Average static Pressure
Does average static pressure=0 at (outlet) signify Neumann boundary conditions. :confused:
If I require Streamwise gradient=0 at the outlet, will the above condition suffice it? Thanks Santhosh |
This is all discussed in the documentation.
But from memory it defines dirichlet BCs on pressure and neumann BCs on velocity and scalars. So yes, a static pressure BC will give you zero streamwise gradient in other variables providing the boundary is perpendicular to the flow. |
Yes,
The Flow is perpendicular, and when the Static pressure is Zero, The only pressure acting is the Dynamic pressure, So, By Bernoulli's equation, the dynamic pressure provided by the velocity of the fluid is constant or the flow is fully developed. Thank you for clearing my doubt. San |
Hello,
I am having another doubt in this regard, even if the total pressure in the system is equal to the dynamic pressure provided by the velocity need not be constant. So how is it possible to apply the boundary condition of zero velocity gradient at the outlet. |
Hi everyone.
I have a problem, I want to set outlet's pressure is 0.8 atm. but I cannot distinguish between "Static pressure and Average Static pressure" Please give me an advice. Thank you very much ! |
Static pressure means the static pressure at all locations is set to the defined value. Average static pressure means that the area average of static pressure over the patch is the defined value.
|
What is the physical significance of both options. When to use static pressure (fixed every where) and when to use the Average static pressure(floating with average constrained)?
In my opinion static pressure condition may be the good for the boundary far away from the object to be studied, in other words flow is more or less uniform at outlet boundary. On the other hand the average static pressure may be more physically correct as it allows the conditions to vary along the boundary. This can be the situation in nozzle (may be diffuser, compressor, turbine, combustion chamber) where the velocity is higher in centre and low in the wall region. In other words pressure may change along the boundary so as to capture the true picture of flow form the interior and at the same it also forces the mass flow to be some average value (also changes from cell to cell) corresponding to average value of static pressure on whole boundary. Sounds logical? |
Yes, you are correct. But also consider the numerical implications - specified static pressure is more numerically stable than average static pressure. For some simulations this may be important.
|
1. Take an example of compressor, we knew from theory that the static pressure increases from hub to casing in response to centrifugal force, so if we restrict the outlet condition to single static pressure condition everywhere, wouldn't it be non-physical. It is also implied that the mass flow rate is also same every where on the outlet plane.
2. More over I have simulated NASA rotor 37 with both conditions and found that the results are overlapping (performance (pressure, efficiency) curve overlapping) , but It is also important to mention they don't have same mass flow rate or location on map for the same pressure ratio or efficiency. This result is also surprising for me that if average pressure in both cases is same then why mass flow rate is different. 3. It is also hard to think how can be static pressure be same throughout the outlet plane, if outlet plane is not far away from the wall. 4. Also I have statistics from the CFX 13 tutorial guide that shows this: Among 32 tutorials there 14 are cases which use the average static pressure and static pressure as boundary condition at outlet (and mostly with zero pressure) 1. eight cases used average static pressure 2. six cases used static pressure (it is not mentioned in tutorials why static pressure or average static pressure condition is chosen) What this shows? 5. There is also one note in tut of coal combustion; reads as : Average pressure boundary condition leaves pressure profile unspecified while constraining the average pressure to the specified value. In some situation, leaving the profile fully unspecified is too weak and covergence diffculties (not talking about accuracy!!!) may result. The pressure profile blend feature woks around this by blending between a unspeficied pressure profile and fully specified pressure profile (so this is not completly loose condition!!!). By default, the pressure profile blend is 5%. For swillring flow, however, any amount a uniform pressure profile is inconsistent with the radial pressure profile which should naturally develope in response to the fluid rotation and therefore pressure profile blend must be set to zero. This statement is also supporting the use of average static pressure either with specified blend or with zero blend. |
Quote:
|
Quote:
Quote:
Quote:
The pressure is essentially constant normal to the flow through a boundary layer. That is why this BC is commonly used. The tutorials teach you which buttons to press. They do not teach you CFD. You need a knowledge of CFD to decide whether averaging is what you want or not. Quote:
|
I have similar results, didn't find in any difference in compressor map. 100% overlap
|
Quote:
Should I put a large tank/reservoir on the outlet and see what type of flow I get on outlet, whether varying static pressure or uniform static pressure? |
If the experiment has downstream equipment to smooth the flow out then you should model it, one bit at a time, until your results start behaving themselves.
