lederhosen

Rather beautiful post on

deadmachinery the other day. Linking image here so I can enthuse about it:

Paint flaking.

There are two great secrets of science. The first is that simple rules often hide within them a recipe for tremendously sophisticated behaviours - see Conway's Life to learn how the rules "If three neighbours, live; if two neighbours, stay as you are; else, die" give rise to the magic of the 'puffer train' and even a Turing machine. (Click on some of the applets, they're worth it.)

The second is that the sophisticated behaviours we encounter in the world around us are often driven by a handful of simple rules.

turnberryknkn can tell you, better than I, how a few amino acids arranged in a certain pattern tell a cell whether to live peaceably amongst its neighbours or to grow and divide without stopping.

theonementat could tell you interesting things about how to make a microbe decide that heavy metals are tasty... but that would involve Asking Him About His Thesis.

Look at this image, the way the cracking paint forms a semi-regular grid. I don't know the exact process involved, but it would be something like this:

Persuading two layers of different materials to stay together is a big problem in applied science. Weathering, temperature, loading, vibrations, all these things affect the natural shape of each layer, causing them to bend and flex - and because the materials are different, each bends and flexes to a different degree.

When the paint's new, it's still quite flexible and the 'glue' holding it onto the metal is strong, so it just goes along with whatever the metal wants to do. But as it weathers, it becomes hard and brittle, and eventually the stresses on it become too much, and it starts to crack at one point.

Now, one thing you have to understand about cracks, is that they propagate. When you have forces to either side of a crack in a material, pulling it apart, those forces concentrate in the material just ahead of the crack. Ever tried tearing open a packet of candy? Tearing a hole in an intact piece of plastic, that's hard work. But start that tear with a tiny little nick, and it's child's play; indeed, sometimes it's all too hard to *stop* the tear before it runs down to the bottom of the packet and everything falls out. (The secret, BTW, is to bring the tear back around in a circle. Circular holes don't focus the force in the way that a thin crack-tip does.) So this crack extends from the original site, at right angles to the stresses that drive it.

Near the center of the photo, the stresses seem to have been greatest in the vertical direction, so the crack runs horizontally.

What happens nearby? Well, in the region very close to the crack, that crack takes a lot of the tension off. The paint isn't being pulled up and down, it's only being pulled away from the crack, so it's not stretched any more. But if you go a bit further away from the crack, that's no longer the case - there's enough paint around, stuck firmly to the metal, that the tension builds up again. And more than that, you now have air and water seeping in through the crack, getting under the paint and rusting it, which increases the deformation.

So, the paint cracks again. This near the first crack site, the stresses are in much the same direction, so you get a set of near-parallel cracks dividing the paint into long ribbons. Note that near the center of the site, those ribbons are all of much the same thickness - it takes a certain distance across for the stresses caused by the metal's movements to reach breaking point for the paint.

Meanwhile, elsewhere on the car, other crack-families have started. And because the stresses aren't the same here, they may not run in the same direction. Indeed, if you look at the edges of the image, you'll see most of the newer cracks there run towards the center, because the rust there has swelled the structure enough to change the stresses.

Just above the central bare patch, you can see how two of these families met. Towards the top, they've hit at right-angles, and whichever one gets there first interrupts the other. (Curiously, sometimes you do get two cracks crossing one another, with neither terminated at the point of intersection - I suspect this has to do with inelastic behaviour of the paint, but that's a topic for another time.) But down and to the right of that, they've come together at a shallow enough angle that they turn towards one another and meet, forming a single bent line.

Now we have strips, but it doesn't end there. Towards the center, the rust has created more and more horizontal stresses - and while these took longer to build up than the vertical stresses, they've now grown strong enough to break those ribbons of paint crossways into neat little rectangles.

Why are these rectangles longer than the ribbons were thick? Well, by this stage the rust is weakening the paint's adhesion to the metal, and the edges are starting to lift up, so there's more leeway now for the paint to 'give' before it breaks.

* * * * *

For Spanish Inquisition fans: the third Great Secret is that when viewed from a great height, these sophisticated behaviours may once again coalesce into a handful of simple rules. Although there are many, many unknowns between the laws that govern individual atoms in the brain, and the levels of organic chemistry, biochemistry, neurology, and psychology above them, we still know that if you walk into a room of geeks and declare "This is an ex-parrot", somebody else will respond "It has ceased to be!"