LASERNET FINES WEAR DEBRIS ANALYSIS TECHNOLOGY:APPLICATION TO MECHANICAL FAULT DETECTION

J. Reintjes1, J. E. Tucker1, S. E. Thomas2, A. Schultz1, L. L Tankersley3, C. Lu4,P. L. Howard5, T. Sebok6 and C. Holloway6

1 Naval Research Laboratory, Washington, DC USA2 Titan Corp, Crystal City, VA USA3 US Naval Academy, Annapolis MD USA

Towson University, Towson, MD USA5 PL Howard Enterprises, Newmarket, NH USA6 Lockheed Martin NESS, Akron, OH USA

ABSTRACT. The operation of LaserNet Fines wear debris analysis technology is described.Application to detection and identification of mechanical faults in diesel engines is presented. Thedevelopment of fault indicators directly from the LaserNet Fines analysis is described and results arerelated to controlled laboratory experiments.

INTRODUCTION

Monitoring of wear debris in the lubricating oil of rotating and reciprocatingmachinery is a well-established method of non destructive testing that has been used formany decades to give machinery operators a measure of the state of wear in engines,transmissions and drive trains, along with early warning of impending catastrophicfailures. Many laboratory analysis tests have been developed to allow assessment ofvarious wear debris characteristics in terms of machinery condition. Among the variouslaboratory analyses conducted are tests to determine particle concentration, elementalcontent, elemental composition, size and various shape and appearance characteristics.Generally speaking, these tests can be time consuming, expensive and require humanexpert interpretation. They also usually involve drawing of a sample of oil from theequipment and sending it to a remote laboratory for analysis, a procedure that can involvesignificant delay from the time the sample is drawn to the time results are received by theoperator. However, some of these tests, especially those involving microscopicexamination [1-7], can often give an excellent analysis of the onset of wear processes thatcan ultimately lead to machinery failure. More immediate results can be obtained from on-line monitors, but on line detectors such as chip detectors and inductive monitors givelimited analysis of particle properties and can fail to give reliable early warnings of theonset of excess wear conditions.

In this paper we describe the LaserNet Fines (LNF) wear particle analysistechnology [8-11]. This technology was developed for on line and on site analysis of weardebris with the goal of providing more immediate, reliable and earlier detection ofmechanical faults without the need for expert human intervention. LNF is an opticallybased wear particle analyzer that determines type, severity and rate of progression of

CP657, Review of Quantitative Nondestructive Evaluation Vol. 22, ed. by D. O. Thompson and D. E. Chimenti2003 American Institute of Physics 0-7354-0117-9

1590

mechanical faults by measuring the size distribution, concentration and shapecharacteristics of the wear particles. In addition to mechanical wear particles, LNF detectsand identifies fiber, abrasive oxide and free water concentration. This capability allowsassessment of system cleanliness and the effects on system health of specificcontamination types (such as the effect of sand on gear wear and failure, or the presence offilter deterioration). LNF has similarities to ferrography and microscopic analysis in that itmakes use of the shape characteristics of individual particles, but goes beyond thesetechniques by developing quantitative fault indicators that can be used to identify not onlythe presence of a fault but its type and level of progression. It is a broadly applicable debrismonitoring technology that can be used not only for the detection of faults in oil wettedmachinery but also for the determination of particulate contamination levels in hydraulicsystems and identification of contaminant type, guiding remedial actions.

LASERNET FINES TECHNOLOGY

The LNF analysis process is illustrated in Fig. 1. A sample of fluid, either a bypassfrom the main flow for an on line system or a bottle sample from a manually-sampledsystem, is illuminated with a laser diode and images of entrained objects are recorded onan electronic camera. LNF measures the size distribution of particles from 4 Jim to above100 jLim, and provides classification based on shape features for objects larger than 20 |im.Analysis of particles in this size range is useful for the early detection of mechanical faultsin diesel and turbine engines, turbochargers and gearboxes. This range of sizes is alsorelevant to particulate contamination levels in hydraulic and fuel systems. Examples ofimages of different types of particles and their implications for mechanical difficulties areshown in Fig. 2. Quantitative records are made of the amount of each type of particle as afunction of size, along with an image map of all particles larger than 20 |iim. These recordsallow fault signatures to be developed directly from the LNF results, providing the basisfor root cause analysis, definition of fault severity, determination of maintenance actions,and remaining useful life prognostics.

