A note. A couple of people have asked why I am not posting case studies from my six years at Cabot Corporation but, instead, am reaching back earlier in my career. This is not because I have nothing to post; rather, because Cabot’s technology is highly proprietary and confidential, I am extremely constrained in what I can say. (One report I wrote detailing a root-cause analysis of a field failure was, by necessity, so laced with the design details of a critical system component that people were ordered to only read it online on the company’s intranet and not even print it, lest a copy somehow fall into a competitor’s hands! Needless to say, while a fascinating experience from which I’m sure people could learn, I don’t think I could write about it in a meaningful way without revealing things I can’t reveal because of an ongoing confidentiality agreement). At some point in the future I will try to write up some things from there, but will need to be extremely careful in not disclosing anything proprietary.
When I joined Ford Motor Company in the picturesque city of Sandusky, Ohio, the best boss I’ve ever had told me to go out to the floor and “get lost”. His idea was to have me become familiar with the plant and the product lines it manufactured: automotive exterior lighting, air cleaners, and carbon canisters for the fuel system. Over time and several projects I became very familiar with leak testing and water ingress into lighting systems. I became one of the few people outside the Quality department involved in examining the lights returned from dealers that had water in them. (One of the recommendations that came out of that effort was passing back information to the dealers that if the water was clearly just condensate from air coming through the vent, cooling overnight, and precipitating dew inside the lamp, just turn it on and drive off the water with the bulb’s heat; nothing was wrong with the light itself.)
It was not long after the launch of the Ford Escort’s new look, which including a new headlight, that we started to get reports from the assembly plant where – once or twice a week – a headlight was filling with water during the post-assembly car wash. So, logically, I got tapped with the problem.
First things first, I asked what we’d done with the returned items as we’d had at least a dozen of them. These headlights had been dried out and leak-tested on our production line. They passed the leak test. We also had a set of testing protocols, done on every light as a part of the pre-launch design validation process, consisting of water spray testing while the light was on, while off, cycled on-off, humidity and dust testing, etc. Every single light we’d gotten back was comprehensively run through these protocols again; they all passed.
The Design Engineer was stumped, as was the New Model Launch Engineer. My immediate conclusion was that it was not the light – it was the light in the environment of the car. I told the “troubleshooting manager at large”, a manager late in his career whose job it was to handle customer issues like this, that we needed to be called the next time it happened and drive up to take a look at the light in the vehicle. We only had to wait a couple of days and we got the call; we drove up immediately. Sure enough, standing water in the headlight.
We drove the car into a side spray booth used for additional spot washing, and I took up a hose. I started spraying a jet of water around the perimeter of the light’s interface with the sheet metal around it. When I got to one point underneath the integral side turn signal, I saw water coming in through the vent and into the lamp. A little fine-tuning of the position and angle and I could reliably get a good stream of water coming in. What was happening was that the water was ricocheting off the sheet metal underneath the light, and into the vent which was shaped like a piece of curved macaroni. Since the opening of the vent faced forward, it acted like a scoop to catch the water. It took me no more than ten minutes to find it, and once I knew how I could reliably do it on any Escort.
The first step, now that we knew how the water was coming in, was to block the flow of water from getting to the vent. I originated the containment action: using a foam strip stuck to the light just in front of the vent, and which would be compressed against the sheet metal when installed and act as a dam. We installed several such lights into a couple of different Escorts in rotation and I tried to spray water in – I couldn’t do it despite a good hour of trying. This was a good containment action to patch the problem.
Our Design counterparts then designed molded-in ribs that created a housing around the vent; this design was mocked up on several more lights. These ribs were intended to shield the vent from any flow of water impinging on the opening to get scooped in. Again, I couldn’t duplicate the failure. With the customer’s approval we changed the mold and did one more proof test series on real cars. Once more, success: I couldn’t get water into the lights with the new molded ribs.
- When presented with a component failure where the component itself, even after the system failure, passes the test in stand-alone tests – look at the systemic environment.
- Only by testing a real failure with the failing component in situ can the failure mode truly be investigated.
- Refine or modify the testing protocols to include the surrounding environment/system when doing validation testing of new, similar components.
- Solicit other opinions early, especially on highly-visible problems. I was only called in when the incidents reached double digits and the assembly plant manager was screaming at our Design team.
- Any failure mode discovered that can be designed out in the next product must be. DFMEA is a great tool for this, as is creating an explicit and formal method for passing information from Manufacturing back upstream. To wit, our specific recommendations to our Design group:
- Obtain or kluge a mock-up of the sheet metal and mounting environment of the light to use in the spray, moisture, and dust testing protocols.
- If using open-ended vent tubes, either have the opening pointed away from any possible water infringement, or proactively shield it to avoid this situation from occurring again.
© 2013, David Hunt, PE