Multi-stations sheet metal assembly modeling and diagnostics

In this paper, a multi-station assembly process is modeled for diagnosing automotive body dimensional faults. The proposed approach enables multi-station assembly process modeling based on design information (CAD) and allows system behavior determination based on the in-line measurements of the final product. The method includes generic modeling of the key assembly process characteristics; tooling types and locations, as well as part-to-part joints. The modeling is based on the pre-determined variation patterns caused by failures of each selected characteristic. The proposed model successfully diagnosed multiple dimensional faults of an automotive body assembly process. Verification of the proposed method is presented through an actual case study. INTRODUCTION Increasing quality and productivity are two major goals in today's automotive industry. Fixture failure and dimensional variation is one of the major factors to decrease productivity (manual adjustments on doors, fenders and hoods) as well as quality (water leaks, closing effort and wind noise). Currently, automotive assembly plants rely on trial and error to locate sources of variation and system faults, which requires years of experience and knowledge of the assembly processlproduct design. Traditionally, SPC (Statistical Process Control) (Faltin & Tucker, 1991) is the standard method for controlling the process and maintaining a high dimensional quality of product. With the recent introduction of OCMM (Optical Coordinate Measurement Machine), and the availability of 100% measurements, advanced quality control techniques are required to take advantage of this information (Hu & Wu, 1990, Roan et al., 1993, Ceglarek et al., 1994). A diagnostic technique based not on heuristic information but on the first principle, system model, is necessary for today's requirement of fast model change over (Davis, 1983, Reiter, 1987). Currently, in the area of sheet metal assembly, a method developed to model fixture failures proves the identical relationship between the Principle Component Analysis (PCA) and the fixture tooling failures (Ceglarek & Shi, 1996). However, this method only focused on the single assembly fixture failure. This paper focuses on the development of a multi-station assembly modeling methodology for diagnostics of the automotive body assembly process. In this paper, a multi-station assembly process is modeled based on critical characteristics of the assembly process, such as the locating mechanism which controls the orientation and location of parts and the part-to-part mating geometry which describes the joining conditions. A diagnostic reasoning strategy is developed by matching model behavior to the real behavior of an automotive body assembly process that obtained through in-line 100% measurements. In this paper, automotive body assembly process will be introduced first. Then, modeling based on the critical feature of the single assembly station is described and followed by the modeling of multistations process. Next, the fault propagation model is introduced. Finally, simulation results of single and multiple faults diagnostics and a case study based on the real industrial data are presented. AUTOMOTIVE SHEET METAL ASSEMBLY This section describes the automotive body structure and assembly process design. Automotive body manufacturing is a complicated sheet metal assembly process because of the complex body structure. A typical automotive body is made of 200-250 sheet metal parts assembled in 60 -100 assembly stations with 1700 to 2100 different types of locators. The Body-In-White (BIW), as shown in Figure 1, can be defined as an automotive body without closure panels such as: doors, hood, fenders, and deck lid, and without powertrain and chassis accessories. A BIW is the basic structure of the whole vehicle. In general, a BIW is made from three major subassemblies such as the underbody, the left and right hand side frames, and the roof. Each major subassembly is made of smaller subassemblies or parts. FIGURE 1. Body-In-White (BIW) The main factors olf the assembly process having impact on the BIW dimensional integrity are: (1) locating mechanisms in geometrical assembly stations, and (2) part-to-part interaction. Locatina mechanism in assemblv Rrocess desian Geometrical assembly stations are designed to assemble two or more parts together. They consist of locating mechanisms, joining equipment, and parts transferring mechanisms. Due to the functionality of a geometrical assembly stations, the most important source of dimensional non-conformance would be contributed by the failure of locators within such assembly stations. Locators and their locating schemes for part positioning consist of one basic building block called "3-2-1" or "17-2-1" schemes designed for rigid parts, and flexible parts respectively. The fixture and locators' function are to position the parts in space correctly. Theoretically, a 3-21 locator scheme as shown in Figure 2 would be sufficient to orient and locate a rigid part in space. FIGURE 2. A typical 3-2-1 scheme for rigid sheet metal The 3-2-1 locating method utilizes fundamental geometry. Three points in space generate a plane, two points in addition to this first plane would generate a second plane perpendicular to it, while one final point generates a third plane perpendicular to the other two. Therefore, an xyz coordinate system would be generated in space to locate and orient a part. This locating method is the backbone of the simulation process where part location is determined by the locator description or design. In addition, the locating scheme depends on the method of joining and their part-to-part interaction, which will be discussed in the next section. On today's manufacturing floor, locators, as shown in Figure 3, are divided into three different categories: (1) 2-way or Cway locating pins; (2) NC locating block with part holding clamp; and (3) the NC locating blocks without clamp. A Gway locating pin gives four direction control. Locating pins are used extensively because the position of locating holes can be stamped very consistently in relation to the part's geometry. An NC block with clamp gives a two direction control while NC block gives one direction control. FIGURE 3. Types of locators Part-toart interaction One of the most critical elements in BIW design is part-to-part interaction. Part-to-part interactions stern from the characteristics of the joining interface. These interactions can be characterized into three main types of joints (Ceglarek & Shi, 1995) in an automotive body as shown in Figure 4: (1) lap-to-lap joints (two flat sheet metal pieces overlap to form a joint), (2) butt-to-butt joints (two flanged parts joined at their flanges), and (3) lap-to-butt joint (one flanged piece and one flat piece joined together at their edges). Each type of joints exhibits a different characteristic and behavior. Lap-to-lap joints tend to allow slip plane movement along the plane of the sheet metal. Butt-to-butt joints tend to constrain the movement in the sheet metal plane. Lap-to-butt exhibits a combination of both characteristics. Behavior resulting from these part-topart interactions will be discussed and modeled in the next section. FIGURE 4. Three main types of joints BuHc-Butt Joint . -------Laptc-Lap Joint . -\