Gas turbines are monitored today by vibration probes (accelerometers). Vibrations are an important indicator for beginning damages ( "Ill. 2.5-4"). The expert can draw conclusions from the recording of the vibration, giving signals as to the possible cause and risks of the affected components. The following data could help him in this exercise:
"Illustration 2.5-3": There are many recommendations, standards and specifications for the use of vibration measurements, monitoring and the analysis in industrial applications, especially for gas turbines. They are published by national and private organisations as well as manufacturers (OEM). In the USA there are:
A german spezifikation is the DIN 1016-4. It concentrates on the measurement at non rotating parts of gas turbines and the evaluation of the results. To selects the suitable standard the scheme is recommended. It considers two effects:
As parameter (vibration magnitude, rms) for the intensity of a vibration, the maximum vibration velocity is applied (rms=0,707 x maximum deflection) in the design relevant frequency range. Also the deflection, acceleration and the maximum values can be used. The estimation of the vibration severity considers maximum values and the time rate of change.
Introductions for vibration measurements as condition monitoring and diagnosis are included in the ISO 13373. There are included:
"Illustration 2.5-4": (Lit. 2-16): Depending from the cause, the vibration monitoring of stationary gas turbines can have different goals ( "Ill. 2.5-5").
The vibration behaviour of a gas turbine depends strongly on the bearing stiffness. The oscillations of the rotor are damped by an oil film between the outer race of the bearing and the seat in the supporting casing structure. This was primarily applied to heavy frame engines with sleeve bearing mounted rotors. Even when derivates used as gas generator have direct in the casing supported antifriction bearings, it is possible that only the hot gas coupled power turbine has sleeve bearings. Derivates of modern aero engines are equipped with antifriction bearings, however often damped with an oil film against the supporting casing.
In both cases the oil film can prevent the early registration of dangerous rotor vibrations. So, when, e.g., a smaller fragment of a blade arises, there is the danger that the damage is not recognized because the trigger was not overstepped. Anyway high dynamic bearing loads can cause extensive secondary damages.
The sketch shows schematic typical sensors for the vibration measurement and their positions in and on the engine. The chart at the right contains a survey for heavy frame engines with sleeve brearings to the rotor.
A particular problem is the possibility of a failure (short term total breakdown) of the measuring system ( "Ill. 3.6.2-1"). That complicates the identification of a true failure. Including damages on the axial thrust bearings. To avoid spurious shutdowns sufficient certain redundant systems with two sensors are used. Their indications will be compared.
For smaller engines some manufacturers (OEMs’) offer as standard equipment only seismic probes on the casing. In those cases also proximity probes as an option are available. The simulteneously use of both sensor types generally is recommended for an engine monitoring.
The API-standard 616 (see also "Ill. 2.5-3") demands as components af a monitoring/guard system proximity probes. This is applied as well for an acceptance run as for the operation. The probes and the monitoring system often conform to the API 670 standard.
Generally measurements of the elastic strain with strain gauges are possible. But they are not mentioned. Apparently the failure probability (mechanical damage, thermal damage, debonding) is to high for a gas turbine specific long time use.
"Example 2.5-2": An accelerometer of a gas turbine showed, over a period of time, a rise in the vibration of the rotor speed. The engine was inspected several times and examined for a rotor unbalance. In no case an unpermitted unbalance could be found. A rotor distortion could also not have been the cause, as the vibrations appeared in normal operation as well. In the last strip, after the vibration level had reached worrisome values, the thrust bearing of the rotor was examined for peculiarities, without result. With the change of the compressor outlet casing, which supports the bearing chamber, no more vibrations arose. On examining for cracks on the compressor outlet casing, one found a crack of several centimeters in a surrounding reinforcement frame ( "Ill. 2.5-4").The examination of the crack showed that the failure was to be traced to vibration fatigue. From the fracture, (beach marks, rust formation), it was recognizable that a slow crack propagation had taken place over a long time. Thus there was a plausible explanation for the slowly rising vibration level: The stiffness of the casing and, with it, it’s own frequency was so widely reduced through the crack formation, that the outcome was a resonance with the rotor. With growing crack lengths, the amplitude of the casing increased and this was registered by the warning system. If the operator had not reacted, a catastrophic secondary damage would have been difficult to avoid.
From this example, we learn how important it is to have a vibration monitor on our gas turbine and to „believe“ it as well.
