Fan flow simulation in open air

idiroft · June 2, 2014, 16:06

Hi guys,

first time posting here.

I am a doing a small study on the air flow past the outlet of an array of axial fans in open air. My major concerns are to be able to determine with a certain degree of accuracy the air velocity profile at different distances from the fan outlet.

I know that fan performance relates the static pressure and volume flow for a given system and I know that the air flow at the outlet has swirl that I am not modelling.

As my first approximation, I have build a model where a cylinder is removed from the solid body/domain. The cylinder has a diameter of 0.6 m and a length of 0.2 m. The domain is 10.5 m long (0.5 m behind the cylinder inlet, 9.8 m in front of the outlet) and has a squared section of 6 m with the cylinder at the center.

In the outlet face of the cylinder I have prescribed an inlet mass flow (Q*rho). On the inlet face of the cylinder I have prescribed an outlet velocity (Q/A). Behind the fan I have defined a wall (the ground). On the sides of the wall and in front of the fan outlet I have prescribed an opening with a total pressure of 1 atm.

I solved for steady state, isothermal ait at 25C, k-Epsilon turbulence model.

My questions are:

- Assuming that I have unrestricted air flow (open air), is it acceptable to prescribed the mass flow at the fan outlet from the CFM value given by a fan manufacturer?

- Are the opening boundary conditions well defined (total pressure = 1 atm)?

- For someone with experience with axial fans, how much is the velocity reduction in the axial direction (axis of the fan) when we later take into account the swirl effect in the flow?

- If I were to study the distance I need from the ground in order to obtain a certain fan performance (mass flow), do I need to model the fan as a momentum source and introduce a fan curve (static pressure vs Q)?

Thanks in advance for your thoughts.
idiroft

ghorrocks · June 2, 2014, 21:39

Read the FAQ on accuracy: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

I can already see your mesh is very coarse so that is going to be a problem. Your boundary conditions are also probably too close to the action as well.

Your questions:
- manufacturers CFM - the fan will have a fan performance curve. The CFM you quote will be one point on that curve. It is better to implement the whole curve. But if the manufacturers single point is close enough to your model then it should be OK.
- Opening boundaries - yes, they are well defined. It would not converge if it was not. But whether they are accurate is another question.
- I am puzzled why you are asking this question. A more important question is probably how turbulent is the jet from the fan as that will determine how the fan's flow jet diffuses. Your simulation is showing high diffusion but I suspect that is because of the coarse mesh.
- How will distance from the ground affect the fan?

idiroft · June 3, 2014, 23:25

Thank you for your reply!

I have reworked my mesh and my domain will dedicate some time to perform sensitivity studies.

Quote:

- manufacturers CFM - the fan will have a fan performance curve. The CFM you quote will be one point on that curve. It is better to implement the whole curve. But if the manufacturers single point is close enough to your model then it should be OK.

- My main question in this problem is to understand if, as a first approximation, I can use the CFM given by a fan manufacturer (the value you often see in the technical specifications on a vendor website). Can I assume that the single point in the performance curve is the one for unrestricted open air flow?

Quote:

- Opening boundaries - yes, they are well defined. It would not converge if it was not. But whether they are accurate is another question.

I have played a little bit with different boundary conditions for the opening boundary conditions (with a bigger domain this time) and I have found that there is not actually difference in the velocity at different point of the domain between defining static pressure to 0 or total pressure to 0 (reference pressure to 1 atm). So I have defined my domain big enough where the dynamic pressure at the walls is practically zero.

Quote:

- I am puzzled why you are asking this question. A more important question is probably how turbulent is the jet from the fan as that will determine how the fan's flow jet diffuses. Your simulation is showing high diffusion but I suspect that is because of the coarse mesh.

That is a very interesting point and I have no idea how to go about it. Can you point me to a bibliographic source on the turbulent behavior at fan outlets?

Quote:

- How will distance from the ground affect the fan?

I don't know. I was wondering if anyone has any idea how to relate the fan performance to the distance to ground...

Thanks!

ghorrocks · June 4, 2014, 07:31

If you have no other data then all you can use is the CFM number. Obviously this is not ideal and accuracy may suffer because of it.

A literature search should find lots of information on turbulence of fan exhausts.

idiroft · June 9, 2014, 15:31

So I have redone my simulation:

- Used symmetry to model only 1/4 of the fan, increased the domain upstream from the fan exhaust and refined the mesh:

- Set the fan exhaust to a mass flow inlet and the fan inlet to a mass outlet with the same value.

- Set the turbulence to Medium intensity (5%) at the inlet.

- Used symmetry BC on the new faces, total pressure = 0 on the left and side faces. Wall on the right.

- Used 1st order upwind to converge the residuals to 1E-4 and then switched to high resolution (2nd order upwind) and converge to 1E-4 again.

- At a distance of 15 m (left boundary) I still have an air velocity that is 1/3 of the speed at the fan exhaust. Should I increase my domain in the flow direction?

