Designing my house! Looking @ wierd alternative way for CPU cooling.

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

gamephile

Member
Jul 10, 2001
162
0
0
Oh man this idea has me foaming at the mouth.....well, maybe not foaming but it's definitely one of the coolest 'mod jobs' I've ever heard of.

I've read your plans and the suggestions of others and they all sound great. Unfortunately I don't have the experience nor the engineering/physics training to come up with any more suggestions. However there is one section of the project that I am interested that has yet to be discussed. And maybe it's so simple I'm not even seeing it but how exactly are the computers going to..ahem...'interface' with the water system. Are you going to put the boxes in a row or a circle or something and have plastic tubes run to a custom copper/plastic/PVC/lead/wicker pipe system? I find this intriguing as even people who are doing this sort of cooling system are, (as I've seen, and Plugers has so nicely linked us to) using only 1 computer. Pretty paltry when compared to 11....damm 11 computers....I have enough trouble getting the heat out of my one box.

[EDIT] - I just erased a paragraph where I asked you if you were going to have a backup system for the pumps should the power fail.....only after finishing did I realize water movement would stop in a power failure....however my feeble brain failed to realize that, so would the damm computers..Oy, it' is late. [/EDIT]

Anyways, Good luck Evadman, not only on this awesome project that makes me proud to be a geek, but also of course on building your new home! New homes are great in that you will most likely not have to worry about fixing crap around the house and worry about the real equipment...your computers! Oh yeah that family is in there somewhere too..jk

Jake


P.S. Can't wait to see this project on slashdot. You'd better log everything online! Hey, I've even got an idea for your next project, set up those 11 boxes as auto-load balancing web servers so you don?t get pummeled by click-happy /.'ers. Phew, ok I'm done, I mean it.

goodnight. or morning....?
Shut up Jake.
right.
 

Locutus4657

Senior member
Oct 9, 2001
209
0
0
You know, this would be a great problem for a GA.

Originally posted by: Evadman
Hey guys. I am actually going to start a topic that belongs here ;)

This is the story. I will be building my house sometime next year. I am currently designing it. This is something I want to do to cool the huge wattage that my fleet puts out.

Currently, I have 11 computers running in a 14 x 16 room. It gets extremely hot in here. In the beginnin of the year I touched on 120 degrees or so. I ened up adding a though the wall 5000 btu AC unit to help cool the room down. It did make a huge difference, but it never turns off. Sucking up the power :)

This is my plan. I want to water cool this fleet. I want to cool it using the naturaly cool ground. The water circuit I have in my head goes like this:

Water is held in a tank 4' or so undergound. From there, it is gravity fed into my basement, where the cooling water would be directed to each computer. From the computer, it would go into a water pit like a sump pump has, but much larger. In that pit will be 3 pumps that will raise the water up to 18" below ground level, and though a long series of pipes. Those pipes would be in a grid like patern ( smallest possable area wise )

After flowing though the grid patern ( where it gives up its heat to the gound ) it would return to the tank about 4' below ground.

I am looking to disapate somewhere in the area of 10,000 watts. ( I disapate somewhere in the area of 3k now, and I have to plan for the future :) )

I was looking for some imput on pipe diameter, location, and area of the "field" that would be needed. Remember to think 3d as I can go up ( within 18" of gorund level ) with no real depth limit, but 10' I would like to be max as that is the same height the basement will be excavated to.

I was assuming black or galvinized pipe for the pipe in the gound, and PVC for interior runs. ( not sure yet, as I may get a voltage difference )

Oh, and the pumps I have selected will pump about 1200 GPH @ a 10' head.

 

Def

Senior member
Jan 7, 2001
765
0
0
I see a few problems.

