Aeroengine Safety

Fig. "Informations of oil analysis" (Lit. 22.3.4-3,-4,-5, -9,-13,-24): The manifold causes and properties of oil contaminations and changes of the oil demand for proof and identification specific processes (Ill. 22.3.4-2 and Fig. "Paricle size and optimal analysis").
Frequently sufficient certain conclusions need several testing methods whose results are combined.

Conclusions at concerned parts: The identification of deteriorated parts before the catastrophic failing is of high importance. Crucial is the determination of the chemical composition from the particles. There are summeries for may types of aeroengines (Lit. 22.3.4-3), which correlate the analysis results with the parts in the oil system. So for example, silver particles hint at a coating/plating of bearing cages. The same is true for the abrasion of bronze. Many parts like race rings and rolling elements of anti friction bearings or gears consist of similar hardened steels. This makes an individual correlation only with the analysis difficult. Chips of light metal hint at a rub process against the wall of the casing from an oil flow through an acessory device (gear, pump, valve).

Conclusions at failure mechanisms: Information can give the outside appearance of a particle (Fig. "What deposits of magnet plugs can tell"). Suitable are as well optical-macroscopic (e.g., binocular) as also microscopic or electron microscopic (SEM = scanning electron microscopy) methods.The expert can identify from surface features, if a particle developed by a fatigue process. Typically the origins of those particles are pitch surfaces like races of anti friction bearings or gear tooth flanks. Fine, needle like chips are rather typical for rubbing processes (Fig. "Specifications for oil contaminations"). Tiny balls indicate a crack formation by fatigue (Fig. "Spherical oil contaminations").

Conclusions at the type and origin of foreign particles in the oil, which can arise from most different sources (Fig. "What deposits of magnet plugs can tell" and Fig. "Spherical oil contaminations").

• Production processes like shot peening, abrasive blasing, machining.
• Components from outside the oil system, like hard facings of labyrinth seals (tips), hairs of brush seals.
• Maintenance and assembly like excessive sealing compound, dust.
• Particles, sucked by the aeroengine, like dust and salt deposits.Appearance (macroscopic, microscopic), chemical composition, hardness (Fig. "Deterioration effect of particles in oil") and magnetic properties are also important features.

Conclusions at deteriorating influences: Concerned are changes of the oil by aging (acid number, Fig. "Influences at oil consumption", viscosity, Fig. "Influences at oil consumption", Fig. "Oil viscosity and lubrication by temperature" and Fig. "Formation conditions shown by coke deposits") and liquid respectively dissolved oil contaminations like water (Fig. "Problems by water in lubrication oil"). An example is the corrosion inside of gear casings from light metal.

Conclusions at timely developments and risks are essential for the decision for a further approach, respectively the introduction of targeted measures. They enable an evaluation of the risks. The lapse of time arises from changes of the particle amount during the control/maintenance intervals. The particle amount (number of particles = particle count, Fig. "Paricle size and optimal analysis") can be determined, depending from the magnet plug deposit, filter residue or floating in an oil sample (Fig. "Ol filter helping for diagnistics" and Fig. "Paricle size and optimal analysis").
The more independent test results are available the higher is the probability of success/safety from the measures.
Timely sequences are especially from interest for fatigue processes like fatigue pittings on the races of anti friction bearings or gear tooth flanks. They should permit an action in time. Also for wear processes like friction wear (fretting) at spline couplings (Fig. "Danger also without abrasion and chips") with a connection to the oil system, the chance of an identification in time exists.

Fig. "Suitable test methods" (Lit. 22.3.4-2 and Lit. 22.3.4-9): The analysis of contaminations, especially od particles in the oil, is of high importance for the identification of failures in time (Fig. "Wear particles point at problems"). With this targeted, suitable measures will be possible (Fig. "Decisions using magnetic plug depositions"). This minimises risks and costs, delay and safety problems. The chart contains typical tests/investigation methodes with characteristic application features. In the following, selected procedures/processes are described for the practitioner.

