Hi, I am using Fluent for hydraulics valves simulations. I would like to correct the Fluent-definition for massflow. Indeed it is defined as: dm/dt=rho*v*A. But in my case there is a correction coefficient which appears.So massflow should be in my case: dm/dt=K*rho*v*A, where K is a known coefficient. How am I suppose to do this in Fluent?

Else I have another question. There is always cavitation in my computed flows, ie I got some domains with negative absolute pressure. But as I have to compute piston force, it strongly influences the results because of the Force definition: F= -integral(p*dA). So I would like to correct those negative pressure in my defined surface,with giving them zero value. How can I do that? Thanks for your help!

Are you sure that mass flow is calculated with the formula: dm/dt = ro*V*A ???

You say "Indeed" it is defined this way, but I don't remember this equation discussed in the fluent manual. The continuity equation is used but not in that form.

If you use the viscous models (laminar or any turbulent model), the Navier-Stokes equations are used in conjunction with the continuity equation, and the fluid friction (the reason why you have the K coefficient) is accounted for.

What viscous model are you using? The K coefficient is the factor for reduced mass flow due to friction, correct?

If you have negative absolute pressure, that is not cavitation, that is an error in the solution. Theoretically the lowest absolute pressure is 0 Pa, correct? Be sure you are not looking at gage pressure.

Cavitation, as I know it, is bubbles or pockets of fluid with different properties, and these pockets impact the compressor and cause damage. For example, air bubbles in water, that hit the blades of a turbomachine. Is my understanding of cavitation incomplete?

You should try to rework your solution to obtain reasonable pressures. I had the same problem, and that problem could in itself have such limitations in CFD. The more complicated the problem, the more error you can expect.

Hi, Mass flow rate is defined as follow: dm/dt=rho*v*A (surface integration panel) I am using standard k-epsilon model. The K coefficient, is a correction due to the friction, you re right! Regarding the negative absolute pressure, it comes from cavitation: with hydraulic oil I reach max velocity about 300m/s! Sure those absolute negative pressures are not correct, but as I don t use cavitation model, Fluent return me those values. And it strongly influences the results. Despite from those crazy pressure values, I have good convergence: all residuals under 10**-5 except continuity which is always remaining between 10**-3 and 10**-4 (ddp solver & all 1st order scheme) I already sent one of my case at Fluent support, and they confirmed presence of cavitation. But I still have problem for computing force on defined surface because of those negative values, which influence drop pressure or massflow calculation too. Regards Maxime

Try to coursen the overall mesh, but then further refine the areas of erroneus pressure. Also try to use simpler discretization and other solution schemes.

Also try it first with maybe the inviscid or laminar formula. With the easiest solution, get realistic results, then go back and modify the models and controls for accuracy.

Hi, I already try everything to correct those pressure. I made benchmarking for simple 2D nozzle. So the mesh is very fine is the high velocity regions, but it still remains cavitation domains. Regarding the laminar or inviscid formula, I already tried, but without reaching any convergence. Regards Maxime

Start over with a relatively course mesh. Use the simplests of models and schemes (inviscid, ist order, simple, etc), decrease the u-r factors to about 3/4 of the original value. try the segregated solver.

Try to patch in initial solutions that may be closer to the actual solution. Maybe modify your boundary conditions so that you have lower Re number or lower compression ratio.

First find what converges, and then lets take it step by step.

Hi,

I am a bit wondering why you try so urgently to prevent negative absolute pressure values. Of course, we do not need to discuss the general phenomenon (even the fact that liquids are able to sustain traction forces at specific circumstances), but I think that the appearance of such values are reasonable in the given case as Maxime describes that cavitation might appear. To prevent missunderstanding: I am not trying to say that you'll see also negative values in the real case. What I want to say is that it is very likely that you will see the creation of vapour bubbles (since at least a major criterion for cavitation onset is fulfilled) in reality. And therefore, the computed negative absolute pressure is just an indication for that--ignoring the physical feasibility.

But, coming to your main problem, the calculation of the surface force. I guess you should be able to fix that by a udf?!

Sure, It is a fact that cavitation appears. So I have to do with it, and I don t have the time to threat it with cavitation model which needs many work for no-better result (I mean negative pressure are still remaining). So I want to treat my surface with UDF, but I am not familiar with it. Do you have any advices? Regards Maxime

Have you already looked into the udf manuals? I think they are quite extensive in Fluent 6, giving you a good starting point. I am not absolutely sure what might solve your problem, but as a rough guess, you could start by employing a cell macro, accessing the pressure in each cell. With that information you could run over a if-else statment that filters out all the cells with negative-absolute pressure. Finally, you might only need to assign a new value to these cells...

Anyhow, use e.g. first a simple 2d model to try different startegies. Good luck!

OK, I will try to do this. Thanks

hi maxime

i also had all the problems you mentioned in your posting.large negative absolute pressure,velocities exceeding 400m/s and the cavitation model in fluent 6.0 didn´t change the whole case for the better (still negative absolute pressure ).now i changed to fluent 6.1 which has a modified cavitation model.first results show no negative absolute pressure.minimum pressure is half the saturation vapor pressure now.but now i´m looking for a source on good experimental data on cavitation to validate the modified model.maybe somebody can help? thanks

mfg klaus

Hi Klaus, I turned to 6.1 too... I quickly made the 2d tutorial. It is quite good, because there is no need to turn the unsteady solver. So I had to find the good settings for calculating my 3d models in steady state WITH cavitation. But I need the right parameters for my fluid: mineral oi @ 40deg the liquid phase is: density 870 kg/m**3, viscosity 0.04 kg/ms the gas phase is: density 1.2 kg/m**3, viscosity 1.2e-5 kg/ms liquid surface tension 0.035n/m non condensablke Gas: ??? but set to 1e-8

Regards Maxime