Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
are the library materials parameters correct?
Posted Aug 19, 2009, 5:55 a.m. EDT 2 Replies
Please login with a confirmed email address before reporting spam
This is Comsol 3.5a on Linux 32.
First, when I choose from Materials Library Titanium elastic properties (an isotropic case), I get 40e9 Pa (which is very incorrect). But who cares much - I can change that, as long as I know that this is incorrect number.
Than I try some anisotropic materials. For instance quartz or silicon single crystals. I get the same values for them. For instance, the first upper left corner matrix element is 166 GPa - this can not be right either. All other elastic matrix coefficients remain the same, too.
And the last problem: it is assumed that elastic constants matrix is symmetric. That is right. It ought to be symmetric when a proper coordinate system is used. However, if I want, for instance, to compute stress on a sapphire with R-plane in (x.y) axis - I am stucked. I can not enter asymmetric matrix parameters into the dialog box!
And now the really last question: what is the indexing method really in that dialog box where elastic matrix parameters should be entered? I do not seem to be able to find any information about that.. OK, I know - 6 only parameters are need, but... what is the meaning of every matrix cell ? What is indexing ij?
First, when I choose from Materials Library Titanium elastic properties (an isotropic case), I get 40e9 Pa (which is very incorrect). But who cares much - I can change that, as long as I know that this is incorrect number.
Than I try some anisotropic materials. For instance quartz or silicon single crystals. I get the same values for them. For instance, the first upper left corner matrix element is 166 GPa - this can not be right either. All other elastic matrix coefficients remain the same, too.
And the last problem: it is assumed that elastic constants matrix is symmetric. That is right. It ought to be symmetric when a proper coordinate system is used. However, if I want, for instance, to compute stress on a sapphire with R-plane in (x.y) axis - I am stucked. I can not enter asymmetric matrix parameters into the dialog box!
And now the really last question: what is the indexing method really in that dialog box where elastic matrix parameters should be entered? I do not seem to be able to find any information about that.. OK, I know - 6 only parameters are need, but... what is the meaning of every matrix cell ? What is indexing ij?
2 Replies Last Post Aug 21, 2009, 3:22 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I have no problems with my material library, but there are, as usual a couple of elemets to always check:
first there are many Titanium's in there, you have all sorts of variants of the "same" material, it up to the user to check which one.
second many have their properties variable with a temperature or sometime other parameters i.e. magnetic saturation. Again it's up to the end user to be sure that these variables are correctly defined. In particular the temperature T is not always pre-defined in Comsol if you are not using a temeprature physics, that's why I always define by default a couple of variables such as T 20[degC] in my constants (another one I like is G0 1[lbf/lb] that defines the gravitational constant, the only way I have found to extract it from COMSOL even if I have to cheat by using imperial units, I'm definitively an SI guy)
To help when you select your material, check the "Hide undefined properties" and check that all your needed items are really there, very important, as you say for anisotropic material.
If you are using with anisotropic materials, know that by default COMSOL has mostly steel as baseline, so if you do not update these values (or if the material properties are not fully defined also as anisotropic) you will need to update them by hand.
Boldface items in the physical properties are from the library, that helps to see which ones are filled in.
now if you find an error in the material data, you should report it back to support@comsol, this might happen, and is important for all of us users
The final vord: model validation is essential (take a look about this on the web site of nafems or just hit on google "FEM Model Validation" (but there is quite a lot out there)
Good luck
Ivar
I have no problems with my material library, but there are, as usual a couple of elemets to always check:
first there are many Titanium's in there, you have all sorts of variants of the "same" material, it up to the user to check which one.
second many have their properties variable with a temperature or sometime other parameters i.e. magnetic saturation. Again it's up to the end user to be sure that these variables are correctly defined. In particular the temperature T is not always pre-defined in Comsol if you are not using a temeprature physics, that's why I always define by default a couple of variables such as T 20[degC] in my constants (another one I like is G0 1[lbf/lb] that defines the gravitational constant, the only way I have found to extract it from COMSOL even if I have to cheat by using imperial units, I'm definitively an SI guy)
To help when you select your material, check the "Hide undefined properties" and check that all your needed items are really there, very important, as you say for anisotropic material.
If you are using with anisotropic materials, know that by default COMSOL has mostly steel as baseline, so if you do not update these values (or if the material properties are not fully defined also as anisotropic) you will need to update them by hand.
Boldface items in the physical properties are from the library, that helps to see which ones are filled in.
now if you find an error in the material data, you should report it back to support@comsol, this might happen, and is important for all of us users
The final vord: model validation is essential (take a look about this on the web site of nafems or just hit on google "FEM Model Validation" (but there is quite a lot out there)
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi,
Thanks for pointing out the problem with the Ti property value.
The property data for Ti in the MEMS material library was originally taken from the book: Microsensors, MEMS and Smart Devices, by Julian W. Gardner et al. (Wiley 2001). Indeed, that book gives Young's modulus E = ~40GPa. It is likely that the book incorrectly refers to the original paper, or the original sources define some special condition for the property value. We have made a problem report of this, and we will check the original sources: Handbook of Chemistry and Physics (CRC Press, Inc.) and MacMillans Chemical and Physical Data (James and Lord, MacMillans Pressd Ltd, 1992).
The elasticity matrix of single crystall silicon is exactly what is given in The MEMS Handbook by CRC Press (ch 16.4: Silicon as a Substrate and Structural Material). But I have seen other values also that usually differ by some decimal, like D11 = 165.7GPa. For general computations, we thought that the rounded values should be accurate enough.
The indexing or ordering of the anisotropic D matrix is same as commonly used in structural mechanics. Stress (s) and strain (e) are related by the 6x6 elasticity matrix D by: s = D e, where s and e are 6 x 1 vectors: s = [sx sy sz sxy syz sxz]' and e = [ex ey ez exy eyz exz]' (' is transpose). This is described for example in the MEMS Module User's Guide chapter: Structural Mechanics Application Modes > Theory Background > Stress-Strain Relationship, and in similar place in the Structural Mechanics Module User's Guide.
Best regards,
Jarmo Ritola
COMSOL
Development, MEMS
COMSOL Finland
Thanks for pointing out the problem with the Ti property value.
The property data for Ti in the MEMS material library was originally taken from the book: Microsensors, MEMS and Smart Devices, by Julian W. Gardner et al. (Wiley 2001). Indeed, that book gives Young's modulus E = ~40GPa. It is likely that the book incorrectly refers to the original paper, or the original sources define some special condition for the property value. We have made a problem report of this, and we will check the original sources: Handbook of Chemistry and Physics (CRC Press, Inc.) and MacMillans Chemical and Physical Data (James and Lord, MacMillans Pressd Ltd, 1992).
The elasticity matrix of single crystall silicon is exactly what is given in The MEMS Handbook by CRC Press (ch 16.4: Silicon as a Substrate and Structural Material). But I have seen other values also that usually differ by some decimal, like D11 = 165.7GPa. For general computations, we thought that the rounded values should be accurate enough.
The indexing or ordering of the anisotropic D matrix is same as commonly used in structural mechanics. Stress (s) and strain (e) are related by the 6x6 elasticity matrix D by: s = D e, where s and e are 6 x 1 vectors: s = [sx sy sz sxy syz sxz]' and e = [ex ey ez exy eyz exz]' (' is transpose). This is described for example in the MEMS Module User's Guide chapter: Structural Mechanics Application Modes > Theory Background > Stress-Strain Relationship, and in similar place in the Structural Mechanics Module User's Guide.
Best regards,
Jarmo Ritola
COMSOL
Development, MEMS
COMSOL Finland