| Summary: | This work presents a decision-making framework for global optimization of detailed renewable energy processes considering technological uncertainty. The critical uncertain sources are identified with an efficient computational method for global sensitivity analysis, and are obtained in two different ways, simultaneously and independently per product pathway respect to the objective function. For global optimization, the parallel stochastic response surface method developed by Regis & Shoemaker (2009) is employed. This algorithm is based on the multi-start local metric stochastic response surface method explored by the same authors (2007a). The aforementioned algorithm uses as response surface model a radial basis function (RBF) for approximating the expensive simulation model. Once the RBFs parameters are fitted, the algorithm selects multiple points to be evaluated simultaneously. The next point(s) to be evaluated in the expensive simulation are obtained based on their probability to attain a better result for the objective function. This approach represents a simplified oriented search. To evaluate the efficacy of this novel decision-making framework, a hypothetical multiproduct lignocellulosic biorefinery is globally optimized on its operational level. The obtained optimal points are compared with traditional optimization methods, e.g. Monte-Carlo simulation, and are evaluated for both proposed types of uncertainty calculated.
|
|---|
