WSRC-RP-2001-00605
Modifications to, and Vibration Analysis of,
Tank 7 Slurry Pumps, F and H Tank Farms
R. A. Leishear and David B. Stefanko
Westinghouse Savannah River Company
Aiken, SC 29808
This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
This report has been reproduced directly from the best available copy.
Available for sale to the public, in paper, from:
U.S. Department of Commerce, National Technical Information Service, 5285
Port Royal Road, Springfield, VA 22161, phone: (800) 553-6847, fax:
(703) 605-6900
email: orders@ntis.fedworld.gov
online ordering: http://www.ntis.gov/support/index.html
Available electronically at http://www.osti.gov/bridge
Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy, Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062, phone: (865 ) 576-8401, fax: (865) 576-5728, email: reports@adonis.osti.gov
1. Summary
Slurry pumps have demonstrated short life spans when operated in nuclear waste tanks. Their life approximates one thousand hours or » 42 days of continuous operation, evidenced by past performance in H-Area and F-Area at the Savannah River Site (SRS). Several investigations over the past six years have isolated the most significant reliability problems. These problems are seal and bearing failures caused by the vibrations of the long drive shafts in the pump, manufacturing tolerance accumulations, failures caused by material incompatibility between the waste and the lowest process bearing that is exposed to the waste, and vibrations which occur when the pump operates at critical speeds. Only vibration and material problems were corrected. Potential bearing and seal degradation still exists for those pumps with a critical speed near the operating speed. Bearing damage can be expected below 700 rpm.
The pumps are used to mix or slurry nuclear waste products contained in waste storage tanks prior to transferring the tank contents for further processing. In particular, Lawrence Pumps, Inc. slurry pumps are installed on Tank 7 in F Tank Farm. Appendix A provides the initial recommendations, and further states that this follow up report would provide detailed descriptions of the pump components, failure mechanisms, and corrective actions which include tilt pad bearings, a Stellite process bearing, and modified split shaft retainers. By testing the pumps in a non-radioactive test facility, these corrections have been shown to significantly decrease the vibrations associated with bearing and seal failures, and consequently are expected to improve reliability.
2. Introduction
Thirteen pumps were purchased from Lawrence Pumps, Inc. (LPI) for installation in F and H Tank Farm. Four were unmodified, and installed in Tank 8, four were modified and installed in Tank 7, and four of the remaining pumps are tentatively planned for modification and installation in Tank 11. Modifications were performed in accord with the initial recommendations made by M. S. Miller, C. L. Sharpe, D. B. Stefanko, and R. A. Leishear [2001], Appendix A. The report herein encompasses all changes made to the pumps.
The recommendations are based on several bearing and seal failure investigations, recorded in Appendices B through D. The history of the failures and investigations is documented by Charles G. Sharpe and David B. Stefanko.1 As early as 1995, E. Kinard Fennel recommended reducing vibration in similar Sulzer-Bingham pumps by mechanisms such as tilt pad bearings.2 Recently, Wes Franklin , of Bently Nevada Corporation, confirmed that identical vibration problems existed in the Lawrence pumps.3 Mark Corbo extended the understanding of these vibration induced seal and bearing failures by not only modeling the vibration problems inherent in the shaft design, but by modeling the tilt pad bearing solution.4
Parallel to the vibration investigation, Operations personnel found that pumps installed in Tank 8 were unable to be rotated by hand. Several studies were performed to address this concern, and are documented in Appendices G through J. Corrosion testing performed by Mickalonis, Imrich, Jenkins, and Wiersma demonstrated that corrosion occurred on the process bearing material.5 Ortner, Sharpe, and Monahon consequently deduced that the process bearings jammed due to corrosion product build-up.6 Then Mickalonis and Bowser performed corrosion testing, followed by J. D. Miller performing abrasion testing. 7,8 The test results were then used to select appropriate materials for the remaining pump bearings.
To clarify vibration and corrosion issues, a more detailed discussion follows.
2.1 Pump Description and Failure Identification
A pump description is required to adequately describe the reliability failures. Both Sulzer–Bingham and Lawrence Pumps Inc. have provided slurry pumps to SRS; each provided reliability which was lower than desired. Although the reliability problems are similar for each pump, the subject of this report is the Lawrence pump.9 (fig. 1) A cross section of the pump 9 (fig. 2) displays the overall construction of the pump, specifically the parts that were changed: the eight inhibited water lubricated bearings referred to as the shaft bearings, the shaft retaining rings, and the single process lubricated bearing. The details of the changes are documented in the design changes to vendor print file for the pump.10
The pumps are installed through risers located at various locations on the tank tops. The risers are cylindrical openings in the concrete tank tops that may be sealed with removable concrete plugs, or in some cases, sealed with equipment such as slurry pumps.
Referred to as a long shaft pump since the shaft connecting the impeller to the motor is approximately 45 feet long, the rotating pump acts as a mixer by drawing the waste into the pump suction and then discharging a high velocity stream or jet back into the tank waste. The waste stream originates at a low pressure and velocity, and then passes vertically up through a protective screen into the pump casing’s suction inlet at the bottom of the pump. The impeller then accelerates the waste through the pump volute and out the two pump discharge nozzles. The jets formed at the nozzles entrain waste as they expand into the tank and either dissolve solid salts or lift sedimented waste, called sludge, from the tank bottom. In either case the jet impingement on the sludge or salt is the motive force to both initiate mixing and to maintain the slurried material in suspension.
The process bearing uses waste as the lubricant, and fits the shaft immediately above the impeller. The process bearing is a journal bearing construction and consists of two parts; the cylindrical sleeve on the shaft, and the bushing which surrounds the sleeve. The bushing was originally made from machined Silicon Carbide while the sleeve was actually a Tungsten Carbide coating, flame sprayed onto the shaft, and machined to its final dimension. As mentioned, binding of the pumps has been attributed to failures of this bearing.
The impeller shaft passes through a mechanical seal at both the top and bottom of the pump column to prevent waste escape from the tank through the pump, which would result in tank top contamination. Effectively a dual seal is maintained by pressurizing the inhibited water between the two seals. Three shaft sections couple together inside the column, using split retaining rings to form a single drive shaft restrained by eight shaft bearings and the process bearing.
The bearings are enclosed within the pump housing or column, and are lubricated by corrosion inhibited bearing water. Essentially stagnant in the column, the inhibited water provides some cooling to the bearings, as well as a fluid film between the shaft and the bearing. This film forms as the shaft rotates and is known as a hydrodynamic film that minimizes shaft to bearing rotational contact which reduces frictional wear between the surfaces, and increases the bearing life. The cylindrical Graphalloy (a nickel impregnated carbon) bushing that was originally installed around the stainless steel shaft acted as a journal bearing. Circumferential shaft grooves below the journal bearing have been observed on LPI slurry pumps, Sulzer-Bingham pumps, and even Hazleton transfer pumps. Apparently, similar vibration problems plague all long shaft pumps used in SRS waste tanks.
A radial bearing is enclosed within the mechanical seal at the top of the column and below the motor coupling. The motor thrust bearing supports the shaft and impeller weight while the motor stand supports the column and pump casing weight. The column and shaft are thus independent except through the film layers between the shaft and process bearings and between the seal interfaces.
The shaft continues from the thrust bearing through a shaft coupling to the one hundred and fifty horsepower motor, operated by a variable frequency drive (VFD). The VFD has the capability to operate the pump from 10% of full speed to a speed above 2000 rpm, even though the operating speed is typically 1600 rpm.
