On the effectiveness of top‐down strategy for forecasting autoregressive demands

We investigate the relative effectiveness of top‐down versus bottom‐up strategies for forecasting the demand of an item that belongs to a product family. The demand for each item in the family is assumed to follow a first‐order univariate autoregressive process. Under the top‐down strategy, the aggregate demand is forecasted by using the historical data of the family demand. The demand forecast for the items is then derived by proportional allocation of the aggregate forecast. Under the bottom‐up strategy, the demand forecast for each item is directly obtained by using the historical demand data of the particular item. In both strategies, the forecasting technique used is exponential smoothing. We analytically evaluate the condition under which one forecasting strategy is preferred over the other when the lag‐1 autocorrelation of the demand time series for all the items is identical. We show that when the lag‐1 autocorrelation is smaller than or equal to 1/3, the maximum difference in the performance of the two forecasting strategies is only 1%. However, if the lag‐1 autocorrelation of the demand for at least one of the items is greater than 1/3, then the bottom‐up strategy consistently outperforms the top‐down strategy, irrespective of the items' proportion in the family and the coefficient of correlation between the item demands. A simulation study reveals that the analytical findings hold even when the lag‐1 autocorrelation of the demand processes is not identical. © 2006 Wiley Periodicals, Inc. Naval Research Logistics, 2007.     

Modeling lead-time demand for lumpy demand and variable lead time

Slow-moving items that occasionally exhibit large demand transactions are known as lumpy demand items. In modeling lumpy demand patterns, it is often assumed that the arrival of customer orders follows a Poisson process and that the order sizes are given by the geometric distribution. This gives rise to a stuttering Poisson (sP) model of lumpy demand. If lead times are constant, the result is a stuttering Poisson model of lead-time demand. Heretofore, authors such as Ward [18] and Mitchell, Rappold, and Faulkner [12] have assumed constant lead times and thus stopped at the sP model. We develop this model further by introducing the effect of lead-time variability. For illustration, we use the normal and the gamma distributions as characterizations of lead time. The resulting models of lead-time demand are referred to as the geometric Poisson normal (GPN) and the geometric Poisson gamma (GPG). For both these models, the article derives tractable expressions for calculating probabilities. Errors introduced by using the sP, constant lead-time model instead of the exact, variable lead-time model are also illustrated.     

An economic sequential screening procedure for limited failure populations

An economic sequential screening procedure is considered for limited failure populations in which defective items fail soon after they are put in operation and nondefective ones never fail during the technical life of the items. A cost model is constructed that involves screening test cost and external failure cost. A sequential scheme that minimizes the expected cost is derived from the solution of a dynamic programming formulation and the optimal decision at each stage is obtained in a closed form. © 1994 John Wiley & Sons, Inc.     

The many-product cargo loading problem

An algorithm, based upon dynamic programming, is developed for a class of fixed-cost cargo loading problems. The problems can be formulated as integer programming problems, but cannot be efficiently solved as such because of computational difficulties. The algorithm developed has proved to be very efficient in an actual operations research study involving over 500 different cargo items, more than 40 possible stops and several types of transportation vehicles. A numerical illustration is provided.     

Optimal clustering: A model and method

Classifying items into distinct groupings is fundamental in scientific inquiry. The objective of cluster analysis is to assign n objects to up to K mutually exclusive groups while minimizing some measure of dissimilarity among the items. Few mathematical programming approaches have been applied to these problems. Most clustering methods to date only consider lowering the amount of interaction between each observation and the group mean or median. Clustering used in information systems development to determine groupings of modules requires a model that will account for the total group interaction. We formulate a mixed-integer programming model for optimal clustering based upon scaled distance measures to account for this total group interaction. We discuss an efficient, implicit enumeration algorithm along with some implementation issues, a method for computing tight bounds for each node in the solution tree, and a small example. A computational example problem, taken from the computer-assisted process organization (CAPO) literature, is presented. Detailed computational results indicate that the method is effective for solving this type of cluster analysis problem.     

