Simulation experiments for maximising the availability of a commercial octene production facility

Abstract

Overall availability of a chemical process is of critical importance in industry. In this paper we evaluate the process design factors that infuence the availability of a new chemical production facility by performing computer experiments on a stochastic simulation model. Experimental designs commonly used in the Design and Analysis of Computer Experiments (DACE) and Classical Design of Experiments (DOE) are evaluated and compared for application by means of simulation experiments. Furthermore, response surface and kriging models are evaluated for the approximation of the input-output relationships. The most accurate experimental design by approximation model combination is used to explore the design space, both in terms of the overall availability and the percentage time offine. We illustrate how the design and analysis of simulation experiments (DASE) are used for minimizing the risks in the design of a new 1-octene production facility in terms of maximising the overall availability and minimizing the percentage time offline simultaneously.

Key words: Availability, design and analysis of computer experiments, kriging, response surface modelling, simulation experiments.