The Relationship Between Component and Product Quality in Manufacturing, with Emphasis on Competitiveness

Yue Wang, Purdue University

Abstract

The capability of continuously producing good quality products with high productivity and low cost is critical for manufacturers. Generally, products are made up of components, which enable the product to perform its purpose. A complex product may be assembled from many components through multiple assembly stages. Any quality defects in a component may build up in the product. A good understanding of how the quality of components impacts the quality of products in a complex manufacturing system is essential for keeping the competitiveness of a manufacturer. In this research, a series of quality management models are proposed based on studying the relationship between component quality and product quality. Optimal quality control leads to increased competitiveness of a manufacturer, since it helps reduce cost, increase production, and limit environmental impact. The research starts from studying the tolerance allocation problem, which is fundamental of managing the tradeoff between quality, productivity, cost, and waste. First, a tolerance allocation method that minimizes cost is proposed. This model jointly considers process variation and tolerance specifications. The relation between manufacturer, user, design, and processing are embedded in the cost model. To solve the tolerance allocation problem from the root cause, i.e., the variations in production processes, a second tolerance allocation model is then provided. This model considers both product design (tolerance selection) and operation planning (or production rate selection). Relations among production rate, production cost, processing precision, and waste are considered. Furthermore, a new process control model that extends traditional statistical process control techniques is proposed. Data acquired from a manufacturing system are usually in the forms of time series, and anomalies in the time series are generally related to quality defects. A new method that can detect anomalies in time series data with long length and high dimensionality is developed. This model is based on recurrent neural networks, and the parameters of the neural networks can be trained using data acquired during routine operation of a manufacturing system. This is very beneficial because often, there are few data labeled as anomalies, since anomalies are hopefully rare events in a well-managed system. Last, quality control of remanufacturing is studied. A component-oriented reassembly model is proposed to manage the varied quality of returned component and varied needs of customers. In this model, returned components are inspected and assigned scores according to their quality/function, and categorized in a reassembly inventory. Based on the reassembly inventory, components are paired under the control of a reassembly strategy. A reassembly-score iteration algorithm is developed to identify the optimal reassembly strategy. The proposed model can reassemble products to meet a larger variety of customer needs, while simultaneously producing better remanufactured products. In summary, this dissertation presents a series of novel quality management models to keep manufacturers’ competitiveness. These models are based on studying factors that impact component and product quality at multiple stages of a product life cycle. It was found that analyzing the relationship between component and product quality is a very effective way of improving product quality, saving cost, and reducing environmental impact of manufacturing.

Degree

Ph.D.

Advisors

Sutherland, Purdue University.

Subject Area

Artificial intelligence|Design|Electrical engineering|Environmental management|Industrial engineering

COinS