Suction Specific Speed, should I worry about it?

(Article by Simon Bradshaw, ITT Goulds Pumps)

(Part 1)

Suction Specific Speed (Nss) limitations for pumps have been around since Jerry Hallam's detailed
analysis of refinery pump reliability in 1982. His study was very comprehensive and covered 480 pumps
over a 5 year period at the Amoco Texas City refinery. He found a correlation of failure frequency to Suction
Specific Speed. Basically above 11,000 Nss the failure frequency almost doubled. I've reproduced his
graph below.

Now I've had the great pleasure of talking to Jerry Hallam about his work and one thing he kept coming
back to was that his conclusion was not that pump Nss should be limited to a single hard number but rather:
"caution should be exercised when purchasing hydrocarbon or small water pumps with a Nss greater than
11,000 unless operation is closely controlled near BEP."
He has been surprised by how people over the years have interpreted the results of his analysis to mean
that a pump with Nss = 11,001 is bad and a pump with Nss = 10,999 is good, since that was never his
intention.
So that was 1982.
Since then many pump users will specify a maximum Nss value without understanding what it means or
where the Nss = 11,000 number originated from...
What about 2014 ?
Pump suppliers have been making the point for a number of years that the 11,000 Nss limit was based on
the performance of pumps designed in the 1950's to 1980's when hydraulic design techniques were much
less sophisticated than they are today. The "gist" of the argument is that modern pumps have better
hydraulic designs and are more robust meaning that they can be reliable at higher Nss values than would
have been possible in the past.
I did a webinar about testing that was done on allowable operating range vs. Nss for an OH2 pump in 1985
and how with modern pumps designs that would look completely different.
Then in 2013 we got around to actually testing this theory and we proved that it was correct. Modern pumps
do indeed have a significantly wider operating range than pumps designed or built prior to the 1980s. Ideally
we'd also perform a repeat of Hallam's large scale reliability study with more modern pumps. Unfortunately
no major oil company seems willing to take this on, which seems incongruent with saying that reliability is
a key factor in their continued success.

For now you might want to consider how a reliable higher Nss pump might contribute to lowering the overall
life cycle cost of your operation.
(Part 2)
Before I discuss the effect of Nss on life cycle cost, I want to continue from part 1 of this topic and reiterate
how far we've come in terms of hydraulic design and pump construction in the 32 years since Jerry Hallam's
study of reliability was published in 1982. I'd continue with the car analogy theme (i.e. Vehicle Fuel
efficiency).
The graph below speaks to how two relevant parameters have changed:

So you can see that since 1982, vehicle fuel efficiency has improved significantly through many small
incremental technological improvements. However the same engineers that happily drive around in cars
capable of 50 MPG/ton, also make the (incorrect) assumption that pump maximum safe Nss is fixed (or
even declining) despite 30 years of similar incremental improvements.
For example in 2012 we did a study on the effect of impeller leading edge profiling and found that just by
applying an optimized parabolic profile we could improve the NPSHr of the pump by 15%. So using just
this one change an impeller with Nss =11,170 was improved to Nss = 12,644 with no adverse effect on
part load performance, vibration or suction recirculation.
We also know that through the deployment of more advanced hydraulic design tools (we utilize several,
some commercial, some developed internally), we can now have much more control over the impeller vane
shape to provide such things as:
 Low blade loadings in the inlet region up to the impeller throat area. These help prevent the formation
of low pressure zones where cavitation will begin.
 S shaped developments of the impeller camber line in order to achieve the required impeller throat
area while minimizing the eye diameter.
 Backward swept blades to reduce the volume of any cavitation that develops at the leading edge.
 Impeller leading edge carried well forward at the impeller hub in order to reduce the formation of
cavitation at part load operation.
Now granted the above might sound like geek speak to anyone who doesn't find joy in hydraulic design,
but bear with me, there is a payoff. What this means in practical terms is that if no Nss limits are specified
by the customer, ITT Goulds is confident in designing OH style pumps with the Nss limits shown in the
chart below. For obscure reasons that are not terribly important the blue line that indicates the limit is what
we call the SGsT curve:

The chart is read by determining the specific speed (Ns) of your pump and finding it on the X axis then
moving vertically until the blue line is intersected, at which point you read off the recommended Nss limit
from the Y axis. So for example a pump with a 2000 specific speed, could achieve acceptable performance
with a 12,000 Nss impeller.
Those readers with sharp eyes will have noticed that above 2500 specific speed, the maximum Nss drops
below 11,000. Why would this be ?
Well the clue is in the shape of the pump impeller, which I have included on the chart.
At 500 specific speed the impeller is like a skinny disc with a tiny impeller eye. For these impellers suction
recirculation is not a problem (in fact the problem is often too little recirculation leading to unstable HQ
curves). Consequently we can confidently design to 15,000 Nss knowing the pump will be reliable.
Contrast this with a 3000 specific speed impeller where the impeller eye is not much smaller than the
outside diameter of the impeller. An impeller like this requires careful design in order to limit the onset of
suction recirculation and consequently we would prefer to limit the Nss to around 10,000 or less.
When viewed in this context having a single Nss value regardless of the pump specific speed makes very
little sense. Regardless of how conservative you want your Nss to be, I'd strongly recommend changing
from a single value to a more graduated approach.
(Part 3)

The effect of Suction Specific Speed (Nss) limits on life cycle costs (LCC).
But before we dive into the arcana of Nss effects I think it is worth recapping what goes into making up a
typical pump life cycle cost. There are a quite a few studies out there. While they vary in the exact values,
most identify that the cost of buying and installing the pump (what I would call 1st cost), is a relatively small
percentage of the total life cycle cost for an industrial pump. The US DOE, Hydraulic Institute and Europump
also concur on this point in their LCC guide for pumping systems.
At Goulds we use the graph shown below as an average for an industrial pump with a 10 year life. If you
are considering an API 610 pump with a 20+ year life, the 1st cost as a percentage of the LCC shown here
will be almost halved

In this example the 1st cost (purchase and installation) is only 23% of the overall LCC. As I mentioned
above for an API 610 pump that value is nearer to 15%.
Hopefully you noted that the energy cost amounts to around 1/3rd of the LCC. I'd also note that this is
based on US industrial electricity prices. If you live in a region of the world where electricity costs
substantially more, this will be a larger portion. For example the US industrial electricity cost has been
around $0.07 per KWh in 2014. In Europe it is around twice this at $0.16 per KWh.
So having established that energy cost is a big chunk of LCC, how does limiting the maximum Nss affect
it ?
First recall that Nss is a function of the pump speed (as well as BEP flow and NPSHr). So the normal way
to achieve a target Nss limit such as 11,000 (when flow and NPSHr are fixed), is to slow the pump down.
However when this is done, the pump Specific Speed (Ns) also reduces. This reduction affects the shape
of the impeller as I discussed in Part 2, but it also affects the efficiency that the pump can attain as I have
shown in the chart below taken from Johann Gülich's book Centrifugal Pumps.

So for any given pump BEP flow rate (pick one of the lines), lowering the pump Specific Speed will lower
the attainable efficiency, sometimes significantly. To illustrate this I have a couple of real world examples.
You can even follow along by going to the Gould’s Pumps website and using our online selection tool
ePrism (registration required).
The first example is for an OH2 pump for the following service:
Head 651 ft (198.4 m), Flow 1598 USGPM (363 m3/hr), Maximum NPSHr 26 ft (8 m), SG 0.754, 60 Hz
The pump sold was a 3x8-27A running at 1785 RPM with an efficiency of 64% and an absorbed power of
310 HP (231 KW). The specific speed of this selection is 482 US units (9 metric)
Because the Nss was limited to 11,000 (US units) a 2 pole selection was not possible. However if that limit
was raised to the SGsT limit talked about in Part 2, a valid selection would be a version of the 4x6-13H.
This would run at 3560 RPM with an efficiency of 79.5% and an absorbed power of 249 HP (186 KW). The
specific speed of this selection is 1164 US units (22 metric).
It is pretty clear from the selections the effect of the Nss limit in terms of a big efficiency reduction. But how
much does that really amount to ?
The power difference is 231 - 186 = 45 KW. In this example with the pump running 8000 hours per year
that amounts to 45 x 8000 x $0.07 = $25,200 per year. For the 20 year service life of this API 610 pump,
the Nss limit will cost the end user an additional $0.5 million in energy costs.
The second example is for a 2 stage BB2 pump for the following service:
Head 1191 ft (363 m), Flow 1940 USGPM (441 m3/hr), Maximum NPSHr 18 ft (5.5 m), SG 0.65, 60 Hz
The pump sold was a 8x10-27CD running at 1785 RPM with an efficiency of 64.5% and an absorbed power
of 574 HP (438 KW). The specific speed of this selection is 598 US units (12 metric)
Similar to the first example, raising the Nss limit above 11,000 would allow a 2 pole selection with an
efficiency of 71.5% and an absorbed power of 533 HP (398 KW). The specific speed of this selection is
971 US units (19 metric).
The power difference is 438 - 398 = 40 KW which with the same run time as the first example, amounts to
40 x 8000 x $0.07 = $22,400 per year. For the 20 year service life, the Nss limit will cost an additional $0.45
million in energy costs.
Now just to reiterate these are both real world examples that customers have purchased. The choice to
limit Nss will cost them $1 million (or more if they were installed in a location with high energy costs), over
the life of these pumps.
Clearly if the higher Nss pumps were unreliable, the potential revenue loss might be more of a concern and
higher than $1 million. But I am not advocating aggressive Nss targets, simply to adjust them to levels
commensurate with 30 years of improvements in pump design.
Interestingly, efficiency optimization still appears to be a lower priority than expected. A few customers
(India located customers being a leader in this practice), will specify minimum efficiency targets and weigh
bids in terms of life cycle energy cost. However even here blanket Nss < 11,000 limits are applied, limiting
the realizable energy savings.
Will this change in the future ? Well the DOE in the USA exempted API 610 pumps from their forthcoming
energy efficiency regulations (which will be broadly similar to the EU ones). But you can envision future
scenarios with global warming and energy efficiency being continued areas of concern to the extent that
the regulations around refinery energy efficiency come to the top of the agenda.