Plan, predict, profit

At Ayrshire-based family-owned distillers William Grant & Sons, the installation of a motor condition monitoring system has seen run times reduced by 20%, due to improved performance and a subsequent reduction in energy consumption. "Through analysis and diagnosis, we were able to make the necessary changes and mechanical responses to shorten the processing time and achieve the reduction," says Andrew Napier, process team leader of the engineering team at the distillery, which produces one of the most sought after grain whiskies in the industry. The changes in question, he explains, followed an in-depth three-month evaluation, which saw intelligent predictive maintenance and condition monitoring technology from Cambridge-based specialist Artesis put through its paces. The motor condition monitoring units were installed on three separate yet critical machinery and plant applications, spanning a wide range of motor sizes – 22kW up to 250 kW, as it turned out. It was these motors, experience had taught, that were most prone to reliability issues and which, in the past, had significantly impacted both productivity and profits. And so, over a three-month period, intelligent sensors monitored the power consumption and load on the three motors selected for the process: an agitator on the batch cooking vessel, the drive motor for hammer milling of malted barley, and the drive motor for hammer milling of wheat. The role of the sensors, explains Andy Bates, a director of Artesis, is to 'learn' the normal operation of the machine in question, non-intrusively recording its performance in minute detail, and reporting back to a database via the plant's own in-house network – be that ethernet, 8.211, Zigbee or GPRS. The database in question, he adds, can be any form of SQL-based database obeying the OBDC standard that the plant happens to have – MySQL, Microsoft SQL, or Oracle, for instance. "So there's no need to buy another licence, which can be quite a saving," he notes. "In effect, we're using the motor as a transducer, which means that the motor itself is the sensor," he says. "And by creating a multivariate model with three axes, load, speed and phase, you can detect a lot of problems – cavitations, flow turbulence, bearing imbalance and so on. Given that 90% of industrial equipment utilises three-phase motors, it's a very useful technique – and significantly cheaper, and less intrusive, than traditional condition monitoring approaches such as vibration analysis." So it transpired. Over the period, the motor condition monitoring produced a significant amount of data, which Napier and his team duly set about processing and analysing, performing the task themselves, albeit with guidance from Artesis as and when needed. With a clear signature of what 'normal' operation looks like, it's now very clear that a condition in either a motor or the equipment that it is driving will be detected in ample time to precipitate maintenance actions – an obvious benefit, and one of the prime objectives of undertaking the exercise. But intriguingly, the analysis – and not unusually so, stresses Bates – highlighted other opportunities for improvement. For example, the distillery could more accurately predict when the contents of the batch cooking vessel were properly cooked, thus implementing a stop time that was 20% sooner – which also usefully triggered a matching reduction in energy consumption. Likewise, it proved possible to change the milling screens in the malt milling application based on electrical signal analysis rather than personal judgment. And further installation of more motor condition monitoring units is planned, says Napier. "We will now be installing the units on two troublesome grain elevators, both of which are outside and go up to around 30 metres in the air," he explains. "They are known to fail up to two or three times a year without warning, so this will give us a real insight and enable preventative action before failure – saving us downtime, expense and loss of production." Writ large, the William Grant example cuts right to the heart of many of the issues facing today's new generation of computerised maintenance management systems (CMMS) and enterprise asset management (EAM) systems. Such systems – essentially handling maintenance, technician and spares scheduling – have been around for several years, gradually spreading their capabilities beyond earlier CMMSs. Now web-enabled, and with links to both ERP systems and plant floor equipment, their promoters enthuse as to their capabilities. Better planned and scheduled maintenance, for example, reducing downtime through both breakdowns and machinery being off-line for maintenance. And better utilisation of maintenance technicians, who find themselves 'level loaded', and responding to maintenance calls, not urgent breakdown calls involving overtime and weekend working. What's more, there's usually a lower inventory of spare parts and maintenance consumables, through a better understanding of what parts and components will be required, and when. "There's a growing acceptance that manual methods, and old-style CMMS tools, aren't good enough," says Matt Muldoon, vice-president for product marketing at ERP vendor Epicor, which offers its customers one of the latest and most capable ERP-based EAM systems. "Increasingly, manufacturers are seeing slickly-managed and scheduled maintenance as a way of generating extra capacity, without having to pay for new machinery. Why buy a £100,000 machine before you've explored how much extra capacity a few thousand spent on some maintenance software can deliver?" Yet such opportunities aren't necessarily available to those businesses taking only a very narrow view of what a CMMS or EAM application might offer them. "Many companies make the mistake of selecting a CMMS or EAM application that addresses the short-term needs of primarily the maintenance department," warns Andrew Kinder, director of product marketing at ERP and EAM vendor Infor. "This is not the best approach given the number and degree of changes that most companies go through over a relatively short period of time. Instead, CMMS and EAM requirements should cover a minimum period of three years, and span the requirements of departments other than maintenance – such as operations, engineering, purchasing, finance, IT and other stakeholder groups." Yet for every manufacturer who views improved maintenance as an opportunity waiting to be exploited, the prosaic reality is that many more see it as an interruption and a hindrance. In short, maintenance remains a Cinderella function. Very much the 'poor relation' to production, maintenance activities are often carried out when production deign to release the machinery – and not when the maintenance department has planned to work on it. The costs of overtime and weekend working might remain the same, in other words, and it's little comfort to know that the maintenance engineers in question are working on maintenance, not breakdowns. And – perhaps most significantly of all – there's been a gradual acceptance that planned maintenance, carrying out maintenance activities at fixed intervals of either time or production volume, isn't necessarily the best or most economic way of scheduling maintenance. The familiar 'bath tub' failure curve, in short, turns out to only apply to a minority of industrial equipment – perhaps as little as 10%. For the other 90%, there's no significant uptick in failure rates when maintenance intervals are extended. It's a reality that has prompted the rise of condition monitoring as a solution. And, as at William Grant, maintenance is done when sensors detect that maintenance needs to be done, rather than to the diktats of a fixed calendar. Worse, without that fixed and planned calendar, some of the assumed benefits of regular planned maintenance also disappear or are reduced: better inventory management, better labour utilisation and less weekend and overtime working. Two particular long-term trends, then, are shaping the future of EAM and CMMS going forward. First, there's a growing understanding of the need to connect condition monitoring equipment and systems directly to factories' EAM and CMMS tools. GE's popular Proficy plant automation and historian platform, for instance, provides precisely that capability in the shape of its Proficy Maintenance Gateway, says Coleman Easterly, product general manager at GE Intelligent Platforms. "Proficy has the capability of tracking process parameters, quality parameters, and a wide variety of events, including alarms and downtime," he says. "And with Proficy Maintenance Gateway, the intelligence derived from this can be used to trigger maintenance activities, based on conditions and rules that can make more sense than intervals of time and cycles alone." Proficy can then trigger maintenance work orders in third party applications such as IBM's Maximo and SAP's EAM solutions at the appropriate state of the equipment, he adds – after as much usage as possible, in other words, but before degradation has advanced to the point that excess power consumption, poor quality, or even failure results. "In this way, Proficy magnifies the value of EAM and CMMS systems with more precise and automatic monitoring and triggering of work orders, while reducing the clerical burden on maintenance and production personnel," notes Easterly. The second long-term trend shaping the take-up of EAM and CMMS applications is a growing acceptance of the need to make sure that planned maintenance actually takes place on time in those instances where it is merited. In some companies, that means adapting EAM and CMMS scheduling to take advantage of smaller windows of opportunity. At GKN Wheels, for instance, plant engineering teams try hard to minimise the impact on production by carrying out maintenance while equipment is down for tooling set-ups. "Changeovers can take six hours – and on older equipment, up to 12 hours," notes continuous improvement project engineer Steve Burgess. "So we try to carry out planned improvement work, or small maintenance jobs, while the equipment is down for a changeover. That way, we get the work done without affecting output." Elsewhere, maintenance scheduling is actually carried out by the system that schedules mainstream production activity – an output of the plant's main scheduling system, in other words, and not an input to it. Take, Lincolnshire-based potato processor Solanum, which supplies potatoes to a number of UK supermarket chains. Conveyor-based sponge drier roller tables handle 57,000 tonnes of potatoes each year – with the bearing races on each table being prone to water and grit ingress, which dramatically shortens their lives. In theory, a twice-yearly bearing re-build would significantly cut unplanned downtime, says Darren Mortimer, Solanum's operations director. In practice, he concedes, temporary repairs were more usually the order of the day. "We were completely reactive," he recalls. "When pieces of equipment broke, we'd repair them – with the engineers under pressure to give them back to production as quickly as possible." No longer. These days, an advanced planning and scheduling (APS) system from Chippenham-based specialist vendor Preactor builds maintenance and improvement activities into the factory production schedule. The result? They happen, and as planned. "We'd never have been able to do that before," says Mortimer. "The pressure to keep up production would have been just too great." The scheduling system even takes into account which lines are the most efficient substitutes for lines taken out of production, he reports, and reallocates work in a strict order of preference – a task that an EAM or CMMS application would find impossible.