Fluid Handling: Avoid repeat pump failures

In brief:

Most pump problems develop gradually. Others manifest themselves sporadically.
The onset of pump problems is not the same for different pumps or different services.
The vulnerability of operating process pumps in parallel is not always appreciated by pump purchasers.

Equipment failures cost money. Repeat failures are an indication that the underlying failure causes either haven’t been discovered or have been ignored.

But if your facility has 600 pumps and they survive an average of three years, there will be 200 repair events per year. Even at a mere $6,000 per repair, that’s $1.2 million each year. Suppose your competitor knows how to avoid repeat failures and sends his pumps to the shop only every six years. His yearly outlay will be $600,000, and his repair and maintenance crews can be deployed for failure prevention tasks instead of fix-up tasks.

So, what exactly is it that best-of-class users do to set themselves apart from the rest?

Pumps have a defined operating range

We start with the obvious. All machines have a defined operating range. A Boeing 787 cannot land on a 300 ft (100 m) landing strip. It cannot fly at 50 knots. A sailplane can do all of that; it’s designed for different performance and seats two people. It’s no different with process pumps; they, too, have operating ranges and performance limitations.

Pump & Turbo

Heinz P. Bloch, owner of Process Machinery Consulting, will be leading a tutorial, “Breaking the Cycle of Pump Repairs,” on Oct. 1 at 2:00 PM and again on Oct. 2 at 10:30 AM at the 29th Pump and 42nd Turbomachinery Symposia in Houston.

Most pump problems develop gradually. Others manifest themselves sporadically. They could be operations-related vibration excursions due to fluid vaporization, but hydraulics influence pump life. Perhaps we can draw parallels with human illnesses. Most serious health issues develop gradually. A few can be easily cured and go away after a while if we make wise decisions. Some illnesses manifest themselves with extreme suddenness and can have devastating consequences. All of these problems could have been diagnosed earlier if we had used more time, money, and well-targeted efforts.

The onset of pump problems is not the same for different pumps or different services. Attempts to identify best practices led Paul Barringer and Ed Nelson to explain the effects of deviations (Figure 1). While focusing on the best efficiency point (BEP), Barringer and Nelson plotted eight traditional non-BEP problem areas on a representative H/Q curve. The plot supports the notion that pump reliability can approach zero as one operates farther away from the BEP. At some combination of age, load, speed, temperature, or whatever, reliability goes to zero with every conceivable creation of man.

Figure 1. The Barringer-Nelson curve shows reliability impact of operation away from BEP. (Source: Paul Barringer)

The implications of the Barringer-Nelson curve are easy to visualize. Just because pumps are able to run at lower-than-BEP flows doesn’t mean it’s good to operate there. Compare it to a vehicle able to go 12 mph in sixth gear, or 57 mph in first gear. It can be done, but will likely prove costly if done for very long. Pioneering efforts to define minimum allowable flows can be traced back many decades and attention is drawn to the sketch by Irving Taylor (Figure 2). His work is worth mentioning because Taylor approximated in a single illustration what others have tried to convey in complex words and elaborate mathematical expressions. Although Taylor’s relationships are typical at best, he deserves much credit because he kept the average user in mind. More scientific approaches were documented by Taylor’s very famous contemporary, Igor Karassik in “Pump Handbook.”

Figure 2. Pump manufacturers usually plot only the NPSHR (subscript R) trend associated with the lowermost curve. At that time a head drop or pressure fluctuation of 3% exists at BEP flow. (Source: “The Most Persistent Pump-Application Problems for Petroleum and Power Engineers,” by Irving Taylor)

Taylor suggested a demarcation line between low and high suction specific speeds (Nss numbers) at somewhere between 8,000 and 12,000. His data are supported by surveys taken after 1977 at Amoco in Texas City, Texas, by Nelson and Jerry Hallam; there also were other plant locations which pointed to suction-specific speeds of 9,000 or 9,500 as Nss numbers that deserve attention. Many pumps with Nss numbers higher than approximately 9,500 will degrade when being operated at flow rates much lower or higher than BEP. By how much the life expectancy or repair-free operating time of these pumps will be reduced is speculative, at best. Whether these life reductions will amount to 10% off normal or 60% off normal is the subject of much debate and requires reviews on a pump-specific basis.

