Comprehensive experimental design for chemical engineering processes: a two-layer iterative design approach

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

A systematic framework for optimal experimental design (OED) of multiple experimental factors is proposed to support data collection in chemical engineering systems with the purpose to obtain the most informative data for modeling. The structural identifiability is firstly investigated through a combined procedure with the generating series method and the identifiability tableau. Next the parameter estimability is analyzed via the orthogonalized sensitivity analysis in order to identify crucial and identifiable model parameters. Traditionally OED treats separate problems such as the choice of input conditions, the selection of variables to measure, and the design of sampling time profile. A new OED strategy is proposed that optimizes these interdependent factors in one framework. An iterative two-layer design structure is developed. In the lower layer for observation design, the sampling profile and the measurement set selection are combined and formulated as a single integrated observation design problem, which is relaxed to a convex optimization problem that can be solved with a local method. Thus the measurement set selection and the sampling profile can be determined simultaneously. In the upper layer for input design, the optimization of input intensities is obtained through stochastic global searching. In this way, the multi-factor optimization problem is solved through the integration of a stochastic method, for the upper layer, and a deterministic method, for the lower layer. Case studies are conducted on two biochemical systems with different complexities, one is an enzyme kinetically controlled synthesis system and the other one is a lab-scale enzymatic biodiesel production system. Numerical results demonstrate the effectiveness of this double-layer OED optimization strategy in reducing parameter estimation uncertainties compared with conventional approaches.
LanguageEnglish
Pages135-153
Number of pages19
JournalChemical Engineering Science
Volume189
Early online date25 May 2018
DOIs
Publication statusPublished - 2 Nov 2018

Fingerprint

Chemical engineering
Design of experiments
Sampling
Biofuels
Convex optimization
Biodiesel
Parameter estimation
Sensitivity analysis
Enzymes

Keywords

  • optimal experimental design (OED)
  • multi-factor optimization
  • input conditions
  • sampling time profile
  • measurement set selection
  • chemical reaction systems

Cite this

@article{4826fc38b0234c8da63138685b0200a0,
title = "Comprehensive experimental design for chemical engineering processes: a two-layer iterative design approach",
abstract = "A systematic framework for optimal experimental design (OED) of multiple experimental factors is proposed to support data collection in chemical engineering systems with the purpose to obtain the most informative data for modeling. The structural identifiability is firstly investigated through a combined procedure with the generating series method and the identifiability tableau. Next the parameter estimability is analyzed via the orthogonalized sensitivity analysis in order to identify crucial and identifiable model parameters. Traditionally OED treats separate problems such as the choice of input conditions, the selection of variables to measure, and the design of sampling time profile. A new OED strategy is proposed that optimizes these interdependent factors in one framework. An iterative two-layer design structure is developed. In the lower layer for observation design, the sampling profile and the measurement set selection are combined and formulated as a single integrated observation design problem, which is relaxed to a convex optimization problem that can be solved with a local method. Thus the measurement set selection and the sampling profile can be determined simultaneously. In the upper layer for input design, the optimization of input intensities is obtained through stochastic global searching. In this way, the multi-factor optimization problem is solved through the integration of a stochastic method, for the upper layer, and a deterministic method, for the lower layer. Case studies are conducted on two biochemical systems with different complexities, one is an enzyme kinetically controlled synthesis system and the other one is a lab-scale enzymatic biodiesel production system. Numerical results demonstrate the effectiveness of this double-layer OED optimization strategy in reducing parameter estimation uncertainties compared with conventional approaches.",
keywords = "optimal experimental design (OED), multi-factor optimization, input conditions, sampling time profile, measurement set selection, chemical reaction systems",
author = "Hui Yu and Hong Yue and Peter Halling",
year = "2018",
month = "11",
day = "2",
doi = "10.1016/j.ces.2018.05.047",
language = "English",
volume = "189",
pages = "135--153",
journal = "Chemical Engineering Science",
issn = "0009-2509",

}

TY - JOUR

T1 - Comprehensive experimental design for chemical engineering processes

T2 - Chemical Engineering Science

AU - Yu, Hui

AU - Yue, Hong

AU - Halling, Peter

PY - 2018/11/2

Y1 - 2018/11/2

N2 - A systematic framework for optimal experimental design (OED) of multiple experimental factors is proposed to support data collection in chemical engineering systems with the purpose to obtain the most informative data for modeling. The structural identifiability is firstly investigated through a combined procedure with the generating series method and the identifiability tableau. Next the parameter estimability is analyzed via the orthogonalized sensitivity analysis in order to identify crucial and identifiable model parameters. Traditionally OED treats separate problems such as the choice of input conditions, the selection of variables to measure, and the design of sampling time profile. A new OED strategy is proposed that optimizes these interdependent factors in one framework. An iterative two-layer design structure is developed. In the lower layer for observation design, the sampling profile and the measurement set selection are combined and formulated as a single integrated observation design problem, which is relaxed to a convex optimization problem that can be solved with a local method. Thus the measurement set selection and the sampling profile can be determined simultaneously. In the upper layer for input design, the optimization of input intensities is obtained through stochastic global searching. In this way, the multi-factor optimization problem is solved through the integration of a stochastic method, for the upper layer, and a deterministic method, for the lower layer. Case studies are conducted on two biochemical systems with different complexities, one is an enzyme kinetically controlled synthesis system and the other one is a lab-scale enzymatic biodiesel production system. Numerical results demonstrate the effectiveness of this double-layer OED optimization strategy in reducing parameter estimation uncertainties compared with conventional approaches.

AB - A systematic framework for optimal experimental design (OED) of multiple experimental factors is proposed to support data collection in chemical engineering systems with the purpose to obtain the most informative data for modeling. The structural identifiability is firstly investigated through a combined procedure with the generating series method and the identifiability tableau. Next the parameter estimability is analyzed via the orthogonalized sensitivity analysis in order to identify crucial and identifiable model parameters. Traditionally OED treats separate problems such as the choice of input conditions, the selection of variables to measure, and the design of sampling time profile. A new OED strategy is proposed that optimizes these interdependent factors in one framework. An iterative two-layer design structure is developed. In the lower layer for observation design, the sampling profile and the measurement set selection are combined and formulated as a single integrated observation design problem, which is relaxed to a convex optimization problem that can be solved with a local method. Thus the measurement set selection and the sampling profile can be determined simultaneously. In the upper layer for input design, the optimization of input intensities is obtained through stochastic global searching. In this way, the multi-factor optimization problem is solved through the integration of a stochastic method, for the upper layer, and a deterministic method, for the lower layer. Case studies are conducted on two biochemical systems with different complexities, one is an enzyme kinetically controlled synthesis system and the other one is a lab-scale enzymatic biodiesel production system. Numerical results demonstrate the effectiveness of this double-layer OED optimization strategy in reducing parameter estimation uncertainties compared with conventional approaches.

KW - optimal experimental design (OED)

KW - multi-factor optimization

KW - input conditions

KW - sampling time profile

KW - measurement set selection

KW - chemical reaction systems

UR - https://www.journals.elsevier.com/chemical-engineering-science/

U2 - 10.1016/j.ces.2018.05.047

DO - 10.1016/j.ces.2018.05.047

M3 - Article

VL - 189

SP - 135

EP - 153

JO - Chemical Engineering Science

JF - Chemical Engineering Science

SN - 0009-2509

ER -