Bottlenecks in parallel algorithms for power system stability analysis

Chai, J.S.; Bose, Anjan

doi:10.1109/59.221242

Cited by 92 publications

(32 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For power system dynamic simulations, the popular VDHN scheme has been suggested as benchmark [31]. This consists of solving at each Newton iteration the integrated linear system stemming from the model (1), with an infrequently updated Jacobian.…”

Section: B Performance Indicesmentioning

confidence: 99%

Power System Dynamic Simulations Using a Parallel Two-Level Schur-Complement Decomposition

Aristidou

Lebeau²,

Cutsem

2016

IEEE Trans. Power Syst.

View full text Add to dashboard Cite

Abstract-As the need for faster power system dynamic simulations increases, it is essential to develop new algorithms that exploit parallel computing to accelerate those simulations. This paper proposes a parallel algorithm based on a two-level, Schurcomplement-based, domain decomposition method. The twolevel partitioning provides high parallelization potential (coarseand fine-grained). In addition, due to the Schur-complement approach used to update the sub-domain interface variables, the algorithm exhibits high global convergence rate. Finally, it provides significant numerical and computational acceleration. The algorithm is implemented using the shared-memory parallel programming model, targeting inexpensive multi-core machines. Its performance is reported on a real system as well as on a large test system combining transmission and distribution networks.

show abstract

Section: B Performance Indicesmentioning

confidence: 99%

Power System Dynamic Simulations Using a Parallel Two-Level Schur-Complement Decomposition

Aristidou

Lebeau²,

Cutsem

2016

IEEE Trans. Power Syst.

View full text Add to dashboard Cite

show abstract

“…Minimum Degree ordering is greedy algorithm that reorders the rows/columns of symmetric sparse matrix so that the row/column with fewest non-zero elements at a given factorization stage is the next one to be eliminated [7]. Physical reordering of the matrix wastes time, so Markowitz ordering strategy is actually to choose pivots with the added constraint of minimum fill-in.…”

Section: Methodology: Lu Factorization Based On Minimum Degree Orderingmentioning

confidence: 99%

LU Factorization Algorithm with Minimum Degree Ordering in Power Distribution Network Problems

Selman

2016

scjournal

View full text Add to dashboard Cite

Power systems computations for nowadays common large distributed systems typically involve the usage of very large sparse matrices, whose analysis and verification is very time and memory consuming. When blocked, sparse matrices can be processed much more efficiently, and this made blocked sparse matrices widely used in acquiring solutions for power system problems. The established sparse matrix storage and reordering techniques however do not fully utilize the existing computer architecture, thus search for efficient sparse system solution is ongoing. This paper presents adjustments of well-known LU factorization algorithm suitable for use in power distribution network applications. LU factorization algorithm processes data in blocks and uses minimum degree ordering to accelerate the computations.

show abstract

“…Some methods, like parallel VDHN [12], Newton W-matrix [13] and parallel LU [14], divide the independent vector and matrix operations involved in the linear system solution over the available computing units. Other methods, like parallel successive over relaxed Newton [15] and MaclaurinNewton [12], use an approximate (relaxed) Jacobian matrix with more convenient structure for parallelization.…”

Section: B2 Fine-grained Parallel Methodsmentioning

confidence: 99%

Dynamic Simulation of Large-Scale Power Systems Using a Parallel Schur-Complement-Based Decomposition Method

Aristidou

Fabozzi

Cutsem

2014

IEEE Trans. Parallel Distrib. Syst.

116

View full text Add to dashboard Cite

Article:Aristidou, P orcid.org/0000-0003- 4429-0225, Fabozzi, D and Van Cutsem, T (2014) Dynamic simulation of large-scale power systems using a parallel schur-complement-based decomposition method. IEEE Transactions on Parallel and Distributed Systems, 25 (10). pp. 2561 -2570 . ISSN 1045 https://doi.org/10.1109/TPDS.2013.252 © 2013 IEEE. This is an author produced version of a paper published in IEEE Transactions on Parallel and Distributed Systems. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Uploaded in accordance with the publisher's self-archiving policy.eprints@whiterose.ac.uk https://eprints.whiterose.ac.uk/ Reuse Unless indicated otherwise, fulltext items are protected by copyright with all rights reserved. The copyright exception in section 29 of the Copyright, Designs and Patents Act 1988 allows the making of a single copy solely for the purpose of non-commercial research or private study within the limits of fair dealing. The publisher or other rights-holder may allow further reproduction and re-use of this version -refer to the White Rose Research Online record for this item. Where records identify the publisher as the copyright holder, users can verify any specific terms of use on the publisher's website. TakedownIf you consider content in White Rose Research Online to be in breach of UK law, please notify us by emailing eprints@whiterose.ac.uk including the URL of the record and the reason for the withdrawal request. APPENDIX A DOMAIN DECOMPOSITION METHODSDDMs were originally used due to the lack of memory in computing systems: data needed for smaller portions of a problem could fit entirely to the memory while for the whole problem they could not. They lost their appeal as larger and cheaper memory became available, only to resurface in the era of parallel computing. These methods are inherently suited for execution on parallel architectures and many parallel implementations have been presented on multi-core computers, clusters and lately Graphics Processing Units (GPUs) [1], [2]. They are mainly distinguished by three features: subdomain partitioning, problem solution over sub-domains and sub-domain interface variable processing [3]. A.1 Sub-domain PartitioningSub-domain partitioning has to be chosen based on the desired sub-domain characteristics for the given problem. This includes choosing the number of sub-domains, the type of partitioning, and the level of overlap between the sub-domains. Each of these choices depend on a variety of factors such as size, type, and geometry of the problem domain, the number of parallel processors, communication cost, and the actual system being solved.When considering spatial domain problems, such as PDEs, the decomposition is usually g...

show abstract

Bottlenecks in parallel algorithms for power system stability analysis

Cited by 92 publications

References 9 publications

Power System Dynamic Simulations Using a Parallel Two-Level Schur-Complement Decomposition

Power System Dynamic Simulations Using a Parallel Two-Level Schur-Complement Decomposition

LU Factorization Algorithm with Minimum Degree Ordering in Power Distribution Network Problems

Dynamic Simulation of Large-Scale Power Systems Using a Parallel Schur-Complement-Based Decomposition Method

Contact Info

Product

Resources

About