WHY PUMP SHAFT FAILURES HAPPEN Heinz P. Bloch *

Transcription

1 WHY PUMP SHAFT FAILURES HAPPEN Heinz P. Bloch * Technical publications thrive on questions from reliability engineers. These questions tend to inform editors and readers about prevailing level of knowledge. Also, the questions often reflect training-related issues confronting industry. In any event, finding answers may require practical knowledge and editors of trade and professional publications must often match-up answer seekers with answer providers. Next, and to the extent possible, editors fold the resulting discourse into presentations or articles of interest to a wider spectrum of readers. This is one of these articles. We were recently asked to provide guidance as a reader pondered over the possible causes of catastrophic pump shaft and impeller failures at his plant. In the course of my work as a Pump and Vibration Specialist the reader said, I have encountered a number of such failures. One root cause of such failures that has been suggested is a defective check valve, which allows the process pump to freewheel in reverse due to back flow. However, I have found no documentation that supports or explains this failure mode. In conversations with various pump and field service technicians, it has been suggested that when a pump is started while spinning in reverse, the starting torque exceeds the shaft strength and catastrophic shaft failure occurs. But this explanation is counter to my understanding of electric motor starting torque. Impellers and Reverse Rotation Our initial reply highlighted first what our reader, of course, knew: There are many cheap pumps and a few well-designed, more expensive pumps. When there was still an abundance of common sense (decades ago, perhaps), some wise man wrote that we always get what we pay for. Of course that is still true today and it applies to entire machines, their component parts, services, employees, contractors, educators, consultants---whatever. Back to the point: User experts Ed Nelson and John Dufour (Ref. 1) noted that nearly all impeller thread arrangements (left hand versus right hand) for single-stage end-suction pumps are such that the impeller fit gets tighter if the pump shaft rotates in the asdesigned direction. Conversely, the impeller gets loose (or spins off the shaft) if the pump shaft is rotated in the opposite, or unintended, direction of rotation. Impellers installed with close-fitting keys (Figure 2, left representation) properly mated to the shaft will not spin-off in case of inadvertent reverse rotation. Properly designed keyed fits make it near-impossible for an impeller to come off in the event of a malfunctioning check valve causing reverse pump rotation.

2 Figure 1: Screw-on (dead-end threaded) impeller hub (Ref. 2) But pumps are occasionally running in the wrong direction and if they do, the reasons are quite easy to find. Pump and steam turbine discharge check valves have been known to leak. Metal distortion, seat erosion and hinge friction issues can occur over time. Auto-start or standby equipment is more vulnerable because block valves are deliberately left open. In another scenario an uninformed pump installer may decide to test for motor direction of rotation after the driver is already coupled up to the pump shaft. That s a risky way to verify direction of rotation which, of course, should have been checked before coupling the driving to the driven shaft. Yet, someone may have decided to save time by installing the motor and pump n the as-shipped (generally mounted on a base plate and coupled to an electric motor) condition. In that case the chances of a non-keyed impeller coming off are, of course, 50%. The damage potential is then several orders of magnitude greater than the hoped-for savings in time.. Know the Impeller Fastening Method No pump manufacturer has a universal impeller securing method suitable for all pump sizes and service environments. In fact, the fastening method is not usually shown on the manufacturer s standard drawings. Also, relatively few user-purchasers include process pumps in thorough MQA, which stands for up-front machinery quality assessment (Ref. 3). Some impellers are fastened to shafts by standard acorn nuts or similar components which the pump manufacturer buys in bulk quantities from a costcompetitive supplier. Wanting to save money up front, some users join pump manufacturers by purchasing parts from the lowest-cost sub-vendor or third-party supplier. That s why a good pump specification usually contains a clause requiring pump cross-section views and parts lists which the purchaser reviews during the bid evaluation

