Heat exchanger for a chiller?

[sir_pyro]

After rebuilding my computer, I not suddenly have the room the to put in a heat exchanger big enough to supply sufficient cooling. So, this means I can FINALLY pull my chiller out of the closet and run it on a computer. Now I just need to get advice on what to do...

For a heat exchanger, I have two options: Buy one, or build one.

Building one, I had originally planed on making a 9" long shell exchanger with 7 tubes. After talking with one of my professors about it for a little while, we ran the math and found out it wouldn’t be big enough for the cool the computer. Now that I have enough room, I can put in one of sufficient length and number of tubes, but I'd need to have the plates that the tubes go through machined, then find out a way to get it all sealed up. My old prototype one leeks like all heck, so it's a pain. this probably would be my best performing option though, because I could make the internal parts entirely out of copper and tune it to handle the heat load of my system.

Buying one, there are numerous plate HX's on ebay for about $100, but I know very little about them. One person seems to be selling these 14-plate SWEP ones every week or so that are small enough to fit in my case, but I dont really know enough about them to make an informed decisions. What would one of these do to my flow, and would the SS that the plates are made off cause me mixed metal problems?

Of course, I can just put in a 3-way valve to direct the flow out of my system, to the chiller to be pumped back in, but that involves using one single coolant (I wanted Zyrix in the main loop, normal AF in the chiller loop), and would hurt my performance when I'm running on the radiator.

[Pf.Farnsworth]

plate hx are the way to go if you have the money.

The only thing you need to know is they have to be positioned vertically in order to work. And if you want to use them to their full potential use crossflow. Refrigerant entry in the bottom exit top, coolant entry at the top, exit from the bottom.

[greenmaji]

Looking at the specs on different plate HX's it seems you only louse about 2psi of preasure pumped though them on average.

The 10 plate ones are rated for around 40K btu/hr that should be about 8-9X what you need (I can't seem to find smaller ones then 10 plate or I would be after one :thup: )

[sir_pyro]

Wait... You have to run them vertically?! I dont have the room for that.... Only place i have room is sitting horizontally along the bottom of the top chamber in my case. That doesn’t really make sense though, why the heck would it matter?

The cross flow thing, got that already. I did a lot of reading about HX's before i built my first one.

2PSI drop? Shoot... Even with the MCP655 I'm getting in on friday I'm not sure if I can pump that. Though If I put a bypass valve on it for when I run on air I think I may still end up ok. My mag 3 on the chiller might have some problems though seeing as its already pushing through two sealing quick disconnects and a TON of tubing. How much flow would that 2psi drop be at?

One last question... Wouldn’t a HX with more plates cool better? I know these thigns are rated with a lot more capacity then I really need, but those capacities are probably at a rather high temperature difference.

[greenmaji]

2 PSI drop is about the same as 2 90degree angles in 1/2 inch pipe, Im not sure if you got that :shrug:

From what I can tell it involves a change in phase (that would be why it needs to be in that orentation), and how in the world you fit a coil and coil in the case if you cant find a place to put a vertical plate HX :confused: how small is that case??

The smallest ones I can find are complete overkill. So a bigger one would just be a waste of money IMHO.

[sir_pyro]

I'm running a wal-mart chiller at the moment, so this is going to be water to water. Thus, I guess it doesn’t actually matter what way it's oriented (it does make sence though if you are running straight phase through it). Sorry I didn’t clarify that, but at least that was good info for when I rebuild the chiller from scratch down the road. It's an Antec P180 case, but it has quite a bit of stuff in it already, and thus the only space big enough that doesn’t have a fan or WC equipment bolted to it is the floor of the case in the top chamber.

My thing is I have to keep my computer so it can work off the chiller when it’s at home or a friend’s house, so I need a way to keep the normal loop separate from the chiller loop to maintain good performance on air. Thus, I need a HX to take heat out of the main loop and dump it into the chiller water when I’m back at school.

And wouldn't 2 psi be the same as 4.8ft of head? That's a lot seeing as my pump maxes out at 10ft.

[greenmaji]

google says 4.6ft/hd
question... why would you need a coolant to coolant HX?

[Pf.Farnsworth]

Yes you have to position them vertically. It has to do with the petterning on the faces of the plates and the general design. If you look at it carefully you will understand. The flow of heat is up and down. If you put the hx on its back then the heat wont be moving up and down the plates, rendering it useless.

By the way with the poor wallmark chiller cooling capasity, and then the water to water transfer your performance will be horrible. Dont even think its worth bothering with. You might as well get some pelts and call it a day.

The pressure drop due to the plate HX is very small and mostly due to friction as its alot less restrictive then the rest of your loop. That said any other hx will be worce.


The smallest ones I can find are complete overkill. So a bigger one would just be a waste of money IMHO.

It is only overkill if your coolant at full load when exiting the heat exchanger is the same temperature as the refrigerant evaporating on the other side. And it wont be. Going from 10 plates to 20 will help. And from 20 to 30 will as well. However its a diminishing gains curve, the gain is not linear to the number of plates. You can eventually archive thermal equilibrium but your hx will be pretty large.

