Multilevel Task Partition Algorithm for Parallel Simulation of Power System Dynamics

Abstract

Abstract. Nowadays task partition for parallel computing is becoming more and more important. Particular in power system dynamic simulation, it is critical to design an efficient partition algorithm to reduce the communication and balance the computation load [1]. This paper presents a novel multilevel partition scheme based on the graph partition algorithm. By introducing regional characteristic into the partition, improving the weights of nodes and edges, proposing an objective function to evaluate the partition results and some other schemes, we can efficiently improve the defects in the traditional partition methods. With 12 CPUs for a large scale power system with 10188 nodes, the parallel efficiency with our new algorithm was 63% higher than that with METIS, a well-known program used for partitioning graphs. The proposed algorithm will satisfy the requirement for large scale power grid dynamic simulation.

High performance computing for electric power systems: Applications and trends

“…The work also makes some further efforts to optimize communication and computation. Overall, the methodology is very similar to that seen in [50], [51]. A speedup of roughly 10 is demonstrated.…”

“…Implemented on a SMP cluster with 705 to 10,108 nodes, each with four processors, a super linear speedup is reported in almost all cases and the method is shown superior to previous methods.This work is further extended in [51] where a novel multilevel partitioning scheme is applied to dynamic power system simulations in order to efficiently map the problem to the HPC domain.…”

Parallel Implementation of Non-slicing Floorplans with MPI and OpenMP

“…But in ideal situations, it is better to use the number of processors that is divisible by the number of clusters to balance the computation load. It is a very vital task to appropriately partition the data so as to map the computational load onto the processors [32].…”

An interactive visual enviroment based on advanced numerical and computer techniques for power systems applications

“…3 shows the use of parallel programming to reduce the computational time required in a complete simulation. To achieve de maximum performance from the parallel scheme the computational load needs to be balanced [11], that is, all the processors must be kept busy all the time. A simple, but efficient way to achieve this aim is by having the same number of elements (in terms of computational effort) in each linked list.…”