Visible improvement

No plan survives its first contact with the enemy – and that enemy isn't your customers, your suppliers, your shopfloor or your employees. It's time." So says Dr Richard Price, a simulation engineer currently working at manufacturing systems developer FactoryMaster on a Knowledge Transfer Partnership scheme with Aston University, aimed at developing next-generation advanced planning and scheduling (APS) tools. "Pause and you never get that five minutes back," he continues, not only making the obvious point about time's centrality to the seven deadly wastes in lean thinking, but also – and just as important – reminding us that things change all the time. So what seemed an optimal plan for production five minutes ago may not be any longer. "That doesn't mean that planning and scheduling are futile. But it does mean that for an APS system to work well, it's critical that it takes into account what has happened as well as what you want to achieve." And we might add, 'and keeps doing so'. Sounds impossible? Not at all. What Price is alluding to is essentially twofold. On the one hand, a good APS system ought to be capable not just of detailed scheduling for best OTIF stats, but also of readily rescheduling as real-world events dictate – identifying resulting issues and solutions on the way. On the other, such a system should additionally help planners recognise where genuine improvements (or the reverse) can be, or have been, achieved – differentiating between what's realistic and sustainable versus optimistic or, frankly, poor practice. This latter capability is, he argues, key to guiding projects for continuous improvement (CI) and, as they build, also to cementing them back into the system – as long as the APS software is integrated with the company's ERP. Equally, it is central to highlighting requirements for training, new jigs, work centres that need some TLC, reworked sequences, revised SOPs, and so on. But getting this right demands that the APS software recognises a level of detail that goes way beyond primary resources. Instead, it needs to take account of secondary constraints, such as the people behind the machines, their skill sets, certifications and expiry dates, but also shared resources, tool limits or variable material handling limits. And the challenge then is complexity. Anyone familiar with establishing the rules for a constraint-based scheduler will understand that the more parameters you model, the longer it takes and the harder it gets to comprehend the end result. And, the more subsequent change is resisted. Hence Price's project, which he describes as building a framework that enables the easy addition of new constraints as they emerge, and then harnessing a constraint propagation algorithm to churn out results. This, he says, should go one better than standard predictive modelling techniques by remaining fully deterministic while also recognising that there are limits – and that time doesn't stand still. "The truth is too complex to capture accurately and you run the risk of becoming a slave to it," insists Price. "That's why we're not building a system that tries to find the optimal solution. We want it to deliver a good solution that's easy to understand. Then, when a production supervisor asks why this operation is scheduled to run on that resource, we can say, 'this is why'." Beyond that, he concedes that the user interface on FactoryMaster's new development – due out this autumn – is much the same as market-leading Preactor's latest, with intuitive 'drag and drop' screens. Why? Quite simply, so that managers can move jobs around on-screen to match their expert views of what works best in the real world. This, he says, is about engaging the workforce, but then advising of sequencing rules being broken, and/or upstream and downstream issues leading to inefficiencies and/or late customer orders. What's more, Price advises that the system is being designed to enable rescheduling around managers' desired changes, always striving for a new 'good solution'. And make no mistake: for Price this is not just about orchestrating the shopfloor. It's also about sensibly optimising work with contractors and JIT suppliers, and ultimately the management team – helping to inform decisions around performance and investment at one level, but also cost-down programmes and continuous improvement at another. But Price is not a lone voice in support of APS. Martin Clocherty, continuous improvement manager with Hayward Tyler, which manufactures large submersible motors to order for the power, and oil and gas sectors, is another firm believer in APS as a powerful CI tool. "I was in the semiconductor industry for 20 years, where all improvement projects are based on data. You need a good scheduling system to get that data. Without it, you're back into guesswork." That's why, when Clocherty joined Hayward Tyler three years ago, he was soon rooting for an APS implementation. What he inherited was Dataworks (now Epicor) Vantage 6.1 ERP, of Y2K vintage, with planning and scheduling only partly implemented at its Scottish site but unused at the Luton plant. In part, he explains, that was because of the sheer diversity of activities at the latter – with four business units looking after new-build, spares and repairs, as well as production of Varley double helical gear pumps. However, he concedes that Luton's perception of its own complexity probably got in the way – leaving site-wide scheduling on Vantage, with no visibility of detail, as the unchallenged norm. "I was quite amazed at not being able to use APS, so that was one of the triggers when we were upgrading to Epicor 9.05 last year," recalls Clocherty. "We had a close look at their scheduling system, learnt how it operates and what it could do for the business, and treated the implementation, including APS, as a fresh start." For him, that meant getting all the basics right first, including, for example, reviewing all parts lead times, and updating Epicor to enable a move to JIT material ordering capable of dovetailing with production in each of the business units. And it was a similar story with capacity