APS is a sure bet but back the right horse

Vendors of advanced planning and scheduling (APS) systems haven't made life easy for themselves. They've stuck the same label on virtually everything from electronic planning boards to systems using clever algorithms to optimise entire supply chains. Confusion results about exactly what's on offer and what it can do for them. It also means a few manufacturers are feeding a racehorse when they could get away with a Shetland pony. Far more, however, are failing even to make it to the starting gate. Which is a great pity because of all the post-MRP technologies, APS is the one that really can make a difference to a manufacturer's performance. Today, APS systems are mature, able and proven, prioritising and rationalising shopfloor processes and taking huge chunks of time and stress out of the race from first enquiry to last lorry load leaving the factory gates. Let's take a look at the basics. Leave aside top-level optimisation systems like i2 and Manugistics: these do an excellent job but are largely the province of Tier One companies that need to drive demand optimally through their supply chains in the most tightly synchronised fashion. Most manufacturers' priorities lie closer to home. When can I promise delivery to win this order, and how do I then get it through my factory floor? How can I load my factory to maximise on-time delivery all round, minimise WIP (work in progress), work to a logical and economical sequence and get by without either incurring overtime costs or killing someone? Day to day realities Enter APS to compensate for MRP's limitations in making longer-term production plans cope with the day-to-day realities of the shopfloor. APS at this level models the capacities and constraints of the shopfloor. It then provides a realistic picture of what can be produced that day/week/month, simultaneously taking into account all available resources including labour, materials, subcontracted assemblies and machine capacity. The better ones make that picture available at order-taking so that promises can be made and achieved ('capability to promise' or CTP). APS allows planners to rejig schedules on the fly to cope with, say, rush orders, delivery shortfalls and machine breakdown. 'What-ifs' allow options to be assessed, including changing sequences, working longer shifts, expediting suppliers or contacting customers to negotiate dates. So far, so good – but there is a wide difference in the way systems provide these capabilities. At one extreme there are the scheduling specialists: most well known among these is Preactor, which sells directly and also to many ERP companies for tight integration into their own software. Sage's Line 500 is among the most recent ERP systems to take this route. Alternatively, some ERP vendors offer their own APS functions as an integral module of their core systems. Two pioneers in this area, Lilly Software and Infor (now under one roof following the acquisition of the former by the latter) built their ERP systems around APS specifically for make-to-order companies. Shortly, Infor will be joined by Mapics, which now owns the Syteline 7 ERP system, itself with embedded APS capabilities from specialist Frontstep. Confusing for potential customers but there's obviously no shortage of scheduling experience under the Infor roof. What does all this mean for potential users? Does it really make a difference whether APS is integrated or part of core software? Graham Hackwell, technical director of Preactor, takes a pragmatic approach. "People looking for all-in-one functionality must accept scrapping everything in their existing systems, including those bits fitting their businesses better in other areas. People invest a lot in getting their ERP up and running. We want to add the extra capability." Watch the quality But that implies integration, and the quality of the integration depends on the work that's gone into it. ERP vendors selling own-badged specialist systems have usually made a good job – it's a lot easier to link to just one system than it is to many. On the other hand, those starting from the ground up would be missing a trick if they didn't take a look at the 'built-in' APS/ERP systems. Anyone treating APS as a subsidiary requirement to MRP functionality at that stage is limiting their future agility and competitiveness. Hackwell makes an important point, however: "To do APS properly, you need memory-resident data. So, if they run off disk-based databases, even all-in-ones are going to be slow and will not have the proper iterative capability." Resident memory creates its own issues, too, since you may end up with two copies of the data: one on the ERP database and another used to create the schedules and simulations. Some APS shopping lists feature 'real-time' as a prerequisite. It is a suggestion that drives Hackwell's colleague Nigel Shires into apoplexy. "What do they mean by real-time? If you are in a plane flying by wire, you talk in milliseconds but in even the most reactive factory – say fresh food – real time is probably more than 10 minutes. In many it could be a day or more because things just don't change that fast. Most classic discreet batch manufacturers don't need something updated every time a product comes off the line. Once or twice a day will probably give them a significant improvement over what they are doing now." Hackwell believes people should work with their existing feedback mechanisms where they can: "Instead of duplicating everything, if they are already feeding back into their ERP when they complete batches and ends of shift, it will still provide all the updates the scheduler needs." Mapics takes a different approach, according to sales director Adrian McNay. Syteline 7 makes its APS-generated production plan available to everyone – sales and purchasing as well as the shopfloor. Sales can get an instant CTP; and sales and even customers can place demand on production, in the latter's case via the web. Mapics is not alone in this but it does handle the synchronisation issue somewhat differently. Take, for example, its customer PBT International. Part of its business is refurbishing print cartridges: as these come from charities, PBTI rarely has a firm picture of its supply. It therefore revisits its plan regularly to see if there is enough to meet new demand. The plan is cached at the end of the day for the next day's trading. As orders and supplies come in, they are taken into a layer sitting notionally above, and both layers are automatically taken into account before promises are made. 