Traffic Signal Volume Warrants A Delay Perspective The Manual on Uniform Traffic Introduction The 2009 Manual on Uniform Traffic Control Devices (MUTCD) Control Devices (MUTCD) 1 is widely used to help determine the need for a traffic signal at an intersection. This manual contains eight traffic signal provides nine traffic signal warrants that warrants to aid in determining give threshold conditions under which traffic signal installation may be appropriate. Of these, the three volume war- the need for a traffic signal rants (8-hour, 4-hour, and peak-hour) at an isolated intersection. are commonly used by traffic agencies and engineers to determine the need to However, the theoretical basis enhance the intersection control from a stop control to a signal control. for determining these warrants Vehicular volume warrants for traffic signals have been included in the is unknown. Using control delay MUTCD since its first edition in 1935, and have been modified in subsequent as measure of performance, we editions. 2 4 The volume thresholds currently used for the 8-hour volume warrant were first introduced in the 1961 evaluate three MUTCD warrants version of the MUTCD and remained for three geometric conditions. unchanged. The documented origins of this warrant show that it was based on engineering judgment, economic feasibility, and consensus of practicing traffic engineers as to what volume conditions seemed to result in improved safety and efficiency. 2 5 Similarly, the current 4-hour and peak-hour warrants first went into effect in 1985 and remained unchanged. These warrants were developed based on the 4-hour and peak-hour volume warrant curves then adopted by Texas, the peakhour delay warrant recommended in the NCHRP project (#3-20), and consensus agreement from the practicing traffic engineering community. 5,6 BY BHARGAVA RAMA CHILUKURI, P.E. AND JORGE LAVAL, PH.D. The literature review reveals some concerns about the criteria specified in the warrants. Simpson 7 indicated that current traffic-signal warrants based on traffic volumes and accidents are inflexible, and a warrant instead based on queue would be sensitive to a wide range of intersection-related variables. Oppenlander 8 suggested adjusting minimum traffic volumes in the 8-hour warrant for complete linear agreement on a log-log relationship. Based on modeling studies, Zhu et al. 9 indicated that traffic volumes specified for a 4-hour volume warrant need to be revisited due to several issues such as control delay inconsistencies along the different volume combinations of major and minor streets and among the different lane configurations. The Texas Transportation Institute (TTI) study 10 recommended that a 30 percent volume reduction factor be used on the volume warrants at intersections in the presence of pedestrian trip generators, such as medical facilities, pedestrian transportation facilities, and activity centers. Thus, the literature suggests the need to revisit the threshold volumes specified in the current MUTCD. A theoretical basis for the engineering judgment used to develop these threshold volumes must be provided before any modifications are made to the current MUTCD. Moreover, considering that these warrants have remained unchanged for a long time (8- hour warrant for more than 50 years and 4-hour and peak-hour warrants for more than 25 years), it is only appropriate to evaluate the pertinence of the current threshold volumes with the recently developed models and methodologies available to analyze intersections. Along these lines, this paper evaluates the three volume warrants to identify the shortcomings based on control delay. Methodology The approach proposed in this paper is based on indifference delay curves. These curves represent all pairs of major and minor street volumes at an isolated intersection such that signal control and stop control yield similar delays. We recognize that total control delay alone is not an adequate MOE since 36 ITE JOURNAL / MARCH 2012
equity issues may arise among the competing approaches. To circumvent this problem, delay curves for both the minor street and the intersection as a whole were used as criteria to modify the indifference delay curves. The final modified indifference delay curve takes the lower envelope of these curves, ensuring that the most restrictive criterion is met. The estimation of control delay was performed with the Highway Capacity Software 2010 (HCS 2010), 11 which uses the Highway Capacity Manual (HCM) methodology to analyze intersections. 12 To develop the indifference delay curves for an isolated intersection, various scenarios were analyzed. A range of volumes were used for both major and minor streets under various geometric conditions, and both two-way stop control (TWSC) and signal control scenarios. Figure 1 presents a schematic of the scope of study. Capacity analysis of the intersection was conducted and the delay was recorded for each combination of volume, geometry, and control, as shown in Figure 1. For a two-lane approach, it was assumed that the approach has a shared through-right turn lane and a shared through-left turn lane with equal lane utilization. Since the volume criteria specified in the warrants is based on a total of both major street approaches, a directional distribution of 50 percent was assumed in the analyses. Also, both minor street approaches were assumed to have equal volumes. For the intersection capacity analysis, turning