Problem 4: Actuated Control

Printable Version

In our previous analysis of the proposed signal at U.S. 95/Styner-Lauder Avenue, we have considered the signal to be fixed time. In reality, most new traffic signals that are installed today are actuated, responding to changing traffic demands during the day. The HCM provides a mechanism to estimate green times for an actuated controller, based on the relative traffic volumes on each intersection approach, and to estimate the delay and level of service that would result from traffic actuated timing plans.

Traffic-actuated control will generally accommodate a given volume of traffic with lower delays than pretimed control, because of its ability to adapt to demand variations.

The effects of actuated control are reflected in the HCM analysis procedure in two ways:

1. The equivalent cycle length and green times produced by the Appendix B procedure are typically lower than their pretimed counterparts, yielding a lower computed value of uniform delay.

2. As illustrated in Exhibit 16-13 of the HCM 2000, the incremental delay factor, K, is given a lower value for traffic-actuated control, depending on the unit extension time and the v/c ratio. Lower K values also produce lower delay estimates.

As we discovered in sub-problem 2d, we must also be aware of whether the intersection is part of a coordinated system of intersections, where a fixed cycle length must be used. The constraint of a fixed cycle length can have a significant effect on the operating characteristics of an actuated controller. We will explore this further in sub-problems 4c and 4d.

Problem 4: Actuated Control

The traffic-actuated timing procedure requires three control parameters:

This problem illustrates some of the important elements of performing an analysis of a signalized intersection operating under actuated control by addressing the following issues as they relate to the proposed signal at the U.S. 95/Styner-Lauder Avenue intersection:

Sub-problem 4a: Estimating phase times for actuated signal control

Sub-problem 4b: Effects of Unit Extension on intersection operating characteristics

Sub-problem 4c: Coordinating a system of actuated controllers

Sub-problem 4d: Coordinated operation of an actuated controller with left-turn protection

Sub-problem 4a: Pretimed Control vs. Actuated Control

Whenever signalized control is considered for an intersection, it is usually worthwhile to spend at least some time thinking about and evaluating the different types of signal control that are available for implementation. Signalized intersection analyses described earlier in this case study have assumed the signal will operate in a pre-timed mode, but most new signals today also have the ability to operate in an actuated mode. Typically, the kinds of questions that the analyst would want to answer before deciding between pre-timed and actuated control would include the following:

Discussion:
Take a few minutes to consider these questions. When you are ready to continue, click continue below to proceed.

Sub-problem 4a: Pretimed Control vs. Actuated Control

In this exercise, we will convert the existing pretimed example from sub-problem 1b (see Exhibit 1-11) to a traffic actuated control to illustrate the differences between the timing and performance measures from the HCM computations. Using Appendix B, Chapter 16 of the HCM with the control parameters set as discussed above, we can estimate the average green times if the signal controller is actuated. Exhibit 1-38 shows the results of this estimation.

Note that the estimated average phase times for traffic-actuated control are shorter than pretimed control. Perhaps even more surprising is the fact that the estimated average phase time for the east-west movement is actually below the minimum phase time, which has been set to reflect the crossing time requirements of pedestrians. This does not reflect an unsafe situation because these are equivalent phase times for purposes of computing delay. They reflect the fact that, because of low vehicular volumes, the east-west phases will sometimes not be displayed. When the east-west phase is displayed, it will always be at least 20 seconds long (i.e., the minimum phase time). When it is not displayed, it will obviously be zero seconds long. In this particular case, the average of these various phase lengths is expected to be about 15 seconds under actuated control. The very short cycle (32 sec) is also an equivalent value, and should not be viewed in the pretimed sense.

Because the east-west phase is sometimes skipped, the average lost time associated with this phase will also be less, during the analysis time period, than would be the case if it were pretimed. Specifically, the average lost time for the east-west phase is equal to the average lost time that would occur if the phase were pretimed, multiplied by the proportion of cycles during the analysis period when the east-west phase is NOT skipped.

The green times shown in the table above for actuated control will now be used in the estimation of the control delay and level of service for the intersection.

Sub-problem 4a: Pretimed Control vs. Actuated Control

Exhibit 1-39 summarizes the level of service and delay at each approach for a pretimed and actuated control.

