“Model or Muddle?” (Thoughts on using structural software)

Computer software packages are powerful, increasingly user friendly but they remain only tools and can only be as good as the engineer using them. It is he or she who still carries the ultimate responsibility. The reliability of the results depends on the engineer’s awareness of structural principles and his or her correct assumptions on the input data – otherwise, Rubbish in equals Rubbish out.

For a start, care is needed when choosing the appropriate software package. For example, a simple steel frame is best analysed using a straightforward 3-D elastic model rather than sophisticated finite element software. Finite element software is more suited for concrete structures with distinctive features of some complexity - e.g.: Flat slab with irregular grids.

Whilst setting up of some parameters may often be assumed to be taken care of by the software defaults, this should not be taken for granted. Important parameters such as material properties, self weights, load combinations, directional axis, support conditions, etc; require checking and confirmation by an engineer. These should be considered at the beginning before even attempting to model the structure.

A model must accurately represent the structure being considered. This involves not only correct geometry but also realistic section properties, support conditions and joint details. One should not use a model that is more complex than necessary. The simplest models which satisfy the purposes of the analysis are normally the most useful.

After these models have been created and within seconds of running, mathematically demanding problems, which require numerous iterations, complex matrices or even differential equations are now easily solved without much effort by the computer. The benefits seem to be overwhelming as the design outputs are neatly displayed on calculation sheets. However, these analyses and designs can have significant shortcomings – e.g.: selection of unsuitable section sizes, impractical steel reinforcement for a concrete beam, wrong usage of connections, etc.
Thus, knowing how to get the computer model right has never been more important. These are some of my suggestions to assist engineers in their use of structural software: -

1. Understand the requirements and choose the right software package.
2. Check the parameters and inputs.
3. Check overall equilibrium: The total applied forces should equate to the overall reactions.
4. Examine the model’s deflected shape, the shape of bending moment diagram and the distribution of forces – Do these agree with what would be expected from the loading? For example, an applied uniform load on a simply supported beam should give a symmetric parabolic bending moment diagram.
5. Compare results using engineering judgement with simplified hand calculations.
   

The final output from the software can only be as good as the engineer’s input, because powerful as it is, the computer cannot think for itself. Effective and reliable use of computer software packages requires a solid understanding of their limitations. The experience of an engineer including an understanding of the engineering fundamentals, the selection of the correct software package and using it correctly all play a critical role in the design of any structures.