Centrifugal Pump Low Flow limit vs. Speed Reduction 1

[mrspcs]

Centrifugal pump manufacturers specify a Low Flow Rate Limit value for a given pump . In addition, there are a number of "rules of thumb" etc. to estimate the value of that low flow limit (e.g. don't go below 30 % of nominal flow etc.).

The manufacturer's figure as well as most of the estimates and calcs. found in literature refer to "choked flow conditions". That is, the pump is running at its nominal speed and the flow is reduced by means of a control valve or some other flow rate reduction device.

However, what happens when the reduced flow condition is the result of a reduction in pump speed ??
It seems logical that the Low Flow Limit value should go down for lower values of pump speed.

An example. A pump train consists of a number of centrifugal pumps in series, one of them is VSD (VFD) driven. Flow through the train is regulated by speed control of the VSD unit (which is assumed to be always -on-).
(There are guidelines as to which pump should be VSD driven, pump sizes and characteristics for adequate series operation etc etc.).

The pump on VSD could probably run safely at lower flow rates than its "nominal low flow limit" since its speed goes down and so it should its low flow limit. In this arrangement however, the critical Low Flow Limit will be determined by the pumps running at nominal speed (the manufacturer's indicated value for them).

In the case of a single pump - single VSD arrangement, it should be possible to run that pump safely at flow rates that may be below the value indicated as "limit" by the manufacturer and based on pump at nominal speed.
(lets assume that all other conditions are met, like seal temp., flow velocity etc).
This possibility may have an influence on control strategy (recirculation flow, low speed rpm limit for the VSD) and on potential energy savings by using the VSD. Therefore, it would help to be able to evaluate it.

Aside from consulting with the pump manufacturer (obvious alternative), -AND THIS IS THE QUESTION-: is there / does anyone know about ways to estimate Low Flow limits at pump reduced speeds from its known value at nominal speed ??
(e.g. a similar calculation to that of using Affinity Laws and nominal values to calculate pump characteristics at different speeds).
(this would of course be "ideal" values. There would be constrains from temp., seals performance, cavitation etc)

One of the few references I have found that hint at something like this calc. suggests that the Low Flow Boundary limit would be a function similar to an iso-efficiency line (e.g. Affinity Law lines). This function would connect all "low flow limit points" in the family of Head vs. Flow for the pump at various speeds.
(See S. Mirsky, "Pump Control Strategies Benefit from Compressor Know-how", Hydrocarbon Processing magazine, Feb., 2005).

By the way,I recently posted a similar question regarding the BEP point. In the article suggested by biginch,

http://www.gambica.org.uk/pdfs/VSD_Pumps.pdf#search="pump efficiency speed variation"

(EXCELENT article, by the way) there are some comments on this but not very specific.

Thanks and regards,


MS

http://www.3rps.com

 

[BigInch]

Do you really need to go lower than the manufacturer's suggested low limit?

While there may be something to your suggestion that lower flow limits can be achieved using VSDs, I would suggest it will be of little practical value for two reasons (talking centrifugals anyway).

#1) The manufacturer's won't warrenty their pumps, if you try to do it.
#2) The head generated at lower speeds where you may possibly gain something from running at rpms that might lower the min flow below 30% BEP would indeed be very very low, so if you have any initial static head to overcome at all, you won't be moving any fluid forward, and you might possibly be moving it backward.

So, I suggest it will be of little practical value to persue this much further, but I'm willing to listen to what others may say about this.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[Artisi]

Without spending a lot of time trying to figure out what you're trying to achieve and why - I will put a question to you;
Q; Why do manufacturers specify a minimum flow, usually at full impeller diameter and design speed?

If you look into this aspect of pump hydraulics / design you might answer the first part of your own question.

As to being able to calculate the Low Flow limits - what is the purpose and what will you gain from this information?

Like BigInch - I'm always willing to listen and learn.

Naresuan University
Phitsanulok
Thailand

 

[quark]

I had a lengthy discussion about this issue with pump manufacturers the first time I attempted to use variable flow pumping systems. Static head, in my case, was not at all a problem because it was a closed loop system for air conditioning. Manufacturers agreed that the safe minimum flow reduces with reduction in pump speed but didn't dare to quantify it on paper. Further, They were reluctant to take chances with motor cooling (TEFC type). There were times when I run the pumps between 15 to 20% of BEP flow but I left the company after one year. No idea about what is happening now.

