AnsweredAssumed Answered

Flow results do not correspond to input fan curve

Question asked by Amit Katz on Dec 10, 2015
Latest reply on Apr 12, 2019 by Bill McEachern

Like • Show 0 Likes0
Comment • 9

Good morning fellow Flow Sim users!

I am working with fan curves for the first time, and I'm a little confused about how the software handles them. I simulated an air manifold with an external outlet fan. When I look at the surface parameters of the face that I assigned the fan to, I get these data:

Local Parameter	Minimum	Maximum	Average	Bulk Average	Surface Area [in^2]
Pressure [in H(2)O]	280.11	289.56	285.97	285.97	2.1425
Density (Fluid) [kg/m^3]	0.83	0.86	0.85	0.85	2.1425

Integral Parameter	Value	X-component	Y-component	Z-component	Surface Area [in^2]
Mass Flow Rate [lb/s]	-0.1263				2.1425
Volume Flow Rate [ft^3/min]	-142.8205				2.1425

However, my fan curve looks like this:

A flow rate of 143 CFM is right around the 172"WG mark on my curve, but according to the software the pressure difference at my outlet fan is 407-286 = 121"WG. So what gives? I should be getting either a lot of more or a much higher pressure difference.

No one else has this question

Outcomes

9 Replies

Kim KyuHyung Dec 11, 2015 11:26 AM
Mark Correct
Correct Answer
If fan curve reference density set to 1.2, i think it was correct value
As i know, Fan curve was corrected automatically consider air density in F/S
Like • Show 0 Likes0
Actions
- Amit Katz @ Kim KyuHyung on Dec 11, 2015 11:43 AM
  Mark Correct
  Correct Answer
  But, the manufacturer's curve is given in aCFM units, not sCFM. How can I correct for this?
  Like • Show 0 Likes0
  Actions
  - Kim KyuHyung @ Amit Katz on Dec 11, 2015 6:27 PM
    Mark Correct
    Correct Answer
    Normally, the pressure of output fan is directly proportional to the density of air.
    if air density set to 1.2 in engineering database, fan curve pressure need to change in your environment of density
    Max. Pressure 185*(0.85/1.2)=131inH2O
    
    but Flow simulation calculated automatically considering density. you don't have any change of fan curve.
    Actually, i would like to show corrected fan curve in flow simulation but not supported yet.
    Like • Show 0 Likes0
    Actions
Dario Denzler Apr 11, 2019 4:07 AM
Mark Correct
Correct Answer
Hello

I kinda have a similar issue with my fan curve. I checked the modelling of a fan in a very simple model (channel with an internal fan) to verify my implementation of an internal fan using a fan curve and it corresponds quite well. So at a given volume flow, I do get the right pressure drop at the fan.

Now in my actual model, I have some doubts, even though I used the same approach as in my verification model. From my understanding, the pressure drop of the system needs to be compensated by the internal fan to satisfy the boundary conditions. In my case, I provided volume flow @ inlet and ambient pressure @ outlet and I tracked the pressure drop of the system and of the fan.
What is interesting, my pressure drop of the fan is 74 Pa but according to the fan curve provided, it should be at 60 Pa for the given volume flow. The pressure drop of the system is reported to be 46 Pa. So the 60 Pa from the fan curve sits right in between the current values.

Density should not be an issue as I have 25°C and the fan curve with a reference density of 1.2 kg/m3.
Like • Show 0 Likes0
Actions
- Bill McEachern @ Dario Denzler on Apr 11, 2019 7:51 AM
  Mark Correct
  Correct Answer
  Your volume flow at inlet - what's with that? Is the fan not responsible for driving the flow? Would not a pressure BC be more appropriate? Maybe explain better whats going on.
  Like • Show 0 Likes0
  Actions
  - Dario Denzler @ Bill McEachern on Apr 11, 2019 8:00 AM
    Mark Correct
    Correct Answer
    Actually, I did start with just ambient pressure BC's at Inlet & Outlet but ran in some convergence issues as the internal fan is not meant to be the driving force according to this thread Convergence Problems . Thinking of the fan curve as a "lookup table" does explain why it is recommended to use a volume flow as BC.
    
    For me it is hard to understand, why in a simple test model (channel & internal fan), the output's seem to be right. The only difference I see between my actual model and the test model, appart from a more complex geometry, is that I now have multiple outlet's which have perforated plates assigned. But I guess the perforated plates just reduce the free area numerically to simplify geometry.
    Like • Show 0 Likes0
    Actions
    - Bill McEachern @ Dario Denzler on Apr 11, 2019 9:25 AM
      Mark Correct
      Correct Answer
      I would disagree with that post in it being a "generally applicable approach". I have solved a wide variety of problems with internal fans and multiple openings. Sometimes it is a much better approach to do it as and external flow if you want a decent approximation. You put ids on the openings and disable them and use them as sensor surfaces for inlet and outlet goals to see what's actually going on. While the problem in that post is ill conditioned it may not be wrong - it has many outlets and many inlets not that you know before hand which one is which as it will depend on the pressures that evolve near those openings. The variation in mass flow that was considered "non converged" had very small values - it looked bad on a chart with a small scale but was it of a magnitude that even mattered in the big scheme of things? Maybe, I didn't look that closely. I did not read the whole post in detail either. Pressure pressure BC's can be difficult and sometimes challenging often it is best to expand the domain as mentioned earlier. The other thing is that the term "converged" is not strictly correct. A more precise and accurate way to look at a goal in a steady steady analysis is - has a quasi steady state solution been achieved? There is no such thing as a steady state in fluid dynamics. The best you can hope for is quasi steady and some solutions just don't turn out that way - they have big eddies fluctuating in the solution that could affect your goals and you will see that as unsteadiness. Does that mean you solution is a bad estimate - maybe not. If you have one inlet and one outlet I would skip the flow rate BC - how do you specify it correctly?. It would be better to design your device with the fan at one boundary or the other in my humble opinion but hey there could be good reasons or not doing so. There is nothing inherently bad about a fan curve and it is strictly a local thing - it has no idea what is going on other than what is happening at the fan inlets and outlets. Fan curves often times have two conditions on the curve that both satisfy it........it can be very flat in certain areas. You don't really want your fan to operate near that area as it can bounce around quite a bit. Good luck.
      Like • Show 0 Likes0
      Actions
      - Dario Denzler @ Bill McEachern on Apr 12, 2019 1:50 AM
        Mark Correct
        Correct Answer
        ...Sometimes it is a much better approach to do it as and external flow if you want a decent approximation. You put ids on the openings and disable them and use them as sensor surfaces for inlet and outlet goals to see what's actually going on....
        I will definitely give this variation a try and also compare the results to an internal analysis. For me it was just kind of logic to go with an internal analysis as I am just interested in what is going on inside my system.
        
        ...The variation in mass flow that was considered "non converged" had very small values - it looked bad on a chart with a small scale but was it of a magnitude that even mattered in the big scheme of things? Maybe, I didn't look that closely...
        Well it is not about the "non converged" mass flow balance. For me, the magnitude of the imbalance is acceptable. The problem was the oscillation of the actual mass flow's at the different openings. I think you agree that the solution was definitely not in a steady state situation. And of course, there are problem's never reaching this state but I doubt that this is the case in my problem.
        
        ...If you have one inlet and one outlet I would skip the flow rate BC - how do you specify it correctly?. It would be better to design your device with the fan at one boundary or the other in my humble opinion but hey there could be good reasons or not doing so...
        Unfortunately, it will not be possible to reach my target with just one inlet/outlet and the inside fan would be a preferred layout to avoid noise problems and having a "clean" looking product. For the flow rate, I can estimate the necessary flow quite close as I know the input of thermal energy into my system.
        
        ...Fan curves often times have two conditions on the curve that both satisfy it........it can be very flat in certain areas. You don't really want your fan to operate near that area as it can bounce around quite a bit. Good luck...
        Thank you for the hint and that is also why we want to use CFD in order to choose the right fan which is not over engineered and is just right for the application. But for that, I need a reliable simulation, meaning the fan curve should be on spot. I will probably go back one step and check if something is wrong with my fan curve...maybe double checking with the nativly implemented fan's in Flow Simulation.
        
        For me it is just hard to understand where to problem arises. The volume flow is given by the inlet BC. Therefore, the volume flow at the internal fan should be the same (given there is no big change in density) and thus, I should get the pressure drop from the fan curve, when my fluid enters the fan. Even a flat region should not be a problem if the curve is just a lookup table as for the flow volume, there is only one associated pressure drop.
        Like • Show 0 Likes0
        Actions
        Bill McEachern @ Dario Denzler on Apr 12, 2019 7:18 AM
        Mark Correct
        Correct Answer
        You can have a totally enclosed system with an internal fan and the net flow is zero. So I agree you can easily calculate what you need but not what you will actually get for the net flow in an open system with an internal fan. If you prescribe the flow you need i am sure your results will be what you want but will they be what you get in an experiment?
        There is no reason the inlet flow will match the fan flow as per the closed system example.
        Like • Show 0 Likes0
        Actions

Flow results do not correspond to input fan curve

Outcomes

Related Content

Recommended Content