lumped model, adjustment of parameters, genetic algorithm, calibration procedure
Comprehensive models are widely adopted to predict the performance of compressors due to their low computational cost and acceptable prediction capability. In general, the accuracy of such models depends strongly on the correct adjustment of some parameters that are of difficult to determine both analytically and experimentally. However, due to the nonlinearities of the compressor model, the tuning of such parameters affects many output variables and hence can be very challenging and time consuming. In this paper, we consider this procedure of adjustment of parameters as a multiple objective optimization problem that can be solved by using an elitist non-dominated sorting genetic algorithm (NSGA-II). The parameters of two simulation models are adjusted following this new procedure. In the first model the suction muffler was neglected and a mass-spring-damper system was adopted to predict the suction valve dynamics. The second model solves the suction valve dynamics by the finite element method and the flow in the suction muffler with the finite volume method. In both models the clearance volume and parameters associated with the suction valve were chosen to be adjusted while the deviations between predictions and measurements for the mass flow rate, indicated power, and suction valve dynamics were defined as the objective functions to be minimized.