Performance Analysis of Assemble-to-order Systems Through Strong Approximations.

Yingdong Lu,Jing-Sheng Song,Weian Zheng
DOI: https://doi.org/10.1145/507553.507572
2001-01-01
Abstract:The core of Assemble-to-Order is a multi-product multi-item inventory system. The set of item usually consists all the component( subassembly, or product). And each product requires a subset of items. The concept of Assemble-to-Order is popularized by Dell Computer's successful business practice. The key element in its practice is to delay the final assemble stage of the product until the order is realized. Hence, the inventory control are realized at component level. The relatively negligible assembly time required for computer products makes it feasible; and the short life cycle of the final product, comparing that to the component, proved the value of the model. As pointed out in [4], the common weakness of the all the above efforts have been lack of efficient computational techniques for dealing with problems other than experimental small scale ones. Meanwhile, the diffusion approximation schemes, in which the state process is being approximated by a diffusion process, has been very proved successful in analyzing queueing and queueing network models. In this paper, we intend to extend the Diffusion approximation techniques, strong approximation, to be exact, to the study of assemble-to-Order systems, as well as explore the structure of the limiting process itself, and then apply them in facilitate the computations of key functionais.
What problem does this paper attempt to address?