The distribution grid is changing as more devices are being connected every day. The IEA predicts that there will be 1 billion smart households, 11 billion connected devices, and 500M EVs by 2040. These distributed energy resources, like solar PV, electric vehicles, batteries, and controllable loads, provide consumer benefits but also interact with the grid. The challenge is that having so many active resources connected on the distribution grid leads to multi-directional power flows and unpredictable grid conditions. In order to ensure reliability and safety with such a dynamic distribution grid, many utilities and grid operators are interested in better monitoring, analyzing, and controlling the distribution grid, but there are limited tools available.
Traditionally, distribution grids are managed somewhat statically, using system studies that are conducted every so often that include information about estimates of voltages and currents in a limited number of scenarios. Operating decisions are made with the best data available but typically without knowing real time distribution grid conditions or being able to accurately predict conditions. Historically, this has been mostly sufficient since the distribution grid has mainly had one-way power flow. Jim See, Wayne Carr, and Steven E. Collier discuss this in a 2008 IEEE conference paper titled “Real Time Distribution Analysis for Electric Utilities”. As they state, given the tools most utilities currently have in place: “decisions to switch lines, control equipment and devices, or manage customer loads must be based on static criteria chosen to allow ample margins of error.”
This historical, static way of operating the distribution isn’t going to work going forward. Many utilities and grid operators would like to orchestrate or better work with the distributed resources on their grids, but it is difficult to know how they should orchestrate resources for active grid management or flexibility. Distribution utilities and grid operators need real-time analysis and active grid management in order to operate the grid as efficiently, reliably, and sustainably as possible. Without more insight, utilities won't know how to most optimally operate resources on the distribution grid.
At the transmission and bulk power generation level, state estimation is used to calculate real time conditions on the grid and to estimate the near future state. State estimation is the process of predicting system conditions in the very near term future. Having this information allows for system operators to take action now to achieve operational goals for the grid like efficiency or reliability in the near future. Though these tools exist at the generation and transmission level, they can’t easily be used for the distribution grid. The number of data points (generators, power lines, devices, etc.) at the G&T level is very small compared to the number of data points that would need to be incorporated at the distribution level. The tools that have been used historically can’t feasibly continuously model the distribution grid in real time due to the sheer number of data points.
Real time distribution analysis is needed for distribution system state estimation. See et al. describe what would be needed to do real time distribution grid analysis:
Detailed circuit model- an accurate and detailed circuit model is the foundation of real time analysis
SCADA data - measured data from sources of power and energy
Smart meter data - measured data from sinks of power and energy, kW loading
Optimal power flow is a critical component of state estimation. The optimal power flow problem was introduced by Carpentier in 1962 (see: https://www.gerad.ca/en/OPF). It involves looking at planned consumption and production to determine how to operate the grid (voltages, active and reactive power, etc.) in a way that minimizes costs or negative outcomes while operating within physical and safety constraints such as Ohm's laws, Kirchhoff's laws, voltage limits, power limits, etc. The classic formulation of this problem, which is known as alternating current optimal power flow (ACOPF), is a non-convex problem that’s very difficult to solve. “In operations research, a convex optimization problem is an optimization problem for which any local optimum is also a global one. ACOPF may be formulated as a quadratic optimization problem with quadratic constraints. In the last decade or so, several convex relaxations (approximations) have been developed, including the second-order conic relaxation (2006) and the semidefinite relaxation (2008). The challenge today is to develop a model that’s as good as semidefinite relaxation and as quick to solve as the second-order conic relaxation” (from https://www.gerad.ca/en/OPF). Research is ongoing.
Economic dispatch is another important optimization problem in grid operations. Real-time economic dispatch optimization is a common market scheduling problem that tries to economically balance electricity system supply and demand and provide locational marginal prices while respecting system reliability requirements. Economic dispatch is a convex optimization problem with a linear or quadratic objective, typically the minimization of generator costs or the maximization of social surplus. Research is ongoing in this area as well.
The OPF problem is a non-convex problem, but the economic dispatch problem is typically convex. Ideally, one would solve these simultaneously for the best possible outcome. It’s very difficult to do all this quickly enough today. There are only a few companies that say they can do all this right now. Opus One Solutions claims to do it, offering planning tools, real time power flow optimization and state estimation, and a transactive energy market platform. Apparently, Omega Grid also claims to have incorporated optimal power flow in their platform. Lastly, there is a startup called ProsumerGrid that is focuses on this and they previously won an ARPA-E grant in this area. Several ADMS vendors also claim to have this capability, including Survalent.
There remains a lot of work to be done in this area. It seems likely that real-time (or near real time) analysis will be needed before energy markets activities at the distribution grid level can really take off. The question is how widespread and good these capabilities will need to be.
Jim See, Wayne Carr P.E., and Steven E. Collier, "Real Time Distribution Analysis for Electric Utilities", Rural Electric Power Conference 2008 IEEE. https://ieeexplore.ieee.org/document/4520136
IEA, Digitalization in Energy, Nov’17