It reminds us of other rules of thumb. We consider automobile tires unserviceable whenever less than 1/32 in. of tread is left. While it may be possible to operate tires with absolutely no tread left on the carcass, it would be risky to use bald tires on a vacation trip with a wife, four kids, and a dog. Traveling at 75 mph on roads full of potholes would increase the risk exponentially. No researcher has quantified it, but common sense tells the story.

So, while no rigorous Nss value exists, cautious reliability professionals observe safe margins. Many users choose Nss = 9,500 as the limit for flows away from BEP. There are, however, some pumps, including certain high-speed Sundyne designs, that will operate quite well with Nss values higher than 9,500. But these are special cases, pumps with relatively low pump specific speeds (Figure 3). Even so, a close pump-user-to-pump-manufacturer relationship is needed to shed light on applicable long-term experience with these pumps.

Figure 3. Impeller cross-sections align with specific speed ranges. (Source: “Pump Wisdom,” by Heinz P. Bloch)

Mechanical issues

The vulnerability of operating process pumps in parallel is not always appreciated by pump purchasers, although API-610 advises against parallel operation for pumps with relatively flat performance curves. Reasonable, yet general, specifications require an 8% to 10% rise from BEP to shut-off. There are other parts of API-610 that raise eyebrows, among them the user’s erroneous perception that back-to-back 40° angular contact thrust bearings must always be used.

There are problems with short elbows near the suction nozzle of certain pumps, and flow stratification and friction losses are sometimes overlooked. Some sources advocate a minimum of five; others advocate a 10-diameters equivalent of straight pipe run at the pump suction. Pump parallel operation and piping issues deserve attention, and reasonable rules of thumb are quite adequate for the reliability-focused. Because the tie-in between the lack of conservatism in piping and issues of less-than-adequate pump reliability is tenuous, the multipoint trouble illustration in Figure 1 is again of interest here. Tight-radius elbows and incorrect pipe reducer orientation can quickly wreck certain pump configurations (Figure 4). Neglecting piping issues can be a costly mistake, if casing distortion causes point-loading in a rolling element bearing. The resulting force per unit area causes extreme pressure, and lubricants can no longer prevent metal-to-metal contact.

Figure 4. Using the wrong reducer type or installing it incorrectly can cause unstable fluid conditions.

Figure 5. An edge-loaded bearing will fail.

Pulling piping into place at a pump nozzle can cause edge-loading of the pump’s bearings, which will surely lead to premature bearing failures (Figure 5). But what if soil settlement under pipe supports played a role in misalignment? Concrete driveway sections often misalign within just a few years of construction, so why would we expect pipe supports to still be straight and plumb decades after they were first installed? As to short-radius elbows, the flow velocity at the small-radius wall of an elbow will differ from that at the large-radius wall. That could certainly cause premature failures in double-flow pumps. Because these facts are generally well known and many symposia have been devoted to them, our discussion focuses on pump mechanical end, or drive end (power end) issues.

Deviations from best available technology

User plants will usually get away with one or two small deviations from best available technology. But when three or more deviations occur, failure risks usually increase exponentially. That said, there are a number of reasons why a few well-versed reliability engineers are reluctant to accept pumps that incorporate the drive end shown in Figure 6. The short overview of reasons is that reliability-focused pros take seriously their obligation to consider the actual, lifetime-related and not just short-term, cost of ownership. Specialists realize that the bearing housing in Figure 6 will work initially and then fail prematurely. The housing is shown exactly as originally provided, including its several risk-increasing features. Allowing these features to exist will sooner or later hurt the profitability of users and vendors alike.

Figure 6: This bearing housing has several potentially costly vulnerabilities.

We can visualize that the angular contact thrust bearings in Figure 6 will probably incorporate cages (ball separators) that are angularly inclined, which means they’re arranged at a slant. These cages often act as small “fans,” and fans promote air flow from fan center toward the fan tip. This air motion can upset the direction of oil flow.

However, pressures inside the bearing housing must be equal in front and behind bearings, and equalization may require internal passageways. Also, pressure balancing is sometimes thwarted by installing bearing housing protector seals. While there are compelling reasons to use bearing housing protector seals, attention must be given to the effects they can have on housing-internal pressure profiles.

Upon examining Figure 6, a careful viewer can be certain of five facts.

