 |
Center for Nonlinear Studies |
Our society depends on electricity for almost any day-to-day activity, and renewable electricity generation is key for a sustainable energy future. While solar cells, wind turbines and overhead lines are visible components of the system, the physics of the grid and the power flows from generators to customers are less tangible. In the first part of this talk, I will give an overview of the electric system and explain how renewable energy is creating both opportunities and challenges for our energy supply. If you know nothing about electricity grids, this is your chance to understand the revolution that is currently happening! The second part of the talk will be more technical, and I will present methods to quantify and manage forecast uncertainty from renewable energy. While the motivation is electric grids, the problem setting is quite general. We consider a large-scale network with non-linear, non-convex physics and uncertain inputs. Uncertainty quantification, i.e. understanding how input uncertainty affects system performance, is integrated within a stochastic optimization problem, where we explicitly limit the probability of adverse outcomes. While solving the full non-linear, stochastic optimization is currently an open problem, I will show two novel solution strategies based on (i) linearization and (ii) an iterative solution algorithm. Numerical results are presented to demonstrate the performance and scalability of the solution methods, along with a discussion of a practical implementation in the real German electricity grid.​
Host: Chris Neale