Parking/Congestion

(updated  31st August 2007)

Keywords: simulating people, simulating crowds, simulating crowd dynamics

A badly managed car park will impact profits for an event. The consequences of congestion/queueing means the fee paying public are sitting in cars rather than spending money at the event. The knock-on consequences of traffic congestion on the local area and the success of an event, future attendance can significantly impact the bottom line. So it is important that event planners get it right.

The annual county fair near Crooklands in Cumbria created chaos for the day, local traffic backed up for over 10 miles, many people abandoned their cars in frustration adding to the congestion. Reports of cars trying to jump the central reserve of the dual carriageway, at high risk of accident were reported on the local news. Locals trying to go about their daily business were also caught up in the severe congestion. That day I had an important meeting to attend (hence the interest in this specific application of Myriad) fortunately this is exactly the reason I have a motorbike. While filtering through the fuming masses it became obvious that SOMETHING had to be done about this awful planning situation

We examine this event and apply the type of analysis that Myriad has been designed to allow us to perform environmental impact analysis.

Step 1. Orientation.

The turnoff to the country fair is via the A65 at Crooklands. Access is also via Milnthorpe. We can see that cars coming from two directions (along the A65 North and South) and cars coming to the site from Crooklands and Milnthorpe converge on the fair. Unless this situation is managed with maximum efficiency the traffic will inevitably back up along the routes to the fair. We can use satellite imagery to assess the local area in more detail.

Hi-resolution map of the area

Satellite Image of the area shown in the map above.                       Zooming into the area of interest.

Locating the road access and assessing the problem at a local area scale.

Locating the site entrance and local road configuration/geometry.

The area to the left of the image above is the sole entrance to the site. This is a single channel queueing problem and we can apply a variety of analysis techniques to the environmental impact of the traffic.

Step 2. Queueing Models

The entrance to the parking area will allow one car at a time to enter. As cars can arrive from two directions we have an arrival rate such that the queue will build exponentially.

We can apply a single channel queueing model to the situation to determine the rate of fill of the queue (traffic). The situation is going to compound very quickly as NON queueing traffic will become entangled in the queueing traffic.

Queueing Theory can predict, with some considerable accuracy the rate, extent and duration of many different types of queueing problem.

See Getting in for further information

Step 3. Applying the Queueing models

We can apply a variety of ingress tests (service rate - rate at which cars are being directed to parking areas) to arrival rates. The Spatial Utilisation map also allow us to determine arrival directions. Basically this is a superposition of local traffic flow rates into a back-propagated queuing model which highlights (red) those routes which will experience congestion. The rate at which (and direction of) the traffic jam builds up is seen when we run this model superimposed on the satellite image.

Once the traffic queue reaches this point (end of the red line) the traffic feeding the motorway from the Kendal by-pass will add to the arrival rates at Junction 36 and the queue grows at a much faster rate. The whole system is governed by the parking efficiency are the car park entrance. A simple model illustrates how this effects the entire system.

If arrivals are 8 per minute and parking is at 10 cars per minute the following graph shows the probability of 1 - 10 cars being in the queue on the approach roads.

If arrivals are 9 per minute and parking is at 10 cars per minute the following graph shows the probability of 1 - 10 cars being in the queue on the approach roads. The curve inverts (ie: traffic builds up faster than parking will facilitate) very quickly - giving us cause for concern in that local traffic is now going to be adding to the congestion problem. People who DO NOT want to go to the event are now PART of the traffic problem.

We can see that if the arrival rate of cars approaches the parking rate then the queue has a 30% probability that there will be at least 10 cars waiting to be parked. These cars start to queue back along the approach roads. By using a simple metric for the average length of a car we can determine the queue length, the time it will take to back-up to the main A65 and onwards to Junction 36 and beyond to Kendal.

When the arrival rate exceeds the parking rate the queue is guaranteed to be extensive. A gambler would like these odds > 10000% certain that the queue will be longer than 10 cars.

Testing a variety of queueing models indicated that this queue would have engulfed the M6 Junction 36 within 15 minute, thereafter it would not clear for several hours which was exactly what happened during the event.

Step 4. Solving the problem

Solving this type of problem is much simpler than creating a model of the queueing congestion rates. The process of parking cars at these kind of events is very much a single channel system (and prone to the above exponential growth of queues). Turning the system into a Multi-Channel problem, filling more than one row of parking spaces at a time, would significantly reduce the congestion. 5 channels are required to remove the queueing/congestion problem at this site. This does NOT require additional man power. In fact it significantly reduces the need for local traffic control.