Predicting the future
June 1st 2006

All plant requires periodic inspection and maintenance to ensure it operates at optimum performance. Castrol's 'Predict' service helps many businesses to reduce maintenance costs, maintain equipment efficiency and increase process productivity. In this article Chris Poole discusses the techniques and the benefits

Imagine how much easier our lives would be if we could predict the future? While most of us accept that the gift is beyond us, it is possible for engineers working in manufacturing industries to know what the future holds – but only for the machines and processes that produce their companies'wealth!

A recent survey of a number of industrial sectors estimated that 89% of maintenance costs are related to the mechanical wear of oil lubricated machines. With oil lubricated machines accounting for such a large proportion of the maintenance effort it's no surprise that the relationship between lubrication and mechanical wear is now used as a tool by services such as Castrol's 'Predict' to inspect and monitor conditions within machines and, with the development of the latest analytical techniques combining the principles of ferrography and spectroscopy, to predict future maintenance requirements.

So, why do we need services such as Predict? How does it work? What are the benefits compared with a reactive regime? And when should engineers move from a passive regime to a pro-active one?

Let's look first at why we need it.

The major culprit is wear in lubricated components, which can be classified into two main types – chemical and mechanical.

Chemical wear arises from reactions on the surface of the wearing material and the subsequent loss of that material by mechanical action. Typical sources of chemically reactive compounds are the liquids or gases being processed by the machine, the coolant fluids and the lubricant itself.

Mechanical wear is associated with friction, abrasion, fatigue and erosion. Under normal operating conditions a regime known as 'benign wear'occurs between oil lubricated moving surfaces. This wear is the result of chemical and mechanical actions releasing small particles from the surfaces.

Abnormal wear – an indicator of incipient failure – is associated with the release of significantly larger particles. There are many factors that result in the transition to abnormal wear conditions. These range from components simply reaching the end of their life cycles due to factors such as fatigue or excess bearing clearances, to overload conditions and lubricant contamination problems.

The particles generated by these different wear regimes have unique features that are directly related to the wear process and the resulting surfaces on the wearing components.

For example laminar particles are associated with fatigue from rolling contact and often contain holes, the result of being squeezed through rolling contacts after their formation.

To understand how predictive maintenance works we need to examine the role of the lubricant, which has many functions. Besides lubricating, it cools, protects against rusting and corrosion and prevents micro seizures occurring under extreme pressure conditions. It also keeps surfaces clean by carrying away debris generated from surfaces as they wear, and it is this characteristic that is used by Predict.

Because much of the maintenance effort is directed at minimising the affects of wear on lubricated surfaces, it makes sense to look closely at these surfaces by inspecting the particles carried away in the lubricant. There are a number of ways of doing this to provide an accurate picture of the key factors affecting machine maintenance.

The traditional analysis of 'used' oil, including wear metal spectrographic analysis, is an invaluable tool for assessing the current condition of the lubricating oil. If the samples are trended over a period of time the technique can predict the future condition of the oil, allowing for the timely changes that are particularly important for large machine systems.

However, predicting the future condition of the oil is the limit of this type of analysis and is of no use in predicting where abnormal wear may occur. This is because spectrographic analysis does not detect particles larger than 10 microns, which are the acknowledged signs of abnormal wear in machines.

Extending analysis beyond conventional oil condition monitoring requires the very latest 'predictive' techniques such as ferrography, which evaluates particles in a 2-stage process that ensures a more complete understanding of the machine condition.

The first stage is a quantitative analysis that provides an index of particles called the wear particle concentration or WPC, which is the sum of the large particles (>5 microns) DL and the small particles (<5 microns) DS. The WPC can be used to determine the percentage of large particles (PLP), which is calculated as follows: ((DL – DS) / WPC) x 100. The results can then be cross referenced against established industry standards.

The second stage is a qualitative analysis in which particles captured on a ferrogramme are examined under the microscope. It is at this stage that the true value of a predictive maintenance service is realised, when highly skilled analysts are able to identify the nature of the wear particles.

This is a critical stage in understanding the nature of the wear in a particular system. Particles generated by different wear methods have unique characteristics that must be identified to ensure that remedial maintenance is scheduled well in advance. The trending of this detailed level of information is what can give engineers an ability to predict the future.

The key benefit in having an effective maintenance strategy is that it is able to predict when the machine condition is going to hit point P, after which the machine will go into exponential condition decline and will need remedial work to ensure it doesn't reach point F – which is machine failure.

Another vital benefit is predicting how long the engineer has between the start of machine decline and reaching machine failure. Using a predictive maintenance service and monitoring its trends over a period of time will give a much better understanding of these important parameters and allow maintenance regimes to be scheduled more accurately and therefore, more cost effectively.

So, is there a right time to make the move from a passive to a predictive regime? As mentioned earlier, simply following oil analysis trends over a period of time will transform lubrication management from a regime focused on fixed and often wasteful oil changes – whether they are needed or not – to one that changes oil 'as required'. Large scale continuous processes are obvious beneficiaries of a predictive regime, which will protect equipment as well as productivity.

However, there are also benefits for medium and small processes, especially where expensive equipment is involved, or where delivery promises are critical.