Instead of sucking a bunch of hot air, try this bong instead!

[Shingoshi]

Your "Mazzei " is just a simple single stage venturi tube using Bernoulli's principles . You need .5 psi (absolute pressure) to boil water @~80F. Even the best venturi tubes won't get you there. There are are a lot of venturi calculators on the web if you want to do the math first before you buy a mazzei that won't boil water. Search under:venturi,venturi tube,aspirator,venturi tube simulation,venturi tube flow calculator.

BTW: water is a refrigerant

Would it be possible to put another form of vacuum device between the reservoirs and the Mazzei injectors? If I could find some kind of vacuum pump which can be driven by the excessive flow of fluid coming from the injector, that vacuum pump would then do the real work. The Mazzei injectors produce such an excessive flow for this system, that they really beg for something to do with that flow. Granted, they don't produce hydraulic pressures. But something should be able to be done with that flow to contribute to the overall design.

The first-stage vacuum device(s) would create the vacuum necessary for boiling the water. The injector then would only have to move that vapor into the fluid stream.

Shingoshi

[Shingoshi]

that was a joke based upon the title of the thread...

I was looking at your comment literally. I'm laughing now, if that's any consolation!



Shingoshi

[Shingoshi]

It involves the creation of a miniature cooling tower:
http://en.wikipedia.org/wiki/Cooling_tower
http://en.wikipedia.org/wiki/Chiller

Then of course I have another route to follow:
http://en.wikipedia.org/wiki/Ultrasonic_humidifier
http://en.wikipedia.org/wiki/Ultrasonic_transducer

And here's something to read just for fun!
http://www.stuff.co.nz/the-press/new...tchurch/242882

I simply have too much to look at...

Shingoshi

[Holst]

I think you're still missing the point here:
INJECTORS ONLY DIRECT THEIR FLOW FORWARD (from the outlet), NOT BACKWARD (through the suction port).
The suction port is called that, because it causes suction to exist. The suction port provides no external pressure, and therefore cannot return fluid to the source. You would have to completely BLOCK the flow from the injector's outlet for the injector to stop working. Then and only then would you see backflow. And that's simply not going to happen.

You keep looking at the injector as the main focus of the system...

You need to look at the entire system and work out the pressure drops.

[Shingoshi]

You keep looking at the injector as the main focus of the system...

You need to look at the entire system and work out the pressure drops.

I believe the pressure will drop shortly after the liquid leaves the injector. So just how much of a pressure drop do you want? It will drop again after leaving the cpus.

Shingoshi

This is an old post. I thought I had already submitted it. I just found it still open. So I'm sending it now. Someone may have already posted, while this was sitting around.

[Shingoshi]

I went to the ScienceMadness.org site to get specific information on the performance of injectors. This is the last post placed there:

http://www.sciencemadness.org/talk/v...rch&pid=158160
First, are you sure the case you picked will stand the negative pressure? Firgure out how much force will be on it under vacuum.

This case is nothing more than a miniature clean-room, in which the components will operate. The coolant-loop is completely closed. If there were any leaks into the cavity of the case, the process would stop. And, if the vacuum system isn't strong enough to boil water, how would it have enough power to impact the structure of the case?

Second, yes you can stack Venturis in series with the 'downstream' one being larger than the upstream one as it has to handle the full flow of the upstream.

I almost didn't understand you here. But now I realize by upstream, you mean the pump. And by downstream, you mean the outlet of the injector. This is precisely how the design is configured

Before going any further, you should analyse the requirements. Don't think about the liquid you will use, you will calculate what it's properties need to be.

How much heat is being generate, and what is the target temperature. How do you plan to dump the heat; remember that pumps will add heat to the system, and dry heat exchanges can not cool below ambient temperature.

This system will have a maximum cpu count of 12 in the present configuration. Given 75 Watts per processor, the load will greater than 900 Watts when the cooling system is included. The cooling system will produce some heat. Though I'm not sure yet how much. I intend to simply provide the largest cooling system to handle any possible load.

