow,

and the average discharge flow) were obtained for each breakdown event. Average values

of each of these measures were obtained separately for the left-lane closure and the right-

lane closure work zone. The remainder of this chapter presents the field data analysis

summary, while comparisons to other methods and techniques are discussed in Chapter 5.












3.2 Left Lane Closure Field Data

This section presents the analysis of the field data collected when the left lane of

the study segment was closed. These consist of capacity-related measures at the work

zone (data collection point 10) and lane utilization data collected upstream (data

collection points 2 and 3). First, time series plots of the data are presented for each day of

data collection, followed by lane utilization data. A summary of the field data is

presented at the end of the section.

In Figures 7 through 16 the time series plot of flow vs. speed is presented for each

of the data collection periods. As shown in Figure 7, on March 27, 2007, congestion

began just after 3 PM and continued until approximately 6 PM. Visual inspection by the

research team using TMC cameras indicated that at 5:00 PM the queue extended more

than a mile upstream.



3/27/07 1-95 Jacksonville WZ Both Lanes

5000 70 00
65 00
4500 65
*. 6000
4000 '. 5500
4000 ".5
3500 ...., .-- 5000
.... / .. ^ -' "'' 4500
3000 4500
4000
2500 35 00
30 00
2000
i-I i1 II 1 25 00
1500 -2000
1000- 1500
1000
500
5 00
0 000
305PM 325PM 345PM 405PM 425PM 445PM 505PM 525PM 545PM 605PM
Time

Figure 7: Plot of Flow and Speed for 3/27/07


The data shown in Figures 8, 9, and 10 do not include a breakdown, as it occurred

before the beginning of the data collection. Thus the data in Figure 8 all represent

congested conditions and visual inspection from TMC cameras indicated that the queue

extended more than a mile upstream at 4 PM. During the data collection on April 13

(Figure 9), a backup from within the work area occurred around 5:30 PM resulting in a

significant speed drop upstream. The event was not visible on the cameras, but there was

significant queuing from within the work zone. On April 19 (Figure 10) there was a speed














drop at 3:45 PM. It was not visible from the TMC cameras what caused it, but it is

possible that it was triggered by significant activity within the work zone. The speed

flow pattern that day was unusual in that there are noticeable speed and flow drops


throughout the data collection period. Visual inspection using TMC cameras indicated

that the queue extended more than a mile upstream at 4 PM.


4/5/07 1-95 Jacksonville WZ Both Lanes

4500 55

~ A~ 50
4000 _, 5"
--. 45
3500 *... ,
40
3000
35
6 2500 30

o 2000 :1 -n, 25
20
1500
15
1000
10
500 5

0 0
310PM 330PM 350PM 410PM 430PM 450PM 510PM 530PM 550PM 610PM 630PM
Time




Figure 8: Plot of Flow and Speed for 4/5/07

4/13/07 1-95 Jacksonville WZ Both Lanes

6000 60


5000 -50


4000- ---- 40
\ ** > '-- .. '" .- "

S3000 30


2000 -20


1000 10


0 0
345PM 405PM 425PM 445PM 505PM 525PM 545PM 605PM 625PM
Time


Figure 9: Plot of Flow and Speed for 4/13/07














4/19/07 1-95 Jacksonville WZ Both Lanes


'IIb


350PM


420PM


450PM


~*. ....
*%


520PM


550PM


Figure 10: Plot of Flow and Speed for 4/19/07


On April 26 (Figure 11) congestion began about 3 PM and recovered around 3:30


for a short time. After 4 PM another breakdown occurred, and congested conditions


continued through the peak hours until 6 PM. Visual inspection using TMC cameras


indicated that the queue extended more than a mile upstream at 5 PM.





4/26/07 1-95 Jacksonville WZ Both Lanes


p'
q d *. *.


A A r--r-. -


.... .
~c...-4
V -.


''-I II I ,,rI,,,,r, II,,,


235PM 250PM 305PM 320PM 335PM 350PM 405PM 420PM 435PM 450PM 505PM 520PM 535PM 550PM
Time


Figure 11: Plot of Flow and Speed for 4/26/07




Congestion on May 14 (Figure 12) began about 3:45 PM, recovered around 4:30


5000 -

4500

4000

3500

3000

2500

2000

1500

1000

500


0


320PM


5000

4500

4000

3500

3000

2500

2000

1500

1000

500


+ :l)l: I:1 I: 1: + -i'lllh,:h I1:) ,!1"::1


,
..
/ .














until just before 5 PM, and then another breakdown occurred with congestion lasting until


6 PM.


5/14/07 1-95 Jacksonville WZ Both Lanes


255PM


325PM


355PM


425PM


455PM


525PM


Time


Figure 12: Plot of Flow and Speed for 5/14/07


On May 22 (Figure 13) a breakdown occurred later than in previous days (about


4:45 PM) because demand that day was much lighter.


5/22/07 1-95 Jacksonville WZ Both Lanes


3 35 PM 3 50 PM 4 05 PM


4 20 PM 4 35 PM
Time


4 50 PM 5 05 PM


5 20 PM 5 35 PM


Figure 13: Plot of Flow and Speed for 5/22/07





On May 31 (Figure 14) there were two breakdowns. The first one occurred around


4:15 PM with a recovery around 5 PM. A second breakdown occurred around 5:15 PM,


5000

4500

4000

3500

3000

S2500
0-
2000

1500

1000

500

0


-. -'


ii I ,,ri,,,.-r, ii,,, i


* ~


-w .


-
.c~


~-c
,


I


.- ... ,.".














with congestion lasting until 6 PM.


5/31/07 1-95 Jacksonville WZ Both Lanes
5000 80

4500 -
A 70
4000.

3500

3000 -.- "- 50

j2500 40

2000

1500
20
1000

500 10

0 0
220PM 240PM 3 00 PM 3 20 PM 3 40 PM 4 00 PM 4 20 PM 4 40 PM 5 00 PM 5 20 PM 5 40 PM 6 00 PM
Time


Figure 14: Plot of Flow and Speed for 5/31/07




Similarly to the data collected on May 22, congestion on June 4 (Figure 15) began

about 4:45 PM and recovered by 5:30 PM. The demand on this day was much lighter


than previous days.


