Thirty (30) some years ago I skimmed through an exhaustive study of the possible modes a prototype nuclear pump could fail and the probabilities connected with these events. In the report, the first reliability study I had seen, there was no discussion on how to influence or change failure probabilities. Absent a linkage to changing real outcomes, I was left with an impression that reliability was little more than an academic exercise.
Years later I came to realize that reliability has operational and financial consequences and encompasses much of the life cycle of a pump beginning with equipment selection and continuing on through operation and maintenance and, as applicable, through modifications of the process and the machinery. As depicted in Figure 1, the scope of reliability is a composite of how the pump is designed, manufactured and assembled, how it is installed, the product being pumped, the pump operating conditions, condition monitoring and maintenance practices. The whole point is missed if one considers that pump reliability is somehow detached from all of the activities that conceive, create, operate and maintain a pumping installation.
Arguably, the greatest technical challenge facing manufacturers and users of pumps is improving reliability. Mechanical seal leakage gradually or exponentially increases to an unacceptable level, seemingly without cause. Bearings fail, often with little advance warning; when bearings fail this can wreck the mechanical seal and a whole lot more. Pump internal clearances, subjected to pressure and erosion, wear out, amperage draw goes up and pumping performance falls off. Shafting and impellers fatigue and break. Flanges leak, gaskets blow. This is all familiar to those who operate and maintain pumps. Is it all kismet or can we be agents of cause over the future of a pump?
Usually a pump ‘fails’ due to the failure of a single component. Moreover, depending on how promptly the pump is shut down after initial symptoms of imminent failure appear, other components may be damaged as well. Depending upon trends in vibration, leakage, power consumption, loss of performance or other indicators, the pump’s symptoms are telling us with some degree of predictability how soon it should be pulled out of service for repair. The less forewarning, the greater the impact of lost production and pump repair costs.
Repairing the pump brings it back to operational status, but this action alone may not improve reliability. The root cause of pump premature failure may be: operating continuously at an excessively low flow or an excessively high flow, some sort of machinery dynamic problem resulting in excessive vibration, excessive piping loads, cavitation, coupling or rotor imbalance or shaft misalignment, a mechanical seal or bearing overheating problem, and others. In such instances, finding and fixing the root cause brings about a dramatic improvement in the pump’s mean time between failure (MTBF).
A pump includes both stationary and moving parts that can fail. Usually stationary pump components are not the cause for pulling a pump out for service, but rings and bushings do wear and occasionally static seals or gaskets leak. Also pump casings can become excessively worn or corroded but seldom cause in-service failure. It is often quipped that a centrifugal pump has a single rotating part. But actually, considering the mechanical seal and bearing assembly, there are some two or three dozen moving parts, any one of which can fail and take a pump out of service. In order of frequency, the most common pump components that fail are mechanical seals, bearings, couplings, shafting and impellers.
Excessive leakage of the mechanical seal is the most common reason why a pump is taken out of service, if even for just a few hours to install a new seal or remedy the specific leakage issue. There are many possible reasons and scenarios leading to excessive seal leakage. Mechanical seals are precision devices that demand careful handling and installation procedures. The sealing gap between faces for liquid seals can vary between 5 and 50 micro-inches (0.13 to 1.3 microns). Dirty fingerprints can ruin sealing faces. Improper or careless installation, improper operation of the seal piping system or the seal support system, inappropriate mechanical seal selection, inappropriate seal piping plan or seal support system, excessive pump piping strain, and excessive shaft misalignment can each or in combination have an adverse effect on the life of a mechanical seal. Ideally a mechanical seal operates in a cool, clean and vibration-free environment, in which case, depending on the specific application and operating parameters, a mechanical seal should provide years of trouble-free operation.
Rolling element bearings are the second most common pump element to fail. Rolling element bearings have a statistical life rating, based on failure due to metal fatigue, known as L10 (aka B10), often expressed in hours. At a given shaft rotational speed and load, 10 percent of ‘identical’ bearings can be expected to fail (or 90 percent will survive) when the bearing manufacturer’s rated L10 hours have elapsed. 50 percent of bearings will survive approximately five (5) times the L10 life rating.
Dimensional standard process pumps have a specified minimum L10 life of 17 500 hours (2 years). So the median bearing life for these pumps would theoretically be about 10 years. But pump bearings most often fail way before achieving their ‘natural’ statistical demise due to metal fatigue. Indeed, observed field failures are often related to particulate or water contamination of the lubricant, abrasive wear, improper lubricant viscosity, mounting damage, misalignment, improper shaft/housing fits, insufficient loading resulting in skidding of rolling elements, shock loading, false brinelling, failure of a bearing separator (cage) or a failure or malfunction of the lubrication system. Debris particle contamination as small as 3 microns can cause bearing damage. Proper bearing storage, handling and installation, as well as the initial bearing housing and assembly cleanliness are factors the pump manufacturer or the repair shop can control. General lubrication cleanliness, control of bearing housing contamination and monitoring of the lubrication system are factors the pump user can control.
