Alternative manufacturing infrastructures for a special class of arborescent production processes.

Abstract

Two major models are developed in this dissertation. The first model provides a framework for the selection of an appropriate infrastructure for a special class of arborescent (tree-like) production processes. An example is provided wherein a significant redesign of a business's infrastructure was required in order for it to effectively support the company's manufacturing strategy. The research methodologies used include an in-depth case analysis and a simulation model of the proposed inventory and scheduling systems that validates the recommended changes. The second model, which is an important component of the above redesign, is an algorithm for the solution of a new one-dimensional cutting stock problem. Cutting stock problems arise when master rolls of material must be cut into smaller rolls with a diversity of widths. The unique aspect of the cutting stock problem considered here is that both the master rolls and customer orders have multiple quality gradations. The solution methodology is two-stage and heuristic. In the first stage, a sequential procedure is used to make slitting decisions for master rolls with variable quality characteristics. In the second stage, a residual problem for any remaining order requirements is solved with a linear programming model using available first-quality rolls. An important characteristic of the approach is its robustness. The model can deal effectively with problems of varying size and complexity, and can also easily adapt to changing circumstances with respect to production quality and dem and . The cutting stock model is tested on a series of simulated problems and the results are compared with answers from a slightly modified second model which mimics but improves upon existing manual approaches. The results clearly dominate the solutions from the second model with respect to yield and nearly do so with respect to the level of inventory created. In comparison to manually generated answers, solutions provided by the model during its early stages of implementation require an average of 8% fewer master rolls to satisfy each batch of orders. The benefits to an organization which applies these research results include improvements in customer service, scheduling flexibility, lower investment in inventories, and reductions in cost.