Citation
Estimation of annual incidents delays on Denver metro highways

Material Information

Title:
Estimation of annual incidents delays on Denver metro highways
Creator:
Marlina, Susi
Publication Date:
Language:
English
Physical Description:
x, 64 leaves : illustrations ; 28 cm

Subjects

Subjects / Keywords:
Traffic congestion -- Colorado -- Denver Metropolitan Area ( lcsh )
Traffic accidents -- Colorado -- Denver Metropolitan Area ( lcsh )
Traffic accidents ( fast )
Traffic congestion ( fast )
Colorado -- Denver Metropolitan Area ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 63-64).
General Note:
Department of Civil Engineering
Statement of Responsibility:
Susi Marlina.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
75391651 ( OCLC )
ocm75391651
Classification:
LD1193.E53 2006m M37 ( lcc )

Full Text
ESTIMATION OF ANNUAL INCIDENTS DELAYS
ON DENVER METRO HIGHWAYS
By
Susi Marlina
B.S., University of Bandar Lampung, Indonesia 2000
Diploma. Eng., University of Lampung, Indonesia 1997
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Civil Engineering Department
2006


2006 by Susi Marlina
All rights reserved


This thesis for the Master of Science
degree by
Susi Marlina
has been approved
by
Stephan Durham
David C. Mays

'ate


ABSTRACT
Traffic congestion is a daily problem around the world. Reducing congestion is
important because congestion is costly, wastes time, and produces pollution. This
study performs analyses of incident occurrences by time of day, day of week, and
month of year on several Denver metro highways for the years 2001 and 2002. This
study focused mainly on estimating vehicle delays caused by crashes due to their lane
positions, capacity reductions, and approximate clearance durations. Crashes have the
longest clearance time among incident types on these highway segments, with 15
minutes 6 seconds in 2001 and 16 minutes 11 seconds in 2002. In this study, there are
two scenarios outcomes for queuing delay prediction. Case I occurs when an incident
cleared before demand decreases and the queue dissipates, and case II occurs when an
incident is not cleared until after demand decreases and the demand reduction must be
estimated in order to predict queue dissipation. The overall finding was that there was
an increase of nearly 20% in the average total delay per crash, an increase of nearly
10% in the number of reported crashes, and an increase of 30% in total vehicle hours
of delay on these highways sections over these two years.
This Abstract accurately represents the content of the candidates thesis. I recommend
its publication.
BruceN.Janson


DEDICATION
I dedicate this thesis to my parents for their unfaltering understanding, praying and
support to reach my dreams.


ACKNOWLEDGEMENT
I am sincerely very grateful to the Almighty Creator of the World.
I want to express my gratitude to the following people: my advisor, Bruce N. Janson,
for his patience with me during these past two years; my other professors in the
University of Colorado at Denver who always encourage me to pursue wisdom and
good understanding of knowledge.
I am very appreciative of the efforts of all of my colleagues and my best-friends who
really supported, gave me data, great ideas, and motivation to create this study. I
could not possibly list all of their names.
My deep thanks for all the people in Indonesia in general, to Bandar Lampung
University, the Technological Professional Skills Development sector Project in
particular, for giving me the courage to look at the truth regardless of the outcome,
and for sending me to study overseas.
And especially, for my special friend, my soul-mate, for gently loving me out of my
pain, for bringing stability into my life and peace into my heart and my mind. Lastly,
I want to thank my family who is always support and prayerful for me.


TABLE OF CONTENTS
Figures.........................................................................ix
Tables.......................................................................... x
Chapter
1. Introduction................................................................ 1
1.1 Background.................................................................. 1
1.2 Objectives Study............................................................ 8
1.3 Organization of Thesis...................................................... 9
2. Literature Review...........................................................10
2.1 General Overview............................................................10
2.2 Incident Types..............................................................11
2.3 Incident Management Activity................................................13
2.4 Estimation of Queuing Delay.................................................19
3. Data Collection.............................................................24
3.1 Description of CDOTs Incident Management Operations........................24
3.2 Data Classification.........................................................25
3.3 Denver Area Highway Incidents in 2001 and 2002..............................27
3.4 Period of Incident Data Collection..........................................28
4. Methodology
37


5. Summary of Data........................................................39
5.1 Incident Occurrences by Time of Day, Day of Week, and Month ofYear.....39
5.2 Incident Types.........................................................43
5.3 Incident Clearance Times...............................................44
5.4 Estimating Volumes and Delays by Time of Day...........................48
5.5 Estimation of Traffic Delay............................................49
6. Conclusions and Recommendations........................................60
6.1 Conclusions............................................................60
6.2 Recommendations........................................................62
Bibliography................................................................63
vm


LIST OF TABLES
Table 1.1 Denver Roadway Congestion Index Value, 1982 1990................. 2
Table 1.2 Incidents on Denver Metro Highways (2001 and 2002)................. 7
Table 3.1 Denver Metro Highway Study Location................................29
Table 3.2 1-25 Highway Crash Data from Arapahoe Rd. to Lincoln Ave..........30
Table 3.3 1-25 Highway Crash Data from Evans Ave. to Arapahoe Rd............31
Table 3.4 1-25 Highway Crash Data from 1-70 to Evans Ave....................32
Table 3.5 1-25 Highway Crash Data from 104th Ave. to 1-70...................33
Table 3.6 1-70 Highway Crash Data............................................34
Table 3.7 1-225 Highway Crash Data...........................................35
Table 3.8 US-6 Highway Crash Data............................................36
Table 4.1 MHCP Incident Types in 2001 and 2002 ..............................38
Table 5.1 Lane Positions of Disabled Vehicles after Crashes..................44
Table 5.2 Average Clearance Times by Incident Type...........................45
Table 5.3 Incident Clearance Times by 1-Minute Interval......................46
Table 5.4 Percentage of Lane Position........................................50
Table 5.5 Hourly Volume Factor Percentages...................................51
Table 5.6 Total Vehicle Delay and Average Total Vehicle Delay per Incident...59
IX


LIST OF FIGURES
Figure 1.1 Causes of Congestion................................................ 1
Figure 1.2 Growth of Peak-Period Congestion in the U.S. (1982-2003)............ 3
Figure 1.3 Mile High Courtesy Patrol Coverage Area............................. 6
Figure 2.1 Reported Incidents by Type..........................................12
Figure 2.2 Phases of the Incident Management Process...........................15
Figure 2.3 Timeline of the Incident Management Process.........................16
Figure 2.4 Schematic of Traffic Flow during an Incident........................21
Figure 2.5 Estimation of Queuing Delays during an Incident.....................23
Figure 3.1 CDOTs Traffic Management Organization Chart........................24
Figure 3.2 Incident Type Classifications.......................................28
Figure 3.3 Map of 1-25 from Arapahoe Rd. to Lincoln Ave......................30
Figure 3.4 Map of 1-25 from Evans Ave. to Arapahoe Rd........................31
Figure 3.5 Map of 1-25 from 1-70 to E^ans Ave................................32
Figure 3.6 Map of 1-25 from 104th Ave. to 1-70...............................33
Figure 3.7 Map of 1-70.........................................................34
Figure 3.8 Map 1-225...........................................................35
Figure 3.9 Map of US-6.........................................................36
Figure 4.1 Data Processing.....................................................37
Figure 5.1 Incident Proportions by Time of Day in 2001 .......................39
Figure 5.2 Incident Proportions by Time of Day in 2001 .......................40
Figure 5.3 Incidents by Day of Week in 2001 ...................................40
Figure 5.4 Incidents by Day of Week in 2002....................................41
Figure 5.5 Incidents by Month of Year in 2001..................................42
Figure 5.6 Incidents by Month of Year in 2002..................................42
Figure 5.7 MHCP Incident Types in 2001 and 2002................................43
Figure 5.8 Average Clearance Times by Incident Type............................45
Figure 5.9 Clearance Times by 1-Minute Interval in 2001 .......................47
Figure 5.10 Clearance Times by 1-Minute Interval in 2002........................47
Figure 5.11 Numbers of Crashes by Time of Day in 2001 .......................48
Figure 5.12 Numbers of Crashes by Time of Day in 2002........................49
Figure 5.13 Hourly Lane Volume Percentages......................................52
Figure 5.14 Hourly Volume Variation Patterns....................................52
Figure 5.15 Case I occurs when an incident cleared before demand decreases and
the queue dissipates..............................................56
Figure 5.16 Case II occurs when an incident is not cleared until after demand
decreases and the demand reduction must be estimated in order for
queue to dissipate..............................................................57
x


1. Introduction
1.1 Background
Traffic congestion is a daily problem in many major cities around the world.
Congestion is costly, wastes time, and produces pollution. Therefore, congestion
reduction is a significant issue. Congestion is the outcome of inadequate capacity and
high-level demand (Ozbay, et.al. 1999). Two categories of traffic congestion on the
freeway are, recurring congestion or predictable congestion caused by the high
demand of vehicles during rush hour, and non-recurring congestion caused by
incidents. Furthermore, there is a correlation between incidents and congestion. As
incidents bring more congestion, more congestion results in more incidents (Ozbay,
et.al. 1999). Figure 1.1 illustrates that the two root causes of congestion are physical
bottlenecks with 40 percent and traffic incidents with 25 percent.
1


Roadway Congestion Index Value is a relative measure of the level of congestion for
given urban area (US DOT, 1993). The roadway congestion index value of 1.0 or
greater indicates an undesirable area wide congestion level. Table 1.1 shows Denver
Roadway Congestion Index Value from 1982 to 1990. Denver congestion has
increased by 21% over 9 years.
Table 1.1 Denver Roadway Congestion Index Value, 1982 1990 (US DOT, 1993)
Year Congestion Index Value
1982 0.85
1983 0.88
1984 0.93
1985 0.96
1986 0.97
1987 0.95
1988 0.99
1989 1.01
1990 1.03
Percent Change 21 %
The most recent data that was reported is from September 1, 2005 by the Cambridge
Systematics Corporation and the Texas Transportation Institute. Figure 1.2 illustrates
that from 1982 to 2003 in the United States largest cities, weekday peak-period
congestion increased in duration from 4.5 hours to 7 hours, increased the percentage
of trips affected from 33 percent to 67 percent, and increased the percentage of travel
2


delay accounted for from 13 percent to 37 percent. Moreover, this total recurring
congestion can be dramatically influenced by non-recurring congestion.
Intensity
Figure 1.2 Growth of Peak-Period Congestion in the U.S. (1982-2003) (Cambridge
Systematicss, 2005)
Nowadays, investment in developing systems to control incidents has increased. The
objective of an Intelligent Transportation System (ITS) is to identify problems
quickly as they develop, manage traffic flow, and provide travelers with information.
The Federal Highway Administration describes incident management as:
the systematics, planned, and coordinated use of human, institutional,
mechanical, and technical resources to reduce the duration and impact of
3


incidents, and improve the safety of motorists, crash victims, and incident
responders (PB Farradyne, 2000).
In order to develop regulations and methods for dealing with a number of types of
incidents, incident management teams are established. Some required parts of these
systems are (1) urgent situation and perilous material response plans, (2) personnel
and equipment resource lists, (3) personnel guidance programs, and (4) agreements
with outside contractors.
Intelligent Vehicle Highway System (IVHS) technology can be used to reduce
congestion on the highways. IVHS comprises several electronic sensing systems,
computing, and communications for managing traffic flow on major highways. IVHS
consists of three basic high-technology approaches to improving traffic flows:
1. Advanced Traffic Management System (ATMS). ATMS assembles
information electronically on congestion and flow situations from many
segments of the highway network. ATMS give information to a control center,
which analyzes it and adjusts traffic signals, ramp entry controls, and lane
direction controls throughout the system to reduce delays.
2. Advanced Traveler Information System (ATIS). ATIS also assembles traffic
data, but gives the information to individual drivers before they depart home
or as they travel so they can adjust routes and timing to prevailing conditions.
4


3. Advanced vehicle control system (AVCS). AVCS is comprised devices in
vehicles or on roadways or both that improve drivers control of vehicle.
Since 1991, the Colorado Incident Management Coalition (CIMC) also began
evaluating traffic congestion on the major Denver area highways. Moreover in 1995,
the Colorado Department of Transportation (CDOT) initiated a courtesy patrol
program as a pilot project to provide incident management on the major roadways,
with the aim of reducing congestion, and handling and clearing incidents faster.
CDOT Region 6 implemented the Mile High Courtesy Patrol (MHCP) service on the
major Denver area highways in 1992. Currently, the service is available on the major
Denver area highways in Colorado as shown in Figure 1.3.
1-25 from 104 th Avenue to Alameda
US 36 from 1-25 to Wadsworth
1-70 from Sheridan Boulevard to Pena Boulevard
1-225 from 1-70 to Parker Road
US 6 ( 6th Avenue 1 from 1-25 to Kipling Street
C-470 from 1-25 to Wadsworth Boulevard
1-270 only on request from law enforcement or if these is a major incident
5


Figure 1.3 Mile High Courtesy Patrol Coverage Area (Nageli, 2003)
6


Table 1.2 shows a summary of all incidents on major Denver area highways in 2001
and 2002. Table 1.2 shows that 1559 crashes, 4974 disablements, and 383 other
incidents occurred in 2001, and 1317 crashes, 3710 disablements, and 118 other
incidents occurred in 2002.
Table 1.2 Incidents on Denver Metro Highways (2001 and 2002)
No. Incident Type Number of Incident in Year 2001 Percent of Incident in Year 2001 Number of Incident in Year 2002 Percent of Incident in Year 2002
1 Crash 1559 22.54 1317 25.60
2 Disablement 4974 71.92 3710 72.11
3 Others 383 5.54 118 2.29
Total 6916 100.00 5145 100.00
There are many strategies that have been used to improve incident management and a
novel approach needs to be implemented to improve incident clearance. This
assessment of incident delays depends on how accurately the crash or incident is
reported. The more time that passes between the date of the incident and when it is
reported, the harder it is to get accurate information, identify the root causes, and
establish response strategies.
7


Overall, no one organization has responsibility of incident management everywhere
in Colorado. Each agency has a significant role in enhancing good incident
management. In a genuine system, the police in Colorado, particularly through the
Mile High Courtesy Patrol (MHCP), are the key service providers that undertake the
crucial duty of supporting incident management.
1.2 Objectives of Study
The objective of this study is to estimate total incident delays on major Denver area
highways as part of the CDOT incident management program, with an emphasis on
CDOTs ability to detect and manage these incidents.
The study will perform the following analyses:
Compare incident occurrences by time of day, day of week, and month of year
for years 2001 and 2002.
Approximate incident duration times, incident lane positions, and the capacity
reductions of these incidents.
Estimate incident delay caused by crashes among these incidents that
occurred.
8


1.3 Organization of Thesis
The thesis consists of six chapters and is organized as follows:
Chapter 1 presents a background of incident management, the objectives, and
scope of this thesis.
Chapter 2 presents a comprehensive review of the literature.
Chapter 3 describes the incident data collected.
Chapter 4 explains the method used to analyze the data.
Chapter 5 presents the results of this analysis.
Chapter 6 presents conclusions and recommendations.
9


2. Literature Review
2.1 General Overview
Incident management is one of the essential elements of freeway or road traffic
operation. A freeway is defined as a divided highway with full control of access and
two or more lanes for the exclusive use of traffic in each direction and freeways
provide uninterrupted flow (TRB, 2000). Incident Management is determined as the
coordination of activities undertaken by one or more agencies to restore traffic flow
to normal conditions after an incident has occurred (Ozbay, et.al. 1999). Traffic
incident management is delineated as an operational strategy for a transportation
network that involves a coordinated and planned inter-jurisdictional, cross-functional,
multidisciplinary, and ongoing approach to restore traffic to normal conditions after
an incident occurs, and to minimize the delay caused by the resulting disruption to
traffic flow (FHWA, 2001).
Due to its unpredictability, nonrecurring congestion is considered high safety
vulnerability, especially on high speed facilities such as freeways, and special
attention needs to be devoted to limiting the number of motorists exposed to this
hazard. The ultimate goal in an incident management plan is to limit the amount of
the time the incident remains, causing a reduction in capacity (Hofener, 2003).
10


2.2 Incident Types
Classifying incident types is a necessary step to estimating incident impacts and
improving an incident management program. Figure 2.1 depicts the percentages of
reported incidents by type, severity and duration from a national study performed by
Cambridge Systematicss (1990). Of the incidents that were reported, 80 percent were
vehicle disablements car and trucks that had run out of gas, had a flat tire, or been
abandoned by their drivers because of some malfunction. During off-peak periods
when traffic volumes are low, these disabled vehicles have little or no impact on
traffic flow. However, when traffic volumes are high, the presence of a stalled car or
a driver changing a flat tire in the breakdown lane can slow traffic in the adjacent
travel lane, causing 100 to 200 vehicle hours of delay to other motorist.
The other 20% percent of incidents consist of 10% crashes and 10% other incidents.
Crashes are mostly the result of minor collisions, such as sideswipes and slow-speed
rear-end collisions. In 60% of crashes, drivers are able to move their vehicles to the
shoulder. Each such incident lasts an average of 45 to 60 minutes. In congested
traffic, they can produce 500 to 1,000 vehicle hours of delay per incident. In the other
40% of crashes, vehicles are blocking lanes and each such incident typically last 45 to
90+ minutes causing 1,200 to 2,500 vehicle hours delay per incident (Cambridge
Systematicss, 1990).
11