|
Reference : J. Turbomach Volume 128 Jan 2006 page 126 NA Cumpsty and J.H. Horlock
If static pressure conditions are to be used at exit (total to static performance measurement) they should be defined where the static pressure is uniform, as in a large plenum or in the open atmosphere. remedy to this problem is that move the outlet where static pressure is uniform, otherwise this approach will introduced error. Non uniform static pressure is clear evidence that the flow is being turned or accelerated and such regions do not lend themselves to providing a good reference condition downstream of a turbine or compressor. While this may be true at the diffuser exit, whose function is reduce the velocity to level (mach 0.1) acceptable to combustion chamber with minimum distortion at exit plane Now we can say if the flow is not uniform then why should we say it so to code? |
Hi,
What I had seen in my past work was that flow eventually mixes out after a certain distance behind the rotor. So as per Cumpsty ( your reference paper) static pressure boundary condition should be stated there. Now as far as specifying average static pressure with pressure profile blend factor goes, I would say it would be farther from reality as compared to specifying only static pressure. Also what I have seen from various experiments and literature was that the pressure profile evens out to a great extent 2.5-3 chords downstream of the rotor. So specifying pressure ( or average static pressure ) at this location would not make a huge difference and should be more appropriate. -D.B |
Quote:
|
Quote:
At a far away boundary where the profile actually is evened out, I would say the difference shouldn't be much.. at a nearer boundary where there actually is a variation in the pressure distribution at the exit plane I would say obviously average static pressure with blend factor is better. But let me point out that it would be far from reality, because of presence of swirling features like passage vortices and tip flow vortices and other non uniformitiesin the flow. If you want a better solution and have experimental data with you, I would say at a nearer boundary a mass flow outlet condition is better suited than the pressure boundary conditions. -D.B. |
so you are ruling out the uniform static pressure boundary condition if it is placed too close?
And from your discussion I also infer that average pressure static with zero blend factor should work good regardless of location of outlet boundary? Mass flow is not a option !!! because some times we have to work with new designs and it is very hard to make map near choke conditions. |
Quote:
Quote:
Quote:
-D.B |
2 Attachment(s)
While looking at the static pressure at different planes from inlet to outlet, it is clearly visible that the average static pressure condition permits the solution to be developed naturally than the enforced uniform static pressure at outlet.
From physical point of view enforced uniform static pressure condition assumes that the flow is completely mixed out which may be true at the outlet boundary placed at far far downstream. Another physical interpretation is : it is basic design philosophy of turbo machine that there should be minimum radial velocity gradient, this can only be minimised if the static pressure increases from hub to shroud. |
Quote:
Quote:
Quote:
But you have to take care some design allow a pressure gradient to induce some radial velocity gradient. i.e designs not sticking with raidal equilibrium design/theory |
Now question comes, when to use static pressure or average static pressure condition. Can you name three three cases for each boundary condition as per your understanding? e.g. what should be boundary condition for fully developed pipe flow with uniform cross section?
