Happiness is a plant that's flexible

Making sense out of complexity, itself brought about by the need to diversify and to create flexibility and responsiveness, is never easy – but it is getting easier. That's certainly what cigar manufacturer Gallaher found when it opted for an advanced planning and scheduling (APS) system to manage its multi-stage, multi-product factory in Cardiff, South Wales. And within a few months of going live it's already seeing costs falling significantly. Julian Morgan, Gallaher's resources planning manager, describes the firm's production processes as very complicated, with dependencies and bottlenecks that made end-to-end planning using spreadsheets a nightmare. In brief detail, we're looking at five distinct sub-processes in separate parts of the plant that take blended tobacco and leaf wrappers from raw materials to packed products – with thousands of routes and options according to the materials themselves, the unique SKU and even its destination. Processing starts with leaf conditioning and blending to create the filler material. Next is production of the cigar rods on high speed rod making machines, with eight lines running four sizes and types. Then comes 'over-rolling' – putting the leaf wrapper around the rod – for which there are some 80 fixed-function machines of varying types running at up to 400 items per hour. Fourth is pressing and trimming, which creates the section and style of cigar, according to national taste and brand, running on a mix of 12 conventional presses and high-speed machines. And finally, we're into the packing hall, with 10 lines. Sounds simple? It's not: over-rolling is the most difficult, with around 900 permutations of route there alone, even with slower, same type machines grouped. Machine choices depend, for example, on operating speed, size of rod, leaf die profile, left or right hand leaf wrappers, and even the blend itself. "Aromatic blends, for example, can only follow certain routes, to avoid cross contamination," explains Morgan. And raw material leaf length, preparation (bobbinised or not) and weight also matter – not only for machine choice, but for order tagging. Hundreds of permutations Clearly, getting lines balanced, with the best mix of machines according to materials and cigar SKUs ordered, is a considerable task for planners – within over-rolling and the other sub-processes. But there's also the matter of optimising for the big picture so that the right product is produced in the right volumes at the right time for each of the processes to match demand at minimum cost consistent with service levels. Until this summer, that was being done on manpower planning spreadsheets – a throwback to when the processes were more labour-intensive. "Each of the stages was managed independently in silos with its own KPIs around output and efficiency, but not quality of adherence to customer delivery performance." And the net result was the usual over-production of some items and stock-outs on others. "We couldn't do better with spreadsheets; it was just too complex." So 18 months ago, Morgan was tasked with finding a solution – and to tackle the separate issues of longer range demand, capacity and business planning. "Expanding into Europe and adding, for example, some of the aromatic blends to our range – and also equipping ourselves with more modern machinery – changed what we needed to do," he says. Morgan worked at spreadsheet planning for several weeks before deciding there must be a better way, and trawling the Internet to find it. And after looking at several potential solutions, including SAP's APO (advanced planner and optimiser), not least because SAP is the corporate choice and APO is in operation elsewhere in the business, his team opted for Workplace Systems' WorkPlanner APS as the best fit. It's been a resounding success – initially quite easy to set up, but also now easy to run – and as a result it's also getting attention and interest considerably beyond the original remit. Like so many that start with APS in the production arena, as the information gets better and more reliable, and the facility starts to perform better, the penny drops and others think about how it could help to improve their performance too. Solves multiple problems Morgan says it took a team of three people six months, doing everything from deriving the spreadsheet data to establishing routes, dependencies and bottlenecks to validating the APS model. The resulting system takes input from Gallaher's UK forecast sales, and SAP APO for the export market, and generates a 36 month plan for business guidance, with year one more fixed, and months one and two fixed quota. "We ended up with a tailor made solution, which was very easy to build and is now also easy to use," he says. "It's spreadsheet-driven, but the power of the APS is in the speed with which it transforms that data into robust, optimised plans." So how's it performing and what does he see as the benefits? "It's very good," says Morgan. "We went live in July and we've been running planning so far on a monthly basis, generating cost-optimised and capacity-constrained production plans for all the sub-processes as they relate to one another – and reaching out upstream and downstream. We'll take that up to weekly planning in December, but already finished goods stocks have reduced by 20%, and that will go to 35% by the end of this year. WIP [work in progress] reduction has not been as significant yet, but it will improve when we take scheduling weekly." But there's more. Planned maintenance is due to be linked into APS, with the objective of getting it away from expensive weekend working and into normal production hours. "The APS system has the freedom to move production up and down across the sub-processes, and to adjust routings and machine utilisation to re-optimise for machines taken out for maintenance," says Morgan. "Without the system, it would be impossible to calculate effective plans on spreadsheets." Beyond that, the system is also enabling planners to respond quickly and accurately to 'what ifs' around forecast inaccuracy, or new business, new brands, campaigns and the like. "In the last 18 months, we've had 30 brand launches into a variety of European markets, where the shape, style and packing of the product have been different. We can now support this kind of development right from the start," says Morgan. "The system can see manpower and machine capacity, so as we build new routes and suggest extra volumes, we can instantly see if production has the capacity and the location of the pressure points… At the moment, we still assume that non-tobacco materials will always be available, but now we're also assessing the case for integrating that into APS. They would just become part of the BoMs." And interest and applications for the APS system are growing, with the system likely to drive better decision support against real numbers. "We can send the requirement for leaf wrappers three months out to our leaf buyers… The system can also deal with recipes, which has beneficial application to our blend requirements planning. This is currently being considered by the leaf purchasing department." Incidentally, Morgan's feedback set-up is instructive. He indicates that shopfloor data collection (SFDC) is limited to machine status: "We don't need SFDC and we don't need full finite scheduling either – weekly buckets are enough. Our manufacturing cycle time is 10 days anyway, mostly because there are two long conditioning periods." In fact, finished goods stocks at the warehouse in Crewe provide feedback. "APS has stock targets and that live loop is enough." It's about meeting the business goals and loading production accordingly. "If sales vary from forecast, that's reflected in the finished goods stocks and the planning system responds automatically at the next run to meet the demand." As for the future, Morgan says he's confident the system will pay for itself quickly and is flexible enough to allow for change. "We have been left out of the SAP system in terms of planning at Cardiff. But if ever we did go for full SAP, our APS system could be integrated with it." And he indicates that for now that would be his preference. "It's an effective solution: it's like a lightweight reconnaissance vehicle as opposed to a Chieftain tank. It's very good for the business now that we're being asked to plan for making more, smaller and more complex batches. But either way, it means we have a choice if and when we get to that point."