A KNOWLEDGE-BASED SYSTEM APPROACH TO WORK SHIFT SELECTION
FOR MULTILANE HIGHWAY RECONSTRUCTION AND
Q. AMIN AHMED
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
To my parents
I would like to extend my deep gratitude to Dr. Ralph D. Ellis,
my supervisory committee chairman, for his guidance and dedicated
support, which made this study possible. A special thank you is due to
Dr. Zohar Herbsman for his encouragement and valuable recommendations
throughout my doctoral studies. My sincere appreciation goes to Dr.
Fazil T. Najafi, Dr. Paul Y. Thompson, and Dr. Ajay Shanker for serving
on my supervisory committee. Additionally, I am grateful to Dr. Ronald
A. Cook for serving as an observer in my committee.
I wish to thank Mr. Ananth Prasad and Mr. Henry Haggerty of the
Florida Department of Transportation for their personal efforts in
providing valuable information for this research. I am grateful to Mr.
Ashish Kumar, graduate assistant in construction engineering, for his
support and advice--particularly for the statistical part of this study.
Finally, much gratitude is owed to my parents for their love,
continuous encouragement, and support during the entire course of my
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ........................................... iii
LIST OF TABLES ......................................... vii
LIST OF FIGURES .......................................... ix
ABSTRACT .................. ................................ x
1 INTRODUCTION AND PROBLEM STATEMENT ................ 1
Work Shift Selection in Highway Construction ....... 1
Need to Identify Alternate Shift Times ............ 2
Day vs. Night Shift .............................. 3
Decision Factors ...................... ........... 6
Need for Decision Model ......................... 6
Research Objectives ................... ............ 9
2 REVIEW OF LITERATURE ............................ 11
Introduction ....................................... 11
Overview of Factors Attributable to Day and Night
Shift Work ....................................... 12
Traffic Volume ..................... .. ............ 13
Work Zone Safety ................................. 21
Public Relations ................................... 29
Noise Level ........................................ 30
Cost .......................................... 32
Quality of Work ................................ 41
Job Types in Highway Construction ................ 46
Human Factors ................................... 50
Productivity ........................................ 54
Other Factors ................................... 55
Summary ........................................... 57
3 KNOWLEDGE-BASED EXPERT SYSTEM TECHNOLOGY .......... 58
Introduction ..................................... 58
Architecture of an Expert System ................. 59
Knowledge Representation ........................ 63
Inference Engine .................................. 69
Knowledge Acquisition ........................... 70
Expert System Development Shells ................. 72
Current Applications in Construction Engineering .. 74
Summary ........................................ 75
4 PROJECT COST AND PRODUCTIVITY DATA FOR DAY AND
NIGHT SHIFTS ................................... 77
Introduction .................................... 77
Project Cost Comparison ......................... 78
Productivity Comparison ......................... 86
Summary ........................................ 93
5 MODEL APPROACH FOR WORK SHIFT SELECTION ........... 94
Introduction .................................. 94
Lane Closure Traffic Analysis ................... 96
Accident Analysis ............................... 99
Project Cost Analysis ........................... 100
User Cost Analysis ............................ 102
Summary ........................................ 105
6 DEVELOPMENT OF KNOWLEDGE-BASED SYSTEM MODEL ....... 107
Introduction ....................................... 107
Knowledge Acquisition ............................ 108
Knowledge Base Objective ........................ 110
Formulation of Rules ............................ 120
External Database Interface ..................... 123
Solution Search Technique ....................... 125
Explanation/Help Facility ....................... 125
Summary ........................................ 127
7 MODEL TESTING--CASE STUDY ....................... 128
Introduction .................................. 128
Case Study I .................................... 128
Case Study II .................................... 132
8 CONCLUSIONS AND RECOMMENDATIONS ................. 135
Summary and Conclusions ......................... 135
Recommendations ................................. 138
PROGRAM INTRODUCTION, QUALIFIERS, CHOICES, AND
KNOWLEDGE BASE RULES ..............................
EXAMPLE OF EXTERNAL TRAFFIC COUNT DATABASE ........
PROGRAM HELP FILES ................................
EXAMPLE OF PROJECT DATA COLLECTION FORM ...........
PROGRAM INPUT DATA AND RESULTS ....................
INTERVIEW RESULTS .................................
BIOGRAPHICAL SKETCH ........................................
LIST OF TABLES
1.1 Work Hour Restrictions (In Accordance to Traffic
Plan Developed by Maryland & Virginia)............... 4
1.2 Highway Construction Work Shift Selection--
Factors To Be Considered .......................... 7
2.1 Lane Capacities for Selected Metropolitan Areas ..... 15
2.2 Number of Accidents Before and During Construction
in Seven States ...................................... 23
2.3 Summary of Work Zone Accidents at Night ............. 24
2.4 Fatal Accidents on Urban Freeways by Time of Day
and Day of Week ................................... 28
2.5 Estimated Increase in Cost Differential
of Night Work .......................................... 34
2.6 1-70 Night Paving Project vs. 1-25 Day Paving
-- Percent Cost Differential ...................... 35
2.7 Costs to Purchase, Install, and Remove Light Posts .. 39
2.8 Paving Quality Test Results from Field Samples ...... 43
2.9 Quality Rating of Nighttime Operations .............. 45
2.10 Florida DOT Projects Using Night Work .............. 48
2.11 Construction Activities Performed Better At Night ... 49
3.1 Difference Between Expert System and
Conventional Program ............................... 60
4.1 Statistical Summary of Unit Costs for Selected Work
Items for All Daytime FDOT Projects in 1990 ......... 79
4.2 Statistical Summary of Unit Costs for Selected Work
Items for All Nighttime FDOT Projects in 1990 ....... 80
4.3 Difference Between Day and Night Unit Costs for
Selected Work Items for All FDOT Projects in 1990 ... 81
4.4 Quantities of Work Items for Eight Selected FDOT
Night Projects ...................................... 83
4.5 Effect of Quantity on Project Costs for Eight
Selected Work Items ............................... 84
4.6 Summary of Productivity Rates for FDOT Construction
Projects ................................... ... 87
4.7 Summary of Productivity Rates for FDOT Nighttime
Construction Project on I-95 in St. Johns County .... 88
4.8 Guidelines for Estimating Production Rates for
FDOT Projects ..................................... 89
4.9 Statistical Test for Day & Night Production Rate
of Plant Mixed Surface ............................. 91
4.10 Statistical Test for Day & Night Production Rate
of Milling Existing Surface ........................ 92
5.1 Decision Factors That Influence Shift Selection ..... 95
5.2 1-75 (North Bound) Average Weekday Hourly Traffic
Volumes. Location: Hamilton County ................ 98
5.3 Personal Cost of Time Delay for Queuing ............. 104
LIST OF FIGURES
1.1 Research Development Flowchart ...................... 10
2.1 Traffic Flow on a 4-Lane Freeway ..................... 17
2.2 Traffic Volume/Capacity Relations .................... 19
2.3 Effect of Lane Closure at Different Hours ............ 20
2.4 California Urban Freeway Fatal Accident Rates ........ 26
2.5 Triad of Shift Work Coping Factors ................... 51
3.1 The Expert System Architecture ...................... 61
3.2 A Frame Representation of Knowledge .................. 64
3.3 Knowledge Representation Using Semantic Network ...... 66
6.1 Decision Tree Segment for Work Shift Selection:
Congestion Analysis ................................. 114
6.2 Decision Tree Segment (Closed) for Work Shift
Selection: Accident Analysis ....................... 115
6.3 Decision Tree Segment (Open) for Work Shift
Selection: Accident Analysis ....................... 116
6.4 Decision Tree Segment (Open) for Work Shift
Selection:Noise, Quality, Productivity Considerations.. 117
6.5 Decision Tree Segment (Open) for Work Shift Selection:
Experience, Supply and Temperature Considerations ..... 118
6.6 Decision Tree Segment (Closed) for Work Shift Selection:
Supervision/Communication and Human Factors ........... 119
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
A KNOWLEDGE-BASED SYSTEM APPROACH TO WORK SHIFT SELECTION
FOR MULTILANE HIGHWAY RECONSTRUCTION AND
Q. Amin Ahmed
Chairman: Dr. Ralph D. Ellis
Major Department: Civil Engineering
The timely, efficient, and quality completion of highway projects
partly depends on the selection of the appropriate work shift. Certainly
there are advantages and disadvantages for both day and night shifts.
Daytime operations are generally considered to be safer for both workers
and motorists because of better visibility and a higher state of
alertness. However, to accommodate the flow of higher daytime traffic
volumes, work zone lane closures may not be possible--and the option to
work during the night becomes a serious consideration. A highway
project's characteristics may favor a particular work shift. Type of
work, lane capacity, average daily traffic (ADT), work zone accidents,
project duration, and project costs are some of the issues that may
dictate shift times.
This research is aimed towards the development of a decision model
that will incorporate all the factors that may influence the selection of
either a day or a night work shift. The process is based on a decision
tree representing qualitative and quantitative factors, ranked in the
order of importance. The shift selection methodology is similar to the
human reasoning process and that is why a knowledge-based system approach
has been chosen for this decision model. An expert system shell has been
utilized to develop the knowledge base, which consists of IF-THEN logic
rules. The rule-based knowledge structure also has the option to
interface with an external traffic count database for the purpose of lane
The model approach to work shift selection includes mathematical
reasoning in the analysis of traffic congestion, vehicle accident numbers,
motorist (user) costs, and project (owner) costs. The final solution
offered by the knowledge-based system model consists of the recommended
work shift, number of daytime lane closures allowed, user cost savings for
a night shift option, and percent change in total owner cost for a
INTRODUCTION AND PROBLEM STATEMENT
Work Shift Selection in Highway Construction
In recent years, in Florida, as in many other states, the emphasis
in highway construction has primarily been on rehabilitating existing
facilities. Resurfacing, widening, multilane reconstruction, and bridge
repair are typical activities currently taking place in various states.
A major concern for state transportation agencies during construction and
rehabilitation of the nation's highway system is to minimize public
inconvenience. Work zone traffic accidents, congestion, noise, and cost
to the delayed motorists are issues that may lead to adverse public
reaction. In the urban areas, required lane closures during
rehabilitation work are resulting in heavy traffic congestion. This
problem is not limited to roads in urban locations, but also includes
freeways that become crowded during peak travel periods. For example, in
Florida, Interstate 75 is so crowded during holiday periods that it
resembles an urban main street. I-4 near Orlando often becomes a "local"
street due to congestion during tourist season.
Agency experience or mathematical reasoning can be utilized to
evaluate the acceptability of this congestion level. A recent survey of
ten state agencies done by researchers indicates a cutoff volume of
approximately 1500 vehicles per hour per lane at which backups are
expected (Shepard and Cottrell, 1985). As the traffic volume exceeds the
cutoff point, the agency is obliged to consider alternate working hours--
namely, night shift. However, noise ordinances in some areas may prevent
usage of heavy and noisy equipment during night hours. The selected shift
hours must be such that traffic congestion and public disturbance are kept
minimal and that overall safety requirements are adequately met during
those periods. After these preliminary criteria are satisfied, the state
agency will have to consider or weigh a variety of other factors (positive
and negative) that are associated with different shift times. In addition
to the many site-related variables, there are human and sociological
factors that can be attributed to a work shift and which in turn may
influence the timely, efficient, and quality completion of highway
projects. Different work shifts have different effects on a worker's
health, attitudes, and performance. Thus, the selection of appropriate
work shift hours is an important project management decision for the state
Need to Identify Alternate Shift Times
Unlike the manufacturing industry where production takes place in an
enclosed, controlled surrounding, highway construction takes place in a
dynamic environment, influenced by many external factors. Although the
highway work shift may be broadly classified by "day" and "night" shifts,
it is necessary also to identify the specific hours that provide the ideal
work environment. These shift times will no doubt vary from project to
project, when all the factors are taken into account. In some situations,
the work shift may even be a combination day and evening hours. When
working in major metropolitan areas, highway agencies typically restrict
their roadway construction and maintenance activities to hours of off-peak
traffic and weekends.
In certain areas, highway construction and maintenance work can
proceed unhindered during "off-peak" hours, i.e., 9 a.m. to 3 p.m.
However, there are situations where lanes cannot be closed during the day
at all due to high traffic volumes. Traffic congestion in some areas may
last up to 12 or 14 hours a day. Nighttime operation is now being
considered as a viable and necessary option, not only from the point of
view of project management, but also from the public relations
perspective. Table 1.1 shows work hour restrictions for varying traffic
conditions in accordance with the traffic control plan developed by
Maryland and Virginia highway agencies (Shepard and Cottrell, 1985). Each
traffic condition represents different lane closure situations.