Note that traffic-actuated control has produced a substantial reduction in delay for east-west traffic accompanied by a very slight increase for north-south traffic. Note also that the delays are now more or less equal on all approaches, whereas the distribution of times under the pretimed design favored the north-south (arterial) movements at the expense of the east-west (cross street) movements. In apportioning the times, the controller has clearly ignored the much longer maximum green times for the north-south phase. In other words, the north-south phase would never reach the maximum time, because the unit extension settings would cause the phase to terminate whenever the queue of vehicles on the approach has been serviced.

While the timing plan presented above produces lower delays than the previous pretimed plan, the local agency may still wish to favor the north/south movements because of the heavier movement on U.S. 95. The results of this exercise suggest that increasing the maximum green time would not accomplish this goal. The maximum green setting will have a definite effect on the green time distribution under high traffic volumes; but when demand is low, the minimum green time is the only parameter that will force a green time distribution that is not in keeping with the distribution of demand.

Sub-problem 4b: Effect of Unit Extension

In this exercise, we will consider the effect of the unit extension variable on the overall operating characteristics of an intersection controlled in an actuated mode.

Consider again the Styner-Lauder/U.S. 95 intersection that is at the center of this case study. If you were going to evaluate the operational effects of an actuated signal controller at this intersection, it would be very important to select an appropriate unit extension for the purposes of the analysis.

This sub-problem will help you address the following issues, each of which is important to understand as you settle upon an appropriate value for the unit extension variable:

Discussion:
Take a few minutes to consider these questions.  When you are ready to continue, click continue below to proceed.

Sub-problem 4b: Effect of Unit Extension

Remember that, at the outset of this problem, we assumed that the unit extension times would be set to 3 seconds for single lane operation and 2 seconds for multiple-lane operation. These are commonly-used values for the unit extension and are consistent with an expected saturation flow rate of about 1,900 vehicles per lane per hour of green. Under these conditions, the 2/3-second unit extension values can be expected to have the effect of extending the green phase beyond its minimum length only so long as vehicle spacing is not much greater than one would expect under saturation flow rate conditions.

As a further illustration of the effect of the traffic actuated settings, let us examine what would happen to the operation of the Styner-Lauder/U.S. 95 intersection if we increased the unit extension times for all phases to 5 seconds each. This would mean that a phase would not terminate until a five-second gap was observed between vehicles on any approach. In other words, the controller would be waiting for stragglers instead of passing control to the next phase after the queue of vehicles has been serviced.

Exhibit 1-40 summarizes the results of this exercise. Using the same Appendix B procedures that we used in sub-problem 4a, we can see that the average phase times have increased from 15 and 16 seconds to 19 and 24 seconds, respectively. The cycle length has increased from 31 seconds to 43 seconds. It is also interesting to note that the apportionment of green time now favors the arterial movement to a greater extent, because the longer north-south maximum green time has allowed this phase to be extended more than the east-west phase. This is also evident in the proportionately greater increase in the east-west delays that are now being predicted.

Based on the information shown in Exhibit 1-40, is it better to use the originally-estimated unit extension values of 2-3 seconds, or the 5-second unit extension value assumed for this sub-problem? The answer depends on what the analyst is trying to accomplish and the environment in which the signalized intersection is located. In the case of the Styner-Lauder/U.S. 95 intersection, it might be appropriate to use the longer unit extensions on the northbound and southbound approaches of the state highway in order to favor and benefit through traffic. The shorter unit extensions might be appropriate to use on the eastbound and westbound side street approaches in order to minimize overall intersection delay and cycle length. An alternative approach, however, could also be used with equal validity.

Sub-problem 4c: Coordinated Operation With Actuated Control

The Styner-Lauder/U.S. 95 intersection is located within a section of U.S. 95 that has the potential to be operated as a coordinated system of signalized intersections. If the decision is made to signalized the Styner-Lauder/U.S. 95 intersection, it will need to be designed and operated in a way that supports and enhances any coordinated signal system that may ultimately be implemented.

The matter of arterial signal system coordination is complicated somewhat when one or more of the system signals operates under actuated control. This sub-problem will introduce you to the key issues to consider to appropriately evaluate the effects that actuated control of the Styner-Lauder/U.S. 95 intersection will have on the overall arterial signal system. Specifically, the following are key considerations that would need to be taken into account:

Discussion:
Take a few minutes to consider these questions.  When you are ready to continue, click continue below to proceed.

