Project Block

[BGP Spook]

Please forgive me for butting in and posting about the off topic.

With some slight modification to that design idea I could see it being reasonable to produce.

As you said, "This is not a block design, only an illustration..." but I could see it becoming a block design without undue trouble and tweaking. It would still require a CNC machine though; so it would not be too practical.

I have had somewhat similar ideas before but never bothered to follow up on it. Other interests always get in the way, and I don't have access to a CNC machine.

[bobo5195]

Sorry I was in abit of a bad mood earlier, too much work these days. You’re right mein deutsch ist nicht so gut and your English is probably better than some of my lecturers so I wouldn’t worry. If your really stuck on a point I can ask some of my mates to translate as they are native german speakers, they are taking a year abroad at Aachen university (saying that it is very easy as well).

The Reynolds number describes a critical state where turbulence starts in a circular tube. The values to calculate the number are velocity, density and viscosity of the fluid and the tube diameter.

Reynolds number actually is actually only connected with characteristic length scales. So it could be for anything. It is commonly used for tubes because that is where good data exists. But can be applied to anything really. Tubes it works as the eddy sizes are limited by the diameter but for a really short fat pipe Reynolds number could equally be a function of pipe length. .

..plus the navier-stokes equation is OK for laminar flow situations but you don't know how accuarte it is for turbulent flow, you can only guess, even CFD is guessing. For turbulence you can't simplify the equation as much and DNS eats a lotta CPUpower. The Solid works plugin can't do that, you need Fluent+Gambit for that (i used Fluent once and it is a thousand times more powerful than the solidworks plugin simulations i have seen so far).

All cfd is approximation. Seeing some of the assumptions made in my CFD lecture course I get really worried even for lamina flow accuracy. Could do some accuracy calculations for turbulent flow using turbulence kinetic energy and see how much it is compared to mean flow. Think the solid works plugins use k-epsilon so it is not that bad. Although definitely not FLUENT level control.

Think you maybe able to actually do DNS for CPU blocks as the Reynolds numbers are fairly low so the eddy sizes are fairly large.

I meant fluid mechanics book (not block). Should be able to use pipe heat transfer correlations to design a maze block.

I disagree with your stuff on boundary layer but need to think about what exactly I am saying. Pure conduction would be a Nusselt number of 1 and we should be getting far far more than that.

Measuring the mean velocity is not trivial for this case as velocities non-normal to the measuring plane are important. Need to look at my boundary layer notes but overall velocity near the surface is only partially important. This definitely needs more beer.

The Rayliegh number is only important for convection in a fluid, not between a solid body and a fluid. Of course it is not bad when Ra is high or rather that there is much Fluctuation (which basically is the same as Convection) and Turbulences can help that but it is not the most important figure in watercooling.
BTW: what software did you use for that, it doesn't seem too accurate

I agree that natural convection isn’t that important in water block design (can measure its importance by doing Gr/Re^2 and comparing the ratios) but it was an example.

The software was developed by a Prof Gosman in our department so we kinda have to use it. Its called STAR CD and is equivalent to Fluent (better at it for some things worse at others). It was originally used to model combustion in engines and how inlet conditions effect turbulence generation. The mech eng department at imperial was the place that invented CFD or so they tell us anyway. We get more lectures than any other courses in the world on it because of this. One Gosmans phd students developed the K-Epsilon turbulence model for his thesis, I think. We still have 5 profs and 6 doctors working on turbulence models.
http://www3.imperial.ac.uk/mechanica...leflowdynamics

The relations are pretty good. The numbers match up with empirical data and CFD is very bad at this stuff as near wall modelling assumptions are important. There is some variation in the data which is to be expected, depending on the models used FLUENT is between 17% and 52% out when compared to real life results. Mainly due to the assumptions it makes of near wall regions. The relation is for heat transfer including a wall. Stuff like this is analytically derived.

Your block design won’t work. Jet impingement dynamics is different to normal heat transfer. You can have effective local h’s of the order of the heat transfer rate of copper. If you PM me your email, I’ll send you a few references to papers.

The jet diameter is very important and there is no there is no jet to speak of in your design. Angles can increase overall heat transfer but only just, and it lowers maximum values. It is a tough trade off and hard to get right.

The recalculating regions in storm style designs at the corner are very good at enhancing heat transfer. They are actually about as good as the jet so you don’t want to get rid of them. Think of it as the wall jet (along the base plate) hitting the side walls is like a jet hitting a plate normally.

The flow is definitely connected to the wall in impinge jet flow. It sticks to it. You can actually derive the flow patterns by using analytical solutions of the navier stokes equation.

Need more bear before I post again.

[davidzo]

Originally Posted by bobo5195

The flow is definitely connected to the wall in impinge jet flow. It sticks to it. You can actually derive the flow patterns by using analytical solutions of the navier stokes equation.

I don't have beer at this time so i won't answer all but only this one first. You read wrong. I wrote that the jet itself is not connected to the baseplate. if you could beam some heat up from the baseplate into the tube of the jet (which of course has to be made of copper too in that case) you can effectively use the jets surface and the turbulant flow inside as dissipating area.

Originally Posted by BGP Spook

Please forgive me for butting in and posting about the off topic.

With some slight modification to that design idea I could see it being reasonable to produce.

As you said, "This is not a block design, only an illustration..." but I could see it becoming a block design without undue trouble and tweaking. It would still require a CNC machine though; so it would not be too practical.