Day vs. Night Shift
The obvious advantage of night work is that there is less
interference from heavy traffic, allowing efficient project scheduling.
According to Lee (1969), a concrete paving operation in California
conducted at night was finished in 16 working days, whereas the same
project would have taken a minimum of 35 days for completion during the
daytime. A day shift would have provided fewer working hours and more
interference from heavy traffic, resulting in delay of material delivery.
Quality of construction may or may not differ during day and night
shifts. Price, in his 1985 report on nighttime paving, has stated that
overall quality on a night paving job (1-70) in Denver, Colorado, was very
similar to a well done day job (1-25 north of Denver). Compaction did not
Table 1.1 Work Hour Restrictions (In Accordance with the Traffic Control
Plan Developed by Maryland & Virginia)
I II III
Monday 6:00 am 9:00 am 9:00 am 3:30 pm 12:00 pm 6:00 am
Thursday 3:30 pm 8:00 pm 8:00 pm 12:00 pm
6:00 am 9:00 am 9:00 am 3:00 pm 12:00 pm 6:00 am
3:30 pm 8:00 pm 9:30 pm 12:00 pm
12:00 pm 6:00 am
Saturday N/A 7:30 am- 8:00 pm
8:00 pm 12:00 pm
10:30 am 10 pm* 12:00 pm 10:30 am
Sunday N/A 10:00 pm 12:00 pm*
10:30 am 8 pm** 8:00 pm 12:00 pm**
Traffic Condition I : No lane closures permitted, all lanes open
Traffic Condition II: Single lane closure (median, center, right)
Traffic Condition III: Three, two or one lane closure permitted.
* After Easter weekend through second Sunday in September.
** Third Sunday in September through weekend before Easter.
Source: Shepard and Cottrell, 1985
suffer despite cooler night temperatures. The only aspect of quality that
deteriorated during night work was aesthetics.
congestion and delay being less during night hours. A user cost analysis
done by Price revealed a total savings of $1,154,900 for the motorists in
an 1-70 night project compared to an equivalent 1-25 daytime project.
However, construction costs tend to be higher during the night shift.
Price (1985), in comparing the 1-70 and 1-25 projects, has stated that the
average cost per item was approximately 40% higher for the night paving
job (1-70). Material acquisition, extra personnel, and lighting fixtures
are some of the items that have additional costs during night work.
The issue of safety, for both construction workers and vehicular
traffic, is an important consideration in deciding shift times. Safety
records in highway work zones can indicate the factors that cause
accidents and whether they can be attributable to certain shift hours.
For a meaningful comparison of safety and accident characteristics of
night and day shifts, it is necessary to consider many variables, such as
duration of lane closure, proportion of closed lanes, job type, closure
length, traffic control devices, etc. At present, there are no general
conclusions regarding safety on various work shifts. According to Lum
(1980), in his study of 7 states, construction during night shift did not
increase the percentage of total night accidents to overall accidents
(constant at 30%). However, Lum's research did reveal that there was a
9.4% increase for night accidents as a result of construction, which means
that there is a similar increase for day accidents due to construction.
Another issue in project management, which may be of concern during
a particular work shift, is communication. Communication between the
highway agency and the contractor may be difficult during night work
hours. If the proper authority cannot be reached for consultation at a
certain time, construction work could come to a virtual halt. A similar
concern arises for equipment maintenance. Equipment failure would result
in production delay if parts or mechanics were unavailable during a work
The highway agency, as a decision maker, will have to consider both
qualitative and quantitative factors in the work shift selection process.
Examples of qualitative factors are agency policy, public relations,
noise, job type, communication, worker morale, work quality, and
temperature (see Table 1.2). Highway capacity, traffic volume, motorist
delay resulting from traffic congestion, and accident rate are typical
quantitative variables. The cost differential factor, if it can be
quantified, may be categorized as quantitative; otherwise, it can be
qualitative in nature, being higher or lower.
Need for Decision Model
Currently there is no decision model available for state highway
agencies that will enable them to select work shifts quickly and
efficiently, taking all possible factors into consideration. For most
highway projects, state agency engineers select the work shift hours based
on judgment and experience, with some usage of mathematical models to
Table 1.2 Highway Construction Work Shift Selection--Factors To
Traffic Volume vs. Roadway
User Costs Due to Delay at
Project Item Cost
Work Zone Accidents
calculate expected traffic volumes and delay times. But there is no
systematic methodology to "optimize" the shift selection process. A
project may be done during the night unnecessarily, resulting in increased
project costs. Work shift selection by the state agency is aimed towards
public satisfaction mainly; owner costs are largely ignored. Some
projects remain behind schedule because of delays caused during a
particular work shift. For example, at night, materials or spare parts
may be available on a limited basis; during the day, work progress may
slow down due to traffic interference. An interactive consulting tool,
such as a knowledge-based expert system, would aid the highway agency in
optimizing the selection of a particular work shift and identify the
reasons for the selection. For instance, if daytime traffic volume
allowed lane closures and night work were determined to be unsafe, then
certainly the appropriate work period choice would be day shift, even if
cooler night temperature is preferred.
The knowledge of the experienced decision makers in various state
agencies is scattered in the form of literature in reports, journals,
survey results, and guidelines. Although various state agencies have
guidelines that determine allowable daytime lane closures based on traffic
analysis, there are many other qualitative factors that are not considered
in a formal manner. This information needs to be compiled and made
available to state highway engineers in the form of a knowledge base. A
rule-based expert system approach is ideally suited for the type of
decision-making problem discussed above. Its interactive feature and ease
of use makes it a suitable substitute for a human consultant.
A systematic, focused research in a narrow domain will enable the
practical development of a prototype expert system model for work shift
selection in highway construction. Figure 1.1 illustrates the research
development flowchart. To this end, the goals of the research study are:
1. To visit several district offices of the Florida Department of
Transportation (FDOT) to investigate the decision-making
process used in selecting work shifts for highway construction
and interview experts in highway construction.
2. To review current procedures used by the FDOT and other state
agencies nationwide for construction projects during day and
night shifts. Cost differentials, accident and safety, human
performance, and case studies of successful night operations
are some of the areas for review.
3. To perform a statistical analysis/comparison to determine if
there are any significant differences in project related costs
and productivity rates between day and night shifts.
4. To identify the various parameters necessary for the
development of knowledge-base rules for an expert system model
for highway work shift selection.
5. To utilize PC-based shell EXSYS Professional in designing a
6. To evaluate the possible uses of the model by using a sample
case study approach.
7. To summarize the final outcomes and state the conclusions
drawn forth from this research.
Figure 1.1 Research Development Flowchart
. Interview Experts
. Review Current Procedures
. Review Research Studies
. Data Analysis
KNOWLEDGE BASE STRUCTURE
Usage of ES Shell
SFormulation of Rules
REVIEW OF LITERATURE
Recent literature on highway construction emphasizes the need to
perform rehabilitation work during shifts other than the typical day shift
utilized in most industries. In urban areas the problem of congestion is
of particular concern. To accommodate rehabilitation and improvement
activities on the freeways, state agencies are required to close traffic
lanes, creating heavy congestion on roads already loaded to capacity. The
congestion has created problems for state agencies and contractors in
terms of safety and project scheduling. The situation is also hazardous,
costly, and inconvenient for the traveling public. As a result, more
construction and rehabilitation work is being performed during the night
shift, when traffic flow is minimum.
This literature search was done to identify the principal factors
involved in deciding shift times for highway construction projects.
Qualitative and quantitative attributes that are linked to activities
during different work periods were obtained from various highway
construction literature. Specifically, unique aspects of day and night
operations were targeted for identification. A computer database search
through the Southern Technology Applications Center (STAC) in Gainesville,
Florida, revealed over 25 published articles related directly to the
subject of nighttime highway construction.
In the following section, an overview of the factors attributable to
day or night shift work in highway construction is presented. The
literature review for each of the workshift selection factors is presented
in detail section by section in this chapter.
Overview of Factors Attributable to Day and Night Shift Work
Several published reports in the transportation area have provided
information on issues relating to the planning, safety, and traffic
control aspects of nighttime maintenance and construction work and their
advantages and disadvantages. Shepard and Cottrell (1985) conducted a
study that compiled information on current practices in nighttime highway
maintenance and construction operations. This information was used to
develop guidelines for determining when night work should be done and what
traffic control devices should be utilized. The authors have researched
the many variables, problems, decisions, and issues involved in nighttime
highway rehabilitation. Results from visits and discussions with 11 state
departments of transportation revealed the different strategies and
philosophies involved in nighttime operations. The major areas covered in
their nighttime construction feasibility study are scheduling of lane and
road closures, work zone costs, safety, public relations/user costs, and
traffic control. Hinze and Carlisle (1990) have done further evaluation
of the important variables in nighttime construction. The authors have
focused their research on rehabilitation and maintenance activities of
major metropolitan highways. Qualitative and quantitative factors related
to nighttime construction have been detailed in their study. Advantages
and disadvantages of a nighttime construction schedule are also discussed.
The following sections review information obtained on traffic
congestion, work-zone safety, public relations, noise, cost, quality
control, job type, human and sociological factors, and productivity. All
these are factors that need to be considered in the shift selection
Typically, daytime traffic volume in metropolitan areas demands a
particular lane capacity to keep congestion low. Lane closures during
construction work are dictated by the comparison of observed traffic
volume and specified lane capacities. The determination of lane
capacities and actual traffic volume is necessary to estimate the level of
congestion created by lane or road closures. If this congestion is
unacceptably high during a certain time period, lane closures for
construction may be limited or not permitted at all, requiring agencies to
decide on alternate work shifts. Low volume periods should be identified
for scheduling work shifts. Night shift offers the advantage of low
Shepard and Cottrell (1985) have discussed in some detail how
traffic volume and lane capacities are determined and how traffic flow can
be synchronized with road construction activities. Agencies rely on
mathematical models as well as past experience to make traffic capacity
forecasts. According to Dudek and Richards (1985), the State of Texas
expects a lane capacity of 1050 vehicles per hour (VPH) on a 3-lane urban
highway with 1 lane open, when resurfacing or removing asphalt. A study
of cumulative distributions and average capacities of similar work zones
allowed the Texas officials to make this capacity forecast. Based on
policies and experience, work location, job type, etc., some state
agencies have developed guidelines for lane capacities. Agency policy may
dictate that nighttime construction is necessary if the traffic volume
during the daytime hours for a certain location exceeds a "cutoff" value,
after which resulting delays are "unacceptable." Table 2.1 shows a
listing of cutoff values adopted by ten agencies surveyed by Shepard and
Cottrell. The Highway Capacity Manual (National Research Council, 1985)
provides one of the empirical formulas to estimate roadway capacity.
VPH (per lane) = Z f, fhv fp
Z = capacity under ideal conditions at design speed.
f, = lane width / lateral clearance adjustment factor.
fh, = heavy vehicle adjustment factor.
fp = driver population factor.
The values for these adjustment factors may be obtained from tables
included in the manual.
To ensure good volume estimates, agencies supplement existing data
with special spot counts. In major cities such as Chicago, Los Angeles,
and Detroit, agencies have supplemented traffic volume data with real-time
traffic flow data from their freeway surveillance system. The estimation
of traffic volumes is also influenced by diversion. Motorists are often
induced to take alternate routes if prior publicity is made by the agency.
According to Shepard and Cottrell (1985), information from California
indicated that a traffic volume diversion up to 30% was possible for
local traffic when prior publicity on alternate routes was done.
Lane Capacities for Selected Metropolitan Areas
VPH per Lane
a) 1500--usually no congestion unless
more than one lane is closed.
b) Sometimes use 1800 (with some backups
to give contractor more time).
a) No daytime closure. Depending on the
area, no daytime closures if 2 or more
lanes need to be closed.
a) Depends on location, number of ramps.
a) 1200--volume before backup starts.
b) 1500--expect serious backups.
a) Depends on area and experience.
a) Closures based on experience.
a) If closure is 2 or more lanes, have to
detour traffic and work at night.
a) In many cases will accept daytime
backups rather than work at night.
a) >1200--start worrying.
b) =1500--requires detailed analysis.
c) >1500--only with special traffic control.
Source: Shepard and Cottrell (1985)
A common method used by agencies to evaluate congestion is simply to
plot the hourly volumes for the work shift time period. Figure 2.1
indicates the volume distribution (curve) on a 4-lane highway during a
probable work period, along with estimated capacities (horizontal lines)
for 3 lane, 2 lane and 1 lane closures. From the graph it is apparent
that for 2 lane closure the traffic volume from 10:00 a.m. to 3:30 p.m.
and from 8:00 p.m. to 6;30 a.m. does not exceed the given capacity.
