Genetic programming-based development of thermal runaway criteria

Alex Kummer, Tamás Varga, J. Abonyi

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Common thermal runaway criteria (e.g., divergence criterion and the Maxi criterion) may predict a thermal runaway unreasonably as the Maximum Allowable Temperature (MAT) is not taken into account. This contribution proposes a method for the goal-oriented construction of reactor runaway criteria by Genetic Programming (GP). The runaway prediction problem is formulated as a critical equation-based classification task, and GP is used to identify the optimal structure of the equations that also take into account the MAT. To demonstrate the applicability of the method, tailored criteria were developed for batch and continuous stirred-tank reactors. The resultant critical equations outperform the well-known criteria in terms of the early and accurate indication of thermal runaways.

Original languageEnglish
Article number106582
JournalComputers and Chemical Engineering
Volume131
DOIs
Publication statusPublished - Dec 5 2019

Fingerprint

Genetic programming
Temperature
Hot Temperature

Keywords

  • Criterion identification
  • Early warning
  • Equation discovery
  • Process safety
  • Safety criteria

ASJC Scopus subject areas

  • Chemical Engineering(all)
  • Computer Science Applications

Cite this

Genetic programming-based development of thermal runaway criteria. / Kummer, Alex; Varga, Tamás; Abonyi, J.

In: Computers and Chemical Engineering, Vol. 131, 106582, 05.12.2019.

Research output: Contribution to journalArticle

@article{f61b39fa3e784cef835f4f40b6e6942f,
title = "Genetic programming-based development of thermal runaway criteria",
abstract = "Common thermal runaway criteria (e.g., divergence criterion and the Maxi criterion) may predict a thermal runaway unreasonably as the Maximum Allowable Temperature (MAT) is not taken into account. This contribution proposes a method for the goal-oriented construction of reactor runaway criteria by Genetic Programming (GP). The runaway prediction problem is formulated as a critical equation-based classification task, and GP is used to identify the optimal structure of the equations that also take into account the MAT. To demonstrate the applicability of the method, tailored criteria were developed for batch and continuous stirred-tank reactors. The resultant critical equations outperform the well-known criteria in terms of the early and accurate indication of thermal runaways.",
keywords = "Criterion identification, Early warning, Equation discovery, Process safety, Safety criteria",
author = "Alex Kummer and Tam{\'a}s Varga and J. Abonyi",
year = "2019",
month = "12",
day = "5",
doi = "10.1016/j.compchemeng.2019.106582",
language = "English",
volume = "131",
journal = "Computers and Chemical Engineering",
issn = "0098-1354",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - Genetic programming-based development of thermal runaway criteria

AU - Kummer, Alex

AU - Varga, Tamás

AU - Abonyi, J.

PY - 2019/12/5

Y1 - 2019/12/5

N2 - Common thermal runaway criteria (e.g., divergence criterion and the Maxi criterion) may predict a thermal runaway unreasonably as the Maximum Allowable Temperature (MAT) is not taken into account. This contribution proposes a method for the goal-oriented construction of reactor runaway criteria by Genetic Programming (GP). The runaway prediction problem is formulated as a critical equation-based classification task, and GP is used to identify the optimal structure of the equations that also take into account the MAT. To demonstrate the applicability of the method, tailored criteria were developed for batch and continuous stirred-tank reactors. The resultant critical equations outperform the well-known criteria in terms of the early and accurate indication of thermal runaways.

AB - Common thermal runaway criteria (e.g., divergence criterion and the Maxi criterion) may predict a thermal runaway unreasonably as the Maximum Allowable Temperature (MAT) is not taken into account. This contribution proposes a method for the goal-oriented construction of reactor runaway criteria by Genetic Programming (GP). The runaway prediction problem is formulated as a critical equation-based classification task, and GP is used to identify the optimal structure of the equations that also take into account the MAT. To demonstrate the applicability of the method, tailored criteria were developed for batch and continuous stirred-tank reactors. The resultant critical equations outperform the well-known criteria in terms of the early and accurate indication of thermal runaways.

KW - Criterion identification

KW - Early warning

KW - Equation discovery

KW - Process safety

KW - Safety criteria

UR - http://www.scopus.com/inward/record.url?scp=85073065555&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85073065555&partnerID=8YFLogxK

U2 - 10.1016/j.compchemeng.2019.106582

DO - 10.1016/j.compchemeng.2019.106582

M3 - Article

AN - SCOPUS:85073065555

VL - 131

JO - Computers and Chemical Engineering

JF - Computers and Chemical Engineering

SN - 0098-1354

M1 - 106582

ER -