6/4/07 1-95 Jacksonville WZ Both Lanes

5000 80

4500
.. 70
4000 .. ... .-- -,....- .. .
-. ,60
3500 ,

3000 50

2500 40

L 2000 30

1500
-. -, 1 z- I r- 1-n l 1 -. -- I 20
1000

500 10

0 0
2 00 PM 2 20 PM 2 40 PM 3 00 PM 3 20 PM 3 40 PM 4 00 PM 4 20 PM 4 40 PM 5 00 PM 5 20 PM
Time


Figure 15: Plot of Flow and Speed for 6/4/07





On June 14 (Figure 16) congestion began before 4 PM. That day there was heavy

rain which slowed the traffic down considerably during the data collection.












6/14/07 1-95 Jacksonville WZ Both Lanes


-. *v
/ -S ..


.4,


3 35 PM


4 05 PM


Figure 16: Plot of Flow and Speed for 6/14/07



Table 1 presents the four flow values for each of the 14 breakdown events

observed during the data collection. The pre-breakdown and breakdown flow rates were

not available (denoted by N/A) for three of the breakdowns, as discussed above (Figures

7-9).


Table 1: Capacity-Related Measures for Each Breakdown (Left-Lane Closure) in VPH

iL Pre- lBriakdm n oii NLi\ Diischary'e .A ei'r;l' Discharlie
Breilkdoi n Dalc Brialikdom n F FI Fo
Flow Flow Flmo

1 3/27/07 3078 4020 4548 4040
2 4/5/07 N/A N/A 4272 3910
3 4/13/07 N/A N/A 4944 4004
4 4/19/07 N/A N/A 4452 3950
5 4/26/07 3768 3636 4236 3939
6 4/26/07 3648 4452 4368 4079
7 5/14/07 4224 4500 4236 4068
8 5/14/07 4188 4500 4524 4180
9 5/22/07 4188 4092 4272 4228
10 5/22/07 4320 4140 4488 4035
11 5/31/07 4116 4416 4140 3830
12 5/31/07 4140 3708 3960 3584
13 6/4/07 3996 4008 4140 4036
14 6/14/07 3912 3672 4344 3899



Table 2 presents the minimum and maximum values, as well as the mean and

standard deviation for each capacity measure for a 5-min aggregation interval. The


5000


4500

4000

3500

S3000
S2500
S2000

1500


A".


W p


.,A -


2 35 PM


3 05 PM


4 35 PM


5 05 PM


5 35 PM










maximum discharge flow rates were the highest, with the breakdown flows the lowest.

The average discharge flow values had the lowest standard deviation (i.e., lowest

variability). The pre-breakdown flows had the highest range at 1350 vph, while the

breakdown flows had a range of about 850 vph.



Table 2: Aggregate Capacity-Related Measures for Left-Lane Closure (in VPH)

Mx P Brcakdo i n Mi NI\ Dicliharll' A cra~ie Dicharlrc
Brellkdoi n Br.akl i dow n
BFIo Fl~o FIoi
FloFr
# of Observations 11 11 14 14
Min Value 3078 3636 3960 3584
Max Value 4320 4500 4944 4228
Mean 3962 4104 4352 3985
Std. Dev. 357 334 237 157


In addition to field data collected at the work zone, a limited amount of data were

collected upstream (data collection points 2 and 3). Table 3 summarizes the lane

utilizations observed during two of the days of data collection. As shown, lane utilization

is very similar between those two days; therefore it is assumed that these numbers are

fairly typical for this particular work zone configuration. In a subsequent chapter of this

report, these numbers are compared to the respective numbers observed in CORSIM.


Table 3: Lane Utilization at Data Points 2 and 3

Left Lane Center Lane Right Lane
Field data (03-27-07) 21% 45% 34%
Field data (04-05-07) 19% 45% 36%


3.2.1 Left Lane Closure Data Collection Summary

Figure 17 provides a summary histogram of the four capacity measures observed

at the study site. The maximum pre-breakdown flows seem to be generally lower than the

maximum discharge flows. Both resemble a normal distribution, with the maximum pre-

breakdown flow centering around 4,250 vph and the maximum discharge flow centering

around 4,500 vph. The average discharge flow has a maximum value of about 4,250 vph.
















Flow Rate Frequency
9
Max. Pre Breakdown Flow

8 Breakdown Flow

Max Discharge Flow
7
Ave. Discharge Flow

6


U5


4













<3000 3000-3249 3250-3499 3500-3749 3750-3999 4000-4249 4250-4499 4500-4749 >4750

Flow Rate (VPH)


Figure 17: Histogram of Capacity Measures Observed for the Left-Lane Closure













3.3 Right Lane Closure Field Data

The results of the right lane closure data collection and capacity analysis are

presented in this section. In Figures 18 through 22 the time series plot of flow vs. speed is

presented for each of the data collection periods. As shown in Figure 18, on September

19, 2007, congestion began just after 5:15 PM and continued until approximately 6 PM.

Visual inspection using TMC cameras indicated that at 5:30 PM the queue extended

approximately one mile upstream.




9/19/2007 1-95 Jacksonvlle Both Lanes

4500 70


4000
60

3500

50
3000


S2500 40
I -.-FlowRate
: Speed
S2000 30
30


1500
20

1000

10
500


0 0
40000PM 41500PM 43000PM 44500PM 50000PM 51500PM 53000PM 54500PM 60000PM 61500PM
Time of Day


Figure 18: Plot of Flow and Speed for 9/19/07













On September 20 (Figure 19) congestion began about 4:45 PM and recovered

around 5:00 PM for a short time. After 5:15 PM, another breakdown occurred, and

congested conditions continued through the peak hour until 5:45 PM. Between

breakdowns there was a significant decrease in the amount of demand indicating that

there may have been an incident upstream that caused demand at the study site to drop in

between the breakdowns. Visual inspection using TMC cameras indicated that the queue

extended more than a mile upstream during the second breakdown.




9/20/2007 1-95 Jacksonville Both Lanes

5000 70


4500
60
4000


3500 50


3000
40

2500 -.-Flow Rate
2500
'Speed
30"
2000


1500 20


1000
10
500


0 . 0
20000 21500 23000 24500 30000 31500 33000 34500 40000 41500 43000 44500 50000 51500 53000 54500
PM PM PM PM PM PM PM PM PM PM PM PM PM PM PM PM
Time of Day


Figure 19: Plot of Flow and Speed for 9/20/07












On September 28 (Figure 20) congestion began about 3:30 PM and recovered

around 5:00 PM for a short time. The breakdown was nearly continuous with two near-

recoveries between 4:15 and 4:30 that were not sustained. Visual inspection using TMC

cameras indicated that the queue extended more than three miles upstream during the

congested period.