1. Your pumps are DEFINITELY not the right type of pumps to be using in a watercooling application. They sound like positive displacement pumps, and you do NOT want these for watercooling. They sound like bilge pumps or something like that. You could probably get by with a big version of the aquarium pumps usually used by watercoolers. Keep in mind this is a CLOSED system, so any pressure difference that the pump can create will be experienced by some point in the system. A resevoir will not "stop" the pressure from reaching the lines. If the pump is compressing the air to 25psi above atmospheric in there, that air will inturn pressurize the rest of the system. It just seems like any restrictions(i.e. barbs, waterblocks and hoses) will be experiencing way more pressure than they can handle. The whole concept behind a computer watercooling setup is to minimize restriction at every step to get good flowrate through the waterblock to maximize heat transfer. Increasing the pressure of the system also won't appreciably increase flowrate, except when it blows a hose off in your computer. ;)

2. You need to make absolutely positive your choosen type of pipe will not corrode underground or create a battery effect. You'd probably want to be looking at aluminium waterblocks since they are much cheaper than copper, and I don't think you'd be missing that 1-2C in your application. So keep that in mind when choosing your pipe material. Galvanatic corrosion is nasty, and if it can happen, it definitely will. While I'm thinking about temps, heck, you should get much better temps than most watercoolers assuming you can keep the water below 75-80F. Which sounds pretty easy.


I'd do it like this.

START
- 1/2 Big aquarium pumps
- 1" or larger inner diameter piping going in a "stepped" square rising up a few feet, (the diameter of this ring can be determined by your dimensions)
- come back into the wall to a resevoir that has a bleed hole at the higest point in the system(can also add water here)
- have a multitude of 5/8" quick disconnect fittings at the bottom of this resevoir
- then you can T those 5/8" fittings off into 1/2" line running to your computers
- I'd use 1/2" aluminium waterblocks for good flowrate and cost(maybe $15-20 max for each).
- Run that 1/2" return line to maybe another resevoir with quick disconnects
- that resevoir to your 1/2 pumps


Things you'd have to watch for
- You'd need a pump with a large inlet and outlet diameter. This is critical to good flowrate.
- Make sure you try to equate cross sectional areas. i.e. 1/2" tubing can flow 70% more than 3/8" tubing, so you can feasibly run 2 3/8" lines off of 1 1/2" line with good flow


Sounds like an interesting engineering problem, and I'd go here to get more help. ProCooling.com Some very knowledgable people in the forum, and some good info on the site to boot. :)

PM me if you want anymore suggestions. I've never actually done a watercooling setup(not enough money to do it right, and I move my computer every few months, ahhh.. downfalls of college life:(), but I've heavily researched things, and consider the efficiency of it a great thought game. I guess I should since I'm hopefully going to graduate as a Mechanical Eng. one of these days. :D

PM me if you want some more suggestions or thoughts. I might look around and see if I can find some useful equations to give you a simple analysis of the surface area required for the pipe in the ground to dissipate ~10-12k watts(always overengineer:)).
 

Belegost

Golden Member
Feb 20, 2001
1,807
19
81
Def, I think you misunderstood the design, the pump will not be pushing the water into the computers, the pumps merely push the hot water through the ground pipes to a holding tank. The holding tank gravity feeds the water to the computers, and then returns to the pump which is in a sump basin.

The only problem I can see is the flowrate due to gravity over the distances, You don't say how deep your basement will be, but I will assume not much deeper than 12', that leaves an 8' maybe 9' difference between the height of the sump basin(roughly) and the tank. 8' isn't going to get you a whole lot of velocity; the resistance inside water blocks is high, not to mention that for every computer there has to be at least a couple feet of tubing, plus whatever distance from the computers to the sump basin. I would leave room on the line that returns to the sump for an inline pump to help pull the water through.

The how big question is hard to answer, the number of variables is high. The heat transfer ability will be dependent on the flow rate through the tube, the surface area of the water/pipe contact, the heat transfer properties of the pipe, the temperature of the water, and the temperature of the ground, and probably more things I've forgotten. I would say as large as you can... :)

Also, I don't know if auto anti-freeze helps, but the galvanic differences between the metals used could become a problem, especially on such a large scale.
 

Evadman

Administrator Emeritus<br>Elite Member
Feb 18, 2001
30,990
5
81
The drop will be about 4' or so from the holding tank to the computers, then another 2-3 to the sump inside the basement floor.

Belegost: what Def is saying, is that if there is not a bleed hole in the holding trank, then the water that is being pumped in will presurize the air which will do the same to the water. I never even thought of that. I had that tank so that it would act as a pressure break. The pumps I will be using will be able to put out over 400+ psi if there is a blockage of some sort. That would instantly blow all the pipes in my basement :)

That tank outside was to act as a pressure break. Looks like I atleast need to add a vent to that.

I may also have to add a smaller pump ( or 2 for redundancy) to pump water from that tank to the computers to help aid flow.

