Predict, react or prevent?
July 1st 2007

Anthony Mayall, process control systems, Siemens A&D, looks at the case for predictive maintenance and the impact this can have on a company's bottom line

For many companies, maintenance is still viewed purely as a cost factor.

However, research shows that it can cost as much as 12.5% of revenue in some industries. For some companies the maintenance budget is therefore higher than their profit margin. So it is all the more surprising that too little attention is still being paid to defining and implementing an optimised maintenance strategy.

What is predictive maintenance? Generally there are th ree approaches to maintenance; Reactive, Preventative and Predictive. To use a motoring analogy, Reactive is when you wait for the car to breakdown before any maintenance is carried out. The cost of this approach is the danger of breaking down (safety), loss of the car whilst an unscheduled repair takes place (downtime) and potential additional damage due to catastrophic failure (consequential damage). Preventative maintenance is when you change the oil and plugs every 10,000 miles, break pads every 20,000 miles and disks every 30,000 miles. The cost of this approach is that you have to pay the parts and labour costs irrespective of how the car has been driven, in other words you might be routinely replacing components that are okay. Predictive maintenance is where your modern vehicle has the technology to monitor the way in which it is being driven, plus enhanced sensors, to enable the prediction of when specific maintenance functions should be carried out. The cost of this is a slightly higher capital investment but you may not change your oil and plugs before 25,000 miles.

Why implement a predictive maintenance strategy? In a real production environment, cost factors of 1 for reactive, 0.5 for preventative and 0.1 for predictive are widely reported. To put this into context, studies show that greater than 55% of plants are operating a reactive maintenance strategy (if you can call that a strategy!). If you took an average maintenance spend of £1 million, there is a potential £500K saving for a preventative approach and a staggering £900K saving for predictive maintenance. A capital investment of £900K would facilitate the technology required for predictive maintenance in many plants. I doubt that more than 10% of businesses operate a predictive strategy, but don't beat yourselves up about it, it's only in recent years that the technology required has been used in earnest. What about the technology? The ability to predict failure requires an intelligent process control system including sensors, actuators, drives and switchgear.

Traditionally, all these components would have been hardwired making the delivery of diagnostic data very expensive, or in many cases impossible. Fieldbus technology, such as Profibus, means intelligent components are 'networked' together – this opens up the opportunity for exchanging a broader range of information. There are now over 14 million Profibus devices installed, more than any other Fieldbus, this means that many plants have the infrastructure required to implement predictive maintenance strategies.

The data from the Fieldbus devices is transformed into information by integrating asset management software into the process control system. An asset is determined by each application, it could be a pump, a tank, a reactor, a distillation column etc. A profile for each asset can be built to deliver maintenance information to the operators and maintenance staff, for instance if an intelligent level probe becomes contaminated (but is still operable), it can trigger a maintenance operation before it fails and production is lost. Similar to the motoring analogy, intelligent switchgear can deliver a predicted "life remaining" figure and maintenance can be scheduled accordingly. Without this, contacts and components would be routinely replaced, or worse replaced on failure.

What are the benefits? In case the £900K saving didn't draw your attention there are additional benefits to predictive maintenance. A tangible benefit is quality – end stops moving on valves, compressed air leaks and poorly performing pumps all have an impact on product quality, keeping your assets within specification ensures consistent quality which is essential industries such as Pharmaceutical. Another measurable benefit is energy savings, the aforementioned simple examples all contribute to inefficient process control which equals wasted utility consumption.

A softer benefit is gaining the involvement of plant operators, if they are trained on a process control platform with integrated asset management, they can be key contributors to a predictive maintenance strategy. Many will identify with a breakdown scenario where the operator kicks back and switches on the kettle "it's a maintenance problem" – with proper training and a collaborative approach this could be a thing of the past.

More articles from Siemens Moore Process Automation: