Nuclear Tower Water Cooling

Add Your Comments

ED NOTE: This article is 5 pages long.

Those of you running peltier cooled T-Bird rigs are probably aware that you are pushing the limits of standard water cooling methods. Many vendors claim their coolers are capable of “handling” several hundred watts of heat output, but what does “handling” mean?

About a week ago, I was cooling 172 and 84 watt peltiers with the large radiator you see in the above picture – it “handled” the heat, there were no catastrophic meltdowns or other disasters. But I was using a reservoir and noticed that the water was particularly warm. I didn’t measure it, but my best guess is that it was between 30 and 35C, and I’ve become pretty good at guessing through weeks of testing.

So are you satisfied if your water is that hot? I personally expect to keep temperatures near ambient with a high performance water rig. That’s why I purchased the massive radiator you see. So what the heck is the other monstrosity in the picture? You’ll find out shortly.

The Contenders

DD Cube

The benchmark unit is the Danger Den Cooling Cube – a very popular unit many of you are familiar with, which is exactly why I chose it. I found it has about 0.6 square meters of surface area.



After countless hours looking up numerous heat exchanger manufacturers, I found what I believe to be the very best radiator in the dual 120mm fan size class. I checked out every type of cooler: heatercore, tranny cooler, oil cooler, sprint car radiator and several others. The best type are generally racing oil coolers.

After checking most of the racing oil coolers on the market, I came across this model: the Fluidyne DB-35017. It’s all aluminum, welded into one solid chunk of it. The cooling core is nearly 4″ thick, so there is more cooling power in a smaller footprint. It holds about one quart of water. After careful measurement, I found it has about 2.6 square meters of surface area.

Obviously, this is very ballsy watercooler; its job is usually to cool the engine oil of Nascar racers. It wasn’t cheap either, MSRP is $400, though I got it for $360. I bought it because I had the money at the time and this is not the sort of thing you ever need to upgrade.

Although I won’t be using it at the moment, I plan on keeping it as backup for when I go to college in September. I can easily see somebody thinking it’s funny to pour beer into the evaporative while I’m not there, or simply knocking it over in a drunken stupor. You can check it out at here.


What’s this? A giant bong? A misplaced piece of sewage equipment? Both very good guesses. After a quick lesson in bong mechanics from one of my more “enlightened” friends, I found that yes, it is in fact physically possible to use this as a bong. What it really is though is an evaporative water cooler.

Now I know this may disappoint some of you, but I feel that it’s necessary that you know it cannot function as a bong and a water cooler at the same time. We’ll get to hear more about how it works later.

The Test

I used a somewhat unorthodox method of testing for these heat exchangers. Oftentimes a radiator that looks as though it will kill another one in a roundup but only performs a fraction of a degree better. I wanted a test that would easily capture even a small performance difference between radiators.

To do this, I set the heat exchangers to cool a reservoir of heated water. I found that 10 liters (2.64 gallons) of water heated to 50C is a good balance between dependable results and monotony of testing. I recorded temperatures at 1 minute intervals for 1.5 hours. This should give reproducible results – you should be able to test your OCWC Big Momma radiator with YS Tech fans against my DD cube with Panaflo l1a even though you may live 2000 miles away.

The water was held in a picnic cooler to prevent heat loss. It proved very effective; several times I left to have lunch in the middle of a test. When I came back, temperatures had hardly budged. Joe seemed to like this idea for testing, we’ll see if he uses it in future radiator reviews.

This method also allows you to deduce thermal resistance. It takes a little bit of calculus and probably the aid of a graphing calculator. I’m trying to find a way to calculate thermal resistance using MS Excel, but I’m not that handy with it. If anybody knows how to get a function from a set of data points using MS Excel, or to simply find area under data curve, please let me know.

I had to use a mercury thermometer for testing. These sorts of thermometers usually vary a couple of degrees when compared to each other, but I used the same one for all tests so that should not be an issue. My reading of the thermometer is probably only good to +/- 0.1C most of the time, but as you’ll see, 0.2C here or there doesn’t have a big impact on comparing heat exchangers.

