What kind of WC disasters have you had?
 

I've been involved in a thread about disaster management for a while, and one of the things we don't seem to have is a collection of data about just what kind of 'disasters' people experience with their WC systems.

I define a WC disaster as a malfunction in the WC system that could (or should) have resulted in permanent damage to the hardware. I want to see what kind of problems people actually encounter, what sorts of protective measures you use, and how effective they are.

Please check the box for the type of problem you've had, and then tell your tale of woe. What happenned? What caused it (if you can figure that out)? If a particular component failed, what kind / brand was it? What kind of damage did you experience? What did you do to protect against this kind of problem, and did it work? Is there something you should have done differently to have prevented the problem? Anything else you think is relevant?

This survey is primarily aimed at WC issues, I'm not particularly interested in pelt, HSF, or other non-WC failures, unless they were caused by WC problems, or caused WC problems.

Who knows, if we get enough responses to be meaningful, it might be possible to make this into a sticky, or turn it into a 'disaster prevention' article.

I voted total pump failure - but take that in a "I forgot to plug it in and so it wasn't working" failure type of way. As I explained in your other thread, no damage to any part of the system - it even ran stable for 5 or 10 minutes while the block and whatever coolant was inside it at the time heated up. Once those were hot though it didn't run very long at all (wouldn't finish booting).

I've also played with rad-fan failure. Actually, I recommend more people experiment with this, if you keep an eye on it you're plenty safe. It takes a CPU a long time to warm up the entire mass of coolant circulating in your system, and the rad still passively dissipates some heat. Turn off / unplug your rad-fan and watch your temps. Try it while idle/under load (I can run forever with no fan at idle). See how long it takes for temps to go up 10C or so. Most of you probably last a lot longer than me with no fan as I don't have a res so I don't have much coolant-volume-buffer to heat up before I crash/need to turn the fan back on.

disasters, ok lets see...

- when i had a bong, my pump came disconected from the outflow tube and stopped pumping water through my system...system overheated and shut down.

- when i had a bong system ran out of water several times, system overheated and shut down.

- My BIX ruptured in two places and water leaked all over my system claiming the life of my geforce3 (smoke poured out from it)

- The seal on my mag drive 3 broke when a metal fitting became corroded.

- The plexi top on my orginal Maze3 spider cracked then completely snapped and leaked water over motherboard and video card, claimed the life of my first A7N8X.

- I've knocked out several breakers fooling around with the wiring for my pumps as well.

The main thing I've learned about watercooling thus far is that if it ain't broke, don't fix it....and right now my system is running better than ever.

MAN! Your a poster child for aircooling. LOL

I have only had leaks that were due to stupid assembly and lack of leak testing on my part. It is amazing how much water a mobo and graphics card can take on and still work. :D

I once blew in one of the pipes. I thought I had the res open but instead I didn't.

Result: Super Wet western digital 1200JB :eek:

Still works

Once I had the coolant (water + antifreeze) freeze up in the tubes. Not enough antifreeze obviously, and flow stopped. The pump made a bunch of weird noises as pressure built up. This was with -15C coolant.

Another time I froze my BP6 dual CPU setup so hard that the sockets and everything one inch around the sockets froze. Capacitors and all had a nice thick frost layer on them. This was with the sub-freezing coolant and dual peltiers. Took a couple of days to dry it all out, but it still worked.

I have also had lots of condensation drips off of tubing, but I was prepared for most of those.

Generally the annoying drips... worst part about that is even with nothing seriously going wrong and a towel there to saok it up... was having to top up the system every 4 hours or my liquid started "foaming" resulting in a rise of temperatures.

I had one of the first Koolance systems, the one with the big aluminum resevoir / heatsink on the bottom of the case but without the watercooled power supply.

The pump in the resevoir was a cheap fountain pump. I had two of them die on me. The first (original) pump had a burned trough coil. It was completely dead. The second pump had its impeller catch something in the housing and stopped turning. My guess is that the resistance of the narrow water hoses is too much for these pumps to perform properly.

