Are you taking your plant seriously

Enterprise asset management (EAM) systems – maintenance management systems to you and I – are well understood. Most organisations with expensive capital assets, whether that's CNC machines or more substantial process plant, run at least scheduled maintenance programmes. The more advanced link those into production planning, for obvious reasons. And the even more advanced also indulge in inventory and procurement management linked to ERP to cut costs against the old ad hoc ways. The very advanced take it even further, and model their risks and plant and factory bottlenecks so that appropriate inventory (with an aware supply chain) and skilled technicians are held and ready near set locations. And they use portal technology to enable visualisation and collaboration across departments: IT-assisted lean thinking in maintenance, if you like. We're also all familiar with predictive, automatic sensor-based maintenance regimes for the very highly asset intensive industries and the utilities – and of intelligent sensing systems that provide diagnostics, not only of specific plant problems, but their own behaviour. Some of that is slowly filtering down the 'food chain' – becoming more affordable and applicable – so it's worth just being aware. Point is, this is a mature field: well, up to a point. When it comes to plant and factory equipment just below that clear 'high capital investment' tag, or where machines are seen as relatively simple, you're far less likely to see investment in maintenance. Manufacturers typically haven't spent a lot at this level because, well, what could any system tell them that they couldn't work out for themselves? The cost/benefit equation just doesn't stack up. Right? No, not any more. Just recently, we've been hearing stories of manufacturers that have applied what some might term MES (manufacturing execution systems), but are actually machine and maintenance monitoring and analysis systems – rather more than SFDC (shopfloor data collection) systems – that can and are paying for themselves very quickly on lower level kit. Effectively, they're creating more capacity by reducing downtime, cutting costs and reining in overtime spend. In the best examples, they're also improving plant flexibility and deferring capex. And the systems involved, and indeed their applications, are remarkably simple. A great example is at Kalon Decorative Products, part of the SigmaKalon Group, which manufactures paints – its own Leyland and Johnstone's brands, as well as own-brands for Homebase, Wickes, Wilkinsons, Focus and B&Q. The firm has two manufacturing sites four miles apart at Birstall and Morley in Yorkshire, and employs over 1,000 staff. Birstall is by far the larger, with around 40 paint filling and lidding machines, for example, strategically dotted around the facility, compared with Morley's 12. We're talking high volume production of a growing range of coatings in smaller and smaller batches to match demand as closely as possible. Giving a quick idea of operations, the plant produces base paint semi-continuously for storage in 30,000 litre vessels, either for white sales or for tinting. The most popular products and colours, such as magnolia, run as large scale, direct feed production, and are set up to provide the required short lead times with minimum stock. "It's a limited colour range, and we don't have to schedule production and filling of that alongside the slower moving products," says John Hart, production manager at the Birstall plant. Which brings us to the rest: intermediate blenders around the plant, each running families of colours and product types – taking the base paint, with manual pigment additions according to recipe under Datastore production management control, followed by discharging to adjacent filling machines in groups. It's a highly scheduled environment under bespoke AS/400 control, with batch quantities ranging from 500 to 12,000 litres requiring a lot of changeovers using flexible hoses to enable the sequences of production, cleaning and the rest. In fact, it was the pressure of smaller batches, more variety and increasing variability of orders that originally challenged Kalon's old operating procedures, led to some restructuring and got the company re-examining its operating costs. Back in 1999, Kalon went through a programme of activity-based costing, and soon discovered a large proportion of downtime on the filling machines that was unallocated – up to 25%. The company already had a maintenance management system scheduling annual refurbishment and service etc, with information on costs and frequencies of machine repair to determine capex justification and so on. But Hart points out that no system can improve operations for downtime that remains unallocated. Kalon devised a spreadsheet system to get a more accurate picture but there were problems, not least that the system could only alert management after the event. Hart: "We'd get to know about it at the weekly review, and it's very difficult to get people to remember detail. So, we did some manual work as 'day-in-the-life' cases to identify the reasons for downtime. But it was very time consuming." Real-time management Last year the team came across MVI Tehnology's Eventsengine, which the vendor describes as: "a real-time performance management system that provides a human context to stoppages." It measures downtime to the second and forces operators to enter reasons by pressing line-side pre-coded touch screen buttons that also prompt corrective action. Operators can also key in other explanations, and buttons change as problems are resolved and others arise. The captured information is also accessible in real time from anywhere and can be subjected to root cause analysis. For these filling lines, it was superb. They're only semi-automatic, low-tech machines under PLC control, each with a team of four operators, but Hart makes the point that there are thousands of similar lines in factories everywhere, all subject to unplanned downtime. "Taking away the assumptions or ignorance of the causes can hugely impact productivity and capacity, no matter how apparently simple the machines." And that's precisely what it's done. Despite reservations by machine operators – evaporated by MVI's insistence on their involvement – the system went live on two filling lines in April this year. Since then it has been successfully identifying and quantifying downtime – enabling prioritised and rapid improvement. There are numerous examples. Hart again: "We suspected the filling machines were having to wait for pre-labelled packaging. The packaging department prints and applies enough labels for the planned run, but if the paint quantity in manufacturing goes up to meet specification, there might be an extra 200 litres, and the operators don't find out till they're out of tins. Before the system, we couldn't quantify the problem." In fact, the biggest problem revealed on one line was end of line pallet changeover. It's on a scissor lift, which lowers as the pallet is stacked: removing, wrapping and changing the pallet was accounting for one third of all unallocated downtime – several hours per week, and solved by introducing a second scissor lift. Maybe that should have been obvious, but as Hart says: "There was an assumption that the pallet changeover time was insignificant – so it wasn't looked at." Just so: that's what happens. But the downtime there alone amounted to 10—15% of weekly output or hundreds of thousands of litres per year in filling capacity. That's the key to this. "It's been one of the most important projects we've done," insists Hart. "The reduction in downtime allows us to produce more for the same, or the same for less. In that department we were close to full capacity so it's freed capacity. There has now been no need for overtime for four months!" Overall, he says: "We've improved machine utilisation by 14% and increased output by 5%. That's a really great performance because we've been running smaller containers with more changeovers and expected output to drop. So the net increase is nearer 8%. We've already saved about 15% on labour costs." In fact, he reckons that with costs of around £5,000 for the system and £2,000 for the hardware, ROI will be well within one year. He also makes the very important point that where factories have similar machines, implementing the system on a few is likely to show the issues on the rest. Quite where the optimum level of implementation is depends upon the degree of similarity both of the machines and their operating regimes – and the value of production going through. But it's clear that a small investment, especially in low tech plant productivity management, can make a big difference indeed.