Other than that, I used an Eheim 1250 pump and 120mm Panaflo l1a’s for testing. Room temperature remained at 24C throughout testing. Relative humidity was 50% in the room during evaporator tests. Fans were set to suck air through the radiators.




Steve Pignato


The easiest way to view results is to simply plot the temperatures on a chart:

Temp Data

The difference between each heat exchanger shows up easily, but to get a more quantitative difference between radiators, the best way is to count the time it takes a radiator to drop the water temp.






DD CubeRadiator










Evaporator, no fan





Evaporator, w/ fan





Quite impressive indeed! But wait, there’s more!

If you look again at the first chart, you’ll notice that the evaporative cooler dipped under ambient temperature. I stopped the fanless test after noticing wood chips had clogged part of the evaporator, but in the test with the fan, water temps settled at 17.6C – 6.4C below ambient! If you would like the full data set to analyze yourself in further detail, you can get it here.

I also tested flow rates of each heat exchanger. Flow rate for the pump with ½” hose and barb was 240 gph, the DD cube cut this down to 80 gph and the Fluidyne cooler cut flow to 156 gph. When I switched the hose and barb to 5/8″, I had a pleasant surprise.

With the hose and barb alone, I got 360 gph, pretty good for a 317 gph rated pump. The evaporator cut this down to 212gph. This makes me glad I chose to go all the way up to 5/8″ for my water system. I’m sure if the barbs on the fluidyne cooler were 5/8″, that it would flow much more water. Nothing can be done for the DD Cube though.


When I first came up with the design for the evaporative cooler, I was hopeful that it would perform somewhere between the DD Cube and the Fluidyne cooler. As I came closer to building and testing it, I became skeptical of my calculations and I was crossing my fingers that it would merely perform on par with the Cube. However, as you can see, it unmercifully kicked the crap out of both the Cube and the Nascar oil cooler.

I actually expected the oil cooler to perform a little better relative to the cube. I suspect that fan power is holding it back. When the water temp is at 35C or so, the air coming out of the Fluidyne is warm; however the air coming out of the dd cube is not.

So how much does performance 7 times better than the Danger Den Cube cost? About 20-30 bucks, less if you are crafty or know the right people. So let’s get to how the heck this evaporative cooler works, and what you need to make one.

Steve Pignato

Evaporative Cooling

For my research project this semester, I am exploring various methods of each cooling stage in a typical water cooled computer, trying to find the best method for each. I analyzed three different types of evaporative coolers for the water heat exchanger. They were:

  • Bubbler type evaporator;
  • Dribbler, one that dribbles water over a surface;
  • Straight waterfall design.

The bubbler had a fatal design flaw that would keep it from cooling much more than a video card, and the dribbler design would probably give similar performance to a standard radiator but costs more to make. So what you see is the waterfall design. After searching on the internet for evaporative cooling info, I found that what I had made is in fact the very same thing as all standard industrial cooling towers! (Although much smaller).


“Look Ma! Just like a real nuclear cooling tower!”

Don’t expect to see this unless you are using a dual Athlon rig outside in the dead of winter. This steam plume only showed up while the water was over 40C. The effect was more impressive with the fan on, but I couldn’t get it to show up on this cheap digicam.

So to cool the water, it is pumped to the top of the tube. It then falls down through cool air and is collected at the bottom. It is so effective at cooling because of the large amount of energy required to convert water into a gas. To increase the water/air contact area, the water needs to be effectively dispersed. Industrial towers use what is called a water deck, but for this cooling tower a shower head is perfect.


Parts you need:

  • 2 or 3 feet of PVC pipe
  • PVC Cap
  • PVC Sanitary T
  • Shower Head
  • Hose Barbs
  • PVC Cement

For my cooling tower, I used 6″ pipe. I chose this because it was compatibly with the shower head I got, and a 172mm fan would fit perfectly into the pipe socket. However, you’ll see in a minute that such a large fan would be a bad idea. So depending on what shower head you get, I highly recommend 4″ pipe. A 120mm fan will fit perfectly onto the pipe joint if you choose to use one, and the overall appearance is much less daunting than 6″ piping.

