Microscopic Model of Road Capacity for Highway
Systems in Port Based Metropolitan Areas
Final Report
METRANS Project 09- 11
March 2011
Principal Investigator
Fei Wang
Electrical Engineering Department
California State University, Long Beach
Long Beach, CA 90840
Disclaimer
The contents of this report reflect the views of the authors, who are responsible for the
facts and the accuracy of the information presented herein. This document is
disseminated under the sponsorship of the Department of Transportation, University
Transportation Centers Program, and California Department of Transportation in the
interest of information exchange. The U. S. Government and California Department of
Transportation assume no liability for the contents or use thereof. The contents do not
necessarily reflect the official views or policies of the State of California or the
Department of Transportation. This report does not constitute a standard, specification,
or regulation.
Abstract
In this report, we present our approach to use microscopic modeling to assess
highway traffic mobility during lane blockage situation. A test microscopic model using
ARENA software is developed. In this model, we specifically aim to simulate the impact
of container trucks on highway traffic capacity and risk of accident during lane blockage.
The test model simulates a 3000 feet distance on a two lane highway, while the
passing lane is blocked at 2500 feet distance from the beginning. Merging sign and taper
zone was used to guide traffic on the passing lane to merge to driving lane. We adjust the
cargo truck traffic ratio to assess the effect of cargo trucks on the road mobility. The road
mobility is assessed using the throughput ratio ( total number of vehicles exited the
system/ total number of vehicles entered the system). The risk of collision is assessed
using the number of un- successful merging attempts at the taper end.
Traffic throughput and risk of collision is simulated as a function of traffic density,
cargo truck percentage as well as number of merging signs placed onsite. We found that:
1) Increasing traffic density significantly reduces the road throughput, while increases the
risk of collision moderately too; 2) Increasing cargo truck percentage reduces road
throughput, while increases the risk of collision; 3) Increasing number of merging signs
increases the traffic throughput and reduces the risk of collision up to a threshold. Under
the traffic condition we studied, this threshold is two merging signs.
Table of Content
Introduction …………………………………………………………………………… 1
Methodology…………………………………………………………………………… 2
Data Collection…………………………………………………………………….. 2
Model Overview…………………………………………………………………… 3
Arena Software Introduction………………………………………………. 3
Road Configuration and Driver Behavior………………………………….. 3
Model Vehicles…………………………………………………………….. 4
Model Car Following and Merging using Network Links…………………. 5
Simulation Results……………………………………………………………………… 9
Effects of Arriving Interval…………………………………………………………. 9
Effects of Cargo Truck Percentage………………………………………………… 11
Effects of Number of Merging Signs………………………………………………. 13
Conclusion……………………………………………………………………………… 15
Reference
List of Tables and Graphs
Table1 RVC of selected highway sections near Port of Long Beach
Table 2 Vehicle Statistics for Initial Set- up
Table 3 Definition of Intersections
Table 4 Description of Network Links
Table 5 Merging Gap
Figure 1 Road configuration of two lane one way highway section with lane closure in
ARENA module.
Figure 2 Distribution Plot of Intersections and Network Links
Figure 3 Model for Merging Behavior at Signal_ 1
Figure 4 Percentage of vehicles leaving the testing zone as a function of total number of
vehicles entering the testing zone.
Figure 5 Total transfer time in testing zone as a function of number of vehicles entering
the testing zone.
Figure 6 Number of Vehicles in taper zone as a function of number of vehicles entering
the testing zone.
Figure 7 Merging attempts in taper zone ( a measure as risk of collision) as a function of
number of vehicles entering the testing zone.
Figure 8 Percentage of Vehicles Leaving the testing zone vs. Cargo Truck Percentage
Change
Figure 9 Total Transfer Time in Testing Zone vs. Percentage Change of Cargo Truck
Figure 10 Average Merging Attempts in Taper Zone
Figure 11 Percentage of vehicle exiting the system as a function of number of merging
signs.
Figure 12 Merging attempts at taper end as a function of number of merging signs
THIS PAGE IS LEFT BLANK INTENTIONALLY
1
Introduction
Although Los Angeles County should be proud of its well- developed, conveniently
accessible highway systems, due to increasing volume of automobiles on road, the traffic
situation still becomes worse and worse each year. According to RAND California
statistic database [ 1], most of the highway sections in Los Angeles County have already
reached their maximum capacity, which is measured by RVC ( Ratio of Volume to
Capacity Factor). RVC equals 1.0 means this highway section is already serving in its full
load. Table 1 gives the RVC of selected highway sections near Port of Long Beach.
