Response surface methodolgy

Research output: Chapter in Book/Report/Conference proceedingChapterScientificpeer-review

Abstract

This chapter first summarizes Response Surface Methodology (RSM), which started with Box and Wilson’s 1951 article on RSM for real, non-simulated systems. RSM is a stepwise heuristic that uses first-order polynomials to approximate the response surface locally. An estimated polynomial metamodel gives an estimated local gradient, which RSM uses in steepest ascent (or descent) to decide on the next local experiment. When RSM approaches the optimum, the latest first-order polynomial is replaced by a second-order polynomial. The fitted second-order polynomial enables the estimation of the optimum. This chapter then focuses on simulated systems, which may violate the assumptions of constant variance and independence. A variant of RSM that provably converges to the true optimum under specific conditions is summarized, and an adapted steepest ascent that is scale-independent is presented. Next, the chapter generalizes RSM to multiple random responses, selecting one response as the goal variable and the other responses as the constrained variables. This generalized RSM is combined with mathematical programming to estimate a better search direction than the steepest ascent direction. To test whether the estimated solution is indeed optimal, bootstrapping may be used. Finally, the chapter discusses robust optimization of the decision variables, while accounting for uncertainties in the environmental variables.
Original languageEnglish
Title of host publicationHandbook of Simulation Optimization
EditorsMichael C. Fu
PublisherSpringer Verlag New York
Pages81-104
Volume216
Edition1
ISBN (Electronic)9781493913848
ISBN (Print)9781493913831
DOIs
Publication statusPublished - 2015

Publication series

NameInternational Series in Operations Research & Management Science
PublisherSpringer-Verlag New York
Volume216
ISSN (Print)0884-8289

Fingerprint

Response surface
Response surface methodology
Polynomials
Uncertainty
Metamodel
Robust optimization
Gradient
Experiment
Heuristics
Bootstrapping
Mathematical programming

Cite this

Kleijnen, J. P. C. (2015). Response surface methodolgy. In M. C. Fu (Ed.), Handbook of Simulation Optimization (1 ed., Vol. 216, pp. 81-104). (International Series in Operations Research & Management Science; Vol. 216). Springer Verlag New York. https://doi.org/10.1007/978-1-4939-1384-8
Kleijnen, J.P.C. / Response surface methodolgy. Handbook of Simulation Optimization. editor / Michael C. Fu. Vol. 216 1. ed. Springer Verlag New York, 2015. pp. 81-104 (International Series in Operations Research & Management Science).
@inbook{612af66208984442a92fa563c7053ef1,
title = "Response surface methodolgy",
abstract = "This chapter first summarizes Response Surface Methodology (RSM), which started with Box and Wilson’s 1951 article on RSM for real, non-simulated systems. RSM is a stepwise heuristic that uses first-order polynomials to approximate the response surface locally. An estimated polynomial metamodel gives an estimated local gradient, which RSM uses in steepest ascent (or descent) to decide on the next local experiment. When RSM approaches the optimum, the latest first-order polynomial is replaced by a second-order polynomial. The fitted second-order polynomial enables the estimation of the optimum. This chapter then focuses on simulated systems, which may violate the assumptions of constant variance and independence. A variant of RSM that provably converges to the true optimum under specific conditions is summarized, and an adapted steepest ascent that is scale-independent is presented. Next, the chapter generalizes RSM to multiple random responses, selecting one response as the goal variable and the other responses as the constrained variables. This generalized RSM is combined with mathematical programming to estimate a better search direction than the steepest ascent direction. To test whether the estimated solution is indeed optimal, bootstrapping may be used. Finally, the chapter discusses robust optimization of the decision variables, while accounting for uncertainties in the environmental variables.",
author = "J.P.C. Kleijnen",
year = "2015",
doi = "10.1007/978-1-4939-1384-8",
language = "English",
isbn = "9781493913831",
volume = "216",
series = "International Series in Operations Research & Management Science",
publisher = "Springer Verlag New York",
pages = "81--104",
editor = "Fu, {Michael C.}",
booktitle = "Handbook of Simulation Optimization",
edition = "1",

}

Kleijnen, JPC 2015, Response surface methodolgy. in MC Fu (ed.), Handbook of Simulation Optimization. 1 edn, vol. 216, International Series in Operations Research & Management Science, vol. 216, Springer Verlag New York, pp. 81-104. https://doi.org/10.1007/978-1-4939-1384-8

Response surface methodolgy. / Kleijnen, J.P.C.

