1980s refrigerator shelf failures

Tom's personal refrigerator history. Used as the canonical teaching parable for "you can't substitute one material for another without redesigning the joints." Mid-1980s threaded-joint plastic failure → 1990 KitchenAid slot-and-drawer design.

Then in the mid-eighties I bought a new refrigerator. The outside was steel and the inside was a panel of plastic. Plastic is a much better material for the inside of a refrigerator because you're going to spill foods on it — it has a low internal surface energy, much lower than metals, and things don't stick to it as easily. But within five years I had to replace the refrigerator, because they used screws to attach the shelves to the plastic, and plastic is a lousy material for threaded joints. You just shouldn't use threaded joints in plastics — maybe in a Hasbro or Fisher-Price toy. And your child is going to come to you and say, daddy, my toy broke, can you glue it together. And you say, of course not, I took Professor Eagar's course on adhesives and we know that plastics have low surface energy and there's no good glue to fix your cheap little toy, child. Well, maybe you won't say it exactly that way to them, but that's in fact the case. Most of these things you just throw away. I had to throw away the refrigerator because the shelves were broken.

Material-substitution case. Manufacturers transitioned to plastic interiors but kept screw-mounted shelves, designing plastic as if it were steel. Plastic's brittleness at stress concentrations caused three-year failure cycles until KitchenAid and others adopted molded slots.

But what bugged me, because I lived through this transition in the 80s: I would go and buy a nice refrigerator, and within three years it was a piece of junk, because they screwed the shelves into the plastic. That's not how you design plastic, folks. Plastic can't take screw holes and sharp stress concentration; it's a somewhat brittle material. I've shown you about fracture mechanics — you put a little notch in something and it breaks easily. Plastics don't have the toughness of steel. If you're going to design a refrigerator — by the 1990s KitchenAid and other people learned — you slide the shelves in, you don't screw them in. Plastic has this wonderful property: it's easy to mold into complex shapes, so you mold a slot in there. Now you go look at virtually any refrigerator, it's made out of plastic but there're no screws on the inside. The other problem with screws on the inside: after time they would start rusting, and you'd have a little rusty drip on the inside. Your refrigerator didn't look very good. Plastic is great — it's got great room-temperature corrosion resistance.

Used as opening example of how naïve material substitution (metal → plastic shelving) creates new stress concentrations at dissimilar-material joints. Holman teaches it as a thermal-expansion-mismatch failure.

One of the main material substitution issues that happened years ago was when they switched from metal shelving to plastic shelving. What they initially did was say, all right, we're going to mold this out of plastic and stick it in, and it'll save us a ton of weight. One of the things here is that when you're working with a system, you can't just look at one property or one particular aspect of the system. You have to look at everything as a whole and how it interacts.