Wear Particle Analysis - A Predictive Maintenance Tool, by Rob Lovicz & Ray Dalley, PREDICT

This presentation was originally made at PdM-2005 - The Predictive Maintenance Technology Conference, September 19-22, 2005 Atlanta Georgia.  Click here for future dates

Abstract: Wear particle analysis and Ferrography in particular is an effective means to identify and respond to maintenance needs. The development of this technology includes image analysis, on-line sensors, automated oil analysis screening tools, electronic transfer of evaluation results, and artificial intelligence. This technical paper describes different ways on how Ferrography works and its application in the real world through case histories. 

Key Words: Ferrography; predictive maintenance; Used Oil Analysis; Passport System V; Wear Particle Types; Fourier Transform Infrared Spectrometer; Viscosity

Wear Particle Analysis or Ferrography 

Ferrography is a technique that provides microscopic examination and analysis of wear particles separated from all type of fluids. Developed in the mid 1970’s as a predictive maintenance technique, it was initially used to magnetically precipitate ferrous wear particles from lubricating oils.

·         This technique was used successfully to monitor the condition of military aircraft engines, gearboxes, and transmissions. That success has prompted the development of other applications, including modification of the method to precipitate non-magnetic particles from lubricants, quantifying wear particles on a glass substrate (Ferrogram) and the refinement of our grease solvent utilized in heavy industry today.

·         Three of the major types of equipment used in wear particle analysis are the Direct-Reading (DR) Ferrograph, the Analytical Ferrograph System and the Ferrogram Scanner.

Direct Reading (DR) Ferrograph: The DR Ferrograph Monitor is a trending tool that permits condition monitoring through examination of fluid samples on a scheduled, periodic basis. A compact, portable instrument that is easily operated even by a non-technical personnel, the DR Ferrograph quantitatively measures the concentration of ferrous wear particles in a lubricating or hydraulic oil. The DR Ferrograph provides for analysis of a fluid sample by precipitating particles onto the bottom of a glass tube that is subjected to a strong magnetic field. Fiber optic bundles direct light through the glass tube at two locations where large and small particles are deposited by the permanent magnet. At the onset of the test, before particles begin to precipitate the instrument is automatically “zeroed” with a microprocessor chip as the light passes through the oil to adjust for its opacity. The light is reduced in relation to the number of particles deposited in the glass tube, and this reduction is monitored and displayed on a LCD panel. Two sets of readings are obtained: one for Direct Large >5 microns (DL) and one for Direct Small <5 microns (DS) particles. Wear Particle Concentration is derived by adding DL + DS divided by the volume of sample, establishing a machine wear trend baseline.

Machines starting service go through a wearing in process, during which the quantity of large particles quickly increases and then settles to an equilibrium concentration during normal running conditions. A key aspect of ferrography is that machines wearing abnormally will produce unusually large amounts of wear particles indicating excessive wear condition by the DR Ferrograph in WPC readings. If WPC readings are beyond the normal trend a Ferrogram sample slide is made with the fluid for examination by optical microscopy.

The Analytical Ferrograph: Additional information about a wear sample, can be obtained with the Analytical Ferrograph system, instruments that can provide a permanent record of the sample, as well as analytical information. The Analytical Ferrograph is used to prepare a Ferrogram -- a fixed slide of wear particles for microscopic examination and photographic documentation. The Ferrogram is an important predictive tool, since it provides an identification of the characteristic wear pattern of specific pieces of equipment. After the particles have deposited on the Ferrogram, a wash is used to flush away the oil or water-based lubricant. After the wash fluid evaporates, the wear particles remain permanently attached to the glass substrate and are ready for microscopic examination.

Ferrogram Maker Instrument

The Microscope: Ferrograms are typically examined under a microscope that combines the features of a biological and metallurgical microscope. Such equipment utilizes both reflected and transmitted light sources, which may be used simultaneously. Green, red, and polarized filters are also used to distinguish the size, composition, shape and texture of both metallic and non-metallic particles.

Types of Wear Particles: There is six basics wear particle types generated through the wear process. These include ferrous and nonferrous particles which comprise of:

1. Normal Rubbing Wear: Normal-rubbing wear particles are generated as the result of normal sliding wear in a machine and result from exfoliation of parts of the shear mixed layer. Rubbing wear particles consist of flat platelets, generally 5 microns or smaller, although they may range up to 15 microns depending on equipment application. There should be little or no visible texturing of the surface and the thickness should be one micron or less.

2. Cutting Wear Particles: Cutting wear particles are generated as a result of one surface penetrating another. There are two ways of generating this effect.

·         A relatively hard component can become misaligned or fractured, resulting in hard sharp edge penetrating a softer surface. Particles generated this way is generally coarse and large, averaging 2 to 5 microns wide and 25 microns to 100 microns long.

