A single‐period inventory placement problem for a serial supply chain

This article addresses the inventory placement problem in a serial supply chain facing a stochastic demand for a single planning period. All customer demand is served from stage 1, where the product is stored in its final form. If the demand exceeds the supply at stage 1, then stage 1 is resupplied from stocks held at the upstream stages 2 through N, where the product may be stored in finished form or as raw materials or subassemblies. All stocking decisions are made before the demand occurs. The demand is nonnegative and continuous with a known probability distribution, and the purchasing, holding, shipping, processing, and shortage costs are proportional. There are no fixed costs. All unsatisfied demand is lost. The objective is to select the stock quantities that should be placed different stages so as to maximize the expected profit. Under reasonable cost assumptions, this leads to a convex constrained optimization problem. We characterize the properties of the optimal solution and propose an effective algorithm for its computation. For the case of normal demands, the calculations can be done on a spreadsheet. © 2001 John Wiley & Sons, Inc. Naval Research Logistics 48:506–517, 2001     

A genetic algorithm methodology for complex scheduling problems

This paper considers the scheduling problem to minimize total tardiness given multiple machines, ready times, sequence dependent setups, machine downtime and scarce tools. We develop a genetic algorithm based on random keys representation, elitist reproduction, Bernoulli crossover and immigration type mutation. Convergence of the algorithm is proved. We present computational results on data sets from the auto industry. To demonstrate robustness of the approach, problems from the literature of different structure are solved by essentially the same algorithm. © 1999 John Wiley & Sons, Inc. Naval Research Logistics 46: 199–211, 1999     

An exact algorithm for the batch sequencing problem in a two-machine flow shop with limited buffer

This paper deals with the problem of makespan minimization in a flow shop with two machines when the input buffer of the second machine can only host a limited number of parts. Here we analyze the problem in the context of batch processing, i.e., when identical parts must be processed consecutively. We propose an exact branch-and-bound algorithm, in which the bounds exploit the batching nature of the problem. Extensive computational results show the effectiveness of the approach, and allow us to compare it with a previous heuristic approach. © 1998 John Wiley & Sons, Inc. Naval Research Logistics 45: 141–164, 1998     

A robust regression technique using compound estimation

Least squares fitting of regression models is a widely used technique. The presence of outliers in the data can have an adverse effect on the method of least squares, resulting in a model that does not adequately fit to the bulk of the data. For this situation, robust regression techniques have been proposed as an improvement to the method of least squares. We propose a robust regression procedure that performs well relative to the current robust methods against a variety of dataset types. Evaluations are performed using datasets without outliers (testing efficiency), with a large percentage of outliers (testing breakdown), and with high leverage outliers (testing bounded influence). The datasets are based on 2-level factorial designs that include axial points to evaluate leverage effects. A Monte Carlo simulation approach is used to evaluate the estimating capability of the proposed procedure relative to several competing methods. We also provide an application to estimating costs for government satellites. © 1998 John Wiley & Sons, Inc. Naval Research Logistics 45: 125–139, 1998     

On the orbit characteristics of the M/G/ 1 retrial queue

In teletraffic applications of retrial queues only the service zone is observable. Another part of a retrial queue, the orbit, which represents the delay before repeated attempts to get service, cannot be observed. Thus, it is very important to get general results about behavior of the orbit. We investigate two characteristics of the orbit, namely, the orbit busy period and the orbit idle period, which seem to be very useful from this point of view. © 1996 John Wiley & Sons, Inc.     

Optimal policies for stochastic clearing systems with time-dependent delay penalties

We study stochastic clearing systems with a discrete-time Markovian input process, and an output mechanism that intermittently and instantaneously clears the system partially or completely. The decision to clear the system depends on both quantities and delays of outstanding inputs. Clearing the system incurs a fixed cost, and outstanding inputs are charged a delay penalty, which is a general increasing function of the quantities and delays of individual inputs. By recording the quantities and delays of outstanding inputs in a sequence, we model the clearing system as a tree-structured Markov decision process over both a finite and infinite horizon. We show that the optimal clearing policies, under realistic conditions, are of the on-off type or the threshold type. Based on the characterization of the optimal policies, we develop efficient algorithms to compute parameters of the optimal policies for such complex clearing systems for the first time. We conduct a numerical analysis on the impact of the nonlinear delay penalty cost function, the comparison of the optimal policy and the classical hybrid policy (ie, quantity and age thresholds), and the impact of the state of the input process. Our experiments demonstrate that (a) the classical linear approximation of the cost function can lead to significant performance differences; (b) the classical hybrid policy may perform poorly (as compared to the optimal policies); and (c) the consideration of the state of the input process makes significant improvement in system performance.     

Properties of a source location estimator in the plane

The statistical properties of an estimator of a source location were established by simulation for the case in which the source location is estimated—using transformation of lines to points—from the angles in which different observers see the source, and both the assumed locations of the observer points and the observed angles are subject to error. It was found that for normal error distributions the estimator is unbiased, and the resulting estimates are approximately normally distributed with a small standard deviation. An easy-to-use and reliable forecasting formula was suggested to forecast the parameters of the distributions of the estimates for different observer-source relationships. © 1993 John Wiley & Sons, Inc.     

Optimizing disaster relief: Real-time operational and tactical decision support

We introduce a real-time decision support system which uses optimization methods, simulation, and the judgement of the decision maker for operational assignment of units to tasks and for tactical allocation of units to task requirements. The system, named ARES for the Greek god of war, accommodates a high degree of detail in the logistics of unit movements during operations, yet separates the assignment and allocation activities in a fashion which naturally accommodates human intervention and judgement—ARES is designed to assist the decision maker, not to replace him. ARES is demonstrated with a hypothetical scenario constructed for 14 Engineering Battalions of the Hellenic Army which are assigned 20 tasks employing 25 resource types in repairing major damage to public works following a great earthquake. (This hypothetical data was prepared prior to the earthquake in Kalamata near Athens on 13 September, 1986, and exhibits uncanny, but coincidental, resemblance to that real situation.) ARES is designed for use in real time, and quick data preparation is aided by the provision from published sources of standard data for many foreseeable tasks; this data can be quickly accessed via visual icons on a computer screen and customized for the actual work at hand. © 1993 John Wiley & Sons, Inc.     

Real-time task scheduling with overheads considered

n periodic tasks are to be processed by a single machine, where each task i has a maximum request rate or periodicity F_i, a processing time E_i, a deadline D_i, relative to each request of task i, a task-request interrupt overhead I_i, and a task-independent scheduling overhead S. Two scheduling strategies are considered for sequencing the execution of an arbitrary arrangement of task requests in time: the preemptive and the nonpreemptive earliest-deadline algorithms. Necessary and sufficient conditions are derived for establishing whether a given set of tasks can be scheduled by each scheduling strategy. The conditions are given in the form of limited simulations of a small number of well-defined task-request arrangements. If all simulations succeed, the schedule is feasible for the given set of tasks. If any simulation fails, the schedule is infeasible. While interrupt handling and scheduling overheads can be handled by such simulations, context switching overhead resulting from preemption cannot. A counterexample illustrates how the simulations fail to uncover unschedulable task sets when context switching overhead is considered.