## Fluids

Topics relate to Fluent, CFX, Turbogrid and more

#### Temp. Divergence Cp = 0.000 J/kgK for Nitrogen [Floating Point]

• ai0013
Subscriber

Hello,

I am simulating an airflow initiated by a pressure discharge from a canister @ 21 bar to a vacuumed vessel at 0.4 bar absolute. At the end of the discharge, I have a uniform atmospheric pressure in all domains, but different temperatures in the reservoirs (as has been already shown in the experiments)

Fluent estimates 145 K in the canister, and from an isoentropic relationship (gamma = 1.4) this should be:

• Rob
Ansys Employee
Looking at the TUI messages the solver starts to struggle with the flow and then temperature diverges. Chances are the temperature hits a nonphysical value somewhere which then returns a zero Cp causing the failure. If the solver hadn't returned the zero value I suspect you'd see temperature limit warnings followed by divergence. Have a careful look at the flow field about 5-10 iterations before the crash (re-run from a saved data set) and see what's going on. nWhat are the particles for?n
• ai0013
Subscriber
Dear Thank you for your kind explanation. The particles are because I'm trying to re-create the 20L Siwek Sphere experiment (used for dust explosions). To understand the flow I started with pure gas simulation, with no physics other than K-eps turbulence and temperature equation. I patched the regions accordingly, and I observe that this produces a shockwave entering into the vacuumed sphere. In this case, the simulation successfully represents the pressure-time history for both reservoirs (I'm comparing with OF too)nMy problem comes when I inject the particles alongside the gas. My coal-volatile-mixture was defined using the coal-calculator, with properties of the mixture defined by;ndensity = ideal-gas & Cp = mixing law, I am using the pressure-based solver with default URF's (except for T = 0.6) , 2nd Order Discretization schemes (except MUSCL for Temperature) and fixed dt = 1e-05s. I still don't understand when my flow hits a non-physical temperature if the gas is perfectly OK (Max temp across the shockwave ~ 550 K). Besides, I read somewhere that it could be due to a bad mesh, but I I already refined it and increased the cell quality up to a min of 0.4. Any other suggestion would be much appreciated.
• Rob
Ansys Employee
Have a look at where the particles are. DPM is intended for a 10-14% max volume fraction, and I suspect your problem is down to having a lot of particle mass somewhere. Are you planning on burning the mixture? If not use inert particles. Note, multiphase and shocks are a very complex subject, and you may find the models don't pick everything up. n
• ai0013
Subscriber
Thank you Yes, eventually I'll need to burn the mixture. At this stage, I'm looking at the particle dispersion only. As suggested I already tried to use inert instead of combusting particles. I suspected I added all the physics (+reactions, +radiation, +DPM interaction) without any benefit, but when I turned off all the extra things it didn't help too much, as I had the same error.nIn fact, a few authors have used fluent to run the same simulation as I, so I guess I'm still doing something wrong. n
• Rob
Ansys Employee
Check mesh resolution, just because the cell quality is good doesn't mean the mesh is suitable. n
• scholar
Subscriber
nI am experiencing the same issue. I have three phases: mixture-air, carbon-solid, and silica-solid. Currently, I am not solving any reactions. I am solving mass, momentum, and energy transport. Due to Cp becoming zero for carbon-solid, I changed Cp to a constant value. The simulation is running fine.nBut when I change the carbon-solid to a mixture-phase (having only carbon-solid as species) and kept Cp as a constant value, the simulation is not running. I see that the temperature is falling to 1 K which is the least set in controls (see the attachment).nPlease let me know your suggestions on this.nn
• Rob
Ansys Employee
If you have a species-mixture with only one material it might be awkward numerically. We solve for n-1 species so if you only have one material I'm not sure how the logic will work out. Add a second species with a small volume fraction, and same properties as the carbon and see what happens. n
• scholar
Subscriber
Hello Rob,nI tried using another species in the mixture-coal. Still, the error persists. Since I kept Cp constant, the Cp value is >0. But the divergence is still there.nThe temperature of phase-2 (mixture-coal) is attached. Also when I change the operating conditions (T = 1023) so that the reactor temperature is 1073 K. But that is not reflected from contours. Should I patch the temperature for the entire domain as an input to the operating temperature of the reactor?nn
• scholar
Subscriber
n
• Rob
Ansys Employee
Operating temperature is for the Bousinesq approximation, click on Help in the panel and it'll explain or link to an explanation of the various fields. n