Related: Automotive Production
Jay S. Baron, manager of Manufacturing Systems,Office for the Study of Automotive Transportation(OSAT), The University of Michigan Transportation Research Institute (UMTRI; Ann Arbor), makes a rather controversial point: "You don’t always need to get parts made perfectly." He’s talking about stamped parts for automotive production.
More than controversial, this may seem somewhat heretical. Haven’t people been working long and hard in order to make sure that things are absolutely, positively, dead-on? Yes. But that may not be the right thing to do, Baron suggests. Supported by the American Iron and Steel Institute(AISI) and several auto companies for over a decade, UMTRI has compiled a database supporting its position.
What’s wrong with this picture? Isn’t it flying in the face of common sense? Yes. But, in fact, what seems like common sense sometimes isn’t. Slowly but surely, an increasing number of auto manufacturers are reaching the same conclusion, as counterintuitive as it might seem.
Researching Worldwide Practices.
It goes back to research that Donald N. Smith, UMTRI’s associate director, Manufacturing Systems, did for the AISI’s Automotive Applications Committee starting in 1986, trying to determine how world-class auto companies were able to produce stamping dies more efficiently in terms of both cost (in 1987 just 35% the cost of the highest cost in North America) and speed than U.S. companies. There were a number of things that world-class companies were doing (and continue to do) that were determined to be advantageous, such as double attaching and integrating multiple parts on one die, and trying out the stamping die on the home line. They’d do experimentation before the start of production: setup the die, try it, take it off. Set it up again. Try it. Change material coil. They’d make measurements of the results. They’d determine what was actually being produced. Then they’d work to control the inputs. They would achieve a stable process, even before measuring how close the panel was to spec.
On the other hand, the common practice in the U.S. is to try to achieve in steel precisely what is shown on the part print. That is, the goal is typically to produce a part so that for every dimension measured on the part, the mean value is at the nominal specification. The focus is on part specification even before getting a stable process. But according to Patrick Hammett, associate research scientist at OSAT, none of the manufacturers that they studied were able to achieve a Cpk > 1.33 on 100% of the points on body side panels. It just can’t be done. Add decreased product development cycle demands to the picture, and you’re talking big Maalox sales.
The thing is, U.S. car makers have tended to chase the 100%, to work to get everything right. (As in the requirements for PPAP—Production Part Approval Process.) But several automakers—most notably those based in Japan—didn’t follow the same procedure. Their approach is that they’d make sure that they understand what they’re making—what, say, the body side die is producing—and then work to (1) maintain consistency and (2) be certain that the rest of the system (i.e., other portions of the vehicle build process, such as welding tools) could accommodate the part size to assure that the vehicle as built met the functional specification.
And guess which automakers were lauded for their superior fit and finish?
The final customer is not really too keen on knowing whether the quarter panel meets the part print. (And even if the mean value is at nominal for all measured dimensions on the quarter panel, it may not make any difference to the final assembly because by the time it has gone through a multitude of clamps and fixtures and undergone welding, its profile undoubtedly isn’t what it was when it was in a checking fixture.) The final customer is certainly going to be looking at the gaps around the closure panels. How that dimensional consistency is achieved is what makes the difference.
The approach that Baron and his colleagues are promulgating is what is known as "functional build." Perhaps one way of thinking about this is to contrast it with the common practice, which might be considered "theoretical build," with the theoretical part being the chase after the specs, and which tends to lead to late parts and rushed work.
What they have determined is that not all parts affect the overall assembly quality. In other words, not all parts need to be made with the mean at nominal. (We’ll pause here for some quality control-minded folks to regain their breath.) There are parts that can be made under the functional build regimen and parts that can’t be. The latter include simple parts that tend to be rigid and greater than 1.5 mm thick. According to Baron, these parts (e.g., strut housings) are typically small and simple, so the traditional approach is still the right way.
Functional build can be applied to two types of parts. The first category includes major outer panels, such as the fenders, outer quarter panels, and body sides. The second includes non-rigid (i.e., less than 1.5 mm thick) inner panels, such as the floor pan, and rigid (> 1.5 mm) complex panels, such as the windshield frame reinforcement. Judgments must be made (but can be guided, based on past experience). This is why Baron calls this approach "criteria-based" functional build.
There are two aspects to keep in mind. One is the idea that what is being sought is an assembly of parts that meets requirements, not a series of discrete parts that individually measure up. Second is that both other parts and the assembly process have an effect on a given part.
Baron provides an example. There are two parts: the center pillar reinforcement and the body side panel. The center pillar is a structural part. It’s a non-functional build part. It should meet nominal. Assume that the body side is off centerline, outboard, by 1 mm. What do you do? The answer for many people is to rework the body side die so that it is on centerline. Which takes a lot of time and money. But if you are practicing functional build, you’ll realize that because the body side is a non-rigid part, the mere act of attaching it to the rigid center pillar is going to have an effect—in this case, to shift the assembly toward nominal, without reworking anything. No time delays. No die rework money.
Building the Screw Body.
A key operation within functional build is the "screw body" evaluation. What occurs here is that prototypes are built from parts stamped with production dies. The assembly is performed using screws or rivets in place of welds. This allows the engineers to determine the effects of the assembly operation on the mating of the parts. As subassemblies are built up to the final assembly, there is, of course, less acceptable variance from design specification, and once the first level subassembly is passed, the print spec comes into play. But early in the build, it is often the case that there is no need to rework dies to bring parts into tolerance. The total assembly is what is important.
One benefit of the screw body process is that it provides the ability for manufacturing engineers to tune-in the placement of spot welds for actual vehicle production.
Baron stresses that functional build is not about passing the responsibility of build quality away from the stamping operation and to the assembly department, nor is it about compromising design intent. But what it is about is not obsessing about things that can’t be obtained, like perfect body panels. (He admits that perfect parts would make making things easier, of course.)
The bottom line here is big savings in terms of time and money. According to UMTRI data obtained from auto makers that are employing functional build, there are up to:
• 90% time savings on body die tryouts
• 48% cost savings on the dies for the total body
• 50% time and cost savings on tryout, automating, and process validation of body dies.
And it is important to stress that this is being realized while providing the customer with high-quality products, usually higher final quality than could have been obtained with theoretical build.