Tag Archives: engineering

How do you hypnotize an engineer?

Please note, I am on a job search for a full-time, direct-hire position.  Ideally between Burlington, MA and Concord, NH, and my focus is in either the medical device or defense industries (doesn’t mean I’m disinterested in other possibilities!).  Positions in which I’ve kicked you-know-what in the past:

  • Plastics Design Engineer
  • New Product Introduction (design – manufacturing handoff)
  • Cost Reduction


Fix the Problem XIII: What’s Different? What changed?

This is the latest of a series of case studies and examples from my career, where I attempt to summarize a problem solved, how I did it, all with an eye of passing along useful information… while still, of course, making a good faith effort to protect any confidential information. I hope this, and the others in the series – found HERE – will prove useful and education, illustrative of my abilities, and inspirational as to how I might fit into a new employer.  I am, after all, on a job search!

And a brief disclaimer: These cases are “a while ago” so it’s possible I am slightly off on some of the exact details – but the broad sweep of each case is correct.


As a part of teaching at U Mass / Lowell’s Plastic Engineering Department, one of the tacks I’ve tried to stress is that the students’ program and thoughts need to be aimed – ultimately – at solving real-world problems.  I’ve told my students that when solving a problem in production, or returns from the field, two critical questions often assist in delving quickly to the root cause(s) of the issue.  Specifically, What’s different? and What changed?.

To that end, I’d like to put forth three examples where these two questions were instrumental in helping to find the issues involved.

Retaining Rings Wouldn’t Retain

In one position in my career I worked as a floor-level Manufacturing Process Engineer. I was constantly challenged (read: hammered) with issues ranging from fixtures to field returns to, well, you get the picture.  Lots of “opportunities”, as my then-boss used to say.

One of these was a long-standing issue predating my arrival in the department. On the smaller, hand-held units there were several pieces in a family of similar products.  Some would be sent out of our department and rarely come back from retaining ring issues (essentially, a female piece installed over a mating male piece and held on by that retaining ring).  Others would come back having “spontaneously disassembled”, sometimes even before having left our factory.  This was a constant sore spot in my area’s weekly quality issues, not to mention warrantee report.  Thus it became a priority to solve and get one high-profile issue off the list.

I sat down with a sample of each product: the male, the female, and the retaining ring, laid out to compare and contrast. Visually, on a first-pass look, there seemed to be no differences aside from subtle variations in overall lengths.  More to the point, arranging the pieces from small to large, and putting either an OK or a NOT OK post-it below the pieces, based on their problem frequency, there wasn’t an obvious pattern (e.g., if it had been the two largest ones, or the two smallest ones, that could have been a clue – but there wasn’t).  Since some worked, and some didn’t… “What’s different?”

Being a fan of the Value Engineering discipline, which drives me to think functionally, I asked “What holds these retaining rings on?”  Answer: The groove geometry.  The groove depth, the central diameter, the width… in point of fact, each assembly used the same retaining ring.  Clearly, the ring was not the issue.

So I pulled prints for everything. One possibility was that there was not enough clearance, or the tolerances were wrong, and somehow the retaining rings were being pushed off.  But looking at the nominal and “worst case” dimensions showed that was not the case.  Also note that the company had good machinists, and a strong SPC program; things were in control on that score… and each groove had the exact same dimensions and tolerances.

But doing the “sit and stare” at the drawings laid out side by side I realized that some of the drawings had a callout for the external corners of the groove: SHARP. Some didn’t.  And the ones that didn’t have that callout were the ones with the issue.  Aha, a clue.

The company had a default callout requiring that all corners be broken by – going from memory – a chamfer of .010-.015 inch unless otherwise specified. In looking at the retaining ring groove design recommendations from the supplier, they stated that the edge at the outer diameter of the groove needed to be called out SHARP and could not have a break, whether radius or chamfer.  And the parts showed it; parts without that callout did, indeed, have that edge broken.

