Starting with Bernoulli's Equation... 2

[Tunalover]

Does anyone out there have a reference showing how to calculate the volume flowrate (cfm) vs. static pressure (inches water) curve for an aircooled electronics enclosure? This would involve consideration of duct losses due to expansion, contraction, splits, turns, etc. Years ago I saw where someone started with Bernoulli's equation but I can't seem to put my hands on it anymore.

My goal is to determine the enclosure's operating point given the fan's published static pressure vs. flowrate curve; I want to overlay the two curves on the same graph and find the intersection (the operating point).

I've been able to find plenty of sources who talk about doing the above but haven't been able to find a source that actually DOES it and shows how to do it!

Thanks in advance for your help!

Regards,
Tunalover

 

[quark]

First of all you have to calculate the system resistance for the maximum flow. All the equations( for straight duct, elbows, bends, tees and reducers etc) to calculate system resistance are basically derived from Bernoulli's equation with some constants factors added. All these details are available from ASHRAE or Carrier data books. Your best choice will be SMACNA's duct design manual.

Once your are done with system resistance at maximum flow, it is a cake walk to construct a system resistance curve as follows.

Take flowrate on x-axis and system resistance on y-axis. Mark the max. flow rate and corresponding system resistance on the graph(say, point A) and drop a vertical line from this point to the x-axis (say, point B. So AB will be parallel to y-axis). Draw a line OA with O as origin of the graph. Now split AB into 5 parts and name them C, D, E and F from x-axis. Draw OC, OC, OE and OF. Draw lines parallel to x-axis from C,D,E and F to cut OA and name them G, H, I and J respectively. Draw vertical lines parallel to y-axis from G, H, I and J to cut OC, OD, OE and OF respectively at K, L, M and N. Join K, L, M and N and you will get a parabloa. This is your system resistance curve. You can take as many no. of point as you wish to make a smooth curve, but I took four points for simple explanation.

Each point on this curve will give you system response if you fix any one parameter. (either flow or resistance) Now you have to superimpose this curve on various fan performance curves to check which fan best suits your application. Maximum possible, try to select a fan which operates at best efficiency point.

The above said example is the simplest one and some corrections are to be done depending upon different applications.

Regards,

 

[Tunalover]

quark-
Looks like you've done this before! I was actually looking for more detail e.g. formulas. Do you know where I might find a worked out example?
Tunalover

 

[Bgoldstein]

There is a good book by Steinberg, called "Cooling Techniques for Electronic Equipment". He goes through some pretty in-depth calculations of exactly what you are looking for, with duct constrictions, bends, expansions, splitting (when passing over several circuit boards), going through filters, etc. Basically end up with a system pressure curve, like you are looking for.

You can get it from Amazon (they have links to used versions of this book too.)

 

[wilg]

Bgoldstein "Cooling Techniques for Electronic Equipment"
sounds like a definite one for my bookshelf,
Thank You,
Wilg

 

[Tunalover]

Thanks folks but I have Steinberg and have never been able to understand his formulas and procedure. To gain a fundamental understanding, I need to see how the formulas are derived from the basic Bernoulli Equation.
Tunalover

 

[Bgoldstein]

Well, as best as I can remember it (I don't currently have a copy of Steinberg - borrowed it from a coworker at my last job) all Steinberg is doing is treating the flow path as a series of pipes. So all of the different parts of the flow, like constrictions and bends, have head losses calculated for them, similar to pipe flow. Any basic fluid mechanics book should explain how to interpret head losses from a Bernoulli standpoint.

Hope that helps.

 

[Tunalover]

Thanks quark and BGoldstein!
Tunalover