|
Quote:
|
Quote:
|
Quote:
|
1 Attachment(s)
please look at this
|
4 Attachment(s)
Finally got the definite answer. The answer is
" For axial compressor rotor, at-least, the uniform static pressure does not represent the true physics and is therefore technically wrong. However, surprisingly, the performance map with both conditions is overlapping with some shift (on same curve) on map for the same back pressure and for both conditions". The reason is unknown and your comments shall be very much helpful to me. For confirmation the following procedure was adopted : 1. Another domain of constant area ratio is constructed at the outlet of rotor extending approximately 8.18 chords from the interface plane. And approximately 10.12 chords from leading edge. The walls of the extended domain are modelled as free slip as frictional effects are not important and to save the computational resources as well. 2. Rotor has Y+ = 1 and around 1.3 million nodes. SST turbulence model and automatic wall treatment. 3. Interface is modelled as mixing plane and frozen rotor and at the exit of extended domain two boundary condition are specified a) uniform static pressure b) Average static pressure (For results shown here it is 115000 Pa) Conclusions are: 1. For four cases mass flow rate is same and convergence is good and residuals converged to 1e-05 -1e-06 2. Results are shown at three planes a) plane upstream of interface at 1.45 chord from leading edge (Plane 1) b) plane of interface at 1.94 chords from leading edge (Interface) c) plane downstream of interface at 5.1 chords from leading edge (Plane 3) 3. Values at the interface (original outlet boundary for the rotor simulation) shows the radial variation of static pressure from hub to shroud. It is natural to have this pressure gradient to account for the radial flow and other effects. 4. Results at interface are interesting to note where for both boundary conditions, we get the similar pressure profile at this boundary, which confirm the validity of average static pressure condition. 5. Even at far downstream plane static pressure variation from hub to shroud is visible and gradually decreases to nearly uniform value at far downstream boundary. Although some variation is still there. |
So what do you conclude ?? From your analysis, it seems that at far far away boundary both B.C's are good enough ( the point I was trying to make earlier ). Now ,at the interface ( original outlet ) I am unable to see much difference in pressure distribution in both cases. but a pressure variance occurs throughout the domain, hence at no point it is logical to apply static pressure unless it is far far away.
It would be interesting to see what is the pressure distribution for both the cases at 8.18 chords downstream of interface ( the current outlet ). Could you post that too ? |
Quote:
Quote:
However if you define the average static pressure = 0 or 101325 or any other value with blend factor = 0 does specify the neumann BC on pressure, velocity and scalars (By default blending factor = 0.05 or 5%). @DB I just ran few more cases to understand it thoroughly and now I have made my conclusions and shall post the comments with pics within few days. @Glenn Can I post my results along with pics so that it can be of help to everyone. Can you please guide me how to do this? |
You have previously posted images in this thread, so I trust you know how to do that. For things like CCL files put them in as attachments.
|
Well, I am not talking about this particular thread.
|
1 Attachment(s)
Quote:
Two important conclusions are: 1. More important is their effect at interface (original outlet), which is very much equal for both conditions. Hence the mass flow rate, pressure ratio and efficiency values are also same. 2. However in original case (without extended domain), they produce very different results for same outlet static pressure condition. |
Hi,
Could you compare the interface results for far far outlet boundary domain with the small domain for both the b.c's ( avg S.P and S.P) and tell which B.C is giving interface values much closer to case of far far outlet boundary case interface value. ANd what is the % for both B.C's as compared to far away boundary. This would let us know which B.C can be used with much greater accuracy for a small domain size ( Again I think Avg Static Pressure would be better ) |
yes you are right, Av Static pressure BC is more realistic. I shall share full analysis here as you have suggested within couple of days.
Thanks for giving me new ideas for data interpretation. :) |
1.If I set Static pressure 1 bar at inlet, and mass flow rate at outlet, my simulation is not converging. Neither vice versa.
2.But if I put mass flow rate as inlet and average static pressure 0Pa. Solution converges. My results are qualitatively perfect, but are 40% of paper. I have thoroughly checked the CEL expressions and others input. (I put 1bar in reaction rate equation, besides Pref as 1bar) Using 2nd option the pressure and total pressure in the domain is like 0-0.2 Pa. So the discrepancy can be regarded to that? Note: Changing Pref changes inlet density though. |
All times are GMT -4. The time now is 12:36. |