Things API 610 got wrong (Part 3)

My goal with this series of posts on API 610 is to inform, spark debate and occasionally be accused of entertainment. I will be assured that I've accomplished this if you the reader find this installment more fulfilling and less olfactorily damaging than feeding Pete the Wombat. (Observant readers may note in my blogs that Wombats have a habit of contributing to many hitherto unexplained pump failures. So if you see one it is always advisable to immediately retreat to a safe distance and call your local pump repair facility).

I'll warn you in advance this isn't going to be an easy read (even by my optimistic standards of technical density), but hopefully worthy of your time.

Caveat Lector

Before I continue with part 3 of the topic, I'd just like to recognize what API 610 got Right and the tremendous value of API 610 in guiding pump manufacturers and purchasers alike into specifying and building safe, reliable products. What I'm writing about in this series are quibbles with what is 99.9% otherwise an excellent standard.

I'd also like to mention the largely unrecognized Engineers who collectively put in many thousands of hours of time and care into crafting the standard. They are the ones very much deserving of your consideration and praise. A few of them such as Ron Adams, Morg Bruck and Bill Goodman I've been very lucky to work with and have benefited greatly from their wisdom.

What API 610 got Wrong #3

6.1.11 - Relying solely on a minimum 10% head rise to shutoff to prevent reliability problems when operating pumps in parallel

This clause of API 610 states:

6.1.11 Pumps that have stable head/flowrate curves (continuous head rise to shutoff) are preferred for all applications and are required if parallel operation is specified. If parallel operation is specified, the head rise from the rated point to shutoff shall be at least 10%...

For the purposes of the following discussion we'll assume that we are dealing with a system with just two pumps in parallel. That may not be the case with your system as I've personally worked on pumping systems with 5 pumps in parallel and all the attendant complexity that causes.

The thinking behind the minimum 10% requirement is to prevent the situation where one of the pumps in parallel delivers most of the flow, while the other pump is operating at a much lower flowrate, further away from BEP and hence subject to more vibration and wear.

While this thinking that "steeper pump curves are better" is directionally correct, once you've achieved a minimum head rise to shutoff, a far more important consideration is the matching of the pumps. If the pumps are not matched to each other (and kept matched), then bad things happen to your MTBF.

I can demonstrate this with a real world example of some pumps I worked on a few years back. These were multistage pumps with an expected Head Rise to Shut Off (HRSO) of 11%. The customer had stated that they would be used in parallel operation. The curve is shown below.

It's Parallel Jim, but not as we know it

To understand why pump matching is so critical when operating pumps in parallel, take a look at the curves below. In this example I've applied the curve from the pump mentioned above without any curve matching - i.e. each of the pumps has curves that achieve 11% HRSO but also meet all the API 610 test tolerances of +/- 3% head at rated point per API 610 Table 16 shown below.

As you can see one curve is stronger (Pump A), than the other (Pump B), which is not uncommon given typical manufacturing tolerances. Note that the allowable head tolerance at the rated point is +/- 3% per API 610 Table 16.

Another common reason for one curve being stronger than the other would be when one pump has much fewer running hours (for example if it was a spare). As this would result in less wear, it can result in that pump being closer to the original factory performance compared to the one with high run hours (more on this later).

The resulting combined Pump A+B parallel curve is discontinuous due to the mismatching of the two pumps. This exhibits itself as a step at around 1000 USGPM. (Below that point Pump B would operate at zero flow resulting in rapid failure).

In this scenario Pump A being stronger will force Pump B to operate back on its curve. If the system is operated at its Parallel Normal Flow, Pump B will be running at around only 50% of BEP. This is well outside the preferred operating range and will result in Pump B seeing higher wear and ultimately needing repair much sooner. Whether the pump has a HRSO of 9%, 10% or 11% doesn't matter so much. A 10% minimum HRSO helps to reduce the necessary matching accuracy a little but the end result is still much the same.

So what could have API 610 have done differently ? It gets back to my original point about curve matching. The only way to ensure reliability in parallel operation is to:

1. Mandate in the API 610 standard, high quality curve matching (to within better than 3%). This 3% matching requirement will thankfully be included in the forthcoming API 610 12th edition (clause 6.1.13c to be precise).
2. Mandate in the API 610 standard, the utilization of individual flow measurement and flow control on each pump. This is the best method for long term reliability as it allows the effects of pump wear to be monitored and corrected for.

That's it for this posting. As always I am very grateful to any of my readers who have made it to this point. If there are any comments, criticisms or questions those too are gratefully received.

Obsesio positivum