Running-in wear

Why does new machinery last longer if its first hours are spent wearing itself down?

A new engine is handed over with instructions that sound like a contradiction: vary the load, do not labour it, do not hold one steady speed, change the oil early because it will be full of metal. Metal, from a machine that has done nothing yet.

The instructions accept that the first hours will remove material at a rate you would call a fault at any other time in the machine's life — and claim that a machine treated this way lasts longer than one that was not. If wear is the enemy, how can deliberately spending some of it early buy you more of the machine?

Start with what the surfaces are actually like when new. A ground or honed surface looks mirror-flat and is nothing of the kind. Under magnification it is a landscape of ridges and hollows — asperities, in the trade — with peaks standing above the mean line. Two such surfaces pressed together do not meet across their apparent area. They meet only where peak lands on peak.

Now ask the consequence. The load is carried by that small true contact area, so the pressure at those points is enormous — orders of magnitude above the nominal figure you get by dividing load by area. Enough to deform the peaks plastically, enough locally to break through the oil film that is supposed to keep the surfaces apart.

So a new machine is not in a gentle state at all. It is a machine with a few tiny, savagely loaded contact points. Wear there is rapid, and it happens exactly where the pressure is highest — which is to say, at the peaks.

Follow that. Removing a peak takes away the point that was carrying load, so the load transfers to the next-highest points, and the number of points sharing it rises. True contact area grows. Contact pressure falls. And falling pressure means falling wear rate. The process attacks its own cause: the geometry that produced severe wear is the geometry that severe wear removes.

That is the whole trick, and it is worth stating plainly. Running-in is not damage that the machine survives. It is a convergence — two surfaces of independent, arbitrary shape becoming complementary to each other, driven by the loads they will actually carry. Each surface ends up bearing the imprint of the specific mate it will spend its life with, and of the way that pair is loaded. No manufacturing process can supply that in advance, because the fit that is wanted is a fit to a particular partner under particular load.

There is a second, less obvious gain. As the peaks flatten while the hollows survive, the surface becomes smooth at its high points and still open at its low ones. Those retained hollows hold oil. A honed cylinder's crosshatch is the deliberate version of the same idea. The run-in surface is therefore better than a merely smoother one: flat where it must carry, grooved where it must store.

Now the instructions decode themselves. Some load is needed, because pressure is the agent that does the bedding — a machine idled gently through its first hours does not run in at all. Too much load and the peaks do not merely deform but seize and tear, scarring the surfaces rather than conforming them. Varying speed and load beds the parts across their whole working range rather than to one operating point. And the early oil change is because the debris is a hard abrasive circulating in the very film meant to separate the parts.

One honest qualification: modern engines are machined and plateau-honed far closer to their run-in state than older ones were, so the ceremony surrounding running-in has shrunk considerably and manufacturers differ on how much still matters. The physics has not changed; the amount of conforming left to do has.

Two things this is easily confused with.

The first is the idea that running-in is a fixed phase — so many hours and it is done. It is better read as a process that ends when it has removed its own driver. That is why it is self-limiting: once contact pressure has fallen far enough, the rate falls away and the machine settles into the slow steady-state wear that occupies the rest of its life. Interrupt it with a heavy load and you may have scuffed the surfaces into a shape that never converges properly.

The second is the assumption that smoother is always better. Push that to its limit and you would polish both surfaces to a mirror before assembly. In practice that removes the oil-holding hollows and, in some pairings, encourages the surfaces to adhere. What running-in produces is not maximum smoothness but a particular distribution: flattened peaks with the valleys intact.

Notice what makes this adaptation rather than mere degradation. Adaptation needs a feedback coupling the damage to its own cause. Here it is direct — wear occurs where pressure is highest, and wear there is exactly what lowers that pressure. Uncoupled wear, the kind that grinds a whole surface away indifferently, has no such feedback and simply consumes the part.

What would refute the account? If running-in were only about clearing manufacturing debris, then a scrupulously cleaned assembly would show no run-in period at all — and it does, which is why the conforming story rather than the cleanliness story is the one to keep.

• Why plateau honing exists, and how far a manufacturing process can pre-empt the conforming that running-in would otherwise do.
• What scuffing is at the surface level, and why exceeding the load limit during running-in produces a fault that no later gentleness repairs.

Every term the collection defines is gathered in the glossary.