3 4
Shayan Khoshmagham, Graduate Research Assistant
5(Corresponding Author)
6Department of Systems and Industrial Engineering
7University of Arizona
81127 E. James E. Rogers Way, Tucson, Arizona, 85721, USA
9Phone: +1 520 4681375
10Email: [email protected]
11 12K. Larry Head, Professor
13Department of Systems and Industrial Engineering
14University of Arizona
151127 E. James E. Rogers Way, Tucson, Arizona, 85721, USA
16Phone: +1 520 6212264
17Fax: +1 520 6216555
18Email: [email protected]
19 20Yiheng Feng, Graduate Research Associate
21Department of Systems and Industrial Engineering
22University of Arizona
231127 E. James E. Rogers Way, Tucson, Arizona, 85721, USA
24Phone: +1 520 2483988
25Email: [email protected]
26 27Mehdi Zamanipour, Graduate Research Assistant
28Department of Systems and Industrial Engineering
29University of Arizona
301127 E. James E. Rogers Way, Tucson, Arizona, 85721, USA
31Phone: +1 520 2034304
32Email: [email protected]
33 34 35 36 37 38Number of Words: 5,489 texts + 2000 (8 Figures) = 7,489 words
3940
Prepared for Presentation at the 95
thAnnual Meeting of the Transportation Research Board,
41January 10-14, 2016, Washington D.C., and for Publication in the Journal of Transportation
42Research Record.
43 44November, 2015
45ABSTRACT 1
2
The purpose of this paper is to define a visualization method to evaluate the performance of a multi-3
modal traffic signal system. Previous studies have concentrated on performance assessment for single 4
modes, such as delay or travel time of passenger vehicles, or transit running times. The methodology 5
presented in this paper considers an integrated approach to multi-modal performance assessment. A tool, 6
called a Multi-Modal Performance Dashboard, is developed to visualize the relationship between various 7
performance measures and multiple modes. Dashboards can be used to characterize the performance of an 8
existing system and also to compare before and after studies when a new design is implemented. Radar 9
diagrams are the basic element of the Multi-Modal Performance Dashboard tool and are constructed for 10
performance measures, e.g. passenger vehicle travel time, transit delay, pedestrian volume, and truck 11
stops, and for each movement at an intersection. An arterial corridor in the Maricopa County Department 12
of Transportation’s SMARTDrive test bed is analyzed using the VISSIM micro-simulation model to 13
study the effects of different designs and signal timing strategies on several performance measures for 14
both vehicles and pedestrians. Based on the results of this study, choosing an appropriate control strategy 15
can impact the different movements of different modes (including pedestrians) in a variety of ways. The 16
more modes involved in the system, the more challenging it is to determine the proper control strategy. 17
Using this comparative tool, alongside statistical models, makes it easier for decision makers to 18
understand, visualize, and analyze data. 19
INTRODUCTION 1
2
Capital improvements to transportation systems, such as rebuilding an intersection, arterial, or corridor, 3
need to be evaluated in terms of the improvements or impacts to the users of the system. Traditional 4
studies have focused on performance measures such as vehicle delay and travel times for a single mode 5
(e.g. passenger vehicles). Recently, there has been a greater focus on mobility and safety of multi-modal 6
travelers. While performance assessment studies for each mode exist, there is a need for a comprehensive 7
approach to modal performance assessment. This paper presents an integrated approach to multi-8
modal performance assessment with a case study of pedestrian phasing. 9
The first step in any improvement project is to identify the project objectives and to define 10
appropriate performance measures and metrics. Poor outcomes are most probably results of poor selection 11
of measures (1). Based on the fact that one of the basic requirements of transportation system 12
management is balancing the needs of all users and multiple transportation modes (2), this study uses both 13
multi-modal and mode-specific measurements not only to evaluate the total effect but also to recognize 14
individual deficiencies. This study considers four different road users with the performance measures as 15
follows: transit delay, truck stops, passenger vehicle travel time, and pedestrian volume and delay. 16
Transportation improvement projects may include widening an arterial corridor to provide 17
additional capacity and reduce travel time and delay for vehicles, but this may impact pedestrians. In 18
other words, this could be an effective objective for many corridors, but in environments where there is a 19
high pedestrian demand (urban locations), the delays that traffic flow improvements impose on 20
pedestrians might be ignored by existing policies (3). These types of strategies can impact pedestrian 21
traffic in a negative way and may result in unbalanced assessment, especially in city centers, such as 22
central business districts. Inclusion of traffic signal priority for transit (or other vehicles such as trucks) 23
could impact both vehicles and pedestrians. Hence, there is a need to understand how improvements to 24
one mode may impact, or benefit, another mode. 25
In this paper, methods of visualization of multi-modal performance measures that can be 26
presented in an easy-to-understand format to gauge and compare different ways of treating all the users of 27
the system are described. For visualizing, at a glance, the relationship between different measures and 28
multiple modes, a tool was developed which reflects the performance of an intersection. This tool utilizes 29
radar diagrams as the basic element, similar to safety viewgrams (4) (a diagram used for safety 30
assessment), and is not intended to replace statistical analysis and modeling. This tool could be used 31
concurrently with other techniques for analyzing the strengths and weaknesses of the existing system. Not 32
only is this tool useful for presenting the results to stakeholders to show them where the benefits and 33
impacts exist, but it is also useful for assessing before and after studies. 34
This study utilizes the Arizona Connected Vehicle Test Bed, to demonstrate the multi-modal 35
performance assessment methodology. The test bed consists of six intersections along a major arterial 36
corridor in Anthem, AZ. It is analyzed using a VISSIM micro-simulation model to study the impacts of 37
alternative signal timing strategies and crosswalk designs on several performance measures for both 38
vehicles and pedestrians. 39
The organization of the remainder of the paper is as follows: A brief literature review of 40
pedestrians and their impact on a multi-modal system is presented in the next section; the scenarios for 41
designing pedestrian crosswalks considering different signal phasing and timings are then presented in 42
detail; the performance measure visualization methods used to assess multi-modal performance measures 43
are presented using a sample radar diagram; a description of the network and assumptions are then 44
explained; and finally, an analysis of results followed by conclusions and recommendations are discussed. 45
LITERATURE REVIEW 1
2
Historically, data visualization has been shown to be critical in transportation studies. In the context of a 3
multi-modal transportation system, there have only been a few examples of studies that consider 4
pedestrian travel costs in the design of traffic signal systems. Noland (5) focused on the potential 5
differences in signal-timing options that would follow from consideration of pedestrian delay and also 6
collected some empirical data in London to determine if optimal signal cycles were being applied in 7
practice (6). These studies considered only fixed time signal control and didn’t address issues associated
8
with actuated traffic signal controls. 9
Many studies have analyzed interactions between pedestrians and turning vehicles at signalized 10
intersections from a safety point of view (7, 8). A LOS (Level Of Service) model for signalized 11
intersections regarding the safety risk based on pedestrian and vehicle demand was introduced by Zhang 12
and Prevedouros (9). In their paper, the conflicts between the movement of through vehicles and the 13
movement of permitted left-turning vehicles and pedestrians were considered. Hubbard et al. (10) studied 14
the interaction between pedestrians and vehicles caused by concurrent right-turning vehicles (right turning 15
on green) at signalized intersections. However, this work has not been extended to consider the impacts of 16
turning vehicles on pedestrians in terms of other performance measures (e.g. delay, travel time, and 17
number of stops for different modes of vehicles). 18
Urbanik, et al. (11) conducted an assessment of different split-phasing options considering 19
pedestrian crossing treatments. The main focus of their study was on the impact of various split-phasing 20
alternatives and alternative pedestrian treatments on traffic operations. Tian et al. (12) studied four 21
different pedestrian timing treatments including (a) no pedestrian timing consideration, (b) pedestrian 22
timing concurrently with vehicle phases, (c) a special pedestrian overlap phase, and (d) an exclusive 23
pedestrian phase. 24
Considering the geometric layout and control types, there are different kinds of pedestrian control 25
