Quantifying the swirl in CFX, Swirl Number/Swirl Angle
1 Attachment(s)
[Note: It might be the case that this information has already been shared on the CFX forum but this is an effort to bring it under one thread.]
Quantifying swirl becomes very important in some situations where you would like to reduce it or even sometimes when requirement of design is to increase the swirl. Swirl Number: The degree of swirl in the flow can be quantified by the dimensionless parameter, Sn, known as the swirl number which is defined as the ratio of the axial flux of angular momentum to the axial flux of axial momentum: http://www.cfd-online.com/Forums/att...1&d=1331804552 To calculate in the CFX, create the following CEL expression; Swirl Number [non dimensional] = areaInt(Density*sqrt(Velocity v*Velocity v)*sqrt((Velocity u*Velocity u)+(Velocity w*Velocity w))*sqrt((X*X)+(Z*Z)))@Plane 1/(maxVal(sqrt((X*X)+(Z*Z)))@Plane 1*areaInt(Density*Velocity v*Velocity v)@Plane 1) Where areaInt = Area Integral sqrt = Square Root Velocity v = Velocity in mean flow direction i.e. Y-axis in this case Velocity u = Velocity in X-axis Velocity w = Velocity in Z-axis maxVal = Maximum Value Y-axis is perpendicular to Plane 1, while X-axis and Z-axis are parallel to the Plane1 in this case. For centrifugal pump impeller design it should be between 0.05-0.1 for good suction performance or 0.01 for excellent suction performance. Swirl Angle: This is again very important for specifying the blade angles in centrifugal pumps. Use following CEL expression; Swirl Angle[radians] = atan2(sqrt(Velocity u^2+Velocity w^2), sqrt(Velocity v^2)) Where atan2 = arctangent sqrt = Square Root Velocity v = Velocity in mean flow direction i.e. Y-axis in this case Velocity u = Velocity in X-axis Velocity w = Velocity in Z-axis Y-axis is perpendicular to Plane 1, while X-axis and Z-axis are parallel to the Plane1 in this case Then create a variable SwirlAngleVariable to calculate an area average over the plane, this would give you a value in degrees. |
Quote:
As I know that the results from cfx is in x,y,z coordiantion, and the velocity component into the above expression is in r, tetha, z coordiante. your expresion can make unaccurate estiamtion of swrling no. You know some times the horizontal plane which include the two component of the velocity, (in r and theta direction or in x&z direction ( considering Z is vertical) is inclined with angule. so the unaccurate estiamtion can come from. If any one has any knowledge about this issue, please share it with us.:rolleyes: Best Regards |
Quote:
does your celexpression is for the swirl coming from rotating configuration or even for the natural induced? Regards |
Quote:
I normally use these expressions for the calculation of swirl at pump suction (i.e. stationery frame of reference), to adjust the blade angles. Never used them for the rotating frames. |
it seems to me that the swirl number as defined above is not dimensionless.
|
Quote:
|
1 Attachment(s)
Hi,
no I didn't check it in CFX, i just looked at the expression. Probably the correct expression includes a r^2 in the integrand on top: |
correct expression
Quote:
Regrads |
Also be aware that there are different definitions of swirl. For internal combustion engines, swirl inside a combustion chamber is usually defined as the angular momentum of the gas divided by the angular momentum of the same gas if it was in solid body rotation at the crank angular velocity. This results in a unitless number as it is the ratioes of angular momentums.
|
1 Attachment(s)
Agreed the correct equation is http://www.cfd-online.com/Forums/att...1&d=1335348574
areaInt(Density*(sqrt(X^2+Z^2))^2*sqrt(Velocity v^2)*sqrt((Velocity u^2)+(Velocity w^2))*sqrt(X^2+Z^2))@Plane 1/(maxVal(sqrt(X^2+Z^2))@Plane 1*areaInt(Density*(sqrt(X^2+Z^2))*Velocity v^2)@Plane 1*1 [m]) Any comments! |
questions
Quote:
|
Quote:
Secondly you take root of squares of velocities to get absolute values, otherwise you might get a negative swirl number. |
Quote:
Most definitions of swirl numbers can be positive or negative so you get the direction of the swirl. If you want the absolute swirl number then you should calculate it including sign, and take the abs value once at the end. Quote:
Your comment about units is puzzling. The equation seems to have the units cancelling giving a unitless number. If you have to put a divide by 1[m] at the end to get the correct units then you have a mistake in the CEL. |
Quote:
|
Quote:
But my key point is about the units - your formula is almost certainly wrong if you have to add a divide by 1[m] at the end to get the units to check out. |
Quote:
areaInt(Density*(X^2+Z^2)*sqrt(Velocity v^2)*sqrt((Velocity u^2)+(Velocity w^2)))@Plane 1/(maxVal(sqrt(X^2+Z^2))@Plane 1*areaInt(Density*(sqrt(X^2+Z^2))*Velocity v^2)@Plane 1) |
to know the direction of swirling flow
Quote:
|
Quote:
|
Quote:
|
Quote:
This paper calls out the exact (corrected) formula for swirl number that tariq wrote down in post# 10. Would this formula still be valid, given the alternate definition that you posted for swirl for a gas inside a combustion chamber?? I am a student and relatively new to CFD so please excuse my ignorance. |
Have a look in my PhD thesis - http://hdl.handle.net/2100/248
See page 74, in the literature review. |
Quote:
|
I mentioned this because I wanted to make the point that there is no universal definition of swirl number. It is defined based on what makes sense in the application.
In IC engines it makes a lot of sense to normalise against crankshaft velocity as it is the scaling factor on all engine flows. But for a GT this may not be the case, so an equivalent swirl number normalised by shaft speed may be meaningless. |
Quote:
I confused for CEL equation. The W in the swirl number equation is tangential velocity. So, why you use sqrt((Velocity u^2)+(Velocity w^2)) for the W ? Why isn't velocity u or velocity w used for W? |
All times are GMT -4. The time now is 04:19. |