Senior structural engineers nowadays express concern that many young engineers do not have a good understanding of structural behaviour. They tend to attribute this to the fact that structural analysis solutions are performed on the computer thus depriving engineers of the experience of dong the calculations through which, they argue, understanding is mainly developed. Their reservations about the degree of understanding demonstrated by young engineers may have validity, but the attribution to the source of this to lack of experience in doing hand calculations is misplaced. In Section 2.4 of the book a range of strategies for developing understanding of structural behaviour is discussed. The main source of understanding of behaviour comes from reflective consideration to the results of analyses. Those in the pre-computer era who did have such understanding tended to derive their insights not so much by doing the calculations, but rather by being closer to the results. When the calculations are done by hand one expects errors and naturally scrutinises the results for them. One says 'Do these results look right? Are they consistent with the expected behaviour?' Answering such questions involves activities which are highly favourable to developing understanding of behaviour.

Although inadvisable, the normal approach to computer results is not to expect errors. The results may not be scrutinised; they are passed on to the next part of the process without being subject to reflection. But just because scrutiny of the results may not be (for some) a natural activity does not mean that it has to be neglected. Those who do not do it naturally must be taught and required to do it.

It is important to recognise that computer solutions deliver results which are potentially much richer in information than those from hand calculations. The potential for understanding behaviour from computer analysis is much greater than from hand analysis.

What is needed is a good process for doing structural analysis. Chapter 3 describes this process.

Table 1: Comparison of traditional and modern issues in structural analysis
# Traditional (mid 20th Century) Structural Analysis Modern Structural Analysis
1 Hand solution was the only option. Hand solution is too inefficient.
2 The scope of structural analysis models was very limited. The scope of structural analysis models is extensive thus modelling is much more effective than in the traditional context.
3 The process of generating hand solutions stimulated:
(a) good understanding of the underlying mechanics.
(b) limited understanding of structural behaviour.
The (computer) solution process is opaque to the user. Understanding of mechanics and structural behaviour does not arise naturally from using software.
4 Those who did the calculations were naturally close to the results since errors were expected. Working on verification of results stimulated understanding of behaviour. It is easy to use software without scrutinising the results. Reflection on the results is not a natural activity for all software users.
5 The potential for generating results was very limited. The potential for generating results is extensive.

From Table 1 we see that in the modern context computer solutions are much more efficient and effective (issues 1 and 2) but there are potential problems (Issues 3 and 4). However issue 5 points towards resolution of the problems outlined in Issue 4. Scrutiny of results is arguably the most important activity for developing understanding of behaviour. If the software user develops the attitude that the results should always be subject to scrutiny, not only will this negative feature of the modern situation be averted but, because of Issue 5, the potential for understanding behaviour will be greatly enhanced.

This leaves issue 3, that using computer solutions does not naturally develop understanding of the basic mechanics. The move to the modern context has changed the requirements for knowledge of mechanics. In the traditional context one had to have a intimate knowledge of the solution process. This is no longer the case. An understanding of the computer solution process can be helpful but one can also draw the analogy with the electronic engineer who specifies a chip which performs a transformation in a circuit. That engineer needs to be sure that (a) the right chip is being specified and (b) that it performs correctly in the circuit. He does not need to know how the chip works. The analogy here with the structural analyst is not complete in that some knowledge of the solution process may be advantageous, nevertheless knowledge of the solution process is not high on the list of requirements. The structural analyst does need to have good knowledge of force, stress, equilibrium, displacements, strain, compatibility, force deformation relationships, etc.

I do not suggest that hand calculations should be avoided. It is important to do 'back of an envelope' checks (as discussed in Section 3.6.3) but whether the calculations are done by hand or by computer is much less important than one's attitude to their validity and to the verification and interpretation of the results.