This project develops and analyzes dynamic robust optimization models and solution methods for electric power systems operation under uncertainty. In particular, this project will develop multi-stage robust optimization models and algorithms for the unit commitment problem, a critical decision-making tool for daily scheduling of power systems. In a multi-stage robust optimization model, both the generation commitment and power dispatch decisions are dynamically adaptive to sequential observation of uncertainty, as opposed to existing models where the decisions are either static or noncausal. Such models involve a large number of mixed integer variables and decision policies in infinite dimensional space. Finding an exact solution for these models is computationally intractable. The challenge is to develop theory and algorithms for large-scale problems along with performance guarantees. Our approach is to develop and apply finitely adaptive decision policies in properly designed multi-stage uncertainty models, and develop efficient decomposition methods for their resolution.
As an effort toward building a sustainable national energy system, electric power systems in the US are experiencing fundamental changes, with large-scale integration of renewable energy resources and the deployment of demand response technologies. As a result, both the energy supply and demand sides have significantly increased uncertainty. Reliable operation of power systems under these uncertainties is critical. The results of this project, if successful, will significantly advance the state of the art in power systems operations under uncertainty, and will also significantly advance the techniques in solving general multi-stage robust optimization problems. Substantial effort will be committed to educating next generation academics and researchers who are proficient in advanced analytics and devoted to improve the national energy sustainability.