TYPE
LOCATION
DU RATIO NlminsK
VEHICLE-HO UR5
OF DELAY (vtuf)
On Shcukitr 15-30 rtnut**;
Al 30% 10C-2CC yhd
Inodtntj
Roora*d
70%
.3ockng L*im 15-30
20% 500-10CC
Accidntt
10%
On Should <5-60 ninutts!
eo% 500-10CC vnc
Oth*r
10%
Bockrg LanM <0% <5-80 rinutts
1200-MODkM
On Shcui 70% 10C-2CC *h Booling Lorn 30-<5 niinotM
30% ICCC-1500 vhd
Unr*oors*s
30%
Figure 2.1 Reported Incidents by Type (Cambridge Systematicss Inc., 1990)
12


2.3 Incident Management Activity
Incident management activities include (Traffic Incident Management Handbook,
2000):
1. Detection:
Incident detection is the process by which an incident is brought to the
attention of the agency and the agencies responsible for maintaining traffic
flow and safe operations of the facility.
2. Verification:
Incident verification entails confirming that an incident has occurred
determining its exact location, and obtaining as many relevant details about
the incident as possible.
3. Motorist Information:
Motorist information involves activating various means of disseminating
incident-related information to affected motorist. Motorist information needs
to be disseminated as soon as possible, and beyond the time it takes clear an
incident. In fact, it should be disseminated until traffic flow is returned to
normal condition. This may take hours if an incident occurs during a peak
period, and has regional impacts.
4. Response:
Incident response includes dispatching the appropriate personnel and
equipment, and activating the appropriate communication lines and motorist
13


information media as soon as there is a reasonable certainty that an incident is
present.
5. Site Management:
Site management is the process of effectively coordinating and managing on-
scene resources. Site management is defined as the process of accurately
assessing incidents, properly establishing priorities, notifying and
coordinating with the appropriate agencies and organization, and maintaining
clear communications with each responder.
6. Traffic Management:
Traffic management involves the application of traffic control measures in
areas affected by incident.
7. Clearance:
Incident clearance is the process or removing wreckage, debris, or any other
element that disrupt the normal flow of traffic, or forces lane closures, and
restoring the roadway capacity to its pre-incident condition.
Effective incident management consists of six main elements, which are frequently
overlapping as illustrated by Figure 2.2. These elements are incident verification,
incident response, incident clearance, recovery, incident site management, and traffic
management/motorist information (FHWA, 2001).
14


Figure 2.2 Phases of the Incident Management Process (FHWA, 2001)
Concisely, there are 4 steps of incident management process: incident detection,
incident response, incident clearance, and recovery (TRB, 1994) as illustrated in the
Figure 2.3.
15


Detection Response Clearance Recovery
v T T
Incident Incident Incident Incident
Occurred Detected Response Cleared
Traffic Flow
Restored to
Normal
Figure 2.3 Timeline of the Incident Management Process (TRB, 1994)
The incident management process encompasses the following four steps:
a. Incident Detection:
Detection Phase the period of time between the occurrence of the incident
and detection by the traffic managers, police, or freeway response team.
Included in this phase is the verification of the incident as severe enough to
warrant a response.
b. Incident Response:
Response Phase the period of time between the detection of an incident and
the arrival of emergency or response vehicles.
16


c.
Incident Clearance:
Clearance Phase the period of time when responding agencies treat victims,
close lanes, and eventually remove vehicles and debris.
\
d. Incident Recovery:
Recovery Phase the period of time after the clearance of an incident for the
traffic flow to return to normal conditions.
Smith, et.al. (2001) developed a forecasting model that can predict the clearance time
of freeway incident with 3 models: a stochastic model, non parametric regression
model, and classification tree model. The stochastic model was not applied to
forecasting future accidents due to inability to accurately fit any probabilistic
distribution to the accident data. The results from the non-parametric regression
models are not encouraging because the model has a very large error and in most
cases is larger than model prediction value. The final feature of classification tree
models that makes them advantageous to traffic managers is the output type. This
model predicts a class or range of values instead of a single numerical value. The
other two models were developed but suffered from poor performance in predicting
the clearance time of future accidents. However, the classification tree model appears
to be well suited for forecasting the phases of incident duration given a database of
incidents with reliable and informative characteristics.
17


Mannering, et.al. (1998) found that hazard-based approaches are well suited to
incident duration analysis because they account for uncertainties in incident duration.
For instance, incident clearance duration-dependence effects shifted from positive to
negative duration dependence (i.e., more or less likely to end soon).
Robinson, et.al. (1993) described that Denver, Colorado began a Mile High Courtesy
Patrol as a 6-month pilot project in August 1992. Two types of patrols were
established:
(1) Colorado State Patrol with push bumpers are providing two vehicles during peak
hours with off-duty officers being paid time and one-half. These vehicles are assigned
to a seven to eight mile stretch, and are dedicated to car assists and freeway incident
management.
(2) Tow trucks are operated on a 15 mile stretch during peak operating hours,
Monday trough Friday. This program was evaluated by the University of Colorado
with regard to cost effectiveness and motorist perception.
PBSJ (2004) developed a manual for a South 1-25 Corridor incident management
program in Colorado with the aim to provide coordination and resources for incident
response on 1-25. The manual contents of incident level descriptions/actions, scene
management guidelines, emergency response staging areas, alternate route
information, existing variable message sign locations, contact and material lists,
18


media packet, accident alert packet, and incident management program evaluation
forms.
Korpal (1992) stated that significant benefits can be produced by a freeway traffic
management system coupled with an effective incident response program. Some
measures can be introduced without major increases in operating budgets and useful
technology and techniques. The key aspects of an effective incident response program
are interagency cooperation, and openness to innovation, and a conviction to reduce
motorists delay.
2.4 Estimation of Queuing Delay
An incident reduces highway capacity and interrupts traffic flow by slowing down or
even stopping the flow of vehicles. The duration of the incident, the number of
blocked lanes, and the volume of traffic approaching the incident location are the
main determinants of total delay caused by an incident (Cambridge Systematics Inc.,
1990).
The total delay caused by an incident depends on the duration of the four incident
phases described earlier, and the number of blocked and unblocked lanes relative to
the traffic demand. An incident causes queuing and vehicle delays because the
19


vehicle arrival rate (hourly vehicle volume) exceeds the vehicle service rate
(unblocked lane capacity) during the first three incident phases.
Hall (2000) developed model to estimate response times and delays for highway
incidents, accounting for the spacing between interchanges and the time needed by a
response vehicle to use these interchanges to change direction and reach an incident
on the opposite side of the highway.
Cuciti, et.al. (1995) explains that the queue dissipation or recovery phase is the time
from capacity restoration to when a normal traffic flow resumes. The dynamics of an
incident are illustrated by Figure 2.4. When an incident occurs, highway capacity is
reduced and traffic queues begin to build. The vehicle hours of delay accumulated by
motorists in the queue are represented by the shaded area between the normal flow
rate and the incident flow rate- the difference between traffic demand and available
road capacity at the incident location. If traffic demand going toward the incident site
is reduced by diverting traffic to alternate routes, the delay will be minimized (dark
gray area, noted as Cumulative with Demand Reduction). However, if traffic is not
diverted, additional delay will accrue (shaded area). Upon incident clearance, traffic
will clear through the incident site until the queue is dissipated. Nevertheless, the
getaway traffic flow will be limited by the maximum capacity of the highway (ATA
Foundation and Cambridge Systematics, 1997).
20


