Fix the Problem II

In Part I on this topic I looked at identifying the source of variation in an assembly cell making automotive headlights, and then outlined the proposal I made to neutralize the effects of the variation on the quality of the assembly since the variation could not be eliminated.

In this installment (of what will probably several case studies of both production issues as well as product failure troubleshooting) I will look at my troubleshooting work on a sterile blister pack consisting of a syringe, alcohol and iodine wipes, and some ancillary items.  The problem was critical: these blister packs are meant to be used in a hospital, and the package contents are supposed to be sterile for immediate use – often in an emergency room situation where there is no time to worry about checking for sterility.  The reality of sterility needed to match the assumption.

These blister packs consisted of a thermoformed polyethylene film pocket, the contents, and a Tyvek film covering which would be peeled back and the contents removed and used (Tyvek was used as it is permeable to the sterilization gas that is used).  The company was receiving field returns where the pockets had punctures in the pocket corners, allowing outside – and obviously non-sterile – air to penetrate into the pocket thus ruining the sterilization and the product’s usability.  (Side note: I use thermoforming as a generic term; thermoforming and vacuum forming are, functionally, the same process IMHO: a film that, through a pressure differential side-to-side, is formed to a shape – whether over a male protrusion or into a female cavity.)

In order to identify the root cause of the problem it was first necessary to figure out where the punctures were being introduced.  Begin at the beginning was my approach.  I obtained, straight off the production line, a sample of blister packs of, probably, a thousand formed pockets.  An exhaustive examination found no perforations.  Since we knew from returns that this was a significant problem, comprising several percent of production, a sample size of 1000-odd pockets should have turned up punctures if this was the source.

The next step was the assembly process – the various pieces were put in manually by a team of operators, each responsible for a couple of items.  Again, a sizable sample (several hundred products) was exhaustively checked to search for punctures.  Note that I did not forewarn people I was grabbing samples off the end of the line.  I merely showed up (with my boss’ written approval of course) and took them, which prevented any special care people might have taken had they known that a certain part of their production would be taken for testing.

This right-off-the-line sample had no punctures.  While not as large a sample size as the unfilled pockets, we should have seen something had the root cause been in the assembly stage.  Clearly the punctures were happening post-assembly.

One theory that had been proposed just before I started there was that the alcohol and iodine swabs, which were in flat foil packets, were creating the punctures through vibration of shipping causing them to shift and slide, and the sharp corners of the packets were – as the theory proposed – cutting the corners.  This was plausible as the two swab packets were placed in the bottom of the thermoformed pocket… although, in looking at the field returns we received it didn’t look like the punctures were clean cuts, so I was personally skeptical about this proposed failure mode.  It still needed to be tested.  So we did a test where the order of placement of pieces was changed to put the two wipe packets on the top of the cavity, thus keeping any foil corners away from the pocket bottom.  These samples were then packed as normal and shipped off to a vibration test simulating shipping and handling; IIRC we had five boxes of these parts with several hundred individual product packets.  To be complete, though, we also put five boxes of “normal” production pieces through the same testing process.  The idea was that if these foil packets were the root cause of the punctures, we should see (theoretically) no punctures in these, and the punctures in the standard product.

The results were not as expected, though.  Both the test sequence and regular production product had, statistically, the same number of punctures.  It wasn’t the product placement – it was the rough-and-tumble vibration of pieces moving in the box.  Aha!  At least we knew where in the product’s life cycle from start to end-user the problem was happening.  But we still didn’t know why.  We only saw punctures in the corners, which begged the question “Why only in the corners?”

Anyone who knows anything about thermoforming (or vacuum forming) knows that the thickness of the part is not uniform, but the wall thickness gets thinner the farther the film has to stretch, and in particular it gets thin in corners.  Measurements showed the thickness in the corners, which had far tighter radii than I would have designed knowing this process limitation, was 2-3 thousandths of an inch (mils).  This is not a robust thickness!

The next step was to outline recommendations for fixing the problem; these could have been used in combination but I’ll address each separately.  Note that all three options would require revalidation of the sealing process.

Option I: New tooling – with, IIRC, triple the radius size at the bottom of the tool cavities that made the pockets.

Advantage: The film itself would not change, so there would be no variable cost-of-material change.

Disadvantage: New capital cost for tooling; only a modest increase in film thickness.

Comment: This did, however, create information useful for the next generation of pockets that needed new tooling; specifically, use generous radii at the bottoms of the tools.

Option II: Thicker polyethylene film.

Advantage: The adhesion between the Tyvek and polyethylene material was known and
quantified.  No new capital cost required for tooling.

Disadvantage: The thicker film was more expensive, and the thicker film created the possibility of requiring a marginal increase in sealing time to get the interface up to the needed temperature.

Option III: New material for the film.

Advantage: There were any number of materials that were more puncture resistant for a given thickness.

Disadvantage: New film materials are more expensive than the old one; stronger materials might also affect cycle time.

The Final Action

Given the long lead time for new tooling and the need for a fast solution, as I recall we adopted a double-hit approach.  A thicker film made from nylon rather than polyethylene was chosen as a “swing for the fences” option.  The advantage to this was a very high probability of success in one iteration; the disadvantage was increased material cost (admittedly not large on an absolute scale) plus a potential for longer cycle times for sealing.

Again, going from memory, the heated sealing plates were so powerful in terms of pressure and temperature driving the sealing that this did not affect the sealing cycle time.

Lessons

1. When faced with a problem coming back from the field, working systematically through the process from start to finish can be a simple way to isolate the location where the failure is occurring, thus minimizing the number of possibilities that need to be investigated.
2. When taking samples of manual work, don’t give warning; just take them.  This avoids workers taking extra care because they know it’ll be looked at more closely.
3. When doing experiments to evaluate an improvement idea, be sure to include production pieces as a control.
4. Sometimes a hard-hitting, fast-to-implement solution is called for – even if it’s more expensive than an optimized, minimal-cost one – especially if the problem is jeopardizing shipments, customer satisfaction, and market share, etc.

© 2013, David Hunt, PE