PSR delivers new model for the energy planning of the West Coast of the United States

Studies – 21.08.2018

In the beginning of August, PSR held a seminar for the delivery of the new Genesys optimization model for energy planning in the Pacific Northwest region. About 100 experts participated in the seminar, representing more than twenty agencies and companies in the region, such as Bonneville Power Administration (BPA), US Army Corps of Engineers and Pudget Sound and Light.

A year ago PSR won the bid for the development of the new Genesys model, promoted by the North West Power and Conservation Council (NWPCC), an agency that coordinates the region's energy development. The region’s 61 thousand MW power system is quite complex, because it includes the Columbia river hydroelectric plants, a large amount of wind generation, thermal gas plants, eleven control zones (balancing markets) and interconnections with the entire West Coast (WECC), with an installed capacity of nearly 400 thousand MW.

According to PSR’s president Mario Veiga Pereira, who coordinated the project, the NWPCC specifications were challenging: the Genesys software should carry out in six hours thousands of hourly probabilistic simulations of the power system operation, from long term stochastic optimization to “hour-ahead” scheduling, including the modeling of renewable generation under uncertainty, forecast models, gas contracts and power exchanges with the WECC. Mario continues: “These probabilistic simulations require the solution of 108 million large scale optimization problems (mixed integer programming), a world record for this type of model. In addition to the computational effort, the volume of results was also a record, about 10 terabytes per execution.”

The solution developed by the PSR team, called PSR Core, consists of a new computational architecture, with a distributed solution of the optimization problems (involving about 16 thousand CPUs in cloud computing); specialized databases for creating scenarios and managing results (using Cassandra / Spark); and new optimization algorithms. This solution considerable interest in the research community: around the same period of the Genesys seminar, PSR research scientist Joaquim Garcia made invited presentations about PSR Core and the large scale optimization solution for the Amazon optimization team in Seattle; in a high performance computing session at the IEEE General Meeting, in Portland; and also in Bordeaux, France, for the development group of the new scientific programming language Julia and in the International Symposium on Mathematical Programming (ISMP 2018).