Stephens, Darrin W; Fawell, Phillip D Process equipment optimisation using CFD and surrogate models Journal Article Progress in Computational Fluid Dynamics, an International Journal, 15 (2), pp. 102–113, 2015. Abstract | Links | BibTeX | Tags: Design Optimisation, Multi-Objective, Process Equipment, Raked Industrial Thickener, RBF, Sediment Transport, Surrogate Modelling, Surrogate models @article{stephens2015process,
title = {Process equipment optimisation using CFD and surrogate models},
author = {Darrin W Stephens and Phillip D Fawell},
doi = {10.1504/PCFD.2015.068818},
year = {2015},
date = {2015-01-01},
journal = {Progress in Computational Fluid Dynamics, an International Journal},
volume = {15},
number = {2},
pages = {102--113},
publisher = {Inderscience Publishers},
abstract = {CFD modelling offers powerful simulation capability, but only provides information for a given design and set of process conditions. While it allows comparison between different outcomes, CFD results alone do not provide an optimised outcome. In many design problems, thousands of function evaluations may be required to undertake an optimisation study. As a result, CFD models are often impractical for design optimisation. In contrast, surrogate models are compact and cheap to evaluate and can, therefore, be easily used for such tasks. This paper applies surrogate modelling techniques to a CFD model of sediment transport in a raked
industrial thickener. The surrogate is used to demonstrate single and multi-objective optimisations for the case study. For multi-objective problems, use from a practical design perspective is illustrated through four examples, in which the best design solution or operational response is sought to different process scenarios. },
keywords = {Design Optimisation, Multi-Objective, Process Equipment, Raked Industrial Thickener, RBF, Sediment Transport, Surrogate Modelling, Surrogate models},
pubstate = {published},
tppubtype = {article}
}
CFD modelling offers powerful simulation capability, but only provides information for a given design and set of process conditions. While it allows comparison between different outcomes, CFD results alone do not provide an optimised outcome. In many design problems, thousands of function evaluations may be required to undertake an optimisation study. As a result, CFD models are often impractical for design optimisation. In contrast, surrogate models are compact and cheap to evaluate and can, therefore, be easily used for such tasks. This paper applies surrogate modelling techniques to a CFD model of sediment transport in a raked
industrial thickener. The surrogate is used to demonstrate single and multi-objective optimisations for the case study. For multi-objective problems, use from a practical design perspective is illustrated through four examples, in which the best design solution or operational response is sought to different process scenarios. |
Stephens, Darrin W; Fawell, Phillip D Optimisation of process equipment using global surrogate models Conference Ninth International Conference on CFD in the Minerals and Process Industries, 2012. Abstract | Links | BibTeX | Tags: Design Optimisation, Global Surrogate Models, Optimisation, Process Equipment, RBF, Sediment Transport, Surrogate Modelling, Surrogate models @conference{stephens2012optimisation,
title = {Optimisation of process equipment using global surrogate models},
author = {Darrin W Stephens and Phillip D Fawell},
url = {http://www.researchgate.net/publication/266138725_OPTIMISATION_OF_PROCESS_EQUIPMENT_USING_GLOBAL_SURROGATE_MODELS},
year = {2012},
date = {2012-01-01},
booktitle = {Ninth International Conference on CFD in the Minerals and Process Industries},
abstract = {Computational Fluid Dynamics (CFD) modelling offers powerful simulation capability, but only provides information for a given design and set of process conditions. While it allows comparison between different outcomes, CFD results alone do not provide an optimised outcome. The computational cost associated with the use of high-fidelity CFD models poses a serious impediment to the successful application of optimisation algorithms in engineering design. Advances in hardware and parallel processing have reduced costs by orders of magnitude over the last few decades, but the fidelity with which engineers desire to model systems has also increased considerably. Evaluation of such models may take significant computational time for complex geometries. In many design problems, thousands of function evaluation may be required to undertake an optimisation study. As a result, CFD models are often impractical for design optimisation. In contrast, surrogate models are compact and cheap to evaluate (order of seconds or less) and can, therefore, be easily used for such tasks. This paper applies surrogate modelling techniques to a CFD model of sediment transport in a raked industrial thickener. The global surrogate produced using radial basis functions is used to demonstrate single and multi-objective optimisation for the case study. For multi-objective problem, use from a practical design perspective of the information contained in the set of optimal solutions is illustrated through four examples. NOMENCLATURE f function f function datâ f model of f k number of folds N integer number s real number x input sample points y true model output y mean true model output ! y predicted model output β RBF model parameters ! model error Subscripts i index INTRODUCTION},
keywords = {Design Optimisation, Global Surrogate Models, Optimisation, Process Equipment, RBF, Sediment Transport, Surrogate Modelling, Surrogate models},
pubstate = {published},
tppubtype = {conference}
}
Computational Fluid Dynamics (CFD) modelling offers powerful simulation capability, but only provides information for a given design and set of process conditions. While it allows comparison between different outcomes, CFD results alone do not provide an optimised outcome. The computational cost associated with the use of high-fidelity CFD models poses a serious impediment to the successful application of optimisation algorithms in engineering design. Advances in hardware and parallel processing have reduced costs by orders of magnitude over the last few decades, but the fidelity with which engineers desire to model systems has also increased considerably. Evaluation of such models may take significant computational time for complex geometries. In many design problems, thousands of function evaluation may be required to undertake an optimisation study. As a result, CFD models are often impractical for design optimisation. In contrast, surrogate models are compact and cheap to evaluate (order of seconds or less) and can, therefore, be easily used for such tasks. This paper applies surrogate modelling techniques to a CFD model of sediment transport in a raked industrial thickener. The global surrogate produced using radial basis functions is used to demonstrate single and multi-objective optimisation for the case study. For multi-objective problem, use from a practical design perspective of the information contained in the set of optimal solutions is illustrated through four examples. NOMENCLATURE f function f function datâ f model of f k number of folds N integer number s real number x input sample points y true model output y mean true model output ! y predicted model output β RBF model parameters ! model error Subscripts i index INTRODUCTION |