'What-ifs', or renegotiations, can happen before the plan is consolidated at the end of the day. In this case, the system can solve a problem specific to an individual business. This is probably the best rule-of-thumb in choosing any APS: make sure it takes away the pain that hurts most. The solution can be cheap, simple yet highly effective. Brunton Shaw, for example, uses Preactor to plan effective production of over 40,000 different wire ropes, each made to a precise formula from a vast potential combination of widths, materials and finishes. Managing multiple sites Many demand a specific manufacturing sequence and will only fit on certain machines. Handled manually, scheduling was a nightmare. Preactor here is not linked to MRP/ERP; it works only with an in-house database containing specific product codes and construction details. But it does what's needed. Hackwell says that this situation is unusual; most integrate to ERP because they already have it. Those that go the stand-alone route either do it because they are small and only need an accounts package and a scheduler, or they intend to buy ERP in the future and are holding the fort today. But let's look at another business requirement: Strix, whose HQ is on the Isle of Man, makes thermostat controls for kettles, and uses Mapics to manage a common problem – efficiently splitting production between two UK and one Chinese plant. APS gives Strix global supply chain visibility of both BoM (bill of materials) and resources. It allows it to balance product complexity against order volume, and cheaper production against longer delivery times to make a sensible loading of capacity to meet customer lead times profitably. On a larger scale, Eliane Revestimentos Ceramicos of Brazil, producer of ceramic coatings, uses Preactor's APS to manage production of over 1,800 products from 12 Brazilian plants some 3,000 km apart. Some coatings can be made in two or more plants; others need to be processed in more than one. Previously, scheduling was decentralised, as was control and purchase of raw materials. Customers would typically order from more than one plant and it was almost impossible to achieve common delivery dates. Eliane decided to centralise scheduling across the whole supply chain with a single customer contact point. APS was integrated with Eliane's ERP system and a Preactor viewer installed on each site. ERP was modified to deal with multi-line sales orders and routings across many plants. Now, when demand cannot be fulfilled by real or projected stock, a new production order is generated and sent to Preactor. It creates several alternative production plans based on all plants' current status, minimised set-ups and delivery horizons. The chosen schedule is sent as a text file to the viewers, and each plant passes suggested changes back to the MPS. Finalised plans go to ERP to synchronise purchasing with individual plant schedules, and actual completion times are then entered via the viewers back to the master system. The result has been smaller production batches, reduced lead times, lower inventory and better utilisation of capacity across the company – all characteristics of a good lean organisation. Who says IT plays no part in lean? Ultimately, the measure of a scheduling system is how well it deals with real issues in your business. For example, Hawes Signs, which makes signage for major refits of petrol stations, car showrooms and retail stores, has completely turned round its customer service, partly due to its use of an InforCOM ERP system. That not only has built-in APS but a module that provides functionality for managing the costs and conflicts over resources afflicting most companies handling large-scale projects. A year ago Hawes was receiving black marks for missing deadlines. This year it received Tesco's highest 'Blue Status' accolade and is top of the supplier league table for facilities and buildings. According to Steve Walters, Hawes' business systems manager, the system has helped to bring order and greater continuity of customer service to the operation. "The system sets the plan, and everyone knows what they have to do to meet it." Significantly, improved knowledge of stock movements and project allocations has also let it rationalise its finished goods stock from £2.3m to £1.3m. The moral is indeed 'horses for courses.' Any APS depends on creating a computer model of your process and exercising it to create your schedule. But as Hackwell says, all models are approximations by definition. The question is, is it good enough to give you a useful, realistic schedule? If the APS salesman begins to shuffle his feet when you describe a process vital to you but arcane to him, show him the door. But don't then assume no-one else can help. There is an APS out there with your name on it, unless you want to be one of the competitors overtaken early in the race.