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.

Heat Transfer in Soils

Hi,
I am a comsol user and I want to have a 2d axisymmetric model. In my model, I have a soil sample that I put a steel bar in the soil. I want to heat the steel bar with a constant temperature 500(k) and then wait for 6 hours to see that how the heat will be conducted in the soil. The problem that I have is that after simulation, I do not know why I could not have good results. The heat will just be conducted at the top of the model, and the rest of the model will remain unaffected. You can see my model in attached file.So if it possible, I want to know that what do I need to do to solve this problem? Where is my mistake?

Regards,
Al


21 Replies Last Post Apr 1, 2017, 11:30 AM EDT
Nagi Elabbasi Certified Consultant
Posted: 5 years ago Nov 28, 2012, 11:35 PM EST
Based on the figure you attached it seems that you are only applying the 500K temperature at the top surface and not along the full bar.

Nagi Elabbasi
Veryst Engineering
Based on the figure you attached it seems that you are only applying the 500K temperature at the top surface and not along the full bar. Nagi Elabbasi Veryst Engineering

Posted: 5 years ago Dec 4, 2012, 6:22 PM EST
Dear Nagi;
First I have to appreciate you because of your kind consideration.I have another problem. I am modeling a pipe in the soil in 3d. When I want to solve the problem, I will see this error:

Failed to find consistent initial values.
Undefined value found.
NaN or Inf found when solving linear system using PARDISO.
Last time step is not converged.



What is my mistake?
In attached you can see the picture of model an error.

Best Regards,
Al
Dear Nagi; First I have to appreciate you because of your kind consideration.I have another problem. I am modeling a pipe in the soil in 3d. When I want to solve the problem, I will see this error: Failed to find consistent initial values. Undefined value found. NaN or Inf found when solving linear system using PARDISO. Last time step is not converged. What is my mistake? In attached you can see the picture of model an error. Best Regards, Al


Posted: 5 years ago Dec 5, 2012, 1:25 AM EST
Hi

often this comes from missing IC or BC values (particularly in multiphysics cases) as you need to be sure that ALL dependent variables have enough BC and coherent IC conditions to give an unique solution to the model.

But you have also sometimes, in time dependent diffusion models, some incoherent IC conditions, typically all default "0" default IC values and then in on place a abrupt temperature change, or velocity change or concentration change. If you do not have a fine enough mesh in this region the solver might crash due to a typical numerical undersampling case.
Often its worth to solve for the initial conditions, and look at the results, see if there are very high gradients, check if one cannot add a little to the BC or IC, or ramp them up with a parametric sweep

--
Good luck
Ivar
Hi often this comes from missing IC or BC values (particularly in multiphysics cases) as you need to be sure that ALL dependent variables have enough BC and coherent IC conditions to give an unique solution to the model. But you have also sometimes, in time dependent diffusion models, some incoherent IC conditions, typically all default "0" default IC values and then in on place a abrupt temperature change, or velocity change or concentration change. If you do not have a fine enough mesh in this region the solver might crash due to a typical numerical undersampling case. Often its worth to solve for the initial conditions, and look at the results, see if there are very high gradients, check if one cannot add a little to the BC or IC, or ramp them up with a parametric sweep -- Good luck Ivar

Posted: 5 years ago Dec 5, 2012, 8:21 PM EST
Hi,
Thank for your responses. I want to model a pipe that is located under the ground in the soil. The temperature of the water in the pipe is higher than the temperature of the soil around the pipe. So when the water flows in the pipe, it should decrease the temperature of the soil and the water become cooler at the outlet of the pipe. Also the soil around the pipe should become hotter. So at the same time I want to simulate flow of water in the pipe, heat that transferring by flow in the pipe and heat that transferring from the pipe to the soil. Now if it possible for you I want to ask you that do you have any model similar to my model or any simulation that could help me progress my modeling or not?