Also the power supply died. The only part of this case that is of some quality is the resevoir / heatsink. Totally useless though because a regular heatercore is much better / lighter / smaller. I will never buy a Koolance product again. Wasted $350 on that piece of crap. :mad:

Had me epoxy on me plexi lid melt as me brother forgot to turn on the pump. Dried everything out beside the fire and it worked grand again.
Also was taking me Pc home from a lan and it got banged about a bit and i had another leak . Dried it out at the fire and it was alright.

None:D I've been WCing for over a year and not a single misshap.;)

One time during chem lab my friends and I were discussing how low my CPU temps were compared to there Dells. Well the idea of seeing how low I could push it came up. The result was us pouring about 4 gallons of finely chrushed ice into my bong. I had condensation on every part of the loop. Took me a few hours to clean up the mess but luckily everything worked fine afterwards.

I had the user, malfunction on me a couple of times...and I kick his sorry little ass every time, but it does not seem to help much... :D

I've only run my Little Giant pump in a test loop, but I believe that it was sucking air in. Still gotta figure that one out...

I had a fan failure (Ok I forgot to connect it). The block was being used with a T.E.C. It got to the point where the water could no longer carry heat away from the T.E.C. The T.E.C failed and started to over heat. The thing became so hot it melted the solder off the water block. The C.P.U was fried and the socket melted. The funny thing was that I replaced the C.P.U and used air cooling and the system started like nothing had happen. Those sockets are a lot more robust than they appear. I was using a power supply that is capable of putting out 40 amps at 16 V combine this with the system running with no fan for over 24 hrs, and now I get to post in this thread. Before the melt down I was producing ice like old man winter very cool stuff (no pun intended).

Same as FRAGN'STIEN, none for a year of wc'ing. Must be cos' of the use of the #Rotor's blocks, seems that systems like 'em:D

Only one failure and no damage from it.

I was playing a game online and suddenly my MBM5 alarm went off... I shut down the game and look at the temps... they were well above the "norm" on my machine. :shrug: I shut it down.

Opened the system and the pump was still running (separate power cord) but I had NO FLOW! :cry:

Further investigation showed that the submerged pump I was using was fine, but the hose connecting it to the "out" barb on my reservior was ruptured. :confused:

I found that the tubing I had been using (medical tubing, tan color, same stuff they use for tornequets, sp?) was deteriorating with submersion in water faster than the rest of the tubing.

It turns out that stuff is made out of almost the same compound as rubber bands (tan ones) and so was subject to the same deterioration.

Bottom line... bad planning on my part. I've had vinyl tubing ever since and have had NO problems at all. :rolleyes:

slow drips...

1) cracked danger den reservoir (old version) mounted above my CD drives... fortunately I had put a piece of foam underneath it which caught the slow drip. Guess I tightened the barbs too much... since then, switched to "T" fill setup.

2) cracked via 1300 pump intake. again, overtightening barb caused this. Very slow leak (drips would dry out by morning...). solution: JB welded entire intake section of the pump... Pump sits at the bottom of case w/ foam so again, no major meltdown...

voted small drips. i was so lucky that the drips were so minor that they evaporated before dripping on components. this was entirely my fault and wouldn't happen when buying a manufactured block from a retailer.

I've only had problems with my old system (that I had about 4 years ago, peltier cooling my Celeron 300A).

Water often dripped from the block (which was nothing more than some pieces of plastic glued to a celeron heatsink). VGA, sound, network, tv tuner and motherboard all got splashed at one time or another but they all still work... The CPU itself (slot1) got pretty wet too, and while the PCB looks really bad in places it still works just fine. I have it running 24/7 in a SMP setup in my server.

I guess they don't build them like they used to :)

You need to have a choice for hose collapse. That is the truly silent killer if you use crap tubing (I lost a sweet pump that way).

hose on NB leaked :( and fried my few month old video card, :cry:

I had a slow leak out of a fitting that went between 2 different hose sizes that dripped water all over the top of my 9700 for about 24hrs until I noticed it, while it was on, but suprisingly nothing ever happened. I've also had a small drip from my pump intake housing that was slow enough it would evaporate before getting anywhere, but after a LAN or two, it got bigger and bigger until the entire bottom of my case was flooded and started spilling out onto the floor. It completely soaked the foam pad stuck to my pump relay and nearly touched the contacts. If I hadn't put my foot in the water that spilled out, I probably would be a very unhappy dude right now. I also keep my relay at least an inch off the floor now.

I had also gotten a bay-res and had taken out the angled barbs to put in straight ones, but ended up using them anyway and since they fit so tightly I didn't think they would need any sealant - wrong! Another slow leak, again fixed with Plumbers Goop.

I've encased the entire front housing of the pump (a Danner) in Plumbers Goop, as well as gone completely to 1/2 fittings all around so I could get rid of the converters and am leak free.

Gooserider

Are we seeing a pattern in most of the replies here?? like I'd indicated in your disaster thread, user error in one form or another seems to be the most common factor from what I've seen here and generally.

This was part of the reason I made an idiot proof system to ensure the pump must be plugged in, powered & monitoring flow once it is.

Been water-cooling about 2½ years now and no leaks or disasters YET. The only mishap was when I refilled the system once and had forgot to re-connect a festo fitting, I paid for that little oversight with a hair dryer & blow gun, but the system as always was completely de-powered before I work on it........ so it was just an inconvenience really.

I'm now pretty "confident" in my block making and system set-up, and I really don't think too much about leaks anymore, this complacency is natural I guess, but dangerous, however I accept sooner or later I will have, or cause a disaster. Can't ever see myself going back to air-cooling though whatever mistakes I may make with water-cooling in the future.

I think the key is to work in a logical fashion, and not be tempted to rush stuff. If something isn't right, re-do it properly, never be tempted to bodge for now, as the bodge will stay as a bodge until it fails. You'll then have to sort it out when you replace the hardware it took with it when it failed. ;)

Lots of interesting patterns
 

Yes BR, I'm seeing lots of patterns including the UE that you mentioned (whether the immediate cause, or a design failure that occurred later)

I'm very glad that I put this survey up, I think that it's a useful collection of information that shows lots of things BESIDES the somewhat expected User Error. (I've been watching the replies as they came in, but not commenting as I didn't want to skew the results....)

Slow leaks from various places seem to be the biggest class of problems, but there don't seem to be many major floods. Arguably this is the class with the fewest direct UE's as direct cause, though most could be argued as UE indirectly.

Somewhat to my surprise, pumps seemed to be another major problem area. Some problems were direct UE (didn't turn on) but there also were more pump functional failures than I would have expected. (makes me glad I got an Iwaki)

I didn't like seeing all the 'Other' responses, but it seems to me like most of those were related to radical cooling failures, which I had deliberately not included.

I also decided to allow multiple choice responses, which may have been a bad choice on my part, though I thought it was best at the time I did the poll. It is interesting to note that 'never had a problem' and any of the major 'had problem' responses total to more than the number of messages in the thread... :confused: Guess this means there were voters who didn't leave messages, or possibly someone's fibbing...

I hope this survey either gets made into a sticky, or an article is created from it, as it taught me things that I haven't seen expressed elsewhere, namely the types of things that actually DO go wrong (Note that we have only had one (so far at least) CPU / Block separation, which wasn't talked about. This was one of the things I had feared most, but now feel much less worried about.)

