Search

This paper presents a solution approach to the joint assortment and inventory planning problem for vertically differentiated products considering dynamic consumer-driven substitution. The demand for each product and the stockout-based substitution rates are derived from a consumer’s utility function and a random market size. We propose a two-step integral solution approach, which uses the substitution characteristic of vertically differentiated products. In a first step, the approach determines the initial purchasing probabilities of all potential products and the substitution rate matrices for all possible product-availability combinations. In a second step, the inventory levels are determined by iteratively solving a sequence of two-product problems. We prove that an optimal assortment only contains products whose critical ratios are decreasing with increasing quality levels. We further compare the proposed approach with the exact results obtained by complete enumeration and find that in almost all instances, the optimality gap is below 0.5%. Numerical experiments reveal that an increasing demand uncertainty leads to first deeper assortments and second a shift of inventory towards lower-quality products. We also compare the proposed integral approach with a sequential assortment and inventory planning approach that first optimizes the assortment only considering assortment-based substitution and subsequently, determines the optimal inventory levels constrained on this assortment. We show that a joint assortment and inventory planning considering stockout-based substitution is particularly essential when both the profit margin and demand uncertainty is high. Moreover, we provide an extension considering shelf-space constraints.

Approximately 89 million of tonnes of food is wasted every year in the EU along the whole food supply chain. The reasons for food waste by retailers include inappropriate quality control, overstocking and inaccurate forecasting. This study shows that food wasted by retailers can be reduced by discounting old products or by applying a dynamically adjustable expiration date (in other words dynamic shelf life (DSL)). We developed a simulation based optimization model to optimize the replenishment and discounting policy of a retailer who sells meat products. DSL outperforms a fixed shelf life (FSL) in terms of profit, waste, shortages and food safety. Furthermore, replenishment quantities can be higher. The benefits of DSL are greater when demand is low or when the shelf life of products is short. Discounting is a successful strategy to reduce food waste for both FSL and DSL. DSL without discounting is more effective than FSL with discounting. Combining DSL and discounting, allows for a further reduction of food waste.

The food industry is confronted with a pressure to reduce waste and to make agreements on donating surplus food to charitable organizations. Charitable organizations such as food banks and soup kitchens can use these donations in preparing food parcels or meals for their clients. For soup kitchens, donation management is strongly influencing menu planning, and conversely, menu planning considerations have a strong impact on donation management decisions. To make the best use of (mostly highly perishable) food donations, we develop an MILP model for integrated donation management and menu planning that proposes a menu plan and suggests which (part of the) donations to accept. The combination of menu planning and donation management is essential for soup kitchens, but has not been studied before. The model is used to assess the impact of contracts on a strategic or tactical level, and captures operational decision making due to the integration of donation management and menu planning. To deal with meal variety considerations and to resemble planning practices, the developed model is solved in a rolling horizon. The results show that (i) the use of donations reduces overall costs for the soup kitchen; (ii) despite the short shelf life of donations, most donations can be used efficiently; and (iii) meal variety can be easily ensured and food donations increase this variety. In addition to the benefits for soup kitchens, the approach has implications for waste reduction in food supply chains, by structural/contractual donations of surplus food by retailers.

Increasing concerns on energy use, emissions and food waste requires advanced models for food logistics management. Our interest in this study is to analyse the benefits of horizontal collaboration related to perishability, energy use (CO2 emissions) from transportation operations and logistics costs in the Inventory Routing Problem (IRP) with multiple suppliers and customers by developing a decision support model that can address these concerns. The proposed model allows us to analyse the benefits of horizontal collaboration in the IRP with respect to several Key Performance Indicators, i.e., emissions, driving time, total cost comprised of routing (fuel and wage cost), inventory and waste cost given an uncertain demand. A case study on the distribution operations of two suppliers, where the first supplier produces figs and the second supplier produces cherries, shows the applicability of the model to a real-life problem. The results show that horizontal collaboration among the suppliers contributes to the decrease of aggregated total cost and emissions in the logistics system. The obtained gains are sensitive to the changes in parameters such as supplier size or maximum product shelf life. According to experiments, the aggregated total cost benefit from cooperation varies in a range of about 424% and the aggregated total emission benefit varies in a range of about 833% compared to the case where horizontal collaboration does not exist. HighlightsA chance-constrained model for the multi-supplier IRP is presented.Model accounts for perishability, explicit fuel consumption and demand uncertainty.Model allows us to analyse the benefits of horizontal collaboration in the IRP.A case study on the distribution operations of two suppliers shows model applicability.Horizontal collaboration contributed to the decrease of total cost and emissions.

The duration of logistics operations, as well as the environmental conditions during these operations, significantly impact the performance of a logistics network for fresh agricultural products. When durations or temperatures increase, product quality decreases and more effort is required to deliver products in time and with the right quality. Different network designs lead to different durations and conditions of transport, storage, processing, etc. Therefore, when making network design decisions, consequences for lead time and product quality should be taken into account. As decay of perishable products, for instance food, is often not uniform, heterogeneity in product quality decay also has to be considered. The aim of this paper is to show how product quality decay as well as its heterogeneity can be integrated in a network design model. A new mixed integer linear programming formulation is presented, which positions stocks and allocates processes to maximise profit under quality constraints. It is applied to several test instances from the horticultural sector. Results show that different levels of decay lead to different network structures. Changing decay rates due to processing particularly affect the level of postponement. Heterogeneity in product quality causes a split in product flows with high and low product quality. All in all, it is shown that heterogeneous product quality decay should be taken into account in network design as it significantly influences network designs and their profitability, especially when the supply chain includes processes that change the level of decay, and product quality differences can be exploited in serving different markets.

