GC31E-1243
Development of a Computational Framework for Stochastic Co-optimization of Water and Energy Resource Allocations under Climatic Uncertainty

Wednesday, 16 December 2015
Poster Hall (Moscone South)
Yi Xuan, North Carolina State University Raleigh, Raleigh, NC, United States
Abstract:
Owing to the lack of a consistent approach to assimilate probabilistic forecasts for water and energy systems, utilization of climate forecasts for conjunctive management of these two systems is very limited. Prognostic management of these two systems presents a stochastic co-optimization problem that seeks to determine reservoir releases and power allocation strategies while minimizing the expected operational costs subject to probabilistic climate forecast constraints. To address these issues, we propose a high performance computing (HPC) enabled computational framework for stochastic co-optimization of water and energy resource allocations under climate uncertainty. The computational framework embodies a new paradigm shift in which attributes of climate (e.g., precipitation, temperature) and its forecasted probability distribution are employed conjointly to inform seasonal water availability and electricity demand. The HPC enabled cyberinfrastructure framework is developed to perform detailed stochastic analyses, and to better quantify and reduce the uncertainties associated with water and power systems management by utilizing improved hydro-climatic forecasts. In this presentation, our stochastic multi-objective solver extended from Optimus (Optimization Methods for Universal Simulators), is introduced. The solver uses parallel cooperative multi-swarm method to solve for efficient solution of large-scale simulation-optimization problems on parallel supercomputers. The cyberinfrastructure harnesses HPC resources to perform intensive computations using ensemble forecast models of streamflow and power demand. The stochastic multi-objective particle swarm optimizer we developed is used to co-optimize water and power system models under constraints over a large number of ensembles. The framework sheds light on the application of climate forecasts and cyber-innovation framework to improve management and promote the sustainability of water and energy systems.