Smart grid dispatch optimization control techniques for transactive energy systems

Abstract

Transactive smart-grid systems require control techniques to manage forecasting, dispatch optimization, feeder dynamic segmentation, interconnect and micro-grid operations analysis, and other services. Two dispatch optimization tools used in a virtual controller are compared using the same input data: a mixed integer linear programming micro-grid dispatch system, and an artificial neural network (ANN) dispatch system, each of which have been developed using specifications compliant with the Pacific Northwest National Laboratory Smart Grid Demonstration (SGD) project transactive energy nodal system model. The characteristics of these separate optimization techniques are documented from a control perspective, and a distribution substation communications compliant service architecture is developed to use either simulator output set based on an operations context or grid-operator preference such as timing or least-cost. Methods for real-time use of both models are reviewed, and the application of test cases specific to comparison of the systems is explored, using historical time-series inputs from the SGD project and other relevant scenarios. A scalable software architecture is recommended which may function as a smart-grid system controller where separate program optimization methods may be applied in parallel, enabling results to be used in real-time for a risk based assessment of competing operations strategies for an interconnect, control area or micro-grid.