it is hard to get what exactly method is meant

some more details about "tried it in hook's laws to calculate the other moduli Ey and Ez" are necessary

Actually you are trying to validate your model with somewhat minimum information.

First, it is possible from the mathematical point of view!!!

Second, the main criteria from a set of various correct mathematical methods

(can be e.g. homogenization methods) is "equivalence of energies"

That means for example you have model No. 1 with full infomation (all possible moduli, geometry etc. etc.l) and model No. 2 with the reduced informations

Then you need to provide "a set of relevant numerical tests" and show that the energy is the same

A good and undestandable example is:

a beam consists of very complex composite material, property of which you hardly can model because of complex geometry of inclusion, many layers etc. etc.

But you want to model this with only one element!

Then you are generating RVE - representive volume element - this only one element, in which you are generating those possible props.

"a set of relevant numerical tests" is only a bending test with this RVE

now it is possible:

a) to validate the bending result by experiment using only the global bending energy for beam (as it would be in the beam theory)

b) to make the average result by several identical statistical computations with RVE and taking each time the same energy as for the beam : actually this method is exactly the homogenization method

Beware of "curve fitting" for another parameters like displacement, stresses at particular point.

it is hard to get what exactly method is meant

some more details about "tried it in hook's laws to calculate the other moduli Ey and Ez" are necessary

Actually you are trying to validate your model with somewhat minimum information.

First, it is possible from the mathematical point of view!!!

Second, the main criteria from a set of various correct mathematical methods

(can be e.g. homogenization methods) is "equivalence of energies"

That means for example you have model No. 1 with full infomation (all possible moduli, geometry etc. etc.l) and model No. 2 with the reduced informations

Then you need to provide "a set of relevant numerical tests" and show that the energy is the same

A good and undestandable example is:

a beam consists of very complex composite material, property of which you hardly can model because of complex geometry of inclusion, many layers etc. etc.

But you want to model this with only one element!

Then you are generating RVE - representive volume element - this only one element, in which you are generating those possible props.

"a set of relevant numerical tests" is only a bending test with this RVE

now it is possible:

a) to validate the bending result by experiment using only the global bending energy for beam (as it would be in the beam theory)

b) to make the average result by several identical statistical computations with RVE and taking each time the same energy as for the beam : actually this method is exactly the homogenization method

Beware of "curve fitting" for another parameters like displacement, stresses at particular point.

Hello Mr Ahmed

Just to clarify .The vibrational experimental result will also depends on the damping of your system or the material itself, in your case the turbine blade itself will have a damping. The amplitude you get from the vibration results are a combination of mass and stiffness matrix . So the results not only depends on the elastic property.

In either case to fit it to your solution instead of doing a trial and error you can set the target amplitude and get the values of either your elastic property or damping using the Surface Response DOE method in ANSYS WB. You can do such sampling design to reach your target amplitude and Frequency.

Use the experimental data for in plane moduli. The normal modulus Ez assume quite equal to Ey (matrix direction) and in first approx use the other elastic constants equal to in plane data. It is a good approx then you can adjust with experimental data. Why do you need 9 constants for blade simulation? You can use also shell simulation in this case so only 5 constants.

regards