Using a step-by-step analysis process, engineers correctly diagnose many gear failures and, more importantly, develop solutions that prevent such failures from happening again. The basic steps in this process, which a previous article (“How to analyze gear failures”) describes in detail, include:
• Inspect failed parts.
• Determine type of failure.
• Conduct tests and perform calculations.
• Form and test conclusions.
The following examples demonstrate how engineers have used this basic approach to solve gear failures in various applications — wind turbines, waste water treatment plants, hydroturbines, and tunneling machines.
Wind turbine gears
Two years after a wind turbine farm began operating, technicians found that six of the 24 gearboxes for the wind turbines were noisy and one of these had jammed so that it couldn’t rotate. The remaining units continued to operate smoothly and quietly.
Investigators first interviewed the owner and service technicians to obtain operating histories of the wind turbines, then they inspected the six noisy gearboxes plus two others that appeared to be operating normally.
The exteriors of the gearboxes showed no signs of oil leaks, overheating, or unusual contamination. After opening the inspection ports, the six noisy gearboxes were found to have broken teeth on the low-speed pinions. The one jammed gearbox had broken teeth on both pinion and gear. Gears from the two units that operated normally had no apparent damage.
After disassembling the eight gearboxes, investigators inspected, documented, and photographed all components.
The oil sump of each damaged gearbox yielded broken teeth, which were collected for later inspection. The jammed gearbox contained so much damage, it would have been difficult to analyze. So attention shifted to the other five damaged gearboxes.
Each of the five pinions had bending fatigue fractures on two adjacent teeth, Figure 1. In each case, beach marks on the fracture surface showed that the fatigue crack originated in the root fillet at the end of the leading side of the tooth, which is in tension, Figure 2.
The inspection revealed several facts:
• On each pinion, the first broken tooth (first one to be loaded as the gear rotates) had a smooth fracture surface, which is characteristic of slow crack growth, whereas the trailing tooth had a rougher surface, indicating faster crack growth, Figure 2.
• Teeth adjacent to the fracture had extensive macropitting, whereas teeth away from the fractures were undamaged, except for mild abrasion.
• Working surfaces of the teeth that were retrieved from the oil sumps had no pitting or other damage.
• Color photos of the failed pinions disclosed copper plating on the tooth ends and in the adjacent radius with the integral shaft, commonly known as the grinding relief radius for the bearing journal, Figure 1. The copper plating was at the same end of the teeth, and in the same area, where the fatigue cracks originated. On the two undamaged pinions, copper plating was present in the grinding relief radius, but stopped short of the tooth ends.
• Gears from the two gearboxes that operated normally had no apparent damage. Contact patterns taken from these gears with marking compound were centered, showing that the teeth were properly aligned.
These facts led to the following conclusions:
• The primary mode of failure was bending fatigue originating at the tooth ends that were copper plated.
• The lead tooth failed first, by slow crack growth. As this tooth cracked, it transferred load to the following tooth. This following tooth cracked more rapidly, because it was overloaded due to transferred load from the lead tooth.
• Macropitting on teeth adjacent to the fractured teeth was a secondary failure mode that was caused by overloading due to loss of load sharing. This conclusion was substantiated by the lack of pitting on teeth located away from the fractures and on the broken teeth retrieved from the oil sumps.
• The gears were properly aligned. Normally, misalignment is suspected when fractures originate at the ends of gear teeth. In this case, however, the wear patterns, tooth contact patterns, and bearing conditions all indicated good tooth alignment.
• Based on these conclusions, investigators formed a hypothesis that the pinion teeth failed, because the copper plating prevented their ends from being properly hardened.
Copper plating is used to mask areas that shouldn’t be hardened. The copper prevents carburizing during heat treatment so these areas develop low hardness. It is likely that the bearing journals of the pinions were plated (to minimize hardness and ease machining) by lowering them into the plating solution. Apparently, some pinions were lowered too far into the solution so the tooth ends were inadvertently plated.
A metallurgical laboratory then tested the gear teeth to confirm (or disprove) the hypothesis. A scanning electron microscope (SEM) equipped with an energy dispersive X-ray (EDX) revealed striations on the tooth fracture surfaces, proving that the fractures were caused by fatigue. An EDX analysis confirmed the presence of copper plating at the fractures.
