Drawbacks to using two reservoirs?

[nascasho]

Before someone thinks I'm a dur durr for even thinking this, hear me out first lol.

I need to explain to you my setup first then explain why I want a second res. Here's a pic I took a couple months back, it's basically the same loop right now, but build is different:

http://nascasho.com/Photos/ReVOLT/ReVOLT3.png

Looking at the picture give you a general idea of how it's setup. The pump (MCP355) is mounted to that dual bay XSPC res which is a little hard to see in the picture. The outlet of the pump goes straight to the GPU and so on and so forth onto the other parts. Initially when I was first putting the build together I wanted a tube res to be mounted right next to the MB and have my 24-pin wrapped around it, but hours of drawing configurations and this ended up being the choice mainly because it would've looked the best in this case since I wanted a clean look.

Now earlier I was just brainstorming with one of my buddies on reservoirs and then it hit me, could I use two reservoirs, keep my setup, and still have it look even cooler? I was thinking to get rid of the bay res and getting a top, keep the pump in the same place. This introduces a fill issue because the bottom of the tube res would be below the pump, so when I fill the res I'm just asking for trouble. I could put the pump below the PSU cover, but then that MIGHT make draining a little annoying. Now if I were to put the tube res between the pump and the GPU (Bay Res> Pump> Tube res> GPU...) would that cause any issues? I'll be filling it the way I always have, from the Bay Res. Also, the bottom port on the tube res will be where the water will enter, horizontal ports will be where it'll exit to the GPU.

Need a second or more brains to help me figure this out and decide. It's purely eye candy the reasoning behind this, thanks ahead of time. :)

[SuperSilent]

you can do it but you loose the energy in the flow when it slows on entry to the 2nd tube res and then is re-accelerates on exit so loop flow rate will drop slightly. secondally the tube res will be pressurised, not by much but it wont be usefull as a bleed/ fill res. as opening it to fill with the pump on will cause it pump the fluid in the 1st res out the top of the 2nd and not round the loop.

[PiLsY]

Rubbish. No effect on flow - its a closed loop. Ignore everything SuperSilent posted.

I use 3 reservoirs in my loop (mcr220-res, pump top res and alphacool mini res) and it makes it way easier to fill and bleed. Go for it.

[meanmoe]

you can do it but you loose the energy in the flow when it slows on entry to the 2nd tube res and then is re-accelerates on exit so loop flow rate will drop slightly.

This is not true... velocity will drop proportional to the increase in cross-sectional area and then increase again proportional to the decrease in cross sectional area...

@op, Just fill it. it will look cool. Don't worry about the bottom of the res being below the pump, just make sure the water level stays well above the pump and it will be gravity fed.

[SuperSilent]

hmmmm..... perhaps i didnt explain my point clearly

Yes the flow has to obay the law of continuity and therefore the flow rate through the two pipes (into res & out of res) has to be the same. BUT because the inlet to the resivoir isnt a well designed diffuser the flow will separate on entry to the resivoir tank and loose enery through turbulance (there is also an energy loss associated wiht the outlet from the res not being a bellmouth), this enrgy has to come from the pressure head of the pump therfore by installing the 2nd res the overall loop flowrate will drop. crucially this drop will be very small

see: http://en.wikipedia.org/wiki/Borda%E2%80%93Carnot_equation

if you want to actually do the calculations

[PiLsY]

Pressure head is irrelevant in a closed loop once youve done the bleeding and filling.

Reservoirs add no appreciable resistance to a loop.

Inlet pressure at the pump can affect flow, yes, but you will be hard pushed to make a noticeable change here by adding a second reservoir anywhere in the loop except directly at the pump outlet.

[SuperSilent]

pressure head is massively relevent in a closed loop, in an equilibrium state the pressure [head] jump accross the pump HAS to be equaled by the head loss across all the fittings and constrictions in the loop [blocks etc.] otherwise there would be a resultant force on the liquid and it would accelerate.