Highway I- 605, I- 710 and I- 405 are three major highways passing close- by Port of Long
Beach. All of their sections have been already fully saturated years ago.
On the other hand, Port of Long Beach, the nation’s largest and most important sea-port,
accommodates almost one- half of the nation’s port- related cargo container
distributions. All those containers need to pass through the Los Angeles metropolitan
area to reach interior areas through either truck or railroad transportation ( Alameda
Corridor). According to statistics provided by Port of Long Beach and Alameda Corridor
[ 5], the daily container traffic at port of Long Beach is nearly 20,000 for year of 2006,
and this number is projected to be doubled or tripled in the next decade. The container
movement paradigm has several requirements. The first is to accommodate projected port
growth so that economic base of the Southern California region, and the entire country,
can continue to grow. The second is to optimize the existing highway infrastructure
which currently cannot well handle its load before alternatives come into implementation.
Table 1. RVC of selected highway sections near Port of Long Beach. [ 1]
The Alameda rail corridor ( the appropriative railroad connects port of Long Beach to
LA downtown railroad transportation center)- developed to help accommodate the
unprecedented growth of container traffic going to and from the Port - has not
significantly reduced the impact of freight movement on the Los Angeles community.
The daily throughput of Alameda Corridor is only 9,600 containers. That leaves more
than half of the containers waiting to be moved out of the port by truck transportation. In
addition to that, a number of terminals at the port must relocate containers to the terminal
of the Alameda corridor, four miles from the port, before they could be shipped out by
train. This additional container truck traffic not only causes significant congestion inside
the port terminals as well as the surrounding community, but also brings up more
concerns on safety issues and speedups the aging of roads.
Recently, CCDoTT ( Center for the Commercial Deployment of Transportation
Technologies) at California State University Long Beach developed the idea of ECCO
( Electric Cargo Conveyor) [ 2] to help improve Port of Long Beach’s daily throughput by
Hwy Grade Terrain Category 1999 2000
405 -- FWY 710 Flat Flat Ratio of Volume to Capacity. 1.00 1.00
605 -- ORA CO L Flat Flat Ratio of Volume to Capacity. 1.00 1.00
710 -- FWY 91 Flat Flat Ratio of Volume to Capacity. 1.00 1.00
2
Maglev freight which can move large numbers of containers quickly and efficiently from
port terminals to transcontinental trains, tarns- shippers, and Inland Empire warehouses
with markedly reduced pollution as well as reduced traffic congestion. The suggested
strategy is to build an appropriative Maglev freight way from Port of Long Beach to a
less condensed inland port. Therefore, some containers could avoid merging into regular
traffics across the population condensed LA base. This strategy has been partially
implemented inside the port terminals and generated positive impacts [ 3]. But larger scale
implementation in entire LA basin area needs further justification in terms of
effectiveness and cost.
Despite of these new ideas, the conventional highway transportation will still be the
primary media for the container distribution for years in the near future due to several
reasons, such as insufficient infrastructures other than highway and union protection of
truck drivers. Therefore efficient utilization of the current limited highway resources
continues to be in urgent need. The congestion and risk for collision becomes more
severe when a road construction or temporary lane closure cannot be avoided. To ensure
safety, lane closure signs need to be placed way ahead of the actual blockage location and
lane closure taper need to be placed as well. Obviously, the longer the notifying distance,
the longer time the drivers get to react, thus the lower the risk of collision. However, lane
closure taper zone virtually lengthens the blocking zone, which makes the highway
resources stingier. This clearly indicates a trade- off between highway capacity and risk of
collision. On a highway section that constantly consumes large amount of cargo truck
traffics, more factors need to be considered in order to build a precise simulation model,
such as vehicle class ( length), percentage of vehicles in each class, following distance for
each class, specific merging distance for each class, deceleration rate for each class, to
name a few. In this project, we aim to study the highway throughput and the risk of
collision during merging as a function of traffic density, percentage of heavy cargo trucks
during a lane blockage situation. Moreover, this study also aim to determine the optimal
number of merging signs to be placed in order to reduce the risk of collision without
heavy delay penalty.
