A Condition-Based Maintenance (CBM) strategy has the potential to give you maximum equipment reliability at minimum cost.
In order to adopt a CBM strategy, you need some means of monitoring the condition of your machines, and detecting the development of faults before the machine fails completely.
This fault detection needs to give sufficient advance warning that you can plan corrective maintenance to be at a convenient, non-disruptive time, and with all the tools, spares and staff all lined up and available to deliver the work efficiently.
While the human five senses are a very valuable, and often overlooked, means of detection and early warning of developing problems (you can hear and feel vibration, you can feel raised temperatures, you can see and possibly smell oil leaks, and you can smell hot electrical insulation or rubbing belt drives for example), these are subjective measures.
Objective, quantitative measures are required if you want to base future maintenance plans on the results, and this means you need some form of condition monitoring measurement.
Condition Monitoring experts like to refer to the “P-F” curve as shown below. If you track machine condition over time, the first point at which you can reliably detect that the condition has deteriorated, is an indication of Potential Failure, designated point P; as you continue to monitor, the condition continues to deteriorate, until eventually the machine is no longer capable of fulfilling its required Function, which is designated point F. Note this is not the same as complete, catastrophic failure; for example a water pump whose required function is to pump 1000 litres/min against a 10m head might still be rotating and pumping water, but if it is only delivering 900 litres/min against this same head, that is already functional failure.
The P-F curve is a useful concept, but is rarely seen exactly like this in practice; in the real world, condition monitoring signals are often noisy, and machines can deteriorate in an irregular manner, so sometimes the readings can look as if condition is improving for a time, before dropping again, creating a wiggly line, rather than the smooth, steadily falling one shown. This unsteady, real-world behaviour makes prediction of when maintenance work is required more tricky.
There are a number of things that can be measured to indicate condition – including temperature, pressure, flow, vibration, noise, motor current, and various others. Whichever measure is selected requires an appropriate condition monitoring system. The system needs to capture the readings, process and store them, display them, and provide a means to analyse the results, drawing conclusions and “so-whats”. Good systems should provide a means of diagnosing the nature of the fault, provide help to planning the appropriate corrective action, and give an indication of how soon this work will be required.
This last element should ideally be based on appropriate predictive algorithms, that take into account the rate at which the fault is developing, and the acceptable limit before intervention is required – which is Predictive Maintenance. Whilst there are many condition monitoring systems, most of them tell you the level of a particular measure, eg overall vibration level, but not many of them track the behaviour of the individual faults that are the cause of these overall signals; and even fewer provide Predictive indications, relying simply on the present level. Both of these elements (the nature of the fault and the time at which that fault will require maintenance intervention) are required in order to plan the appropriate work at the appropriate time, ie to adopt a fully Predictive Maintenance strategy. Faraday Predictive systems all provide both these capabilities, supporting you to adopt the most efficient maintenance practices.