A HYBRID IMPULSIVE SCHEME FOR FASTER THAN REAL-TIME VEHICLE LOADS PREDICTION

Faster than real-time and real-time vehicle dynamic models run with fixed time step integrators, without accuracy control and usually apply numerous approximations to obtain a stable solution. These models generally provide very good descriptions of the system behavior, capturing the gross motions and character of the response. However, the consequence of approximation and the lack of error control is that the resulting loads cannot be trusted for structural design analysis. Nonetheless, these lightweight faster than real-time models are indispensible in concept development where an unlimited number of designs may be considered in the automated exploration of the design space. This work investigates a novel simulation technique in an attempt to converting the family of real-time vehicle dynamics models into reliable first order structural load predictors. The method applies an error estimator to the real-time fixed step integrator to identify loss of accuracy in stiff models with resolution to the offending degree(s) of freedom. Inaccurate and potentially unstable time steps are then replaced by an impulse-based solution and the time-step recomputed. The required impulsive load can then be transformed (on or off-line) into a repeatable and accurate load pseudo-time history through integration of an independent nonlinear contact event. A simple tracked vehicle model example is used to demonstrate the features of the solution which is validated by comparison to results obtained from fully integrated trajectories.