With closed loop systems, I am ready to try any time. Savings may not be substantial due to decreased efficiency but you have room for lesser power consumption. BigInch is dead right if it is an open system.

 

[mrspcs]

There certainly are limitations to the use of VSDs with Centrifugal pumps. However, I wouldn’t put it in terms of closed vs. open systems. It seems to have more to do with system characteristics like static vs dynamic dominant heads. VSDs obviously don’t go well in applications like boiler feed water or oil well water injection systems (high static heads) but should do very well in oil terminal applications (assuming low or even slightly negative static head. I have a good example of one of those). And of course in the process industries there are many cases that offer advantages with VSDs.

Part of the problem is that, as quark correctly points out, manufacturers are not willing to provide consistent data on some aspects of VSDs applications. Their position is simple and very comfortable: the pump is designed to work at fixed speed and all operational data provided is under that assumption (period). Anything else will void the warranty.
However, it is the pump users who pay expensive bills for electricity and pump spare parts (the result of operating under choked flow conditions).

The motors are a different story. The Converter and the Motor manufacturers are usually one and the same. There is consistent data on operational specs and limitations for motors in VSD applications (vs. fixed, nominal speed operation).

It isn’t common for a pump manufacturer to have solid experience with VSDs or for a Converter manufacturer to have solid experience on pumps. Result: users are often times left to their own devices.

At current energy prices, there are significant payback advantages in the use of VSDs derived from energy savings and from improvements in control systems with the VSD as final element.
It is the task of engineering to take these concepts to practical applications.
Systems and operating practices and limitations need to be designed based on adequate engineering data and calculations (and some inevitable assumptions). And that’s what I am after.

Simply sticking to very conservative figures and huge safety margins (e.g. “not below 30 % of BEP flow”…hell, be safer, make it 50 % !!!!) based on conditions different from actual intended cases won’t do it. -Over- or -under- designing seems to me to be equally poor engineering.

An example based on the case I mentioned above (oil shipping). For tanker topping (end of a load) only one of the smallest available pumps is left -on- and that is at near 15 to 10 % of its rated flow capacity (depends on tanker’s requested flow). Under that condition, often times the HH discharge pressure interlock is activated, the operation is stopped and may require re-starting (time wasted).
Recirculation is not practical. Ideally, it would be to the same origin tank but that requires tanks capable of doing two operations at once (in and out lines, not often the case). Recirculating to other tanks brings about mixing problems. It would also need control valves on the recirc. lines (currently only on the loading lines).
True, the design of the facility has a number of legacy shortcomings. For the most part however, a VSD in one of the pumps would solve most of them. The potential payback is huge just by avoiding wasted time (energy savings is not the big issue in this case).

For the control system design it would be nice to know the true low flow limit for that pump working at reduced speed (and NOT at the old ultra-high margin choked flow condition). One of the alternatives could be LL speed interlock in the Converter and/or a small recirc. line with on-off valve, not CV, going back to the suction manifold ways upstream … etc etc ).

(kindly excuse the length of the post. It won’t happen again)



MS

http://www.3rps.com

 

[DesignerMike]

I agree with mrspcs. I have quite a few customers using VFDs to control the performance of centrifugal pumps with good results.
None of them are using multiple pumps is parallel, but the pressure and flow control (with the appropriate controls) is easy to adjust and the pumps can be used for many more applications that just what it was sized for "the CIP pump for tank 45"...with the VFD, now this pump can do many more jobs (and do them better).

My vendors can give me the curves for my different duty points and I will pick the most versitile pump for each app.

Very much agreed that most manfacturers don't want to hear about using a VFD for this...probably because they want to sell you another pump and it is more work for them to size it properly (if they can)

 

[ccfowler]

mrspcs,

I browsed the reference that you cited (

http://www.gambica.org.uk/pdfs/VSD_Pumps.pdf#search="pump efficiency speed variation"),

and found one potentially misleading item. Figure 3.1 shows a charcteristic curve for NPSHR that seems atypical for centrifugal pumps--it decreases progressively for flow rates below the best efficiency point. Normally, the minimum NPSHR is found to match the flow rate at best efficiency point and then rises for decreasing flow as well as for increasing flow. Another (and somewhat more "fussy") item to consider in Figure 3.1 is that the head vs. flow curve of many pumps does not rise continuously to shut-off.

Another aspect of NPSHR that has been "beaten to death" in numerous discussions is that the NPSHR curves represent flow conditions where performance is already impaired by either 1% or 3% depending upon the standards applied. The "cavitation free NPSHR" can be much greater (multiple times the "curve NPSHR"--not just increases of a few percentage points). This is why NPSHA always needs to be much greater than NPSHR for good pump operation. In reality, the tolerable NPSHA vs. NPSHR relationship is usually an unpleasant compromise between initial cost, operating costs, maintenance & repair costs, physical space constraints, etc. (I've known cases where seemingly great periodic maintenance costs were gladly absorbed because of the even more burdensome costs that other alternatives would have imposed. It is a rare engineering decision that doesn't involve substantial compromises!)

In defense of the pump manufacturers (please note that I have never been affiliated with any pump manufacturer), users must recognize that the variety of conditions that any pump model may be presented with are nearly infinite, and the potential for legal liability issues and customer dissatisfaction is also nearly infinite if the pump manufacturer would risk attempting to somewhat wildly (and perhaps somewhat blindly) extend the nature of their generic performance claims. When a customer (or potential customer) presents a reasonable range of conditions to a pump manufacturer, then the manufacturer can address these needs in a coherent manner. Like it or not, the pump(s) and their connected system ALWAYS operate as an integrated entity. Pump failures and operating problems do happen, but most often, the problems arise from the system presenting the pump with conditions that exceed the pump's designed capabilities in some manner.

Oh, well! Enough ranting by an old geezer on these subjects for now.

 

[BigInch]

Total head includes the velocity term V^2/2, hence it is reasonable that a portion of total suction head, and NPSHR, reduces with flow.

Theoretically, NPSHR should correspond to the suction specific speed at the new rpm.

I've actually had 1 major pump mfgr resist changing their minimum flow requirement, even when it was shown to be obviously "wrong". After proving to them the number was unreasonable, they eventually caved in and lowered the number, but it still wasn't easy to get them to do it.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[Artisi]

The comments re NPSHr and the H/Q curve posted by ccfowler (Mechanical)are absolutely correct but is something most people are unaware of and the first time I have seen it mentioned in this forum, especially the point of NPSHr INCREASING as the duty moves to left of BEP.
Actually I prefer to call it "apparent" NPSHr, actual NPSHr is falling but extra energy is required at the impeller inlet as the flow rate reduces resulting in a mismatch of flow related to impeller rotation. It can or should be reflected as a required increase in inlet pressure and best represented as NPSHr. Some manufacturers of large pumps will show NPSHr as a falling / rising line on the graph with an additional line rising left and right of the point of lowest NPSHr.

"Normally, the minimum NPSHR is found to match the flow rate at best efficiency point and then rises for decreasing flow as well as for increasing flow. Another (and somewhat more "fussy") item to consider in Figure 3.1 is that the head vs. flow curve of many pumps does not rise continuously to shut-off."

This does not necessarily address in full the original posting but is one of the reasons that most pump companies will set or show a minimum flow on their pump performance curves.

I also fully agree with ccfowler and his posting regarding his defense of the reputable pump manufacturers and the misapplication of pumps as probably the major cause of operating problems.

Naresuan University
Phitsanulok
Thailand

 

[BigInch]

Art, I didn't mean to imply that NPSHR should fall continuously with reducing velocities, only a component thereof, which affects I agree could be easily overwhelmed by other factors, but generally correspond to the typically falling NPSHR characteristics (at least to a point).

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[Artisi]

BigInch
Comments were of a general nature as information only. I hadn't / didn't connect your comments as being specific to my reply.

I will also reword my earlier comment from the following:
from;

"Some manufacturers of large pumps will show NPSHr as a falling / rising line on the graph with an additional line rising left and right of the point of lowest NPSHr."

to
........ rising left and right of the point of BEP on the NPSHr curve.



Naresuan University
Phitsanulok
Thailand

 

[mrspcs]

ENG TIPS ANSWER

At this point, I'd like to try to make a summary of my understanding of the answers / subject (effect of flow control by speed reduction on some pump characteristics).

1) Objective
How to calculate/estimate operational characteristics and limitations of Centrifugal pumps at various speeds from known data and values at nominal speed (thread focus: “Low Flow Limit” at reduced speed) with the objective to:

a) Properly design their control systems (based on speed control as opposed to throttled flow) to obtain the maximum possible operational and energy savings advantages.
b) Do so in a SAFE manner for the process and the equipment.

(Note: perhaps for the sake of discussion the type of pump and nature of the fluid should be further limited, otherwise the possibilities could be too wide).

2) Applicability of literature references
An example to illustrate the point. A literature ref. in connection with “pump speed control” indicates something like -at low flow conditions, the pump is operating far from the BEP and, therefore, at low efficiency. This increases heat dissipation at the pump which could cause fluid temperature rise thus increasing NPSHr-.

That is true in the case of throttled flow / pump at nominal speed. However, the ref. in question seems to “extrapolate” the concept to low flow rates resulting from pump speed reduction where, I think, that may not apply.

In fact, there are cases where a low flow rate value, resulting from pump speed reduction, may be closer to the BEP (e.g. at higher efficiency) than the same flow rate value when the pump is at nominal speed and choked flow.

Most of the literature material on pumps operational characteristics and limits is based on throttled flow conditions. It seems to me (personal opinion) that some times this information is applied to speed reduction conditions without truly considering its validity in those conditions.

Also, many literature ref.’s on VSD-driven pumps seem to be totally off reality. Many go on to describe what would happen in a system “at twice the pump speed”. In reality, most applications of VSDs in connection with pumps are to use them instead of Control Valves as flow control final element. In that context, the idea is to move -down- from nominal speed. I believe a VSD could drive the pump over its nominal speed but that would be for limited amounts of time (determined by VSD rating) and by a small % over the nominal speed.

3) Low Flow Limit at different Speeds
To determine possible flow rates and flow control ranges for a VSD-driven pump, the true Low Flow Limit at reduced speeds is needed. Using the same value as for nominal speed is a very conservative approach that precludes some advantages when using variable speed.
I have found ONLY ONE literature ref. that could be used to calculate this.
It is the one in the ref. mentioned in the opening post (S. Mirsky, "Pump Control Strategies Benefit from Compressor Know-how", Hydrocarbon Processing magazine, Feb., 2005). The following graph is from that article:

One possible interpretation (my opinion) is that the Low Operating Boundary (LOB) line follows the shape of an Affinity Law iso-efficiency line. These laws could then be used to calculate the Low Flow Limit point (on the H-Q curve) at different pump speeds from the known (manufacturer supplied) Low Flow Limit value at the pump’s nominal speed.

QUESTION: Would this be an acceptable approach ? Any alternative suggestions ? (would appreciate comments)

The point so calculated would be an IDEAL value and should be subject to a number of constraints (depending on pump and fluid type, system configuration etc.).
Some of them are (excluding electrical issues, usually addressed by the VSD and Electric Motor manufacturers) :

a) Shaft Critical Speed(s) and/or Shaft Deflection
There may be critical speed values within the proposed speed range for pump operation.
Operation too far from the BEP may cause shaft deflection (need different L3/D4, gear boxes etc.)

b) Pump Seals: must be adequate for the proposed speed range.

c) Fluid Properties: are they likely to change with speed ? (e.g. viscosity of non-Newtonian fluids).

d) NPSHr
The following quotes are from material specific for VSD-driven pumps:
From McNally Institute:
“At higher shaft speeds, the NPSHr is increased to prevent cavitation”
From the Gambica / BPMA article:
“NPSHR increases as flow through the pump increases”
“NPSHR varies approx. with the square of speed in the same way as pump head and conversion of NPSHR from one speed to another can be made using the following:
Q [proportional to] N and NPSHR [proportional to] N^2 (where N = pump speed).
It should be noted however that at very low speeds there is a minimum NPSHR plateau, NPSHR does not tend to -zero- at -zero- speed.”

QUESTIONS (for ccfowler):
interesting comment that NPSHR “rises for decreasing flow as well as for increasing flow” (e.g. flow rates away from BEP in either direction).
Is this connected (I think it should) to the fact that Efficiency -decreases- in either direction from the BEP ??
Is that statement true whether increasing and decreasing flow rates are the result of throttled flow (e.g. pump at nominal speed) or pump speed changes ??

(Note: your points ref. Fig. 3.1 in the cited article. Perhaps this has to do with the previous note above. Seems that this Fig. is based on -radial flow- centrifugal pumps, different from -axial- and/or -mixed- flow types. In fact, maybe the entire article is based on that premise.).

