[jins9158]

I want to simulate ceiling fan of air conditioner.

But I do not want to use conventional rotation analysis techniques(MRF, SMM) because I need to determine the rotation speed of the ceiling fan.

For example, there is an air conditioner attached to the ceiling, and directly beneath it, there is a ceiling fan. When the air conditioner operates, the ceiling fan starts rotating due to the air conditioner's airflow. In this scenario, I want to determine the rotational speed of the ceiling fan.

"In typical CFD simulations, a rotating domain is defined, and the rotational effect is imposed within that domain with the rotational speed specified as an input parameter. However, my goal is not to set the rotational speed as an input but rather to determine it. I want to find out how many RPMs a ceiling fan is rotating at by observing it in my home, where it is driven by an air conditioner. Is it possible to achieve this through conventional CFD analysis methods, or do I need to use a different tool? I've been searching for related papers but haven't been able to find any. I appreciate your assistance."

[ghorrocks]

The approach in this case is to do a series of CFD simulations of the fan at various operating speeds and get the speed versus pressure curve (or alternately the speed versus flow rate curve) for the fan. Then if you know the pressure difference across the fan (or the flow rate) then you can read the speed off from the curve you established.

[Opaque]

Originally Posted by jins9158

"In typical CFD simulations, a rotating domain is defined, and the rotational effect is imposed within that domain with the rotational speed specified as an input parameter. However, my goal is not to set the rotational speed as an input but rather to determine it. I want to find out how many RPMs a ceiling fan is rotating at by observing it in my home, where it is driven by an air conditioner. Is it possible to achieve this through conventional CFD analysis methods, or do I need to use a different tool? I've been searching for related papers but haven't been able to find any. I appreciate your assistance."

That statement is theoretically correct. However, keep in mind there are several ways to solve a problem.

For simplicity, we can reduce any CFD problem to F(X) = 0, that is "find the roots of my model". Now you can solve that problem with sophisticated algorithms, say Newton, or even more simplistic graphically (like smart people used to do before computers ever existed).

In the case of a simple 2x2 linear system, the solution is the intersection of the two lines. For a non-linear system, the two equations may not be straight but still cross somewhere --> the solution.

Glenn's suggestion is the practical, engineering battle-tested approach used for decades in all areas of engineering.

[jins9158]

Originally Posted by ghorrocks

The approach in this case is to do a series of CFD simulations of the fan at various operating speeds and get the speed versus pressure curve (or alternately the speed versus flow rate curve) for the fan. Then if you know the pressure difference across the fan (or the flow rate) then you can read the speed off from the curve you established.

"Thank you for answering my question. However, I would like to know the rotational speed directly based on the airflow volume as an input for the air conditioner. I may need to analyze the ceiling fan's shape in various cases, and if that's the case, the number of analyses required could become quite extensive. (Each different fan shape would require its own performance curve) ... Wind turbines rotate in response to the wind, driven by the aerodynamic forces. I would like to determine how fast they rotate due to the wind. In my opinion, I wonder if it would be possible with a standard CFD tool..."

[jins9158]

Originally Posted by Opaque

That statement is theoretically correct. However, keep in mind there are several ways to solve a problem.

For simplicity, we can reduce any CFD problem to F(X) = 0, that is "find the roots of my model". Now you can solve that problem with sophisticated algorithms, say Newton, or even more simplistic graphically (like smart people used to do before computers ever existed).

In the case of a simple 2x2 linear system, the solution is the intersection of the two lines. For a non-linear system, the two equations may not be straight but still cross somewhere --> the solution.

Glenn's suggestion is the practical, engineering battle-tested approach used for decades in all areas of engineering.

I appreciate your response, but I'm still having trouble understanding your question. Let me clarify what I want to know using wind power generation as an example. When the wind blows, the wind turbine generates lift on its blades, causing it to rotate. I want to know how fast it will rotate when the wind blows. In my opinion, if we know the material properties of the blades, we should be able to predict the forces acting on the blades and calculate the rotation speed.

[ghorrocks]

I may need to analyze the ceiling fan's shape in various cases, and if that's the case, the number of analyses required could become quite extensive.

That is true, but they are simple simulations which will give you a result quickly.

I wonder if it would be possible with a standard CFD tool...

Yes, you can do a rigid body simulation of it and it will theoretically adjust the speed of the rotor until it finds the correct speed. I say theoretically as this is a difficult simulation to set up, and convergence is going to be slow and difficult to achieve. So feel free to do it this way if you like, but I bet you never get it working so never get a useful result.

I appreciate your response, but I'm still having trouble understanding your question. Let me clarify what I want to know using wind power generation as an example. When the wind blows, the wind turbine generates lift on its blades, causing it to rotate. I want to know how fast it will rotate when the wind blows. In my opinion, if we know the material properties of the blades, we should be able to predict the forces acting on the blades and calculate the rotation speed.

Yes, that is what Opaque and I are talking about. As I just said, getting the solver to work out the rotation speed based on a rigid body solution approach is hard to get working and difficult to get accurate; whereas simulating a series of fixed speeds and plotting the performance curve is easy and accurate enough in most cases.

[jins9158]

Originally Posted by ghorrocks

That is true, but they are simple simulations which will give you a result quickly.



Yes, you can do a rigid body simulation of it and it will theoretically adjust the speed of the rotor until it finds the correct speed. I say theoretically as this is a difficult simulation to set up, and convergence is going to be slow and difficult to achieve. So feel free to do it this way if you like, but I bet you never get it working so never get a useful result.



Yes, that is what Opaque and I are talking about. As I just said, getting the solver to work out the rotation speed based on a rigid body solution approach is hard to get working and difficult to get accurate; whereas simulating a series of fixed speeds and plotting the performance curve is easy and accurate enough in most cases.

Okay I understand. Thak you for your kind