However, these work periods are very short in duration to allow major
repair activities. Statistics have shown that under normal daytime
operations, there are two peak traffic loads that actually cut the work
period to an inefficient 5-1/2 hour day (Abbott, 1978). Figure 2.1
indicates a lengthy work period during the night when 3 lanes are closed,
with only 1 lane required to remain open to handle traffic. By noting
when the traffic demand and the capacity are in the same range, it is
possible to determine the times at which the lanes can be closed and
reopened to traffic.
Vehicle delay costs is an important quantitative factor in the
workshift selection process. Highway construction literature has
indicated that nighttime construction greatly reduces user costs related
to vehicle delay. A research report by Price (1985) comparing two similar
paving projects (one nighttime, the other daytime) in Colorado, reveals a
reduction of vehicle cost from $119,100 (daytime) to $10,100 (nighttime).
The author obtained the dollar estimates from a 1977 report titled A
Manual on User Benefit Analysis of Highway and Bus Transit Improvement.
The values were estimated showing excess cost of vehicle speed change
cycles above cost of continuing at initial speed. For the daytime cost,
Volume VPH (thousands)
12 1 2 3 4 5 6 7 8 9101112 1 2 3 4 5 6 7 8 9101112
A.M.<--Time of Day-->P.M.
-Volume Curve 1-LANE CLOSURE
2-LANE CLOSURE --- 3-LANE CLOSURE
Source: Hinze & Carlisle (1990)
Figure 2.1 Traffic Flow on a 4-Lane Freeway
the author assumed an initial vehicle speed of 55 mph and a 10 minute
delay due to traffic stoppage, followed by resumption of speed. For the
daytime paving project analyzed by Price, a cost of $64.68 per 1000
vehicles was estimated. By using daytime traffic counts and considering
a project duration of 37 days, a cost of $119,100 was calculated (Price,
1985). For the nighttime paving project, traffic speed was assumed to
slow down to 30 mph, and the estimated vehicle cost was $10,100 for the
change in speed over the project duration.
Vehicle delay, length and size of vehicle queue can be calculated
from a mathematical model, as shown graphically in Figure 2.2 (Shepard &
Cottrell, 1985). The graph shows the arrival curve and the departure
curve indicating the upstream traffic volume and bottleneck capacity,
respectfully. At time t2 the number of vehicles in queue, Q, equals to the
corresponding ordinate on the arrival curve minus the corresponding
ordinate on the departure curve, the ordinate value being the total number
of vehicles. The queue begins to form when the arrival curve slope is
greater than the departure curve slope. The queue is at its maximum when
the two slopes are equal, and it starts to dissipate after that point
(Mannering & Kilareski, 1988). As shown in Figure 2.2, the longest
vehicle delay occurs at time t3. The length of this delay is given by the
horizontal line D, joining the two points where the two curves have equal
slopes. Agency planners can utilize the quantitative information from
this model to determine whether the traffic characteristics of a
particular workshift is acceptable or not. Figure 2.3 illustrates an
example usage of this model, based on a study done in California (Gillis,
Source: Shepard & Cottrell (1985)
t2 t3 t4
Figure 2.2 Traffic Volume / Capacity Relations
mu mum. I -,
10 AM Lane
No Delay A
9 AM L
S2 lanes cl
mne Closure --Begin Delay
JI/ Cum. Vehicle a Using Rojad
Hour of Day
Source: Gillis (1969)
Figure 2.3 Effect of Lane Closure at Different Hours
Vehicle delay in hours can also be measured by the following
equation noted by Hinze and Carlisle (1990):
Delay = D, / S, D/S
D, = work zone route (detour) distance in miles.
S, = average work zone route (detour) speed in m.p.h.
D = normal route distance in miles.
S = average normal route speed in m.p.h.
This equation takes into consideration the fact that vehicle delay
can occur by speed changes through the work zone and also by distance
changes resulting from a detour. During highway construction work shift
selection, this type of detailed analysis allows planners to evaluate the
traffic impact objectively.
Work Zone Safety
One of the most important factors that state agencies consider when
deciding highway work shift periods is safety. The issue of safety during
nighttime construction has been addressed in several research reports.
These reports note the controversy over which work shift is safer than the
other. The obvious hazards during the night shift are reduced visibility,
higher speeds, and a higher frequency of drunk or inattentive drivers on
the roadway. But the night shift also offers a reduction of traffic
volume, which in turn results in safer working conditions. From reviewing
existing literature, it is difficult to compare the safety aspects of
night shift as opposed to those for day shift simply because of the lack
of comparable data.
The subject of highway work-zone accidents is covered by several
published reports. A report prepared by Graham, Paulsen, and Glennon
(1977) for the Federal Highway Administration includes results of a study
of construction zone traffic accidents. The study involved analysis of
traffic accidents occurring in 79 zones in seven states. In two states
the total number of accidents actually decreased during construction.
However, when data from all seven states were combined, results indicated
an overall "before work" to "during work" average accident rate increase
of 7.5%, with a corresponding 9.4% increase for night accidents (Graham et
al., 1977). Lum (1980) has shown these results in a tabular format (Table
2.2) in a paper published in the December issue of Public Roads.
Breakdowns by accident type, severity, light condition, roadway type, area
type, work area, job type and location are also included in the results.
The authors concluded that the number of traffic related night accidents
increased during construction, but the proportion of night accidents to
total accidents remained the same at 30% for before and during
Nemeth and Rathi (1983) published a work-zone accident report in the
Transportation Quarterly that contains accident data collected through a
28 month study by the Ohio Turnpike Commission. The report includes the
number and percent of total accidents in different portions of the work
zone that occurred during the night shift. Table 2.3 summarizes the
percent of night accidents in the work zone. The study concluded that the
crossover zone, where traffic is forced to shift over to other lanes, has
the highest accident rates.
Table 2.2 -
of Accidents Before and During Construction in Seven
State and Before Construction During Construction % Change
* = Statistically
L _______________ I
significant at the 5 % level
Source: Lum (1980)
Table 2.3 Summary of Work Zone Accidents at Night
(Ohio Turnpike Commission)
Zone Total Number Number of Percent
of Accidents Accidents Night
at Night Accident
Advance 12 1 8.3
Taper 17 6 35.3
Single Lane 43 13 30.2
First Curve 49 34 69.4
Total 63 39 61.9
Bi-direction 41 18 43.9
Other Work 9 3
Zone Total 185 80 41.6
All Turnpike 3,429 1,431 41.7
Source: Nemeth and Rathi (1983)
A 1985 California accident study reveals that although the lightest
traffic congestion occurred between the hours of 11:00 p.m. and 6:00 a.m.,
the fatal accident rate (per 100 MVMT, million vehicle miles of travel)
during this period was very high. Figure 2.4 (Highway Maintenance
Activities During Low Volume Traffic Hours, Report to the Legislature,
1988) illustrates the California fatal accident rate expressed as a
function of hours of the day. The graph shows how the rate increases
sharply at night, peaking at about 3:00 a.m. A work zone vehicle accident
report by Hargroves and Martin (1980) indicates similar results in a 1977
study in Virginia, where for all accident categories the lowest number of
accidents occurred during the late night hours.
Shepard and Cottrell conducted interviews on the subject of safety
at night work zones with state agencies from Texas, California, Illinois,
Michigan, New York, and Pennsylvania. Texas is taking extra measures to
ensure safety during night operations; for example, in Houston a special
crew was formed to manage traffic on high volume freeways. Another urban
area in Texas is more ready to accept long daytime traffic backups rather
than accept the high risk of night construction. The California
Department of Transportation (DOT) is prepared to accept daytime vehicle
delay up to a certain cutoff point, beyond which night work is considered.
In the Los Angeles area, the hours between 2:00 a.m. and 3:00 a.m. are
considered unsafe for workers due to the presence of irresponsible drivers
on the road. Safety is given the most priority in Chicago when the agency
selects work shifts for road maintenance operations. Illinois DOT
officials reported that nighttime accidents were more severe because of
higher vehicular speeds and more frequent encounters with drunk drivers.
URBAN FREEWAY FATAL ACCIDENT RATES PER
100 MILL. VEH. MILES OF TRAVEL (1985)
FATAL ACCIDENTS PER 100 MVMT
121 2 3 4 5 6 7 8 91011121 2 3 4 5 6 7 8 9101112
A.M.(--TIME OF DAY --> P.M.
Source: California DOT (1988)
Figure 2.4 California Urban Freeway Fatal Accident Rates
In Chicago, diversion of traffic volume by detours during highway work is
not considered very often because of the potential safety hazard in "bad"
neighborhoods. However, in Detroit, providing detours is not a problem,
due to the availability of convenient alternate routes such as express
lanes and service roads. As a result, Michigan DOT has been able to
close freeways entirely for nighttime maintenance, diverting traffic
through detours successfully. In New York State, agency officials realize
the importance of providing nighttime safety and the need for additional
spending on safety measures. Pennsylvania DOT, although initially
skeptical of the safety of nighttime operations, discovered that night
shift work could be less accident-prone if vehicular speeds are reduced by
the presence of a police car with flashing lights at the beginning of the
lane closure. Also, advance publicity through news media increased safety
for both day and night shift work.
According to the California DOT report "Highway Maintenance
Activities During Low Volume Traffic Hours," nighttime construction
accounted for 60 percent of the injuries/fatalities sustained in highway
work zones. Table 2.4 shows the percentage of fatal accidents on urban
freeways by the time of day and day of week. A high percentage of fatal
accidents were noted during nightime and early morning hours. Most worker
fatalities (and major injuries) are related to errant drivers causing
these accidents. The California report concludes that the high fatal
accident rate during night hours creates a very hazardous environment for
the workers on site. The report cautions that a large scale move from day
maintenance to night maintenance could mean a significant increase in
worker injury and fatality rates.
Table 2.4 -
Fatal Accidents on Urban Freeways
and Day of Week (1986)
by Time of Day
Time No. of Percent Day of No. of Percent
of Day Accidents Week Accidents
* High Percentage of Freeway Fatal Accidents
Source: Report to the Legislature, California DOT (March 1988)
The most effective way to protect workers is, of course, to divert
vehicular traffic away from the work zones through detours provided by
city streets. Although this will cause some inconvenience to the
motorists (5 to 10 minutes added to their travel time), it is not a high
price to pay to eliminate daytime congestion caused by maintenance
activities. Another protection method is to construct a physical barrier
separating the workers from the traffic. Better illumination of work
areas, highly visible clothing for workers, reflective cones/barriers,
police presence, etc., are additional ways to increase worker safety.
Public relations is an important element in the workshift selection
process for highway construction. Public inconvenience during road
construction operations is a major concern for the state agency. Public
acceptance of a particular work shift (day or night) means it warrants
high consideration by the decision makers. Good public relations can
facilitate the success of an operation in terms of reduced congestion,
increased safety, and goodwill (Shepard and Cottrell, 1985). Shepard and
Cottrell discuss various means of informing the public about nighttime
operations. The size of the project, location, traffic volume,
experience, and so forth dictate how extensive the coverage should be.
Certainly, the amount of publicity for a night project should be more
extensive for safety reasons. Survey results indicate that personal
contact (special signs, door-to-door contact, letters) is thought to be
the most effective method for publicity. The city of Chicago has an
elaborate plan to provide the public with information on upcoming freeway
operations. Local radio, television, and newspapers are effectively
utilized to increase public awareness. Detroit has successfully used
changeable message signs, permanently installed along freeway sections, as
a method of informing the public of lane closures.
A high level of noise, resulting from operation of heavy equipment
during road construction activities, may adversely affect public opinion.
Thus, noise is one of the important qualitative factors attributable to
shift work. During the night shift, the public might be particularly
sensitive to construction noise. Little information on the subject of
noise levels during road construction was found in the literature review.
According to Shepard and Cottrell, many states have noise ordinances that
limit noise levels in urban areas. Agency officials who were interviewed
by the authors reported that if equipment such as a jackhammer is
necessary during work near a residential area, the night shift is avoided.
Another cause for concern for the agency is that diversion of traffic
during freeway construction may increase noise levels on the detour
A research study (Hinze and Carlisle, 1990) included a survey of
project planners of various state highway agencies that found that noise
is of high importance next to congestion and safety. However, the same
study also concluded that resident engineers of the agencies do not
consider noise an important factor compared to other factors when planning
partial or full closure of freeways during construction.
