THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER Finding Bearing Failure Wally Lastinger, Naval Aviation Depot, Jacksonville, FL Richard Overman and Lynwood Yates, Wyle Laboratories, Orange Park, FL and the RIAC Bearings give many different warnings when they are failing. Some of these warnings include increased vibrations, noise, and the loss of bear- ing material. Each warning can be detected by different techniques. Vibration can be detected by vibration analysis, noise by listening, and bear- ingmateriallossbyFilterDebrisAnalysis(FDA). The vibration analysis is commonly used and well known,listeningfornoiseisoftentoolate,FDAis notcommonlyknownandisapowerfultool. This article reports on an analysis of a jet engine bear- ing and the use of FDA to detect impending fail- ures. ThepaperdescribestheFDAprocess,bear- ing failure stages, and how all of this information was pulled together through Reliability-Centered Maintenance analysis to develop a cohesive func- tion preservation strategy. Bearingsareusedinmanyapplicationsfromsmall toys to jet engines with varying degrees of failure consequences that range from a nuisance to a catastrophe. Bearing failures in a jet engine can often result in catastrophe. This was the case for the bearing, which was the subject of this analy- sis. The bearing had failed causing the engine to come apart and the aircraft to crash. Obviously, efforts were needed to prevent future failures. A Reliability-Centered Maintenance (RCM) analy- sis was performed to identify the appropriate ac- tions to predict bearing failure. RCM identifies policies that must be implement- ed to preserve the function of an asset in its oper- ating context. RCM considers scheduled main- tenance, design changes, training improvements, operationalchanges,run-to-failure,andotherone- time changes as function preservation strategies. Eachoptionisevaluatedforitstechnicalfeasibility and its impact on safety, environment, operations, and cost. The option that meets safety and envi- ronmental requirements with the least impact on operations and cost is selected and implemented. Each of these options are explained more thor- oughly later in this article. With RCM, scheduled maintenance is based on the failure characteristics of the component in its operating context. Scheduled maintenance takes many forms that include visual inspections, sophisticated inspection techniques such as vibra- tion analysis and thermal analysis, scheduled over- hauls, and others. RCM provides the analytical technology to get the most out of these tools. By ensuring that the equipment's operating context drives scheduled maintenance, the right mainte- nancecanbeperformedattheright,mostefficient time. RCM recognizes that "fix when fail" or "run-to- failure" is a valid option when it is cost effective and there is no potential impact on safety or the environment. Precious maintenance resources are often used to maintain primary systems when back-upsystemsareavailableandcapableofmain- taining safe operations if the primary system fails. Depending on the failure modes, much of this maintenance may be unnecessary. One-time-change considerations are valuable con- siderationsintheRCMprocess. Throughoutthe process, operations and maintenance personnel are working together to ensure that the use and maintenance of the system is proper and correct within the operating context. Oftentimes, the analysis process clears up misunderstandings by or between operations and maintenance that lead to more efficient system use. Experience has shown that these increased efficiencies can pay for the entire analysis. Hence, RCM provides a proactive approach to maintenance planning. This proactive approach isdesignedtoensurethattherightpeopleperform therightmaintenanceattherighttimeintheright way with the right training and equipment to gain the maximum benefit of the system or process. This bearing analysis focused on the various con- dition monitoring (CM) techniques that may be useful in detecting bearing degradation. In gen- eral CM techniques are technically feasible when certain criteria have been met. The Society of Automotive Engineers (SAE) RCM standard provides the following criteria for the technical feasibility of a CM task . 1. "There shall exist a clearly defined potential fail- ure" (point B on Figure 1). 2. "There shall exist an identifiable P-F interval" (P-F stands for "potential to functional failure" in- terval and is the same as the degradation interval ABSTRACT INTRODUCTION RELIABILITY-CENTERED MAINTENANCE APPLICATION OF RCM TO THIS BEARINg ANALYSIS FIRST QUARTER - 2006 Through Filter Debris Analysis on Figure 1, interval from point B to point C). 