What do you think of this new setup?

ghorrocks · June 10, 2014, 08:34

Your first post suggests you are looking to get the flow strength at various distances away from the fan. So you should do some checks to see if what you have modelled is accurate. Compare the results from this model to the results from the last model? I bet they are very different. So you will need to refine the mesh until you get the same answer - this is called mesh convergence. Likewise, have a look at the proximity of your boundary condition and your convergence tolerance.

June 2, 2014, 16:06	Fan flow simulation in open air	#1
idiroft New Member Coiso Join Date: May 2014 Posts: 3 Rep Power: 12	Hi guys, first time posting here. I am a doing a small study on the air flow past the outlet of an array of axial fans in open air. My major concerns are to be able to determine with a certain degree of accuracy the air velocity profile at different distances from the fan outlet. I know that fan performance relates the static pressure and volume flow for a given system and I know that the air flow at the outlet has swirl that I am not modelling. As my first approximation, I have build a model where a cylinder is removed from the solid body/domain. The cylinder has a diameter of 0.6 m and a length of 0.2 m. The domain is 10.5 m long (0.5 m behind the cylinder inlet, 9.8 m in front of the outlet) and has a squared section of 6 m with the cylinder at the center. In the outlet face of the cylinder I have prescribed an inlet mass flow (Q*rho). On the inlet face of the cylinder I have prescribed an outlet velocity (Q/A). Behind the fan I have defined a wall (the ground). On the sides of the wall and in front of the fan outlet I have prescribed an opening with a total pressure of 1 atm. I solved for steady state, isothermal ait at 25C, k-Epsilon turbulence model. My questions are: - Assuming that I have unrestricted air flow (open air), is it acceptable to prescribed the mass flow at the fan outlet from the CFM value given by a fan manufacturer? - Are the opening boundary conditions well defined (total pressure = 1 atm)? - For someone with experience with axial fans, how much is the velocity reduction in the axial direction (axis of the fan) when we later take into account the swirl effect in the flow? - If I were to study the distance I need from the ground in order to obtain a certain fan performance (mass flow), do I need to model the fan as a momentum source and introduce a fan curve (static pressure vs Q)? Thanks in advance for your thoughts. idiroft

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Forced Convection: 2D fan (forced air flow source)	acdc_cfd	FLUENT	0	March 17, 2014 09:34
how to stimulate unsteady flow to steady using fan	kelbros	Phoenics	4	December 12, 2013 04:50
Flow meter Design	CD adapco Group Marketing	Siemens	3	June 21, 2011 09:33
How to model flow of centrifugal fan?	Peter	Main CFD Forum	0	April 2, 2008 07:07
axial flow in counter rotating ducted fan	Vishu	FLUENT	4	January 13, 2004 18:52

June 2, 2014, 21:39		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	Read the FAQ on accuracy: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F I can already see your mesh is very coarse so that is going to be a problem. Your boundary conditions are also probably too close to the action as well. Your questions: - manufacturers CFM - the fan will have a fan performance curve. The CFM you quote will be one point on that curve. It is better to implement the whole curve. But if the manufacturers single point is close enough to your model then it should be OK. - Opening boundaries - yes, they are well defined. It would not converge if it was not. But whether they are accurate is another question. - I am puzzled why you are asking this question. A more important question is probably how turbulent is the jet from the fan as that will determine how the fan's flow jet diffuses. Your simulation is showing high diffusion but I suspect that is because of the coarse mesh. - How will distance from the ground affect the fan?

June 4, 2014, 07:31		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	If you have no other data then all you can use is the CFM number. Obviously this is not ideal and accuracy may suffer because of it. A literature search should find lots of information on turbulence of fan exhausts.

June 9, 2014, 15:31		#5
idiroft New Member Coiso Join Date: May 2014 Posts: 3 Rep Power: 12	So I have redone my simulation: - Used symmetry to model only 1/4 of the fan, increased the domain upstream from the fan exhaust and refined the mesh: - Set the fan exhaust to a mass flow inlet and the fan inlet to a mass outlet with the same value. - Set the turbulence to Medium intensity (5%) at the inlet. - Used symmetry BC on the new faces, total pressure = 0 on the left and side faces. Wall on the right. - Used 1st order upwind to converge the residuals to 1E-4 and then switched to high resolution (2nd order upwind) and converge to 1E-4 again. - At a distance of 15 m (left boundary) I still have an air velocity that is 1/3 of the speed at the fan exhaust. Should I increase my domain in the flow direction? What do you think of this new setup?

June 10, 2014, 08:34		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	Your first post suggests you are looking to get the flow strength at various distances away from the fan. So you should do some checks to see if what you have modelled is accurate. Compare the results from this model to the results from the last model? I bet they are very different. So you will need to refine the mesh until you get the same answer - this is called mesh convergence. Likewise, have a look at the proximity of your boundary condition and your convergence tolerance.