res. WHY?

nemaste · 06-20-2002, 10:24 AM

i've been doing some thinking about the methods people use in designing h2o systems & the subject of "res" has come to the forefront. i just knew this was the best forum to post this thread, so here goes:

i'm somewhat confused why so many ppl call anything containing a large amount of water a resevoir. some of these "resevoirs" are actually sumps & some are more so expansion tanks, but i have rarely seen a true res in a PC. from my understanding, a res is not a part of a flowing loop, but more so a stagnant "side storage" that allows for expansion & gives evaporation replenishment. the key in the definition, is that a res does not add to the effective cooling mass. therefore we should all say that resevoirs bite, but sumps & e-tanks own.

secondly, i'd like to ask a question about water mass (below). i was wondering if the water mass contained in a typical inline watercooled tower (only effective mass is in the ~10' of lines, rad, pump, & blocks) is enough to eliminate thermal spikes. if for some reason the rad could not remove the heat faster than the blocks add it, having extra cool water in the loop would give headroom that will allow me more time to react to the situation before the water gets too hot & things start to fry.

i wasn't sure if the small mass of h2o held in my inline system is already well into the law of dimishing returns or not (10' of 1/2" line, pump, blocks, rad). if not, i thought about adding an inline tank to add to the effective cooling mass. this would of course have no trapped air, but it will more so be a "fat" region in the loop that does nothing more than increase the total mass of circulating water. i know i could calculate the time rate of change in temperature using fancy math & detailed specs, but i just want preliminary thoughts on this subject.

please give reasons:
more mass is usually needed?
or
typical inlines already have mass overkill?

webmedic · 06-20-2002, 10:41 AM

Obviously the more h2o mass the better but In my systems using with a res and without it takes at least 5 minutes at full load to reach max temps.

Now this is using toast wich will run 2c hotter than prime95 and therefore will be much more of a load then any normal aplication you can find or run. This means that any normal apps working out your system will probably not get to max temps faster than 20 or 30 minutes.

Is this enough of a buffer for you? If not then you will probably want your fat area for more h2o mass. Remeber this is max, I whould probably say that under normal operating it would take at least 45min to reach max temps if a costant load is put on the cpu.

gmat · 06-20-2002, 11:23 AM

The only 'buffer' you might want ot consider is the static water mass in the CPU waterblock. In the case of a coolant loop / pump failure it's what will make the difference.
Now for a res in the coolant loop. It's only a 'capacitor' if you make the electrical analogy. It wont change your min/max temps (unless the res itself is cooled down, like a chiller or a rad) but just 'delay' them.
My personal taste goes to a small bubble trap ala Turbokeu. I like hauling my PC to LAN parties.

bigben2k · 06-20-2002, 12:09 PM

The additional water mass adds nothing but a delay in reaching max temps. Once max temp is reached is where your rig is running at the "balance point" where the heat from the CPU is dissipated by the rad at the same rate.

The reason that it takes time to reach that point, is that the rad needs to have a difference in temperature between the coolant and the air flowing through it.

Think about it: the instant that you turn on your computer (assuming everything was off), the coolant temp is the same as the ambiant air, so what's a rad going to do? Absolutely nothing, but when the coolant temp increases (thank you CPU!) then the rad can start doing some actual work.

As for spikes in temps, it can only be absorbed by the waterblock, if it allows the heat to be transmitted that fast (and believe us, you won't find anything better than copper for the price).

Temp spikes is not something to worry about, unless you're running some new weird type of cooling.

All in all, the heat flow is this:
CPU to [copper] waterblock
waterblock to water
water to [copper] rad
rad to air.

As for the res, it would be more appropriately called an air trap. It doesn't need to be a square box, but it does need to be able to seperate the air from the water, as the coolant passes through it.

Some people run a submerged pump, so they put it in the air trap, and call it a res.

