Location of retail facilities under conditions of uncertainty

Models for the optimal location of retail facilities are typically premised on current market conditions. In this paper we incorporate future market conditions into the model for the location of a retail facility. Future market conditions are analyzed as a set of possible scenarios. We analyze the problem of finding the best location for a new retail facility such that the market share captured at that location is as close to the maximum as possible regardless of the future scenario. The objective is the minimax regret which is widely used in decision analysis. To illustrate the models an example problem is analyzed and solved in detail.     

Retail facility location under changing market conditions

In this paper we investigate the location of retail facilitiesunder changing market conditions when market conditions areexpected to change during the planning horizon. Three modelsare presented: (1) the minimax regret model where the objectivefunction is to minimize the maximum possible loss under differentmarket scenarios, (2) the Stackelberg equilibrium model wherebya future competitor enters the market and one wishes to maximizethe market share captured throughout the planning horizon incorporatingthe market share loss to the competitor, and (3) the thresholdmodel in which we consider a minimum threshold level such thatif the market share captured by the new facility fails to reachthat threshold, the facility will not survive. Our objectiveis to minimize the probability of failure.     

Decision/forecast horizon for a single facility dynamic location/relocation problem

Procedures to solve finite horizon dynamic location/relocation problems have been reported in the literature by many authors. This paper provides several decision/forecast horizon results for a single facility dynamic location/relocation problem; these results are helpful in finding optimal initial decisions for the infinite horizon problem by using information only for a finite horizon.     

井位不确定环境下的油田仓库选址问题

油田工作中,合理的仓库选址决策不仅能节约物流成本,而且能提高油田作业效率。现有研究通常基于当前的油井位置(简称井位),没有考虑未来井位变化对仓库选址的影响。同时井位受到地下储层条件及油气公司远景规划等因素影响,未来的井位具有很强的不确定性。此外,仓库选址决策属中长期决策,将长期影响油田的物流费用、管理工作甚至开发工作,而且油田生产要求仓库能持续供应物资,所以油田仓库选址应考虑井位的不确定性和仓库服务中断等因素。本文首先根据油田井位分布和钻井规划采用随机模拟方法模拟未来井位,建立并求解考虑设施中断的离散选址模型。然后以鄂南油区物资仓库选址问题为例,模拟井位并求解仓库选址问题,从8个候选点中选出3个建库/租库。最后,分析井位和需求量变化对仓库选址结果的影响。井位不确定环境下油田物资仓库选址问题的研究,不仅对油田物流系统管理的研究具有一定的理论意义,也对油田的物流决策有重要的现实意义。     

最小最大后悔准则下新增设施选址策略研究

基于新增设施选址问题,考虑网络节点权重不确定性,以设施中最大负荷量最小为目标,提出最小最大后悔准则下的新增设施选址问题。在网络节点权重确定时,通过证明将网络图中无穷多个备选点离散为有限个设施候选点,设计了时间复杂度为O(mn²)的多项式算法;在节点权重为区间值时,通过分析最大后悔值对应的最坏情境权重结构,进而确定最大后悔值最小的选址,提出时间复杂度为O(2ⁿm²n³)的求解算法;最后给出数值算例。     

Strategic facility location: A review

Facility location decisions are a critical element in strategic planning for a wide range of private and public firms. The ramifications of siting facilities are broadly based and long-lasting, impacting numerous operational and logistical decisions. High costs associated with property acquisition and facility construction make facility location or relocation projects long-term investments. To make such undertakings profitable, firms plan for new facilities to remain in place and in operation for an extended time period. Thus, decision makers must select sites that will not simply perform well according to the current system state, but that will continue to be profitable for the facility's lifetime, even as environmental factors change, populations shift, and market trends evolve. Finding robust facility locations is thus a difficult task, demanding that decision makers account for uncertain future events. The complexity of this problem has limited much of the facility location literature to simplified static and deterministic models. Although a few researchers initiated the study of stochastic and dynamic aspects of facility location many years ago, most of the research dedicated to these issues has been published in recent years. In this review, we report on literature which explicitly addresses the strategic nature of facility location problems by considering either stochastic or dynamic problem characteristics. Dynamic formulations focus on the difficult timing issues involved in locating a facility (or facilities) over an extended horizon. Stochastic formulations attempt to capture the uncertainty in problem input parameters such as forecast demand or distance values. The stochastic literature is divided into two classes: that which explicitly considers the probability distribution of uncertain parameters, and that which captures uncertainty through scenario planning. A wide range of model formulations and solution approaches are discussed, with applications ranging across numerous industries.     

