Research Terms
Keywords
Data Analytics Energy Systems Utilities
Researchers at the University of Central Florida have developed an online self-healing power grid tool that responds to power outages faster and more efficiently than current restoration methods. The new Adaptive Restoration Decision Support System (ARDSS) uses novel computational methods to holistically assess and respond to power outages, even as conditions affecting the grid change.
The invention encompasses a system and methods for enabling self-healing power grids after a power outage. When integrated with a company's energy management system, the unique ARDSS combines the many stages of power restoration into one holistic problem, and then decomposes it into a two-stage mixed-integer linear programming (MILP) problem. Once the first-stage problem is solved, the ARDSS addresses the second-stage problem using an integer L-shaped algorithm, auxiliary variables and optimality cuts to ensure the restoration feasibility.
To run the necessary computations, the system employs two types of functions: optimal planning (OP) and optimal real-time (OR) functions; each is executed at specific time periods. The OP function is executed at the early stages of restoration, and the OR function is run periodically at each restoration time step. Moreover, the OP function solves the two-stage problem as a whole, while the OR function only solves the second-stage problem with the first-stage decisions fixed. Both functions may be implemented via hardware or software.
Two-Stage Adaptive Restoration Decision Support System for a Self-Healing Power Grid, IEEE Transactions On Industrial Informatics, Vol. 13, No. 6, December 2017
The University of Central Florida invention enables companies to maintain a nominal alternating current (AC) frequency in an electrical power grid, providing cost savings for consumers and companies. AC flowing through an electrical power grid has a nominal frequency rating of approximately 50 or 60 hertz (Hz). Yet, in most cases, too many loads draw too much power from the grid, causing the AC frequency (generated by AC power sources) to fall below the nominal value. Such deviations can cause system malfunctions and damage turbine generators (for example, the blades). Existing methods for maintaining nominal AC frequency have proven costly and inefficient.
In response to these issues, the UCF invention provides techniques for applying stochastic control of one or more loads at consumer premises in a distributed manner. The capability helps to ensure that the frequency of the AC electrical power through the grid does not exceed a nominal frequency for an extended period. Thus, the UCF technology offers a cost-effective solution for companies such as power utilities, energy aggregators, appliance manufacturers, and building construction companies.
Technical Details: The UCF invention comprises techniques for stochastically determining whether to reduce AC provided to a load. Techniques are employed at consumer premises, enabling companies to aggregate stochastic control of AC electrical power to loads at multiple consumer premises. As long as sufficient consumer premises have stochastically controlled AC power consumption load(s), the AC returns to the nominal frequency when the load(s) in an initial set of consumer premises have their AC consumption diminished or reduced, for example, to zero.
In one example application, a control system stochastically controls AC electrical power to a load. The control system includes a deterministic controller (or controller circuitry) and a stochastic filter (or stochastic filter circuitry). Optionally, the deterministic controller and the stochastic filter are stored and executed in one or more processing systems or circuits. Each processing circuit may comprise state machines or neural networks.
The stochastic control system can include an optional communications circuitry, a frequency detector (or frequency detector circuitry), or an optional power switch. Alternatively, each communications circuitry, frequency detector, and power switch may be external to the system or not used.