In the fresh produce wholesale market, the market price is determined by the total demand and supply. The price is stochastic, and either wholesaler or retailer has few influence on it. In the wholesaler’s inventory decision, the price’s uncertainty plays an important role as well as the uncertainty from the demand side: the wholesaler makes his decision based on the retailer’s ordering, which is influenced by the stochastic market price and the distribution of the consumer’s demand. In addition, when at the wholesale stage, the products show a similar quality of similar appearance. With more efforts being input, the wholesaler could detect and record more additional information than that reflected from the appearance. Based on this, he can classify the quality into different levels. No experience shows how the wholesaler could use the underlying quality information and how much this information could improve his profit. To describe and explore this problem, a bilevel dynamic programming approach is employed. We evaluate different strategies of using the underlying information, show the features of the optimal policy, develop heuristics, and discuss the influence of factors such as quality and market price. We also develop the managerial principles for the practical use.

The objective of this paper is to develop a systematic overview of interventions for preventing chilled-food waste at retail outlets, and to assess the impact of these interventions for a particular case of fresh-cut iceberg lettuce at a Dutch retail outlet. The structure of the simulation model as presented in this paper is generic, hence suitable for other retailers and other chilled-food products as well. The generated systematic overview focusses on interventions that do not require a system change. A distinction is made into technical, logistical and marketing interventions. Model simulations show the effectiveness of these interventions. It is concluded that a number of ‘waste drivers’ exists, such as a low and varying consumer demand, high selection behaviour, the order lead time, a fixed order unit, and a short use-by date. The retailer can fine-tune the replenishment level of his order policy and the way of rounding to the given order unit, but by doing so he is at best able to exchange waste for out-of-stock or the other way around. The systematic overview of interventions is valuable input to future research on defining and estimating the effectiveness of combining interventions, and interventions that do require a system change.

Many supermarkets and retailers use computer assisted ordering (CAO) or automated order systems (ASO) to determine replenishment quantities. Traditionally, these systems are designed for non-perishables, and order quantities are set by (modified) base stock policies, determined according to the total number of units in stock, regardless of the ages of the products in stock. Today's technology allows the user to keep track of the age of inventories. In the last decade, a few so-called stock-age dependent order policies have been proposed. An optimal stock-age dependent policy may be computed by stochastic dynamic programming (SDP), but an SDP policy is hardly ever adopted in practice for two reasons: (1) the policy can only be computed if the underlying state space is not too large, and (2) the resulting policy can have a complicated structure. In practice, logistics managers prefer well structured policies they do understand, such as base stock policies (BSP). In this paper, we give an overview of existing stock-age dependent order policies and provide new stock-age dependent order policies. We present a simulation-based search procedure and suggest search ranges for setting optimal parameter values. For a broad set of 11,177 instances, we test the algorithm, compare these policies by simulation, and discuss the value of stock-age information. A new policy, named BSP-low-EW, performs close to optimal (SDP) and turns out to outperform all other policies in virtually all cases.

As a large part of world food production spoils/expires before consumption, reduction of food waste by optimizing order policies in retail is of importance. We sketch here the computational burden of trying to obtain the optimal order quantities with the process of value iteration for a retailer situation with highly perishable products. It appears that waste can be reduced if retailers follow the optimal policy compared to an easier order‐up‐to‐level policy. When there is willingness to substitute a product for another if the desired product is out of stock, the complexity increases drastically. Basically, one has to take the stock of both products into account. The necessary probabilistic analysis and computational aspects are described in this paper.

Management objectives of the European Union for North Sea fish stocks are shifting towards considering both biological sustainability and economic benefits. As part of multiannual management plans, an adjustment restriction on fish quota has been introduced. Its objective is to obtain an efficient fish stock and to reduce overcapacity for the fishing industry. We develop and apply a bi-level stochastic dynamic programming model to study the effect of a quota adjustment restriction on the net present value of resource rents, overcapacity and fish stock, when the system is recovering from a downward environmental shock. At level one, a policy maker sets the quota, considering fishermen behavior, stochastic fish stock dynamics, capital stock dynamics and a quota adjustment restriction. At level two, fishermen harvest myopically and make long-term investment decisions, assuming that fish stock and quota do not change over time. The two levels are linked by the quota, which is optimized by the policy maker at level one and becomes a restriction for myopic harvest and long-term investment decisions of fishermen at level two. Our analysis suggests that in the long run, overcapacity can be reduced by 54 % at modest costs, namely at a 1 % reduction in the net present value of resource rents. Long and short run sustainability of the fish stock is not affected.

We study the practical decision problem of fresh food production with a long production lead time to decide every period (e.g. week) how many items to produce. When a batch is ready for use, its items have a fixed shelf life, after which the items become waste in the sense that they cannot be sold anymore. The demand for (fresh) food products is uncertain and highly fluctuating, mainly caused by price promotions of retail organisations. We focus on cases where a so-called cycle fill rate service level requirement applies. We investigate the generation of a production plan that fixes the timing and quantity of the production for a finite time horizon. To minimise waste, one issues the oldest items first, i.e. a FIFO issuing policy. In case of out-of-stock, sales are lost. We model this decision problem as a Stochastic Programming (SP) model. The objective of our study is to find order quantities for the SP model, that approximately meet cycle fill rate service level requirements while keeping outdating low. To find approximate solutions for the SP model, an MILP model is developed. The MILP model is a deterministic approximation that generates feasible replenishment quantities in less than a second. With a scenario-based MINLP approach, optimal solutions are generated for a large sample of demand paths as a benchmark for the MILP solutions. We show that the MILP model is suitable for practical use if the setup cost is such that the replenishment cycles in the production plan are close to or of the same length as the maximum shelf life. In those cases, the expected total costs are close to the costs of the optimal solution and the average fill rate is close to the required one.