Simple dynamic location problem with uncertainty: a primal-dual heuristic approach

Abstract

In this paper, the simple dynamic facility location problem is extended to uncertain realizations of the potential locations for facilities and the existence of customers as well as fixed and variable costs. With limited knowledge about the future, a finite and discrete set of scenarios is considered. The decisions to be made are where and when to locate the facilities, and how to assign the existing customers over the whole planning horizon and under each scenario, in order to minimize the expected total costs. Whilst assignment decisions can be scenario dependent, location decisions have to take into account all possible scenarios and cannot be changed according to each scenario in particular. We first propose a mixed linear programming formulation for this problem and then we present a primal-dual heuristic approach to solve it. The heuristic was tested over a set of randomly generated test problems. The computational results are provided.     

Online stochastic optimization under time constraints

This paper considers online stochastic combinatorial optimization problems where uncertainties, i.e., which requests come and when, are characterized by distributions that can be sampled and where time constraints severely limit the number of offline optimizations which can be performed at decision time and/or in between decisions. It proposes online stochastic algorithms that combine the frameworks of online and stochastic optimization. Online stochastic algorithms differ from traditional a priori methods such as stochastic programming and Markov Decision Processes by focusing on the instance data that is revealed over time. The paper proposes three main algorithms: expectation E, consensus C, and regret R. They all make online decisions by approximating, for each decision, the solution to a multi-stage stochastic program using an exterior sampling method and a polynomial number of samples. The algorithms were evaluated experimentally and theoretically. The experimental results were obtained on three applications of different nature: packet scheduling, multiple vehicle routing with time windows, and multiple vehicle dispatching. The theoretical results show that, under assumptions which seem to hold on these, and other, applications, algorithm E has an expected constant loss compared to the offline optimal solution. Algorithm R reduces the number of optimizations by a factor |R|, where R is the number of requests, and has an expected ρ(1+o(1)) loss when the regret gives a ρ-approximation to the offline problem.     

Application of decision analysis techniques to the Weber facility location problem

A location problem with future uncertainties about the data is considered. Several possible scenarios about the future values of the parameters are postulated. However, it is not clear which of these scenarios will actually happen. We find the location that will best accommodate the possible scenarios. Four rules utilized in decision theory are examined: the expected value rule, the optimistic rule, the pessimistic rule, and the minimax regret rule. The solution for the squared Euclidean distance is explicitly found. Algorithms are suggested for general convex distance metrics. An example problem is solved in detail to illustrate the findings, and computational experiments with randomly generated problems are reported.     

Opportunity Losses Due to Uncertainty in Multiobjective Decision Problems

A multiobjective maximization problem is considered in which at least one objective function f_i(x, C) depends on a random parameter C. If a single-valued measure, such as weighting or an l_p distance, is used to determine the preferred solution among the nondominated solutions, then standard decision-theoretic methods can be used to determine the expected opportunity loss (EOL). By an example hydrologic problem, it is shown that EOL is highly dependent on the single-valued measure selected to solve the multiobjective problem. The expected multiobjective opportunity loss (EMOL) is developed as a vector-valued measure of the effect of uncertainty on the problem which is independent of the technique. Finding the decision point with minimum EOL or EMOL is a possible way of selecting the preferred point. Problems pertaining to a multiobjective formulation of the EOL concept are examined.     

