Justification for machine condition monitoring

Justification for machine condition monitoring

Continuous operational condition monitoring implies that machine condition data are collected and evaluated in real time. Of special interest here are data that could rapidly change and are tell-tale signs of unexpected degradation of the condition of the machine. Vibration amplitude excursions indicate rapid changes and will give early warning of disaster. A timely shutdown will minimize possible consequential damage and the resulting extended downtime.

Both frequency and duration of shutdowns can be reduced with carefully targeted planning of maintenance work. Such detailed planning allows streamlined manpower and spare parts allocation. But, intelligent allocation requires information on machine condition to project wear progression and probable time to failure.

In condition-based maintenance, machines are only shut down if their condition demands it. Parts are only changed if a damage criterion is reached and the total anticipated survivability of parts in terms of remaining life and wear reserves are being exploited. Thus, then, achieving a reduction in material costs may only be possible if one has reasonably accurate data on the machine's condition. One obviously gages the existing wear potential in "traditional" wear parts, e.g., piston rings or packing, and thereby optimizes machine operating life.1

Efficiency is another indicator of machine condition. The primary purpose of efficiency monitoring in fluid machinery is to record changes in process or machine parameters that influence energy transfer to the medium being moved or compressed. Low efficiency causes fluid temperature rise and increases power consumption.

Reciprocating gas compressor efficiency may be determined, for example, by finding how indicated compression power relates to the driver power consumption.

What to monitor and why. Recall that the objectives described here are general and not specific to reciprocating compressors; they apply to numerous industrial machines. However, designing an effective strategy for machine condition monitoring systems parallels those used for reciprocating compressors. In these positive-displacement machines, their mechanical features—and perhaps their most vulnerable parts—must be analyzed in terms of maintenance frequency.

This was done by Dresser-Rand who, in 1997, made a survey of200 operators and designers on the causes of unscheduled shutdowns of reciprocating compressors. The results (Fig. 1) clearly indicate the relative maintenance intensity of certain component groups.
These statistics show that eight component groups were responsible for 94% of unscheduled shutdowns. Valve defects are obviously responsible for most of the unscheduled maintenance events. More recent experiences indicate similar results, even though the absolute involvement ratio of valves has dropped slightly due to the use of new materials.2

Monitoring systems are being marketed for every component listed in the statistics for rotating machinery, but the specific machine requirements must be considered when choosing a system. For example, analyses of entire subassemblies or operating point-specific threshold checks are appropriate to detect valve damage on reciprocating compressors.

As of2006, the three most important methods are:

•    Valve pocket temperature measurement

•    Vibration analysis

•    p-V diagram analysis.

Using suitable temperature probes and measuring the temperatures upstream and downstream of the valves is the simplest and most cost-effective method of determining valve condition.

LITERATURE CITED
1    Klein, U., "Vibration diagnostic assessment of machines and plant," Verlag Stahleisen GmbH, Dusseldorf, 1998.

2    Leonard, S. M., "Increasing the Reliability of Reciprocating Compressors on Hydrogen Services," Presented at the NPRA Maintenance Conference, May 20-23, New Orleans, Louisiana, 1997.

3    Nickol, J., "Reciprocating compressors in process installations: Availability of 24.000 hours, Utopia or reality?" Presented at the 1st EFRC-Conference, Nov. 4-5, Dresden, 1999.