Incident Duration Recovery
Figure 2.4 Schematic of Traffic Flow during an Incident
(Cambridge Systematics Inc., 1990).
Ozbay et. al (1999) illustrates on the Figure 2.5 the effect of an incident on traffic
using the deterministic queuing approach. When an incident occurs, it blocks one or
more traffic lanes, and a queue starts building upstream of the incident due to the
reduction of the capacity. The incident flow is a fraction of the total flow and is
function of the number of lanes available and the type of incident. Thus, quick
incident clearance using effective incident strategies will reduce the clearance time
and time to normal-flow condition. The total vehicle-hours of delay experienced by
the motorists in the queue are represented in the figure. This delay will continue to
21


build until the incident cleared and traffic flow is restored to its normal conditions. If
the traffic demands normally, present at the incident site is reduced by diverting
traffic to alternate routes, then the total vehicle-hours of delay will be minimized.
Following total blockage of the lanes, the capacity will be partially increased by re-
opening some lanes to traffic. This capacity increase will still be less than full
capacity which will be realized when the incident is completely removed from the
incident site. Following the complete clearance of an incident, the traffic waiting in
queue will try to go through the incident site as soon as possible. Flowever the
number of vehicles that can clear the incident site will be limited by the maximum
capacity of the roadway at that point. This Time to Normal Flow (TNF) is also
depicted as the incident recovery time. In Fact, TNF needed to clear the very large
queue of vehicles formed during the clearance operations is the reason for long delays
following the complete clearance of major accidents on an urban freeway (ATA
Foundation and Cambridge Systematics 1997). In some cases, it may even take hours
after the incident is cleared to dissipate this queued traffic. Once the queue is
completely cleared, the traffic is considered to have returned to its normal flow
conditions. Thus, quick incident clearance using effective incident management
strategies will reduce the queue and the time to return to normal flow condition.
22


Figure 2.5 Estimation of Queuing Delays during an Incident (Ozbay et.al, 1990)
23


3. Data Collection
3.1 Description of CDOTs Incident Management Operations
Generally, Colorado Transportation Management Center (CTMC) operators monitor
Closed Circuit Television Cameras (CCTV), as well as listen to police radios to
inform the Mile High Courtesy Patrol (MHCP) operators of incidents. The MHCP
operators monitor and respond to incidents on location as well. Figure 3.1 shows
organization chart of Colorado Department of Transportation for Intelligent
Transportation System (ITS) branch, CTMC, ATMS, ATIU, and the MHCP.
Figure 3.1 CDOTs Traffic Management Organization Chart
24


The Advanced Traffic Management (ATM) Unit constantly scrutinizes traffic
conditions and responds to incidents, especially during the peak hours. The ATM
operates seven days a week, 365 days a year, and also provides information on road
conditions and critical traffic situations.
The Mile High Courtesy Patrol dispatchers use two-way radios to communicate with
the ATM operators. The ATM operators are composed of 16 recovery vehicles
patrolling key areas of 1-25, 1-70, 1-225, and 6th Avenue during morning and
afternoon peak hours. The primary motive of the patrol is to provide immediate
incident response such as the removal of minor accidents and stalls to facilitate
smoother and faster rush hours. Patrol vehicles are equipped to handle minor
problems, including flats and stalls, and they bring spare fuel for stranded drivers. If
the Courtesy Patrol cannot get a car or truck started, then recovery vehicles can tow
disabled vehicles to nearby drop points.
3.2 Data Classification
Some of the criteria that transportation agencies use to classify incidents include the
following (Balke et.al., 2002):
Number of lanes blocked;
Estimated duration of blockage;
Severity and/or number of injuries involved;
25


Time-of-day;
Presence of hazardous materials;
Degree of damage to vehicles and/or infrastructure;
Type of vehicles involved (e.g., trucks, buses, etc.); and
Number of vehicles involved.
Examples of data collected by CDOTs MHCP are the following:
1. Incident Data
Date of incident
Time that service vehicle arrived
Time that incident was cleared
Type of incident
2. Position of Vehicle(s): right or left shoulder, lanes 1-5, median or gore area
3. Road Location: including road direction, milepost or nearest cross road
More detailed data for all crashes on the Denver area freeway system was obtained
from CDOT. The data for these crashes included:
1. Crash date and notification time
2. Route and milepost location
3. Crash severity
However, the MHCP database contained a better classification of incident types,
service times, and positions of vehicle(s) after a crash. As described later, the lane
26


positions of vehicles involved in crashes, and the service and clearance times of these
vehicles, were estimated from the MHCP data and applied to all crashes on the area
freeways as reported by CDOT to estimate delays.
3.3 Denver Area Highway Incidents in 2001 and 2002
The MHCP classified 20 types of incidents in its database. However, it was found
that 8 types were not incidents but rather incident causes or service provided such as
wrong direction, used phone, protect scene, private in route, tow to drop site, help-in-
route, service refused, and medical emergency. After reviewing the MHCP incident
types, this study reduced the MHCP incident types into three main categories and 11
subcategories as listed below and shown in Figure 3.2:
1. Crashes are incidents causing injury, fatality, or property damage only (PDO).
2. Vehicle disablements are incidents caused by or needing:
Flat tire: vehicle needed tire repair to continue
No fuel: vehicle needed fuel to continue
Water transfer: vehicle needed water to continue
Jump start: vehicle needed a battery jump to continue
Vapor lock: vehicle needs fuel line cleared to continue
Abandoned stall: vehicle left because of some mechanical failure
Occupied stall: vehicle stopped but driver remained with vehicle
Other mechanical failure
27


3. Other types of incidents are:
Debris: scattered remains of objects lying on the roadway
Spilled load: vehicle contents spilled to the roadway from a vehicle
still present
Flat tire
No fuel
Need water transfer
Jump start
Vapor Lock
Abandoned stall
Occupied stall
Other mechanical failure
Debris
Spilled load
Figure 3.2 Incident Type Classifications
3.4 Period of Incident Data Collection
Two years of incident data recorded by the MHCP on several Denver area highways
in the years 2001 and 2002 were used in this study. In 2001, MHCP reported 6,916
incidents, with an average of 19 incidents per day. In 2002, MHCP reported 5,145
incidents, with an average 14 incidents per day. The sections of roadway included in
28


this study were along 1-25, 1-70, 1-225, and US-6. The total length of roadway
sections was 61.40 miles as shown in Table 3.1. The MHCP operation was staffed 12
hours a day (6:00 AM 6:00 PM).
Table 3.1 Denver Metro Highway Study Location
Location Begin End Length
1-25 Lincoln Ave. MP 192.98 104th Ave. MP 221.027 28.05
1-70 Sheridan Blvd. MP 270.49 Chambers Rd. MP 283.52 13.03
1-225 Tamarac Pkwy. MP 0.66 1-70 E Interchange MP 12.14 11.48
US-6 1-70 NE Interchange MP 275.64 1-25 N Interchange MP 284.48 8.84
Total 61.40
The Statistics Section provides summary roadway statistics compiled by CDOT
annually, such as lane miles from each county, city streets, county roads, and vehicle
registrations, etc. For mapping accident locations, the CDOT website provides
metadata containing table.
Tables 3.2 through 3.5 show examples of crash records for the segments of 1-25
included in this study and shown by Figures 3.3 through 3.6. Tables 3.6 through 3.8
show examples of crash records for the segments of 1-70,1-225, and US-6 included in
this study and shown by Figures 3.7 through 3.9, respectively.
29


Table 3.2 1-25 Highway Crash Data from Arapahoe Rd. to Lincoln Ave.
(CDOT, 2001 and CDOT, 2002)
MP Section Length Func Class Section Description Year 2001 Year 2002
MVMT AADT MVMT AADT
192.98 3.22 R-4-R Lincoln Avenue Interchange CO Rd 8 E 0.01 65,760 0.01 75,744
193.53 0.53 R-6-R Change Median type enter Denver urbai 0.06 92,620 0.06 92,467
195.12 1.52 U-6-R County Line road interchange Rd E and W 0.02 99,458 0.02 99,343
196.71 1.58 U-6-R Leave centennial denter greenwood 0.01 130,599 0.01 130,311
197.18 0.48 U-6-R Arapahoe road interchange arapahoe 0.04 135,937 0.04 135,310
Fie Ed* V*w Add Toe* ttdp
Figure 3.3 Map of 1-25 from Arapahoe Rd. to Lincoln Ave.
30


Table 3.3 1-25 Highway Crash Data from Evans Ave. to Arapahoe Rd.
(CDOT, 2001 and CDOT, 2002)
\1P Section Length Func Class Section Description Year 2001 Year 2002
MV'MT 0.04 AADT 135,937 MVMT 0.04 AADT 135,310
197.18 0.48 U-6-R Arapahoe road interchange arapahoe
198.28 1.07 U-6-R Orchard road interchange Rd E dan W 0.02 158,893 0.02 158,789
199.37 1.08 U-6-R Belleview Avenue interchange SH 88 W Rd 0.01 180,723 0.01 178,359
199.71 0.33 U-6-R Arapahoe-Denver County Line enter 0.04 195,230 0.04 195,207
200.08 0.42 U-6-R 1-225 interchange SH 225 NE Overpass Str 0.03 195,410 0.03 195,416
201.57 1.46 U-6-R Hampden avenue interchange southmoor 0.01 182,639 0.01 182,515
202.63 1.05 U-6-R Yale avenue interchange Rd E and W 0.01 174,898 0.01 174,617
203.53 0.87 U-6-R Evans avenue interchange rd E and W 0.02 167,858 0.02 167,863
FI* Ei* VWt Add Tods H*t>
Figure 3.4 Map of 1-25 from Evans Ave. to Arapahoe Rd.
31


