Text Box: Research Report No. UVACTS-15-0-110
Date: 7/17/2007

Evaluation of the Adaptive Maximum Feature in the EPAC300 Actuated Traffic Controller Using Hardware-in-the-Loop Simulation

By:

Byungkyu “Brian” Park, Ph.D.

University of Virginia

Thornton Hall

351 McCormick Road

P.O. Box 400742

Charlottesville, VA 22904

Phone: 434-924-6347

Fax: 434-982-2951

E-mail: bpark@virginia.edu

Ilsoo Yun, Ph.D.

University of Virginia

Thornton Hall

351 McCormick Road

P.O. Box 400742

Charlottesville, VA 22904

Phone: 434-825-9271

Fax: 434-982-2951

E-mail: iy6m@virginia.edu

Matthew Best

University of Virginia

Thornton Hall

351 McCormick Road

P.O. Box 400742

Charlottesville, VA 22904

M.S., Civil Engineering, May 2007

Phone: 434-996-9118

Fax: 434-982-2951

E-mail: mgb3e@virginia.edu

A Research Project Report for the Intelligent Transportation Systems Implementation Center (ITS)

A U.S. DOT University Transportation Center

Dr. Byungkyu “Brian” Park, Ph.D.

Department of Civil Engineering

Email: bpark@virginia.edu

Center for Transportation Studies at the University of Virginia produces outstanding transportation professionals, innovative research results and provides important public service. The Center for Transportation Studies is committed to academic excellence, multi-disciplinary research and to developing state-of-the-art facilities. Through a partnership with the Virginia Department of Transportation’s (VDOT) Research Council (VTRC), CTS faculty hold joint appointments, VTRC research scientists teach specialized courses, and graduate student work is supported through a Graduate Research Assistantship Program. CTS receives substantial financial support from two federal University Transportation Center Grants: the Mid-Atlantic Universities Transportation Center (MAUTC), and through the National ITS Implementation Research Center (ITS Center). Other related research activities of the faculty include funding through FHWA, NSF, US Department of Transportation, VDOT, other governmental agencies and private companies.

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, in the interest of information exchange. The U.S. Government assumes no liability for the contents or use thereof.

Text Box: CTS Website Center for Transportation Studies
http://cts.virginia.edu University of Virginia
351 McCormick Road, P.O. Box 400742
Charlottesville, VA 22904-4742
434.924.6362

1. Report No.	2. Government Accession No.			3. Recipient’s Catalog No.
UVACTS-15-0-110
4. Title and Subtitle				5. Report Date
Evaluation of the Adaptive Maximum Feature in the EPAC300 Actuated Traffic Controller Using Hardware-in-the-Loop Simulation				July 17, 2007
				6. Performing Organization Code

7. Author(s)				8. Performing Organization Report No.
Dr. Byungkyu “Brian” Park, Ph.D., Ilsoo Yun, Ph.D., Matthew Best
9. Performing Organization and Address				10. Work Unit No. (TRAIS)
Center for Transportation Studies
University of Virginia				11. Contract or Grant No.
PO Box 400742 Charlottesville, VA 22904-7472
12. Sponsoring Agencies' Name and Address				13. Type of Report and Period Covered
Office of University Programs, Research Innovation and Technology Administration US Department of Transportation 400 Seventh Street, SW Washington DC 20590-0001				Final Report
				14. Sponsoring Agency Code

