You only know what you know

By now, you'd think, advanced planning and scheduling (APS) systems would be familiar factory furniture. But APS is anything but a mature market. The market leader, Preactor, is reporting record sales amid a truly awful recession, and surveys seem to show that APS tops many companies' wish lists. It's not difficult to work out why. Mike Evans is IT administrator at TRP Sealing Systems of Hereford, which makes gaskets for plate heat exchangers: "Our customers are increasingly demanding shorter lead times and quicker, more accurate delivery dates." Seriously-flawed alternatives to APS persist. Until TRP adopted Preactor, it used a Lakeview ERP suite, supplemented by handwritten schedules and Excel spreadsheets, to update an entire wall of T-cards once a week. Planning manager Alan Lewis would then write the next week's schedule, machine by machine. Urgent orders, machine failure or tool damage meant starting all over again. Timings were approximate and some jobs took longer than planned. On-time deliveries were in the low 60-per- cents. In the end, a main customer sent in a consultant. Using Preactor, the schedule can now be updated in seconds, and the delivery record is at or near 100% for a growing number of customers. Silvergate, a Wrexham MBO which supplies accurate-colour plastics for mouldings, extrusions, sheets and films, used to shuffle coloured plaques around a large rubber mat. The resulting rough working schedule was typed into a spreadsheet to be photocopied and handed round. As Brian Tinham explained here 18 months ago, it's clear that APS is capable of humungous sophistication in very complex production environments. One example is the Syspro ERP and APS combination that K3 has installed at make-to-order precision engineering company Axiom Manufacturing Services in Newbridge, South Wales. Axiom's system adds specific machine-capacity data to bills of materials, routing information and standard times to turn sales orders from its medical, industrial and military customers into capacity and material requirements. The jobs show up on a real-time Gantt chart available to touchscreens across the shopfloor. World railway-electrification equipment supplier Brecknell Willis (BW) in Chard, Somerset, uses Seiki Systems software to schedule 1,800 jobs at a time. Graphical views of each machine show the operator a queue of the work due on the machine. Production and material controller Steve Crimp can manipulate this queue from his office, so rush orders can be dropped in and projects on longer delivery deadlines can be pushed back. Before APS, "we were constantly fire-fighting and taking jobs off machines before the batch was finished. [Now] we let the jobs run because the software gives us the ability to look ahead and forecast our production performance. It's very visual and predictive, which allows us to act very quickly." APS thrives on complexity. TRP has to juggle eight extruders, 24 presses and over 700 tools in bewildering permutations. So the technology, unlike some others, seems to work. But not by itself. There are traps for the unwary before, during and after implementation. The first trap, now beginning to disappear, is to think you can be competitive with manual systems. Most companies that abandon them do so for one of two reasons. A planner who carries everything round in his head is about to retire. "All of a sudden," says Preactor International president and CEO Mike Novels, "they realise they've got a problem, that they need to capture his expertise, his knowledge, in something that would do something close to what he did automatically and then they can do some smaller changes manually." The second is that two companies merge and the size of the planning and scheduling task overwhelms whoever's doing it at the moment. Novels's number one competitor is the spreadsheet: "It's free, and it does do a job, up to a certain point," he says. But keying errors are a constant danger, and the spreadsheets may include a large number of macros that no-one understands once their originator leaves the company. Another trap is to use technology to automate old ways. Successful implementers simplify their operations before buying anything. In 2005, Silvergate identified up to 60 process steps taking eight hours from receipt of an order through to order delivery. Only after identifying and eliminating non-essential steps did Silvergate business manager Tony Bestall identify Preactor as a possibility. In-full and on-time service levels were 79% with a lead time of between five and seven days. Now there's no lead time and 96-99% in full and on time. Aerial and communications equipment maker Blake UK adopted just in time (JIT) manufacturing before it adopted Preactor. Blake relied on a combination of its Sage CS/3 business management system and a collection of custom Excel spreadsheets. As the business grew, it became increasingly difficult to to allocate demand across cost centres: "It was getting very difficult for a human to do it," says managing director Paul Blake. Preactor can process a lot more information than a human. Telford, Staffs, based Foilex, which laminates window, door, cladding and roofline products, standardised products into a family grouping code to enable products of like families to be optimally scheduled. Novels has another take on complexity. At first Foilex used Preactor to schedule operations that didn't affect the schedule. It was easier and quicker just to put a short delay in there and schedule only the critical items. There may be 25 operations, he says but only 10 of them need a schedule. Novels stresses the need to understand the differences between planning and scheduling. One is timescale. Planning systems tell you the total time-capacity available. They don't tell you how you have to divide it up. Scheduling is the minute by minute allocation of work to resources – workcentres and people – while retaining the sequence of jobs within that time bucket. The key question for anyone considering buying one of these systems, says Novels, is whether the vendor is offering a system that delivers a 'work-to' list for today. Steven Littlewood of Mirfield, West Yorkshire, based SL-ECT, is a former planner, now a Sage and Preactor integrator, who worked with Blake to install Preactor: "People often don't know what the software does," he says. "Even production directors are often embarrassed to talk about it." The chief benefit of APS is that it tells the planner immediately of any problems, including wrong data. Blake now has complete visibility of not just its daily production plan but also its next seven days' projected workload. Blake now spends half an hour a week on planning instead of two days. Next-day deliveries are up at 85%. Many companies tell Novels that the first thing they discovered on installing Preactor was how poor their data was. One customer found a three-day process that should have taken three hours had not only been throwing all the plans out of kilter because of a decimal point error in the ERP system, he says, it had been costed that way and the capacity also miscalculated accordingly. In a dynamic Gantt chart system like Preactor, that becomes immediately visible as an over-long bar. It's easy to type wrong data in, says Novels, but in a graphical system it stands out. BW's Crimp says that, before installing the Seiki software, everything was done manually on Excel spreadsheets with paper 'work-to' lists. Now individual operators can see an electronic work queue for their machine. Crimp isn't specific, but talks of "benefits all the way round on our performances" from using the scheduler. "It's so much easier to monitor the work and the workflow." There is no such thing as vanilla APS. You have to pick the one that suits your kind of operation. Novels's advice to users, after making sure that they know which system they need, is to ask to see it running with their own data. "If sequencing is a key issue for them, then they need to make sure that the planning system they are looking at does give you despatch ['work-to'] lists for each resource, for each time of the day." Littlewood says the next danger is that, once that exercise opens users' eyes, they start wanting too many bells and whistles: "They want to go to the nth degree. [But] it's very important to collect all those ideas then to rein it in and make a compromise that gives you a simplified phase one implementation. That is absolutely the key to successÉ Then you can go back to the wish list and make a judgement that's really pragmatic and practical and will give you some benefits." Most manufacturers, alas, have a budget for part one, but not for part two. Another Novels tip, based on years of experience, is to get the planner involved very early: "Not just IT, not just senior management, but the actual planner who's going to use it. We've had situationsÉ where we had a lot of problems in producing a schedule that met what they required until we started to talk to the planner. He pointed out that the way they actually ran the plant was totally different to the way the IT people thought. You need to get the people at the coal face involved as part of the project team." Some planners see APS as a threat to their jobs when in fact it will make the job easier. But there's sometimes a fear of being found out. Planners may not be that well trained, and what they do often isn't that visible to others. They often enter the role from some other job, perhaps as shopfloor supervisors, says Littlewood: "They can get found out when you systemise it." Littlewood says it's not that they can't do the job properly. It's that, no matter how competent they are, it's not easy to express the rules they use clearly enough to transfer it to a machine in a form the computer can use to deliver answers. If they don't have that mentality, they won't enjoy using the software and they won't use it to explore what-ifs and improve the operation: "They'll try to repeat what they are already doing on the spreadsheet." Novels says that some resistance is inspired by the reduction in the number of planners Preactor makes possible. But it also happens that someone who is very powerful in the company suddenly becomes less powerful. That has to be managed. Teething problems are as likely to be about the ERP system as the APS itself. Data integration, the passing of information between systems, isn't an issue, says Littlewood. But to model someone's factory, the integrator needs a flexible interface with the ERP system and some ERP systems have fixed or 'hard-coded' interfaces. Blake is still struggling with this and is using Littlewood's expertise. There's a health warning about all software, says James Burwood of Goldratt UK. Goldratt has provided software to support the customers who adopt its theory of contraints (TOC) approach. But Burwood says founder Eli Goldratt now downplays the role of software: "People started thinking of the software as the answer," neglecting the other things they needed to do to eliminate waste. So they'd get immediate improvements, but once the implementation consultants had gone and people started reverting to old habits, the improvements evaporated. You have to know what you want the software to achieve, says Burwood, and you then have to ask whether the software achieves it. "Or are you just going to get a computer to help you make the same mistakes more efficiently." He cites a manufacturer of sealed units for double glazing which used software to cut glass more efficiently. But the constraint wasn't the glass cutting process and the amount of glass waste cost wasn't that significant. The problems lay elsewhere. In Littlewood's view, "if you can get that two-phase approach, the second round of improvements, then you'll really get a happy customer that's going to be reference-able and [the system] will achieve whatever they want it to achieve".