Table 3.4 1-25 Highway Crash Data from 1-70 to Evans Ave.
MP Section Length Func Class Section Description Year 2001 Year 2002
MVMT 0.02 AADT 167,858 MVMT 0.02 AADT 167,863
203.53 0.87 U-6-R Evans avenue interchange rd E and W
204.03 0.49 U-6-R Colorado Boulevard iiterchange SH 2 N 0.03 173,161 0.03 173,166
205.05 1.00 U-6-R University boulevard interchange Rd N 0.02 171,204 0.02 171,209
205.91 0.85 U-6-R Downing street interchange Rd N and S 0.02 174,280 0.02 174,285
206.61 0.70 U-6-R Logan street separation Rd N and S 0.02 164,823 0.02 164,780
206.96 0.34 U-6-R Interchange strs (F-16-DH) (F-16-DI) 0.05 169,064 0.05 168,944
207.48 0.48 U-6-R Santa Fe Drive interchange SH 85 S Rd N 0.04 156,984 0.04 156,871
207.98 0.36 U-8-R Alameda avenue interchange SH 26 W Rd E 0.04 198,248 0.04 198,121
209.20 1.20 U-8-R 6th avenue interchange sh 6 W rd E 0.01 216,230 0.01 210,774
209.47 0.26 U-8-R 8th avenue interchange rd E and W 0.05 216,450 0.05 216,456
210.3 0.78 U-8-R Interchange strs (F-16-JX) Underpass SH 40 0.02 219,945 0.02 219,825
Pi* Ed* view Add Took help
Figure 3.5 Map of 1-25 from 1-70 to Evans Ave.
32


Table 3.5 1-25 Highway Crash Data From 104th Ave. to 1-70
(CDOT, 2001 and CDOT, 2002)
MP Section Length Func Class Section Description Year 2001 Year 2002
MVMT 0.01 AADT 243,308 MVMT 0.01 AADT 243,315
213.62 0.85 U-8-R 1-70 interchange SH 70 EB underpass strs
213.95 0.21 U-8-R 48th ave interchange rd E and W BN RR 0.07 199,261 0.07 199,122
214.45 0.49 U-8-R Denver-Adams county line leave denver 0.03 199,261 0.03 199,122
215.23 0.78 U-8-R 58th avenue interchange SH 53 W Rd E 0.02 199,261 0.02 199,122
216.39 0.09 U-8-R 1-76 Interchange SH 76 WB underpass strs 0.13 237,538 0.13 237,377
217.00 0.62 U-8-R Milepost 217 0.02 238,002 0.02 237,839
218.19 1.22 U-6-R Enter Thornton city limits 0.02 147,414 0.02 147,310
218.45 0.25 U-6-R 84th avenue interchange rd E and W 0.07 147,414 0.07 147,310
220.31 1.79 U-6-R Leave thomton city limits enter 0.01 139,722 0.01 139,169
221.02 0.70 U-6-R 104th avenue interchange rd E and W 0.03 136,823 0.03 136,832
H* Edt V*w Add Teds HMp
Figure 3.6 Map of 1-25 from 104th Ave. to 1-70
33


Table 3.6 1-70 Highway Crash Data (CDOT, 2001 and CDOT, 2002)
MP Section Length Func Class Section Description Year 2001 Year 2002
MVMT AADT MVMT AADT
270.49 0.49 U-6-R Sheridan boulevard interchange SH 95 N 0.07 81,838 0.07 81,359
271.54 1.04 U-6-R Lowell boulevard interchange rd N and S 0.03 96,815 0.03 100,808
271.99 0.48 U-6-R Milepost 272 0.05 105,887 0.05 105,267
273.00 0.99 U-6-R Milepost 273 0.02 110,920 0.02 110,270
274.05 0.98 U-6-R Interchange strs (E-16-ET) EB Lane 0.02 117,577 0.02 116,825
274.6 0.53 U-6-R Washington street interchange rd N 0.03 151,026 0.03 151,030
274.75 0.14 U-6-R Major strs (E-l 7-FX) Overpass RD S Brighton 0.14 136,800 0.14 136,705
275.17 0.52 U-6-R Ramps on and off 0.04 126,242 0.04 130,609
275.69 0.51 U-6-R Sign bridge str E-17-KK EB Lane 004 132,283 0.04 132,249
276.07 0.35 U-6-R Ramps on and off to US and SH 6 0.06 124,725 0.06 124,639
276.56 0.48 U-6-R Colorado boulevard hterchange SH 2 0.05 120,806 0.05 120,723
276.99 0.46 U-6-R Dahlia street/Holly Street/Monaco 0.05 129,490 0.05 126,543
Ns EsK View Add Took HHp
Figure 3.7 Map of 1-70
34


Table 3.7 1-225 Highway Crash Data (CDOT, 2001 and CDOT, 2002)
MP Section Length Func Class Section Description Year 2001 Year 2002
MVMT AADT MVMT AADT
0.66 0.65 U-4-R Tamarac Parkway/Yosemite Street 0.04 112,003 0.04 112,006
1.32 0.52 U-5-R Yosemite street/Tamarac Paikway 0.06 92,073 0.06 92,076
1.61 0.28 U-4-R Denver Arapahoe County Line Leave 0.09 111,981 0.09 111,984
1.79 0.17 U-5-R Leave Greenwood Village City Limits 0.14 112,199 0.14 112,202
3.94 2.14 U-5-R Parker Road Interchange (1-225) nine mile 0.01 112,199 0.01 112,202
3.96 0.01 U-4-R Milepost 4 2.73 100,449 2.73 100,452
5.36 1.45 U-4-R lliff avenue interchange (1-225) Rd e and W 0.02 100,449 0.02 100,452
6.88 1.51 U-4-R Mississippi avenue interchange (1-225) Rd E 0.02 100,445 0.02 100,448
8.94 2.00 U-4-R 6th avenue interchange (1-225) SH 30 E 0.01 100,021 0.01 100,024
9.89 1.00 U-4-R Colfax avenue interchange (1-225) 0.03 100,650 0.03 100,653
9.99 0.03 U-4-R Milepost 10 0.98 93,149 0.98 93,151
10.03 0.03 U-4-R Ramps on and off 0.98 92,881 0.98 92,883
R< Edft vtow Add Toot* Help
Figure 3.8 Map of 1-225
35


Table 3.8 US-6 Highway Crash Data (CDOT, 2001 and CDOT, 2002)
IMP Section Length Func Class Section Description Year 2001 Year 2002
IMVMT AADT MVIMT AADT
275.64 0.31 U-5-R 1-70 Interchange SH 70 NE and SW 0.28 32,115 0.3 29,431
276.30 0.74 U-4-R Indiana street interchange rd N and S 0.06 63,759 0.06 63,766
276.90 0.59 U-6-R Enter Lakewood city limits 0.07 66,975 0.07 65,681
278.22 1.25 U-6-R Simms street/Union boulevard 0.03 66,997 0.03 65,694
279.30 1.06 U-6-R Kipling street interchange SH 391 N and S 0.03 94,993 0.03 94,999
280.82 1.51 U-6-R Ramp on 0.02 113,165 0.02 113,177
282.32 1.46 U-6-R Sheridan Boulevard Interchange SH55N 0.02 120,644 0.02 120,550
283.85 1.50 U-6-R Federal Boulevard Interchange SH 88 N 0.01 131,818 0.01 131,795
284 48 0.73 U-6-R 1-25 Interchange SH 25 N and S 6h ave E 0.03 146,815 0.03 146,831
Me Ed* Add Tocfc Help
Figure 3.9 Map of US-6
36


4. Methodology
Figure 4.1 shows the basic steps used to model total incident delays per year in this
study. Figure 4.1 shows the three basic steps of data processing and analysis.
Basically, all of the data is organized in a spreadsheet, and then processed through a
series of formulas to estimate total incident delays as the final result.
Figure 4.1 Data Processing
Table 4.1 lists the number of incidents by type responded to and recorded by the
MHCP in the years 2001 and 2002.
37


Table 4.1 MHCP Incident Types in 2001 and 2002
Incident Type 2001 2002
Abandoned stall 206 103
Crashes 1559 1317
Debris clean up 124 106
Flat tire 1145 1189
Fuel transfer 767 678
Jump start 126 127
Occupied stall 1502 394
Other mechanical 986 881
Spilled load 259 12
Vapor lock 74 138
Water transfer 168 200
Total 6916 5145
38


5. Summary of Data
The MHCP reported 6,916 incidents in 2001 and 5,145 incidents in 2002. The
average numbers of incidents per day were 19 in 2001 and 14 in 2002.
5.1 Incident Occurrences by Time of Day, Day of Week, and Month of Year
When comparing the incident proportions by time of day in 2001 (Figure 5.1) and in
2002 (Figure 5.2), it is seen that crashes and other incidents occurred with about the
same frequency in the two periods 6 AM to noon and noon to 6 PM. Overall, about
45% of all incidents occurred in the morning and about 55% of all incidents occurred
in the afternoon.
Incident proportions by time of day in 2001
100%
80%
O) 60%
3
c
0)
S 40%
Q.
20%
0%
Crash
Disabled
Type of incident
853 2704 190


193
706 2270
!