A multi-objective integrated facility location-hardening model: Analyzing the pre- and post-disruption tradeoff

Two methods of reducing the risk of disruptions to distribution systems are (1) strategically locating facilities to mitigate against disruptions and (2) hardening facilities. These two activities have been treated separately in most of the academic literature. This article integrates facility location and facility hardening decisions by studying the minimax facility location and hardening problem (MFLHP), which seeks to minimize the maximum distance from a demand point to its closest located facility after facility disruptions. The formulation assumes that the decision maker is risk averse and thus interested in mitigating against the facility disruption scenario with the largest consequence, an objective that is appropriate for modeling facility interdiction. By taking advantage of the MFLHP’s structure, a natural three-stage formulation is reformulated as a single-stage mixed-integer program (MIP). Rather than solving the MIP directly, the MFLHP can be decomposed into sub-problems and solved using a binary search algorithm. This binary search algorithm is the basis for a multi-objective algorithm, which computes the Pareto-efficient set for the pre- and post-disruption maximum distance. The multi-objective algorithm is illustrated in a numerical example, and experimental results are presented that analyze the tradeoff between objectives.     

Simulated N-Body: New Particle Physics-Based Heuristics for a Euclidean Location-Allocation Problem

The general facility location problem and its variants, including most location-allocation and P-median problems, are known to be NP-hard combinatorial optimization problems. Consequently, there is now a substantial body of literature on heuristic algorithms for a variety of location problems, among which can be found several versions of the well-known simulated annealing algorithm. This paper presents an optimization paradigm that, like simulated annealing, is based on a particle physics analogy but is markedly different from simulated annealing. Two heuristics based on this paradigm are presented and compared to simulated annealing for a capacitated facility location problem on Euclidean graphs. Experimental results based on randomly generated graphs suggest that one of the heuristics outperforms simulated annealing both in cost minimization as well as execution time. The particular version of location problem considered here, a location-allocation problem, involves determining locations and associated regions for a fixed number of facilities when the region sizes are given. Intended applications of this work include location problems with congestion costs as well as graph and network partitioning problems.     

Climate policy under fat-tailed risk: an application of FUND

We apply four alternative decision criteria, two old ones and two new, to the question of the appropriate level of greenhouse gas emission reduction. In all cases, we consider a uniform carbon tax that is applied to all emissions from all sectors and all countries; and that increases over time with the discount rate. For a one per cent pure rate of the time preference and a rate of risk aversion of one, the tax that maximises expected net present welfare equals $120/tC in 2010. However, we also find evidence that the uncertainty about welfare may well have fat tails so that the sample mean exists only by virtue of the finite number of runs in our Monte Carlo analysis. This is consistent with Weitzman’s Dismal Theorem. We therefore consider minimax regret as a decision criterion. As regret is defined on the positive real line, we in fact consider large percentiles instead of the ill-defined maximum. Depending on the percentile used, the recommended tax lies between $100 and $170/tC. Regret is a measure of the slope of the welfare function, while we are in fact concerned about the level of welfare. We therefore minimise the tail risk, defined as the expected welfare below a percentile of the probability density function without climate policy. Depending on the percentile used, the recommended tax lies between $20 and $330/tC. We also minimise the fatness of the tails, as measured by the p-value of the test of the null hypothesis that recursive mean welfare is non-stationary in the number of Monte Carlo runs. We cannot reject the null hypothesis of non-stationarity at the 5 % confidence level, but come closest for an initial tax of $50/tC. All four alternative decision criteria rapidly improve as modest taxes are introduced, but gradually deteriorate if the tax is too high. That implies that the appropriate tax is an interior solution. In stark contrast to some of the interpretations of the Dismal Theorem, we find that fat tails by no means justify arbitrarily large carbon taxes.     

A Linear Programming Approach to the Solution of Constrained Multi-Facility Minimax Location Problems where Distances are Rectangular

The problem of locating new facilities with respect to existing facilities is stated as a linear programming problem where inter-facility distances are assumed to be rectangular. The criterion of location is the minimization of the maximum weighted rectangular distance in the system. Linear constraints which (a) limit the new facility locations and (b) enforce upper bounds on the distances between new and existing facilities and between new facilities can be included. The dual programming problem is formulated in order to provide for an efficient solution procedure. It is shown that the duLal variables provide information abouLt the complete range of new facility locations which satisfy the minimax criterion.     

