Discussion Forum

Posted: 1 decade ago Mar 1, 2011, 12:52 a.m. EST

Hi in V4 to add a study you right click the main "root" node (all on the top) and add a Study, then you can select different items. When you have several physics some solvers are not implemented for all so the do not directly appear, they are there but under Custom Studies ... but if you select them you should ensure that you uncheck the corresponding physics int he solver node, which do not accept the specific solver case. This is typically used in segregated solver set-up -- Good luck Ivar

Posted: 1 decade ago Mar 1, 2011, 5:18 a.m. EST

Hi Ivar! Thanks a lot. I checked according to your suggestion. But still couldn't find quasi-static option for the structure mechanics module. Actually my question is about evaluate the strain rate using d(solid.evol,t) as a source term in Darcy's law in poroelastic material. I noticed that in Version 4.0a, there is predefined module "poroelastisity", it works. But when I want to see the value d(poro.evol,t), it's always zero. It might doesn't matter if i used the default module (since it works although d(poro.evol,t)=0, maybe i didn't use the correct expression here), but If I want to construct poroelastic material using Darcy's law coupling with Structure mechanics myself, this strain rate term d(solid.evol,t) should appear as a source term in Darcy's law. Or my problem might be, is it correct to use d(poro.evol,t) as time differentiation of the strain in the predefined "poroelastisity"? Thanks a lot. BR, XLi

Posted: 1 decade ago Mar 1, 2011, 9:18 a.m. EST

Hi indeed I believe we do not fully understand each other, if you are looking for the time derivatives, indeed you should use the time dependent solver. Furthermore, I do not have the earth science modulus so I'm not familiar with Porosity nor Darcy modeling. My thoughts would rahter go for a variable (name it "myvar_") with a global equations and use the COMSOL calculated time derivative "myvar_t" In any case get hands on V4.1 the 4.0 was an early version it has been greatly improved since, speak with your local rep. -- Good luck Ivar

Posted: 1 decade ago Mar 1, 2011, 11:06 a.m. EST

Hi Ivar! Thanks for your reply. The reason why I didn't use version 4.1 is because I have shell elements in my model, and FE mesh was imported. Unfortunately Shell physics interface cannot be used for an imported mesh in 4.1. I asked Comsol support, this is what they told me. But version 4.0a works for imported mesh for shell. XLi

Posted: 1 decade ago Mar 1, 2011, 1:44 p.m. EST

Hi Ivar, I think I am close to solve the problem now. Say that we have a structure mechanics module using time-dependent analysis, In Comsol version 3.5a, if want to use volumetric strain rate, it can be expressed as d(smsld_evol,t), where smsld_evol = ux+vy+wz, (u,v,w are x, y , z displacement respectively, ux = du/dx). However, this expression doesn't work in Comsol version 4.0a any more. d(solid.evol,t) this is always zero when evaluate it in the postprocessing. Just found that directly add a t afterwards is the time derivative, and solid.evol = uX+vY+wZ, Thus, what I want to use d(solid.evol, t) should be equal to (uXt+vYt+wZt). This should be correct? Thanks for help. /XLi

Posted: 1 decade ago Mar 1, 2011, 3:46 p.m. EST

Hi indeed that could be it, as the derivative of a sum is the sum of the derivatives, no ? One thing indeed too, in 3.5, for structural, the deformed frame was off and you had only one frame, in V4 the spatial frame is driven by the displacemnts ad you have the lower case x,y,z and upper case X,Y,Z variables for the spatial, respective material frame. Its important to define from which frame you derive, so indeed many lower case ux,uy,uz in 3.5 become uX,uY,uZ in v4 to state you are deriving w.r.t the material frame. This is easily overlooked. You should also check the equations, turn the Preferences "show equations" on Now finally you should be able to solve it in v4.1 too then -- Good luck Ivar

Posted: 1 decade ago Mar 1, 2011, 4:06 p.m. EST

"indeed that could be it, as the derivative of a sum is the sum of the derivatives, no ?" ==== I think so too. ==== Thanks a lot for reminding me the different frames for material and spatial. BR, XLi

Posted: 1 decade ago Mar 2, 2011, 7:12 p.m. EST

Might be useful too, here is what i got from the Comsol support. +++++++++++++++++++++ The Poroelasticity physics interface is defined on the material frame. This means that you can not differentiate with respect to the time variable t. If you try differentiating other variables, for example u, v and w, you will see the same behavior: The expression evaluates to 0. Instead, you can differentiate with respect to the time variable TIME: d(poro.evol,TIME). This is unfortunately not that intuitive. I have asked the developers to look into the possibility of differentiating with respect to t also for the material frame. +++++++++++++++++++++++ Thanks. XLi

Posted: 1 decade ago Mar 3, 2011, 12:58 a.m. EST

Hi Indeed, I had forgotten about that one ;) well you cannot change it, it must be TIME because time t and the material frame are fixed. TIME is the moving reference, check carefully the doc. It's important to understand the frames and their implications. This is well known by most fluid experts, but never used by structural experts, although it's the same physics we are studying. I agree it's not intuitive at first, but once understood it becomes natural (still easily forgotten ;) -- Good luck Ivar

Posted: 1 decade ago Mar 4, 2011, 12:07 p.m. EST

[QUOTE] Hi! Click on the module, "Advanced settings - Equation form (Automatic, stationary, time dependent ......)" I couldn't find Quasi-static analysis in Structure mechanics module in version 4.0a, it exists in version 3.5a. === Does this time-dependent by default same as quasi-static? === In some predefined module, such as poroelastisity, thermal stress module, where structure mechanics module is included, there is "quasi-static" option in the "Elastic transient behavior". Thanks. /X Li [/QUOTE] Finally I found the answer to my question, and I did make a big mistake here. === Does this time-dependent by default same as quasi-static? === The answer to my question is: NO. The time-dependent is dynamic analysis, check the equation in Comsol V4.1 (it's more clearly shown than version 4.0a), there is intertia part included in the equation. Quasi-static is actually STATIC equation. I should switch to Version 4.1a if the shell interface works for imported FE, it is really much better and clear especially the equation form. XLi

Posted: 1 decade ago Mar 6, 2011, 7:02 a.m. EST

Now I have a new problem regarding different "time frame". To use structure mechanics strain rate, I should use d(solid.evol,TIME). But if I want to add a time-dependent water source for Darcy's law, then I should use 't' instead? For example, for the coupling to model poroelastic material, fluid source in the Darcy's law interface should be Qs = sourceConstant1*t - d(solid.evol, TIME). Otherwise, if I use the same TIME, Qs = sourceConstant1*TIME - d(solid.evol, TIME). It doesn't work. Error is something like couldn't recognize 'TIME'. What is the difference between "t" and "TIME"? Rather confused. Thanks. BR, XLi