FEA as a method of code checking is a can of worms. There are two general points that some argue make FEA not a valid code check of any kind.
1. A structural engineer who does do code checks must decide on what an acceptable stress level is for a given situation. Then you can check the simulation stress levels. For steel the values of a desirable max stress can range from 5KSI to 15 KSI. Typically steel design is limited by deflection and not stress levels.
2. Creating proper load condition for a simulation is very tricky and takes years of experience & feedback to do correctly. Until you close the cycle with tests and real life failures, it is difficult to know if the models are set up right to have usable values.
What it is good for is identifying the weak spots of a design.
I disagree. Simulating reality in FEA is a bit tricky for those that do not understand how it differs from a conventional hand calc which is what it is often compared to. The problem with this is that most people do not appreciate the inherent assumptions in a hand calc. Simulating a simple bending problem can be done with ease in FEA and by a hand calc. However, the FEA solution (assuming continuum elements here as is common with SWX Sim) will not only nail the bending stress in the middle but it will also highlight the stress risers at the restraint which the hand calc conveniently ignores by its nature because it is not a realistic depiction at that location for simply supported bending. You can't really test a hand calc because you have to have a realistic holding system. The hand calc misleads you into not even measuring the response at the restraint as it isn't highlighted by the analysis for design of the test set up. This situation often leads to posts on this forum of "FEA doesn't match my hand calc?". Nothing could be further from the truth. As far as code checks go, it is used extensively in many industries including civil structural codes though the vastly conservative nature of those codes as you mention limits it ability to add high value over other more conventional and familiar methods. In many other industries such as more complicated pressure vessels, aerospace structure and others it may offer the only way to get where you want to code and to demonstrate code compliance. Specifically with 1 above you are referring to conventional structure in steel buildings and other building code driven design where you do not need FEA unless you are going outside the box of conventional construction - which you can certainly do with evidence of compliance with the intent of the code and not its prescriptions. FEA maybe your only option in certain cases. And with 2 above FEA is not so much about the loading it is more about the restraints. How the load gets into the structure has far less impact on peak response than how you hold it. The load application location is typically not where the peak response is located. As long as the load gets the bending moment about right at the restraint a reasonable estimate is likely, assuming the load paths are realistic modeled. Further, everything is an estimate no matter what method you use. You can't learn how to ski powder unless you get in it. The same applies to FEA, CFD or other methods. In the modern era your best option is figure it out and have a quality system that does not let a "bad" analysis out of the office. I do agree with your last point which tends to be a key thing in any analysis but you have to do one to see it..
Josh Barnett wrote:
Good afternoon all,
This is a two-part question...
First I have a profile I want to extrude in 2 ft sections, and then run an analysis, using 1 fixed and 1 roller fixture. So for example, extrude 2ft then run sim, extrude to 4ft, all the values stay the same, rerun simulation, extrude to 6ft, etc...
Seems simple enough, but I am getting non-consistent values. The Von Mises stress starts high, goes lower, then shoots up to a value which triggers the large displacement calculations.
You can create configurations increasing the lengths by 2 foot each time and copy the loads & fixtures from the old study to the new study. I'm pretty sure you can do a design study to vary the length in the model and monitor the displacement and stresses in the different lengths but that will take more than a license of Simulation Standard to do that. The vonMises stress looks reasonable from the loading and fixtures applied to the split lines, but if the the large displacement is activated you may be past yield or need to check that adequate fixtures are applied.
For the second part of your question, if you think of your deck panel like a beam you can get the moment of inertia of the area from the section properties of your model and calculate what equally distributed load would result in a bending stress equivalent to the yield stress for your material and then go from there. Hope that answers your questions. Congrats on the CSWP!