Droop Instability 3

[PUMPDESIGNER]

Curious,
I strive to eliminate droop on all pumps in our system.
I notice that many pump companies do not do so, some of the largest companies included.

I figured it was always due to their excessive drive for maximum efficiency for sales purposes, you know, "Our pump pumps more than their pump", mindless dribble.

Am I correct about this assumption?
Other factors could be laziness, not caring, or don't know, or they know but do not think it important.

I figure a few percentage points of efficiency is well spent to eliminate instabilities.

What do you guys think or know.

PUMPDESIGNER

 

[Artisi]

Pumpdesigner
Not sure exactly what you mean by eliminating droop. Are you refering to extending the pump curve beyond BEP.

Naresuan University
Phitsanulok
Thailand

 

[Artisi]

or the droop towards shut head on some units.

Naresuan University
Phitsanulok
Thailand

 

[PUMPDESIGNER]

Droop I suppose is a colloquialism referring to drop in head as the flow approaches 0. More formally I suppose this would be called a flow instability approaching shutoff.

Usually takes some casing mods to eliminate, costs some efficiency points.

PUMPDESIGNER

 

[jay165]

Would not a minimum flow spillback line solve the problem?

 

[TZellers]

I've seen many performance curves are more artwork than reality. Take the NPSH curves for instance. They always draw them straight back to shutoff, when in reality they should rise.

Sales and marketing folks, eager to get their numbers, will always fudge things to achieve their goal. A drooping curve is a no-no generally speaking, but is more prevalent than most think.

Also, many curves are estimates, or calculated, not entirely derived from empircial tests.

Why? Because they can. Sad but true.

 

[scalleke]

Hey guys,

Having worked for a pump manufacturer I can guarantee that we NEVER fudged our curves.

If you blaim someone for straight curves, blaim API for asking only 5 points to determine the curve.

Droop is a design issue. Make your unit concentric (easy to machine) and the curve will droop. Mchine a volute, (more difficult) you get more efficiency and rize to shutoff.

Pay the small price difference and you will get a rizing curve.

Also Concentric = drooping curve = fixed direction load on shaft = easy bearing selection for the complete curve.

Volute = rizing curve = changing direction of load on shaft for different pump loads = difficult bearing selection.

So either you have a rizing curve and your operation point might be limited by bearing selection or drooping curve and your operational range might be limited by instabilities of the curve.

The unit that will work best is concentric, use a discharge orrifice to get a rizing curve to shutoff, loze some efficiency and have a mechanically solidly selected bearing.

Long life and stability are often paid for with some efficiency points.

est regards.

Scalleke

 

[vanstoja]

Stepanoff,A.J.,1957, "Centrifugal and Axial Flow Pumps" has a curve of typical head vs flow for 7 specific speeds showing that the two lower double suction specific speeds, 900 and 1500(RPM.GPM,Ft.) have drooping heads toward shutoff. This figure is in the 1976 Pump Handbook on page 2-132. Since specific speed is essentially an "impeller-only" characteristic that does not account for pressure recovery design features downstream of the impeller, it suggests that low specific speed impellers are more prone to mechanical/frictional losses that tends toward a parabolic head vs flow characteristic shown in another figure in Stepanoff. At higher specific speeds, pressure recovery design features like vaned and vaneless diffusers are prone to flow stalling at flowrates below best efficiency initiating reverse flows through the impeller, prerotation in the pump inlet piping and coswirling inlet flow which unloads the impeller and causes reduced head.
From the above, it might be concluded that avoidance of head droop in centrifugal pumps at low flows represents an extraordinary challenge to pump designers not particularly well-solved by past pump design practices but most likely solvable with current Computational Fluid Dynamics (CFD) design methods that can minimize targeted losses and avoid flow stalling in blade channels by judicious contouring of impeller blades, etc. Pump manufacturers might not have had sufficient incentive to correct drooping head problems, if found during engineering testing, unless the customer had specified a rising head-flow curve toward shutoff such as is done in some Military Specifications for centrifugal pumps. With CFD tools in hand, the manufacturer may now be more inclined to eliminating drooping heads as an improved interpretation of "good design practice". I'm inclined to agree with Scalleke, that HI Standard allowance of only 5 test points is not conducive to finding drooping head characteristics during tests or the critical phenomenon of possibly multiple unstable flow regimes between best efficiency and shutoff or minimum thermal flowrates. For more on this subject see vanstoja's 8/1/02 input to thread407-27055.

 

[vanstoja]

Correction-The thread discussing testing practices is thread407-29954 with vanstoja's 8/21/02 and 9/22/02 inputs.

 

[vanstoja]

Two different centrifugal pump conventional design method,ie,specific speed based and area ratio based both indicate that drooping head flow curves are inevitable for certain design parameters. Lobanoff and Ross, 1985, "Centrifugal Pumps-Design and Application" recommend combinations of impeller blade number and discharge angle and show in their Fig.3-2 that 3,4 and 5 blades with 17(or 15),20 and 23 degree blade angles do not droop but that 6,7 and 8 blade impellers with blade angles of 25,27 and 28 degrees do droop. They use specific speed to select the blade number/angle for a specific head rise percentage from bep to shutoff. For 3 to 8 blades, the associated specific speeds (scaled,RPM-GPM-Ft.) for a 15% head rise are about 338,488,650,881,1313 and 2475,respectively. Their Fig.3-16 suggests that efficiency losses are of the order of 9 to 7 percent for each stepwise reduction in blade number/angle to maintain the same flowrate by increasing blade discharge width progressively.
Anderson.H.H.,1961,"The Hydraulic Design of Centrifugal Pumps and Water Turbines, ASME Paper 61-WA-320 has data on the ratio of impeller discharge to volute casing throat area showing that for area ratios,Y, of 0.75 to 2.0 head-flow curves rise without drooping toward shutoff but for area ratios of 3 and 4 they do droop. The scaled percentage head rise from bep to shutoff for Ys of 0.75,1,1.5,2,3 and 4 are respectively, 13.5, 15.5, 10.5, 6.2, 2.5(with droop) and 1.7(with droop).
Not being a pump designer myself, I would be interested in knowing just how a legitimate one like Pump Designer actually designs for zero head droop in typical designs or corrects existing designs that display head droop on test without resorting to advanced flowfield analysis techniques such as CFD. I've seen a little bit of how experienced pump/compressor CFD practitioners can conjur up exotic impeller blade shapes to reduce blade loading in critical parts of a flow channel where recirculations and backflows are initiated to possibly thwart drooping heads ordained by conventional design methods. However, as a semi-ignorant bystander, I can't offer any useful help to solve this problem.

 

[smckennz]

Hi Pumpdesigner

have a look at Pumping Manual (9th ed, Elsevier) chapter 5, "head characteristics". It presents Vanastoja's concepts, in graphical summary form.

In relation to sacrificing a couple of percentage points to eliminate the droop, this decision may be outside the designers brief and could belong in the specification. There are many applications with large pumps where a couple of percent is quite important but the droop effect is largely irrelevant. Consider a 1000kW pump running 24/7 and power costs 10 cents per kW-h. The cost of a couple of percent is around $17,500 p.a.

Cheers

Steve

 

[PUMPDESIGNER]

Sorry guys, I started this thread, then work smashed me.
I agree with everything in general that has been said, but specifically,
The most efficient volute pump design will have droop, reducing droop costs efficiency points.
In many applications efficiency is paramount, as smckennz points out, and the pump can be implemented properly.
My narrow focus for a few years has been smaller stuff, standard production pumps, used in highly controlled systems, droop is deadly in those cases.
But even in those areas droop is not a crime necessarily, pumps with droop have many uses if properly implemented, especially in water transfer, water features, etc. Big IF though, and don't expect the sales person to warn you about the droop, "customer beware", they say.

My gripe comes when marketing gets hold of it and attempts to hide droop by publishing deceptive curves and vague rough curves where the curve line is thick enough to cover 25 feet of head, on purpose? Not sure. I have one graph in my possession, published by an old name in the business, available on the internet, where the flow rate starts at 40 gpm, I never noticed it. Get the pump, wow, big droop from 40 gpm down to 0. When asked about that, their engineers tell me that they knew about it, has been that way forever, too late to change it now, no use starting trouble.

Problem 1 with Droop - The pump will have higher efficiency than any competitor without droop. Lose lots of sales due to less efficient pump. Educate customers and sell the non-droop pump, but cannot be everywhere and lose sales where we are not close to the customer.

Problem 2 with Droop - The pump cannot be used easily in multiple pump situations, or in systems that are highly controlled. Controls engineer must properly account for the droop, controls must be more complex sometimes, or at the very least controls must be more robust, which is what we try to do. Still, our lives are made simpler and more predictable when designing controls, to make sure no pumps enter the system with any flow instabilities. And lets not forget about all the pumps with midrange instabilities out there.

Finally, I am considering changing my handle. Look at my profile and you can see that I never claimed to be a pump designer. When I first came across this group I was only interested in looking at something quickly and I had to come up with a handle quickly, not thinking I would remain for any length of time. All caps is no good either. I am not a pump designer by training. I would not classify myself as expert. Have thought about changing that handle, but hated to lose familiarity with so many nice people.