strategies that are possible at signalized intersections. Some of the common ones are fixed-time pedestrian 26
operation, coordinated operation, and push-button actuation (13, 14). Lead and lag phasing on the side 27
street of an intersection have been investigated as a way to lessen the impacts of pedestrians on 28
coordinated signal systems (15). 29
In the German Traffic Control and Traffic Safety Guidelines for Traffic Signal (16), three 30
different categories for signalization of two successive pedestrian crosswalks were introduced based on 31
local boundary conditions or other considerations of traffic operations including: Simultaneous 32
Signalization, Progressive Signalization, and Separate Signalization. Their recommendation was that two 33
successive crosswalks can be treated as two independent crosswalks only if the width of the central 34
refugee island is more than 4 m. 35 36 37 METHODOLOGY 38 39
The basic concepts in this multi modal performance assessment methodology are based on the idea that 40
each mode of vehicles and travelers in a network, section, or an intersection can be assessed both 41
separately and as part of the overall population being serviced. In addition, the performance of a facility 42
can vary significantly by movement. For example, trucks are one mode of vehicles that might traverse a 43
signalized arterial. Trucks are clearly impacted by other vehicles, including passenger vehicles and transit 44
vehicles, but understanding that the number of stops is more important to trucks and delay is more 45
important to transit, is vital when considering multi-modal performance. Trucks that are turning left at 46
one intersection might incur significantly more stops as compared to trucks that are passing through the 47
same intersection. 48
The ability to understand, visualize, and analyze different measures for different movements of 49
different modes is an important capability in today’s complex transportation systems. It is believed that 50
operating agencies might want to establish multi modal operating policies that favor one mode over others 51
during certain times of the day or under certain conditions. For example, assume that one section of a 1
network is heavily traveled by pedestrians and transit vehicles and another section of the same network is 2
heavily traveled by trucks that are moving goods from factories and warehouses to an interstate system. 3
Similarly, the pedestrian and transit corridor might be heavily used by passenger vehicles in the AM peak 4
period when commuters go to work. The ability to observe, analyze and understand the impact of traffic 5
management strategies on the different modes and movements is necessary if a multi-modal policy is to 6
be implemented. 7
The methodology presented in this paper is illustrated by comparing three different alternative 8
pedestrian phasing strategies at an intersection. The data for the study is generated from a micro 9
simulation model, but the methodology has been developed and is intended for deployment as part of the 10
Multi Modal Intelligent Traffic Signal System Pooled Fund project (17). 11
12
Pedestrian Crossing Scenarios 13
14
In the following analysis of three different pedestrian phasing strategies the selected split-phasing design 15
is Protected Left-Turn Arrow Display. Based on a previous study (12) drivers prefer this type of phasing 16
design which requires that a green arrow be displayed for each left turn movement. The investigated 17
pedestrian crossing designs are as follows: 18
19
Scenario 1: Single-Stage Crossing with Pedestrian Timing Concurrent with Phases
20
Single-stage crossing refers to the case when the pedestrian crossing is accomplished in one stage without 21
requiring/allowing the pedestrians to wait at a central refugee island (if available). As shown in Figure 22
1(a), pedestrians are served in parallel with the moving traffic. For example, the pedestrians using the 23
west crosswalk receive a Walk indication concurrently with the adjacent through-vehicle movement 24
(Phase Ф4), and the pedestrian on the other side of the street would be served simultaneously with the 25
northbound through movement (Phase Ф8). The dashed right-turn arrow in Figure 1 represents a 26
permitted movement in which the vehicles are required to yield to the pedestrians. 27
The most common signal indications at crosswalks consist of three main intervals: “Walk”, 28
“Flashing Don’t Walk” (FDW), and “Don’t Walk”. When a pedestrian is detected by push-button 29
actuation in the system, the minimum phase duration is determined by the time required to traverse the 30
crosswalk and the time needed to serve the vehicular traffic. It is possible that the phase duration is longer 31