The transition to sustainable food supply chain management has brought new key logistical aims such as reducing food waste and environmental impacts of operations in the supply chain besides the traditional cost minimization objective. Traditional assumptions of constant distribution costs between nodes, unlimited product shelf life and deterministic demand used in the Inventory Routing Problem (IRP) literature restrict the usage of the proposed models in current food logistics systems. From this point of view, our interest in this study is to enhance the traditional models for the IRP to make them more useful for the decision makers in food logistics management. Therefore, we present a multi-period IRP model that includes truck load dependent (and thus route dependent) distribution costs for a comprehensive evaluation of CO2 emission and fuel consumption, perishability, and a service level constraint for meeting uncertain demand. A case study on the fresh tomato distribution operations of a supermarket chain shows the applicability of the model to a real-life problem. Several variations of the model, each differing with respect to the considered aspects, are employed to present the benefits of including perishability and explicit fuel consumption concerns in the model. The results suggest that the proposed integrated model can achieve significant savings in total cost while satisfying the service level requirements and thus offers better support to decision makers.

In retail, it is usual to measure the performance of inventory control by a so-called service level which measures the number of days (probability) that demand is fulfilled for a certain product, i.e. the product is not out of stock. Part of the customers that do not encounter the product in stock will look for a substitute. To measure the performance, we introduce a so-called δ-service level which measures the availability of the substitute. We investigate the effect of safety levels and substitution probability on the δ-service level and illustrate them based on the case of poisson demand.

Network designs can strongly depend on product quality decay.Product quality constraints added to MILP for hub location and process allocation.Simulation of MILP solution determines detailed product quality service levels.Quality service levels are updated in an iterative optimization-simulation approach.Results depend on whether locations or allocations are varied across iterations. Dynamics in product quality complicate the design of logistics networks for perishable products, like flowers and other agricultural products. Complications especially arise when multiple products from different origins have to come together for processes like bundling. This paper presents a new MILP model and a hybrid optimization-simulation (HOS) approach to identify a cost-optimal network design (i.e. facility location with flow and process allocation) under product quality requirements. The MILP model includes constraints on approximated product quality. A discrete event simulation checks the feasibility of the design that results from the MILP assuming uncertainties in supply, processing and transport. Feedback on product quality from the simulation is used to iteratively update the product quality constraints in the MILP. The HOS approach combines the strengths of strategic optimization via MILP and operational product quality evaluation via simulation. Results, for various network structures and varying degrees of dynamics and uncertainty, show that if quality decay is not taken into account in the optimization, low quality products are delivered to the final customer, which results in not meeting service levels and excess waste. Furthermore, case results show the effectiveness of the HOS approach, especially when the change from one iteration to the next is in the choice of locations rather than in the number of location. It is shown that the convergence of the HOS approach depends on the gap between the product quality requirements and the quality that can be delivered according to the simulation.

The floricultural sector is facing market developments that have forced a redesign of the European logistics network. Via workshops and interviews with key stakeholders the main developments and industry needs are identified. These are then summarised in three central themes that require further investigation, i.e. decision problems (e.g. network design and control), context factors (e.g. demand uncertainty and product perishability), and objectives (e.g. efficiency and product quality). Thereafter, 17 articles that review Supply Chain Management (SCM) research are analysed to obtain more insight into the state-of-the-art on these themes and to identify the main issues within the themes and their interrelationships. This resulted in a conceptual research framework in which particular attention is given to how decision problems could be modelled and solved in order to get quantitative insights into the impact of logistics network redesign. Successively, 71 SCM articles were analysed in depth to classify current SCM research and to determine research gaps and challenges. Results show that Floricultural SCM research challenges can be found in integrated, quality-driven and responsive network design and control using hybrid optimisation and simulation.

We study the practical production planning problem of a food producer facing a non-stationary erratic demand for a perishable product with a fixed life time. In meeting the uncertain demand, the food producer uses a FIFO issuing policy. The food producer aims at meeting a certain service level at lowest cost. Every production run a setup cost is incurred. Moreover, the producer has to deal with unit production cost, unit holding cost and unit cost of waste. The production plan for a finite time horizon specifies in which periods to produce and how much. We formulate this single item—single echelon production planning problem as a stochastic programming model with a chance constraint. We show that an approximate solution can be provided by an MILP model. The generated plan simultaneously specifies the periods to produce and the corresponding order-up-to levels. The order-up-to level for each period is corrected for the expected waste by explicitly considering for every period the expected age-distribution of the products in stock. The model assumes zero lead time and backlogging of shortages. The viability of the approach is illustrated by numerical experiments. Simulation shows that in 96.4% of the periods the service level requirements are met with an error tolerance of 1%.

The dynamic adaptive control of traffic lights can be formulated as a Markov decision problem (MDP). This framework is hardly used, as solving an MDP can be very time-consuming and is only possible for simple infrastructures with a small number of traffic flows. Nevertheless, we show that the MDP framework can be used to construct control policies that approximately minimize the long-run average waiting time at intersections. The MDP-based approach is fast and thus allows real-time use of actual information on traffic conditions, like queue lengths and the arrival times of near-future arrivals. Simulation of an isolated intersection as well as a small network shows that the new policies with arrival information improve pretimed as well as exhaustive control. The new control policies and underlying algorithms scale up well to control networks.

