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.
volume quantity, mass variation
Posted 9 apr 2012, 07:31 GMT-4 Fluid & Heat, Porous Media Flow, Geometry 3 Replies
Please login with a confirmed email address before reporting spam
Hi everybody,
We have a simulation of darcy flow in a porous media:
1. The darcy simulation gives us as a result, for each node, the time variation of the piezometric head (imagine pressure). But this is a nodal quantity.
2. This variation of pressure is associated with a corresponding variation of water storage or mass. But this (mass variation) is, we believe, a quantity distributed by volume, not by nodes.
3. We need to decide which is the volume linked to each node, to calculate there the mentioned mass variation to use in other processes.
So the questions are:
1- How does Comsol atribute a volume to each node?
2- How can we have access to this distribution to perform calculations on it?
3- Would it be possible to use this "volume quantity" distribution as input material parameter for new simulations?
Thanks in advance,
Adriana
We have a simulation of darcy flow in a porous media:
1. The darcy simulation gives us as a result, for each node, the time variation of the piezometric head (imagine pressure). But this is a nodal quantity.
2. This variation of pressure is associated with a corresponding variation of water storage or mass. But this (mass variation) is, we believe, a quantity distributed by volume, not by nodes.
3. We need to decide which is the volume linked to each node, to calculate there the mentioned mass variation to use in other processes.
So the questions are:
1- How does Comsol atribute a volume to each node?
2- How can we have access to this distribution to perform calculations on it?
3- Would it be possible to use this "volume quantity" distribution as input material parameter for new simulations?
Thanks in advance,
Adriana
3 Replies Last Post 9 apr 2012, 13:06 GMT-4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
interesting question, but when I look at COMSOL my dependent variable i.e. "p" I could expect from "classical" FEM theory to be defined on nodes, is influenced by the model dependent variable discretization function and power used, AS WELL as from the postprocessing rendering averaging used. So for me the "densities" you mostly get as outcome are already averaged and meaned over the respective mesh volumes, mesh element nodes and "discretisation nodes"
But agree I havnt really found this clearly written ou in the doct, I would need to go through a simple example from one of my older FEM books and compare the results with COMSOL, something I havent really done so far, as the global results so far have always agreed, or been far more precise ;)
--
Good luck
Ivar
interesting question, but when I look at COMSOL my dependent variable i.e. "p" I could expect from "classical" FEM theory to be defined on nodes, is influenced by the model dependent variable discretization function and power used, AS WELL as from the postprocessing rendering averaging used. So for me the "densities" you mostly get as outcome are already averaged and meaned over the respective mesh volumes, mesh element nodes and "discretisation nodes"
But agree I havnt really found this clearly written ou in the doct, I would need to go through a simple example from one of my older FEM books and compare the results with COMSOL, something I havent really done so far, as the global results so far have always agreed, or been far more precise ;)
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hello Ivar,
thank you for your quick answer.
You seem to suggest that the node results from the FEM calculation are already averaged values.
If so, I don't understand very well the point.
Nevertheless, my main concern is to get the mass variation of fluid as a spatial distribution, kind of matrix.
Let us assume that you get approximate point values of the true solution. Then, let's take an element, for instance a triangle. Over this element of the mesh, you would have a polynomial approximation of the solution depending on the degree of the finite element (for instance, a quadratic approximation). You could use this to calculate the fluxes THROUGH each of the element edges (the three segments of the triangle) and then calculate, by difference the NET FLUX which is the amount of fluid that "dissapeared".
My question is if it is possible to calculate this automatically with COMSOL. I mean, if i can have access to a matrix with these values.
Best regards
Adriana
thank you for your quick answer.
You seem to suggest that the node results from the FEM calculation are already averaged values.
If so, I don't understand very well the point.
Nevertheless, my main concern is to get the mass variation of fluid as a spatial distribution, kind of matrix.
Let us assume that you get approximate point values of the true solution. Then, let's take an element, for instance a triangle. Over this element of the mesh, you would have a polynomial approximation of the solution depending on the degree of the finite element (for instance, a quadratic approximation). You could use this to calculate the fluxes THROUGH each of the element edges (the three segments of the triangle) and then calculate, by difference the NET FLUX which is the amount of fluid that "dissapeared".
My question is if it is possible to calculate this automatically with COMSOL. I mean, if i can have access to a matrix with these values.
Best regards
Adriana
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
that is what I understand COMSOL is doing anyhow, the only thing is that you do not access the mesh elements as such but the geometric entities (domains, boundaries ...) and you get the the resulting dependent variables displayed such that globally they map your domains and boundaries, to solve the underlaying equations.
Now what I do not fully understand is what you are searching by "mass variations" of your fluid. COMEOL obeys the physicsals laws it's based on, including mass conservation, ...
--
Good luck
Ivar
that is what I understand COMSOL is doing anyhow, the only thing is that you do not access the mesh elements as such but the geometric entities (domains, boundaries ...) and you get the the resulting dependent variables displayed such that globally they map your domains and boundaries, to solve the underlaying equations.
Now what I do not fully understand is what you are searching by "mass variations" of your fluid. COMEOL obeys the physicsals laws it's based on, including mass conservation, ...
--
Good luck
Ivar