15. Supplementary Notes

16. Abstract
Several actuated traffic controllers contain the Adaptive Maximum feature, with which the controllers can adjust the maximum green intervals for actuated phases within specified upper and lower limits according to fluctuating traffic demand. It has been difficult, however, to evaluate the merits of this and other features available in modern traffic controllers. Recent developments in traffic signal control systems hardware and software technologies have now made it possible to evaluate controller features in a realistic and risk-free environment using hardware-in-the-loop simulation (HILS). HILS is a method of simulation in which one or more actual traffic signal controllers are physically linked with a microscopic traffic simulator. This paper demonstrates the performance of the Adaptive Maximum feature using HILS, which consisted of an EPAC300 traffic controller and the VISSIM microscopic simulation model. The demonstration was conducted at an isolated, fully actuated intersection in Richmond, Virginia. In a feasibility test of the Adaptive Maximum feature, the HILS results indicated that the Adaptive Maximum feature was able to provide traffic signal control operations as efficiently as normal maximum green intervals optimized by SYNCHRO. However, in a robustness test, where fifteen percent changes in traffic volumes were considered, the Adaptive Maximum feature outperformed the normal maximum green intervals.
17 Key Words Traffic Controller, Hardware-in-the-Loop Simulation			18. Distribution Statement
			No restrictions. This document is available to the public.
19. Security Classif. (of this report)		20. Security Classif. (of this page)			21. No. of Pages	22. Price
Unclassified		Unclassified			22	N/A

Abstract

Several actuated traffic controllers contain the Adaptive Maximum feature, with which the controllers can adjust the maximum green intervals for actuated phases within specified upper and lower limits according to fluctuating traffic demand. It has been difficult, however, to evaluate the merits of this and other features available in modern traffic controllers. Recent developments in traffic signal control systems hardware and software technologies have now made it possible to evaluate controller features in a realistic and risk-free environment using hardware-in-the-loop simulation (HILS). HILS is a method of simulation in which one or more actual traffic signal controllers are physically linked with a microscopic traffic simulator.

This paper demonstrates the performance of the Adaptive Maximum feature using HILS, which consisted of an EPAC300 traffic controller and the VISSIM microscopic simulation model. The demonstration was conducted at an isolated, fully actuated intersection in Richmond, Virginia. In a feasibility test of the Adaptive Maximum feature, the HILS results indicated that the Adaptive Maximum feature was able to provide traffic signal control operations as efficiently as normal maximum green intervals optimized by SYNCHRO. However, in a robustness test, where fifteen percent changes in traffic volumes were considered, the Adaptive Maximum feature outperformed the normal maximum green intervals.

INTRODUCTION

As traffic congestion becomes more severe in its intensity and duration, traffic engineers have the challenging task of operating traffic signal systems more efficiently. To accomplish this, they have attempted to use green times more efficiently to reduce vehicle delays. For example, the implementation of isolated or coordinated-actuated traffic signal control systems over pre-timed signal control has improved the efficiency of green time allocation. In isolated, fully actuated signal control, detectors are placed on the approaches of an intersection to sense the presence of vehicles from the subject movements. They send these vehicle actuations to the controller, which then adjusts the subject phase length to serve current traffic demands. Under normal actuated operation without the volume-density control, three main settings govern the length adjustment of a phase—minimum green, maximum green, and vehicle extension. For example, if recall is not set for a phase and the phase receives no detector calls during its yellow and red intervals, it is skipped. An advantage of actuated traffic signal control over pre-timed traffic signal control is the ability to terminate the green time of the active phase immediately and to provide additional green time to the phases with demand. This function—gap out—is intended to serve approaches with demand if additional vehicles are not arriving on the current approach within the predetermined gap out time. Gap out provides flexibility in the allocation of green times when needed. Due to these merits, actuated traffic signal control is widely deployed across the U.S. (1).

Modern actuated traffic controllers have additional features that add to the overall efficiency of actuated traffic control. For example, the Adaptive Maximum feature (also known as Dynamic Maximum) is available in several actuated traffic controllers (2, 3, 4). Using this feature, controllers can adjust the maximum green intervals according to traffic fluctuations for actuated phases within specified upper or lower limits. This feature is especially useful at an isolated-actuated signal control.

It has been difficult in the past, however, to evaluate such a feature in the field or laboratory environment. Recent developments in hardware and software technologies in the transportation industry have introduced hardware-in-the-loop simulation (HILS), which allows for the evaluation of traffic controller features in a realistic and risk-free environment. In the engineering industry, HILS is a new method of simulation in which a physical device is added to simulation software to provide an effective platform for developing and testing real-time embedded systems (5). In the case of traffic simulation, a traffic signal controller is connected to a microscopic traffic simulation software program, such as CORSIM or VISSIM, in a personal computer. Here, the external controller operates the traffic signal in the microscopic simulation model instead of the software.