The system, process and specifying engineers play crucial roles in determining the ultimate life cycle cost and equipment reliability by ensuring the pump is properly specified for the properties of the pumped product, the equipment’s operating environment and the system’s hydraulic characteristics.
Equipment installation also plays a critical role in pump reliability. Proper setting of the anchors, leveling and grouting of the baseplate, piping up to the pump while minimizing strain, elimination of soft foot and accurate alignment of the drive train contribute to smooth running. Installation and alignment requires care and precision and are best done by or under the supervision of an experienced millwright.
Ideally a pump is operated at or near its best efficiency point (BEP). The Preferred Operating Range (POR) is typically 70 or 80 to 120 percent of BEP flow according to Hydraulic Institute, API or ISO specifications. The Allowable Operating Range (AOR) is defined by the manufacturer. Graphically the concept of POR and AOR ranges are shown in Figure 2. The AOR is typically bounded by flows at which the vibration level exceeds a specified value. Pump NPSH-Required may be the limiting factor at the high flow extent of the operating range. Turbulence, recirculation and cavitation come from operating a pump at off-BEP flows and result in energy dissipation forces that act upon the rotor and the pump stationary components. These forces cause displacements and strain, fatigue and wear and shorten the pump’s MTBF.
The standards for pump maintenance and repair activities should be no less rigorous than would be expected of newly manufactured equipment. Proper cleanliness, handling and assembly practices for critical tolerance mechanical seals and bearings can avoid maintenance induced failures and at least give these components an opportunity to achieve long life. Inspection of the disassembled pump also provides an opportunity to discover the root cause of failure.
A thorough treatment of pump reliability encompasses nearly the entire life cycle of the equipment, including selection, design, specifications, manufacture, inspection and testing, shipment and storage, installation, commissioning, operation and maintenance. When viewed in this context, reliability is a real world activity relevant to all hands involved with the pumping machinery.
For an independent evaluation of pump reliability, contact an experienced consulting engineer who can help with your specific application.
Thanks Randal for the great article!
Kelvin, you’re welcome and glad you liked it! I’m sure reliability is of critical importance to APG as it must be for any company making products for industrial and commercial processes.
love reading these subject matter expert stories with loads of engineering know how and true facts that never are debated due to mechanical / electrical logic. A small suggestion from the today technology puts a bit of pressure on the story. In fact 60% of pump failures happen due to scheduled maintenance. Quite an eye opener for those plant owners that are under pressure of OPEX spendings.
The solution to that would be to practice machine learning technology that is able to predict scheduled maintenance before the event of failure starts. Let me explain;
“A Recipe for Failure”
The Potential Failure (PF) curve illustrates how equipment degrades over time, ultimately resulting in failure. The “P” indicates the first detectable symptoms of failure. Before this point, damage is slight, gradual, and unmeasurable. Traditional condition monitoring indicates an issue only after damage has occurred, increasing both losses and repair costs.
“A Recipe for Success”
Smart Machine Learning redefine the PF Curve. Our solution discovers minuscule changes that other condition monitoring solutions are unable to detect, moving the “P” in the curve far earlier. At the earliest point of detection, simple maintenance fixes for small mechanical issues are serviced prior to machine failures that interrupt production. ”
Mike,
With critical production machinery, there is a case to be made for investing in instrumentation, software and training for detecting the first symptoms of failure as early on as possible. This provides valuable information for planned equipment shutdowns, as well as avoiding unscheduled and expensive process downtime.
Thank you for your comments.
Randal
Hi Randall,
Very informative article. Thank for sharing your expert views on pump reliability. as engineering consulting side it give more insight in reliability.
Regards
vishal
Hi Randal,
Not sure this is the right place for my question but can try…
If we are talking about modification to reduce force and flow on existing water pump with semi closed impeller and 8 vanes by factory, in order to get less force (save energy) on pump rotating and in the same time less flow, in general power saving wise and flow wise what I can expect if I just remove 4 vanes and keep other 4 only?
Thanks in advance
Milos,
For the benefit of others who would not have seen our correspondence since you originally submitted this comment…
Cutting out every other vane would reduce impeller input power but this reduction would be guesswork without a good meanline analysis. This is a moderately low specific speed design, and all of the shroud skin friction, representing significant input power, would remain.
The better solution is to remove the impeller and trim the OD. The amount of diameter trim, of course, depends on where you wish to intersect the system resistance characteristic.
Regards,
Randal