Robust and stochastic multistage optimisation under Markovian uncertainty with applications to production/inventory problems

A novel robust optimization approach for an integrated municipal water distribution system design under uncertainty: A case study of Mashhad

We present a new methodology to support an energy trading company (ETC) to devise contracting strategies under an optimal risk-averse renewable portfolio. The uncertainty in the generation of renewable energy sources is accounted for by exogenously simulated scenarios, as is customary in stochastic programming. However, we recognize that spot prices largely depend on unpredictable market conditions, making it difficult to capture its underlying stochastic process, which challenges the use of fundamental approaches for forecasting. Under such framework, industry practices make use of stress tests to validate portfolios. We then adapt the robust optimization approach to perform an endogenous stress test for the spot prices as a function of the buy-and-sell portfolio of contracts and renewable energy generation scenarios. The optimal contracting strategy is built through a bilevel optimization model that uses a hybrid approach, mixing stochastic and robust optimization. The proposed model is flexible to represent the traditional stochastic programming approach and to express the ETC's uncertainty aversion in the case where the price distribution cannot be precisely estimated. The effectiveness of the model is illustrated with examples from the Brazilian market, where the proposed approach is contrasted to its stochastic counterpart and both are benchmarked against observed market variables.

Robust optimization and stochastic programming approaches for medium-term production scheduling of a large-scale steelmaking continuous casting process under demand uncertainty

Coordinated operations of electricity and district heating networks offer a potential for mitigating inherent variability of renewable energy sources (RES) in the ongoing transition to smart grids. This paper proposes a two-stage distributionally robust optimization (DRO) approach to determine the optimal day-ahead unit commitment in coordinated electricity and district heating networks with variable RES power output. The proposed formulation is to minimize the worst-case expected total cost over an ambiguity set comprising a family of probability distributions with given support and moments of RES power output. As such, the proposed DRO approach can overcome the limitations of stochastic programming in its inherent dependence of exact probability distributions along with a huge computational burden, but also becomes less conservative than classical robust optimization. The pertinent DRO model is eventually reformulated as a tractable mixed-integer second-order cone (SOC) programming after employing linear decision rules and the SOC duality. Simplified affine policies are utilized to further improve computational tractability and performance. Finally, case studies are conducted based on Barry Island electricity and district heating networks. The numerical results demonstrate the decision-making superiority of the proposed method as compared with deterministic, stochastic programming, and robust optimization approaches. They also validate the computational improvement of the proposed approach by employing simplified affine policies.

This thesis considers the network capacity design problem with demand uncertainty using the stochastic, robust and distributionally robust stochastic optimization approaches (DRSO). Network modeling in itself has found wide areas of application in most fields of human endeavor. The network would normally consist of source (origin) and sink (destination) nodes connected by arcs that allow for flows of an entity from the origin to the destination nodes. In this thesis, a special type of the minimum cost flow problem is addressed, the multi-commodity network flow problem. Commodities are the flow types that are transported on a shared network. Offered demands are, for the most part, unknown or uncertain, hence a model that immune against this uncertainty becomes the focus as well as the practicability of such models in the industry. This problem falls under the two-stage optimization framework where a decision is delayed in time to adjust for the first decision earlier made. The first stage decision is called the "here and now", while the second stage traffic re-adjustment is the "wait and see" decision. In the literature, the decision-maker is often believed to know the shape of the uncertainty, hence we address this by considering a data-driven uncertainty set. The research also addressed the non-linearity of cost function despite the abundance of literature assuming linearity and models proposed for this. This thesis consist of four main chapters excluding the "Introduction" chapter and the "Approaches to Optimization under Uncertainty" chapter where the methodologies are reviewed. The first of these four, Chapter 3, proposes the two models for the Robust Network Capacity Expansion Problem (RNCEP) with cost non-linearity. These two are the RNCEP with fixed-charge cost and RNCEP with piecewise-linear cost. The next chapter, Chapter 4, compares the RNCEP models under two types of uncertainties in order to address the issue of usefulness in a real world setting. The resulting two robust models are also comapared with the stochastic optimization model with distribution mean. Chapter 5 re-examines the earlier problem using machine learning approaches to generate the two uncertainty sets while the last of these chapters, Chapter 6, investigates DRSO model to network capacity planning and proposes an efficient solution technique.

Due to large thermal inertia of buildings and flexibility of interruptible loads, smart buildings pose a remarkable potential for developing virtual energy storage systems (VESSs). However, current literature lacks advanced models to quantify and thus properly optimize available capacity of VESS for power system ancillary services, especially frequency regulation services (FRS). This paper, firstly, presents a novel probabilistic model for explicitly quantifying the VESS capacity in charging and discharging modes, which can further be optimized by scheduling building loads. While the optimized capacity is fully available for FRS, building customer economic benefit and comfort could considerably be impaired by various uncertainties including weather and user related parameters, as well as FRS requests. Due to limited information of the uncertainties, common model-based stochastic and robust optimization approaches are inefficient, leading to either poor out-of-sample performance or over-conservative solutions. Thus, secondly, this paper develops a data-driven distributionally robust optimization (DRO) method to robustly optimize the capacity of VESS against the worst distribution of the uncertainties with a finite training dataset. The proposed data-driven DRO-based model can well match modern data-driven operation environment. Numerical simulations validate the high efficiency and out-of-sample performance of the proposed VESS capacity optimization method under the uncertainties.

