Performance of deflecting concrete highway barriers

Shaped concrete barriers are employed as roadside protection in many highway jurisdictions. Those found in British Columbia, Canada, are assemblages of 2.5 m precast concrete segments. Two types of barrier are currently employed: a 810 mm high barrier used for median protection and 690 mm tall roadside barriers installed on the shoulders of the roadway. The latter represents a unique roadside protection system to B.C. Both the median and roadside barriers are free standing structures that can deflect during a vehicle impact. The collision performance of these barrier was examined experimentally and through a mathematical model developed herein. The experimental test program consisted of 26 vehicle crash tests of both barrier types. Principal findings of the testing were that the vehicle was redirected to the road for all test conditions and that the barriers could separate at their joints without allowing the vehicle to penetrate. Specialized tests on the joint structures were conducted, to assist in developing a mathematical description of the barrier dynamics. A mathematical model was developed to simulate a vehicle-barrier collision. This model was based on a planar representation of the collision event. The vehicle was modelled as a deformable body using structural deflection characteristics developed for accident reconstruction programs. The dynamics of the barrier were modelled as a multibody system of rigid links. Connections between the barrier segments were defined with a five phase joint deflection behaviour. These different phases represented (with increasing barrier deflections): 1) the slack between the segments, 2) joint rotations about the hook and eye, 3) initial binding contact of the concrete segment corners, 4) rotation about the point of binding, and 5) joint separation. Friction forces were calculated from a model of two planes in contact. Static and dynamic coefficients of friction were incorporated into this friction model. The simulation used analytical descriptions of the impulsive or discontinuous events occurring during the collision event. These involved transitions between different joint or friction regimes. Numerical integration was suspended while an impulse analysis of the event was conducted. The post impulse parameters were used as the starting point for the resumption of numerical integration. This approach allowed the simulation to proceed efficiently without excessive iteration about these transition periods. The resulting mathematical model was programmed into a personal computer program. This program allows arbitrary barrier and vehicle definitions to represent a vehicle-barrier impact. Results of a sensitivity analysis and comparison to test data were very encouraging. The vehicle model appears to reasonably represent a vehicle collision with a rigid vertical wall. Simulations with the deflecting barrier definition did not have as good agreement with test data, but showed a reasonable replication of the collision event. The sensitivity analysis showed that the vehicle redirection variables describing the vehicle yaw and yaw rate at exit were the most sensitive to perturbations in the input parameters. The analysis suggests that the redirection is closely related to the impact of the rear axle and rear bumper with the barriers. Better representation of this event will improve the model's performance.Applied Science, Faculty ofMechanical Engineering, Department ofGraduat