Hi!
We tried to combine the non-constant bodyforce with a constant bodyforce to add gravity to our model.
We did this by applying a fixed constraint to one face of a cylinder and then ramping up a "normal" bodyforce, thus getting a a hanging cylinder. We then ramped up the non-constant bodyforce while leaving the gravity constant. This new force is directed against the gravity, ergo pushing the cylinder back up.
Calculating the same model without gravity leads to a certain displacement. After adding the gravity we expected that the displacement relative to the state with full ramped up gravity but no non-constant bodyforce should be more or less the same as in the model without gravity. However, we found that the displacement increases significantly (factor ~2.5).
Since we have no explanation for this behavior we did some tests. The effect stays exactly the same with different elementsizes. So we tried different material models and did the exact same thing with an isotropic elastic material. Surprisingly we found that in this case the displacement decreases (up to a factor of ~0.5). Since the displacements caused through the gravity are not that big we believe neo-hookean and isotropic elastic material models should behave almost the same.
All these calculations were static calculations with a nearly incompressible (v=0.499) neo-hookean material with a Young's-Modulus of 10kPa and a density of 1 g/cm³.
We also did some simplified tests. We did the same thing with a cube with one face constraint in all three spatial dimensions. This time we used E =2000kPa and v=0.3 to get rid of the problems with nearly incompressible materials. We then did the same thing as before, but this time we used a more simple non-constant bodyforce (e.g. we set it to a constant). Although the effect this time was really small it is still there.
Do you have any suggestion where this behavior might come from? We are out of ideas at the moment.
Thanks for any input,
Sebastian
P.S.: This is the way we configured the forces:
<Globals>
<body_force type="non-const">
<z lc="1" data="exp(-12.5*(x^2+y^2))*0.007355*exp(-0.345*z)">-1000.0</z>
</body_force>
<body_force>
<z lc="2">981</z>
</body_force>
</Globals>
<LoadData>
<loadcurve id="1" type="linear" extend="constant">
<loadpoint>0.010,0.0</loadpoint>
<loadpoint>0.020,1.0</loadpoint>
</loadcurve>
<loadcurve id="2" type="linear" extend="constant">
<loadpoint>0.000,0.0</loadpoint>
<loadpoint>0.010,1.0</loadpoint>
</loadcurve>
</LoadData>
We tried to combine the non-constant bodyforce with a constant bodyforce to add gravity to our model.
We did this by applying a fixed constraint to one face of a cylinder and then ramping up a "normal" bodyforce, thus getting a a hanging cylinder. We then ramped up the non-constant bodyforce while leaving the gravity constant. This new force is directed against the gravity, ergo pushing the cylinder back up.
Calculating the same model without gravity leads to a certain displacement. After adding the gravity we expected that the displacement relative to the state with full ramped up gravity but no non-constant bodyforce should be more or less the same as in the model without gravity. However, we found that the displacement increases significantly (factor ~2.5).
Since we have no explanation for this behavior we did some tests. The effect stays exactly the same with different elementsizes. So we tried different material models and did the exact same thing with an isotropic elastic material. Surprisingly we found that in this case the displacement decreases (up to a factor of ~0.5). Since the displacements caused through the gravity are not that big we believe neo-hookean and isotropic elastic material models should behave almost the same.
All these calculations were static calculations with a nearly incompressible (v=0.499) neo-hookean material with a Young's-Modulus of 10kPa and a density of 1 g/cm³.
We also did some simplified tests. We did the same thing with a cube with one face constraint in all three spatial dimensions. This time we used E =2000kPa and v=0.3 to get rid of the problems with nearly incompressible materials. We then did the same thing as before, but this time we used a more simple non-constant bodyforce (e.g. we set it to a constant). Although the effect this time was really small it is still there.
Do you have any suggestion where this behavior might come from? We are out of ideas at the moment.
Thanks for any input,
Sebastian
P.S.: This is the way we configured the forces:
<Globals>
<body_force type="non-const">
<z lc="1" data="exp(-12.5*(x^2+y^2))*0.007355*exp(-0.345*z)">-1000.0</z>
</body_force>
<body_force>
<z lc="2">981</z>
</body_force>
</Globals>
<LoadData>
<loadcurve id="1" type="linear" extend="constant">
<loadpoint>0.010,0.0</loadpoint>
<loadpoint>0.020,1.0</loadpoint>
</loadcurve>
<loadcurve id="2" type="linear" extend="constant">
<loadpoint>0.000,0.0</loadpoint>
<loadpoint>0.010,1.0</loadpoint>
</loadcurve>
</LoadData>
Comment