Bayesian policy support for adaptive strategies using computer models for complex physical systems

In this paper, we discuss combining expert knowledge and computer simulators in order to provide decision support for policy makers managing complex physical systems. We allow future states of the complex system to be viewed after initial policy is made, and for those states to influence revision of policy. The potential for future observations and intervention impacts heavily on optimal policy for today and this is handled within our approach. We show how deriving policy dependent system uncertainty using computer models leads to an intractable backwards induction problem for the resulting decision tree. We introduce an algorithm for emulating an upper bound on our expected loss surface for all possible policies and discuss how this might be used in policy support. To illustrate our methodology, we look at choosing an optimal CO₂ abatement strategy, combining an intermediate complexity climate model and an economic utility model with climate data.     

Tools of the mind: Techniques and methods for intellectual work: V. STIBIC North-Holland,Amsterdam, 1982, xiii + 297 pages,Dfl.85.00

Many location problems may be separated into a series of interrelated macro, meso and micro decision-making states. The macro scale decision determines the type, capacity and number of facilities, the meso scale decision determines the location and allocation of facilities and the micro scale decision determines such considerations as routing and scheduling of service vehicles. This paper concerns the first two levels of decision-making.The present paper demonstrates the use of two models: (i) an analytical model that uses continuum approximations and methods of calculus to determine the number of facilities, the capacity and the approximate location of each that minimizes the sum of the transportation and facility costs for a slowly varying demand rate, and (ii) a traditional location-allocation model that determines more exactly the resulting locations and allocations. These two approaches have specific requirements in terms of data input, cost of data collection and cost of solution and, consequently, yield unique insights and benefits for practising planners. The strengths and weaknesses of the two models are complementary. This thesis is developed with an analysis of the Calgary, Alberta refuse collection and disposal system.     

A Lagrangian Relaxation Heuristic for Capacitated Facility Location with Single-Source Constraints

Facility location models are applicable to problems in many diverse areas, such as distribution systems and communication networks. In capacitated facility location problems, a number of facilities with given capacities must be chosen from among a set of possible facility locations and then customers assigned to them. We describe a Lagrangian relaxation heuristic algorithm for capacitated problems in which each customer is served by a single facility. By relaxing the capacity constraints, the uncapacitated facility location problem is obtained as a subproblem and solved by the well-known dual ascent algorithm. The Lagrangian relaxations are complemented by an add heuristic, which is used to obtain an initial feasible solution. Further, a final adjustment heuristic is used to attempt to improve the best solution generated by the relaxations. Computational results are reported on examples generated from the Kuehn and Hamburger test problems.     

A lexicographical compromise method for multiple criteria group decision problems with imprecise information

This paper addresses multiple criteria group decision making problems where each group member offers imprecise information on his/her preferences about the criteria. In particular we study the inclusion of this partial information in the decision problem when the individuals’ preferences do not provide a vector of common criteria weights and a compromise preference vector of weights has to be determined as part of the decision process in order to evaluate a finite set of alternatives. We present a method where the compromise is defined by the lexicographical minimization of the maximum disagreement between the value assigned to the alternatives by the group members and the evaluation induced by the compromise weights.     

A multi-objective heuristic approach for the casualty collection points location problem

In this paper, we formulate the casualty collection points (CCPs) location problem as a multi-objective model. We propose a minimax regret multi-objective (MRMO) formulation that follows the idea of the minimax regret concept in decision analysis. The proposed multi-objective model is to minimize the maximum per cent deviation of individual objectives from their best possible objective function value. This new multi-objective formulation can be used in other multi-objective models as well. Our specific CCP model consists of five objectives. A descent heuristic and a tabu search procedure are proposed for its solution. The procedure is illustrated on Orange County, California.     

An expected regret minimization portfolio selection model

Fuzzy portfolio selection has been widely studied within the framework of the credibility theory. However, all existing models provide only concentrated investment solutions, which contradicts the risk diversification concept in the classical portfolio selection theory. In this paper, we propose an expected regret minimization model, which minimizes the expected value of the distance between the maximum return and the obtained return associated with each portfolio. We prove that our model is advantageous for obtaining distributive investment and reducing investor regret. The effectiveness of the model is demonstrated by using an example of a portfolio selection problem comprising ten securities in the Shanghai Stock Exchange 180 Index.