Valve_Spring_Fault_Detection_4.pdf

The article presents problems related to vibration
diagnostics in reciprocating compressors. This paper presents
the evaluation of several techniques of the digital signal processing,
such as the spectrum calculation with the Discrete Fourier
Transform (DFT), Continuous Wavelet Transform (CWT), Segmented
Analysis for detection the spring failure in reprociating
compressor valve with the help of the vibration monitoring. The
experimental investigation to collect the data from the compressor
with both the faultless valve and the valve with spring
failure was conducted. Three 112DV1 vibration acceleration
probes manufactured by TIK were mounted on the cylinder
of the compressor. The keyphasor probe was mounted on the
compressor’s flywheel. The signal of the vibration acceleration
probe mounted on the top of the cylinder was used for the
Condition Monitoring and Fault Detection of the valve. The
TIK-RVM system of monitoring and data acquisition was used
for gathering the signal samples from the probes. The sampling
frequency was 30193.5 Hz, signal length was 65535 samples. To
imitate the spring fault, the exhaust valve spring was replaced
by the shortened one with the same stiffness. It has been shown
what the signal acquired from the vibration acceleration probe
mounted on the reprociating compressor cylinder can be used
for reliable detection of the spring failure of the compressor’s
valve with the help of the mentioned signal processing techniques.
As it can be seen from the signal processing results
in the article, the tecniques used are showing quite different
results for the cases of the normal valve spring and the short
one. It seems what for this type of the compressor and valve,
the valve sping failure can be quite reliably detected with the
help of the vibration monitoring. To see if this is a case for
other compressor types and other valve types, the additional
experiments are needed.