Toward a scalable chance-constrained formulation for unit commitment to manage high penetration of variable generation

In this work, a risk-averse optimization model is applied to the security constrained unit commitment problem. The optimal day-ahead scheduling of the system generators is formulated as a chance-constrained optimization model in which the load balance constraint is satisfied with a user-defined probability level. The assumption of a specific underlying distribution is avoided and a flexible data-driven uncertainty set is used to obtain a feasible risk-averse scheduling of the system. Results on a test-scale system show the flexible and effective nature of this approach and indicate significant potential for application to large scale instances.