The primary noise sources in road construction and maintenance can
be identified as (1) diesel engines, powering drive trains and cutting
heads; (2) the vacuum/blower system; and (3) grinding heads in contact
with pavement. According results obtained by the Arizona DOT (Kay, 1985),
these sources combine to produce a noise in the range of 82 to 95 dB (A)
at 50 feet from construction. Typically, noise generated during concrete
pavement construction is higher than noise during flexible pavement
construction. Many highways in the metropolitan areas are of rigid
pavement, thus the potential for construction noise in those areas appear
to be significantly high. The study done by the Arizona DOT (Kay, 1985)
notes an upper limit of 86 dB (A) for an acceptable noise level at a
distance of 50 feet. Other states also supported this finding, verifying
the acceptability of the limit. By using a full complement of silencers
it was estimated that noise levels could be reduced to 80 dB. Noise
reduction is also possible by the insertion of a portable sound barrier
between the source and any sensitive receivers.
Kay discusses about the problems encountered in setting and
enforcing noise-level limits for nighttime construction operations. In
the state of Arizona, a noise-abatement incentive was established to
encourage potential bidders to silence noisy construction equipment,
specifically the grinding machines. This noise-abatement incentive plan
was considered to be an alternate approach to including a relevant
specification in the contract. In some occasions, contractors have
threatened not to bid if specifications on noise levels were maintained or
tightened. The objective of the incentive plan was to provide
compensation for the additional cost incurred by the contractor. The
Arizona DOT set the ceiling for the incentive amount at $ 50,000 -- where
as the actual amount turned out to be $ 12,684 (for the first contract
using this plan) to maintain an average of 82 dB (A) noise level
throughout the project.
Some of the important findings in the research done by the Arizona
DOT, as mentioned by Kay in his paper published in the 1985 issue of the
Transportation Research Record, are as follows:
1. Fast completion of project results in less noise complaints
2. Accessory equipment, such as jackhammers, have a noise impact due
to their higher frequencies. Retrofitting on machines can result
in significant dampening of noise.
3. For large scale projects, it may be too expensive to
retrofit major components to maintain an overall
construction noise level to 86 dB(A). In such cases, public
relations can play a major role.
4. Typically, high productive equipment make more noise than
low productive equipment. Therefore, strict noise
restrictions may cause decrease in productivity and longer
project duration. A "trade-off" between project duration
and acceptable noise level can be considered.
Like any decision making process, the selection of day shift over
night shift or vice versa is influenced by the economic result of the
alternatives. Literature review indicated a lack of cost information for
an effective comparison between daytime and nighttime construction. In
highway construction, the costs attributable to a particular workshift can
be categorized into the following: owner/construction costs, user costs,
accident costs, and pavement maintenance costs. The latter cost refers to
post construction maintenance costs which depends on quality control
during a shift.
Owner costs are essentially the costs borne by the state highway
agency, resulting from the construction of the specified facility The
cost of contract (labor, material, equipment and contractor) and agency
costs (planning, evaluating, monitoring) are included in this category.
Construction costs can vary from shift to shift. Additional construction
costs that can be attributed to night work include: lighting, additional
traffic control, inspection, labor premiums, overtime, and increased
material costs. A recent survey of agencies and contractors helped
identify the cost difference by project element for night construction
(Hinze & Carlisle, 1990). The results are shown in Table 2.5.
Material costs may be higher during the night shift due to batch
plants charging higher rates. Material prices are increased because of
hiring night employees at the plant and also because of shift differential
and overtime. Price (1985) in his report entitled Niqhttime Paving,
investigated a cost comparison between two similar overlay projects in the
Denver area. One job was done during the day on 1-25 north of Denver, and
the other was done during the night on 1-70. Table 2.6 (Price, 1985)
shows the material cost percentage that the night (1-70) job was over that
of the compared day (1-25) job. Although asphalt prices are subject to
variation, this table gives an idea of expected material cost increases.
Table 2.5 Estimated Increase in Cost Differential of Night Work
Project Element Increase in Cost
Lighting 63 %
Traffic Control 28 %
Engineering Inspection 22 %
Labor (Shift Premiums) 18 %
Overtime (Agency Personnel) 16 %
Material 5 %
Total Contract Amount 9 %
Source: Hinze and Carlisle (1990)
Table 2.6 1-70 Night Paving Project vs. 1-25 Day Paving
Percent Cost Differential
Cost Item 1-70 Night Paving
HBP (Patching) (H-Asphalt) / ton 23 % Higher
HBP (GR EX) (H-Asphalt) / ton 23 % Higher
EMUL ASPH (CSS-1 H) / gallon 42 % Higher
Flagging / hour 71 % Higher
Average / Item
= 159 % higher at night
= 40 % higher at night
Source: Price (1985)
The table shows that for the night project, the price for hot bituminous
pavement (HBP) (patching) (haul and asphalt) per ton was 23% higher than
that of the day project. The price for emulsified asphalt (CSS-1H) per
gallon was 42% higher. From a survey of various state highway agencies,
Hinze and Carlisle (1990) concluded that the average percent cost
differential for materials during nighttime construction is approximately
5% higher. Special materials required for night work may also contribute
to greater cost. Conditions of the night, such as cooler temperature may
require a higher cost specific material.
A cost comparison study of asphalt and base widening roadway work
for day shift versus night shift was done by R. G. Layfield, resident
engineer of the Florida Department of Transportation (FDOT). The study
indicated a 2-3% increase in material costs for asphalt roadway work
during the night (Layfield, 1988). The daytime unit price for limerock
base (4 inch) was $4.50 per square yards compared to its nighttime unit
price of $6.34 per square yards, indicating a 3% increase in night cost.
The nighttime unit cost for asphalt was $33.07 per ton compared to $31.50
per ton for its daytime unit cost, resulting in an increase of 2% in night
For scheduling reasons, special material (e.g. rapid curing
concrete) may be utilized (Fallin, 1990). On the other hand, if the
contractor owns the batch plant, material acquisition costs may be lower
during the night shift due to less traffic interference on the road. When
materials can be stockpiled during the day, the only additional cost may
be the cost to double handle the material. Thus, material costs may or
may not be a factor during night operations.
Labor cost differential is another item that may increase during the
night shift. Literature review did not reveal detailed information on
this area. According to Willis (1982), labor is considered to be one of
the premium costs for nighttime paving. It could be expected that some
incentive should be used to influence the performance of nighttime
overlay. For example, if the contractor works 8 hours per night, the
contractor's workforce could be paid for 9 hours. Willis states that
there should not be any across-the-board pay increase for nighttime work.
If the contractor works 6 nights a week, there will be overtime on the
sixth night. The contractor may be subjected to specific union rules for
its use of labor force during nighttime work. This possibility should be
thoroughly investigated before preparing labor cost estimates (Willis,
1982). Labor and inspection costs are additional cost items for night
construction. Shift premiums accounted for an increase of 18% in direct
labor costs, while overtime costs for agency personnel required an
additional 16% (Fallin, 1990). A Florida contractor agreed to pay $0.50
per hour extra for all of his personnel involved in nighttime operations
(Layfield, 1988). The shift differential can vary depending upon the
specific time period. According to the 1988 report, "Highway Maintenance
Activities During Low Traffic Hours", prepared by the California DOT,
there is a 25 cent per hour differential if four or more hours are worked
in the swing shift (6:00 p.m. to 12 midnight). The pay differential for
four or more hours worked in the graveyard shift (midnight to 6:00 a.m.)
is 35 cent per hour.
The survey done by Hinze & Carlisle indicates that artificial
lighting is one of the most significant items in terms of differential
cost between day and night construction. Table 2.5 shows the estimated
increase (63% for lighting) in the cost differential of night work. Since
the cost of artificial lighting is unique to nighttime work, it can be
regarded as a project specific cost, when rented or leased lighting
equipment are used on an "as needed" basis. To minimize setup costs and
maximize efficiency, equipment mounted lighting is recommended. For
contractors involved regularly in night work, it is best to purchase their
own lighting systems. Fallin (1990) mentions the potential use of laser
technology to reduce the need for artificial lighting. Although currently
this alternative is very expensive, it may be a viable option in the near
future. Lum (1980), in his report published in Public Roads, investigates
the economic feasibility of lighting an entire construction zone to reduce
night traffic accidents. Table 2.7 (Lum, 1980) shows the costs to
purchase, install, and remove light posts. The cost information was
obtained from a number of utility companies and municipalities, and hence
not representative of cost data nationwide. Lum concludes from his
analysis that the benefit/cost ratio for lighting a construction project
is, at all situations, less than 1.
Another major cost factor in night work zone construction is traffic
control. According to Shepard and Cottrell (1985), the cost involved in
traffic control can approach an upper limit of the cost of the permanent
construction work itself. For nighttime construction, there is an added
cost for traffic control, due to the need for additional signs in a low
visibility environment. Additional signs for night shift work may include
changeable message signs, arrow boards, warning signs, and channelizing
devices. Homburger (1989) has written an investigative report detailing
Install, and Remove Light Posts
1. Unit cost of purchasing and
installing wooden pole,
foundation and bracket.
2. Unit cost of luminaire.
3. Unit cost of purchasing and
...........$ I/ lin. ft.
4. Unit cost of energy and
5. Cost of poles, foundations,
brackets, & luminaires.
6. Cost of wiring.
......... ..$ 5,280
7. Cost of energy and
8. Cost of removing poles.
9. Total cost of lighting 1 mile
of roadway for 230 days.
Information Source / Assumptions
Mounting height of 40 ft.; poles
located on one side of roadway.
High pressure sodium, 400 watt.
Overhead as opposed to underground
wiring; costs about twice as much.
Includes energy & minor replacement
or repair of lamps.
100 ft. spacing for 1 mile and one
side only, for a total of 53 units.
1 mile of wiring.
For 230 days.
Estimated at $ 125 per pole
Sum of items 5-8. Figure rounded to
the nearest $ 100.
Source: Lum (1980)
Table 2.7 Costs to Purchase,
traffic control measures in construction zones with nighttime activities.
Apart from signs, other traffic control measures are emergency/enforcement
controls (police and ambulance), and deterrent controls (barricades,
vehicle-mounted barriers). Literature review did not specifically reveal
cost differentials attributable to traffic control during the night.
However, Price (1985) in his analysis of two (night and day) paving
projects in Colorado, mentions that the flagging per hour cost was 71%
higher for the I-70 night paving project.
The cost incurred by travelling motorists due to ongoing
construction and maintenance work on the roadway, can be classified as
"user" costs. Vehicle operating costs, personal costs, and accident costs
are the primary components of this category. Literature review indicates
the availability of detailed information on the first two components, but
very little on accident costs. It is difficult to estimate a cost for the
difference in accidents for night operations versus day operations. There
is a lack of nighttime accident data and considerable variability in
accidents between the types and locations of night projects. Vehicle
delay costs investigated by Price have been discussed previously in the
traffic volume section of the literature review. The calculation of unit
user costs is presented in A Manual on User Benefit Analysis of Highway
and Bus Transit Improvements (AASHTO, 1977). Tables for operating costs
and time value are utilized to determine unit user costs. Traffic volume
and work-zone capacity are used to calculate the vehicle operating cost
and the queuing cost due to time delays. This methodology allows the
comparison of alternate work shifts to evaluate the work zone cost impact
and the optimal shift time to affect a lane closure.
A method which combines user operating and time costs has been
discussed by Lytton and McFarland (1975). Using workzone location as a
parameter, tables are provided to calculate the user costs. The
limitation of their cost model is that it is applicable only for an
overlay type of road work. Dudek and Richards have developed a computer
model, using data from 14 work zone sites in Texas. Hourly traffic
volumes, capacity, traffic control method, and other parameters can be
entered into the QUEWZ model to determine user costs with precision and
efficiency (Dudek and Richards, 1985).
Quality of Work
In some situations, the quality of work during the night shift may
be lower than that on the day shift. Poor visibility during the night may
result in less than perfect finishing of paving jobs. Worker morale,
unsafe working conditions, and quality control for materials are factors
that are shift dependent and these may affect construction quality.
According to Shepard and Cottrell, the key word is "acceptable quality",
when considering the quality of work during night shift. Even if quality
is suspect during night operations, the product may be accepted under the
From the literature review it is not possible to conclude that
quality is always adversely affected during the night shift. Some
resident engineers of the Arizona DOT have found better quality control
during the night for certain projects requiring cooler temperatures for
material and equipment (Kay, 1985). One state official noted that quality
during the night was better because the night hours allowed total road
closure, which provided safer working conditions.
The comparison of day and night paving jobs in Colorado done by
Price indicates that the overall quality of the nighttime work was very
similar to that of the daytime project. Test results from the 1-70 night
paving job and from a similar daytime job on 1-25 are compared in Table
2.8 (Price, 1985). The results indicate that compaction on the night job
was as high as that of the day job, and in some cases better. Quality was
affected mostly in aesthetics, such as roller marks being more apparent on
the finished product. The Colorado study was an indication of the success
of night paving operations from a quality standpoint regarding smoothness.
However, pavement densities during nighttime work were more difficult to
attain, possibly due to cooler temperatures. Because of limited
visibility during the night, crack filling and rolling operations were
identified as being more challenging tasks. High levels of illumination
were recommended for these activity types. One other point mentioned by
Price in his Colorado study was the difficulty in detecting approaching
rain storms which could affect the pavement material quality.
The night shift provides cooler temperatures which permits easy
placement of cement concrete or paving in certain locations during summer
time. High temperatures during the day may cause increased evaporation
and faster rate of set (Hinze & Carlisle, 1990). During the placement of
26 miles of single lane concrete roadway in California, it was realized
that the quality of concrete placed at night was better. The cooler night
temperatures allowed the concrete to be placed closer to its ideal
temperature of 70 degrees Fahrenheit. This resulted in good strength for
concrete and minimized any surface evaporation (Fallin, 1990).
Paving Quality Test Results from Field Samples
Test Night Paving (1-70) Day Paving (1-25)
No.of Samples = 62 No.of Samples = 62
Compaction Mean = 96.1 Mean = 95.6
Specified = 95.0 Specified = 95.0
STD.* = 0.810 STD.* = 0.735
Asphalt No.of Samples = 47 No.of Samples = 47
Content Mean = 5.49 Mean = 5.81
Specified = 5.0-6. Specified = 5.3-6.3
STD.* = 0.267 STD.* = 0.222
Field No.of Samples = 62 No.of Samples = 62
Specific Mean = 2.22 Mean = 2.23
Gravity Specified = 2.31 Specified = 2.33
STD.* = 0.019 STD.* = 0.018
Standard Deviation of Samples
Source : Price (1985)
Tabl e 2.8 -
A recent survey of state highway agencies revealed that general
results of paving operations conducted during the night shift produced
less quality results (Hinze & Carlisle, 1990). Both portland cement
concrete and asphalt had apparent finish defects when placed during the
night. However, the results were acceptable to the state agencies and
satisfied the specification standards. The quality rating (on a scale of
1 to 7) of certain tasks performed during the night, is shown in Table 2.9
(Hinze & Carlisle, 1990). Compaction of subgrade is the highest rated
night construction task and all the crack sealing related activities are
Literature study indicated the following problem areas in paving
quality during the night shift :
roughness of paving surface
S inconsistency in the mix
S poor compaction
S cold joints
S inspection of work
S less control on tack spread
S repairing roller marks
Quality of work is highly dependent on visibility. Almost all of
the above quality related problems result from inadequate lighting at the
project site. In the Colorado project, shadows were reported as a problem
in the rolling and crack filling operations. Proper illumination with
additional portable flood lights can reduce many of these defects.
Table 2.9 Quality Rating of Nighttime Operations
Activity Performed At Night Quality Rating
Compaction of Subgrade 4.88
Surface Compaction of Asphalt Concrete 4.77
Installing Guard Rails 4.73
Placing PCC Pavement 4.54
Pavement Marking 4.36
Crack-Sealing on PCC Pavement 4.19
Crack-Sealing on AC Pavement 4.10
Source: Hinze and Carlisle (1990)
Job Types in Highway Construction
The type of highway work to be done often dictates the selection of
workshift hours. Highway construction and rehabilitation activities can
be classified into two types: fixed-location projects and moving projects
(Homburger, 1989). Fixed-location projects are more commonly found --
they involve work on highway and freeway structures, new paving which
proceeds relatively slowly (e.g. adding lanes, completely rebuilding
existing pavement), revising interchanges, etc. Moving projects are
typically pavement overlay projects which move relatively quickly from one
end of the job to the other.
Moving projects require special alertness from both motorist and
worker since the control conditions change quickly. If traffic control is
not a factor, then moving projects are undoubtedly preferred to be
undertaken during daytime hours. At a fixed-location site, the control
change can be programmed in advance and signs can be posted alerting
motorists to the changes ahead of time. Thus, fixed-location projects are
relatively safer and suited for night shifts.
The following fixed-location and moving job types were identified
from literature review as being more appropriate for the night shift
because of comfortable working temperatures, scheduling, and ease of
(1) Widening and replacement of concrete deck (using precast slabs) for
multi-lane divided bridge in a metropolitan location. Usage of
polymer concrete -- which sets to 4000 psi in one hour at
temperature range of 20 to 100 degrees F (Fahrenheit) -- can allow
the bridge to remain fully open during the day (Pasko, 1985).
(2) Interstate widening and reconstruction (e.g. from 4 lane to 6 lane)
through a metropolitan city.
(3) Concrete median barriers safety upgrading, pavement (concrete joint)
patching, shoulder work, on multi-lane divided urban highway, with
one or two lanes closed.
(4) Resurfacing operation on a multi-lane divided urban freeway, with
all lanes closed, and detour set up to follow parallel arterial
routes (Strakovits, 1974).
(5) Installation of raised pavement markers on freeways.
(6) Grooving of concrete and asphalt pavement to improve friction
(7) Placing of D-mix asphalt on an urban interstate highway. D-mix
placement requires minimum temperature of 60 degrees F (Hudson,
1991). During cold seasons this operation may need to be done
during the day shift.
(8) Pavement overlay on multi-lane freeway (approximately 10 mile
stretch) involving an asphalt overlay on existing portland cement
Table 2.10 contains a list of highway projects undertaken by the
Florida DOT for night schedule at various counties as of March, 1991.
Approximately half of these projects involved resurfacing operations.
Table 2.11 shows a list of tasks which have the potential of better
performance during the night shift. These activities are shown in
descending order of preference (Hinze and Carlisle, 1990).
Shifting all maintenance work in a given area to a night schedule
may not be considered practical. Certain types of roadway repair
activities which are temperature sensitive such as asphalt paving,
Table 2.10 : Florida DOT Projects Using Night Work
Type of Work Road County
Repair 18 Bridges
Skid Hazard Resurfacing
Add Turn Lanes
Milling and Resurfacing
Milling and Resurfacing
Paving Shoulders and Resurfacing
Widening and Resurfacing
Add Lanes and Reconstruct
Construct Grade Separation
Add Lanes and Resurface
Add Lanes and Resurface
Replace Low Level Bridge
Add Lanes and Resurface
Skid Hazard Resurfacing
Source : F.D.O.T. (1991)
Table 2.11 : Construction Activities Performed Better At Night
No. of Respondents
Construction Activity Naming The Activity
Asphalt Concrete Paving 6
Bridge Deck Rehabilitation 5
Pavement Marking 5
PCC Paving 4
Setting Girders 2
Concrete Joint Repairs 2
PCC Pavement Spall Repairs 2
PCC Pavement Slab Repair 2
Latex Modified Concrete 2
Grooving PCC Concrete 1
Large Concrete Pours 1
Asphalt Concrete Removal 1
Loading and Hauling Dirt 1
Preparing for Structural Pours 1
Backfilling Structures 1
Shifting Traffic 1
Installing Guard Rails 1
Street Lighting 1
Grading/Crushing Aggregate 1
Source : Hinze and Carlisle (1990)
chip seals, and slurry seals may still need to be done by day shift crews.
Other activities, such as landscape work, can continue to be performed
during the day without causing traffic congestion. And certainly, daytime
emergency response to repair needs must continue. Work activities on the
pavement and shoulders are the obvious candidates for night schedule.
Worker behavior during a particular work shift is an important
consideration for project management. The human factor has a high
potential to adversely affect both productivity and safety. Worker
performance is significantly affected by the choice of work shift.
Several studies on human performance (related to shift work), particularly
in the manufacturing industry, were found during the literature search.
Based on this review, three principal factors were identified as having
the most impact on worker performance: 1) sleep; 2) human circadian
rhythms; and 3) social/domestic issues.
Monk uses a model for coping with shift work to illustrate the
relationship between the three human factors. The worker's internal
system and the condition of his external surroundings dictate his ability
to sleep. The human biological clock depends on the degree of
fragmentation of sleep and the daily social demands. Finally, the
worker's social harmony is dependent upon the degree of sleep deprivation
(Monk, 1989). Monk stresses the importance of understanding the
relationship between sleep, the worker's biological clock, and the
domestic aspects of the worker's life. Figure 2.5 (Monk, 1989) presents
a triad of factors for coping with shift work. The three interrelated
Source: Monk (1989)
Figure 2.5 Triad of Shift Work Coping Factors
factors determine the individual's ability to cope with shift work. All
three components complement each other, a failure in one can negate
advances made in the other two. During an atypical working environment,
such as nighttime construction, the project management should try to
maintain a proper balance of these factors among the workers.
Night shift workers have faced the problem of insufficient sleep
during the day because of the difficulty in physiological adjustments and
also because of the heat, light, and noise distruptions of daytime hours
(Delisle, 1990). Lack of adequate sleep can lead to physical and mental
disorders, which can lead to on-the-job accidents (Finn, 1981). A recent
survey has indicated that over 20% of night shift workers in the U.S.
suffer from insomnia, fatigue, digestive disorders, loss of alertness, and
other sleep-related problems. From the point of view of safety and
productivity, it is necessary for the project management to identify those
individuals who are more adaptable to the stresses of night shift work.
According to a study by Vidacek (1986), night productivity rises with
successive night shifts, but eventually declines, while afternoon shift
productivity remains high and consistent. In construction, safety is
dependent on the work habits of the crew. Rotating shift workers often
resort to usage of drugs and alcohol to overcome their sleep deprivation.
Compared to other industries, construction has produced higher accident
results, and thus, the need to have a well rested and properly adjusted
night worker cannot be over-emphasized.
Research indicates that most workers placed on a night shift,
eventually becomes adjusted, as long as the same pattern is followed every
day. Delisle (1990) has examined two basic issues regarding a rotating
shift system: the length of time that workers should be kept on a
particular shift before rotating, and, the steps that management should
follow when implementing a shift schedule to meet their needs. Opinions
are divided between rapid and slow rotating shift schedules. Rapid
rotation means that a person works two or three nights in a row before
going back to standard hours, while slow rotation involves working many
more shifts (20 or greater, with intervening rest days). Scheduling the
shift rotations should depend on the type of work to be performed. A
rapidly rotating shift is best for tasks requiring complex cognitive
performance, involving a high short-term memory load (Delisle, 1990). For
highway construction, however, slow rotating schedule is more appropriate,
since it involves the performance of more simple, non-memory type tasks.
This allows the worker's circadian rhythm (biological clock) adequate time
to adjust to different shift periods, and, as a result, increase work
performance. Delisle emphasizes that, in order to successfully implement
a slow rotating schedule, the project management should encourage the
workers to try to keep on a reversed schedule on their days off. A
complete internal readjustment, at the start of each new work week, will
present added stresses on the human body and negate the benefits of a slow
rotating scheduling system.
Management should consider the type of specialization workers are
trained for, when assigning them to certain shifts. Members of an asphalt
crew, who are accustomed to lengthy work periods, would probably be more
adaptable to atypical schedules such as night shifts. Also the night
shift provides a more comfortable temperature for the crew placing hot
Additional guidelines for management regarding rotating shifts are
outlined by Benjamin (1984). Night schedules should be planned in
advance, so the workers can prepare themselves adequately. The shift
workers require uninterrupted weekends off as often as possible, for
social reasons. Shift duration should be limited according to the type of
tasks performed. It is recommended that tasks involving intensive
physical labor be limited to 9-hour shifts. Management should avoid
scheduling peak work loads on the job site between 3 a.m. and 6 a.m for
safety reasons. Construction scheduling should be such that the critical
activities occur at the start of the night shift, when workers are at
their peak mental alertness.
The unique aspects of night construction can have both negative and
positive effect on productivity. Productivity during a particular work
shift is impacted by several factors, such as traffic volume, job type,
material delivery, lighting, supervision, communication, worker morale,
etc. During typical daytime construction operations there are two peak
traffic loads that actually reduce a work day to a 5 1/2 hour work shift.
Whereas, during the night, the work shift and actual daily working hours
are extended (Layfield, 1988). Availability and supply of material,
equipment spare parts at night also has an effect on productivity.
Artificial lighting, which may vary with the type of project, has a
potential impact on the output of night construction workforce. Certain
human factors, such as the biological clock, also govern the crew
productivity during the night shift. Productivity also depends on the job
type. Projects allowing total road closure may save more than 50% of the
construction time. A state transportation official notes that for a job
requiring large number of road patches, daytime patching (closing one lane
at a time) would take much longer than nighttime patching, when total road
closure is permitted (Shepard & Cottrell, 1985).
