Fix the Problem VII: Product Launch Crisis

At Ford Motor Company in Sandusky, Ohio, the flagship product of the plant was exterior lighting: headlights, taillights, turn indicators, and high-mount stop lights.  During launch there would inevitably be problems – sometimes in molding, or decorating, or assembly.  Things would, of course, get solved, but sometimes things were so bad that it became an all-hands-on-deck affair.

One product launch was going particularly badly.  Despite every shift in every department related to this product – a high-end luxury-car headlight – going 24/7, we literally could not get enough product out the door.  With every department scrambling, with the assembly plant screaming, the group I was in was also asked to contribute.

At the time the methodology called Theory of Constraints, developed by the late Eli Goldratt, was in vogue at the plant (I believed it then, I still believe it).  I believed there had to be one or two critical points in the process of going from resin pellets to finished headlights on pallets that were the prime constraints.  My task was to study the process from a 10,000 foot view and attempt to identify what those critical points were, so as to permit a focused and systematic attack on the problem rather than everyone milling around everywhere.  (Please keep in mind that I’m “making up” most numbers – they’re close, but after a decade and a half since I was in the situation, my memory is rather fuzzy.  But the overall analysis technique will hold.)

Let me be clear in saying my goal was simple: identify (and prove with data) the one or two process steps where the fallout was the most significant, and which was preventing meeting shipment quantities and deliveries.  In other words, identify the “worst of the worst” for more focused attention.

A Group Effort

One other thing: I must give credit to the team.  Although I gathered a lot of information that was useful to focusing attention, and chimed in with many suggestions, many people contributed to attacking the problem.  This is critical – the aggregation of multiple minds, with different perspectives and experiences, was the “magic factor” in making such enormous progress.  Vendor participation and vesting in the success of this effort were also critical, as in many instances they had expertise we did not have in-house.

Beginning at the Beginning

I started with our customer’s demand.  We had a volume of X lights that had to go out the door every day.  I worked backwards from there.  I quickly found that we easily had enough purchased-part hardware for the back of the light.  We also had enough of the internal reflectors that handled the headlight beams; bulbs were, of course, standard and plentiful.  Leak testing was the primary failure point for a light once it started going through the assembly cell – leaks coming from bad glue pours.  The failure rate through the assembly line, almost always at the leak tester, meant that for every light that got packed ready to go, we had to try to assemble 1.2 lights.  This was well within the capacity of the assembly line to handle – although the failure rate was not sustainable in the long-term as the gluing operation, once done, could not be reworked or repaired… thus creating an expensive piece of scrap.

There were two main components to the light: the lens and the body.  I had a hunch where the problem was, and anecdotal feedback from the floor, but hunches and anecdotes are just that; I needed data.

The Body

The body consisted of injection molded polycarbonate.  The body was metalized to reflect the light from the main lamp and the two integrated signal lamps.  Masking was tricky on metalizing for a complex, curved piece like this one… in order to get 1.2 bodies ready for assembly, we had to attempt to metalize 1.4 pieces.  This was not a problem for the metalizing department’s capacity.

Swimming further upstream, the fallout from the molding process meant that we needed 1.5 molding cycles for every light being shipped.  Again, this was well within the capacity of the molding machine.  Clearly the body was not the issue.

The Lens

The lens was also a complex, 3D curved piece that wrapped around the front corners of the car.  Arriving at the assembly line, it required two secondary operations after molding.  Again swimming upstream towards raw materials, the next process was the painting operation.

In this instance there was a thin strip of paint applied to the back of the flange that wrapped around almost the whole exterior of the lamp.  An after-thought, it was not believed to be necessary initially as the back of that flange had some texture which was thought, during initial design, to be sufficient.  However, the prototype and first-run pieces showed that paint was necessary; thus it was kluged into the production sequence (at a vendor, as we didn’t do painting in-house).  Here was the first bottleneck – the fallout was huge.  To have 1.2 lenses ready-to-assembly, we needed 5 lenses shipped to the painting vendor.

Proceeding further upstream, the polycarbonate lens needed a clearcoat protection on the outside.  A scratch-and-UV protection, this too ended up with a huge fallout.  To have 5 lenses ready to get painted, we needed to have 7 lenses delivered to the clearcoat process.

Lastly, the injection molding process.  This lens was a multi-shot lens; what this means is that a large, multi-station mold was used to overmold one material over another to form a single piece with the amber turn signal plastic integrated into the lens.  Looking at the bins full of scrap at the machine, and gathering data, I found that to have 7 lenses going to clearcoat, we had to try to mold 11 lenses.  Fallout was primarily due to bleeding of one material, the amber, into the space which should have been clear.  Given the cycle time of the mold to produce a lens and the volumes required, this was not possible in a 24 hour day.