This paper aims to evaluate the Adaptive Maximum feature available in the EPAC300 actuated traffic controller using the VISSIM microscopic simulation model-based HILS. The feasibility of the Adaptive Maximum feature was first investigated. In this test, operations with the Adaptive Maximum settings were compared with the (1) normal maximum green settings optimized by Synchro (6), a macroscopic signal timing optimization model, and (2) excessively large maximum green settings. It is noted that the latter setting was tested to evaluate the performance of actuated control using only the gap out feature. For the feasibility test, observed traffic counts were used as the traffic demands in the VISSIM-based HILS; however, fifteen percent increases and decreases in traffic demands were assumed in the robustness test, which was intended to verify the proper operation of the Adaptive Maximum feature in different volume conditions.

LITERATURE REVIEW

Adaptive Maximum

As discussed, the Adaptive Maximum feature—or a similar one under a different name—is available in several different actuated traffic controllers. According to NTCIP 1202: Objective Definition for Actuated Traffic Signal Control Units (7), the Adaptive Maximum (referred to as Dynamic Maximum in NTCIP 1202) operation can be defined as a cycle by cycle maximum green interval adjustment within an upper and lower limit. The advantage of this feature is that the controller can adjust maximum green intervals according to the degree of traffic demand and fluctuations.

In the EPAC300 traffic controller, shown in Figure 1, the Adaptive Maximum operation has three main parameters: Maximum Green (seconds), Dynamic Maximum (seconds), and Dynamic Step (tenths of seconds). Here, Maximum Green is different from a normal maximum green time setting and works as the lower or upper limit in the operation. For example, after a phase maxes out on two consecutive cycles, and after each successive max out thereafter, the Dynamic Step value is added to the current normal maximum green time until before it is greater than the larger of Maximum Green or Dynamic Maximum. In a similar way, after a phase gaps out on two consecutive cycles, and after each successive gap out thereafter, the Dynamic Step value is subtracted from the current maximum green until before it is less than the smaller of Maximum Green or Dynamic Maximum. If a phase gaps out in one cycle and maxes out in the next cycle or vice versa, the current maximum green will not change (2, 7).

FIGURE 1. EAGLE EPAC300 traffic controller (http://www.itssiemens.com)

Engelbrecht et al. (8) investigated the improvement of diamond interchange operations using advanced features of several different traffic controllers. They tested eight features using HILS, which consisted of CORSIM and actual traffic controllers. The Adaptive Maximum feature, available in Naztec’s Model 980 controllers, was among those features studied at a diamond interchange. The authors applied this feature for a single phase with sudden fluctuations in demand and demonstrated that it can improve traffic operations. A similar concept to the Adaptive Maximum feature is Dynamic Split, which is implemented in coordinated-actuated signal systems and is sometimes referred to as Coordinated Adaptive Split or Critical Intersection Control in different controllers (2, 3). The feature changes the non-coordinated phase splits while maintaining coordination. Sunkari et al. (9) provided analyses of the Dynamic Split feature available in the EPAC300 controller.

Hardware-in-the-Loop Simulation

HILS is one of the most advanced forms of microscopic simulation for traffic signal control systems. This is due to the physical link between the microscopic simulation model and the traffic controller via a controller interface device (10). Engelbrecht et al. refined a methodology to integrate signal controllers into the simulation process using the VISSIM and CORSIM models (11, 12). Examples of microscopic simulation models that are commonly used in HILS include CORSIM, VISSIM, and SimTraffic (13). Currently, several commercial CIDs have become available for use in HILS (13, 14).

CASE STUDY

Site Selection and Data Collection

The test site was located at an urban corridor in Richmond, Virginia. Shown in Figure 2, it is the intersection of Dumbarton Road and Lakeside Avenue and has isolated, fully actuated signal control. This site was chosen due to the availability of assistance from Virginia Department of Transportation (VDOT) personnel in charge of the traffic signal. VDOT provided the geometry, detector layout, and traffic signal controller settings for the test site.

