This invention relates to a method and apparatus or system for monitoring oil-operating equipment to identify potential equipment failure and is more especially concerned with such a method and apparatus for predicting gas turbine engine failure. BACKGROUND ART

Equipment of various types employs oil in its operation, especially circulating oil; such oil may circulate as a result of forced or natural circulation forces. Typically equipment of this type includes engines in which the oil serves as a lubri¬ cant and electric transformers in which the oil serves as a coolant and as a dielectric medium. Such oil operating-equipment is subject to failure for various reasons and such failure can have dire consequences; for example, in the case of aviation engines, engine failure may result in a crash and fatalities; in the case of transformers failure may result in temporary shortage of electri-" city supply which at best results in inconvenience to users and at worst may result in fatalities.

Ongoing maintenance of such equipment is routine in attempts to minimize failure. in order to comply with the need for safety in aviation gas turbine engines and to reduce the costs of maintenance, periodic sampling of oil used in the operations of the engine and analysis of the oil is carried out in an attempt to diagnose mechanical and oil related problems.

The method currently used by commercial airlines and military bodies for diagnosing mechanical problems involves the measurement of small

metal particles suspended in the oil using atomic emission or atomic absorption spectroscopy and the detection of larger metal particles by filter analysis, chip detector analysis or ferrographic analysis.

All of these methods rely on the principle that a mechanical component of a particular metallic composition has already begun to fail sufficiently to generate metal particles in the oil, which particles can be detected by one of the aforementioned methods. It is well known that at best these methods provide only a short warning period before failure of the engine. Often, there is no warning period whatsoever. Chip detector warning is characteristi- cally of the order of several hours; filter analysis warning is at best 18 hours; traditional oil analysis using atomic spectroscopy is unreliable since it is limited to very small particle detection.

Infrared analysis of oil has been used to monitor the quality of the oil in order to determine its ability to effectively lubricate the engine. However, it has not been used as a diagnostic tool to determine the state of health of mechanical com¬ ponents of the engine. Engine oils employed to lubricate engine parts during use contain a number of chemical additives in relatively small amounts. Common additives include oxidation inhibitors, rust inhibitors, anti-wear agents, detergents, dispersers, pour-point depressants, viscosity index improvers and foam inhibitors. The selection and amount of additive depends on the nature of the engine and the environment in which it will be employed.

Tri-cresyl phosphate is widely employed as an anti-wear additive in gas turbine engine oils, especially for gas turbine engines employed in aircraft and industry, and functions by forming a protective sacrificial film on contacting mechanical surfaces which are lubricated by the oil.

Phenyl-^-naphthylamine and dialkylpheno- thiazine are employed as antioxidants in engine oils; and 2, 6-ditertiary-butyl para cresol and 2,6-di- tertiary-butyl phenol are employed as antioxidants in electrical insulating oils. DISCLOSURE OF THE INVENTION

The invention seeks to provide a method and apparatus or system which will provide relatively long term warning of failure in oil-operating equip¬ ment.

In accordance with one aspect of the invention there is provided a method of monitoring oil-operating equipment to identify potential equipment failure comprising: i) determining the concentration of an identified additive in an oil employed in the operation of equipment, ii) comparing the concentration determined in i) with a reference value to provide a comparison, and iii) evaluating the comparison as an indication of the state of the equipment. Conveniently steps i), ii) and iii) are repeated sequentially during continued operation of the engine.

In accordance with another aspect of the invention there is provided an apparatus or system for monitoring oil-operating equipment to identify potential equipment failure comprising: a) means for determining the concentration of an identified additive of an oil employed in the operation of equipment, b) comparator means for comparing a

determined concentration with a reference value and providing a comparison, and c) evaluating means for evaluating the comparison as an indication of the state of the equipment. The method and apparatus or system of the invention employ determination of the concentration of an additive in the oil, more especially a normal oil additive, for example, tri-cresyl phosphate, and which oil has been used in the operation of an engine under investigation.

The invention relies on the surprising observation that oil in an engine which is approach¬ ing failure, suffers a significant drop in its concentration of an additive, such as tri-cresyl phosphate prior to failure.

Indeed for the particular case in which the additive is tri-cresyl phosphate, if the oil in an engine which exhibits this drop in concentration, is replaced by fresh oil, the concentration of the tri-cresyl phosphate in the fresh oil rapidly drops. This drop in concentration occurs in a time signifi¬ cantly shorter than, and much more dramatically than, the expected drop in concentration of the phosphate which occurs during the normal operation of a pro- perly functioning engine.

This observed drop in concentration of an additive such as tri-cresyl phosphate has been observed to occur long before actual failure of the engine, and before development of metal particles from the metallic components of the engine, in an amount detectable by the traditional methods described above.

Although the reason for this phenomenon has not been established it is possible that a developed mechanical defect or problem in the engine results in the creation of one or more hot spots in the engine, and that the heat from these hot spots rapidly degrades or destroys the additive thereby reducing its concentration.

The determination of the additive concent¬ ration of the oil may be carried, out by a number of analytical techniques, however, infrared spectro¬ scopy, and in particular fourier transform infrared spectroscopy has been found to be especially useful for this purpose.

By means of the present invention it is possible to predict engine failure based on the analysis of the engine oil, and this prediction can be made within a reasonable warning period. Further¬ more, while an additive can be deliberately incor¬ porated in the oil for the purposes of this invention, the invention is applicable to additives normally incorporated in such oils based on parti¬ cular properties, for example, the anti-wear additive tri-cresyl phosphate.

It is likewise possible to predict failure of other oil-operating equipment, for example, electrical transformers, in which hot spots which degrade or thermally decompose oil-additives, may result from early stage electrical shorts or other electrical failures which forebode a breakdown or failure. By way of example other additives which might be exploited in accordance with the invention include phenyl-o-naphthylamine, dialkylphenothiazine, 2,6-ditertiary-butyl para cresol and 2,6-ditertiary butyl phenol.

In this invention "oil-operating equipment" means equipment which employs an oil in its operation and includes mechanical equipment, for example, engines and gear boxes and electrical equipment, for example, transformers.

BRIEF DESCRIPTION OF DRAWINGS

Fig. 1 illustrates schematically an apparatus or system of the invention for use in carrying out the method of the invention; Fig. 2 illustrates in flow chart form an algorithm for use in the invention. MODES FOR CARRYING OUT THE INVENTION

The levels of tri-cresyl phosphate for normally functioning gas turbine aviation engines can be shown to be relatively stable based on analysis of samples of the oil over an extended period of time; in particular analysis shows a level of about 90% of the original content of tri-cresyl phosphate when the engine functions normally. It has been observed that the levels of tri-cresyl phosphate are abnormally low for an engine which subsequently fails due to mechanical breakdown. An engine employed in the invention failed due to mechanical breakdown several weeks after a consistent drop in concentration of the tri-cresyl phosphate below the 80% level, exhibited between 9,000 and 10,000 hours of use of the oil.

The traditional tests for metal particles failed to give any indication of the imminent f ilure at the stage at which the consistent drop in concent¬ ration of tri-cresyl phosphate was observed.

The early warning of failure provided by the method of the invention provides ample time for more extensive tests and inspection of an engine so

that the possibility of a disaster such as an in¬ flight failure can be reduced or avoided. This early warning permits a range of more extensive testing for a specific engine for which failure is predicted, which would not be practical on a routine basis for all engines.

The invention is more particularly des¬ cribed by reference to the embodiment in which the additive is a normal additive, particularly tri- cresyl phosphate. In carrying out the method of the invention samples of an engine oil containing tri- cresyl phosphate as an additive are obtained periodi¬ cally and preferably at regular intervals throughout the working life of the oil. The oil sample is analyzed to determine the concentration of tri-cresyl phosphate and the con¬ centration is compared with a reference value. The reference value may suitably be a predetermined value based on the normal levels of tri-cresyl phosphate observed in properly functioning engines or the level in unused or fresh oil. It will be recog¬ nized that the predetermined reference value may be associated with a particular oil or brand of oil and therefore there may be a number of reference values with the appropriate reference value being selected for the oil under investigation.

On the other hand, particularly in the field of aviation oils, there is a relatively small selection of oils available and the composition of such oils, including the tri-cresyl phosphate con¬ centration, is established within fairly narrow ranges by standards set by the International Civil

Aviation Organization ( CAO), which standards are followed internationally, similar standards are set by military establishments.

