Strange 80mm Tube Axial Fan Curve - Flowrate vs RPM

[Saladsamurai]

I am looking at fan curves here at work and I stumbled upon this one. Its for an 80mm diameter tube axial fan.

What's up with the Flowrate vs RPM curve (the dashed line)?
In the flowrate interval between 12 - 23 CFM there are instances of the same RPM for different flowrates.

That doesn't make sense to me For a fixed geometry and a fixed RPM how on Earth can we have different flowrates?

Unless it is accounting for blade deformation?

Any thoughts?

 

[Mech_Engineer]

That seems atrange to me as well. It seems to me that generally the flow rate should increase with fan RPM so either something is wrong with the test, or something unexpected is going on with their fan.

 

[Bob S]

The fan is not a constant displacement "pump" like a water or hydraulic pump. With different static pressures, there is different "leakage".
Bob S

 

[Saladsamurai]

The fan is not a constant displacement "pump" like a water or hydraulic pump. With different static pressures, there is different "leakage".
Bob S

I am not sure that I follow you BoB S. Could you elaborate? I am under the
impression that these tests are done under fixed conditions. Some sort of tunnel with
a fixed resistance.

What are these leakages? And how can they cause the RPM to Flowrate relationship
to not have a 'one-to-one' correspondance?

Thank you,
Casey

 

[mikeph]

From a physics background this looks like hysteresis, where there is more than one stable solution to your conditions (eg at 3100 RPM you can have two different flow rates), and the certain state that you are presently is dependant on history. I can't think of a physical explanation for it, though.

 

[minger]

Maybe at those conditions the blades were in some sort of stall or surge condition.

 

[Saladsamurai]

Ah. I think that could make sense. So I've seen other fan curves that explcitly
show a 'stall region.' This one does not. Is there a way to get that info
from this plot? Or do I need p_stag to know that (or even more than that)?

 

[russ_watters]

I am not sure that I follow you BoB S. Could you elaborate? I am under the
impression that these tests are done under fixed conditions. Some sort of tunnel with
a fixed resistance.

No, the test is done under a varying resistance in order to generate the static pressure vs cfm curve. You use a damper in a duct and as you throttle it back, you measure the cfm, static pressure and rpm.

Maybe at those conditions the blades were in some sort of stall or surge condition.

Yes, what that curve is telling you is that the fan at its nominal 12v can maintain a roughly constant 3,000 rpm, but that changes in the airflow and static pressure will cause various aerodynamic effects such as stalling and surging. Note that the difference in rpm across the operating range is only about +-5%. That's a pretty small variation.

 

[russ_watters]

From a physics background this looks like hysteresis, where there is more than one stable solution to your conditions (eg at 3100 RPM you can have two different flow rates), and the certain state that you are presently is dependant on history. I can't think of a physical explanation for it, though.

That's the "surging" that minger was referring to. Under some conditions a fan can indeed have two different flow rates at the same rpm. The flow rate will actually jump back and forth between the two and that oscillation will make the duct rumble like there is a freight train running through it and can physically damage a fan.

Here's a long article on various causes of flow instabilities:

http://www.achrnews.com/Articles/Feature_Article/ac51bcf96e75a010VgnVCM100000f932a8c0____

 

[Bob S]

I am not sure that I follow you BoB S. Could you elaborate? I am under the
impression that these tests are done under fixed conditions. Some sort of tunnel with
a fixed resistance.

What are these leakages? And how can they cause the RPM to Flowrate relationship
to not have a 'one-to-one' correspondance?

Thank you,
Casey

Hydraulic pumps with pistons might pump 50 or 100 cc's per revolution, independent of RPM. Air compressors (piston type) will take a specific amount of air at 1 atm per revolution and compress it. These are constant displacement pumps. A fan blade is not like that. It depends on the viscosity (Reynolds number) of air. Air has a very low viscosity, and flows around the fan blade. If you mount the fan on a closed box and run it, the air pressure inside will rise, but the CFM (air flow) will equal zero.
Bob S

 

[Saladsamurai]

No, the test is done under a varying resistance in order to generate the static pressure vs cfm curve. You use a damper in a duct and as you throttle it back, you measure the cfm, static pressure and rpm.

Interesting. That makes sense. Russ, does this particular test have a name? I would like to look into it some more to get a better understanding of fan curves in general. If I knew the specific name of the test, it would make my researching a little easier. Thanks.

 

[FredGarvin]

Usually we just refer to these tests as fan mapping tests. I don't know of an "official" name for them. Usually, you will see a "whoopy-doo" in the flow line in the same area as the rpm increase. That is where the stall region is. It doesn't really show in that plot which adds to the confusion. Perhaps they don't have the fidelity in the flow data to pick it up.

 

[Saladsamurai]

Usually we just refer to these tests as fan mapping tests. I don't know of an "official" name for them. Usually, you will see a "whoopy-doo" in the flow line in the same area as the rpm increase. That is where the stall region is. It doesn't really show in that plot which adds to the confusion. Perhaps they don't have the fidelity in the flow data to pick it up.

Whoopy-doo. Awesome.

Alright so, I have been googling the hell out of fan curves, since I have never seen one before this internship. Let me see if I am catching on, or if I am totally lost.

On the abscissa we have the volume flow rate Q
On the ordinate we can have the total pressure drop or the static pressure drop, depending on the fan manufacturer.

Either way, ΔP is defined as ΔP=P_in - P_out

I gather from the fan curve and Russ' explanation of the test, P_in is fixed. By varying the damper, both P_out and Q will change accordingly.

I assume that the 'damper' is just some sort of 'mesh' screen of sorts that allows one to constrict and loosen the openings; one extreme being 'completely open' and the other being 'completely closed.'

Questions:

What kind of devices do we use to measure the P_out and Q?

I am also currently baffled by fan curves with positive slopes, but we'll get to that later. I'm sure it has more to do with stalling.


EDIT

Also, I am still confused by the fan curve in post #1. Is it in this 'sloppy' region over the entire fan curve? Why wouldn't they adjust the power input such that the fan runs at a nominal RPM < 3000?

 

[minger]

we can have the total pressure drop

The fan is doing work on the fluid. How do you figure there can be total pressure drop?

What kind of devices do we use to measure the Pout and Q?

There are tons of ways to measure flow rate, see
http://en.wikipedia.org/wiki/Flow_measurement

See the following for some information on measuring static and total pressures
http://en.wikipedia.org/wiki/Pitot_tube

 

[Saladsamurai]

The fan is doing work on the fluid. How do you figure there can be total pressure drop?

I am not sure what you are asking. Surely there is a difference in the total pressure at the inlet of the duct and outlet of the duct (before fan, after fan).

So said it yourself, the fan does work on the fluid. How can there not be a drop?

 

[FredGarvin]

They are measuring the pressure rise across the fan. It should make sense where you have a fan in a duct and you close off the duct, you should get the highest static pressure rise. At the other end it is similar to having the fan outside the duct in that there is no resistance to flow so you get the highest flow from the fan.

I attached an article from a fan manufacturer that should help.

 

[Saladsamurai]

They are measuring the pressure rise across the fan. It should make sense where you have a fan in a duct and you close off the duct, you should get the highest static pressure rise. At the other end it is similar to having the fan outside the duct in that there is no resistance to flow so you get the highest flow from the fan.

I attached an article from a fan manufacturer that should help.

Well, I thought that I demonstrated that I understood that in my post#13 . My question was to minger wrt post #14.

He seems to imply that there is no total pressure drop, which I do not understand or I am interpreting him incorrectly.

This is how I picture the test:

where the the fraction of 'free' airflow out of the duct at 2 can be adjusted by messing with the damper.

 

[Saladsamurai]

Sorry. Still lost. If fan laws say that Flowrate scales with RPM, then why does this graph imply that for roughly constant RPM of 3000 there are flow rates all over the map?