than the required pedestrian crossing time due to vehicular actuations or the assigned phase split time, but 32
sufficient crossing time must be assured for the pedestrians. 33
34
Scenario 2: Two-Stage Crossing with Simultaneous Signalization 35
In this scenario pedestrians can traverse the street in two stages; the first stage consists of moving from 36
the sidewalk to a central refugee island and the second stage from the center refugee island to the far side 37
sidewalk. In this case the pedestrian signal heads must be split to control pedestrian movement in each 38
stage. From a signalization standpoint, the signals shown both on the edge of the curb and on the central 39
refugee island should display the same indications, so pedestrians can cross to the central refugee island 40
and wait for the next Walk indication to finish the path. This type of design seems to be more efficient for 41
vehicles and is widely used in some European countries, mainly because the impact of long pedestrian 42
crossing time on phase duration can be minimized. The shorter the length of the crosswalk, the less 43
pedestrian clearance time is needed. Figure 1(b) illustrates the phasing scheme associated with this 44
scenario and depicts the pedestrian phases for both stages. Consider the east crosswalk where pedestrians 45
move concurrently with the northbound traffic movement (Phase Ф8) in two stages. 46
47
Scenario 3: Two-Stage Crossing with Combination of Simultaneous and Separate Signalization
48
By combining two of the main types of signalization, simultaneous and separate (as stated in the literature 49
review), improvements can be applied to the system. Generally, separate signalization refers to the case 50
when the indication for one of two successive crosswalks displays the Walk interval earlier than the other 51
(18). As shown in Figure 1(c), for both stages on each crosswalk, the pedestrian intervals are timing 1
concurrently with their adjacent vehicle through movement (exactly the same as scenario 2). The option 2
added to this design is that the pedestrian timing for one stage of each of the crosswalks can overlap with 3
the left-turning vehicles that do not conflict with the pedestrian movement. For example, for the first stage 4
of the west crosswalk, pedestrians can start walking concurrently with the southbound through movement 5
(Phase Ф4) and the eastbound left-turn movement (Phase Ф5). While vehicles are completing Phase 5, 6
there is no conflict between vehicles and pedestrians in the upper stage of the west crosswalk. This makes 7
it possible for both the pedestrians waiting on the northwest corner planning to move south and those on 8
the central island intending to reach the northwest corner to walk simultaneously with eastbound left-9
turning vehicles. 10
Figure 1(d) shows the controller phase and ring configuration associated with all the scenarios. 11
Pedestrian overlaps A and B corresponds to phase 1 and phase 7 with overlap of northbound through 12
movement, phase 8 (A = 1+8 and B = 7+8). Accordingly pedestrian overlaps C and D corresponds to 13
phase 5 and phase 3 with overlap of southbound through movement, phase 4 (C = 5+4 and D=3+4). 14 15 16 (a) 17 18 19 (b) 20 21 22 (c) 23 24 25 26 27
N
N
N
1
(d) 2
FIGURE 1 Phasing scheme of (a) single-stage crossing, (b) two-stage crossing with simultaneous 3
signalization, (c) two-stage crossing with simultaneous and separate signalization, (d) Phase Overlaps 4
5
Visualization Method: Radar Diagrams 6
7
Performance assessment in this paper is a combination of performance measures of multiple 8
transportation modes in the context of the three pedestrian scenarios. Radar diagrams are utilized to 9
visualize the performance measures in an easy to read and understand way. 10
Radar diagrams, which are widely used in the field of organizational development (19), 11
demonstrate multivariate data in the form of a two-dimensional diagram. In the context of transportation, 12
this is a tool that could help monitoring the improvements at intersections. Figure 2 is an illustrative 13
sample of the radar diagrams used in this paper. On each axis of this diagram one of the twelve possible 14
movements at an intersection is shown and the selected performance measure is average travel time in 15
seconds for passenger vehicles on each movement. For example, passenger vehicles on Northbound Right 16
Turn (NBRT) approach spend the least amount of time (17 seconds), and cars on Southbound Left Turn 17
(SBLT) movement have the longest travel time (78 seconds). In the next section the three pedestrian 18
design scenarios will be analyzed using performance measures capturing the performance of multiple 19