The dynamic and stochastic knapsack Problem with homogeneous‐sized items and postponement options

This article generalizes the dynamic and stochastic knapsack problem by allowing the decision‐maker to postpone the accept/reject decision for an item and maintain a queue of waiting items to be considered later. Postponed decisions are penalized with delay costs, while idle capacity incurs a holding cost. This generalization addresses applications where requests of scarce resources can be delayed, for example, dispatching in logistics and allocation of funding to investments. We model the problem as a Markov decision process and analyze it through dynamic programming. We show that the optimal policy with homogeneous‐sized items possesses a bithreshold structure, despite the high dimensionality of the decision space. Finally, the value (or price) of postponement is illustrated through numerical examples. © 2015 Wiley Periodicals, Inc. Naval Research Logistics 62: 267–292, 2015     

Bayes sequential design of stock levels

Bayesian determination of optimal stock levels is studied for the case of Poisson distribution of the demand variable, and prior gamma distribution of the expected demand. Bayes sequential procedure is derived, assuming that stock level can be adjusted at the beginning of each period so that a shortage can be immediately replenished and an overstock can be corrected. The Bayes sequential procedure is more difficult to obtain if this assumption is removed. A dynamic programming method for obtaining the general Bayes sequential procedure is outlined. Finally, an empiric Bayes estimation procedure of the optimal Bayesian stock level is presented.     

A time dependent continuous review inventory model with compound poisson demand

We discuss a time dependent optimal ordering policy for a finite horizon inventory system for which the provision of service is essential and thus no stockout is allowed. It is assumed that the system can place an order at any point in time during the horizon when it cannot meet the customer's demand and that lead time is negligible. The demand is considered to be distributed as a compound Poisson process with known parameters and the functional equation approach of dynamic programming is used to formulate the objective function. An algorithm has been developed to obtain the solution for all the cases. In addition, analytical solutions of the basic equation under two limiting conditions are presented.     

Algorithms and heuristics for variable-yield lot sizing

We consider the multiperiod lot-sizing problem in which the production yield (the proportion of usable goods) is variable according to a known probability distribution. We review two economic order quantity (EOQ) models for the stationary demand continuous-time problem and derive an EOQ model when the production yield follows a binomial distribution and backlogging of demand is permitted. A dynamic programming algorithm for an arbitrary sequence of demand requirements is presented. Heuristics based on both the EOQ model and appropriate modification of the underlying perfect-yield lot-sizing policies are discussed, and extensive computational evaluation of these heuristics is presented. Two of these heuristics are then modified to include the notion of supply safety stock. The modified heuristics consistently produce near-optimal lot-sizing policies for problems with stationary and time-varying demands.     

Cyclical schedules for the joint replenishment problem with dynamic demands

The Joint Replenishment Problem (JRP) involves production planning for a family of items. The items have a coordinated cost structure whereby a major setup cost is incurred whenever any item in the family is produced, and an item-specific minor setup cost is incurred whenever that item is produced. This paper investigates the performance of two types of cyclical production schedules for the JRP with dynamic demands over a finite planning horizon. The cyclical schedules considered are: (1) general cyclical schedules—schedules where the number of periods between successive production runs for any item is constant over the planning horizon—and (2) power-of-two schedules—a subset of cyclical schedules for which the number of periods between successive setups must be a power of 2. The paper evaluates the additional cost incurred by requiring schedules to be cyclical, and identifies problem characteristics that have a significant effect on this additional cost. © 1997 John Wiley & Sons, Inc. Naval Research Logistics 44: 577–589, 1997.     

