A systems approach for assigning printed circuit board inner layers to product clusters based on product attributes.

摘要

Production and capacity planning are critical processes that require a high level of detail to provide accurate and precise unit hour measurements. However, as new products are introduced and the development of existing products is enhanced, the ability to track, monitor and maintain these standards becomes a more difficult task. Product complexity, customer requirements and advances in technology have all contributed to the diversity of product lines. Furthermore, traditional manufacturing processes of products have yielded to more tailored operations for the new product sets. In addition, previous research attempts have focused on Group Technology (GT) and Design for Manufacturing (DFM) activities that accounted for a limited set of product attributes. Therefore, a need exists to cluster products in a more efficient manner for a variety of applications.; This thesis is one of the first attempts that utilized a systems approach, namely General System Problem Solving (GSPS), to cluster Printed Circuit Board (PCB) inner layers. Products were grouped based on the most similar and significant product attributes for a unit hour development. Also proposed in this thesis is to reduce the number of Industrial Engineering (IE) groups for the PCB inner layers. A PCB inner layer hierarchy was developed to systematically identify the critical factors. The systems approach attempted to account for systems that: (i) minimized the amount of uncertainty, and (ii) maximized the relationship to the response variables.; A primary benefit in utilizing the GSPS framework is that complex problems are reduced to a more simplified and manageable state. Moreover, it systematically reduced the amount of uncertainty contained in the problem as it progressed through the source, data, generative and structure system states. The Design of Experiment (DOE) utilized in this research highlighted the following product attributes: (i) test, (ii) core type, (iii) board format, (iv) copper weight, and (v) repairs as having the largest affect on Raw Processing Time (RPT). Equivalent classes were constructed, on a three level state, based upon the aforementioned factors. Product clusters based on the product attribute methodology reduced the number of IE groups from 273 to 81, a large contribution of this thesis.; The generative system state identified that the product clusters needed to be analyzed periodically. A Decision Support System (DSS) was developed to facilitate with the assignment process. Furthermore, information on PCB inner layers was stored and used for analytical purposes and resulted in a reduction in the amount of uncertainty. Therefore, a more timely and informative decision can be achieved when a new product is introduced. This research would help in increasing the accuracy of product clustering by utilizing the most significant attributes to determine the IE group. Information distance is calculated from deviations in the RPT. In addition, maintenance requirements and variability (or RPT) within each IE group would be minimized as a result.