APS: Band-Aid or keyhole surgery?

Finite capacity scheduling (FCS), or more accurately advanced factory planning and scheduling (APS) systems, have achieved huge gains for those few that have implemented them. Our June 2002 feature on the subject was considerable testament to that (www.mcsolutions.co.uk, hit ‘reference’, search ‘real difference’). But the vast majority remain either sceptical or downright dismissive of FCS’ and APS’ capabilities and the systems’ ability to turn their companies around. An MCS Forum last summer for senior manufacturing users examining how best to improve on their existing ERP implementations revealed industry’s misgivings – and it has to be said, prejudice, ignorance and perhaps cynicism. A straw poll ranked APS bottom of the classic ERP improvement options like business intelligence add-ons, shop floor data collection (SFDC) and so on. Similarly, two months later at the MCS Forum on supply chain improvement technology, APS was down there in the third division alongside investing in real-time automated order promising technology – ATP (available to promise), CTP (capable to promise) and PTP (profitable to promise). So what’s the problem? History, hype and scepticism borne of users’ experience with old, oversold, much less capable technologies in the late ‘80s and early ‘90s, seems the answer. That and some software companies’ failure at the time to point out – and users’ reluctance to accept – the paramount importance of process change, and change management, to mend what, in hindsight at least, were clearly poor manufacturing practices belonging to a past era of profligate make-to-stock. Hence the all too frequently held belief that ‘finite scheduling is the last resort of a company whose processes are tortuous, ill-considered and inflexible – an IT Band-Aid applied to a festering wound’. Views critical of FCS are strongly held and variously expressed. Malcolm Bruce, for example, European operations director with radar and navigation equipment maker Litton Marine Systems: “We would never allow capacity to be a limiting factor for our business.” Steve Read, manufacturing manager with aircraft ducting systems maker Meggitt Aerospace: “Things are moving and changing so fast we’d spend as much time keeping the finite scheduler up to date as getting the job done.” And other’s say: “The last thing we need is a system that tells us we’re going to fail.” To be fair, most have found ways around their problems with a mix of pragmatic production management and common sense, dressed up as popular initiatives or not. But are they enough? Are they limiting themselves because of old, outdated perceptions? We’ve moved on guys The plain fact is that in 2003 this is a very different world to that of the ‘80s when packaged FCS was born. Best practice processes, expectations, supply chains, competition and absolutely manufacturing technologies, including IT, have all moved on – big time. Few would dream today of a business strategy founded on charging higher prices when capacity is tight. And inventory? Standard lead times? Expediting? These have long since stopped being anything like adequate answers. And hence the improvement initiatives, ranging from Ollie Wight’s sales and operations planning at one end, to lean, Six Sigma, TQM and, more controversially, applied Theory of Constraints (TOC) at the other. For most, being successful today ought to be about taking forecast, order book and real orders as they arrive, and being able to flex capacity – allocating materials, subcontracting and so on, weighing up priorities, due dates, costs and preferred sequences to minimise set-ups and change-overs. But that’s not trivial. And with the pressures towards make-to-order, more product variants, smaller batches and shorter lead times, it’s getting increasingly difficult to do so fast enough. Which is where modern schedulers come in. Because changing requirements have not passed the IT development community by unnoticed. Gone are the days of crude ‘solvers’ – slow systems running in batch mode linked to MRP and considering capacity first, then materials if you were lucky! As Doug Miles of SME ERP developer Inforswan says, those were at best point-solutions for production that meant they “worked in isolation, adjusting and optimising the load ‘imposed’ upon them by sales and expecting materials to turn up on-time from purchasing.” Which didn’t happen. Today, APS developers have an array of advanced algorithms, fuzzy logic, artificial intelligence and web technologies to call upon that can reflect reality quickly and dynamically, and find order out of chaos. As Ed Stubbs, presales consultant for ERP software vendor JD Edwards, says: “Heuristics may have been among the best solver technologies in the ‘80s, but increasing complexity on the shop floor with mass customisation, the need for higher velocity, more configure- or engineer-to-order, have all long since pushed the boundaries.” Simulation and optimisation software specialist Ilog is behind many of today’s APSs – and the good news is they’re far more capable, much faster, far cheaper and considerably easier to implement. Notwithstanding process reconsideration, the data (BoMs, routings, machine capacities, inventory policies) often