1. Problem 3: Shenendehowa Campus
3a: AM & PM Peak Hour - Existing Conditions
3b: PM Peak Hour - With Conditions
To examine (1) PHF, (2) heavy vehicles & (3) impact dilution.

2. Characteristics of Moe Road Intersection
Signalized & fully actuated
2 lanes EB (left-through & exclusive right)
2 lanes WB (left & through-right)
2 lanes NB (left & through-right)
1 lane SB (left-through-right)

3. The volumes to/from the north are extremely small because a church is the only building generating traffic on that approach
Traffic leaving the Shenendehowa campus uses the NB left and right
Arrival Patterns
Large volumes exist on the EB, WB and NB approaches

4. Sub-problem 3a: Shenendehowa Campus AM & PM peak - Existing Conditions
We will use the highly peaked flows at the entrance to the Shenendehowa campus to show how the peak hour factor works and the effects it has.
What does the PHF account for?
Variations in flows that occur during the heaviest hour of traffic
The peak hour factor (PHF) accounts for variations in flows that occur during the heaviest hour of traffic. If the volume for the hour is 800 vehicles and the heaviest volume duLevel of Serviceny one 15-minute period is 250 vehicles, then the peak hour factor is 0.80 (800/(4*250)). When you input the hourly volumes and the peak hour factor, you will evaluate the conditions that exist during the peak 15 minutes, the time when the facility is most heavily loaded.
We can use the highly peaked flows at the entrance to the Shenendehowa campus to show how the peak hour factor works and the effect it has. Using data for this intersection will show how the typical method for applying the peak hour factor might or might not lead to the right assessment of the performance conditions in some situations.
Discussion: Traffic engineers hold different perspectives on the peak hour factor. Some compute values for each clock hour (3-4, 4-5, etc.). Some consider each sequence of four 15-minute time periods and use the sequence with the maximum total volume for the peak hour factor. Some predicate PHF calculation on the sequence of 15-minute time periods that has the maximum flow for the movement, others, the maximum total intersecting volume for the intersection. Still others focus on demand, not volume as the basis for computing the PHF. What do you do? What do you think should be done if the data were available? 
When you input the hourly volumes and the peak hour factor, you will evaluate the conditions that exist during the peak 15 minutes, the time when the facility is most heavily loaded.

5. What other observations should be made?
The AM peak volumes by 15-minute interval are presented in Exhibit The peak hour is highlighted in yellow. As can be seen, there is high variability in most of the flows. Only the eastbound through is relatively consistent. For example, during the peak hour, the eastbound right ranges from 54 to 111 vehicles in a 15 minute period. The westbound left ranges from 106 to 45 (because of the school traffic), while the northbound right ranges from 41 to 79.
If we do a standard peak hour analysis for the AM peak hour, we get an overall level of service C. Dataset 22 contains the complete input data for the AM peak hour analysis. Exhibit 2-28 shows the delays and levels of service for each of the movements. The largest delays are associated with the westbound left and the conflicting eastbound through. The westbound left movement clearly has the worst LOS (D).
When is the AM peak hour?
What other observations should be made?
Doing a standard peak hour analysis we get a LOS = C
High variability

6. What other observations should be made? When is the PM peak hour?
The PM peak is similar. The 15-minute counts are shown in Exhibit 2-29, and the peak hour is highlighted in yellow. Comparing the PM peak to the AM peak in Exhibit 2-27, we can see that the PM peak eastbound rights and the westbound lefts are significantly less then in the AM peak. Also, the westbound through volume is much larger. Finally, there’s a major change in the percentage of heavy vehicles. In the AM peak, the percentages were 15% for the westbound left and 9% for the eastbound right. In the PM peak, they are 26% for the westbound left and 41% for the eastbound right.
What other observations should be made?
When is the PM peak hour?
- Not as much variability as the AM peak
- Less EB rights & WB lefts than in the AM peak
- Major increase in % of heavy vehicles, relative to AM peak, on some approaches

7. Comparison of delays & LOS for the AM & PM peaks
What observations can be made?
Questions to consider:
- Are either of the conditions shown in this table likely to occur?
- Are these good representations of the conditions in either peak hour?
- Are they pessimistic? Optimistic?
The points we are trying to emphasize with this slide are the following:
Using a PHF predicated on the total flows by 15-minute period smoothes out the peaking and produces a result which is optimistic.
Using the PHF for each movement individually, which the procedure and software allows, produces a very pessimistic result because it asserts that the peak 15 minute flows occur for all movements simultaneously.
Looking at the individual 15 minute results demonstrates that neither the conditions predicted by 1) or 2) above ever really occurs. The LOS are all better than 2) and closer to 1) but they never match 1) because the 15-minute flow combinations are always slightly different than those implicit in 1).
Perhaps the most accurate way to model peak 15-minute flow rates is to identify the 15-minute period when the sum of all entering movements is greatest, and then multiply each of the movement volumes by a factor of 4 to yield hourly flow rates. In an actual analysis based on this approach, the PHF factor should be set to 1.0, since the peaking effects have already been taken into account in the flow rates that are being used. This method does not use the PHF factor in an explicit way, but may nevertheless accurately reflect the intent of the factor.
- Similar LOS
- There is less delay in the PM Peak