2.2 Failure Mechanisms
To date, the actual and potential failures are attributed to vibrations. On the one hand, the bearings fail as a direct result of increased vibrational impacting between the bearing components. Spalling of the bearing surfaces and consequent bearing wear follow this loosening of small chips, leading to bearing degradation. On the other hand, seal failure results from overheating. Within the seal, a flat annular rotating surface contacts a mating stationary face.9 (fig.5) Separated by a fluid seal, one rotates with the shaft while the other remains stationary with respect to the column. The fluid seal is created by pressurizing the water inside the column. As the seal rotates the pressurized water travels radially outward between the seal faces. A radial pressure gradient exists through the water as it travels across the seal face. This pressure gradient prohibits waste migration or diffusion from the tank inward to the pump column. Ideally the water moving across the seal face will evaporate as it approaches the outer edge of the seal due to frictional heat. This evaporation prevents inhibited water leakage into the tank so that the seal will be effective in both directions. Excessive vibration increases the friction on the seal surfaces, and increases the available heat for evaporation. Typically separated by distances of 6 – 20 millionths of an inch, the seal surfaces are sensitive to the fluid film between them. Increased evaporation eliminates the film and permits contact between the rotating faces. The greater the vibration, the greater the contact, the greater the surface wear, the shorter the seal life, and consequently the greater the leak rate of water into the tank. The goal is to remove the tank contents, not to add inhibited water back into the tank. Also, the two failures may combine. If the bearings are damaged, greater radial shaft motion may be permitted which increases the vibration across the seal surface with consequent seal wear increase and seal damage. Since vibration causes the failures, a discussion of vibration is warranted.
3. Vibration Reduction Using Tilt Pad Bearings and Modified Split Shaft Retainers
3.1 Problem Definition
A brief discussion of testing methodology and rotational vibration follows to explain the vibration induced bearing and seal failures. More detailed vibration discussions are available in Karrassik’s, et. al., Pump Handbook 11 or James E. Berry’s, Vibration Analysis I and II.12,13
Pumps were installed at the SRS Full Tank Facility to facilitate testing. The pump was mounted to an overhead structural steel platform that spans an eighty-five foot diameter, eight foot deep tank. 9 (fig. 6) Operational vibration readings were obtained with the impeller submerged in water.
To identify the initial pump vibrations, the vibrations were measured at various locations on the pump column and at the exposed shaft near the impeller. The cascade plots are similar for each location except that the vibration quantities shown on the right side of the cascade plots differ. The shaft deflection is quantified 9 (fig. 8), and is measured in mils (thousandths of an inch). Column vibration amplitudes are measured in velocity (inches / second). 9 (fig. 7) From basic physics, the deflection, velocity, and acceleration are mathematically interchangeable quantities, implying that any one of the three may be measured to obtain the other two. Eddy current proximity probes were used for the shaft measurements. Electrically measuring the changes in resistance between the probes and the shaft as the shaft rotates, these transducers provide continuous measurement of the shaft deflection. Piezoelectric transducers were used for the column to measure the force that is proportional to the acceleration. Similar to the crystals used to ignite propane gas cooking grills, these transducers contain a piezoelectric crystal that emits an electrical signal when a force is applied to it. The piezoelectric signals, as well as the eddy current signals, are mathematically interpreted by vibration analysis systems to provide graphic displays of displacement, velocity, and acceleration.
Two different vibration analysis systems were used throughout testing to collect the vibration signals. The raw data made available from each of these systems is tabulated according to serial number in Appendices F and G. The CSI system is a hand held analyzer actuated by the user, whereas the Bentley Nevada ADRE system provides continuous sampling. Of the two, ADRE creates the cascade plots. Of note, the vibration data readings in the Appendices are referred to as Flange Long and Flange Short, referring to the orientation of the transducers with respect to the platform’s long or short direction, and if required the flange numbers are indicated in Appendix E sketches. Sampling dates and the analyzer type are provided on each plot.
Figures depicting each of the vibrations as the pump operates at different speeds are called cascade plots due to their waterfall appearance. 9 (fig. 7,8) The left axis of the plot lists the motor speeds, the lower axis lists the frequencies, and the right axis lists the vibration amplitude. The vibration at 1900 rpm can be inspected using the three plot axes to see the vibration effects. Moving to the top of the left axis at 1900 rpm, the top vibration record is observed to move left across the length of the plot. Looking at the plot center, and panning either left or right from 0 Hz., several significant vibration spikes or peaks can be observed. The first peak observed occurs at 15.833 Hz. or ½ X, followed by a 1X, a 3X, and a 5X frequency. Each of these frequencies has a specific cause, and understanding each cause may lead to a method to reduce that specific vibration.
When equipment rotates numerous vibrations are created, and their vibration energies at a point on the machine add together to create the overall total machine vibration.
3.1.1 Resonant Frequency / Critical Speed
The LPI slurry pump will be used as an example to clarify this last statement, and the first vibration to be considered is referred to as the 1X, or one times, synchronous frequency. The term 1X refers to the fact that this frequency occurs at one times the running speed of the machine. If the machine speed is 1600 rpm, which happens to be the maximum permitted running speed for this pump, then the 1X frequency is 1600 rpm / 60 seconds / minute = 26.67 Hertz (Hz). Any imbalance or misalignment of the shaft, the impeller, or other rotating components directly affects the magnitude of the 1X frequency. A shaft imbalance bends the shaft, and both radial and axial vibrations are created each time the bent shaft rotates, whereas misalignments essentially cause the shafts ends to impact each other during each rotation, creating axial vibrations. If no imbalance or misalignment exists, there would be no 1X vibration, virtually impossible. When the machine speed is changed through VFD control, the 1X frequency changes proportionally.
A shaft imbalance and deflection problem was observed on 91103-10. When the pump was first operated, the shaft deflections actually caused binding of the shaft and the coupling. Audible rubbing between components was noted. Subsequent changes to the tolerances within the tilt pad bearings improved pump performance by reducing the vibration at the critical speed, but vibration levels were still unacceptable. Operation at the critical speed is not recommended since bearing degradation will be accelerated once degradation begins. A tolerance study of the assembled parts may be performed, using the dimensions from the VPF. The accepted statistical technique of the root sum of the squares shows the tolerance accumulations will approximate 0.015 inches, which is of course the bending deflection. Unfortunately, the assembly process of aligning registers, or bosses, from column to column does not permit correction of a built in tolerance problem.
During pump operation, it was also noted that the bearings were heard to tap against the shaft, indicating that the film layer between the bearing and the shaft was insufficient to prevent metal to metal contact below approximately 700 rpm. Minor shaft damage can be expected when the pump is operated below this speed.
Another vibration aspect typically related to the 1X frequency is resonance, which occurs when the natural frequency of a component equals the operating frequency or speed. Similar to the audible frequency of a tuning fork, the shaft and the column each have a natural frequency. If this frequency is excited, excessive vibrations are readily observed. Exciting the shaft resonance by an external force, excessive shaft vibration can be expected. On these pumps, significant vibration increases are noticed near 850 and also at 1600 to1800 rpm. Graphically displayed throughout Corbo’s report4, these mode frequencies coincide with the first and second mode natural bending frequencies of the pump column. Observed higher mode frequencies such as the second, third, fourth, etc. are multiples of the first mode frequency, and are documented in Appendix E. Those frequencies above the second mode are out of the operating range of the pump and thus not of concern.
To approximate these resonant frequencies, the pump column is struck with a heavy plastic hammer, and the vibration frequency is then measured. The resonant frequency approximates the critical speed of the pump. The critical speeds change somewhat from the resonant frequency when the pump is operated. Coupling of the shaft through the bearing film increases the column stiffness, and consequently the critical speed. In short, the critical speed is better defined through measurements during operation, rather than through static measurements. Operational speeds are frequently restricted to within 10 % of the critical speeds, since process bearing and seal damage can be expected from operating at these speeds.
These resonant frequencies are also intolerant of minor structural differences in pumps. The bending frequency was measured, and found to be different in different directions on the same pump. Although the frequency only varied by about 10%, the cause was isolated to the positioning of rather small fins on the pump column. This small structural difference resulted in a significant frequency change. From this conclusion, the difference in critical speeds of » 1600 to 1764 rpm for various pumps can be easily attributed to minor changes in pipe wall thickness. The critical speed also varies by about 10 % along the length of the pump due to variations in construction along the pipe column length.
The frequency will be lowered when the pump is installed at the tank top with mounting cans. These cans surround the pump column, and combinations of different length cans can be used to elevate the pump. The addition of cans to the pump lowers the pump frequency, since the system stiffness may decrease. A decrease in frequency is expected due to the increased flexibility of the mounting cans. The frequency increase occurs with a stiffness increase according to
where w equals the frequency, k equals the stiffness, and m equals the mass.
