Impeller ratio in VSD application 5

[virata]

Hello guys,

Our specification specified that the ratio of rated impeller to its max impeller size is to be less than of equal to 98%. Our vendor has offered with 98.4% impeller ratio. Also its pump motor are VSDs and thus the speed can be increased to mitigate this as Vendor stated. Is this acceptable? What is its relation to VSD application?

Regards,
V

 

[LittleInch]

In theory, the fact it is VSD has no impact. You are asking the vendor to give you a pump to do a certain duty (differential head and flow) at a certain speed. He is very close to the limit, but I've never seen this 98% rule and given that most vendors allow something for errors and manufacturing, it would seem irrelevant.

In practice clearly the speed is not fixed and will vary based on process parameters pressure, flow etc so the fact the impellor is close to the maximum allowable would appear to be irrelevant.

The only thing is you would not be ale to replace the impellor with a bigger one, but if you've sized the pump right then there should be no reason to. ).4% of a difference is not going to change anything.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[BigInch]

API standard is to allow a 10% increase in impeller size for new pump installations.

you must get smarter than the software you're using.

 

[1gibson]

API 610 also allow a variable speed drive to be used to satisfy future head rise, or a change in hydraulics (swap out to a higher head/flow impeller) so yes it is relevant. The intent here is to avoid a complete model change if the pump underperforms in the field due to bad system curve calculations. We all know that everyone who touches it adds a fudge factor, so the pumps end up oversized anyway, but that's the intent.

As much as it pains me to say it, just ask the vendor the increase the speed a few rpm on the proposal curve, so it reduces the impeller diameter to 98% of full. As long as the slip is realistic then it won't cause any problems. For an 1800 rpm motor, 1785ish rpm +/- 5 rpm, maybe a bit slower if it is less than 25hp, a bit faster if it's a couple hundred hp or more and the rated point of the pump is 3/4 load or less.

Now if you get a curve that's 70 rpm less than synchronous and just barely meets your specifications, then you would want to ask some questions, because it will make a complete mess out of all the submittal documents (not to mention the performance guarantees) once you receive the official motor data package with the "real" rpm.

 

[DubMac]

Are you really serious about the specification? 98% requirement is a complete joke; like trying to figure out if a rock is 200 million years old, or 201 million years old.

Must be a typo. Probably ink smeared on the original 90% and it now looks like 98%. Otherwise, check the spec writer's BAC.

 

[Artisi]

Agree with DubMac, 98% is nonsense, needs checking and clarification - plus, is VSD part of the spec. or a means of the suppler trying to meet spec.

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.)

 

[1gibson]

95% of max diameter is a typical value, so 98% could just be less conservative. Maybe a bad experience with high radial thrust at full diameter, but not too worried about future head rise. Agree that 98% is an unusual number, but it could be a personal touch in the spec writing.

 

[virata]

Yes it is 98% ratio of the rated impeller to the max impeller as per spec and it is ASME pumps by the way (3600 rpm).

 

[Artisi]

As you seem to not answer reasonable questions which might help you all that can be said is "it doesn't meet spec."

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.)

 

[GPRnD]

Hmm, so the API 610 standard is for a 5% head increase with a larger impeller (API 610 11th edition 6.1.4), not the 10% that was stated earlier. The translates to needing an impeller diameter of less than 97.5% of maximum diameter due to the affinity laws.

I suspect the 98% specification the OP encountered is a rounding of the 97.5%. We see people specifying 95% or even 90% impeller diameter limits, completely oblivious to the API 610 intent or affinity laws.

If you have a VFD and if the VFD can accomodate a 5% head rise by overspeed (without equipment damage), I'd say the 98% spec limit is pointless. On many pipeline pumps now it is routine to go with maximum impeller diameter and compensate with the VFD since this yields higher efficiency.

 

[BigInch]

GPR Thanks very much for that. You are absolutely correct. Makes a big difference.

you must get smarter than the software you're using.

 

[DubMac]

OP stated the 98% ratio was for impeller SIZE, not head increase; making the specification even more ridiculous.

If in fact, as GPRnD says, the spec rounded up from 97.5% to get the 98% requirement, why doesn't the OP just round down from the 98.4% impeller offered and be done with it..... Crap in, crap out.

I wonder if there is still enough brainpower around to send a rocket to the moon........just wonderin.......

 

[1gibson]

In my first post, I suggested a solution to eliminate the non-compliance, so all academic discussion aside, that is what probably should happen.

Make the curve 5 rpm faster, this should drop the diameter by maybe 1/8" or so (surely more than the 0.4% diameter change required) then it will be compliant. 5 rpm is about the max I would go when "cheating" a spec requirement. I've been stuck holding the bag too many times when someone decides to cheat by 50 rpm or more, to get past actual hydraulic requirements and force a bad selection. Then you can end up with a real problems, like exceeding a flange rating at shutoff when the pump runs at a faster RPM in the field.

So that's why I don't like the approach, but as long as it is realistic, go for it. 0.4% on the diameter for a 20" impeller is 0.080." If everything else meets the spec, it is an otherwise good selection, and you can't bend the rules (no deviations allowed) then cheat a little bit. It will make everyone's life easier, no harm no foul.

 

[BigInch]

GPRnD

Could you please explain how higher efficiency is obtained,
"On many pipeline pumps now it is routine to go with maximum impeller diameter and compensate with the VFD since this yields higher efficiency."

you must get smarter than the software you're using.

 

[GPRnD]

So with most impeller designs, the maximum efficiency is attained at maximum diameter. (there are a few with the so called 'bullseye" iso efficiency lines where the maximum efficiency occurs as you trim, but that tends to be due to poor impeller design).

