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.
in situ stress conditions
Posted 2011年1月21日 GMT-5 09:09 Porous Media Flow, Structural Mechanics Version 4.1, Version 5.0 4 Replies
Please login with a confirmed email address before reporting spam
Dear all,
in a poroelastic / structure mechanics model, that only consists of a material block (x=1000, y=1000, z=3000 meters, where z=direction of gravity vector), a stationary solution will always contain some initial displacements especially along the gravity axis (z).
The lateral boundaries are set on "roller", the bottom one to "fix constraint" and the surface one is "free".
The deformations happen only due to gravitational effects of its own weight until the model reaches an equilibrium.
Is there a way to tell Comsol, that the defined initial conditions already reflect an equilibrium and that there should be NO displacements at all in the solution?
One approach was:
First create a stationary solution, which contains this strange initial displacements.
Using a second physics module in a second study with exactly the same definitions as the first one but defining -u, -v and -w in the initial displacement field (where u, v and w are the components of the displacement field of the first stationary solution) should avoid any initial displacements in the second study.
This does not work. The deformations still appear in the second study.
Does anyone have an idea, why this work around fails?
Anyone has another idea or work around?
Thanks a lot and have a nice weekend everyone,
Juergen
in a poroelastic / structure mechanics model, that only consists of a material block (x=1000, y=1000, z=3000 meters, where z=direction of gravity vector), a stationary solution will always contain some initial displacements especially along the gravity axis (z).
The lateral boundaries are set on "roller", the bottom one to "fix constraint" and the surface one is "free".
The deformations happen only due to gravitational effects of its own weight until the model reaches an equilibrium.
Is there a way to tell Comsol, that the defined initial conditions already reflect an equilibrium and that there should be NO displacements at all in the solution?
One approach was:
First create a stationary solution, which contains this strange initial displacements.
Using a second physics module in a second study with exactly the same definitions as the first one but defining -u, -v and -w in the initial displacement field (where u, v and w are the components of the displacement field of the first stationary solution) should avoid any initial displacements in the second study.
This does not work. The deformations still appear in the second study.
Does anyone have an idea, why this work around fails?
Anyone has another idea or work around?
Thanks a lot and have a nice weekend everyone,
Juergen
4 Replies Last Post 2015年7月23日 GMT-4 10:19
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I would say it's slightly difficult for us to reproduce your model, is it possible for you to make a/the simple model and publish it here, to allow us to take a look ;)
Not necesarily all details of your model, just what puzzles you
I believe we will then better understand whats happening
--
Good luck
Ivar
I would say it's slightly difficult for us to reproduce your model, is it possible for you to make a/the simple model and publish it here, to allow us to take a look ;)
Not necesarily all details of your model, just what puzzles you
I believe we will then better understand whats happening
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Ivar,
ok, so here is an example model.
There is a block with certain hydraulic head or pressure conditions, which already should represent an equilibrium state.
Which is sad, because there is no way to tell Comsol that the model should already be in equilibrium from the beginning.
Comsol computes deformations due to the weight of the whole block until an equilibrium is reached.
There seems to be no way to tell Comsol anything about the initial in situ stress conditions.
So, running a stationary solver ("initial stationary solution") on poroelasticity (poro) creates z-displacements (w) of about 2.5 meters.
I ran a second stationary solver on poroelasticity 2 (poro 2). Poro 2 is exactly like poro. It only has a defined displacement field in its initial values, which substracts the u, v and w component of the displacement of the first stationary solution.
This should delete any initial displacements and in the end the displacement of the 2nd stationary solution should be zero. But they are still about 2.5 meters.
This idea is from comsol support and I think it is very promising. I just seem to make a stupid mistake somewhere.
The initial displacements are gravity driven (gravity acts along z axis).
Thanks a lot for any ideas on this.
Have a nice weekend,
Juergen
ok, so here is an example model.
There is a block with certain hydraulic head or pressure conditions, which already should represent an equilibrium state.
Which is sad, because there is no way to tell Comsol that the model should already be in equilibrium from the beginning.
Comsol computes deformations due to the weight of the whole block until an equilibrium is reached.
There seems to be no way to tell Comsol anything about the initial in situ stress conditions.
So, running a stationary solver ("initial stationary solution") on poroelasticity (poro) creates z-displacements (w) of about 2.5 meters.
I ran a second stationary solver on poroelasticity 2 (poro 2). Poro 2 is exactly like poro. It only has a defined displacement field in its initial values, which substracts the u, v and w component of the displacement of the first stationary solution.
This should delete any initial displacements and in the end the displacement of the 2nd stationary solution should be zero. But they are still about 2.5 meters.
This idea is from comsol support and I think it is very promising. I just seem to make a stupid mistake somewhere.
The initial displacements are gravity driven (gravity acts along z axis).
Thanks a lot for any ideas on this.
Have a nice weekend,
Juergen
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I hope you find out finally, as I do not have the geo module and it's not my most familiar domain, it's difficult for me to test; still, from the methodology, I'm convinced that (at least in >=4.1) there are a way to get a stable condition with one or two stationary solvers (or a segregated one) and then start a transient from there. I do not see why it should be different from classical structural, or ACDC, or ...
COMSOL proposes "0" as default starting initial values, but for moast true "engineering" cases this is far from true, so one needs first to set the system in static equilibrium and this is not always a trivial case with a linear distribution of some parameters/variables
--
Good luck
Ivar
I hope you find out finally, as I do not have the geo module and it's not my most familiar domain, it's difficult for me to test; still, from the methodology, I'm convinced that (at least in >=4.1) there are a way to get a stable condition with one or two stationary solvers (or a segregated one) and then start a transient from there. I do not see why it should be different from classical structural, or ACDC, or ...
COMSOL proposes "0" as default starting initial values, but for moast true "engineering" cases this is far from true, so one needs first to set the system in static equilibrium and this is not always a trivial case with a linear distribution of some parameters/variables
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Hello,
I have the same problem in my poroelastic models (currently using COMSOL 5.0). Did you ever find a working solution for this?
Thanks a lot
Karen
I have the same problem in my poroelastic models (currently using COMSOL 5.0). Did you ever find a working solution for this?
Thanks a lot
Karen