Abstract:
A theoretical
design and study of production of various types of ultrasound transducers
for acoustic emission research into the fatigue failure mechanism and application in
machine condition
monitoring of rolling element bearings and gears are reported.
The transducer transmission modelling was presented and verified, theoretical
transducer
responses were exercised and their comparisons were made with
experimental results. Transducer size, construction and location were considered
important in the application of ultrasound. Advanced method of metallurgical attachment
improved the bond durability and reliability, the signal transmission as well as the
capability of withstanding a hostile environment (117 °C) within the vicinity of the
component being monitored. In-Situ calibration provided a means of ensuring the in-
service transducer
performance and the sensing circuitry as a whole. _
The smallest fabricated transducer had a circular sensor of 2.6 mm diameter and 0.5
mm
thick, which was highly sensitive to a few milli-meter stress wavelength. The ring
type transducer was preliminarily found to improve the amplication range. r
Five
rolling element bearings were tested using the purposely designed ultrasound
transducers, the experimental observations of the detected signal were then time-
amplitude distribution displayed. The changing stages of failure development from a
single fatigue spall of less than 3 mm to 1.5 mm in length and down to surface
crackings of 25 m long, were appropriately explained.
Two
pairs of test gears having three different fault conditions (good, moderate and
severe) were conducted on a Roll-Royce back-to-back gearbox test rig at 1000 RPM to
2000 RPM
approximately. Transducers implanted inside the gearbox sensitively showed
the
gradual changes in amplitude distribution resulting from the fault incremental
category. A study of signal transmission of the fault in a gear assembly revealed that
half of the elastic stress wave energy was lost by the time it arrived at the support
bearing.
Statistical methods
using RMS and kurtosis parameter a fault indicator were employed
to evaluate the
bearing and gearbox data. High frequency analysis confirmed that failed
bearing or gear signatures were ultrasonic, in addition the signal frequencies below 50
kHz were also detectable
by the purposely designed transducers. A defect simulation
was also established to
study the gearbox failure, its usefulness was then discussed.