9/28/2007 1-95 Jacksonville Both Lanes

5000 70

4500
60
4000

3500 50


3000
40
E -FlowRate
2500




1500 20

1000
10
500


23000PM 24500PM 30000PM 31500PM 33000PM 34500PM 40000PM 41500PM 43000PM 44500PM 50000PM
Time of Day


Figure 20: Plot of Flow and Speed for 9/28/07



On October 2 (Figure 21) there was very heavy rain in the Jacksonville area with

4.55 inches of rain recorded for that day at the National Weather Service Station at

Jacksonville Naval Air Station (source: www.wunderground.com). The rain was constant

throughout the day and affected operations within the work zone. At three times during

the data collection (3:00, 5:15, and 6:30) very heavy rain significantly affected the flow

and speed within the work zone. Demand was much lower than usual as motorists

avoided the roadways. A breakdown occurred just after 5:00 PM and continued until 6:15












PM with one very short recovery resulting by a break in the rain. Visual inspection using

TMC cameras indicated that the queue extended more than one mile upstream during the

congested period.




10/2/07 1-95 Jacksonville Both Lanes

4000 70


3500
60


3000
50

2500
40
S2000 Flow Rate
2000
~ -Speed
30
1500

20
1000


10


23000 24500 30000 31500 33000 34500 40000 41500 43000 44500 50000 51500 53000 54500 60000 61500 63000
PM PM PM PM PM PM PM PM PM PM PM PM PM PM PM PM PM
Time of Day


Figure 21: Plot of Flow and Speed for 10/2/07




On October 3 (Figure 22) congestion began about 5:15 PM and recovered around

5:30 PM. Traffic demand was not as high as many motorists were still recovering from

the effects of the heavy rain and flooding of the previous day. Visual inspection using

TMC cameras indicated that the queue extended approximately a mile upstream at 5:20

PM.












10/3/2007 1-95 Jacksonvlle Both Lanes


13000 14500 20000 21500 23000 24500 30000 31500 33000 34500 40000 41500 43000 44500 50000 51500 53000
PM PM PM PM PM PM PM PM PM PM PM PM PM PM PM PM PM
Time of Day


Figure 22: Plot of Flow and Speed for 10/3/07



3.3.1 Right Lane Closure Data Collection Summary

The results of the data collection from the right lane closure are presented in Table

4. Table 4 presents the four capacity measures for each breakdown observed during the

data collection. A total of 6 breakdown events were observed. Breakdown 5 is shown in

bold because it occurred during heavy rain conditions.


Table 4: Capacity-Related Measures for Each Breakdown Observed (Right Lane Closure)
in VPH

Bre.ivado n 1 Aia\ Pre- Brnvahido n Mait Dichariir A e. Dichainr,
E lenl Br.cekdoi n FIhl FloiF FloI FloI
1 9/19/2007 3912 4092 4044 3904
2 9/20/2007 4128 4236 4416 4099
3 9/20/2007 3924 4452 4260 4074
4 9/28/2007 4284 4044 4296 3944
5 10/2/2007 3672 3504 3624 2863
6 10/3/2007 4308 4428 4260 4014


5000 70

4500
60
4000





40
2500

30 '
2000

1500 20

1000
10












Table 5 presents the minimum and maximum values, as well as the mean and

standard deviation for each capacity-related measure for a 5-min aggregation interval for

all days of data collection with a right-lane closure (including the day with heavy rain).

The standard deviation for the average discharge flow is relatively high because of the

inclusion of the relatively low values observed during the heavy rain. Since this heavy

rain is a rare event, subsequent analyses are conducted separating the rain-related data

from the other capacity values.



Table 5: Aggregate Capacity-Related Measures for Right-Lane Closure, With Rain Events
in VPH
Max Pre-
Max P Breakdown Max Discharge Ave. Discharge
BraIldo(mn n
Flom Flom Flom
Flom
# of Observations 6 6 6 6
Min Value 3672 3504 3624 2863
Max Value 4308 4452 4416 4099
Mean 4038 4126 4150 3816
Std. Dev. 247 348 284 473


Table 6 presents the same information as Table 5 but excluding the rain-related

data. The average max discharge flow rates was the highest (4255 vph), with the average

discharge flow the lowest (4007 vph). The average discharge flow values had the lowest

standard deviation (i.e., small variability). The breakdown flows had the highest

variability and range between the minimum and maximum values.



Table 6: Aggregate Capacity-Related Measures for Right-Lane Closure,
Without Rain Events in VPH
Max Pre-
Breakdown Max Disharge Ave. Discharge
Flohm Flo Flohm
Flom
# of Observations 5 5 5 5
Min Value 3912 4044 4044 3904
Max Value 4308 4452 4416 4099
Mean 4111 4250 4255 4007
Std. Dev. 189 187 134 83







The data collected meets the required sample size, which was estimated to be 10 samples
(for a confidence level of 95 percent and an acceptable error of +100 vehicles), which is
lower than the actual number of samples obtained (20 samples).
Figure 23 provides a histogram of the four flow parameters observed in the work
zone (which correspond to the values presented in Table 4).


Flow Rate Frequency


Max. PreBreakdown
4 U Preakdl'v.n. Fln,..
3.5
r [h lt JuitFl",',


3


2

1.5


0. s

<3000


3000-3249 3250-3499


- -_ -


11 1111


3500-3749 3750-3999 4000-4249
Flow Rate (VPH)


4250-4499 4500-4749


Figure 23: Histogram of Capacity Measures Observed for the Right-Lane Closure (in
vph)


>4750


tt~










3.4 Summary, Comparisons, and Conclusions from the Data Collection

The results of the entire data collection are presented in Table 7. A total of 20

breakdown events were observed during the data collection. The shaded entries represent

the right-lane closure data collection. Breakdowns 12 and 19 are in bold because of rain

events that occurred on those two days. Breakdown 12 occurred in a light to moderate

rain and breakdown 19 occurred during heavy rain.

