Problem 3: Analysis of oversaturated conditions - U.S. 95/Styner-Lauder Avenue Intersection

Printable Version

In the two previous problems, we considered traffic volumes only for the afternoon peak period. This allowed us to focus on several important issues relating to the application of the signalized and unsignalized intersection procedures of the Highway Capacity Manual. In problem 3, we recognize that there are other time periods and traffic conditions that also need to be considered. Just as we learned more about the signalization decision when we considered the U.S. 95 corridor and the related system issues beyond the single intersection of U.S. 95/Styner-Lauder Avenue, we will see that there is more to learn about the variation in traffic volumes, providing us with an even better perspective on the signalization decision.

The University of Idaho has a number of special events during the year that attract large crowds to its sporting arenas and performing arts venues. During these periods, demand often exceeds capacity along the U.S. 95 corridor in ways that do not happen during normal weekday periods. What tools are needed to assess the operation of the U.S. 95 corridor in general and the intersection of U.S. 95/Styner-Lauder Avenue in particular? And what analytical issues must be faced when demand exceeds capacity at a traffic facility? In sub-problem 3a, we will use the HCM methodologies to determine the level of service of the intersection under these high volume conditions. In sub-problem 3b, we will use a microscopic simulation model to assess these conditions.  Finally, in sub-problem 3c, we will use another method, critical movement analysis, to determine whether the intersection can accommodate the demand during these high volume events. These sub-problems will address the following issues:

Take a few minutes to think about each of these questions before proceeding. Click on continue when you are ready.

Sub-problem 3a: Oversaturated Intersection Analysis

Step 1. Setup

The University of Idaho has periodic special events that attract large crowds to sporting arenas and performing arts venues. During these time periods, demand often exceeds capacity along the U.S. 95 corridor. What tools are needed to assess the operation of the U.S. 95 corridor in general and the intersection of U.S. 95/Styner-Lauder Avenue in particular? And, what analytical issues must be faced when demand exceeds capacity at a traffic facility?

In sub-problem 3a, we will consider the conditions that occur when traffic is leaving a football game at the University of Idaho. Demand is high for about an hour after the conclusion of the game, and the U.S. 95 corridor experiences a high level of congestion during this period.

Consider these questions:

Discussion:
Take a few minutes to consider these questions.  When you are ready, continue to the next page.

Sub-problem 3a: Oversaturated Intersection Analysis

Now let's review each of these questions, and how they are important to this analysis.

What is the difference between volume and demand, and why is it important to distinguish these two terms? Most field studies at intersections include the measurement of volume. Volume, or sometimes service volume, is the traffic flow at the point where vehicles are entering the intersection just past the stop-line. Demand (or sometimes demand volume), by contrast, is the traffic flow desiring to enter the intersection. When we measure demand in the field, we must be at a point upstream of any queues that form at the intersection. When we conduct analyses using the Highway Capacity Manual, we must always use demand volumes and not service volumes. If demand is less than capacity, then demand volume equals service volume, and we can collect stop-line counts for use with the HCM. However, if demand exceeds capacity (as evidenced by continuing queues that don't dissipate at the end of each cycle), then we must collect traffic flow data upstream of the intersection to account for all vehicles desiring to use the intersection during a given time period. Special care must be taken when collecting turning movements when queues extend upstream of the intersection as it is sometimes difficult to see the final direction that a given vehicle follows when the observation is conducted upstream. Videotaping might be considered as an aid in the data collection process when these conditions occur.

Can the intersection operate at level of service F when demand is less than capacity? Level of service for a signalized intersection is defined by average control delay; and while it is somewhat dependent on capacity, it is often more dependent on other factors such as arrival type. So it is possible for an intersection to operate at level of service F (when delay exceeds 80 seconds per vehicle) while demand is less than capacity. The reverse is also true: lane groups, approaches, and intersections can sometimes be found to operate at levels-of-service better than F, even while the computed v/c ratio is greater than 1.0, especially in situations where the cycle length is short and/or progression is very good. The point to remember from all this is that LOS is not, by itself, a sufficient measure of the operating status of the lane group, approach, or intersection: other factors like v/c ratio, queue length, and cycle length must also be considered when forming an overall judgment.

Sub-problem 3a: Oversaturated Intersection Analysis

What is the appropriate value of the duration of analysis when demand exceeds capacity? The default value of the duration of an operations analysis is 15 minutes, or 0.25 hours. This value should be used for most analyses. However, when demand exceeds capacity for a 15-minute period, it may be necessary to expand the analysis period to ensure that all demand can be accommodated. Another alternative to be explored in this sub-problem is to conduct a multiple time period analysis.

