[ghorrocks]

The fan will have a performance curve. This is a chart of pressure rise over the fan versus flow rate (called the operating line). The rpm and power of the fan will also vary along the operating line. When you put the fan in a system it will find equilibrium where the power input into the system equals the power dissipated by the system, and that will give a point on the operating curve.

You will also find many systems have a system curve - that is a pressure drop versus flow rate line as well. Where the system curve and the fan curve intersect is the operating point.

So to do anything useful with your fan you need to specify a fan curve and a system curve. Then you can define an operating point which satisfies both lines.

Here's a more detailed description of it: http://www.greenheck.com/library/articles/10

[miquiita]

Hi,

Thanks I know these curves, my thought was to put different flow rates and calculate the pressure drop from inlet to outlet of the fan in a room then. Then put the fan curve into the CEL in ANSYS and run the simulation.

Would this still work? Would I get the operating point then?

Thanks,
Michelle

[ghorrocks]

I do not understand your question. How can you define inlet and outlet when the fan is in a room? And what do you mean by putting the fan curve into CEL? Do you want to replace the fan with a source term?

[miquiita]

Hi,

I been doing some thinking and wondering why I cant have a open duct to the atm? Then for the room model I do not want to have a source term, but I was thinking then what I should put in as the BC inlet and outlet, I have no velocity and or massflow from the system that you talked about.

How would I then model the problem? I need to se how the fan creates the flow velocity in the room.

[ghorrocks]

This is why I have been saying the room model is simpler. It has no inlets or outlets as it is an enclosed volume. So the external boundaries are all walls and the fan is inside it generating the flow. No inlets or outlets are required.

[miquiita]

Hi yes but I model the room with the same concept as the duct but with a smaller duct as interface between the room volume and the rotor of fan. Still need a inlet then and outlet of the fan to the room. But you telling me I dont? I need to specify something to go through a rotating fan, the air. Or just inlet if fan as static pressure as 1 atm and outlet 0 static pressure?

[ghorrocks]

It appears to me that you are getting confused between what you are trying to model and the models available in CFX. The first step is always to carefully define what you are trying to model. Do not think about the available options in CFX yet. Can you draw an image of what you want to model and what flow parameters you know about the flow?

[miquiita]

Hi I had a meeting today with my supervisor (not a pro on ANSYS) but my model that I attach a pic of now is to plot a curve of the pressure vs. flow rate. This is to see so that my model is valid interms of the geometry of the blades etc.

But when I calculate the pressure change from inlet to outlet, it is negative. Inlet BC is set to normal velocity of 1-5 m/s (different times) and the flow does nearly not change velocity as in the picture. He said this approach of modelling a fan in a duct is okey for what I am doing, checking so my model is good of the fan. But my question is still why is my velocity not changing in the fan and also why is the pressure upstream negative? Should I measure the pressure differently? (taking the total pressure now inlet)

[miquiita]

this is the picture

[ghorrocks]

Your image shows massive recirculations at the fan. This means you are miles away from the operating point of the fan and anything could happen. So I would not worry about the results from this model, I would adjust things so you are closer to the fan operating point.

[miquiita]

Hi,

So adjust it, you mean the rpm or geometry or what do you mean? My supervisor wanted me to plot inlet velocity to pressure in the fan, but which pressure should I calculate in the function calculator? Please I do not understand all of the functions there, if I should use ave after the fan, but where?

Thanks

[miquiita]

Thanks so much for the help before I figured some stuff of it out thanks. I am just a bit confused about this image when I have a inlet velocity of 0.25 m/s and 0 atm static of outlet in the duct. The other higher inlet velocities gives me the flow through the whole duct, do anyone have a clue why?

Also I cannot seem to get the simulation down to below 1e-4 and I got a good mesh and BC and iterations. But can I fix this to go quicker somehow?

Thanks

[ghorrocks]

The difficulty in convergence is the first FAQ on the list: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

Please post images or the different configurations you have run and the streamlines they give.

The pressures you should be comparing are probably area averaged pressures over a cross section of the flow a short distance up and downstream of the fan. If you are comparing to published data then use the locations they define as their pressure points.

[miquiita]

This image is with inlet velocity of 0.25 m/s inlet velocity. outlet pressure is 0Pa static pressure, like gauge pressure zero.

I have frozen rotor and interface between the inner cylinder, domain of rotation and the walls are stationary of the rotor and the outer duct. As I explained before.

Then my thought is why the velocity stops after the rotor with the inlet velocity of 0.25 and then if the velocity increase it looks like the picture of 1 m/s inlet velocity.

My mesh might not need to be medium, instead maybe I need to have it coarse to have a lower convergency.

I try the tips in the link you sent me and see, so for example the average residence time I take the length of the duct/ velocity of the stream?

Thanks

[ghorrocks]

This does not look adequately converged to me. You will need to converge tighter. If you are having problems converging tighter have a look at the tips in the FAQ.

[miquiita]

Thanks I will have a look then, it is just converging to 1e-2 so

[ghorrocks]

That is very loose convergence. You are unlikely to have useable results with convergence that loose. So I am not surprised your flow fields look very weird.

[miquiita]

Hi,

I tried the physical timestep, by taking the average residence time as the lenght of the duct divided by the velocity. and then take that times 0.3 to get the physical timestep according to a guide of ANSYS CFX. (=0.8)

Then I fixed the mesh to more coarse of the duct and fine on the rotor domain.

But I get this as a msg:
A wall has been placed at portion(s) of an OUTLET |
| boundary condition (at 59.8% of the faces, 60.5% of the area) |
| to prevent fluid from flowing into the domain. |
| The boundary condition name is: Downstream. |
| The fluid name is: Fluid 1. |
| If this situation persists, consider switching |
| to an Opening type boundary condition instead

Should I just change the outlet of the duct to open?

Also the flow is looking like this and the residual is still above 1e-4.

[miquiita]

And also if I change it to an opening, the residual does not go under and keep going straight between -2 and -3.

[miquiita]

And when I put the outlet as an opening with 0 static pressure I get this result and I have the timestep as local time step factor 5. What is the definition of this factor and why 5? As it recommends in the solver helper guide.

Thanks