Industrial Gas Turbines

"Illustration 0-1": After a failure the question after prevention respectively remedies arises. Those are usually prepared/developed by the OEM and authorised from the agency in charge. The assignment occurres in the respectively suitable form conform to the mandatory procedures.

The picture shows on typical measures how it is possible to minimise the risk of failures. So especially in many cases so a further operation or short time recommissioning is possible. Measures first are based on instructions, respectively overhaul and maintenance manuals. Ar the definitions and descriptions of toöeerable flaws respectively specific approaches unsatisfactorily the OEM must be consulted.

Supervisory measures come into consideration, when the failure did not yet lead to impermissible consequences or an unbearable risk. Examples are slow propagating or resting thermal fatigue cracks in static parts (e.g., combustion chamber, "Ill. 3.2.3-1"; or turbine guide vanes, "Ill. 3.3-9").

Boreskopinspektions permit preventing surveys, e.g., FOD at the blading.

Non destructive testing (NDT) like inspections with penetrant, eddy current or ultrasonic can ( "Ill. 5.3-3") be carried out in the assembled situation when the position and size of the failure is convenient or the part is tempeorarily disassembled (e.g., at blade roots).

Chip control (oil, fuel) of filters ( "Ill. 3.5-3")and magnetic chip plugs ( "Ill. 3.5-5") can as well show the begin of a failure, as permit conclusions on concerned parts.

Operation limitations will be normally necessary, if at least over a certain time, we „can/must live with the failure“. That is the case when there is no time for specific measures or the lack of spare parts. The constraints comply with the mechanism and the progress of the failure.

During fatigue problems in the LCF- region, like under cyclic laoads of discs or thermal fatigue of hot parts, frequently the number of starts till the exchange will be limited. Thereby the crack propagation ( "Ill. 5.3-1"), in relation to the starts and/or operation time must be known. That requires an analysis of the fracture surfaces of the failed parts. Also an abatement ot the thermal stresses and temperature peaks with an optimised, respectively „cautious“ start procedure can be promising ( "Ill. 3.3-5").

Problems with impermissible heavy oxidation will rather demand a constraint of the operation lifetime.

Elevated temperatures of the blading by clogging of cooling passages or unfavourable temperature distribution in the hot gas ( "Ill. 3.3-11") can make it inevitable to limit the performance.

Repairs on site are limited to manual instructions. To this belongs, e.g., the smoothening of mechanical damages (FODs) on the blading ( "Ill. 3.1.2.2-4"). Cracks in gas ducts behind the gas turbine can often be repaired by welding. In some cases, in which no unacceptable leakage is to be expected, the crack propagation can be slowed by drilling a hole.

Maintenance can prevent failures with additional measures. To these belong especial efforts for the cleaning of the compressor blading ( "Ill. 4.2-1.1" and "Ill. 4.2-1.2") and/or the turbine. Further, frequent checks of the oil (oil analysis, "Ill. 3.5-4") can assure the operation. Also a lubrication, improved or adapted to the problem, can ease difficulties with sluggish or stuck parts.

Exchange and reconditioning of aggrieved components. To those belong blades with critical fretting, or hot parts (turbine blades, combustion chamber) with damages by thermal fatigue or oxidation. If there are not yet fatigue cracks at potential endangered parts, shot peening can minimise the risk.

"Illustration 0-2": (Lit. 0-1): This relevant literature (Lit.0-1) from the year 1994, regarding the evaluation of failures on industry gas turbines in the USA, enables us to draw some interesting conclusions. Such a statistic is dependent on the number and the type of individual engines and is, therefore. limited in it’s general validity. It is connected with cause related costs of individual failures, divided between engine derivates and heavy frame engines ( "Ill. 2.1-7" and "Ill. 2.1-8").

Combined causes of failure:

Stall in the compressor: This is apparently much more expensive for aero engine derivates as for heavy frame engines. Up until 5 years ago the failures on derivates often occurred in relation to steam injection in order to uprate performance. This shifted the operation point of the compressor to the surge line ( "Ill. 3.1.1-1"). As a result, the stall was so violent that massive destruction in the compressor took place. Through appropriate monitoring and control of the engine such problems can be overcome. This can serve as an example for the problem of uprating performance ( "Ill. 2.1-3.4").

Fatigue failure (HCF) on compressor blades: This concerns a type of failure for which the manufacturer is responsible in most cases. The compressor guide vanes in base load engines seem to be especially affected, which is actually surprising as the engines run in a small operation speed band, with relatively few start cycles. Thus, a dangerously long operation in the resonance area of the blades should be simpler to avoid during design, as derivates with typical peak load operation. The extensive trials of the engine variations, with their typically large number of engines, possibly play a part in the safe control of the frequent load changes and start cycles.

