Convection-Diffusion with Heat Sink - Mesh Effect
Hello Again,
I try to model the temperature change (of water) through an insulated pipe in time. So I solve: https://i.postimg.cc/59nmWr8r/2019-0...-docx-Word.png The thermal resistance for the pre-insulated pipe is based on x: https://i.postimg.cc/HxjkcjPX/2019-0...yses-Gener.png The problem is that whenever I change the mesh size (&number) in space x while solving the convection diffusion equation (with heat sink) the overall thermal resistance Rt is changing so the water temperature profile through the pipe. For Example (According to steady-state calculations the plateau should be 68.07°C!): https://i.postimg.cc/CxBCDHzL/2019-0...s-Figure-1.png QUESTION: How to avoid the mesh size effect on the heat sink? Are my formulations true in principle? |
Quote:
It is by your definition depending on the mesh size... Are you sure that your problem has the resistence depending on the mesh size and not on a physical lenght width? |
Quote:
https://i.postimg.cc/4N0Q2FcT/2019-0...s-Figure-1.png |
I don't know the physics of your problem but I cannot understand this problem that shows this PDE with a forcing term depending on the mesh size. First of all, I would understand the meaning of the solution for dx->0, that is the exact solution you would compute in the continuous meaning.
|
Quote:
I could not find any resource explaining the convection-diffusion equation with heat sink. Could you suggest any? |
Convection-Dominated Flow | Finite Difference | Sink Term as Heat Loss
I found the mistake in my formulation:
There is a need to correct the heat sink term in the convection-diffusion equation, solving the temperature propagation through an insulated pipe segment in time (no thermal inertia considered). This way the value of c (check the formulation below) becomes constant whatever the spatial mesh size is! However, now I have trouble in my numerical results (oscillation, abnormal results). https://i.postimg.cc/tCzKhdc1/2019-0...-docx-Word.png Questions: 1) My results are oscillating when I remove the sink term. So can you please check the approximations? Is using FTCS with Upwind Treatment correct for this PDE (Check Figure1)? 2) With the sink term, the result is not realistic! Especially I want to ask you 'How to approximate T itself (in the sink part)?'. Is it true to have the approximation of T as Ti,n? Check Figure2! Thanks in advance. Figure1: https://i.postimg.cc/Ls6SwCcF/2019-0...s-Figure-1.png Figure2: https://i.postimg.cc/CL1jMSbR/2019-0...s-Figure-1.png |
The scheme that use the Forward Time and the Upwind discretization is denoted FTUS while the FTCS stands for centred discretization in space.
The FTUS is monotone, therefore you have a bug in your code. Is the transport velocity positive everywhere? |
Quote:
|
Quote:
If the FTUS solution (without the source term) shows numerical oscillations in the range of the numerical stability you have a bug in the code. What about the numerical condition at the outlet? |
Quote:
|
Quote:
However, you must assess that the solution is monotone in space, not in time at the outlet |
All times are GMT -4. The time now is 18:56. |