The Confidence to Build

The author acts as a consultant of last resort on arching structures, chiefly in the UK. He has been involved on the periphery of a number of major infrastructure projects where interaction with historic structures has been an issue. Mostly, this work is in unlocking restraints that have been imposed as a result of lack of understanding. Modern computer analysis is very demanding of knowledge, skill and time. Developing analytical skills seems to develop a corresponding belief system that the computer knows better than the structure, but any analysis is only as good as its input. The input requires understanding of:

The analytical tools

The likely behavior of the structure

The interaction of what can be modeled and the real behaviour The values and confidence levels of the input

The values and confidence levels of the output.

If any of these is missing, the results are unlikely to be useful. Unquestioning confidence in the output of analysis may lead to either grossly un-economic or even dangerous designs.

The paper discusses a number of examples where complex analysis has yielded false results. Most often this is because the model constructed falls far short of representing the real structure. Each example is tested against the fundamental principle of statics, that every force must be traced from point of application to an adequate foundation, while remembering that “every structure is a system of interconnected stiffnesses” (Happold).