Fatigue failures (HCF) on turbine blades: The cost of damages for heavy frame engines are clearly higher than for derivates. It appears that the derivates also profit here from the extensive trials of engine variants. These failures usually fall in the area of responsibility of the manufacturer. The cause is mainly the excitation of the blade in the resonance ( "Ill. 3.1.2.1-9"), through periodical disturbances of flow in the gas duct. Even flutter excitation, (self exciting aerodynamic process), was identified as a cause of failure. In the blade tip area, relatively thin, turbine rotor blades, (narrow chord), were endangered by engines whose performance was uprated. Here also, the problem of uprating is clearly seen.

Failures on engine derivatives ( "Ill. 2.1-7").

Thermal crack initiation (thermal fatigue, "Ill. 3.3-16"): In connection with the steam injection to increase performance, crack initiation followed, in the area of struts, at the turbine casings of a particular manufacturer. More precise details are not given in the literature: it seems to concern crack initiation that occurred because of higher thermal stresses.

Turbine disc failure: This has apparently to do with a design weak point (notch, "Ill. 3.1.2.1-5 and 6" to "Ill. 3.1.2.1-7 and 8") on the discs of a certain type of engine that led to LCF fatigue ( "Ill. 2.2-5" and "Ill. 3.1.2.1-0" ), through start cycles, and is thus life limited.

Damages on heavy frame engines ( "Ill. 2.1-8")

Cooling air loss on T-rotor blades occurred apparently only on heavy frame engines with an extreme external air/air cooling system, that, after malfunctioning of the steam injection, developed internal icing on being shut down. Thus, there was insufficient cooling air for the hot parts and the outcome was creep failure, apparently first noticed after longer runs. Not only were turbine rotor blades affected but also the turbine discs belonging to them. In such an arrangement, a monitoring of the cooling air flow is urgently recommended to stop this kind of ‘creeping’ damage’ in time.

Creep fractures ( "Ill. 2.3-1" and "Ill. 2.3-2") on turbine blades: They are often the consequence of insufficient cooling air. This situation can enter when the manufacturer had underestimated, e.g., the deterioration of the sealing systems ( "Ill. 1.1-3" and "Ill. 3.3-11") during operation, or material data was unduly optimistically extrapolated concerning long running times.

Also clogging in the cooling system by intake of dust or forbidden intensive oxidation can trigger creep damages ( "Ill. 3.3-12"). A similar effect can be expected if material data were i mproper optimistic extrapolated for long durations or the operator ignored the advised (OEM) lifetime.

Internal fire and explosions: These events are closely connected to the extinguishing of the combustion or instant shut down, and insufficient drainage of the residue fuel at the end. A special problematik are backlashs of the flame ( "Ill. 3.2.1-5.2") or self-ignition in premixers of combustion chambers. In derivates of aircraft engines with titanium alloy compressor blades a titanium fire as secondary damage (for example after a blade failure) can not be ruled out (in this Ill. not documented).

"Illustration 0-3": In this picture (Lit 0-1), the percentage of the engines and the distribution of the failures according to performance classes is represented. The sum of the percentage of all performance classes is 100%. Further, the failure costs pro kilowatt, given delivered output, is shown. The evaluation is related to the statements of many insurance firms in the USA.

Performance range up to 9000 Kilowatts: Contrary to expectations, clearly a lesser number of derivates in the relatively lower performance range up to 9000 kilowatts is insured than heavy frame engines (HFM). If one relates the failure costs of the installed kilowatts and compares the range of both engine concepts up to 9000 kilowatts, one notices that the failure of a HFM causes higher average costs as those for a „derivate“. Against that, the percentage of the HFM failures in the performance area up to 9000 kilowatts is lower than the comparative value for derivates. This can be due to the different operation cycles. For the HFM, the base load operation is typical, for the derivate, the peak load operation. These differences are most noticeable in engines up to 1500 kilowatts. Apparently, derivates have less frequent failures. If a HFM breaks down, one has to reckon with higher costs. The costs of the unavailability of the engine, surely higher for engines that deliver base load than for those with peak load, are not given.

Performance class 9000 to 25000 kilowatts: Almost 80% of all derivates find themselves in this performance range, against only approximately 20% of the HFM. In this performance range, the failure rate and costs of moderate failures between both engines are absolutely comparable.

Performance range above 25000 kilowatts: It is noticeable that the share of the affected engine on the current total number is many times higher in derivates as in HFM (700000 kilowatts). Against that the moderate failure costs are thoroughly comparable.

Conclusion: Heavy frame machines show comparatively fewer failures in the lower and higher performance range, but are very costly in lower classes relative to output performance. In the middle performance range, there is hardly any difference between the failure performances of HFM and derivates. It is to be hoped that a levelling of failures to a lower niveau will be reached for all classes of performances.

"Illustration 0-4": This illustration touches upon works from J.Leopold (Lit.0-2) from the years 1980 (time period 1970 up to 1979 with approx. 240 failures) and 1987 (time period 1981 up to 1986 with over 180 failures). 110 engines were observed (1980) with a performance of 900 kilowatts up to 88 MW. The variation of the distribution over the years is of interest next to the distribution of the causes of failure and affected parts, from which, information regarding the developments and successes of failure prevention are to be expected.

Causes of failure: The biggest share of failures is to be traced back to so called product defects. These are faults that, true to definition, are the responsibility of the manufacturer, planning and assembly. Around 20% of the causes of failures are due to operation faults, divided equally between maintenance and service failures.

Affected parts: Hot parts in the combustor and turbine are most affected, whereas turbine rotor blades and guide vanes take the lion’s share. In contrast, failures in the compressor area are clearly more seldom. The relatively high share of bearing failures is remarkable (Chapter 3.5.2). Valves, air passages and pipes are not to be overlooked in this regard.

Trends and conclusions: Product defects have slightly decreased with time, but are still numerous. What J.Leopold found already in 1980 may still be valid: „Despite extensive efforts to repair, on the grounds of earlier failure trends in new developments, the improved version was not always defect free. Weak points in the design were clear only after many more running hours and starts“. The increase of service faults to above 20% was traced back to a clear increase in maintenance faults, while service defects drop under 5%. This spoke of an increase in the competence of the service personnel. The share of foreign objects (Chapter 3.1.2.2) as a primary cause of failure was given as approximately 10%, which justified higher efforts towards their avoidance.

"Illustration 0-5": The background to this evaluation is described in Fig.-3. This is related to statements from J.Leopold from the year 1980 (Lit.0-2), but can still be evaluated as a remarkable and impressive confirmation of the bath tub shape of failure frequency being above that of operation time ( "Ill. 4.1-9").

Conspiciuous is, that the installation and early guarantee phase shows an especially high failure incidence on hot parts, although one should actually expect long term failures at these parts at a later point of time. There follows a phase with a relative low failure rate, which then clearly increases during the time of big inspections (Chapter 4.1.2) between 15000 and 20000 operation hours. As reason for these observations, Leopold says that one recognizes these weak spots only on opening the engine. In the following phase, new improvements and alterations in the form of a lower failure rate seem to prevail.

"Illustration 0-6": In this presentation, according to T.D.Matteson (Lit0-4), it is recognizable that at least in flight engines, most failures (about 90%) do not show any definite increase with operation time. It is, however, to be seen that cost intensive, hot part and compressor failures like fatigue ( "Ill. 2.2-5" ) and creep ( "Ill. 2.3-2" and "Ill. 2.3-3") show a typical operation time dependency.

Despite that, maintenance and inspection procedures needs to be worked out ( "Ill. 4.1-1") which minimize revisions with disassembly. Progress in the monitoring sensors allow new perspectives also in industrial use. The progress in control/ monitoring (Chapter 5.1.1) and the sensor technology allow also in the industrial use new prospects.

"Illustration 0-7": This workout (Lit.0-5) was carried out by an overhaul shop and originated as early as 1974. It shows experiences with the aero engine type JT-4 and compares this with the derivate GG4.

With these types of engines, except for bearing and oil system problems, there seem to be serious differences between derivate and aero engine. Thus, the reasons for the repairs on the engine version are the given lifetime limitations. In derivates, the problems relate to the compressor and turbine. The high share of foreign object damages is noticeable when it comes to industrial use, whereas it is rather to be expected that aero engines have more problems due to the intake of foreign objects. That there was a lack of filters in front of engines in the early seventies, or that they were less effective is no plausible explanation for the high foreign object attack.

Conclusion: The weak spots of engines and their derivates can be very different.The steady state operation in industrial use strains the engine apparently differently. These problems show up only in long term use. The stationary use in the industry application seems to demand the engins different than in air traffic. Often the difference becomes clear in the temporal distribution ( "Ill. 0-5").