Centrifugal Pump Low Flow limit vs. Speed Reduction

[Dokuen]

In regards to thread407-167783.


I too have not seen anything in writing about variable speed and the minimum flow requirement for pumps; however, the "standard" practice I have seen in the field is to take the same percentage of the new BEP at the reduced speed.

For example if the pump company states 10% of nominal flow for their minimum. Look at the performance curve at lowest speed you will operate in the application and take 10% of that reduced speed curve's BEP.

As flow is linearly related to speed this is simply the percentage of speed times rate minimum flow. So if standard min. flow is 10GPM then at 80% speed you could consider the minimum flow to be 8GPM. Most variable speed pumps I have seen that come from the factory with drive installed still use the rated minimum flow of the pump. But the mfr will always lean conservative.

Please also keep in mind this is all assuming cool clean water. If this is a boiler feed application with high temp water then your minimum flow will be higher to begin with.

 

[Artisi]

I would suggest that in all cases such as this an engineering decision based on all factors pertaining to your particular pump / installation is the best approach.

For instance - at zero speed the minumum flow is zero and between this point and the minimum flow at the design speeed set by the manufacturer / designer should or can only be a decision made if you have all the facts in front of you - the manufacturer doesn't have this information and can only give a guideline when setting their limits. Hence the comment you have made about not seeing anything in writing.

 

[ozmosis]

Dokuen
Maybe it's just me but I'm not sure what your question is. What do you want to know?

 

[BigInch]

Douken, (sed2),

If I may. He's saying, when mfgrs give a minimum flow, it is customary for them to give that value as a percentage of BEP rated flow, so I think the question is, "what is the pump's minimum flow at a reduced rpm setting".

Douken, yes, its typically assumed to follow the reduced equivalent "BEP-eq" at a lower RPM.

I think that its true in general, but have doubts as to how far down into the lower flows that the affinity laws can be extended. I think other factors can enter at lower flows that obviously would cause "minimum flows" to approach some limiting value > 0. The lubrication effect, for example, as well as some others.

Pumps components are hydraulically optimized for BEP conditions, which define the differential head and the flowrate, so if one wants to deviate from the BEP condition, one should try to evaluate possible effects. As I understand this, determining a minimum flowrate should be based on 3 primary factors, temperature, lubrication and unbalanced forces.

Pumps components are hydraulically optimized for BEP conditions, which define the differential head and the flowrate, so we should assume that any deviation from that operating point will result in reduced performance in one respect or another.

The reduced performance in terms of lubrication should be evaluated. At reduced rpm, will there be reduced lubrication? Flow through the pump is reduced linearly with rpm, so yes it is possible that lubrication will be reduced as well, depending on the type of lubrication provided for the particular pump. If lubrication is dependent on internal flowrate, beware. However, as rpm is reduced, it is possible that friction is reduced. If friction is reduced by the square of the speed, a net gain in lubrication effects could be possible.

The reduced performance, in the case of pressure imbalance might best be correlated to increased bearing wear, for example, due to higher thrusts in the axial direction, and increased lateral loads to the shaft. In the case of a variation in RPM, one should therefore examine the resulting force imbalances in relation to their values at BEP. Are imbalanced forces reduced proportionally with reduced rpm? No. Discharge pressure would most likely be reduced by the square of the speed, and assuming suction pressure remained more or less constant, a higher net thrust force could be developed towards the discharge end, unless the pump somehow compensated for the change in the differential pressure at the new operating point.

Lastly, temperature build-up at lower flowrates should be evaluated. Temperature is a function of both hydraulic efficiency and the cooling effect of fluid flow through the pump. At a lower rpm a lower product flowrate with consequent lesser cooling action would be expected. Efficiency may also vary. While efficiency is generally assumed to follow the pump affinity law, that may not be entirely the case, since pumps with higher BEP rpms typically reach higher efficiencies, should not some reduction in efficiency somehow be expected to occur with a VSD reduced rpm? I expect that there is some, but have not tried to evaluate how to predict what it would be. In any case, at very low flows, in the range of what would be a typical "minimum flow" region, the affinity laws might not be entirely valid for correlating efficiency, so I'm hesitant to push a possible analytical solution, but believe it could be at least partially evaluated with a heat balance calculation. If you assumed that efficiency followed the affinity laws, the mfgr's minimum BEP flowrate would follow the "BEPr" flow wher r=> BEP op point at a reduced rpm. There does appear to be a limitation to the validity of the assumption that efficiency follows affinity laws, since the frictional component of power at shutoff on the BEP curve is not entirely dependent on rated rpm. The static friction component of total friction, would not necessarily change in total proportion to lower rpm's, so caution at low flows would be prudent.

I would really like to hear from some pump designers on this subject myself, but until they say otherwise, then I certainly beleive you are correct, at least until you approach the limitations I listed above.

http://virtualpipeline.spaces.msn.com

 

[TenPenny]

Let's consider the Goulds 3196 ANSI pumps.

In the tech data available to sales engineers, they have a table of minimum flow vs speed for each pump. These figures are provided for 2 pole and 4 pole speeds for 60 and 50 Hz.

They are based on max diameter impellers, for other impeller sizes, the printed recommendation is to use the affinity laws.

Without more explicit data, using the percent BEP ratio as a guideline is 'sound engineering practice', as they say.

 

[BigInch]

10P, is that all the way down to zero flow?

I think where the mfgr voids the warranty is the cutoff point. I've never had them say anything less than 10% for short term (start-up) operation, moving to 20% min ASAP.

http://virtualpipeline.spaces.msn.com

 

[TenPenny]

I'm not sure what you mean about down to zero flow...how do you envision this coming into play, based on what I posted?

 

[ozmosis]

BigInch
Thanks for the explanation.
As an electrical sticking his nose into this I am particulary interested in the aspects of VFD control of pump-motors.
It makes me think that the VFD we design is somehow missing a few key points when it comes to monitoring minimum flow. We incorporate a macro that determines the motor load (torque). In our load monitoring section within the VFD we measure torque and, where required, minimum torque based on a measurement of torque at minimum flow/speed. At minimum torque, our macro then provides the user with a facility to enter a hysterisis. The basic theory is that measured torque moving below the stated hysterisis is either no-flow or flow at an unwanted level. We also incorporate a timer that kicks in once this point is reached (to prevent it constantly coming in and out).
Have I missed the point?

 

[BigInch]

Heh heh. That's part of my beef with VSDs. I don't think they're all they're cracked up to be, when to compared to the everyday typical type pump installation with a well designed steady speed pump matched to the system curve. Don't get me wrong, VSDs have their place, but for the typical installation, they shouldn't really be needed at all. Especially true when the system curve has a static head of > 50% operating head. Most of the time I can do just as well with an off and on switch and/or adding a tank. They're great on systems with wide flow variations and no to little head variation, or for running different products with widely varying viscosities and specific gravities (ie, multiproduct pipelines and hot heavy crudes in a hot pipeline) but for water 24 hours a day to an elevated tank or to a boiler with constant high inlet pressure, forget it. Most VSD proponents emphisize the power savings at low flows to make the sale, then forget to tell you that you probably won't get the head you need when you're lower than 50% rpm anyway. IMO, there is a very narrow range in which VSDs can really do what they're good at (saving power cost) in relation to a well designed pump (& a control valve) matched to the piping system. In those cases I love VSDs, but ... I still like the diesel type better. They make more noise.

http://virtualpipeline.spaces.msn.com

 

[BigInch]

And as you were surmizing from my post above, operating outside BEP (even if BEP is proportionally reduced with rpm) may have some other penalties and limitations to consider.

http://virtualpipeline.spaces.msn.com

 

[ozmosis]

oh blimey BigInch, don't say things like that..VFD's are my business!
I know what you mean though. My particular branch in VFD design is HVAC and about 80% are fans. I agree, a well designed pump system will not see the benefits of VFD's for a while but thank goodness there are still a lot of badly designed systems out there and moreover, people not able to do it properly. A VFD will cover up many a badly designed system!

