Use of simulation to test the validity and sensitivity of an analytical model

摘要

The purpose of the research reported here was to test formally the validity of some assumptions made in solving models by analytical techniques and to test the sensitivity of the system to the arrival distribution. The presentation refers to a simplified real-time model. The analytical solution in the literature is to derive shown results for each stage of the system and add them up to obtain the behavior of the system. Assumptions must then be made at each stage. Simulation has been used to solve the same system in terms of the same measure of efficiency, i. e. the response time. However, the system is solved as a whole, the output from one stage becoming the input to the second stage. Confidence limits have been obtained for the response time in order to test the results obtained from analytical techniques. Simulation has also been used to test the sensitivity of the system to a change in the arrival distribution. Using analysis of variance, the effect of the arrival pattern and of the interaction is determined.