A very useful tool in examining production problems of discrete items is the “Measles Chart” – marking the location of defects on a drawing or schematic of a product. In this essay I’ll discuss my first, instinctive, use of this technique, and then briefly show how I used it on the three radiators discussed in a prior Fix the Problem posting (link will be given below, close to the analysis), asking the readers to coordinate the images here with the text in that other essay.
Bins of Scrapped Parts
When I first started at Ford Motor Company in Sandusky, Ohio, my new boss told me to go out into the plant, explore, and start to learn my way around the 1.1 million square foot facility. One day I was out wandering and came across the newly-launched Ford/Taurus headlight line, and noticed that every few minutes a shiny component would arc out from the central repair area inside the line and land in a large basket, something like 5 feet square, by 4 high; at that moment in the mid-morning the basket was half-full. So I went into the repair area, introduced myself as a new engineer out exploring, and asked about the basket full of scrap parts.
I learned that the repair loop’s primary job, as expressed by-the-numbers, was to take assemblies that had failed the leak test, remove the reflector which formed the back of the headlight, replace it, and send it back into the loop to go through the leak test again. The presumption, therefore, was that the leak was the reflector’s fault. At something like $15 a piece, with baskets being filled almost daily for return to the supplier, this was adding up to big bucks fast.
I asked if there were a way to determine the location of the leak, as I realized that if the reflector were defective the leak would be on the inside of the rubber molded “boot” around the edge of the reflector, while if it were a seating problem it would be on the outside. Sure; just spray soapy-water “snoop” and watch for bubbles while the lamp was under pressure.
Because I was new, and this line was not my responsibility, I sought out the Line Engineer and – again – introduced myself. I brought him back to the line, explained what I wanted to do, and got his assent which he communicated to the repair operator. Here’s what I wanted:
- I drew an oval representing the perimeter of the reflector, and asked the repair operator to make a mark where the leak was found
- Explain to the next shift repair operator to do the same thing
- I’d come by, collect the sheet, and give another one every morning.
What Did The Data Say?
This first image was typical. Almost all the marks were on the outside, indicating some kind of a seating issue. With the cooperation of the Line Engineer, we instituted a new rework procedure:
- Use “snoop” to determine if the leak was inside or outside.
- If inside, remove and replace the reflector.
- If outside, make a mark with a wax pencil on the bottom of the lamp, back the screws holding the reflector off, jiggle and reseat the reflector, and redrive the screws.
- If a lamp with a mark came back, remove the reflector, replace it, and make another mark on the bottom of the lamp to form an “X”. If a lamp came back in even with a new reflector, scrap the whole thing because driving and redriving the self-tapping screws – should the lamp come back yet again – would weaken the plastic where they were driven if they were cycled too many times.
The reseating operation almost always worked. Instead of shipping a full basket back to the supplier virtually every day, returns fell to – going from memory – something like one basket every 2-3 weeks. Furthermore, almost all returns were due to issues with the glue joining the hard plastic reflector to the soft rubber boot… which became the subject of an improvement project to find a replacement glue and more reliable / repeatable dispensing system compatible with that to-be-identified glue. (But that’s a tale for another essay.)
To my recollection no full assembly was ever scrapped.
In my essay Fix the Problem IX I discussed problem solving on three different radiators. In this case I added a new wrinkle: color coding for each shift (as this department was a round-the-clock operation). Specifically, black for first shift, red for second, and green for third. Please look at the images, below, and then read the essay about these radiator trouble-shooting cases. (Note that these drawings are representative, and are instructional only.)
For the first radiator, the measles chart looked like this:
There was no clear pattern in either location or shift.
The second radiator had this chart:
Definitely a location dependency, but no shift dependency. All high fins were adjacent to the Z-notch.
And the last one had a chart looking like this:
Both a location and shift dependency. All high fins were right next to the notches for the braze bars.
When faced with a production defect issue a Measles Chart, especially with the added shift color differences, can be enormously useful – yet very simple – in ferreting out patterns in time, personnel, and location. I strongly recommend its use as a standard, first-pass tool in collecting data before more advanced, time-and-resource-intensive data collection is pursued (again, assuming the product being produced is amenable to such charts, of course).