Spektrometric oil analysis allow also for a high sample number, a fast analysis of the chemical composition. Because it can be applied for small metallic particles (Fig. "Paricle size and optimal analysis"), the possibility of an early failure detection exists. Typical examples are fatigue failures and wear failures at anti friction bearings. This early detection is charcterised with the keyword Spectrometric Oil Analysis Program (SOAP, Lit. 22.3.4-14, Fig. "Material specific particle content in oil" and Ill. 25.2.2.2-7).

Fig. "Analytic ferrography samples" (Lit. 23.1-29): The ferrography requires a magnetic field during the oil flows over a small glass plate. This servs as sample carrier for a microscopic analysis. The paricles, selected with this device in size/weight and magnetism, enable the expert conclusions at the formatin/failure mechanism. To this belong sliding wear, wear by seizing/galling, rubbing, abrasion, corrosion and different forms of (vibration) fatigue (fatigue pittings). For example, there are besides other magnetic particles from wear or fatigue, small balls which can be related different failure mechanisms (Fig. "Spherical oil contaminations", Lit. 22.3.4-28).

Fig. "Paricle size and optimal analysis" (Lit. 22.3.4-9): The amount and size distribution of particles (particle count) can supplement the analysis for the evaluation of oil cleanliness and the risks<FONT COLOR=“#ffffff”> </FONT>(diagram below). For example, the demand for assuring samples can be justified (Lit. 22.3.4-16). Should the size distribution serve the assessment of unusual wear/particle formation, the chronological distribution, compared with the normal case, can be significant.
The method specific proof range of particles show the figures above.

Fig. "Specifications for oil contaminations" (Lit. 22.3.4-16): It is obvious, that OEMs and/or operators (e.g., military) edit aeroengine type specific informations. Those contain the material composition of the oil system components/parts. With this the particular causative components/parts can be brought faster and more certain in connection with the particles in the oil (example in chart below, see also Fig. "Metallic contaminations in oil show problems"). Such a program is the Joint Oil Analysis Program (JOAP) at army, navy and air force of the U.S armed forces. In this, in the internet available paper (Lit 22.3.4-16), helpful informations for a large number of aeroengine types, of which many are also in civil use/airliners, can be found. In the upper chart for an in former times very frequent used aeroengine type for airliners, decision support for assuring sampling is given.

Fig. "Tests for the footprint in oil" (Lit. 22.3.4-23): For oil analysis applies especiallv, that the quality of the result depends from every single step. This begins already with the sampling. Basically the instructions in the maintenance manual must be exactly kept, in case of doubt the OEM must be consulted. For the maintenance (dark areas) applies:

• The suitable extraction location for sampling must be planned already during the development of the aeroengine.
• Approved sample extraction according to the instructions in the maintenance manual.
• Use clean sample containers.
• Transfer of the oil samples to the investigation as fast as possible. At least the oil sample must reach the investigating laboratory in the specified time.

Also for the laboratory apply the basic rules of a „good housekeeping“.
From experts the influence of the cited rules at a suitable investigation result are higher estimated than the usual accuracy of the devices/equipment.

Fig. "Material specific particle content in oil" (Lit. 22.3.4-14): This diagrams show for 3 incidents/failures the chronological particle formation („footprint”) in the lubrication oil.

Example „1“(sketch and diagram): Concerned is a (sliding ring) face seal for the oil (Ill. 23.4.2.2-1) in a bearing chamber of an elder civil aeroengine type. Because of the contamination trend from thirteen succeeding samples, the exchange of the aeroengine was decided after a SOAP recommendation. During the disassembly arose, that two segments of the carbon seal ring (sliding against a surface from Cr-Ni-Steel?) had failed.

Example „2”: At an elder civil aeroengine type, after this trend the main oil pump was exchanged (like in example in the middle). The disassembly of the pump showed heavy wear of the bush bearings (Cu) from the gear wheels. Obviously the gear wheels had machined the pump casing (Al). A seal at the chromium plated drive shaft (Cr) had failed. This caused an oil loss of about 1 l/h.

Example „3“: After 8 succeding oil samples, this „footprint” of an elder civil aeroengine type lead to its exchange and disassembly. Cause for the particle formation was a failure of the anti friction bearing (Fe and Cr of the bearning steel) from the gear wheel shaft.

Fig. "Monitoring particle formation in oil" (Lit. 22.3.4-18, Lit. 22.3.4-20 and Lit. 22.3.4-27): This concerns the monitoring of the oil for magnetic particles. Applied is a continous operating magnetic sensor (quantitative debris monitor = QDM, sketch above left). Its position is located at the fan casing, directly behind the scavenge pump. It can send a warning signal dirctly to the cockpit. Additionally it can separate particles like an usual magnet plug and so gather for an investigation. Above this it is in the position to distinguish and separately count large and small particles (chips, Fig. "Comparison of sizes from particles in oil"). A warning signal can be identified by the mechanic from the outside. The well accessible position of the sensor at the aeroengine near the oil tank facilitates the control.
For the removing and check of the plug, the sensor is located at a suitable position inside the oil filter (detail below right). The rotating oil flow centrifuges sufficient big particles. These enter through an opening and so come in contact with the active face side of the cylindric sensor. During start and cruise 75 % of the particles above a certain mass are centrifuged/separated. This is a multitude of an usual magnet plug, which reaches only 5 %. During the removing of the sensor, a valve prevents the exit of oil. The example, displayed in the diagramm below left, shows a comparison of QDM sensor and SOAP monitoring. This concerns the increase of particles of the main bearing failure of a large civil fan engine. Of importance is the markedly sooner identification of the developing failure with the QDM sensor.
Naturally an early warning is only possible, if recordable paticles arise early enough. This is the case for the fatigue of pitch surfaces on bearing races and gear tooth flanks (Fig. "Fatigue pittings at bearings" and Fig. "Development of fatigue puittings"). However, during fractures of gear teeth frequently a crack respectively fracture starts from inside (chapter 23.2). In such a case an early warning can not be expected. There are so called on-line diagnostic sensors (ODM's) in the series introduction which monitor the oil for magnetic and nonmagnetic metallic particles (Fig. "Sensor in oil stream ODM for metallic particles" up to Fig. "Benefits by unloading of a rolling bearing").

Fig. "Monitoring by magnetic detector" (Lit. 22.3.4-22): The collected abrasion on the magnetic detector consists of ferromagnetic particles of diverse origins and formation mechanisms (Ill. 3.5-5). Differ the agglomerations from the assessment scale of the OEM in type and amount from the acceptable limits its, favorable to clear the further actions with a decision tree (Ill. 4.1-4). Frequently after a review or an overhaul the amount of abrasion is the highest, then declines significantly.
A rerise is, corresponding to the manufacturer instructions, to monitor attentively. For this a dokumentation like in Ill. 3.5-4 is helpful. Imprints of accetions secured with a special adhesive film. If not available, also a adhesive film from the household can suffice. The imprints will be collected on a sheet of paper with the necessary comments. These include the datum and the cause of the sampling, the operation hours and the identification of the magnetic detector if there are more. So the findings are available to possible later investigations. The paper with the collected abrasion films will be added to the documentation of the engine. With this it is possible to check the point of time at which a bigger failure developed. So the chance for conclusions at the sequence of the failure exists.

Fig. "Wear particles point at problems" (Lit. 22.3.4-1 and Lit. 22.3.4-15): An oil analysis can serve the identification and early detection of problems. This applies also, when fine abrasion/wear particles got into the oil circuit (size < 0,001 mm = ca. 1/50 of a hair daiamer!). These behave as floating material in the oil and pass the oil filters. Concentrations are in the range of millionth weight proportion (ppm). Typical metals as finest abrasion in the oil circuit are:

• Fe from steels of gears and bearings.
• Ni from Ni alloys of labyrinth tips.
• Cr from cromium plated seal sliding surfaces of bearing seats.
• Ti from fretting wear between steel and Ti alloys (e.g., bearing seats of shafts or casings from Ti alloys.
• Cu or brass from bearing cages (Fig. "Metallic contaminations in oil show problems").

It is important for the use of this monitoring possibility, that the OEM specifies informations from experience and limits. The two typical analysis methods are atomic absorption (AA) and optical emission spectrometry (Ill. 22.3.4-2.1 and Fig. "Paricle size and optimal analysis"). In both cases a little oil sample is required. For a successful analysis of These methods; the point of time of the sampling respectively the knowledge of the related operation hours is a requirement. Attention must be payed, that refuelling oil (“A”) and change of oil (“B”) influence the curve progression. A sudden climb of the curve (sampling “1”,“2”, “3”) must be observed especially attentive. If the trend does not recover to normal; a failure may be announced. Filter residues (Fig. "Ol filter helping for diagnistics") and magnet plug deposits (Fig. "What deposits of magnet plugs can tell" ) must be ananlyzed for a certain identification of the problem .
Location and period of the oil sampling is aeroengine specific. The test result and its reliability depends essentally from the sampling, according the specifications (Fig. "Tests for the footprint in oil"). ensuring oil samples usually are taken according to a decision systematic in the maintenance manual (Fig. "Decisions using magnetic plug depositions"). A schedular example shows Fig. "Specifications for oil contaminations".

Fig. "Decisions using magnetic plug depositions" (Lit. 22.3.4-16): It is to be about an example for a structured decision (decision tree). Even the deposits on the magnetic chip detectors (Ill. 22. 3.3.1-4) provide informations as about the condition of the engine and /or the trend of possible problems (Fig. "Monitoring by magnetic detector" and Fig. "Oil coaking endangers main bearings"), as does the filter residues. A combination with oil filter examinations (Ill. 22.33.1-2) can clearly improve the surety of statement and is, therefore, often demanded from the manufacture. Magnetic chip detectors, (Fig. "What deposits of magnet plugs can tell") that show, electrically, non permitted deposits, should be checked for damaged indicators.

Relative to the findings of the filter and the magnetic chip detector deposits, the represented decision tree is a help. Black areas show results, indicating that a further operation without manufacturer's recommendation is not advisable. Grey areas demand ensuring actions suggested by the producer. White areas allow unlimited use of the engine, according to the findings.

Fig. "Time limits of chip detector warning" (Lit. 22.3.4-12): Magnet plugs and magnet-Chip sensors are like other sensors absolutely not necessarily reliable (chapter 19.2.1, Fig. "Chance to identify a bearing failure"). To this also count failures in the evaluation of warning signals. In the chart typical failure causes of the magnet plug control are related to cases (see illustrations).

Fig. "Chance to identify a bearing failure" (Lit. 22.3.4-17): Maintenance technicians of this fighter aircraft overlooked the early warning sign of a bearing failure. Obviously this was announced by chip formation in the oil. Thereby with high likelihood, the fatigue failure at the race of an anti friction bearing was concerned. Such a failure annouces itself with increasing chip formation (fatigue pittings by break-outs, Ill.22.3.3.1-4). The failure started with a loud bang and severe vibrations of the aircraft. After this further bangs occurred. This hints at a compressor surge (volume 3, Ill. 11.2.1.3-1). It can be triggered by the bearing failure which decelerates (by rubbing?) the rotor. The aircraft crashed.
The investigation showed, that a main bearing (No. 3, location see detail) had failed. In the scavenge filters chips resided. Probably they have been overseen from the technicians during the filter control at the morning before the accident.

Fig. "Failing of oil monitoring" (Lit. 22.3.4-1 and Lit. 22.3.4-11): The oil system of the aeroengine is equipped with a magnetic chip detector (= MCD) in the casing of the propeller gear (reduction gearbox = RGB, see sketch of the engine). Concerned is a dipole magnet (Fig. "Monitoring particle formation in oil"). Magnetic chips from the oil, which bridged the ring gap between the poles, showed the pilot by means of a warning light a problem with the aeroengine.
The disassembly of the aeroengine after the accident unfolded a cage of the ball bearing No.1 (sketch), which has fractured into several pieces. So the balls could exit into the bearing chamber. Metal chips covered the wall of the accessory gearbox (= AGB). Chips also have been found in the oil filter and the propeller gear. However these are not sufficient to bridge the sensor to trigger a warnig signal. Further the flanges of the gear teeth in the AGB showed many fatigue out-breaks, which obviously started at craters of local overheating (electric sparking).
An investigation arose, that about 700 operation hours before the starter generator had failed. Obviously this caused a deteriorating sparking at the wheels of the AGB. During the prescribed 100h inspekction after the generator failure, the specified check (patchtest) of a filter sample was not carried out. During this the filter must be washed out with a resolvent. The washing fluid is separated from the particles with a little paper filter (patch). This test traces every material. So it is not limited at magnetic chips. Howevre with such a test, according the informations from the OEM, at this aeroengine type never a beginning anti friction bearing failure was found.
However a spectrometric oil analysis of the SOAP (Fig. "Suitable test methods") took place. Even this showed about 30 operation hours before the accident no alarming trend.
To this the OEM explained, that for the concerned aeroengine type this investigation has already proven as not reliable to detect beginning failures of antiim friction bearings. The bearing failure could be interpreted as a result of chip formation from the gear failures. For the unsufficient display of the MCD the following plausible explanation was found:
The MCD does not monitor the scavenge oil from the AGB and the main bearings. From the AGB the oil is pumped back into the tank. From here the oil gets through the main filter into the fresh oil circuit. Thereby all dangerous particles are withholded. So the MCD in the propeller gear, which is located within the front no chips from the AGB can be recorded. The pilot gets no warning signal. So it emerged, that the oil monitoring with the built-in MCD system is unsuitable, especially for a single-engined aircraft.

Fig. "Chip detector warning signal" (Lit. 22.3.4-19): The warning light for the chips in the oil flashed several times. However the pilot continued the flight after a short intremediate stop and check of the magnetic sensor. Shortly after this the aeroengine failed.
After the accident at the OEM in presence with a authority representative, the gear and the bearing chamber at the compressor entrance (bearing No. 1) have been investigated. A function test with a following inspection of the gear components showed no anomalies.
In contrast the bearing No. 1 showed wear damages at the cage and the rollers. Additionally two cage fillets missed. This was interpreted as secondary failure. A causative material failure or shortage of the oil flow was not found.
The maintenance manual of the concerned aeroengine type gives exact instructions, how to approach in case of a display of the chip warning light in the cockpit:
Inspection of the magnetic chip sensors as soon as possible. Thereby the findings must be evaluated as follows (see also Fig. "Decisions using magnetic plug depositions"):

• Magnetic particles (Fig. "Ol filter helping for diagnistics" and Fig. "What deposits of magnet plugs can tell") like small fragments (debris), splints (slivers) as well as break-outs (chips with features of a fracture) are possible indications of a bearing failure or gear failure and/or unusual wear in the aeroengine.
• Break-outs or flakes larger than 0,8 mm or more than 4 chips at a warning display are forbidden. This engine must be send to an repair shop, approved by the OEM.
• Break-outs or flakes smaller than 0,8 mm or less than 4 chips at a warning display are acceptable under requirements. This also applies for fine accumulations of abrasion/wear (fuzz).

After the reassembly of the magnetic sensor, according to the maintenance manual (operation and maintenance -O&M- manual) the following tasks must be carried out:

• After 30 minutes run with power output without warning light display, the magnet sensor must be checked for chip accumulation. If the warning light during the test run flashs, the aeroengine must be dismounted.
  Is the indication, according the criteria mentioned before acceptable, the sensor must be cleaned and reassembled. After this the aeroengine can get into the air traffic.
  Does the warning light flash up again during the next 8 operation hours and indicates the magnet sensor particle accumulations, the aeroengine must be dismounted.

Summarised conclusions:

• The outage of the aeroengine is traced back at a wear failure fot the cage and rollers from the bearing at the compressor side.
• The pilot classified the engine misleadingly as airworthy, although during few hours the warning light two times flashed
• The pilot was not informed about the meaning of repeated warning signs and the then necessary test run. He also got no formal training about the inspektion of the magnet sensor. However this is demanded in the manual of the operator (maintenance control manual).
• The flight operations manual of the operator says nothing about such a situation.
• The flight manual of the concerned helicopter type allows, if there are warning displays, the interpretation of measures as less urgent.