Cyco-Dude · 06-20-2002, 08:04 PM

as for why ppl call it a res, its just a general term given to something that holds additional coolant. a 'res' can also double as an 'air-trap' as previously mentioned, where it also serves to help remove air bubbles from the system.

Brad · 06-21-2002, 04:22 AM

Quote:

The additional water mass adds nothing but a delay in reaching max temps. Once max temp is reached is where your rig is running at the "balance point" where the heat from the CPU is dissipated by the rad at the same rate.

like he says

the more water you have the longer it will take, but eventually it will get to the same temp

nemaste · 06-21-2002, 11:15 AM

great, thanks for all the advice guys; you all hit the bullseye of what i was asking.

i mentioned the word mass, because i knew i'd get the right train of thought going. i knew mass increases the temp rise time; analogous to a capacitor & voltage. from the answers i got & after thinking about it, the normal operating "spikes" i was thinking about are left up to the copper to "filter" out (like a PSU filter cap). i wasn't clear if the 30 to 40min heat up times assumed inital conditions of "normal idle" (no sleeping apps) or cold boot, but even the worst case 30min from cold boot is plenty (considering even the high temps would be not far from ambient).

as for the spike created by a failed pump, the change in thermal conductivity would preclude calling it a "heat spike" as opposed to "sub-system failure". this brought me to another question about redundant pumps:

I am starting with an ehiem 1200 pump, which i feel already has overkill power. if i wanted redundancy:
?will 2 series ehiem 1200 pumps create an obscene amount of unwanted water & case heating?

BTW,i guess i'm not alone when in my attempts to use better definitions to help get a point across more clearly. an air-trap is the more appropriate term (putting pumps in a box to make it a sump is a fringe benefit). i might need to reconsider my airtrap design, since at higher flow rates a vortex might suck air-bubbles back in the lines (like a bathtub drain sometimes does):

flowing R->L a vortex could form at the exit (left side). flowing the other direction might impede the flow somewhat. any ideas?

SCompRacer · 06-21-2002, 12:07 PM

My main reason for a plexi reservoir was to put lights under it and have the water do more than cool. I believe it is easier to fill and bleed the system too.

Due to the inline design of your res, I don't think you'd have to worry about a vortex forming. Just keep it full. I would also mount it on the suction side of the pump. In my first water rig, I soldered fittings on a horizontal mounted radiator for 1/8" pipe plugs on the radiator tanks opposite the inlet/outlet. I could remove the suction side while the pump was on to top it off without water surging out.

Now the placement of the fittings on my res invites a vortex if the res is not at least 7/8 full. Inlet is at the side, outelt at the bottom. I leave a slight airspace at the top so I get an everchanging light show from the air bubble moving around the top. At 3/4 full, I will get an attention getting air sucking vortex once it meanders over to the outlet and the pump gets a hold of it.

bigben2k · 06-21-2002, 12:17 PM

Glad to help.

That's a very original air trap, and I think it would work very well, maybe with a little tweaking...

As for the vortex, forget about it.
1-The vortex won't happen naturally, because there would be too much turbulence.
2-Vortex require a flow design to exist, and your air trap isn't designed to accomodate one.

As for your design, I'd recomend moddin it like this:
Tap it for a new water inlet (at either side).

nemaste · 06-21-2002, 06:14 PM

my bad, the pic didn't show how it functions very well. the below pic shows the inlet to the left, outlet to the right, & brass filling cap (i might reverse the inlet & outlets, which would result in the flow i erroneously thought might invite a vortex. there's just too much turbulence in the small hi-flow container & the inlet is perpendicular to the likely axis of rotation of the drain flow. a tangential inlet & large free volume (no baffles) would certainly invite the vortex demons into the tank.

the bad news: i filled it up, plugged the ends, & "simulated" a bleed. i had to shake the thing back & forth violently to get most of the air out. i might have to break my silicone seal & chop that stem down to a nut with my Dremel (AHHG!!! *crouches in fetal position protecting g401n*).