The dynamic transportation problem: A survey

The dynamic transportation problem is a transportation problem over time. That is, a problem of selecting at each instant of time t, the optimal flow of commodities from various sources to various sinks in a given network so as to minimize the total cost of transportation subject to some supply and demand constraints. While the earliest formulation of the problem dates back to 1958 as a problem of finding the maximal flow through a dynamic network in a given time, the problem has received wider attention only in the last ten years. During these years, the problem has been tackled by network techniques, linear programming, dynamic programming, combinational methods, nonlinear programming and finally, the optimal control theory. This paper is an up-to-date survey of the various analyses of the problem along with a critical discussion, comparison, and extensions of various formulations and techniques used. The survey concludes with a number of important suggestions for future work.     

Diagnostic analysis of inventory systems: An optimization approach

The objective of a diagnostic analysis is to provide a measure of performance of an existing system and estimate the benefits of implementing a new one, if necessary. Firms expect diagnostic studies to be done promptly and inexpensively. Consequently, collection and manipulation of large quantities of data are prohibitive. In this paper we explore aggregate optimization models as tools for diagnostic analysis of inventory systems. We concentrate on the dynamic lot size problem with a family of items sharing the same setup, and on the management of perishable items. We provide upper and lower bounds on the total cost to be expected from the implementation of appropriate systems. However, the major thrust of the paper is to illustrate an approach to analyze inventory systems that could be expanded to cover a wide variety of applications. A fundamental by-product of the proposed diagnostic methodology is to identify the characteristics that items should share to be aggregated into a single family.     

A linear programming model for design of communications networks with time varying probabilistic demands

In this paper marginal investment costs are assumed known for two kinds of equipment stocks employed to supply telecommunications services: trunks and switching facilities. A network hierarchy is defined which includes important cases occurring in the field and also appearing in the literature. A different use of the classical concept of the marginal capacity of an additional trunk at prescribed blocking probability leads to a linear programming supply model which can be used to compute the sizes of all the high usage trunk groups. The sizes of the remaining trunk groups are approximated by the linear programming models, but can be determined more accurately by alternate methods once all high usage group sizes are computed. The approach applies to larger scale networks than previously reported in the literature and permits direct application of the duality theory of linear programming and its sensitivity analyses to the study and design of switched probabilistic communications networks with multiple busy hours during the day. Numerical results are presented for two examples based on field data, one of which having been designed by the multi-hour engineering method.     

Outsourcing warranty repairs: Dynamic allocation

In this paper we consider the problem of minimizing the costs of outsourcing warranty repairs when failed items are dynamically routed to one of several service vendors. In our model, the manufacturer incurs a repair cost each time an item needs repair and also incurs a goodwill cost while an item is awaiting and undergoing repair. For a large manufacturer with annual warranty costs in the tens of millions of dollars, even a small relative cost reduction from the use of dynamic (rather than static) allocation may be practically significant. However, due to the size of the state space, the resulting dynamic programming problem is not exactly solvable in practice. Furthermore, standard routing heuristics, such as join‐the‐shortest‐queue, are simply not good enough to identify potential cost savings of any significance. We use two different approaches to develop effective, simply structured index policies for the dynamic allocation problem. The first uses dynamic programming policy improvement while the second deploys Whittle's proposal for restless bandits. The closed form indices concerned are new and the policies sufficiently close to optimal to provide cost savings over static allocation. All results of this paper are demonstrated using a simulation study. © 2005 Wiley Periodicals, Inc. Naval Research Logistics, 2005     

Expected time delay in multi‐item inventory systems with correlated demands

This paper considers multi‐item inventory systems where a customer order may require several different items (i.e., demands are correlated across items) and customer satisfaction is measured by the time delays seen by the customers. Most inventory models on time delay in the literature assume each demand only requires one item (i.e., demands are not correlated across items or are independent). In this paper, we derive an exact expression for the expected total time delay. We show that when items are actually correlated, assuming items are independent leads to an overestimate of the total time delay. However, (1) it is extremely difficult in practice to obtain the demand information for all demand types (especially in a system with tens of thousands of part numbers), and (2) the problem becomes too complicated to be of practical interest when the correlation is considered. We then explore the possibility of including the demand information partially and develop bounds for the time delays. © 1999 John Wiley & Sons, Inc. Naval Research Logistics 46: 671–688, 1999     