Your kind consideration of these matters are highly appreciated,
Best Regards,
Nasirian
Hi, Thank for your responses. I want to model a pipe that is located under the ground in the soil. The temperature of the water in the pipe is higher than the temperature of the soil around the pipe. So when the water flows in the pipe, it should decrease the temperature of the soil and the water become cooler at the outlet of the pipe. Also the soil around the pipe should become hotter. So at the same time I want to simulate flow of water in the pipe, heat that transferring by flow in the pipe and heat that transferring from the pipe to the soil. Now if it possible for you I want to ask you that do you have any model similar to my model or any simulation that could help me progress my modeling or not? Your kind consideration of these matters are highly appreciated, Best Regards, Nasirian

Posted: 5 years ago Dec 6, 2012, 1:06 AM EST
Hi

if you have the required modules, I would suggest to use the conjugated heat flow physics, and model your fluid (if possible laminar) the pipe and the soil. Now depending on which soil model (I do not have the geo modules) you might use a specific physics for the HT in the soil, or the porous media or something else. But I would start with a simple NITF and adapt the soil material parameters, then later once, this works consider, more complex physics.

You have also the pipe flow module for long pipes where you do not want to study the details of the flow but want the global exchange possibilities. Look at the model library and the model gallery, there are examples close therein

--
Good luck
Ivar
Hi if you have the required modules, I would suggest to use the conjugated heat flow physics, and model your fluid (if possible laminar) the pipe and the soil. Now depending on which soil model (I do not have the geo modules) you might use a specific physics for the HT in the soil, or the porous media or something else. But I would start with a simple NITF and adapt the soil material parameters, then later once, this works consider, more complex physics. You have also the pipe flow module for long pipes where you do not want to study the details of the flow but want the global exchange possibilities. Look at the model library and the model gallery, there are examples close therein -- Good luck Ivar

Posted: 5 years ago Dec 10, 2012, 2:46 PM EST
Hi,

During my simulation, when I want to calculate my model (pipe in a soil, thermal heat pump) I will see this error :


Failed to evaluate variable Jacobian.
- Variable: v0
- Geometry: 1
- Point: 6


what is the problem of my simulations?

Best Regards,
Alex
Hi, During my simulation, when I want to calculate my model (pipe in a soil, thermal heat pump) I will see this error : Failed to evaluate variable Jacobian. - Variable: v0 - Geometry: 1 - Point: 6 what is the problem of my simulations? Best Regards, Alex

Posted: 5 years ago Dec 10, 2012, 4:39 PM EST
Hi

probably your v0 is not well defined, or it cannot be derived once, the Jacobian uses probably dv0/dx somewhere. if v0 is defined with Boolean conditions, try to replace them by step or rect functions that can be derived.

or you need to study the nojac() operator in the doc



--
Good luck
Ivar
Hi probably your v0 is not well defined, or it cannot be derived once, the Jacobian uses probably dv0/dx somewhere. if v0 is defined with Boolean conditions, try to replace them by step or rect functions that can be derived. or you need to study the nojac() operator in the doc -- Good luck Ivar

Posted: 5 years ago Dec 10, 2012, 7:26 PM EST
Dear Ivar;
According to the picture that I attached, where I can define V0? I did not change any initial value or assumption in the model.

Regards,
Alex
Dear Ivar; According to the picture that I attached, where I can define V0? I did not change any initial value or assumption in the model. Regards, Alex

Posted: 5 years ago Dec 11, 2012, 2:09 AM EST
Hi

you image just tells us that there is something incomplete in the IC's, unfortunately one cannot say more from the screen, any other COMSOl user must dig into your model to understand it and then perhaps he might come with some suggestions.

I agree there are really many tabs and nodes to handle in COMSOL, and the error messages are rather short, we all have to live with that. Is it a weakness of the tool, or of the mkethodology ? I'm not sure, as I have no real suggestion I can give to "support" I have given up complaining, also after some years using COMSOL one get the tricks how to search.

One thing: run for the initial conditions before running a full solver, and plot these out, and ask yourself are these coherent ?, and for each depednet variables are all fully constrained ? And for physics with dependent fields, are the gauge values set (if required) ?