Indeed most of the things I mentioned didn't get much response, especially the more dramatic failures. Pumps seem to go out or not get turned on alot (which shows the need to have a fail safe to ensure they are turned on, and the flow monitored) and we have lots of slow drippy leaks, which seem to be hard to catch, but not do a great deal of expensive damage otherwise. (shows need for good sealing and clamping practices)

Gooserider

I suppose it depends on how you view it, but to me the majority of the above descriptions indicate user error of some type or form. Leaks from tube to barb joins are user error in my opinion, because if they are done right they will not leak. A non user error would be if the tube actually split, but I'd consider this user error also if it were either the wrong type, bad fitting or cheap tube. For instance if I got a leak at one of the festo fittings because I didn't push it all the way in and pull it out to grip as designed, then that would be my user error, not a simple drip leak fault as such.

Again the drip type leaks are avoided using quality tube that is smaller than the barb, (but keeping the ID the same throughout). There are product faults like real pump & block failures, but these are rare and can be partially linked to user error for choosing cheap products, or ones that are maybe not the best suited.

Its hard to categorise every happing, but maybe to be of real world use your poll should distinguish between real pump failure and the usual user error induced "not turning it on". My guess is the majority of that vote % on pump failure is actually user error, but that wouldn't be obvious to someone just looking at the poll.

I agree, depends on how one defines UE.
 

I agree, in the same sense that driver error is the cause of nearly all car accidents. But note that the person experiencing the disaster might not be the person making a design or assembly error that caused a later problem (IE the person who had multiple problems with a Koolance case) However it is more likely that the person who forgot to turn a pump on is also the user experiencing the result.

So there are multiple sorts of 'user errrors' ranging from the direct action that caused a near immediate failure (forgot to turn the pump on) to the somewhat indirect effect of poor workmanship leading to a delayed problem, to the very indirect effect of making poor design choices (possibly because of lack of knowledge) that cause later problems by component failure.

While all can quite accurately be described as 'User Error', just lumping them all together doesn't really seem to be useful as an educational tactic.

I feel it is better to find out what kind of problems people had, what the causes were in as many cases as people are willing to admit to having a problem at, and find out what specific failures are occurring. This tells you what kind of user errors are being experienced, and points to the defensive strategies giving the most effectiveness for the effort. (or showing which ones aren't worth the trouble...)

For instance, CPU / Block separations are seemingly rare, which means there isn't much payback in trying to do extreme measures to prevent it, or do detection and handling when it does. Drippy leaks are much more common, which says that putting more efforts into assembly quality, sealant application, clamp useage, etc. will have a payback that makes the added effort worth while.

Again, I sort of agree, but bear in mind that I am limited by the board to 10 responses. I made a list of every failure type I could think of, or recollect someone complaining about having had on the board, and got about 15 possible items. I had to whittle things down, so I took those items that I thought were less likely and combined them, hoping that the stories people posted would tell enough detail to figure out how to break down the combined responses. In the case of the pump failure, you might just be wrong! I just did a search on 'pump' and found 11 described failures out of 16 reported for that response.

2 drippy leaks
3 'induced' failures where something else caused the pump to die
2 failures of a connection to the pump
2 total failures (both in the same case)
and just TWO 'Forgot to plug it in' UE failures.

(Some of the classifications might be subject to interpretation, but that's what I would call them)

The problems with the poll as I see it,

1. Not enough granularity in the responses (not enough choices).
2. Not all responders told their tale of woe.
3. Some responders may have checked multiple boxes for the same failure (i.e. forgot to plug in pump, so the block melted, so it leaked water all over the place, so it electrocuted the cat ;) should have gotten one response, but might have gotten four)
4. While I tried to be specific in my questions, people may have interpreted them differently.
5. Nothing prevented votes being cast for 'extreme cooling' failures which, while interesting, were NOT the subject of the poll and could skew the results. (IE plumbing freeze-ups, or condensation drips are not problems that a non-extreme cooler would ever have to worry about)

The only way I see that could realy get around some of the problems is if people just told their failure stories, and ONE person analyzed them and put them into a chart so that the classifications were as consistent as possible, but that would be harder to manage.