Overview of Construction Zone Control
The lane closure due to construction or accidents often cause capacity bottleneck of the
freeway. During lane closure, traffic needs to be redirected and forced to merge into open
lanes. There are several merging strategies that are either implemented or under study
now. These strategies follow either an early merge concept or a late merge concept. An
early merge concept encourages drivers to merge ahead of the work zone. The advantage
of this concept is reduced safety concern as all drivers are provided sufficient time to plan
merging action. On another hand, the early merging virtually reduces the road capacity
because the road section ahead of the work zone is not effectively used. Therefore, the
throughput is sacrificed. The late merge concept encourages drivers to use both the open
and closed lane until the beginning of work zone. This concept clearly takes full
3
advantage of road resources, but the tradeoff is risk of collision since the drivers are not
provided very much time to plan for merging.
The traditional strategy uses advanced warning signs ( i. e. merging signs) placed ahead of
the work zone together with blockage tapers. This is still the most frequently used
method due to its easy implementation. This strategy follows a static early merging
concept, which encourages drivers to merge to the open lane way ahead of the working
zone. Obviously, how far ahead the working zone we should place the merging sign
affect the overall throughput as well as the risk of collision.
A dynamic early merging strategy a merging notice is posted at a variable distance ahead
of the work zone based on the traffic density. The strategy seems to overcome the
shortcoming of the traditional early merge. However, simulation study conducted by
Tarko et al indicates that the travel time through the work zone is actually longer. Plus,
the implementation of this strategy is more complicated and cost significantly more than
the traditional one.
The static late merging strategy place “ USE BOTH LANES TO MERGE POINT” sign
on both side of the roadway. Near the beginning of taper zone, “ MERGY HERE” signs
are placed. The intention of this strategy is to efficiently use both open and closed lanes.
McCoy et al. showed that the capacity of the late merge strategy is significantly higher
than the conventional merge. However, the concerns here are about the risk of collision in
heavy traffic environment. Beacher et al. evaluated the late merge strategy under 3- to- 1,
2- to- 1 and 3- to- 2 situations. They claim that the late merge strategy should be considered
for 2- to- 1 and 3- to- 2 situations only when the heavy truck percentage is higher than 20%.
While the implementation of late merge strategy to 3- to- 1 situation needs more evidence
from field study. They also showed concerns by claiming more field studies are needed
for all situations before wide implementation.
Due to the concerns regarding static late merge strategies, many studies are devoted to
dynamic late merge strategies. Wei et al. suggested a Dynamic Merge Metering Traffic
System ( DMM- Tracs), in which a meter is installed at the work zone location, whose
actions are controlled by the traffic density, detected a certain distance ahead of the work
zone. If the traffic volume from upstream exceeds a threshold, the meter is triggered to
instruct the vehicles in the advance areas of the work zone. They found the DMM- tracs
strategy reduces delay for heavy traffic density. Kang et al. suggested a strategy that
combines the early merge and late merge concept. Under light traffic, an early merge
concept is implemented, while under heavy traffic, a late merge concept the
recommended. They also suggested an algorithm to determine the optimal threshold
dynamically considering the moderate flow rate, optimal merging capacity and the
maximum flow rate. Their simulation shows the advantage of this dynamic strategy on
4
throughput becomes significant only when the traffic density increases above 40
vehicles/ min.
Despite the performance advantages of dynamic merging strategies, they all require more
complicated hardware implementation, which may not be always possible for lane
blockage due to traffic incidents. The most frequently used merging strategy is still the
traditional method mentioned above. There are still things we can do to improve the
throughput under the traditional method, such as adjusting the location and number of
merging signs. Also, the effect of heavy truck to the construction area is also of interest
and is studied in this work.
Methodology
 Data Collection
Collecting real world traffic data, such as incoming traffic flows, class of vehicles and
percentage, speed of movement, traveling time in construction zone, is necessary to build
a model that produce reliable results. Data collection is usually time and money
consuming. Therefore, limited by the scope of this project, we could not afford to collect
real world data ourselves. Due to the scope of this project, we suggest a test microscopic
model that is capable to be expanded to analyze mobility and risk problems as a function
of vehicles types and construction zone layout provided the traffic data. We will build the
test model based on an estimated data from the Caltrans live camera videos on highways.