--
Good luck
Ivar
Hi you image just tells us that there is something incomplete in the IC's, unfortunately one cannot say more from the screen, any other COMSOl user must dig into your model to understand it and then perhaps he might come with some suggestions. I agree there are really many tabs and nodes to handle in COMSOL, and the error messages are rather short, we all have to live with that. Is it a weakness of the tool, or of the mkethodology ? I'm not sure, as I have no real suggestion I can give to "support" I have given up complaining, also after some years using COMSOL one get the tricks how to search. One thing: run for the initial conditions before running a full solver, and plot these out, and ask yourself are these coherent ?, and for each depednet variables are all fully constrained ? And for physics with dependent fields, are the gauge values set (if required) ? -- Good luck Ivar

Posted: 5 years ago Dec 12, 2012, 1:03 PM EST
Dear Ivar;

I have two questions: First, I want to know that can I simulate a pipe flow that the temperature of my inlet flow would be a time dependent temperature or not? for example can I use this formulation for the temperature of my inlet : T(i)=T(0)*SIN(w*t) ?
Secondly, I want to know that for simulating the isothermal pipe flow in the soil to have a results for heat transferring in the soil, with modulus of the software do I need to buy?

Regards,
Al Nasirian
Dear Ivar; I have two questions: First, I want to know that can I simulate a pipe flow that the temperature of my inlet flow would be a time dependent temperature or not? for example can I use this formulation for the temperature of my inlet : T(i)=T(0)*SIN(w*t) ? Secondly, I want to know that for simulating the isothermal pipe flow in the soil to have a results for heat transferring in the soil, with modulus of the software do I need to buy? Regards, Al Nasirian

Posted: 5 years ago Dec 12, 2012, 2:36 PM EST
Hi

I'm nt geoephysicist and I do no have those modules, the best is to ask your local COMSOL rep, he can give you the best examples.

From my understanding you need the pipe flow, and probably geomechanics, if not also solid and / or HT

--
Good luck
Ivar
Hi I'm nt geoephysicist and I do no have those modules, the best is to ask your local COMSOL rep, he can give you the best examples. From my understanding you need the pipe flow, and probably geomechanics, if not also solid and / or HT -- Good luck Ivar

Posted: 5 years ago Feb 15, 2013, 4:35 PM EST
Dear Ivar;
I have 2 physics for my model, isothermal heat flow and heat transfer in solids. the hot water is coming through a pipe and the pipe is located underground. how do i can define the input temperature of my heat transfer in solids, based on the temperature of the water in the pipe? for example, in Cooling of an Injection Mold example, In the T-in edit field, type T_inlet or In the T2 edit field, type T_init_mold. how do i have to define my initial temperature for the software?

best regatds
Dear Ivar; I have 2 physics for my model, isothermal heat flow and heat transfer in solids. the hot water is coming through a pipe and the pipe is located underground. how do i can define the input temperature of my heat transfer in solids, based on the temperature of the water in the pipe? for example, in Cooling of an Injection Mold example, In the T-in edit field, type T_inlet or In the T2 edit field, type T_init_mold. how do i have to define my initial temperature for the software? best regatds

Posted: 5 years ago Feb 17, 2013, 11:46 AM EST
Hi

But are you shure you need 2 physics each solving for T ? its much easier to use i.e. NITF and combine the fluid sde, the "solid" side and have only one T variable. Else you need to define manually the continuation conditions on all relevant boundaries, possible, but trickier than using a combined solid+fluid HT physics module

--
Good luck
Ivar
Hi But are you shure you need 2 physics each solving for T ? its much easier to use i.e. NITF and combine the fluid sde, the "solid" side and have only one T variable. Else you need to define manually the continuation conditions on all relevant boundaries, possible, but trickier than using a combined solid+fluid HT physics module -- Good luck Ivar

Posted: 5 years ago Feb 17, 2013, 9:16 PM EST
Hi,
where can I find an example of NITF that combine the fluid side, the "solid" side and have only one T variable?

regards
Hi, where can I find an example of NITF that combine the fluid side, the "solid" side and have only one T variable? regards

Posted: 5 years ago Feb 18, 2013, 1:03 AM EST
Hi

NITF is the tag for Conjugated Heat flow, it comes with several modules (not sure exactly which have to check the COMSOL main web site on the modules) There are several examples in the model library, I can sort them with the search keyword "NITF"

--
Good luck
Ivar
Hi NITF is the tag for Conjugated Heat flow, it comes with several modules (not sure exactly which have to check the COMSOL main web site on the modules) There are several examples in the model library, I can sort them with the search keyword "NITF" -- Good luck Ivar

Posted: 5 years ago Feb 18, 2013, 11:43 AM EST
Dear Ivar,
In attached you can find my model. The problem is that I do not know how to apply the temperature to the water that i want to pump through the pipe. the temperature is only affects the inlet not the whole water in the pipe.
Dear Ivar, In attached you can find my model. The problem is that I do not know how to apply the temperature to the water that i want to pump through the pipe. the temperature is only affects the inlet not the whole water in the pipe.


