This photo is not a ball mill. It is a very naughty expansion in a piping system.

At first glance, the photo looks convincingly like a ball mill. It has a cylindrical body, conical ends, flanges, bolts and smaller pipes entering at each end. Ignore the surrounding pipework, add two bearings and rotate it slowly, and most people in the mineral-processing industry would begin speaking about it in hushed tones.

For many years, I would explain large grinding mills to clients by saying: “It is just a pipe”. This was not always well received. The room would fall silent. People would look at me as though I had uttered the name Mufasa. Then, almost collectively, they would tremble: “Oooh … Mufasa.” It couldn’t just be a pipe. It was a mill! Mills had pedigree, proprietary knowledge, accumulated experience, highly specialised calculations and, apparently, things that could not safely be explained to ordinary engineers.

Calling a mill “just a pipe” is a deliberate engineering reduction. Engineers routinely reduce structures to their governing mechanics before adding back the detail. A bridge may begin as a beam, a pressure vessel as a shell and a grinding mill as a pipe. The simplification is not the conclusion; it is an engineering reduction to the governing mechanics.

The discovery of magic

Metso contributed a particularly useful phrase to the discussion in an article [1] titled “The magic of bolting for a grinding mill assembly.” The article states: “Bolting is magical.” It then discusses factors that may “disrupt the magic” and the steps required to obtain its full benefit.

ABB has identified another source of enchantment. Discussing gearless mill drives in [2], it says: “The magic behind the GMDs’ efficiency lies in attention to detail.” Thus, Metso has located magic in the bolts, while ABB has located it in the motor.

What remains for the purchaser is to determine whether the magic enters through the lubrication system or is installed later during commissioning.

A brief structural description

A large grinding mill is not a jet engine.

A jet engine contains rapidly rotating components operating under severe thermal gradients, aerodynamic instability, vibration, combustion, creep, fatigue and extraordinary consequences for small departures from design intent.

In general, for global structural analysis, a mill can safely be treated as a large, slowly rotating, quasi-static structure. Structurally, it is substantially a pipe with:

a large-diameter central section;
restrictions at each end;
a distributed internal load;
some torque;
support reactions; and
a regrettably large number of holes.

The load changes as the mill rotates, but it does so very slowly. We are not trying to predict the flutter boundary of a turbine blade at 12,000 rpm. We are trying to determine stresses in a large steel or iron structure moving at approximately the speed at which committee decisions are made. That does not mean mill design is trivial but its level of design difficulty should be described accurately.

Large mills involve fatigue, welding, castings, bolted joints, load estimation, fabrication tolerances, inspection reliability and the interpretation of finite-element results. Each of these requires competent engineering. None requires supernatural intervention. The engineering may be demanding. The governing mechanics are not supernatural.

Science during the week; magic on weekends

In [3] the authors write that a good failure analyst should remember: “the laws of physics apply everywhere.” That seems a sound principle. Physics has the considerable advantage of continuing to operate after the design meeting has finished.

Another paper [4] warns engineers: “Don’t fall in love with the model.” It also emphasises scepticism about finite-element results because models can contain incorrect elements, unsuitable meshes, bad boundary conditions, numerical problems and meaningless stress singularities. The “cleverness” was said to lie in producing a good model, recognising a bad one and interpreting the output competently.

An earlier mill-design paper [5] is even more direct: “Graphics should remain a tool, not a dazzling substitute for engineering expertise.” It compares an automatically generated mesh containing 2,890 elements with an engineered mesh containing only 176. Both achieved comparable accuracy, but the smaller model required engineering judgement rather than computational decoration.

Elsewhere in [4], readers are warned that modern graphics can display correct and incorrect results with equal splendour and that evaluators should not be “dazzled and sold” by coloured animations. This is all very sensible.

We therefore have a rather delightful progression within the same organisation:

The laws of physics apply everywhere.
Do not fall in love with the model.
Do not be dazzled by coloured pictures.
Bolting is magical.

James Randi would have asked for a controlled demonstration. MythBusters would at least have installed strain gauges.

