A
Multidisciplinary Approach Gets to the Root Cause of a Failed Shaft
|
|
An 8-inch diameter pinion shaft,
which was driving a 12-foot diameter bull ring gear on a ball mill, failed
after a period of rough operation. The crack originated in the area of a
keyway at a clutch hub, Figure 1. This geometry acts as a stress
concentrator and crack nucleation, in this region, is not uncommon. |
|
|
 |
| Figure 1. Location of failure. |
|
|
|
|
During our examination of the broken
shaft, we observed that the surface of the fracture was relatively smooth,
with characteristic “beach marks” indicating the direction of crack
propagation. Using SEM (scanning
electron microscopy), we found fatigue striations to confirm the
suspected nature of crack propagation. |
|
These findings, combined with the
45-degree slant of the fractured end of the shaft, implied that torsional
fatigue stresses drove the failure. However, our analysis of the shaft
loads indicated that the operating stresses on the shaft were considerably
below the fatigue endurance limit of the material, even after the stress
concentration factors in the keyway were taken into consideration. |
|
As the team members discussed the
conflict between design and actual operating experience, we considered
approaches to “on-line” evaluation of shaft stresses that combined two
traditional measurement techniques: strain gauging and vibrational
analysis. |
|
Because of the uncertainty in the
source of the shaft stresses, we decided to attach strain sensing devices
(Note: Strain gauges are typically spot welded or glued to the part.) to the
pinion shaft in order to measure bending and torsion in the area of the
failure. |
|
The gauges stretch or compress in
concert with the surface to which they are affixed. The stretching and/or
compression changes the electrical resistance of the wire grid imbedded in
the gauge. The change in resistance is directly related to the strain (or
movement) of the gauge and the underlying metal. By measuring the change in
resistance, the strain, Δε, is derived. In the elastic strain range of a
material’s mechanical properties, there is a direct correlation between the
strain and the applied stress, Δσ. This alternative stress provides a
measurement of the fatigue stresses on the component. |
|
Since the changing strain appeared
to be periodic, we decided to analyze the strain gauge data in terms of a
vibration phenomena. |
|
Putting all of the data together,
the source of the torsional stress on the shaft was discovered. It was
determined that the periodic torsional stress had a frequency which
corresponded to the rpm of the ball mill. It was concluded that a defect in
the ball mill ring gear was exciting a system torsional resonance. Although
the information was detectable from the vibration spectrum, it had been
overshadowed by the gear mesh problem. Without the spectrum analysis of the
strain gauge data, we would not have pinpointed the problem and would most
likely have had similar shaft failures in the future.
|
|
Additional lessons learned from this
case: |
|
The first step toward getting away
from simply reacting to equipment failures is to treat the failed component
as a valuable source of information. An incident occurred at the time of
the initial failure that, unfortunately, is a typical experience in the
field. The pinion gear was reversible, end for end. The failed end of the
shaft in the clutch hub was totally useless to the area and was set aside. |
|
The maintenance crew very carefully
placed the shaft so that the flat butt of the shaft was in the air and the
end with the fatigue crack, with all of the failure information, was
embedded in the mud and stones outside of the mill shed door, thus
damaging valuable evidence. |
|