For those who are wondering, the pumps are actually compercial duty sump pumps. Vane type, 100% duty cycle. 5 year warranty @ 100% duty. 1200 GPH with a 10' head ( water rise ) Max head was 30+ feet, but flow dropped to about 100 GPH. I am using 3 for redundancy, not for more flow. 500 GPH should be more than enough.

I was going to make my own aluminum water blocks for about $5. I have a metal worker who is going to help :)
 

Evadman

Administrator Emeritus<br>Elite Member
Feb 18, 2001
30,990
5
81
Also, I don't know if auto anti-freeze helps, but the galvanic differences between the metals used could become a problem, especially on such a large scale.

Autos use Aluminum heads and Iron blocks. AF should be enough, but I can always add some of that GM stuff that is supposed to remove all voltage differences in the water.
 

Ilmater

Diamond Member
Jun 13, 2002
7,516
1
0
The 70/30 Water/AF mix that I stated before will not erode aluminum or copper. 50/50 wouldn't either. Some things I guess at; other things I know...
 

Locutus4657

Senior member
Oct 9, 2001
209
0
0
Genetic Algorithm, once you know where the frost line is all you need to do is input the various paramaters into it and have a custom black box evaluate the solutions. There should be a pretty efficient solution to the problem fairly quickly this way. If you're really interested in what GA's are and how they work just do a google search on it. There's a wealth of information out there.

Carlo

Originally posted by: Evadman
You know, this would be a great problem for a GA.
GA?

 

Den

Member
Jan 11, 2000
168
0
0
There are lots of home heating and cooling sytems that work this way. Both with a grid of pipes buried fairly shallow, and with 2 pipes in a u shape buried in a single deep (100 feet or so) small (4 to 6 inches) hole. I would sugest you research that to figure out what kind of grid you need (IE, correct size and depth to disipate the amount of heat you make)
 

SWScorch

Diamond Member
May 13, 2001
9,520
1
76
Forgive me if I have overlooked this fact in the thread, unless it is missing as I presume.

How are you going to hook up the hoses to the computers? What I mean is, will you have a single hose coming from the holding tank, then running to the first computer, then to the second, then to the third, and so on, and then back into the tank? Or will you have a seperate inlet/outlet hose for each computer? The first setup would result in fewer hoses, but then the heat drawn from each CPU would go to the next one, and so on, so that by the time the water got to the last CPU, it would be pretty hot from all the heat it absorbed, and the last few CPUs wouldn't recieve much benefit.

I assume you've taken this into consideration; I'm just curious as to how you will set this system up.
 

Jerboy

Banned
Oct 27, 2001
5,190
0
0
Originally posted by: Evadman
Hey guys. I am actually going to start a topic that belongs here ;)

This is the story. I will be building my house sometime next year. I am currently designing it. This is something I want to do to cool the huge wattage that my fleet puts out.

Where do you live? Depending on the climate, good ventilation might do.


Currently, I have 11 computers running in a 14 x 16 room. It gets extremely hot in here. In the beginnin of the year I touched on 120 degrees or so. I ened up adding a though the wall 5000 btu AC unit to help cool the room down. It did make a huge difference, but it never turns off. Sucking up the power :)

This is my plan. I want to water cool this fleet. I want to cool it using the naturaly cool ground. The water circuit I have in my head goes like this:

Water is held in a tank 4' or so undergound. From there, it is gravity fed into my basement, where the cooling water would be directed to each computer. From the computer, it would go into a water pit like a sump pump has, but much larger. In that pit will be 3 pumps that will raise the water up to 18" below ground level, and though a long series of pipes. Those pipes would be in a grid like patern ( smallest possable area wise )

After flowing though the grid patern ( where it gives up its heat to the gound ) it would return to the tank about 4' below ground.

I have a feeling that such underground facility requires a permit. If it does, who knows if such plan would be approved.

I am looking to disapate somewhere in the area of 10,000 watts. ( I disapate somewhere in the area of 3k now, and I have to plan for the future :) )

If you mean 10KW from computers, you have some serious electrical planning to do as well. Computers have a power factor of ~0.6, so to get 10KW of high tech loads from 120V, 10,000/120/0.6=140A. You're going to have to install 7 20A circuits to accomodate this load.

