Modelling congestion. Most micro-simulation models use simple car following and lane changing algorithms to determine vehicle movements. During congested conditions these do not realistically reflect driver behaviour. For example complex situations can occur when queues block back across junctions or when parked or broken down vehicles block a lane producing shock waves. The way such congested situations are modelled is often critical to the results obtained.
Environmental modelling. Considerable effort is being directed at producing emission models for incorporation into micro-simulation models. Many of these have been reviewed during the DRIVE II KITE project. For some emissions this is straightforward but for others complex chemical reactions are taking place within car exhausts making predictions difficult. It is also proving difficult to get reliable emission data for a reasonable mix of traffic.
Integrated environments and common data. Micro-simulation models are often used with other models such as assignment models. There are common inputs required by all these models, such as origin-destination data, network topology, bus route definitions. However, each model often requires the data in a different format so effort is wasted in re-entering data or writing conversion programs.
Safety evaluation. Safety is a very complex issue. We are still a long way from being able to accurately predict the safety implications of implementing many transport scheme options. Most safety prediction models are very crude, being based on vehicle flows on given road types or on changes in mean vehicle speeds. Most micro-simulation models completely ignore vulnerable road users such as cyclists or pedestrians. Greater effort should be directed at this area.
Responsive systems. Many new systems have been developed to respond to changing conditions within the road network, such as accidents, roadworks, levels of flow or joy-riding. Assessing the likely performance of such systems obviously requires a suitable amount of variability to be input into the models. There is a lack of data about such levels of variability and there is no standard methodology for including it within the assessment. How drivers respond to such systems is also an issue.
Level of detail. Often compromises have to be made between the level of detail modelled (roads to include, mix of traffic etc.) and acceptable simulation run times. Guidelines indicating the consequences of such approximations would be useful.
Validation and Calibration. Few models have been rigorously calibrated and validated. More data needs to be collected so that suitable ranges of model parameters can be determined This has consequences for transferability of the models across Europe.
Standard procedures and indicators for evaluation. The micro-simulation has to produce outputs which will rank the schemes realistically. Scheme rankings are a function of the chosen performance indicators and the weights used. Standard sets of performance indicators and procedures to apply need to be produced. Some of these problems will be answered by task 7.3/19 in the 3rd call.
Public transport. Public transport vehicles behave in a different way to other vehicles but are often not modelled in sufficient detail to reflect these differences.
The SMARTEST project will be questioning current and potential users of micro-simulators across Europe to find out what areas they particularly want improving. This list of problems will be prioritised and solutions to the most important ones will be tackled by using and enhancing the simulation tools within the SMARTEST consortium. Data to validate the models will be collected from sites around Europe.