I made a simulation of long blade of wind turbine 7 meters length. I cannot make a prototype to invalidate results of FEM of mode shapes and so on.
I have an idea of using small similitude made by rapid prototyping using mathematical relations between similitude and real blade so I can measure natural frequencies.
Use manual calculations. For example you can estimate deflections and stresses by means of the principle of structural mechanics. This can be a rapid and first approximation check of FEM results.
I suggest a validation based on experimental and theoretical data available in the scientific literature.Try to simulate some literature examples and compare the results with your FEM method.
Firstly, check these publications below, and than check their references to find other ones.
https://www.isma-isaac.be/past/conf/isma2010/proceedings/papers/isma2010_0627.pdf
http://orbit.dtu.dk/fedora/objects/orbit:90910/datastreams/file_7712483/content
http://orbit.dtu.dk/fedora/objects/orbit:88210/datastreams/file_7711587/content
http://scholarworks.umass.edu/cgi/viewcontent.cgi?article=1010&context=windenergy_report
I agree with Wederley, your best bet is to mine literature for published experimental data. Usually FEM packages come with validation data-sets that follow this pattern. Maybe your package contains some similar scale problems to your turbine blade?
Dear Ahmed Hesham,
You can validate the results obtained by your code with the benchmark problems available in the literature. Otherwise you can validate your code by considering the problems experimentally performed by researchers available in literature. And the best way to compare the results of developed code is by conducting experiments for the same.
I think the LS DYNA software gives best results for FEM
Also you can use Hyper works Or Comsol or ABACUS
Dear Ahmed
Suggest the validation approaches as suggested by many experts as above...
The success is, whether in non-linear or linear, static or dynamic problems of real life also depends on many assumptions, which gets reflected in terms of constraints and restraints. Each one problem is unique by itself and can not find an direct example... You need absorb the functionality of the product and synthesise... One of the example in my work on medical device, you will find a close correlation of 95% at various locations of an assembled product with multiple components... similarly for non-linear model... On modal, there was an example of frequency predicted and measured... best way is to do a simple "hammer Test"
"Validation" can mean different things to different people. Often you will see the term "V&V" which stands for Verification and Validation. In this setting there are two distinct pieces of testing of your code:
1. Verification: Knowing that you mathematical model is being solved correctly (i.e. that you don't have any code bugs).
2. Validation: Understanding whether your mathematical model, as solved by your code, simulates reality.
For #1 (Verification) you can use some of the suggestions from above (comparing to analytic / published solutions) for simple models... but in more complex cases I highly recommend looking at the Method of Manufactured Solutions (MMS). This is a decent introduction to the idea: http://prod.sandia.gov/techlib/access-control.cgi/2000/001444.pdf
For #2 there is simply no substitute for experiments. Either find some in the literature or buddy up to an experimentalist to figure out whether or not you can do some sort of reduced model that will approximate the full behavior you are simulating.
Usually having an error below 10% in all compared parameters makes the simulation more realistic.
- Badges
- Science topic