Automatic differentiation approach for solving one-dimensional flow and heat transfer problems. (15th September 2021)
- Record Type:
- Journal Article
- Title:
- Automatic differentiation approach for solving one-dimensional flow and heat transfer problems. (15th September 2021)
- Main Title:
- Automatic differentiation approach for solving one-dimensional flow and heat transfer problems
- Authors:
- Niu, Yuhang
He, Yanan
Xiang, Fengrui
Zhang, Jing
Wu, Yingwei
Tian, Wenxi
Su, Guanghui
Qiu, Suizheng - Abstract:
- Highlights: High-order spatial and temporal discretization was implemented in single-phase and five-equation two-phase flow models. The ADOO method was used to form the Jacobian matrix automatically. Effects of ADOO on convergence performance and calculation accuracy were investigated. A certain amount of experimental data and numerical results using RELAP5 were adopted for validation and verification in two-phase problems. Abstract: Traditional reactor system analysis tools are confronted with challenging difficulties in model development, low accuracy, and poor convergence. To solve these problems, the automatic differentiation (AD) method that allows the automatic numerical calculation of derivatives of functions was adopted to develop the reactor system code in this paper. For the simulation of single-phase models, the steady and transient responses were presented to investigate the effects of the spatial and temporal discretization schemes on modeling accuracy and efficiency. Meanwhile, the comparison of convergence performance between the automatic differentiation using operator overloading (ADOO) and the traditional hand-coded method was completed. Further, in the case of two-phase flow problems, the high-order discrete schemes were applied in this code. It was demonstrated that the reactor system code with single-phase and five-equation two-phase flow models, which adopted the high-order discretization and the ADOO method, performed very well for one-dimensional flowHighlights: High-order spatial and temporal discretization was implemented in single-phase and five-equation two-phase flow models. The ADOO method was used to form the Jacobian matrix automatically. Effects of ADOO on convergence performance and calculation accuracy were investigated. A certain amount of experimental data and numerical results using RELAP5 were adopted for validation and verification in two-phase problems. Abstract: Traditional reactor system analysis tools are confronted with challenging difficulties in model development, low accuracy, and poor convergence. To solve these problems, the automatic differentiation (AD) method that allows the automatic numerical calculation of derivatives of functions was adopted to develop the reactor system code in this paper. For the simulation of single-phase models, the steady and transient responses were presented to investigate the effects of the spatial and temporal discretization schemes on modeling accuracy and efficiency. Meanwhile, the comparison of convergence performance between the automatic differentiation using operator overloading (ADOO) and the traditional hand-coded method was completed. Further, in the case of two-phase flow problems, the high-order discrete schemes were applied in this code. It was demonstrated that the reactor system code with single-phase and five-equation two-phase flow models, which adopted the high-order discretization and the ADOO method, performed very well for one-dimensional flow and heat transfer problems. … (more)
- Is Part Of:
- Annals of nuclear energy. Volume 160(2021)
- Journal:
- Annals of nuclear energy
- Issue:
- Volume 160(2021)
- Issue Display:
- Volume 160, Issue 2021 (2021)
- Year:
- 2021
- Volume:
- 160
- Issue:
- 2021
- Issue Sort Value:
- 2021-0160-2021-0000
- Page Start:
- Page End:
- Publication Date:
- 2021-09-15
- Subjects:
- Automatic differentiation -- Single-phase and two-phase flow -- High-order discretization -- Flow and heat transfer
Nuclear energy -- Periodicals
Nuclear engineering -- Periodicals
621.4805 - Journal URLs:
- http://www.sciencedirect.com/science/journal/03064549 ↗
http://catalog.hathitrust.org/api/volumes/oclc/2243298.html ↗
http://www.elsevier.com/journals ↗ - DOI:
- 10.1016/j.anucene.2021.108361 ↗
- Languages:
- English
- ISSNs:
- 0306-4549
- Deposit Type:
- Legaldeposit
- View Content:
- Available online (eLD content is only available in our Reading Rooms) ↗
- Physical Locations:
- British Library DSC - 1043.150000
British Library DSC - BLDSS-3PM
British Library HMNTS - ELD Digital store - Ingest File:
- 17261.xml