FIGURE 2 Test site in Richmond, Virginia

Traffic counts were collected directly from the site on a weekday in 2006. Table 1 summarizes the fifteen-minute aggregated traffic counts collected from the site visit during the morning peak hour (7:00 a.m.–8:00 a.m.) using a JAMAR hand-held manual traffic data collector. During the data collection, there were several pedestrians; none, however, pushed the pedestrian call button, so pedestrians were not considered in this analysis.

TABLE 1. Traffic Counts during Morning Peak Hour

Time	From North			From East			From South			From West
Time	Right	Thru	Left	Right	Thru	Left	Right	Thru	Left	Right	Thru	Left
7:00 - 7:14	23	199	39	33	153	18	9	120	32	88	63	16
7:15 - 7:29	15	190	32	51	144	22	11	110	51	71	66	11
7:30 - 7:44	9	218	22	14	100	23	17	132	45	58	74	18
7:45 - 7:59	10	158	26	17	73	19	16	138	45	48	82	14
Sum	57	766	119	115	470	82	53	500	173	265	285	59
Peak Hour Factor	0.62	0.88	0.76	0.56	0.77	0.89	0.78	0.91	0.85	0.75	0.87	0.82

VISSIM Network Building

The traffic network for the test site was prepared in VISSIM 4.10-09 (15). To draw the VISSIM network, an aerial photograph of the intersection was used as a background image. Detailed attributes of the intersection, such as detector locations and lengths, were provided by VDOT personnel. To consider the fluctuations in traffic volumes, vehicles in the VISSIM network were generated based on fifteen-minute-interval traffic volumes. Figure 3 shows a screen capture of the VISSIM-based HILS where the traffic signal was controlled by an EPAC300 controller.

FIGURE 3. Screen capture of VISSIM animation

Optimization of Maximum Green

To measure the relative benefits of the use of the Adaptive Maximum feature at the test site, the normal maximum green interval of each phase was optimized using Synchro based on the observed traffic counts (6). Figure 4 shows the network drawn using Synchro, Version 6 (Build 610), where the geometry and traffic signal control settings—other than the maximum green intervals—were identical to those in the VISSIM network. The optimization process in Synchro involved optimizing the intersection cycle length, followed by optimizing the intersection splits. Additionally, the maximum green intervals were optimized as the normal maximum green intervals because there are currently no tools available in traffic simulation and optimization models for optimizing timing plans under the Adaptive Maximum feature. The numbers in Figure 4 indicate the volume-to-capacity ratios based on the traffic volumes collected from the field and the signal timing optimized by Synchro.

FIGURE 4 Synchro network with volume to capacity ratios.

Implementation in the VISSIM-based HILS

Since no existing microscopic simulation program contains the Adaptive Maximum feature, VISSIM-based HILS was used in this study. The HILS interface used a CID developed by the University of Idaho and McCain Traffic Supply, allowing the EPAC300 controller to communicate with the VISSIM simulation software program and control the intersection signal in the simulation (2, 13). Figure 5 shows a data flow diagram of HILS.

FIGURE 5. Data flow in HILS

Sensitivity Analysis of Settings for Adaptive Maximum

The main parameters for the Adaptive Maximum operation are Dynamic Maximum, Maximum Green, and Dynamic Step. The performance of the Adaptive Maximum operation is problem-specific and dependent on these three parameters, so it is important to determine the values for the three parameters for the isolated-actuated signal under consideration. However, the use of the Adaptive Maximum operation is uncommon, so a methodology for finding the proper values of Dynamic Step, Dynamic Maximum, and Maximum Green for the Adaptive Maximum operation is not currently available. It should be noted that in the Adaptive Maximum operation, Maximum Green intervals work as a lower or higher limit for the current maximum green, which is continuously changing according to fluctuations in traffic demand. In this paper, the values of Maximum Green work as the lower limits.