"Example 2.5-3": (Lit. 2-15): Midst the 90s two rotorblades of a 100 MW-class gas turbine failed in the last stage. These bladefailures produced against the rotating direction secondary damage on ten succeeding blades. This lead to a heavy unbalance. The failure occurred when the engine reached the continous power rating. The result were heavy vibrations, causing additional damages (example 2.5- 4):
A following investigation showed vibration fatigue in the HCF region ( "Ill. 3.1.2.1-0") as cause of the blade failures. As reason random occurring blade vibrations in the fundamental bending mode or 1st bending mode have been declared. Both vibration modes stress the transition of the blade leading edge to the root platform significantly. This region is already subject to high prestresses from the centrifugal force, aggravatet by the notch effect of the transition radius. This influences act additionally as mean stress and lower the usable fatigue strength. The reason of the vibration exitation were not harmonic rotor vibrations. As remedy the critical blade area was strenthened by design that the stress niveau caused by centrifugal force dropped and the vibration frequences appropriate changed. It looks as if those measures have been successfull.
Comment: The conclusion of random arising vibrations seems not fully satisfying. The fact of two by HCF cracked but not broken blades rather indicate a LCF-load. Otherwise, from experience, it is very likely that the first superficially cracked blade breaks befor a HCF crack starts in a second one. If the reason was a HCF load the designed medium stress in the crack area was dangerously high. Regarding the vibration exitation the suspicion exists that this was either not found during the following tests or was not released by the investigator.
By the way it seems that also engines of other manufacturers (OEMs’) had similar failures on the turbine end stage (Lit. 2.5-16). In this light a cause for the vibration exitation may exist behind the turbine.
"Example 2.5-4": (Lit. 2-15): In context of the "Example 2.5-3" secondary failures that are rather dedicated at derivates/aero engines were observed. Concerned are containment (holding fragments inside) and oil fire.
The oil fire developed in the outer casing and in the exhaust area. Cause was the breaking of a pressure oil pipe due to dynamic overload by unbalance vibrations of the rotor. Apparently a pipe connection failed. Remedy was a design with matching and optimzation of the mountings, the elasticity of the pipe and fixed masses (flanges). Even when tubes not fail observable during so extreme vibrations, they must be controlled very closely (severe fretting maks, cracked clamps, even slight deformations). In case of doubt a non visible fatigue damage is to assumpt. That means such pipes must be changed.
Containment means that even small fragments are not allowed to exit the engine during the damage. It ‘s interresting that this is explicit mentioned, that means it is no matter of course. In the case of this failure the blade fragments have been contained. The rotor merely showed rub marks which seamed easy to repair .
Comment: It can not be seen from the existent documentation if the oil fire at the broken pipe is in causal connection with the fire in the exhaust area. Against this indicates, that the pipe broke outside the gas channel. Rather a leak on the main bearing can be expected. The rub traces on the rotor, accordingly to experience, harm the material deeper so that they can not be fixed with a ‘light’ rework. So even after this repair a remaining damage must be considered.
"Illustration 2.5-5": In this illustration, typical causes for unusual heavy vibrations of a gas turbine are collected.
Oscillations in the combustion chambers (combustion instabilities, "Ill. 3.2.2-5"): They normally vibrate with the relatively low frequency (‘buzz’) of the gas stream during the combustion. Those vibrations should be considered and mastered by the designer. However it’s possible that due to water or steam injection ( "Ill. 3.2.2-4") or problems with the fuel injection, abnormal high vibrations occur. Especially frequent oscillations are observed at Dry-Low-Nox combustion chambers which operate with a very lean "Ill. 2.5-5" fuel/air mixture. Even gas from an other source can stimulate instabilities.
The pre mixing with a large quantity of air favors such a self amplifying instability of the combustion ( "Ill. 3.2.2-5"). Those violent vibrations, e.g. can lead to unexpected heavy abrasive wear (fretting) of the assembly connections and the break away of chamber shingles ( "Ill. 3.2.1-4").
Surge in the compressor can in the formation phase (eg. rotating stall, "Ill. 3.1.1-5") trigger vibrations and fatigue failures of blades and vanes. A complete stall of all blades (surge) with the disruption of the air stream leads to heavy abrupt impulses with the threat of a short-term dynamic overload of
the blading and/or intense rub ( "Ill. 3.1.1-6"). Not dissipated friction heat with the air (‘blunger loss’) in the momentary lack of an air stream can short-term overheat and damage the rotor blades. The lack of cooling air in the hot parts, together with a shortage of combustion air, can lead in very short time to grave overheating damages.
Rotor bow and rub events: The differentthermal inertia of the rotor and the casings, even hours in the cooling phase can lead to gap bridging and jam of the rotor ( "Ill. 2.2-2" and "Ill. 3.1.2.4-2"). The warm air in the shut down, standing still engine rises up and heats the upper part of the rotor more intensive than the zones beneath. The consequences are self reinforcing vibrations in the frequency of the rotor speed with dangerous damages. This includes the weakening of the rotor by abrasive wear and the overheating as well as overloading the blading.
Air seals: Vibrations of labyrinths can be excited in manifold ways. Examples are excitations by the leakage stream or gas oscillations in the ring chambers around the rotor shaft, formed by the seals. The results can be fatigue cracks and break-outs (plate vibrations) at thin walled components. Endangered are e.g., baffles or support cones of the labyrinths.
Casings: The stiffness of supporting casing structures can change as result of a crack formation. Normally this leads to the weakening of the cross section with a drop in the natural frequency. That permits increased vibrations ( "Example 2.5-2"), especially in case of resonance. Corresponding with the crack growth, normally the vibration level rises with the operation time.
Turbine vanes, turbine nozzles: Damages of the turbine nozzles (high pressure turbine entrance vanes) can excite the rotor via the blades to vibrations that also show outside the engine.
Turbine rotor blades: In the turbine the breaking of a blade can quite happen, especially if only a part of the blade which is not at once noticed, is concerned. On the other hand in a compressor substantial secondary failures can be expected when a blade breaks. This means the immediately breakdown of the engine ( "Example 2.5-2").
Smaller turbine defects can be recognized by the unbalances. This is not easy when the engine has elasic suspended, damped main bearings. We find such a design in aero engine derivates. In this case the unbalance forces are only little noticeable at the acceleration probes on the outside of the engine. Heavy damages can be expected from an operation over longer time with internal unbalances. Therefore it is important that the trigger intensity settings of the probes are satisfactory sensitive without activation of false alarm.
Main bearings:
Anti friction bearings (chapter 3.5.2.1): Failures cause remarkable, mostly high frequency vibrations. There are failures which show early enough before a catastrophic vibration. To this belong fatigue outbreaks (fatigue pittings, pittings) in the bearing races. Is there the suspicion of a bearing failure, magnet plugs and oil filters must be controlled. On the other hand, in case of indications at a magnet plug, a very close look for vibrations is recommended ( "Ill. 5.1-1").
There is quite a realistic chance to recognize at big engines a bearing failure in time with a relatively low number of rotation. In contrast smaller engines with higher speed sustain catastrophic bearing failures in short time. Small gas turbines may suffer this in seconds during the observable end phase. Comprehensible, an engagement to capture such a failure in time is hardly possible.
An intermittent oil supply can also provoke vibrations in the bearing area. Thereby also exists the danger of the breaking of the oil jet with extensive second failures.
Friction bearings/journal bearings (Chapter 3.5.2.2): In gas turbines of the heavy type, normally journal bearings are used. They have specific failure modes (Lit. 2-10 und Lit 2-11). Typical causes of damage are:
Accessory equipment (Chapter 3.6.1) can be excited to vibrations by the failure of components like gears (Chapter 3.7-2), couplings/insert shafts ( "Ill. 3.6.1-7") or bearings (Chapter 3.5-2). In an extreme case this leads to the fracture of the mounts and/or destruction of the device. It is also possible that vibrations are routed into the engine and produce damages. So, e.g., a case emerged, at which a small inaccuracy of a gear wheel lead to the fatigue fracture of the central interlocking bolt of the turbine rotor. Emanating from the teeth of gear wheels are extremely high frequent vibrations (up to the ultrasonic region) because the number of teeth and the high rotation speed. At the arising vibration modes even smallest amplitudes in the region of 0.1 mm are enough for a damage by vibration fatigue.
Subsequent installed accessory drives with alternating torsion moment like cardan shafts or universal couplings with elastomers can excite vibrations through a gear in the engine and produce failures. In such a case it came several times to the fracture of a drive shaft in a control unit which was driven by a gear. This potential transmission of vibrations could easily be overlooked. It was therefore discovered not until several failures.
Vibrations of the drive train: Insufficient aligned shafts and couplings can trigger vibrations. The neccessity to drive through resonances has the risk of high amplitudes with an over load of the schaft up to a fracture. Also vibrations of the powered aggregates (e.g., pump or generator) can be fed into the engine by couplings. Are vibrations of the rotor excited, an increased damage hazard exists. Therefore special regard to the correct funktion of the drive side is needed.
"Illustration 2.5-6": Through crack formation or too high operation temperatures, a casing can become more elastic and can trigger the drop of the natural frequency and get in resonance with the rotor speed. This will be shown by the vibration warning system. Such effects (increase of the amplitude with the operation time) are to be followed exactly, even if the amplitudes are in a acceptable range. So it should be possible to take remedy measures if necessary (see "Example 2.5-2").