[CFXMUFFIN]

Hi everybody !

I'm dealing with a calculation and cant find the right way to deal with the problem. Maybe the main approach is incorrect, therefore i*m asking for your advice

Let me introduce the situation first (see sketch):

There is a big volume of air with presribed pressure level (for example 5[bar] . The Air is is entering a pipe of Diameter D.
To enter the pipe the fluid has to pass a small orifice of Diameter d.

The target is, to calculate the pressure drop generated by the orifice for different mass flow rates. And later the influence of different orific diameters.

Or short: delta_p as function of m_dot and d.

Whats is known:

prescribed fluid pressure level at the inlet
Different mass flows through the orifice/pipe

My approach:

Inlet: Total pressure at presribed pressure level.
(i set reference pressure as the presribed pressure level and Total
Pressure at Inlet as 0[bar])
Wall: smooth or rough wall
Sym: symmetry
Outlet: mass flow rate
compressible calculation (Air)
steady state

The idea is to to vary the mass flow rate at the outlet for different calculations. For the given Inlet pressure level, that should result in different total pressure drops (Inlet -Outlet) depending on the different massflow rates.

This is what happens:

At the first look, convergence looks fine.
All CFX standard Residuals (Rms, Rms_max etc.) are dropping to levels of
e-5 - e-7. Mass Imbalance is about 0.0002%.

But if i take a look at my variable of interest:

areaAve(Total Pressure)@Outlet,

this value is converging really really slow.

slow means: RMS dropped from e-2 to e-5 in 300 Iterations.
During this 300 Ieterations the Total Pressure at Outlet droped only by 100 [Pa].
I ran the simulation for further 3500 Iterations but There is no sign that this slow dropping will stop, since the inclination is more or less constant.

I had a look into the result file, evaluating the total pressure drop in planes along the pipe length. Bizarelly the Total Pressure drop is biggest in upper parts of the pipe and decreases until it reaches the Outlet plane.

For example:

Pressure drop at outlet: 31400 [Pa]
Pressure drop in the middel of the Pipe: 31550 [Pa]

I already tried to use
MassFlowAve(Total Pressure)@Outlet,
for the calculation of the pressure drop.

Here the pressure drops are sligtly smaller but the tendency looks better:
(biggest pressure drop at Outlet)
Pressure dropd at outlet: 28900 [Pa]
Pressure drop in the middel of the Pipe: 27340 [Pa]

But even if the use of massFlowAve shows the right trend of decreasing total pressure along the pipe, i still have no idea if the solution did converge or not. (Since pressure was still slowly decreasing at Outlet)

Is this maybe a known problem fot that BC pair? (Inlet: Total pressure , Outlet: mass flow) ? Maybe there is a problem for the calculation of the necessary pressure drop for a prescribed mass flow rate in a system.

Or is the approach the problem?
Mass flow at Inlet would be a possibility but there is no possibility to define the pressure level at Inlet then.

I hope the post is not too confusing and you got a idea what i'm trieng to do

I would be very gratefull for some advices.
Have a good knight and many thanks in advance.

greetz
Muffin

[ghorrocks]

I understand your question - nice user name by the way

Your question really boils down to this FAQ: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

I think it covers the key area you need to look at.

Also be aware that a given oriface will have a maximum flow rate it can provide (choked flow). If you define a flow rate greater than the choked flow it will never converge, because you have asked it to solve the impossible.

[CFXMUFFIN]

Hey ghorrocks,

thanks for your rapid answer.

I had a look at the FAQ, thanks for that.
It gave me some ideas how to solve the problem. Since i want to avoid a transient calculation, maybe a coarser net will give me a steady state solution. It wont be the most accurate pressure drop, but i think for my usage as a characteristic in system simualtion it will be accurate enough.


What do you think about the averaging problem?

massFlowaverage Total pressure drop or areaaverageTotal pressure drop?

When i have a look at the velocity profiles in planes along the pipe, the flow is not fully developed. 1/3 of pipe lenght above the OUTLET ,the flow just beginns to attach to the walls, but still has a faster core flow. Maybe thats the reason why the areave function calculates a bigger Total Pressure drop in the upper parts of the pipe than at the outlet?

I cant extrude the pipe lenght, since in my real geometry the pipe is short too, followed by a big volume.

Haha thanks for the compliments according my name
I was just thinking, a Muffin is something, everybody likes

greetz
Muffin

[ghorrocks]

If the real domain then goes into a tank then model the tank.

Regarding averaging:
Think about what makes this parameter physically useful. Pressures are important because that is what pumps need to push against to make the flow move - and the areaAve makes sense in this case. If you are talking about enthalpies then you are interested in the heat input into the fluid - and massflawAve makes sense.