Office furniture maker triumphs with TOC

How brave do you feel? Merthyr Tydfil-based Triumph Business Systems (TBS) started its career of bravery back in 1989 using Theory of Constraints (TOC) to manage production. Within weeks it saw inventory slashed from £1,800,000 to £750,000, lead times cut from four weeks to less than seven days and due date performance up from 75% to 95%. And that was just the start! The last 10 years have seen even that look no more than good. This is a fascinating success story, made all the more so by the directors’ sheer guts in breaking with tradition, throwing out their failing MRP system and relying instead on faith, imagination and ultimately unconventional common sense – all the ingredients of just about any good business and IT strategy. TBS is a £23 million company that produces steel office furniture – mainly filing cabinets, pedestals and storage, with hundreds of permutations of colours, sizes and types. It’s one of the older fashioned ‘total process’ firms, doing everything from raw steel coil receiving to slitting, blanking, forming, drilling, painting, assembly, testing, packing, shipping and distribution – and releasing about 37,000 components a day. Raw material content is relatively high (40% – the firm only buys in paint, locks and plastic fittings) and the business is both labour and capital intensive. It employs around 500 and has a substantial investment in a mix of standard and heavily automated production, with CNCs, punching machines and bending machines, as well as powder painting lines and so on. The story starts back in 1987, when TBS was turning over £12 million and needing better systems to take it into the ‘90s. The firm hired Coopers & Lybrand who recommended MRP and helped TBS with Impcon. Says Andrew Jackson, operations director at TBS: “We fought with Impcon for about two years constantly in the belief that it is possible to succeed with MRP: you’ll emerge with the business under control, you’ll have good due date performance, falling inventory and so on. “It didn’t happen. We were told we weren’t committed; the team wasn’t capable; the inventory records weren’t accurate enough… There was always some excuse.” Eventually, TBS was running with it, but Jackson says, “we still had lousy performance against due date, even more inventory, and we had built up a huge infrastructure to support it all – progress chasers, master schedulers, WIP (work in progress) chasers.” Enter Goldratt TOC By 1989, and with a turnover now of £14 million, the recession was hitting, and Jackson says management knew it had to find a way to support the business better. Enter Goldratt and TOC. Jackson, who had been impressed with his first exposure to TOC thinking in Eli Goldratt’s book ‘The Goal’, sent a team to a Goldratt course in London and decided to start again with production control using TOC logic. “We thought ‘we’re on the wrong boat’. We couldn’t dump the Impcon system – it was running a lot of our business on the commercial side. But we switched off MRP.” So what about the additional investment in IT? Initially there was none. Says Jackson, “We installed a manual DBR (drum buffer rope, the classic metaphor for TOC-based production): it was very easy and it worked.” He explains, “DBR was set up in the business using the assembly area as the drum, and launching material to arrive in time to meet the due date.” Like many factories of its type, the capacity limiter is the final assembly area after the paint line. Despite the variety of products, Jackson says it was actually quite simple. Lead times were estimated for machine shop, painting operations and the like, with some assistance from simulation IT. And the result was that due date for shipping became the driver for everything. “That was a major transformation,” says Jackson. “Works orders went out of the window. Before, we were launching to forecast and pulling through to order after the paint shop. With TOC/DBR, we would only launch onto the shopfloor the precise materials we needed to meet the assembly schedule with a slight time buffer.” In action, materials to fill a trailer full of orders – released according to manual BoM explosion – would be tagged with a reference code indicating date and time period for shipping, and this would travel everywhere with the work. “We were aggregating into batches [that] could contain anything.” When the components come together at final assembly, fitters would do their job according to work-to lists matching the line items ordered. And if a component failed QA along the line? “Replacement parts would be issued from material release with the same reference code, so get squirted fast through the factory floor system [to catch up],” says Jackson. In effect, it was manual decision making on the shopfloor within the broad rules of sequencing. “In fact we changed the bonus scheme. It used to be based on piece work, but we introduced a system so that everyone was bonused on on-time deliveries.” And the results were nothing short of outstanding. “In 1990 we had peaked at an inventory of £1.8 million, and we reduced that within three or four weeks to £750,000. We had mountains of empty pallets outside our building!” TOC and the DBR system had released a huge hidden capacity in the factory which meant there was no need to worry about finite capacity scheduling, booking work against resources, bar coding or shopfloor data collection. Says Goldratt TOC consultant John Tripp, “Effective buffer management was all that was needed.” But little in manufacturing stands still. By 1994, with the success of the factory floor, business had grown and TBS needed to up the ante with a computer-based scheduler. And then came the next surprise: the firm went for the Goldratt-preferred ‘Disaster’ finite capacity scheduler (so called because of the disaster that Eli Goldratt says is the rest of manufacturing IT) for the princely sum of £12,000. Says Tripp: “It only schedules against key constraints, but resolves temporary constraints using delivery date. It checks everything, and pushes some things slightly earlier, reporting to the operator if there are problems and adding machinery and capacity as needed to catch up.” Jackson claims it wasn’t difficult to set up. “You simply input files for orders, WIP, the BoMs and the routings, the materials and the resource file using a local modelling tool. You plug that into Disaster and it then issues the schedules.” And that gave TBS dynamic, computer-controlled buffering for the constraints. The firm had to modify the numbering system, but beyond that it stuck with its time periods, work-to lists and tickets for most of production. The result: production management became more robust and that enabled the firm to support further increases in volume without increasing resources. Turnover increased from £14 million to £23 million while prices for an average trailer load of manufactured goods (the only effective measure of work) fell 30%. From then on it’s been relatively plain sailing. With the coming of Y2k, TBS was encouraged to upgrade its Disaster implementation, and used the opportunity to make some improvements – with, in particular, the introduction of replenishment buffers of specific components at what were determined to be strategic stock locations within the plant. Says Jackson, “One of the pressures at the end of the ‘90s was from key clients wanting to reduce lead times further, even though volume was up.” And that had meant cheating the scheduler. Putting replenishment inventory around the shop floor and allowing Disaster to schedule that as well, with re-build only as it was consumed, resolved the problem – and most importantly, without incurring additional WIP. It did mean re-engineering the factory and the software, but Tripp says it wasn’t difficult to see the critical categories of product for the buffers. “Clearly with screws, bolts and so forth, straightforward replenishment is fine, but for more specific, higher value, longer lead time parts you have to be more intelligent.” And the results: lead time was reduced to such an extent that key customers could get a 72 hour delivery service without fiddling the system. Other key benefits include due date performance improved “to the extent that 100% is the norm and 99% is failure”. And, adds Jackson: “We can predict with accuracy when things are going to happen and we can see what we need to do to make them happen as things change.” That in itself will be astonishing music to many ears.