8. How would the results change if only the peak 15 minute period was considered?
In the original analysis, the average delay per vehicle is 25.9 seconds. When we use movement-specific PHF values, the average delay is 38 seconds, or 47% higher. Is this realistic? We’ll see. The average delays on a 15-minute basis range from 17.8 to 29.2 seconds per vehicle. So the 38.0 seconds is clearly too high. The original analysis underestimates the delays during the peak 15 minutes where it is 13% higher.
How does the delay change between base case and by-movement?
The by-movement delay is highly over estimated based on the data collected for each 15 minute time period

9. How would the results change if only the peak 15 minute period was looked at?
A detailed look at the individual 15-minute intervals from Exhibit 2-31 is also instructive. The eastbound through-and-right delays range from 20.8 to 34.9 seconds; the northbound approach delays range from 19.6 to 32.6 seconds; and the westbound lefts range from 28.4 seconds all the way to 51.6 seconds per vehicle. For the rest of the movements, the delays are more consistent. Notice that in all of the interval cases and in the base case, each of the delays is lower than those produced by the analysis done using PHF by movement.
When comparing the original peak hour analyses to each of the 15-minute interval analyses, it is obvious that the actual intersection performance levels are not consistent with the predicted AM or PM peak hour conditions. In the worst 15-minute interval (8:00 to 8:15), the overall LOS is C. For this interval, there are significant increases from the PM peak hour results for the eastbound right, the westbound left, and all three of the southbound movements. Looking at the best performing 15-minute interval (8:30 to 8:45), the overall LOS is B. For this interval there are significant decreases (from the base case) in the delays for the eastbound though-left, the westbound left, and all three northbound movements.
Is there consistency between the original peak hour analyses and each of the 15-minute interval analyses?
No, there are significant differences

10. Effects of Heavy Vehicles
What would happen if the heavy vehicle percentages were ignored?
What would happen if the heavy vehicle percentages were ignored? Let’s compare the results from the base case AM and PM peak hour analyses with results if the correction factors were left out. For complete input data for each of these analyses click here.
Exhibit 2-33 demonstrates the differences in delay that will be obtained by neglecting the percent heavy vehicle correction. In both the AM and PM conditions, the delays are smaller when the correction factors are omitted. The differences in the AM peak are slightly larger than they are during the PM peak. This is due to the slightly higher volumes that occur during the AM peak.
A sensitivity analysis was not conducted. It might be useful to see how much the performance predictions change if the percentage of trucks grows. Then you could understand how important it is to have an accurate estimate for the analysis and the sensitivity to variations that occur in normal traffic.
The result would be a significant decrease in delay, but of course the intersection would not be accurately evaluated

11. Sub-problem 3b: Shenendehowa Campus PM peak - With Conditions
The PM With condition is a 2004-forecasted condition that considers the impacts of the traffic generated by the Maxwell Drive site development. As we move further from the actual site development, the impact of the site-generated traffic diminishes. We’ve seen the impacts at Maxwell Drive and Moe Road. Let’s now look at the impacts of this site-generated traffic at the current intersection.
The volumes that will be used to analyze the 2004 PM With and PM Without conditions are shown in Exhibit 2-34. There is a small estimated growth on the eastbound through movement, and the westbound through movement and the rest of the movement volumes are unaffected by the site development
Exhibit 2-35 shows what we find from analyzing each of these three conditions. Comparing the with and without conditions, the changes in overall delay are quite small. To check the robustness of this comment (i.e., the sensitivity to uncertainty in the site development volumes), we looked at an additional with condition with 30% more site-generated traffic. The delays still have not changed much. This tells us that this intersection is not significantly affected by the site development at Maxwell Drive.
What would the effects be at this intersection if the traffic at Maxwell Drive were increased?
The sensitivity analysis suggests the changes in overall delay are quite small

12. What have we learned?
We’ve seen that you have to be careful in using the peak hour factor.
It’s good to incorporate a PHF so that the conditions in the peak 15 minutes are examined. But unless you know the flows all peak simultaneously, it’s not good to use peak hour factor values that are movement specific.
We’ve also seen that it is important to pay attention to the heavy vehicle percentages.
We’ve seen that there are ways to check for impacts from site-generated traffic.
So what have we learned? We’ve seen that you have to be careful in using the peak hour factor. It’s good to incorporate a peak hour factor, so that the conditions in the peak 15 minutes are examined. But unless you know the flows all peak simultaneously, it’s not good to use peak hour factor values that are movement specific. You’re better off using the value that pertains to the intersection as a whole during the peak hour. Even that value can lead to delay estimates that are higher than any real values obtained during the actual 15-minute intervals. The reason is that the overall peak hour factor, applied to all of the flows, still assumes implicitly that all of the movements peak simultaneously and proportionally as well. Sometimes, as is the case here, that doesn’t happen. If you find this is a significant issue, you might want to do analyses for each 15-minute period individually.
We’ve also seen that it is important to pay attention to the heavy vehicle percentages. This may be of particular importance in a situation like this, where the Shenendehowa intersection serves a lot of school buses. We might not initially realize the importance of accounting for their presence in the traffic stream, but doing so changes the delays considerably.
Lastly, we’ve seen that there are ways to check for impacts from site-generated traffic. We were relatively formal about that, doing the performance assessment with and without the site-generated traffic, looking at the resulting changes in delay, and deciding that the impact was insignificant. Sometimes, for expedience, analysts make a decision based on the percentage increase in intersecting traffic that results from the site-generated traffic.