Centrifugal pump impeller design

[neomoses]

I'm working on an interesting problem which involves cooling very small areas, specifically electronics components. The current method involves using water circulating through copper heat exchangers (waterblocks). The constraints I must work in demand:
1. small size. Everything must fit within 1 ft^3
2. quiet. cooling equipment will be near people for prolonged periods, so it must be relatively quiet.
3. High heat transfer rates at the waterblock. Currently heat loads can reach above 100 W/cm^2.
4. Low temperature. The temperature of the cooled area should never exceed 50C.
5. Low cost. This limits pump price to around $50-$75 max.
6. low vibration. the pump cannot cause excessive noise or vibrations.

Right now I have been using aquarium pumps to drive the flow. These have all been centrifugal pumps with very simple, straight vane impellers. These are typically high flow, low pressure pumps(around 2500 LPH max flow, 3m max head). To increase the cooling at the waterblock, more pressure is desired. 5-7m head at 700-1200 LPH flow would be a huge improvement. I'm looking at 2 possible avenues.

1. Choose a different style pump.
2. Modify existing pump to deliver higher pressure.

I have been unsuccessful in finding a different style pump that is within this project's price range, so I am looking more into option 2.

is it possible to re-design the impeller of these pumps to deliver more head, possibly at the expense of some flow? I would like to stick with a single impeller if possible, but multiple stages may be necessary.

I know only basics in this area, so if you could help me out with some good textbooks or online references, I would be very thankful.

 

[insideman]

I have pronted your question, will study it and get back to you later. Meanwhile. have you considered using heat pipes as part of your cooling scheme?

 

[neomoses]

heat sinks that utilize heat pipes have been used successfully with high RPM fans, but even the best aircooled heatsinks have trouble keeping temperatures low enough at high heat loads. In order for aircooling solutions to be viable, they require very loud fans or blowers (50-70 DbA), which is undesired.

It has been found that watercooling is a much quieter solution because heat is rejected from the water in a radiator, which can be larger because it can be located somewhat remotely, usually within 1-2 feet. Bryan

http://www.logiccooling.com

 

[insideman]

Neomoses writes that he wants to increase pressure in order to improve cooling. Also says that less flow would be OK. However, reflection indicates that he actually must INCREASE flow if he wishes to improve cooling.

Quickest way is to change to a higher speed motor. If the present rig is using an induction motor it will likely be spinning ar about 1750 rpm. Changing to a universal motor that will operate at 5000 rpm will increase the pressure. Pressure is proportional to rpm squared, flow is directly proportional to rpm. Power is proportional to rpm cubed. A 1/10 hp motor should do the trick.

I thnk this will stay within your cost parameters!

If a more advanced answer is needed, Neomoses should measure the flow and pressure in the present set-up. Measure flow by catching it in a bucket for timed interval. This would allow us to calculate the system characteristic. Flow is directly proportional to pressue squared. With this in hand we will have all the information needed to improve the cooling.

Please let us know what happens.

 

[neomoses]

QUOTE: "reflection indicates that he actually must INCREASE flow if he wishes to improve cooling."

Yes, this is obviously true. after reading my initial post, I see that I was unclear in my statement. My pump currently can deliver a MAX flow of 2650 LPH at 0 head and a MAX head of 3m at 0 LPH. I do not have a PQ curve for this particular pump, but I do know that these centrifugal aquarium pumps produce good flows at low head, but flowrate drops off quickly with any type of restriction.

When I say I don't mind sacrificing flow, I really mean MAX flow. I wouldn't mind if I went from a max flow of 2650 LPH to 1500 LPH as long as I could increase the flowrate at my operating point. I'll try to do some measuring with a stopwatch and bucket this weekend to get a rough estimate of the flowrate of the system.

I like the idea of increasing the pump speed, that seems very viable. I am currently using mag-drive pumps, which are nice because they are very easy to seal. Is there a way to increase the speed of a mag drive pump? Or will I need to go with a different type of pump and just increase the motor speed?