Sub-problem 4c: Coordinated Operation With Actuated Control

In the first two sub-problems involving traffic-actuated control, we have assumed that the controller at the Styner-Lauder/U.S. 95 intersection will operate in an isolated mode, independent of any other intersections on U.S. 95. It is, however, more likely that the local agency would want to establish a coordinated system of actuated controllers on this route to take advantage of the improvements that could be derived from the progressive movement of traffic on U.S. 95. This would require a full timing plan design, including cycle length, phase split times, and offsets. Coordinated arterial system timing design is beyond the scope of the HCM arterial analysis procedure. It would therefore be necessary to use one of the several available arterial signal timing software products for this purpose.

We will consider this example in the context of a coordinated system operating on a 90 second cycle as was introduced in Problem 2d (coordination effects of a new signal). Coordinated traffic-actuated control systems are generally modeled as pretimed systems for purposes of timing plan design. This is because actuated controllers within a coordinated environment are forced to operate under a constant cycle length. As a coarse approximation of their internal logic, traffic actuated controllers are usually represented as devices that will assign enough time to the cross street to maintain a reasonable degree of saturation, with the remaining time given to the arterial movements. 

If we follow that logic here, we must first determine the maximum amount of green time that the side street approaches can be expected to require. The maximum green time must be at least long enough to accommodate pedestrian crossing time requirements, which we have previously determined to be 20 seconds. If approaching traffic volumes are very high on the side street, then the maximum green time may need to be even longer to assure vehicle needs are also being met.

Applying the HCM procedure under an initial presumption that pedestrian needs will control side street maximum green times (i.e., a maximum green+yellow+all-red time of 20 seconds in this case), we find that the v/c ratio for the most critical movement (WB through and right) is 71% for a 90-second cycle. Thus, vehicle needs are also being adequately met, so we can conclude that a timing apportionment of 70 seconds to north-south traffic and 20 seconds to east west traffic would be an appropriate estimate of the average green+yellow+all-red times for a coordinated traffic-actuated system. This apportionment would be implemented by coordination hardware that would impose a background cycle of 90 seconds. 

The presumption of a 70-sec/20-sec split of time is adequate for conducting an operational analysis. In reality, however, the east-west phase would be traffic-actuated with a maximum phase time somewhat longer than the 20 second pedestrian requirement to provide for the occasional cycles that inevitably occur with heavier-than-normal cross street demand.

Sub-problem 4d: Coordinated Operation With Left-Turn Protection

An issue that commonly arises when actuated controllers are used within a coordinated system is whether or not to provide protected phasing to the major street left-turn movements. This issue is appropriate to address for the proposed new signal at the Styner-Lauder intersection.

In sub-problem 4d, we will demonstrate how to address this question at the U.S. 95/Styner-Lauder Avenue intersection.  Here are some issues to consider as you prepare to conduct this analysis:

Discussion:
Take a few minutes to consider these questions.  Click continue when you are ready to proceed.

Sub-problem 4d: Coordinated Operation With Left-Turn Protection

The left turning volumes are very light at this intersection, and all of the previous problems have assumed a simple two-phase operation. On the other hand, some agencies prefer to provide left turn protection even when left-turning volumes are low with the idea that the left-turn phases will be skipped if there is no demand during a given cycle.

We will therefore consider the addition of protected phases for the arterial (north-south) left turns, retaining a 90-second cycle. We will keep the same settings for the cross street through phases and add a phase for the northbound and southbound left turns with a 12-second maximum phase time consisting of 8 seconds green and 4 seconds yellow plus all red. The minimum green will be 8 seconds, so the full 12-second phase time will be displayed on cycles during which any left-turning vehicles arrive on the approach served by the left-turn phase.

In summary, a pretimed representation of the proposed signal timing would be 12 seconds for NB and SB left turns, 58 seconds for NB and SB through, and 20 seconds for all EB and WB movements. Since the left turns are very light, the phase will be skipped on some cycles and the unused time will be added to the through movements. The objective of this exercise is to determine how much green time on an average cycle will be added to the arterial through traffic phase. The amount of added green time must be based on the probability of one or more left-turn arrivals during any given cycle.

Sub-problem 4d: Coordinated Operation With Left-Turn Protection

In order to estimate the proportion of skipped left-turn phases during the analysis time period, we need to estimate the probability that no left-turn vehicles will arrive during a cycle. For this purpose, it is common to assume that the left-turning vehicles will arrive at the intersection according to a Poisson distribution (a Poisson distribution reflects random arrivals). Assuming a Poisson distribution of arrivals, the probability of zero arrivals on any cycle may be computed as:

where a represents the average number of arrivals per cycle, based on the hourly volume.

