Files in this item



application/pdfLu_Shen.pdf (2MB)
(no description provided)PDF


Title:Decomposition-based, complementarity models for renewable energy generation system design optimization
Author(s):Lu, Shen
Director of Research:Kim, Harrison H.M.
Doctoral Committee Member(s):Thurston, Deborah L.; Pang, Jong-Shi; Ha, Christopher
Department / Program:Industrial&Enterprise Sys Eng
Discipline:Industrial Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Renewable Energy Generation System
Decomposition-based Design Optimization
Multidisciplinary Design Optimization
Mathematical Program with Complementarity Constraints
Hybrid Power Generation System Design
Wind Farm Design
Abstract:With the development of modern technologies, the design community nowadays often faces engineering systems too complex to be addressed by a single design team. In order to solve such design problems, a common approach is to divide a complex system into coupled, yet smaller subsystems manageable by individual design entities, and coordinate the coupled subsystem design decision making towards an overall optimal system design. Decomposition-based design optimization techniques provide the theoretical and computational framework for implementing such design processes. Existing models in the decomposition-based design optimization literature lie largely within the realm of nonlinear programs (NLP). Yet, the increasing intricacy and complexity of design problems often require generalization to more complex models. This dissertation presents decomposition-based, complementarity models for renewable energy generation system design optimization. A generalization of the decomposition-based design optimization, the subsystems of which involve complementarity conditions, is studied under the concept of multidisciplinary design optimization with complementarity constraints (MDO-CC). Such complementarity usually arises from physical, economic or procedural considerations that cannot be described well by NLP models. Due to the ill-posedness associated with the complementarity models, many existing decomposition-based optimization approaches may have numerical difficulties in solving this class of problem. To address this challenge, two decomposition-based approaches for MDO-CC are proposed along the directions of augmented Lagrangian decomposition and regularized inexact penalty decomposition respectively. For each presented algorithm, a correspondence is established between the stationarity conditions of the original all-in-one formulation and those of the decomposed formulations. The design optimization of renewable energy generation system is considered in demonstration of the presented concepts and techniques. Specifically, two research areas are investigated: hybrid power generation system (HPGS) design optimization and renewable energy farm design optimization. A decomposition-based complementarity model is proposed for HPGS design optimization using deterministic simulation-based reliability analysis. In the proposed model, a reformulation technique is introduced to capture the nonsmooth time-dependent battery update with complementarity constraints composed of smooth functions; and a multistage decomposition scheme is applied to the complementarity formulation so that the problem can be solved by the presented MDO-CC algorithms. As a logical extension of the deterministic scenario, HPGS design optimization under probabilistic settings is also studied and an optimization formulation based on a Markovian model for HPGS reliability assessment is proposed. The model estimates the HPGS’s power output using statistical approaches and describes the stochastic state of charge of the battery with a Markov chain. Numerical comparisons with a well-accepted HPGS analysis software as well as the Monte Carlo simulation indicate that the presented model is suitable for HPGS design optimization. Finally, another decomposition-base complementarity model is proposed for wind farm layout design optimization. Complementarity constraints are introduced so that the nonsmooth wake effect can ultimately be considered in a continuously differentiable optimization formulation; and a decomposed formulation is derived through multi-scenario decomposition. It is shown that the design problem can be effectively solved by a hybrid optimization approach which combines the global exploration capacity of a genetic algorithm and the local optimization capability of an MDO-CC approach.
Issue Date:2012-02-06
Genre:Dissertation / Thesis
Rights Information:Copyright 2011 Shen Lu
Date Available in IDEALS:2012-02-06
Date Deposited:2011-12

This item appears in the following Collection(s)

Item Statistics