Reciprocating compressor is one of the key machines in chemical and petrochemical industries. The most common failure mode in the compressor is valve leakage. Generally, leakage fault is considered to be of little harm to machines. However, it was found that the serious leakage of the valve would cause abnormal bending vibration of the piston rod and accelerate the formation of fatigue cracks. Most researchers utilized the signal of cylinder dynamic pressure, valve temperature or acoustic emission to diagnose valve leakage fault. However, each of these methods has disadvantages.

In this paper, a new method is proposed to diagnose valve leakage fault using vibration signal. The main idea is that the severity of valve leakage can be assessed by analyzing energy of the main frequency band during compression process and the delay of discharge valve opening. Firstly, the vibration signal in the time domain is segmented into several angle-domain signals according to the keyphasor signal. Each of the angle-domain signals corresponds to one cylinder working cycle. Then, the time-frequency analysis is conducted with the Gaussian window, and the main energy frequency band and the compression process are determined. Filtering the signal with a bandpass filter based on the main energy frequency band, and the root-mean-square (RMS) of the filtered signal during compression process is calculated in a crank revolution. Besides, the angle of discharge valve opening is detected. Based on the two indicators, the severity of valve leakage can be estimated.

To verify the effectiveness of the method, keyphasor of the flywheel, cylinder dynamic pressure and vibration acceleration of valve are acquired on a double-acting oil-free air compressor under different degrees of suction valve leakage. The experimental results and analysis show that the proposed method can well identify the valve leakage fault, even in the case of weak leakage, and is very effective in quantifying the severity of leakage.

This content is only available via PDF.
You do not currently have access to this content.