Traffic lights are put in place to dynamically change priority between traffic participants. Commonly, the duration of green intervals and the grouping, and ordering in which traffic flows are served are pre-fixed. In this chapter, the problem of minimizing vehicle delay at isolated intersections is formulated as a Markov Decision Process (MDP). Solving the MDP is hampered by a large multi-dimensional state space that contains information on the traffic lights and on the queue lengths. For a single intersection, an approximate solution is provided that is based on policy iteration (PI) and decomposition of the state space. The approach starts with a Markov chain analysis of a pre-timed control policy, called Fixed Cycle (FC). The computation of relative states values for FC can be done fast, since, under FC, the multi-dimensional state space can be decomposed into sub-spaces per traffic flow. The policy obtained by executing a single iteration of Policy Iteration (PI) using relative values is called RV 1. RV1 is compared for two intersections by simulation with FC, a few dynamic (vehicle actuated) policies, and an optimal MDP policy (if tractable). RV1, approximately solves the MDP, and compared to FC, it shows less delay of vehicles, shorter queues, and is robust to changes in traffic volumes. The approach shows very short computation times, which allows the application to networks of intersections, and the inclusion of estimated arrival times of vehicles approaching the intersection.

Traditional wholesalers of fresh produce in China display and sell all the products that they have brought to the wholesale market every day. Products left by the end of the day are deteriorated due to high temperatures at which products are displayed, and are thus sold at a low price or disposed. Modern wholesalers can preserve product quality and carry over unsold products to the next day, by using a back storage with cooling facilities at or close to the market site. Modern wholesalers thus face a multi-period decisions problem, with decision related to amongst other the number of products to keep behind in the cooled back storage, and the number of products to display in the open air from each quality class. During a market day these decision can be revised. Hence the decision problem is not only a multi-stage problem because of carrying over products from one day to the next, but also because of intra-day periods at which the display decision is updated. The price decreases over the day. So does the demand, and the demand depends on the quality of the displayed products. Products that are of (too) low quality are sold/disposed to a secondary market. We modeled the inventory control problem as a Markov decision process model that maximizes the profit of the wholesaler. Based on the model and numerical result for a realistic setting, we analyze the structure of the optimal policies and derive heuristics for practical use. The heuristic methods perform close to the optimal policy, resulting in managerial insights to practitioners.

High effectiveness and leanness of modern supply chains (SCs) increase their vulnerability, i.e. susceptibility to disturbances reflected in non-robust SC performances. Both the SC management literature and SC professionals indicate the need for the development of SC vulnerability assessment tools. In this article, a new method for vulnerability assessment, the VULA method, is presented. The VULA method helps to identify how much a company would underperform on a specific Key Performance Indicator in the case of a disturbance, how often this would happen and how long it would last. It ultimately informs the decision about whether process redesign is appropriate and what kind of redesign strategies should be used in order to increase the SC's robustness. The applicability of the VULA method is demonstrated in the context of a meat SC using discrete-event simulation to conduct the performance analysis.

The production and inventory management of blood products at blood banks and hospitals is it problem of general human interest. As a shortage may put lives at risk, shortages are to be kept to a minimum. As the supply is voluntary and costly, any spill of unused blood (products) is also to be minimized.Blood platelets (thrombocytes), which are the most expensive and perishable blood product, have the complication of a limited "shelf life" of only 5-7 days. A general figure in the Western world (the USA and Western Europe) for the spill of blood platelets is in the order of 15-20%.A combined new approach is therefore presented which combines stochastic dynamic programming (SDP) and simulation to provide:(i) Practical simple order-up-to rules that are nearly optimal.(ii) Formal theoretical support.The approach has been applied to a Dutch regional Blood bank. Numerical results show it significant reduction of the figures from: 1%, to 1% for shortages; 20% to 1% for spill.A practical question for blood bank managers that still remains is: "How to anticipate irregular production breaks like at Easter and Christmas?"The present paper therefore will extend the combined SDP-Simulation approach to include such breaks. The main findings are:Also for these breaks a simple order-up-to rule remains to be nearly optimal.Also for these breaks the outdating and shortages call be kept less than M.The (stationary) periods with production and the (non-stationary) breaks without can be integrated.The approach thus seems suitable for practical implementation. (C) 2007 Elsevier B.V. All rights reserved.

This article presents a novel approach for the dynamic control of a signalized intersection.At the intersection, there is a number of arrival flows of cars, each havinga single queue (lane). The set of all flows is partitioned into disjoint combinationsof nonconflicting flows that will receive green together. The dynamic control of thetraffic lights is based on the numbers of cars waiting in the queues. The problemconcerning when to switch (and which combination to serve next) is modeled as aMarkovian decision process in discrete time. For large intersections (i.e., intersectionswith a large number of flows), the number of states becomes tremendouslylarge, prohibiting straightforward optimization using value iteration or policy iteration.Starting from an optimal (or nearly optimal) fixed-cycle strategy, a one-steppolicy improvement is proposed that is easy to compute and is shown to give a closeto optimal strategy for the dynamic problem.

In practice, traditional stock-level dependent policies, like base-stock policies (BSP) and constant order policies (COP), are commonly used for replenishing inventories of perishable products at retailers. These policies are preferred for being easy to implement: they only require information on the total number of products in stock, but not differentiated by their age. In this paper, we analyze a number of new and existing hybrid BSP-COP policies. These policies have different complexities, but, so far, they have not been systematically compared with respect to their performance. By simulation-based optimization, the parameter values of the policies are determined. For this purpose, search ranges for the parameter values are provided. Based on an extensive set of experiments, insights are gained on when to apply which policy. The results show that two newly proposed enhancements of traditional base-stock policies, in particular, perform well and can be recommended for practical implementation.