modelling, recognising constraints for each of the resources – machines and people – to reveal the reality of overloads and under-utilisation. Effective planning, he asserts, needs to factor in everything from individuals' skill sets and holidays to planned plant maintenance. "Then you can use it for planning training," states Clocherty. "You can also see how much flexibility you have; how many hours you need for machines and machining centres; how many machine minders with what training you need... It's helped us make major improvements, in terms of direct hours wasted, because people didn't understand where material was and which machine it was against. Now we get the right people scheduled for the right machines with the right material at the right time." And he cites other benefits, ranging from live and visual supply chain purchasing, including subcontractor operations, to improved machine utilisation. In fairness, the latter has been partly due to an in-house developed machine performance management system, but it's also the result of Epicor's ability to display opportunities for improvement. It's a similar story at Great Yarmouth-based subcontract machine shop AK Precision – although, in this case, not driven by an ERP upgrade, but a requirement to escape from spreadsheets and tee cards. The company, which serves the sub-sea, offshore, aerospace, and oil and gas sectors, went for a Seiki scheduler to tame the classic challenges: products varying in size and complexity; orders from single items to batches of 10,000; a range of different machine centres; and flexible routing requirements. Add to those constant changes in product specifications, massively variable set-up and running times, and skill set variables, and you get the picture. Production co-ordinator Rebecca Dack says the firm needed shopfloor visibility, and describes the impact of Seiki as massive. "Now we can see not just every order on the schedule, but how they interact with each other. Previously, we might have made changes to address an issue, only to find out later that these were detrimental. Now we can make much better informed decisions." And she adds that while the preference is always to group products – to minimise set-up and changeover times and maintain smooth flow – the system's 'what-if' capabilities now show the whole picture, including changing commercial considerations. As for the benefits, operations director Anthony Taylor says that improved human and machine resource use has brought direct time and cost savings, including an increase in turnover and WIP down by 50%. That said, he also points to the system's ability to highlight where manual routings for new or existing products turn out to be wrong or inefficient. And as each machine now has an accurate work-to list, operators can see which job needs to be done and when, as well as work coming down the line. "Not only does this make the transition smoother from one job to another, but it also gives production staff an opportunity to input suggestions," comments Taylor. And he adds that, because Seiki enables managers to compare projected and actual schedules in real time (the system automatically logs start and stop times, and updates the schedule), it helps steer continuous improvement in several ways. With its ability to run 'what-if' scenarios, the company can investigate opportunities not only to move orders forward but also to embark on capital purchases. Already, it has installed a large screen in the planning office to assist with production meetings, and next up will be screens on the shopfloor. "By enabling everyone to see the schedule, they can see the impact of what they do – not just the job they're working on, but on the schedule and the company as a whole," comments managing director Wayne Read. But complexity isn't just the preserve of machine shops and project-based engineering houses: iconic food and confectionery manufacturer Mars has also recently invested in scheduling software to improve operations at its Basingstoke plant. Paul Hazelwood, supply logistics manager, explains that it's not just about wide product ranges and variants here, with multi-shifting machines and lines covering the permutations and volumes. It's also about countries requiring different formulations as well as different pallet sizes and quantities, and the way demand is specified. "When dealing with the UK and EU, we receive a forecast breakdown to an individual product level within a given time period, typically rolling 20 weeks," states Hazelwood. "For the US and Japan, however, we work on a make-to-order basis, again over 20 weeks, but with a four-week locked window." And he also throws in product grouping and allergen control, different storage requirements, as well as the need to minimise changeover and clean-down times. For this company, the combination of the Mars IS department withdrawing support for its previous scheduling system and its desire to become leaner led to the selection of Preactor APS and GMPS. And while that project is still in its early days, Hazelwood believes he has already seen marked improvements. Apart from lowering stock levels across the entire pipeline – on one SKU, reducing stock from 10,000 pallets to 4,000 – the system is also driving internal improvements, he says. "Preactor GMPS is driving internal behaviour change and we are definitely already seeing lean improvements. We have a long road ahead but there are undoubtedly big benefits in the future." But FactoryMaster's Price has the last word: "You shouldn't underestimate the power of data that helps everyone see this is what we planned, but this is what we actually achieved – and then validates the 'this is what we plan to do about it' part of the equation. That's how a good APS system becomes a driver for the business." And he adds: "If something has slipped in the plan and you can see why, then you also have the benefit of 20-20 hindsight. You can investigate, see how to do better and communicate that with everyone that needs to know, using real data. Then you can feed that back into the next round of scheduling. Good APS is a powerful tool for CI."