Salads on schedule

Adjust font size:

Magazine Article, Source : The Manufacturer
Zone : IT in manufacturing
Published : September 2004

Malcolm Wheatley takes a look inside the Geest Group to see how APS has been helping smooth the food industry’s scheduling problems

W ingland Foods, part of the Geest Group, produces pre-packed leafy salads for supermarkets such as Waitrose. Based in Sutton Bridge, Lincolnshire, the company faces some unusual scheduling challenges, as Wingland’s planning manager Richard Thorold explains. Demand is volatile: a spell of warm sunny weather can see consumers’ menu planning swing away from comforting casseroles to lighter dishes incorporating salad items. And, just as quickly, demand can slump once the warm weather ends, which in Britain, can mean anything from two days to a week after it started. To cope, supermarkets place their orders as late as possible, which places its own challenges on a supply chain based around sourcing freshly-grown salad items from farms and specialist providers across the UK and the continent. Wingland’s orders, for example, arrive the day before they must be despatched. Inventories aren’t really an option. Undeniably, it’s a tough industry in which to work: companies accustomed to long order books and demand forecasts stretching weeks or months ahead would blanche to know the realities under which many businesses in this sector operate. But it’s at the shopfloor where the most exacting challenges are found, and where the rubber really meets the road. As Thorold explains, synchronised streams of differing items – cooked ingredients such as rice and pasta together with dressings, mixed and prepared leaves – must combine at the packing line in the exact proportions required at the exact times required. The trick, he says, is to hit the ‘outlet time’: the time at which the trucks heading for particular distribution depots must leave in order to be cross-docked on roll cages in time to reach the supermarket shelves. And the further away the distribution depots are from Wingland’s base in Lincolnshire, the earlier the outlet time. Unlike in the automotive industry, where the regular beat of the assembly line provides a helpful rhythm, the extreme volatility of demand provides little such comfort. In other words, just because you managed to meet yesterday’s demand, doesn’t mean that the same tactics will enable you to cope with today’s demands. It’s shopfloor scheduling at its most challenging, walking on a daily basis a tightrope with a slip-up meaning disaster. And it’s a challenge that for some years, Wingland, like other Geest factories in similar lines of business, had become used to meeting with the world’s most popular scheduling software tool: the humble spreadsheet. Not that using spreadsheets is wrong: countless factories use them. Both in terms of the graphical representation of their results, and in the way they work, spreadsheets mimic the ‘master schedules’ that were beloved of factory schedulers in the pre-computer era. But if the name of the game is producing truly optimised manufacturing schedules, spreadsheets have two weaknesses. The first is data integration. In an environment with complex bills of materials, or worse, complex bills of materials that are constantly changing, then keeping the information in the spreadsheet synchronised with the information in the rest of the business’ enterprise systems can be difficult. At Wingland, for example: “There was quite a bit of manual adjustment required on a regular basis,” concedes Thorold. “If someone made alterations to information stored in the Bills of Materials in the MRP system this data had to be noted in Excel. But because this couldn’t occur in an automated fashion much time was wasted in adding the information afresh by hand and trying to follow through links with other systems. It was quite a task making sure these links were robust.” And despite the undoubted power of the formulae that one can build into a spreadsheet to process, the algorithms required for a spreadsheet to become a truly capable shopfloor scheduling system are in another league, way beyond what Microsoft Excel or Lotus 123 can offer. These tools can do the job, but when it comes to optimised shopfloor scheduling, the key word is optimise, and that calls for software of a different breed. Software with such a capability is increasingly known by the term Advanced Planning and Scheduling (APS), and it was one such system, from UK-based vendor Preactor International, of Chippenham, to which Wingland turned. Preactor, explains its managing director Mike Novels, “is a scheduling company, not an ERP company. Many of our customers are ERP companies, in fact, who use our scheduling capabilities to provide APS functionality to their customers.” Such specialism, in fact, is not unusual in the APS field. Unlike other areas of functionality, where the major enterprise systems vendors – companies such as SAP, Oracle and PeopleSoft – have been able to spot emerging markets and capitalise on them with systems of their own, APS systems (and particularly