So you end up with a heat source of X watts that you want to keep at temperature Y while rejecting that heat into a sink of temperature Y. This lets you calculate the thermal impedance, and the size of the heat sink if you are rejecting heat into the ambient air - heat sink vendors should show you how to calculate that.

Venturis using liquids will give you no lower vacuum than the vapour pressure of the liquid at its working temperature.

Shingoshi

[Shingoshi]

I am pleased to say that I spoke with Celia Mazzei this evening about the Mazzei pumps. Taking her suggestions into account, I am now looking over some of the specifics that need to be dealt with. I now need to consider the flow rate of the pump I intend to use, and possibly change it if I need to. I think more than anything else, I may need a pump that not only delivers high flow rates, but high pressure as well.

The injector that she has suggested, is the Model 1584. It has a 1.5" intake and outlet, with a 3/4" suction. She actually took the time to look online at the specifications for the pump I specified, and made her suggestion based on it. But as has been mentioned here in this thread, injectors are poor candidates for creating vacuums. While Celia discourages using the injectors in series, she did acknowledge my alternate idea of running them in parallel as valid.

I've just sent off another email to Mazzei asking if they have better suggestions regarding the selection of pumps for the pressures their injectors are capable of. We'll see.

Hey! Friday I get my case. Go Pelican!

Shingoshi

[Shingoshi]

You'd have to admit having two very large flasks in your system with rotating fluid under UV light would be extreme!

http://jnaudin.free.fr/html/vtxbottle.htm
Just visualize that vortex happening inside of the reservoir below!
With UV for added effect...

Shingoshi

[Shingoshi]

My last posting led me off on yet another diversion. I began wondering whether a vortex as pictured above could be made to behave as a cork-screw. So I searched for corkscrew vortex. The proper term seems to be helical vortex. The intuition here being that if another point of drainage were introduced in close relative proximity to the main drain point, the orientation of the primary vortex would be influenced by the second point of exit. The same may apply to multiple points of drainage symmetrically placed in close proximity to each other. This should be apparent as a form of deflection on the spin of the primary vortex from the expected radius.

I've found some information that supports this assumption. But most of the data is contained in patent documents and related matter which require subscriptions to obtain. But I think I have a simple method for the observation of a helical vortex in my application here. So I'm going to investigate a means to do this, as it will assist in the mixing of fluid in the reservoirs, allowing for greater heat transfer to any cooling method at work here.

To visualize this, think of a single strand (not both pairs) of DNA in how it moves around a common radius.

Shingoshi

[Shingoshi]

I'm considering this Shertech CHMNA56T for my pump:

Part No. 0474056
Vendor No. CHMNA56T
Horsepower 2
Voltage 115 / 230
Flow Rate 96 gpm
Maximum Working Pressure 91 Ft. Head
Height 7-1/2"
Length 14-7/8"
Material Noryl
Weight 39 lb

Performance - Standard Models (Water at 70˚)
--------------------------------------------
Model CHMNA5X
HP** 2
GPM of Water at Total Head in Feet*
10 96
20 89
30 83
40 77
50 68
60 58
70 47
80 31
Max Head* 91
(Note: “X” refers to the motor supplied on the pump head. Pump performance is based on pump head, not the motor.)

Based on the (anticipated) number of processors in this system,
I think will be the best pump I could use.

Calculating the expense and performance of this pump against supporting five motherboards, and the cost of building comparable individual cooling-loops and cases for each, the efficiency of this build becomes immediately apparent.
--------------------------------------------

Let's start with the orientation of the case in which these parts will fit. The case will be in a vertical position (standing against a wall) so that the inside of the top (back)* is facing the user, with the bottom (front)* removed.

