Stress wave analysis (SWAN) provides real-time measurement of friction and mechanical shock in operating machinery. This high-frequency acoustic sensing technology filters out background levels of vibration and audible noise, and provides a graphic representation of machine health.
By measuring shock and friction events, the SWAN technique can detect wear and damage at the earliest stages, prior to increased vibration, and can track the progression of a defect throughout the failure process. This is possible because, as the damage progresses, the energy content of friction and shock events increases. This “stress wave energy” is then measured and tracked against normal machine operating conditions. Several types of aircraft and industrial gas turbine engines were tested to demonstrate SWAN’s ability to accurately detect a broad range of discrepant conditions and characterize the severity of damage.
SWAN is a state-of-the-art instrumentation technique for measuring friction, shock and dynamic load transfer between moving parts in rotating machinery. These events produce structure-related ultrasound waves (stress waves) that are detected and analyzed electronically by the SWAN system.
An externally mounted sensor on the machine’s housing detects stress waves transmitted through the machine’s structure. A piezoelectric crystal in the sensor converts the stress wave amplitude into an electrical signal, which is then amplified and filtered to remove unwanted low-frequency sound and vibration energy (Figure 1).
The output of the signal conditioner is a stress wave pulse train (SWPT) that represents a time history of individual shock and friction events in the machine, such as those influenced by lubrication. A digital processor then analyzes the SWPT to determine the peak level and the total energy content generated by the friction or shock event. The computed stress wave peak amplitude (SWPA) and stress wave energy (SWE) values are displayed and stored in a database for comparison and historical trending with normal readings. SWAN measures even minor shock and friction events that occur between contact surfaces. The level and pattern of anomalous shock events become a diagnostic tool.
The digital analysis of stress waves consists of computing both the amplitude and the energy content of detected stress waves. The amplitude (or peak level) of a stress wave is a function of a single friction or shock event’s intensity. The SWE is a computed value (the time domain integral) that considers the amplitude, shape, duration and rates of all friction and shock events that occur during a reference time interval. In a spalled bearing, for example, the peak level of the detected stress waves is primarily a function of the spall depth, whereas the SWE is a function of spall size (Figure 2).
Figure 2. Stress Wave Energy (SWE)
Because SWAN can detect stress waves at a lower energy than other diagnostic tools, it can usually diagnose problems early. In most cases, during early component fatigue, the energy released between the contact surfaces is too small to excite gearbox or engine structures to levels significantly above background vibration levels. Fatigue is observed only when catastrophic failure or extensive secondary damage occurs. However, with SWAN, SWE can be detected and analyzed early in the failure process (Figure 3).
Figure 3. SWE Operating History
As machine parts come in contact with the defect, even at the earliest stages, shock and friction events generate SWE. SWAN detects and measures this energy at damage levels well below the degradation required to excite vibration sensors, and before sufficient damage has occurred to activate metal chip detectors in lubrication systems.
Acronyms Commonly Used in this Article:
SWAN - Stress Wave Analysis
SWE - Stress Wave Energy
SWPA - Stress Wave Peak Amplitude
SWPT - Stress Wave Pulse Train
FOD - Foreign Object Damage
Case History No. 1: Roller Bearing Roller End Wear
The benefits of SWAN technology vs. vibration analysis were clearly demonstrated during the monitoring of an industrial gas turbine in a power generation application. SWAN indicated that there was damage from either the No. 1 or No. 2 roller bearing. A spectral analysis from the sensor in that location showed the spectrum to be riddled with 105.8 hertz (Hz) and harmonic spectral lines. According to the bearing defect frequency tables, the cage rotational frequency relative to the stationary outer race of the No. 1 and No. 2 bearings was 105.8 Hz. Thus, it was apparent that a bearing defect was present, and that the probable root cause was roller end wear. Spectral content was strongest from sensors close to the No. 1 and No. 2 bearings, and subsided the farther away the sensors were positioned - further evidence that there was bearing damage.
This turbine was also equipped with oil debris and vibration monitoring equipment. Niether of these conventional monitoring systems indicated there was a problem. This is typical of early levels of damage that do not release enough kinetic energy to excite the structural dynamics of the machine’s stiff spring-mass system, or generate enough debris to be captured by chip detectors. However, because SWAN filters out background noise and vibration, the signal-to-noise ratio is high enough to detect problems while they are small, so the machine can be safely operated until a repair can be efficiently scheduled. In this case, a disassembly inspection was scheduled to coincide with previously planned downtime.
The teardown and inspection did in fact confirm damage on the No. 2 bearing, just as predicted several months earlier by the first SWAN indication. The photo in Figure 5 clearly shows roller end wear, as suggested in the initial diagnosis. In addition, abrasive wear on the race shoulder was reported.
Figure 5. No. 2 Bearing Showing
Roller End Wear
Case History No. 2: Labyrinth
Data was collected from a large industrial gas turbine at an electric power utility for five months before it was removed from service for a scheduled overhaul. During this period, there was no indication of a problem with any of the gears and bearings, but a sensor located on the low-pressure turbine (LPT) did show a significant increase in friction levels (Figure 6).
Figure 6. Labyrinth Seal Wear
The erratic nature of the SWE readings and the absence of any clear indication of damage to gears or bearings, plus the fact that lube changes had no impact on the SWE, led maintenance and engineering staff to the conclusion that some sort of rubbing (or similar constant wear mechanism) was at fault. A thorough disassembly inspection conducted during overhaul revealed the cause to be excessive wear to the faces of the labyrinth seals in the LPT area.
Case History No. 3: Seeded
Fault Engine Test (SFET)
A series of discrepant conditions was intentionally built into a Pratt & Whitney F100-PW-100 engine, which was then operated over a range of operating conditions to test the detection capability of various diagnostic techniques. (Author’s note: This test was sponsored by the Joint Strike Fighter program.)
Figure 7. Bearing Race Damage
Figure 8. September 30 (Good Oil), SWE = 16153
Figure 9. October 3 (Degraded Oil), SWE = 22175
Figure 10. October 13 (Good Oil), SWE = 21941
These case histories document the ability of the Stress Wave Analysis technique to:
SWAN is a new diagnostic and prognostic tool for the engine maintainer’s toolbox. It is superior to vibration analysis for detecting and quantifying discrepant conditions that generate friction and shock. This includes not only localized fatigue damage to bearings and gears, but also includes lubrication problems, abrasive wear, abnormal dynamic loading and foreign object damage. The basic analysis tools in SWAN provide accurate detection, quantification and fault isolation through an easy-to-use graphical user interface.