The shower head is the most important part of the cooling tower, so choose wisely. So what should you be looking for? Well, Seinfeld fans should be thinking “black market elephant shower”. Everybody else should be looking for high-flow, many-holed shower heads. I bought my shower head from HERE. The “down under shower” has almost no flow restriction and 127 evenly spaced holes. It was only 13 bucks w/ free shipping, very good for a shower head. However, it’s 3.5″ in diameter, which means you probably can’t use it with 4″ pipe.

Anybody planning on making a 7′ tall 8″ 10″ PVC tower should check out the “euro-rainshower” from neat items. If there is such a thing as a black-market elephant shower, this is it. If you used it for actual showering, it would probably drain your hot water tank in about 30 seconds. I opted for the down-under shower because frankly, I don’t think my Eheim 1250 can push enough water for that head, and it is also 5.25″ in diameter.

I recommend you test lots of shower heads before making a final decision. Lots of shower heads that have a flow restrictor will cut flow to unacceptable levels and not disperse water correctly. If you just pick a shower head off the shelf, chances are there will not be enough flow after the water passes through your 3/8″ hose barbs. There is a shower head at my local hardware store (I tested all of them, it pays to know the hardware guy) that is perfect for a 4″ tower. If you guys want to see pictures of it or know the brand, let me know and we’ll post them.

Once you have all the parts, putting it together should be pretty straight forward. Try to pick a height that still has a fair amount of flow with your waterblocks in use. Depending on your pump, you should be able to put it inside of the pipe. My Eheim 1250 will fit horizontally into a 6″ pipe and vertically into a 4″ pipe.

Don’t throw away your DD cube or Big momma quite yet!

There are, of course, sacrifices you have to make for the performance of a cooling tower. First and foremost, the cooler will have to be mounted outside your case. The only case I can think of that might fit a cooling tower inside would be The Coldforge case, soon to be released by Faraday Cases.


Second, some people might not like the look. You may not want such a beast next to your computer, but I bet a lot of people will love it. My tower is just under 30″ made from 6″ pipe. Most of you will want to use 4″ pipe. The tower is probably still very effective between 18″ and 24″ high. Now I know somebody out there is going to make a quad-shower tower 7 feet tall. When you do, E-mail me! I want to see it in action.

Is it quiet? Not really, but it has a waterfall noise and I happen to like it. Most of you will probably agree with me – it’s a much more soothing sound than a Delta black label fan. I don’t have multiple YS Tech 120’s like some people do, but I would assume the waterfall sound is quieter than several YS Techs decibel-wise. I’m going to try installing a fine mesh window screen just above the water level to try and cut noise. I think this will be very effective, but I’ll have to see if I prefer to compute with or without the waterfall noise.

The biggest problem that may scare some of you away is the moisture. The air coming out of the tower is very humid. Not only that, but it gives out a very light mist. The mist was much more prevalent in the final testing that it was on some of my prototypes. I think this largely because I was using 5/8″ hose and barbs and was pushing huge amounts of water through the shower head. It was coming out at a higher velocity and made more mist. Even so, the mist is extremely light. During the fanless test, it was too light to feel with your hand – you had to put your face in front of it to know it was there.

I have a few idea’s to remedy this:

First is that fine mesh screen I mentioned earlier: I think a lot of the mist is coming from the splashing at the bottom of the tower.

Second, I suspect putting a fan filter (foam or aluminum mesh) at the output will catch most of the water. If you plan on using a fan for your tower, make sure it is an extremely weak one – you don’t need lots of air to get lots of cooling. The Panaflo L1a produced an unacceptable amount of mist at full power, but cutting power to 6 volts eliminated almost all of the mist.

Third, some water would collect at the lip of the pipe joint and sputter off a couple ml’s of water every 5 or 10 minutes. My final cooling tower is going to have a 22.5 degree elbow at the bottom so this will not be an issue.

