[miquiita]

Hi ghorrocks,

I did what you told me I created a cylinder and a rotor inside, see pic. Then I put them as a rotating frames of reference, one solid domain(rotor) and one fluid domain. The inlet is stat. frame total pressure and outlet is average static pressure. But I get a rotation of the flow but not how the fan is pushing/draging the air through the duct.

Why is this?

Is it becasue I created a wall to rotate?

Thanks

[ghorrocks]

This FAQ explains what is going on: http://www.cfd-online.com/Wiki/Ansys...f_reference.3F

[miquiita]

Thanks so much for the answer and help.

I ended up with a frozen rotor and interface and optained a solution I am happy with. But I am confused, I have these BC:
-rotating frame of reference (rotor) has a constant rpm
- inlet has a mass flow
- outlet has a total pressure 0 atm
-Then the interface has a frozen rotor.

I am worried that the mass flow of the inlet will change the fan to a turbine and not give me velocity of the fan.

Can ANSYS CFX solve the problem without using mass flow as the inlet BC? Or should I just calculate it with rpm*r (radius of blades). and assume that there is only the circumference velocity to take into account?

Kind regards,
Michelle

[ghorrocks]

You need to think a little more deeply about what is going on.

When you set up a simulation you do it at a fixed rpm and flow rate. The simulation then gives you the torque on the rotor (assume it is accurate). When you add the load on the rotor you will get a net torque. If the net torque is zero then the system is in equilibrium and the rotor will turn at that speed forever. If the net torque is negative then the rotor will decelerate, if positive it will accelerate.

If the rotor accelerates or decelerates then you can try again with a rotor speed which is higher or lower......

Repeat this cycle until you find a rotor speed where the rotor torque equals the load to give zero net torque - and then you have found your operating point.

[miquiita]

Hi,

What you say make sense but for my simulation (I am looking at the air velocity created by the fan), it feels like putting a constant volume-flow at inlet as a BC would not be correct with a constant rpm of the fan. Also the Q (volume-flow) would be affected by the fan velocity and to calculate the Q, or just guess and see what the power is (negative or positive) is that a valid solution?

My main question is if the BC as Q is valid to use and how to choose it when I already have a fan design and a rpm with it.

Thanks so much.

[ghorrocks]

it feels like putting a constant volume-flow at inlet as a BC would not be correct with a constant rpm of the fan.

Can you explain which bit of it "feels" wrong?

The principle I explained is general to different fan driving situations. It is basic stuff - if there is a positive net torque the rotor will accelerate, if negative it will decelerate and if zero net torque the speed will be constant and the machine has found steady state. You can just as easily adjust the flow rate until you find the zero net torque point as well.

[miquiita]

I mean that setting a constant flow rate will give the model no freedom to create one. But as you say I can also change the flow rate, which is what I am intending to do. I am just wondering if there is a way to calculate it with velocity triangles of the blade and the rpm of them.

Then I was thinking if the meshing is needed to be finer, or how fine is the blade and wall needed to be? Is it just the elements I need to change?

Thanks

[ghorrocks]

Your comments about accuracy are an FAQ: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

Don't forget that the air speed through this thing will also be a function of the whole air duct system. The pressure loss though the entire system is what the fan is pushing against.

[miquiita]

What do you mean by "Don't forget that the air speed through this thing will also be a function of the whole air duct system. The pressure loss though the entire system is what the fan is pushing against."

Kind regards,
Michelle

[ghorrocks]

You appear to be modelling a short section of duct. There will be more ducting up and down stream of this, of maybe it just expands into a large volume. Either way these features have resistance and that resistance contributes to the total flow resistance. The power input by the fan is balanced by the flow rate pushing air through the resistance, so changing the total resistance will change the flow rate. This is why modelling the full flow path is important in your case.

[miquiita]

Hi,

yes of course it is. But in my case that does not matter, we assume a circular duct open to the ambient air. I am later modelling the fan as a ceiling fan in a room. This is mostly a "step on the way". My main thought was then how to get the net power from the model.

Kind regards,
michelle

[ghorrocks]

My point is the contraction from a large reservoir into the duct and the expansion from the duct to the outlet reservoir will have a larger loss than the duct itself. Therefore you cannot model this system without the transitions to the large reservoirs on the inlet and outlet.

[miquiita]

Sorry, I do not understand you know.

[miquiita]

so what should I model it as then? thanks,

i mean why cant i model it as the duct too?

and if I would change duct to a room, what would change in the model?

[ghorrocks]

The fan puts a certain amount of power into the flow. That power causes one reservoir to have flow through a contraction into a duct, along a duct, then out an expansion to another reservoir. So the power input by the fan is balanced by the power consumed by the entire flow path from input reservoir to output reservoir. Your model just appears to have modelled the duct.

You seem to mention that this is leading to modelling a fan in a room. Of course that is quite different as the fan then generates a flow field in the room and the power input from the fan is dissipated by the flow field in the room.

[miquiita]

Yes all of that is true about the fan path but if I model it as air with no velocity so it has no flow exept what the fan is creating. Is this not correct? With the BC I putin, Q inlet and pstat outlet?

More importantly if it is not, what does it mean for the model?

I also was looking into the net power but cant seem to calculate it, do you have an idea how to?

Then about the room, I do not see what I have to change, same set up but with a bigger domain as walls and square....?

[ghorrocks]

No, it is not correct. The flow losses will not be correct.

To model the room you just model a fluid domain shaped like the room, and put the fan in a rotating frame of reference. If the room has no significant air flows in or out then this is all you need do.

[miquiita]

Hi the room is already modelled the way you suggested, I just need to create the fan into the domain.

But I am not sure how I will create the reservoir in the model I have then. As you said before "my point is the contraction from a large reservoir into the duct and the expansion from the duct to the outlet reservoir will have a larger loss than the duct itself. Therefore you cannot model this system without the transitions to the large reservoirs on the inlet and outlet."

How should I take this into account?

Plus the inlet and outlet of the duct, BC, is making the flow look like this picture. I got static pressure in outlet as 0 and inlet as a pressure of 100 static and changed it to velocity, the same result.

My main thought is though that the inlet condition should not be a velocity because I want the fan to create the velocity, I have no data there to put in. But I am unsure of what then the BC in the inlet of the duct should be. I am sorry for the many msg, I just are very new at this and I tried doing all of the tutorials too, but I know I always question everything to be sure that I do it correct.

Please help me with the BC of inlet also. the fan should create the pressuer difference and velocity of air to stream through it. But what BC should I put in (both in the air model and duct)

[ghorrocks]

I suggest you have a think about what you are trying to do with this simulation. I think you said you really wanted to model the fan in a room, and wanted to model this as a simpler introduction. As you can see, the model with the duct is in fact more complex than the room model, as the room model does not require flow boundary conditions. SO I recommend you forget about modelling the duct and just model the room.

If you still want to model the duct then carefully define exactly what you are modelling, and what flow conditions upstream and downstream you know. Then we can look at appropriate choice of boundary condition.

[miquiita]

Thank you for answering me fast and so clear and understandable.

I will model the room instead then, but then I am still confused of the BC for the fan inlet and outlet. I have still no velocity except the fan rpm and the dimension of it. What would you suggest the approach to be, is it doable to simulate this with no velocity of the air or pressure values of inlet and outlet?

Also the duct, I wanna do this too but I want to see the flow coming from the atm then through a duct and to the atm again, or the open air as the duct then is a passage for the air with a fan that pushes it through without any air velocity at the start. Would that be ok?

Thanks so much