modes using a single radar diagram. 20
Different visualization tools can be used to construct the radar diagrams including InDesign, 21
Matlab, Excel, CADD, and LaTex. InDesign and Excel were used in this paper. Radar diagrams have 22
been implemented in a dynamic web application using “Highcharts” and “Chart.js” packages in java. 23
FIGURE 2 Sample radar diagrams used to visualize traffic performance 24
ANALYSIS OF RESULTS 1
2
Micro-simulation Network Description 3
Based on a traffic simulation software comparison study (20), micro-simulation is a powerful tool that 4
captures driver behavior and simulates the movement of individual vehicles on a network. Another 5
advantage of micro-simulation is that it assigns each vehicle a realistic performance characteristic while 6
traveling through the network. Similarly, individual vehicle performance can be measured and used in 7
analysis. Kaseko et al. (21) concentrated on micro-simulation’s pedestrian modeling capabilities that are 8
useful for analyzing the pedestrian and vehicle interactions in an urban traffic environment. 9
Figure 3 shows the layout of a section of Daisy Mountain Drive, an arterial in Anthem, Arizona, 10
which was coded in the VISSIM (Version 6) microscopic simulation model. One intersection was 11
selected as the focus of this study. Four modes were considered in the model: passenger vehicles, 12
pedestrians, buses, and trucks (Heavy Good Vehicles - HGVs). The traffic composition used was: 84% 13
passenger vehicle, 11% truck (HGV), and 5% transit for vehicular traffic. Pedestrian flows are considered 14
for each crosswalk with an average of 170 ped/h assumed. The pedestrian mean speed was set to 3.5ft/s in 15
all crosswalks. In order to capture vehicle queues during congested conditions, the roads were extended 16
800ft beyond the intersection in the simulation model. 17
The duration of the simulation was set to 60 minutes plus 15 minutes of a warm up period. The 18
simulation resolution chosen for all the scenarios is 1 second per time step. Based on Fellendorf et al. 19
(22), in order to get sufficiently precise results for almost all the traffic management applications, this 20
resolution is appropriate. Ten (10) unique random seeds were used for experimental replications to make 21
allowances for the stochastic variations. All the simulations are conducted with actuated traffic signal 22
controls for both vehicles and pedestrians to match the current field operations. The performance 23
measures reported in the study were generated by the VISSIM model. 24
25
26
FIGURE 3 Test Bed of six intersections along Daisy Mountain Drive in Anthem, AZ 27
28
Three analysis approaches were considered: 29
Comparing the performance measures of each scenario for a specific mode, 30
Assessing the performance measures of one traffic signal phase in each scenario and comparing 31
all the measures for each mode associated with the traffic signal phase. 32
Comparing performance measures of one mode in one specific scenario, and trying to determine 33
the correlation between different measures related to that mode. 34
35
One Mode, Different Approaches, Different Scenarios 36
In this analysis one specific mode (pedestrians) was analyzed. Figure 4 depicts the pedestrian travel time, 37
in seconds, for pedestrians to walk from one corner of the intersection to another. For example, the 38
average travel time (including waiting times) varies for moving from the SouthWest corner (SW) to the 1
NorthWest corner (NW) under different scenarios, labeled as SW-NW with the square (red) representing 2
the 2-stage crosswalk with simultaneous signalization, triangle (green) representing the 2-stage crosswalk 3
with simultaneous and separate signalization, and the circle (blue) representing the 1-stage crosswalk. The 4
single-stage crosswalk has an average travel time of 103 seconds, while the second scenario (two-state 5
crosswalk with simultaneous signalization) has an average of 179 seconds, and for the third scenario 6
(two-stage crosswalk with combination of simultaneous and separate signalization) has an average of 114 7
seconds. 8
9
FIGURE 4 Pedestrian Travel Time including Waiting Times for Different Movements and 10
Scenarios 11
Pedestrian travel time for the second scenario (two-stage crossing with simultaneous 12
signalization) has the highest value for almost all the pedestrian movements. The third scenario (two-13
stage crossing with a combination of simultaneous and separate signalization) is placed roughly in 14
between the other two, and could be interpreted as the trade-off scenario. The travel time for the second 15
and third scenarios are higher than that of first scenario since pedestrians have to wait at the central 16
