Combustion optimization of ultra supercritical boiler based on artificial intelligence

Yan Shi, Wenqi Zhong, Xi Chen, A. B. Yu, Jie Li

Research output: Contribution to journalArticleResearchpeer-review

Abstract

A method for optimizing the combustion in an ultra-supercritical boiler is developed and evaluated in a 660 MWe ultra-supercritical coal fired power plant. In this method, Artificial Neural Networks (ANN) models are established for predicting the boiler operating and emission properties. To enhance the generalization of the ANN models, Computational Fluid Dynamics (CFD) simulation is performed to generate some data as training samples for ANN modeling, together with the historical operating data. The inputs of the ANN models are unit load, coal properties, excess air and air distribution scheme, and the outputs are thermal efficiency and NOx emission. Based on the ANN models, Genetic Algorithm (GA) is used to optimize the air distribution scheme to achieve a higher thermal efficiency and lower NOx emission simultaneously. The predictions of the thermal efficiency and NOx emissions show a good agreement with the plant data, with mean errors of 0.04% and 3.56 mg/Nm 3 , respectively. The results indicate that the use of CFD data can help generalize the ANN models. The application to a practical plant demonstrates that the proposed approach provides an effective tool for multi-objective optimization of pulverized-coal boiler performance with improved thermal efficiency and NOx emission control.

Original languageEnglish
Pages (from-to)804-817
Number of pages14
JournalEnergy
Volume170
DOIs
Publication statusPublished - 1 Mar 2019

Keywords

  • ANN-GA
  • CFD simulation
  • Coal combustion
  • Multi-objective optimization

Cite this

Shi, Yan ; Zhong, Wenqi ; Chen, Xi ; Yu, A. B. ; Li, Jie. / Combustion optimization of ultra supercritical boiler based on artificial intelligence. In: Energy. 2019 ; Vol. 170. pp. 804-817.
@article{f37b8300912c4f4bb86c660c73c4fc0b,
title = "Combustion optimization of ultra supercritical boiler based on artificial intelligence",
abstract = "A method for optimizing the combustion in an ultra-supercritical boiler is developed and evaluated in a 660 MWe ultra-supercritical coal fired power plant. In this method, Artificial Neural Networks (ANN) models are established for predicting the boiler operating and emission properties. To enhance the generalization of the ANN models, Computational Fluid Dynamics (CFD) simulation is performed to generate some data as training samples for ANN modeling, together with the historical operating data. The inputs of the ANN models are unit load, coal properties, excess air and air distribution scheme, and the outputs are thermal efficiency and NOx emission. Based on the ANN models, Genetic Algorithm (GA) is used to optimize the air distribution scheme to achieve a higher thermal efficiency and lower NOx emission simultaneously. The predictions of the thermal efficiency and NOx emissions show a good agreement with the plant data, with mean errors of 0.04{\%} and 3.56 mg/Nm 3 , respectively. The results indicate that the use of CFD data can help generalize the ANN models. The application to a practical plant demonstrates that the proposed approach provides an effective tool for multi-objective optimization of pulverized-coal boiler performance with improved thermal efficiency and NOx emission control.",
keywords = "ANN-GA, CFD simulation, Coal combustion, Multi-objective optimization",
author = "Yan Shi and Wenqi Zhong and Xi Chen and Yu, {A. B.} and Jie Li",
year = "2019",
month = "3",
day = "1",
doi = "10.1016/j.energy.2018.12.172",
language = "English",
volume = "170",
pages = "804--817",
journal = "Energy",
issn = "0360-5442",
publisher = "Elsevier",

}

Combustion optimization of ultra supercritical boiler based on artificial intelligence. / Shi, Yan; Zhong, Wenqi; Chen, Xi; Yu, A. B.; Li, Jie.

In: Energy, Vol. 170, 01.03.2019, p. 804-817.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Combustion optimization of ultra supercritical boiler based on artificial intelligence

AU - Shi, Yan

AU - Zhong, Wenqi

AU - Chen, Xi

AU - Yu, A. B.

AU - Li, Jie

PY - 2019/3/1

Y1 - 2019/3/1

N2 - A method for optimizing the combustion in an ultra-supercritical boiler is developed and evaluated in a 660 MWe ultra-supercritical coal fired power plant. In this method, Artificial Neural Networks (ANN) models are established for predicting the boiler operating and emission properties. To enhance the generalization of the ANN models, Computational Fluid Dynamics (CFD) simulation is performed to generate some data as training samples for ANN modeling, together with the historical operating data. The inputs of the ANN models are unit load, coal properties, excess air and air distribution scheme, and the outputs are thermal efficiency and NOx emission. Based on the ANN models, Genetic Algorithm (GA) is used to optimize the air distribution scheme to achieve a higher thermal efficiency and lower NOx emission simultaneously. The predictions of the thermal efficiency and NOx emissions show a good agreement with the plant data, with mean errors of 0.04% and 3.56 mg/Nm 3 , respectively. The results indicate that the use of CFD data can help generalize the ANN models. The application to a practical plant demonstrates that the proposed approach provides an effective tool for multi-objective optimization of pulverized-coal boiler performance with improved thermal efficiency and NOx emission control.

AB - A method for optimizing the combustion in an ultra-supercritical boiler is developed and evaluated in a 660 MWe ultra-supercritical coal fired power plant. In this method, Artificial Neural Networks (ANN) models are established for predicting the boiler operating and emission properties. To enhance the generalization of the ANN models, Computational Fluid Dynamics (CFD) simulation is performed to generate some data as training samples for ANN modeling, together with the historical operating data. The inputs of the ANN models are unit load, coal properties, excess air and air distribution scheme, and the outputs are thermal efficiency and NOx emission. Based on the ANN models, Genetic Algorithm (GA) is used to optimize the air distribution scheme to achieve a higher thermal efficiency and lower NOx emission simultaneously. The predictions of the thermal efficiency and NOx emissions show a good agreement with the plant data, with mean errors of 0.04% and 3.56 mg/Nm 3 , respectively. The results indicate that the use of CFD data can help generalize the ANN models. The application to a practical plant demonstrates that the proposed approach provides an effective tool for multi-objective optimization of pulverized-coal boiler performance with improved thermal efficiency and NOx emission control.

KW - ANN-GA

KW - CFD simulation

KW - Coal combustion

KW - Multi-objective optimization

UR - http://www.scopus.com/inward/record.url?scp=85060870879&partnerID=8YFLogxK

U2 - 10.1016/j.energy.2018.12.172

DO - 10.1016/j.energy.2018.12.172

M3 - Article

VL - 170

SP - 804

EP - 817

JO - Energy

JF - Energy

SN - 0360-5442

ER -