Q₁/Q₂ = RPM₁/RPM₂

 

[FredGarvin]

Sorry. Still lost. If fan laws say that Flowrate scales with RPM, then why does this graph imply that for roughly constant RPM of 3000 there are flow rates all over the map?

Q₁/Q₂ = RPM₁/RPM₂

They are, essentially, doing a single speed line from a compressor map. The fan is held at a constant speed (as close as you can) and the outlet of the fan is varied from wide open (highest flow, lowest pressure rise) to fully closed (zero flow and highest pressure rise). The flow and static pressure is measured on the outlet of the fan.

The affinity laws are used in the "if everything else remains constant" scenario. In this case, the outlet conditions of the fan change so the affinity laws don't apply.

 

[Saladsamurai]

Okay. Thanks for that explanation FredGarvin. I just don't understand what other fan curves are doing now. For example, in this curve, how did they do the mapping? Did they hold RPM constant or not? Is it assumed that they did? I just don't see how one would know if they do not publish that detail.

That's why I tried Googling 'fan mapping test' so I could better understand the test. But I did not get any returns on 'fan mapping test.'

 

[minger]

My question was to minger wrt post #14. He seems to imply that there is no total pressure drop, which I do not understand or I am interpreting him incorrectly.

There is no total pressure drop. The fan is performing work on the fluid, so there is a total pressure rise.

 

[FredGarvin]

Okay. Thanks for that explanation FredGarvin. I just don't understand what other fan curves are doing now. For example, in this curve, how did they do the mapping? Did they hold RPM constant or not? Is it assumed that they did? I just don't see how one would know if they do not publish that detail.

That's why I tried Googling 'fan mapping test' so I could better understand the test. But I did not get any returns on 'fan mapping test.'

That's exactly the same thing. In your second plot, the show 2 different fan speed lines. Both of them show the stall/surge area right around the 75 CFM mark. Each line should be labeled with a speed. Line 4 is a lower speed line than 5.

Have you seen this link?: http://en.wikipedia.org/wiki/Compressor_map

 

[Saladsamurai]

That's exactly the same thing. In your second plot, the show 2 different fan speed lines. Both of them show the stall/surge area right around the 75 CFM mark. Each line should be labeled with a speed. Line 4 is a lower speed line than 5.

So in general, fan curves represent a 'constant rpm' line? That is the curves were mapped using a fan whose speed was maintained?

Have you seen this link?: http://en.wikipedia.org/wiki/Compressor_map

No I have not; reading now though. Thanks

 

[FredGarvin]

So in general, fan curves represent a 'constant rpm' line? That is the curves were mapped using a fan whose speed was maintained?

When presented in this particular format, yes.

 

[Saladsamurai]

You know what I don't like? They are called Laws yet after an hour of searching I have yet to find one derivation. I am sure it is out there (maybe), but I have had no luck.

I found one mention that the Bernoulli conditions need to be assumed.

Edit: I guess that it makes sense. A law is not actually derived so to speak. It is observed. I really just need to get my hands on some test results.

 

[Saladsamurai]

So let me ask those in the field this: when, if ever, are the fan laws of any use?

FredGarvin had this to say:

They are, essentially, doing a single speed line from a compressor map. The fan is held at a constant speed (as close as you can) and the outlet of the fan is varied from wide open (highest flow, lowest pressure rise) to fully closed (zero flow and highest pressure rise). The flow and static pressure is measured on the outlet of the fan.

The affinity laws are used in the "if everything else remains constant" scenario. In this case, the outlet conditions of the fan change so the affinity laws don't apply.

So if the exit pressure is not maintained, the affinity laws do not hold, correct?

So the affinity laws will not hold in a fan mapping test even if the fan speed is not held as a constant.