One of the (possibly) easiest ways to improve the flowrate would be to modify or replace the impeller. After looking at many of these pumps, they all seem to be very similar. The impellers have 5 straight blades. The OD of the impeller is approximately 0.4 inches less than the ID of the housing, leaving a good deal of room between the impeller tips and the housing walls.

Assuming I left the motor the same:
1. What effect would a larger impeller have? For example, if the blades were only about 0.050" from the side of the housing instead of 0.200".

2. would more blades on the impeller increase maximum flow? maximum pressure?

3. How would curved blades affect the performance of the pump?

although changing the motor and increasing the RPM is one way to accomplish this task, I would like to find the most economical way as well. Thanks for all the help!

Bryan

http://www.logiccooling.com

 

[ClydeMule]

Neomoses,

When you say, "Modify the pump impeller," what do you really mean? Are you going to take the impeller into your shop and modify it yourself? Or do you mean you need 2,000 of these pumps and you are looking for a differenet design?

If you are doing this to one or two pumps, modifying the impeller seems impractical. Possibly you could change impellers, but these things aren't really designed to be a modular reconfigurable unit. This is based on cost. You would be better off just getting a bigger pump. The easisest modification would probably be as someone suggested, change the motor to a higher speed. If this is onesy twosie, maybe you could rummage around the net and find a different impeller. But again, I don't know if you can change the impeller without breaking the thing.

Other sources of high flow pumps for cheap are something you could rip out of an old dishwasher or washing machine. They are designed to be no frills machines. The usually aren't what goies bad in a washing machine. An old junked one will usually still have a good water pump. Something you buy out of a catalog has a lot of money put into the casing, UL listing, power cord, etc.


If this is for a whole bunch of them, there are a number of OEM pump manufacturers in the appliance business who can build you exactly what you want.

Change of subject, have you looked at TEC (thermo electric cooling)? It is very quiet other than some small fans. Many people are doing this for electronics cooling. Check out melcor.com.

Good Luck,

Clyde

 

[HomeMadeSin]

neomoses:

I think more info is necessary. Some of the things to consider, mose smaller pumps, particularly the aquarium variety run at 2 pole speeds, so at 60hz that's 3450-3550 rpm. Speed makes a significant difference on noise, so if possible find a low speed pump that will do the trick. Unfortunately, thanks to physics, the smaller the speed, the larger the pump has to be to deliver the same flow and head (pressure). And since flow is something that most can calculate, you need to realize that the flow is the resultant of the system (i.e. the system resistance curve "intersects" the pump Q-H curve and provides that flow). Low speed pumps have "flatter" curves which dictates that the flow will drop off more with the same increase in pressure. For example, a high speed (2 pole) pump may only drop 100 lph with a 1kg/cm2 increase in pressure, whereas the low speed (4 pole/1750 rpm @ 60Hz) pump may drop 300 lph with the same backpressure increase.

In reality, the price range limits you to the smallest of aquarium pumps and I don't think that many companies in that market do any curved vane impellers (which would improve efficiency).

 

[zouzo]

hello
centrifugal pump running at 1489 tr/min by electrical motor(50 HZ) recorded each time high level of vibration on velocity mesurement as 12 mm/sec.impeller of the pump have 4 vanes that's the frequency appear at 100 HZ.
THANKS.

 

[LoonWatcher]

I have a similar problem as neomoses. I see these aquarium pumps, all with straight vanes, and wonder what would be the effect of changing the contour of the vanes.

 

[canox23]

Hello
I think you can't change the impeller because inside the case of the pump ther is a impeller tongue. from the tongue to the imapeller is just a litter gap that permit the pump be a pump and also the imapeller is not in the center of the volute is closer to the discharge that is to the suction. if you mean same diameter but more vane in the impeller well it could work but i think that is not a big different. what type of pump is split case or close end cause those are very efficient. and Pressure don't mean that there is flow. is that a brand new pump you using.