I showed this to the Design Engineer who acknowledged the issue, concurred with doing an ECO, and I wrote it up to put through.

Result: The problem went away… because I laid the good and bad parts out, l considered multiple potential factors, but the thing that was different was a design detail missed on the problem assemblies.

It Was Good, Now It’s Not

At that same company, in the same position, another product had a significant percentage of leak test failures. A far more complex assembly than the one above, it had multiple potential leak path failure locations.  Again, an inherited sore spot.

My path was to systematically take failed torches and block off one possible leak path after another, attempting to isolate which of the multiple possible potential leak locations it could have been was the culprit. My goal was to systematically examine the leak location(s) once I’d identified them.

But in one meeting in discussing the larger area, of which mine was a part, one of the managers said “We used to have no leak failures. Let’s find out what changed and change it back.”  A detail I had not known at the time.

It turns out that a design change had been made, with the best of intentions, that resulted in moving several o-rings axially by – IIRC – about 20 thousandths of an inch, which created the potential for them to move under pressure and thus lose their sealing ability.

However, ECOs are not done for no reason.  As I recall, a more careful re-examination of the initial ECO and its reason for being done found that the change could be made with a reduction  in the positioning and seating of the seals – leaks not being considered the first time – and thus maintain the ECO but eliminate the ripple effect that created the leak test failure issue.

Result: The problem went away. And the key lesson learned is to identify the time frame when things change for the worse, and ask “What changed?”  Not just materials, personnel, processes, etc., but consider Design changes too.

Cracked Handles… Sometimes

In one company, the plastic we sold was injection molded into large trash cans – the kind that are used for homes and often picked up by an arm to be dumped into a truck. The company that made them was receiving complaints from customers that the handle used by the truck lifter was cracking.  This was creating quite a problem for them and we were asked to investigate.

My initial role was to do a stress analysis of the handle. I obtained load forces, etc., and built several finite element models of the handle simulating both centered and eccentric loadings.  None showed stresses high enough to create cracks although the stress hot spots were where the cracks were forming.  I also considered impacts, not just static loads – again, while the stresses were in the right places for the cracks they weren’t high enough to exceed the yield strength.  I increased the mesh density – refining the model as sometimes that can affect the stress levels, but the numbers held.  Based on the force and impact load cases I was given, there was no reason for the handles to be cracking.

Our initial thought was that some kind of chemical might be attacking the material, and we started to inquire about possible chemical contacts, but then a clue arrived through our sales group. Only one color of the several the company offered to the end users was having the issue.

Aha! “What’s different?”  We supplied the base resin; at the molding location the customer would blend in colorant masterbatches to create the color variations.  Pursuing this further, we learned that the base resin used by the masterback colorant provider was, for this one color, significantly lower in average molecular weight.  I.e., by blending in this color masterbatch the molder was introducing a weaker and less impact-resistant material to be blended into the base resin.  (Material note: plastic strength is directly related, as a general correlation, to molecular weight of the polymer chains.)

Looking back with more experience under my belt, I’d ask two questions: 1: “Can the failure be reproduced?” And 2: If the answer to the first question is YES, “What happens to products molded from uncolored resin?”  Assuming YES was the answer to the first question, and the failure didn’t happen with uncolored resin, that would at least have eliminated the base plastic as the sole cause of the problem.

When a different masterbatch blend with a material molecular weight like the other masterbatch colors was tried, the problem went away.

Again, the key clue was learning that one color didn’t work while the others did… “What’s different?”  (And in the back of your mind you can add the potential for different colors to be questioned.)

Adding to the Toolbox

Multiple tools exist to aid in systematically looking for the root causes of a problem, e.g., Ishikawa/fishbone diagrams, the Five Whys (and Two Hows) questioning, etc.  Adding “What’s different?” and “What changed?” as appropriate can add an important new tool to your problem-solving toolbox.


© 2016, David Hunt, PE