When should multiple time periods be considered?  If demand exceeds capacity for a given 15-minute period, the excess demand cannot be served during this period. In reality, this demand is shifted to the next 15-minute period. The analyst has a choice of considering a longer duration of analysis (see above), or conducting a multiple time period analysis. In this latter case, we would need to shift the excess (or unserved) demand from the first time period (when demand exceeds capacity) to the next 15-minute period. In addition to considering this demand shift, we must also take account of the initial queue to make sure that our estimates of control delay are realistic.

We'll now consider how to setup this problem.

Sub-problem 3a:  Oversaturated Intersection Analysis

Data were collected for a typical post-football game period at the U.S. 95/Styner Avenue-Lauder Avenue intersection. Exhibit 1-27 shows the demand (flow rates in veh/hr) for the four 15-minute periods immediately following the football game:

How will the intersection perform, under both signal control and stop sign control, for these demand conditions?  How should we proceed with this analysis?

We will first consider the operational performance of the intersection operating under today's control conditions (two-way stop-control) to confirm what we suspect: the high volumes on the main street (U.S. 95) will cause long delays on the side streets (Styner and Lauder Avenues). If this is found to be the case, it may further support a recommendation that signalization is a valid option to handle current and future traffic volumes. Even so, it is important to note that event-related conditions would not, by themselves, be sufficient justification for installing a traffic signal at this intersection.

We will then consider the operation of the intersection under signal control. Here, we will break down the analysis into four steps, one for each of the four time periods, in sequence, starting with the 4:00 p.m. to 4:15 p.m. time period. 

Sub-problem 3a: Oversaturated Intersection Analysis

Step 2. Results

Exhibit 1-28 shows the results of evaluating the performance of the existing U.S. 95/Styner-Lauder Avenue intersection when the traffic volumes are at the peak level after a football game. As we would expect, the delays and level of service estimated by the TWSC intersection capacity model are high, almost unreasonably high. The heavy volumes on the main street (U.S. 95) make it extremely difficult for any minor street vehicle to enter the intersection. Capacities are estimated to be at or near zero. We should not be too surprised to see that the model is unable to estimate the delays. For example, for the through and right-turn movements on both minor street approaches, the volume/capacity ratios are 68.33 (westbound) and 75.00 (eastbound). Clearly the intersection can be expected to perform quite poorly during these high volume periods.

How will the intersection perform under signal control?

Sub-problem 3a: Oversaturated Intersection Analysis

We now consider the performance of the intersection for each of the four 15-minute time periods with the proposed signal control. The results for the first 15-minute time period (4:00 pm to 4:15 pm) are shown in Exhibit 1-29:

For this first time period, while there is some delay on the side streets, the intersection operates at acceptable levels of service. Continue to the next page to see the results of the analysis for the second time period.

Sub-problem 3a: Oversaturated Intersection Analysis

The results for the second 15-minute time period (4:15 pm to 4:30 pm) are shown in Exhibit 1-30:

The increase in traffic flow rates during this second time period results in a much poorer performance at the intersection. The major thing to point out from Exhibit 1-30 is that the volume/capacity ratio for three lane groups exceeds 1.0.

What is the practical implication of this result? The excess demand for each of these lane groups (that is, the difference between the lane group capacity and the adjusted flow rate) is transferred to the next time period (4:30 pm to 4:45 pm). The excess demand is shown in Exhibit 1-31. Note that since the demand is shown first in vehicles/hour that we also show the actual number of vehicles that would be transferred to the next time period.

Sub-problem 3a: Oversaturated Intersection Analysis

Exhibit 1-32 shows both the initial and revised demands for time period 3 (4:30 - 4:45 pm):

For this analysis, we will use the initial demand. In addition, we will use the excess demand as the initial queue, or unmet demand, that is one of the required parameters in the computation of control delay in the HCM Chapter 16 procedure. The results of this analysis, for time period 3 (4:30 - 4:45 pm), are shown in Exhibit 1-33:

The unserved demand that was present at the end of time period 2 (4:15 - 4:30 pm) was successfully served during time period 3 (4:30 - 4:45 pm). The delay for the NB LT movement is still quite high, but all volume/capacity ratios are less than 1.0. Therefore, the analysis of the effects of oversaturation can be considered to be complete.

It is worth noting that the control delay reported in Exhibit 1-33 includes the additional delay experienced by vehicles in the residual queue that was transferred to time period 3 from time period 2. The HCM procedure accounts for the carryover of excess demand through the third (d3) term of the delay equation.

We  will still want to analyze the last 15-minute time period in order to determine the average control delay for the hour, and this is done on the next page.

Sub-problem 3a: Oversaturated Intersection Analysis

Exhibit 1-34 shows the results of the analysis for time period 4 (4:45 - 5:00 pm):

The results for the fourth time period are much the same as for the third time period. All volume/capacity ratios are less than 1.0, so all demand is served during this 15-minute period. As before, delays are high for the NB LT movement.