Table 7: Capacity-Related Measures for Both Work Zone Configurations
Max Pre-
Breakdown Max Pre Breakdown Max Discharge Ave. Discharge
Eicn Date Bieal.l(mn lo Fn
E 'II hr Flh Floh Floh
1 3/27/2007 3078 4020 4548 4040
2 4/5/2007 N/A N/A 4272 3910
3 4/13/2007 N/A N/A 4944 4004
4 4/19/2007 N/A N/A 4452 3950
5 4/26/2007 3768 3636 4236 3939
6 4/26/2007 3648 4452 4368 4079
7 5/14/2007 4224 4500 4236 4068
8 5/14/2007 4188 4500 4524 4180
9 5/22/2007 4188 4092 4272 4228
10 5/22/2007 4320 4140 4488 4035
11 5/31/2007 4116 4416 4140 3830
12 5/31/2007 4140 3708 3960 3584
13 6/4/2007 3996 4008 4140 4036
14 6/14/2007 3912 3672 4344 3899
15 9/19/2007 3912 4092 4044 3904
16 9/20/2007 4128 4236 4416 4099
1" 20 ii" '1 24 4452 42 ;'10 40"4
Is i 2 21' 4214 41,44 42' 3''44
1'9 111/2,211117 36'2 35114 3624 2863
20 10/3/2007 4308 4428 4260 4014



Table 8 presents the minimum and maximum values, as well as the mean and

standard deviation for each flow parameter for a 5-min aggregation interval including the

days with the rain events. The average max discharge flow rates were the highest, with

the average discharge flows the lowest. The maximum discharge flow values had the

lowest standard deviation (i.e., small variability).










Table 8: Aggregate Capacity-Related Measures for Both Work Zone Configurations, With
Rain Events (in VPH))
Max Pre-
Breakdown Max Disharge Ave. Discharge
Biv akdo n
Flo(m Flo( Flo(
Flom
# of Observations 17 17 20 20
Min Value 3078 3504 3624 2863
Max Value 4320 4500 4944 4228
Mean 3989 4112 4291 3934
Std. Dev. 316 328 262 286



Table 9 presents the minimum and maximum values, as well as the mean and

standard deviation for each flow parameter for a 5-min aggregation interval, without the

days with the rain events. The average maximum discharge flow rates were the highest,

with the average pre-breakdown flows the lowest. The average discharge flow values had

the lowest standard deviation (i.e., small variability). Table 10 presents the comparison

of the capacity measures for rain and non-rain events. The table shows that the rain

events have lower values than the non-rain events for all capacity measures.



Table 9: Aggregate Capacity-Related Measures for Both Work Zone Configurations,
Without Rain Events
Ma,\ Pir.-
Max Pn.- Brciakdoi n MNla Dikchairin A.s c. Dischainr-e
BiFl(m Floh Flo(
Flo(
# of Observations 15 15 18 18
Min Value 3078 3636 4044 3830
Max Value 4320 4500 4944 4228
Mean 4000 4179 4347 4013
Std. Dev. 325 283 202 102











Table 10: Comparison of Capacity Measures for Rain and Non-Rain Events
Rain Events Summary
M;i\ P,'e-
SPe- BreaLkIdo n .A e. Didchaire
Br'eakdomi n M!l l\ Disclhilre Flow
FIoiw FlIo
Fhl F
# of Observations 2 2 2 2
Min Value 3672 3504 3624 2863
Max Value 4140 3708 3960 3584
Mean 3906 3606 3792 3224
Std. Dev. 331 144 238 510
Non-Rain Events Summary
Max Pre-
Max Pre- Breakdown Ave. Di schiarie
Br'eakdomin ii \I;I\ Discharl-re Flow
FIoiw Fhlo
Fhl F
# of Observations 15 15 18 18
Min Value 3078 3636 4044 3830
Max Value 4320 4500 4944 4228
Mean 4000 4179 4347 4013
Std. Dev. 325 283 202 102


A statistical comparison was conducted to determine whether the four capacity

measures are affected by the side of the work zone closure (left vs. right). The two

sample t-test assuming unequal variances was used at the 95 percent confidence level.

Based on this test it was concluded that there is no significant difference in capacity

measures between the left- and right-lane closures.

The total sample size for rain events was too small (only two data points) to

conduct a meaningful statistical test between two samples. A statistical test (t-test of

means for one sample) was conducted to compare the average values for the four

capacity-related measures during non rain events to the corresponding average values

during the rain events. It was concluded that there is a significant difference in three of

the four capacity measures at the 95 percent confidence level between the rain and non-

rain events. The only capacity measure that was not shown to be statistically different

was the pre-breakdown flow.

Figure 24 provides a histogram of the four capacity measures observed in the

work zone. The maximum discharge flow resembles a normal distribution centered at

4500 vph. The breakdown flows also approximate the normal distribution centered at

4250 vph.










Flow Rate Frequency


< 3000 3000-3219 3250-3199 3500-3719 3750-3999 4000-1219 4250-1499 4500-4719 >1750
Flow Rate (VPH)

Figure 24: Histogram of Capacity Measures Observed for Both Work Zone
Configurations


The following conclusions can be drawn from the data collection:

* An adequate sample size was collected to estimate four different capacity
measures at a work zone in Jacksonville, FL;

* There was no significant difference in the capacity measures between the left and
right lane closures;

* Rainy conditions significantly reduced three of the four capacity measures in the
work zone:
o Breakdown flow (from 12 to 17 percent for the moderate and heavy rain
respectively)

o Maximum Discharge Flow (from 9 to 17 percent for the moderate and
heavy rain respectively)

o Average Discharge Flow (from 10 to 29 percent for the moderate and
heavy rain respectively)









The average discharge flow had the smallest standard deviation of all the capacity
measures. The other three measures showed higher variability.


4. SIMULATION OF THE 1-95 SITE IN JACKSONVILLE, FLORIDA
The field site operations were next simulated using CORSIM to compare the field
capacity measures to those obtained by CORSIM, which was the simulator used in FDOT
Project BD-545-51. The main objective of this analysis was to evaluate the CORSIM
simulator with respect to its ability to replicate work zone operations and assess its
flexibility to evaluate the impacts of various factors that affect work zones (such as left
vs. right-lane closures). This section first describes the replication of the study site in
CORSIM, the simulation approach used, and summarizes the assumptions employed.
Next, it discusses the results of the simulation. The last section compares the field
capacity estimates to those provided by CORSIM.


