Parallel processing in a cluster of microcomputers with application in contingency analysis

This work is about the use of a microcomputer network as a parallel processing environment. In order to allow the communication, the synchronization and the distribution of the tasks among processes, the PVM (parallel virtual machine) and MPI (message passing interface) systems, were used. These computational systems allowed for parallel processing in the area of electric power systems, namely the contingency analysis. This application, usually found in energy management systems, requires the use of a large quantity of resources in a short time. The aim is to analyze thousands of disturbance scenarios, in a preventive mode in real time. Two parallel programming paradigms classified as Master/Slave, in the synchronous and asynchronous modes, were implemented. Case studies with sequential and parallel implementations were carried out to validate the algorithms used to analyze elaborated contingencies, including the use of data from the interconnected Brazilian electric power system, allowing the analysis of single and multiple disturbances.