Design of Experiments in Simulation

SYNOPTIC ABSTRACT

Many simulation studies may be viewed as experimental situations in which a number of factors (independent variables) are to be investigated. There are, however, a number of special characteristics that distinguish simulation experiments from statistical experiments in general. In this paper we discuss four topics which are of particular interest and relevance in the simulation context: (1) Identification of the important factors (i.e., input variables); (2) Investigation of the statistical relationship between the output and input variables; (3) Determination of the combination of factor levels for which the response (i.e., output variable) is optimized; and (4) The use of variance reduction techniques.

Key Words and Phrases:

• experimental design
• computer simulation
• factor screening
• optimization of simulation outputs
• investigating statistical relationships
• variance reduction