refugee island, which increases their delay and, subsequently, their travel time. 17
18
One Approach, Different Modes, Different Scenarios 19
In this analysis one approach is analyzed by considering the different modes and the different design 20
scenarios. The eastbound through movement was selected to analysis since the 2-stage crosswalk impacts 21
the east and west-bound movements which have the highest volume of traffic demand. 22
As shown in Figure 5, two modes, trucks and passenger vehicles, were considered by comparing travel 23
time and delay. 24
1
FIGURE 5 Eastbound Through Movement Analysis for Different Scenarios and Different Modes 2
The Radar diagram shows that travel time and delay (in seconds) for both passenger vehicles and 3
trucks in the first scenario (single-stage crossing) are higher than that of the second and third scenarios. 4
Generally, considering a measure like vehicle travel time, two-stage crosswalks are better than single-5
stage crosswalks since the pedestrian intervals are shorter and the signal can cycle faster. In the previous 6
section, it was demonstrated that pedestrian travel time in one-stage crossings had the lowest value, and 7
among two-stage crossings, the third scenario is preferred because of its lower travel time and delay. 8
Again, there exists a trade-off between the travel time of pedestrians and vehicles, depending on the 9
modal demand, and an appropriate strategy could be chosen to achieve the objectives of the system 10
operators. For example, in an intersection where the volume of vehicles is much higher than the volume 11
of pedestrians, a two-stage crossing is recommended. Also, when the street to be crossed is wide, it’s 12
safer to have a two-stage crosswalk (23), and an appropriate phasing design (two-stage crossing with 13
combination of simultaneous and separate signalization) could provide the best service for all road users. 14
15
One Scenario, Different Approaches, Different Measures 16
In this analysis two performance measures for different movements of a signalized intersection, are 17
analyzed for one scenario (one-stage crossing). Figure 6 illustrates the relationship between two 18
performance measures: travel time in seconds and vehicle throughput for passenger vehicles. 19
20
FIGURE 6 Passenger Vehicles Analysis with Different Measures 21
The axes of this diagram don’t have specific units, so that both travel time and the number of 1
vehicles can be compared in one diagram. Combining measures help identify the factors that can 2
contribute to the observed performance. For example the northbound right-turn movement (NBRT) serves 3
a relatively a large volume of vehicles, but their travel time is quite short. The primary reason is that there 4
is a dedicated right-turn lane on this approach, which allows the vehicles to discharge during the red as 5
well as during the green indications. The westbound right turn (WBRT) and the eastbound right turn 6
(EBRT) movements exhibit the same behavior, but the southbound right-turn (SBRT) exhibits a different 7
behavior. The SBRT movement doesn’t have an exclusive right turn lane, hence the travel time is higher. 8
Construction of a dedicated SBRT lane would reduce the travel time. The through movements all have 9
similar travel time despite the higher volume on the westbound though approach. 10
This type of diagram could be extended for other scenarios and transportation modes as well. 11
Hence, the concept of a multi-modal transportation dashboard consisting of different radar diagrams was 12
developed to help understand the overall situation of the system and to assist the decision makers in 13
choosing the most appropriate design depending on the actual condition of the intersection and the trade-14
offs between performance measures, modes, and movements. 15
Dashboard 16
17
When considering the complex interactions of the different transportation modes in complicated network 18
geometry, there is a need for a tool that allows visualization of all the modes at the same time. The 19
dashboards provided in this paper are able to show different measures of various modes under specific 20
scenarios. They are presented as follows: first, performance measures of all the travelers at only one 21
intersection are investigated and then, radar diagrams are deployed to show the efficiency of this tool 22
even in a corridor with numerous intersections. 23
24
Intersection-level
25
Figure 7(a) is an example of a dashboard under the two-stage crossing with simultaneous and separate 26
signalization design but at two different operational hours (peak vs. off-peak). Vehicle and pedestrian 27
demands are assumed to be 825 vehicles per hour per lane and 350 pedestrian per hour per crosswalk 28