movement percentages of 10 percent left, 80 percent through, and 10 percent right were utilized. For the two-phase signal used in the paper, the cycle length and splits were calculated using the HCM time-budget method. It was found that the optimal cycle length is less than 60 sec. for all the scenarios; therefore, a cycle length of 60 sec. was used in the analyses. Splits were calculated based on critical-lane analysis. However, a minimum split was used to accommodate pedestrian crossing time. Indifference Delay Curve The isolated intersection was evaluated under 468 different scenarios (see Figure Moreover, considering that these warrants have remained unchanged for a long time (8-hour warrant for more than 50 years and 4-hour and peak-hour warrants for more than 25 years), it is only appropriate to evaluate the pertinence of the current threshold volumes with the recently developed models and methodologies available to analyze intersections. Figure 1. Scenarios evaluated in the study. 1) and both the minor street approach control delay and the intersection control delay for each of the scenarios were recorded separately. The results showed that for lower volumes, the control delay under stop control was lower than the control delay under signal control. However, at higher volumes, the delay under stop control increased exponentially and signalization yielded significant delay reduction. It was found that traffic signal is more efficient than stop control when the traffic volumes on the minor streets are higher than 200 vehicles per hour (vph) and/or the major street volumes are 450 vehicles per hour (vph). This means that if one were to install a traffic signal based solely on reduced delay, all the driveways along a major street that has volume more than 450 vph would need a traffic signal, which is practically infeasible. Moreover, installation of a new signal is expensive, increases responsibility to an agency, and comes with several issues such as increased rear-end collisions, increased delay during offpeak hours, and increased concentration of activity in the vicinity. Therefore, the above indifference delay curves need to be modifi ed before evaluating critical volumes in the warrants. ITE JOURNAL / MARCH 2012 37
Figure 2. Indifference delay curves based on both criteria. Table 1. Three Conditions of the Eight-hour warrant. 1 Tolerance Criteria It is reasonable to assume that signalization should decrease the intersection delay by at least 15 sec., a minimum of one alphabet improvement on the level of service of a TWSC intersection. Therefore, it was decided to re-create the indifference delay curves with a tolerance of 15 sec./ veh. This means that the intersection is assumed to operate better under signal control than stop control if the control delay under signal control is at least 15 sec./veh. less than that under TWSC. As a result of this criterion, the indifference delay curves shifted upward, indicating that higher volumes are required on both the major street and minor street to warrant a traffic signal. Cutoff Criteria The need for a traffic signal at an existing unsignalized intersection is typically driven by the excessive delay incurred to the minor street under stop control. Therefore, it was decided to incorporate the control delay for the minor streets into the analysis. The HCM states that the control delay greater than 35 sec./ veh. corresponds to a level of service of E for the unsignalized intersections, so it was determined to develop indifference delay curves with a cutoff control delay of 35 sec./veh. for the minor streets. This means that the intersection will be signalized only if the control delay for the minor street is greater than or equal to 35 sec./veh. As a result of this criterion, the indifference delay curves shifted further upward some times and lower other times. Figure 2 shows the indifference curves developed using the lower envelope of the two criteria: intersection delay tolerance and minor street delay cutoff. Because the traffic signal warrants use smooth and continuous curves, the indifference delay curves were smoothed. These curves are used in the rest of the analysis for the warrant evaluation purposes. Figure 2 shows that the indifference delay curves for the one-lane major and one-lane minor scenario and two-lane major and one-lane minor scenario are very similar. This is because the intersection delay is mainly contributed by the excessive delay on the minor street and therefore did not significantly vary with adding more lanes to the major street. In fact, the delay for the minor street approach sometimes reduced slightly when the number of lanes on the major street increased (for the same combination of major and minor street volumes) possibly due to an increased number of gaps. The indifference delay curve for the two-lane major and two-lane minor scenario was significantly higher than the other two developed in this study because of the increased capacity and increased number of gaps for both major and minor streets. It should be noted that even though the assumptions of 35 sec./veh. delay cutoff and 15 sec./veh. tolerance seem reasonable, the indifference delay curves will be affected if a different set of cutoff and tolerance values are used. Discussion The 8-hour, 4-hour, and peak-hour warrants are evaluated against the indifferent delay curves developed in this paper. Each of these warrants specify threshold hourly volumes to satisfy the warrant, so the data collection time frame does not matter and the indifference delay curves developed in this paper can be used to compare the warrants. 38 ITE JOURNAL / MARCH 2012
