Methods of Model Verification

Abstract

“Verification” is the word used to describe the process that attempts to establish that a computer simulation model is consistent with the underlying conceptual or mathematical model upon which it is based. It involves trying to ensure that the computational model contains no errors in terms of logic and coding, including the choice of the numerical algorithms being used for integration and other operations that are important for simulation applications. The three most commonly-used approaches to the detection of run-time errors in simulation models are code reviews, static-analysis methods and dynamic testing for carefully chosen cases. A further aspect of the verification process involves demonstrating the accuracy of data used for the simulation and estimating errors in numerical solutions and this is usually approached through tests involving well-understood situations. The chapter also includes discussion of more specific issues concerning the verification of simulation models based on ordinary differential equations, models involving differential algebraic equations, models based on partial differential equations and models involving discrete-event or hybrid descriptions. The chapter concludes with a brief discussion of the role of formal methods in simulation model verification.