Engelbrecht et al. (8) recommended the following rules of thumb for Adaptive Maximum operations. The lower limit of the current maximum green should be set as low as possible; however, it should be high enough to serve queues approximately 1.3 times the average peak queue length during the period the Adaptive Maximum operation will be in effect (16). In addition, the upper limit of the current maximum green can be set equal to the optimized maximum green interval, provided that the volume-to-capacity ratio is not greater than 0.85 (17).

Even with such rules of thumb, the selection of the three parameters has to be problem-specific. Therefore, this paper conducted sensitivity analyses for the following combinations of the three parameters as indicated in Table 2. The following rules were considered in the parameters selection.

For Dynamic Step—referred to as DS 1, DS 2, and DS 3 for each case, respectively, in Table 2—2.0 or 3.0 seconds, or a combination thereof, were selected. In DS 3, a Dynamic Step of 3.0 seconds was assigned to phases 2 and 6 (through movements on the major street), and 2.0 seconds was set for the other phases.

As discussed by Engelbrecht et al., the values of Maximum Green were set as low as possible (8). To this end, this study applied the following equations:

· MG 1: Minimum green interval

· MG 2: Minimum green interval + vehicle extension interval

To select the Dynamic Maximum, the optimized maximum green interval was considered. However, preliminary evaluations with this setting produced too frequent max outs and vehicle stops. Thus, this study used the following equations:

· DM 1: Optimized maximum green interval + 10 seconds – vehicle extension interval

· DM 2: Optimized maximum green interval + 10 seconds

· DM 3: Optimized maximum green interval + 10 seconds + vehicle extension interval

TABLE 2 Selected Adaptive Maximum Parameters for Sensitivity Analyses

Parameter	Case	Selected Value for Phase (sec)						Remark
		Phase 1	Phase 2	Phase 3	Phase 4	Phase 5	Phase 6
		NB Left	SB Through	EB	WB	SB Left	NB Through
Dynamic Step	DS 1	2.0	2.0	2.0	2.0	2.0	2.0
	DS 2	3.0	3.0	3.0	3.0	3.0	3.0	Default
	DS 3	2.0	3.0	2.0	2.0	2.0	3.0
Dynamic Maximum	DM 1	17	29	23	25	18	28
	DM 2	20	35	26	28	21	34	Default
	DM 3	23	41	29	31	24	40
Maximum Green	MG 1	6	12	6	6	6	12
Maximum Green	MG 2	9	18	9	9	9	18	Default

During the sensitivity analyses of a parameter, the values of other parameters were fixed with those marked as “default” in Table 2. Based on the parameters under a sensitivity analysis, the VISSIM-based HILS was conducted ten times with random seeds. The resulting network performance data are summarized in Tables 3, 4, and 5.

TABLE 3. Sensitivity Analysis Results for Dynamic Step

Parameter	Statistics	Average Delay (sec/veh)	Average Number of Stops (stops/veh)
DS 1	Average	28.34	0.78
DS 1	Standard Deviation	0.86	0.01
DS 2	Average	28.07	0.77
DS 2	Standard Deviation	1.17	0.02
DS 3	Average	28.51	0.78
DS 3	Standard Deviation	1.08	0.02

Note: in this analysis, DM 2 and MG 2 were used as default values

TABLE 4. Sensitivity Analysis Results for Maximum Green

Parameter	Statistics	Average Delay (sec/veh)	Average Number of Stops (stops/veh)
MG 1	Average	28.26	0.77
MG 1	Standard Deviation	0.84	0.01
MG 2	Average	28.07	0.77
MG 2	Standard Deviation	1.17	0.02

Note: in this analysis, DS 2 and DM 2 were used as default values

TABLE 5. Sensitivity Analysis Results for Dynamic Maximum

Parameter	Statistics	Average Delay (sec/veh)	Average Number of Stops (stops/veh)
DM 1	Average	28.41	0.78
DM 1	Standard Deviation	1.28	0.02
DM 2	Average	28.07	0.77
DM 2	Standard Deviation	1.17	0.02
DM 3	Average	28.43	0.78
DM 3	Standard Deviation	1.21	0.03

Note: in this analysis, DS 2 and MG 2 were used as default values

Through the sensitivity analyses, Dynamic Step was found to play a relatively important role during the Adaptive Maximum operation. It was also found that the values of Dynamic Maximum and Maximum Green are less important if they capture an appropriate range for the current maximum green interval. The final selections of the three parameters are listed at the bottom of Table 6.