The critical speed is unaffected by bolting the pump to the cans. The typical radial deflection of the pump at each bearing is approximately 0.005". Using this as the deflection at a distance five foot from the mounting surface, and using a 14" diameter column section, along with Table 3, Equation 1a, from Roark's Formulas for Stress and Strain, the moment at the top of the pump may be determined. With this moment, and Table 24, Equation 23, from Roark, the pressure distribution on the flange face may be calculated. This calculation provides a maximum stress on the flange surface of approximately 20 psi. The 13,000 pound pump will pivot slightly due to this stress. This pivoting action alters the column frequency by less than 1%, and is considered negligible.
The critical speed is also unaffected by water levels in the tank. Resonant frequencies were measured on the pump for two different conditions: eight feet of water in the tank at the Full Tank Facility, and no water in the tank. The resonant frequencies increased by about 1 – 6% along the pump length. The increase in frequency occurs due to the stiffening effect that the fluid exerts on the pump as the pump vibrates. The denser water increases the stiffness more than air. Fluid properties differing by orders of magnitude have a maximum effect of 6%. The difference in water levels are within one order of magnitude, and are thus expected to have a much smaller effect on the stiffness. The effects of water level in the tank can be neglected when considering critical speeds.
3.1.2 Shaft Whirl Vibration
The ½ X frequency results from a condition known as shaft whirl. The comparative data is available in Appendix G, S/N 91103-09. 9 (fig. 7) Along with the shaft vibration at the 1X frequency, another slower vibration is superimposed on the shaft between the bearings and is known as the shaft whirl. Whirl is somewhat difficult to visualize. Thomson 14 provides a vector discussion of whirl, but analogy can be used to explain whirl. If the pump or shaft is assumed to be a single line fixed at its two ends, and a point on the line is selected, its motion can be visualized by comparing this point to the orbit of the moon around the earth, and the earth around the sun. The position of the sun represents the shaft at rest, the centerline of the shaft. As the shaft rotates with a 1 X imbalance, the centerline moves out to the earth’s position and the rotation here equals the 1 X speed. If a ½ X whirl component is added to the vibration, the shaft then rotates as the moon would rotate about the earth centerline. The shaft is then circling the imaginary shaft center at a 1 X speed, and also spiraling about the shaft orbit at a ½ X speed. For the long shaft pumps, the ½ X component has been attributed to fluid instability associated with journal bearings by both Franklin and Corbo. 3,4 The unstable fluid film permits abrasive contact between the shaft and bearing as the film breaks down.
3.1.3 Other Vibration Components
Other motions are also superimposed on the shaft, the 3X is attributed to parallel misalignment of the shaft at the retaining rings. Shaft misalignment may result in any combination of the 1X, 2X, and 3X bending vibrations. In this case the 2X is negligible.
The 5X frequency is also superimposed on the shaft and is attributed to impeller induced vibrations. Since there are five vanes in the impeller, five distinct vibrations due to bending are measured by a stationary transducer as the shaft moves through one complete rotation. Hence, the 5X frequency is observed, and is usually caused by either hydraulic forces or cavitation on the vane surfaces.
Attributed to cavitation, minimal broad band, axial, random vibration was observed on LPI91103-01 and – 03 near the lower end of the pump. And, incipient cavitation has been heard on - 03. Vapor bubbles are created along the vane surfaces of the impeller where the high velocities of the water create low pressures. If these low pressures drop below the vapor pressure of the water, bubbles are created. Considered to be a low temperature boiling process, the created bubbles travel with the flow until they reach an area of higher pressure where they implode. Since the implosion occurs at a point, the energy rapidly dissipates. If these implosions occur near material surfaces cavitation erosion may result. The axial nature of the cavitation vibration indicates that it occurs in the volute of the pump casing, which is the expanding section of the pump casing preceding the discharge nozzle. Operation of LPI 91103, serial number – 01 at the Full Tank Facility for over 500 hours demonstrated a complete lack of cavitation damage, during visual inspection.