Anyway your not going to beleave my argument so here is some hard data from martins liquid lab.

http://img403.imageshack.us/img403/5760/35xreservoircomparison.png

notice the loss in dynamic head avalable from the green line [pump only] to blue line [Pump + Extra Resivoir] you will see a similar additional head loss going from 1 res to 2. this reduction in dynamic head will reduce your loop flow speed.

In short - if you just want it for looks go for it, but many people in these forums will spend $$$ to get a new waterblock that is 0.5oC better than the one they had and i just wanted to try and make you aware of the potential minor impact it will have on your system incase you cared about that. Will you notice any difference in you overclock? NO. but it will have a fractionally lower flow rate.

[Polizei]

If you read the review for the Swiftech 35X pump top/res, you would have read that Martin never used TWO reservoirs in the same loop for that test. He used a reservoir with his modification (the down tube like that found in the XSPC res top), the pump only with probably a T-line, the pump with Swiftech's tube res and the additional sponge, and the pump top with a standard tube res. Never did he use two reservoirs in any test, so you can't use that to further your argument.

It anything, the additional tube res the OP wants to use will act as a minor restriction, altering flow NEGLIGIBLY, and thus negligibly altering temperatures.

[meanmoe]

@ Supersilent - very good. I blew the dust off a John D Anderson book. I'm not used to being schooled. However, the velocities are so low though that I doubt the effect will be noticed, ie negligible. Still, point taken about the 0.5deg C delta argument although I seriously doubt that it will make that type of difference, the energy loss is so small due to the low inlet velocity.

[nascasho]

Wow, you guys know a lot more than I thought about water cooling!

I really appreciate the responses even though a bunch of it went way over my head lol. I'll definitely give it a shot though. Worse case scenario, if it does do what Supersilent says, if the water peaks out at the top of the res, maybe it'll even out or something?

I was going to use this res (http://www.performance-pcs.com/catalog/index.php?main_page=product_info&cPath=59_318_666&products_id=28835) since I think it looks the coolest imo with all that shiny-ness.

Again guys I really appreciate the help.

[meanmoe]

if I was buying another tube res, it would be an iandh hands down.
http://www.sidewindercomputers.com/iast225mure.html
http://www.sidewindercomputers.com/iast225mure.html

Second choice would be EK although I've not owned one.

[PiLsY]

People seem to forget were pushing water through about 5 feet of 11mm ID tubing, not trying to empty a flooded mine 2km below the earths surface :D.

Because of the scale of what we do you really dont need to overcomplicate stuff.

If this was relevant we would have issues with running multiple rads (an end tank is for all intents a reservoir and you have 3 per rad).

You wont be able to measure the difference between using multiple reservoirs with LC flowmeters and temp gauges. I doubt even martin/skinee would be able to measure such differences on their test rigs.

[Petra]

People seem to forget were pushing water through about 5 feet of 11mm ID tubing, not trying to empty a flooded mine 2km below the earths surface :D.

Because of the scale of what we do you really dont need to overcomplicate stuff.
So true...


You wont be able to measure the difference between using multiple reservoirs with LC flowmeters and temp gauges. I doubt even martin/skinee would be able to measure such differences on their test rigs.
The pressure drop incurred by a reservoir is easily measurable (Martin and I have both done such tests) and that information can be used to determine the reservoir's impact on a given loop's coolant flowrate. That said, the impact on flowrate from a reservoir is usually negligible... so, unless you're looking to run like 50 reservoirs in your loop, it really doesn't matter.

[PiLsY]

Ah so it is actually measurable then - fair play. I'll take that one on the chin! ;)

[Annoyingmouse]

How about a plugged, non functional res? Fill it with whatever fluid looks cool with no worries about it's affect on the loop.

[nascasho]

How about a plugged, non functional res? Fill it with whatever fluid looks cool with no worries about it's affect on the loop.

I was thinking about that and it's an awesome idea, but the XSPC is an absolute headache to bleed, so might be able to run the tube res with the top off and maybe make my life a little easier. To bleed the XSPC res I use a cut up piece of tubing and cover the inlet hole to prevent the bubbles from getting in, still wonder why they designed it like that.