Is Micro-simulation a Waste of Time?
K Fox, Halcrow Group, UK
This paper explores the problems of using micro-simulation to perform transport assessments. It also describes new products that integrate macro, micro and meso models allowing the best approaches to be combined.
Micro-simulation models are computer models where the movements of individual vehicles travelling around road networks are determined by using simple car following, lane changing and gap acceptance rules. They are becoming increasingly popular for carrying out transport assessments and for the evaluation and development of traffic management and control systems.
A guiding principle in choosing a model for a transport assessment is the appropriateness of the modelling tool for the job taking into account the scope of the task, the needs of the stakeholders, and the resources available. When considering the use of a micro-simulation model, a key question to ask is: Would a more traditional (non micro-simulation) model meet the requirements?
Despite the claims of micro-simulation developers, for the vast majority of transport assessments there is no evidence that a pure micro-simulation approach is the most appropriate. For most cases greater understanding of the operation and performance of the network can be obtained better and much faster from a more traditional approach.
A major flaw with micro-simulation models is their inability to consider capacity explicitly ? which is a key indicator when you want to learn about your network. With a traditional traffic assignment model you can see where the network is over capacity and where there is spare capacity that you can utilise. It is very difficult to answer such questions with a micro-simulation approach. Micro-simulation models have no concept of capacity, it is not an input and it isn?t calculated, so it is very difficult to understand what is happening in the network from running a micro-simulation model. You can?t work out where the bottlenecks are or where they would shift to if any were removed. All they will tell you is that there is a queue if you send traffic down a street and it can?t all get out the other end. They don?t tell you how much spare capacity there is if there is no queue and they don?t tell you if there will be other problems in your network if you fix a capacity problem on one link and allow the traffic to get through the first bottleneck. To get a proper understanding of what is going on in the network you need a model that does look at capacity explicitly, i.e. a traditional traffic assignment model.
The increased detail offered by a micro-simulation is spurious ? modelling in finer detail does not necessarily produce more accurate answers - and is not worth the hugely increased computer run-times. Quite often when observing animations of micro-simulation models you see cars going through one another or running into the back of queues or lane changing at inappropriate places or even doing 360 degree turns as they travel along. This makes it very difficult to get models accepted as being valid as it just isn?t realistic behaviour. The look-ahead capabilities also often don?t work very well which causes problems when modelling small junctions, particularly mini-roundabouts. Claims that micro-simulation models pollution emissions more accurately than traditional approaches based on average link speed are doubtful given these other fundamental problems.
Another major weakness with micro-simulation models is their lack of any academically accepted route choice methods. The ad-hoc route choice algorithms utilised by the commercially available micro-simulation models are not based on any sound principles such as Wardrop Equilibrium. They are also often unstable particularly in congested networks where gridlock can occur and they have problems with convergence. They also can?t deal with fundamental issues such as elastic assignment or variable demand modelling. This makes micro-simulation inappropriate for any networks where route choice is a key issue. Certainly not for any large models such as whole town centres.
This paper explores these problems in more detail and provides advice on when micro-simulation is most effective. It also describes how new products are coming to the market that provide tools that seamlessly integrate macro, micro and meso modelling together allowing the best approaches to be combined effectively.
Association for European Transport