Modelling Group Investigates…Pedestrian Crossings
Introduction
Welcome to the latest Modelling Group post and something a bit different from our usual blog posts.
In this post, we are going to look at the modelling of pedestrian crossings and do an experiment to understand how different set-ups perform as the number of pedestrians increase.
The aim of the post is to provide an insight into the optimum pedestrian crossing set-up, depending on the likely flow of pedestrians.
The Experiment
The experiment undertaken focuses around a single, 6m wide pedestrian crossing, as shown in Figure 1.
Figure 1 – Pedestrian Crossing Modelled
The pedestrian crossing has been modelled using a Puffin/Toucan crossing VisVAP template, with both stop-line detectors and on-crossing detectors.
Static assignment has been used for the traffic flows, with vehicle inputs of 300vehs/hr used for both traffic lanes.
The model has been configured to run for 600 simulation seconds (10 minutes), with a simulation resolution of 5.
Priority rules have been added at the stop-line locations, to ensure that vehicles stop and wait for any ‘late-crossing’ pedestrians to cross before proceeding with their journey.
For the pedestrian flows, three different flow sets are to be tested:
1) 900 peds/hr (150 peds/10 mins (length of simulation)) in both directions
2) 1800 peds/hr (300 peds/10 mins (length of simulation)) in both directions
3) 2700 peds/hr (450 peds/10 mins (length of simulation)) in both directions
To assess the effectiveness of each crossing set-up, the pedestrian journey times and volumes will be collected for those travelling eastbound and westbound.
Figure 2 – VisWALK (Basic) Set-Up
Pedestrian Crossing Set-Ups
5 different crossing set-ups have been configured, which are detailed as follows:
Set-Up 1 – VisWALK (Basic)
The wireframe view for the basic VisWALK set-up is shown in Figure 2.
As detailed in Figure 2, a basic VisWALK pedestrian crossing set-up has been developed. This allows a simple model operation where pedestrians enter the network and then look to reach the other side of the road, following a single static route.
Figure 3 – VisWALK (Dynamic Potential) Set-Up
Set-Up 2 – VisWALK (Dynamic Potential)
The wireframe view for the VisWALK set-up with dynamic potential is shown in Figure 3.
The only difference in this model from the VisWALK (Basic) set-up is that the use of dynamic potential is enabled. This means that pedestrians should be more reactive to crowded spaces and look to take routes that may be slightly longer, but take less time to reach their destination. In this instance, it would not be expected that all pedestrians should look to walk from one corner of the crossing to the other diagonally to reach the other side.
Figure 4 – VisWALK (Partial Routes – Static) Set-Up
Set-Up 3 – VisWALK (Partial Routes - Static)
The wireframe view for the VisWALK set-up with static partial routes is shown in Figure 4.
As detailed in Figure 4, partial routes have been added, which utilise new pedestrian areas which have been added either side of the crossing stop-lines. These make pedestrians walk straight across the crossing and should therefore lead to less conflicts (than pedestrians walking diagonally across the crossing). The partial route assignment has been set to static, to allow a simple set-up to be modelled.
It should be noted that for the static routes, dynamic potential has not been enabled, to allow an understanding of only the introduction of the partial routes to be understood.
Figure 5 – VisWALK (Partial Routes – Density) Set-Up
Set-Up 4 – VisWALK (Partial Routes - Density)
The wireframe view for the VisWALK set-up with static partial using ‘density’ for their route choice is shown in Figure 5.
The only difference in this model from the VisWALK (Partial Routes – Static) set-up is that density has been chosen for the route choice. This means that the volume of pedestrians in the areas of the partial routing decisions is calculated, with 90% of the people then assigning to the area with the lowest density (the best route). Therefore, it would be expected that people would spread out over the stop-line as the individual areas became busier.
As in the other model, dynamic potential has not been enabled for the static routes, to allow an understanding of only the introduction of the ‘density’ route choice to be understood.
Figure 6 – VISSIM (Pedestrians) Set-Up
Set-Up 5 – VISSIM Peds
The wireframe view for the VISSIM pedestrians set-up is shown in Figure 6.
To understand the impacts of simply using the VISSIM pedestrians at the crossing, new links were added and the same pedestrian flow inputs assigned to these links. All associated network elements were transferred to the new links, to allow a VISSIM vs. VisWALK comparison to be made.