Posted: 5 years ago Feb 20, 2013, 5:03 PM EST
Hi

your mesh is far too coarse for the fluid domain, and you should use some initial velocity and pressure drops more realistic than all "0".
You should also round off the sharp edges at the fluid pipe angles (if possible) as you get singularities and potentially vortices at these sharp corners hence further solver issues.

Then if you are not really interested in the flow in detail, the "pipe flow" module is better suited for long piping with different exchanges,such as HT, it solves far easier and far quicker, for far less RAM, and is mostly fully "correct" and highly representative

--
Good luck
Ivar
Hi your mesh is far too coarse for the fluid domain, and you should use some initial velocity and pressure drops more realistic than all "0". You should also round off the sharp edges at the fluid pipe angles (if possible) as you get singularities and potentially vortices at these sharp corners hence further solver issues. Then if you are not really interested in the flow in detail, the "pipe flow" module is better suited for long piping with different exchanges,such as HT, it solves far easier and far quicker, for far less RAM, and is mostly fully "correct" and highly representative -- Good luck Ivar

Posted: 5 years ago Feb 22, 2013, 7:19 PM EST
Dear Ivar;
Thank for your kind consideration in advance. I have a probleml. I have a phase change martial and time dependent temperature . the time is SIN formulation. For example for water and ice, the transition temperature is 0 degC. When i heat my model from T=-5 to T=+5, my ice will become water and i can see the phase change procedure. but when my temperature oming from +5 to -5, i do not see the phase change procedure. so what do i have to do? in attached, you can find the picture from results for 2 points in the model.

sincerely yours,
Ali Nasirian
Dear Ivar; Thank for your kind consideration in advance. I have a probleml. I have a phase change martial and time dependent temperature . the time is SIN formulation. For example for water and ice, the transition temperature is 0 degC. When i heat my model from T=-5 to T=+5, my ice will become water and i can see the phase change procedure. but when my temperature oming from +5 to -5, i do not see the phase change procedure. so what do i have to do? in attached, you can find the picture from results for 2 points in the model. sincerely yours, Ali Nasirian


Posted: 5 years ago Feb 23, 2013, 9:04 AM EST
Hi

phase change is a delicate physics,as you have large discontinuities due ti the latent heat you need to know locally if you are in one or the other state, and you must know if you go up or down the temperature scale

Now you do not tell us much how you have programmed your phase chage model

--
Good luck
Ivar
Hi phase change is a delicate physics,as you have large discontinuities due ti the latent heat you need to know locally if you are in one or the other state, and you must know if you go up or down the temperature scale Now you do not tell us much how you have programmed your phase chage model -- Good luck Ivar

Posted: 5 years ago Feb 24, 2013, 1:19 AM EST
dear Ivar;
in attached you can find my model. I know that it is important to know that we go up or down the temperature scale, but i do not know how to define both for comsol. because i have a Sinusoidal time dependent temperature, sometimes i go up and sometimes down.

regards,
Ali
dear Ivar; in attached you can find my model. I know that it is important to know that we go up or down the temperature scale, but i do not know how to define both for comsol. because i have a Sinusoidal time dependent temperature, sometimes i go up and sometimes down. regards, Ali


Posted: 7 months ago Apr 1, 2017, 11:30 AM EDT
Dear ivar
kindly, can you have a model for the effect of heat transfer in collapsible soil using steel bar heating?
or any model close ?

best regard
bilal

Dear ivar kindly, can you have a model for the effect of heat transfer in collapsible soil using steel bar heating? or any model close ? best regard bilal

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.