By virtue of the fact that in a fan mapping, there are varying resistances being applied, there will always be a pressure change. Thus, the conditions are not always constant and

\frac{Q_1}{Q_2}\ne\frac{\text{RPM}_1}{\text{RPM}_2}


EDIT Here is a piece of a document put out by Greenheck. It seems to fly in the face of everything I just said

I feel an experiment coming on

 

[FredGarvin]

When I say the outlet conditions, I mean the piping, etc... that the fan is connected to. In other words, your system doesn't change, you just want to increase the flow of your same fan or what have you. That is when the affinity laws come into play. These are just looking at how the fan/pump will change. They do not consider the affects your system will have. You need to revisit your system curve and see where the new flow puts you and start the iterative process again.

 

[russ_watters]

You know what I don't like? They are called Laws yet after an hour of searching I have yet to find one derivation. I am sure it is out there (maybe), but I have had no luck.

I found one mention that the Bernoulli conditions need to be assumed.

Edit: I guess that it makes sense. A law is not actually derived so to speak. It is observed. I really just need to get my hands on some test results.

The affect of changing rpm is linear with fan speed because of geometry. When you spin a fan, it takes a certain size slug of air and moves it past the fan. Sping it twice, and it takes two slugs of air of that size and moves them past the fan. So there isn't really anything to derive for that part.

The part about the effect that has on pressure is "derived" (if you can even call it that) by taking two examples of Bernoulli's equation and setting them equal to each other (V and 2V, for example), similar to the way the various forms of the ideal gas law are generated.

The part about the effect that has on fan power is derived from the equation for power, in the same way.

They are just ratios of two applications of the appropriate equations.

So let me ask those in the field this: when, if ever, are the fan laws of any use?

I use them all the time to answer the following questions:

-Can I increase the airflow of this fan on this system without over-amping the motor, exceeding 60hz on the vfd, etc?
-Can I add a heat recovery coil to this system and still generate the airflow I need (without overamping the motor, exceeding 60 hz, etc...)?
-Why does lowering the resistance on this fan cause it to over-amp? (that's a really counterintuitive one)
-If I shut my outside air damper at night and circulate return air through the unit, what is my fan energy penalty?
-If I reduce my room air change rate by half at night due to a night set back mode, how much fan energy do I save?
-The fan is running at 50 hz, 80% of airflow and 60% of max fan rpm. What happens when I up it to 60 hz?
-How much energy/airflow does this broken damper, clogged coil, etc. cost me?
-If I cut my airflow in half, does the dP sensor on my air measuring station still have enough dP to measure the airflow?

Now sometimes you can read these directly from the fan curve, but particularly when you don't have a fan curve at hand, the affinity laws can be very useful for answering these questions.

So the affinity laws will not hold in a fan mapping test even if the fan speed is not held as a constant.

A fan mapping test is a test! You measure every possible parameter you can to avoid errors from calculations. For example, as you close that damper and map the performace of that first fan at a constant rpm, applying the affinity law will give you a straight line for rpm vs airflow. But this fan doesn't perform exactly like the affinity laws predict (no fan does). So you want to actually measure all of these parameters for the test.

By virtue of the fact that in a fan mapping, there are varying resistances being applied, there will always be a pressure change. Thus, the conditions are not always constant and...

Well hold on - the fan laws still apply, they just aren't absolutely perfect. If you crank down on the damper for the test, the result you get should should follow the applicable law to within a few percent. You just chose a fan law that doesn't describe the situation you are dealing with, that's all! The law you want is the one that relates flow and pressure.

The paremeters being discussed are:
rpm
pressure
velocity (cfm)
power

When you apply the affinity laws, you have to choose the law that describes the situation you are analyzing. If you make an adjustment that changes the resistance in the duct, you need to relate pressure, velocity and power. RPM hasn't changed, so there isn't any reason to use that law. Changing the resistance in the duct and changing the rpm are two completely separate ways of affecting the airflow.