Based on our observations through Caltrans live video, the following estimated data
is used in our initial set- up.
Table 2 Vehicle Statistics for Initial Set- up
Heavy Traffic Light Traffic
Driving Lane Passing Lane Driving Lane Passing Lane
Mean Vehicle
Arrival Interval
( Exponential
Distribution)
1 second 2 second 4 seconds 5 seconds
Cargo Truck
percentage
40% 8% 40% 8%
 Model Overview
5
We considered a scenario of a two lane one way highway section where one of two
lanes is blocked. The highway section under consideration has a container truck flow
merges with urban vehicles. A real world prototype of this situation is the I- 710 south
bound from I- 91 ramp all the way to the Port of Long Beach. Due to the funding and time
limitation, only a two- lane one way test model is developed in this project. However, this
model could be expanded into multi- lane system as a second phase project.
• Arena software introduction
Arena simulation software ( by Rockwell Automation) will be used to build this
traffic model. This software is designed to model queues. The program uses Entities to
represent dynamic players in system, such as people, parts or cars. Entities enter the
system for some kind of service, but there are only limited resources to provide this
service. The service will be provided on a first come first serve base, the later comers will
form a queue waiting for the resources to be available. After the entities received the
service, they leave the system.
Arena provides statistical analysis tools to assist user analyzing the simulation
results, such as queuing time, length of queue, waiting time in system etc. It also provides
a nice animation option, which makes it possible to visualize the system’s operation.
 Road Configuration and Driver Behavior
Figure 1 shows the configuration of the road section to be modeled. The model
begins 500 ft before the first “ Merge” sign and stop approximately 500 feet after the
blocked zone. We assume the passing lane is blocked, while all traffic need to merge to
driving lane. Three “ Merge” signs and 1 taper zone are place one the passing lane, which
are used to direct the traffic. The distance between the “ Merge” signs are 500 ft and the
taper zone is 500 ft long. We assume vehicles do not change lanes before the first
merging sign. After the first “ Merge” sign, vehicles on the blocked lane start to look for
opportunities to merge onto the passing lane. The acceptable merging condition depends
on the vehicle type, speed of vehicle on blocked lane and speed of traffic on the passing
lane. Once the merging conditions are satisfied, the vehicle will be placed on the passing
lane and removed from blocked lane.
Chances are not all drivers can successfully merge after they see a “ Merge” sign;
some of them have to wait for an opportunity to come up. At the beginning of taper zone,
if the merging condition is still not satisfied, the vehicle has to stop until there is
opportunity for it to merge. In ARENA software, this module can be realized by
assigning each point of interest ( each merging sign, beginning of taper, end of taper) with
a “ station”.
6
Figure 13 Road configuration of two lane one way highway section with lane closure in
ARENA module.
 Model Vehicles
Major components of this ARENA module are “ Entities”, representing different
types of vehicles. They will be paired with “ Transporters” that travels on the links
between intersections. In this proposed module, we will consider two types of vehicles:
- Cars or light trucks ( length < = 20 ft)
- Cargo trucks or very heavy vehicles ( length= 65ft)
Entities are generated by Create module in ARENA based on the distribution of each
type of vehicles in the entire traffic. Once generated, entities are assigned to different
types of Transporters. Two types of transporter are used in this project, CT ( Car
Transporter) and TT ( Truck Transporter). Attributes of transporters, such as speed and
length, are assigned based on the vehicle type ( cars, trucks or cargo trucks). The needed
merging distance will be determined based on vehicle type and its relative speed to the
passing lane traffic. Each entity- transporter pair represents one vehicle traveling on the
highway.
This model allows user to adjust input variables, such as inter- arrival time, speed
distribution, vehicle type distribution, to simulate for different situations. .
 Model Car Following and Merging using Network Links
In Arena software, there are two ways to move a transport. One is using free- path
transporter, the other is through guided transporter. Free- path transporter moves from
point 1 to point 2 at a specified speed, but the exact path is not defined. Therefore, the
time it takes for free path transporter to reach a destination only depends on the distance
between two destinations and the speed of the vehicle. However, the vehicles on highway
is bounded by the road, therefore, they are not really moving freely. Therefore, we chose
guided transporter to model the highway vehicles.
