As suggested by several past studies (Barkan and Hinckley, 1993; Eking, 1988; Gebala, 1992; Beiter et al., 2000; Shibata et al., 2001), complexity in the assembly process has a strong correlation with the occurrence of defects. The authors propose a new method that uses a product’s complexity to predict defect rate (Shibata et al., 2001 and 2003). This method provides metrics for assembly complexity using two engineering measures: 1) assembly time estimates and 2) ease-of-assembly ratings. Extensive field data for consumer audio equipment assembled at various manufacturing sites around the world provide the means to validate the proposed metrics. A new process-based complexity factor uses a “time standard” defined for a set of assembly tasks. Predicted defect rates, based on this process complexity, exhibit a significant correlation with actual defect data. Another factor, the design-based complexity factor, uses the “Design for Assembly” method for evaluating an assembly and allows the user to predict defect sources not captured by the process-based complexity factor. Combining these complexity factors not only improves prediction accuracy but also provides guidelines for improving the original design concept as well as each process step. The authors conclude with an example of implementing the Assembly Quality Methodology using the new complexity factors for globally distributed manufacturing.

Mixed-model production is the manufacture of similar products on a single assembly line (see Fig. 1). This assembly technique is gaining popularity in a multitude of production environments. Benefits of this include: reduced investment costs, reduced fluctuation in production due to customer demand, and smaller production facilities. For some world-class manufacturers, mixed-model production also causes increased commonality amongst products on each assembly line thus leading to reduced inventory levels and number of stock-keeping units (SKU). However, a systematic approach is lacking in most companies leading to an increase in human assembly errors due to the increase in process complexity. In response, many companies focus on automation, lean-manufacturing, and JIT parts delivery along with other forms of technology and error-proofing devices. Unfortunately, there are two problems with this approach. First, how does a company with low production volume justify the investment in automation, technology, and error-proofing devices to alleviate these types of errors? Second, employing these fixes after the problems exist leads to sub-optimal designs. With product life-cycles shrinking (Shibata, 2001) and development times shortening even more severely (see Fig. 2), companies cannot afford to fix problems after they find them in production. Companies need to address the problem in the design stage. The authors propose a multi-phase approach to managing mixed-model assembly errors in low to mid-volume assembly environments. First, companies should employ mixed-model FMEA during preliminary design to drive commonality and eliminate errors at the systems level. Second, a geometry-based comparison function based off CAD system data that can identify similar parts during detailed design is needed. An automated comparison technique alerts the designer to potential assembly problems of individual parts. Third, when parts must be physically similar, this information must be transferred to manufacturing so appropriate technologies are employed to error-proof the assembly operation.

This paper describes the development of the quantitative predictor of defects that works not only on the whole product but also on its modular units. The authors divide the assembly process of a product into small modules, and apply the Assembly Quality Methodology (AQM) to determine the assembly complexity of the modules. The authors also verify the correlation between complexity factor of the modules and their defect rates. The newly proposed floating threshold assembly time improves the correlation between complexity factor and defects per unit (DPU). The authors discuss the comparison of case studies and the interpretation of their results for the future improvement of the method.

Manual assembly errors are a significant source of manufacturing defects. Therefore, an efficient method is needed to identify and alleviate potential assembly defects. Process Failure Modes and Effects Analysis (Process FMEA) is one technique used to anticipate, evaluate, and resolve potential manufacturing and assembly issues. However, performing FMEA is widely considered to be tedious and time-consuming, and not always worth the effort. In response, many researchers have attempted to automate FMEA using Artificial Intelligence (AI) to make it less arduous. Unfortunately, automated techniques are limited to systems with predictable behaviors (e.g., electronic circuits) and are rarely used on unpredictable processes such as manual assembly. “Assembly FMEA” is a novel technique developed specifically to identify manual assembly errors. Assembly defect levels are related to assembly complexity, which can be estimated using “Design for Assembly” (DFA) time penalties. Hence, Assembly FMEA uses a series of DFA-related questions to elicit potential assembly defects. The questions help to focus, standardize, and expedite the FMEA process. Assembly FMEA quickly identifies a large number of assembly errors with significantly less effort than conventional FMEA. This paper describes the Assembly FMEA procedure and illustrates its use on a conceptual design and on an existing product.

The complexities in assembly processes have a strong correlation with the occurrences of defects. This paper develops a design-based complexity factor derived from the “Design for Assembly” method for evaluating assembly to augment factors not captured by the process-based complexity factor proposed by the authors. The authors collected extensive field data including consumer audio equipment assembled at various manufacturing sites around the world, and used defect data for validating the metrics. The quantitative correlation between the design-based complexity factor and defect rates will provide mechanical designers with guidelines for improving the original design.