Sub-Problem 3b: Using a Microscopic Simulation Model

While the HCM model for signalized intersections can be used to evaluate the performance of the U.S. 95/Styner Avenue-Lauder Avenue as we showed in sub-problem 3a, we should also consider the use of a microscopic simulation model. Why? The HCM model is only an approximate solution for conditions when demand exceeds capacity. Since it considers only macroscopic conditions, the HCM cannot provide the same level of accuracy as can a microscopic simulation model for high demand periods or when the flows from one intersection interact with the flows from an adjacent intersection.

There are several microscopic simulation models that can be used for this problem.  The application of microscopic simulation is covered in Chapter 34 of the HCM 2000.  You are encouraged to review this chapter to learn more about some of the advantages and disadvantages of using microscopic simulation models.  Here, we will use one such model, the CORSIM model developed by the Federal Highway Administration. It should be emphasized that the CORSIM model is being used here for illustrative purposes, and other microscopic simulation models are available that would work equally well.

Microscopic simulation models consider much more detail than macroscopic models such as the HCM. Individual vehicle interactions and the detailed operation of traffic controllers are two of the more important features of these microscopic models. This more detailed treatment of traffic flow and controller operations allows the models to consider more directly the oversaturated conditions that we found in sub-problem 3a. In addition, we can input the traffic flow data for all four time periods for this problem at once and not have to run several analyses as we did in sub-problem 3a. The models are also stochastic in nature, which means they explicitly account for the probabilistic nature of traffic flow and driver behavior. But there is a cost for this additional detail: more input data, more time required to ensure that the model is calibrated for local conditions, and multiple simulations in order to assure that statistically valid results are obtained.

Sub-Problem 3b: Using a Microscopic Simulation Model

CORSIM produces a very detailed output, covering several measures of delay and travel time, as well as other measures such as vehicle emissions and fuel consumption. We will consider the average control delay measure, similar to the measure produced by the HCM signal model.

The results from the CORSIM simulation model are shown in Exhibit 1-35. We've also included the results from the HCM analysis for easy comparison between these two models.

Exhibit 1-35 shows that there are some significant differences between the estimates of average control delay produced by the two models, HCM and CORSIM. While we have no way of verifying the quality of either estimate, we can say that CORSIM and other similar microscopic simulation models have the potential to produce more realistic results. Why? Because the microscopic simulation models are better at accounting for the microscopic interactions between vehicles, they are more likely to better represent conditions of oversaturation than the macroscopic approach taken by the HCM. But this is true only if the model is correctly coded, calibrated, and applied (including multiple runs to account for the stochastic nature of the simulation). Our major advice is this: if you are dealing with conditions in which demand exceeds capacity, queues from one intersection interacting with an adjacent intersection, intersections that are closely spaced, or multiple time periods, you should consider the use of a microscopic simulation model to produce estimates of control delay. Just like the HCM procedures, use of any microscopic model requires special care and expertise to assure that the accuracy of the results is commensurate with user expectations.

Sub-problem 3c: Critical Movement Analysis

In sub-problem 3b, we saw how we can use a microscopic simulation model to address conditions in which demand exceeds capacity. The model produces detailed statistics on the performance of the U.S. 95/Styner-Lauder Avenue intersection under these conditions. We'll now see in this sub-problem that we can use a much simpler approach, critical movement analysis, to determine whether the demand will exceed capacity for this intersection during the high volume conditions present after football games at the University of Idaho.

Discussion:
Take a few minutes to consider these issues before we proceeding with this analysis.

Sub-problem 3c: Critical movement analysis

What is critical movement analysis? Critical movement analysis is a method to determine whether the projected volumes at a signalized intersection will be under, near, or over the intersection's capacity to accommodate them. The method is fully documented as a planning-level procedure in Transportation Research Circular 212. The method considers each of the four conflicting movement pairs at the intersection (for example, the NB LT and the SB TH movements). The critical movements for each intersection phase (for example, the maximum of either the NB LT/SB TH movement and the SB LT/NB TH movement) are summed. This sum is compared with the following standards:

What data are needed to conduct critical movement analysis? The approach volumes, the number of lanes, and the lane configuration on each approach are the data needed to conduct a critical movement analysis.

What outputs are produced by critical movement analysis? Critical movement analysis produces only an assessment of the intersection's sufficiency to accommodate the projected volumes. It does NOT provide estimates of delay, LOS, or queue lengths.

Are the results from critical movement analysis any more or less valid than the results produced by the HCM or by microscopic simulation models? The results of a critical movement analysis help to determine whether the intersection will operate under, near, or over capacity. It is more of an approximation than either the HCM or other models, but in many cases the results from a critical movement analysis are sufficient to answer the question at hand.

