PSCAD经典案例--Electric Network Interface(ENI) Breakthrough

Electric Network Interface (ENI) Breakthrough

By Craig Muller, Rajendra Singh, Bathiya Jayasekera, and Rohitha Jayasinghe, Manitoba HVDC Research Centre

The PSCAD Development Group is always exploring new ways to expand and enhance our tools

capabilities, one of these enhancements has been in the area of High Performance Computing (HPC). Over the last three years the group has achieved multiple milestones in this domain. The first core development was the integration of a new workspace and simulation sets that allows the user to develop coordinated runs in sets using the Xoreax Grid Engine (XGE). Last year a successful implementation of the Electric Network Interface (ENI) was demonstrated and developed to a

commercial grade for deployment to end-customers. ENI effectively enables a large system to be broken into independent, but interconnected sub-networks [1]. This enables a large system to be broken into multiple projects, one per sub-network, so that each project can be dedicated to its own CPU and improve performance.

As the number of breaks increases, such as the 9 Terminal HVDC case, each terminal requires a

dedicated CPU core to execute plus one for PSCAD itself. Most hardware does not have 10+ cores in a single machine (internally we have some single machines with 8, 12, 24 cores that have been used for testing). Our goal is to bring the application to common networked hardware (typically 4 cores/machine). In order to improve overall performance. Recent successes in non-blocking EMTDC allows the process to pick-up the values from its peer the moment the data becomes available rather than waiting to un-block. This means the next release of PSCAD will have the potential to use XGE for this feature without the need for specific configurations.

Recent development in the group have created an even more advanced build of the software that is now capable of analyzing a single project network, computing suitable breakpoints, separately build individual EMTDC processes and launch those processes in concert to work together on a single project. This means the power of (ENI) can be deployed to an end customer with absolutely no changes required by the user. Essentially, a customer is not required to learn how to break networks apart to support

(ENI), rather it will support the customer in their existing implementations. To the best of our knowledge this has never been done before in an offline Electromagnetics Transients tool, making us the first in the world to have this capability.

It is expected that this work will continue to ensure its functionality with control signal carriers,

increasingly complex split patterns and performance mapping. The next phase of the work will be

to develop algorithmic methods to include intelligent optimization of the network splitting. Stay tuned for further advances and release information.

References

[1] In 10th International Conference on Power Systems Transients (IPST 2013),18th July – 21st July 2013, Vancouver, BC, Canada.

本文来源：https://www.bwwdw.com/article/uawp.html