SAMPLE ANALYSIS WEAR MODECLASSIFICATION

MACHINE MAINTENANCECONDITION ACTIONASSESSMENT

FIGURE 1. Schematic illustration of the operation of LaserNet Fines and its application to fault analysis.

1591

WMM?M®im*

Fatigue spall. Excess load orend of life

Air bubble. Possible leak in hydraulicsystem.

c Water bubble. Free water

Fiber. Artifact in bottle sampleanalysis, filter deterioration, compositebearing wear.

Cutting wear. Shaftmisalignment or secondarydamage from debris buildup orsand contamination.

Severe sliding wear. Oil filmfailure or inadequate lubrication.

Sand. Contaminant, source ofsecondary damage.

FIGURE 2. Images of objects detected and identified by LaserNet Fines.

LNF can be configured as an off line batch sample analyzer or an on line wear or hydrauliccleanliness monitor. First and second generation batch processor models are incommercial production by Lockheed Martin. An on-line system is undergoing laboratorytest and evaluation.

DIESEL ENGINE CLASS STUDY

As an example of the ability of LNF to find and identify specific mechanical faults,a study was made of 32 medium speed diesel engines. The engines were put into service atdifferent times, but were all subject to the same operational profile. A class study of thistype provides a snapshot of the diesels over their life, allowing the establishment of normaland abnormal operation.

Total particle concentration for the engines is shown in Fig. 3, and concentration ofparticles in the different wear categories are shown in Fig. 4. The majority of the enginesshowed consistently low particle concentrations, allowing the establishment of a baselinefor normal wear. However, significantly elevated particle concentrations were found on 6engines. Of the 6 engines that showed elevated particle concentrations, one (#18) showedsignificantly higher concentrations than the other engines in all LNF categories (fatigue,cutting, sliding and non-metallics). This engine was singled out for detailed study. Duringthe study it was made known to us that a complete failure of a piston wrist pin hadoccurred in one of the engines (#14). Our initial sample was taken for that engine after thecylinder assembly was replaced. Subsequently a sample was obtained from that enginethat had been taken 14 months prior to detection of the failure.

Engines with elevated total particle concentrations but low concentrations ofparticles larger than 20 microns are indicated by open symbols in Figs. 3 and 4. Thischaracteristic can indicate the presence of faults such as grinding, that do not produce largeparticles or of heavy sooting. In either case, it is indicative of a second type of fault, notrelated to the severe wear seen on the engines with elevated large-particle concentrations.

1592

Total Particles

•

o o

V*%« **«*x±* » « V %»*»*

0 5 10 15 20 25 30 35Diesel Number

FIGURE 3. Total particle concentrations for the set of medium speed diesels.

Severe sliding Wear Cutting Wear

i:•§nn

"Inn|nn

0

^800

C 200

0

«

m

® $

«g»»Mtity»tt*..o. •?••?•?•••?,??. .5 10 15 20 25 30 3

Diesel Number

Fatigue wear

ti 'tttt'ttttt* - t* ,»tt.ttt,tt

120-

*— 100-

1 6°

I 4°

& 2°

5

s1̂ 60

©

J »•0-

•

••

•+ o*

t t TM i» t f » t t f r . . » . r , . » t »%»» t t t t . . lD 5 10 15 20 25 30

Diesel NumberOxides

•+

•

o

FIGURE 4. LNF wear class analysis for the set of diesel engines showing particle concentrations for eachengine in the various wear types. Top left: severe sliding wear; top right: cutting wear; bottom left: fatiguewear; bottom right: non-metallics.

Validation of LNF Results

A lab analysis consisting of optical microscope examination and scanning electronmicroscope energy dispersive x-ray (SEMEDX) measurements was made of the debrissamples from engines 18 and 14 and also of the failed parts from engine #14.

The optical microscope examination confirmed that the oil samples from engines18 and 14 contained large concentrations of metal chips of various sizes and shapes inagreement with LNF analysis. A high-resolution microscope image of debris from engine

1593

#18 (Fig. 5) shows the presence of flakes, slivers and curled cutting wear confirming theLaserNet Fines analysis of debris type.

The SEMEDX analysis of the failed parts from engine #14 showed that the wristpin was steel with chromium and nickel components, while the bearing sleeve was mildsteel with no chromium or nickel. The debris from the failed engine showed piecesmatching the wrist pin and others matching the sleeve. All of the particles in the oilsamples from #18 and #14 matched only the bearing sleeve material, not the wrist pin.

These results confirm that LaserNet Fines is indeed measuring heavy wear when itoccurs, and is not reporting it in engines without heavy wear.

LNF Fault Signatures

Analysis of the LaserNet Fines records indicates that the quantitative distributionsof particle concentration by size in the various wear categories can be combined to give asignature for fault presence and severity (Figs. 6 and 7). The distributions for the enginewith normal wear show low particle concentration, with a steeply declining concentrationwith particle size for the individual wear categories (Fig. 7). For engine #18, the particlesize distributions are significantly different, being skewed to the larger particle sizes

I ^———— 500 jimFIGURE 5. Magnified images of debris from engine #18 showing fatigue flakes, sliding slivers and cuttingcurls.

Fatigue wear

i

20-25 25-50 50-103 >100see rang e Jjm)

Severe sliding wear

t20^25 25-50 50-100 >100

size range (|jn)

0 20

S:

Cutting wear

LI2D-25 25-50 50-100 >100

size range (\m)

FIGURE 6. LNF particle distributions for engine #18 showing signature for heavy wear.

1594

Fatigue wear Severe sliding wear Cutting wear

I20-25 25-50 50-100 >100

size range (jjii)20^25 2550 50-100 >100

size range (|jm)20-25 25-50 50-100 >10D

size range (jjm)

FIGURE 7. LNF particle distributions for an engine with normal operation

(Fig. 6). The relative amounts of particles in the various wear categories, along the sizedistributions provide the information needed to interpret presence and severity of the fault.LNF thus provides quantitative analysis that can be used for recommendation of remedialaction. Furthermore, the LNF fault signatures are robust against maintenance actions suchas topping off of oil supplies because they are based on relative distributions, not justabsolute particle concentrations.

Early Fault Detection

Following this analysis it was discovered that a sample from engine #14 that hadbeen drawn 14 months prior to detection of the failure was available. LNF analysis of thatsample showed an abnormal signature in the fatigue and sliding categories (Fig. 8),indicative of an early stage of the mechanical fault that eventually resulted in total failureof the wrist pin. This demonstrates that LNF has the capability for providing aconsiderable advanced warning of potentially serious wear conditions.

Validation of the LNF Fault Signature

Tests have been carried out in laboratory controlled environments to obtain LNFsignatures for specific mechanical wear processes in various stages of development. Theresults show that, for severe sliding test conditions, the fault signature described abovedevelops in the LNF sliding category after sliding seizure events. (Fig. 9) No similardevelopment is evident in the fatigue category, confirming the ability of LNF todistinguish different mechanical wear conditions. (Fig. 10)

Fatigue wear Severe sliding wear Cutting wear

20-25 25-50 50-100 >100

size range (Mm)

FIGURE 8. LNF analysis of oil sample from engine 14 taken 14 months prior to failure detection showingearly detection of heavy wear.

1595

Sliding Pre-seizureSliding Onset of seizure

t 30,-

m 7°C 60

- [ :- 2 30

i -§ "

size (urn)

Sliding During seizure

I

I ™ 3Ui !«8"FIGURE 9. LNF analysis of wear particles from a controlled laboratory severe sliding test showingdevelopment of the abnormal signature in the severe sliding wear category following a sliding seizure event.

Fatigue Eire-seizure Fatigue Onset of seizure80

70

60

50

40

& 30

20

10

Ii

J 20- —

Fatigue Post seizure

25 50size (urn)

25 50size (urn)

FIGURE 10. LNF analysis of wear particles from controlled laboratory severe sliding test showing a normalsignature in the fatigue wear category following the severe sliding seizure.

This result ties the LNF fault signature for heavy wear in the operating equipmentto fundamental mechanical wear processes.

1596

ACKNOWLEDGEMENTS

The authors acknowledge the support of the Office of Naval Research.

REFERENCES

1. T. P. Sperring, J. Tucker, J. Reintjes, A. Schultz, C. Lu and B. J. Roylance," WearParticle Imaging and Analysis - a contribution towards monitoring the health ofmilitary ships and aircraft," International Conference on Condition Monitoring, pp.539-546, University of Wales, Swansea, UK, April 1999.

2. B. J. Roylance and G. Pocock, "Wear Studies through Particle Size Distribution,"Wear, 90, pp. 113-136 (1983).

3. A. Albidewi., A. R. Luxmore, B. J. Roylance, and G. Wang, "Determination ofParticle Shape by Image Analysis-the Basis for Developing an Expert System," in"Condition Monitoring '91," M. H. Jones, J. Guttenberger and H. Brenneke, eds.,Pineridge Press, Swansea, UK, 1991, p. 411.

4. B. J. Roylance and S. Raadnui, "The morphological attributes of wear particles - theirrole in identifying wear mechanisms," Wear 175, 115 (1994).

5. B. J. Roylance, I. A. Albidewi, A. R. Luxmoore, A. L. and Price, "The Developmentof a Computer-Assisted Systematic Particle Analysis Procedure - CASPA," Lub. Eng.,48, pp. 940-946 (1992).

6. B. J. Roylance, I. A. Albidewi, M. S. Laghari, A. R. Luxmoore and F. Deravi,"Computer-Aided Vision Engineering (CAVE) - Quantification of Wear ParticleMorphology," Lub. Eng., 50, pp. 111-116 (1994).

7. T. G. Barraclough, T. P. Sperring and B. J. Roylance, "Generic-based Wear DebrisIdentification - the First Step Towards Morphological Classification", InternationalConference on Condition Monitoring, University of Wales, Swansea, UK April 1999.

8. J. E. Tucker, J. Reintjes, A. Schultz, C. Lu, L. L. Tankersley, P. L. Howard, T. Sebok,and C. Holloway, "LASERNET Fines Shipboard Wear Debris Monitor", TechnologyShowcase 2000, JOAP International Conference, Condition, Mobile AL, April 3-6,2000.

9. J. E. Tucker, A. Schultz, C. Lu, T. Sebok, C. Holloway, L. L. Tankersley , T.McClelland, P. L. Howard and J. Reintjes, LaserNet Fines Optical Wear DebrisMonitor," International Conference on Condition Monitoring, pp. 445-452, Universityof Wales, Swansea, UK, April 1999.

10. J. E. Tucker, J. Reintjes, T. L. McClelland, M. D. Duncan, L. L. Tankersley, A.Schultz, C. Lu, P. L. Howard, T. Sebok, C. Holloway, and S. Fockler, "LaserNet FinesOptical Oil Debris Monitor," 1998 JOAP International Condition MonitoringConference, pp. 117-124 April, 1998, Mobile AL.

11. J. Reintjes, R. Mahon, M. D. Duncan, L. L. Tankersley, J. E. Tucker, A. Schultz, V. C.Chen, C. Lu, T. L. McClelland, P. L. Howard, S. Raghavan and C. L. Stevens, "RealTime Optical Oil Debris Monitors," Proceedings of 51st Meeting of the Society forMachinery Failure Prevention Technology, April, 1997, pp. 443-448, Virginia Beach,VA.

1597