Modelling motorway merge: The current practice in the UK and towards establishing general principles

Motorway merging has been regarded as a major source of conflicts and congestion, but has long been recognised as an area in which modelling has been relatively weak. The current traffic models represent the traffic operations at a merge are based on gap-acceptance approach, with the merging traffic giving-way to traffic on the mainline carriageway and imposing little or no delay to the mainline traffic. The results tend to overestimate delays to the merging traffic, and underestimate delays and interruptions to the mainline traffic.This paper investigates the current practice in modelling motorway merge in the UK and highlights some major conceptual and methodological differences to those in the US and Germany. Studies based on empirical observations and a dynamic, behaviour-based microsimulation modelling were carried out to improve understanding of the factors involved in the performance at merge. The results demonstrate the sensitivity of model predictions – and perhaps policy decisions – to the behavioural assumptions made in traffic models. The paper provides simple, practical recommendations to improve the current practice on modelling merge by the consideration of peak flow profile in the estimation of merge capacity, realistic local and variable pcu values for heavy vehicles, and a more confined merge influence area than currently advised in the UK guidelines. It calls for further empirical and behavioural-based microsimulation analysis and for the explicit consideration of the variability of traffic flows and dynamic congestion effects on merge, to improve the model predictions and provide better investment decisions.

► Existing models based on gap-acceptance rule underestimate delays and overestimate capacity.
► UK COBA method was based on out-of-date data, and lacks behavioural representation.
► Capacity at merge can be higher than that downstream. Use of wrong value leads to investment bias.
► Delays to mainline vs. merge traffic are different; presence of HGVs makes prediction complicated.
► Motorway merge is best modelled with dynamic, microsimulation models.