An electrically excited vibration was also observed on the motors of all pumps for Tank 7 and Tank 8. The electrical induced vibrations are documented in Appendix F, S/N 91103-10. High frequency vibrations were proportional to multiples of the electrical frequency. These numerous vibration components changed frequency as the VFD changed the pump speed. Since the vibration was a function of speed, bearing failures were eliminated as a potential vibration cause. The electrical excitation was attributed to either an eccentric air gap between the rotor and stator of the motor which causes an imbalance, or an unfiltered VFD signal which caused electrical pulsations of the motor. Regardless of the specific cause of the electrically induced vibration, the observed vibration levels were considered to be acceptable.
3.1.4 Vibration Acceptance Criteria
Once the vibrations are identified, the vibration magnitudes need to be considered. What is an unacceptable vibration? In general, this question requires a case by case evaluation. What are the maintenance issues? For these pumps, their installation in a nuclear environment prohibits maintenance. What is the life expectancy? As noted, the present life was unacceptable. At SRS, typical vibration velocity criteria for pumps is 0.2 inches / second using CSI equipment, 0.3 inches / second using the ADRE equipment which is more sensitive to lower frequencies. These values are readily acceptable peak vibration values. Beyond these values pump tests are suspended pending vibration evaluation.
The presented initial pump vibration data leads into a discussion of the actual pump testing and operating results.
3.4 Theoretical, Experimental, and Operational Results
Eight pumps were initially tested, four were considered unacceptable based on Appendix F, CSI vibration data which indicated vibration velocities in excess of 0.2 inches /second. To validate the recommendation based on vibration, one of the pumps considered unacceptable, 91103-01, was operated at the Full Tank Facility. As a result of a bearing failure, shaft damage occurred in less than five hours at the resonant frequency. 9 (fig. 4) This failure validated the application of vibration analysis to the pumps. The four pumps installed in Tank 8 with journal bearings were expected to have short lives. They have operated approximately 1500 hours, and only one pump experienced seal leakage before the tank contents were emptied during waste removal operations. The present pump status is then: four pumps are installed, four pumps considered unreliable and were modified, one pump was modified and still had high vibrations at the critical speed, and four pumps are untested and need modifications.
The initial vibration analysis provided a tangible basis from which to validate a model. Detailed in Appendix C, the rotordynamic finite element model was not only validated, but the model provided the minimum design requirements for a tilt pad bearing. With these requirements and the existing bearing housings supplied with the pump, Turbo Components completed the final design for the tilt pad bearings.
Corbo’s predicted vibration reductions are clearly demonstrated.9 Comprehensive raw data for the vibration results after testing, for pumps 91103-01, -02, -09, -10, and -11, are available in Appendix F. The data consistently indicates that the ½ X vibration is virtually eliminated along the shaft between tilt pad bearings, and the vibration is nearly eliminated near the impeller where the shaft is supported by one tilt pad bearing and the process journal bearing. The bearing life should improve, since the ½ X spiraling vibration component would no longer repeatedly impact the shaft against the bearings during rotation. The seal life should increase since the vibration is reduced.
The column resonant frequencies are solely dependent on the pump structural construction. Since the column remains unchanged, the resonance remains unchanged. But of course, the resonant frequencies are affected by the installation of cans on the pumps at the tank, as discussed above.
The vibration effect of modifying the retaining ring is questionable, but the mechanical looseness associated with the original installation bore correction. The 3x vibrations decreased but the opposing 1X increase cancelled the vibration improvement. 9 (Fig. 7,8,19,20) Also of note, column alignment improvements are unavailable for the existing pumps.
Since hydraulic issues were not addressed, the 5X vibration components are unchanged.9
3.5 Tilt Pad Bearing Conclusions
Journal bearings for use in long shaft pumps have potential vibration instabilities that have plagued slurry pumps. In particular, the Tank 8 pumps have a journal bearing construction that cannot be operated above 1600 rpm due to the fluid instabilities and their consequent high vibrations. Replacement of the shaft journal bearings with tilt pad bearings partially reduced vibrations at the resonant frequencies. Shaft whirl vibrations are eliminated or reduced to improve pump reliability by reducing the vibration failure mechanism of the shaft bearings and seals in the four installed Tank 7 pumps. This vibration reduction would permit operation of the pumps in the range above 1800 rpm: 1900 rpm ideally. Operating below 700 rpm is discouraged, since bearing damage may occur.
