Hüseyin Sarper

Minimizing the sum of absolute deviations about a common due date for the two-machine flow shop problem

Abstract This paper deals with the minimization of the sum of absolute deviations (SAD) of job completion times from a common due date for the two-machine flow shop problem. The problem is mathematically modelled using a mixed integer programming formulation. Heuristic solution methods are developed and the performance of each heuristic is compared with the optimal solution for data sets with a small number of jobs. Three heuristics are selected as the best performing and are compared with each other using larger data sets. The methodology is also compared with available two-machine methods for other criteria. It is shown that the SAD criterion is not generally compatible with other criteria. This paper is the first extension of the SAD criterion to the two-machine case.

CAPITAL RATIONING UNDER RISK: A CHANCE CONSTRAINED APPROACH USING UNIFORMLY DISTRIBUTED CASH FLOWS AND AVAILABLE BUDGETS

ABSTRACT This paper presents pure capital rationing and selection of the best capital project mix when cash flows and available budgets are uniformly distributed random variables. Chance-constrained programming is used by approximating uniformly distributed random variables into normally distributed ones. Thus, a methodology is proposed to handle uniformly distributed random variables in capital rationing. The accuracy of the results is checked using the Monte-Carlo simulation technique. A literature review on the inclusion of risk into capital investment decisions using chance-constrained programming is also provided.

Estimating the effects of parameter variability on learning curve model predictions

Abstract The learning curve concept has proven to be a valuable management tool. However, regardless of which learning curve model is used, uncertainty is inherent in the forecast due to the empirical nature of learning curve theory and complications with establishing model parameters. Such variability is often ignored but can greatly affect the reliability of the models predictions. Thus, as a means of approximating the effects of such uncertainty on model predictions, this paper proposes an analytical stochastic approach to estimating the precision of learning curve forecasts and provides an illustration of the technique with actual product cost data. The example shows that this analytical stochastic approach can provide accurate cost predictions with reliable prediction interval estimates.

Comparison of equivalent pure cellular and functional production environments using simulation

Abstract Two alternative production environments are compared by allocating equivalent machining capacity to each one. Equivalent capacity of each environment has been found by minimizing the total daily production cost under each choice. SIMAN simulation language is used in determination of non-cost performance measures for each environment. This study assumes that lot sizes are fixed and cannot be reduced under the cellular option. Results support the recent findings that the cellular manufacturing or group technology method of production is not superior to traditional functional or job shop layout.

Probability Distribution Function of the Internal Rate of Return for Short-Term Projects with Some Random Cash Flows and Extensions

This article reports closed-form probability density function and related expressions of the internal rate of return for one- and two-period pure and simple probabilistic engineering economy problems. The random cash flows are either uniform or exponential variables. In addition, there is no correlation between the cash flows. The probability density function of the internal rate of return is an important tool that helps decision makers estimate the probability of the internal rate of return exceeding the minimum attractive rate of return. Hilliers method (1963) using normal cash flows is also evaluated when the cash flows are uniformly distributed.

Evaluation of simple dispatching rules for dynamic single-machine earliness/tardiness problem

Abstract Abstract. This paper investigates the effects of four simple dispatching rules on just-in-time production related performance measures of mean and maximum absolute lateness. The rules used are modified due date (MDD), shortest processing time (SPT), earliest due date (EDD), and first in first out (FIFO). A single machine is used under three utilization levels. Due-dates are set according to total work content rule. The results indicate that each rule performs well under certain conditions. The MDD rule is the best one to minimize mean absolute lateness. The EDD and FIFO rules do well in minimizing the maximum absolute lateness. Economic interpretation of these performance measures are also discussed.

Monte Carlo simulation for analysis of the optimum value distribution in stochastic mathematical programs

This paper shows how simulation can be used to quickly solve an otherwise complex mathematical problem of derivation of the optimum value distribution function in stochastic mathematical programs. Two software packages (LINDO and UNIFIT II) are used along with a FORTRAN code in simulating and analyzing a given linear program with stochastic coefficients. Two previously analytically solved problems are resolved using simulation to show that simulation is indeed an effective method of solution.

Reliability analysis of descent systems of planetary vehicles using bivariate exponential distribution

This paper evaluates the accuracy of two Monte-Carlo simulation methods that draw the minimum of two correlated exponentials. This evaluation is done by comparing the simulation results against the results from two well-known analytical bivariate exponential distribution (BED) sources. Then, the future systems are simulated to obtain the probability density functions (PDF) of the system lives from which the reliability expressions are derived. Results show that simulation is a viable tool in analyzing the reliability of systems composed of correlated components. Positive correlation seems to increase both the mean and the standard deviation of the lives of both systems.

Scheduling for the maintenance of completely processed low-demand large items

Abstract This paper presents a mathematical model to load the existing maintenance/repair capacity with low-demand large items that are classified for various reasons. The objective is to minimize the lost sales cost while ensuring that no job is left unfinished. This model could be used in making scheduling decisions involving expensive and classified items that must be delivered by a self-imposed common due date. A sample problem is provided.

The reliability of correlated two unit systems

The classic problem of the extremes of two unit systems is revisited by using an explicit correlation coefficient (rho) between the units. Two unit systems have long been studied in reliability literature as these systems serve as the foundation of the serial and parallel systems. Often, the units are assumed to be independent, but this is not always the case in reality. The effect and inclusion of correlation in reliability analysis is often an unclear area. The purpose of this paper is two-fold: i) to provide clear probability density, mean, and variance expressions for two unit systems where the bivariate exponential and bivariate normal distributions hold and ii) to show how the available simulation schemes compare to some known analytical solutions. Results are stated in terms of the correlation coefficient, and include probability density function (PDF) and parameter expressions needed for reliability analysis and comments on the performance of simulation methods. As simulation works well for known cases, it can be used with high confidence in cases where analytical solutions do not exist