A modelling approach of two-echelon inventory systems.
Item
-
Title
-
A modelling approach of two-echelon inventory systems.
-
Identifier
-
AAI9807959
-
identifier
-
9807959
-
Creator
-
Liu, Hua.
-
Contributor
-
Adviser: Georghios P. Sphicas
-
Date
-
1997
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Business Administration, Management | Engineering, Industrial
-
Abstract
-
To provide good service to customers, many firms use multi-echelon inventory systems. There are two types of inventory replenishment policies in multi-echelon inventory models. The one-for-one replenishment policy has commonly been applied. The batch order policy is much more complicated and more challenging. In our study, several typical multi-echelon inventory models are presented. The differences among these models have been elaborated. We proposed a new approach to model two-echelon inventory systems.;Most two-echelon (Q, R) studies have similar assumptions. The major difference lies in the modelling of the depot demand process. In our study, the depot demand is approximated by a Poisson process. We investigate the steady state of the two-echelon inventory system. The depot demand process in computer simulations is tested with several goodness of fit tests against a Poisson process and a normal process. We also run the run-ups test and correlation test for the independence of base orders at the depot. The number of bases and the base order size affect the approximation of the demand process at the depot. As the number of bases increases and/or the base order size decreases, the approximation of the demand process at the depot by a Poisson process is improved. We compare the test power of these goodn ess of fit tests. Our conclusion is that the most commonly used {dollar}\chi\sp2{dollar} test has a very low test power and should not be used alone.;The inventory system performance is not very sensitive to the demand process. We compare our simulation results with the results of our analytical model when the approximation of depot demand process by a Poisson process is not good. We conclude that the results with our approach of modelling the two-echelon inventory systems are consistent with simulation results.;In order to obtain the global minimum total inventory cost, the total inventory cost should be a convex function of the decision variables. Conditions of the convexity of the total inventory cost are studied in detail. We point out that certain parts of the inventory system cost, such as the holding cost at bases, are not convex with respect to some decision variables. The sufficient condition of the convexity of the total holding cost with respect to the depot reorder point is obtained. Since the proof of the convexity of the total cost is very long, we put the complete proof in the appendix. An algorithm to obtain the optimal values of decision variables and the minimum total inventory cost is developed. Numerical examples show that computer simulations and our approach to model the inventory system produce very close results.;We study the effect of system parameters on the minimum total cost and the optimal values of decision variables. A table to show the general trend of changes in decision variables with respect to changes of the system parameter is obtained. Most results in that table are consistent with what we expected. Because of the interactions among the system decision variables and the assumption that decision variables only have integer values, some results of our algorithm do not follow the general trend. These intuitive results are explained in detail at the related chapter. We also determine the priority of the decision variables on the system total cost. Possible future research topics are discussed at the end of our study.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
