Skip to main content
Seminar | Mathematics and Computer Science

Modelling and Solving a Large-Scale Distribution Network AC Optimal Powerflow Problem

LANS Seminar

Abstract: The increasingly wide-spread penetration of distributed energy resources (DERs) in distribution networks is creating scenarios where these parts of the grid are gradually becoming important players in electricity generation. However, with DERs proliferation utilities need to ensure that the distribution network can be operated stably, reliably and safely, and at the same time efficiently utilize their capabilities through market participation. Therefore, to be able to carefully orchestrate the dispatch of DERs while ensuring distribution network’s operational requirements are met, an unbalanced AC optimal powerflow problem (ACOPF) has be solved.

Using physics to model the system containing millions of DERs, several thousands of capacitors, regulators and load tap-changers helps utilities to effectively plan, monitor and control their DERs. While unbalanced ACOPF accurately models powerflows, voltage drops, congestion and losses required for decision making, it is a computationally challenging problem to solve because by nature it is nonlinear and non-convex. The scale of the nonlinear program that needs to be solved further complicates the process of finding a (local) solution.

With clever heuristics to prune the network -- thus reducing its size, solver algorithm informed modelling and input data scaling, Opus One’s optimization engine (OE) is currently able to solve real-world problems that have approximately 3 million of variables and equations in about 10 min (on an average), just in time for distribution grid operators to make real-time decisions. However, as the problem size continues to grow for a multitude of reasons, the use of problem decomposition for parallelization while retaining model accuracy becomes important in the future. The talk will concentrate on the current focus for OE and the directions of interest to develop a product for the grid of the future.

Bio: Pranav Bhaswanth Madabhushi is an optimization engineer at Opus One from GE Digital’s DER Optimization Team. He holds a Ph.D. in Chemical Engineering with a specialization in Process Systems Engineering from McMaster University.