Metallurgical sections were prepared by cutting and polishing the ends of teeth from both damaged and undamaged pinions. Hardness surveys on these sections confirmed that the copper plating prevented carburization of the ends of the fractured teeth, so that hardness was limited to 40 HRC. On the undamaged teeth, however, the hardness gradient ranged from 60 HRC at the surface to a core hardness of 40 HRC, which is normal.
Based on these tests, it was concluded that the failures were caused by improper heat treatment due to copper plating on the tooth ends. The resulting low hardness and low strength allowed the bending fatigue failures to occur.
The wind turbine owner subsequently found copper plating on the tooth ends in four more gearboxes. Technicians replaced these pinions two years ago, and the gears have operated with no further failures.
Waste water aeration blower
At a waste-water-treatment plant, there are two blower drives, each using a diesel engine operating through a speedincreasing gearbox to drive a roots-type aeration blower. Each gearbox has double helical gears on parallel shafts and converts the 900-rpm engine speed to 3,600 rpm at the blower.
After only 3 months operation, both gearboxes were replaced because of noise. The replacement gearboxes also became noisy after only 8 weeks, prompting the operators to investigate.
Continue on Page 2
The gear noise could be heard over the roar of the diesel engine and the scream of the blower. After shutting down the drive, investigators found severe macropitting on both pinion and gear teeth, Figure 3. Pitting was uniform across the face, indicating that the gears were properly aligned. It was obvious that the gears were noisy because of the pitted tooth surfaces.
To prevent further damage, the owners shut down the drives and sent both gearboxes to a vendor for immediate repair.
Disassembly of the gearboxes revealed that, except for extensive pitting on one side of the gear teeth, all components were in good condition. Tooth hardness was within specifications. Measurement of the undamaged, backside of the gear teeth showed that tooth accuracy, AGMA Quality 11, and surface roughness, 25 μin, were appropriate for the application.
Laboratory analysis of the gearbox lubricant showed that its viscosity was ISO VG 68, which conforms to the gear vendor’s recommendation.
Calculations according to ANSI/AGMA standard 2001-B88 (spur and helical gears) showed that the gears were lightly loaded, so their pitting life should have been more than adequate. However, the calculated specific film thickness ( l) of the lubricant was only 0.4, indicating that the gears had only boundary lubrication. The estimated probability of wear, Pw, was calculated as over 80%.
The ANSI/AGMA lubrication standard 9005-D94 recommends a viscosity of ISO VG 150 for such an application. This would increase the specific film thickness to 1.0 and decrease the wear probability to less than 5%.
Investigators concluded that the gear teeth pitting and resultant noise were caused by the low viscosity of the lubricant, and recommended changing the viscosity to ISO VG 150. The plant replaced the gears and switched to the higher viscosity lubricant over 5 years ago, and the gearboxes have operated since then without a reoccurrence of pitting or noise.
Hydroturbine gears grumble
In a pit Kaplan type hydroturbine, water flowing through the turbine supplies power to a planetary gearbox, which increases the speed to drive a generator. Before installing one of these gearboxes in a 10-MW generator application, the manufacturer spin-tested it under low load and found it to be relatively quiet, emitting 82 dBA sound pressure level at full speed. After installation, however, the gearbox ran only 2 hr before a 100% load rejection test was performed on the turbine assembly. Immediately after the test, the gearbox was very noisy.
A load rejection test consists of opening a switch between generator and electrical grid while the generator is producing power, which causes the turbine to speed up. The governor must respond quickly, adjusting water-flow-control gates and turbine blades to limit overspeed and shut down the system. In this case, the tests imposed high loads on the gearbox shortly after startup and before the gear teeth had run-in.
The turbine supplier and gear manufacturer found nothing wrong with the gearbox. But a consultant measured the sound pressure level, which had risen to 110 dBA, and concluded that the gears were the source of the noise.
Believing there were resonance conditions in the pit structure, the turbine supplier and gear manufacturer added beams to stiffen the structure. But, this modification didn’t help.
After one year with no improvement, the plant operator had another consultant inspect the gears (without disassembling the gearbox). Unfortunately, planetary gearboxes have limited accessibility. Using a borescope, the consultant discovered that the planet gears had scuffing on the lower part of the tooth (dedendum) and the sun pinion had scuffing on the upper part (addendum), Figure 4.
Drawings of the gears, plus data on surface roughness of the gear teeth asmanufactured, and startup loads were used to calculate the contact temperature between gear teeth and estimate the risk of scuffing based on equations from ANSI/AGMA standard 2001-B88. These calculations predicted a 70% risk of scuffing, which is considered high.
A review of the inspection data, calculations, and drawings led to the following conclusions:
• Scuffing was the primary mode of failure. This conclusion was supported by the fact that the gearbox was quiet during tests at the factory and noisy immediately after the load rejection tests. Scuffing is an instantaneous failure that usually happens the first time a high load is applied.
• Certain gear design and manufacturing parameters were inadequate for use as a speed increaser rather than a reducer.
• The scuffing failure was caused by a combination of inadequate gear-tooth design for a speed-increasing application and inadequate run-in before applying full load. This conclusion was supported by the contact temperature calculations, which indicated a high risk of scuffing, and the operational logs that showed the load rejection tests were performed after only 2 hr of running.
Based on these conclusions, the consultant recommended gear design changes to reduce the risk of scuffing, plus gear run-in under reduced loads to improve the tooth surface finish before applying full load.
After several years of litigation, the turbine supplier and gear vendor finally conceded that scuffing caused the gear noise. The gears have been redesigned, and replacement gears were recently applied to the drive.
Tunneling machine gears that shine
A tunnel boring machine uses several planetary speed reducer gearboxes to transmit power from electric motors to the final drive gears. After only 300 hr of service, an analysis of the gearbox lubricant showed 2,400 ppm of iron, an unacceptably high value.
Investigators found severe polishing wear on the planet gear teeth and planet bearings, Figure 5 and Figure 6. The planet tooth surfaces had a mirror-like, but wavy finish with no machining marks. Surprisingly, only one side of each tooth was polished; the side that contacted the mating ring gear tooth. The other side, and the mating sun pinion teeth were not polished.
Rollers from the planet support bearings were severely polished in the area where they contacted their bronze retainer, Figure 6. The polishing wear had reduced the diameter of the rollers over two-thirds of their length. Consequently, loads from the inner and outer race were borne by only one-third of the roller length. Macropitting was found on the inner raceway of one bearing.
By load-testing one of the gearboxes in the lab, the gear vendor was able to duplicate the wear conditions, which meant that the wear was caused by something other than the severe environmental conditions of the tunneling operation. To determine why the polishing occurred, investigators conducted several laboratory tests:
• Spectrographic, X-ray fluorescence, and ferrographic analysis of the used lubricating oil.
• Measurement of the planet gear teeth accuracy on a gear inspection machine.
• Metallurgical inspection of a planet gear tooth for microstructure and chemical composition.
• Scanning electron microscope (SEM) analysis of a planet gear tooth, bearing roller, and bearing retainer.
• Tribometer experiments to investigate the mechanism of polishing wear. These tests yielded the following results:
• The lubricating oil contained 1.7% sulfur (an anti-scuff additive), and a high concentration (over 2,000 ppm) of small metallic particles, many of which were metal chips from abrasive wear.
• Polishing reduced the gear tooth accuracy significantly — from an original condition of AGMA Quality 12 to a worn condition value of 8.
• Metallurgical properties of the gear teeth showed no abnormalities.
• Polishing on the gear teeth and bearing rollers consisted of fine-scale abrasive wear.
• Particles of silicon, aluminum, and iron were embedded in the surface of the bronze bearing retainer. The silicon and aluminum probably came from environmental dust, and the iron particles were probably fine wear debris.
• Tribometer experiments showed that polishing wear is fine-scale abrasion promoted by a combination of a fine abrasive and a gear oil with chemically active additives.
These results led to the conclusion that abrasives became embedded in the teeth of the ring gear and in the bearing retainer, causing wear on the planet gear teeth and the bearing rollers.
It was recommended that the currently used sulfur-phosphorous lubricating oil be replaced by a less-aggressive borate lubricant with a higher viscosity to increase the lubricant film thickness. A second recommendation was to change the oil frequently to remove abrasives. Subsequent lab tests followed by several months of service have shown that these measures eliminated the polishing wear.
Note: This case history is reported in AGMA paper No. 90 FTM 5.
Robert L. Errichello is president of Geartech, a gear research, analysis, and design consulting firm in Albany, Calif.