Exact analysis of a two-echelon inventory system for recoverable items under batch inspection policy

We investigate a two-echelon (base-depot) inventory system of recoverable (repairable) items. The arrivals of demand at the bases are in a Poisson manner and the order sizes are random. The failed units can be repaired either at the base or at the depot, and the units beyond economic repair are condemned. Inspection of the failed units is carried out in the batches they arrive, that is, arrival batches are not broken up. The exact expressions for stationary distribution of depot inventory position, and of the number of backorders, onhand inventory, in-repair inventory at all locations are derived under the assumptions of constant repair and lead times. Special cases of complete recoverability, nonrecoverability, and of the unit order size are also discussed.     

Assemble to order and assemble in advance in a single-period stochastic environment

Faced with stochastic demand, a firm may decide to assemble its products in advance or assemble them once actual demand is realized. In general, the production cost for items assembled in advance (AIA) is lower than for items assembled to order (ATO), because there is no need to expedite, and the production process can be planned and executed well in advance. On the other hand, items assembled in advance (AIA) for which there is no demand incur excessive and unnecessary assembly costs. The two policies, AIA and ATO, as well as a composite one, are compared and analyzed in light of these trade-offs. The composite model, which is shown as the dominating policy, is also extended to deal with the following two scenarios. The first assumes a loss of a fraction of the demand when demand cannot be satisfied from the shelf but rather through ATO. The second considers the effects of budget constraints on the total inventory cost. © 1995 John Wiley & Sons, Inc.     

Demand shuffle—A method for multilevel proportional lot sizing and scheduling

This contribution acquaints the reader with a model for multilevel single-machine proportional lot sizing and scheduling problems (PLSPs) that appear in the scope of short-term production planning. It is one of the first articles that deals with dynamic capacitated multilevel lot sizing and scheduling, which is of great practical importance. The PLSP model refines well-known mixed-integer programming formulations for dynamic capacitated lot sizing and scheduling as, for instance, the DLSP or the CSLP. A special emphasis is given on a new method called demand shuffle to solve multilevel PLSP instances efficiently but suboptimally. Although the basic idea is very simple, it becomes clear that in the presence of precedence and capacity constraints many nontrivial details are to be concerned. Computational studies show that the presented approach decidedly improves recent results. © 1997 John Wiley & Sons, Inc. Naval Research Logistics 44: 319–340, 1997     

Note: On the set-union knapsack problem

We consider a generalization of the 0-1 knapsack problem called the set-union knapsack problem (SKP). In the SKP, each item is a set of elements, each item has a nonnegative value, and each element has a nonnegative weight. The total weight of a collection of items is given by the total weight of the elements in the union of the items' sets. This problem has applications to data-base partitioning and to machine loading in flexible manufacturing systems. We show that the SKP remains NP-hard, even in very restricted cases. We present an exact, dynamic programming algorithm for the SKP and show sufficient conditions for it to run in polynomial time. © 1994 John Wiley & Sons, Inc.     

Scheduling of depalletizing and truck loading operations in a food distribution system

This paper studies a scheduling problem arising in a beef distribution system where pallets of various types of beef products in the warehouse are first depalletized and then individual cases are loaded via conveyors to the trucks which deliver beef products to various customers. Given each customer's demand for each type of beef, the problem is to find a depalletizing and truck loading schedule that fills all the demands at a minimum total cost. We first show that the general problem where there are multiple trucks and each truck covers multiple customers is strongly NP‐hard. Then we propose polynomial‐time algorithms for the case where there are multiple trucks, each covering only one customer, and the case where there is only one truck covering multiple customers. We also develop an optimal dynamic programming algorithm and a heuristic for solving the general problem. By comparing to the optimal solutions generated by the dynamic programming algorithm, the heuristic is shown to be capable of generating near optimal solutions quickly. © 2003 Wiley Periodicals, Inc. Naval Research Logistics, 2003