The analysis is suitably carried out using an infrared spectrometer, preferably employing fourier transform infrared spectroscopy and a spect¬ rum of the oil is developed. The spectrum is decon¬ volved using a second derivation spectrum. The predetermined reference value involves the similarly deconvolved spectrum derived from oils in properly functioning engines. The deconvolved spectrum of the oil sample under investigation is subtracted from that in the reference value and an integration is made of the difference in the spectral region corres- ponding to the tri-cresyl phosphate wave length or wave number. This integrated value is compared with a previously prepared calibration curve for the reference value, which calibration curve relates concentration and integrated value in the spectral region under consideration. In this way a comparison is made of the concentration of the tri-cresyl phosphate in the sample of oil under investigation and the normal concentration in use.

These operations can be performed in a computer system for which the algorithm is readily developed by routine experimentation. For example, a drop in the tri-cresyl phosphate concentration to a value of 80% or less, and especially a drop to 50-70%, of the normal value is indicative of potential engine failure, and the computer system produces an evaluation and issues a warning signal when a drop into this region occurs.

It is appropriate to maintain a history or record of the results of determination of tri-cresyl phosphate concentration of oil for each engine so that any change or initiation of drop in such phos- phate concentration can be more readily identified as an early warning of engine failure. This history can be compiled automatically in the computer by appro¬ priate software, so that a history of results for each engine is available to the algorithm programs which have the ability to evaluate significant changes in the additive level.

In particular the analysis is conducted so as to identify a statistically significant drop in the concentration of the additive under investi- gation, in the oil. This may be achieved by drawing the best line possible through a series of plotted results of additive concentration for normally functioning engines over a period of time; and determining the standard deviation and coefficient of correlation. The standard deviation may be deter¬ mined by drawing a perpendicular line from the afore-mentioned best line to each plotted value, determining the length of each perpendicular line, taking the square of each length, summing the square values and taking the square root of the summation. A drop in the concentration of the additive of statistical significance would be one having a given number of standard deviations, and is valid when the coefficient of correlation is close to unity.

Aviation oils contain 1 to 5%, usually 2 to 3%, by weight, of tri-cresyl phosphate. A decrease in the concentration of 50 to 80% of the normal value would represent a drop of greater than 2 to 3 times

the standard deviation. Determination of such a drop, in accordance with the invention, is an indication of a problem with the engine.

Thus by means of the invention an engine can be monitored and prediction of engine failure can be obtained well in advance of such failure relying on analysis of levels of normal additives present in the oil. The invention avoids the need to introduce foreign materials into the oil or engine parts for the sole purpose of analysis and can rely on deter¬ mination of materials routinely employed as additives in commercial oil formulations. On the other hand, it is feasible to practice the invention with inclu¬ sion of special additives in the oil which additives are added solely for the purpose of the invention. This can be achieved by incorporating in the oil a substance which does not interfere with the proper operation of the oil or with the functioning additives of the oil, and which will remain sub- stantially inert in the oil under normal conditions, but which will suffer a drop in concentration, for example, as a result of hot spots causing thermal degradation or decompositon, in the event of a problem developing in the engine, which problem ultimately will result in engine failure.

With further reference to Fig. 1, an apparatus or system 10 of the invention comprises a detector 12, a computer or computer network 14 and an output terminal 16. Computer network 14 includes a spectral storage system 18, a relational database 20, a deconvolver 22, a subtracter 24 and a comparator 26.

Computer network 14 additionally includes an evaluator or expert system 30.

An oil sample is designated 32; the straight lines with arrows indicate signal flow and the broken lines indicate information access.

Detector 12 is in particular an infrared spectrometer, more especially a fourier transform infrared spectrometer; the spectral storage system 18 contains infrared spectra of the oil in previously tested engines with identifying parameters for such engines; and relational data base 20 contains the references to the spectra of fresh oil, base oil spectra, data of oil additives and functions, spectral data of additives, including spectral regions, calibration models and historical norms.

With further reference to Fig. 2, there is illustrated in flow chart form an algorithm 28 for use in the invention.

In algorithm 28 a spectrum 34 of oil under analysis is deconvolved in step 36 and the deconvolved spectrum is subtracted from a known spectrum of unused oil of the same type and an integration is made of the difference in spectral region corresponding to the tri-cresyl phosphate wave length in step 38. The integrated value is compared in step 40 with a calibration curve from 42 and the result evaluated in step 44.

In operation a sample 32 of a known oil containing tri-cresyl phosphate is taken from an engine, and the sample is identified by reference to the specific engine from which it is taken and suitably by reference to other parameters such as the number of flight hours of the engine. The sample 32 is introduced into detector 12 which may be, for example, an infrared spectrometer, and a deter¬ mination of the tri-cresyl phosphate content of the sample 32 is made. In the case in which detector 12

is an infrared spectrometer, an infrared spectrum is developed at least for the regions in which the characteristic peaks for tri-cresyl phosphate occur. The resulting spectrum is fed into computer i4 where the algorithm of Fig. 2 is employed to successively deconvolve the spectrum, substract and integrate, and compare, as described hereinbefore.

The data for the comparison as well as for the deconvolution and subtraction and integration is taken from the reference data base 20.

The reference values in data base 20 for the comparison are representative of a normal or acceptable level of tri-cresyl phosphate in engine oil. The comparison is fed to evaluator 30 which issues a signal to ouput terminal 16. The evaluator 30 may, for example, be programmed to issue a warning signal if the evaluation shows an unacceptable deviation of greater than, for example, 2 times the standard deviation.

EXAMPLES Example 1

In a one case history a PW120 turboprop aircraft engine manufactured by Pratt & Whitney was operated continually using Castrol (Trade Mark) 5000 engine oil. For convenience this engine is identified as engine 293.

A total of 23 similar PW120 turboprop aircraft engines each identified by a similar 3 digit number were operated under similar conditions using the same engine oil.

Oil samples were taken from all the engines at periodic intervals during operation of the engines and an infrared analysis was taken to deter-

mine the tri-cresyl phosphate concentration of the oil for each engine, in accordance with the specific procedure described herein.

Table I below sets out the results of the analysis for engine 293 and comparison results for 7 of the other 23 engines. In each case the number of hours of operating time when the sample of oil was taken is indicated, together with a percentage value for the tri-cresyl phosphate remaining in the sample, this percentage value being a percentage of the normal concentration in the oil.

TABLE I ENGINE NO,

293 313 323 550

I

ENGINE NO.

830 839 850 555

The seven comparison engines for which results are shown in Table I are representative of the 23 engines and were selected for Table I to demonstrate results for a range of different flight hours. Thus engines 550, 555 and 830 were investi¬ gated through flight hours similar to engine 293, engine 839 and 850 demonstrate results for flight hours in the 2000 to 4000 hour range; and engines 313 and 323 demonstrate results for flight hours in the 6000 to 8000 hour range. Engine 830 additionally shows results first in the 2000 to 3000 and then in the 8000 to 10000 hour range.

The results for engine 293 indicated very low % remaining at the 9346 and 9997 flight hours analyses, i.e., 67.31% and 69.14%, respectively so that the oil was replaced by fresh oil. The engine failed following 9 flight hours after the 10254 flight hours oil sample was taken. The analysis after 10263 flight hours was thus carried out after failure of the engine.

During the course of these analyses con¬ ventional spectrometric analyses were conducted for the content in parts per million of iron, copper, lead, tin, aluminum, chromium, nickel, titanium, silver, magnesium, silicon, boron, sodium, barium, calcium, phosphorus, manganese, molybdenum, zinc and vanadium. In no case did the levels measured for these metals indicate abnormal contents such as would indicate approaching engine failure. These results demonstrate that an analysis of the oil for metals such as iron and copper, within nine flight hours of the failure would not have predicted imminent engine failure. On the

other hand the analysis at 9346 flight hours followed by oil replacement and the subsequent analysis hours at 9997 flight hours can be seen as having provided a warning of the approaching failure at 917 and 266 flight hours prior to failure, respectively. In each case a rapid drop can be seen in the concentration of tri-cresyl phosphate.

Upon disassembly of the engine following failure it was found that two lubrication nozzles supplying a bearing and gear assembly in the auxiliary drive unit were partially blocked. The balls and races of the bearing were deformed and scored and the pinion shaft could not be held in place. As a result the gears disengaged from one another and no fuel or oil could be supplied to the engine.

The traditional methods of filter analysis, chip detectors and atomic absorption spectroscopy (by the Strategic Oil Analysis programme SOAP) did not indicate any operating or mechanical abnormalities.

Following repair of the engine it can be seen that normal levels of tri-cresyl phosphate in the engine oil were maintained. As indicated, other engines in Table 1 are indicative of results for the 23 engines which functioned normally, and in all cases the lowest values recorded are still close to 90%.

In this case the detected low levels of tri-cresyl phosphate concentration preceded failure of the engine and thus provided a prediction of the subsequent engine failure.