TABLE 6 Signal Control Settings for Three Scenarios

Signal Control Settings		Phase-Related Data Settings (sec)
		Phase 1	Phase 2	Phase 3	Phase 4	Phase 5	Phase 6
		NB Left	SB Through	EB	WB	SB Left	NB Through
Yellow Time		3.5	4.0	3.5	3.5	3.5	4.0
All-Red Time		2.0	2.0	2.0	2.0	2.0	2.0
Minimum Green		6.0	12.0	6.0	6.0	6.0	12.0
Extension		3.0	6.0	2.5	2.5	3.0	6.0
Added Initial		-	1.5	-	-	-	1.5
Time-Before-Reduction		-	20.0	-	-	-	20.0
Time-To-Reduce		-	15.0	-	-	-	15.0
Minimum Gap		-	4.5	-	-	-	4.5
Recall		-	Ö	-	-	-	Ö
Large Maximum	Maximum Green	35	40	35	35	35	40
Optimized Maximum	Maximum Green	10	25	16	18	11	24
Adaptive Maximum	Maximum Green	9	18	9	9	9	18
	Dynamic Max	20	35	26	28	21	34
	Dynamic Step	3.0	3.0	3.0	3.0	3.0	3.0

Feasibility Test

This feasibility test was intended to verify whether the operation of an isolated, fully actuated traffic signal with the Adaptive Maximum settings was more or less efficient than with the optimized maximum green settings or with the large maximum green settings. In the experiment, three different signal control scenarios were simulated—large maximum green settings, optimized maximum green settings, and Adaptive Maximum settings. The large maximum green settings were based on the assumed non-optimized maximum green settings, where the controller mainly terminates green time because of gap outs. For the feasibility test, the observed traffic counts were used as the traffic demands in the VISSIM-based HILS. Table 6 shows the details of the three different signal control scenarios.

Given that microscopic simulation model performance measures sometimes show significant variability, it is important to consider such variability during evaluation. Thus, ten randomly seeded VISSIM-based HILS runs were conducted; the average and standard deviation values of two measures of effectiveness (MOEs)—average delay per vehicle and average number of stops per vehicle—are shown in Table 7

TABLE 7 Summary of Feasibility Test Results

Signal Control Setting Scenario		Average Delay (sec/veh)	Average Number of Stops (stops/veh)
Normal Large Maximum Green	Average	31.30	0.74
Normal Large Maximum Green	Standard Deviation	1.21	0.02
Normal Optimized Maximum Green	Average	28.29	0.80
Normal Optimized Maximum Green	Standard Deviation	1.50	0.03
Adaptive Maximum	Average	28.07	0.77
Adaptive Maximum	Standard Deviation	1.17	0.02

It can be seen that the operation of the isolated, fully actuated traffic signal with the optimized maximum green settings and Adaptive Maximum settings produced less delays than with the large maximum green settings. However, the operation with the large maximum green settings generated fewer average numbers of stops than the others. This was due to the operations with the optimized maximum green settings and the Adaptive Maximum settings being able to terminate green times earlier than those with the large maximum green settings. For example, if there was a demand for ten seconds of green time on a phase, the operation with the large maximum green settings gapped out after the vehicle extension time passed. As a result, queued vehicles in other phases waited during the vehicle extension time even though no vehicles were using that green time. If the operations with the optimized maximum green settings and the Adaptive Maximum settings had the ten seconds of green time as the maximum green interval for the phase, the signal controller maxed out at the tenth second. As a result, the operations with the optimized maximum green settings and the Adaptive Maximum settings were able to give green time to other phases three seconds—equivalent to the vehicle extension time—sooner by minimizing the maximum green intervals.