during the peak period and to be 325 vehicles per hour per lane and 180 pedestrian per hour per crosswalk 29
during the off-peak period. This dashboard consists of 4 radar diagrams, one for each mode concentrating 30
on the following measures: travel time for passenger vehicles, delay for transit, number of stops for 31
trucks, and volume for pedestrians. The pie chart in the middle depicts the distribution of modes in the 32
system. 33
The impact of peak hour demand on vehicle performance measures is clearly shown. The increase 34
in travel time for passenger vehicles and number of stops for trucks in the southbound through and 35
southbound right-turn movements is significantly more than the increase in the other movements. This 36
shows that there is a bottleneck on this approach during the peak period. Actually, the reason for this 37
observation is that there is no dedicated right-turn lane for vehicles moving southbound at the 38
intersection, as discussed above. 39
Figure 7(b) shows a dashboard consisting of 12 radar diagrams and represents different measures 40
considering the alternative pedestrian crossing treatments: travel time for passenger vehicles in blue, 41
number of stops for trucks in orange, delay for transit vehicles in green, and travel time for pedestrians in 42
red. 43
1 2 3
FIGURE 7 (a) Peak vs. Off-peak Performance Observation, (b) Comprehensive Multi-Modal 4
Assessment under the Three Pedestrian Crossing Designs 5
6
Section-level
1
Evaluating the performance of the road users along an arterial corridor is of interest for traffic engineers, 2
planners, and researchers. Figure 8 shows different measures of multiple modes under congested (Peak 3
hour) and uncongested (Off-Peak hour) conditions in the test bed corridor which consists of six 4
intersections, as mentioned previously. The pie chart in the middle shows the distribution of traffic 5
demand in the corridor. 6
7
8
FIGURE 8 Hexagonal Radar Diagrams to Show the Performance Measures in a Corridor 9
consisting of Six Intersections 10
11
The main advantage of using these dashboards as a visualization technique over the typical data 12
demonstration methods is that the same data showed in Figures 7 and 8, requires the total number of 96 13
rows and 17 columns in Tables or 238 vertical rectangular blocks in bar charts. Finding the value of one 14
specific measure for a particular mode under one of the design scenarios would be challenging. The 15
dashboards provide a quick means to visualize the relationships between different measures, modes, and 16
movements as well as consideration of alternative control scenarios can impact the users of the system. 17 18 19 CONCLUSION 20 21
This research has developed a method that can support the visual assessment of a multi-modal traffic 22
system associated with both vehicular and pedestrian flow. An arterial corridor in Maricopa County is 23
analyzed considering three different pedestrian crossing design scenarios and multiple traffic conditions. 24
Generally, according to the radar diagrams generated, strategies that benefit the mobility of 25
vehicular traffic may not be beneficial from a pedestrian point of view. There is almost always a trade-off 26
between vehicular and pedestrian measures. Although improving either of these two modes most probably 27
results in impacting the other, ignoring pedestrians in assessments of the system leads to unintended 1
impacts to some of the road users. Similarly, there will be trade-offs between different modes, such as 2
transit buses, trucks and passenger vehicles, that also need to be considered. Scaling plays an important 3
role in the dashboards, like every other visualization tool. Choosing a meaningful way to show the values 4
of performance measures (e.g. proportions, percentages, normalized values, etc.) on the diagrams should 5
be given careful consideration. The proposed dashboards presented in this paper play important roles to 6
bridge the gap between the end users of such tools and the presenter of the data analytics. 7 8 9 ACKNOWLEDGMENT 10 11
This work has been supported by Arizona Connected Vehicle Initiative which is a collaboration between 12
the Maricopa County Department of Transportation SMARTDrive Program, the Arizona Department of 13
Transportation, and the Cooperative Transportation Systems Pooled Fund Multi-Modal Intelligent Traffic 14
Signal Systems Project. 15 16 17 REFERENCES 18 19
1. Pratt, R., and T. Lomax. Performance Measures for Multimodal Transportation Systems. 20
Transportation Research Record: Journal of the Transportation Research Board, Vol. 1518, No. -1, 21
Jan. 1996, pp. 85–93. 22
2. Hubbard, S., D. Bullock, and C. Day. Integration of Real-Time Pedestrian Performance Measures 23