 

[russ_watters]

Hmm...actually, I might have another way of explaining this that might help. RPM is a physical condition - an input condition, not an output. It is a physical property of the system, not a resulting air behavior. Though it isn't done this way, you could include a term for damper position or duct diameter or however you wanted to vary resistance in the duct. That would give you, for example:

D1^2/D2^2= P2/P1

Ie, as you decrease the duct diameter, the pressure rises in a square function. But since you can also relate RPM and pressure, you might write an equation that implies that an increase in rpm decreases your duct diameter - but that's nonsensical because both are physical parameters.

 

[mikeph]

That's the "surging" that minger was referring to. Under some conditions a fan can indeed have two different flow rates at the same rpm. The flow rate will actually jump back and forth between the two and that oscillation will make the duct rumble like there is a freight train running through it and can physically damage a fan.

Here's a long article on various causes of flow instabilities:

http://www.achrnews.com/Articles/Feature_Article/ac51bcf96e75a010VgnVCM100000f932a8c0____

So when we look at the curve, we are not correct in assuming the data has been taken when the fan was in a steady state? The fan operating at one RPM can have a variety of speeds, and it is just a statistical effect that we see the curve going up and down, as a result of the state of the flow at a certain time of sampling?

This makes the data meaningless to some degree... surely you need a stable flow before you can obtain measurements!

Seems to me to be analogous to plotting the angle of a pendulum against its length, when the pendulum is moving- it makes no sense because the angle is a function of time which is not acknowledged in the graph, leading you to deduce the angle has some bizarre dependence on the length!?

 

[FredGarvin]

The fan will be in as much of a steady state as can be controlled. When you approach surge you start to get rapid pressure excursions and thus large rpm changes due to the fan loading. You have to understand that, in these curves, you can get different flows at the same speed because one is changing the downstream conditions.

 

[Saladsamurai]

So in my scouring the internet for information about fan laws and curves, I came across a source (a snipping of a textbook it appeared) in which the author stated

...the fan laws can be readily identified by doing a vector analysis of a fan wheel ...

But he doesn't actually bother to do this. So if anyone knows of a source where they have seen this done, please let me know.

Also:

1 How do they keep the fan running at constant speed during the fan mapping? I assume they monitor the speed and control the voltage accordingly?

2 This is not generally how a fan operates in an actual application right? The RPM's vary with pressure. For example: when I have my window fan cranking along in the summer, when I close the door to the room, I can distinctly hear the fan slow down.

 

[FredGarvin]

I have never heard of the vector deal with the affinity laws. I was looking through the pump handbook and they don't mention that. I'll have to check out my fan reference at home.

1) Yes. They will monitor the voltage and probably have some form of speed sensor on the test article. It's very easy to do from a magnetic prox probe or even a simple strobe light.

2) This is true. I would question Russ as to how many times he has seen a previously designed system change to the point it would affect a fan. I would imagine it would have to be a rather large change to cause an appreciable change in operation.

Are you sure your fan slows down? I know my fans will change pitch because of, what I perceive as lower flow or such. I can't say it sounds like they slow down, but maybe they do. I'll have to see if I can take a strobe home and do a quick test.

 

[Saladsamurai]

I have never heard of the vector deal with the affinity laws. I was looking through the pump handbook and they don't mention that. I'll have to check out my fan reference at home.

1) Yes. They will monitor the voltage and probably have some form of speed sensor on the test article. It's very easy to do from a magnetic prox probe or even a simple strobe light.

2) This is true. I would question Russ as to how many times he has seen a previously designed system change to the point it would affect a fan. I would imagine it would have to be a rather large change to cause an appreciable change in operation.

Are you sure your fan slows down? I know my fans will change pitch because of, what I perceive as lower flow or such. I can't say it sounds like they slow down, but maybe they do. I'll have to see if I can take a strobe home and do a quick test.

I am not sure that I understand how 'lower flow' causes a pitch change. I am not doubting that it can, but the lower fan speed was the more intuitive explanation for me. It never even occurred to me that lower flow might cause a change in pitch.

Now I am curious.