06-20-2002, 10:24 AM	#1
nemaste Cooling Neophyte Join Date: May 2002 Location: CA Posts: 78	res. WHY? i've been doing some thinking about the methods people use in designing h2o systems & the subject of "res" has come to the forefront. i just knew this was the best forum to post this thread, so here goes: i'm somewhat confused why so many ppl call anything containing a large amount of water a resevoir. some of these "resevoirs" are actually sumps & some are more so expansion tanks, but i have rarely seen a true res in a PC. from my understanding, a res is not a part of a flowing loop, but more so a stagnant "side storage" that allows for expansion & gives evaporation replenishment. the key in the definition, is that a res does not add to the effective cooling mass. therefore we should all say that resevoirs bite, but sumps & e-tanks own. secondly, i'd like to ask a question about water mass (below). i was wondering if the water mass contained in a typical inline watercooled tower (only effective mass is in the ~10' of lines, rad, pump, & blocks) is enough to eliminate thermal spikes. if for some reason the rad could not remove the heat faster than the blocks add it, having extra cool water in the loop would give headroom that will allow me more time to react to the situation before the water gets too hot & things start to fry. i wasn't sure if the small mass of h2o held in my inline system is already well into the law of dimishing returns or not (10' of 1/2" line, pump, blocks, rad). if not, i thought about adding an inline tank to add to the effective cooling mass. this would of course have no trapped air, but it will more so be a "fat" region in the loop that does nothing more than increase the total mass of circulating water. i know i could calculate the time rate of change in temperature using fancy math & detailed specs, but i just want preliminary thoughts on this subject. please give reasons: more mass is usually needed? or typical inlines already have mass overkill? __________________ StrongShockModz

06-20-2002, 10:41 AM	#2
webmedic Cooling Savant Join Date: Dec 2001 Location: Cheney, Wa Posts: 367	Obviously the more h2o mass the better but In my systems using with a res and without it takes at least 5 minutes at full load to reach max temps. Now this is using toast wich will run 2c hotter than prime95 and therefore will be much more of a load then any normal aplication you can find or run. This means that any normal apps working out your system will probably not get to max temps faster than 20 or 30 minutes. Is this enough of a buffer for you? If not then you will probably want your fat area for more h2o mass. Remeber this is max, I whould probably say that under normal operating it would take at least 45min to reach max temps if a costant load is put on the cpu. __________________ www.water-cool.com

06-20-2002, 12:09 PM	#4
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	The additional water mass adds nothing but a delay in reaching max temps. Once max temp is reached is where your rig is running at the "balance point" where the heat from the CPU is dissipated by the rad at the same rate. The reason that it takes time to reach that point, is that the rad needs to have a difference in temperature between the coolant and the air flowing through it. Think about it: the instant that you turn on your computer (assuming everything was off), the coolant temp is the same as the ambiant air, so what's a rad going to do? Absolutely nothing, but when the coolant temp increases (thank you CPU!) then the rad can start doing some actual work. As for spikes in temps, it can only be absorbed by the waterblock, if it allows the heat to be transmitted that fast (and believe us, you won't find anything better than copper for the price). Temp spikes is not something to worry about, unless you're running some new weird type of cooling. All in all, the heat flow is this: CPU to [copper] waterblock waterblock to water water to [copper] rad rad to air. As for the res, it would be more appropriately called an air trap. It doesn't need to be a square box, but it does need to be able to seperate the air from the water, as the coolant passes through it. Some people run a submerged pump, so they put it in the air trap, and call it a res. __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

