The London Millennium Bridge was designed using FEA, which proved that the bridge would be stable. After completion it was found to have the famous wobble. The FEA software was used to ensure stability at 2 Hz, but not at 1 Hz. I read, but cannot find any relevant source, that the grid spacing in the model used was such that the 1 Hz problem would not have been detected.

If FEA has been used elsewhere and the results verified then it may be possible to use it again with lesser need for experimental verification. If yours is the first time I would park it as interesting but needs verification. Credible, but not verified credible.

Your question [3] would expose the weaknesses of FEA  - are you sure of how the material properties change with ageing? The effects of corrosion, or thermal cycling? I think (= I'm guessing) that either generous safety factors are used, or accelerated ageing physical tests (as found in the nuclear and aeronautic industries).

The question is related to my following publication

Conference Paper The switch from epoxy lined carbon steel fermentation and st...

Dear John,

I used to manage 29 design engineers at Rolls-Royce (aero-space) including containment casings (pressure vessels).

When used in the hands of an expert CFD can give remarkably accurate results. Indeed  some fan blades now can have reduced testing as some certification can be done analytically.

Equally, when used by a novice they can give answers that are orders of magnitude in error.

We used to use the Demming cycle on design, i.e. one of continuous improvement from analysis to experiment back to analysis.

The tests were used to validate and improve the CFD model, then future designs started on a closer difference between prediction and experiment.

A wise professor on mine once said that: experiments are always right, as long as you know what you are measuring, analyses are usually wrong, until proven by experiment.

So, if you have a CFD expert trust them. If you are starting afresh, do lots of experimental validation to prove the model.

Let me know if I can be of further assistance.

Regards

Paul

The London Millennium Bridge was designed using FEA, which proved that the bridge would be stable. After completion it was found to have the famous wobble. The FEA software was used to ensure stability at 2 Hz, but not at 1 Hz. I read, but cannot find any relevant source, that the grid spacing in the model used was such that the 1 Hz problem would not have been detected.

If FEA has been used elsewhere and the results verified then it may be possible to use it again with lesser need for experimental verification. If yours is the first time I would park it as interesting but needs verification. Credible, but not verified credible.

Your question [3] would expose the weaknesses of FEA  - are you sure of how the material properties change with ageing? The effects of corrosion, or thermal cycling? I think (= I'm guessing) that either generous safety factors are used, or accelerated ageing physical tests (as found in the nuclear and aeronautic industries).

Dear Paul

Many thanks for your sharing your experiences in engineering and using CFD, normally used to solve fluid dynamical problems. In my experience we have used FEA normally associated with solid mechanics. In both techniques of analysis we have an opportunity to predict designs and their potential problems over a specific time period. Would you say this an accurate definition of these computerised modelling technique's?

Dear Jeremy

Thanks for your observations and answers.  Your comments to question 3 are important. The impact of corrosion has had dramatic effect on the life cycle of the assets. In some case equipment has lasted 6 weeks and some over 20 years. How does one enter these safety factors to predict divergent life cycles?, May be by collating the experiences and feeding them in to the model giving them a relative safety factor?