Sub-problem 3c: Critical Movement Analysis

Why is there virtually no difference between estimated delay on the eastbound and westbound approaches to the intersection? The volume to capacity ratio of the intersection with a traffic signal installed is 0.35. Essentially, the intersection is operating well below capacity as you would expect because a signal was installed. The addition of heavy vehicles will reduce the saturation flow rate at the intersection. The presence of heavy vehicles may also have an effect other factors such as lost time (this will be discussed later on this page). Specifically for this intersection, the impact of 25% heavy vehicles on the traffic stream is to reduce the saturation flow rate by 20%, (the f_HV for 25% trucks is 0.80). This is applied within the signalized intersection methodology and results in a new v/c ratio at the intersection of 0.43, which is approximately 20% higher than the existing scenario.

What is the effect of grade and heavy vehicles? Similar to the effect heavy vehicles have on the saturation flow rate, grades are treated by a separate factor f_g.

How do changes in vehicle mix affect the intersections when the intersection operates near or at capacity? The HCM delay equation is more sensitive to changes in saturation flow rate when an intersection is near or at capacity. For instance, under post-football game traffic volumes, assuming the heavy vehicle percentage as a part of that analysis would yield the results described on the next page.

What effects do heavy vehicles have on the intersection beyond changes to saturation flow rate? Beyond design issues that must be considered for accommodating the heavy vehicles at an intersection, the presence of heavy vehicles may have an impact on the lost time, lane utilization, and arrival type at an intersection.  Heavy vehicles will have an effect on lost time at the intersection because of the acceleration characteristics of these vehicles. This will be exacerbated on a uphill grade. 

Heavy vehicles will also affect the arrival type for each approach to the intersection by increasing the amount of time that a platooned flow may arrive on a standing queue. Simulation models will explicitly model the vehicle characteristics and performance on the approach to consider the relationships between the factors.

We will now consider how to set up this problem.

Sub-problem 3c: Critical Movement Analysis

Application of the critical movement analysis methodology to the Styner-Lauder/U.S. 95 intersection yields the following results, in vehicles per lane:

The results shown in Exhibit 1-37 are consistent with those that we obtained in earlier sub-problems from both the HCM methodology and from the application of a microscopic simulation model.  It should be noted that all volumes presented in this Exhibit are expressed in terms of vehicles per hour per lane, or vphpl. Specifically, the sum of the critical movements (1,790) is greater than the estimated capacity threshold of 1,400. Thus, the intersection demand exceeds its capacity and we can conclude with a fair amount of confidence that the intersection is above capacity.

If the sole question we were trying to answer was whether or not the intersection has sufficient capacity in its current configuration to accommodate the projected traffic volumes, the critical movement analysis might have been the most appropriate analysis tool to use, because it is able to provide the answer with much less effort and time than would be required by the other methodologies we have explored. Thus, the critical movement analysis methodology can be an effective and efficient way to answer some questions, as long as its limitations and constraints are always kept in mind.

Problem 3: Analysis

We will now consider the concepts presented in each of the three sub-problems.

In sub-problem 3a, we consider the special event-related traffic conditions that necessitate a second look at the intersections during event traffic. In this analysis we considered key concepts such as oversaturated conditions and demand volume. Specifically, we analyzed traffic across multiple scenarios, where demand exceeds capacity and arrivals from one scenario are served in the adjacent time period. In this case, we are able to analyze a scenario that provides an insight into oversaturated conditions. Where volume exceeds capacity, the collection of demand volumes becomes critical to our analysis in order to capture the effect of oversaturated conditions. This is a challenge that requires significant investment in data collection.

In sub-problem 3b, we utilize a simulation model to analyze the conditions. Our analysis using the simulation model provides us with an opportunity to test a variety of variables to determine the effect of specific measures and changes.

Finally, in sub-problem 3c, the concept of critical movement analysis is used to exhibit another methodology for determining whether an intersection is at or near capacity.

Discussion:
The ability of a traffic signal to handle fluctuations is a function of the signal timing that is in the controller in the field. In time period 3 (4:30 - 4:45 pm) of our previous analysis, we changed the green ratio slightly to serve the traffic at the post-game traffic at the intersection. Would this green ratio be possible under the existing pre-timed control? Take a few minutes to identify other factors that you think should be considered in this analysis, and continue to the next page when you are ready.

Problem 3: Discussion

What is next to consider in this problem, after looking at the effects of varying traffic intensities presented by event traffic, is the signal timing settings that affect the green ratio for the intersection. Note that we have focused our analysis so far on pre-timed control, but most signalized intersections today feature detection that can respond to traffic based on signal timing settings. 

In problem 4, we will consider these settings and determine the effect they have on performance of the U.S. 95/Styner-Lauder Avenue intersection.