Saturday, 20 July 2013

Engineering and The Mind's Eye

3rd Century Roman Bronze Water Pump





















I'm just finishing "Engineering and The Mind's Eye" by Eugene S Ferguson, a truly excellent book which crystallises a misgiving I have been developing about the teaching of engineering for some time now.

Basically, the engineering curriculum emphasises mathematics above all else, next science, then language. Visual reasoning and intuition tend not to get much of a look in. It is as if engineering were a child of science and mathematics, rather than their precursor, as illustrated by the Roman pump above. Professional engineers still make far more use in professional practice of analysis by drawing, by analogy and using experience-based intuition than they do of mathematics and science.

As Ferguson says: "the art of engineering has been pushed aside in favour of the analytical "engineering sciences" which are higher in status and easier to teach...an engineering education that ignores its rich heritage of nonverbal learning will produce graduates who are dangerously ignorant of the myriad subtle ways in which the real world differs from the mathematical world their professors teach them"

One of the many great quotations in the book: "It is usually a shock to [engineering] students to discover what a small percentage of decisions made by a designer are made on the basis of the kind of calculation he has spent so much time learning in school"

This knocks on to discussions I have been having recently about the possibility of teaching creativity, or development of a personal style by beginners, as advocated by Gibbs.

The constrained creativity of professional designers is often grounded in recombination of things they have used before, or have seen others use. For example, I have a favoured way of combining static mixers, piston-diaphragm pumps, loading and pressure relief valves in order to create a robust, economical and reliable way to dose acids and alkalis into water under control of a pH probe.

I know from multiple experiences which areas of this design need the greatest attention in order to construct a system which control pH to within 0.1 pH units. I (and other professional designers) can reliably do this based on very sketchy information about the water which is to be treated.

In no real case is there sufficient sampling data to even determine a statistically significant estimate of the values of key parameters. Furthermore, the buffering capacity of the water is a key determinant of how much acid or alkali will be required. No scientifically valid estimate of a representative range of buffering capacities is ever available to the professional designer.

The way we do this is based partly on some mathematical analysis, and partly on an old scientific paper, but the end result is based at least as much on a feel for the data and certain qualitative aspects of situation as is its on science and maths. My choice of technology is based on a personal style, grounded in repeated experience, and to some extent the preferences of those who taught me the art of engineering.  
    
The details of how I put the system together in space, considering that it contains strongly corrosive chemicals under pressure, requires manual interventions and maintenance from time to time, and that it carries a client expectation of a neat and professionally designed appearance has very little at all to do with science and maths.

I reviewed the book on layout we used to use at Nottingham, and was amused to note that back in the 80's people were already speculating that computer programmes would soon be laying out plants, just as there are those now who think that process design will soon be done by computer programmes.

If someone put in sufficient effort, I am sure some sort of programme could be written to lay out plant and make process design choices. In fact I know that some already have been. They are not used much by practitioners because they are inferior to the art of a real engineer (and always will be) - they are at best only as good as their programmers, who are not professional process engineers.

To give a simplified example, I have a friend who used to cut fabric for upholstered furniture. This was a very well-paid job, as the material was very expensive, and a good cutter could get a few more suites out of a roll of fabric than the cutting pattern suggested by the best computer programme. They could also do it faster, as it was piece-work. My friend went into this job straight out of school, and served an apprenticeship with experienced cutters which gave him an extraordinary visual reasoning ability in this one application.

Being a process design engineer is orders of magnitude more complex than this. One learns to do it by doing it, and in doing so, building up reliable approaches to problems, tried and tested sub-assemblies of components, a feeling for appropriate margins of error and so on.

I am now teaching our students process design in this way. I had been considering attempting to teach formal approaches to creativity, but I see now that I am already teaching real engineering creativity by teaching process design based on my professional experience. This is how I learned, so this is how I will teach.

This is harder than teaching science and maths, and the process being taught is too complex ever to be practically replaceable with software. When we allow our students to use such software, it reliably causes them to have less understanding of the process. They get very precise answers, but hidden in the black box of the software are hundreds of tiny assumptions which no-one (perhaps not even the programmer) really understands.

The key skill of the process engineer is an intuitive grasp of the ways in which a complex system fits and works together. If process design were ever to be handed over to software, process plants would be produced than no-one really understood. This would be very bad from a safety point of view. 

Academics love the precision of maths, and the certainty of pure science, but real engineers know that there is far more to reality than these disciplines can usefully express, and we can meet society's needs better with rough calculations and our mind's eye than scientists and mathematicians ever will.

No comments: