Take it to the limit one more time

5 mins read

Cosworth Racing is driving its finite capacity scheduler further and faster than most would contemplate – with surprising extra business benefits. Brian Tinham investigates

Forget guesstimates and fire fighting: consider the value of definitely deliverable production plans that you can quickly change and optimise no matter what's thrown at you, using 'what-ifs' to help get decisions right. Cosworth Racing did, and so impressed was the company with the power of its advanced planning and scheduling (APS) systems that it's spread them across all manufacturing cells, and made them the heart of its very impressive and fast engineer- and build-to-order business. And it hasn't stopped there. Cosworth has used the same tools to improve internal engineering and production collaboration and the manufacturing processes themselves, as well as developing means to provide instant and meaningful production and inventory measures in pounds – even improved warehouse management and customer service. This is an eye opener. Cosworth Racing is a legend in its own time. It's a world leader in design, development and manufacture of high performance engines for racing, rally and road cars (who remembers the Ford Cosworths?): everything from Formula One to CART, World Rally Championship and motorbike racing. Now owned by Ford, it's also transferring race engine technology to Ford's Premier Automotive Group. So it's a dynamic and innovative environment. Something of a surprise, then, to find that until 2000, production management was still down to an ageing ManMan infinite capacity MRP system (dating back to 1971) helped by Excel spreadsheets and, in 1998 and up to 2000, also Microsoft Project. Cosworth's production scheduling manager Darren Dowding explains that everything started to change when in the late '90s, the firm decided to move its manufacturing into several separate core teams, each with its own machines and people – including production engineers and schedulers – specialising in particular components and assemblies. "When we did that, we went for quite aggressive lead time reduction targets. For example, we wanted to reduce machine times for crankshafts from 17 weeks to six." Hence initially Microsoft Project, which was purchased to provide better planning automation and visibility for the team schedulers. "We spent two years with Project finding out how things could be done, what could be achieved and what we really needed," explains Dowding. And he says it was a significant step forward, but still time consuming. "For example, when a job had to be moved, the links between all the processes associated with that job had to be re-established manually. "Our big problem is that we're constantly having to change the plan. Without a quick and visual system you're constantly guessing and fighting fires. But having used Project we knew finite capacity scheduling [FCS] was the way to go. We knew that would allow us to say 'This is the effect of what you've asked me to do'." Take the high road So in 2000 Cosworth went for Preactor, initially a P300 FCS system, and initially also on what Dowding describes as the most challenging section. "We purposely chose the most complex cell – the cylinder head section – to start with Preactor, because we knew if we took the pain at that end, we'd be able to get finite scheduling working in the rest relatively easily." And it was complex: by any standards, Cosworth Racing is an extreme of mixed-mode manufacturing. The firm has all the usual issues around machine specifics and operation times, but with the added complications of very short lead times, a lot of unpredictable changes and engineering intervening with new twists to production requirements. "We have to be very flexible, but we're also a business," observes Dowding. "So we need to maximise capacity, and that's all about product mix. There are times when, due to differences in operation times, for example when we're building Formula One and American Racing heads, there are choices over best fit around maintaining flows in logical sequence or overlapping." He concedes it took around eight months to set up the APS because of the sheer detail and scale of rules required to get the cell production model right. In fact, during the process, he had to upgrade to Preactor's full APS system. But the team did get it right, and early in 2001, when a new scheduler was appointed and trained solely on Preactor, the section was quickly running with all the expected benefits. Two more APSs and two P300s followed for the other cells, and the entire implementation rolled out on time and on budget. Dowding recalls one of the early extra successes: "We were having a core team meeting in our ARP [Ancillaries, Refurb and Prototype] section, which has to handle the most immediate changes. Before Preactor, the meetings would be full of 'what if' questions that were answered by gut feel or intuition. This time, the scheduler simply took the 'what if' scenario to Preactor and returned a few minutes later with a hard copy of the actual consequences. "In that meeting we saw not only the transition from 'guesstimates' to 'definites', we also saw the beginning of a very short process by which all the teams came to trust the information from Preactor. 'What if' scenarios are now so easy to work through that we exercise this function just about every day. Further and faster "For example, on Thursday there was a change required to a cylinder head, so the assembly needs to be reworked, and it's got to be done for shipping to America on Monday. That immediately impacts the plan, because the rest of our customers want their orders by the delivery dates. The beauty of Preactor is you can tell it what's now wanted and it shows you straight away what can still be done. If you can't achieve what you need, you can suggest alternatives, like calling in overtime or working a weekend, or moving some jobs around the machines, and see what works best. You think of the scenarios and it shows you the results." All good stuff, but last year Cosworth went on to link its systems within the electronics assembly section. What was again entirely spreadsheet-based was clearly inefficient, but because it's entirely assembly work, scheduling needed to be of operators – so resources and their skill sets are named accordingly. "It generates in effect a work-to list for each person." Beyond that, Cosworth also went live with the system's Dynamic Materials Control feature to auto-peg individual orders within an assembly. Says Dowding: "Previously, planners manually linked item A to item B, but with the pace we work at there were bound to be errors. Preactor now recognises the link as one work order, so we never get a lower-level component arriving after the assembly delivery date!" And there's more: Preactor is also being used to help drive business understanding and decision-making. Dowding: "On ManMan we can look at the value of WIP [work in progress], the value of outstanding purchase orders and stores value in pounds. From Preactor, we export the finite schedule to an Excel spreadsheet: it can export such detail as hours required for each of the machines, the value and rates of those machines, then determine how much money we're due to earn that week. That then transposes into the value we will receive into finished goods inventory." It's an excellent way of getting around the strangulation of standard costing systems and finger pointing over machine utilisation – particularly where lean initiatives are concerned. "Either way, you don't want to end up with piles of inventory that's not going to be sold, and the system helps put real figures on the value – in production terms – of what's due to go into the finished goods warehouse according to the plan. "It's a very precise financial tool, and provides us with a lot of visibility... The by-product is we have a production forecast tool that works in pounds." Not bad eh? Cosworth also now uses it to peg material purchasing back to when it's actually needed. Dowding agrees that's MRP's job, but points out that ManMan orders against required date, not achievable date. Now MRP is only used for pricing and rough cut planning, while Preactor triggers purchasing. There seems no stopping the firm. "There's so much extra potential. Preactor could control part of the stores, and eventually we would like to add shipping activity and forecasts as well," says Dowding. "For example, if we can see that we're not scheduling much for finished goods, we don't need to run stores overtime. We can change working patterns and use our annualised hours scheme better."