Last, if you are planning on using a powerful fan, a 45 degree elbow at the top pointing away from your workstation would be a good idea.

I am really making this sound like a bigger issue than it is. I only ran into as much mist as I did during testing because I was pushing 211 gph through the shower head. In a regular system with 60 or so gph, mist will not be an issue. Build a tower and you’ll see for yourself.

The final problem is a minor one:

You’ll have to fill the tower with water every so often. How often depends on how big the reservoir is at the bottom and the humidity in your house. Because the water evaporates into the air, you are limited to the additives you can use. I would assume anti-freeze is out of the question, but you are going to need an biocide because germs will be sucked out of the air and into your water.

You are limited to using soluble solids to kill biologicals, but a quick conversation with my biology teacher showed me there is a very wide selection. You just need to make sure you choose one that won’t corrode your waterblock.


Obviously this solution is not for everybody, but those who are willing to take the plunge will reap the benefits. People with heat loads of more than a couple hundred watts should see very noticeable temperature drops. I would love to hear how a cooling tower works on your system, so please e-mail me and let me know. I’ll add updates on some of the developments I mentioned, as well as testing with a constant heat-load.

Steve Pignato

By popular demand, here are some answers to many of the emails I am receiving:

Q: What kind of temp drop can I expect on my system?

A: Ask decodeddiesel

Q: Would it be a good idea to recondense the
evaporated water and reduce coolant loss?

A: NO!! I cannot stress this enough – once the water
has evaporated, you want to get rid of it and not take
it back. It takes tremendous energy to convert water
into vapor, and to return it to liquid it takes MORE
ENERGY than it did to vaporize it. This is one of the
fundamental Laws of Thermodynamics. To recondense the
vapor, you’d need a cooling system MORE POWERFUL than
the cooling tower, and if you had one, you might as
well attach THAT to your CPU and throw out the cooling
tower. You guys are trying to create perpetual motion
which (I hope) you know is not possible.

Q: Can I use antifreeze or water-wetter with a cooling

A: Only if you feel like inhaling their vapors. Also,
eliminating the surface tension of the water in the
system would cause the droplets to be extremely small;
a fine mist would blow right out of the tower
with very little airflow.

Q: How far under ambient can it go?

A: In the test with 50% humidity, 6.4C, though I’ve
taken it as much as 8C under. It really depends on the
relative humidity in your area. Also, it will only go
under ambient under little or no load. I’m not sure
exactly what “little” is yet though, could be 5 watts,
could be 200.

Q: I like stealth mods, how small of a tower can I

A: I would say the practical limit for piping is
probably 3″; you’d need to have some innovative water
dispersion methods smaller than that. The shorter it
is, the less performance it will have, but it would not
surprise me if a 12″ tall 3″ tower still handily beat
a Danger Den Cube.

Q: Is it portable?

A: Yes! It’s not small, but it is light and sturdy. I
am building a case that has in/out hose barbs on the
back next to the IO ports. The barbs are above the
reservoir waterline. All I have to do is undo the
hoses, empty the tower, and throw it in my car. A
quick trip to the sink at the LAN, and I’m back in

Steve Pignato

After reading about Steve’s design, I decided to go out and make one for myself. I can remember looking at something like this in my Thermodynamics Text book and thinking “Gee something like that would dissipate a lot of heat, but would it work on something putting out as little wattage as my CPU Cooling system?”

Well, Steve proved without a doubt that you could. I had decided to do it, and after a quick trip to Home Depot the next day to pick up my materials, I was ready to go town. The materials I purchased, as well as their costs, are listed below:

  • PVC Piping…$20.00
  • Showerhead…$15.00
  • Fan…$10.00
  • Tap water…free
  • Building a system that will blow away any radiator system out there…priceless

I got 4″ PVC Piping, including one std tube, the sanitary tee, and an endcap, all of which came to about $20. Mine was about as exact to Stevens as you can get, except I used a bit more tubing below the sanitary tee to form more of a reservoir.

Steve's Tower

Steve Pignato’s Cooling Tower.

