Electrical Grids of the future: GIS-integrated Power-flow as an Online Service

Authors: Dr. Joymala Moirangthem, Mr. Krishnanand K.R., Prof. S.K. Panda, and Prof. Gehan Amaratunga

An electric power system is an interconnected network to carry electric power from points of production to points of consumption. The electrical grid is often represented as a graph with nodes and branches, and their physical interconnection implies that a local change could have non-local impacts. The modelling of the grid using electrical quantities produces nonlinear equations, the solutions of which inform us about the state of the grid. The basic task in the power-flow analyses is to find the mathematical solutions to those equations while ensuring their viability in the physical world. The modernization of the grids through changes in physical infrastructure and the objective of having smarter grids – both of them place new expectations on this indispensable solution-finding step.

The electrical grids of the future are going to be inevitably more complex than what they have been till now. The simple and necessary change of accommodating renewable sources of energy will have not so simple changes in structure and corresponding operations of the grid. Change in power-flow within the grid is one of the prime items of interest for confident expansion and predictable grid-behaviour, especially when parts of it are intermittent. What changes in the system states do the changes in structure of an electrical grid entail? Is the grid stable? How far away is it from instability? What are the permissible changes for the next hour? What are the expected voltages at the nodes within a grid if the solar irradiance for a particular area decreases for a few minutes? What are the optimal power-flow scenarios possible for a given grid, so that it requires minimum operating costs and the carbon emissions are within acceptable limits? There are many questions that require power-flow analyses to answer.

Solar energy, which is the choice renewable energy of Singapore, when accessed through photovoltaic conversion brings higher and faster variability in power than before. Also, the mechanism of distributed generation rather than having a centralized renewable energy source would mean that the grid would become less radial and more meshed.  The paradigm change in the way power-flow would require faster and error-free solutions that facilitate practical decision-making, especially for a computer loaded with algorithms whose operations are contingent on the power-flow solution. The combination of these requirements point at having a power-flow method that is fast for a given precision, works for grids with time-varying dimensions, makes the grid more analysable, and is accessible even at the distribution level by a prosumer. Along with distributed power-flow, the idea of having power-flow as a near-real-time online service has its own appeal in the above described scenario; with advances in reliable connectivity and faster communication speeds being supporting factors.

In this context, we identify Holomorphic Embedded Power Flow (HEPF) proposed by Antonio Trias, Aplicaciones en Informatica Avanzada S.A. to have both mathematical and computational merit to realize such a service. This power-flow method is non-iterative and fast, unlike the conventional Newton-Raphson based power-flow or other gradient-based search methods. HEPF reliably constructs the solution in terms of a power-series after embedding a complex variable ‘s’ to the power-flow formulation, instead of performing a search which is prone to failure. A Geographic Information System (GIS) integrated automation of HEPF, with Singapore’s Jurong Island as the location, has been tested for validating the idea of online power-flow. This scheme enables easier optimization of power-flow problems with ad-hoc AC network structures and ad-hoc objectives, especially if evolutionary algorithms which are parallelizable are used. While the power-flow method acts as the core of the online service, it is the union of it with various sub-technologies (such as GIS libraries, evolutionary algorithms, GPU-based processing, interpreted languages, run-time code-generation, machine-learning, etc.) that empower the service to have expandable smartness. The sum would be then greater than its parts.

Reference

Joymala Moirangthem, Krishnanand K. R., S. K. Panda, and Gehan Amaratunga. “GIS integrated automation of a near real-time power-flow service for electrical grids.” In Sustainable Energy Technologies (ICSET), 2016 IEEE International Conference on, pp. 48-53. IEEE, 2016. (https://doi.org/10.1109/ICSET.2016.7811755)