Tales from the Quality World — Equipment validation
Nearly twenty-five years ago I was working at an automotive supplier as a process engineer. That was the term that company used for our roles which were a blend of quality engineering, manufacturing engineering, supplier management and launch engineering. At the time, one of the product lines for which I was responsible was the GM trunk latch mechanism. This was a commmon part across numerous GM cars including the Chevy Impala. As you can imagine, the volume increased quite a bit.
Given the increased volume, the sole rivet machine we had could not keep up with demand. A second rivet machine was warranted for which I put in a capital request which was approved per the quote from National Rivet. [1] They had built our current machine years ago.
Once they notified me that the machine had been completed, I gathered 300 sets of parts which go into making the latches and took an overnight drive from west Michigan to National Rivet’s location in Wisconsin. I also brought our most experienced operator.
The next day we were in their facility and the operator took a seat at the machine and began assembling latches from the parts we brought to validate the operation of the equipment. Everything worked as required except a timing issue in the controls. There was a delay in the controls such that the operator could have pressed both palm buttons to activate the machine and still been able to reach in and get pinched by the rivet application- safety issue. National Rivet’s controls engineer immediately corrected it in his ladder logic program and we signed off on the machine and had it shipped back to our plant.
Just a few years ago as the manufacturing engineer for the company I was with then commissioned rivet installation equipment for a crash can which was shipped to the Corvette plant.
The nature of the assembly was that the can was placed into the fixture and then a plate with rivnuts was inserted into the can and then the rivets were fired into the prescribed holes to hold the plate into the can. GM used the rivnuts in their assembly process. Our rivet machine was supposed to have sensors to ensure all four rivnuts from the supplier’s part were present.
I took a drive to the supplier to see their press and operation and immediately found that they did not have sensors to verify all four rivnuts were present in their stamping. I explained to them that we were supposed to pokeyoke that issue on our equipment, but that if for some reason it escaped us that their plate was missing nuts, it could cost them the price of a Corvette frame. They immediately began working on an off-the-press detection method.
One problem I discovered as I reviewed was on their PFMEA from the PPAP package which had been approved by my company’s corporate supplier quality department. Their detection levels on the PFMEA were all at “2.” Visual detection only should have been at least a “7” or “8.”
Unfortunately, a couple weeks later, sure enough, I get a call from my contact in Bowling Green. A crash can had reached them and been welded onto the frame of the vehicle. The can was missing one of the four nuts. I immediately called the supplier and had them certify all stock we had in inventory as well as product at their location. Knowing they didn’t have pokeyokes in place yet, their detection was manual.
I then went back to my facility to determine how the parts escaped our process when pokeyokes were supposed to be in place. I discovered that the pokeyoke input-outputs were not enabled for that aspect of the assembly. The machine should never have been brought into the plant or allowed to run production until its operations were fully validated!
Validate before production!
More tales in the future.