Then there was the showerhead. In buying the components this is what’s gonna’ make or break one of these things. The one I decided to go with is a Crometta-1 “Essential Full Shower” made by Hansgrohe. It can be found HERE.


I already had the fan (a 120mm Panaflow HA1) just sitting around from another project, so I decided to use it. Also, I bought the necessary PVC cement, the 3/8″ barbs I needed, and the fittings for the showerhead (which is ½” NPT by the way).

Eagerly awaiting results, I flew home from Home Depot (avoiding a cop or two along the way) and quickly got to unpacking and assembling my new toy.

After quickly setting it up and testing it for leaks, I fired her up while attached to H2O cooling system with my waterblock and peltier disconnected from my CPU. I am using a Danger Den MAZE2 with 3/8″ fittings cooling a 156 watt Peltier. I use this on the intended target for this system, a 1.0GHz AXIA which I have running smoothly at 150×10 =).

I let it run for 2 hours and, much to my delight, noticed a layer of frost QUICKLY forming. This was the great part…the bad part however – the flow rate wasn’t all that great with my 380 gph inline pump outside of the Tower, and most importantly, I had almost blown all of my coolant out of the tower.

Oh no, what happened??

Well it was pretty obvious: The 120mm, 105 cfm Panaflow had blown all the coolant out, and from what I figured, it was pushing 1 liter per hour out of the top!! Aside from these problems, this thing was working like a charm, I knew these were small bugs to iron out.

Solving the showerhead flow rate was nothing more than taking it apart and CAREFULLY drilling out the flow restrictor. Just make sure you don’t use it as a showerhead later, cause it definitely is above the 2.5 gph limit after this mod!! Well, more importantly, I thought of a few ways to cut down the loss of coolant.

I searched around my workshop looking for a few things. First thing I tried was placing a 92mm fan on it. While it worked OK, but I was still loosing coolant at an unacceptable rate; however, it was much better than before and it was cooling the coolant just fine. Well at this point I made a bold move and decided to hook it up to my system. OK, I wanted to go to default voltage, so I started her up at 1450. Oh my gosh…this thing really does work!

After a 1hour test to see if these temps were a fluke, I cranked her up to 1.5 GHz again. Well, to be blunt, it spanked my previous setup, the Danger Den standard. Essentially, the aforementioned components and a DD Cooling Cube (still love this setup, though, and will definitely be using it for my portable box). I kept my room at 21C and I took these temps after letting the system idle for 60 minutes, then stressed it with Prime95 and SiSoft Burn-In for 90 minutes. Yes I used the motherboard probe on my KT7A (I did bend it up so it’s touching), but it shows the point.

Temps with Danger Den cube “wind tunneled” with 2x120mm Panaflow drawing and pushing air:

  • Idle: -3C
  • Load: 17C
  • Coolant was about 10C-12C above ambient.

OK results, these results are at 150×10 with the old Danger Den Radiator rig.

Temps with the “Bong” Tower and a 92mm Generic fan:

  • Idle: -13C
  • Load: 6C
  • Coolant was 2C BELOW ambient.

Not bad!!!

Bad thing: After the testing, my reservoir was running low on coolant – damn, the fix didn’t work. I knew I was gonna have to get a bit more creative.

OK, this is when I remembered Steve mentioning a screen on the exhaust. Just so happens I had a spare aluminum mesh fan filter for a 120mm fan laying around. I cut a hole in it, fed the hose through it, and low and behold – it caught a TON of coolant. The airflow was hardly cut at all.

Also I couldn’t feel the “sprites” popping out like before. Steve and I found there are 2 tests to check this: The face and hand tests. They are exactly what they sound for the most part. If you can feel “sprites” with your hand, you are loosing WAY too much coolant. If you can only feel a little on your face when right next to it, and NONE on your hands, you’re good to go.

Water Trap

I found another place I was loosing coolant – around the fan. This was due to back pressure and the splashing coolant drops working its magic on coolant loss. I needed to drop the coolant level and add the 120mm fan. OK, I knew the HA1 was just too powerful at full blast, so I cut it to 5v and placed a little tape on it to restrict air flow even more.