Today’s business environment and harsh competitiveness force companies and entire supply chains to increase their efficiency as much as possible. As a consequence, supply chains have become highly sensitive to disruptions and less tolerant to deviations in operations, that is, supply chains have become more vulnerable (see Kleindorfer and Saad, 2005). Vulnerability of supply chains may result in less consistent supply chain performances, and consequently, their competitive power in the market may diminish. In order to maintain stability of supply chain performances, it is necessary to design robust supply chains. Robust supply chains should be able to continue to function well in the event of a disruption as well as in the normal business environment (see Dong, 2006; Tang, 2006; Waters, 2007). Generally, robustness of the supply chain depends on its capability to respond adequately to different kinds of risks of disturbances. Recently, supply chain vulnerability and robustness has become a hot research topic, and as such, it is still in its infancy. With this paper, we aim to contribute to the existing knowledge in these areas.

This paper studies the computation of so-called order-up-to levels for a stochastic programming inventory problem of a perishable product. Finding a solution is a challenge as the problem enhances a perishable product, fixed ordering cost and non-stationary stochastic demand with a service level constraint. An earlier study [7] derived order-up-to values via an MILP approximation. We consider a computational method based on the so-called Smoothed Monte Carlo method using sampled demand to optimize values. The resulting MINLP approach uses enumeration, bounding and iterative nonlinear optimization.

The determination of order quantities in an inventory control problem of perishable products with non-stationary demand can be formulated as a Mixed Integer Nonlinear Programming problem (MINLP). One challenge is to deal with the \(\beta \)-service level constraint in terms of the loss function. This paper studies the properties of the optimal solution and derives specific algorithms to determine optimal quantities.

This chapter aims to promote and illustrate the fruitful combination of classical Operations Research (OR) and Computer Simulation. First, a highly instructive example of parallel queues will be studied. This simple example already shows the necessary combination of OR (queueing) and simulation that appears to be of practical interest such as for call center optimization. Next, two more ’real life’ applications are regarded: - blood platelet production and inventory management at blood banks, and - train conflict resolution for railway junctions. Both applications show the useful combination of Simulation and optimization methods from OR, in particular Stochastic Dynamic Programming (SDP) and Markov decision theory (MDP), to obtain simple rules that are nearly optimal. The results are based on real life Dutch case studies and show that this combined OR-Simulation approach can be most useful for ’practical optimization’ and that it is still wide open for further application.

Perishables, such as packed fresh food and pharmaceutical products (a.o. blood products), typically have a fixed shelf life set by a fixed use-by date or sell-by date. Despite their limited life time, orders in practice are usually based on the stock level irrespective of the ages of the products in stock. The management of inventories of such products can be improved by applying stock-age dependent ordering, issuing, and disposal policies. This paper investigates cost reductions that can be achieved by an optimal stock-age dependent ordering, issuing, or disposal policy as obtained by Stochastic Dynamic Programming. Orders are made before the uncertain demand is revealed. When demand turns out to be relatively low, a disposal policy enables to get rid of excess (old) stock. Disposal decisions are an understudied area, but may be relevant to retailers for which displaying the freshest items is of high importance. Also blood banks prefer not to issue products that are about to expire as transfusion of younger blood products is more effective. This paper fills a research gap identified in Karaesmen et al. (2011) : the paper appears to be the first to report optimal stock-age dependent disposal decisions, both under a base stock policy and under optimal stock-age dependent ordering. Results of optimal stock-age dependent ordering, disposal, and issuance are compared to a base stock policy, which is commonly used in practice. Under FIFO issuance, the added value of an optimal disposal policy is high. An optimal disposal policy in combination with optimal ordering reduces the average costs only when issuing old products is penalized, e.g. by selling at a discounted price. Under LIFO issuance, an optimal disposal policy has significant impact when orders are set by a BSP, but not under optimal stock-age dependent ordering. When no penalty or discount applies, disposals reduce costs only in case of suboptimal ordering, e.g. by a BSP.

Traffic lights can be controlled dynamically through rules reacting on the number of waiting vehicles at each light. A rule can be captured by a so-called Traffic Control Table (TCT). The Value Iteration method from Stochastic Dynamic Programming has been used for simple networks to derive a TCT. This work studies the generation of a TCT-based rule that takes the arrival information of new vehicles into account. The question is how to generate such a table for simple intersections (or a network of these). The generation is particularly difficult due to the computational work involved in the Value Iteration process. The problem is formulated as a Markov Decision Process and the parallelization of the Value Iteration method for this problem is discussed. We are specifically interested in exploiting the structure of the problem for simple infrastructures, with only a few traffic lanes, using a parallel algorithm.

Many perishables such as fresh food and blood platelet concentrates are characterized by a short maximum shelf life. As demand is often highly uncertain the outdating and shortages figures can be very high, especially when frequent replenishment is not possible or inefficient due to fixed ordering cost. We present a new class of stock-level dependent ordering policies: the (s,S,q,Q) policy, which is a periodic review (s,S) policy with the order quantity restricted by a minimum (q) and maximum (Q). Optimal (weekday dependent) parameter values are derived by dynamic programming and simulation. The (s,S,q,Q) policy performs nearly optimal and improves the (s,S) policies in many cases by 4–25%.

North Sea fisheries are managed by the European Union (EU) through a system of annual quota. Due to uncertainty about future fish stocks, yearly revisions of these policies lead to fluctuation in quota, which in turn affects harvest and investment decisions of fishermen. Determination of quota requires high management costs in terms of obtaining information and negotiations between experts and policy makers. To reduce both quota fluctuation and management costs, the EU has proposed a system of multiannual quota. In this paper we study the effect of multiannual quota on quota volatility and resource rents, while accounting for management costs. We develop a bi-level stochastic dynamic programming model, where at level one, the EU determines the quota that maximizes resource rents. At level two, fishermen decide myopically on their harvest and investment levels, subject to the quota. Results show that policy makers can reduce quota volatility and improve resource rents from the fishery with multiannual quota. Important trade-offs are involved in the accomplishment of these objectives: fish stock and investments become more volatile, which leads to more overcapacity.

This paper describes and analyses a Stochastic Programming (SP) model that is used for a specific inventory control problem for a perishable product. The decision maker is confronted with a non-stationary random demand for a fixed shelf life product and wants to make an ordering plan for a finite horizon that satisfies a service level constraint. In literature several approaches have been described to generate approximate solutions. The question dealt with here is whether exact approaches can be developed that generate solutions up to a guaranteed accuracy. Specifically, we look into the implications of a Stochastic Dynamic Programming (SDP) approach.

After years of emphasis on leanness and responsiveness businesses are now experiencing their vulnerability to supply chain disturbances. Although more literature is appearing on this subject, there is a need for an integrated framework to support the analysis and design of robust food supply chains. In this paper we present such a framework. We define the concept of robustness and classify supply chain disturbances, sources of food supply chain vulnerability, and adequate redesign principles and strategies to achieve robust supply chain performances. To test and illustrate its applicability, the research framework is applied to a meat supply chain.

The management of inventories of perishable products with a short maximal shelf life takes a good issuing policy next to a good ordering policy. Ordering policies of non-perishables are well studied in literature and implemented in Automated Store Ordering (ASO) systems and Computer Assisted Ordering (CAO) systems. These ordering policies are stock-level dependent and do not take the product ages into account. As a consequence when applied to perishable products they do not anticipate expected future outdating of products and thus result in unnecessary outdating and shortages. Improved ordering policies are proposed in the literature but hardly implemented in ASO and CAO systems as these are designed for the management of non-perishables. Without changing the order policies of such systems one may reduce outdating and shortages by issuing the products in a sophisticated way. We present therefore a stochastic model that is solved by stochastic dynamic programming.

BACKGROUND: The challenge of production and inventory management for blood platelets (PLTs) is the requirement to meet highly uncertain demands. Shortages are to be minimized, if not to be avoided at all. Overproduction, in turn, leads to high levels of outdating as PLTs have a limited "shelf life." Outdating is to be minimized for ethical and cost reasons.STUDY DESIGN AND METHODS: Operations research (OR) methodology was applied to the PLT inventory management problem. The problem can be formulated in a general mathematical form. To solve this problem, a five-step procedure was used. This procedure is based on a combination of two techniques, a mathematical technique called stochastic dynamic programming (SDP) and computer simulation.RESULTS: The approach identified an optimal production policy, leading to the computation of a simple and nearly optimal PLT production "order-up-to" rule. This rule prescribes a fixed order-up-to level for each day of the week. The approach was applied to a test study with actual data for a regional Dutch blood bank. The main finding in the test study was that outdating could be reduced from 15-20 percent to less than 0.1 percent with virtually no shortages. Blood group preferences and extending the shelf life of more than 5 days appeared to be of marginal effect.CONCLUSION: In this article the worlds of blood management and the mathematical discipline of OR are brought together for the optimization of blood PLT production. This leads to simple nearly optimal blood PLT production policies that are suitable for practical implementation.

Blood platelets are precious, as voluntarily supplied by donors, and highly perishable, with limited lifetimes of 5–7 days. Demand is highly variable and uncertain. A practical production and inventory rule is strived for that minimizes shortages and spill. The demand and production are periodic, as varying over the seven days of the week. Demand for ‘young’ platelets (oncology and hematology) and demand for platelets of ‘any’ age up to the maximal shelf life (traumatology and general surgery) are distinguished. A combined Markov dynamic programming (MDP) and simulation approach is presented and applied to a real life case of a Dutch blood bank. By down-sizing the dimension and applying this combined approach it is shown that order-up-to type replenishment rules that perform quite well can be found. Particularly, a double-level order-up-to rule, so-called 2D rule, is derived, with one level corresponding to ‘young’ platelets and one to the total inventory. This rule is easy to implement and is shown to be ‘nearly optimal’. This approach and the double order-up-to rule seem to be new. The results are most suitable for sensitivity analyses such as with respect to shortages and production costs for blood platelet inventory management.

