Multivariate Analysis and Evaluation of Adaptive Sheet Metal Assembly Systems

The flexibility of sheet metal assembly processes is one of the most critical issues in the design stage of assembly systems. Currently in the automotive industry, flexibility and adaptability of assembly systems are mainly considered as the capability to assemble a family of products. The notion of assembly system flexibility, as understood from the point of view of error compensation by upstream processes, has not been widely researched. This paper presents a systematic method of flexibleladaptive assembly system evaluation, based on its ability to compensate fo r part dimensional variability caused by upstream processes. This allows for the expansion of the present design for manufacturability approaches by applying multivariate analysis and fixture diagnostic techniques. The proposed method was applied to evaluate a flexible assembly system at a US manufacturing plant. K e v w o r d s :Flexible assembly, error compensation, dimensional error 1. In t roduc t ion : Ongoing developments in the field of flexibleladaptive assembly technology have resulted in the automation of increasingly complex assembly processes such as sheet metal assembly. Moreover, traditional application of flexible assembly systems for middle range production volume and product complexity (Makino and Arai, 1994) has expanded to application in high volume production, thus increasing the demands on assembly systems, including tighter assembly tolerances and growing number of component variants. Currently, many of the problems being encountered in automotive sheet metal assembly stem from the fact that the products were designed with a shallow understanding of assembly processes, and that the assembly tooling was designed with a lack of knowledge about product variability. In order to achieve a reliable assembly process and a high degree of assembly quality, it is necessary to examine the automation-related design of both product and process in the earliest phases of product and tooling design (Kroll et al., 1988). Until now significant results were achieved in the area of integrating product and processes during design phase through DFAIDFM (Boothroyd; Miyakawa et al., 1990), design for producibility (Suh, 1988), assembly-oriented design (Warnecke and Bassler, 1988; Milberg and Diess, 1988), feature-based design, and others. Currently in the automotive industry, the areas of flexibility and adaptability of assembly systems are mainly addressed as the capability to assemble a family of products by using statically and dynamically reconfigured systems (Makino and Arai, 1994). The notion of assembly system flexibility, understood as er ror compensation caused by upstream processes, has not been widely researched. The current DFAIDFM approaches do not emphasize the impact of incoming part variability on the design of flexible assembly systems (Warnecke and Bassler, 1988). However, the importance of the product error compensation in flexible assembly systems has been observed by a few researchers (Arai and Takeuchi, 1992; Reinhart et al., 1996). Reinhart et al. (1996) emphasized the importance of minimizing scatter in the performance and fault characteristics of an assembly system. Arai and Takeuchi (1992) suggested to use "adjustable assembly" or "selective assembly" in order to assemble a product with high accuracy using parts manufactured with limited accuracy. This paper presents a systematic method of flexible assembly system evaluation during the design phase, based on its ability to compensate for part dimensional variability caused by upstream manufacturing processes. This allows for the expansion of the present design for manufacturability approaches by applying multivariate analysis to model dimensional faults Annals of the ClRP Vol. 47/7/1998 17 (variation patterns) of product based on the CAD design information. The proposed methodology is based on three steps presented in three consecutive sections of the paper: (1) modeling of variation patterns of pre-assembled parts and subassemblies (components); (2) modeling of assembly system capability to compensate fo r components error; and (3) evaluation of assembly system capability to compensate for p reassembled component variability (Fig 1 ) The methodology is applied to sheet metal assembly of door and automotive body using Net Form and Pierce (NF&P) system (Fig. 2; layer 3). The NF&P system has the ability to compensate fo r position and orientation errors of the door and door openings during assembly process to minimize door-to-body gap variation. Information about Information about Flexible Assembly Station Assembly Process/Product Section 4 Section 3 Model of all Section 5\ / Determine assembly system evaluation index p Figure 1. Outline of the methodology 2. Automotive bodv assemblv s v s t e m with and without error compensat ion An automotive body assembly system is a mult ifixture hierarchical system with over 200 stations/fixtures. Sheet metal components are joined together at each station to form higher layer subassemblies, which then become input components for the next layer of the assembly (Fig. 2b). Study on the automotive assembly process indicate that as many as 72% of all root causes of faults are attributable to failures of assembly fixtures (Ceglarek and Shi, 1995). The largest identifiable portion of all assembly f ixture failures is attributable to process variabi l i ty, which takes the form of dimensional variation of the product. Thus, one of the principal quality/productivity parameters for sheet metal assembly process is dimensional variation of the final product (body with installed doors). Traditional automotive assembly systems are designed without error compensation capabilities. Each station fixture uses a fixed set of locators to position each part independently from the other using, for example, a 3-2-1 fixture layout principle. This means, that errors from each station can accumulate f o r over long and complex assembly lines. Since not all dimensional variations are of equal importance for overall performance of the vehicle, the compensation of errors in cr i t ical stations can significantly improve the final quality of the product. Layer 8 Layer 5 Layer 4 Layer3 Layer2 Layer1 (a) Assembly process layout Body Complete Layer I