High Cycle Fatigue Testing
|
High Cycle Fatigue Failure PreventionIntroductionHigh Cycle Fatigue (HCF) is one of the leading causes of engine failure in aircraft, and thus has received considerable attention in recent years. Unfortunately, inexpensive and rapid test methods which provide useful information about potential HCF problems have not been available until recently. Therefore many HCF problems are only discovered in the field, long after completion of the engine design and development. Fixing these problems can be extremely expensive, often affecting the entire fleet and substantially reducing readiness of military aircraft. Recently a new HCF test method has been developed under a joint government-industry program called Atomized Liquid Jet Excitation spin testing. This test procedure is an adaptation of a spin test, a relatively inexpensive test commonly used for Low Cycle Fatigue (LCF) development and proof (also known as over-speed) testing. The rotor is spun through its full operating speed range while being excited in a manner that simulates the vibratory response in an engine that can lead to HCF failures. In the controlled environment of a spin test, design and operating conditions can be carefully controlled, allowing rapid characterization of the behavior of engine components under realistic operating conditions. This method has been successfully demonstrated for several turbine engine components and offers great promise to help avoid or resolve HCF problems in current and next generation engines. The HCF ProblemHigh Cycle Fatigue (HCF) in turbine engines occurs when resonance or vibration modes of parts like fans, compressors or turbine rotors are excited under certain circumstances, sometimes leading to very sudden and catastrophic failure. HCF is a significant problem for any operator of turbine powered aircraft; responsible for as much as 50% of all engine failures. The Air Force and Navy estimate that HCF-related problems are the leading cause of engine failure and cost approximately $400 Million per year.1 Furthermore, as turbine designs become more aerodynamically efficient and component designs become thinner and more flexible, vibration response increases. In recognition of this problem, since 1994 the DoD has allocated in excess of $100 Million of research activity under the National HCF S&T Program to better understand and prevent HCF problems in the nation’s fleet of military aircraft.
Figure 1 The earlier in the development and deployment of an engine that HCF problems can be detected, the less disruptive and expensive it is to find and implement a solution. . Unfortunately, by their nature HCF problems can be elusive and often don’t show up until years after an engine is placed into service. Diagnosing and correcting an HCF problem in the fleet can be very expensive, often requiring that aircraft be grounded, engines removed and components replaced. Occasionally these problems are uncovered during engine development tests. But at a cost of up to $1 million per week, full scale engine testing on a dedicated test stand is quite expensive and is often scheduled years in advance. The next best alternative is a rotating test of a portion of the engine (such as a compressor) in a whirligig rig with simulated airflow. Even whirligig tests are expensive (at about $100K per week) and there are limited facilities to perform these tests. Spin testing is commonly used to determine the Low Cycle Fatigue (LCF) characteristics of single stage rotor parts like fans, compressors and turbines, but until recently very little HCF or dynamic information was available from a spin test. Occasionally a rotor is spun at operating speed in an evacuated spin chamber, and then air is introduced via nozzles pointing at the rotor in a manner designed to excite specific vibration modes. While this method can excite a vibration mode, it is a transient test because the rotor slows down rapidly from the drag created as the spin chamber fills with air. This method cannot be used to evaluate HCF crack initiation or propagation which requires maintaining the resonant speed for extended periods of time. Furthermore, the damping of a particular vibration mode, one of the most important characteristics of HCF behavior, cannot be accurately measured during such a transient test. Novel Solution DevelopedRecently, a novel and promising spin test method for determining HCF characteristics of rotors has been developed. Funded jointly by the Air Force Research Laboratory and the Naval Air Warfare Center2, the method was developed by Test Devices of Hudson, MA. US engine makers Pratt & Whitney, General Electric Aircraft Engines, Allison Advanced Development Company and Honeywell Engines also collaborated in the development and application of the HCF spin test technique. The method is called Atomized Liquid Jet Excitation and is shown conceptually below. A series of nozzles pointing, in this case, radially inward are arranged around the periphery of the test rotor. During test, a liquid is continuously pumped through these nozzles such that the resulting atomized stream impinges on the rotating blades at specific locations. Generally the fluid stream is directed at an anti-node (region of maximum deflection) for the particular vibration mode of interest. When the high speed rotating blade, contacts the fluid stream, momentum is transferred to the fluid and lost by the blade. This momentum transfer imposes a reaction force on the blade, of a magnitude established by the mass of the fluid impacted and the velocity of the blade. The number of fluid streams and the speed of the rotor determine the frequency of the applied force.
Figure 2 By controlling the number of jets and the point of contact, a specific mode of interest (frequency and deflected shape) can be selected for excitation. Furthermore, by controlling the fluid flow rate and velocity, the magnitude of the forcing function, hence the resonant response (i.e. strain level), can also be controlled. During the course of the two year development program, Test Devices refined the test method into a practical technique. Test capabilities include:
During the program Test Devices solved many important testing issues. First, the momentum of the impinging jet could cause erosion of the blade surface if not properly controlled. However, through careful management of jet distribution and droplet size, it is possible to maximize excitation while avoiding erosion. HCF modes have been excited in test articles to high strain levels and tip speed in excess of 1600 fps for over 2 million cycles without erosion damage. Second, strain gages cannot be located in liquid impingement areas. In practice, there is no significant limitation because the fluid impingement zone is typically at a displacement antinode (highest deflection) location, while maximum strain locations are typically elsewhere. Finally, the use of fine liquid spray or mist in a closed chamber can lead to conditions that will support rapid combustion. To perform a safe test requires that these conditions be monitored and controlled. To prevent a disaster in the event of inadvertent ignition, the spin chamber must be capable of sustaining in excess of 100 psi deflagration pressure. Well designed spin test chambers can contain this pressure, but many older chambers cannot. Spin test systems should be evaluated carefully for safety before installation of this kind of excitation system. Test Devices has applied for a United States Patent for this novel test method. HCF Test Examples TF41 FanAn Atomized Liquid Jet Excitation test was performed for the first stage fan of an Allison TF41 engine, a 38 inch diameter rotor consisting of 25 shrouded titanium blades as shown on the next page. In this case four nozzles were arranged around the rotor at 90 degree intervals inward to generate a 4EO (engine order) excitation. The nozzles were pointed radially at the leading edge, the location of an anti-node for the chordwise tip bending. Sixteen strain gages recorded blade response at various locations. At 9,100 RPM a sharp resonant peak was detected which corresponded to the chordwise-bending mode. In this particular case, the peak-to-peak vibratory strain levels reached +600 microstrain (~inches/inch). The rotor can be held at resonant conditions indefinitely and the strain level can be “dialed in” to determine how many fatigue cycles are needed before damage (i.e. a crack) is initiated.
