Highlights From a Keynote - Strategies for Successful DMAIC - Part VII
Good morning, and thanks for tuning in!  This is the last in a series of posts regarding strategies for successful DMAIC.  I hope you've enjoyed the first six discussions, and I think you'll find this one of interest, as well--don't become the babysitter.

Often times in our projects, we get so excited when we've identified root cause and have implemented the improvement action, that we forget all about the control plan.  Or, even if we have thought about the control plan, we implement it haphazardly or inefficiently.  Don't think this only happens to inexperienced Belts--even as a practitioner for nearly 20 years, I have still fallen into this trap now and again.  We all know what happens when there is no control plan or when the control plan is not effective--either the failure mode comes back or we are forced to monitor the process continuously.  I've worked on hundreds of projects during my career as a Black Belt and Master Black Belt.  Can you imagine how much time would be wasted if I continually had to monitor and revisit each one of these projects?

The most recent time I was unintentionally trapped into being the babysitter was on a project I had recently completed on one of our tail lamp assemblies.  The particular process I was helping to improve was the vibration welding process, which welds the lamp housing and lens together to create both a stronger structure and a good seal.  After an extensive DOE, we found the manufacturing parameters that controlled the weld quality and weld consistency--the most important of which was temperature of one of the platens.  Even better, this setting was controlled by a control panel with digital entry.

To measure the strength and consistency of a lamp weld, we use a burst pressure test.  It's exactly as it sounds--we fill the lamp with air and build up the pressure until the lamp bursts.  It's actually a pretty exciting test to watch!  Higher burst pressures indicate a stronger weld, and if the weld is consistent, the burst origin location should be at a random point around the perimeter of the lamp.  The control plan we implemented was focused around the KPOV (the key process output variable--the burst pressure) because it was our response in the DOE and, after all, the KPIV (key process input variable--platen temperature) was a digital setting on a computer screen.  For a few weeks after the process settings were locked in, burst pressures were great.  But then they started to drop, and supplier called me in desperation because none of the process settings had changed.  I drove the two hours out to the supplier to check things out, only to find out that while the temperature setting was indeed not changed, the actual temperature achieved in the machine was no longer reflected by the digital reading (we had measured the actual temperature with a thermocouple during the DOE).  A preventative maintenance issue was causing the readings to differ.  Our problem was that we were focusing on the outputs of the process instead of the inputs of the process, because had we noticed that temperature dropping, we would have caught it so much sooner and more efficiently.  We installed a permanent thermocouple in the machine and the issue never returned.  No more babysitting for me!

So remember, don't forget about the control plan, and certainly, don't neglect it by making it non-robust!  Focus on the KPIVs, not the KPOVs.  If you choose error proofing, choose actions that are dedicated to prevention, not detection.

I really hope you've enjoyed this series of posts.  With them, I wish you successful DMAIC problem solving.