shopfloor APS systems, as opposed to the subtly different problems of supply chain APS), have proved trickier to master. In part, admittedly, this has been because for a large number of manufacturing companies, the basic capabilities of MRP-based scheduling have proved adequate, despite its limiting assumptions of infinite capacity, backwards scheduling and fixed-duration lead times. As a result, the market historically hasn’t been huge, especially given the investment necessary to create a solution capable of winning out over a specialist APS vendor in a head-to-head selection exercise. And that investment shouldn’t be underestimated: the algorithms involved in APS systems are complex, and in many cases protected by patents. Manugistics and i2, two other leading vendors, have for example both made acquisitions in recent years in order to acquire the patent-protected intellectual property of smaller competitors who had come up with better algorithms. Not that the complexity of the solution is necessarily manifest to those who use an APS system. Despite the mathematics embedded in the software code, users generally find APS quite intuitive. “The drag and drop function means schedule adjustments can be made very quickly, which is a great help both for our schedulers and the personnel on the shop floor when planning their day’s activities,” says Wingland’s Thorold. “The colour coding is also very useful. It’s probably the best visual tool I’ve come across in this industry: it tells us when to start manufacturing orders in order to meet the outlet time, and everybody knows where they need to be at a given point in time in order to hit the schedule.” Not surprisingly, after a slightly rocky start – despite being twenty years old, APS systems first came to real prominence just as sales of enterprise systems plummeted in the run-up to Y2K and the subsequent dotcom crash – experiences like Wingland’s are now encouraging others to take the plunge. “The electronics sector is coming back, as companies start once again to optimise their operations, and the automotive sector is starting to use APS to improve vehicle availability,” says Michael Gamper, vice-president of strategic consulting at Manugistics. “Increasingly, industries such as consumer products manufacturing and pharmaceuticals view APS as something that they simply have to have.” But the news is not all good. “Aerospace and defence are still very MRP-centric,” he concedes. And in industries where APS has been common for some time: “The market is maturing: APS has become something of a commodity, and the deal size is becoming smaller,” he notes. Even so, observes Sharon Crawford, product manager for System 21 Aurora with enterprise system vendor GEAC: “More and more manufacturing companies are waking up to the fact that an APS capability would be useful.” As a result, while GEAC has hitherto licensed APS technology from Frontstep (now part of process-industry enterprise system vendor MAPICS), the company is once again contemplating whether creating its own APS solution would make commercial sense. Other vendors are likely mulling similar steps. Typically, adds Preactor’s managing director Mike Novels, the companies that are waking up to what APS can offer, and so creating this interest among system suppliers, tend to be those manufacturing companies where sequencing is an issue. In other words, where dependent set-ups are involved, or where pipelines or vats must be purged, or where the ratio of changeover time to production time is high. But while APS can help in these situations, APS technology also places demands on the manufacturer. Chiefly in the form of a need for accurate data. With complex algorithms making exacting calculations in order to generate an optimised schedule – whether in terms of inventory reduction, due date adherence, profitability or capacity utilisation – APS is a classic instance of the need to bear in mind the aphorism ‘garbage in, garbage out’. And costly garbage, at that, if the resulting schedules actually result in decreases in profitability, due date adherence, and capacity utilisation. It’s chiefly for this reason, explains Steve Mabbott, supply chain manager at Macclesfield-based HW Plastics, that HW Plastics’ own implementation of GEAC’s APS solution has been delayed. “We realised that the data we had on things like structures, bills of material, and production rates simply wasn’t good enough,” he says. More than that, the data was overly complex, with approximately 17,000 separate part numbers on file. “We needed to re-think the whole way that we conceived and planned structures,” he adds. Now, thanks to concentrating on identifying the business’ runners, repeaters and strangers, and clearing out unnecessary complexity, the number of part numbers is down to just 7000, a reduction of almost 60 per cent. “At the end of the year, we’ll take these simplified process and structures and start using APS,” says Mabbott. “We’ve seen some reduction in inventories just from the simplification exercise, but we expect a lot more once we’re running APS.”