Globally, around one-third of the edible parts of perishable food products is wasted every year. Adequate logistics management of the food supply chain is of importance. Inventory control of processed fresh food with a best-before or use-by date deals with the questions how much to order and how often, and how to review the inventory position. For perishable products, there exists a trade-off between desired product-availability and waste. A high order quantity or frequent ordering may result in high product-availability but also waste. A small order quantity or less frequent ordering results in lower inventory levels, but may cause out-of-stock. For the inventory control of a perishable product the age-distribution of the items should be considered. This is influenced by order picking at a food producer or warehouse, or consumer behaviour at a supermarket, picking First In First Out (FIFO), where first the oldest items are used, or Last In First Out (LIFO), where first the freshest items of the product are used. This thesis investigates periodic review order policies for a food producer and a retailer for a perishable product with a fixed lifetime and a non-stationary demand. For a food producer as well as a retailer, it is not always economically feasible to reorder in every period. For planning purposes it is desirable to set the timing of the orders in advance. The aim of this thesis is to contribute to better decision making regarding inventory control in the food supply chain from food producer to supermarket, in the trade-off between product-availability and waste. Methods are designed to generate practical order policies using commercial solvers for business rather than custom made solution procedures, where at least the timing of ordering or production is set beforehand. The practical inventory control problems discussed in this thesis are characterised by a perishable product with a fixed lifetime, non-stationary stochastic demand, a single-echelon production/inventory situation in a finite time horizon. The product has a service level requirement to ensure a certain product-availability. The decision problems deal with fixed setup or ordering cost, holding cost and disposal cost for wasted items. The age-distribution of the items in stock is considered in specific theoretical Stochastic Programming (SP) problems that deal with the service level requirement as a chance constraint or fill rate constraint. The studied problems vary in planning characteristics. Fluctuations in demand combined with fixed setup or ordering costs imply that regular production or ordering is probably not optimal. This situation requires a strategy to deal with the fluctuations. In inventory literature, non-stationary demand for a perishable product with a fixed lifetime is hardly studied, and consequently also strategies to deal with it are little investigated. In this thesis, methods have been designed to generate practical order policies using commercial solvers, where at least the timing of ordering or production is set beforehand. The practical order policies have been benchmarked with custom made solution procedures. The developed practical policies are applicable in the practice of fresh food producers and in retail organisations, on the level of the store, as well as on the level of the distribution centre. An interesting conclusion is that the policies with fixed order timing are hardly more costly than the investigated policy with flexible order timing. The fixed order timing increases the ease of planning. The studied cost functions in the models comprise disposal cost for which the value of the disposal cost in the experiments was varied. The results show that in the models with fixed order timing and corresponding order-up-to levels, increasing the disposal cost can reduce the percentage of waste, maintaining the service level. In setting a (fictitious) value for the disposal cost, one can steer the percentage of waste in determining the order policy, and thus contribute to better decision making in the food supply chain from food producer to supermarket in the trade-off between product-availability and waste.

The world’s marine fisheries are characterized by declining global catch, an increasing number of overexploited stocks and high natural variability in fish stocks. Policy makers are becoming more aware that effective management systems have to be implemented to rebuild overexploited fish stocks. The success of a management system is specified in terms of biological, economic, social and political objectives of policy makers and the fishing industry. However, the combination of these objectives makes the implementation of policies one of the main challenges in fisheries management. To prevent overfishing, different measures have been applied that limit catch and/or fleet capacity. Yet, the same management systems and economic behavior of fishermen may lead to an increase in investments in the fleet capacity, causing overcapacity. Decisions of policy makers and fishermen are made under uncertainty, such as uncertainty about fish stock dynamics and/or fish prices, and this can affect optimal management, investment decisions, healthy and productive fish stocks and resource rents. The objective of this thesis is to study the impact of different fisheries management systems on resource rents from the fishery, investment in the fleet capacity and fish stock under uncertainty about fish stock growth and the ex-vessel price of fish. Bi-level stochastic dynamic programming is used to model the interaction between the quota decision of a policy maker and fishermen behavior. The fisheries management systems of multiannual quota and a quota adjustment restriction are introduced and analyzed in terms of resource rents, overcapacity and fish stock. Real Options theory is used to determine for a fisheries management system of limited entry at what ex-vessel prices it is optimal for fishermen to make investment and disinvestment decisions. De zeevisvangst in de wereld wordt gekarakteriseerd door afnemende vangsten, toenemende overbevissing en grote natuurlijke variatie in visstanden. Beleidsmakers zijn er zich steeds meer van bewust dat effectieve managementsystemen moeten worden ingevoerd om uitgeputte visbestanden weer op te bouwen. Het succes van een managementsysteem wordt gespecificeerd in termen van biologische, economische, sociale en politieke doelstellingen van beleidsmakers en de visserij industrie. Echter, de combinatie van deze doelstellingen maakt het invoeren van beleid ´e´en van de belangrijkste uitdagingen. Om overbevissing te voorkomen zijn er verschillende maatregelen genomen om de vangst te beperken en/of de vlootcapaciteit te verminderen. Toch kunnen deze managementsystemen, als zij onvoldoende rekening houden met het economisch gedrag van vissers, leiden tot toenemende investeringen in vlootcapaciteit en mogelijk overcapaciteit veroorzaken. Beslissingen van beleidsmakers en vissers worden gemaakt op basis van onzekerheid over dynamiek in visstanden en/of visprijzen en dat kan invloed hebben op optimaal management, investeringsbeslissingen, gezonde en productieve visstanden en winsten. Het doel van dit proefschrift is om de invloed te bestuderen van verschillende visserijmanagementsystemen op winsten, investeringen in de vlootcapaciteit en visstanden onder onzekerheid over de groei van visbestanden en de prijs van vis. Een stochastisch dynamisch programmeringsmodel met twee niveaus is gebruikt om de interactie te modelleren tussen de quotabeslissing van een beleidsmaker en vissers. De visserijmanagementsystemen van meerjarenquota en een aanpassingsrestrictie voor quota zijn ingevoerd en bestudeerd in termen van winsten, overcapaciteit en visstand. De Re¨ele Optie Theorie is gebruikt om te bepalen bij welke visprijzen het optimaal is voor vissers om investerings- en desinvesteringsbeslissingen te maken.