4. Material Modifications to the Shaft and Process Bearing
4.1 Problem Definition
Material selection for both the process and shaft bearings needed reconsideration. The process bearings had jammed, and the shaft bearing failure cause had not yet been isolated to vibration.
The lowest bearing on these pumps is the process bearing and is lubricated by the waste in the surrounding Waste Tank. This hostile caustic environment includes sodium nitrate, sodium nitrite, sodium aluminate, sodium hydroxide, sodium chloride, abrasive Zeolite particles, and small oxide and hydroxide concentrations of the following materials in solution: iron, manganese, aluminum, calcium, nickel, and uranium. In the original design, the bearing was essentially a journal bearing. The inner sleeve of the bearing was not an independent part, but was in fact a Tungsten Carbide coating flame sprayed onto the shaft, allowed to harden, and machined to the appropriate dimensions. The outer bushing of the journal bearing was fabricated from silicon carbide and interference fitted into the bearing housing.
The shaft bearings enclosed in the pump column are subjected to an inhibited water solution of pH = 12, containing Sodium Nitrite, and Sodium Hydroxide.
4.2 Experimental Results
To select the appropriate materials, corrosion tests were performed for both the shaft and process bearings, while abrasion testing was performed for only the process bearing.
The corrosion test data presented in Appendix I details the performed testing and chemical composition of the materials. Tested materials are used as electrodes and inserted in an electrolyte, and are performed in accordance with national standards. This accelerated corrosion is used to establish corrosion rates. Test results were satisfactory for Stellite 12, Stellite 712, and Waukesha 88.
Abrasion tests were also performed in accordance with national standards and are detailed in Appendix J. A block of material is rubbed back and forth over a flat substrate that is immersed in the fluid of concern. The test fluid contains corrosion inhibitors and abrasive material to imitate abrasive frit found in some waste tanks. For the purposes of this report the abrasive frit parallels the expected abrasiveness of the Zeolite found in Tank 19. Several materials were tested: Silicon Carbide, 410 stainless steel, Nitronic 50, Stellite 12, and Stellite 712. Of these materials, Stellite 712 was found to have the best abrasion resistance.
Two consecutive tests were performed to validate the pump’s performance after the changes were implemented. A preliminary, 288 hour, pump test was performed on 91103-01, and a 72 hour acceptance test followed. Each of these tests was performed at 1600 rpm. At this speed, over 27,000,000 cycles occurs. Since cyclic failures due to fatigue generally do not occur in metals above one to ten million cycles, this test duration was expected to identify all fatigue failures. Thorough disassembly and inspection of parts was performed after the preliminary test.
An unsatisfactory observation following the tests at the Full Tank Facility was that the process bearing Tungsten Carbide sleeve was worn in the area contacted by the Silicon Carbide bushing. A square groove was cut into the sleeve by the bushing. Informal testing showed that the sleeve could be scratched with a file of a lower hardness than the published hardness value of the sleeve. Apparently the coating not only failed corrosion requirements, but it did not meet hardness requirements either. Regardless of the hardness question the materials needed replacement due to the material test results.
Some minor scratching on the bearing surfaces was attributed to overly tight bearing clearances. The bearing clearances were increased to solve the potential bearing to shaft interference. Subsequent testing demonstrated that the vibration levels were reduced at resonance due to the increased clearances.
4.3 Process and Shaft Bearings, Materials Conclusion
Waukesha 88 is the preferred material for the tilt pad bearings and Stellite 712 is the preferred material for the process bearings. The tilt pad bearings provided by Turbo Components were made from Waukesha 88. The coating on the shafts was removed and replaced with a Stellite 712 press fitted bushing. And, the Silicon Carbide process bushing was replaced with a press fitted Stellite 712 bushing.
References