Assessment of the Suitability of Microsimulation as a Tool for the Evaluation of Macroscopically Optimized Traffic Signal Timings Aleksandar Z. Stevanovic 1 and Peter T. Martin 2 Abstract: In practice, traffic signal timings are derived using macroscopic tools that are essentially deterministic. Traffic flows, signal phasing, and street geometry are processed to deliver optimized signal timings. Objective functions strive for efficiency through mini- mizing measures such as delay and journey time. We now have traffic microsimulation tools that model traffic by imitating its stochastic nature. This paper looks at microsimulation as a means of testing optimized signal timings. We assess the suitability of evaluating signal timings optimized macroscopically through microsimulation. We analyze a range of traffic demand and traffic control scenarios. A real-world arterial with 12 signalized intersections serves as a test bed for the experiments. The results show that when macroscopically optimized signal timings are subject to extensive evaluation through microsimulation, their efficiency is shown to be inconsistent. The paper concludes that the traffic microsimulation tools cannot always be relied upon to evaluate macroscopically optimized traffic signal timings because these timings sometimes perform worse, in microsimulation, than the nonoptimized signal timings. DOI: 10.1061/ASCE0733-947X2008134:259 CE Database subject headings: Traffic signals; Simulation; Optimization; Traffic delay; Assessments. Introduction Background Traffic signal control optimization is one of the most important requirements for successful traffic control. Programs such as TRANSYT-7F Hale 2005 and SYNCHRO Husch and Albeck 2003b enable the solution of complex signal timing problems. According to Tarnoff and Ordonez 2004, TRANSYT-7F and SYNCHRO are the most popular traffic control optimization tools among traffic signal practitioners in the United States. The same study reported that 52% of surveyed respondents use manual techniques, such as field observations and fine tuning, during the traffic signal timing process; and that traffic engineers often use microsimulations to check and fine tune signal timings before they apply them in the field. The reasons for using microsimula- tions when fine-tuning signal timings are twofold: simulating the stochastic nature of traffic flow and testing the quality of the optimal signal timings. Further, researchers often use microsimulations to evaluate the effectiveness of various innovative and adaptive traffic control strategies. The effectiveness of these strategies are commonly compared to “the best” signal timing plans generated by the tra- ditional macroscopic optimization tools. However, one can question the suitability of evaluating timings optimized by tradi- tional tools by microsimulation. Should microsimulation tools as- sess macroderived signal timings? The best signal timings from SYNCHRO, for example, may not be identified by CORSIM FHwA 1999 as optimal. The objective of this paper, therefore, is to investigate the suitability of evaluating timings derived macro- scopically TRANSYT-7F and SYNCHRO through microsimu- lation CORSIM, SIMTRAFFIC, and VISSIM. Review of Past Research Investigation of the reliability of macroscopically optimized tim- ings in microscopic environments has drawn some attention. Park et al. 2001 evaluated the reliability of TRANSYT-7F optimiza- tion schemes for both uncongested and congested conditions. They compared TRANSYT-7F’s measures of effectiveness MOE to those resulting from CORSIM’s output using the same TRANSYT-7F timing plans. This is the first explicit approach of assessing macroscopically optimized timings within a microsimu- lation environment. Comparing the performance measures from TRANSYT-7F and CORSIM, they reported high correlations of MOEs for the uncongested condition and low correlations for the congested condition. In both cases, the performance measures for the macroscopic and microscopic models were quite different. They associated these differences to a low fidelity traffic model within TRANSYT-7F. Rouphail et al. 2000 recognized the need for a better optimi- zation tool to evaluate performance measures for microscopic simulations. They developed direct CORSIM optimization, a type of stochastic optimization, using the genetic algorithm GA method for optimization of performance measures. Their findings showed that direct optimization consistently provided better 1 Research Associate, Dept. of Civil and Environmental Engineering, Univ. of Utah, 122 S. Central Campus Dr., Rm. 104, Salt Lake City, UT 84112-0561. E-mail: aleks@trafficlab.utah.edu 2 Associate Professor, Dept. of Civil and Environmental Engineering, Univ. of Utah, 122 S. Central Campus Dr., Rm. 104, Salt Lake City, UT 84112-0561 corresponding author. E-mail: peter@trafficlab. utah.edu Note. Discussion open until July 1, 2008. Separate discussions must be submitted for individual papers. To extend the closing date by one month, a written request must be filed with the ASCE Managing Editor. The manuscript for this paper was submitted for review and possible publication on October 3, 2005; approved on August 15, 2007. This paper is part of the Journal of Transportation Engineering, Vol. 134, No. 2, February 1, 2008. ©ASCE, ISSN 0733-947X/2008/2-59–67/$25.00. JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / FEBRUARY 2008 / 59