·         Hard abrasive particles in the lubrication system, either as contaminants such as sand or wear debris from another part of the system, may become embedded in a soft wear surface (two body abrasion) such as a lead/tin alloy bearing. The abrasive particles protrude from the soft surface and penetrate the opposing wear surface. The maximum size of cutting wear particles generated in this way is proportional to the size of the abrasive particles in the lubricant. Very fine wire-like particles can be generated with thickness as low as .25 microns. Occasionally small particles, about 5 microns long by 25 microns thick, may be generated due to the presence of hard inclusions in one of the wearing surfaces.

·         Cutting wear particles are abnormal. Their presence and quantity should be carefully monitored. If the majority of cutting wear particles in a system are around a few micrometers long and a fraction of a micrometer wide, the presence of particulate contaminants should be suspected. If a system shows increased quantities of large (50 micrometers long) cutting wear particles, a component failure is potentially imminent.

3. Spherical Particles: These particles are generated in the bearing cracks. If generated, their presence gives an improved warning of impending trouble as they are detectable before any actual spalling occurs. Rolling bearing fatigue is not the only source of spherical metallic particles. They are known to be generated by cavitation erosion and more importantly by welding or grinding processes. Spheres produced in fatigue cracks may be differentiated from those produced by other mechanisms through their size distribution. Rolling fatigue generates few spheres over 5 microns in diameter while the spheres generated by welding, grinding, and erosion are frequently over 10 microns in diameter.

4. Severe Sliding: Severe sliding wear particles are identified by parallel striations on their surfaces. They are generally larger than 15 microns, with the length-to-with thickness ratio falling between 5 and 30 microns. Severe sliding wear particles sometimes show evidence of temper colors, which may change the appearance of the particle after heat treatment.

Severe Sliding Wear

5. Bearing Wear Particle: These distinct particle types have been associated with rolling bearing fatigue:

·         Fatigue Spall Particles constitute actual removal from the metal surface when a pit or a crack is propagated. These particles reach a maximum size of 100 microns during the microspalling process. Fatigue Spalls are generally are flat with a major dimensions-to-thickness ratio of 10 to 1. They have a smooth surface and a random, irregularly shape circumference.

·         Laminar Particles are very thin free metal particles with frequent occurrence of holes. They range between 20 and 50 microns in major dimension with a thickness ratio of 30:1. These particles are formed by the passage of a wear particle through a rolling contact. Laminar particles may be generated throughout the life of a bearing, but at the onset of fatigue spalling, the quantity generated increases. An increasing quantity of laminar particles in addition to spherical wear is indicative of rolling-bearing fatigue microcracks.

6. Gear Wear Two types of wear have been associated with gear wear:

·         Pitch Line Fatigue Particles from a gear pitch line have much in common with rolling-element bearing fatigue particles. They generally have a smooth surface and are frequently irregularly shaped. Depending on the gear design, the particles usually have a major dimension-to-thickness ratio between 4:1 and 10:1. The chunkier particle result from tensile stresses on the gear surface causing the fatigue cracks to propagate deeper into the gear tooth prior to spalling.

·         Scuffing or Scoring Particles is caused by too high a load and/or speed. The particles tend to have a rough surface and jagged circumference. Even small particles may be discerned from rubbing wear by these characteristics. Some of the large particles have striations on their surface indicating a sliding contact. Because of the thermal nature of scuffing, quantities of oxide are usually present and some of the particles may show evidence of partial oxidation, that is, tan or blue temper colors.

Many other particle types are also present and generally describe particle morphology or origin such as chunk, black oxide, red oxide, corrosive, etc. In addition to ferrous and non-ferrous, contaminant particles can also be present and may include:  Sand and Dirt, Fibers, Friction polymers, and Contaminant spheres. 

Contaminant particles are generally considered the single most significant cause of abnormal component wear. The wear initiated by contaminants generally induces the formation of larger particles, with the formation rate being dependent on the filtration efficiency of the system. In fact, once a particle is generated and moves with the lubricant, it is technically a contaminant. 

Passport System V Software and Instruments

The combination and enhancement of WPA and UOA within the past few years have been oriented towards managing a predictive maintenance program efficiently with the advent of software and high tech tools. Of the recent development of our Passport System V software and instrumentation allows the user to incorporate all different types of predictive maintenance tools with a customized approach. The Passport System V is sophisticated, yet simple to use, state of the art data management and report writing tool, which provides users the most advanced capability available for computerized storage, comparison data, and evaluation of lube and wear data. The software design makes the creation of tables, charts, digitized pictures, drawings, and qualitative reports, previously produced manually, a faster and easier task, with more accurate results.