06-21-2002, 11:15 AM	#7
nemaste Cooling Neophyte Join Date: May 2002 Location: CA Posts: 78	great, thanks for all the advice guys; you all hit the bullseye of what i was asking. i mentioned the word mass, because i knew i'd get the right train of thought going. i knew mass increases the temp rise time; analogous to a capacitor & voltage. from the answers i got & after thinking about it, the normal operating "spikes" i was thinking about are left up to the copper to "filter" out (like a PSU filter cap). i wasn't clear if the 30 to 40min heat up times assumed inital conditions of "normal idle" (no sleeping apps) or cold boot, but even the worst case 30min from cold boot is plenty (considering even the high temps would be not far from ambient). as for the spike created by a failed pump, the change in thermal conductivity would preclude calling it a "heat spike" as opposed to "sub-system failure". this brought me to another question about redundant pumps: I am starting with an ehiem 1200 pump, which i feel already has overkill power. if i wanted redundancy: ?will 2 series ehiem 1200 pumps create an obscene amount of unwanted water & case heating? BTW,i guess i'm not alone when in my attempts to use better definitions to help get a point across more clearly. an air-trap is the more appropriate term (putting pumps in a box to make it a sump is a fringe benefit). i might need to reconsider my airtrap design, since at higher flow rates a vortex might suck air-bubbles back in the lines (like a bathtub drain sometimes does): flowing R->L a vortex could form at the exit (left side). flowing the other direction might impede the flow somewhat. any ideas? __________________ StrongShockModz

06-21-2002, 06:14 PM	#10
nemaste Cooling Neophyte Join Date: May 2002 Location: CA Posts: 78	my bad, the pic didn't show how it functions very well. the below pic shows the inlet to the left, outlet to the right, & brass filling cap (i might reverse the inlet & outlets, which would result in the flow i erroneously thought might invite a vortex. there's just too much turbulence in the small hi-flow container & the inlet is perpendicular to the likely axis of rotation of the drain flow. a tangential inlet & large free volume (no baffles) would certainly invite the vortex demons into the tank. the bad news: i filled it up, plugged the ends, & "simulated" a bleed. i had to shake the thing back & forth violently to get most of the air out. i might have to break my silicone seal & chop that stem down to a nut with my Dremel (AHHG!!! crouches in fetal position protecting g401n). __________________ StrongShockModz

06-20-2002, 11:23 AM	#3
gmat Thermophile Join Date: Mar 2001 Location: France Posts: 1,221	The only 'buffer' you might want ot consider is the static water mass in the CPU waterblock. In the case of a coolant loop / pump failure it's what will make the difference. Now for a res in the coolant loop. It's only a 'capacitor' if you make the electrical analogy. It wont change your min/max temps (unless the res itself is cooled down, like a chiller or a rad) but just 'delay' them. My personal taste goes to a small bubble trap ala Turbokeu. I like hauling my PC to LAN parties.

06-20-2002, 08:04 PM	#5
Cyco-Dude Cooling Savant Join Date: Feb 2002 Location: Dayton, Ohio Posts: 836	as for why ppl call it a res, its just a general term given to something that holds additional coolant. a 'res' can also double as an 'air-trap' as previously mentioned, where it also serves to help remove air bubbles from the system.

06-21-2002, 12:07 PM	#8
SCompRacer Cooling Neophyte Join Date: Apr 2002 Location: Illinois Posts: 94	My main reason for a plexi reservoir was to put lights under it and have the water do more than cool. I believe it is easier to fill and bleed the system too. Due to the inline design of your res, I don't think you'd have to worry about a vortex forming. Just keep it full. I would also mount it on the suction side of the pump. In my first water rig, I soldered fittings on a horizontal mounted radiator for 1/8" pipe plugs on the radiator tanks opposite the inlet/outlet. I could remove the suction side while the pump was on to top it off without water surging out. Now the placement of the fittings on my res invites a vortex if the res is not at least 7/8 full. Inlet is at the side, outelt at the bottom. I leave a slight airspace at the top so I get an everchanging light show from the air bubble moving around the top. At 3/4 full, I will get an attention getting air sucking vortex once it meanders over to the outlet and the pump gets a hold of it.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)