In the comparison between the Adaptive Maximum and the optimized maximum green settings, there was no significant difference based on a t-test (p-value of 0.36 for the average delay) (18). However, the Adaptive Maximum settings produced significantly fewer vehicle stops than the optimized maximum green setting (p-value of 0.01 in t-test). Thus, it can be concluded that the Adaptive Maximum settings work as efficiently as the optimized maximum green settings.

Robustness Test

For the feasibility test, the observed traffic counts were used as the traffic demand in the VISSIM-based HILS. However, fifteen percent increases and decreases in traffic demand were assumed in the robustness test to verify the proper operation of the Adaptive Maximum feature in different traffic patterns. These fluctuations in traffic volumes were assumed to be caused by normal time-to-time or day-to-day changes in traffic patterns.

To consider variability in microscopic simulation models linked to HILS, ten random-seeded replications were made for the three scenarios with two sets of traffic volumes. The resultant MOEs, including average delay and average number of stops, are summarized in Table 8.

TABLE 8 Summary of Robustness Test Results

Signal Control Settings		15% Increased Traffic Volume		15% Decreased Traffic Volume
Signal Control Settings		Average Delay (sec/veh)	Average Number of Stops (stops/veh)	Average Delay (sec/veh)	Average Number of Stops (stops/veh)
Normal Large Maximum Green	Average	39.44	0.84	29.16	0.75
Normal Large Maximum Green	Standard Deviation	1.00	0.03	1.34	0.03
Normal Optimized Maximum Green	Average	36.44	1.00	26.32	0.77
Normal Optimized Maximum Green	Standard Deviation	2.34	0.12	1.21	0.02
Adaptive Maximum	Average	34.86	0.85	25.66	0.76
Adaptive Maximum	Standard Deviation	1.46	0.02	1.06	0.03

In both traffic volume cases, the same results as those in the previous section were produced, i.e., the operations of the isolated, fully actuated traffic signal with the optimized maximum green settings and the Adaptive Maximum settings produced fewer delays than those with the large maximum green settings. Additionally, operations with the large maximum green settings generated a fewer average number of stops than the others. Further, the MOEs from the Adaptive Maximum settings were significantly better than those from the optimized maximum green settings (p-values of 0.02 for average delay in the decreased volume case and 0.04 for the increased volume case). The operation with the Adaptive Maximum settings also reduced the number of vehicle stops.

Based on the summary shown in Table 8, it can be concluded that operations with the Adaptive Maximum settings handles fluctuating traffic patterns more efficiently than those with the optimized or large maximum green settings.

Conclusions and Recommendations

This paper evaluated the feasibility and robustness of the Adaptive Maximum feature available in the EPAC300 actuated controller using the VISSIM-based HILS, which was able to provide a risk-free and realistic evaluation environment by physically coupling a microscopic simulation model and an actual traffic controller. To investigate the potential benefits of the Adaptive Maximum feature, the two alternatives—large maximum green times and optimized maximum green times—were compared in an isolated, fully actuated traffic signal. The comparison was conducted via multiple runs of VISSIM-based HILS.

The results of the feasibility test indicated that the Adaptive Maximum feature has benefits over both the large and optimized maximum green settings. The results of a robustness test also showed that operations with the Adaptive Maximum settings are more flexible and efficient at adapting fluctuations in traffic patterns than the others.

The Adaptive Maximum feature also holds the following benefits:

· One Adaptive Maximum setting is able to cover different times of the day and days of the week because of its ability to adjust the maximum green time according to fluctuations in traffic demands. As a result, less time and effort is needed to develop different signal timing plans for different time periods

· Due to its flexibility in traffic operations, the Adaptive Maximum feature can be applied to any isolated, fully actuated signal without an extensive engineering study

To spread the use of the Adaptive Maximum feature in the field, the following is recommended for future research:

· An intensive study for identifying the traffic patterns (volumes, fluctuations, etc.) under which the Adaptive Maximum feature can produce the greatest benefits

· A methodology for finding the proper values of the three parameters—Dynamic Maximum, Maximum Green and Dynamic Step—should be established. There is currently no tool to find the optimal values for these settings. As an alternative, a stochastic optimization method coupled with software-in-the-loop simulation (SILS) should be considered (19). SILS links a microscopic simulation model and a traffic signal controller software emulator. Here, SILS is able to emulate the feature in a microscopic simulation environment with reduced required computational time.

· A feasibility study of the Adaptive Maximum feature for coordinated-actuated signal control—sometimes referred to as Dynamic Split these systems—should be conducted.

ACKNOWLEDGMENTS

The authors thank Mr. Mike Goodman and the other staff members at the VDOT Richmond District for their support during the data collection and network coding used in this research.

REFERENCES

1. Tarnoff, P. J. and J. Ordóñez. Signal Timing Practices and Procedures – State of the Practice. Institute of Transportation Engineers. March 2004.

2. EAGLE Traffic Control Systems. EPAC300 Actuated Controller Unit - Product Manual. EAGLE Traffic Control Systems. Austin, Texas. 1997.

3. Naztec Incorporated. Intersection Control Software, TS-2 Version 50.x. Sugarland, Texas, 2001.

4. Econolite. Advanced System Controller ASC/3 Programming Manual. Anaheim, California. 2005.

5. Mehalso M. Development of Software-In-The-Loop Simulation with the VISSIM and CORSIM Simulation Models. Thesis. University of Virginia. March 2006.

6. Husch, D. and J. Albeck. Trafficware SYNCHRO 6 User Guide. Trafficware, Albany, California, 2004.

7. National Electronic Manufacture Association. NTCIP 1202: Object Definitions for Actuated Traffic Controller Units. Washington D.C. 2004.

8. Engelbrecht, E., S. Venglar and Z. Tian. Improving Diamond Interchange Operations Using Advanced Controller Features. Research Report 4158-1, Texas Transportation Institute, College Station, Texas, October 2001.

9. Sunkari. S. R., R. J. Engelbrecht, and K. N. Balke. Evaluation of Advanced Coordination Features in Traffic Signal Controller. Report 0-4657-1, Texas Transportation Institute, College Station, Texas, September 2004.

10. Bullock, D., and A. Catarella. A Real-Time Simulation Environment for Evaluating Traffic Signal Systems. In Transportation Research Record: Journal of the Transportation Research Board, No. 1634, TRB, National Research Council, Washington, D.C., 1998, pp. 130-135.

11. Engelbrecht, R. J. Development of a Distributed Hardware-In-The-Loop Simulation System for Transportation Networks. Presented at 78th Annual Meeting of the Transportation Research Board, Washington, D.C., 1999.

12. Engelbrecht, R. J. Using Hardware-in-the-Loop Traffic Simulation to Evaluate Traffic Signal Controller Features. 27th Annual Conference of the IEEE Industrial Electronics Society. 2001.

13. McCain-NIATT. Controller Interface Device (CID II): User Manual Version 2.0. McCain Traffic Supply. May, 2002.

14. Advanced Traffic Analysis Center. ATACid User’s Manual. North Dakota State University, Fargo, North Dakota, 2005.

15. Planung Transport Verkehr AG. VISSIM 4.10 User Manual. March 2005.

16. Orcutt, Jr., F. R. The Traffic Signal Book. Prentice Hall, Englewood Cliffs, NJ, 1993.

17. Venglar, S. P., P. Koonce, and T. Urbanik II. PASSER III-98 Application and User’s Guide. Texas Transportation Institute, Texas A&M University, College Station, Texas, 1998.

18. Devore, J. L. Probability and Statistics 5 for Engineering and the Sciences. Duxbury Thomson Learning, 1999.

19. Park, B. and I. Yun. Evaluation of Stochastic Optimization Methods of Traffic Signal Control Settings for Coordinated Actuated Signal Systems. Presented at 85th Annual Meeting of the Transportation Research Board, Washington, D.C., 2006.