I have had somewhat similar ideas before but never bothered to follow up on it. Other interests always get in the way, and I don't have access to a CNC machine.

Interesting that you thought of this way too. Actually a friend of mine told me about a similar idea long ago, so it is not 100% mine idea, i just pursue the goal of finding a block design whith lots of turbulence and dissipating surface which has its restrictivity only direct inside the Block where it can be of greater use and not in some plastics with jets inside.
Thats right, it is possible to manufacture, but i can't mill it on my 3 axis CNC, i need at least a 4-axis cnc to make a very simple version. I could mill the Top on a CNC machine and center punch the holes and then mill it on a manual machine. I just have to mount it on the millingtable 45°rotated around the y-achsis and need a mill which can be adjusted to the side that i can drill the holes in an angle of 60 or 70°. Then mount 180° rotaded and drill the connecting holes from the other side. a little countersinking should reduce pressuredrop. I would go for a 20*20mm surface with 32x 2mm V-Channels (64 holes) and 2*7mm distributing Channels in the top to the V-holes. The flow and water distribution should be absolutely even in every V-Channel when machined correctly.
That block would sure be something really new and the more i think about it the more potential do i see in it...

[bobo5195]

Ah is see you mean jet as part of thing.

Jets themselfs generate far more heat transfer than standard convection so i'm unsure if it is worth it.

These are some ideas i did for a multi jet array ages back.IN industry they are called CHICs i think and are a common way of doing things. They use multipul rows of jets and can be made using a manual mill and drill press. They probably perform less than a storm block as confined jet (jet shooting into a hole) as 3 times the heat transfer of an unconfined jet.

h1 and h2 are different designs. these are high restriction devices.

[phide]

Interesting illustrations. I really can't offer any comments (technical aspects currently beyond my understanding), but my primitive mind can grasp what you've done here and it all looks very tasty.

I don't mind the hijacks. Hijack away - it all concerns waterblockin'.

[phide]

Alright, I thought I'd go ahead and update on recent progressions.

Simply put, there are none. My machinist has given me some sort of bizarre run-around, saying that he did not want to risk machining the pins. He didn't really elaborate further, and I didn't peg him on the subject, as I was considering going with someone else anyway. The guy was, in so many words, a dick, and I wasn't particularly interested in paying his bills with my project. So... the process begins again.

To make matters worse, my financial situation has come to a slight turn. The block/testbed budget I've alotted has since moved to fulfill a more pressing matter. The funds will return again, but not for a few months, and since the money went to something far more important that machining costs, I don't feel at all badly about it. That's simply the way things go.

I suspect this thread will lie dormant for a while, but I do intend to wake it once I have some updates worth adding. To those I have not thanked for their input, and to those I already have, thanks for the assistance. I'm sure I'll be asking many more questions when the time comes.

[BGP Spook]

Hey, you know, life happens.

This delay could best be used for further research and discussion, since there is always something else to look up and compare against.

[bigben2k]

We'll be waiting.

If it's any help, you can always run prototypes in plastic (acrylic) for flow observations. Any shop can do that cheaply.

[Talcite]

Phide, i see your dilemna, and i want to engineer a test block quickly, cause drawing my waterblock and simulating it going to take over a month. If you want to send me your CAD drawings, i'll have them machined on my 4axis CNC. Please have it in a format compatible with QuickCAM. I can't send you the block, but i'll set it up on my test system when its up, and i can tell you how the numbers turn out.

[phide]

I haven't found it to be anything of a real dilemma (family is more important than waterblocks by a long strech), but I still would very much like to see this thing finished in one state or another. Though this is the case, I'm not sure I'm entirely comfortable with your proposal. In the end, I'd be providing about $100 worth of materials (the .010" slitting saw is about $75 alone), and I'm not entirely confident they'd be put to good use. I'd also provide copper stock, Delrin flat bars and other components. It's not typical for me to doubt anyone, but the idea seems a bit uncomfortable to me at the moment as finances are still pretty tight, and the loss would be something of a downer. As it stands, I'm currently just scraping by (damn these gas prices!), working when I can and going to class when I'm able. This is no personal knock on you or your intentions, but you're rather new here and I simply don't have enough confidence to really proceed with this.

I'm more than happy to provide all of my CAD files, and further, work with you on your own projects and components to the best of my ability, but I really can't provide the materials for you at the moment. If you, or others you know, are willing to pony up, I have absolutely no problem with you making the block, modifying it, testing it or using it. I'll get back to you concerning what I can and can not export to in SolidWorks - if all else fails, it can export to simple AutoCAD-compatible formats, and you should have no problem importing those in QuickCAM.

Although it seems pretty simple, if you should manage to pull it off without breaking the pins, you'd have a pretty nice piece to show off the capabilities of modern CNC machining technology, and your classmates would think you pretty bitchin'.

[Talcite]

hahaha yeah. are you sure you need the slitting saw? if i can do it with a CNC milling machine, i'd be happy to. I really would prefer if you didn't give me materials lol. I wouldn't know what to do with them, unless it was a block of copper or something. I have a 1/16th" end mill, so that should be better than the slitting saw maybe? Oh wait i just looked up slitting saw... I have one of those lol. Isn't that just a table saw? Don't know if its .1" though. I just saw it in the machine shop. I'll look it up. But yeah, please send me the CAD files. I'd be happy to try machining it.