Nearly-perfect Compensation for Time Delay in Real-time Hybrid Simulation

SUMMARY: Real-time hybrid simulation (RHS) is a powerful technique to evaluate structural dynamic performance by combining physical simulation of complicated and rate-dependent portion of a structure and numerical simulation of the rest portion of the same structure. This paper shows that the stability of RHS with time delay is not only related to compensation methods but also to the integration methods. With conventional compensation method, even when the time delay is exactly known, some combinations of numerical integration and displacement prediction schemes may reduce the response stability, and lead to unconditionally instability in the worst cases. To deal with the inaccuracy of prediction and the uncertainty of delay estimation, a nearly-perfect compensation scheme is proposed, in which the displacement is over-compensated and then the datum that is closest to the desired displacement is picked out by an optimal process. Advantages of this scheme over conventional compensation have been shown through actual tests. Blank line 10 pt Keywords: real-time hybrid simulation; delay compensation; overcompensation; control error., as a novel technique for evaluating dynamic responses of structures, draws much attention in the past two decades. In a hybrid simulation, the structure is divided into physical and numerical parts. The coupling between the two parts is handled by one or several numerical coordinators and physical transfer systems. The synchronization at the interface is critically important for an accurate RHS. However, the delay is inevitable owing to the time spent on the computation of numerical coordinator and the actuation of physical transfer system. The time delay will reduce the accuracy, and cause the instability of RHS in the worst case. Blank line 11 pt Most research efforts to reduce the negative effect of time delay focus on the development of various compensation schemes for transfer systems. These schemes basically can be classified into two types: i) to send the command in advance; ii) to add a compensator with positive phase. In a hybrid simulation, the command sent to physical substructure is actually the calculated response at the interface of physical and numerical parts, so it can not be prescribed as the external loading is. Then response prediction is needed for the first type of compensation schemes. All the available prediction methods are based on polynomial extrapolations, among which the third-order Lagrange polynomial proposed by Horiuchi for RHS is most widely applied. Extrapolations based constant and linearly varying acceleration were also studied [Horiuchi & Konno …