Eight-hour volume The 8-hour warrant consists of three independent conditions, 1 Condition A, Condition B, and 80 percent volumes of both Condition A and Condition B (hereafter called Conditions A and B), as shown in Table 1. Any one of the conditions must be satisfied for the warrant to be satisfied. Figure 3 shows the critical volumes specified in Table 1 against the indifference delay curves for three geometric scenarios. Based on Figure 3, it can be noted that the current warrant has lower volume thresholds compared to the indifferent curves. This means that the critical volumes specified in the current 8-hour warrant are lenient in satisfying the warrant for a certain range of volumes (shaded areas in the figure). Also, it should be noted from Figure 3 that the 8-hour warrant does not encompass all the possible ranges of volumes. For one-lane major and one-lane minor and two-lane major and two-lane minor scenarios, the volume combinations overlap for Condition A, Condition B, and Conditions A and B. However, for two-lane major and one-lane minor scenario, the volume combinations for Condition A and combination of Condition A and B overlap, but the volume combination for Condition B does not overlap with either of the conditions. This results in a discontinuity in the warrant. The implications of this discontinuity can be explained using an example. If an intersection has at least 900 vph on the major and 110 vph on the minor streets, then the intersection satisfies Condition B. Similarly, if an intersection has at least 850 vph on the major and 120 vph on the minor streets, then the intersection satisfies Conditions A and B. However, if an intersection has volumes between 850 vph and 900 vph on the major and 110 vph and 120 vph on the minor streets, then the intersection does not satisfy any of the three conditions. This discrepancy arises for a two-lane major and a one-lane minor scenario due to the unique values of the critical volumes specified for this geometry in this warrant. Therefore, it is recommended that the threshold volumes be revised or volume curves be developed to encompass all possible ranges of combinations of major street Figure 3. Critical volumes for eight-hour vehicular volume warrant. Figure 4. Critical volumes for the four-hour vehicular volume warrant. and minor street volumes. Oppenlander 8 suggested replacing the major street and minor street combination with a product of major street and minor street volumes as a criteria to provide a continuous warrant. This is an option, but we believe that this option should be supplemented with criteria specifying minimum threshold volumes for the minor street. Four-hour volume Figure 4 shows the minimum vehicular volume curves for the 4-hour volume warrant and the indifference delay curves. It can be observed that the warrant curve for two-lane major and one-lane minor is significantly higher than that for the onelane major and one-lane minor. However, based on the delay analysis it was found that the delay pattern for a two-lane major and one-lane minor matches closely with one-lane major and one-lane minor. Moreover, the warrant curves seem to be shifted in both x and y directions compared with the corresponding indifference delay curves for all three geometries. The warrant curve for the one-lane major and one-lane minor scenario seems to be shifted by almost 100 vph in the x direction and 50 vph in the y direction, when compared with the corresponding indifference delay curve. This means that ITE JOURNAL / MARCH 2012 39
Figure 5. Critical volumes for peak-hour volume warrant. for a given major (minor) street volume, the corresponding minor (major) street volume on the warrant curve is 50 (100) vph more than the corresponding minor (major) street volume on the indifference delay curve. Similarly, the warrant curve for the two-lane major and one-lane minor scenario seems to be shifted by almost 175 vph in the x direction and 75 vph in the y direction. The warrant curve for the two-lane major and two-lane minor scenario seems to be shifted in x direction by 150 vph and in the y direction by 70 vph. This shows that the values on the warrant curves are consistently higher than the corresponding values on the indifference delay curves for all geometries. Thus, it can be concluded that the 4-hour warrant is relatively strict in satisfying the warrant when evaluated based on the indifference curves developed in this study. Peak-hour volume This warrant has two independent categories. The first category consists of three conditions, all of which need to be satisfied to satisfy the first condition. The three conditions of the first category are: (1) Total stopped time delay experienced by the one minor-street approach equals or exceeds 4 vehicle-hours for a one-lane approach or 5 vehicle-hours for a two-lane approach; (2) The volume on the same minor-street approach equals or exceeds 100 vph for one lane of traffic or 150 vehicles per hour for two lanes; and (3) The total entering volume equals or exceeds 650 vph for intersections with three approaches or 800 vph for intersections with four or more approaches. Figure 5 shows the peak-hour warrant and indifference curves. As shown in Figure 5, three Category 1 curves were