The operation of supply chains (SCs) has for many years been focused on efficiency, leanness and responsiveness. This has resulted in reduced slack in operations, compressed cycle times, increased productivity and minimised inventory levels along the SC. Combined with tight tolerance settings for the realisation of logistics and production processes, this has led to SC performances that are frequently not robust. SCs are becoming increasingly vulnerable to disturbances, which can decrease the competitive power of the entire chain in the market. Moreover, in the case of food SCs non-robust performances may ultimately result in empty shelves in grocery stores and supermarkets. The overall objective of this research is to contribute to Supply Chain Management (SCM) theory by developing a structured approach to assess SC vulnerability, so that robust performances of food SCs can be assured. We also aim to help companies in the food industry to evaluate their current state of vulnerability, and to improve their performance robustness through a better understanding of vulnerability issues. The following research questions (RQs) stem from these objectives: RQ1: What are the main research challenges related to (food) SC robustness? RQ2: What are the main elements that have to be considered in the design of robust SCs and what are the relationships between these elements? RQ3: What is the relationship between the contextual factors of food SCs and the use of disturbance management principles? RQ4: How to systematically assess the impact of disturbances in (food) SC processes on the robustness of (food) SC performances? To answer these RQs we used different methodologies, both qualitative and quantitative. For each question, we conducted a literature survey to identify gaps in existing research and define the state of the art of knowledge on the related topics. For the second and third RQ, we conducted both exploration and testing on selected case studies. Finally, to obtain more detailed answers to the fourth question, we used simulation modelling and scenario analysis for vulnerability assessment. Main findings are summarised as follows. Based on an extensive literature review, we answered RQ1. The main research challenges were related to the need to define SC robustness more precisely, to identify and classify disturbances and their causes in the context of the specific characteristics of SCs and to make a systematic overview of (re)design strategies that may improve SC robustness. Also, we found that it is useful to be able to discriminate between varying degrees of SC vulnerability and to find a measure that quantifies the extent to which a company or SC shows robust performances when exposed to disturbances. To address RQ2, we define SC robustness as the degree to which a SC shows an acceptable performance in (each of) its Key Performance Indicators (KPIs) during and after an unexpected event that caused a disturbance in one or more logistics processes. Based on the SCM literature we identified the main elements needed to achieve robust performances and structured them together to form a conceptual framework for the design of robust SCs. We then explained the logic of the framework and elaborate on each of its main elements: the SC scenario, SC disturbances, SC performance, sources of food SC vulnerability, and redesign principles and strategies. Based on three case studies, we answered RQ3. Our major findings show that the contextual factors have a consistent relationship to Disturbance Management Principles (DMPs). The product and SC environment characteristics are contextual factors that are hard to change and these characteristics initiate the use of specific DMPs as well as constrain the use of potential response actions. The process and the SC network characteristics are contextual factors that are easier to change, and they are affected by the use of the DMPs. We also found a notable relationship between the type of DMP likely to be used and the particular combination of contextual factors present in the observed SC. To address RQ4, we presented a new method for vulnerability assessments, the VULA method. The VULA method helps to identify how much a company is underperforming on a specific Key Performance Indicator (KPI) in the case of a disturbance, how often this would happen and how long it would last. It ultimately informs the decision maker about whether process redesign is needed and what kind of redesign strategies should be used in order to increase the SC’s robustness. The VULA method is demonstrated in the context of a meat SC using discrete-event simulation. The case findings show that performance robustness can be assessed for any KPI using the VULA method. To sum-up the project, all findings were incorporated within an integrated framework for designing robust SCs. The integrated framework consists of the following steps: 1) Description of the SC scenario and identification of its specific contextual factors; 2) Identification of disturbances that may affect KPIs; 3) Definition of the relevant KPIs and identification of the main disturbances through assessment of the SC performance robustness (i.e. application of the VULA method); 4) Identification of the sources of vulnerability that may (strongly) affect the robustness of performances and eventually increase the vulnerability of the SC; 5) Identification of appropriate preventive or disturbance impact reductive redesign strategies; 6) Alteration of SC scenario elements as required by the selected redesign strategies and repeat VULA method for KPIs, as defined in Step 3. Contributions of this research are listed as follows. First, we have identified emerging research areas - SC robustness, and its counterpart, vulnerability. Second, we have developed a definition of SC robustness, operationalized it, and identified and structured the relevant elements for the design of robust SCs in the form of a research framework. With this research framework, we contribute to a better understanding of the concepts of vulnerability and robustness and related issues in food SCs. Third, we identified the relationship between contextual factors of food SCs and specific DMPs used to maintain robust SC performances: characteristics of the product and the SC environment influence the selection and use of DMPs; processes and SC networks are influenced by DMPs. Fourth, we developed specific metrics for vulnerability assessments, which serve as a basis of a VULA method. The VULA method investigates different measures of the variability of both the duration of impacts from disturbances and the fluctuations in their magnitude. With this project, we also hope to have delivered practical insights into food SC vulnerability. First, the integrated framework for the design of robust SCs can be used to guide food companies in successful disturbance management. Second, empirical findings from case studies lead to the identification of changeable characteristics of SCs that can serve as a basis for assessing where to focus efforts to manage disturbances. Third, the VULA method can help top management to get more reliable information about the “health” of the company. The two most important research opportunities are: First, there is a need to extend and validate our findings related to the research framework and contextual factors through further case studies related to other types of (food) products and other types of SCs. Second, there is a need to further develop and test the VULA method, e.g.: to use other indicators and statistical measures for disturbance detection and SC improvement; to define the most appropriate KPI to represent the robustness of a complete SC. We hope this thesis invites other researchers to pick up these challenges and help us further improve the robustness of (food) SCs.