Am I estimating my airflow needs right?
 

I have been working on figuring out the cooling reqirements for the system I'm working on, and want to see if there are any obvious problems with the way I'm doing it.

I am water cooling the all the major heat producers in the system. The radiator will be mounted in the front of the case, with the fans behind it sucking through the radiator, and dumping into the case. The air will then pass through the case and exit out the back. To control the airflow and contain the noise, I intend to seal all case openings other than the radiator and the 3 fan outlets on the back of the case.

Some of the 'radical coolers' I've talked to don't think sending the air from the rad through the case is a good idea, but I don't think it will hurt anything since the figures I've seen say air only gets warmed 2-3*C max on its way through a WC rad. I don't see that as a significant diference from ambient.

For the purposes of making my estimate, I am trying to be very conservative. I think it is better to estimate to high and have to turn fans down / off, rather than not having enough cooling power available when it's needed.

I am making two assumptions, 1. That all the heat to be dissipated will be coming out of the radiator (I know it's false, but it makes the numbers much easier...) 2. That my maximum allowed temperature rise in the rad is 10*C. (I probably could get away with a bigger Delta-T, but that's part of being conservative

My heat budget is as follows:

2 X CPU @ 100W = 200W
1 X pump @ 50W = 50W
4 X HDD @ 18W = 64W
1 X FGF @ 100W = 100W (FGF = Fudge and Growth Factor :wink: )
----------------------
Total Heat out = 414 W

According to the NMB website, to calculate required air flow, one uses the following formula

required CFM = (1.7 X Watts) / max allowed Delta-T *C

or

required CFM = 1.7 X 414 / 10 = 72.4 CFM

Does this seem like a valid figure?

Gooserider

Got a link?

The formulae can only apply to a specific model of core, no?

Re: Am I estimating my airflow needs right?
 

I suspect (haven't worked it out) that the equation from the NMB website refers to the Delta-T between incoming air temperature and outgoing air temperature.

My impression is that you might be treating that Delta-T as if it is the differential between the water temperature and the air temperature.

These are significantly different issues. The water temperature will be somewhat higher than the temperature of the air exiting the rad.

If you put 414W into your rad and 72.4 cfm air at 100 kPa 22Deg. C

Your exit air temp will be 32.2 deg C or 90 Deg F

I will be glad to discuss this.

The delta T (air temp) was 10.2 deg. C

My assumption was the same as Since87's so I checked it...

Hmmm... Interesting!
 

There was an article either here or on Overclockers that said the temp change for air going through a WC rad would be 2-3* C max, based both on math and actual measurments.
I believe the reasoning was based at least in part on the idea that thermal transfer takes time, and the air wouldn't be spending long enough passing through the rad to do the complete heat transfer.

The article was by BillA, or one of the other senior cooling gurus, so I'm inclined to believe he knows what he's talking about...

One of the ideas I've had, is that since I'm using a single pass rad (FedCo 2-342) there will be a significant temp. gradient between the inlet at the bottom, and the outlet at the top. Per heat xfer laws, the greater the temp difference, the more heat will be transferred. Since I'll be using two fans in parallel, the lower fan will get hotter air than the top fan, so why not duct it out the bottom of the case? This would keep the worst of the heat load from getting blown back through the case.

Gooserider

Would you please elaborate on this forumla. Or at least show the formula. That would be really, really helpful.

Sure, I will post a spreadsheet tonight that anyone can use.
Do you have Excel?

Re: Hmmm... Interesting!
 

Find the article and see what heatload that 2-3°C figure was based on.

I'll bet it wasn't 414 Watts.

Yes I do, could you please post it. Thanks a lot (i'll have to save it because its helpful.)

That would be great. I tend to avoid doing these calculations because of all the conversions that need to be looked up.

Yeah but it would be good just to know it in the first place. Formulas are always helpful, wait, they should have a thread of all the equations that could be used. Like C/W, airflow, and stuff like that.

I am starting a new one tonight. The first post will relate total watts dissipated from the radiator, cfm, and the temperature change of the air through the radiator. Keep in mind that this has all been covered before, but is burried in the forum.

Yeah but having it in one place would be helpful to noobs like me.

There's a flaw in your heat calcs- the pump/HDs will not make 114 watts of heat. Most of this energy is converted to kinetic energy moving either the water or platters.

I'm guessing 250W would be much more realistic, and thats including growth factor.

I have uploaded the work that relates airflow to temperature rise through a radiator. Since I have to host the information, I am going to hold off on creating a new thread for formulas and spreadsheets until I complete more of them.

This example assumes no pressure drop across the radiator. This is not true, but the pressure difference is insignificant and I will do the thermo the hard way to prove it. I am doing this for myself anyway, b/c I want to model my whole system.

Here is an example of the work. You can launch the excel spreadsheet from the link at the top of the page -or- click here.

edit: let me know if that is what you wanted.
The variables are kind of messy b/c I haven't found the right ones in Excel yet.

Thanks man! I've been waiting for this.

No problem -That was an instant response! I hope it is not too messy. I am sure it will look terrible when I critique it tomorrow!

Yeah, it was quick, I have been waiting for this almost all day (sad...) That looks pretty good and it makes sense. So if you can improve on that, that is pretty impressive.

ouch! i'm looking at 10-15 degree increase in my delta

enter temp is 22c and my watt load is about 660 which yields an exit temp of about 35c

ouch

Some sums previously shown in a "Water Cooling Simulator Discussion" thread http://forums.procooling.com/vbb/sho...5&pagenumber=4

http://www.jr001b4751.pwp.blueyonder.co.uk/Air1.jpg

Some reference stuff...
 

Well, I've been searching, haven't found it yet, but here are some other things I found on Overclockers that appear relevant...

"Radiator System Resistance" This is the source for my assumption of 50% of rated fan performance.

"Pump-Off"
Commentary on the influence of system resistance on waterflow from pumps. (Note that I'm designing for a comparatively high flow volume.)

"Waterblock Bench Testing Results"
Discussion of waterblock testing.

"Water Cooling Flow Rate and Heat Transfer"
Zhentar, see the bottom part of this one, it addresses your comment about pump heat. (sorry, you got it wrong...)

"Measuring Airflow Resistance"
An interesting background peice on how airflow resistance is measured.

"Radiator Heat Dissipation Testing"
This is a primary info source, though a bit dated. It does it's measurements in BTU's, and I forget offhand how they xlate into Watts, but an interesting read. In terms of the rad that I'm using vs. this article; my rad is bigger than either the Serck or the Big Momma. It is the same type of construction as the BM, but is a single pass design which gives it low flow resistance like the Serck. Some crude testing w/ just the rad shows my pump can push about 6GPM through it, showing ~6' of head loss, out of a rated 14' head and 8 GPM flow )

It might also be worth reviewing the Water Cooling Simulator Discussion message area here on Pro-Cooling, there is alot of good info there.

Gooserider

In what sense is pressure drop across the radiator insignificant? (In terms of power dissipated due to the flow through it?)

I haven't had a chance to look at the spreadsheet. (Computer is under construction and doesn't have Excel ATM. Will take a look when I get to work.)

How about posting your simulation work in the Water Cooling Simulator Discussion forum? It will make it easier to find at a later date. BTW, you can attach small files to your post, (like a zipped spreadsheet) if long term hosting is problematic.

Re: Some reference stuff...
 

Yes, all the power going into the pump in a closed loop system exits the pump as heat. (The proportion that exits through the pump casing and through the water will vary depending on the system.) Only in a system where water is pumped from one location to another, is power exiting 'the system' in the form of kinetic energy.

The maximum pressure generated by the fan is only around 1" W.C. or .036 PSI. For comparison: the atmosphere we are working in is 407" W.C. or 14.7 PSI. In this context, the temperature change of the air due to the change in it's internal pressure is small. I said that I am going to figure this the hard way (which includes the air's internal pressure drop) to validate this. I am trying to make the math easier...

This is actually irrelevant -What is important is that I am assuming "Constant Specific Heats" for air at room temperature. This is acceptable for the range of -20 to 70 Deg C with an error of less then 1/2 a percent.

edit: This is a little vague...please take a look at my post when you get your computer running...and tell me what you are asking. ;)

I might post in the "Water Cooling Simulator Discussion" when I get some more significant relations completed. This is toy stuff and probably the easiest model in thermodynamics (Constant Pressure, Ideal Gas, etc.) Some people here wanted to use it -so I put it here.

Now I have to figure out this "attachment thing" with a post...

Off the immediate topic, but another way to look at it is from an 'overcoming friction' viewpoint. This has the advantage of applying to the power consumption of HDD's as well. In the ideal 'Newtonian perpetual motion universe' once we got our disk drive spinning, or our coolant circulating, they would continue to do so as long as we didn't do anything to stop them, and without further energy input.

However, we have the problem in this universe of friction, which will make our coolant stop flowing, and our drives stop spinning UNLESS we keep constantly putting energy into them. Friction expresses itself as heat.

Thus the power of the pump, or the HDD is really the amount of energy which must be constantly added to replace the heat losses caused by friction. This means that those power numbers show up as HEAT in the system, which must be dissipated. In a WC system, that heat ends up in the radiator...

Gooserider

Okay, I'm going to break away from any science here: when planning for cooling, NEVER go for exact amounts. If you think you need 79 or so CFM, plan on using around 200 or so. Really: you can never have enough. You should be able to get one large blower or something that will provide that airflow in near silence if you look and tune hard enough.

NEVER try to skimp on airflow. There is more producing heat in your system than what you are measuring ... think bridges, mosfets, ram, IC's, Fred the Cockroach, etc ....

I updated the spreadsheet today to convert temps to farenheit and to post the temperature rise (dT)

edit: Whew...I haven't been able to post since the update last night. I was playing with my account when the update started and lost all privelidges to my account. I couldn't even open the members page. I tried resetting my password, reactivating my account, and emailing the staff several times. I even sent a screen shot of the flags I was getting. That really sucked b/c I was in the middle of discussions....and had to watch other people answer questions directed to me...

I thought I had been banned from here and didn't even know that I was "on the bubble".

edit: I also added a disclaimer about actual fan CFM.

I'm doing just that, according to my estimates, I 'need' 72.8 CFM. Keep in mind that all the numbers that went into my heat budget were either MFG specs (pump label) or intentional overestimates (100W/CPU, 2X number of drives, etc) and then included an extra 100W as 'fudge factor'. If I was to be more 'realistic' on my heat budget, it would have been on the order of 160 Watts instead of 414

Then I used the '50% of rated CFM through a rad' figure that BillA and Joe mentioned in their articles. Since I'm going to be using two 120mm fans in parallel, I started looking for high Static Pressure (38mm) fans in the 80-100+ CFM (each) range. Currently, I'm looking at Sanyo-Denki 102.3 CFM fans for $5.00 each @ Electronic Goldmine.

In addition, while I want my case to have SLIGHT positive pressure to keep the dust bunnies out, I'm also putting two additional lower volume 120mm X 25mm fans on the case exhausts. For those I'm looking at a couple of Delta WFB1212M fans, (2.95 ea @ Electronic Goldmine) good for a nominal 72CFM each. Since I'm planning to pretty much seal the case (except for whatever leaks through drives and whatnot) the exhaust fans will be in series with the top intake fan, which in effect makes the two pressures add together.

I would hope I never need to run it this way, since I'm sure the noise would be tremendous, but I'm guessing that if I cranked all the fans to max, I'd be doing serious negative case pressure, and sucking about 100CFM+ through the case, and another 50 through the bottom fan / duct combo.
My plan is to get overkill grade fans, then seriously undervolt them w/ temp sensitive controls. I've had some trouble finding a good temp sensitive controller that I liked. My current plan is to make a hybrid unit of a DigiDoc front end feeding the fans through an NTC thermistor controlled FET power ckt.
Agreed, part of why I did such an over estimate.
'Fred's not here Man...' He prefers Windows systems... (more bugs!):D

I think I may have missed the link on this one, where is it, and what's the title? - I have a Linux box, but StarOffice opens .xls files pretty well. (albeit slowly - one of our target metrics will be if the girlfriend and I can both open and run SO at the same time) :cool:

Gooserider

Earlier in this thread, but here it is again:

HTML

Spreadsheet