Challenges of Generation Dispatch for Smart Grid
- Written by Kwok W. Cheung
As power resources become more distributed, systems more conducive to demand-response, and generation more intermittent, efficient and robust system operation will depend critically on the ability of new dispatch methods to provide a better predictive, forward-looking and holistic view of system conditions and generation patterns.
Restructuring of the electric power industry has made the secure operation of stressed power systems more challenging than ever. On top of that, systems are undergoing fundamental changes in response to concerns about carbon emissions and energy conservation in advanced and emerging economies. The upshot is that the inherited centralized infrastructure is being profoundly transformed with generation facilities dispersed, demand made responsive and controllable and networks actively managed system-wide.
A particular challenge connected with monitoring and controlling the demand-supply balance, when it comes to renewable resources, is that their outputs often do not follow traditional generation/load correlation but have strong dependencies on weather conditions. At the same time, a desire for transparency and market liquidity on the part of regulators and market participants naturally drives systems towards larger trading areas and towards creation of new incentives for end users to consume their energy smartly.
Around the world, very large power grid operators like PJM, Midwest ISO or North China Grid are fundamentally reliant on generation dispatch. In order to face the challenges posed by the smart grid, regional system operators (RTOs) and transmission system organizations (TSOs) are designing next-generation dispatch systems with broader and more sophisticated capabilities to handle greater uncertainties than ever before.
Notably, the limited dispatchability and intermittent nature of wind and solar generation will likely require grid operators to supply additional ancillary services needed to maintain reliability and operational requirements. Just as important, efficient system operation will depend critically on the ability of new dispatch methods to provide a better predictive, forward-looking and holistic view of system conditions and generation patterns. That in turn requires a more capable predictive model, one that incorporates better modeling of transmission constraints, better modeling of resource characteristics — such as capacity limits and ramp rates — and more accurate demand forecasting.
The dispatch system of the near future will economically and simultaneously manage rapid changes in load, generation, interchange and transmission security constraints on a real-time or near real-time basis. The system will flexibly incorporate various power forecast data sources, including demand and renewable generation outlooks. A new time-coupled dynamic dispatch engine will provide a desired dispatch profile for any specified time frame. The scheduling solutions of different timeframes addressing different system scenarios are consolidated to form a comprehensive operating plan, which has a continuum of gradual and harmonized transitions from one time frame solution to another. The plan provides operators with a holistic, forward-looking view that continually updates trends in the dynamics of generation profiles and general system conditions hours ahead of real-time.
Many RTOs, such as PJM, Midwest ISO, and ISO New England, have recently implemented a comprehensive dispatch process that includes a set of modular components to address input data quality, multiple dispatch time periods, and visualization and automation requirements. As a critical component, RTOs deploy multiple short-term commitment and dispatch engines based on mixed-integer programming; these are time coupled to produce a continuously forward-looking and updated dispatch trajectory for each flexible resource. This integrated dispatch process has proven to significantly improve operations efficiency and situational awareness.
Another way to cope with uncertainties is to make dispatch solutions more robust. Optimality or even feasibility of dispatch solutions can be very sensitive to system uncertainties. Reserve requirements and “n-1” contingency analysis (accounting for the loss of single piece of equipment) are traditional ways to ensure a certain robustness of a given system.
Scenario-based simulation is another common technique to address uncertainties. Multiple scenarios, each associated with probability, can be simulated based on Monte Carlo techniques; individual scenario solutions or a composite solution derived from scenario probabilities can be used to assess sensitivity of system economics and reliability to uncertainties, such as demand forecast. Specifically, the uncertainty for demand forecast is one of the most critical factors influencing the uncertainty of generation requirements for system balancing.
Assuming there is no transmission congestion due to renewable generation, it is viable to simplify the analysis by treating wind generation as a negative load and incorporate its uncertainty as a part of the so-called uncertainty of net demand. Researchers and practitioners are working hard together to make use of more advanced techniques, such as stochastic optimization and robust optimization, to address uncertainties for the problems of unit commitment and economic dispatch.
My company, Alstom, has been partnering with other organizations to research and develop new commitment and dispatch tools, using advanced techniques of stochastic optimization and robust optimization. For example, Alstom has recently been working with Sandia National Laboratories, the University of California-Davis, Iowa State University and ISO New England on the ARPA-e GENI project entitled Improved Power System Operations Using Advanced Stochastic Optimization, which aims to develop scalable algorithms for advanced stochastic unit commitment for cost-effective and low-risk scheduling and dispatch solutions in scenarios where there is high penetration of renewables.