 

[BigInch]

Ya I know you're in that business. I Didn't mean to be critical of VSD, just truthful. And you hit the nail on the head. There are a lot of badly designed systems out there and there are a lot of systems where the flowrate has changed and the system was not up/down graded and optimized for the new flowrate. That is where VSDs can really fill in the gap. THEY ARE A GOOD THING .. when used in the right application.

http://virtualpipeline.spaces.msn.com

 

[Tmoose]

I'd be looking at a flow meter as an option if too little flow is really that risky, or simply starting at 25 or 30% speed can't be evaluated. The output from the flow meter to a VSD would directly bypass all the rpm measurements and assumptions. At some point a fan cooled motor is going too slow to cool itself anyhow.

Has there ever been a system that could be estimated so closely as to be "well designed?", and, at least in HVAC, without some extra capacity available for future expansion. Prudent engineering seems to invariably end up with "too big" a pump/blower.
At least in the 80s and 90s adjustable sheaves were pretty much standard for most HVAC, understandably recognizing that betting heavily on the accuracy of calculated (estimated?) system characteristics just is not worth the risk. With 40 HP VSDs available today for as little as $2000, and capable of using a variety of inputs for automatic or remote speed control, fooling with adjustable sheaves and input dampers loses a lot of appeal.

 

[BigInch]

I think most of the time, a possible increase in future capacity, beyond a size or two up, is best handled now with a tee, a blind flange and leaving space for train #2, as too big can be just as bad or probably even much worse than too small.

http://virtualpipeline.spaces.msn.com

 

[quark]

BigInch,

Good post and what we in these forums do if there are no bad designs(?) and certainly you don't want to snatch bread from other engineers.

I came across the VSD guys that say they reduce power consumption by reducing the speed of anything that rotates.

The other day the big speciality pump company guy came and suggested not to use variable speed device for a perfectly recirculating variable flow (requirement) system. The reason he says "Boss! you know, when you reduce the speed by half, the static head reduces by one fourth". I asked as to what happens to the frictional losses, simultaneously. He asked one week time and replied "no change". He said that he was using one of the best software packages for piping design.

Now, as we both are sticking to our ideas, they are giving me the pump on a trial basis for 1 month.

So, there are stupids on both sides, who boast VSDs and who don't.

 

[BigInch]

Well, don't know, but depending on the system curve, he might be right, or you might be right.

Let me know how it turns out.

http://virtualpipeline.spaces.msn.com

 

[Artisi]

Looks like we have discovered something new in the area of physics - "Boss! you know, when you reduce the speed by half, the static head reduces by one fourth".
Certainly would be nice if it did.

 

[quark]

Artisi got it Ofcourse, I took that he meant discharge head of the pump.

 

[Artisi]

The important thing to consider when reducing speed and its relationship to minimum flow is that, the flow is reducing by the ratio of the speed change while head (pressure) is reducing by 1 fourth and power by 1 eighth, therefore the factors which influence minimum flow are reducing at a far greater rate than the speed change - things like radial load / stress on the shaft, internal recirculation, mis-match of flow onto the impeller and pre-rotation of the inlet flow etc etc -

I guess at some point (speed) the minimum flow really becomes irrelevant as it causes no real problem - not even heat build up as the inefficiency from the power input in relation to the mass of the pump and volume of water is next to nothing. Therefore I would like to repeat my earlier statement - "I would suggest that in all cases such as this an engineering decision based on all factors pertaining to your particular pump / installation is the best approach."

For me - a line drawn from the minimum flow rate point on the full dia/ design speed curve to the 0/0 point on the H/Q curve axis would probably be more than sufficient and on the safe side for all normal purposes - looking for factors or de-ratings etc is just an academic discussion which doesn't lead anywhere.

 

[TenPenny]

Artisi got it Of course, I took that he meant discharge head of the pump.

And if the speed reduces by one half, I would hope even the discharge (which technically, should be differential) head doesn't reduce BY one fourth, or I've been doing something wrong for years.

Speed x 0.5 gives Head x 0.75???