Simulation solves bottlenecks at Nissan Sunderland

Simulation software was behind Nissan Sunderland’s huge achievement last year of shoe-horning production of three models into its two Primera and Micra lines – with live planning and scheduling subsequently managed so successfully by Ilog APS (advanced planning and scheduling) software (MCS, April 2001, page 20). You’ll recall that when the Almera hatchback was launched in 2000, to minimise capital expenditure Nissan decided to integrate its production into the two-model plant rather than add a third line. The premise was that the Primera and new Almera would move down separate production lines, while the Micra would share the Almera line but be switchable between both. It was plant modelling that helped establish the lines’ detailed capability against production targets and identified bottlenecks and areas where engineering improvements were needed to ensure optimal unconstrained throughput of the three cars. It was also used to calculate and validate best sequences for the cars and their numerous variants throughout the assembly stages to minimise disruption of the subassembly facilities, the recycling and so on. In the Body Shop, for example, a key objective was to maintain sequences at the merge-points with the engine compartment and manual assembly lines (particularly an issue for the Micra), and to identify associated bottlenecks. These could be caused by the longer assembly cycle times involved in the Almera build, or sequences falling out of synch. Then in the Paint Shop, Nissan had to look at questions of reject rates after touch-up, and the timely delivery of jigs to carry individual vehicle panels through the process. Again sequencing was critical. Impossible manual task Nissan production controller Paul Wreglesworth says it would have been impossible to do this manually. “It was too big a problem. Creating a computer model to simulate the processes was the only sensible solution. We’d already used [Lanner’s] Witness software for some engineering projects, so it was the obvious choice.” Initially, the team created individual models for the main line and sub-assembly facilities – the Body Shop, Paint Shop and Trim Shop – on both the lines. Says Wreglesworth: “Because the task was so enormous we had to build separate models with a view to integrating them later for the big picture. Also, it made sense because we could do testing and validation on the shops without running the whole thing.” There were three distinct stages. First was establishing which elements needed to be modelled, and to what level of detail, across the shops and the buffers protecting downstream facilities from historically expected breakdowns. Wreglesworth says his team worked with data supplied by engineering and validated by production, on cycle times and what happened as cars and assemblies went through . Then the ‘inter-element’ logic – body movements in the Body Shop, sequencing in the Paint Shop and prioritisation protocols in the buffers to maintain sequences and handle line balancing and the rest – was put in, along with the shop layouts themselves. Second, Nissan used simulation to examine proposed weekly working patterns. From a modelling standpoint that meant taking into account strip-out operations for shift-end clearing of the paint ovens and spray booths, and the impact of plant breakdowns on downstream facilities. For the latter, the team initially used historical data gathered from its Demag plant maintenance management system, and then developed predicted breakdown profiles based on a 5% overall downtime plan taking into account the Almera assembly sequences. Says Wreglesworth, “We also had to identify variables needed for logical scheduling decisions to meet the required output of the three cars – for example, to help with prioritisation when it makes sense to put another body in the paint line if we’re already doing that colour and there’s spare capacity.” That done, Nissan moved onto the third stage, creating input schedules to run the models, at first manually, but then as complexity grew, generated by the Ilog APS system being developed concurrently. This in turn led to the development of Nissan’s work in progress (WIP) schedules. Says Wreglesworth, “By the time we arrived at this stage we had a clear understanding of the output data required of the model to quantify production volumes, to enable us to identify the efficiency of the various facilities and predict stoppage times, and to assess WIP levels – the number of vehicles and subassemblies in the facility and in the buffers at any time.” It was a great success. He says that the model revealed several issues for planning sequences based on capacity and cycle times. “We found some of the assembly production wouldn’t be able to keep up – for example, if we ran more than six five-door cars down the line in a sequence. The model helped prove this and demonstrate the impact on the other parts of the plant. It meant we could predict the most efficient sequencing of the three models down the two lines.” For the Body Shop operations Witness enabled Nissan to assess the impact of schedule mix and loss of sequence, and to devise action plans to handle breakdowns. In the Paint Shop it helped raise ‘first time OK’ rates to minimise disruptions and, again, the capacity to accommodate and resolve breakdowns was improved. Recirculation levels in the Trim Shop were also cut and the levels of vehicles held reduced. Says Wreglesworth: “Our douki seisan [continuous improvement] concept aims to reduce lead-time throughout the supply chain. By accurately predicting the time and sequence in which vehicles are built, order lead times, finished vehicle stocks, inventory and distribution costs were all reduced. It was invaluable in enabling us ... to get it right.”