Handbook of Simulation Optimization. ed. / Michael C. Fu. Vol. 216 1. ed. Springer Verlag New York, 2015. p. 81-104 (International Series in Operations Research & Management Science; Vol. 216).

Research output: Chapter in Book/Report/Conference proceedingChapterScientificpeer-review

TY - CHAP

T1 - Response surface methodolgy

AU - Kleijnen, J.P.C.

PY - 2015

Y1 - 2015

N2 - This chapter first summarizes Response Surface Methodology (RSM), which started with Box and Wilson’s 1951 article on RSM for real, non-simulated systems. RSM is a stepwise heuristic that uses first-order polynomials to approximate the response surface locally. An estimated polynomial metamodel gives an estimated local gradient, which RSM uses in steepest ascent (or descent) to decide on the next local experiment. When RSM approaches the optimum, the latest first-order polynomial is replaced by a second-order polynomial. The fitted second-order polynomial enables the estimation of the optimum. This chapter then focuses on simulated systems, which may violate the assumptions of constant variance and independence. A variant of RSM that provably converges to the true optimum under specific conditions is summarized, and an adapted steepest ascent that is scale-independent is presented. Next, the chapter generalizes RSM to multiple random responses, selecting one response as the goal variable and the other responses as the constrained variables. This generalized RSM is combined with mathematical programming to estimate a better search direction than the steepest ascent direction. To test whether the estimated solution is indeed optimal, bootstrapping may be used. Finally, the chapter discusses robust optimization of the decision variables, while accounting for uncertainties in the environmental variables.

AB - This chapter first summarizes Response Surface Methodology (RSM), which started with Box and Wilson’s 1951 article on RSM for real, non-simulated systems. RSM is a stepwise heuristic that uses first-order polynomials to approximate the response surface locally. An estimated polynomial metamodel gives an estimated local gradient, which RSM uses in steepest ascent (or descent) to decide on the next local experiment. When RSM approaches the optimum, the latest first-order polynomial is replaced by a second-order polynomial. The fitted second-order polynomial enables the estimation of the optimum. This chapter then focuses on simulated systems, which may violate the assumptions of constant variance and independence. A variant of RSM that provably converges to the true optimum under specific conditions is summarized, and an adapted steepest ascent that is scale-independent is presented. Next, the chapter generalizes RSM to multiple random responses, selecting one response as the goal variable and the other responses as the constrained variables. This generalized RSM is combined with mathematical programming to estimate a better search direction than the steepest ascent direction. To test whether the estimated solution is indeed optimal, bootstrapping may be used. Finally, the chapter discusses robust optimization of the decision variables, while accounting for uncertainties in the environmental variables.

U2 - 10.1007/978-1-4939-1384-8

DO - 10.1007/978-1-4939-1384-8

M3 - Chapter

SN - 9781493913831

VL - 216

T3 - International Series in Operations Research & Management Science

SP - 81

EP - 104

BT - Handbook of Simulation Optimization

A2 - Fu, Michael C.

PB - Springer Verlag New York

ER -

Kleijnen JPC. Response surface methodolgy. In Fu MC, editor, Handbook of Simulation Optimization. 1 ed. Vol. 216. Springer Verlag New York. 2015. p. 81-104. (International Series in Operations Research & Management Science). https://doi.org/10.1007/978-1-4939-1384-8