As you trim, the D3/D2 ratio increases, which has a beneficial effect on pressure pulsations and vane pass vibration, but reduces the efficiency due to increased losses and flow mismatch in the transition between impeller and casing volute (or diffuser). As a rule of thumb (if test data is not available), a 5% reduction in diameter (i.e. a trim from 100% to 95% impeller diameter), will reduce the efficiency by 1% for a volute pump and 2% for a diffuser pump.

Some pump suppliers will use this effect to reduce the D3/D2 ratio closer to 1 and gain efficiency at the expense of vibration. I personally view this as cheating the customer since the resultant pump will often not be reliable. A good example is the case study at the link below. Although it isn't stated in the presentation, I believe the D3/D2 ratio was in the region of 1.02, when it should have been at least 1.06.

http://turbolab.tamu.edu/proc/pumpproc/P27/Rotor Retrofits Improve Pump Station Vibration.pdf

Sorry that was a rather long rambling answer to your question, but hopefully suffices.

 

[BigInch]

Not sure. That explanation appears to throw some efficiency increase to impeller trimming, although you also appear to conclude saying trimming reduces efficiency by 1-2% because of flow mismatches, so where's the increase?

The first explanation gave the impression that VFD increase efficiency, which may be somewhat true for the pump, if impeller speed increases past rated speed, however simply adding a VFD tends to reduce efficiency by the efficiency of a VFD, which can be relatively high when fully loaded, but reduces considerably at partial load, where historically it has been said that VFDs are of most use. Truth is that when I run the numbers in explicit detail, including all efficiencies of pumps, other devices and line losses, adjusted proportionally for partially loaded motors and VFDs when at lower flow rates, I find that VFDs rarely if ever increase efficiency for my systems at all (over that of a pump and control valve, or pump alone), and in fact making the system cost more to run, except within a possible range of 85%-95% of pump's BEP flow rate. So if I notice that is true, I generally also find I have chosen the wrong system flow and I reduce that by 10% and wind up using a pump with a discharge control valve, or a pump alone.

you must get smarter than the software you're using.

 

[BigInch]

I think it can also be said that the pumps in the linked study were not the best choice for the job. The secondary causes given on page 225 were actually the root cause. I find it a bit difficult to select the wrong pump for the operating conditions (perhaps a too long a shaft) and operate it too much on recycle or with inappropriate recycling valves, or piping configuration, then throw all the blame on the pump for failing.

you must get smarter than the software you're using.

 

[GPRnD]

So in the past I've read some of your posts regarding VFD misapplication and the inefficiency added by the VFD and I'd agree with them.

Taking pipeline applications, the choice to use a VFD from the outset is driven by the steep system curve (mostly friction loss), as well as the wide range of different products and flows they want to achieve with the pumps as close to peak efficiency as possible. Ralph Dickau who is far more knowledgeable than I on this topic has a good presentation on it at the link below.

http://calgarypumpsymposium.ca/uploads/2013PDFs/EPC/Pipeline_Pumps.pdf

So starting with the base assumption that a VFD is required, it makes more sense just to achieve the future head requirement by over speeding the VFD than building trim into the pump. Take for example a pipeline BB1 18x20-23A selected for a rated condition of 13,000 USGPM, 1660 ft, 0.85 SG:

Scenario 1, maximum VFD Speed for present head requirement 3560 RPM, 97.5% of max diameter impeller trim (86.1% pump efficiency), power consumed 4010 KW

Scenario 2, maximum VFD Speed for present head requirement 3560 RPM, overspeed VFD to 3648 RPM (86.5% pump efficiency), power consumed 3991 KW

That small uptick in efficiency will reduce the LCC for power by about $200,000 over the life of the pump.

 

[BigInch]

It is generally assumed that the pump efficiency doesn't change with change in speed, but that is probably an erroneous assumption that more exact data might rectify. The problem is that the data describing the variance of efficiency with speed typically can't be found anywhere.

Hence my question now is, are you referring to the change in efficiency due to change in speed, or change in efficiency due to a change in impeller diameter, or both, and can these be calculated, or estimated using some methods of which I am not aware?

virata, sorry I'm hijacking your thread. Do you mind if we continue?

you must get smarter than the software you're using.

 

[GPRnD]

In my previous reply I was considering only the efficiency change due to impeller diameter (sorry that I wasn't clear). However in my experience both speed and impeller trim affect efficiency.

Efficiency change with speed
HI assumes in its test code 14.6 that speed does not have an effect on efficiency for a 20% speed change. I'd disagree. Based on a lot of test data we'd conservatively expect about 0.4% reduction in efficiency going from 60Hz 2 pole to 50 Hz 2 pole. The formula I prefer to use (for flows up to 10,000 USGPM) is Effy2 = 1-(1-Effy1)*(Q1/Q2)^0.16, where Effy2 and Effy1 are the pump efficiencies and Q1 and Q2 are the corresponding flowrates.

That said since the efficiency change is small it is buried in the test variability so it isn't always clear for small speed changes. For API 610 pumps such a small change is unlikely to matter, but if it is a critical pump quoted to HI 14.6 1U or 1E, the penalties for missing efficiency can be unpleasant. Hence the desire to be conservative.


Efficiency change with impeller trim
As I mentioned previously each 5% impeller trim will reduce the efficiency by approximately 1% for a volute pump and 2% for a diffuser pump. This again is based on an aggregate of lots of trim testing. You could look at any individual pump curve and find it deviates some from this, but as a rule of thumb it is a good starting point. Clearly it is better to perform actual trim testing, but for a new design we need to account for the effect of trimming when predicting the performance curve for a customer.