* This case was meant to lay flat. So the bottom and top sections in a vertical orientation become the back and front sections. I originally intended to have the case simply tipped upward, with the bottom of the case (with it's wheels) facing the user. That would allow for easy movement from one location to another, by simply tipping the case towards you to roll it. But now that I've looked at the need to have a vertically oriented radiator for this build, I'm considering reversing this. So that the wheels would be facing the wall against which it is stood.

The pump will be mounted lengthwise along the bottom. Having the pump head to the left, and the motor towards the right. This will have the outlet of the pump pointing in the vertical. Mounted directly to the pump's outlet will be the 2083X injector (also in vertical orientation).

Drilled and mounted into the bottom of each reservoir will be a banjo bulkhead fittings. On the inside (reservoir) fitting will be mounted a nozzle. On the outside bottom of each reservoir will be the fittings to attach them inline (by one tee and one elbow) and will drain down into the suction of the injector. I'm planning on using a cone with three ports in it allowing for what I hope will be helical vortex.

A vortex in which that twisting stream of liquid flows to drain, will be modified somewhat to look like a corkscrew. This hopefully will happen due to the increased forces pulling the walls of the vortex off-center and downward, causing it to compress. So instead of being the near vertical column we expect, it should look more like a spring. Though the degree of compression in that helical vortex will likely vary.

What will happen is that the turbulent movement of liquid will cause more mixing and thus transfer of heat to the cooling process applied to this reservoir. This will happen because the vortex stream will move into regions it would not have otherwise entered. So while it will be very pretty to look at, it will also be very functional as well.

This is ironic. It could be that the computer which is meant to be a computational environment, becomes the target of it's own analysis.

Shingoshi

[Shingoshi]

So it finally arrived. It was probably the last delivery of the day. But I got it!

I'm just taking the time now to assess what all I can do with it, and how and where I will place things inside of it. While being large, it's still is going to be a very tight squeeze to get everything inside, while being able to service any parts freely. But I'll get it done. Now, it's just a matter of appreciating this one step.

Shingoshi

[Naja002]

[Dark Bishop]

X2, more pichers!!!

[msull]

You really need to look into thermofluid science. The system you are describing breaks the laws of thermodynamics if it actually lowered the temperature of the working fluid. If you consider your entire setup (pump, fittings, reservoir, ect.) a system, you have work input (work done by the pump to the fluid) with no energy leaving the system (the boiled liquid stays inside the closed loop). If anything, your liquid will actually gain heat from the energy input by the pump. Also, your not going to lower the pressure enough to boil water with a venturi nozzle.

[Shingoshi]

You really need to look into thermofluid science. The system you are describing breaks the laws of thermodynamics if it actually lowered the temperature of the working fluid.

This was something that plagued me. I realized that I would have to remove the heat, but wasn't sure about the placement of radiators to accomplish this.

If you consider your entire setup (pump, fittings, reservoir, ect.) a system, you have work input (work done by the pump to the fluid) with no energy leaving the system (the boiled liquid stays inside the closed loop).

If anything, your liquid will actually gain heat from the energy input by the pump. Also, your not going to lower the pressure enough to boil water with a venturi nozzle.

The venturi issue came up elsewhere, having it already shot down on the ScienceMadness site for the same reason as you've stated. But then, I've wondered about not using water as the base liquid, and using a liquid more volatile, like ethanol.

This was something that I realized also. It's inferred from the statement above. So I accept this as well. Correct me here. I would have to remove the heat from the liquid that evaporated, right? But since I would be drawing that liquid back into the main flow, I would have to cool the main flow after the point where the liquid entered it. None of this seems practical in the long run. Which is why I've pretty much given up on it. Simply because it would require too much space in the confines of my larger project.

So thanks for the feedback. I already began questioning how this could possibly work.

Shingoshi

[msull]

What it really sounds like you are trying to do is make a vapor compression refrigeration cycle (what your A/C unit and refrigerator use). You should look in the phase change forum after doing some background reading.