QUESTION (for biginch):
“Total head includes the velocity term V^2/2, hence it is reasonable that a portion of total suction head, and NPSHR, reduces with flow.”
I can see that at higher flow rates, the friction in the suction line will be greater and this will reduce suction pressure (e.g. NPSHAvailable is reduced, hence the difference between Required and Available is also reduced).
But how can that reduce NPSHRequired (which seems to depend more on fluid temperature, for instance) ??
I think your second comment on that point clarifies the idea some but I am still not sure I understand.

(thanks to all for the comments and time dedicated to them)


4) Pump manufacturers (my view, based on my own experience).
It would be unreasonable to ask a manufacturer to put in writing commitments or guarantees of pump performance outside their recommended limits (if any of them did, probably half their legal dept. should be fired).
On the other hand, it does not seem reasonable to take the position that pump operational characteristics and limits are the same in the case of flow regulation by speed control and the case of throttled flow with the pump running at nominal speed (in this case, half their engineering dept. should be questioned -maybe not fired… trade sympathy-. Either that or stop talking about “innovation”, “engineering leading the way”, “ new technology”, etc.).
Instead of reluctance to discuss the subject (some cases) a more reasonable attitude, in my opinion, would be to discuss the subject under the admission that maybe there isn’t enough data or experience to even suggest figures. A hint that maybe some talent, time and money should be invested to that effect ?

Perhaps a joint program [ pump manufacturer / user / AFD manufacturer ] ??
(notice how the “user” is “caught in between”….).





MS

http://www.3rps.com

 

[BigInch]

Fluid temperature concerns related to NPSHA (or R, its all relative) are better treated separately when condering the product's vapor pressure. NPSHR is <I think always> referenced to clear cold water, so its probably better that everybody should keep discussing it within that framework. I suppose that a highly specific pump design for one particular product could warrent a specific product NPSHR test, which would be nice to have, but that's Space Shuttle stuff as far as I know.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[ccfowler]

mrspcs,

Usually, the NPSHR curve will be somewhat symmetrical about the minimum NPSHR point, and this is due to the less favorable flow patterns away from the BEP. This is at CONSTANT shaft speed. Usually, the NPSHA increases as flow decreases due to reduced frictional losses in the suction piping, so the NPSHA and NPSHR curves will diverge less severely than they will for flows greater than BEP. The NPSHR characteristics of radial, mixed-flow, and axial pumps are generally very similar in nature, but obviously, the precise quantities vary.

When the pump is operated at adjustable speed and the connected system has only frictional losses (such as a closed circulating system), a pump can operate over a very wide speed (and flow) range without difficulty. In such as system, the NPSHR at BEP (minumum) decreases as the shaft speed decreases, and usually, the NPSHA increases as the flow decreases. Note that such systems have very distinct maximum flow limits usually most distinctly set by the capability of the driver since the power required varies with the cube of shaft speed. Where effectively infinite power is available, the pump (and connected system) will suffer progressively greater wear rates (and other problems) as the shaft speed and flow increase beyond any reasonable design point.

Pumps don't understand throttling. They simply see an increase or decrease in the system head. The adjustment of a throttling valve serves to shift the system's head vs. flow curve to match a different point on the pump's head vs. flow curve. Similarly, the adjustment of the pump's shaft speed serves to shift the pump's head vs. flow curve to match a different point on the system's head vs. flow curve. Pumps AND systems ALWAYS operate on their respective head vs. flow curves, and the NPSHR, NPSHA, power, & efficiency then are determined by the actual operating point.

This leads to the matter of a common misunderstanding involving the application of an adjustble speed drive to a pump. The expected energy savings of such applications may not be realized if the pump is sized so that it is not operating reasonably close to its BEP. For example, let's assume a pump has been selected "conservatively" (the pump has substantial excess capacity) for a circulating system without a throttling valve. This causes the pump to always operate at a flow substantially below its BEP wasting power and energy all the while. The main lesson here is that the use of an adjustable speed drive does not reduce the need for thorough understanding about the pump and system performance characteristics and needs.

In general, circulating systems with varying flow rate needs are an ideal case for a properly sized adjustable speed drive for minimized operating costs. Where a system has closely fixed flow requirements, the application of an adjustable speed drive most likely represents a waste of materials since a constant speed drive can serve well at comparable efficiency. Where the pump must operate against a very large fixed head that varies little with flow rate (boiler feed pumps are an excellent example), the costs and benefits of an adjustable speed drive vs. fixed speed drive are not readily obvious. Only careful study of the specific application can clarify which type of drive is most beneficial.

As always, enginering desicions involve substantial compromises, and it is a rare situation where there is only one possible solution.