3 Figure 2: Different impellers and bores give clues regarding their attachment methods. The bore on the left impeller has a keyway. process. The commercial parts a pump manufacturer obtains from third parties must be identified in exact detail. The purpose is not a secret: Years later, an equipment owneroperator may have cause to buy replacement parts ( buy-out parts ) directly from their respective manufacturers or from entities that produce superior, upgraded, components. Fasteners can be another source of problems. Hardness and metallurgy must be observed, which brings us back to MQA. Usually, an impeller spins off only if it is not properly secured. But even a keyed impeller fit jeopardizes reliability if the key is loosely fitted. And, whatever their size and speed, pump impellers secured by castellated nuts or tab washers must retain impellers in a manner that does not allow them to come off while operating in any direction. That's why we should examine drawings before we purchase; we should also know how parts or machines work before we buy parts. While this does not mean that one needs 50 years of experience to buy a $10,000 pumps, it does mean that, in the interest of reliability and safety, one must ask lots of relevant questions before buying plant assets. The majority of superior pump designs use keys to secure impellers. A good key fit is a snug fit, which means that hand-fitting is generally advantageous (Ref. 2). The more vulnerable sharp-cornered keyways should be avoided. Half-keys are often superior to full keys. Keyways with generous bottom fillet radii and so-called sled runner geometry have low stress concentration and a more desirable (higher) shaft factor of safety. Well-designed shaft ends also have a generous fillet radius at the shaft shoulder (see Figure 4, later). As strength considerations prompt us to maximize the various radii, their contours must not interfere with the mating radius at the bearing inner ring. Verification takes time and is time well spent.

4 Shaft Deflection Induces Fatigue Failures Hydraulic forces act on pump shafts; the magnitude of these forces determines shaft deflection. Shaft diameter, hardness, metallurgy, fillet radius at shaft shoulder and distance to nearest bearing also influence how much the shaft will deflect. Shaft Figure 3: Force vectors (arrows) indicate how a bending load acts on the pump shaft and how the magnitude and direction of this force changes at low flow (left image) versus flow at design throughput conditions (right image). deflection is greater when centrifugal pumps operate at throughputs below design point or in excess of design point. Because the hydraulic force action (Figure 3) is of unequal magnitude and the shaft rotates, reverse bending will take place and fatigue failures are possible if the design is marginal. Examining the fracture surface and performing a simple stress calculation will give focus to weaknesses and available remedies. Possible risk reduction steps will present themselves, and future failures will be less likely when we thoughtfully select one or more experience-based upgrade option. Motor Torque and Shaft Stresses For the record: The starting torque of many motors is as high as 7-times the full normal running torque. Agreed, discharge check valves rarely leak to the point of allowing substantial reverse flow. But lean and mean plants don't always install these check valves--and if they do, they sometimes forget to include these valves in their preventive maintenance scheduling. In any event, a pump reliability and/or failure analysis review should include the piping and all systems in the pumping loop. Pumps, controls, valves piping and other elements mutually interact and all must be considered.

5 All mechanical parts failures attributable to FRETT Through diligent training and reading we learn about FRETT. We come to accept that any and all mechanical parts can fail only due to one or more of the four cause categories Force, Reactive Environment, Time and Temperature, FRETT. So, because a pump shaft is a mechanical part it can only fail due to FRETT. An excessive reverse bending force will exist if the shaft is too slender, if its allowable bending moment or twist-inducing torque input is exceeded, if its metallurgy is not suitable for the fluid environment, if it has been in service for an abnormally long time, or if it was operated at an excessively high temperature. The reader who expressed his opinion about shaft strength and motor torque would have to examine shaft shoulder radii (fillet radius dimension) and the resulting stress intensification factors, Figure 4. He would eliminate Figure 4: Stresses are intensified at higher D/d and lower r/d ratios (Source: Peterson, Formulas for Stress and Strain ) some of the FRETT causes, perhaps T for Time and T for Temperature. It might be evident to the reader that no Reactive Environment ( RE ) existed in the failed shaft system and his investigative efforts for finding the root causes of the shaft failure might come back to F = Force. Examination of motor starting torques may indeed show that starting torques can reach seven times normal running torque. Applying an inrush current while a shaft system spins in reverse can cause shaft breakage. In the reader s repeat failure example, it is also possible that several seemingly small deviations had combined. One can easily get away with one or even two deviations, but one rarely gets away with four or five. And what does this tell us as we ponder that, next to an electric motor, a pump is the simplest machine used by modern industry? Say a pump typically has 40 parts and yet fails relatively often. An aircraft jet engine has more than 7,000 parts and fails rarely. Why? We believe the jet engine and aircraft manufacturers strive for perfection; they disallow every known deviation. In contrast,

6 pump engineers strive for low cost because that s what the pump purchaser often seems to prefer over pump reliability. But top manufacturers and good engineers know that normalization of deviance can cause disasters. Their quest to find root causes of failure and not tolerating known deviances requires training and discipline, strict adherence to checklists and procedures, and allocating the time needed to do things right. In this instance we were not given enough information to accurately determine (from behind our desk) why the reader s pump shafts failed. We can only vouch for the greatly increased probability that a few seemingly minor deviations from best practice combine. Together, these deviations cause trouble. Their collective safety factors will vanish, impellers come off, and shafts break. There will never be a good substitute for following Figure 5: Sharp corners and edges negatively affect a machine s factor of safety procedures and for uncovering what happened in this instance. Replacing parts and restarting the rebuilt machine without addressing and eliminating the true root cause will set owners up for repeat failures. Finding root causes of failure and implementing sound remedial steps is the common sense course of action. That s exactly what aircraft jet engine manufacturers do, and with great success. From our response, the reader will have inferred that any and all verbal hints at what may have caused impellers to spin off or shafts to fail must be supported by factual observations and evidence. There will always be a cause-and-effect relationship and it should be our task to uncover this relationship. For the benefit of the reader asking the original question, numerous texts show motor torque curves and explain why starting torques and inrush currents can differ greatly from normal operating torques and

7 currents. Check valve action and condition may indeed be relevant. Impeller attachment is always important. Finally, pump shafts have safety factors which range from none to considerable. Safety factors are affected by shaft stresses and high stresses can exist in shafts with discontinuities such as steps, keyways, etc. Radiused corners (Figure 5) will reduce stress intensification in keyways; these and other upgrades can be found in sturdy machinery and are readily duplicated by diligent reviewers. We hoped that our response prompted the original reader to expand his horizons. He, and others, should start by considering better specifications. They should become familiar with keys or half-keys, keyways and how these blend back into the shaft with a sled runner contour. Failure risk can be affected by fillet radii, stress intensification factors and the many other well documented contributors to safe pump design and pump failure avoidance. References 1. Dufour, J. W. and Nelson, W.E., Centrifugal Pump Sourcebook, (1992) McGraw-Hill, New York, NY 2. Bloch, H. P., Pump Wisdom: Problem Solving for Operators and Specialists, (2011) John Wiley & Sons, Hoboken, NJ 3. Bloch, H. P. and Budris, A. R., Pump User s Handbook: Life Extension, 4 th Edition (2014), The Fairmont Press, Lilburn, GA 4. N L Pedersen - Stress concentrations in keyways and optimization of keyway design Dept. of Mech. Engineering, Solid Mechanics, Univ. of Denmark, (2010) Heinz P. Bloch resides in Westminster, Colorado. His professional career began in 1962 and included long-term assignments as Exxon Chemical s regional machinery specialist for the US. He has authored over 600 publications, among them 19 comprehensive books on practical machinery management, failure analysis, failure avoidance, compressors, steam turbines, pumps, oil-mist lubrication and practical lubrication for industry. He holds BS and MS degrees in mechanical engineering, is an ASME Life Fellow and maintains registration as a Professional Engineer in New Jersey and Texas.