dc.description.abstract | Chatter is the condition where unwanted vibrational motion exists between the
cutting tool and the workpiece during machining. Chatter is highly undesirable
because it severely affects the quality of machined surfaces and causes high rates of
wear of cutting tools amongst other detrimental effects. Therefore, elimination of
chatter would yield advantages such as reduced tooling costs, downtime and scrap
generation. In this project, acoustic emission ( AE ) signals generated during turning
in the lathe machine were sampled and analysed with the view to establishing the
onset of chatter. The effects of chatter were also evaluated.
The signals were detected by a piezo-electric transducer mounted on the tool holder.
They were sampled by a microcomputer via a pre-amplifier and filter and then stored
in the hard disk. Signal analysis was carried out off-line after downloading the
sampled data from the hard disk using floppy diskettes. Frequency analysis, count
and amplitude distribution analysis as well as time series analysis were used to
evaluate the AE signals.
It was established that the cutting speed, the feed rate, the depth of cut, the amount of
the tool wear and the cutting geometry all influence the stability of the cutting
process. Further, the changes occurring to AE signal parameters can be used to
identify the onset of chatter and hence plot stability charts. Apart from showing the
borderlines of stability, such charts help in identifying the changes required to
eliminate chatter with minimum or no loss of production.
The sensitivity of AE signal parameters to the cutting conditions was maximum in
the frequency range 100-500 kHz. AE mean intensity level and AE count rate (at a
properly chosen threshold voltage) increased almost linearly with the cutting speed,
cutting depth, feed rate and tool wear up to the onset of chatter when the pattern was
lost. During chatter the AE signal parameters fluctuated significantly. Higher
statistical moments ( skew and kurtosis) of the AE signals were found to be sensitive
to tool wear and the subsequent chatter. In general, frequency techniques yielded
better results compared to count techniques. The fractal order showed little
correlation with the intensity of the cutting conditions. The fractal order failed to
identify the onset of chatter.
Results of the principal component analysis conducted on a zo" order autoregressive
( AR ) time series model of the AE signals were found to be sensitive to
the amount of tool wear. The first principal component showed a decreasing
tendency with increase in flank wear. The tendency was lost at the onset of the
chatter. A scattergram of the first two principal components showed outliers
corresponding to adverse wear conditions, which were accompanied by chatter.
It was concluded that AE technique offers a good adaptive method for on-line
monitoring of the cutting process since AE signal parameters are sensitive to the
cutting conditions. It is recommended that this research should be extended to
designing an on-line closed-loop system to monitor chatter and its effects. | en |