Others
6 AM 12 noon 12 noon 6 PM
Figure 5.1 Incident Proportions by Time of Day in 2001
39


Incident proportions by time of day in 2002
100%
80%
60%
40%
20%
0%
Crash
Disabled
Type of incident
6 AM -12 noon 12 noon 6 PM
661 a - 1903
454
: .... 1

666 1172 299

Others
Figure 5.2 Incident Proportions by Time of Day in 2002
Figure 5.3 and Figure 5.4 show that incidents occurred with greatest frequency on
Wednesdays in 2001 and on Fridays in 2002. The reasons for these differences by
day of week in 2001 and 2002 are unknown.
The number of incident by day of the week year 2001
Day
Figure 5.3 Incidents by Day of Week in 2001
40


The number of Incident by day of the week year 2002
Day
Figure 5.4 Incidents by Day of Week in 2002
Figure 5.5 and Figure 5.6 show the numbers of incidents by month in 2001 and 2002.
A few key observations are:
January had the most incidents, with 15% in 2001 and 13% in 2002.
December had the least incidents, with 5.5% in 2001 and 5.9% in 2002.
Since December had the fewest reported incidents and January had the
most reported incidents, the warmer months of May through October had
roughly the same number of incidents (48% overall) as did the colder
months of November through April (52% overall).
41


Figure 5.5 Incidents by Month of Year in 2001
Figure 5.6 Incidents by Month of Year in 2002
42


5.2 Incident Types
As mentioned earlier, this study divided the MHCP incident types into three main
categories. Figure 5.7 shows that disabled vehicles accounted for 6916 incidents in
2001 and 3710 incidents in 2002 and comprised 72% of all incidents. There were
1559 crashes in 2001 and 1317 crashes in 2002 and comprised 24% of all incidents.
There were 383 other incidents in 2001 and 118 other incidents in 2002, which
comprised 4% of all incidents in the database for these two years.
MHCP Incident Types in 2001 and 2002
J3 6000
c 0) O 5000
o c 4000
o 3000
k. E 1000
z 0
Crash
Disablement
Incident Types
Others
2001 2002
Figure 5.7 MHCP Incident Types in 2001 and 2002
This study focuses mainly on estimating vehicle delays caused by crashes. Table 5.1
shows that over 60% of all crashes reported by MHCP in 2001 and 2002 did not have
a vehicle stopped in a lane of traffic. The lane positions of stopped vehicles in 2001
43


after the crash were 54% on the right shoulder, 5.7% on the left shoulder, and 1-2% in
a gore point area or center median (if space allowed). The lane positions of the
stopped vehicles in 2002 after the crash were 58% on the right shoulder, 8.2% on the
left shoulder, and 1-2% in a gore point area or center median. Roughly one-third of
all crashes resulted in a vehicle blocking lanes 1 through 5, and roughly half of those
(16%-l 7% of all crashes) blocked lane 1 nearest to the right shoulder. Progressively
fewer crashes resulted in vehicles blocking lanes 2 through 5.
Table 5.1 Lane Positions of Disabled Vehicles after Crashes
Vehicle Position 2( )01 21 302
Right shoulder 747 54.3 % 716 58.4%
Left shoulder 79 5.7% 100 8.2%
Center 14 1.0% 11 0.9%
Gore point 23 1.7% 14 1.1%
HOV 1 0.1% 4 0.3%
Lane 1 239 17.4% 189 15.4%
Lane 2 115 8.4% 91 7.4%
Lane 3 95 6.9% 66 5.4%
Lane 4 50 3.6% 30 2.4%
Lane 5 13 0.9% 6 0.5%
Total 1376 100% 1227 100%
5.3 Incident Clearance Times
The elapsed time from when an incident occurs to when the incident is sufficiently
cleared to restore the highway to the normal capacity is called the incident clearance
time. As will be discussed later, the built-up queue of vehicles does not disappear
immediately at the end of the incident clearance time. Table 5.2 of data collected by
44


the MHCP shows that the average incident clearance time was 10 minutes 36 seconds
in year 2001 and 10 minutes 57 seconds in year 2002.
Table 5.2 Average MHCP Clearance Times by Incident Type
200 1 2002
Incident Type Number of Incidents Avg. Time Number of Incidents Avg. Time
Crash 1559 15:06 1317 16:11
Disablement 3192 8:42 3075 10:07
Others 2165 8:00 753 6:35
Total 6916 10:36 5145 10:57
Among all incident types, crashes have the longest average clearance times, which
were 15 minutes 6 seconds in year 2001 and 16 minutes 11 seconds in year 2002 as
shown by the Figure 5.8. In the later estimates of traveler delays caused by crashes,
this study used an average incident clearance time of 15 minutes.
Figure 5.8 Average MHCP Clearance Times by Incident Type
45


When grouped into 1-minute intervals, the distribution of clearance times shown in
Table 5.3, Figure 5.9, and Figure 5.10 show that clearance times were mostly less
than 20 minutes.
Table 5.3 Incident Clearance Times by 1-Minute Interval
Minutes Number of crash Minutes Number of crash
Year 2001 Year 2002 Year 2001 Year 2002
1 22 9 31 14 16
2 50 36 32 9 7
3 80 59 33 7 12
4 74 71 34 18 8
5 78 61 35 11 10
6 76 67 36 8 16
7 84 63 37 8 6
8 81 78 38 6 6
9 77 49 39 5 6
10 70 49 40 5 6
11 62 60 41 3 6
12 58 38 42 4 7
13 67 50 43 5 4
14 46 37 44 6 5
15 57 28 45 2 6
16 51 47 46 2 6
17 45 34 47 2 4
18 49 21 48 3 1
19 45 38 49 4 5
20 23 29 50 1 4
21 27 34 51 0 4
22 22 25 52 1 6
23 36 16 53 1 1
24 20 18 54 4 0
25 18 16 55 1 1
26 21 13 56 2 2
27 16 17 57 1 0
28 21 14 58 0 3
29 18 19 59 1 1
30 13 14 60 1 0
31 14 16 >60 17 48
46


Figure 5.9 Clearance Times by 1-Minute Interval in 2001
Group Clearance Time into 1 minutes interval Year 2002
c
0)
5
o
o
<
0)
Q
E
3
z
90
80
70
60
50
40
30
20
10
0
.Ooofl...
a. ,Q..
Tfh-OCOCOCDCNLOCOO
Minutes
Figure 5.10 Clearance Times by 1-Minute Interval in 2002
47


5.4 Estimating Volumes and Delays by Time of Day
This study focused mainly on estimating vehicle delays caused by crashes due to their
lane positions, capacity reductions, and approximate clearance durations. The CDOT
data included the approximate time and location for each crash on 1-25, 1-70, 1-225,
and US-6. The hourly traffic volume at each location was approximated by
multiplying the percentages in Table 5.4 by the AADT for the crash location based on
the approximate time of the crash. Figures 5.11 and 5.12 illustrated cumulative crash
by time of day in the years 2001 and 2002. It can be seen that most crashes occurred
during rush hours around 8:00 AM and 6:00 PM.
Year 2001
Time
1-25 1-70 I-225 US-6
Figure 5.11 Numbers of Crashes by Time of Day in 2001
48


Year 2002
Time
1-25 -m1-70 1-225 US-6
Figure 5.12 Numbers of Crashes by Time of Day in 2001
5.5 Estimation of Traffic Delay
In order to estimate queuing delays, the following approximations were used.
- Lane capacity = 2600 passenger cars per hour per lane (pcphpl)
- Average response time = 10 minutes from time of crash
- Average service time = 15 minutes = 25/60 = 0.42 hours
- Lane position = seven lane positions as reported by the MHCP
Table 5.4 shows the percentages of the seven lane positions calculated from the
dataset of crashes reported by the MHCP. Table 5.4 also shows the number of
blocked lanes assumed for vehicles in these lane positions.
49


