ADAPTIVE RUNGE-KUTTA ALGORITHMS FOR DYNAMIC SIMULATION

The authors present the development of an adaptive algorithm for the solution of ordinary differential equation systems. The methods used are based on two distinct classes of Runge-Kutta processes which are chosen to match the characteristics of the problem being solved. The principal characteristics related to the problem stability are assessed during the solution procedure and decisions are made on the type of numerical method to use at a given point in time. It is shown that an adaptive algorithm using Runge-Kutta methods gives modes to significant computational savings over the use of a single implicit Runge-Kutta solution method.