Optimization-Based Energy System Planning Under Long-Term Uncertainty: Rapid Monte Carlo Analysis Using Linear Regression

International Conference on Efficiency, Cost, Optimization, Simulation and Environmental Impact of Energy Systems (ECOS), pp. 2480–2490

Abstract

Decision-making for distributed energy systems (DES) is subject to significant uncertainties.Therefore, assuming perfect foresight for long-term system planning might result in suboptimal decisions.Long time horizons result in a variety of possible scenarios.One way of considering uncertainty in DES design is Monte Carlo analysis (MCA).However, MCA suffers from the computational burden of repeatedly evaluating energy system models.Furthermore, MCA is sensitive to distribution assumptions of uncertain parameters.In this work, we combine linear regression and systematic uncertainty modeling in MCA to consider uncertainty effectively.We propose a method for comparing different DES designs regarding total annualized cost (TAC) while reflecting uncertain parameters.We model uncertainties by introducing a small number of representative factors that scale reference parameters.We distinguish between constant and long-term uncertainties increasing over time.We use mixed-integer linear programs (MILPs) to minimize yearly costs of DES.We solve the burdensome MILPs only for Latin Hypercube samples to parameterize linear surrogate models.We use the obtained linear surrogate models during MCA to accelerate the computation of the TAC.We apply our method to a case study adapted from the literature to compare promising DES designs while considering multiple sources of uncertainty.We compute the TAC of the DES designs for thousands of longterm scenarios and identify the design, which results most frequently in the lowest costs.We show that using linear regression can reduce the computational time by more than 99 %, while maintaining a high accuracy measured by the goodness of fit of the linear regression.

Authors 4

1. Hagen Seele Aachen Institute of Technical Thermodynamics

   RWTH Aachen University

   Affiliation as printed

   Institute of Technical Thermodynamics, RWTH Aachen University, 52062 Aachen, Germany

2. Jan Tautorus Aachen Institute of Technical Thermodynamics

   RWTH Aachen University

   Affiliation as printed

   Institute of Technical Thermodynamics, RWTH Aachen University, 52062 Aachen, Germany

3. Christiane Reinert Aachen Institute of Technical Thermodynamics

   RWTH Aachen University

   Affiliation as printed

   Institute of Technical Thermodynamics, RWTH Aachen University, 52062 Aachen, Germany

4. Niklas V. von der Assen Aachen Institute of Technical Thermodynamics

   RWTH Aachen University

   Affiliation as printed

   Institute of Technical Thermodynamics, RWTH Aachen University, 52062 Aachen, Germany