This research aims to adopt two robust optimisation approaches for a real-world flow shop manufacturing system with batch processing machines, where the processing time and size of job are non-deterministic and uncertain. Each machine can process multiple jobs simultaneously as long as the machine capacity is not exceeded. Two important decisions are required: (1) grouping jobs into batches and (2) scheduling the established batches on machines. A mathematical optimisation model is presented, and then two famous robust optimisation approaches are adopted for the purpose of converting the deterministic model to the robust one. An efficient particle swarm optimisation (PSO) algorithm is developed to solve the problem in a reasonable time. In order to verify the developed model and evaluate the performance of our proposed algorithm, a set of small to large test problems are generated, and a simulation approach and a commercial optimisation solver are used to solve these problems. Analysis of the implementation of two independent robust optimisation methods is performed by the paired t-test on all of the test problems. Furthermore, the Taguchi method, as a statistical optimisation technique, is employed to investigate the appropriate level of PSO parameters.

A robust convex optimization approach to design a hierarchical organ transplant network: A case study

Chapter 4 - Data-driven robust stochastic optimization for power systems operations

Multi-objective transient peak shaving optimization of a gas pipeline system under demand uncertainty

Proton plans are subject to a number of uncertainties which must be accounted for to ensure that they are delivered safely. Misalignment resulting from residual errors in daily patient positioning can result in both a displacement and distortion of dose distributions. This can be particularly important for intensity modulated proton therapy treatments where the accurate alignment of highly modulated fields may be required to deliver the intended treatment. A number of methods to generate plans that are robust to these uncertainties exist. These include robust optimisation approaches which account for the effect of uncertainties on the dose distribution within the optimisation process. However, robustness to uncertainty comes at the cost of plan quality. For this reason, it is important that the uncertainties considered are realistic. Existing approaches to robust optimisation have neglected the role of fractionated treatment deliveries in reducing the uncertainties that result from random setup errors. Here, a method of robust optimisation which accounts for this effect is presented and is evaluated using a 2D planning environment. The optimisation algorithm considers the dose in the estimated upper and lower bounds of the dose distribution under the effect of setup and range errors. A comparison with plans robustly optimised without consideration of the effect of fractionation and conventionally optimised plans is presented. Fractionation incorporated robust optimisation demonstrates a reduced sensitivity to uncertainty compared to conventionally optimised plans and a reduced integral dose compared to robustly optimised plans.

A new robust optimization approach for scheduling under uncertainty:: I. Bounded uncertainty

Robust optimization based optimal chiller loading under cooling demand uncertainty

Robust-optimization models belong to a special class of stochastic programs, where the traditional expected cost minimization objective is replaced by one that explicitly addresses cost variability. This paper explores robust optimization in the context of two-stage planning systems. We show that, under arbitrary measures for variability, the robust optimization approach might lead to suboptimal solutions to the second-stage planning problem. As a result, the variability of the second-stage costs may be underestimated, thereby defeating the intended purpose of the model. We propose sufficient conditions on the variability measure to remedy this problem. Under the proposed conditions, a robust optimization model can be efficiently solved using a variant of the L-shaped decomposition algorithm for traditional stochastic linear programs. We apply the proposed framework to standard stochastic-programming test problems and to an application that arises in auctioning excess electric power.

One of the main focuses for network planning is on the optimization of network resources required to build a network under certain traffic demand projection. Traditionally, the inputs to this type of network planning problems are treated as deterministic. In reality, the varying traffic requirements and fluctuations in network resources can cause uncertainties in the decision models. The failure to include the uncertainties in the network design process can severely affect the feasibility and economics of the network. Therefore, it is essential to find a solution that can be insensitive to the uncertain conditions during the network planning process. As early as in the 1960's, a network planning problem with varying traffic requirements over time had been studied. Up to now, this kind of network planning problems is still being active researched, especially for the VPN network design. Another kind of network planning problems under uncertainties that has been studied actively in the past decade addresses the fluctuations in network resources. One such hotly pursued research topic is survivable network planning. It considers the design of a network under uncertainties brought by the fluctuations in topology to meet the requirement that the network remains intact up to a certain number of faults occurring anywhere in the network. Recently, the authors proposed a new planning methodology called Generalized Survivable Network that tackles the network design problem under both varying traffic requirements and fluctuations of topology. Although all the above network planning problems handle various kinds of uncertainties, it is hard to find a generic framework under more general uncertainty conditions that allows a more systematic way to solve the problems. With a unified framework, the seemingly diverse models and algorithms can be intimately related and possibly more insights and improvements can be brought out for solving the problem. This motivates us to seek a generic framework for solving the network planning problem under uncertainties. In addition to reviewing the various network planning problems involving uncertainties, we also propose that a unified framework based on robust optimization can be used to solve a rather large segment of network planning problem under uncertainties. Robust optimization is first introduced in the operations research literature and is a framework that incorporates information about the uncertainty sets for the parameters in the optimization model. Even though robust optimization is originated from tackling the uncertainty in the optimization process, it can serve as a comprehensive and suitable framework for tackling generic network planning problems under uncertainties. In this paper, we begin by explaining the main ideas behind the robust optimization approach. Then we demonstrate the capabilities of the proposed framework by giving out some examples of how the robust optimization framework can be applied to the current common network planning problems under uncertain environments. Next, we list some practical considerations for solving the network planning problem under uncertainties with the proposed framework. Finally, we conclude this article with some thoughts on the future directions for applying this framework to solve other network planning problems.