The site manager who told John Karadelis (Letters last week) that 'analysis skills need no longer be taught-because a complex analysis can be done by pressing a button' was right, up to a point.
But rst the data needs to be correct. Secondly it is no use choosing the wrong form of analysis, such as a linear one when a non-linear analysis would be more appropriate.
Thirdly, the output has to be verified - do the reactions confirm the static equilibrium of your original problem; do the deflections make sense?
What needs to be taught is not analytical technique so much as underlying knowledge of material behaviour, the response of many different kinds of structure to load, and above all an ability to use simple statics and graphic imagination to see what kind of result a finite element analysis ought to produce in any given situation.
Howard G Allen, emeritus professor of structural engineering, University of Southampton, 38, Glenwood Avenue, Southampton, Hants, SO16 3QA