In a recent ENR article "Barriers Remain to Safe Work Zones" a
contractor in Tampa, Florida stated that "productivity is 28 to 30% less
at night due to set up and take-down time" (Stussman, 1988). On the
positive side, night shift was reported to have allowed a certain project
in Long Beach, California to be completed in 16 working days, whereas it
would have taken at least 35 working days to complete the same project
during the day shift (Lee, 1969). During a Florida road project, asphalt
was laid down during the night shift at a rate of 147.03 tons per hour,
compared to 98.09 tons per hour for another daytime project (Layfield,
1988). In his feature ENR article "Paving After Dark Turns Profitable",
McConville stated that in a Indiana road project, the contractor reported
10% improvement in speed of hauling cycles and tonnage during night shift.
Another contractor in a Pennsylvania Interstate project reported that
nighttime production increases have been logged as high as 30%
(McConville, 1991). Cooler temperatures and less traffic interference
during the night shift allowed for this increased productivity.
When deciding work shift periods, other miscellaneous factors may
also need to be considered by the state agency. Since each construction
or maintenance task differs in some respect from others, these "other"
factors vary in importance. The areas covered in this section are
supervision/communication, labor unions, parts availability, and liability
(Shepard & Cottrell, 1985).
The ability to supervise and communicate effectively during night
shifts depends largely on the agency's experience. Communication between
on site personnel and higher authority is vital during late shifts. A
person capable of making decisions should either be present at the site or
be "on call", responding as needed. Another problem mentioned by Shepard
and Cottrel is the difficulty in communication after the transition
between night and day personnel. The night engineer-in-charge may be fast
asleep during the daylight hours when project management may require his
Based on information obtained from literature review, labor unions
do not present big problems for night operations. However, if there are
any differences, they should be negotiated and any stipulations should
reflect in the contract bid price.
Availability of spare parts for construction equipment is necessary
in order to maintain steady production at the site. During the night
shift it may be difficult to obtain spare mechanical parts because of
equipment dealers being closed after normal business hours. Shepard and
Cottrell suggest that the highway agencies have extra parts and equipment
available at the project site to ensure continuity of work.
Work-zone safety is a matter of concern during both day and night
shifts. The possibility of lawsuits, resulting from accidents due to road
construction, is an important consideration for agency planners.
Selecting the "safe" working period for both the worker and motorist
reduces the owner's liability.
The number of available references relating directly to highway
nighttime construction, as a whole, are limited. Only a few studies
provide a comprehensive approach towards night shift operations and the
agency shift selection process. Most of the literature related to one
specific aspect of night work, such as traffic control, safety, lighting,
human factors, etc. Numerous research studies pertaining to shift work
have been conducted in the manufacturing industry, only a few are
applicable to the construction field. Several published reports in the
highway construction area have provided information on issues relating to
the planning, safety, and traffic maintenance aspects of nighttime
It can be concluded from the literature review that the work shift
selection by the agency is mostly based on personal experience of traffic
data and congestion analysis. There is no formal step-by-step methodology
that takes all the factors into consideration in order to arrive at a
particular decision. Most of the information obtained from the literature
study is based on opinions, and is not quantitative. Information on shift
work for the manufacturing sector is readily available, but other than the
data on human factors, it cannot be related to highway construction.
Research has been done in specific areas of highway nighttime construction
which can be used as a guideline for state transportation agencies.
However, a more comprehensive study of all the factors involved in day and
night operations needs to be conducted.
KNOWLEDGE-BASED EXPERT SYSTEM TECHNOLOGY
Knowledge-based expert system technology is a branch of artificial
intelligence (AI) that has seen rapid advancement in recent years. An
expert system is essentially a computer program using AI techniques to
assist users in solving complex problems involving knowledge, heuristics
(rules of thumb), and decision-making. The knowledge-based system
approach has received broad attention in construction engineering
literature. In a typical construction engineering environment there are
decision problems that simply cannot be solved by procedural, algorithmic
computer models. In construction, knowledge and experience are used more
often than complex mathematical formulas, making this field ideal for
expert system application.
A knowledge-based system has the ability to provide explanations of
its reasoning making it useful as a management decision making tool.
The objective of this approach is to create intelligent behavior on the
computer through stored knowledge acquired from human experts. The
knowledge-based system or expert system is an "interactive" program,
playing the role of a human expert by utilizing heuristic knowledge.
Heuristics allows the system to make educated guesses, recognize promising
approaches, and narrow down the search process in a solution space.
The goal during the development of an expert system is to capture
specialized knowledge. This knowledge must be within a narrow, well
defined domain, and it should simulate the expert's reasoning process to
provide consultation about a difficult task.
An important difference between an expert system (ES) and a
conventional program is the representation of knowledge or data.
Knowledge in an expert system is usually divided into separate entities or
rules, shielded from the application methodology. Conclusions are reached
from the knowledge-base by invoking inference reasoning techniques. In a
conventional program, data is typically stored in a data base and the data
is manipulated by usage of algorithms, giving numerical results. Table
3.1 shows the difference between expert systems and conventional programs.
Architecture of an Expert System
The typical expert system (ES) structure consists of four primary
components: a knowledge-base, an inference mechanism, a working memory,
and an input/output interface with explanation and help facility (Adeli,
1988). In addition, features to facilitate knowledge acquisition,
debugging, editing, and intelligent interfacing may also be included. The
structure is schematically shown in Figure 3.1.
The knowledge-base is a repository of information available in a
particular domain. It consists of well-established and documented
definitions, facts, rules, as well as heuristics or judgmental information
associated with the problem domain. Knowledge acquisition is the process
by which expert knowledge is obtained from various sources for the
representation in a knowledge-base. The structuring and development of
the knowledge-base is aided by the knowledge acquisition facility.
Table 3.1 Difference Between Expert System and Conventional Program
L. Representation and use of
. Knowledge-base and control
1. Inferential (heuristic)
i. Effective manipulation of
i. Developed by knowledge
engineer with or without
i. Midrun explanation desirable
'. Modifications, additions
1. Oriented toward symbolic
Representation and use of data
Data and control
Effective manipulation of
Developed by programmer
Midrun explanation not
Not as flexible to changes
Oriented toward numerical
Source: Expert Systems for Civil Engineers, ASCE (1987)
Figure 3.1 The Expert System Architecture
The inference mechanism (or reasoning mechanism) controls the
reasoning strategy of the ES by attempting to match the input data with
the information available in the knowledge base, and subsequently draw
conclusions and produce explanations. In a rule-based ES, the inference
mechanism determines the order in which the rules should be fired, and
resolves any conflict among rules when several rules are satisfied. The
mechanism seeks to solve the problem by chaining the rules together, and
eventually provide a conclusion.
Working memory (or context database) is a temporary storage of
information pertaining to the state of the specific problem currently
being solved. It is a flexible database, its contents changing
dynamically. Included in the working memory is the problem information
provided by the user as well as data derived by the system. The context
database contains all the intermediate results of the problem solving
process as well as the solution upon completion of the ES processing.
The user interface provides a link between the user and the expert
system. The user may create or modify a knowledge-base through the
interface by using a editor facility. Access to the knowledge-base for
information utilization is governed by the user interface. The
explanation facility and the help facility are both attached with the
input-output interface. The former provides answers to questions and
justifies answers. The latter guides the user to use the system
effectively and easily. The intelligent interface is a feature that
allows the user to interact with the ES and query the ES. This may
include natural language processors, menus, multiple windows, icons or
The development of the knowledge-base is an important step in the
creation of an expert system. The developer must decide by which method
the knowledge is to be represented in the knowledge base. Procedural and
declarative representations are two different ways to represent knowledge
(Adeli, 1988). In procedural representation, the knowledge is context
dependent, unintelligible and difficult to modify--it is commonly used in
traditional algorithmic programming. Declarative representation permits
knowledge to be context independent, more understandable, and easily
accessible for modifications. For these reasons, expert systems usually
use declarative knowledge representation. The three most widely used
declarative knowledge representation approaches used in current expert
systems are rule-based, frame-based, and logic-based. Other schemes
include semantic networks, and object-oriented methods. The choice of
representation will depend on the type of problem to be solved and the
inference methods available.
The frame system is a network data structure that represents the
relations between concepts, objects, or events, and their attributes. A
frame consists of a number of attributes, called slots, in which different
characteristics of an object or a piece of information are described.
Slots may contain default values, pointers to other frames, or procedures.
A procedure consists of a set of instructions for determining the value of
the slot, which is known as procedural attachment. The frame structure
has advantages in representing sequences of events, and for knowledge
acquisition and modification (Adeli, 1988). An example of a frame
representation is shown in Figure 3.2, where the object represented is:
Location Alachua County
Highway 1-75 North
Number of Lanes 2
Project Length 5 miles
Project Duration 250 days
Material Default: Asphalt Conc.
Traffic Analysis : If needed, determine
roadway capacity and
compare with traffic
Figure 3.2 A Frame Representation of Knowledge
"resurfacing project". The properties (location, highway name, number of
lanes, project length, duration, material used, & traffic analysis) of
this object are contained in the slots. The material slot allows for a
default value. Default values are typically used when representing
knowledge in domains where exceptions are rare. The last slot (traffic
analysis) in the frame in Figure 3.2 illustrate a procedural attachment,
in which instructions for determining an entry are contained.
For representation of concepts, objects, events, etc., semantic
networks can be used. The network consists of a collection of nodes and
connecting links. Like frames, this system also has flexibility for
modification, allowing addition of new nodes and links. In this type of
representation, each node can "inherit" the characteristics of its
connected nodes. Figure 3.3 illustrates an example of a semantic network.
For example the statements "Resurfacing is_a Highway Construction
Activity" and "Vehicle Delay caused_by Lane Closures" can be represented
in a simple semantic net by using the is_a or caused_by relations.
Knowledge can also be represented by logic. The two most common
forms are: propositional logic and predicate calculus (Harmon and King,
1985). Each basic element of the proposition can be either true or false.
Propositions can be connected to each other by the connectives AND, OR,
NOT, EQUIVALENT, IMPLIES, etc. This type of logic is concerned with the
truthfulness of compound sentences. For example, if proposition A is true
and proposition B is false, then "A AND B" is false, but "A OR B" is true.
Predicate calculus is a special subset or extension of propositional logic
in which propositions can contain variables. In this scheme, the
knowledge is represented through a programming language (e.g. PROLOG) to
Figure 3.3 Knowledge Representation Using Semantic Network
describe the facts and relationships in the problem domain (Chen, 1987).
In PROLOG, a fact statement such as "Highway rehabilitation/construction
activity is a state agency responsibility" can be expressed as:
"is a (rehabilitation/construction activity, state agency
This is an example of a predication that is presented by a predicate name
(i.e., "is a") followed by a list of arguments. A set of clauses
represents this rule.
The following form depicts each clause:
consequent: [antecedent-1, antecedent-2,...antecedent-n
Logic-based knowledge representation is shown in the example below:
is a (resurfacing, state agency task) :
[ is a (resurfacing, transit construction activity),
is a (transit construction activity, state agency task) ].
The antecedents and consequent in each clause are predictions. The
consequent is true if the antecedents are true.
The rule-based (or production rules) system has been the most
popular representation approach for developing expert systems. It is the
chosen scheme for the knowledge base structure of the ES decision model
described in this thesis. The knowledge base is a collection of rules
which consist of IF-(antecedent)-THEN-(consequent) statements. The
general form for this type of representation is as follows (Adeli, 1988):
IF [(antecedent 1) ......................(antecedent n)]
THEN [(consequent 1 with certainty ci)...................
(consequent m with certainty cm)
( c = certainty factor )
Each rule is unique by its rule number. The order of rule application is
not specified by the rule number. A rule represents an independent piece
of knowledge. The antecedent can be regarded as a pattern and the
consequent as a conclusion reached or action to be taken. The antecedent
part of the rule is linked to the working memory of the ES. The rule is
fired when all the conditions of the antecedent part are satisfied. Since
the antecedent-consequent or IF-THEN rules can easily be transformed into
questions, this type of representation can facilitate the generation of
explanations. Certainty or confidence factors can be attached to rules in
a rule-based system. Each rule may be assigned a certainty factor
typically in the range of 0 to 10 or 0 to 100. These factors simply
indicate the level of confidence in a piece of information.
The following is an example of a production rule:
IF: Roadway type is 4-lane freeway
and location is urban
and construction activity type is resurfacing
and number of required lane closures >2
and expected day traffic VPH per lane > 1500
THEN: Work should be performed during 7 pm to 5 am.
