This paper presents the calibration and validation of the INSECT model, which is a vehicle by vehicle simulation model for handling sign controlled intersections, roundabouts, and more recently traffic signals; the paper covers the validation of INSECT for non-signalised intersections. INSECT is a suite of computer programs, designed to allow experienced users to undertake detailed analyses of intersections; however, inexperienced users can apply it to undertake investigations of most intersection design problems likely to be encountered. The calibrations and validation have indicated that the model gives good results across the whole range of intersections tested; for most cases, the INSECT model predictions are accurate within acceptable bounds of variation. On average, INSECT underestimates queue delay by approximately 4.5%, But there are greater differences at individual intersections and on individual approaches. It has been shown that fixed gap times are not applicable, either to individual intersection types or within these types. This is mainly because sign controlled intersections and roundabouts have different gap acceptance characteristics; in addition, the gap accepted by a motorist is related to the time that he has waited. It is believed that INSECT is a useful tool for evaluating all types of intersections, and its recent extension to intersections controlled by traffic signals will be discussed in a future paper. It is also intended to extend the INSECT model to the area of accident prediction.
Samenvatting