Development of a predictive electrical motor and pump maintenance system

Abstract

This dissertation covers the development and implementation of a predictive maintenance monitoring programme for the Water Supply Directorate of the Department of Water Affairs, Namibia. The maintenance policy in the Directorate was based on a combination of breakdown maintenance and preventative maintenance. Thus maintenance was carried out when a specific type of equipment was forced out of production. The cost of the replacement and repair of equipment increased substantially and a condition-based maintenance system was investigated and implemented. The purpose of condition monitoring maintenance is to find a convenient time for maintenance to be carried out. Different types of condition monitoring technologies exist. After the different types of technologies have been investigated, vibration-based predictive maintenance was chosen. The project includes results from a number of field case studies and proves that vibration analysis can be used to determine the mechanical condition of electrical motors and pumps. The monitoring programme covers a total of 80 pump sets comprising mainly of electrical motors and pumps ranging from 45 to 2 400 kilowatt. In general, the programme is based on the determination of suitable monitoring parameters by taking measurements at regular intervals of the vibration characteristics of a machine. The generalised approach to vibration analysis in a predictive maintenance programme of machinery requires a sound understanding of fundamental theoretical concepts associated with machine. element dynamics and the nature of the dynamic forces and instabilities which excite vibration in electric motors and centrifugal pumps, together with the ability to plan concise experiments to obtain practical data regarding the cause of failure. Machine faults will cause a change in the shape of the vibration frequency spectrum. The cause of the fault can be diagnosed by determining which frequency components have increased and to match them with the different characteristics of vibration. Basically, all machines vibrate at the same characteristic level depending upon the machine's design and operation. As a machine begins to age and deteriorate, vibration increases sporadically or gradually and each machine, regardless of its mechanical design, creates its own unique vibration. A vibration problem can be analysed by reviewing its component frequencies and determining at what frequency the vibration occurs. Using a vibration analyzer, it is possible to measure the frequency and corresponding amplitude of each component. It was found that the greatest vibration normally occurs at the running speed of the machine. It can be concluded that unbalance could be a major cause of this. Misalignment was normally identified at two or three times running speed. Rolling element bearings produce their own high frequency with low amplitude vibration. Defects in rolling element bearings can be separated from the vibration produced by other mechanical components. On sleeve bearings, excessive clearances were found to be the main cause ofvibration, producing many harmonic-related frequencies. Another problem which may arise, is mechanical looseness, of which the amplitude is normally dependent on the amount of looseness and the mechanical design of the machine. This was characterised at twice the running speed with higher than usual harmonics. Resonance is another problem that could cause excessive vibration. Each part of a machine, as well as the machine itself, has a natural frequency and this frequency, relative to a machine's running speed, is of great importance since no machine should be operated in a resonant condition. By utilising a predictive maintenance programme such as vibration monitoring, the condition of vital machinery can be determined effectively. This monitoring system can give early warnings of impending failures, determine the cause of fault and can be used to schedule repairs. Such a system can therefore prevent catastrophic failure, lengthen the life of machinery and reduce maintenance costs. Since installation of the programme, the number of unexpected failures on monitored machines has been greatly reduced and the savings gained from the programme (savings associated with maintenance costs) enabled a pay-back on investments within 18 months of installation.