The PSA analysis of a real plant represent a formidable com-putational task usually afforded either with an analytical approach based on the theory of the Markov chains or with a Monte Carlo simulation. In our opinion this latter methodology, thanks to its unique flexibility features, represents the only viable approach to the problem when time dependencies have impacts on the analy-sis: examples are time dependent transition rates (ageing), timing of the protection, control and safety systems, operator actions etc. Moreover the PSA analysis of a real plant demands taking into account the process variable dynamics when the evolution of the underlying physical process interacts with the system hardware configuration, e.g. when the process variables influence the failure rates or activate the protection systems. The inclusion of these dy-namic aspects dramatically burdens the analysis: a solution could be presently attempted only through short cuts to the solution of the deterministic equations governing the evolution of the process variables. In the present paper we consider the application of a multi-layered neural network for the solution of the mathematical mod-els related to the core behaviour of a PWR under varying thermal-hydraulic conditions. Since the neural network works very rapid-ly, this approach seems to be a good candidate for being included in a Monte Carlo dynamic PSA code which requires solving thou-sands of times the model equations relating to the different hard-ware plant configurations. Possible approximations thereby intro-duced could be tolerated if comparable with those following from the uncertainties of the stochastic parameters. The time reduction advantage is expected to increase when the future parallel com-puters become widely available.

Monte Carlo Approach to Dynamic PSA: Neural Solution of Equations Describing Core Transients

PADOVANI, ENRICO;RICOTTI, MARCO ENRICO
1996

Abstract

The PSA analysis of a real plant represent a formidable com-putational task usually afforded either with an analytical approach based on the theory of the Markov chains or with a Monte Carlo simulation. In our opinion this latter methodology, thanks to its unique flexibility features, represents the only viable approach to the problem when time dependencies have impacts on the analy-sis: examples are time dependent transition rates (ageing), timing of the protection, control and safety systems, operator actions etc. Moreover the PSA analysis of a real plant demands taking into account the process variable dynamics when the evolution of the underlying physical process interacts with the system hardware configuration, e.g. when the process variables influence the failure rates or activate the protection systems. The inclusion of these dy-namic aspects dramatically burdens the analysis: a solution could be presently attempted only through short cuts to the solution of the deterministic equations governing the evolution of the process variables. In the present paper we consider the application of a multi-layered neural network for the solution of the mathematical mod-els related to the core behaviour of a PWR under varying thermal-hydraulic conditions. Since the neural network works very rapid-ly, this approach seems to be a good candidate for being included in a Monte Carlo dynamic PSA code which requires solving thou-sands of times the model equations relating to the different hard-ware plant configurations. Possible approximations thereby intro-duced could be tolerated if comparable with those following from the uncertainties of the stochastic parameters. The time reduction advantage is expected to increase when the future parallel com-puters become widely available.
Proceedings of the ASME-JSME 4th International Conference on Nuclear Engineering 1966
9780791812266
File in questo prodotto:
File Dimensione Formato  
I4NEUUSL.pdf

accesso aperto

: Post-Print (DRAFT o Author’s Accepted Manuscript-AAM)
Dimensione 970.26 kB
Formato Adobe PDF
970.26 kB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11311/569437
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 3
  • ???jsp.display-item.citation.isi??? ND
social impact