End of Curves 1

[ghoshtathagata2000]

It is being seen as a trend in the Middle East Oil & Gas companies specification that the drivers shall be designed for the ënd of curve operation. Also I understand that it is common practice to size the driver for end of curve also pumps operating in parallel.

What is the definition of end of curve and how is it achieved in practice? I am sure it is not 120% BEP as defined by the max. allowable flow in API 610 . During a test I witnessed I observed that the OEM described the end of curve even with the discharge valve not fully open which for me is not really the "ënd of curve"

By the way, if the driver is oversized, is there any way the pump operation is limited other than high vibration concerns?

Kindly share your experience

 

[DubMac]

My understanding of "end of curve" for specific applications has always been the point where pump performance begins to degrade due to NPSH deficiency.

Specifically, where the the pump's head is degraded by 3% (may be 1% now) due to insufficient NPSHA is the end of the curve. Note this will change with the NPSH available, so when defining end of curve, it is really application specific. There is also nothing wrong with a spec that asks for a driver coverage up to 120%BEP as long as it is clear that pump must have sufficient NPSH.

In general, the manufacturer's head capacity curve ends on the right side where the pump started to cavitate during original pump performance test. NPSH will typically be the limiting factor on the right side of the curve.

 

[1gibson]

NPSH will be the limiting factor in a lot of applications, but not during a factory performance test.

For vertical pumps, most will be tested in an open sump, even if the final configuration is in a barrel. NPSHA is head due to elevation (32ft at sea level) plus the submergence, so not very likely to limit the flowrate.

Horizontal pumps being tested for performance, test loop capability should be able to provide ample NPSH for "end of curve" operation.

You can argue all you want, but "end of curve" is either a defined flowrate (120% of BEP, or the flowrate with no NPSH margin) or in the absence of an agreed upon flowrate, it is where the line shown on the submitted proposal curve ends.

Some ways that pump flow rate can or should be limited: NPSH, vibration, thrust, fluid velocity, noise, common sense, "recommended operating region" and "allowable operating region"

 

[JJPellin]

I agree with 1gibson. The end of curve is the end of the published curve. The max flow with no back pressure is stonewall. The limit in a given service for NPSH, HP or head are defined by the interaction of the pump with the system, so they really aren't simple properties of the pump.

Johnny Pellin

 

[Artisi]

In practice, selecting a motor for end of curve operation can be overkill. The maximum motor size selected should be an engineering decision based on many parameters.
1. can the pump achieve end of curve operation in any abnormal operation, a pump with high static head may not be able to achieve end of curve.
2. What is the pump / system doing at end of curve, NPSH may well limit its operation.
3. selecting a non over-loading motor may well put you into a vastly over sized motor, needing bigger cables and start gear,heavier base plate, bigger coupling, more space etc

The motor should be selected for normal operation plus some margin for upsets, in many cases the power requiremnents might fall between 2 motor sizes with the larger motor having sufficient margin - any increased power requirement due to abnormal operation could be catered for by overload protection in the control panel.

Oversizing the motor doesn't cure any hydraulic operating problems.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[ghoshtathagata2000]

Thanks for your opinions.

DubMac, I checked the NPSHr curve for the pump. Even at very high flow rates the NPSH difference was very high. So I don't think NPSH was the problem

1gibson, the estimated curve was only upto 630 m3/hr but the OEM tested it at 810 m3/hr. You are very right in your opinion but I cannot interpret why was that test point so very different from predicted performance curve.

By the way, is there any way to know when the stonewall starts fior pumps. This phenomenon is known to me only for cent. or axial compressors and not for pumps.

Artisi, I can mention one hydraulic case where the end of curve is probable i.e. when thwo pumps are operating in parallel and one of them trips. In this scenario I believe we have to size the driver considering the end of curve criteria.

 

[GPRnD]

I've also seen this end of curve trend among EPCs working with middle east oil companies.

IMO its a pretty mindless practice with little thought put into the process. I've been asked to revise curves to reduce the end of curve point so as to limit motor power.

If you want to size for "end of curve" then that should be defined as 120% of rated or 120% of BEP whichever is greater.

For most pumps in API 610 applications you should not be operating past 120% of BEP from a vibration and hydraulic thrust standpoint.

As to the "stonewall" point, that depends greatly on the specific speed of the pump and the impeller design. What will happen is either the flow incidence angle will cause pressure side vane cavitation in the impeller, or the volute/diffuser throat will cavitate (so called discharge side cavitation). In both cases this will choke the flow.

A low specific speed could choke at 110% of BEP, a high specific speed pump with impeller vane incidence angles set for an excessively high flow, could run out to 150% of BEP.

 

[CaracasEC]

If the pump is auto started then the motor should be sized according to end of curve.

 

[Artisi]

CaracasEC - Why??

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[JJPellin]

It looks like the last reply by the OP defines the need. If both pumps are running in parallel with a total flow greater than either pump could produce individually and one pump trips off, it may be important that the other pump does not trip off on overload.

This is a common need within our refinery. Perhaps an example might help. In most of our services, the flow is controlled by a control valve attempting to achieve set-point. For one of our crude units, we have three large crude booster pumps running in parallel. If one of them were to trip off, the other two would take up the load as the control valve opened to hold the total flow constant. If the motors on the pumps were not sized for end of curve, then the two remaining pumps would overload and trip off and all crude flow to the unit would be lost. This would result in a disastrous unit shut down costing many millions of dollars.

In my example, there would also be another economic driver to size the motors for end of curve. When one of the three pumps is down for maintenance, we would want to maximize the flow from the remaining two pumps. Even if this put the pumps at a flow rate that was not within the preferred operating range, we might decide to operate there for a few hours in order to capture a high unit incentive.

In the end, most of the comments above are correct. In some services, there is a very great need to size the motor for end of curve. In other services, there might be no need at all. Each service should be analyzed. But, for our plant, the default answer is to size the driver for end of curve operation. We can decide to deviate from that based on the results of our analysis. But, in case a project slipped through without being fully analyzed, we would prefer to have the driver slightly oversized rather than slightly undersized.

Johnny Pellin

 

[CaracasEC]

I strongly agree with JJpelin. On our current construction, Ethylene Cracker Plant, as part of the owners representative, we require that all auto-started pump motor be sized according to end-of-curves. This is to avoid plant shutdown due to motor overloading.

 

[1gibson]

"1gibson, the estimated curve was only upto 630 m3/hr but the OEM tested it at 810 m3/hr. You are very right in your opinion but I cannot interpret why was that test point so very different from predicted performance curve."

I'd refer to that as a "proposal curve" which is slightly more accurate description than "predicted performance curve" in the following regard: what you were quoted and you purchased is a pump with an "end of curve" at 630 m^3/hr. If you operate beyond that flow, the pump warranty might not cover anything that could *potentially* be caused by that operating scenario. Meaning, just about any pump problem you can think of.

Out of curiosity, what % of BEP is the 630 m^3/hr? And 810 m^3/hr?

 

[Pumpsonly]

1gibson, the estimated curve was only upto 630 m3/hr but the OEM tested it at 810 m3/hr. You are very right in your opinion but I cannot interpret why was that test point so very different from predicted performance curve

The below statement from bradshsi should give you the answer.

IMO its a pretty mindless practice with little thought put into the process. I've been asked to revise curves to reduce the end of curve point so as to limit motor power.

 

[Chance17]

When the discharge pipe is empty, all pumps initially operate at end-of-curve.
Fluid friction backs the performance up the curve.
The motor is sized at end-of-curve to avoid overload when pipe is empty.
The practice also provides a motor that matches the entire operating range of the pump.

 

[GPRnD]

JJPellin gives a good explanation regarding the incentives to operate a pump outside of the recommended range for short periods of time.

We know it happens particularly when operators make trade offs between pump longevity vs maximizing production.

The question though is what is the "end of curve". It certainly isn't the exact point the manufacturer tested to. Since end of curve will often be limited by NPSHa and since that NPSHa (at higher than rated flowrates), is seldom provided when the pump is specified, the EPC is never really sure where that is.

The EPC fallback of course is to size the motor for whatever the end of the published curve shows. However as discussed above this is not where the pump could potentially runout to... This does not seem a satisfactory way to size motors since your motor could still be undersized.

Only proper system analysis can determine true end of curve and in my experience it is seldom performed. More often the operators use the smoke method. That is they run the pumpset until something starts smoking. That's your EOC


Sizing your motor to fill empty discharge pipes is bad practice IMO since this condition is seldom present and can be better regulated by using a discharge valve.

 

[DubMac]

The concept of "End of curve" coverage for the driver is not at all anything new; refinery and chem plant end-users have had the concept in their specs for years. As the term "end of curve" is not very quntitative and needs to be defined much more clearly; most seasoned end-user's specifications have tried to nail that point down in several different fashions.

Even though a pump may runout beyond the end of published curve, the manufacturer is in no way bound to any extrapolations of the curve by varying operating points. There are a lot of commercial considerations to opening yourself up to that kind of vague wording.

I'm not sure if the same is true today, but I was trained at Worthington years ago by I. Karassik and I do remember one of the immutable rules for pump manufacturers was that we only will guarantee ONE rated point for each pump sold. Although the pump may operate all over the place on the curve, ONLY ONE POINT was guaranteed for pump performance. Any "full coverage" driver language should be tied to that ONE rated operating point rather than the vague "end of curve" terminology.

You can really open yourself up to all sorts of liability claims if you allow such open-ended language as the "end of curve" only approach. As stated earlier, if there is not sufficient NPSHA to operate at end of the published curve, what good is it to provide driver coverage there??

API 610 uses a very sensible approach that requires drivers to handle X% of Rated design capacity with X varying according to driver HP (this is a bullet point and needs clarification by specifier).

O&G end-users specs vary all over the place but have generally accepted the "% over rated capacity" approach to size driver along with other variants thrown in from time to time. Another approach used in specs is to size the baseplate to accept the largest driver possible even if that size motor is not supplied. It sure saves money not to have to dig up foundation and replace baseplate if larger driver is req'd later.

At the end of the day, you will provide driver for whatever the customer determines is "end of curve" coverage; but contract should be carefully worded to protect the supplier from unintended liabilities from "rogue" operation.

 

[SeanB]

End of curve conditions are sometimes specified if the pump is on level control. If there is an upset in the column or vessel the level control valve can go fully open to reduce the level. In such a case the system pressure is reduced and the pump can actually go to the end of its performance curve.

 

[BigInch]

The examples given are for appropriately sized pumps with proper curves for the design process flowrate, in which case there are enough pumps running such that if one trips the others can handle the design flow within say a reasonable 120% maximum load, not for the cases where a trip of one will throw 150% load on one pump and requiring that the pump be selected with that requirement in mind. That is obviously a mindless applications of a generally good rule of thumb and that is where a lot of mideast specifications go wrong. Taking a good general rule of thumb, then extending it to the point where it makes no sense. There are many examples of these that I have found during my mideast work. One is combining B31.4 with B31.8 to make a hybrid design code not strictly, or sensibly, applicable to either, making it required to use highly conservative gas pipeline design factors for liquid lines, which can result in some pipe walls that would otherwise be considered as 144% to 180% overdesigned than if B31.4 code was used alone. But ... if you're the biggest and the only elephant in the room, you get to make the rules ... any rules you want. The benefit is that it makes selling VFD's to go with them that much easier.

We are more connected to everyone in the world than we've ever been before, except to the person sitting next to us. Lisa Gansky

 

[stanier]

In a webinar the Hydraulics Institute reckoned that 80% of pumps have been incorrectly selected. With emphasis on reducing energy consumption with carbon taxes etc. People are now starting to look at pump energy, carbon footprints, embodied enerrgy etc etc.

Perhaps now engineers will specify the pump for the duty without rules of thumb being applied by using more science. We have the tools to correctly size pumps.

What is the point of blindly specifying a motor for a duty point that cannot be achieved? Or a VFD when a control valve would be far more efficient in a system?

Oversize motors have higher embodied energy and are less efficient.



"Sharing knowledge is the way to immortality"
His Holiness the Dalai Lama.

http://waterhammer.hopout.com.au/

 

[Artisi]

Only 80%, I would think this is underestimated.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)