come largely from whatever ERP and shop floor systems you have. With this, modern APSs build their own big, but also detailed, picture, with rules and dependencies, so they can optimise production sequences not in isolation, but considering current orders, status, preferences, consequences, everything that’s going on. For firms that want to keep it simple, there can be as few as five or 10 ‘toggles’ to manipulate to get going – and then improvement and detail. And you don’t need to worry overly about that old chestnut, data accuracy either. Of course you need to know that the BoMs and routings are there and that material consumption and capacity is adequately mapped. But as Kyle Sanford, product manager for ERP firm Made2Manage, says: “Users shouldn’t get too hung up on whether set-up times are 20 or 25 minutes – just put 25 minutes and move on. When things are working better, then you can go back and improve. It’s got to be better than nothing.” Astonishing successes Either way, the point is modern schedulers are now accessible, capable tools – and about as far a cry from Band-Aid as you could get. They work – almost too well for their own good. The MCS 2002 feature on APS painted a picture of successes so far removed from the negative perception that, paradoxically, it almost invited disbelief and denial. Ricoh Products reckoned it achieved 30% savings in work in progress (WIP) and 70% in finished goods safety stocks, while also cutting planning time by more than 95%. AWS Electronics said it was expecting “up to 20% improvement in profitability” from handling ATP/CTP order promising on-line and optimising manufacturing cell usage on the fly. HW Plastics was looking at 20% reduction in finished goods inventory and return on investment (ROI) “many times over”. Hamble Structures claimed 30% reduction in WIP and a 10—15% improvement in capacity usage with “phenomenal secondary benefits with regard to our [lean] takt time concepts.” And the latter was with a low cost Preactor system. Sounds too good to be true? They may not be achieving benchmarked ‘industry bests’. (Who can tell? Every factory has its differences and practical and economic difficulties, and we often have to make the best of ill-considered, inflexible set-ups.) But they are undeniably extremely impressive. And Hamble, with its reasonably large machining shop making structural components in a range of materials for aerospace and defence, is singularly instructive. Because it absolutely flies in the face of the doubters that insist benefits in these situations came not from the IT, but the (mostly lean) process initiatives being adopted and the management focus they engendered (see panel on TOC and APS). Or that it’s only successful because the manufacturing is relatively stable with long lead times. Not so. First, this IT can actually complement lean processes – both enabling and supporting continuous improvement as combined development, analysis and automation tools. And second, good APS applies just as much to fast changing environments as to non – arguably more so, even where change happens faster than capacity can apparently be ‘created’. As Preactor managing director Mike Novels puts it: “The greater the complexity of a product and the more uncertainty in demand, the more IT solutions have a part to play.” For high speed, you will almost certainly need automatic SFDC and electronic production management systems for your cells to close the ‘loop’, but APS specifically finds the kind of optimised ‘agility’, amid the complexity and chaos that is so many manufacturing organisations, that we need today. And that applies just as much to detailed scheduling – immediate machine loading – as to short/mid term materials and capacity management. It is precisely because in those environments we are not in control of customers’ demand patterns that APS is so helpful. Hard won results As Novels says: “This is what users tell us is the major step forward. Before, they had no real idea if they could deliver, and if not, what they could do about it. They could change the priority of a batch at one resource only to find that this had an impact on other batches later in the production process. Our APS tools support their decision making processes.” And Andy Ferrar, former production manager turned managing director of APS software vendor Profax, adds: “How else can a manufacturer schedule 800 items from 280 customers with 20 processes for each works order, resulting in 70,000 operations being scheduled in minutes – and do this at least five times per day?” For that you need a system that can look across all the jobs and all the machines and see the consequences and impacts and provide quick, dependable decision support. “Far from a Band-Aid, this is precision surgery,” he insists. He’s right. Here’s how it worked for Moss Plastics, a Midlands-based manufacturer of a huge range of plastic components for industry, which implemented the ST Point TOC-based APS (originally from STG, but now Manugistics) some 18 months ago. Planning supervisor Giles Guthrie says: “Our on time delivery in full has risen from 65% two years ago to 99% now – and that’s with less people and more complex, shorter run, shorter lead time work.” Other top