Table 5.4 Percentages of Lane Positions
ID Lane Position Number of Crashes Percentage Avg Number of Blocked Lanes
1 Left Shoulder 179 6.89% 0.7
2 Left Most Lane 99 3.81% 1.7
3 Left Middle Lane 175 6.74% 2.3
4 Right Middle Lane 217 8.35% 2.3
5 Right Most Lane 428 16.47% 1.7
6 Gore Point 37 1.42% 0
7 Right Shoulder 1463 56.31% 1.7
2598 100.00%
- Arrival time = time when the service provider reaches the incident,
assumed to be 10 minutes after the crash because the
exact time of the crash is unknown.
- Departure time = time when the service provider leaves the scene of the
incident; departure time = arrival time + clearance time
- Clearance Time = time required for service provided to clear the incident;
It takes 15 minutes on average for the MHCP to clear a
crash that occurred.
Estimated volume (hourly volume) is computed by multiplying Average Annual
Daily Traffic (AADT) by the percentage factor. These percentages are shown by the
Table 5.5 and presented in the Figure 5.13. These percentages are based on hourly
traffic variations at urban locations given by the Highway Capacity Manual (TRB,
2000) and shown here by Figure 5.14.
50


Table 5.5 Hourly Volume Factor Percentages
Time Volume Percentage
1 100 0.82%
2 50 0.41%
3 50 0.41%
4 50 0.41%
5 100 0.82%
6 400 3.27%
7 900 7.35%
8 1000 8.16%
9 900 7.35%
10 600 4.90%
11 550 4.49%
12 600 4.90%
13 600 4.90%
14 550 4.49%
15 600 4.90%
16 700 5.71%
17 800 6.53%
18 850 6.94%
19 800 6.53%
20 700 5.71%
21 600 4.90%
22 400 3.27%
23 200 1.63%
24 150 1.22%
Total 12250 100%
51


Conversion Factor Graph
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time
Figure 5.13 Hourly Lane Volume Percentages
{!') Daily Variation in Volume!! l hwr Urban I Volume Variation Patterns ( Used with permission ofTransporuaian Research Board, Highway
Vashinglon IX!, 2000. F.xhibits 8-6 and 8-7, pgs. 8-6 and 8-7, repeated friwn Reis 2 and 3.)
Figure 5.14 Hourly Volume Variation Patterns (TRB, 2000)
52


The following list of calculations presents the procedure used to estimate the total
vehicle delay of each crash. Figures 5.15 and 5.16 depict these calculations
graphically.
- Event volume = min (estimated volume, number of lanes (lane capacity 100))
- Number of lanes depends on the crash location and direction of travel
- Capacity reduction factor computed with the following formula:
If (lane position = 1, -0.7, if (or(lane position=2, lane position=5), -1.7, if
(or(lane position = 3, lane position = 4), -2.3, if (lane position=7, -0.7), if (lane
position = 8, -0.3, if (or(lane position = 6, lane position > 3), 0, **)))))
- Event capacity @to = (number of lanes + capacity reduction) lane capacity
- Excess volume - event volume event capacity
- Event time = time 0 = to
- Time until new hour = time 1 = ti
- Response + service = time 2 = t2 = service time + clear code
- Time of new hour - ti = event time + time until new hour
- Volume @ti - min (event volume, event capacity@to)
- Excess volume @ti formula is on the below:
If (ti < 12, volume@ti event capacity@to, volume@ti clear capacity)
- Volume @t2 min (ti, volume @X\, event volume)
- Excess volume @t2 = volume @t2 clear capacity
- End of queue t3 formula if ti > t2 is following:
53


If (ti < t2, minimum(2, t2 + (ti excess volume @to+(t2 ti)
* excess volume@ti) / (clear capacity volume@ti, ti>t2)
End of queue t3 formula if ti < t2 (ti > t2, ti> t3) is following:
If (ti < t2, minimum(2, t2 +12 excess volume@to) / (clear capacity -
volume@ti, ti End of queue t3 formula if ti < t2 (ti > t2, ti< t3) is following:
If (and (ti < t2, ti < end of queue t3),min(2, ti +12 excess volume@to + (ti -
t2) excess volume @t2) / (clear capacity volume@ti, not app)
Delay X (veh-hrs) formula:
If (ti < t2, 0.5 ti2 excess volume @ to), 0.5 t22 excess volume@to)
Delay Y (veh-hrs) formula:
If (ti < t2, 0.5 (end of queue t3 -12) (clear capacity volume @X\), if (tj <
end of queue t3, 0.5 (end of queue t3 -12)2 (clear capacity volume @ti)
Delay Z (veh-hrs) formula:
If (ti < t2, (t2 ti) (ti excess volume @ to + 0.5 (t2 ti) volume @ tj)),
if (ti < end of queue t3, (ti -12) (t2 excess volume @to) + 0.5 (ti -12) *
excess volume at t2), 0))
Total Delay = Delay X + Delay Y + Delay Z
If total delay negative, delay not happened during the time
If total delay positive, delay can caused congested
Usable Delay= If (delay X < 0.05, 0, total delay Y/N calculation)
54


In this study, there are two scenarios for queuing delay prediction. Case I occurs
when an incident cleared before demand decreases and the queue dissipates. Case II
occurs when an incident is not cleared until after demand decreases and the demand
reduction must be estimated in order for queue dissipated.
Figure 5.15 defined case I, incident occurred when lane position at left shoulder and
right shoulder. When the highways are saturated, delays in incident are felt over a
wide area before the incident cleared. Incident cleared and diverting traffic in this
study has two conditions: first condition, if ti < t3, and second condition, if t3 < ft.
Delay in Y area is zero for the second condition because it is assumed that t2 equals tj.
c> blocked capacity of highway
c2 = open capacity of highway
Iftl X Vi t2. q2
Y /2 (qi +q2) (ti -t2)
Z 'A (t3 ti) qi
Ift3 X '/2 t2. q2
Y 0
Z lA (t3 ti) qi
55


Figure 5.15 Case I occurs when an incident cleared before
demand decreases and the queue dissipates
Time (hours)
A different version of the queuing diagram is proposed in Case II when an incident is
not cleared until after demand decreases and the demand reduction must be estimated
in order for queue dissipated as shown in Figure 5.16. Queues build up and spread
rapidly illustrated in area Y. Incident dissipated after response and service time over.
Arrival rate at time incident happened between to = event time equal to zero and ti =
1.0, and arrival rate after one hours come from ti to t2 continue until t3. Both of those
56


two cases based on different numbers of traffic demand and freeway capacity. Also,
the significant things to estimate delay are the consideration of time of incident and
the lane position.
Figure 5.16 Case II occurs when an incident is not cleared until
after demand decreases and the demand reduction
must be estimated in order for queue to dissipate
Time (hours)
c, = blocked closed capacity
c2 - open capacity
V, = arrival rate at time of incident
V2 - arrival rate after one hour
X = A ti.qi
57


Y '/2(qi+q2) (t2 ti)
Z = Vi (t3 -12) q2
Table 5.6 summarizes the total delay and total average delay per crash estimated by
this study. Within two years, there was an increase of nearly 20% (105.72 to 125.51)
in the average total delay per crash on these highway sections. Within two years,
there was an increase of nearly 10% (4793 to 5244) in the number of reported crashes
on these highway sections. Taken together, there was an increase of 30% (506,715 to
658,199) in the grand total hours of vehicle delay estimated to have been caused by
crashes on these highway sections in these two years. The highest average total delay
per crash occurred on 1-70, the most traveled highway in the Denver metro area.
These estimated delays were computed on the basis of input assumptions regarding
crash service times, hourly volumes, and the numbers of lanes effectively blocked by
these crashes. Assumption made of incident detection time is zero due to lack of data.
Incident service time was estimated to be 15 minutes. Identifying seven lane
positions for which to estimate delay was another essential assumption. While these
assumptions may change the specific results, the procedure and overall finding (i.e.,
that total vehicle delay caused by crashes in the Denver metro area is increasing)
remains unchanged.
58