In order to guide the transporter, we need to specify the paths that are allowed for
the transporter to take. These paths ( named “ network links” in ARENA) could be linked
as desired based on real world scenario. Usually the linked paths build a network ( name
7
“ network” in ARENA. One could define more than one network, which specifically
designed for each type of transporter. However, in our scenario, the highway is
considered public and any vehicle can choose any path it wants. Therefore, one network
is necessary for our problem.
In ARENA, network links are links that connects “ intersections”, which are defined
specifically for each event that needs to be considered in the problem. We defined the
following intersections in this project:
Table 3 Definition of Intersections
Intersection Number Description of Intersection
Intersection- 1 ( I- 1) Starting point of Passing Lane
Intersection- 2 ( I- 2) First “ Merge” sign on Passing Lane
Intersection- 3 ( I- 3) Second “ Merge” sign on Passing Lane
Intersection- 4 ( I- 4) Third “ Merge” sign on Passing Lane
Intersection- 5 ( I- 5) Starting point of taper zone on Passing Lane
Intersection- 6 ( I- 6) End of taper zone on Passing Lane
Intersection- 7 ( I- 7) Starting point of Driving Lane
Intersection- 8 ( I- 8) Corresponding location of 1st “ Merge” sign on Driving Lane
Intersection- 9 ( I- 9) Corresponding location of 2nd “ Merge” sign on Driving Lane
Intersection- 10 ( I- 10) Corresponding location of 3rd “ Merge” sign on Driving Lane
Intersection- 11 ( I- 11) Corresponding location of start of taper zone on Driving Lane
Intersection- 12 ( I- 12) Corresponding location of end of taper zone on Driving Lane
Intersection- 13 ( I- 13) Car’s virtual destination ( 500 ft from I- 12)
Intersection- 13 ( I- 14) Truck’s virtual destination ( 500 ft from I- 12)
The following network links are defined in this project:
Table 4 Description of Network Links
Network Link Number Description of Links
8
Network Link- 1 ( L- 1) Initial position for Truck Transporter
Network Link- 2 ( L- 2) Initial position for Car Transporter
Network Link- 3 ( L- 3) Link I- 1 to I- 2
Network Link- 4 ( L- 4) Link I- 7 to I- 8
Network Link- 5 ( L- 5) Link I- 2 to I- 3
Network Link- 6 ( L- 6) Link I- 8 to I- 9
Network Link- 7 ( L- 7) Link I- 3 to I- 4
Network Link- 8 ( L- 8) Link I- 9 to I- 10
Network Link- 9 ( L- 9) Link I- 4 to I- 5
Network Link- 10 ( L- 10) Link I- 10 to I- 11
Network Link- 11 ( L- 11) Link I- 5 to I- 6
Network Link- 12 ( L- 12) Link I- 11 to I- 12
Network Link- 13 ( L- 13) Link I- 12 to I- 13
Network Link- 14 ( L- 14) Link I- 2 to I- 8 ( merging link)
Network Link- 15 ( L- 15) Link I- 3 to I- 9 ( merging link)
Network Link- 16 ( L- 16) Link I- 4 to I- 10 ( merging link)
Network Link- 17 ( L- 17) Link I- 5 to I- 11 ( merging link)
Network Link- 18 ( L- 18) Link I- 6 to I- 12 ( merging link)
Network Link- 19 ( L- 19) Link I- 12 to I- 14
Network Link- 20 ( L- 20) Link I- 1 to I- 2 ( start point link)
Network Link- 21 ( L- 21) Link I- 2 to I- 1 ( start point link)
The following ( Figure 2) is a distribution chart of the network containing link 1- 21.
9
Figure 14 Distribution Plot of Intersections and Network Links
In the following figure ( Figure 3), we show a sample model set- up at merging
signal # 1.
In the Signal_ 1 sub- model, incoming traffic comes from both driving lane station
( Sig1_ D) and passing lane station ( Sig1_ P). We first assess the vehicle type ( car or truck)
by using “ T_ C_ Sig1_ P” module. This is necessary to determine proper merging
conditions for the vehicle. The Car Transporter will first check the traffic conditions on
both lanes (“ D- lane Crowdy?” module). If passing lane has less traffic, the car will
choose to stay on passing lane, otherwise, it will seek merging following the merging
condition set by “ Continue?_ Sig1_ P” module. This module sets the merging condition
considering the vehicle speed. As far as the Truck Transporters, since most of cargo
trucks are less aggressive than cars, they will seek merging once they see the merging
sign. Therefore, we didn’t set any check traffic module for the truck transporters.