You should start looking into pulling a 208Y/120V 3ph service. Most computer labs are setup this way(schools, library, office, etc). As a matter of fact, they have 208Y/120V exclusively for 120V loads since not many items use 208V. Of these, computer and peripherals represent large portion of 120V loads.

Once you have 3ph service you can use air conditioners, water pumps, shop machinery utilizing three phase motors. They're quieter and more efficient.

BTW, your cooling system needs to be able to handle about 35,000BTU just for the computers. What kind of min. temperature are you expecting?
I was looking for some imput on pipe diameter, location, and area of the "field" that would be needed. Remember to think 3d as I can go up ( within 18" of gorund level ) with no real depth limit, but 10' I would like to be max as that is the same height the basement will be excavated to.

I was assuming black or galvinized pipe for the pipe in the gound, and PVC for interior runs. ( not sure yet, as I may get a voltage difference )

Oh, and the pumps I have selected will pump about 1200 GPH @ a 10' head.

You might want to consult a professional on that. You don't want to make a mistake that costs you 10' of digging.


 

Evadman

Administrator Emeritus<br>Elite Member
Feb 18, 2001
30,990
5
81
Where do you live? Depending on the climate, good ventilation might do.
Outside Chicago. Vetilation wors just fine for the winter months. I just leave the windows open. Summer can get nasty.
I have a feeling that such underground facility requires a permit. If it does, who knows if such plan would be approved.
Already taken care of. I talked to the counties around here. It is possable, but I have to follow certain guidelines. I can fit within them without a problem.
If you mean 10KW from computers, you have some serious electrical planning to do as well[/q[
Heck yes I do :) I currently have my house wired for 300a 2 phase, with a 60 amp sub pnl for my computer room. I have blown it once by plugging a vacuum into the wrong outlet. ( blew the 60a breaker ) I will look into 3 ph. Tahnks!
BTW, your cooling system needs to be able to handle about 35,000BTU just for the computers. What kind of min. temperature are you expecting?
I was looking for something in the rance of 40-45c max. I could live with 50c but I would not want to. I hope to get it down to 35c or so, but that is pretty optomistic.
You might want to consult a professional on that. You don't want to make a mistake that costs you 10' of digging.
that's why I am asking here :) the advice is free and good! What else can I ask for? :D

For those that asked, the distrobution to the computers will be though quick connect fittings in the back of the cases. It will be a parallel setup. A large ( 1 - 1 1/2" pipe) will feed into the basement where I will have "taps" for the computers. From there the water will drain into another large pipe that collects the waste water and drains it back to the pit.
 

Jerboy

Banned
Oct 27, 2001
5,190
0
0
Heck yes I do :) I currently have my house wired for 300a 2 phase, with a 60 amp sub pnl for my computer room. I have blown it once by plugging a vacuum into the wrong outlet. ( blew the 60a breaker ) I will look into 3 ph. Tahnks!

Most houses do not have a 2ph. Residential services are usually fed from a single ph center tapped x-0-y transformer. 0 to either is 120, x to y is 240. So this is single, split phase(180 deg out of phase, even though opposite, not two phase)

Commercial/heavy user power is p-p-p-N p=phase, N=neutral, this is the common 3ph 4 wire system.
between any two p's=208V, any p to N is 120V. When you have two of three and N it's 2ph(each phase is 120deg apart)



BTW, your cooling system needs to be able to handle about 35,000BTU just for the computers. What kind of min. temperature are you expecting?
I was looking for something in the rance of 40-45c max. I could live with 50c but I would not want to. I hope to get it down to 35c or so, but that is pretty optomistic.[/quote]

How about using an old A/C cooling unit with a blown compressor?
Modify the cooling coil a little so that compressor is bypassed. Then feed the water through condenser coil. You're going to use the affixed fan to cool water instead of refrigerant. If you call HVAC contractors, I'm sure they can hook you up with a A/C outside unit that has been taken out of service due to system upgrade or compressor failure.

Flush the tube out with acetone to remove all refrigerant oil before use, then with water and it should be ready to go. Dig a 5' hole in the ground and use a five gallon bucket as a thermal reservior.

Install a probe inside the reservior and you can control the fan (i.e. 45deg C on, 40C off).

The purpose of reservior is to ensure stability and prevent frequent cycling of fan.