Clearly, the lens was the bottleneck component.  This confirmed both my hunch and the anecdotes – but now I had evidence.

Where to Begin?

My first visit was to the painting vendor seeing as they had a 5:1.2 needed-in to coming-out ratio.  I found that there were two primary scrap issues.

The first issue was that overspray was creating scrap.  Since this was a kluged-in-at-the-last-minute process step, masks had been hastily fabricated in order to meet deadlines.

This highlighted the need to invest in new “best available” masks.

The second issue was, interestingly enough, also related to masking – in clearcoat.  The clearcoat overspray would wrap around from the front of the lens to spatter on the back side of the flange where the paint would be applied.  The chemistry of the clearcoat was such that it had a very low surface energy – surface energy being critical in good adhesion – and the paint would not stick to it; this failure mode comprised, per the vendor’s data, about 40% of the fallout.

New masking for the clearcoat operation was also ordered.

Handling Damage

Another area of fallout was between the molding and clearcoat operations.  The lenses were placed in large baskets, layered one on top of the other with cardboard in between the layers.  These baskets would then be conveyed from molding to clearcoat; movement of the lenses, even just a few dozen yards across the plant, resulted in scratches and scuff marks making lenses unusable.  In a sharp-pencil look at this, we were having a 10% fallout from this short travel distance alone.

Stackable soft-foam-lined packing boxes, customized to this specific lens, were immediately ordered.

Low Hanging Fruit

The new masks for both painting and clearcoating took a week or two, as did the new protective containers for transporting the lenses from molding to clearcoating.  The results between these changes were nothing short of amazing.  The paint fallout fell from 5:1.2 to 2:1.2 from the combination of fighting paint overspray as well as better protecting the paint surface from clearcoat overspray.  Fallout from transportation damage fell to zero.  These simple changes resulted in a sea-change situation: the molding process, still strained, was now able to keep up with demand.

Additional process changes to the molding process – IIRC slowing down the injection process slightly – lowered pressures and reduced the bleeding occurrence enough so that even with a slightly longer cycle time, it more than made up for that with significantly-reduced scrap.  (After all, it’s not how fast you can crank out parts – it’s how fast you can crank out good parts.)

Further Steps

According to Theory of Constraints, the process with the longest cycle time is the process constraint – it is the drum that defines the pace of production; a defect in the constraint operation can never be made up by anything done in the downstream operations.  Further, any defect downstream resulting in a scrap piece can be considered a wasted constraint cycle.  In this case the molding of the lens was that constraint process.  Thus it was critical to address the molding defects.

With the improvements downstream there was an ability to build up an inventory of lenses to allow work on the mold – and fix the issues that resulted in scrap lenses.  This had the benefit of being able to revisit the injection rates, speeding them up again now that the mold sealing against leaks was more robust.

After this intense effort production was still not in great shape; we still had unacceptable levels of scrap fallout at multiple places in the pellets-to-pallets flow.  However, this concentrated focus – facilitated by my fallout map – was able to move this product from burning crisis to a more standard launch-issues mode.

Lessons Learned:

1. Starting at the end of the production line and working backwards through the various components’ fallouts can show where the critical issues lie.  A graphical approach like the one I used can help.
2. Last-minute additions to the required processes – in this case painting – should be scrutinized in particular as they are often (based on my experience here and in other such situations) a huge contributor to launch problems.
3. Masking for any paint or coating operation is critical: don’t be cheap, get the best masks possible even if they come at a premium cost.  (We also eventually invested in better masks for metalizing the body – another area of significant, though not critical-path, fallout.)
4. Handling can create scrap; in this case, polycarbonate is great for impacts but has poor scratch and scuff resistance without a coating.  Better to overprotect than underprotect.
5. A few low-hanging fruit improvements were the seed crystals in a progression of changes that allowed major progress.  Often simple things can make huge differences.
6. In the case of the painting operation, we leveraged the supplier’s expertise and involvement in the project’s success.  Good suppliers are not only worth a premium in terms of quality and time, but in terms of having very specific knowledge about their core business.
7. A critical piece of information was that we needed to anticipate – in both design and production planning – painting the backs of such lamp flanges in the future.  Part of the masking issue was that this had not been anticipated, and designing in a small lip/other feature for better mask seating would have helped enormously.  This information was passed back to our Design group.

Addendum: I realized that most of these in this series are manufacturing-related.  I would love to write about design aspects of my career, but in most instances these fall under NDAs I signed.

© 2014, David Hunt, PE