The consequent part is executed when the condition part provides a match
with the available facts in the working memory.
The rule-based knowledge representation allows the ES to provide
explanations to its conclusions rather easily. This scheme also permits
a natural way for the developer to describe complex knowledge. However,
one drawback of the rule-based approach is that the addition of new rules
or modification of existing rules may introduce contradictions.
The inference engine or mechanism is the heart of an expert system.
It is a built-in reasoning process that determines which rules are to be
fired to reach a conclusion. The three common search techniques used in
an inference mechanism are: forward chaining, backward chaining, and the
hybrid approach. The selection of a particular search strategy depends on
the application area. The details of these techniques will be discussed
in this section.
In the forward chaining technique, the rules are scanned until one
is found whose antecedents (IF-parts) match the information entered in the
working memory. The rule is then fired, updating the working memory. The
process is repeated until a goal state is reached. This search strategy
is recommended when the goal state is unknown and has to be constructed or
the number of possible outcomes is large. Complex planning problems,
particularly in construction management, are well suited for this
Backward-chaining is a goal-driven strategy, in which the rules are
scanned for those whose consequent (THEN-parts) actions lead to the goal
state. These rules are then checked to determine whether their
antecedents (IF-parts) match the information in the working memory. When
a match is obtained, the rule is applied and the solution is reached. For
an unmatched antecedent, a new subgoal is defined as "arrange conditions
to match that antecedent" (Adeli, 1988). The process is applied
recursively. This strategy is particularly efficient when the values of
the goal state are known and its number of possible outcomes is small.
Backward chaining strategy is appropriate for diagnostic expert systems.
The hybrid approach combines both forward-chaining and backward-
chaining to yield conclusions. The "blackboard" environment, as described
by Harmon and King (1985), utilizes this combined approach. The
blackboard model is essentially a central global database maintaining a
two-way communication with independent rule groups (knowledge sources).
An agenda-based control system continually examines all of the possible
pending actions and chooses the one to try next. Processing in the
blackboard model is based concept of independent cooperating experts.
This type of model is appropriate for structuring complex, problem-solving
tasks that require multiple experts.
Knowledge acquisition is the process by which the ES developer
collects knowledge from expert sources. Personal contact with domain
experts, reviewing literature documenting the experiences of the domain
experts, are ways to acquire the required knowledge. This process may be
divided into five steps, as described by Hayes-Roth (1983).
The initial step is identification and characterizing the important
aspects of the problem: the problem domain, knowledge sources, and goals.
The next step is conceptualization, during which the key properties and
relations are made explicit. Knowledge representation ideas and tools are
considered at this stage. The formalization process is the third step.
The model of the task is mapped, its key properties and relations are
expressed into some representation schemes. The fourth step is
implementation, which involves mapping the knowledge into the
representation method associated with the tool (ES shell) chosen for the
model. The testing and revision of the prototype system is the final
step. The knowledge base may be refined, or the knowledge representation
scheme may be redesigned, depending on the results of the sample run.
Personal interview with the domain expert is a direct and efficient
way to extract information from the domain expert. The interview can
be in the form of questions and answers and example problem-solving
sessions. This process entirely depends on the amount of time and effort
the expert is willing to spend. Also, the domain expert may not know how
to describe the decision process, or may simply misunderstand the
questions. To minimize this difficulty, it is best to narrow the focus of
the problem domain.
Protocol analysis is another way to obtain knowledge (Hart, 1985).
Instead of making direct contact with the domain expert, the knowledge
engineer allows the domain expert to perform and record problem solving
procedures. The recorded transcript is then analyzed by the knowledge
engineer with assistance from the domain expert. The advantage of this
technique is that it gives the domain expert more freedom to express his
knowledge. However, in this process, a communication gap between the
developer and the expert will most likely exist.
Automated knowledge acquisition methods, such as the induction
method, may reduce the gaps between the domain expert and the ES developer
(Chen, 1987). This process uses algorithms to obtain the knowledge. The
system typically uses a spreadsheet interface to get examples (describing
a particular problem) in tabular format from the knowledge-base builder.
In this method, the expert provides a set of examples of different types
of decisions and the corresponding attributes influencing the decisions.
Rules are induced by the algorithm using the examples. By this method, it
is not necessary for the domain expert to detail the decision making
Expert System Development Shells
Expert system programming environments or shells are recent
commercial developments, intended to facilitate the building of knowledge-
based expert systems. Programming languages, system-building aids can
also be used as ES development tools. Languages such as LISP, PROLOG, C,
PASCAL, or FORTRAN are typically used in AI applications. The ES
developer is expected to be proficient in computer programming in this
situation. Although LISP and PROLOG have been the most chosen languages
in ES development, currently the usage of C language is gaining popularity
since it reduces operational times for the ES (Barber, 1987). ES-building
aids have not seen much usage to date. Usually, these are knowledge
acquisition systems geared toward assisting the developer in obtaining and
structuring knowledge. Examples of existing knowledge acquisition systems
are ETS, TIMM, and RULEMASTER (Adeli, 1988).
The shell is a commercial ES development tool which is intended to
provide an opportunity for non-programmers to build prototype expert
systems. With shells, the developer can concentrate on the knowledge
representation and not be concerned about mastering a programming
language, or deal with complex inference strategies. The ES shell
consists of a domain-independent inference engine, an empty knowledge
base, and an user-friendly interface. Shells contain specific
representation methods and inference mechanisms, thus they are less
flexible than an AI language such as LISP or PROLOG. Adeli has provided
a detailed evaluation of a variety of shells (both personal computer-based
and main frame-based) currently in use.
ES shells are usually developed by taking out the original knowledge
content from domain-dependent expert systems. As a result, these shells
are suitable only for problems similar to that of the original ES. The
developer must bear this in mind when he or she selects a shell for a
particular application. For the decision problem described in this
dissertation, a PC-based ES shell, EXSYS-Professional, was selected as the
frame for the prototype model development. The software is available at
the University of Florida, Department of Civil Engineering.
EXSYS Professional is a generalized expert system development
package. The shell developed by EXSYS Inc., is written in C programming
language. The principal development tool in Professional is the Rule
Editor. The Rule Editor can be utilized to construct simple rule-based
systems or complex modular blackboard systems. The editor provides menus,
prompts and help -- eliminating the need to memorize complex rule syntax.
All input is in the form of normal English text or algebraic expression.
A command language can be used to control the rule execution, allowing the
developer increased flexibility and control for complex applications.
Hypertext, rule compiler, blackboarding, agenda managers, table lookup,
and interface with external databases (Lotus or dBase) are some of the
useful features of the EXSYS shell. EXSYS programs are able to respond to
an end user's query with a full explanation of the logic used to arrive at
a particular conclusion. The knowledge base of this shell can accommodate
upto 2500 rules in a personal computer. Programs written with EXSYS are
directly compatible between the IBM PC/XT/AT, VAX/VMS and UNIX computers.
Current Applications in Construction Engineering
Several successful ES prototypes have been developed in the area of
civil/construction engineering in recent years. Experts systems are most
effective in non-theoretical applications where judgment and experience
plays an important role and in which there are no single solutions. Some
of the areas which are appropriate for ES applications in construction
are: 1) diagnosis, 2) fault detection, 3) prediction, 4) interpretation,
5) monitoring, 6) instruction, 7) planning, and 8) design.
Levitt (1986) has developed an expert system, "Howsafe", that
evaluates the safety-related aspect of a construction contractor's
organization and operating procedures. A personal computer-based ES
shell, Deciding Factor, was used for this system. The knowledge base
contains documented results of construction safety studies obtained from
various technical reports and journals. By using Howsafe, a construction
manager can evaluate a project's or company's safety practices.
An expert system for the selection of materials handling equipment
for construction of concrete frame buildings is described by Wijesundera
and Harris (1985). The system suggests suitable categories of equipment
for materials handling based on factors such as ground type, soil
condition, structural features, site accessibility, etc. The ES utilizes
an external database containing supplemental equipment information, which
is able to provide more specific recommendations. The ES shell SAVOIR has
been used to develop this system.
The area of construction project monitoring has received some
attention by ES developers. McGartland and Hendrickson (1985) have done
research on cost/time control, and purchasing/inventory control. An
expert system would analyze and verify weekly input to a database of
activity schedules and estimates, recognize cost overruns, time slippage
problems, and diagnose probable causes and offer solutions such as
activity duration/cost adjustment. For purchasing and inventory control,
an ES would be able to minimize the overall materials cost and assist a
project manager to determine the most economical inventory levels. A
forward chaining inference mechanism has been suggested by McGartland and
Hendrickson for this type of application.
Stone & Webster have developed expert systems to solve a variety of
problems in the area of welded construction (Hathaway and Finn, 1986).
A PC-based ES available to field engineers on site allows the appropriate
selection of welding procedures, reducing potential construction delays
and problems. Other ES applications include welder qualification test
selection, weld estimating, and weld defect diagnosis.
The current ES technology has not yet reached the stage where expert
systems are able to completely substitute human experts. Only a few
systems have performed rather close to a human expert. MYCIN, a medical
diagnostic program, is considered to be the first major ES to perform at
a human expert's level. The term "knowledge-based system" may be more
appropriate for present day technology.
The architecture of an expert system differs from that of a
conventional computer program. The knowledge base of an ES is separate
from the methods of applying the knowledge to the problem, contained in
the inference engine. In a conventional program, the problem related
knowledge and the methods for using the knowledge are inter-mixed, which
makes it difficult to modify the program for changes and additions to the
knowledge. Processing in a conventional program is algorithmic in nature,
symbols are used to represent numbers, arithmetic properties, and
mathematical operations. In a knowledge-based system, the inference
engine governs the sequence of rules that are fired to lead to multiple
actions or to no action at all. The knowledge rules may include
heuristics as well as mathematical reasoning.
There are three ways knowledge can be represented in an expert
system: rules, frames, logic, and semantic network. Each method is suited
for a particular type of problem. The frame method is appropriate for a
complex knowledge system. The rule-based system is the most popular
method for knowledge representation, more applicable to a decision problem
with a narrow domain.
Due to availability of micro computer-based ES shells, knowledge
engineers are able to develop decision models without the usage of
programming languages such as PROLOG and LISP. However, shells are only
geared toward specific representation methods and are less flexible than
AI languages. There have been several successful knowledge-based systems
developed in the construction engineering field. The knowledge-based
approach represents engineering expertise, theory, and judgment in an
integrated manner and also offers flexibility in data modifications.
PROJECT COST AND PRODUCTIVITY DATA FOR DAY AND NIGHT SHIFTS
The selection of day shift over night shift or vice versa may be
influenced by the project costs of the alternatives. Literature review
indicated a lack of project cost information for an effective comparison
between daytime and nighttime construction. Project related cost is
essentially the contract cost (total work item cost) plus agency
administrative costs (planning, evaluating and monitoring). Most highway
projects are unique, and usually consist of different sets of work items.
This makes it difficult to compare the construction costs for day and
Productivity may also influence shift selection. Shift productivity
is affected by several factors, which include traffic volume, type of
work, material delivery, lighting, supervision, communication and worker
morale. High daytime traffic volume affects productivity negatively.
During the day, the work shift is reduced to a 5 or 6 hour period due to
the morning and evening rush hours. While during the night, the actual
working hours are extended. However, poor lighting and low worker morale
during night can decrease crew productivity. In the following sections,
unit project costs and productivity rates are compared between day and
night shifts, based on data obtained from the Florida Department of
Project Cost Comparison
Since no two highway projects are exactly the same, work items
may differ accordingly. To make an effective cost comparison, a set of
typical work items have been selected for this study. These work items
were selected based upon: a) their usage during a typical day as well as
night project, b) the significance of their contribution in project cost,
and c) their large quantities. The 8 common items are listed as follows:
1) Removal of existing pavement (unit of measure = square yard, SY)
2) Regular excavation (unit = cubic yard, CY)
3) Bituminous material-prime coat (unit = gallon, GA)
4) Bituminous material-tack coat (unit = GA)
5) Milling existing asphalt pavement-2" depth (unit = square yard, SY)
6) Class I concrete-miscellaneous (unit = CY)
7) Type S asphalt concrete-including bitumen (unit = ton, TN)
8) Asphalt concrete friction course-including bitumen (unit = SY)
The unit prices for the above work items were obtained from the FDOT
official cost estimate for road projects done in 1990. Table 4.1 shows
the statistical summary of the rates for these selected work items
performed during daytime (Ellis, Herbsman, & Kumar 1991). A similar
statistical summary for the eight items done during nighttime is shown in
Table 4.2. In both tables, for each work item, columns 4 to 8 contain:
1) number of samples, 2) mean unit cost, 3) standard deviation of unit
cost, 4) highest unit cost, and 5) lowest unit cost, respectively.