Make the rest of piping PVC.

 

rgwalt

Diamond Member
Apr 22, 2000
7,393
0
0
How about using an old A/C cooling unit with a blown compressor?
Modify the cooling coil a little so that compressor is bypassed. Then feed the water through condenser coil. You're going to use the affixed fan to cool water instead of refrigerant. If you call HVAC contractors, I'm sure they can hook you up with a A/C outside unit that has been taken out of service due to system upgrade or compressor failure.

Flush the tube out with acetone to remove all refrigerant oil before use, then with water and it should be ready to go. Dig a 5' hole in the ground and use a five gallon bucket as a thermal reservior.

Install a probe inside the reservior and you can control the fan (i.e. 45deg C on, 40C off).

The purpose of reservior is to ensure stability and prevent frequent cycling of fan.

Make the rest of piping PVC.

Jerboy- Are you talking about dumping the heat into the surrounding air via the radiator on the AC unit? I don't know how practical this would be. The water going into the water blocks would have to be cooler than his desired temperature. It also must be hotter than the temps outside when it leaves the block. This wouldn't be a problem in the winter, but summer-time temps in Chi-town could be a problem. Right now the weather is great, but it doesn't necessarily stay that way. Also, air coolers (such as AC units) need a decent temperature difference to transfer heat effectively. This delta T should be at least 5 or 10 degrees C. The water blocks themselves would need a delta T of between 3 and 5 degrees C. So, if the outside summer temps didn't get worse than 40C, the water coming out of the AC unit would probably be around 43 C to 45 C. This would mean the processor temps would probably be around 46 C to 48 C. The water must stay between the operating temperature of the processors and the outside temperature.

However, I suppose these pinch-point problems could be solved by putting Peltier coolers on the processors and then using the water cooler to cool the Peltier's. However, if a cooler went out during the summer, you would fry the processor. The whole idea behind burying the cooler underground is to dump heat into a cool environment. If evadman lived near a decent sized lake or a small river, he could use that.

Ryan
 

Jerboy

Banned
Oct 27, 2001
5,190
0
0
Jerboy- Are you talking about dumping the heat into the surrounding air via the radiator on the AC unit? I don't know how practical this would be.

Yes I am and I am not sure what kind of ambient temperature it's going to experience in his area, so it might not be practical. It is a very economical solution to digging 10' of ground and bury piping if workable.

The water going into the water blocks would have to be cooler than his desired temperature. It also must be hotter than the temps outside when it leaves the block. This wouldn't be a problem in the winter, but summer-time temps in Chi-town could be a problem.

He is looking for 45C ish water temp I believe.

Right now the weather is great, but it doesn't necessarily stay that way. Also, air coolers (such as AC units) need a decent temperature difference to transfer heat effectively. This delta T should be at least 5 or 10 degrees C. The water blocks themselves would need a delta T of between 3 and 5 degrees C. So, if the outside summer temps didn't get worse than 40C, the water coming out of the AC unit would probably be around 43 C to 45 C. This would mean the processor temps would probably be around 46 C to 48 C. The water must stay between the operating temperature of the processors and the outside temperature.

If the outside temp won't go above 40C and stays in the high 30C's it should work.

However, I suppose these pinch-point problems could be solved by putting Peltier coolers on the processors and then using the water cooler to cool the Peltier's.

How about a laboratory/light industrial/aquarium water chiller? Do you have any idea how inefficient Peltier is? The rate of absorptin is about 0.5x the input(It takes 100W of electricity to suck 50W of heat). A decent refrigerant cooling system has cooling factor of 2.0x.

A chiller has water input and output. Some model has a builit in pump. you simply pull the plumbing into the building, and bring the return to the unit. Evadman can build a small shed around the cooler to protect from weather. I would say unit go for $1000-2000, a fairy good competition pricewise against burying pipes capable of releasing 30,000BTU 10' down the ground.

To keep the power cost down, it will be supplemental to modified AC cooling unit(the AC unit is simply used to air cool water, refrigeration system is bypassed).

Using two thermometers, one measuring water temp and other measuring ambient air temp the system is quite easily controlled. When ambient temperature exceeds desird temp-whatever Delta T specified, the control circuits shuts off the cooling fan and turns on the chiller. Once the day cools down, chiller is shut off and cooling fan cools the water.