4.1 Modeling of the 1-95 Field Site with CORSIM 5.1
The software package CORSIM was used to simulate the study site, since that
was the simulator selected in FDOT Project BD-545-51. The software, originally
developed by FHWA, has been widely used and validated in the past twenty years, and it
is available to the University of Florida through McTrans, allowing for a high level of
software support in understanding the software's algorithms. This section discusses how
CORSIM was utilized to recreate the field site, the modeling assumptions employed, as
well as the number of runs required per scenario.
There is no explicit simulation of a work zone in FRESIM (which is the freeway
analysis component of CORSIM); instead, there are two alternative techniques that allow
FRESIM to approximate a work zone lane closure. The first of these is identified as a
lane drop. The options allow up to three lane additions or drops to occur within the same
link. To simulate a right-lane closure, the rightmost lane would be dropped at a point
specified at a distance from the upstream node, and then it would be added at another
specified point designated again by its distance from the upstream node. The second
technique that can be used to simulate a lane closure is identified as an incident. The user
can create multiple incidents during different times of the simulation on the same link.









Each of these can occur simultaneously and on several lanes if desired. Neither
technique can simulate the taper section prior to the lane closure.
The previous project evaluated the two simulation alternatives and concluded that
the performance of the freeway segment was very similar. Because the two alternatives
produce almost identical values, the one which provides the most flexibility in the
simulation, namely incident analysis, was selected and used. The incident technique
allows for the effects of "rubbernecking" to be simultaneously implemented with a lane
closure, which relates to the worker/equipment activity present at the work zone. To
maintain consistency with the previous project this method was used for the simulation of
the 1-95 work zone.
Figure 25 presents an aerial photograph of the 1-95 work zone. The black lines are
the links that were modeled in CORSIM, while the white circles represent the nodes.


Figure 25: Aerial Photograph of Field Site with Overlay of the Simulated Freeway
Network









Figure 26 shows the simulated test segment and the network view in CORSIM. The
following is a discussion of the function and characteristics of each of the freeway links:
Link (1, 5) This link is the work zone link, with two through lanes that are open
and one lane that is closed. The left and right lane closure configurations were
modeled separately. The work zone was modeled as an incident as detailed
previously. One of the inputs in the simulation of an incident is the distance to the
first upstream warning sign. This was entered as 10,000 ft, to replicate the field
operations. Detectors were placed on the link to collect speed and flow data and to
replicate the data collection location in the field. The detectors were placed 250 ft
from the downstream end of the link. The free flow speed on this link was set at 60
mph. The length of the link was 1,012 ft.
Link (2, 1) This link has three through lanes and starts where the queue begins to
form. There were no detectors placed on the link. The free flow speed on this link
was set at 60 mph. The length of the link is 1,000 ft.
Link (3, 2) This link has three through lanes and is located between the exit ramp
and entrance ramps for the Golfair Blvd. interchange. There were no detectors
placed on the link. The free flow speed on this link was set at 60 mph. The length
of the link is 2,000 ft.
Link (4, 3) This link has three through lanes and one auxiliary lane that connects
the entrance and exit ramps for adjacent interchanges at MLK Parkway and Golfair
Blvd. There were no detectors placed on the link. The free flow speed on this link
was set at 60 mph. The length of the link is 1,000 ft.
Link (6, 4) This is the first upstream link in advance of the work zone. Since the
distance from the work zone to the first upstream work zone warning sign is 10,000
ft. (as indicated above), simulated vehicles entering this link have already had one
warning of the work zone ahead. This link has three through lanes. The length of
the link is 5,000 ft.


Link (8001, 1) This is the feeder link for the network.


















5 1 2 3 4 6

Direction of
Traffic & North


Figure 26: The Simulated Work Zone in CORSIM


The following inputs, based on field data, were entered into CORSIM:
Presence of trucks (percentage) 5 percent

Rubbernecking factor (percentage) 0 percent (There were no workers in the
vicinity of the lane drop)
Free flow speed 60 mph


The following assumptions were also used in the simulation:
Demand flow rate 1,800 vph at the beginning of the simulation, increasing by
600 vph every five minutes for an hour of simulated time. This technique was
used so that all four capacity measures (pre-breakdown flow, breakdown flow,
maximum discharge flow, and average discharge flow) could be obtained, as
the network was loaded gradually to simulate the process of breakdown.

Entering and exiting traffic on each ramp upstream of the work zone- 100 vph.
Field data were not available for those ramps; however their presence and
respective demands do not affect the capacity of the work zone. Therefore,
their demand was assumed to be 100 vph.

Another consideration in modeling work zones relates to the driving behavior of
trucks. CORSIM provides three choices for truck behavior: not biased or restricted to
any lanes, biased to a set of lanes, and restricted to a set of lanes. The lanes to which a
truck is biased can be specified. For this project, trucks were modeled as not biased to
any lanes as observed in the field. There were approximately 40 percent trucks observed
in the left lane and 60 percent in the right lane. This is a different assumption then in the









modeling effort from the previous FDOT project where trucks were assumed to be biased
to traveling on the rightmost lane of the freeway.
Sample size calculations were conducted to determine the number of simulation
runs necessary to obtain a level of confidence in the simulation to account for simulator
variability. The sample size was estimated based on throughput using a 95 percent
confidence interval with an acceptable deviation of 100 vehicles per hour. For each of
the left lane and right lane closure scenarios, 20 runs were conducted which provided a
confidence level well above the 95 percent confidence level with an acceptable error
level less than the 100 vph that has been used on the project in estimating the required
sample size. The results for each are reported in the next section.


4.2 Simulation Results
Two primary sets of output data were collected from the simulation experiments:
a) discharge flows by lane in link (1,5), and b) speed-flow time series, similar to those
obtained for the field data, and the respective capacity measures for each run. A summary
of those results is provided in the remainder of this section.
To obtain discharge flows by lane in link (1, 5), the simulation model was run 20
times for each of the left and right lane closures. CORSIM's default output provides link
flows, however the field data collected were flows measured using detectors at specific
locations. To replicate the field data collection as much as possible, in the design of the
CORSIM simulation detectors were placed on link (1, 5) and these data were compared
to the link flows generated from the simulation to determine whether there was a
significant difference between those two values. Table 11 presents the results of each of
the runs as well as summary statistics for average discharge flows for the left- and right-
lane closures. It was concluded that the two methods of obtaining flows do not provide
significantly different results. Also, the differences in the mean discharge flows between
the left and the right lane closure were found to not be significant at the 95 percent
confidence level. As shown in Table 11, the variability in flows is slightly higher for the
right-lane closure.










Table 11: Results of the 1-95 CORSIM Simulation
Left Lane Closure Right Lane Closure
Link Flow from Point Deiector Link Flow from Point Deiector
CORSIM Oul tpu Film CORSIM Onl pil FIlo
1 3716 3704 3916 3876
2 3820 3828 3860 3851
3 3788 3768 3860 3876
4 3792 3792 3876 3849
5 3824 3836 3812 3808
6 3780 3784 3972 3956
7 3908 3912 3868 3840
8 3804 3796 3680 3672
9 3748 3756 3828 3808
10 3852 3860 3720 3736
11 3808 3780 3856 3844
12 3804 3820 3784 3788
13 3860 3840 3756 3765
14 3748 3760 3840 3828
15 3824 3828 3780 3752
16 3748 3748 3800 3792
17 3780 3800 3828 3808
18 3724 3720 3764 3760
19 3732 3740 3936 3937
20 3840 3816 3772 3764
Mean 3795 3794 3825 3816
Std. Dev 49.89 50.19 71.87 67.51


Next, time series of speed-flow data were obtained to "observe" the process of

breakdown in the simulator. To accomplish this, data were obtained using spot detectors

from the left lane closure in CORSIM every two minutes for the duration of the

simulation. Since the difference between a left and a right lane closure was statistically

not significant, the experiments only focused on the left-side closure.

Figure 27 presents the data collected from the detectors for one of the 20 runs.

The horizontal axis represents 2-min time intervals. The results of the 20 runs are fairly

similar therefore, only one of them is presented here. As shown, the general shape of the

speed and flow time series around the breakdown is similar to those observed in the field

data (shown in Chapter 3 of this report).











Breakdown Flow 4640 vph
5000 / 70.00
Max. Pre- Breakdown Flow 4230 vph
4500 60.00

4000

3500 50.00

2,3000
3000 40.00o

S2500
E
30.00 S
2000 C.
1500 20.00

1000 -
1000 Max. Discharge Flow 4190 vph 10.00
500

0 I I I I I I I I I I I I I I I I I I I I I 0 .0 0
1 2 3 4 5 6 7 8 9 1011 1213141516171819202122232425
Time Interval
-- Volume -- Speed



Figure 27: Time Series Diagram of Simulated Breakdown in CORSIM



Table 12 provides a summary of the four capacity measures obtained from

CORSIM for each of the 20 runs. As shown, the pre-breakdown flow rates were the

highest, with the average discharge flows the lowest. The maximum discharge flow

values had the lowest standard deviation (i.e., small variability).













Table 12: Summary of the Capacity Measures Obtained From CORSIM (in VPH)
Run Max Pre-Breakdoi n Breakdomin Max Dikciarge A e. Discharge
Floh Fhlo FloF Floi
1 4470 4440 4200 3777
2 4620 4200 4170 3917
3 4440 4240 3900 3827
4 4080 4200 4220 3873
5 4380 4020 4110 3957
6 4470 4590 4290 3870
7 4500 4240 4110 3998
8 4320 3780 4530 3960
9 4650 3990 4050 3847
10 4230 4650 4190 3837
11 4470 4440 4200 3932
12 4200 4590 4640 4037
13 4320 4350 4230 3948
14 4290 4260 4290 3980
15 4290 4260 4500 4035
16 3630 3830 4320 4031
17 4340 4380 4320 3922
18 4290 3990 4260 3968
19 4170 4650 4260 3968
20 4530 4230 4440 3948
Mean 4335 4267 4262 3932
Std. Dev. 222 255 171 73
Minimum 3630 3780 3900 3777
Maximum 4650 4650 4640 4037


4.3 Comparison of CORSIM to the Field Data
The field data were compared to the results obtained from CORSIM to determine

whether CORSIM can adequately replicate the work zone operations observed in the

field. The analysis compared a) the capacity measures obtained from CORSIM to those

obtained in the field, and b) the lane distributions observed in the field upstream of the

work zone to the respective ones from CORSIM.

Table 13 presents a summary of the capacity measures obtained from CORSIM

and the field data. As shown, CORSIM predicts a higher maximum pre-breakdown flow

on average, than that observed in the field. The other three capacity measures are

comparable, with differences in the order of less than 100 vph. Generally the variability

(i.e., standard deviation) is higher in the field than that predicted by CORSIM, and the

standard deviation of the average discharge flow is the lowest both in the field, and in










CORSIM.

Statistical testing was conducted using the t-test for means comparing each of the

four capacity measures to the respective field data. The tests showed that the maximum

pre-breakdown and average discharge flows were statistically different at the 95 percent

confidence level, while the breakdown and maximum discharge flows were determined

not to be significantly different at the 95 percent confidence level. Note that for the

average discharge flow, the statistical test fails despite the fact that the difference between

the field and the CORSIM value is only 81 vph. This occurs because the standard

deviation for this measure is very low.



Table 13: Comparison of Capacity Measures in CORSIM to Field Data (in VPH)
y TMr Pre-BriieaiklhImn ii Biraiikoh ii ]T Ma Dichaire 4. D chnre
CORSIMI Floh Flo Flhm Flo
# ofRuns 20 20 20 20
Minimum 3630 3780 3900 3777
Maximum 4650 4650 4640 4037
Mean 4335 4267 4262 3932
Std. Dev. 222 255 171 73
Field Datla Maix Pre-Breaklido n Breaikdow n 1Mi\ Dischal:rye A c. Discharue
(Eweluidingi R:ain Eient 17 Fln F1lnn Flo Flom _
# of Observations 15 15 18 18
Min Value 3078 3636 4044 3830
Max Value 4320 4500 4944 4228
Mean 4000 4179 4347 4013
Std. Dev. 325 283 202 102

Next, the lane utilization data obtained from data collection points 2 and 3 were

compared to the respective ones from CORSIM. Table 14 presents the field-observed

lane utilization and the lane utilizations produced by CORSIM (with a left lane closure

through the work zone for both cases). As shown, in the field, the lane utilization of the

center lane is higher than that of the other lanes, as well as that predicted by CORSIM.