The results of an item-by-item comparison of unit prices are
tabulated in Table 4.3, so that the variation in means between day and
night rates can be determined. Columns 4 and 5 of Table 4.3 show mean
Table 4.1 Statistical Summary of Unit Costs for
All Daytime FDOT Projects in 1990
Selected Work Items for
Pay Item Number Mean Std.Dev. High Low
Number Name of Item Unit of
Samples $/unit $/unit $/unit $/unit
110-4 Rem. exist. pavt. SY 104 10.52 10.99 100.0 0.39
120-1 Regular excavation CY 151 7.41 7.71 60.0 0.42
300-1-1 Bit. mat'l-prime GA 55 2.30 1.70 6.5 0.01
300-1-3 Bit. mat'l-tack GA 190 1.36 1.36 12.6 0.01
327-70-5 Milling existing SY 23 0.68 0.26 1.5 0.32
400-1-15 Class I concrete CY 70 348.38 234.42 1050 10
5331-2 Type S asph. conc. TN 188 45.88 34.14 382 19
5337-1-2 Asph. conc. fric. SY 102 1.26 0.82 5.54 0.65
Source: Ellis, Herbsman, & Kumar, 1991
Table 4.2 Statistical Summary of Unit Costs for
All Nighttime FDOT Projects in 1990
Selected Work Items for
Pay Item Number Mean Std.Dev. High Low
Number Name of Item Unit of
Samples $/unit $/unit $/unit $/unit
110-4 Rem. exist. pavt. SY 22 9.54 9.48 50.0 1.84
120-1 Regular excavation CY 20 4.59 2.83 14.1 1.13
300-1-1 Bit. mat'l-prime GA 12 5.19 3.89 15.0 1.00
300-1-3 Bit. mat'l-tack GA 26 1.00 0.33 2.1 0.68
327-70-5 Milling existing SY 19 0.81 0.49 1.65 0.27
400-1-15 Class I concrete CY 17 401.48 153.11 800 125
5331-2 Type S asph. conc. TN 25 34.06 11.93 75 22
5337-1-2 Asph. cone. fric. SY 23 1.27 0.50 2.65 0.75
Source: Ellis, Herbsman, & Kumar,
Table 4.3 Difference Between Day and Night Unit Costs for Selected
Work Items for All FDOT Projects in 1990.
Pay Item Name of Item Unit Mean Mean Test
Number (night) (day) Amount Per- Result
($/unit) ($/unit) cent
110-4 Rem. exist. pavt. SY 9.54 10.52 -0.98 -9.3
120-1 Regular excavation CY 4.59 7.41 -2.82 -38.1 *
300-1-1 Bit. mat'l-prime GA 5.19 2.30 2.89 125.6 *
300-1-3 Bit. mat'l-tack GA 1.00 1.36 -0.36 -26.5
327-70-5 Milling existing SY 0.81 0.68 0.13 19.1
400-1-15 Class I concrete CY 401.48 348.38 53.1 15.2
5331-2 Type S asph. cone. TN 34.06 45.88 -11.82 -25.8 *
5337-1-2 Asph. conc. frict. SY 1.27 1.26 0.01 0.8
asterisk (*) in the
the hypothesis test
column indicates significant difference
that work item at 95% confidence level.
Source: Ellis, Herbsman, & Kumar, 1991
rates for nighttime and daytime projects respectively. Column 6 of the
table contains the difference in amount for means, where a negative value
indicates lower nighttime costs. For item, bituminous material-prime
coat, percent difference is as high as 125.6%, as shown in column 7. For
item, regular excavation, percent difference is -38.1%, indicating a lower
nighttime unit cost. Although the percent differences have high
variations, these variations are not necessarily conclusive from a
statistical point of view. The significance of these differences can be
tested by performing statistical t-tests for the eight work items at a 95%
confidence level. Many of these differences appear to be inconclusive
because of high standard deviations. The null hypothesis was rejected for
only three items: regular excavation, bituminous material-prime coat, and
type S asphalt concrete.
The total project item cost depends on what items are involved and
on the quantities of those work items. For this reason, further study of
the impact of shift work on project costs was done. Quantity data from
eight selected projects, utilizing most of the above mentioned work items,
was obtained from the FDOT. Table 4.4 lists the corresponding quantities
of work items of the selected projects. Respective item costs for the
eight projects were determined by multiplying the unit costs from columns
4 and 5 of Table 4.3 with the quantities from Table 4.4. The probable
difference in project costs for night and day operations due to the eight
work items, is given by the summation of such products. The total costs
of the work items for each of the selected eight projects is listed in
columns 2 and 3 of Table 4.5. Column 4 shows the difference of day and
night total costs, while the last column shows the percentage difference
with respect to daytime total cost.
Table 4.4 Quantities of Work Items for Eight Selected FDOT Night Projects
Name of Item Unit Projects
#1 #2 #3 #4 #5 #6 #7 #8
Rem exist. pavt SY 416 50 1156 532 96 263 2200
Reg. excavation CY 13398 17909 9952 8279 9306 8595 8137 110041
Bit. mat'l prime GA 86 8888 50 50 50 50 660 -
Bit. mat'l tack GA 18205 44551 41197 13874 19211 31116 8748 74469
Milling asphalt SY 3228 553530 20613 20620 15523 7563 63601 84307
Class I concrete CY 5 1.4 4.58 0.8 9.6 3.2 5 50
Type S asph.conc TN 4403 23284 50010 22889 33935 40972 9275 2912
Asph. conc. fric SY 84008 465686 250955 99472 220267 230100 88005 348080
Source: Florida Department of Transportation, 1990
- Effect of Quantity on Project Costs for Eight
Selected Work Items
Project Total Cost of Eight Work Items Difference of Percentage
Night & Day Difference
# Night Cost ($) Day Cost ($) Costs of Day Cost
1 345,399 440,413 -95,014 -21.6
2 2,006,764 2,246,199 -239,435 -10.6
3 2,138,746 2,768,313 -629,567 -22.7
4 984,170 1,279,185 -295,015 -23.1
5 1,727,683 2,228,934 -501,251 -22.5
6 1,768,486 2,284,877 -516,391 -22.6
7 530,719 655,124 -124,405 -19.0
8 1,230,145 1,586,748 -356,603 -22.5
Source: Ellis, Herbsman, & Kumar, 1991
According to the study described above, the degree of variation in
unit costs for the eight selected work items, for both daytime and
nighttime construction, is very high. Tables 4.1 and 4.2 reveal that the
standard deviation is in most cases nearly 100% of the mean. This
confirms that unit costs in highway construction are highly project-
oriented and are influenced more by project-related conditions rather than
on type of work shift (day or night). The study also rules out the
speculation that nighttime costs are exceptionally higher than daytime
costs. As seen in Tables 4.1 and 4.2, for most work items the maximum
daytime unit cost is higher than the maximum nighttime unit cost, and
conversely the minimum daytime costs are lower than the minimum nighttime
Table 4.3 shows that four items (remove existing pavement, regular
excavation, bituminous material-tack coat, and type S asphalt concrete)
have higher daytime mean unit cost. The other four items have a higher
nighttime mean unit cost. T-tests have confirmed significant differences
for only three work items. Two of these items (regular excavation, and
asphalt concrete) appear to have significantly lower mean unit costs
during nighttime. The third item (bituminous material-prime coat) has a
significantly higher nighttime unit cost. It is reasonable to conclude
that the work item characteristics are responsible for such a manner of
In this study, it is seen that the percentage difference is negative
for all eight of the selected FDOT projects. When the total costs of the
eight items for selected projects are compared, nighttime costs are
observed to be lower than the corresponding daytime costs in the range of
10 to 20%. Although the participation of these item costs to the total
contract cost varies, the pattern allows a prediction of a probable
nighttime cost. This prediction theory is utilized in the cost model of
the proposed expert system as described in a later chapter.
A comparison of day and nighttime productivity rates for typical
highway construction activities was possible by obtaining information from
the Florida Department of Transportation (FDOT). Daytime production rates
were collected from a 1988 report, "Establishing Contract Duration Based
on Production Rates for FDOT Construction Projects", prepared by the
University of Florida Civil Engineering Department. The information
includes: 1) number of observations, 2) mean production rate, 3)
standard deviation, 4) maximum and 5) minimum production rates for each
operation, which are further categorized by project type, local
conditions, traffic conditions. Table 4.6 summarizes this data.
The FDOT provided nighttime production data, in form of daily
reports, for a highway project located on 1-95 in St. Johns County,
Florida. A summary of this data is presented in Table 4.7. Plant mixed
surface and milling existing pavement are the two work items for which
data was collected. The mean and standard deviation of production rates
for these work items are shown in columns 3 and 4 of Table 4.7. The
combined results of all the projects, as listed in this table, are used
for nighttime productivity values. Only limited access facility (e.g.
interstates) observations from Table 4.6 are utilized for daytime
productivity values, in order to make an accurate comparison with the 1-95
nighttime productivity values.
Table 4.6 Summary of Productivity Rates for FDOT Construction Projects
PLANT MIXED SURFACES: STRUCTURAL COURSE
Number of Mean Standard High Low
Category Observa- (Tons/ Deviation (Tons/ (Tons/
tions Day) (Tons/Day) Day) Day)
Reconstruction 147 833 533 2,359 6
Construction 27 623 639 2,863 114
Intersection 15 122 111 356 10
Bridge 9 178 70 274 84
Rural 111 855 616 2,863 6
Urban 72 436 387 1,638 17
Limited 15 1,090 157 1,247 582
Light 20 1,189 761 2,359 119
Medium 81 822 562 2,863 14
Heavy 97 539 426 14 6
Total Combined 198 720 565 2,863 6
MILLING EXISTING PAVEMENT
Number of Mean Standard High Low
Category Observa- (SY/ Deviation (SY/ (SY/
tions Day) (SY/Day) Day) Day)
Reconstruction 94 12,350 7,429 32,028 444
Construction 1 2,274 0 2,274 2,274
Intersection 0 0 0 0 0
Bridge 0 0 0 0 0
Rural 48 14,850 8,108 32,028 444
Urban 32 8,987 4,847 20,533 2,351
Limited 15 10,854 6,765 26,422 3,833
Light 14 20,306 8,159 32,028 5,488
Medium 32 12,137 7,680 29,376 444
Heavy 49 10,011 5,180 26,422 2,274
Total Combined 95 12,244 7,461 32,028 444
Source: Herbsman & Ellis, 1988
Table 4.7 Summary of Productivity Rates for FDOT Nighttime
Project on 1-95 in St. Johns County
PLANT MIXED SURFACES: STRUCTURAL COURSE
Project Number Mean Standard High Low
Number of (Tons/ Deviation (Tons/ (Tons/
Samples Day) (Tons/Day) Day) Day)
78080-3420 14 950.68 348.49 1,428.15 320.86
78080-3421 32 1,110.31 327.06 1,602.68 327.08
78080-3422 29 1,093.52 422.97 1,871.21 196.59
78080-3424 20 1,043.45 379.43 1,644.97 325.92
Total Combined 95 1,067.59 378.57 1,871.21 196.59
MILLING EXISTING PAVEMENT
Project Number Mean Standard High Low
Number of (SY/ Deviation (SY/ (SY/
Samples Day) (SY/Day) Day) Day)
78080-3420 7 11,246.42 5,582.80 16,840 2,766
78080-3421 10 7,379.40 1,061.80 9,080 5,333
78080-3424 12 8,256.40 2,957.20 13,864 4,170
Total Combined 29 8,675.70 3,711.90 16,840 2,766
Source: FDOT (1990)
Table 4.8 Guidelines for Estimating Production Rates for FDOT Projects
WORK ITEM DAILY COMMENTS
Clear & Grub
Normally for all projects unless specific
circumstances justify additional time.
Medium clearing (50 to 100 acres)
Small quantity jobs under 100,000 CY
Medium quantity jobs 100,000 300,000 CY
Large quantity jobs over 300,000 CY
Small quantity jobs under 100,000 CY
Medium quantity jobs 100,000 300,000 CY
Large quantity jobs over 300,000 CY
Double lift installation.
Single lift installation.
Average quantity jobs (under 25,000 CY)
Large quantity jobs (over 25,000 CY)
Average jobs (less than 50,000 Tns)
Large jobs (over 50,000 Tns)
(less than 5000 LF)
(over 5000 LF)
Source: Herbsman &