Now with the showerhead flow rate mod (already done in the previous temps), the filter, and the 120mm fan barely pushing air…well lets say, this thing is absolutely amazing. I’ll let the temps speak for themselves.

“Bong Cooler” final mods and 120mm Fan
  • Idle: -14C
  • Load: 4C
  • Coolant 4C below ambient in reservoir.

Bottom line: I am now running stable at 1580MHz on a 1GHz chip!!!

I used to be running 1500 MHz max…these new temps gave me 80 more MHz!!!!!! This thing really works – we are seeing the infancy of something that is going to throw all the rules of H2O cooling right out the window. But then, what were rules besides something to be broken?

Wes Bower

SUMMARY: Powerful Tower cooling 12 inches high.

Steve’s Cooling Tower is fueling a feeding frenzy among Tower builders, and I’m no exception. I wanted to see if I could put one together that would be compact and yet deliver acceptable performance, and I think this design meets these criteria.

Desktop Tower

The finished unit – 12 inches high with an Eheim 1046 in-line pump and a Be Cooling waterblock.

The key is using ScothBrite Pads to maximize water surface area in a small space. By dribbling the water over a stack of pads, you get a lot of surface area over which the water evaporates. ScotchBrite Pads are made of plastic, so there is no problem with rust. They are also very porous, so getting air through them is possible, although you can’t use too many; the water does quickly impede air flow.

Let’s Build a Tower!

The pic below shows the basic building materials you will need: I used what’s called a WYE 4″ in diameter and about 8″ high (Home Depot). You will also need an adapter so you can screw in an end cap to close off the bottom – the end cap is the last item on the right. I had a bunch of ScotchBrite pads, but as it turned out, you may only need five or six 4″ diameter pads at most.


I spent about $30 in materials shown here.

First thing I did was drill a hole in the end cap and insert a nipple into it for the drain plug; this is where I attached to the intake side of the water pump. To do this, I drilled a 7/16″ hole in the side and tapped it for the nipple’s thread. You can also just drill the hole and screw the nipple into it.


I then attached the middle piece shown above to the WYE. This piece is threaded so you can attach the end cap to the main body. For this, you will have to buy some PVC Cement. This stuff is very quick drying and absolutely permanent. Dab some on both pieces to be joined, slip it on, give it a little twist, and in about 30 seconds, it’s glued together.

I did the same thing with the end cap; only this time I just applied glue on the end cap’s threads and then quickly screwed it in. By the time I was finished screwing it in, it was glued together – don’t take too long!

Next, I used an aluminum coil that Chip Eckert (Overclock-Watercool) had sent me a while ago, although no longer available. I drilled a series of 1/16″ holes into the coil


as shown below. This functions like the shower head that Steve uses in his tower and also holds the pads in place. I’m sure you can also use something like copper tubing for the same purpose.

Coil Holes

Then I cut a bunch of pads into 4″ circles and slit one side so I could intertwine the pads in the coil. The idea here is to get the water flowing onto the pads and then force air through the pads for evaporative cooling. With the large surface area, we should be able to get good performance in a package small enough to fit on the desktop.

Coil + Pads

Looking down into the tower, you can see how the pads fit into its body – snug enough so that air does not leak around the edges.


I then taped a Panaflo 92 mm fan I got from the Heatsink Factory onto the “Y”; this fan is rated at about 57 cfm @ 35 dBA. I also tried a more powerful Delta that Swiftech uses on its 462A heatsink; this is an 80 mm fan rated at 69 cfm and 48.5 dBA. The key difference between the two fans is rated maximum air pressure; the Delta spec is 15mm H²O while the Panaflo is 4mm H²0; this means that the Delta can push more air through the tower than the Panaflo, more than you would expect just based on cfm ratings.

I used an EHEIM 1046 in-line pump and a Be Cooling waterblock mounted on an ABIT KT7 running a Duron 800 @ 1000, 1.91 volts, 63.1 watts. A thermocouple is drilled into the base of the waterblock above the CPU to monitor CPU temps. I also recorded intake and exit air from the cooling tower. I measured noise 8″ from the fan’s intake.

