Yeah. There's no chance that this is going to work out. They're planning on hanging it out a window, basically. That shit's gonna freeze. Did find out the rads on each system is so it can be disconnected and work independently, though. Thats a good idea.
But with them moving soon, and The freezing temps, and the ridiculousness of the set up, i dont see it ending well.
Yeah, and on hot days the system really wont do much to remove heat either.
At the scale they are working at they need to use real chillers/heat exchangers not consumer grade stuff.
Nice spending all the money we donated on something completely useless and extravagent
So I looked up the parts prices it would have cost around 500-600 for a chiller. About the same price as the their radiator/fan setup. and running one well water pump would have been cheaper than all those case pumps.
^ Agreed.
As well as something that will actually Slow down the thing they asked for donations for, like moving. Why set all of this up just to move to a new office?
Do they not have any engineers in their forums they could have consulted when planning this project?
I wouldn't doubt that there are none there, honestly.
I guess they will have to learn their lessons the hard way.
I'm pretty sure they said at some point that the radiators on the PCs are not being used while they are in the big loop, they are just there so you can pull the PC out of the loop and connect it to its own radiators (so each of them can still work as a standalone without the big loop). Mind you I have clue about watercooling and I couldn't tell you how exactly they're doing it.
The bare copper piping, yeah that's far from ideal. But it looks really fancy and I don't think that the actual objective here was to make the room cooler but rather make a series of videos about something silly and interesting.
Regarding the pipes freezing: Haven't they mentioned that they barely get negative °C where they live? If they leave the pumps running then I don't think it's going to be a huge issue.
1. It was mentioned earlier in this thread that those radiators are only used when separate from the main loop.
2. They have stated repeatedly since the beginning was to make the room cooler. I guess the looks of bare copper is subjective, I would personally rather not see any piping. I'm sure the ability to make the video provided some motivation.
3. They live in Canada, their radiator will freeze. It only takes one day of unusually cold weather to freeze the coolant and burst the radiators. Even on days where it doesn't freeze, the coolant will be chilled sufficiently to cause condensation to form on the piping and inside the computer cases on cooler days.
Didn't read the whole thread, sorry.
He has also stated repeatedly that a simple AC unit would do the job, and that this is highly inefficient. It's one of those "because we can" things, I'm fairly sure of it. Not to mention that they're getting a complete makeover for all their workstations (including a hyperthreaded 12-core XEON) from their sponsors + the add revenue from the video series. They'll probably also gain a considerable amount of subscribers over this since it is fairly unique and attention-grabbing (and it seems to work since we're talking about it on a completely different tech forum).
Vancouver (where they have the office) has a pretty mild climate: http://en.wikipedia.org/wiki/Climate_of_Vancouver Just because it's Canada doesn't mean that it's freezing cold all the time, heck it gets colder where I live. A bit of antifreeze should keep their coolant (and their pipes) safe. Condensation, on the other hand, might be a serious issue. Once the PCs are running all the coolant will probably be warmer then their ambient temps, but over night I definitely see this as an issue.
I understand not reading 3 pages of comments, just letting you know I was aware of the purpose of those rads.
As for the freezing, yes vancouver is mild, but it still will get down into the teens. Don't be fooled by the "Average" lows. Many days are colder, and warmer, than the average day of the month. Anti-freeze would work, but considering how they haven't bothered to insulate their copper piping, I don't think they've considered putting anti-freeze in their coolant if they haven't mentioned it.
I would like to see a build log that documents the kind of planning necessary to successfully complete some of these more ambitious projects. Part of the problem I see with the linus tech tips project seems to be that they did not put very much thought into this project ahead of time. I don't think their project was doomed from the start, it was just poorly planned and executed.
I wasn't referring to the average lows, I was referring to this here: "The coldest temperature ever recorded in the city was −17.8 °C" which really isn't all that cold compared to many other places. The video was shot in August or September, I think? So antifreeze definitely wasn't necessary by then and could have easily been added later.
I still agree with you that the whole project is a mess and that the way they implemented it is far from ideal, but at the same time you have to take into account all the other factors: "free" makeover for all their workstations, good content for their channel (even though the timing is pretty bad and many people are criticizing it), lots of attention from many other places. At the end of the day it's a calculated risk that they took, even if they completely fuck up, the pipes burst in winter and everything falls apart they can just leave all the PCs running as standalone, watercooled workstations. The only think they're actually putting at risk here is the time they invested vs the benefits from the video series.
All they need is a general idea in order to get sponsors for it (ie free stuff for the build), so that is likely all they did. I don't think any one on the team has ever actually built anything by hand before. At least not from what I've seen in that video.
I didn't intend to make it seem like their entire operation would self destruct if this project failed. It just really bothers me that they have taken so many wrong turns.
As for the temperatures, if that water isn't treated with anti-freeze, temperatures in the low 20's (Fahrenheit, 20F = -6.67C) would be cold enough to freeze the coolant, especially if they lose flow overnight (shut down computers, power outage, etc). Technically, freezing is 32F (0C), so it should at least be a concern if you live in an area that frequently sees temperatures in the low 30's. Since vancouver averages lows just above freezing in the winter, you can bet your bottom dollar that they see plenty of nights cold enough to freeze plain water in the radiator. You don't have to live in the tundra to be worried about freeze protection. They could have added anti-freeze, but with no mention of it I will assume they haven't. If it was something they thought about, I am positive they would have mentioned it somewhere since freeze protection would be a major design consideration. They still have one more video to prove me wrong, but based on the first two, I doubt it.
We should do one. Who wants to volunteer their house?
I don't think my landlord would approve.
They only glaring aspect i've observed is using the same loop throughout the whole system.
My suggestion:
Make a main Loop that leads outdoors, then make each workstation its own loop connected to primary loop via a heat exchanger.
Example: Note: I didn't research particular exchangers, linus' project would likely need different exchangers, this is just an example: http://koolance.com/hxp-193-compact-plate-heat-exchanger
The advantages of this:
- Secondary loops (each workstation) can control individual cooling by modulating its own pump.
- Less headaches trying to calculate flow dynamics of one giant complicated loop by splitting it up into smaller more managable loops.
- Not sharing all that coolant, less chance of damage by electrolysis and chemical reactions.
- A leak in the main loop won't drain the more sensitive secondary loops
- A leak in a secondary loop won't compromise the whole system.
- Easier to remove and reconnect each workstation for any repairs or procedural maintenance.
- Use higher percentage or even pure glycol in your primary loop to elimate freezing issues
- Modulate pump flow and fans of primary loop separate of the needs of secondary loops. Allowing a higher delta-t and elimating any condensate issues. The capability to preheat (heat tape) your primary loop in extremely cold weather.
- Ability to use automative grade radiator and fans designed to be outdoors for your primary loop and appropriate PC grade components for the secondary loops. Best of both worlds!
We do this in our data center on a much larger scale using Trane Screw chillers which involves tertiary and quaternary loops.