developed based on the condition that the total stopped delay experienced by the one minor-street approach equals or exceeds 4 vehicle-hours for a one-lane approach or 5 vehicle-hours for a two-lane approach. The other two conditions were incorporated in Figure 5 by drawing two regions, one with boundaries at 100 vph minor street volume and 600 vph major street volume, and the other region with boundaries at 150 vph minor street volume and 500 vph major street volume. Thus, the shaded regions in Figure 5 show the volume combinations that satisfy the first condition of the peak-hour warrant. Figure 5 shows that the warrant curves are consistently higher than the Category 1 curves developed for this warrant, which in turn are higher than indifference curves. When compared with the indifference curves, the warrant curve for the one-lane major and one-lane minor scenario in the x-direction seems to be shifted by an average of 375 vph and in the y-direction by an average of 175 vph. However, the Category 1 curves seem to be shifted by only an average of 120 vph in the x-direction and 50 vph in the y- direction. Similarly, the warrant curve for the two-lane major and one-lane minor scenario and two-lane major and two-lane minor scenario seems to be shifted by an average of 600 vph and 250 vph in x and y directions, respectively. But, the Category 1 curves seem to be shifted by only an average of 200 vph in the x-direction and 50 vph in the y-direction. Thus, it appears that there is lack of consistency in the Category 1 delay curves developed using the first category and the critical volume curves of the second category with the latter curves consistently higher than the former. Therefore, it can be concluded that the first category of the peak-hour warrant is lenient in satisfying the warrant compared with the second category. This means that the threshold volumes for Category 1 are lower than that of Category 2. Similar to the 4-hour volume warrant, the curves for this warrant are consistently significantly higher (sometimes as much as the ordinate itself) than the corresponding values on the indifference delay curve. Thus, it can be said that this warrant is critical in its evaluation and does not easily favor installation of a traffic signal. This is reasonable since the signal installation based on peak-hour traffic volumes need strong justification. Conclusions This paper has presented results from an evaluation of the critical volumes of the 8-hour, 4-hour, and peak-hour warrants based on the control delay. TWSC and signal control operation of an isolated intersection for three-lane configurations under various combinations of major and minor-street volumes was evaluated using HCS 2010. The results indicated that: in satisfying the warrant for some volume combinations and requires modification of the existing warrant or development of threshold volume curves similar to the 4-hour and peak-hour warrants to encompass all the possible ranges of volumes. rant curves set the bar for the minimum volume higher than required based on indifference delay curves to satisfy the warrant. Moreover, unlike what is found during this study, the threshold volume curves for two- 40 ITE JOURNAL / MARCH 2012
lane major and one-lane minor are significantly higher than the curve for the one-lane major and one-lane minor. Therefore, there is a need to revisit the critical volumes and delay curves for two-lane major and onelane minor and one-lane major and one-lane minor scenarios. volume warrant are not consistent in evaluating the need for a traffic signal, with the first category more lenient than the latter in satisfying the warrant. The above findings were based on the tions). Although not evaluated in this study, the results from the scenarios for to be similar. shortcomings of the existing warrants warrants to aid agencies and engineers with signalization decisions. However, tive to install a traffic signal when one or more of the traffic signal warrants are satisfied. Agencies and engineers should safety and efficiency before selecting the References 1. Federal Highway Administration. Manual on Uniform Traffic Control Devices, 2009 Edition. - MUTCD: Part 1 - Early Standards for Traffic Control Devices. ITE Journal, Vol. 62, No. 7 MUTCD: Part 2 - The Early Editions of the MUTCD. ITE Journal War II. ITE Journal - Warrants Result in More Signals? ITE Journal, ITE Journal for Traffic Signals. ITE Journal Simulation Evaluation of Four-Hour-Volume Traffic Signal Warrant. CD-ROM. Washington, DC: Academies, 2007. ter Accommodate Pedestrians and Cyclists. College 11. Highway Capacity Software 2010 User Manual. McTrans Center, University of Florida, 2010. 12. Highway Capacity Manual, 2010 Edition BHARGAVA RAMA CHILUKURI, P.E., is a doctoral student at the Georgia Institute of Technology (GaTech). He worked as a consulting traffic engineer prior to joining the Ph.D. program at GaTech. His primary interests include traffic control, traffic flow theory, and simulation. He is a member of ITE. JORGE LAVAL, Ph.D., is an assistant professor at the School of Civil and Environmental Engineering at Georgia Institute of Technology. He obtained a Ph.D. in civil engineering from the University of California at Berkeley. His research interests include traffic flow theory, numerical solution methods and simulation of traffic flow models, queuing theory in transportation, and dynamic congestion pricing. ITE JOURNAL / MARCH 2012 41