Process companies’ reluctance to adopt MRP or its successors over the last two decades is hardly surprising. Even the least important differences between the discrete and process sectors – those in terminology – are far from trivial. In a paper published last year, Wonderware (once a manufacturing execution systems (MES) software supplier and now, with its absorption of fellow Invensys subsidiary and process-ERP developer Marcam, a key player in ‘shop-floor to top floor’ manufacturing management systems) lists almost two dozen terms with different equivalents in process and discrete manufacturing. But even the terminology matters less than the differences in architectures. Discrete systems are ‘convergent’ – they make one product or product variant from many, perhaps hundreds, of components. Process systems are ‘divergent’ – they make lots of products from one or a few raw material(s). Crude oil can be made into natural gas, rich oil, light cycle oil, grease, LPG or petrol. A chicken, bless it, can produce 20 to 30 separate end products which can be either sold as they are or used for further processing. In fact, there are dozens of other detailed differences. Process systems have to deal with: sequencing; hugely different units of measure; catch-weight, variable weight or random weight; lot and batch traceability; lot splitting and combining; concentrations; yields; and, last but not least, regulatory compliance – whether Hazard Analysis And Critical Control Points (HACCP) or Material Safety Data Sheets (MSDS). As if all that were not enough, the costing regime – a key ERP function – is particularly onerous. A bill of materials (BoM) based system can’t normally cope with more than one output at any level in the BoM. Process products are multi-output – co-products, by-products, waste products, recyclable products – defined by market value. Since the value of various outputs can be so different, cost analysis is central to the process definition in a way that it isn’t in engineering, and cost allocation must be more rigorous than in engineering. This introduces different complexities, particularly where co-products or by-products are kept for reuse either later in the same process or in another process. Then again, input materials vary so the process too has to vary. Unlike in discrete manufacturing, you can’t give each product variant its own part number. The same recipe or formulation is capable of so many variants – both currently and over time (beer, cheese) – that the list would quickly become unmanageable. Nor can you assign a different process description to every process variant. The product code has to identify the generic product and be able to add qualifiers for the variations. This allows producers to identify: how much of this spec is in stock; the master formulation; how a particular lot was made; what it should have cost; what it actually cost; and what it was sold for. “One distinct advantage of cost analysis designed for process manufacturing is that it is not limited to just analysing history,” says Wonderware. “The same rules for distributing actual overhead over actual production can be used to distribute budgeted overhead over the projected volume and mix of planned production. With simulation capabilities, the system can support all types of cost modeling for decision support.” Could this be something engineering companies could benefit from too? Cost is far from the only driver, however, and not every process company is a huge conglomerate. They all have to keep costs down, service levels up and quality spot on. But some, like Kettle Foods of Norwich, find they manage well without huge IT investments until they get to a certain size. Kettle UK employs more than 270 people at Bowthorpe, near Norwich, to make a range of natural-food snacks. Kettle’s best-known product is Kettle Chips which, with sales of about £30 million, Kettle says is one of the fastest-growing snack brands in the UK. Kettle’s finance director, Lyn Daplyn, thinks now that Kettle was slow to take on IT: “I think a lot of small companies tend to grow a bit ahead of themselves before they actually take breath and say, ‘We’ve got to go back and put some systems infrastructure in place’. Bigger, established companies will already have that because they’ve been operating at that level for so many years.” Kettle was running a fast-changing business with a sales order processing system and financials plus any number of spreadsheets. “These systems were not integrated,” says Kettle’s system supplier, IFS, “and the lack of central control meant that overall there was inflexibility and duplication of effort [which] led to errors and a waste of resources.” “We deal with all the major multiples,” says Daplyn. “And we have to keep ahead of the competition. The number one reason for this whole exercise was customer service. It wasn’t driven by financial reporting or any other reason. It was all about maintaining and improving upon our customer service levels.” She won’t say what customer service levels were, except to comment that they were already “very good”. But Kettle was concerned, she says, that its high growth and the “increasing complexity of the business could threaten those good service figures.” Nevertheless, at the other end, though the main Kettle ingredient is potatoes – it processes several hundred tonnes of them every week to make crisps – the products also contain ingredients from several other raw material suppliers. Says Daplyn, who led the company’s IT project: “In terms of our own stock management it would be useful to have external interfaces with our suppliers as well as with the customers.” Kettle’s criteria were that a new system had to: be integrated, user friendly, provide immediate information, match Kettle’s growth, provide the flexibility the business needed, and be capable of e-commerce transactions with customers and suppliers. IFS, the winner in competition with SAP and JD Edwards, came out on top in terms of usability, says Daplyn. She and her team thought IFS “was more cost effective in terms of implementation and the cost of maintaining and developing the system.” Flexibility was also important. The IFS offering can be built up in stages and altered later: “If you do want to change something it’s not such a massive job,” Daplyn comments. Implemented in two stages, the first used IFS for financials, order processing, inventory management and purchasing. In the second stage, now under way, IFS will add demand planning, MRP and HR. Implementation for phase one began in June 2000 and went live on 1 October 2000. IFS isn’t the first name you think of in process-industry ERP: top of the list would surely be Ross Systems, then SSI (Chelford) and what was Marcam. Then there’s SAP and Oracle at the higher end, and inevitably Fourth Shift (now AremisSoft) because of its sheer range, although Geac, JDE and Sanderson are among others with notable systems in the sector. Actually though, says IFS’ UK managing director Alastair Sorbie, IFS’ process and food users include Kazoo Nobel, Becker Acroma and Masonite. He emphasises: “Although we have a strong engineering product it would be completely wrong to suggest we were selling that functionality to process users. Our technology allows us to move into specialist markets quickly. “We can offer typical process features such as recipes, batch balancing and the like to process users together with the rest of the product offering, [and we can] combine process features with the appropriate costing features, financials, preventative maintenance, equipment monitoring, quality management, document management and so on.” IFS says Kettle now makes information about stock, creditors, debtors and financial position available throughout the company which is “consistent across all functions”. The same core data is used whether an enquiry comes from customer service or finance.” Better decision making Daplyn says these management reports used to take weeks to produce, “but now it’s available on day 10 of the month. Some reporting is [now] available day one in terms of general sales reporting information,” she notes, “but the actual full financials don’t come through until day five or day six.” The information comes from the factory floor and across all the departments in the business. “There’s still one or two manual processes that take a little while to finish off at the end of a month,” Daplyn says, “and we’ll be looking to focus on reducing that in the future. But there is almost an immediate impact.” Daplyn emphasises that most of the major benefits will follow stage two. But she already sees some: “Visibility of raw materials and finished goods stock has obviously made a big difference in terms of purchasing and also just servicing customers. “And purchase-order processing has removed an awful lot of paper from the business now that that’s all automated. This makes the whole process of purchasing much more simple.” Kettle was already fairly good at tracking costs. Says Daplyn, but: “We certainly will understand more what the cost drivers are in the business.” Her advice to any process company? “I would say you certainly won’t regret it, so go for it. But do not underestimate how much time is involved from your key staff. And I think that’s the other thing, to make sure you do use your key staff for the project. It’s the commitment that has to be made by the senior managers in the business, predominantly, in terms of the original setup – mapping out processes, training – that’s where the really heavyweight involvement of time is.”