Modelling World 2024

When conducting appraisal and scheme testing, it is important to be confident that the change you observe between scenarios is caused by the scheme test rather than noise. This requires converged low noise models. Whilst having accurate realistic journey time and detailed effects are often important to understand the full benefits or disbenefits in a scheme. Dynamic modelling allows for detailed modelling over time however  It has often been said that dynamic models cannot converge. This statement is made as if it is obvious that this is the case. In this presentation, we will look at why it is difficult to achieve convergence in dynamic assignment and how convergence can be achieved through a change in methodology and careful decision making in design.

A case study of the Aimsun Next ONE model will be presented showing how we were able to achieve substantial improvements in stability within a single replication and show minimal noise in scheme testing.

Through the creation of robust dynamic assignment models, more detailed schemes can be tested for the emerging modern world such as bus priority schemes, segregated cycle paths, and dynamic traffic management such as smart motorways, ramp metering, dynamic road pricing, Understanding peaks in congestion, where buses may outperform cars. This leads the way to working together with both traditional and agent-based demand models.