Comparition of Numerical Methods for Stiff Problems in SimInTech

In the problems of modeling various classes of technical systems, it often becomes necessary to numerically solve stiff systems of differential-algebraic equations. The efficiency of numerical simulation is determined by the total computational costs of the integration method used, which consist of the need to calculate the right parts of the system of equations, including the need to recalculate the system function in the process of calculating the Jacobian matrix. Accordingly, numerical methods for modeling stiff systems should provide a minimum number of system function calculations while ensuring the required specified accuracy. This article compared the performance parameters of integration methods implemented in the SimInTech software package and the world collection of integration methods, include BIMD, DDASSL, GAMD, MEBDFI, PSODE-1, RADAU, VODE solvers. A conclusion is made about the effectiveness of the Adaptive 4 and DIRK4 methods in SimInTech on the example of the problem of the kinetics of the Robertson autocatalytic reaction.