The impact of component commonality in an assemble-to-order environment under supply and demand uncertainty

Abstract

A material requirements planning simulator with a two-level bill-of-material is used to study the impact of introducing component commonality into an assemble-to-order environment when demand is subject to random variations, and component procurement orders experience random delays. By using simulated data, our ANOVA results show that component commonality significantly interacts with existence of demand and supply chain uncertainties, and benefits of component commonality are most pronounced when both uncertainties exist.

Introduction

The assemble-to-order (ATO) strategy emerges in manufacturing environments where many finished products are assembled from a relatively small set of standard components and subassemblies. In a typical ATO manufacturing environment, components and subassemblies are acquired according to a forecast, while finished products are assembled only after actual customers' orders have been received. In other words, component and subassemblies are replenished in a make-to-stock (MTS) fashion, but finished products are assembled in a make-to-order (MTO) manner. Such a hybrid planning approach is particularly advantageous in situations where the assembly time of a product is considerably shorter than the procurement and/or manufacturing time of its components and subassemblies; thus, making a tradeoff between inventory holding cost, product variety, and delivery time achievable [1]. Examples of ATO systems can be found in various industries producing consumer goods such as automobiles and personal computers where customers are offered a variety of product options with a relatively short delivery time.

It is well known that the on-time delivery of assembly system in general is diminished by the shortages of components. Experiencing such shortages often results in production capacity losses and/or customers' goodwill loss induced by missing planned product delivery dates. The primary cause of component shortage is the inherent uncertainty associated with procurement and/or suppliers' manufacturing lead times, which can be attributed to a variety of reasons ranging from unexpected delays in shipping, transportation, and receiving times to variable setup, processing and inspection times. Yano [2], Hopp and Spearman [3], and Hegedus and Hopp [4], among others, discuss the issue of lead-time uncertainty in assembly systems.

Another factor that is known to influence the performance of an assembly system is the commonality of components among products. Unlike the element of lead-time uncertainty which is an operating characteristic of a manufacturing/supply system, the commonality of components is an attribute of product design decisions. For instance, one way for increasing component commonality is by designing multiple-feature components that can be used in different products. Often all products do not need all the features and each product requires only a subset of the designed features. A car dashboard, for example, can be designed with multiple features to fit a variety of models without increasing the number of stock keeping units (SKUs) and the associated ordering and holding costs. It should be noted that increasing the number of components' features may increase their production costs; however, the resulting productivity gains from the economy of scale in procurement as well as from cost savings in warehousing and manufacturing operations in most cases offset the potential increase in production costs [5].

This paper investigates the desirability of increasing component commonality in ATO systems when product demand and component procurement lead times are random variables. We conduct a comprehensive simulation study to reveal the level of complex interactions among factors affecting the performance of an ATO system. Our findings provide significant insights in managing component inventories in ATO systems.

The organization of the paper is as follows. In Section 2, we identify the research gap in the literature that motivates our study. Section 3 contains a detailed description of our simulation experiment. The experimental results are presented in Section 4, followed by a summary of our findings in Section 5. Section 6 outlines the limitations of the present study and proposes a few related directions for future research.