Different strokes for different folks

APS (advanced planning and scheduling) software, as the manufacturers quoted below demonstrate, can solve very difficult problems and bring about vast improvements in many business and production metrics – in particular (but by no means only) on-time delivery performance – as well as substantially reducing stress and hassle in a factory. But in spite of these benefits, APS doesn't always have the best of names. We at MCS have spoken to manufacturers who have never quite got APS working; to companies that are turning off their APS; and to others that have still not convinced even some of their senior employees to abide by its decisions, or to provide the feedback it needs to keep functioning. What's absolutely key to ensuring a successful result seems to be understanding precisely which flavour of APS is most appropriate to your situation, and then how the project should be managed and implemented. Because the cause of APS hasn't been helped by some of the software vendors, who have rushed to apply the label 'APS' to quite a range of applications, that probably ought to have different, more useful definitions. But rather than be purist about it, a more sensible approach here is to look at different APS software and understand what might be appropriate and in which situations. Fundamentally, APS is about producing plans and/or schedules (in theory, both) that are realistic and optimum. And to be realistic means taking into account real-world constraints: the availability of raw materials, plant and machinery, personnel, warehouse and distribution capacity, and so on. In reality, however, not every constraint is important in every set of circumstances. So first and foremost the need is to understand what is and isn't relevant. Once you know what constraints need to be taken into consideration, you can turn your attention to the most appropriate algorithm to produce the most useful plans and schedules in the most appropriate timescales. Dogged determination At the ASG Group, for example, a supplier of automotive accessories such as roof racks, dog guards and side steps, one of the challenges was scheduling the manufacture of the myriad parts that go into the retail package to achieve on-time delivery. As Bruce Allman, managing director, explains: "Sawing, drilling, bending, welding, coating, injection moulding and packing materials all have to be co-ordinated to come together in the assembly area at the right time. We used to have, for example, the dog guards arriving in packing, but the brackets wouldn't be ready." In theory, the problem is solvable manually. "But we have 50 or 60 different product lines going through the factory each week, needing well over a million individual parts. To plan all that through the works you would need five or six guys writing a production plan," says Allman, "We never had those five or six guys. We have good staff with a lot of product knowledge who found the time to produce a plan, but no sooner had they done it then an unexpected event would occur, such as a customer wanting to bring the due date forward for an order, and it would be wrong." ASG recently installed Syspro APS from Information Engineering, and now, says Allman, "You can reschedule the entire plan or do 'what-if' scenarios in 30 seconds." Its system optimises production based on a number of criteria, with customer due date the priority. But it also takes into account the attributes of the various shops, such as the size of the steel used, the tooling on a machine, the colour for coating and so on, to minimise setups. Now, if a machine somewhere stops, or a customer order changes, the entire production plan can be re-calculated and machine operators automatically given the most appropriate new jobs. ASG provides a good illustration of one kind of APS working hand-in-hand with MRP. While its APS takes care of what happens on the shopfloor, MRP calculates the materials required and the dates it needs to be there. Currently, Syspro APS works on the basis that all the raw materials are available: while it would be nice to take into account supplier delivery dates – and that might be something for the future – the benefits already being achieved are sufficient to prove the value of APS and to obtain staff buy-in. Capable to promise The challenge for Oliver Valves, manufacturer of valves and assemblies for a range of industries including oil and gas, was completely different – even though both companies manufacture everything to customer order. "We are really an assembly shop," explains Kevin White, IT manager. "While the lead time might be 16 weeks for a particular valve, all of that, apart from one week's assembly, is the time it takes our subcontractors to produce the parts." MRP wasn't much help. "If a supplier rescheduled delivery of a part by one week, MRP would just tell us to schedule it back in order to meet the customer due date," says White. "We were unable to see the impact of late supply or to give customers accurate delivery date." And delays in this industry can turn out to be rather expensive. On one notable occasion, Oliver Valves had to spend £50,000 on a helicopter to ship a delayed valve worth just £1,000. But the alternative – one more day's shut-down of the oil rig in question – would have cost the customer another £1m. Now, with ERP system developer Mapics' SyteLine APS in place of its MRP, Oliver Valves can give accurate capable-to-promise (CTP) dates to customers and, more importantly, give them plenty of notice if there's going to be any change. Updated supplier schedules are received electronically on a weekly basis, and any variations are entered into SyteLine to allow it to recalculate availability. Then, unlike at ASG, on its shopfloor