Table 5.6 Total Vehicle Delay and Average Total Vehicle Delay per Incident (2001 and 2002)
Percentage of Lane Position 6.89% 3.81% 6.74% 8.35% 16.47% 1.42% 56.31% 100.00% Avg. Delay per Crash
Highway Crashes Lane Position 1 2 3 4 5 6 7 Total
1-25 2993 Total Delay (veh.hrs) 58,123 553,583 1,241,456 1,241,456 553,583 0 58,123
* lane position percentage 4,005 21,095 83,624 103,694 91,198 0 32,731 336,346 112.38
1-70 1024 Total Delay (veh.hrs) 22,412 189,902 433,812 433,812 189,902 0 22,412
* lane position percentage 1,544 7,236 29,221 36,234 31,285 0 12,621 118,142 115.37
1-225 436 Total Delay (veh.hrs) 2,894 89,525 213,273 213,273 89,525 0 2,894
* lane position percentage 199 3,411 14,366 17,814 14,749 0 1,630 52,169 119.65
US-6 340 Total Delay (veh.hrs) 0 7 376 376 7 0 0
* lane position percentage 0.0 0.3 25.3 31.4 1.2 0.0 0.0 58.15 0.17
Total 4793
200! Grand Total Delay (veh.hrs) 83,429 833,017 1,888,917 1,888,917 833,017 - 83,429
* lane position percentage 5,748 31,743 127,237 157,773 137,233 - 46,981 506,715 105.72
Percentage of Lane Position 6.89% 3.81% 6.74% 8.35% 16.47% 1.42% 56.31% 100.00% Avg. Delay per Crash
Highway Crashes Lane Position 1 2 3 4 5 6 7 Total
1-25 3160 Total Delay (veh.hrs) 66.583 647,356 1,424,326 1,424,326 647,356 0 66,583
* lane position percentage 4,588 24,668 95,942 118,968 106,647 0 37,495 388,307 122.88
1-70 1146 Total Delay (veh.hrs) 55,873 298,585 595,726 595,726 298,585 0 55,873
* lane position percentage 3,850 11,378 40,128 49,758 49,190 0 31,464 185,767 162.10
1-225 522 Total Delay (veh.hrs) 3,098 109,185 257,730 257,730 109,185 0 3,098
* lane position percentage 213 4,161 17,361 21,527 17,987 0 1,745 62,994 120.68
US-6 416 Total Delay (veh.hrs) 18 32,129 96,780 96,780 32,129 0 18
* lane position percentage 1.2 1,224.3 6,519.1 8,083.6 5,293.0 0.0 10.1 21,131.4 50.80
Total 5244
2002 Grand Total Delay (yeh.hrs) 125,572 1,087,255 2,374,562 2,374,562 1,087,255 - 125,572
* lane position percentage 8,652 41,431 159,949 198,337 179,117 - 70,713 658,199 125.51
Total Crashes 10037 Total Delay in Both Years 209,001 1,920,272 4,263,479 4,263,479 1,920,272 0 209,001 0
* lane position percentage 14,400 73,174 287,186 356,110 316,350 0 117,694 1,164,914 116.06
59


I
6. Conclusions and Recommendations
6.1 Conclusions
This study introduces many different incident types that must be managed in order to
reduce their frequency and mitigate their consequences. In this study, incident types
were simplified into three categories: crashes, vehicle disablements, and others.
This study estimate the delays caused by crashes on the four major freeway sections
of the Denver metro area: 1-25, 1-70, 1-225, and US-6. The total length of these
sections is 61.40 miles. In 2001 and 2002 combined, there were 12,161 incidents
reported by the Mile High Courtesy Patrol (MHCP) of which 2,876 (23.6%) were
crashes. However, there were many more crashes along these highway sections not
responded to by the MCHP. CDOT reported a total of 10,037 crashes (or an average
of 14 per day) along these highway sections in these two years, and there were 3 to 4
times this many other incidents along these highway sections in these two years not
classified as crashes for which this study estimated vehicle delays.
When comparing the incident proportions by time of day, it is seen that crashes and
other incidents occurred with about the same frequency in the two periods 6 AM to
noon and noon to 6 PM. Overall, about 45% of all incidents occurred in the morning
and about 55% of all incidents occurred in the afternoon.
60


The greatest frequency of incidents occurred on Wednesdays in 2001 and on Fridays
in 2002. Reasons for differences by day of the week in 2001 and 2002 are unknown.
A few key observations of the numbers of incidents by month in 2001 and 2002.are:
January had the most incidents, with 15% in 2001 and 13% in 2002.
December had the least incidents, with 5.5% in 2001 and 5.9% in 2002.
Since December had the fewest reported incidents and January had the
most reported incidents, the warmer months of May through October had
roughly the same number of incidents (48% overall) as did the colder
months of November through April (52% overall).
Among all incident types, crashes have the longest average clearance times, which
were 15 minutes 6 seconds in year 2001 and 16 minutes 11 seconds in year 2002. In
order to estimate of traveler delays caused by crashes, this study used an average
incident clearance time of 15 minutes.
Estimated volume (hourly volume) is computed by multiplying Average Annual
Daily Traffic (AADT) by the percentage factor. In this study there are two scenarios
outcomes for queuing delay prediction. Case I occurs when an incident cleared before
demand decreases and the queue dissipates and case II occurs when an incident is not
61


cleared until after demand decreases and the demand reduction must be estimated in
order for queue dissipated.
Within two years, there was an increase of nearly 20% (105.72 to 125.51) in the
average total delay per crash on these highway sections. Within two years, there was
an increase of nearly 10% (4793 to 5244) in the number of reported crashes on these
highway sections. Taken together, there was an increase of 30% (506,715 to
658,199) in the grand total hours of vehicle delay estimated to have been caused by
crashes on these highway sections in these two years.
6.2 Recommendations
The ability to estimate incident delay is limited because of only two years data and
the data reported by MHCP can not fulfilled for further analysis. The incident
detection time is not available in this study. Courtesy Patrol needs to develop more
effective of ITS technology service, expand this service to other areas in the Denver
Metro area, and continue documenting the benefits of courtesy patrol annually. The
benefits should be expanded to include quicker response times, air pollution benefits
and reduction of secondary accidents.
62


BIBLIOGRAPHY
Balke, K.N., David W.F., and Brooke, U. (2002). Incident Management Performance
Measures. Texas Transportation Institute.
Cambridge, Systematics Inc. and The Texas Transportation Institute (2005). Traffic
Congestion and Reliability. Trends and Advanced Strategies for Congestion
Mitigation. Final Report for Federal Highway Administration.
Cambridge, Systematics Inc.(1990). Incident Management. Final Report.
Colorado Department of Transportation. (2002). Crashes and Rates on State
Highways. Transportation Safety and Traffic Engineering Branch.
Colorado Department of Transportation. (2001). Crashes and Rates on State
Highways. Transportation Safety and Traffic Engineering Branch.
Cuciti, P. and B. Janson. (1995). Incident Management via Courtesy Patrol:
Evaluation of a Pilot Program in Colorado. Transportation Research Record
1494. TRB, National Research Council, Washington D.C., pp. 84 90.
FHWA-SA-91-056. (1991). Freeway Incident Management Handbook. U.S.
Department of Transportation.
FHWA (2001). Regional Traffic Incident Management Programs. Implementation
Guide.
Hofener, Michael S. (2003). A Check List For Work Zone Incident Management
Plans. Departement of Civil Engineering Texas A&M University.
Korpal, P.R. (1992). Incident Management: The Key to Succesful Traffic
Management in Toronto. ITE Journal., pp. 58-61.
Mannering, F and Nam, D. (1998). An expiatory Hazard-based Analysis of Highway
Incident Duration. Transportation Research Part A 34 (2000) 85 -102.
Nageli, Navin. (1993). Draft the Mile High Courtesy Patrol Report.
63


Ozbay, K. and P. Kachroo. (1999). Incident Management in Intelligent Transportation
System. Boston London: Artech House Inc.
PB Farradyne (2000). Traffic Incident Management Handbook.
PBSJ. (2004). South I- 25 Incident Management Response Manual. Colorado
Department of Transportation, Region 1.
Robinson, M.D and Nowak, P.M. (1993). An Overview of Freeway Incident
Management in the United States. Michigan Department of State Police and
Michigan Department of Transportation.
Smith, B.L. and Smith, K. (20011 Forecasting the Clearance Time of Freeway
Incident. Smart Travel Lab Report No. STL-2001-01. Center for
Transportation Studies University of Virginia.
Transportation Research Board (2000). Highway Capacity Manual (20001.
Transportation Research Board (TRB). (1994). Special Report 209: Highway
Capacity Manual. Washington, DC: National Research Council.
U.S. Department of Transportation. (1993). Estimates of Urban Roadway Congestion
- 1990. Publication No. DOT-T-94-Ol .
64