The Passport System V incorporates a video camera to capture and transmit the particle image magnified on the microscope to a personal computer. The data management features enable the technician to rapidly prepare a report and compare the current machine condition with the previous analysis history. The computer screen act as a regular display for report writing or reviewing information, while another part of the screen high resolution images are display from the microscope or from earlier reports, or pictures from the Wear Particle Atlas. With these combined features and having predictive maintenance information at your fingertips allows the technician to provide a comprehensive report with quality condition monitoring recommendations. 

Passport System V

The Passport System V is a significant enhancement to predictive maintenance monitoring. The data management capability is obviously valuable to industry today. The technology to capture, digitize, transfer, and store images has progressed rapidly such that the quality is now equal or even better than still photography. The system enables the user to establish their own predictive maintenance reference guides for their machinery. The Modem/Internet e-mail system enables the technician to communicate with distant sister facilities' sites and with other technicians including our own experts. Perhaps most exciting of all is the prospect of developing an artificial intelligence base for diagnosis and decision making.

Case History

Safety Components Fabric Technologies, Inc. is a major worldwide producer of material used in the manufacture of air bags for passenger vehicles.  Safety Components Fabric Technologies, Inc. instituted a predictive maintenance program in February 1998.  Their program consisted of vibration analysis and Ferrographic Wear Particle Analysis.  Included in their program were 205 weaving machines (the focus of this paper), air compressors, chillers, motors, pumps, and fans.  Vibrational Analysis is performed on a 90-day cycle on all units.  All units found to be outside of specified Vibrational limits are sampled immediately and sent to Predict/DLI for Wear Particle Analysis.  The Wear Particle Analysis consists of Direct Reading Ferrography and Analytical Ferrography.  Initially, the standard gearcase program was used to detect premature failures in these units, but was found to be inadequate, as the machines would fail long before expected.  With the help of the customer and Predicts Ferrographic expertise, an innovative plan was developed that best incorporated the unit’s design, sampling oddities, and the condition monitoring tools employed.  This plan, or Modified Program, allowed for accurate detection of premature gearcase failures in these machines long before any unexpected downtimes could occur.

Safety Components Fabric Technologies, Inc. employs dual rapier-weaving machines to weave yarn into cloth.  A rapier weaving machine works as such:  The warp (lengthwise) threads are secured on the loom through the heddle eyes (a thread, wire, metal or Texsolv polyester loop held by the shaft sticks with eyes for threading the warp ends[i]) and attached to the loom beam located in the rear of the loom.  The dual rapiers are fingerlike arms that carry filling yarn halfway through a shed of warp yarns (a shed is a separation of the warp ends into an upper and lower system of threads that permit the rapiers to pass through the space that has been formed).  The filling (crosswise) thread is placed by the two-rapier system between the warp thread shed.  The rapiers are located directly across from each other on each side of the loom [see figure 1]. 

The rapiers meet one another halfway through the shed and the filling yarn carried by the left hand rapier is transferred to the right hand rapier and is carried the rest of the way across the loom where it is cut and the process is repeated at a constant rate[ii].  As the warp ends are drawn through the heddles, via the rotation of the loom’s lower drive gearcase, the shed is formed with each turn and the fabric is woven concurrently as the rapiers add the fill yarn inside the shed[iii].  The loom in general and the rapiers are driven by identical transmission gearcases on each side of the loom. 

The weaving machines are driven by a continuously running AC motor over a magnetic clutch-brake assembly.  This motor is connected to a drive shaft by three V-belts.  Gearcase speed depends upon the size pulley on the motor shaft.  This drive shaft transmits power to the loom’s left transmission gearcase.  The left hand transmission gearcase [see figure 2] then powers a shaft that is connected to the right hand gearcase and drives it in unison. 

Each of the right and left hand transmission gearcases contains a lower and an upper drive gear assembly.  The lower drive gearing primarily contains a pinion gear and a bull gear.  The pinion gear [arrow F2g], which is attached to the drive shaft [arrow F2e], transmits power to the left hand gearcase’s lower bull gear [arrow F2d], which drives the right hand gearcase, the loom, and the double cam follower.  The double cam follower [arrow F2a] transmits the power generated by both of the lower drive gearcase’s bull gears to each of the upper drive gear assemblies via the toothed segment gear [arrow F2c].  The double cam is a vital component of this unit because it transmits the power very evenly and smoothly to the upper gearcase and thereby insures that the filling is inserted gently by the rapiers.  The gearing and cams in the loom are expertly engineered and synchronized with the connecting shafts so that all componentry moves in a specific sequence of motions and the fabric is woven precisely and efficiently. 

The upper drive gearcase [arrow F2b and figure 3] powers each of the rapiers.  Based on the high rapier speeds and loading, the major stress loading in this unit is found in the upper drive gear assembly [see figure 3].  Based on that fact, the upper drive gear

assemblies are the most susceptible to abnormal wearing in both the cylindrical antifriction bearings and gearing. 

The spur gear, located on the upper drive gear shaft, is driven by the toothed segment gear.  Looking at Figure 2, the upper drive gear has been removed from its original position.  When in operation, the upper drive gear assembly is located in the opening above the toothed segment gear [Figures 2 and 3: arrows F2c and F3d] where the spur gears are intermeshed.  This spur gearing [arrow F3c] controls how far the rapier arm swivels back and forth.  The spiral bevel bull and pinion gear set [arrows F3b and F3e] transmits the power generated by the upper drive gear assembly shaft spur gear to the final gear that maneuvers the rapier arm back and forth in a rack and pinion type assembly.  The upper drive gear assembly employs single and double cylindrical roller antifriction bearings to support loading on all applicable shafts. 

The oil reservoir [Figure 2: arrow F2f] for both gearcases is located at the bottom of the gearcase, partially immersing the lower drive bull gear.  The reservoir volume is 2 gallons.  The lubricant used in this unit is ISO 150 grade EP gear oil.  An electronically controlled pumping system applies and recirculates (through a fine filter) the lubricant to the upper gearing and bearings.  After lubricating the upper portion of the gearbox, the oil cascades down to lubricate all remaining componentry.  In addition, the bull gear is further protected from potential abnormal wearing because it is partially immersed in the lubricant and the oil clings to the teeth as it rotates. 

Actual Case History of a “Standard Gearcase” Failure 

In October of 2000, an extruder reduction gearcase was determined to be undergoing a major to catastrophic wear mode via Ferrographic Wear Particle Analysis and was rated CRITICAL.  This case history includes all six sample points taken during the monitored history of this unit and comments about how the gearcase went from a normal wear mode to a catastrophic wear mode in a six-month period. 

As the Direct Reading (DR) Ferrography graph [figure 14] indicates, the gearcase, from April to August (five samples), was operating "Within Limits" based on the Wear Particle Concentration. 

In other words, the samples in the selected timeframe were within the numerical limits of the Mean of all sample points plus or minus two Standard Deviation units.  On all five samples, Analytical Ferrography observations indicated only normal rubbing wear on the ferrogram.  These five samples, from April to August, were rated NORMAL based on the Wear Particle Concentration and the Analytical Ferrography results. 

In contrast, the October sample’s DR result was very high.  Looking at the graph, October’s DR Ferrography testing result of 537 was virtually ten times higher than the Alarm BH point (54) and 500 points higher than when previously sampled in August.  The Alarm BH point on the graph denotes the mean plus two standard deviation points.  Every wear particle concentration value above the established BH Alarm point is considered "Out of Limits".  This result was of great concern.  The next step was to perform Analytical Ferrography.  Analytical Ferrography indicated large amounts and sizes of case hardened steel, low alloy steel, and medium alloy steel abnormal gear and bearing wear particles up to 120 microns in size [figure 15]. 

When comparing October’s Analytical Ferrographic results to previously trended results (where no abnormal wear particles had ever been detected), it was confirmed that this unit was undergoing a major to catastrophic abnormal wear mode.  As a result, based on the combination of the very high Wear Particle Concentration along with the Analytical Ferrography results, this sample was rated CRITICAL and the customer was notified immediately.  The customer inspected the unit and determined that the unit had undergone abnormal gear and bearing wear and the unit was overhauled. 

This case history for “standard gearcase” failure is an excellent demonstration of how a typical gearcase can move from a normal wear mode to a catastrophic wear mode over a period of time.  This case history indicated a definite point (the October sampling) where the unit in question went from a normal wear mode to a critical wear mode based on both DR testing and Analytical Ferrography.  An inspection of the gearcase confirmed what Analytical Ferrography had predicted.  The customer planned for the downtime and the unit was repaired.  Ferrographic Analysis assists the customer in eliminating unplanned downtimes.  Unplanned downtimes are very expensive and detrimental based on the loss or reduction of production and excess man-hours expended to correct the problem.  Not all gearcases undergo such dramatic changes in wear modes (normal to critical) in such a short period.  It is also typical for a gearcase to alternate between a normal to marginal wear mode and vice versa over a long period.  This rating alternation is due to unit loading during a specific period, speed of operation, oil changes, etc.  However, if Ferrographic Analysis indicates wear being generated far in excess of what trending has shown to be as typical (as recorded by previous normal and marginal ratings), that unit will be rated critical and the appropriate steps will be taken to assure that the gearcase is scheduled for appropriate maintenance actions. 

Safety Components Fabric Technologies, Inc. Typical Gearcase Failure 

Safety Components Fabric Technologies, Inc. weaving machine gearcase failures do not conform to the “standard” case history gearcase failure described in the previous section.  Looking at the Wear Particle Trending graph for a typical weaving gearcase, it is unapparent that this unit is in danger of imminent failure.  From May to October, three lubricant samples were sampled from this gearcase. 

The Wear Particle Concentration (WPC) did not vary to any great degree with each sampling.  The results were 4.6 in April, 9.1 in May, and 11.9 in October (all in 1999).  Analytical Ferrography results indicated normal rubbing wear in the April and May samples.  The October sample indicated a small amount of gear and bearing wear particles up to 120 microns in size [figure 17].  Originally, the October sample was rated marginal based on the Analytical Ferrographic results only; the DR results were well within what is expected to be normal for a “standard” gearcase.   

The Analytical Ferrographic results for the October sample were rated marginal based on the small amounts of abnormal wear particles.  In a “standard” gearcase, the observed small amounts of abnormal wear particles and the relatively low WPC would typically constitute a minor overall abnormal wear mode.  As stated previously in the description of equipment condition ratings, assumptions were made that similar equipment would be rated marginal.  A further illustration of the differing amounts of abnormal wear particles is to observe figures 15 and 17.  Figure 15 illustrates a large amount of abnormal wear particles and an obvious high wear mode; so much so that the magnetic flux lines are piled up on one another and are individually indistinguishable.  Figure 17, on the other hand, illustrates a small amount of abnormal wear particles along with a small to moderate amount of normal rubbing wear in clearly distinguishable magnetic flux lines.

Figure 17: low and high alloy steel gear and bearing wear particles (120mm max.) 200X

In reality however, it was found that this unit, along with several others that were rated similarly, should have been rated critical because the upper drive gear assemblies in these gearcases were undergoing a

high to catastrophic wear mode and were failing unexpectantly.  These gearcases obviously did not conform to standards set for units that were assumed similar.  Based on this situation, it was unclear whether Ferrographic Wear Particle Analysis could be employed to accurately predict premature failure in these gearcases.   

Vibrational Analysis was shown to be successful in identifying gearcases that were undergoing some form of an abnormal wear mode.  Many of the weaving machine gearcases, found to be outside of typical predetermined vibrational limits, were undergoing an abnormal wear mode.  However, the disadvantage of Vibrational Analysis was that it was not specific in determining what degree of wear was ongoing in each of these units.  Not all units found to be outside of vibrational limits were undergoing a catastrophic wear mode.  In order for Ferrographic Wear Particle Analysis to be successful in this application, it would have to be capable of differentiating the wear mode ongoing in each unit (if any) where Vibrational Analysis was unable to distinguish.  It would be very valuable to find a complementary relationship between Vibrational Analysis and Ferrographic Analysis.  Ideally, if this relationship could be employed, Vibrational Analysis could flag the weaving machine gearcases that were potentially undergoing an abnormal wear mode, a sample of the lubricant could be pulled from each of the flagged unit reservoir(s), and Ferrographic Analysis in turn would determine the severity of the wear mode in each weaving unit.  Obviously, the missing link in this relationship was the Ferrographic Analysis.  Steps had to be taken to assure that Ferrographic Wear Particle Analysis could accurately predict abnormal wearing in these units.  Once it was proven that Ferrography could be utilized, the method would have to be customized and developed specifically to determine the severity of the wear mode ongoing in each of the gearcases.  This new method would have to be consistent and accurate in the determination of any ongoing wear modes sent in for analysis.   

GEARCASE INVESIGATION, FAILURE ANALYSIS, and ASSESSMENT of FINDINGS 

Patrick Kilbane, a Predict Machine Condition Analyst, was sent to Safety Components to assess why the gearcases had failed prematurely.  First, a failure analysis was performed on the gearcase to estimate how much metal is actually being worn off the internal componentry that led to a premature failure.  If no more than a small amount of wear is generated when the gearcase fails, it will be difficult for Ferrographic Analysis to accurately predict a catastrophic wear mode.  If the failed gearcase is found to generate a large amount of abnormal wear, another avenue must be investigated to explain the ferrographic anomaly.  The investigative team would then have to consider anything relatively unusual in the makeup or sampling intricacies in the gearcases that would explain why they do not conform to standard ferrographic analysis methods.  After all investigative and failure analysis information was completed, it would be compiled.  This compiled information would help the investigative staff determine whether Ferrographic Analysis could be employed and customized to effectively, consistently, and accurately predict what type and degree of a wear mode is ongoing in this unit. 

Failure Analysis: Sixteen failed weaving machine gearcases were opened and inspected.  Very little to no abnormal wearing was found in the lower drive gearing [Figure 2: arrows F2a, F2d-F2g]. 

Conversely, when the toothed segment gear and the upper drive gear assembly [Figure 2: arrows F2b-F2c, Figure 3: all arrows] were inspected, a large amount of abnormal wear was found on most to all componentry.  This is clearly illustrated in figures 18 to 20.  Figure 18 is an image of a severely worn upper drive gear assembly shaft cylindrical roller antifriction bearing.  As observed in the image, a large amount of fatigue spalling was discovered on both of the races and all of the rollers.  All failed gearcases showed this type of wear on every one of the upper drive gear assembly shaft cylindrical roller bearings.  Figure 19 is an image of a severely worn upper drive gear assembly shaft spur gear from a failed weaving unit gearcase.  As the image illustrates, a large amount of pitch line pitting and spalling along with scuffing and scoring was discovered on many of the gear teeth.  All failed gearcases demonstrated some degree of this type of wear on this specific gear.  Figure 20 is an image of a spur gear and shaft found in the upper drive gear assembly pinion gear set.  This gear is also from a failed weaving unit gearcase.  As the image illustrates, a large amount of pitch line pitting and spalling along with scuffing and scoring was discovered on many of the gear teeth.  Many of the failed gearcases demonstrated some degree of this type of wear on the pinion gear set. 

The investigators determined in the failure analysis of sixteen (16) upper drive gear assemblies that the units failed in the same manner.  Every upper drive gear assembly shaft cylindrical roller bearing [figure 18] was severely worn: more so than any of the other gearbox componentry.  Based on that fact and operation history on the failed unit, it was determined that the upper drive gear assembly shaft cylindrical roller bearing would loosen and misalign under very high loads and speeds, initiating an abnormal wear mode in the gearbox.  As the bearing loosened further, the upper drive pinion gear also became misaligned.  The catastrophic wear mode commenced in these units when the misaligned shaft gearing began to wear abnormally. 

As described and illustrated, a large amount of abnormal wearing was discovered in the upper drive gear assembly on every failed gearcase.  Based on that fact, quite a lot of abnormal wear would be present in the lubricant.  Ferrographic Analysis should theoretically be capable of differentiating the degree of abnormal wear ongoing in the weaving unit gearcase.  Because there was so much abnormal wear debris generated in a failure mode and that standard ferrographic methods were not identifying it, there obviously was an alternate reason why Ferrographic Analysis was not accurately diagnosing the problem.  The weaving machine gearcases were further investigated.

Investigation of the weaving machine gearcase: Ferrographic Analysis is dependent on several factors:

One common gearcase factor that may affect Analytical Ferrography results is sampling location.  Ideal sample points can be found in several spots in a gearcase.  The best point is found at the lubricant return line right after the oil has lubricated the gearing and bearings.  This sample is well mixed, uniform, and representative of the overall lubricant circulating in the system.  If it is impossible or impractical to take a sample at that point, the next best sampling point can be found one to two inches deep in the reservoir very close to lubricant return line.  This sample is well mixed and uniform, but care has to be taken to assure the sample is taken in the same place every time to assure consistent trending results.  It is also advisable that the same person takes the sample each time.  If it is impossible to sample at either of those sampling points due to unit design, the sample should be taken at the most favorable location by the same person, in the exact same sample position, and utilizing the same sampling technique every sampling period.  That way the ferrographic trending results are consistent from sample to sample. 

Another common gearcase factor that may affect Analytical Ferrography results is the return flow of the lubricant.  It should be confirmed that the returning lubricant flow is completely homogenous and well mixed once it returns to the sump.  If the sample is not well mixed and uniform, the amount of wear particles in the lubricant will be diluted and any sample taken and sent in for Ferrographic Analysis will not be representative of the ongoing wear mode in the unit.  The most representative samples are ones that return in whole to the reservoir via a return line.  Some of the least uniform and representative lubricant samples are found in units where the oil is sprayed over a large surface area and is allowed to fall over the length of the sump via gravity.  Samples taken from this type of system may be taken in an area that is wear particle lean or rich compared with the mean amounts of particles generated by the unit.  Ferrographic Analysis of these types of samples has a lower probability of accurately identifying the ongoing wear mode. 

A final common gearcase factor that may affect Analytical Ferrography results is the effect of differing loading and speeds on each individual gearcase.  Each weaving unit runs at differing speeds.  In addition, woven fabric size and yarn type creates differing loading on a unit.  As an example, a heavy rope type thread is much heavier and more difficult to weave into cloth.  This gearcase is powering the weaving process under a great deal more loading than an identical unit that is weaving lighter weight thread into cloth.  Therefore, the loading and speeds should also be investigated on all failed gearcases looking for common failure modes. 

