(189s) Stochastic Optimization to Reduce Cost of Energy for Parabolic Trough Solar Power Plant | AIChE

(189s) Stochastic Optimization to Reduce Cost of Energy for Parabolic Trough Solar Power Plant

Authors 

Diwekar, U. - Presenter, Vishwamitra Research Institute /stochastic Rese
The need for clean and cheap renewable energy is on the rise. Solar energy is one of the cleanest and readily available technologies with almost zero carbon emissions. Optimizing the resources to produce efficient power at low costs is the need of the day. However, solar energy power plants face a number of uncertainties like the weather. Since the technologies are new, cost uncertainties are common. In this work, we optimize a solar thermal power plant with two different capacities using the novel Better Optimization of Nonlinear Uncertain Systems (BONUS) algorithm. We use the System Advisory Model (SAM) system from NREL to model performance and economics of the power plant. Since this is a black box model, optimization and optimization under uncertainty becomes difficult. Unlike the deterministic optimization problem, in stochastic programming (or stochastic optimization), one has to consider the probabilistic functional of the objective function and constraints. The generalized treatment of such problems is to use computationally intensive probabilistic or stochastic models instead of the deterministic model, inside the optimization loop. BONUS can circumvent the problems associated with black box models, computational intensity of sampling, and perturbation derivative costs. Instead of the stochastic modeling loop, BONUS uses a statistical reweighting approach to obtain the probabilistic information. It has been shown that BONUS reduces computational intensity by 98 to 99% for large scale problems like this problem of optimization of a solar thermal power plant. We found that the solution of expected value cost minimization for the two different capacity solar plants is robust in the face of uncertainties.