Table 14: Lane Utilization in the Field and in CORSIM

Left Lane Center Lane Right Lane
Field (3-27-07) 21% 45% 34%
Field (4-5-07) 19% 45% 36%
CORSIM 28% 28% 440 ,









In summary, the comparisons reported above show that there are some similarities
and some differences between the field data and the simulator. CORSIM can replicate
some aspects of work zone breakdown, and the general shape of the speed and flow time
series relationship is similar to that in the field. Also, it can reasonably predict the
breakdown and maximum discharge flows. However, the pre-breakdown and average
discharge flows are not statistically equal, despite the fact that the numerical differences
for the average discharge flows were relatively low (below 100 vph).
Generally, CORSIM appears to over-predict the pre-breakdown flow, while the
differences for the remaining three capacity measures are relatively low. With respect to
lane utilization, CORSIM does not predict very accurately the percent of traffic on each
lane, and it does not have the flexibility to allow the user to enter these as inputs into the
simulation for a particular site.


5. COMPARISONS OF FIELD-OBSERVED CAPACITIES TO OTHER
TECHNIQUES
This chapter first presents a comparison of the field capacity measures to those
estimated by the analytical models previously developed in FDOT BD 545-51. The next
section presents a comparison of capacities obtained by CORSIM, the HCM 2000, and
the existing FDOT lane closure methodology.


5.1 Comparison of Field Data to FDOT BD 545-51 Models
This section presents a comparison of the field estimated capacities to the models
developed in the FDOT BD 545-51 project. Two sets of models were developed in that
project: one for planning and one for traffic operations applications. Each of these was
applied to estimate the capacity of the 1-95 study site. Of the four capacity measures
obtained in the field, the average discharge flow was selected to be compared to these
models because the FDOT BD 545-51 project defined capacity as the average discharge
flow, based on the CORSIM output.
Based on the field data the following inputs were used (items in bold were used
for both the planning and the operations models, while the remaining ones were required
only for the operations model):
1) Lane Width, LW = 12 ft









2) Lateral Clearance, LC = 6 ft
3) Rubbernecking factor = 0
4) Passenger Car Equivalency Value for Trucks= 2.4
5) Heavy Vehicle Percentage = 5 percent
6) Sign Distance = 2.0 mi
7) Speed in Lane 1 (unadjusted) = 45 mi/h
8) Lane Distribution in to-be-closed lane = 0.35
9) Lane Distribution in open lane = 0.25
10) Rain Conditions = 0.95 for moderate rain and 0.90 for heavy rain.


In addition, the following were assumed:
1) No adjustments for driver population, light conditions and presence of ramps
are necessary


Table 15 presents the results of the comparison of the models to the field data. The
operations model yielded a value of 4046 vph, or 2023 vehicles per hour per lane (vphpl),
for the no-rain days, while the respective field observations ranged between 3830 and
4228 vph. The predicted value is nearly in the middle of the field observations, with a
very small average difference between the two (33.06 vph). A statistical t-test was
conducted to compare the field data to the operations model estimates (for no rain events
only), and it was concluded that there was no significant difference at the 95 percent
confidence interval. However, the planning model produced a value of 3660 vph (1830
vphpl) for the no-rain days, which was lower than any of the values in the field data. Note
that the models were built with an acceptable level of error of 100 vph.
For observations 12 and 19 the rain factors were applied for moderate and heavy
rain respectively. In both cases the rain had a more detrimental effect on capacity than
that predicted by the operations model. The difference was particularly high for the heavy
rain event (777 vph). Even the planning model, which underestimated the capacity for all
days, overestimated the predicted capacity during the heavy rain event.










Table 15: Comparison of Field Data to FDOT BD 545-51 Models (in VPH)

Field DiIta Ai era.e i OpI'eioin. Difference PI.l.iiiiin Model DiTl'Trece
Ohws', aeililon ] lodtlel Predicltd
Dischrll;'e Flow (. i.i hi Piredicted C iaciri Iph)

1 4040 4046 6 3660 -380
2 3910 4046 136 3660 -250
3 4004 4046 42 3660 -344
4 3950 4046 96 3660 -290
5 3939 4046 107 3660 -279
6 4079 4046 -33 3660 -419
7 4068 4046 -22 3660 -408
8 4180 4046 -134 3660 -520
9 4228 4046 -182 3660 -568
10 4035 4046 11 3660 -375
11 3830 4046 216 3660 -170
12 3584 3842 258 4-~; -108
13 4036 4046 10 3660 -376
14 3899 4046 147 3660 -239
15 3904 4046 142 3660 -244
16 4099 4046 -53 3660 -439
17 4074 4046 -28 3660 -414
18 3944 4046 102 3660 -284
19 2863 3640 777 3294 431
20 4014 4046 32 3660 -354
Average Difference without Rain Observations (vph) 33.06 -301.5
Average Difference for the Rain Observations (vph) -517.5 161.5
Note: Rain events shown in italics



Table 16 summarizes the average discharge flows by level of flow difference. As

shown, 50 percent of the observations are within 100 vph from the estimated capacity,

while 90 percent are within 200 vph.












Table 16: Differences Between Operations Model and Field Data

Slllinallil'n ol Dil'Tl'IInces in A. criIe Dischai._r FhoI
Number of Percent of
Flow Difference (vph) Observations Observations
0-100 10 50%
101-200 8 40%
201-300 1 5%
301-400 0 0%
401-500 0 0%
501-600 0 0%
601-700 1 5%


In summary, it can be concluded that the operations model can reasonably predict

the capacity of this freeway work zone, while the planning model (which requires fewer

inputs) under-predicts it. The effect of rain is not captured very well by either of these

models. Additional research should be conducted to determine the effects of rain for

various levels of rainfall, and for various sites.



5.2 Comparisons to CORSIM, HCM 2000, and the Existing FDOT Lane Closure
Methodology
This section presents a comprehensive comparison of the field data to several

other available methods for estimating capacity including a) the FDOT BD 545-51

project models, b) the CORSIM results, c) the HCM 2000, and d) the existing FDOT lane

closure method. The first two methods were discussed in previous sections of the report

along with the respective results. The methods and results for the last two methods are

presented below, followed by a summary and comparison of the results from all methods.