Therefore, the next logical step in this investigation was to determine lubricant sampling locations, return flow, and/or unit loading and speed anomalies that would explain the small amounts of wear debris that represented a catastrophic wear mode observed by means of Ferrographic Analysis.  The entire lubrication system and reservoir were thoroughly investigated.  Varying degrees of all three factors were discovered. 

As stated above, a common contributing factor that typically affects Analytical Ferrography results is the sample point and sampling techniques.  If the sample is taken in the incorrect location or in an incorrect manner, the Ferrographic results are also typically incorrect.  In the weaving unit’s gearcase, the only location available for sampling was found at the drain cap, which is located approximately one inch above the bottom of the reservoir at the front end.  Due to the gearcase design, there were no alternative sampling points.  The same operator took the samples at the exact same location and utilized the same sampling technique at all times.  This ensures, even though the sample point is less than ideal, that the Analytical Ferrography results will be consistent from sample date to sample date.  In other words, the precision of all samples is excellent while the accuracy may be suspect based on how representative the lubricant is of the wear mode ongoing in the unit (return flow).  Therefore, the sample point location and sampling technique were the best that were practical for this application.  The investigation revealed that sampling and sampling techniques were not likely to be a major factor inhibiting accurate Analytical Ferrography testing results. 

The failed gearcase loading and speeds were investigated and compared to units that were operating within limits.  It was discovered that a large portion of failed gearcases were under high loads and/or speeds at some time in their history, but correlations were not always as would be expected.  Some units under lower loads and/or speeds would exceed Vibrational limits and begin to fail while other units that were under higher loading and/or speeds remained within Vibrational limits and were not sent in for Analytical Ferrography testing.  It was apparent that other enigmatic factors were affecting these gearcases (such as a slight misalignment) and not others.  The investigation revealed that excessive loading and/or speeds was a factor in gearcase failure and should indicate a higher amount of abnormal wear particles via Ferrographic Analysis.  However, this did not correlate with every unit and was not readily apparent in either Vibrational or Ferrographic Analysis.  Therefore, excessive loading and/or speeds could not be easily utilized to aid in the early detection of a gearcase problem due to potential concealed and enigmatic factors ongoing in a gearcase. 

The return flow of the lubricant was then investigated.  The oil is pumped to the upper drive gear assembly, where it is sprayed onto the gears and bearings.  The lubricant returns over the entire length of the sump by gravity.  In analyzing the sampling techniques of these weaving machines, it was found that any sample taken would not be completely homogenous and representative of the wear mode.  The sump, which is long, narrow, and shallow, acts to disperse wear particles generated by the machine because the returning lubricant does not drain into a single point in the sump via a return line.  Rather, the return flow cascades over the length of the sump.  It would be expected that the amounts of wear particles would be much smaller than expected.  Therefore, the investigation revealed this to be the crucial factor that explained why so few abnormal wear particles were being observed via Ferrographic Analysis.  In fact, the failing gearcase was generating a large amount of abnormal wear particles.

The returning lubricant flow was found to be the primary reason that the application of specific alarm limits was needed on these weaving machine gearcases.  To a lesser extent, the sample point was also a contributing factor because of its less than ideal location.  However, because the samples were taken by the same operator in a consistent location and utilizing the same sampling techniques, the trending results would at least be consistent.  The limitations on sampling locations were unavoidable; the weaving units could not be redesigned.  The typical alarm limits for standard gearboxes would not apply in these weaving machine gearcases due to these factors.  In any ongoing wear mode, the amounts of abnormal wear particles and/or the wear particle concentration would be much lower than expected for gearcases in general.

The gearcase failure analysis and further investigation yielded an understanding of the intricacies inherent in these gearcases.  These units would generate far less abnormal wear particles than typically observed in an average gearcase.  Analytical Ferrography could be employed on these units under the proper specifications, but a new method needed to be developed to compensate for the differences between this unit and a typical gearcase.  A new and customized method would now be developed specifically to determine the severity of the wear mode (if any) ongoing in each of the gearcases.

MODIFIED FERROGRAPHIC WEAR PARTICLE ANALYSIS METHOD

The first step in developing the customized Analytical Ferrography method was to try to determine when an abnormal wear mode begins in these units.  The Analytical Ferrography data from the sixteen failed gearcases was taken.  The wear particle concentration and size of particles was plotted and compared to Vibrational analysis data and failure analysis data.  After reviewing this comparison, it became obvious that when the Wear Particle Concentration (WPC) rose above twenty (20) and/or if any abnormal wear particles over 15 microns in size were observed via Ferrographic analysis, that Vibrational readings and pre-failure analysis indicated a problem in the gearcase.  This finding determined the point where these weaving machine gearcases entered into an abnormal wear mode.

The second step in developing the customized Analytical Ferrography method was to attempt to decipher the point when the gearcase enters into a catastrophic wear mode.  It was determined that when the abnormal wear particles reached sizes of 70 microns or higher, failure was imminent.  This value was discovered after comparing the failure inspection and further investigation results with the Analytical Ferrography abnormal wear particle sizes.  Because the amounts of abnormal wear particles were always going to be small, the abnormal wear particle sizes were the most important factor in determining the severity of the ongoing wear mode.  This was also based on comparing the Analytical Ferrography results with the failure analysis and the further investigation results.

The final step in developing the customized Analytical Ferrography method was to set Analytical Ferrography specifications so that the weaving machine gearcase can be accurately rated.  These specifications were set according to the two steps listed above.  The ratings are listed below:

The unit was rated NORMAL if the DR Ferrography results were less than 20 and the Analytical Ferrography results indicated only normal rubbing wear (particles less than 15 microns in size).

The unit was rated MARGINAL if the DR Ferrography results were greater than 20 and/or the Analytical Ferrography results indicated abnormal wear particles (regardless of type) in the range of 15-65 microns in size.

The unit was rated CRITICAL if the Analytical Ferrography results indicated abnormal wear particles (regardless of type) equal or greater than 70 microns in size.

Since the implementation of these specifications, they have been shown to be very accurate in determining the severity of a wear mode in these weaving machine gearcases.  In addition, there have not been any unplanned downtimes due to gearcase failure since the specifications were set.  Predict has made timely predictions of three known premature failures since implementation.  These predictions saved Safety Components fabric Technologies, Inc. the aggravation of unplanned downtimes along with the additional costs of parts and labor.

Aircraft Gas Turbines

Aircraft and aircraft-derivative jet engines are subject to various failure mechanisms. Some of these failure modes proceeded very rapidly, whereas others can be detected hundreds of operating hours before a shutdown condition is reached. Most failures of gas turbines occur in gas path. Gas-path failures frequently, but not always, cause an increase in wear particle size and concentration in the oil system, probably due to the transmittal of imbalance forces to turbine bearings and other oil-wetted parts. The resulting bearing or gear wear is then detected by both Used Oil Analysis and Wear Particle analysis.

Determining the exact source of wear problem can be difficult in a gas turbine because of complexity of the oil-wetted path. Typically several cavities, housing bearings, or gears will be force lubricated through individual return lines connected to a tank from which the oil is pumped (at a high rate), then pass through a filter and heat exchanger, and the cycle repeated. Magnetic chip detectors or magnetic plugs are often installed in the return lines from various engine parts. These can help to pinpoint the source of generation in cases where particle metallurgy, as determined by heat-treating ferrograms, is similar for various engine parts. However, chip detectors will not give a warning until the wear situation is so severe that extremely large particles are being generated. By this time, the opportunity for predictive maintenance may be lost. Other analytical techniques, such as vibration analysis, may help to pinpoint the part in distress utilizing expert system software that provides recommendations for action. In any case, predictive maintenance tools integrated together offer the maintenance engineer the best decision making tool.

Conclusion

The benefit of automation is in the use computer programs and emerging software technologies of artificial intelligence to assist in determining when to remove equipment from service for maintenance. These case histories provide a real world scenario that indicates it’s not that easy to put artificial intelligence to make maintenance decisions. However, this does not mean we do not try. For example, an advanced system, which integrates emerging technologies in vibration, motor current analysis, Thermography, ultrasonic, electronics, microprocessing, graphics, and data management, could regularly sample a number of machines. From a sampling device, compare the samples to previous samples for trend information (along with other Data parameters), make the decision to schedule the machine for maintenance, generate a work order for the maintenance team and send a purchase/work order to accounting for needed repair parts.

The maintenance manager/engineer could have almost instantaneous reports on the condition of each machine, along with a dollar figure indicating the optimal dates for shutdown and other maintenance requirements, basically, a financial decision.

Technology advances oriented toward maintaining and incorporating all production data serve as an efficient assessment of manufacturing equipment. Companies as we know it today can ill afford any shutdowns what so ever due to a tremendous amount of re-engineering or downsizing occurring worldwide. Therefore, predictive maintenance tools working in conjunction with production efficiency, analyzed through a cash flow model are the decisions making tools of today and tomorrow. 

Acknowledgments 

Wear particle analysis and Used Oil analysis information were extracted from the wear particle atlas and extensive experience of Predict employees. Other contributors to the preparation of this technical paper were Rob Lovicz, Mike Cannon, Pat Kilbane, Carolyn Martovitz, Dr. Rod Bowen, Vernon Westcott, and Bill Hoskins.

Contact Robert Lovicz or Raymond Dalley, Predict, 9555 Rockside Road #350, Cleveland, OH 44125; (216) 642-3223, or e-mail rjdalley@predictusa.com