there's little requirement for internal scheduling or constraints management. "We are a fairly low-tech assembly plant, and the only constraint we need to take into account is the test rig," explains White. What these two stories illustrate is the need for software that is right for the job – and there are a number of ways of ascertaining that. Paul Williams, sales director at Information Engineering, suggests that software vendors should be both honest and consultative. They can't simply sell a product and hope for the best. "It is impossible for any supplier to say their solution will definitely work with you," he insists. "There's also little point in the customer going through a checklist process – most software has similar functionality." Williams warns that the devil is in the detail. "A supplier can have a gut feel that it will work, but after they have done all the hard work of implementing the solution, you may still find it doesn't work for you." What he recommends is a partnership approach to a two-week realistic trial that will prove whether or not a solution is appropriate. Yes, there's a cost involved, but it's a lot less than installing a product and finding it doesn't do the job. Another problem that manufacturers will be all too well aware of is that of situations and priorities changing. So while the APS solution might be right for today, it may need to take different constraints into account tomorrow. And that, says Christophe Gasc, managing director of optimisation software developer Ilog, can be a huge issue. "The problem with scheduling and planning systems is that as soon as you take off or add a constraint, or approach a scheduling problem in a different way, you can easily go from a one minute response time to two days," he says. "That means you have to rethink the most appropriate modelling to make sure you get back to acceptable response times." Heat of the moment Flexibility and adaptability can thus be crucial – and it's all in the model/rules. At Pechiney Aviatube, for example, the key factor was not so much a constraint as the optimum batch size. This company produces aluminium extrusions, and uses furnaces for heat-treatment. These can take a load of eight tonnes, whereas the average order size is just one tonne. "We therefore needed a solution with the adaptability to express the batch capacity of a heat treatment furnace," says Ian Bell, supply chain manager. "Previously, our managers would look at the work in front of them and try to work out in their heads the best furnace load, but they didn't know what was coming the next day. So by waiting a day they might have been able to increase the furnace load," says Bell. Now, a Preactor APS solution optimises the furnace load, at least as far as it can within the constraint of reaching customer delivery dates. "If holding something up until tomorrow would make the delivery late, it will force underutilisation of the furnace." Like the ASG Group, Pechiney Aviatube's solution doesn't take into account raw materials. Unlike ASG, however, this is one extension that the company isn't interested in. "We remelt on site and cast our own aluminium alloy, so raw materials aren't an issue," says Bell. "There is no bill of materials." Preactor's solution here did require modification to suit Pechiney Aviatube's precise requirement, but the underlying flexibility meant that wasn't a problem. "They were able to show us some examples that did broadly what we were looking for, so it was just a question of adapting it to our requirements," says Bell. With this APS solution, there are two approaches a manufacturer can take to ensuring the solution is right. Resellers offer a consultancy service and turnkey solutions: alternatively, those manufacturers prepared to do some of the implementation themselves can attend free seminars and see successful solutions. That should provide them with enough information to assess suitability of the solution. Sometimes, the simplest solution is the best. David Trowell, general manager of Seiki Systems, says his company offers a range of solutions, including automatic scheduling algorithms, but starting with a graphics planning board. "This is not a scheduler, but a graphical display that shows each job for each machine," he explains. What's key is that this isn't a plan, but a real-time display. Direct links to the shopfloor, and good data capture, ensure that the display really does represent reality. APV Baker, a supplier of the machines and process systems to make biscuit and confectionary products, uses the Seiki planning board, as Bob Collins, senior production engineer, explains. "We could have quite a few jobs in contention for the same capacity. We want to focus on certain parts of contracts to enable better assembly, but to do that with the business system would be a horrendous task. It is not geared up to do that on a daily basis." So APV Baker uses the electronic planning board to display the job queue for each machine. "Production staff can then re-order jobs according to different criteria, or indeed select jobs from anywhere on the queue," says Collins. "Our range of work is quite varied, and although certain work does repeat, there is a lot of new work coming through and we can never be 100% sure of how long it will take. There's therefore limited opportunity to use automated scheduling – we can't work on that tight a rein. "For the system to schedule automatically, it would need an awful lot of data just to tell us what we already know. With the planning board we're perfectly able to manage unexpected changes: the foreman just tells the machine operator that one job is on hold, so to start on the next. He can just look down the queue and move to another job from anywhere in that queue."