#3 - Empirical Models
As I mentioned in the last post, models are what engineers use to simplify reality into manageable chunks. When I say models, I don’t just mean complex formulas and fancy computer programs. In my definition, a model can be something as simple as copying a design. If the idea works for a lot of people, it’s usually adopted organically as a rule of thumb, or heuristic. In these empirical models, knowledge flows from the practical rather than the theoretical. (Empirical just means based on observation or experience.)
For example, a current industry standard in aerospace is that the closest you should put a fastener (bolt) to the edge of a metallic part is 2 x D (the diameter of the fastener). Any closer and you risk having the fastener tear out, like a sheet in a three-ring binder. This heuristic, straightforwardly called “edge distance”, was apparently raised to 2D from 1.5D after a series of crashes of the de Havilland Comet jet in the 1950s. A practical finding from a failure led to the evolution of the knowledge that is accepted today.
You can imagine how much time and effort having the edge distance heuristic can save. For the vast majority of the hundreds and thousands of fasteners on an airplane or spacecraft, if you are above 2D, you can rest easy and not calculate anything else. And yet, despite the effectiveness and continued use of rules of thumb (rule of thumbs?), it seems to me that modern engineers tend to turn up their nose at them. Take, for example, this excerpt from J.G. Landels’ Engineering in the Ancient World, where he describes the development of a water pump in ancient Rome:
Vitruvius recommends that the rotor should slope upwards at an angle of 37°. Though this is arrived at from the well-known construction of a right-angled triangle with its sides in the ration 5:4:3, it probably represents an approximation to an optimum angle which has been arrived at empirically. (‘Us put ‘un up like this-yur, an’ ‘ee’ wurked allright; us put ‘un higher an’ ‘ee didden wurk so gude, so us put ‘un back where ‘ee be, an’ let ‘un bide’.)
Otherwise a fantastic book, but I don’t understand why empirical development is shown as uneducated. When engineering, you just need to make something that works. When you offer your creation up to Mother Nature, she doesn’t much care whether you based your design on observations or on equations. As Landels himself wrote later in his book:
Since the Roman contribution to technology, though considerable, was almost entirely in the field of practical application, the state of Greek theoretical knowledge may be regarded, for all intents and purposes, as that of the whole Mediterranean world and the Roman Empire, down to the fifth century A.D. or even later.
When a society revered for its engineering and architectural prowess contributed next to nothing to theoretical knowledge, how can we disregard such a practical approach as backwards?