The analysis of used lubricating oil as an Engine Health Monitoring (EHM) technique has found widespread acceptance. The fundamental principle behind this is that oil is considered to be an effective carrier of information of wear particles, dirt and other oil breakdown products. In the early days, the physical properties of lubricating oils were considered good indicators of component health but over the years technology and experience have shown that the analysis of particles carried by the oil gave a better representation of the engine health. The particles in suspension in the liquid oil, wear particles, could be analyzed further to quantify the health of the component they originated from and predict any impending failures. Condition monitoring of Gas Turbines by oil analysis can be broken down into three main categories. These are: 1. Oil /Wear Debris monitoring 2. Oil Condition monitoring 3. Oil system operation monitoring The groups are further divided into two more sections, namely on-line and off-line systems. On-line being constantly monitored systems, which would alert the operator of any no go situations and off-line involving diagnosis during scheduled maintenance checks. OIL/WEAR DEBRIS MONITORING: Gas turbines incorporate bearings which operate under extremely severe conditions of load, shaft speed and oil temperature. Under these conditions, bearing damage can progress undetected until the occurrence of catastrophic engine failure. The challenge is to detect bearing failures early so as to prevent costly secondary damage to the Gas Turbine engine and also to be able to cater for procurement of a replacement engine. These failures can be predicted only by the use of early detection methodologies. MONITORING OF BEARING CONDITION : The starting point in the process of considering techniques for early detection of rolling element bearing failures is to understand the underlying reasons for bearing failure. Since the design of bearings is such that contact stresses are maintained below the fatigue limit of the materials, bearing failure should, in theory, not occur. However, other factors in a real application suggest otherwise. The frequency of mechanical bearing failure for rolling element bearings in general, is classified by cause in Table 1. Note that classical fatigue failure (cracks initiated at surface or sub-surface) constitutes only 2% of the total. Close to 90% of the occurrences are due to either corrosion, overrolling of debris (solids contamination in the oil), or dimensional discrepancies (misalignment, or manufacturing error, etc.). Research has indicated that failure due to corrosion is more common in engines that are subject to stop-start operating cycles versus continuous operation, while engines that are subjected to more frequent disassembly and certain maintenance activities are more susceptible to dimensional discrepancy and debris overrolling failures. In fact, the cause of the failure is not particularly significant if one considers that the mechanical failure of the rolling element bearing is ultimately associated with fatigue at the interface between the rolling elements and the race. The nature of the failure, regardless of the initiation mechanism, is a surface spall. Furthermore, once the spall appears on the bearing surface, the combined effects of stress fatigue and temperature inevitably result in the growth of the spall towards catastrophic failure, if the bearing remains in operation in an engine.
Table 1: Engine Rolling Element Bearing Failures In a gas turbine engine the bearings serve the critical function of supporting and controlling the position of the shaft and associated rotor. In modern high performance engines, the clearance between the blades, the casing and stators is kept at a minimum under all operating conditions. It follows that the degradation of the contact surfaces within the bearing can eventually lead to degraded shaft position control which in turn, can cause secondary damage within the engine when blade contact takes place. Once a spall has been initiated, the primary factors which influence the time remaining to complete failure are the speed of the bearing and its load. Depending upon the combined severity of speed (commonly represented by the expression dN, where d is the bearing diameter and N, the rotation speed) and load, the progression of damage to serious failure ranges from 10s to 100s of hours. The monitoring challenge presented is two fold. Firstly, to provide early detection it is necessary to be able to detect the initiation of the surface spall. Secondly, in order to prevent secondary damage to engine, it is necessary to monitor and quantify the degree of damage as it progresses and compare this against predetermined operational limits. OIL DEBRIS MONITORING The basis of oil condition monitoring as applied to rolling element bearing failure detection is that a failing bearing generates metallic debris in sufficient detectable quantities and that the debris is washed into the lubricating oil system downstream of the bearings. Without exception, both the presence and detectability of debris is confirmed in operational engines by the fact that debris is always found to accumulate in the downstream strainers and filters of the lubrication system. In some circumstances, it is possible that some debris may not make its way all the way to the filter element due to the specific flow traps inherent in any lubrication system. In practice, however, it has been found that the quantity of trapped debris is a very small fraction of the total debris that is found in the filter. It is also interesting to consider the special circumstance of the intershaft bearing design that is used on several modern engines. In such applications, tests have shown that due to the centrifugal force effects of the rotation of both the inner and outer bearing races, that some of the larger debris becomes trapped within the bearing housing, and is only expelled during low speed transients such as during start up or shut down. Tests have demonstrated that even for these designs detectable quantities of debris are still flushed out of the bearing cavity, even at spall initiation, though, in somewhat smaller quantities. FAILURE CRITERION Research programs over the last few years have led to three significant conclusions. 1. Firstly, oil debris monitoring can be effectively applied to detect the on-set of damage on rolling element bearings. 2. Secondly, that significant quantities of what are considered in the industry to be very large particles (>250 microns) are generated even at the initiation of bearing failure (tens of particles for spall initiation versus many thousands of particles for a damaged bearing). 3. Thirdly, total debris quantity (either particle counts or mass) are effective measures of the degree of damage that has occurred on a bearing. Research has also provided valuable insight into establishing alarm limits for engines based on the specific rolling element bearings used in the engines. Based on the debris data, the industry has developed a methodology to estimate for most rolling element bearing types and size, the relationship between the detected debris and the size of the spall (on the bearing race). From the estimate of the damage to the bearing, it is then straight forward to establish alarm limits, which, once exceeded, suggest the risk of secondary damage to the engine. One approach that has shown promise is to establish the alarm level at the debris quantity equivalent to the damaged area that is just large enough to accommodate two rolling elements (balls or rollers). At this point, one can conservatively estimate that there is some risk of loss of position of the shaft and thus, potential for secondary damage. Obviously availability of new data from field experience and research will subject the methodology to continuous improvement. Over the life of a particular component in a gas turbine, the rate of wear can be described as wear in normal operational and abnormal wear leading to failure, shown in figure 1. Abnormal wear is illustrated by a sharp increase in the production rate as shown in figure 2.
Figure 1: Normal Operation Wear History
Figure 2: Abnormal Wear Though useful as an indication of the general health of the engine, the production rate graph does not give any information as to the physical properties of the particles whereas it is important to study the particle size and shape. Particles generated by different wear mechanisms have different characteristics and these serve as a means of identification of the mechanism. Rubbing or adhesive wear particles have the form of platelets and are indicative of normal permissible wear. Cutting or abrasive wear particles take the form of miniature spirals and loops and these are indicative of a serious abrasive wear process. Particles consisting of compound can result from oxidizing or corrosive environment. Steel spherical particles are an indication of fatigue crack propagation in rolling contacts. In summary one can say that the following are characteristic that can determine the technique applied to test for wear. Quantity and concentration Size and distribution Composition and elemental analysis Load, Speed W e a r
Mild Wear Severe Wear Time W e a r
R a t e
Run In Steady Wear Wear Out Morphology DEBRIS MONITORING METHODS: The common techniques and devices employed for this category include spectrometric oil analysis program (SOAP) which in turn includes atomic emission spectroscopy and atomic absorption spectroscopy. The other methods are ferrography, magnetic chip detector, magnetic plugs and microscopic analysis of filter debris. It must be pointed out that no single method of these is suitable for a complete analysis and one would require using all the techniques for accuracy. This is mainly because of the capability of these methods for detection of particles of different sizes. Figure (3) shows the particle size versus detection efficiency for them:
Figure 3: Particle Size range Efficiency Of Different Techniques Of Wear Debris Analysis 1. Spectrometric oil analysis program (SOAP) is generally applied to particles of less than 10 micrometer in size. The name suggests that oil is analyzed when it is the oil wear debris and not the oil itself. This was first used by the American railways in 1954 and is now the most widely used technique. It provides a quantitative, multi- elemental analysis. The elemental concentrations of as many as twenty elements are reported in parts per million. Wear elements such as iron, aluminium, chromium, copper, tin, lead, silver, titanium, and nickel are detectable, as well as lubricant additives such as calcium, barium, zinc, phosphorus, magnesium, boron, and molybdenum. Trends are used to determine the mechanical health of a system. Principle: Depending on the energy state, metallic atoms and ions absorb or emit characteristics light spectra. Since the spectra of each metal are unique the spectral wavelength will identify a metal and the intensity of absorption or emission will show the quantity of the metal present. The two main instruments used to measure the concentrations of metallic elements in SOAP are Atomic emission spectrograph: Essentially it consists of a grating or a prism to separate the wavelength of radiation of the flame, exit slit, photoelectric system to detect and measure the wavelength of radiation. A schematic layout of the equipment and the system is presented in figure 4:
Figure 4: Atomic Emission Spectroscopy In this a very high voltage excites the debris particles, and their characteristic radiation is spectrally analysed. This is a costly method and is generally used in the off-line mode at a central laboratory. However it has the advantage of a fast automatic read out of results when integrated with modern data procuring devices. A technique similar to the emission spectroscopy is the XRF spectroscopy. The difference is that XRF excites the atoms with a bombardment of X-rays. In return, the atoms emit x-rays characteristic of the chemical element at an amplitude that corresponds to the mass of the chemical element in the sample Atomic Absorption spectroscopy: This consists of a hollow cathode discharge lamp as a source to emit light characteristics of the element, an air-acetylene or nitrous oxide flame source to transport and atomize metallic elements. In the sample a grating or prism and optic system (a lens and slit) to serve as a wavelength selector, a photomultiplier and amplifier circuitry to convert light energy to an electric signal and readout device to measure the electrical signals quantitatively. A schematic layout is shown in figure 5:
Figure 5: Atomic Absorption Spectroscopy
This process is relatively cheaper and less time consuming than the former, but only the concentration of one element can be determined at a time. Another method that has been studied is the use of a plasma arc to excite the sample. This is more costly than the previous two methods, however its results are quantitatively superior. This process requires the sample to be filtered or centrifuged to remove large particles in order to get more consistent results. This is because large particles of wear debris are not uniformly dispersed in the oil system. So even if the overall concentrations are small but the presence of large particles can make a major difference to the result. Based on the results of the spectroscopy, the evaluation process can either be manual or computer assisted. It is based on the wear metal guidelines for a particular piece of equipment, although other factors such as operating conditions, environment and the time operated since last oil change are also important. Since the spectrometer is able to distinguish between the different metals, it can indicate which metals make up the total wear content in the oil. For example if only iron and aluminium are present in abnormal amounts, the analysts job becomes easier as the entire system does not need to be stripped and only those components that have a large amount of these metals need to be inspected. 2. Ferrography appeared as a method for wear particle analysis in the early 1970s. This technique is particularly suited for the analysis of particles in the lubricating oil system. Its effectiveness in determining the deterioration of turbine bearings has made it widely acceptable for the aero, the marine, and the industrial gas turbines. The main types of ferrography are: Analytical Ferrography Direct reading Ferrography On-line Ferrography
Analytical Ferrography: This is a form of technique where ferrous particles are magnetically collected from a sample of oil and deposited onto a thin microscope slide. The ferrogram is then analyzed by optical or electron microscopy. The principle together with a ferrogram is illustrated in the figure 6 and 7 below:
Figure 6: A Typical Ferrogram
Figure 7: The Analytic Ferrograph Principle Although ferrrography is designed primarily for steel particles, some non-ferrous particles are also included as they pick up small inclusions of steel and become weakly magnetic and also get trapped in some of the steel particles. The presence of these non-ferrous particles can be distinguished from the metallic particles by the manner in which they reflect or transmit light.
Direct Reading Ferrography: The direct reading ferrography (typical system shown in figure 8) is very similar analytical ferrography in that it uses a magnetic field to collect ferrous wear particles from a sample of lubricating oil. However unlike the latter, in which case the particle morphology can be studied under a microscope, the direct reading instrument quantifies only the amount and approximate size distribution of the wear particles.
Figure 8: The Direct Reading Ferrography Compared with the analytical technique the direct reading ferrography is cheaper and can be operated by less skilled personnel. A good technique is to use the direct reading ferrography to monitor a severity of wear index and when it becomes large, then apply analytical ferrography. Temper colours: The use of temper colours as a means of extracting further information from a ferrogram is the latest advancement in ferrographic technique. A simple heat treatment of ferrograms in a laboratory hotplate can provide useful means of identifying the different materials present. The magnetic particles used on iron or nickel alloys generate temper colours that can be used to divide them into four types- carbon steels, cast iron, nickel and stainless steel. The majority of non- magnetic engineering alloys such as aluminium, magnesium, chromium, cadmium and silver do not form colours under this procedure. The copper-based alloys may form temper colours, but can be readily identified by their yellow bronze colour prior to the heat treatment. The procedure involves heating the ferrogram for 90 sec on a laboratory hot plate and photograph the large particles in the entry zone on cooling. A four-stage process is followed. First the ferrogram is heated to 330 0
C, the carbon steel then turns blue and cast iron a straw to bronze colour, with no change to nickel or stainless steel. In the second stage the temperature is raised to 400 0 C. The result is a light grey colour of the carbon steels, deep bronze with some mottled blueing of the cast iron with no change to nickel. A further heating to 480 0 C leaves cast iron and carbon steel to grey, the nickel turns to bronze colour with some blueing while the stainless steel turns straw to bronze colour On-line Ferrography: This is installed on the engine or machine and its output is in an electrical form. The main aim is to protect the apparatus or the system from deterioration due to excessive debris wear with an automatic safety arrangement. Instead of taking fixed volume of sample and determining its content of debris the online system continues to pass fluid until a prescribed amount of debris has been precipitated. A capacitance sensor measures this and depending on the concentration of the debris the measuring cycle can vary from 30 seconds to 30 minutes. Ferrography is usually applied to particles of sizes from 1-100 micrometer. 3. Magnetic chip detectors and magnetic plugs: These have proved to be a very reliable indicator of incipient failure of gas turbines. These are capable of detecting particles of sizes 50-1000 micrometer. The magnetic plugs consist of a magnetic stub, which is introduced into the oil flow that continuously attracts magnetic material; the debris is then removed and examined. The magnetic chip detector which consists of a set of magnetic electrodes that act like a switch and activate a chip warning light when a metallic wear particle comes in contact with them, have been found to be more effective and reliable than the plugs. The major drawback of the chip detectors is the number of false alarms indicated. This problem can be resolved by sending a strong current pulse through the chip gap every time there is electrical continuity. The pulse is generated in a small capacitor network which can be located in the chip detector or elsewhere. It causes local melting of the kind of particles that are responsible for these indications which interrupt the current path. Debris particles with substantial cross section conduct the current pulses unharmed and the continuity remains. The operator can initiate the current pulse discharge automatically or manually. In conventional usage, these magnetic chip detectors, have poor capture efficiency, and thus, are often not effective at providing early detection of bearing damage. Moreover, they are not capable of capturing non-ferromagnetic metal debris and are prone to false alarms due to the build up of fine ferromagnetic debris or fuzz.
Figure 9: Chip Detector Operating Principle
NEW TECHNOLOGY : Most of the oil analysis methods indicated upto now are used for ferrous particles and also there are problems of poor efficiency of magnetic plugs, inability to develop a trend of the number of particles etc. Improvements due to demand from the industry has led to newer techniques in order to overcome these drawbacks. One such device is a through flow sensor for monitoring the debris. Such a device can be installed in-line with the bearing oil return lines and allows the entire oil flow to pass without restriction.
Fig. 10 Through flow debris monitoring device
Fig. 11 Sensor Cross Section The sensor functions by monitoring the disturbance to an alternating magnetic field caused by the passage of a metallic particle through the sensing coil assembly. The particle couples with the magnetic field to varying degrees as it traverses the sensing region, resulting in a characteristic output signature as shown in Fig. 12.
Fig 12 - characteristic output signature
The amplitude and phase of the output signature is used to identify the size and nature of the particle. The amplitude of the signal is proportional to the mass of the particle for ferromagnetic metals and to the surface area of the particle for conductive non- ferromagnetic metals. The phase of the signal for non-ferromagnetic metals is opposite to that of ferromagnetic metals allowing a distinction to be made between the types of wear particle metal. Signal conditioning using a threshold algorithm is used to categorize the particles that pass on the basis of size. Multiple size categories can be configured which allows the tracking of the distribution of debris size. The minimum size particle detectable by the sensor is a function of the sensor size, which is typically around 70 microns. Research and field experiences with the sensor have led to development of setting of warning and alarm limits for the LM 2500 and the industrial RB 211 as shown in table 2 below:
Table 2: Typical operating Limits for the GE LM2500 and Rolls Royce RB211 Engines
Research carried out using this sensor has led to the following conclusions: 1. An aero-derivative gas turbine with bearings in good condition with a well maintained oil system will not produce debris in the oil that is detectable by the device. In actual field applications it is common that after a new engine or newly overhauled engine is installed, a small quantity of debris is detected over the first few hours of operation. The rate of accumulation is usually very low and goes to zero shortly after initial startup. This debris is the result of engine build debris, or debris left over from a previous engine failure event and is usually insignificant relative to the alarm levels. Common practice is to zero out the accumulated debris after a few hours of running after the first engine start. 2. The time from initial spall to significant damage varies significantly and is dependent upon engine type, specific bearing design and operating profile of the engine. An example of a particularly rapid bearing failure on a P&W FT8 engine is shown in Fig. 13. In this case the initial spall progressed to the alarm limit value in about 80 hours of continuous engine operation. A less rapid bearing failure event on a GE LM1600 engine is presented in Fig. 14, in which case the initial spall to alarm period took about 300 hours (or 12 days) of continuous engine operation. The throughflow oil debris monitor is effective at identifying other engine component problems including gear failure, spline failure, shaft/seal rub and erosion. In some cases these failures have their own identifiable debris signatures which may be characterized by particle size distribution and/or particle type composition
Fig. 13 Failure progression in the P&W FT8
Fig. 14: Failure progression in the GE LM1600 OIL CONDITION MONITORING: In the gas turbine engines the deterioration rate of the physical and chemical properties of oil is dependent on the aeration, oil consumption, oil system capacity and oil formation. A change in the engine operating condition resulting in increased aeration or higher oil temperatures can lead to an increasing rate of oil degradation and/or consumption. Tests for oxidation, additive depletion, solids content, fuel dilution, viscosity and total acid number are performed to ascertain the lubricant serviceability. This is an off-line process in which samples of oil are withdrawn at prescribed intervals and tested either on the spot or in the laboratory. Spot checks are usually carried out in the field and consist of: Blotting paper test This is a visual check of the clarity and the condition of the oil. Degraded oil or one containing dark contaminants will give a dark brown spot or ring when a drop of oil is placed on a blotting paper. New oil without any insoluble additives will give a uniform pale yellow spot. Capacitance test It determines contamination in the oil. The test involves measurement of the capacitance as a result of static electric charge generated between two plates immersed in used oil. The presence of water, methane debris or acidic oxidation products will reduce the capacitance. Viscosity Test Typically a decrease in viscosity would indicate fuel dilution and an increase, an indication of contamination. Detailed tests can be carried out in the laboratory and physical and chemical changes in the oil composition can be determined. These tests are: Emission Spectroanalysis: This uses high-energy excitation to determine the concentrations of submicroscopic wear metal particles in the oil. Part-per-million concentrations can be measured of such wear metals such as iron, aluminium, copper, silver, chromium, magnesium, nickel, lead, tin and titanium. Other elements like sodium, barium, and zinc, which derive from oil, are also measured. Infrared spectroanalysis: this measures the contamination and degradation of the oil. Difference between the infrared spectra of the sample and an unused reference sample of the same oil represents contamination or chemical change in the used oil. Contaminants commonly detected by this technique include water, blow-by products, unburnt fuel etc. Degradation of oil through nitration and oxidation, polymerization can be measured directly. Viscosity Analysis: Viscosity readings are obtained by an oscillation viscometer at ambient temperature. These are then converted to the required standard. OIL SYSTEM OPERATION MONITORING: Oil system operation monitoring is the third classification for oil monitoring. This is the oldest technique and involves monitoring of temperature, pressure and oil quantity among others. This can also be considered as mechanical parameter for trending and will help in giving an idea of an impending failure. One test carried out for the LM 2500 marine engine is that the engine is started at idle and accelerated in increments to full power. How the individual parameters recorded behave as the engine goes to full power largely determines the direction of the troubleshooting. The parameters (such as compressor discharge pressure) taken during the test can also be compared to engine performance model predictions to determine if an engine performance or a controls issue is the cause of the problem. For example a rapid rise in the scavenge temperature as the engine nears full power has shown that there is a problem of flooding of the accessory gearbox caused by the inability of the scavenge element to keep up with the amount of oil entering the accessory gearbox
LUBRICATING OIL ANALYSIS INDIAN NAVY: Though the use of lubricating oil analysis as an ECM technique for the Gas turbines has been going on for quite a while, but it is now felt that a fresh look has to be taken so as to get the best out of such a tool. The reasons for this fresh approach are: 1. Advances in such technologies as a tool for predicting failures have made it possible to predict quite precisely the likely remaining life of the Gas turbine once failure has been initiated. Such techniques have also made detection of such failure initiation quite accurate. 2. Even the most advanced Gas Turbines such as the the LM 2500 have been able to achieve a mean demonstrated life of around 16,000 hrs as against 25,000 hrs as quoted by the manufacturers. With the navy going for such advanced engines and also the use of higher life engines on the Talwar class and the Delhi class of ships, it becomes all the more important to use techniques that give a better indication of likely failures. 3. The move towards a Blue Water Navy will put pressures to improve EHM techniques so as not to have mid ops cycle failures. 4. Cost pressures due to reducing budgetary allocations will make it all the more important that we manage our resources to the optimum and the earlier we start the better. Techniques used for oil analysis in the Navy are mainly monitoring of temperatures, pressures, viscosity, SOAP, magnetic plugs and magnetic chip detector. Of these the most important ones i.e. debris monitoring tools have not been able to serve the purpose they are meant to. As mentioned earlier on, Magnetic plugs/chip detectors are not a very useful tool, as these tend to give warnings, when it is too late and bearings etc. are in a very advanced stage of failure. The application of SOAP also has not helped matters much due to the following: 1. Oil for analysis is used from the circulating tank i.e after the filters, which tend to take up most of the particles and also some of the particles get stuck in the flow traps. 2. Meaningful analysis is not possible without taking into account the addition of oil since the last test. 3. SOAP is useful for debris of around 1-5 micrometer, whereas for failure analysis it is better to have a technique that detects debris of around 50-60 micrometer. Implementation for the Navy could include Ferrography together with temper colours. This would definitely be more useful than SOAP as its detection capability in terms of size is in the region where failure takes off from. The disadvantages of SOAP as indicated above namely not taking additional oil into account and where to take the oil from would still persist, but this could be overcome by use of an online system. The use of a Through Flow Device would necessarily overcome all these drawbacks and when implemented with a PC based system would be able to warn the user of any likely problem. Data from ships could also be transferred to GTTT for detailed analysis and obviously more field experience will help in varying the alarm limits and also help in predicting likely remaining life. Additional points to be taken care of for using through flow devices is the likely pressure drop through such devices. Implementation of a lubricating oil analysis technique in the Navy should be based on a cost benefit analysis. Capital ships that have longer life of the Gas Turbines and need to stay at sea for extended deployments need to have an online analysis system and specific recommendations are: 1. Use of Through Flow Devices or 2. On line ferrography Ships that have Gas Turbines with limited running hours such as the 1241 RE need to have a system using Through Flow Devices. Detailed analysis of oil could be carried out in an oil analysis laboratory.
REFERENCES AND FURTHER READING: 1. BRUCE D. THOMPSON, BEN WAINSCOTT. SYSTEMATIC EVALUATION OF US NAVY LM2500 GAS TURBINE CONDITION. ASME 2000-GT-667. 2. DUKA KITALJEVICH, GERRIT J. VAN VELDHUIZEN. ADVANCED OIL DEBRIS MONITORING FOR PIPELINE MECHANICAL DRIVE GAS TURBINES. ASME 2000-GT-0347. 3. GREITZER, F. L., L. J. KANGAS, K. M. TERRONES, M. A. MAYNARD, B. W. WILSON, R. A. PAWLOWSKI, D. R. SISK, AND N. B. BROWN. GAS TURBINE ENGINE HEALTH MONITORING AND PROGNOSTICS. SOLE 99 SYMPOSIUM, AUGUST 31- SEPTEMBER 2, 1999, LAS VEGAS, NEVADA 4. ANDERSON, DANIEL, ROTRODE FILTER SPECTROSCOPY, PPM MAGAZINE, 1992 5. HUMPHREY, GARY R., "CHARACTERIZATION OF DEBRIS FROM F404 ENGINE OIL FILTERS BY ENERGY DISPERSIVE X-RAY FLUORESCENCE", JOINT OIL ANALYSIS PROGRAM TECHNICAL SUPPORT CENTER, JOAP- TSC-TR-96-02, JUNE 1996. 6. WHITLOCK, R., CHURCHILL D. AND G. HUMPHREY "THE PATH TO AFFORDABLE LONG TERM FAILURE WARNING: THE XRF-WEAR MONITOR," PROCEEDINGS OF THE JOAP INTERNATIONAL CINDITION MONITORING CONFERENCE, APRIL 19-24, 1998, MOBILE, ALABAMA. 7. FITCH, J. C. OIL ANALYSIS FOR MAINTENANCE PROFESSIONALS (COURSEBOOK), NORIA CORPORATION, 1998.