Gooserider

I'm not trying to be negative as I think your idea of a disaster data base would be good for worried water-cooling virgins to check out. It however needs to be based in reality to be valid, and although what you say about the 11 described failures out of 16 reported pump leaks or failures maybe true, when displayed as a poll like that doesn't tell the story and could lead to people thinking pumps are a real problem part, (which they are not if you accept it is the heart of the system, and buy & use accordingly).

I think there is mileage in something useful though & wonder if Joe & the Pro-cooling team could give any input to a way to do this?

I still stand by my statement that most of the problems are caused by the user in some way, and this is something anyone thinking of water-cooling should take on board first.

http://www.overclockers.com/articles772/

This link goes here I think. Very funny read.

Exactly my concern, and why I started this exercise... I had seen many mixed reports of what one needed to be concerned about in setting up a system. They were all probably sincere, but each was based on the author's individual experiences. IMHO a collection of what sort of things ACTUALLY happenned to a large number of people would help to demonstrate where the priorities should be placed, and what sorts of issues are 'must do right or else' vs. what one might decide to let slide a bit. (i.e. I am building a system to operate 24/7, which means often unattended. I arguably need a better quality pump, and more fault detection than a gamer who only has the system on when sitting in front of it. OTOH, we both need to use good quality tubing and good sealing technique when building our systems.)
True, and it is worth noting that while there were only two "didn't plug in" direct cause user errors, one could argue that the rest were indirect errors - The leaking pumps were cheap pumps that the user over-torqued fittings on and cracked. The induced failures were either bad material choices, or operator error in a different area. Even the two total failures were arguably the result of a case supplier using a poor choice of pump.
Good thought, any ideas about the best way to approach them with the idea? (I have a great idea for a title - 'Water Cooling without Diapers' with a photo of a mobo sticking out of a diaper.... ;D )
Agreed, and it should be a prime focus of the survey / articles. Probably something to the effect of 'these are the most common USER INDUCED problems, and how to best avoid them.

Agreed, certainly a wonderful case study of how to do things wrong. While he certainly appears to be learning, and his latest setup is a BIG improvement over his first effort, I still see a lot of problems with it. One of the biggest is probably that he is still engaged in that evil 'mixing of metals' practice. He got rid of the copper pipes, but nothing in the article suggested he wasn't still using the copper rad, not to mention those brass barbs sticking out of the aluminum block. :eek:

I would suggest that he should consider signing up with 'Silicone Abusers Anonymous' - There is a place for sealants during assembly of things like pipe threads, and possibly as 'insurance' when fitting a hose to a barb or the like; but IMNSHO globbing sealant onto an existing leak is an abuse of the technique which might buy a bit of time but is fundamentally another leak waiting to happen.

With the exception of sealing fitting threads and other APPROPRIATE applications, if your system isn't leak free without sealants you have fundamental problems that no amount of external sealant application will solve long term!

The only thing I'll add is this poll here isn't the be all and end all, it would need to be far larger to even hint at that. Generally from topics I have seen in the community that indicate some form of water-cooling disaster pointing to the pump, when you read through it it's really the user, not turning it that is the error. A lot of the time people getaway with it because they are using the system and it either crashes when the CPU gets hot or shuts down at a certain temp.

If the pump actually fails because you skimped on cost and bought a cheap and nasty one, that is 100% user error imo, and you would also deserve a slap :p, (a pump is the Heart of any water-cooling system). If you bought a known quality, (expensive), one and it fails, then you at least did the right thing and man made things sometimes do go wrong, however good they are. A pump relay is quite a simple thing to do & it should be on every water-coolers list of must do's. I do not personally favour 24/7 pump operation, for the same reason I don't favour 24/7 PC operation.

Apart from the possible fire risk leaving a PC unattended for long periods, (especially when sleeping), my opinion is when it doesn't need to be on TURN IT OFF!. If everybody worldwide left their PC and Pumps on all the time think how many more unnecessary polluting power stations we would need in the world. Parts of the US are struggling to cope with record electricity demands atm anyway. :eek: