Wear, 44 (1977) 173 - 182 0 Elsevier Sequoia S.A., Lausanne -- Printed in the Netherlands

173

PREDICTIVE MAINTENANCE BY FERROGRAPHY*

D. SCOTT

National Engineering Laboratory, East Kilbride, Glasgow (Gt. Britain)

V. C. WESTCOTT

Foxboro/Trans-Sonics Inc., Burlington, Mass. 0183 (U.S.A.)

(Received April 1, 1977)

Summary

Ferrography is a technique by which wear debris and contaminant particles are separated from a lubricant and analysed. The apparatus used and complementary investigational techniques are described. The use of Ferrography for machinery condition monitoring to prevent failure and to allow a safe change from expensive periodic dismantling of machinery for maintenance to the more economical predictive maintenance is outlined.

Introduction

Machines wear in service. Under normal working conditions the rate of wear is low and of little concern but, as the cumulative action of wear can lead to some deterioration in performance, regular inspection of machines and attention to the lubrication system is essential to maintain effective operation. Abnormal wear may cause premature failure.

Technological progress has created increasingly complex machines, the working parts of which are usually totally enclosed. Thus, in order to examine the working parts and to determine the type and rate of wear, the modern maintenance engineer periodically dismantles the machinery for inspection. Such a procedure is time-absorbing and costly in labour and lost productivity. The risk of failure is also much greater at the start-up of machinery than when it has been run-in and running for some time. Thus increasing emphasis is now being given to on-line monitoring techniques as a

*Paper prepared for presentation at the 2nd National Industrial Tribology Conference, Powai, India, December, 1976.

means of detecting deterioration of machinery so that remedial action can be taken before the breakdown point is reached. Economic pressures are also demanding a switch from regular periodic dism~tlin~ of machinery for maintenance to “on condition” maintenance which has accentuated the need for simple methods of machinery condition monitoring. Various diagnostic systems [ 1 - 121 have been developed involving techniques such as vibration analysis, magnetic plugs and the spectrographic oil analysis procedure (SOAP). The last technique is extensively used to detect abnormal wear from the amount of debris which enters the lubricant from relatively moving surfaces. Although the technique has proved effective in providing some warning of changes in a system, it has limitations [ 131. It provides a knowledge only of the quantity of metal in the oil but no information on the size and shape of the metallic particles.

Wear particles are unique, having individu~ characteristics which bear evidence of the conditions under which they were formed [ 14 - 161. Thus careful examination of the morphology and determination of the composi- tion of the wear particles can yield specific information about the condition of the moving surfaces of the machine elements from which they were produced, the mechanism of their formation, and the mode of wear in operation in the system from which they were extracted.

Ferrography [l7 - 201 is a technique developed to separate wear debris and contaminant particles from a lubricant and to arrange them on a transparent substrate for examination. As the wear particles are precipitated magnetically, virtually all the unwanted carbon dirt particles are eliminated. The precipita~d particles, deposited according to size, may be individu~ly examined since large particles are not obscured by agglomerates of smaller particles, and the unique characteristics of all sizes of particles can be estab- lished.

This paper describes the Ferrograph analysis system and outlines its use in predictive maintenance and it describes the monitoring of machinery condition so that remedial action can be planned and taken before the break- down point is reached.

The Ferrograph

The Duplex Ferrograph analyser (Fig. 1) consists of two particle sep- arators; on the right is a standard Ferrograph analyser, and on the left a direct reading (DR) Ferrograph.

The standard Ferrograph analyser consists of a pump to deliver the lubricant sample at a slow rate, a magnet that develops a high gradient magnetic field near its poles and a treated transparent substrate on which the particles are deposited. The lubricant sample, diluted with a special solvent to promote the precipitation of wear particles, is pumped across the transparent substrate which is mounted at a slight incline. The magnetic particles adhere to the substrate, being distributed according to size; a

175

Fig. 1. The Duplex Ferrograph.

washing and fixing cycle removes the residual oil and causes the wear particles to adhere permanently to the slide. Using the ferrogram reader attached to a bichromatic microscope [ 16 - IS], the optical densities of the deposits may be observed at various distances along the ferrograms to determine the percentage area covered and thus the amount and particle size distribution.

With the direct reading section, 1 ml of lubricant is syphoned through the precipitation tube and the optical densities of the deposits at selected distances are observed.

Ferrographic analysis

When successive lubricant samples yield ferrograms with essentially constant density readings it may be concluded that the machine is operating normally and producing wear particles at a steady rate. A rapid increase in the quantity of particles, and in particular in the ratio of large to small particles, indicates the initiation of a more severe wear process.

Significant numerical data can be derived from the direct reading section of the ferrograph, such as the fractional area coverage As of particles in the 1 - 2 I.rrn (S) size range and the area coverage AL of large (L) particles. If normal rubbing wear is the predominant wear mode and most of the particles are small, the L readings will be comparable with the S readings. If the operative wear modes are severe, the L reading will be large compared

176

with the S readings. Use of the formula AZ - A; gives a single figure 1, which is designated the severity of wear index. Is increases with an increase in the severity of wear. Experience can determine ranges of Is to cover adequately the modes of wear; this allows a code of monitoring practice to be laid down.

Table 1 gives some summarized results of DR Ferrograph readings of lubricant samples taken at various times during the life test of a large gear- box. The comparatively large severity of wear index of 45 from sample 1 and the miscellaneously shaped wear particles indicate that the gearbox was running in. The reduction of the area of coverage of particles in samples 3 - 12, particularly that of the larger particles, the low severity of wear indices and the presence of only normal rubbing wear type particles indicate steady normal running. The accelerated increase in the area coverage and in the ratio of large to small particles observed in samples 14 - 16, which gives an increasingly large numerical value for the severity of wear index, shows that a new more severe wear mechanism had been initiated. Cutting-wear- type and spherical particles were found, indicating serious abrasion and fatigue crack propagation. Most machines in the field, where filtering is not so efficient, operate with higher values of A, and As but the increase in 1, as failure approaches is characteristic.

TABLE 1

The severity of wear index I, from DR Ferrograph readings during a life test of a gearbox

Sample Test duration Area coverage (%) Severity of Comments no. life of wear

gearbox (%) Large particles Small particles index I

AL AS S

1 7 7.1

3 13 0.8 5 27 1.2 7 40 1.2

10 53 1.0 12 60 0.9

14 15 16 ____

73 3.2 1.0 9.2 Cutting-type 77 3.7 1.3 12.0 wear particles and 80 5.0 1.6 22.4 spherical debris

2.3 45.1

0.2 0.6 0.2 1.4 0.5 1.2 0.2 1.0 0.2 0.8

Miscellaneous debris

Small normal rubbing wear platelets

Information on the morphology of the deposited particles is obtained with the aid of a bichromatic microscope which uses simultaneously reflected red light and transmitted green light. Metal particles reflect red light and block green light and thus appear red. Particles composed of compounds allow much of the green light to pass and appear green or, if they are rela- tively thick, yellow or pink.

Particles generated by different wear mechanisms have characteristics which can be identified with the specific wear mechanisms [ 151. Rubbing

Fig. 2. Cutting wear particles: (a) optical micrograph (X 250); (b) scanning electron micrograph (X 1800).

(a) tb)

Fig. 3. Strings of platelet-type normal rubbing wear particles: (a) optical micrograph (X 1200); (b) scanning electron micrograph (x 6000).

wear particles found in the lubricants of most machines are platelets and indicate normal permissible wear. Cutting or abrasive wear particles take the form of miniature spirals, loops and bent wires similar to swarf from a machining operation. A concentration of such particles is indicative of a severe abrasive wear process; a sudden increase in the concentration of such particles in successive lubricant samples signals imm~ent machine failure. Particles consisting of compounds can result from an oxidizing or corrosive environment. Steel spherical particles are a characteristic feature associated with fatigue crack propagation in rolling contacts [21, 221. The concentra- tion of spherical particles indicates the extent of crack propagation [ 231.

178

(a) 0))

Fig. 4. Spherical metallic particles: (a) optical micrograph (X 1 ZOO); (b) scanning electron micrograph (X 4500).

Fig. 5. A fatigue chunk (X 2000).

Fig. 6. X-ray energy analysis trace from a large cutting wear particle (Fig. Z(b)), indicating a particle composition of principally Fe and Cr and a background of Si, K, Au and Zn from the substrate.

Examination of ferrograms with the aid of a scanning electron micro- scope can reveal specific characteristic details of wear particles [ 241. Figure 2 shows typical cutting or abrasive wear particles. Figure 3 shows strings of platelet-type normal rubbing or sliding wear particles. Figure 4 shows typical spherical particles and Fig. 5 shows fatigue chunks, X-ray energy analysis in the scanning electron microscope and electron probe X-ray microanalysis provide means of determining the composition of the individual particles and thereby determining the material of the wearing component (Fig. 6). The mechanism of wear by which the particle was formed establishes the particle

179

morphology. Progressive heating of ferrograms causes changes in the appear- ance of particles, which allows additional information on particle composi- tion to be deduced [ 251. An operational procedure has been proposed to separate metals and alloys into characteristic groups. Use of this procedure can, for instance, differentiate between particles from a cast iron cylinder liner or a steel piston ring operating within the cylinder.

The specific regimes of wear can be classified by the nature of the particles produced by surfaces in sliding contact [26]. Six regimes of rubbing wear which generate characteristic particles have been identified. Determination of the operative wear regime allows postulation regarding the condition of relatively moving surfaces that are inaccessible for direct examination. Regimes 1 and 2 represent normal wear conditions correspond- ing to hydrodynamic and boundary lubrication. Evidence of one or more of the higher regimes (3, 4 or 5) indicates that some parameter of the system has changed unfavourably. The occurrence of regime 6 indicates impending catastrophic failure. Free metal particles are produced in regimes 1, 2, 3 and 6, and these wear regimes may be identified by the particle size. In regime 4 a mild form of oxidative wear dominates, and the majority of wear particles are hematite (a-FezOs). Regime 5 generates black oxides which indicate a severe form of oxidative wear.

Lubricants, lubricant additives, lubricant contaminants and their influence on machine condition can be assessed by Ferrography [14,19, 271. The friction polymer formed by surface interaction may differ with different lubricants, and with lubricants to the same specification but with different additive packages [ 28,291. An amorphous polymer, termed polymud (Fig. 7), and a tough surface film which subsequently rolled up (Fig. 8) were formed in polyester lubricants with the same specification but containing different additive packages. The surface polymer reduced wear.

Fig. 7. Amorphous polymer (polymud) (X 2000).

Fig. 8. Rolling-pin-type polymer (X 6000).

Fig. 9. Basket-type polymer particles (X 8500).

The same type of lubricant has produced spherical basket-type particles (Fig. 9) in a jet engine. Critical filter performance can be readily assessed by Ferrography [14,28, 291.

Discussion

The study of particles generated by solid surfaces in relative motion can yield significant information for the elucidation of wear mechanisms, lubricant and additive behaviour and environmental influence. The condition of inaccessible mechanisms may be assessed by the amount, size, nature and morphology of the particles of wear.

Ferrography is a versatile refined technique developed for particle tribo- analysis. The relatively inexpensive DR Ferrograph can be applied to the prognostic approach to failure prevention and the safe change from regular dismantling of machinery for routine maintenance to the more economic on- line, machinery condition monitoring. The introduction of a code of practice based on a single number severity of wear index allows monitoring by relatively unskilled labour. A more comprehensive ferrographic oil analysis can be obtained by the analytical Ferrograph. Although the use of the complete Ferrographic analysis system requires personnel with a background knowledge of particle tribology, the economic benefits from using such a system may be substantial. In some instances a complete ferrographic analysis may indicate that an increasing severity of wear index is due to a change in a single operating parameter in the system such as lubricant degradation, contamination or corrosive environment which may be rectified without dismantling the machine.

In machine or systems monitoring, ferrography can indicate premature distress of a specific component such as a bearing, gear, cam, spline or filter element. When numerous bearings or similar self-contained units are vital

181

components, the rinsing of these with particle free lubricant and the subsequent Ferrographic analysis of the rinse reveals the wearing condition of each unit without dismantling [30]. Confirmation of DR Ferrograph readings by systematic Ferrographic analysis permits determination of the factors which could cause failure and thus allows the prediction of the remedial measures that are required and when such maintenance should be carried out.

Acknowledgment

This paper is published by permission of the Director of NEL. It is Crown copyright and is reproduced by permission of the Controller, Her Britannic Majesty’s Stationery Office.

References

1 M. K. Eberie, The Sulzer engine diagnostic system, The Institution of Engineers and Shipbuilders, Paper no. 1400 (Part 2), 1976.

2 C. B. Redgate, The airline viewpoint, Symp. on Gas Turbines - Status and Prospects, 1976, Inst. Mech. Eng. preprint C9/76.

3 R. C. Hunter, Engine early warning failure detection systems, R. Aer. Sot. Air Trans- port Group Symp., Feb. 1975.

4 W. H. Dainty, Gas turbines in the RAF from a maintenance engineering viewpoint, Symp. on Gas Turbines - Status and Prospects, 1976, Inst. Mech. Eng. preprint C4/76.

5 M. H. Piper, The Royal Navy’s experience of Rolls-Royce Olympus and Tyne main propulsion gas turbines, Symp. on Gas Turbines - Status and Prospects, 1976, Inst. Mech. Eng., preprint C5/76.

6 A. V. Cooke, Engine health monitoring for the gas turbines in Royal Navy ships, ASME, paper 73-GT-81,1973.

7 T. E. Thoren, A new maintenance concept applied in the design of a new industrial gas turbine in 100 MW class, Symp. on Gas Turbines - Status and Prospects, 1976, Inst. Mech. Eng. preprint C8/76.

8 T. E. Thoren and R. L. Duncan, Performance and maintainability objectives in the design of a new industrial gas turbine of the 100 MW class, Congres International des Machines 1 Combustion, CIMAC, Barcelona, 1975.

9 D. C. Johnson and D. Dickenson, Monitoring for preventive maintenance in the small fleet, Symp. on Gas Turbines - Status and Prospects, 1976. Inst. Mech. Eng. preprint C11/76.

10 L. H. Chittenden and A. V. Cooke, The development and experience of engine health monitoring for gas turbines of the Royal Navy, Symp. on Gas Turbines - Status and Prospects, 1976, Inst. Mech. Eng. preprint C21/76.

11 A. E. Davies, Principles and practice of aircraft power plant maintenance, Trans. Inst. Mar. Eng., 84 (14) (1972) 441 - 464.

12 F. S. Nowlan, A current turbine engine maintenance programme and the experience and logic on which it is based, ASME paper 73-GT-81,1973.

13 W. W. Seifert and V. C. Westcott, Investigation of iron content of lubricating oils by Ferrograph and emission spectrometer, Wear, 23(2) (1973) 239 - 249.

14 D. Scott, Particle tribology, Chairman’s Address to Tribology Group of Inst. Mech. Eng., NEL Rep. 627, NEL, East Kilbride, 1975, in the press.

182

15 D. Scott, Debris examination - a prognostic approach to failure prevention, Wear, 34 (1975) 15 - 22.

16 D. Scott, W. W. Seifert and V. C. Westcott, The particles of wear, Sci. Am., 230 (5) (1974) 88 - 97.

17 W. W. Seifert and V. C. Westcott, A method for the study of wear particles in lubricat- ing oil, Wear, 21 (1972) 22 - 42.

18 R. Bowen, D. Scott, W. W. Seifert and V. C. Westcott, Ferrography, Tribol. Int., 9 (3) (1976) 109 - 115.

19 R. Bowen and W. W. Seifert, Ferrography - a new tool for analysing wear conditions, Proc. Fluid Power Symp., 1976, in the press.

20 V. C. Westcott, Ferrographic oil and grease analysis as applied to earth moving ma- chinery, Sot. Auto. Eng., 1975, preprint 750555.

21 D. Scott and G. H. Mills, Spherical particles in rolling contact fatigue, Nature (London), 241 (1973) 115 - 116.

22 D. Scott and G. H. Mills, Spherical debris - its occurrence, formation and significance in rolling contact fatigue, Wear, 24 (1973) 235 - 242.

23 J. L. Middleton, V. C. Westcott and R. W. Wright, The number of spherical particles emitted by a propagating fatigue crack in rolling bearings, Wear, 30 (2) (1974) 275 - 278.

24 D. Scott and G. H. Mills, Debris examination in the SEM - a prognostic approach to failure prevention. In Scanning Electron Microscopy, IIT Research Inst., Chicago, 1974, pp. 883 - 888.

25 F. T. Barwell, E. R. Bowen, J. P. Bowen and V. C. Westcott, The use of temper colors in Ferrography, Wear, 44 (1977) 163 - 171.

26 A. A. Reda, R. Bowen and V. C. Westcott, Characteristics of particles generated at the interface between sliding steel surfaces, Wear, 34 (1975) 261 - 273.

27 A. A. Reda, A note on the investigation of friction polymer rolling pin formation, Wear, 32 (1975) 115 - 116.

28 D. Scott, W. W. Seifert and V. C. Westcott, Ferrography - an advanced design aid for the 80’s, Wear, 34 (1975) 251 260.

29 D. Scott, Improved reliability by particle tribology, Proc. Hungarian Conf. on Tribology in Motor Vehicles, Budapest, 1976, in the press.

30 Ferrographic Technical Bulletin No 3, “Non-destructive inspection procedure for mechanical parts”, Foxboro/Trans-Sonics Inc., August 1974.