Is there know-how?

Of course there is know-how in mill design.

There is also know-how in designing a garden shed. The existence of know-how does not transform either structure into a mystery.

Experience tells an engineer where failures have occurred, which assumptions are unreliable, what fabrication shops can realistically achieve, which defects matter and which apparently impressive calculations should be ignored. That experience has genuine value.

But experience should lead to better explanations. It should not lead to exemption from explanation. “Proprietary know-how” can mean several different things:

sound engineering developed through years of work;
useful empirical information not available publicly;
calculation procedures that have been tested against measurements;
inherited rules whose origins have been forgotten; or
in some cases, simply something nobody wishes to expose to independent examination.

These are not all equivalent.

A calculation does not become correct because it is confidential. A design method does not become profound because it has been used for 30 years or because it has used three-dimensional finite element analysis. It becomes credible when its assumptions are disclosed, its predictions are checked against measurements, and its failures are studied honestly.

That is science.

Magic, like marketing, works differently. It depends upon controlling what the audience is allowed to see.

The allegedly complicated parts

The most technically sophisticated components of a conventional mill are probably its bearings and lubrication system.

Hydrostatic and hydrodynamic bearings involve pressure distributions, oil-film thickness, pad geometry, thermal effects, deformation and changes in alignment under load. These are real interacting phenomena.

Some years ago, I had an undergraduate thesis student write a MATLAB program that could model essentially any bearing arrangement we wished to define. It could accommodate different pocket configurations, varying pocket pressures and oil flow rates, the deflected shape of the mill and deformation of the bearing pad.

He was an exceptional student and is now a professor at Oxford. That exceptionality is important. However, he did not arrive wearing a cloak, carrying a wand or demanding that the source code be stored at Hogwarts. The point is not that bearing analysis is trivial or that any undergraduate could have done it. The point is that even one of the more sophisticated parts of a mill could be reduced to geometry, fluid-film equations, equilibrium and structural deformation.

The problem was solved by formulating the physics, writing the equations, checking the numerical method and comparing the output with what the bearing ought physically to do.

That is the important distinction.

Complicated designs are not magic. A problem may contain many variables and still be understandable. A problem may require experience and still be explainable. A program may be proprietary and still be subject to verification.

Why the pipe image matters

The image is useful because it removes the industrial mythology. Without context, it appears to show a ball mill. In reality, it is an industrial pipe expansion.

The visual similarity is not accidental. A mill is fundamentally a pressure-vessel-like structure without significant internal pressure. It must carry distributed loads, transfer torque, survive fatigue and remain manufacturable. Its size makes the consequences of poor design expensive, but size alone does not create new laws of mechanics.

We should respect the engineering of large mills because their failure matters, not because they are mysterious.

There is a danger in presenting ordinary engineering as something close to alchemy. The owner becomes reluctant to ask basic questions. The independent reviewer is told that the true method cannot be disclosed. The coloured stress plot acquires authority merely by existing. Experience is invoked where evidence would be preferable.

And, eventually, someone writes that the bolts are magical.

The test

Perhaps the James Randi test for mill design should be simple.

Whenever someone attributes a result to special knowledge, ask:

What physical mechanism produces it?
What assumptions were made?
What measurements support it?
What would disprove it?
Can another competent engineer reproduce the result?

If those questions can be answered, we have engineering.

If they cannot, we may still have a perfectly good mill. But we should check whether the purchase order included a rabbit, a hat and an allowance for fairy dust.

References

[1] Metso, “The magic of bolting for a grinding mill assembly.”

[2] ABB, “Stator bars in Gearless Mill Drives — This is Engineering.”

[3] V. Svalbonas, “Accomplishments of Mill Design Specifications and Audits.”

[4] V. Svalbonas and M. Fresko, “How Safe Are Your Recent Mills? The Compatibility between FEA and Welding Codes.”

[5] V. Svalbonas, M. Fresko and C.C. Golembewski, “Correlations of Grinding Mill Structural Design Procedures.”