Figure 3 HCF usually occurs at high resonance frequencies (1KHz to 20 kHz) where a large number of damaging fatigue cycles can be rapidly accumulated, even with relatively short dwell times. This can lead to crack initiation, crack growth and, in the worst cases, catastrophic failure. Therefore, a key element of the Atomized Liquid Jet Excitation HCF test is the ability to dwell at resonance under a realistic combination of centrifugal and vibratory loads in order to determine crack initiation and growth behavior. Another key asset developed by Test Devices that enables this capability is a very sophisticated air turbine drive speed control system Many of the most difficult HCF problems in today’s high performance engines involve high order resonant modes with very low damping. The high Q (dynamic response) of these modes means that resonance exists over a very narrow frequency band. As a practical matter, that is a mixed blessing. On the plus side (from the perspective of engine life), this mode is not excited except when the engine is operating over a very narrow speed range. On the minus side, however, should the engine be within that particular frequency band, any excitation will lead to high vibratory strains and reduced component life. It is not unusual for modes to exist with a Q value of 500 to 1000. At a Q of 800, for example, the half power width of the resonance peak is typically only on the order of +10 RPM. This means that to accurately hold at a resonant condition, air drive turbine speed control must be accurately held to levels of +2 rpm. Test Devices has long developed and supplied such precision air turbine systems, and has further refined this technology for HCF testing purposes as discussed below. Blisk ApplicationAn engine fan stage has also been excited using the Atomized Liquid Jet Excitation test method. This fan stage consisted of low aspect ratio blades integrally machined with the disk (so-called blisk). While the design offers improved aerodynamic efficiency, the elimination of joints between the blades and hub also removes a source of friction damping. In a conventional bladed rotor, energy is dissipated through vibratory motion at the blade root-to-disk slot interface. Combining the blade and disk into a single component eliminates this energy dissipation mechanism. Blisks therefore exhibit very low damping (high Q), and the magnitude of any vibratory excitation at a given resonant mode is directly proportional to the value of Q. Therefore, a good measurement of the Q value is important to understand how the rotor will respond to excitation sources. For the fan stage tested, the chosen vibration mode was the chordwise bending mode. This mode excited the blade at a relatively high frequency. A ring consisting of equally spaced nozzles was fabricated, and the target vibration mode was successfully excited. Vibratory strain levels over +1000 microstrain were achieved during a slow sweep of rotor RPM, as noted below. The red trace is rotor speed and all other traces are strains at specific gage locations.
Figure 4 The importance of precise rotor speed control is apparent from an examination of the strain response peak shown below. It is difficult to calculate Q accurately using the half power method with a very sharp peak (such as shown in figure 5) since the slope of the curve at the half power point is very steep. When this data is expanded in the frequency scale (shown below in figure 6) it is readily apparent that the half power points lie only a few RPM apart. Small errors in measuring these points can lead to large potential errors in the value calculated for Q, with a corresponding error in the predicted forced response of the rotor.
Figure 5
Figure 6 As previously discussed, air drive turbine speed control is essential for HCF work. This allows the HCF test method to excite discrete HCF modes, and also monitor developing cracks associated with that mode. As a crack in the component develops and/or grows, the stiffness (compliance) of the component changes, if ever so slightly. Typically component stiffness decreases with crack growth, leading to a small but perceptible shift downward in the resonant frequency. The net effect is that while the excitation speed is held constant, the vibratory response will change and eventually the mode frequency will drift below the excitation frequency. At this point, the excitation frequency can be changed, to “follow” and continue to excite the mode, or the component can be physically examined well before fracture occurs. By varying excitation conditions, a great deal of valuable information can be learned about the HCF behavior of the test article. The photos below illustrate a crack that was developed and grown under controlled HCF dwell conditions such that crack growth characteristics could be studied.
Figure 7
Figure 8
Figure 9 Benefits and ApplicationsThe newly developed HCF spin testing method offers two primary advantages over more complex tests: (1) it can be performed quickly and (2) at much lower cost offering greater value than other test methods. As a rough idea, definition of the problem to first test can be accomplished in as little as 8 weeks, at a cost of about $50K - $150K, depending upon specific details. In the design stage, HCF spin testing can help develop and test rotor designs:
Should an HCF problem manifest itself in service (as often happens), the HCF spin testing method can be used to:
___________________________ 1 Garrison, B, “High Cycle Fatigue (HCF) Science and Technology Program Report”, AFRL-PR-WP-TR-20012010, Feb. 2001. For more information, please contact one of our sales engineers.
|
|
|
Copyright© 2003, 2004, 2005, 2006, 2007 Test Devices Inc | 571 Main Street | Hudson, MA 01749-3035 U.S.A. |