Exhibit 1-41 shows the results of these computations. If the average length of a left-turn phase is less than 12 seconds, the difference is added to the phase time for the opposing through movement.

As an example, we will examine the average northbound phase times. The average cycle length at this intersection is 90-seconds, or 40 cycles per hour. The northbound left-turn volume is thirty-one vehicles per hour, which produces an average of 0.78 vehicles per cycle (31 vehicles per hour/40 cycles per hour). From the above equation we estimate zero arrivals on 46-percent of the cycles (e^-0.78), giving us an average phase time of 6.5 seconds (12 sec * (1-0.46)). Because this is less than the assumed maximum of 12-seconds green, we may add the difference in time (5.5 seconds) to the conflicting southbound through phase, producing 63.5 seconds of green (58 sec+5.5 sec).

The delay computations may use the average phase times for the through and left-turn movements determined from the table shown above.

Problem 4: Analysis

In Problem 4, we explored the effects of traffic-actuated control on the performance of the U.S. 95/Styner-Lauder Intersection. Traffic-actuated control is a function of the detection system and the signal timing active within the signal controller at the intersection. In detailed operations analysis, the use of detector timing presents an additional level of complexity for our analyses and thus, may more closely reflect field conditions.

In sub-problem 4a, we considered the effect of actuated control that revisits our assumption regarding cycle length and phase time. This analysis assumes fully-actuated control typical of an isolated operations.

In sub-problem 4b, we examined the effect of unit extension on phase length and its corresponding effect on cycle length at a fully-actuated intersection. We learned that as the unit extension (commonly referred to as passage gap in signal controller manuals) increases, the phase length and cycle length increase.

In sub-problem 4c, we revealed the implications of a coordinated operations on an actuated signal. Specifically, the introduction of a fixed cycle length may result in increased delay to the side street but improved operations and progression along the arterial.

Finally, in sub-problem 4d, the effect of left-turn protection is considered. In this analysis, we learn how to estimate the percentage of cycles that will result in left-turn actuation.

Problem 4: Analysis

The traffic-actuated timing procedure requires three control parameters. 

1. Minimum phase times, considering driver expectancy and pedestrian requirements. The minimum times from the pretimed example (Sub-problem 1b) were used here.
2. Maximum phase times to assure a reasonable distribution of green time on cycles with heavy demand. The literature contains a variety of techniques for setting maximum green times. For purposes of this example, the maximum green times were set at 20% longer than the pretimed values.
3. Unit extension times to determine the length of the gap in arriving traffic at which a phase will terminate. Most traffic models, including the HCM, will yield lower delay estimates with lower unit extension times. As a practical constraint, however, the unit extension must be slightly longer than the maximum expected gap between vehicles departing from a queue, otherwise premature phase terminations will occur. For purposes of this example, the unit extension times were set to 3 seconds for single-lane operation and 2 seconds for multiple-lane operation.

[ Back ] [ Continue ] to Problem 5

Problem 4: Discussion

The decision to signalize an intersection brings with it many additional issues that also need to be addressed. Should the signal operate in a pre-timed, semi-actuated, or fully-actuated mode? Should it be coordinated with adjacent signalized intersections or should it be designed to operate as an isolated intersection? How should the left-turns be handled on each approach -- should they be permitted, protected, or both? What should be the phasing sequence? What are the appropriate settings for the various signal timing parameters, including in the case of actuated control the factors of minimum green time, maximum green time, and unit extension? The answer to each of these questions is affected by the answers to all the others, and so it is typical that multiple scenarios will need to be investigated before deciding upon a particular implementation plan.

A fairly comprehensive method is presented in Chapter 16, Appendix B of the HCM2000 for reasonably approximating traffic-actuated operation. This method takes explicit account of many phasing variables, detector design parameters, and controller settings. Its application in Problem 4 provided considerable insight into the relative merits of pre-timed versus actuated signal control at the U.S. 95/Styner-Lauder intersection.

The next problem steps away from the locale of the U.S. 95/Styner-Lauder intersection and recognizes that the U.S. 95 facility includes both urban and rural segments. In particular, Problem 5 focuses on a rural segment of U.S. 95 located south of Moscow, and identifies a type of two-lane highway analysis that is not explicitly addressed within the current version of the HCM.