View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000833 | OpenFOAM | Bug | public | 2013-05-03 12:03 | 2013-05-06 21:40 |
| Reporter | Juho | Assigned To | henry | ||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | resolved | Resolution | fixed | ||
| Platform | Linux | OS | CentOS | OS Version | 6.2 |
| Summary | 0000833: compressibleTwoPhaseEulerFoam: fluidisedbed: Excessive temperature variations | ||||
| Description | When both phases of the compressibleTwoPhaseEulerFoam tutorial fluidisedbed are set to the same initial temperature (300K) with adiabatic walls, the results show temperature variations near volume fraction gradients. Especially in the gas (2) phase (~5.4K after 1 s of simulated time). The solid phase temperature variations are smaller, approx 1.3 K). The temperature variations are still present, if the gas phase equationOfState is switched from perfectGas to rhoConst (gas ~4.2, solids 1.1 @ 1 s). The issue is more clearly visible if the heat transfer coupling between the phases is disabled. Solid phase variations decrease to ~0.2K, but the gas phase variations increase to ~230-250K (@ 1 s) level! There's a mall difference depending on the choice of the equationOfState (perfectGas-rhoConst or rhoConst-rhoConst). If the compressibility corrections of the energy equations are also disabled, the temperature variations drop to ~4 K for gas and 0.4 K for the solids. | ||||
| Steps To Reproduce | Take the fluidisedbed tutorial and change the solids initial temperature to 300K. Run the simulation. The heat transfer coupling was disabled by commenting lines 37 and 59 in the EEqns.H. The compressibility correction was disabled by commenting the lines 26-33 and 48-55 in the EEqns.H. | ||||
| Tags | No tags attached. | ||||
|
|
> The issue is more clearly visible if the heat transfer coupling between the > phases is disabled. This is numerically unstable as it is not then possible to define the temperature of a phase for which the phase fraction is 0. What level of temperature variation would you expect from the pressure and mechanical energy changes? |
|
|
The pressure drop between the inlet and outlet is some 6000 Pa. If we assume isentropic expansion and no heat transfer between the phases, this should result in a temperature drop of some 5 K. The peak gas velocities are in the 2.5 m/s range, so I would assume the mechanical energy terms to be small. In all the simulations, the outlet temperatures of both phases match the inlet temperatures almost exactly. Even the decoupled simulations are within 1 K of the inlet temperature (alpha1 = 0 in the freeboard region near the outlet). However, the minimum and maximum temperatures occur very locally on the edges of the bubbles in the bed. The pressure difference between the locations of the minimum and maximum temperatures is about 600 Pa (gas velocity change ~0.15 m/s), which in decoupled isentropic expansion would result in a temperature drop off approx. 0.5 K. In the simulation, even with the damping effect of the interfacial heat transfer the drop is about 5 K. In addition, the magnitude and behavior of the temperature peaks is not significantly affected even if both phases are treated as incompressible (rhoConst equationOfState) In the bed where the minimum and maximum temperatures occur, alpha1 = 0.1...0.7. The decoupled simulations do have a few troublesome cells with low temperatures on the free surface, but they show how no obvious symptoms of numerical problems in the bed region. Without the regions near or above the free surface, the temperature variations of the gas phase are still in the 200K - 230K range. |
|
|
My understanding from you first post is that "gas phase are still in the 200K - 230K" occurs when you do not have any/sufficient coupling between the phases. For stability you will need to use sufficient inter-phase thermal coupling particularly in the case of very high phase-fraction gradients and phase-fractions which aproach zero. |
|
|
In the region where these 200-230K variations occur, the volume fraction is generally between 0.1-0.7. It is not close to zero. The temperature variations occur within one phase (i.e. not between the phases). (I can simulate a case without any near zero volume fraction regions next week to see if there is a difference.) Even if we ignore the demonstration simulations without the interfacial heat transfer, in my opinion the local temperature variations do seem excessive even with the unmodified solver. In addition, this behavior is not significantly affected by the choice of perfectGas or rhoConst as the equationOfState for the gas phase. |
|
|
Thanks for the detailed analysis of the problem. After further investigation I found that part of the pressure-work term due to interface motion was not handled correctly and I have updated the source terms for h and e accordingly. I have re-run the tests you performed and now with the two phases having the same initial temparture the variation due to pressure is as experted and there are no spurious variations around the interface. Please test the fis and report back if the behaviour is not as experted. Resolved by commit 85a28128e25caf9c072479fe4a21d60cecc7b9ad |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2013-05-03 12:03 | Juho | New Issue | |
| 2013-05-03 12:38 | henry | Note Added: 0002154 | |
| 2013-05-03 14:44 | Juho | Note Added: 0002155 | |
| 2013-05-03 14:57 | henry | Note Added: 0002156 | |
| 2013-05-03 15:12 | Juho | Note Added: 0002157 | |
| 2013-05-03 15:17 | Juho | Note Edited: 0002157 | |
| 2013-05-06 21:40 | henry | Note Added: 0002170 | |
| 2013-05-06 21:40 | henry | Status | new => resolved |
| 2013-05-06 21:40 | henry | Resolution | open => fixed |
| 2013-05-06 21:40 | henry | Assigned To | => henry |