Table 5 listed the merging gap data used to determine the necessary merging gap at
different vehicle speed. We applied linear regression on the data to determine the
merging function in our model ( Speed* 1.81+ 75 for cars and Speed * 1.81+ 120 for trucks).
10
Figure 15 Model for Merging Behavior at Signal_ 1
Table 5 Merging Gap
Car Speed
( ft/ sec)
Merging Gap
( ft)
Truck Speed
( ft/ sec)
Merging Gap
( ft)
75 215 75 260
55 170 55 215
35 140 35 185
10 95 10 140
The merging modeling for Signal 2, Signal 3 and taper zone are similar to the one
listed in Figure 3. The only difference is at Taper zone end, the vehicle will be delayed
for a certain time and check merging condition again, as they already reached the end of
the passing lane.
11
Simulation Results
 Effect of Arriving Interval
First, we tried to assess the effect of traffic flow on the mobility of the road.
The mobility of the road is measured by percentage of vehicle leaving the system
and the transfer time in the system. The simulation results are shown in Figure 4
and Figure 5. The simulation time is 1 hour ( 3600 seconds).
From Figure 4, we can see that, the percentage of vehicles that leaving the
system is decreasing as a function of number of vehicles entering the system. The
latter is a direct measure of the traffic flow. Intuitively, the more vehicles entering
the system, the busier is the traffic.
Figure 16 Percentage of vehicles leaving the testing zone as a function of total number of
vehicles entering the testing zone.
From Figure 5, we observe that the total transfer time is showing an overall
decreasing trend as a function of number of vehicles entering the system. This is
kind of contradicts with common sense. Our explanation to this is for less traffic
situation, more vehicles wait till the taper to start their merge. Because in the
model, the driver will assess which lane has less traffic. For the first few vehicles,
the passing lane is less traffic, therefore, they go on on that lane until they hit taper
end. At that time, the driving lane is also at a relatively high speed, because the
overall traffic is light. Then they need larger merging distance to complete the
merge, so they wait longer at the taper end. This enlarged the overall average of
transfer time in the system. On the other hand, when the traffic is normal or heavy,
more vehicles start their merging earlier. This basically reduced the number of
taper end merging ( which is hardest location to merge since your speed is way
12
lower than the driving lane), therefore, reduced the average traveltime. This can be
seen by looking at the number of cars in taper zone ( Figure 6). When the traffic is
light, we have more cars gathering in the taper zone waiting to merge. According
to our simulation, threshold of light traffic is less than 2000 vehicles on the 3000
feet that is being simulated. We do see the transfer time gradually builts up as the
traffic getting more and more severe.
Figure 17 Total transfer time in testing zone as a function of number of vehicles entering
the testing zone.
Figure 18 Number of Vehicles in taper zone as a function of number of vehicles entering
the testing zone.
13
We also simulated the number of car’s attempts to merge at the taper end. The
results are shown in Figure 7. The attempts of merging at taper end could be
account for a risk. This is because the cars at taper ends are desperate in merging
and they are likely to conduct a merging when merging condition is not met. This
may result in collision. From Figure 7, we can see that this risk of collision is
increasing with the traffic flow, that is, heavier traffic results in higher likely- hood
of collision.
Figure 19 Merging attempts in taper zone ( a measure as risk of collision) as a function of
number of vehicles entering the testing zone.
 Effect of Cargo Truck Percentage
The next question in mind will be whether the cargo truck traffic has an effect
on the traffic efficiency as well as the risk of collision. Therefore, we also
simulated the mobility of the road as a function of cargo truck percentage. We set
the overall traffic at medium light rate, i. e. mean arrival interval to be 4 seconds.
Figure 8 shows the percentage of vehicles leaving the testing zone as a
function of percentage change of cargo trucks. The original percentage of cargo
truck is 8% at passing lane and 40% at driving lane. We adjusted the percentage
from - 5% to + 7%. We observe that the number of vehicle leaving the testing zone
in the 1 hour simulation limit reduces as we increase the cargo truck percentage.
