passive radiator design help needed

[BillA]

again, small pump; NOT big

[metarinka]

okay, I'm getting closer the physical construction of this radiator. I just finished welding up the steel frame, and I should be geting the acrylic water jet cut this week.

I'm having a little difficulty thinking about pumps. First of all as the deisgn has it now I have the ability to split the "radiator" into 2 equal sized lengths of pipe into about 27 feet. OR keep it on one pump at a 55' foot length. I'm assuming most smaller garden pumps or 12v computer pumps would not be able to push 55' of water and that it would be much more effecient to push two 27' lengths on smaller pumps than one bigger one at 55ft. The radiator will be 3/4" pipe and 3/4 fittings, it will operate independently of the computer loop as that would be too big of a restriction.
The res I was hoping to be around 1 gallon, but I a feeling that just won't physically fit in my case I believe around .5 gallon will and in the end it will be some arbitary size. I'm thinking of creating it out of bent and welded aluminum which would be seperated with nylon or brass fittings would this be enough to prevent galvanic corrosion since there will be copper in the loop? at any rate draining and filling should be interesting, and I'll probalby end up throwing in a few check valves or ball valves as the radiator will hold 1.3 gallons and the res will be significantly smaller.
In terms of calculating head though, the pumps will be located at the bottom of the loop to be self priming, They will will then pump approximately 2' up and the rest of the loop will be lower, although referencing the original design half of that will be up hill (for 13.5' or 27' on a single pump). I suppose calculating head is mostly for academics but the pump has to be able to at least circulate this water at some rate. Okay here's the real question (forgive me on the thermodynamics) with the radiator flow rate operating independently of the waterblock loop, is there any particular advantage or disadvantage at running it at a certain speed? this will be passively radiated pipe with no fins so already it won't be that effecient, which makes me think that a slower flow rate with a smaller pump might be better. also each pass will be 27 or 55ft and will total 1.3 gallons I'm guessing that with a slow coolant flow rate so close to ambient it might actually get relatively close to ambient before the end of the loop.
Any additional comments on this radiator design? does two smaller pumps sound more practical? or effecient. how high would this theoretical head be and how should I use that to pick a pump or 2.

[bobo5195]

as a genral rule more flow rate the better

ignore all sugestions about turbulators, turbulence is not a factor as the tube is long. I should (once i am sober) be able to find some maths about how much heat transfer there is from a lenght a tubing and corelate to your design the math is hardcore and my time is limited but i will have a look.

I recommend welding fins on there to increase heat transport (more project marks lower C/w)

If you want to calculate pressure drop i recommend the hazen williams equation which should be sufficent for your uses http://www.lmnoeng.com/HazenWilliamsDesign.htm

[Brians256]

Cleaning up old email, noticed that bobo never got sober.

[Edge]

Hey, I've been working on my current drunk for 11 months and it'll probably last until I'm no longer supporting a family of lawyers from my divorce. Cut bobo some slack, its been less than two weeks so he's probably just catching his stride .

[bobo5195]

Ah forgot about this (wonder why) and i do need alot of slack at the moment.

Simple answer that saves me work go to library and get a good heat transfer book out they should have the equations in there. I'll repeat my recommendation of fundementals of heat and mass transfer by incopera sp? thought i saw it in there.

heat transfer form that type of cooling has been done dusted and commiteed to hell.

[Brians256]

Originally Posted by Edge

Hey, I've been working on my current drunk for 11 months and it'll probably last until I'm no longer supporting a family of lawyers from my divorce. Cut bobo some slack, its been less than two weeks so he's probably just catching his stride .

Well, I'd laugh, but that's some pretty dark humor. I wish you well, sir. Divorce is hell on earth. Hope the lawyers die of a thousand paper cuts dipped in acid. If it wasn't for the four kids I'm trying to ride herd on, I'd probably be working on an epic fit of chemical amnesia too.

Bobo is indeed busy and there is plenty of slack. Bobo, I just noticed your reply... um, yup. RTFM is a decent answer for such a well worked problem. I just wanted to tease you a bit.

[bobo5195]

Thought i should do something rather than sit on my ass. Sorry about my spelling, lack of spell checker.

For a horzontal cylinder i believe heat transfer is this (not slept in 24 hours and book is a little cryptic). Based on an analytical formula, accurate to maybe 20%. Adding fins make things a little more complicated so i'll leave that for now.

Mean Nu = 0.5*(Gr*Pr)^1/4

think this is a standard forumula with a correction factor for it being a cylinder and not really like the formula.

As far as i can punt guess L= pipe OD, outer diameter. L is the characteristic length scale.

All properties are for air at room temp (or where ever you plan to run it)

Nu=h*l /k (nusselt number, basically how good your convection is)

Pr= mu*Cp / k (prandlt number, how good your fluid is at transfer heat)

Gr = L^3*rho^3*beta*delta T *g/ mu^2 (grashoff number. Kinda like reynolds number for natural convection)

Where:
h is heat transfer rate per unit area degrees w/(m^2 *delta T)
mu = absolute fluid viscosity
rho = density
beta = 1/T Where T is your average air temp say. Its alot more complicated than this in real life but that number will do.
Cp = fluid specific heat at constant pressure
g= gravity (9.81)
k= thermal conductivity of air.

Happy variable hunting.

[bobo5195]

hmm forgot abput pipe lenghth. Umm assume unit length ie that these are values per m

[bobo5195]

Oh and it gets more complicated if you want to consider the thermal resistances of the pipe and water pipe wall interface

[Long Haired Git]

Remember that head is only created if the loop is open at the top, and thus the coolant "free falls" to the bottom (like a Bong evaporative cooler). If enclosed, then the restriction is only from the friction on the tubing and the velocity of the coolant.
So, long, narrow pipes are restrictive. Short large-bore pipes are not.

If I understand the design, then once the loop is primed the rise and fall of the height over the length of the rig is irrelevant: weight of water being lifted is offset from the "push" of the water falling back down. Net zero effect.

I side with BiilA. Normally radiators exhibit very little gain for increased flow rate. Even in non-passive rigs, the difference between MCP350 with low LPM and Iwaki RT30 with mega LPM is typically less than one degree Celsius due to the heat of the pump offsetting any gains. Suggest you try a low wattage pump like MCP350 and see what flow rate you get.

If you have a long, straight pipe of constant diameter and section, then laminar flow will develop unless velocity gets high enough to cause turbulence. When laminar flow, water is a poor conductor of heat: the outer sections touching the sides get warm and don't pass this warmth on to the water in the middle of the tube. Its the diference in temp between the water and the copper pipe that causes energy (heat) to flow. So, hot water on edges is a bad thing. Now, as long as you have some change, mixing should occur. T sections, bends and even gentle (within reason) curves all cause the coolant to mix and this average the temp. Another way to introduce mixing is with "turbulators" etc, which can just be strips of material inside the pipes causing mixing to occur.

[bobo5195]

Avoid turbulators(increases back pressure significantly). Turbulence will just fade away after the turbulator.

LHG turbulent mixing is indeed desirable but i'm not sure it is desirable enough to add turbulators, especially as the pipe is relatively thin. Heat wil move to the center of the tube by conduction. Will check the maths. I'm guessing after 10diameters = 7.5 inches (amount of time a thermal gradient needs to develop in the pipe) the extra heat transfer from the turbulators will be around 5% in other words neearly worthless compared to the added friction

[metarinka]

yah I was going to avoid turbulators I think the fact the rad is a design out of circles that have several 180 degree returns so that alone should mix it up enough. I'm still stymied on pump selection, I don't want to buy a pump just to find it is going to stall trying to push a 25' column of water. That makes sense about the verticle area being negated by the return pressure. Although more and more so this design is going to a beast to bleed and fill. also I still have to look into check valves or ball valves as right now I would have no way of dissasembling the case once it is filled without loosing all pressure.

In case your wondering I started building the case a few weeks ago and it will probably be around a month until I get started on the radiator.
linky here http://forums.bit-tech.net/showthread.php?t=109384

[metarinka]

did some more math, and with no fins I get a surface area of 1.1 cu M or 12 cu ft which is 87% of the radiated surface area of one of those zalman "reserators" but mine is made out of thinner walled copper (not au) and my tube dia is 3/4" and flowing where as the zalman has a dia of 5.8" and it looks like no turbulence outside what ever the pump would create down at the bottom. Further more my rad design holds 1.97 times as much water as the reserator. which makes me think I can cut my res size down to what ever I can find that is convenient to use.

I can't find any performance numbers of the "reserator" but on paper it looks like I'll be getting at least similar cooling power or better, especially if I have some sort of flow through my rad.
any one know how those reserator's performed?

[bobo5195]

can work out the performance of the resterator to within 10% using standard formula.

The main performance effector is the fin surface vertical length. It is very small on your design compared to the resterators very long fins so your going to need more area than a restorator.

[metarinka]

oy I didn't think about that, I suppose I'll have a go over the formula although I'm a welding engineering student so surprisingly thermal I haven't gone over yet, and fluid I don't think I ever will.