Test 1 – 900 peds/hr
The first test includes a pedestrian flow of 900peds/hr (or 150peds/10 mins) to understand the performance of the different crossings. The performance from Random Seed 42 for each of the crossings is shown below.
Results
The journey time results have been extracted from the average of 10 random seed runs and are summarised in Table 1.
Table 1 – Pedestrian Results – 900peds/hr
From Table 1, it can be seen that in terms of the VisWALK options, there is not much difference between the different set-ups, both in terms of journey times and volume of pedestrians. This indicates that a simpler configuration will give a suitable pedestrian crossing representation.
The VISSIM pedestrian option provides the lowest journey times and highest volume of pedestrians, but this is without any interaction between the pedestrians and as can be seen from the video, pedestrians just stand and walk over/through each other.
Test 2 – 1800 peds/hr
The second test increased the pedestrian flow to 1800peds/hr (or 300peds/10 mins) to understand the performance of the different crossings. The performance from Random Seed 42 for each of the crossings is shown below.
Results
The journey time results have been extracted from the average of 10 random seed runs and are summarised in Table 2.
Table 2 – Pedestrian Results – 1800peds/hr
From Table 2, it can be seen that the use of the partial routes is starting to show a benefit, with the journey times being slightly lower and more pedestrians completing their journeys. As can be seen from the video, this is the result of the partial route configuration providing a more uniform route across the crossing and less interactions with other pedestrians.
As in Test 1, the VISSIM pedestrian option provides the lowest journey times and highest volume of pedestrians. However, as previously commented, this is without any interaction between the pedestrians and as can be seen from the video, pedestrians just stand and walk over/through each other.
Test 3 – 2700 peds/hr
The third test increased the pedestrian flows further to 2700peds/hr (or 450peds/10 mins) to understand the performance of the different crossings. The performance from Random Seed 42 for each of the crossings is shown below.
Results
The journey time results have been extracted from the average of 10 random seed runs and are summarised in Table 3
Table 3 – Pedestrian Results – 2700peds/hr
From Table 3, it can be seen that there is now a clear benefit with the use of partial routes for the crossing. The journey times are nearly 10 seconds quicker than the basic set-up, with more pedestrians also completing their journeys. It can also be seen that the use of ‘density’ over ‘static’ for the route choice has a small benefit, particularly in regard to the pedestrian volumes.
When reviewing the videos, it can also be seen how the basic VisWALK set-up affects the ability of vehicles to continue their journey, as people become ‘stuck’ at the corners and don’t clear the crossing sufficiently (the priority rules then forcing the cars to stop and wait). Whilst this is appeased with the use of ‘dynamic potential’, the use of the partial routes shows a more favourable set-up for vehicle progression, as well as pedestrians.
Finally, as expected from the previous tests, the VISSIM pedestrian option provides the lowest journey times and highest volume of pedestrians. However, as can be emphatically seen in the video, this is without any interaction between the pedestrians, who just stand and walk over/through each other.
Summary & Conclusions
This Modelling Group investigation post has looked at the modelling of different pedestrian crossing set-ups, to understand how they perform with different pedestrian demands.
The most obvious conclusion drawn is that, if you are not interested in the pedestrian interaction and simply want to model the most efficient layout for the crossing, then this can be done in VISSIM using links and vehicle inputs (set to pedestrians).
However, if your study needs to take into account the operation of the crossing and the interaction of pedestrians, then if you have a low demand, the basic set-up is more than adequate to meet your demands. This could be the case if you have remote, standalone crossings in your model.
If your model includes high volumes of crossing pedestrians (for example, located outside of a large public transport interchange), then you should consider the use of the partial route set-up for your crossings. This will configure your pedestrians to cross in a uniform manner and reduce the number of interactions (which can otherwise cause unrealistic behaviour using a simple VisWALK set-up).
It should be noted that there are obviously more variables that can be tested (larger crossing width, larger waiting areas by the crossings, more entry/exit points and a greater number of pedestrian demand increments). However, we hope that this test has helped provide some guidance on what set-up to use to configure your crossings, depending on the nature of the crossing and the likely demands (both baseline and in the future).