into Existing Infrastructure of Traffic Signal System. Transportation Research Record: Journal of
24
the Transportation Research Board, Vol. 2080, No. -1, Dec. 2008, pp. 37–47. 25
3. Ishaque, M. M., and R. B. Noland. Trade-Offs Between Vehicular and Pedestrian Traffic Using 26
Micro-Simulation Methods. Transport Policy, Vol. 14, No. 2, Mar. 2007, pp. 124–138. 27
4. Kim, K., I. Brunner, and E. Yamashita. Use of Safety Viewgrams to Visualize Driver and 28
Pedestrian Interactions. Transportation Research Record: Journal of the Transportation Research
29
Board, Vol. 2002, No. -1, Jan. 2007, pp. 72–77. 30
5. Noland, R. Pedestrian Travel Times and Motor Vehicle Traffic Signals. Transportation Research
31
Record: Journal of the Transportation Research Board, Vol. 1553, No. -1, Jan. 1996, pp. 28–33. 32
6. Noland, R. B. Optimal Travel Time Trade-Offs at Signalized Pedestrian Crossings. Center for 33
Transport Studies, Department of Civil and Environmental Engineering, Imperial College London, , 34
2003. 35
7. Akin, D., and V. P. Sisiopiku. Modeling Interactions Between Pedestrians and Turning Vehicles at 36
Signalized Crosswalks Operating Under Combined Pedestrian-Vehicle Interval. Presented at the 37
Transportation Research Board 86th Annual Meeting, 2007. 38
8. Quaye, K., L. Leden, and E. Hauer. Pedestrian Accidents and LeftTturning Traffic at Signalized
39
Intersections. AAA Foundation for Traffic Safety, Washington, D. C., 1993. 40
9. Zhang, L., and P. Prevedouros. Signalized Intersection Level of Service Incorporating Safety Risk. 41
Transportation Research Record: Journal of the Transportation Research Board, Vol. 1852, No. -1, 42
Jan. 2003, pp. 77–86. 43
10. Hubbard, S., D. Bullock, and F. Mannering. Right Turns on Green and Pedestrian Level of Service: 44
Statistical Assessment. Journal of Transportation Engineering, Vol. 135, No. 4, 2009, pp. 153–159. 45
11. Urbanik, T., Z. Tian, and R. Engelbrecht. Pedestrian Timing with Split Phasing Operations. Proc.,
46
70th ITE Annual Meeting, ITE, Washington, D.C., 2000. 47
12. Tian, Z., T. Urbanik, R. Engelbrecht, and K. Balke. Pedestrian Timing Alternatives and Impacts on 48
Coordinated Signal Systems Under Split-Phasing Operations. Transportation Research Record:
49
Journal of the Transportation Research Board, Vol. 1748, No. -1, Jan. 2001, pp. 46–54. 50
13. Wang, X., Z. Z. Tian, P. Ohene D. Eng., and P. J. V. Koonce. Pedestrian Delay Models at 1
Signalized Intersections Considering Signal Phasing and Pedestrian Treatment Alternatives. 2
Presented at the Transportation Research Board 88th Annual Meeting, 2009. 3
14. Ma, W., Y. Lin, and X. Yang. Design Method of Pedestrian Phases at Multi-phase Signal 4
Intersection. Journal of Traffic and Transportation Engineering, Vol. 4, No. 2, 2004, pp. 103–106. 5
15. Tian, Z. Z., K. K. Kacir, M. A. Vandehey, and H. Long. Signal Timing Strategies in Dealing with 6
Pedestrian Crossings. Presented at the Transportation Frontiers for the Next Millennium: 69th 7
Annual Meeting of the Institute of Transportation Engineers, 1999. 8
16. Traffic Control and Traffic Safety Guidelines for Traffic Signal. Road and Transportation Research 9
Association Steering Committee, Cologne, Germany, 1992. 10
17. Cooperative Transportation Systems Pooled Fund Study – Research | Center for Transportation 11
Studies. http://www.cts.virginia.edu/cooperative-transportation-systems-pooled-fund-study-12
research/. Accessed Aug. 1, 2014. 13
18. Ma, W., Y. Liu, H. Xie, and X. Yang. Multiobjective Optimization of Signal Timings for Two-14
Stage, Midblock Pedestrian Crosswalk. Transportation Research Record: Journal of the
15
Transportation Research Board, Vol. 2264, No. -1, Dec. 2011, pp. 34–43. 16
19. Kaczynski, D., L. Wood, and A. Harding. Using Radar Charts with Qualitative Evaluation 17
Techniques to Assess Change in Blended Learning. Active Learning in Higher Education, Vol. 9, 18
No. 1, Mar. 2008, pp. 23–41. 19
20. Jones, S. L., A. J. Sullivan, N. Cheekoti, M. D. Anderson, and D. Malave. Traffic Simulation
20
Software Comparison Study. University Transportation Center for Alabama, Jun. 2004. 21
21. Kaseko, M. S. Comparative Evaluation of Simulation Software for Traffic Operations. Nov. 2002. 22
22. Fellendorf, M., and P. Vortisch. Validation of the Microscopic Traffic Flow Model VISSIM in 23
Different Real-World Situations. Presented at the National Research Council (U.S.). Transportation 24
Research Board. Meeting (80th : 2001 : Washington, D.C.). Preprint CD-ROM, 2001. 25
23. Alexson, P. W., D. Y. Chesney, D. V. Galvan, J. B. Kirshbaum, P. E. Longmuir, C. Lyons, and K. 26
M. Wong. Designing Sidewalks and Trails for Access. FHWA, U.S. Department of Transportation, 27
1999. 28
