The aim of the Response Surface Methodology (RSM), usually applied to chemical, physical and biological systems, is to represent a reliable response surface with a strong adherence to the described reality. In order to achieve this target RSM researchers focused on experimental designs useful to calculate a regression meta-model featured by a good fit to the experimental data and a sort of stability of the prediction variance. Furthermore the design should allow to evaluate two experimental error components, namely pure error and lack of fit. Achieving this with the smallest number of experimental runs it is important to save experimentation time and costs. In classic RSM approach the Experimental Error (EE) is considered fixed and dependent only from the physical system under study. Several experimentations, though, have been performed on simulation models representing the physical system, as it isn't possible to experiment in a direct way. Complex industrial plants belong to this category because of the difficulty (i.e. time and investments required) to modify them in order to perform an experiment. In Discrete Event Stochastic Simulation (DEVS) the EE is simulation time or replications dependent. This requires a substantial change of approach for EE estimation. In this work Authors propose an improved approach and outline limitations of traditional ones.

Discrete and stochastic simulation and response surface methodology: An approach to a varying experimental error

CASSETTARI, LUCIA;MOSCA, ROBERTO;REVETRIA, ROBERTO;TONELLI, FLAVIO
2007

Abstract

The aim of the Response Surface Methodology (RSM), usually applied to chemical, physical and biological systems, is to represent a reliable response surface with a strong adherence to the described reality. In order to achieve this target RSM researchers focused on experimental designs useful to calculate a regression meta-model featured by a good fit to the experimental data and a sort of stability of the prediction variance. Furthermore the design should allow to evaluate two experimental error components, namely pure error and lack of fit. Achieving this with the smallest number of experimental runs it is important to save experimentation time and costs. In classic RSM approach the Experimental Error (EE) is considered fixed and dependent only from the physical system under study. Several experimentations, though, have been performed on simulation models representing the physical system, as it isn't possible to experiment in a direct way. Complex industrial plants belong to this category because of the difficulty (i.e. time and investments required) to modify them in order to perform an experiment. In Discrete Event Stochastic Simulation (DEVS) the EE is simulation time or replications dependent. This requires a substantial change of approach for EE estimation. In this work Authors propose an improved approach and outline limitations of traditional ones.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11567/852896
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 6
  • ???jsp.display-item.citation.isi??? 2
social impact