You need JavaScript to view this

Transport-diffusion coupling for Candu reactor core follow-Up

Abstract

We couple the finite reactor diffusion code DONJON and the lattice code DRAGON, called for simplicity DD, to perform reactor follow-up calculations using a history-based approach. In order to do this, a new DD module is developed. This module manages the transfer of information between standard DONJON and DRAGON data structures. Moreover, it stores in a history data structure the global and local parameters required for cell calculations as well as the isotopic composition of the various materials present in each cell of the reactor. We then implement in DD a parallel algorithm to perform history-based Candu reactor calculations. Here, we assign to each processor a specific number of fuel channels to be analyzed. The DRAGON cell calculations for each of the fuel bundles associated with the specified channels are performed on the same processor in order to minimize communication time. Only the macroscopic cross section libraries are exchanged between the processor. Since the amount of data exchanged is relatively small, we expect to obtain an ideal speed-up. The coupling is tested for the analysis of a simplified Candu reactor model with 4 x 4 channels each containing 4 bundles. A 100 full-power days core tracking sequence with 16 refueling  More>>
Authors:
Varin, E; Marleau, G; Chambon, R [1] 
  1. Ecole Polytechnique de Montreal, IGN, Montreal (Canada)
Publication Date:
Jul 01, 2003
Product Type:
Conference
Report Number:
INIS-FR-2742
Resource Relation:
Conference: International conference on supercomputing in nuclear applications SNA'2003, Paris (France), 22-24 Sep 2003; Other Information: 18 refs; PBD: 2003
Subject:
21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS; CANDU TYPE REACTORS; COMPUTER CALCULATIONS; COMPUTERIZED SIMULATION; D CODES; PARALLEL PROCESSING; REACTOR CELLS; REACTOR FUELING; TASK SCHEDULING
OSTI ID:
20542450
Research Organizations:
CEA Saclay, 91 - Gif-sur-Yvette (France)
Country of Origin:
France
Language:
English
Other Identifying Numbers:
TRN: FR0401318106380
Availability:
Available from INIS in electronic form
Submitting Site:
FRN
Size:
11 pages
Announcement Date:
Dec 24, 2004

Citation Formats

Varin, E, Marleau, G, and Chambon, R. Transport-diffusion coupling for Candu reactor core follow-Up. France: N. p., 2003. Web.
Varin, E, Marleau, G, & Chambon, R. Transport-diffusion coupling for Candu reactor core follow-Up. France.
Varin, E, Marleau, G, and Chambon, R. 2003. "Transport-diffusion coupling for Candu reactor core follow-Up." France.
@misc{etde_20542450,
title = {Transport-diffusion coupling for Candu reactor core follow-Up}
author = {Varin, E, Marleau, G, and Chambon, R}
abstractNote = {We couple the finite reactor diffusion code DONJON and the lattice code DRAGON, called for simplicity DD, to perform reactor follow-up calculations using a history-based approach. In order to do this, a new DD module is developed. This module manages the transfer of information between standard DONJON and DRAGON data structures. Moreover, it stores in a history data structure the global and local parameters required for cell calculations as well as the isotopic composition of the various materials present in each cell of the reactor. We then implement in DD a parallel algorithm to perform history-based Candu reactor calculations. Here, we assign to each processor a specific number of fuel channels to be analyzed. The DRAGON cell calculations for each of the fuel bundles associated with the specified channels are performed on the same processor in order to minimize communication time. Only the macroscopic cross section libraries are exchanged between the processor. Since the amount of data exchanged is relatively small, we expect to obtain an ideal speed-up. The coupling is tested for the analysis of a simplified Candu reactor model with 4 x 4 channels each containing 4 bundles. A 100 full-power days core tracking sequence with 16 refueling steps is studied. Results are coherent with those obtained using more approximate approaches. Parallel speed-up is near optimal indicating that the use of this approach for more realistic reactor calculations should be pursued. (authors)}
place = {France}
year = {2003}
month = {Jul}
}