So it needs to be programmed like this:
assuming required delta T between water temp and air for air cooling to work is five deg and desired water temp is 45 to 50:

ambient temp <40C, water temp 50C, activate cooling fan. Fan off when water temp goes down to 45.

ambient temp >40C, water temp 50C, cooling fan off, chiller on. when water temp goes down to 45, chiller off.

after water temp has reached 45C and while it's climbing up to 50, both chiller and cooling fan off.


Update: heheh I came up with my lovely diagram:

http://www.geocities.com/jerboiat/untitled.jpg

Sorry about the piss poor quality. It somehow became like that when I saved it.


However, if a cooler went out during the summer, you would fry the processor. The whole idea behind burying the cooler underground is to dump heat into a cool environment. If evadman lived near a decent sized lake or a small river, he could use that.

Ryan

A well engineered system will not have such problem. How do you think nuclear plants prevent itself from going kaboom in case of a cooling system failure?


The use of flowing stream and public body of water of any size for a private purpose is purely an approval nightmare.


Evadman, another option is thermal storage reservior. You can use something like a 500liter insulated reservior.

To raise it's temp by 10C it takes 10463kJoules.
500,000cc and delta T=10C=5million calories=20927kJ.
Say each of his computer puts out 200W and he has 11 units.
2200watts=2.2kJ/s. It gives his computers about 3hrs of cooling on reserve.

That's 3hrs through above water temp ambient temp.

By using a fan unit capable of dissipating alot more than the computer produces, you can cool the water at night for day use by letting the fan run through out the chilly night.


 

rgwalt

Diamond Member
Apr 22, 2000
7,393
0
0
Jerboy-

Your idea of using a chiller is a good one. However, if the chiller goes during the summer, he will cook all of the computers. Sure a pelt is inefficient, but the likelyhood of all of them going out at once is very low.

Also, one thing you are forgetting is that you have to have a temperature difference to transfer heat. I don't know what home AC units are rated for, but industrial units need a 20C temperature difference for good heat transfer. If that difference isn't achieved, the unit won't transfer enough heat to drop the temperature the desired amount with the given surface area. In the winter it won't be a problem, but in the summer you could get a pinch-point during the heat of the day.

One more thing... Evadman will need to look at the pressure drop vs flow rate across the radiator. The refridgerants used in AC units are much less viscous than water. Using an old AC unit is a good idea, if the summer climate would cooperate.

Ryan
 

Jerboy

Banned
Oct 27, 2001
5,190
0
0
The pumps I will be using will be able to put out over 400+ psi if there is a blockage of some sort. That would instantly blow all the pipes in my basement :)

This is a good time to consider adding a safety relief valve.



For those who are wondering, the pumps are actually compercial duty sump pumps. Vane type, 100% duty cycle. 5 year warranty @ 100% duty. 1200 GPH with a 10' head ( water rise ) Max head was 30+ feet, but flow dropped to about 100 GPH. I am using 3 for redundancy, not for more flow. 500 GPH should be more than enough.

Why are you using a vane pump? They're not as efficient or reliable as a centrifugal type. Warranty means they'll replace the pump, not pay you for consequential damage.

GPH? What an icky unit. Hourly rating is too wide of a time frame isn't it? I'm more familiar with liters/min. even 500GPH is an overkill. I read somewhere that excessive water flow has negative influence on heat transfer, because it doesn't give enough time for heat to get into water. 500gph is 31.5l/m which is about 3 times the flow of typical aquarium pump.

Just how much power is your pump going to use?

 

rgwalt

Diamond Member
Apr 22, 2000
7,393
0
0
I read somewhere that excessive water flow has negative influence on heat transfer,because it doesn't give enough time for heat to get into water.

High flow rate doesn't impact heat transfer performance significantly. However, 500 GPH is pretty high.

Ryan
 

lbmcleod

Senior member
Jan 30, 2002
210
0
0
get some of that supter-cooled liquid metal stuff (non-conductive e.g argon) an shove ur computer in there!!!
 

Evadman

Administrator Emeritus<br>Elite Member
Feb 18, 2001
30,990
5
81
Just how much power is your pump going to use?

They will pull about 6 amps @ 120v @ full 27' head.

I chose these specific pumps the the warranty. The reason I am getting 3 is for redundancy, not more flow.