[sir_pyro]

What I'm trying to do here is a little odd for extreme cooling. All I'm trying to do is build a cooling system that will lower my temps to just below ambient when I'm at my apartment (at college), but allow me to run on the internal radiators when I'm not. My goal coolant temps are only about 17c, so I don’t need that much performance. Thus, I need the coolant to coolant HX to allow the easy transition between air and chiller cooling the water in the main loop. To swap to chiller cooling, I’ll just have to plug in two quick connect fittings to the back of the case and turn the main radiator's fans off. Think plug and play chiller here.
Ok, on the positioning of the thing, isnt it the pump's job to move the coolant through and facilitate the movement of heat? All the HX is doing is transferring heat between two liquids that are moving on there own. When you go phase, the boiling gasses will want to go up, so that makes sence there.

[XeonStrikeForce]

Use a two coil, tube in shell exchanger with an intermediate exchange fluid.

One coil is the evap, it goes into a shell that is filled with a pure Anti frees or oil, any thing that can go ultra cold and not freeze. Coil too is submerged into this and sealed in. so the chiller cools the in between liquid, thus when you want to run the chiller just wrap the rad in a towel.

The only thing the shell does is hold the fluid there, nothing circulates inside it.

[Pf.Farnsworth]

What I'm trying to do here is a little odd for extreme cooling. All I'm trying to do is build a cooling system that will lower my temps to just below ambient when I'm at my apartment (at college), but allow me to run on the internal radiators when I'm not. My goal coolant temps are only about 17c, so I don’t need that much performance. Thus, I need the coolant to coolant HX to allow the easy transition between air and chiller cooling the water in the main loop. To swap to chiller cooling, I’ll just have to plug in two quick connect fittings to the back of the case and turn the main radiator's fans off. Think plug and play chiller here.
Ok, on the positioning of the thing, isnt it the pump's job to move the coolant through and facilitate the movement of heat? All the HX is doing is transferring heat between two liquids that are moving on there own. When you go phase, the boiling gasses will want to go up, so that makes sence there.

Well there is a better way of doing this. Why dont you just have the coolant exit your case and have like you said two quick connect fittings for in and out on the side of your case. Then you can either connect your loop to a radiator (internal or external, wherever you put it) or the chiller. That way you dont have to sacrifice chiller performance or the versitility. Just my 2 cents.

[sir_pyro]

Eh, I may just do that (in a way). I dont really want to push my main fluid through the quick connectes because of the high flow loss, but I'm starting to think that I should just put in a 3-way valve so I can switch on or off the chiller and not loose much of any performance. It would also only cost $17 for a NICE 3-way valve (built in fittings) vs $75 + fittings ($15?).

Only problem I have is this: How do I balence my system with such a setup? I'm going to have a MCP350/355/655 in the main loop (whichever I decide to go with eventually), and in the chiller I have a Mag 3 with another res. The quick connects are restrictive as all heck, so I'm not sure where I'd drop in the bypass/3-way system into the computer's loop to give me the best flow.

[XeonStrikeForce]

Schematic time, lets see a blu print of the system as it is and a blu print of what you want.

[sir_pyro]

Ok, I have four picutres here of what the heck I'm wanting to do. I'll laso include a "blank" one so if any one wnats to make one to point out somthing, all you have to do is "drag and drop" the parts. Nothe that there will be at least 5ft of 3/8" hose between the back of the comp to the inlets on the side of the chiller. All ovther tubing is 3/8" except for the feed lines for the pumps (1/2" there instead).

This is how things are now:
http://www.utc.edu/Walter-Hooper/Chiller_now.GIF

This is how things would be with a HX:
http://www.utc.edu/Walter-Hooper/Chiller_HX.GIF

This is one of my ideas on how this would be with a 3-way valve, any one elses ideas on placement of the valve-t inlet combo, tell me. 3-way idea:
http://www.utc.edu/Walter-Hooper/Chiller_idea.GIF

Then here is the blank one. The DDC is in there as an option to allow the D5 to go in the chiller. Also if any of you feel up to playing with this, arange the loop however (except in the chiller, that's fixed):
http://www.utc.edu/Students/Walter-Hooper/Chiller_blank.bmp

[Pf.Farnsworth]

Big no no to your last idea. If you are using the chiller then you will have to close out the radiator out of your loop, or else you will get condensation which will drop all over your nice computer.

More like this. After the hot coolant comes out of the gpu block it can either go to the rediator or the chiller, the controll is done by a two way valve. When the valve lets the coolant go to the radiator the chiller can be disconnected. And when the valve diverts the coolant to the chiller none will go though the radiator and cause condensation on it.

It would also be a good idea to do it like I did the in the wrawing where the incomming tube from the radiator is formming from the top and should be empty of water when the chiller loop is in use and the switch is in the chiller position, thus preventing the water in the tube from pulling the heat from the radiator and causing possible condensation.

http://img136.imageshack.us/img136/6928/chillerblankcopy3ug.jpg

[sir_pyro]

Looks like my school webspace went down...

[Pf.Farnsworth]

http://www.imageshack.us/ :rolleyes: