WHAT PUMP TO USE.... CENTRIFUGAL, POSITIVE DISPLACEMENT, MULTI STAGE, MAGNETIC DRIVE??????

[valmeidan]

I am not sure what pump to use. I need a pump for continuous use, that can handle around 10-20m head pressure, and around 70l/min. The problem is that I have round maybe 6-7" overall height and around maybe 10" for length overall. I am using an RD-40X pump now but with the pressure drop in a few components including some quick connects and a plate heat exchanger I am only gettting around 20 l/min which is not enough flow to the server rack. Does anyone have any suggestions on this? I was thinking of a positive displacement pump as they usually have good head pressure, but sometimes the flow suffers and also are very expensive. I would like to stay under $600 per pump.

 

[valmeidan]

Also I would prefer stainless steel, as I am running coolant through it and would like to keep the pump the same as the fittings

 

[valmeidan]

similar to these except these are 8" on the diameter.. most of them

http://www.pronto.com/shopping?o=10...=sdb&qo=semQuery&ad=semD&am=broad&an=google_s

 

[EmmanuelTop]

You should probably stick to a centrifugal pump if this was the original design. Positive displacement pumps will introduce additional issues that would likely affect the rest of the circuit design.

According to the manufacturer data the model RD-40X provides 70 l/min maximum flow and around 8m of head which is definitely insufficient for your application.

Try with Verder - they seem to have what you need. We have used some of their models as Glycol booster pumps (suction pressure atmospheric, discharge pressure 2-3 barg, range of flows from 50 l/min to 10 m3/hr). They also come in SS option.

Make sure you calculate pressure drop in the circuit accurately. If you fail to do that, you might end up with another "wrong" pump.

Dejan IVANOVIC
Process Engineer, MSChE

 

[EmmanuelTop]

Here is the link:

http://www.verder.ro/fileadmin/user...rdermag/mag_prod_broch_rev01_2013-ZA-_uk_.pdf

Dejan IVANOVIC
Process Engineer, MSChE

 

[valmeidan]

cost is also an issue. I have found a few namely ones like this... which seems to have decent head pressure and flow, and the price is around 460usd, which is a little high. it is also stainless. Only down side is I will have to increase my metal housing to over 8" which sucks, I would prefer under 6" overall height. what do you think about these pumps? I will have a look at your site also. thanks

http://www.amazon.com/AMT-370B-98-C...r=1-6&keywords=centrifugal+impeller+stainless

 

[valmeidan]

I do not see any options for stainless on the model V-MD 100

 

[EmmanuelTop]

V-MD 100 appears to be almost the same thing you have now (~7m head at 70 l/min flow). You need something bigger than that, a pump that can develop higher head.

In addition look at Ebara's website - they have Stainless Steel (AISI 304) off the shelf:

http://ebara.com.my/pic/EPM/upload/20091118144730_Content (SG).pdf

Prepare a specification sheet (min/normal/max flow and the required head) and shoot the spec sheet to the vendor. Remember to specify the maximum pressure that the downstream equipment can tolerate.

Dejan IVANOVIC
Process Engineer, MSChE

 

[LittleInch]

Ok, I think you need some guidance here on pump and system basics, so apologies if this is too simple.

Toget a happy system you need to match your pump to your system and its needs. So some basic decisions need to be made such as:

Is the flow constant or does it change?
Is the friction / back pressure constant or does it change, but you still need the same flow?
What is it you are pumping?
How much power do you have?

The fact you say you need a head of 10 to 20m shows that you don't really know what is required as this is 100% difference for the same flow and for a pump that is quite a big deal. For this level of head and what sounds like a small bore tubing and fitting system it will be difficult to calculate accurately and even a small component could make a big difference.

Centrifugal pumps are in essence constant pressure / head units with a wide flow range, but only a small change in head. Hence if your system curve intersects at a flow less than you want, you're stuffed. But they are cheap and small and can be swapped out for a bigger one quite easily. Hence you buy a pump rated for a certain head at a certain flow - it just sounds like you don't know what these two are.

PD pumps ( a gear pump springs to mind for this application) are essentially constant flow pumps. To avoid the pump stalling or creating a large pressure, they often have inbuilt pressure regulators at this size of unit which limits the pressure and re-cycles some of the flow back to the inlet. Hence you buy a pump rated for a certain flow and able to produce a maximum head of X, limited by the motor power and the relief valve.

Why do you need Stl steel? ?Beats me. You can have the rest of your system in this and the pump as cast iron or steel without any big issue IMO. So long as your fluid is not particularly corrosive or needs to be very clean then why buy a nice shiny pump?

So try and figure ut what is your requirement then go looking for a pump. This is small stuff where it is difficult to get ti right without a bit of experimenting and only by measuring preformence can you work out if the pump is correct or not as you will probably have no flow meters or other instruments common in bigger plant.

good luck

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[valmeidan]

LittleInch

Head pressure of 10m is what I will need. Greater head pressure is always nice to have. We are pumping glychol through a server rack full of wallet size heat exchangers ran in parallel. The loop is circulating through the pump a larger plate to plate heat exchanger and through some quick connects. I believe the whole system pressure drop is around 7psi. A typical server rack is around 6 feet tall, with about 42 sleds with two smaller heat exchangers in each. glycol will be circulated through these. currently we are getting only 20 l/min with the RD-40X with about a 7psi pressure drop which is not ideal. We are somewhat limited by space. running 24v power supply.

 

[LittleInch]

That's what I don't get though - you currently have a pressure drop / differential head of about 5m with a flow of 20l/min (how are you measuring flow and pressure drop?)

You then state 70l/min is what you want. Unless you're going to change something quite radically that implies you're going to need about 60m head as friction increases as a square of the flow rate. where does 10m come from?

How is your system configured? It could be you are just flowing through a few HXs at high rate and not all.

I'm only asking these questions as you could easily spend a lot of time and money changing your pump to find out it makes no difference if your system is set up all wrong. A simple sketch would do wonders for the understanding of your system.

What ever it is you're going to need a substantially bigger pump and 24V might start to struggle....

It might be you just need some more pumps in parallel, but without seeing how your flow system is set up it is impossible to tell

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[georgeverghese]

Couldnt find this pump RD-40X on thre internet, but from Dejan's earlier reply for the V MD-100 ( which produces 7m at 70 l/min, and using the head profile for this pump), would approximate the head at 20 l/min to be some 13m (ie 18psi). This is roughly what the system dp is at the moment at 20 l/min.

Which means that at 70 l/min, the system dp will go up by the square of the flow ratio ie

H2 = 13 x (70/20)^2 = 160m = 15.7bar differential

Would imagine this to be a pd pump

From Valmeidans later reply to Littleinch , at 20 litres / min, pumping dp = 7psi = 5m

So, at 70 litres / min , pumping dp = 5 x (70/20)^2 = 61m = 6bar diff

This would also be a pd pump ?

For a small system like this, there may be some components that are burning a lot of dp.

 

[valmeidan]

As I had mentioned,


"Head pressure of 10m is what I will need. Greater head pressure is always nice to have. We are pumping glychol through a server rack full of wallet size heat exchangers ran in parallel. The loop is circulating through the pump a larger plate to plate heat exchanger and through some quick connects. I believe the whole system pressure drop is around 7psi. A typical server rack is around 6 feet tall, with about 42 sleds with two smaller heat exchangers in each. glycol will be circulated through these. currently we are getting only 20 l/min with the RD-40X with about a 7psi pressure drop which is not ideal. We are somewhat limited by space. running 24v power supply. "

also here is the pump curves.

http://www.iwakiamerica.com/Literature/RD/RD_Family_Curves_2012.pdf

now the thing is I will have about 42 sleds I am pushing coolant through, thats about 84 cpu coolers, plus the quick connects, as they all go to a manifold then down to the reservoir into a pump and then into a larger plate to plate heat exchanger. the total pressure drop since the sleds are in parallel is about 7psi-10psi which equals about 7m of head pressure loss on the top end. If you have a look at the graph on the 40x and look at 7m head you end up with around 22l/min which is what we were testing.

what I would like is a pump that has higher flow with higher head pressure capability that fits within the envelope I mentioend earlier as 20l/min is not enough to effectively remove the heat from the system. this works out to about 0.5l/min through each CPU. I want to increase that to around 1 l/min. Hope this makes sense

 

[georgeverghese]

Think you mean you have 0.5 l/min per sled ( a bank of 2 cpu ), and that you want to increase it to 1 litre /min / sled which will bring it up to 44 litres / min (not 70litres/min). So current head for this pump is 6.7m

That would increase pumping head to 6.7 x 2^2 = 26.8m = 38psi

From the description of the glycol coolant closed loop, it sounds like the pump is somewhere close to grade, plate HX cooler is close to grade, and then the glycol goes up to the top of the server at elevation 6ft to these 42 sleds (each in parallel at different elevations) before it goes into the collection manifold.

Have you checked for air traps in the piping / tubing runs up at elevation 6ft ? - with air trapped, pressure drop will increase significantly. Also check the plate HEX for air traps on the glycol flow passes - any trapped air should free vent out of the HEX. There may be air trapped in each of the CPU cooler passes also. Also check the arrangement of the inlet and outlet at this reservoir to confirm that air wont entrain down into the pump suction. Ask some one to help if you arent sure.

Things should improve once you've fully liquid flooded the whole recirculation path.

 

[georgeverghese]

If this glycol reservoir is closer to grade while the sleds / cpu coolers are above, then the glycol feed to the reservoir should always be below the liquid level in the reservoir, else air will fill up this line from the collection manifold down to this reservoir.

 

[LittleInch]

Valmeidan,

If you want double the flow (~50l/min) and not 70? for the same system I think you're going to need quite a bit more than 10m head.

your initial question asked what type of pump. Based on the fact that you have a good handle on the volume you want, but not the differential head, I would go for some sort of PD pump with a higher maximum head of up to 20-25m. The other actor though that may cause difficulties is that the power you need is climbing up past 250W (70l/min @ 10m) and possibly higher. At 24V this is over 10 amps which is starting to get a bit high for this type of motor.

Have you not thought about just adding more pumps and supplying 20CPUs off each pump instead?

Sometimes you need to think about alternative systems and not just making something bigger.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[valmeidan]

georgeverghese......there will be a few pressure relief valves throughout the system that will release any air in the system. We are also using an accumulator instead of a reservoir to ensure that there is constant pressure throught the system to ensure that pumps will always be primed. It will also help with any air caught in the system before it hits the pump.



LittleInch... how are you getting 250 watts, just curious of the math. would you suggest maybe a higher amperage pump?

 

[LittleInch]

Shaft power (kw) = (Q X h x density x G) / (3.6 x 10^6 x eff(fraction))

In metric so m^3/hr, m, kg/m^3, 9.81

efficiency of these types of pumps is often quite low so maybe 0.5. 70l/min = 4.2 m3/hr

So for 10m shaft power is ~230W. Motor efficiency of 0.9 means a power input of 250W.

My point is that 24V DC might easily be at or coming to the limit of what people actually make and maybe 230V AC is a better voltage to use as it would need a smaller motor.

You don't seem to be getting my point that your required head figures seem unrealistic.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[georgeverghese]

It does seem to me that the root cause of this high dp in this system is either one or more components burning up a lot of pressure and / or air traps currently not identified. This is something you may need to get down to yourself.
What is the manufacturer's design flowrate through a single cpu cooler - is it 1.0 litres/ min ?
and what dp does it incur at this flowrate ? Check out dp for each and every component in this loop.
Once you have assured yourself that this has all been cleaned up, tubing runs modified and may be more air bleeders installed, then with the revised pumping flow and imposed differential head, you can get down to the business of setting pump duty / selection / whether you want to break up the pumping duty into smaller units.