Tower Test

I then ran Prime 95 to stress the CPU and recorded the following data:


CPU Temp

Ambient Temp



Exit Air

MBM Temp


Tower – Panaflo 92mm Fan

25.1 C

22.1 C

3.0 C


20.6 C

34 C

60 dBA

Tower – Delta 80mm Fan

23.8 C

22.4 C

1.4 C


18.7 C

33 C

72 dBA

Delta = CPU temp – Ambient Temp
C/W = Delta / CPU Watts

Interpreting C/W: For every watt the CPU radiates, the heatsink will cool the core by the (C/W x watts) plus ambient temp. For example, at an ambient temp of 25 C, a C/W of 0.25 with a CPU radiating 50 watts means that the CPU temp will be 50 x 0.25 = 12.5 C over ambient temp, or 37.5 C. The lower the C/W, the better.


Note especially that the air temp coming OUT of the tower is lower than the air coming IN.

The more powerful Delta did do a better job, but the NOISE! In my book, not worth it. I also found that I only needed five pads – as the water saturates the pads, getting airflow through them is tough. There was not a lot of air flowing out of the tower’s top. In fact, I felt back pressure at the fan’s intake. A more powerful fan obviously increases flow and performance (as the Delta showed) but with increased noise.

There are probably other materials that might better balance out airflow and water surface area, so I look forward to any help others can give on this and will post accordingly.

Some Caveats

OK, before we all go out and start building towers, we should realize that humidity plays a BIG role in cooling effectiveness. I measured the humidity of the air coming out of the tower at 80% and the ambient air at 60%. As the ambient air gets more saturated with moisture, cooling effectiveness decreases. Using this setup on a very humid day will not deliver the same level of performance as on a dry day.

Second, you really have to monitor water levels; the tower is a humidifier, which means you are pumping water into the air. I would add a “sight glass” to the tower; drill and fit an 90 degree elbow into the base and add a length of clear plastic tubing so you can see how much water is in the system.

Third, for stability, you’ll have to build some kind of base – a chunk of wood that the end can fit into will do the trick.


If you live in a dry climate, a cooling tower along the lines shown here can be an effective alternative to a radiator and a great conversation piece. If you are in air conditioning during the summer, then you can consider this as a year round cooling solution.

I’m sure there are a number of different approaches to building towers and look forward to seeing what folks come up with.

Email Joe


Here’s my “nuclear” tower. I used a showerhead that my dad got from work (He works at MOEN). I used 4 inch diameter PVC pipe. I glued a small section of pipe between the bottom cap and the “T” to form a reservoir. I left the top part of the “T” unglued so I can swap it out for a larger section of pipe. The tower is 3 feet tall with the short tube and around 4.5 feet with the taller section.


I used a 92mm fan grill with the center cut out to hold the showerhead.


I added a small section to the side part of the tower for the fan. This prevents water from getting on the fan or dripping out.


Next is the holder/stand. I got an old square bucket and cut a circle in the middle. The tower sits in the middle and doesn’t move at all.

For testing, I used a 120gph pump. I will be upgrading this to a Pondmaster Magdrive as soon as it arrives – expect a comparison between the pumps and the tower’s efficiency.

To test it, I heated 12 cups and 6 cups of water in a coffee maker. Both results are using 160F water and the exact same tower. A huge steam plume came out of the top until the water went under 95F. During the tests, my pump can’t really pump enough water, so the water kinda’ trickled out of the showerhead. I DID remove the flow restrictor.

The 3 foot tube with 12 cups of water took 35 minutes to cool down to ambient. I let it go for a couple hours and it dipped about 4F below ambient. With 6 cups of water, it took 22 minutes and went about 8F below ambient after left alone for a few hours.

I can’t test the 4.5 foot tube because my pump can’t pump high enough. 🙁 I also can’t test it under load because I don’t have a waterblock either.

If you have any questions, please email me.


Leave a Reply

Your email address will not be published. Required fields are marked *