Oil rings are used to lift oil from the sump into the bearings. These oil rings tend to skip and jump at progressively higher shaft surface speeds, or if not perfectly concentric, or if not operating in perfectly horizontal shaft systems. Worse yet, they will abrade and seriously contaminate the lube oil.
As the pump is transported from shop to field, an oil ring can become dislodged and get caught between the shaft periphery and the tip of the long limiter screw.
The back-to-back oriented thrust bearings aren’t located in a cartridge. This limits flinger disc dimensions, if they were to be retrofitted, to no more than the housing bore diameter.
Bearing housing protector seals are missing from the picture. Advanced bearing housing protector seals are recommended. However, bearing protector seals in this housing will change the flow of venting air, and internal pressure balancing is needed.
Although the bottom of the housing bore (at the radial bearing) shows the needed oil return passage, the same type of oil return or pressure equalizing passage seems to have been left out near the 6 o’clock position of the thrust bearing. A small pool of oil can accumulate behind the thrust bearing, and this oil will probably overheat. Carbon debris will then form.
No particular constant level lubricator is shown, and there is uncertainty as to the type or style of constant level lubricator that will be provided. Unless specified, OEMs rarely supply the best available constant level lubricator.

Failures may originate from age-old practices that have become tradition, and using cooling water on pump bearing housings is one of these traditions. Actually, bearing housing cooling is not needed on process pumps which incorporate rolling element bearings. Cooling is harmful if it promotes moisture condensation (water cooling coils) or restricts thermal expansion of the bearing outer ring (water cooling jacket). In 1967, these concerns were seen to influence pump reliability. The jacketed cooling water passages should be left open to the ambient air environment. The decision to delete cooling water from pumps with rolling element bearings was first implemented in 1967 at an oil refinery in Sicily. The owner’s engineers had recorded bearing lube oil in four identical pumps reaching an average of 176 °F (80 °C) with cooling water in the jacketed passages. Without cooling water, the lube oil averaged 158 °F (70 °C), which is 18 °F (10 °C) cooler. The bearings now lasted much longer.

As of 2012, some process pumps continue to experience costly repeat failures. Motivated reliability professionals and informed users can avoid these and will appreciate recommendations on failure risk reduction. For the truly reliability-focused pump users, a number of conclusions and upgrade recommendations may be of interest.

Discontinue using maintenance-intensive oil rings and, if possible, constant level lubricators.
As a matter of routine, the housing or cartridge bore should have a passage at the 6 o’clock position to allow pressure and temperature equalization and oil movement from one side of the bearing to the other. Note that such a passage was shown in Figure 6 for the radial bearing, but not for the thrust bearing set.
With proper bearing housing protector seals and the right constant level lubricators, breathers (or vents) are no longer needed on bearing housings.
If constant level lubricators are used, a pressure-balanced version should be supplied and its balance line should be connected to the closest breather port.
Bearings should preferably be mounted in suitably designed cartridges, and loose slinger rings (oil rings) should either be avoided or, in some higher shaft peripheral speed cases, disallowed.
Suitably designed flinger discs can be secured to the shaft whenever the oil level is lowered to accommodate the need to maintain acceptable lube oil temperatures (for pumps operating above a particular shaft circumferential velocity).
Modern and technically advantageous versions of bearing housing protector seals should be used for both the inboard and outboard bearings. Lip seals are not good enough, and neither are outdated rotating labyrinth seal designs.
Understand that the implementation of true reliability thinking strongly supports moves away from traditional bearing housings and conventional modes of lubricant application. Reasonable upgrade alternatives are among the tutorial subjects.

Heinz P. Bloch, P.E., is owner of Process Machinery Consulting (www.heinzbloch.com) in Westminster, Colorado, and the author of numerous articles and books, including “Improving Machinery Reliability” and “Pump Wisdom.” Contact him at [email protected].

Knowledgeable engineers can show that some widely accepted pump components tend to malfunction in the real world. Moreover, as industry often moves away from solid training and from taking the time needed to do things right, designing out risk and designing out maintenance become attractive propositions.

References

Bloch, Heinz P. and Allen Budris; “Pump User’s Handbook—Life Extension,” 3rd Edition (2010), Fairmont Press, Inc., Lilburn, GA 30047; ISBN 0-88173-627-9

Taylor, Irving; “The Most Persistent Pump-Application Problems for Petroleum and Power Engineers,” ASME Publication 77-Pet-5 (Energy Technology Conference and Exhibit, Houston, Texas, September 18-22, 1977)

Igor Karassik et alia; “Pump Handbook,” 2nd Edition (1985), McGraw-Hill, New York, NY, ISBN0-07-033302-5
Bloch, Heinz P.; “Pump Wisdom,” (2011), John Wiley & Sons, New York, NY (ISBN 9-781118-041239)