This indicates the number of cargo truck on road has a negative impact on road
mobility.
14
Similar trend is observed in travel time result, as shown in Figure 9. The travel
time is increasing as number of cargo trucks grows, indicating slower traffic and
more traffic congestion.
Figure 20 Percentage of Vehicles Leaving the testing zone vs. Cargo Truck Percentage
Change
Figure 21 Total Travel Time in Testing Zone vs. Percentage Change of Cargo Truck
15
Figure 22 Average Merging Attempts in Taper Zone
The risk of collision is also assessed by simulating number of merging attempt
in taper zone ( Figure 10). The overall merging attempts at taper end grow with
cargo trucks traffic as well. This indicates more cargo trucks on highway results
in higher risk of collision during lane blockage.
 Effect of Number of Merging Signs
In the initial set- up, we designed 3 merging signs to guide drivers to merge to
the driving lane as it is possible. The number of merging signs warns the drivers
ahead of time, so that they get more chances to seek for a successful merge.
However, the more merging signs we put on the road, the more disturbance we
introduces to the traffic. Therefore, it is of interest to study how many merging
signs should be applied to the road without wasting the road resources.
In this experiment, we studied the percentage of vehicles exiting the system as
a function of merging signs. We set the average interval between incoming
vehicles to be 5 seconds, and we reset the percentage of cargo truck to the original
setting. ( 39% for driving lane and 9% for passing lane). As shown in Figure 11, as
we increase the number of merging signs, up to a certain point, it does help the
vehicle getting out the of system faster. But above that threshold, we do not see
significant correlation between number of signs with percentage of vehicle out.
This suggests that under the current traffic situation, three merging signs will be a
waste; two merging signs will be a more efficient choice.
16
Figure 23 Percentage of vehicle exiting the system as a function of number of
merging signs.
Another concern maybe whether reducing the merging signs increase the risk
of collision at the taper end. Figure 12 shows our test of merging attempts at taper
end as a function of number of merging signs. We can see that increasing merging
signs does reduce the risk of collision; however, the difference is not significant as
we go above two signs. Therefore, both the throughput and the risk test confirms
that under the current traffic condition, two merging signs will be more efficient
than 3 signs.
Figure 24 Merging attempts at taper end as a function of number of merging signs
17
Conclusion
In this project, we built a simulation model using Arena Software to model a two lane
highway section under road blockage. Traffic through put and risk of collision is
considered as a function of traffic density, cargo truck percentage as well as number of
merging signs placed onsite. We found that: 1) Increasing traffic density significantly
reduces the road throughput, while increases the risk of collision moderately too; 2)
Increasing cargo truck percentage reduces road throughput, while increases the risk of
collision; 3) Increasing number of merging signs increases the traffic throughput and
reduces the risk of collision up to a threshold. Under the traffic condition we studied, this
threshold is two merging signs.
The model built in this project is a test model, therefore, only two lanes are
considered. It could be expanded into multi- lanes in a future work. Therefore, more
realistic questions can be answered, such as number of lane blockage, length of lane
blockage. Also, a real field data collection could help in building more realistic statistical
distribution of traffic densities as well as vehicle type percentage.
Reference
[ 1]. RAND California traffic congestion database. http:// ca. rand. org.
[ 2]. K. James, and S. Gurol, “ Urban Maglev Freight Container Movement at the Ports of
Los Angeles/ Long Beach,” in Proc. of National Urban Freight Conference, Long
Beach Convention Center, Long Beach, CA, February 1- 3, 2006.
[ 3]. K. James et. al. “ Demonstrator Plan for ECCO System Implementation” in IMPACT
2008 conference, Long Beach, CA, January 16, 2008.
[ 4]. W. David Kelton, R. P. Sadowski, and D. T. Sturrock, “ Simulation with Arena,” 4th
Edition, McGraw Hill, 2007.
[ 5]. www. acta. org and www. polb. com.
[ 6]. John El Khoury and Antoine Hobeika, Journal of Transportation Engineering, vol 133,
6, 2007
[ 7]. A. Polus, M. livneh and B. Fischer, Transportation Research Record. 1701,
Transportation Research Board, National Research Council, Washington, D. C., 53- 60,
2000