Computer models have been widely and commonly used on all projects nowadays due to its relative ease and low cost compared to 1-2 decades ago. Structural/geotechnical modelling is the default norm for analysis on all projects. I am definitely not a modelling specialist but I consider myself to be a casual modeller. There are a few quick tips that I have learned over the years and I would like to share them.
Engineers take pride in the models we do. Some actually fall in love with their models – there is nothing like the feeling of completing a hugely complex and beautiful model and seeing its animations flashing on the screen. Engineers naturally take on challenges. It is naturally so because it is how we improve. We also wish to give the best value to the client by trimming off all unnecessary bits, and this must be done by conducting more sophisticated modelling. However, caution must be taken when we do this, as it sometimes goes against our ultimate goal of delivering things to time, budget and quality.
Engineers have ‘optimism bias’ (according to psychologists) when conducting models. In other words, we tend to be over-ambitious about our models, or our ability to manage and complete sophisticated models, on several fronts, as listed below.
Scale
Engineers have the tendency of making a model all-inclusive, once and for all. This can be risky at times, because a small problem in the model will cause the whole model to fail to run. Avoiding disproportionate impact is something engineers ought to implement in their design, and not just in modelling. The bigger it is, the harder it falls.
Level of detail
There is a what I call the ‘eye-candy fallacy’, where engineers take the pleasure of building up a super model that looks just like the real thing. This is particularly the case nowadays with the software capability of converting BIM models directly into geometric input files. Close to 100% of the time, this attempt is very risky and unnecessary. The level of complexity increases the risk of the model not running parabolically.
Complexity
Complexity could take the form of geometry, such as curves and bends, or material/support properties such as non-linearity. Experienced modellers know that it is hugely risky to conduct a complex model, because the amount of time to de-bug the model to make it run properly can take multiple times the amount of time it takes to build it up in the first place. What is initially estimated to be 4 weeks’ worth of work could take 6 months in the end and still not giving you satisfactory results. If we wish to be reliable in our promises and consistent in our performance, we must know the optimum amount of complexity to take on.
What to do then?
1, In general, I believe we engineers should approach the structure at a macro level and gradually zoom in the lens to attend details. There should be a global model supported by multiple local models where possible. For example, for a bridge, there could be a wireframe model representing the whole structure, supported by local models for the deck, piers, bearings, etc. separately. Engineers must be in possession of the ability of abstracting the primary elements out of a structure. In other words, model the things that really matters.
2, Always consider cost and risk against benefit when cranking the model up on the sophistication scale. Is it really worth the effort and risk? Will this detail really affect the results significantly? We must give up perfectionism when conducting modelling. We must know when to stop, after sufficient accuracy has been achieved. This calls us engineers to know exactly, where laser accuracy must be achieved, and where it can be relaxed.
3, See whether it is absolutely necessary to put everything in a single model? Can the whole analysis be broken down into smaller more manageable models? Even if the interactions between various parts call for an integrated model, can the interactions be modelled in simpler ways, e.g. converted into imposed forces and bending?
4, Think twice and three times more before attempting to build a complex model. Avoid the ‘eye-candy fallacy’ by restraining ourselves from the impulse of taking on challenges or showing off our models. Try to be incremental when building a complex model. Break down the model into manageable steps and make sure each and every step is checked before moving into more details.
Good luck and happy modelling!
Comments