Multiscale Stochastic Simulation of the US Pacific Northwest Using Distributed Computing and Databases with Integrated Inflow and Variable Renewable Energy

TL;DR

This paper presents a multiscale stochastic simulation framework for the US Pacific Northwest power system, integrating distributed computing, databases, and renewable energy considerations to improve system modeling and reliability analysis.

Contribution

It introduces a novel five-step simulation approach combining mathematical programming, high-performance databases, and parallel computing to address complex hydro and renewable energy integration challenges.

Findings

Enabled large-scale stochastic simulations with millions of mathematical programs.

Improved modeling of renewable energy impacts on system reliability.

Demonstrated scalability using distributed computing resources.

Abstract

Modelling challenges of the United States Pacific Northwest system have grown in the last decade. Besides classical modelling difficulties such as a complex hydro cascade with many operational constraints, we have seen higher penetration of variable renewable energy inside and outside the system leading to internal issues and completely different power exchanges with the West Coast System. The analysis of adequacy and reliability of this system motivated the design and implementation of a five-step simulator including four planning phases and an operation step. The five levels were modeled as mathematical programs that are linked from top to bottom by fixed decisions and improvements in forecasts, on the other hand they are also linked from bottom to top by system updated states. The solution of the millions of resulting mathematical programs was made possible by applying state of the art optimization techniques and high performance data bases in a massively parallel environment.