Discussion Forum

Hi

buld your geometry, set your materials, define your BC (boundary conditions such as fixed part, load part ...) Use a stationary solver with a continuation sweep mode, define a sweep parameter and multiply your force load by this parameter to sweep the load from 0 to max value, define your "probe" or average position displacement set up (average boundary entoity operator), then solve with sweep setting in your stationary solver case.

For the postprocessing, use the derived variables and build a table with aveop1(u) then v,w and add a value for "F0x*Param/aveop1(u)" respectively y and v, z ,w etc this should give you the stiffness. You can also turn on the non linear solvng to check the linearity of your load response.

Another way: run an eigenfrequency solver, extract the participation mass normalisation and get the approximate stiffness from the 2*pi*f = sqrt(k/m) relation or the equivalent 2*pi*f0 = sqrt(k0/m0) = sqrt)g_const/DZ0) where f0 is the fundamental frequency in one direction, k0 the equivalnet stiffness, m0 the equivalent participation mass, "g_const" the gravity constant of COMSOL 9.81... m/s^2 and DZ0 the maximum gravity sag in the Z (or X,Y) direction

--
Good luck
Ivar

Hello Ivar,

Thanks for sharing these tricks. However, I have few confusions and would like to share them with you.

1) When I use F=k*x method and vary the magnitude of 'F' what type of BC (for load) should I use? I tried using Edge load and Boundary load and the results tallied with analytical solutions. But, when I use Body load of same magnitude the numerical result do not match with theoretical predictions. This is something I am unable to comprehend.

2) I am unable to implement the your second trick. How do I extract 'Equivalent participation mass'? Under the Eigen Value solver node I used 'MASS MATRIX" as scaling of eigenvectors and used "mod1_u" as participation factor field. Now, if I use the value of "MPF_mod1.u" as equivalent participation mass, I get weird result. Can you explain what is this participation mass and how can I calculate it?

3) Would the above mentioned methods work if I work with some weird geometry? Say Hollow cylinder with fluid inside and one hollow face fixed?

-Sankha

Hello Ivar,

I am really stuck at this. It would be very kind of you to guide me through this.

Thanks,

Sankha

Hi

I mostly use boundary loads and try to avoid edge or point loads (I restrict my cases to "boundaries" for all dimensions hence I use onle the two highest dimenssions for defining physics, as far as possible.

You may vary your boundary load total force F by defining a Parameter "Param" and then define the BC Force as F0*Param, and define Param as a Stationary solver continuation sweep of range(0,0.1,1). Then compare linear and non-linear geoemtry solver cases (just as an example). In the Derived Variable section you can everage the displacement of the loaded boudary and get the stiffness as an "F0*Param/average(w)" (if the load is along "Z")

For the participation mass of an eigenvalue solver, check the doc, you ask for the specific normalisation in the lower solver sub-node "Eigen value solver" (Output - Scaling of eignvectors RMS, Max, Mass matrix...)

This all works for "solid". Now if you mix in a fluid you need to get the effect of the fluid to correctly act on the solid, and the other way, this is a non trivial case, because you use different math representation to analyse fluids (and flow) and solid (bending)


--
Good luck
Ivar