3. "The task interval shall be less than the shortest likely P-F interval" (inspection interval < degrada- tion interval). tional failure (the degradation interval minus the task interval) shall be long enough for predeter- mined action to be taken to avoid, eliminate, or minimize the consequences of the failure mode." This technically feasible criterion was applied to all CM technologies that might be used to detect a potential bearing failure. An inspection interval is identified for each one. The inspection inter- val may be different because different CM tech- nologies can detect the onset of failure at different places along the degradation curve. This will be discussed in more detail later in the paper. As noted above, the first things that have to be de- finedarethepotentialfailureandfunctionalfailure conditions. After the initial aircraft crash, a con- servative Filter Debris Analysis criteria and other actions were taken to try to identify distressed or failed bearings in the entire fleet. Engines with suspect bearings were removed and the bearings replaced. The replaced bearings were used in this analysis to refine the inspection criteria. Figure 1 - Criteria for Technical Feasibility of a CM Task continued on page 10 Figure 1 Inspection Interval 0 % 100 % FAILURE RESISTANCE 10 % FUNCTIONAL FAILURE POTENTIAL FAILURE WHAT CHARACTERISTIC WILL INDICATE REDUCED FAILURE RESISTANCE? B A C DEFINED POTENTIAL FAILURE CONDITION TASK INTERVAL FEASIBLE? DEGRADATION INTERVAL Time DEFINED FUNCTIONAL FAILURE CONDITION STAgES OF FAILURE 4. "It shall be physically possible to do the task at intervals less than the P-F interval." 5. "The shortest time between the discovery of a potential failure and the occurrence of the func- ability predictions overcomes many of the limita- tions of other current approaches because: It is based on the electronic failure rate data contained in the RIAC (previously RAC) data- bases as of September 2005, believed to be the largest in the world It accounts for "system level effects"on assem- blies of all levels It includes the effects of failures due to software It allows the effects of the detailed environmen- tal conditions to be taken into effect, rather than using gross "environmental K-factors" It includes non-operating and duty cycling effects It covers all major part classifications included in other comprehensive methodologies It includes a handbook of the failure rate models used in the software, promoting an understanding of the development and use of the methodology PAGE What is 217Plus 217PlusTM is the latest reliability prediction methodology available from the DoD-funded Reliability Information Analysis Center. The approach highlights the RIAC's commitment to deliver a comprehensive replacement methodol- ogy to the outdated MIL-HDBK-217 "Reliabil- ity Prediction of Electronic Equipment." While previous products of the former Reliability Analysis Center (RAC) only addressed about half of the part types covered by the MIL-HDBK, 217PlusTM now covers all major part types. The 217PlusTM methodology for performing reli- TM THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER Finding Bearing Failure Through Filter Debris Analysis continued from page 9 Figure 4 - Stage 3 Bearing Figure 5 - Stage 4 Bearing Bearingswerereceivedwithpartnumbersandas- sociated engine numbers identified. These bear- ings were visually checked for wear/damage and categorizedasStage1to5accordingtothefollow- ing conditions: Stage 1: Bearing begins to skid and wear on roll- ers (Figure 2) Stage 2: Bearing roller wear becomes noticeable or silver plowing occurs (Figure 3) Stage 3: Cage begins to crack or is cracked through (Figure 4) Stage 4: Severe cage wear (Figure 5) Stage 5: Bearing "self destructs" Forthepurposesofthisanalysis,itwasdetermined thatacrackedbearingcagewouldbeconsidereda failedbearing. Therefore,whenabearingreached Stage 3 it was considered to be functionally failed. The techniques available to detect various stages of this bearing failure were the Joint Oil Analysis Program (JOAP), Filter Debris Analysis (FDA), Chip Collector inspection and Freedom of Rotation check. The relative ability of each tech- nique to detect a potential failure condition is dis- cussed below and illustrated in Figure 6 . JOAP: a. Oil sample taken every 10 operating hours and analyzed for metal content b. Should be able to detect particle sizes ranging from dissolved metal to 10 microns c. Has the ability to detect bearing degradation, but is dependent on how often the filters are replaced. Filter Debris Analysis: a. Oil filter analyzed for metal content of the oil every 50 operating hours b. Should be able to detect particle size of 20 mi- crons and above c. Has demonstrated ability to detect bearing degradation in all stages d. Detection limits discussed later Chip Collector: a. Collector is removed from the oil sump and vi- sually inspected for metal particles b. Should be able to detect Stage 3 or 4 failures c. Newly installed with no track record d. May be used as a confirmation tool Freedom of Rotation: a. Maintenance technician rotates the turbine section of the engine and evaluates freedom of movement b. Should be able to detect late Stage 3 and Stage 4 failures c. Has found degraded bearings on a limited basis Gary Humphry proposed the use of Filter Debris Analysis for detecting degraded bearings. Papers with details of the process have been published in the proceedings of the JointOilAnalysisProgram International Conference, Mobile, AL. April, 2002. For this analysis, oil filters were removed from the aircraft periodically and sent to the labo- ratory for debris analysis. This analysis identifies the metals in the oil filter by percentage and pro- vides the total mass of the sample. The mass of each metal was determined by multiplying its per- centage times the sample mass. Element percent- age and mass was then tracked. The metals of interest in this bearing were Iron (Fe), Molybdenum (Mo), Vanadium (V), Silver (Ag) and Chromium (Cr). Chromium and Iron were not reliable markers because other compo- nents that the oil contacts contained these ele- Figure 6 - Detection Techniques a. Maintenance technician rotates the turbine section of the engine and evaluates freedom of movement b. Should be able to detect late stage 3 and stage 4 failures c. Has found degraded bearings on a limited basis Relative Time Failure Resistance 0 % 100 % Functional Failure Freedom of Rotation Chip Collector Filter Debris Analysis JOAP Figure 6 - Detection Techniques FILTER DEBRIS ANALYSIS Gary Humphry proposed the use of filter debris analysis for detecting degraded bearings. Papers with details of the process have been published in the proceedings of the Joint Oil Analysis Program International Conference, Mobile, Al. April, 2002. For this analysis, oil filters were removed from the aircraft periodically and sent to the laboratory for debris analysis. This analysis identifies the metals in the oil filter by percentage and provides the total mass of the sample. The mass of each metal was determined by multiplying its percentage times the sample mass. Element percentage and mass was then tracked. The metals of interest in this bearing were Iron (Fe), Molybdenum (Mo), Vanadium (V Silver (Ag) and Chromium (Cr). Chromium and iron were not reliable markers becaus other components that the oil contacts contained these elements and any of thes elements in the oil could not be correlated with the condition of the bearing in question Silver was also determined not to be a reliable marker because silver plating wears o early and was not available to identify later stages of bearing life. Therefore, the be indicators of bearing wear were Mo and V. To perform this analysis, a number of bearings were removed from engines. Th bearings were visually inspected and assigned a stage number (i.e. stage 3.5 or 2.3, etc This was a subjective application of the general bearing stage definitions above. Th Figure 2 - Stage 1 Bearing Figure 3 - Stage 2 Bearing CONDITION MONITORINg TECHNIQUES FILTER DEBRIS ANALYSIS FIRST QUARTER - 2006 ments and any of these elements in the oil could notbecorrelatedwiththeconditionofthebearing in question. Silver was also determined not to be a reliable marker because silver plating wears off early and was not available to identify later stages of bearing life. Therefore, the best indicators of bearing wear were Mo and V. To perform this analysis, a number of bearings were removed from engines. The bearings were visually inspected and assigned a stage number (i.e. Stage 3.5 or 2.3, etc.) This was a subjective application of the general bearing stage defini- tions above. The same evaluators were used for consistency. The next step was to determine which FDA analyses were associated with the engine from which the categorized bearings were removed. Withthisinformation,thebearingstage and FDA results could be correlated. The V%, V Mass, Mo% and Mo Mass numbers were used in the analysis of the oil filter debris to determine if the engine is flightworthy or requires replacement. However, if more than one filter was analyzed for a particular engine, cumulative V Mass, Mo Mass were compared to the bearing stage. The FDA results were initially analyzed using the predetermined limits of V% (0.268), Mo% (2.68),VMassCumulative(0.0055)andMoMass Cumulative(0.035). TheV%andMo%limitswere determinedbystatisticalanalysisofdataextracted directly from the FDA results and represent the average plus one standard deviation. V Mass Cumulative and Mo Mass Cumulative limits rep- resent the lowest accumulated mass that would identify all Stage 3 bearings from the first 38 bear- ings to be visually checked. Figures 7 through 10 display the results of filter data analyses for 50 bearings. The graphs display the metal percentage or mass (horizontal axis) by bearing stage (vertical axis). A Stage 3 or higher bearing was considered failed. The percentage and mass limits identified were those determined in the final analysis. The graphs identify the failed bearingsfoundbythefinallimitsoftheFDAanal- ysis, those that were missed, and the non-failed bearings identified as failed (false positives). The small vertical lines are the initial statistical limits representing the average plus 1, 2, 3 or 4 standard deviations. In the final analysis, the data was used to deter- mine which element (V, Mo) or combination of these elements and values would identify all Stage 3through5bearingswiththeleastnumberoffalse positives. ThisresultedbyapplyingbothV%(0.30) and Mo% (2.80) to the filter sample. If the FDA contained V% of 0.30 or greater and Mo% of 2.80 or greater, then 100% (10 of 10) of Stage 3 through 5 (failed) bearings and 10% (5 of 50) false positives were identified. These limits have been adopted for evaluating engine FDA effective 7 Oct. 2002. Limits are continually updated as more data is re- ceived.Wehavealsoperformedvariousregression analyses that have validated this approach. Figure 7 - Vanadium Percentage Figure 8 - Molybdenum Percentage continued on page 12 debris to determine if the engine is flight worthy or requires replacement. However, if more than one filter was analyzed for a particular engine, cumulative V Mass, Mo Mass were compared to the bearing stage. The FDA results were initially analyzed using the predetermined limits of V% (0.268), Mo% (2.68), V Mass Cumulative (0.0055) and Mo Mass Cumulative (0.035). The V% and Mo% limits were determined by statistical analysis of data extracted directly from the FDA results and represent the average plus one standard deviation. V Mass Cumulative and Mo Mass Cumulative limits represent the lowest accumulated mass that would identify all stage 3 bearings from the first 38 bearings to be visually checked. DETECTING BAD BEARINGS Figures 7 through 10 display the results of filter data analyses for 50 bearings. The graphs display the metal percentage or mass (horizontal axis) by bearing stage (vertical axis). A stage 3 or higher bearing was considered failed. The percentage and mass limits identified were those determined in the final analysis. The graphs identify the failed bearings found by the final limits of the FDA analysis, those that were missed, and the non-failed bearings identified as failed (false positives). The small vertical lines are the initial statistical limits representing the average plus 1, 2, 3 or 4 standard deviations. Vanadium (Percentage/50 Samples) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 0.900 1.000 1.100 1.200 1.300 1.400 V Percent Bearing Stages Missed Found False positive .300 Figure 7 - Vanadium Percentage Molybdenum (Percentage/50 Samples) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00 6.50 7.00 7.50 8.00 8.50 9.00 9.50 MO Percent Bearing Stages Missed Found False positive 2.8 Figure 8 - Molybdenum Percentage Vanadium (Mass Cum/50 Samples) DETECTINg BAD BEARINgS Vanadium (Percentage/50 Samples) Molybdenum (Percentage/50 Samples) PAGE 11 continued from page 11 Figure 9 - Vanadium Mass Figure 10 - Molybdenum Mass This analysis shows that FDA can be a powerful tool in identifying degraded bearings so they can be removed from service prior to failure. It meets all of the criteria established by the SAE standard for a condition-monitoring task. The analysis was an integral part of a formal RCM analysis process and provided the engineering support for identify- ing the potential and functional failure conditions the RCM analysis requires. A separate analysis was performed to determine the appropriate in- spection interval. Figure 8 - Molybdenum Percentage Vanadium (Mass Cum/50 Samples) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.0000 0.0050 0.0100 0.0150 0.0200 0.0250 0.0300 V Mass Bearing Stage Missed Found False positive 0.0055 Figure 9 - Vanadium Mass Molybdenum (Mass Cum/50 Samples) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 0.260 MO Mass Bearing Stage Missed Found False positive 0.035 Figure 10 - Molybdenum Mass Chart In the final analysis, the data was used to determine which element (V, Mo) or combination of these elements and values would identify all stage 3 through 5 bearings with the least number of false positives. This resulted by applying both V% (0.30) and Mo% (2.80) to the filter sample. If the FDA contained V% of 0.30 or greater and Mo% of 2.80 or greater, then 100% (10 of 10) of stage 3 through 5 (failed) bearings and 10% (5 of 50) false positives were identified. These limits have been adopted for evaluating engine FDA effective 7 Oct. 2002. Limits are continually updated as more data is received. We have also performed various regression analyses that have validated this approach. CONCLUSIONS This analysis shows that FDA can be a powerful tool in identifying degraded bearings so they can be removed from service prior to failure. It meets all of the criteria established by the SAE standard for a condition-monitoring task. The analysis was an integral part of a formal RCM analysis process and provided the engineering support for identifying the potential and functional failure conditions the RCM analysis requires. A separate analysis was performed to determine the appropriate inspection interval. Finding Bearing Failure Through Filter Debris Analysis CONCLUSIONS Vanadium (Mass Cum/50 Samples) Molybdenum (Mass Cum/50 Samples) THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER The appearance of paid advertising in the RIAC Journal does not constitute endorsement by the Department of Defense or the Reliability Information Analysis Center of the products or services advertised. FIRST QUARTER - 2006