Feb-2013

Reciprocating compressor suction and discharge valve monitoring

A decision making guideline evaluating the strengths and weaknesses of the most common online monitoring technologies

Daniel Goebel
PROGNOST Systems GmbH

Article Summary

PROGNOST Systems failure mode survey of most common reciprocating compressor failure modes in 2009 is showing valve failures as the most frequent root cause for unplanned compressor shutdowns. New valve designs and improved materials have been introduced in the past 10 years and have reduced the percentage significantly. However, for many compressor operators valve monitoring is a main concern when evaluating condition monitoring systems to reduce unplanned downtime.

While the high percentage against other machine failures calls for further improvement, valve leakages in an early stage are usually not safety relevant. Undetected suction valve failures might lead into a complete loss of compression causing more dangerous failures, e.g. seizing crosshead wrist pins resulting from missing rod load reversal.

In the early days of reciprocating machine monitoring, maintenance strategies were mainly based on temperature measurement. Today, different methods of online condition monitoring can be applied to create precise diagnostic information of the valves and other components.

By comparing strengths and weaknesses of temperature monitoring with cylinder acceleration vibration measurement and pV diagram analyses this article provides a decision making guideline to identify the best suitable permanent monitoring technology for a specific compressor.

Description of the methods
Highlighting these completely different methods, it becomes obvious that valve monitoring can be improved regardless whether it is delivered with a new machine or retrofitted to an existing machine. In any case it is the main objective to detect a leakage of a valve that is causing a loss in efficiency of the compressor. Other damages such as broken valve springs or cracks in the valve plate or rings are considered as early stages of a leaking valve.

• Valve temperature monitoring
In compressor thermodynamics it is considered that a gas leakage in either a suction or discharge valve is causing an increase of the gas temperature in the valve pocket. Two installation options are common to monitor the valve temperature:
a) Temperature sensor mounted into the valve cover
b) Temperature sensor mounted in a sleeve installed in the valve pocket through a drilled valve cover.

In both cases, typically one temperature sensor is installed on each valve cover. The valve pocket temperature b) provides a higher quality of the measurement in terms of early detection of the leakage. Valve cover temperatures are subject to bigger influences from environmental conditions such as sunlight or wind. Temperature sensors installed through the valve cover into the valve pocket provide an earlier indication of changing temperatures. The signal coming from e.g. RTD or thermocouple sensors can be transferred to the Distributed Control System (DCS), to a PLC or machine monitoring system (MMS). In the DCS or MMS, temperatures can be trended to generate long-term information about the valve condition. Condition monitoring systems provide additional analyses for the signals such as the group deviation analysis to maximize the value of valve temperature monitoring, no matter at which position (cover or pocket) the sensor is mounted.

• pV diagram monitoring pV diagram analyses requires the installation of one pressure sensor for each compression chamber to monitor the condition of suction and discharge valves. The sensors can be installed on indicator taps prepared by the machine manufacturer and may not be mixed up with the suction and discharge cylinder pressure sensor installed in the pulsation dampers or piping. Such taps are required for API 618 machines and typically indicator valves are installed between the sensor and the cylinder to allow easy replacement of sensors without machine shutdown. If indicator taps are not available, e.g. old machines, suction or discharge valves can be modified with a special centre bolt.

pV diagram analyses are based on dynamic pressure measurement and require sensors that allow sampling rates in the kHz range to allow detect small leakages and high frequency pressure pulsations caused by valve dynamics or stepless unloaders. The pV diagram can be visualized with suitable software. Intelligent diagnostic systems automatically monitor the pV diagrams. In addition to the suction/discharge valves, the pV diagram analysis indicates leakage of various other sealing elements such as piston sealing rings, piston rod packing etc.

Furthermore, the dynamic cylinder pressures in conjunction with other parameters, e.g. speed of the compressor, connecting rod ratio and weight of the piston, allows the calculation and monitoring of the dynamic piston rod load and its reversal periods. Piston rod load is amongst the most critical when monitoring the condition and integrity of a compressor and help to identify critical overload conditions.

• Cylinder acceleration monitoring The third method of interest involves acceleration sensors typically mounted on the cylinder. For permanent installation, cylinder vibration sensors can either be installed by screwing them to a drilled thread hole in the cylinder or using a drilled mounting pad that is glued to the surface of the cylinder.

Combining this piezoelectric sensor with suitable software enables users to identify failures of suction and discharge valves based on segmented vibration monitoring, making use of individual threshold monitoring for each segment. The segmented monitoring requires only one vibration sensor per cylinder to monitor four groups of valves (suction and discharge valves on head end and crank end side) on a double-acting cylinder.

Strengths and weaknesses
What are the criteria to be evaluated when deciding about the best suitable method for valve monitoring? Starting point, typically, is a cost - benefit analysis. Examining solely the investment for sensors and the monitoring system is insufficient. Additional effects have to be taken into account such as MTTR optimization, increase in machine uptime or the system’s ability to detect critical failures other than valve problems.