| Abstract: | This paper presents a new approach to the modeling of congested traffic loading events on long span bridges. Conventional traffic load models are based on weigh-in-motion data of non-congested traffic, or something similar to a Poisson Arrival process. In neither case do they account for the mixing between lanes that takes place as traffic becomes congested. It is shown here that cars move out from between trucks as traffic slows down which results in a higher frequency of long platoons of trucks in the slow lane of the bridge. These longer platoons increase some characteristic load effects under the slow lane by a modest but significant amount. Micro-simulation, the process of modeling individual vehicles that is widely used in traffic modeling, is presented here as a means of predicting imposed traffic loading on long-span bridges more accurately. The traffic flow on a congested bridge is modelled using a random mixing process for trucks and cars in each lane, where each vehicle is modelled individually with driver behaviour parameters assigned randomly in a Monte Carlo process. Over a number of simulated kilometres, the vehicles move between lanes in simulated lane-changing manoeuvres. The algorithm was calibrated against video recordings of traffic on a bridge in the Netherlands. Extreme value statistics of measured strains on the bridge are then compared to the corresponding simulation statistics to validate the model. The micro-simulation algorithm shows that the histograms of truck platoon length are moderately affected by lane changing. This in turn is shown to influence some characteristic load effects of the bridge deck. |