line benefits include customer overdues down from 173% of daily turnover two years ago to 27% today – and in December 2002 one quarter of that. Then schedule adherence, which was 35% two years ago, now sits at 80% and savings in raw materials and finished goods inventories are around 20%. Guthrie emphasises that this was in no way an ‘out of the box’ implementation, nor an isolated initiative, but part of a strategic plan to improve production, cut costs and ramp up throughput. He admits to spending months “working with the engineers and operators trying to see how to make it happen,” but without initial success. Eventually, he changed direction and developed a new model of the production floor by rethinking, looking for commonality and simplifying. “You have to get away from what ‘experienced’ people tell you,” he warns. “They’ll see machines as all very different, when you can think of some, from a scheduling point of view, as the same. I redesigned the work centres so that 99% of the time, within each, work can go onto any one of several machines.” That did force some tooling management – manifolds and the like, for example, had to be tied to work centres and, to maintain the model, only moved with authorisation. So changed routes and processes absolutely played their part – and his system helped. With rules developed to minimise set-ups and set-up hours by pooling tools, materials and colours, and driving for customer due date, while focusing on constraints, like available machine setters, the transformation has been achieved. “Now the system can schedule 400 jobs on 150 machines in an hour. That used to take two days. And I can look ahead three or four weeks to see problems.” For Moss, it percolates everywhere. APS has been linked to its SSA GT BPCS ERP system, extracting consolidated customer orders in terms of BoMs, routings, material details, inventory, etc, and returning the plan to drive shop orders, materials purchasing and so on via MRP, as well as the machine controls. But beyond that: “Sales can see the schedule so we’ve got much better agility. We can create daily setting schedules and respond far more quickly and efficiently to opportunities without disrupting the plant, or pushing out other orders,” says Guthrie. And he adds that lead times are also substantially down, not least as a result of de-cluttering the factory which is also now better able to handle the variability of production increasingly required. He’s not saying there isn’t room for further improvement: schedule adherence could be better, although few would argue with the realities of ‘Murphy’ on the factory floor. With its link into the BPCS’ HR module for the critical skills matrix, next up might be running that more frequently to reflect absence daily. It’s a question of balancing diminishing returns. It’s worth noting here that FCS and APS systems are primarily designed to achieve Moss’ top line result – better and more efficient on time delivery in full performance and maximised plant utilisation come what may – with the more advanced systems extending that to supply chain order promising and optimisation. APSs weren’t conceived primarily as tools for reducing WIP, materials, or finished goods inventories, or lead times, nor directly for process improvement: those are all secondary, but very welcome, additional ROI factors. Get what you pay for In a sense, scheduling technology today provides for a continuum of capabilities and you pretty much get what you pay for, although £10,000 gets a lot, and £50,000 is serious stuff for complexity typical of a Tier One automotive supplier. And what’s becoming progressively more attractive – at least to manufacturers in the more demanding supply chains – is the internal and supply chain order promising components. Stubbs reports that around 30% of JDE’s ITTs in the upper mid market now specify ATP/CTP, even if it’s not initially implemented. The prospect of a system able to take late orders, look at inventory and planned production in real time and suggest, cost and implement re-optimised schedules on the fly seems compelling. The bottom line: scheduling itself is a non value-add process – but proactive, optimised production can save cost and add value in spades. If you want the kinds of benefits already being achieved by your peers, then forget trying to compartmentalise APS’ applicability by industry, manufacturing style, throughput, complexity, product rates, lead versus order times, whatever. Success comes from very diverse industries and companies, and steadfastly resists generalisation. And try to put behind you some of the dogma of others – and even your own. The game is no longer about new processes or IT: it’s both absolutely working together. With a dash of objectivity, it’s hard to escape the conclusion that documented cost savings way beyond simple scheduling – and providing for the coming requirements of ‘collaborative commerce’ – are there to be had, and worth the effort and cost. For your company’s sake, press for a pilot and consider it a key element in continuous improvement that not only supports initiatives on the shop floor, but spreads the thinking and the capability into sales and purchasing straight away, and then on into the supply chain.