INTRODUCTION
Historically, the use of atomic emission spectroscopic instruments as a condition monitoring technique for oil wear debris monitoring of reciprocating engines such as diesels originated in the railroad industry in the mid 1940’s with the advent of diesel engines to power locomotives [1]. In the mid 1950’s, the US Navy began to evaluate the use of similar spectroscopic techniques for its aircraft engines which, at that time, included gas turbines but predominantly reciprocating engines [2]. In 1961, a progress report on the spectrographic oil analysis program summarized its success for identifying reciprocating engine problems; however, it could not comment on its usefulness regarding the identification of gas turbine engine problems, since the engines being monitored had not been plagued by oil wetted component problems.
In 1968, the US Navy reported again on the spectrometric oil analysis (SOA) program for oil wear debris monitoring of various aircraft jet engines, reciprocating engines, transmissions, and gearboxes [3]. The program had been successful at identifying discrepancies in all of these types of equipment in service up to that time. The report mentioned that the US Army and US Air Force began their own program in the early 1960’s based upon the success achieved by the US Navy.
In the late 1960’s, although the US military was now using atomic emission spectroscopy (AES) and magnetic chip detector (MCD) technologies to detect oil wear debris particles in engine lubricating oils, the US military was also experiencing aircraft gas turbine engine failures due to surface fatigue in rolling element bearings which were not being detected soon enough prior to failure [4]. It was found that AES and MCD technologies did not detect the onset and progression of the wear debris important for certain types of failure modes such as surface fatigue. Therefore, by the time AES and MCD provided any significant indication of a problem, the aircraft gas turbine engine bearing damage had already progressed to an impending catastrophic failure mode state.
Scanning Electron Microscopes have been commonly used since the 1970’s in identifying the larger wear debris particles trapped in filters, screens or MCDs which was found to be useful in identifying the source of the debris. However, given the late notice of an impending failure from MCDs, this method still posed a risk of events going undetected. Historically, SEMs have been used for manual imaging/chip analysis. However, recent technology advancements to SEMs have made them capable at automatically analyzing wear debris in an industrial setting [5]. [C1] [SD2] This advancement has now enabled alternative oil analysis methods to be possible by filtering the oil sample and characterizing the wear particles. This approach improves the sensitivity of the oil analysis on the order of 500X compared to traditional methods with the added benefit of characterizing each wear particle individually on the prepared patch.
In this study, we will compare the effectiveness of traditional SOA methods to this alternative oil analysis technique for providing condition monitoring of gas turbine engines.
- Method
The sample preparation for oil analysis per AES is well documented: a sample is oil is taken from the engine. After dilution with a solvent, the sample is introduced into the AES source (mainly plasma) after being nebulized. The nebulization process is critical and a high percentage of non-dissolved particles are removed [6]. It is recognized that the oil spectrometric analysis detection efficiency decreases as the sizes of particles increase [7][8].
An alternative oil analysis has been developed to predict failure [9]. An oil sample is taken from the engine according to specific instruction (location, time after engine operation…). Particles are then extracted from a known volume of oil by filtration on a media that will capture all particles of interest. The volume used can vary according to the nature of the mechanisms that are expected to be found. The size of the filter patch and the volume of solution needed have to be suitably selected to avoid overlapping of particles.
A portion of the filter is then analyzed using SEM. The larger the portion, the better is the accuracy of the analysis but the longer is the analysis. Evaluations need to be performed to determine the smallest volume that can be analyzed without losing key identifiers to optimize the analysis process. Considering that all wear particles collected in the filter patch are also included in the area analyzed, the efficiency of the method is not decreasing with particle size.
After SEM analysis, the specific chemical composition data obtained for particles are processed through software algorithms in order to classify them in specific alloy categories or material classes. Unclassified particles having a mix of elements are
[C1]Not sure that I will made a reference to this paper written by Aspex, which is in my opinin a violation of the non-disclosure agreement we have with Aspex.
[SD2]We changed the reference to Bill Hergurth paper in Journal of ASTM. Let me know if you are ok with this one