There’s no trick to FEA, it’s just a simple trick! :)
Seriously, there are a few tips that I’ve picked up over the years and here is my top ten favourites.
1. Build the Model in Baby Steps Keep it simple to start with and add detail in steps as you go along. Simple for FEA Analysis could mean breaking up a larger model into smaller more manageable models. This often can be done with hand calculations or single component linear analysis. The margin of error increases a lot with model complexity and nonlinearity. So start simple, check, then add complexity in baby steps, checking each step as you go along. When you’re debugging the model you’ll be glad you did this!
2. Use the Right Hardware Buy a decent workstation with plenty of RAM. You can do it on a cheap laptop but it will be a painful experience and you will probably need to spend time thinking of ways to trim a few thousand elements from your model just to get it to run.
3. There's No Such Thing as the Perfect Model Don’t aim to build the perfect model. It will take a long time and it will never be perfect. That’s because the perfect model doesn’t exist. Instead focus on statements for example, ‘the stress is below yield and the margin of error due to mesh and boundary conditions is less than 10%.’
4. Use the Right Type of Element Choose elements wisely. Beams and shells are great for fabricated steel structures. Solid elements are ideal for bulky components. Avoid solid tetrahedral elements without a mid-side node if possible.
5. Save, Report, Archive You can't save your models enough! Save them frequently as you go along . When it’s time to debug the model, you’ll be glad you did! Reference model file names in your reports so that others can find them in the future. Files and folders get moved, archived and sometimes lost over time. The 100 FEA results you got right will soon be forgotten when only one of them is wrong and it may not be your fault. Good record keeping could save your bacon one day!
6. Only Help People Who Want To Be Helped Don’t offer advice unless you’re asked. It’s sometimes seen as interfering. I learnt this one the hard way!
7. Be Honest We all make mistakes, it happens! Be the first to own up to it and put it right and fast!
8. Be Pragmatic Not Dogmatic There’s more than one way to skin a cat! Use all of the resources at your disposal, product return data, test data, the person in the corner who’s worked there for 30 years. Ask around but don’t believe everything you hear.
9. Assume All FEA Models Are Wrong Slightly change the boundary conditions of your model and you'll get a different answer. Does it make any significant difference? That is the real question. The same applies to material properties and loading. Have you checked what the difference is and if it's important or not?
10. Keep Smiling Don't forget to have fun! This is me before I decided to become an engineer. :)