How do I apply fan performance curves to a heat sink?

[mrpi]

I have designed an air-cooled motor housing equipped with cooling fins. Due to the short time available from concept to prototype, I did very brief analysis before choosing the fin height, width, and length. I chose these to be quite conservative based on the amount of heat loss expected from the motor.

Our customer has recently asked us to evaluate the possibility of operating the motor at much higher load. This of course generates much more waste heat that I need to get rid of.

While shopping around for fans, I have found several that appear to meet the environmental and voltage requirements. The fan manufacturers publish performance curves and this is where I have my question:
The curves are Static Pressure on the y-axis and Volumetric Airflow on the x-axis.
I am not familiar with fan design, so I'm not sure exactly what I'm looking at here.
I plan to shroud the heat sink fins in a sheet-metal duct to help direct the airflow. I expect a certain amount of pressure loss across the fins. Can I use the graph to determine the expected airflow rate? (so I can estimate convection heat transfer coefficient at the fins).

I don't know how to use the performance curves with the heat sink geometry to see if its enough fan for the job.

Any ideas or suggestions are appreciated.

Beat to fit, paint to match.

 

[MintJulep]

Well, sort of.

In theory yes, if you know the expected pressure drop at the desired airflow across the fins and through your duct (remember than the drop will be a function of flow) then yes, just use that value on the pressure axis of the fan curve and check that the flow will be what you need.

But, the fan curves were developed using idealized fan inlet and exit configurations - nice long straight sections of duct. I suspect that your configuration will be less than ideal, so the fan performance will be derated.

 

[zekeman]

Well, if you have the fan curves, all you need is the duct system pressure drop vs flow rate curve that you either get empirically or estimate.It is parabolic starting from 0,0; it is of the form p =K*CFM^2 where K is your duct system characteristic; the intersection of the 2 curves gives you the flow rate. From this you get the Nusselt Number , hD/K and ultimately your answer. No big deal. All you need are only a few empirical points to get K.

 

[Twoballcane]

To use the fan curve, you will have to calculate and compare your system's (heat sink with shroud) impedance curve to the fan's curve. Where the two curves meet is what CFM you will have going across the heat sink and then from there you can calculate Q. However, calculating the system's impedance curve by hand is not as easy to explain in a few sentences. I suggest getting "Cooling Techniques for Electronic Equipment" by Dave Steinberg. In it has a whole chapter dedicated to this with cursory loss coefficients table to choose for different entrances and elbows.

Tobalcane
"If you avoid failure, you also avoid success."
“Luck is where preparation meets opportunity”

 

[mrpi]

Thanks for the quick replies!

I don't have any measuring capability at the moment. I don't have the fan in-hand, and the prototype shroud I built is with the customer. I'll have to estimate through calculation.

I had a look through my fluid dynamics books and this is what I figure I'll need to calculate to get the cooling fin pressure-loss curve onto the fan performance curve:

Reynolds number (Re)
Friction coefficient (h)
Velocity range (from volumetric flow data)

From that I should be able to estimate a pressure drop curve.

Whew!


Beat to fit, paint to match.

 

[IRstuff]

Note that typical PC processor cooling is done with the fan mounted directly onto the fin assembly. Are your fins close enough together to mount the fan directly?

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[mrpi]

The motor housing is turned on a lathe, so the fins look like stacked plates. To get mounting clearance, I cut the front and back off the fins, so they don't go all the way around the motor.

I am treating them as two sets of "straight" heat-sinks.

I plan on using two fans, one to cool each bank of fins through two wrap-around shrouds. The fans would be very close to the cut-ends of the fins.

However, I'll need to work with out customer to see where the ducting can come from and go to. I may have to mount the fans remote and duct in the cooling air, although I suspect this won't be necessary.

Beat to fit, paint to match.

 

[Twoballcane]

I suggest purchasing an anemometer so that you measure the final design to make sure what you are getting in terms of CFM is what you wanted for your cooling scheme.

Tobalcane
"If you avoid failure, you also avoid success."
“Luck is where preparation meets opportunity”

 

[ione]

Regards

http://www.novelconceptsinc.com/calculators-forced-convection-heat-sink-thermal-resistance.cgi

 

[mrpi]

Tobalcane, I think we have an anemometer around here someplace and testing the finished designs is something I've been pushing for.

Ione, I've been using the Novel Concepts website for some time. The results they produced have proven to close to the results we've received from third party CFD guys. HOWEVER, for this case the pressure drop value I get is 100x larger for the Novel results than for my hand calcs...

I'm trying to figure that one out right now.

Beat to fit, paint to match.

 

[mrpi]

I fat-fingered a number, it looks like things are correlating better now.

Beat to fit, paint to match.