Think global act local

Engineering lead times have fallen from 30—40 weeks to 10; projects that consumed 20—30,000 engineering man hours now take 8—10,000, and are heading for 3,000; and inventory turns are up from 2 to 3.5, which in sub-sea oil and gas extraction systems engineering is excellent. This is the picture of achievement at $70 million turnover FMC Energy Systems (formerly FMC Kongsberg), based in Dunfermline, following four years of development with advanced planning and scheduling (APS) systems, Theory of Constraints (TOC) processes, Critical Chain multi-project management tools and inspirational internal and supply chain synchronisation projects. The company is FMC Technologies’ well head division, making surface and sub-sea ‘Christmas trees’ for the offshore oil and gas exploration industry. It’s all big ticket, serious engineering and project management stuff. The ‘Christmas trees’ are the sets of valve gear and manifolds that sit on top of well heads and control the flow of hydrocarbon out and injection fluids in – engineered to order, valued at $750,000 to $1,500,000 each, and with build periods of six to 12 months. Andrew Cort, operations manager, says: “We’re in the reliability business. It’s all highly engineered and highly reliable. The systems are designed for 10—25 years of operating life at depths down to 4,500ft, and the environment is extremely harsh down there – 10 to 12,000 barrels of crude a day at 5—10,000psi. Also, they’re very expensive to install and to work on… You don’t want one of these to break loose.” FMC has a large engineering office, and manufacturing facility that operates out of an aircraft hangar size 46,000sq ft factory with a suite of machine tools – seven horizontal boring machines, two large flat bed lathes, numerous CNC machines and specialist welding – as well as assembly and test, including a 30ft water test pit with two 40tonne cranes for gas integrity testing. “We do the highly critical value add production here. The rest is done through our supply chain,” says Cort, which he adds is very highly managed: traceability goes all the way back to original materials procurement. Finite capacity scheduling Handling all this efficiently with such long project cycles hasn’t been trivial. In terms of IT, management was originally on Mapics MRP, which switched to CA (now SSA GT since the change of ownership) ManMan MRPII in 1993. Cort admits that it wasn’t ideal: “It involved a lot of development to make it suit our business, with a lot of bespoke code.” One of the real issues, he says, “was only having an infinite scheduler… We operate in a highly cyclic industry – it can go from very quiet to extremely busy and everything in between. So we needed predictability and control to avoid wasted resource. We needed a constrained finite capacity scheduling system.” In 1998 FMC bought the then ThruPut Resonance TOC-based ‘Drum Buffer Rope’ APS (now owned by Mapics and substantially redeveloped), following its successful implementation at a sister plant in Houston, to support redesigned constraint-based manufacturing processes. “A number of our people had experience of Goldratt [TOC],” says Cort, so – unlike for the vast majority of doubters – it wasn’t a difficult decision. At Dunfermline, Resonance was overlaid on ManMan. Its MRP ASCII files – standard BoM and routing information – were linked to Resonance for schedule generation, and it in turn fed back to ManMan for material requirement planning which was loaded into the company systems in the usual way. Go live was April 1998 and the results were as impressive as had been promised. Inventory was cut by about $1.4 million and subcontract costs fell by about $1.5 million in that year – some 10%. Beyond the headlines, FMC had also achieved controlled release of raw materials and work in progress (WIP) so that WIP and lead times were reduced – these alone more than paying for the system. But after a while improvements slowed. A valve block, for example, that had been on a 21—22 week build cycle was down to 16 weeks, but it wasn’t enough. FMC was realising that it was as much an engineering project company as a production company, and it was that aspect that now needed attention. The firm purchased the Concerto multi-project ‘Critical Chain’ scheduling software tool, also from the Goldratt stable, to help. But it still wasn’t making headway. Cort says that the learning curve had become too steep and that, although he had several TOC specialists in house, “we struggled for 15 months. We needed help.” So in December 1999, Goldratt Institute associate John Tripp was hired to help the firm re-invigorate its TOC DBR programme and integrate it with its Concerto project engineering front end. One of the problems was that Resonance simply wasn’t being used consistently. Operators were launching work in parallel to the DBR scheduler which it didn’t know about. So it was trying to use capacity to do what it knew about while they were pushing through additional orders around it. People and processes Cort says MRP was what they ‘understood’, and in the users’ terms, even though it might deliver late, it was ‘successful’, so they were not keen to let it go. “It was all they had to rely on.” Additionally, there had been some issues with Resonance’s schedules due essentially to the MRP data not being good enough – a common enough story. “We were having to show them that it wasn’t robust enough, but people were looking at Resonance and didn’t understand its results. So it was a culture and knowledge problem.” It’s a very easy trap to fall into, and, as a result, Tripp’s first job was all about cleaning up the data. His second was to provide fundamental TOC education at the grass roots level to help the operators understand how they should be using the systems. As a result, FMC implemented structural changes. “We set up a master scheduling group of five people – updating and using Resonance. So now we had a dedicated team. We modified the schedule and the manufacturing operations. Then we brought in Concerto – tying the company back into Critical Chain, multi-project management scheduling.” This was fundamental. Tripp says: “FMC is a major production company but, as with many big engineering project companies, its process times are very close to its lead times. With pure DBR you need process time to be 1—2% of lead time. But here, process time is 60—80% of lead time, so there’s very little protection if there are welding problems, for example. There is very little time to recover and it’s at the extreme end of TOC applicability. So getting control of this with Concerto, and changing the buffers accordingly was key.” Cort agrees: “It unified deadlines for the whole organisation,” he says. How? “Concerto is about monitoring what’s left to do. It tells you ‘if you don’t take action now this is what will happen; change the resource or you will miss your dates’. It’s especially good for high value, heavy engineering projects. It doesn’t allow us to hide.” By April 2000 both Concerto and Resonance were working properly, engineering projects were being scheduled and integrated with the production schedules, and due date performance was much better, up about 10—15% at 85%. Then came the next revolution. Says Cort, “In August 2000 we started to develop a new business strategy based on linking our three production plants in Houston, Brazil and Dunfermline. We wanted to make them look and work like one plant from anywhere in the world, and at the same time maintain quality and improve lead times further.” Under John Tripp’s guidance, the firm went through a Goldratt ‘four by four’ global analysis to make this happen – discovering things that would have to change and finding out how best to do it. Cort: “We used the Goldratt processes – generating requirements lists, looking at consequences, countering obstacles and agreeing estimates of what we could achieve. We’re monitoring against that now, and we’re about 40% of the way through the plan.” Singing from one hymn sheet The result has been a high degree of integration across all three sites, covering design, construction and procurement. Considerations have included defining components with the same part number and the same spec, enabling project and product transfers and mapping machine tool suites across all three sites. Beyond these, FMC has been looking at subassembly standardisation to enable mass customisation through late configuration processes, and it’s been moving to improve sharing of manufacturing resource, including its supply chains in the three geographies, to further reduce lead times and costs. Cort says the next goal is forecasting together and investing in FMC’s future by working with its supply chains. “Standard product will be held by our supply chain companies, not by us, allowing us short term call-offs. We’ll bring forward investing and manufacturing to a particular point, harnessing the supply chain, and put more of our business through them for premium price, short lead time, high value projects.” It’s working. By engaging the supply chains in this way lead times can already be slashed from 30-40 weeks to around 10. And significant savings have been made in inventory and subcontract expenditure. Inventory turns have improved from 2 to 3.5 – which may not sound brilliant, but in this industry is world class. Last year, inventory was $6 million against an average of $19 million, although this is something of a moving target in a cyclical industry. Now the firm says it’s focusing on using its TOC experience to help supply chain participants to improve their efficiencies and delivery performance. And it’s looking at web-based SCEM (supply chain event management) using portal technology, so that it can post requirements and changes in real time online, put procurement online and so on. It’s a long journey, but rewarding. Cort’s advice: “It’s not easy. You need to set courageous goals and don’t ever give up. Small changes are not enough: you need step changes. We’re now thinking about putting in true manufacturing production cells to reduce cycle times – we reckon we could squeeze another five-fold reduction.”