PUMPDESIGNER

 

[PUMPDESIGNER]

To continue the discussion, what do you guys think of this quote from Stepanoff:
"A theoretical investigation has shown that the actual head-capacity curve is a parabola with its apex displaced to the right of the zero capacity Axis."

My first question is what study?
Then my opinion of the statement is that it is vague, but appears to imply that the "natural" curve of the most efficient design is droop towards shutoff.

Am I wrong about that? I have never been sure of what Stepanoff is saying there.

PUMPDESIGNER

 

[vanstoja]

The version of Stepanoff (1957?) that I have only part of says on page 167, "The Q-H curve on Fig.9.8 is a parabola with its apex displaced to the right of the axis of heads. Such curves are observable on low and medium specific speed pumps. But, when found objectionable, the droop of the Q-H curve at shutoff can be eliminated by special design, as discussed in Chapter 14. On medium and high specific speed pumps the Q-H curve rises constantly toward zero capacity, thus indicating the inaccuracy of the method outlined for establishing the total head-capacity curve from the input head-capacity characteristics." Fig.9-8 is captioned "Q-H curve is obtained by subtraction of hydraulic losses from input head." Shock losses dominate over friction losses in the flow range near shutoff where the droop occurs. I don't hold Chapt. 14 now so I don't know what Stepanoff's design solution is to drooping head at lower specific speeds.
I can't speak to the issue that devious manufacturer marketing practices attempt to hide evidence of undesirable drooping head or other performance anomalies like instabiliy but am more inclined to believe that such results are due to uninformed testing technicians who might skip over preplanned test points when they can't get stable readings of head, flow, power, etc. This is the kind of dangerous omission discussed in my previously cited thread input on testing. A case in point is the Fig. 29 performance curve in Pump Testing Chapt.14 of the 1976 Pump Handbook. Ten test data points are shown for BHP and head curves where the latter is drooping from around 1250 to zero GPM. The capacity interval is 500 GPM for 6 of 10 mid-flow points and 250 GPM at both ends of the flow range. There is no data point at 750 GPM between the two test points at 250 and 1250 GPM. Also, the nominal progression 2750 GPM reading has been displaced to roughly 2800 GPM. This curve tells me that the 750 GPM flowrate near the middle of the droop was too unstable over a rather wide range to operate the pump safely there. Also, near 2750 GPM there may be some other kind of "high-flow" instability that disappeared at the tested 2800 GPM flowrate. Was the manufacturer's pump designer told about these instabilities by the testers and did he elect to disregard their possibly adverse implications? As an evaluator of test data with an acute and bloody appreciation of the problems of hydraulic instabilities anywhere in the operating range, I would have demanded a retest of this unit and possibly followon units with obligatory data acquisition of some kind at or near these "skipped" test points so as to characterize the flow range and assess the prospective damage possibilities of prolonged operation at dangerous off-design flowrates.
Don't change your name, Pump Designer. We'd all be too confused to continue in the Pump Forum without your guidance and astute observations. I should have read your bio before today.

 

[PUMPDESIGNER]

Thank you vanstoja,
I'll keep the name.
I have Stepanoff 1957.
Nothing about this on page 167 in my book.
Only place I know of is page 293.
Just for clarity, here is more of the quote:
"A theoretical investigation has shown that the actual head-capacity curve is a parabola with its apex displaced to the right of the zero capacity Axis. The shutoff head is lower than the maximum head. Actual test curves of low specific speed pumps (Ns =1000) approach this form. Under certain operating conditions part AB of the head capacity curve is unstable, and this instability results in head capacity fluctuations or failure to pump entirely."

Then Stepanoff starts a very good section about fluctuations, swings, vibrations, when parallel pumps are involved, most often observed when free surface water source and destination are involved. Scary thinking about having two poorly characterized pumps start something weird, and with Murphy's law, that would be overseas on a high profile job.

I don't generalize about pump manufacturers either, but I am aware of bad information and incompetence. Large pump companies can be schizo, one guy is perfect and careful, the next engineer a screw up.

I'll digest all this thread later.


PUMPDESIGNER

 

[vanstoja]

The undated portions of Stepanoff I have must be from another (later?) edition rather than the 1957(?) edition. My quote came from Chapter 9 on "Hydraulic Performance
of Centrifugal Pumps" and seems to be a modified version of your quote. I lost my 40 year access to the complete book when I retired last year and am now reluctant to buy the latest edition going for about $87 on Amazon.