Originally Posted by CharlieVT
If they conclude, as I have, that the problem isn't with the design but rather a QA manufacturing/assembly issue then they don't really need to redesign it. They just need to build 'em better. And witness: the Evo drive was a new design and those had a failure rate too. It isn't the design. Don't call me a BMW apologist, but I think the BMW engineers knew what they were doing. The machinists and assembly guys were making mistakes for whatever reason, maybe they were hungover, were commiting sabotoge, or, my favorite theory: former Soviet block East German workers with bad work ethic (that's a good theory if you go back and read some management articles about problems with the work ethic in the labor force following reunification in Germany).
Well, having been an engineer for the past 27 years, I have a passing acquaintance with design concepts and principles. Design is a many faceted exercise and designing a component or system that works well if perfectly assembled is relatively easy. The difference between a bad design, a good design and an excellent design is as follows:
bad design - doesn't meet the requirements no matter how well it is made.
good design - meets the requirements if properly assembled.
excellent design - meets the requirements and virtually can't be assembled incorrectly.
The BMW deisgn may be a good design, but it certainly isn't an excellent design. The issues may well be assembly related and the root cause of the assembly errors may well be as you hypothesize. However, if BMW had used appropriate DFA/DFM, then the variation in assembly outcomes would not be nearly so great.
This reminds me somewhat of the Hyatt Regency walkway collapse in Kansas City a number of years ago as chronicled in the book "To Engineer is Human." As an aside, this is a great book and you don't have to be an engineer to read and appreciate it. The walkway as originally designed wasn't compliant with the prevailing structural code, but very likely would have never failed due to the margin of safety in such codes. However, the problem was that the original design was essentially impossible to build. This resulted in a field change to the design which inadvertantly changed the load path such that one part of the structure now carried twice the load it would have carried in the original design. This was the primary cause of the structural failure, with the inadequate capacity of the original design being a secondary factor. The engineer who made the field change screwed up by not properly analyzing the new load path, but in my opinion the original engineer is the one to blame. Making a design that can't reasonably be built is simply not acceptable design practice.
It may well be that the design is adequate if perfectly assembled, but the assembly process is sufficiently complex and with enough "degrees of freedom" in assembly that it simply can't be reliably assembled correctly in a typical production environment with typical factory labor. And the tolerance for improper assembly (preload, etc.) isn't very good.
To me, this is as much of a design error as is a design that is inadequate even if perfectly assembled. They are design errors of a different type, but design errors nontheless.