4.1.3 Pre-modelling : Experiment Design

This chapter describes the definition of the experiment design process, the content of the process and discusses the issues of each step. Additionally, it provides support and references for scientists
Icône de l'outil pédagogique Authors

Jean-Paul Bousset, Marie Taverne, Etienne Josien, Olivier Thérond


Icône de l'outil pédagogique Definition

In the SEAMLESS-IF impact assessment procedure, experiment design is part of the pre-modelling phase – the so called “project definition” in the Seam:GUI, also including problem definition and indicator selection.

In the context of discrete-event simulation modelling, the design of simulation experiments (DOE) has long been instituted as a crucial and demanding task facing the analyst (Chen et al., 2003). Discrete-event simulation modelling is an effective method for predicting the performance of complex systems. The simulations are used to conduct experimental studies of the modelled system. Simulation runs consist of using mathematical models of a system in the form of a computer program, in which input factors (independent variables) are combined to produce an output or response (dependent variable). Experiment designs are carefully planned simulation runs, which determine

  1. The experimental conditions (independent variables) to be manipulated, i.e. the input variables defining the policy options to be tested in SEAMLES-IF
  2. The measurement (dependent variables) to be recorded, i.e. impact indicators which come from the Indicator Selection process
  3. The extraneous conditions (nuisance variables) that must be controlled, i.e. the agro-environmental and economic context and other outlooks.

Régi par la licence Creative Commons Attribution Non-commercial Share Alike 3.0 License

copyright Seamless 2009