The HCM 2000 method is presented in Chapter 22 of the Highway Capacity

Manual (TRB 2000). The following equation is used to estimate the capacity of a freeway

work zone:

Ca= (1,600 + I R) fHv* N


where









Ca = adjusted mainline capacity (vph)
fHV = adjustment for heavy vehicles; based on the field data (there are 5 percent
trucks) this factor was estimated to be 0.975
I = adjustment factor for type intensity and location of the work activity
(ranges from -160 to +160 pc/h/ln); there is no specific guidance in the HCM
2000 regarding this factor; since there was no work activity during the data
collection, this factor is assumed to be +160 pc/h/ln or +320 pc/h.
R = adjustment for presence of ramps; there are none within the study area,
therefore this factor is zero.


Using the HCM 2000 method, the capacity is estimated to be 3,440 vph.


The existing FDOT lane closure methodology estimates the capacity of the work
zone as follows:
Capacity = 3600 OF


where
3600 = Base capacity for a 3 to 2 freeway lane closure
OF = Obstruction factor, which reduces the capacity of the remaining travel
lanes to account for lane width less than 12 ft, and lateral clearance less than 6 ft;
in this case, this factor is 1.00


Using the existing FDOT lane closure methodology the capacity of the work zone
is estimated to be 3,600 vph.


Table 17 summarizes the field capacity estimates along with the results from each
of the five alternate models. The field capacity measure used in the comparison is the
average discharge flow for all days of data collection (both left- and right-lane closures)
excluding the rain events. As shown, the field data were best approximated by the FDOT
BD 545-51 operations model. CORSIM's results were also fairly close to the field values
(within 2% of the field observations). It should be noted that the FDOT BD 545-51









models were developed based on CORSIM, but the scenarios simulated assumed all
passenger cars in the traffic stream; the method then adjusted the capacity when trucks
were present using Passenger Car Equivalency (PCE) values developed specifically for
trucks on work zones (FDOT BD 545-51 Final Report). Thus the small discrepancy
between the two models is probably due to the differences in handling truck presence
through the work zone.



Table 17: Differences Between Field Data, CORSIM, HCM 2000, and the Existing
FDOT Lane Closure Methodology
Capacity Difference % Difference
Field 4013
FDOTBD 545-51
Operations 4046 33 0.8%
FDOTBD 545-51
Planning 3660 -353 -8.8%
CORSIM 3932 -81 -2.0%
HCM2000 3440 -573 -14.3%
FDOT 3600 -413 -10.3%


6. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS
A previous FDOT project (FDOT BD 545-51), titled "Impact of Trucks on
Arterial LOS and Freeway Work Zone Capacity," developed analytical models for
estimating the capacity of freeway work zones based entirely on simulated data.
Simulation was used because there were no freeway work zone field data available at the
time. The models developed estimate the capacity of various freeway work zone
configurations as a function of prevailing traffic, design, environmental, and work zone
characteristics. As that project was nearing completion, it was determined that field data
could be obtained from a recently installed freeway work zone along 1-95 in Jacksonville,
Florida, using the Jacksonville Traffic Management Center (TMC) cameras. Therefore
this project was initiated to collect field data and based on these re-calibrate the models
formulated using simulation.
The following were concluded from the research:
The average discharge flow had the smallest standard deviation of all the capacity









parameters. The other three parameters showed higher variability.
The average per lane discharge flow (i.e., capacity during congested conditions) at
the study site under good weather conditions was found to be 4013 vph, or 2,007
vphpl.
Rainy conditions reduced the average discharge flow in the work zone by 10-29
percent; heavy rain had a much greater impact on capacity than moderate rain.
There was no significant difference in the work zone capacity between a left- and
a right-lane closure.
CORSIM appears to over-predict the pre-breakdown flow, while the differences
for the remaining three capacity measures are relatively low. With respect to lane
utilization, CORSIM does not predict very accurately the percent of traffic on
each lane, and it does not have the flexibility to allow the user to enter these as
inputs into the simulation for a particular site.
An operations and a planning model were developed under FDOT BD 545-51.
That project defined capacity as the average discharge flow based on the
CORSIM output. Therefore, that capacity measure from the field data was
compared to the results provided by the two models. It was concluded that the
operations model can reasonably predict the capacity of the work zone, while the
planning model under-predicts it. The effect of rain is not captured well by either
of these models. The existing FDOT lane analysis method underestimated the
discharge flow for that site by approximately 10 percent.


The following are recommended:
The operations model from FDOT BD 545-51 provided good estimates of the
discharge flow for this site over several days of observation. Since the data
were collected only at one site, the model should be evaluated in future
freeway construction sites, and adjusted if necessary based on additional field
data, before it is officially adopted.
The planning model from FDOT BD 545-51 may also be further tested in
future construction sites; however it does not seem to provide results as close
as those of the operations model.









As additional data are gathered, it may be possible to develop and apply a
"Rain Factor" in traffic operations applications around work zones.


The following recommendations are provided regarding possible improvements to
CORSIM with respect to freeway work zone simulation:
The software should consider developing algorithms specifically applying to
work zones, and replicating the use of taper sections.
Guidance should be provided regarding the use of the "rubbernecking" factor
and its relationship to worker and equipment presence in the work zone.
Various geometric elements (such as lane width and shoulder width) are
currently not considered within CORSIM. Its algorithms should be modified
to consider such factors generally, as well as with respect to work zones.




REFERENCES

Elefteriadou, L.; D. Arguea; A. Kondyli; and K. Heaslip, "Impact of Trucks on Areteral
LOS and Freeway Work Zone Capacity (Part B: Freeway Work Zone Capacity); Florida
DOT Contract BD-545-51, July 2007.

Elefteriadou, L., and P. Lertworawanich, "Defining, Measuring and Estimating Freeway
Capacity", Transportation Research Board Meeting, Washington DC, January 2003.

Florida Department of Transportation, 2007. Design Standards.
http://www.dot.state.fl.us/rddesign/rd/RTDS/06/2006_Standards_600.htm. Access date:
February 15, 2008.

Highway Capacity Manual 2000, Transportation Research Board, Washington, D.C.
2000.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs