Saving a packet on the digital factory

Computers today have sufficient power to simulate anything from a production process, to a complete factory, or even an entire supply chain. And clued-up manufacturers are reaping significant benefits from using them. By investing the time and money to build and test virtual (digital) factories electronically before committing to real physical outlay, they have been achieving substantial savings – and making better decisions. “The savings you can attain by increasing the utilisation of equipment, and having the knowledge that that it is going to work as predicted, are a major benefit,” says Mike Allan, simulation manager at Durr Automotion. The second largest facility supplier to the automotive industry, Durr uses AutoMod software to simulate paint shops before they are physically built. Allan also says there are several other advantages to simulation. “It isn’t political – it doesn’t have an axe to grind. Because there are no personalities involved and it is completely objective you can quantify the benefits of each alternative without having to take into account whom it might upset or please.” Using a 3D graphical system such as AutoMod is also particularly useful for visually presenting your ideas – whether to the board of directors or, in Durr’s case, a customer in the automotive industry. “With a graphical simulation, the customer can see exactly what’s happening, and comment on whether they agree with the way it’s working and want any changes. If you just send a letter, they can’t visualise what the problem is exactly.” Simulation can also help pick up CAD problems, says Graham Carter, joint managing director of simulation software developer AutoLogic Systems. “We can import the CAD files and build the model from that, so a simple mistake on the CAD layout would be spotted.” Simulation also helps allow for future expansion, explains David Jones marketing director of rival Lanner Group. “You can model over a period of time – perhaps 12 months. You can run that in a matter of minutes and see the impact of future growth or change on that plant or process.” In fact, simulation can be applied to many different types of manufacturing, according to Eric Gautier European marketing director for one of the best known ‘digital factory’ software vendors, Tecnomatix. “It can be flow simulation, discrete event or material handling. We have customers in heavy machinery manufacturing, electronics, automotive, aerospace and even pharmachem and logistics.” And although AutoLogic’s roots are in materials handling and manufacturing, it’s products are suitable for service industries too. “Our customers include fast-food restaurants – we can model the car parking places and the number of servers, for example,” says Carter. According to Jones, the Lanner Group is also doing a lot of work in the services sector. “We’re tackling bottlenecks in financial institutions,” he says. Customers are reporting dramatic savings. “On one particular job at what was then known as Rover Oxford, a quick and dirty simulation agreed with the customer’s own conclusion – that Rover needed to spend £5 million on building a new paint store,” says Allan. “But once we did a simulation of the entire process, we found the existing store was adequate – and saved £5 million.” Some saving. “On another occasion,” continues Allan, “a system that balanced V8 engines was a bottleneck, so it was decided to build another one at a cost of £500,000. About three days’ work with simulation came up with a small change to a conveyor system instead – and saved that half-million.” There are many other examples in quite different industries. Richards Bay Coal Terminal (RBCT), for example, the world’s biggest coal export terminal with a capacity of 60 millions tons a year, implemented knowledge software firm Gensym’s G2 expert system to develop a detailed dynamic simulation model of its existing complex. The simulation runs showed that if RBCT changed its operating strategy, production could increase by 13%. Then at Biwater, a company that produces iron pipes and fittings for the water industry, Quest (supplied by Delmia) was used to analyse a problem with a bottleneck in the finishing stages. Ben McIvor, Biwater’s Development Engineer says: “This allowed us to determine what percentage improvement we would gain from developing each individual station on the production line in a given sequence, so realising the maximum return on investment for each stage.” And although these examples relate to improvements to existing plants, one of the most common uses of simulation is to test designs for new production facilities before they’re built. “When planning a layout, and the design and operation of a new facility, you first want a fairly high level simulation to prove that the concept is valid and check that there are no unexpected consequences,” explains Allan. “That is just a two to three day quick and dirty simulation to give the confidence that you’re heading in the right direction. If you wait until you’ve designed the full plant, then it’s too late – and you have to live with any fundamental design errors.” He continues: “At the next step, you are committing yourself to a specific implementation of that design, and getting down to the nitty-gritty of cycle times, process times, where conveyors are going to be placed and so on. At that stage, we get down to a much more detailed simulation – it can take between one week and three months.” This second stage allows you to look in detail at effective utilisation of a transfer car, for example. “You can also examine the way the control rules operate,” says Allan. “You can design a store, but you don’t know how it will work in real life until you put the rules in – so simulation gives you a chance to try out the rules and refine them off-line.” To identify savings means trying out different possibilities. “There are so many options in optimising a scenario,” explains Jones. “You have literally millions of options. With technology such as ours, you put in the options and it will go away and seek the best optimisation depending on your criteria – for example, maybe you want to maximise throughput but minimise headcount.” Similarly, Quest features an optimiser that uses goal-seeking to find the best solution to suit your particular constraints. “You can set an objective – say, cost or throughput – and then define the constraints, such as the cycle times or human resource limitations,” explains Delmia technical team leader Tracy Doran. “The optimiser uses some of the leading algorithms and search mechanisms, and employs neural networks to add some intelligence.” So far so good. When simulation is used to test improvements and changes to an existing plant, the first step is to model what’s there already. “Quite a lot of processes haven’t been fully understood,” says Jones. For example, one of the problems faced by Peugeot when wanting to expand its Ryton plant, owing to the success of the Peugeot 206, was that of building a model to simulate the changes needed to accommodate greater production volume. 3D modelling was required, but the firm’s existing CAD data was only two-dimensional. In the end, Peugeot was able to use a laser 3D modelling system from UK Robotics – known as LFM – to create a comprehensive model of three cells in the line where the body sides are manufactured, and five cells in the line where the underbody is manufactured, with the minimum of fuss. For many manufacturers, simulation is now compulsory, says Doran. “They insist that any changes made to existing lines must be accompanied by a simulation model up front to ensure that it is the right thing to do.” Jones is seeing the same caution being observed. “Quite a lot of customers have an always-simulate policy. If you don’t simulate first, you don’t invest.” However, simulation must be approached with care. “It’s a critical-path activity,” explains Allan. “Once the layout is finished they want to commission it, but between layout and commissioning you have to have a simulation, so you must make sufficient time available to perform the simulation and learn from its outcome. You might have to go back to the drawing board.” One of the improvements being made by software vendors is to make it faster to build the models. “Our latest technology allows you to import operating characteristics such as conveyor speeds directly from CAD layouts, allowing you to convert the models directly into simulation instances.” Meanwhile, although all this might sound expensive and complex, it transpires that simulation is appropriate even for smaller companies and plants with commensurately smaller problems. “One customer has a cake-making plant – quite a small outfit,” says Doran. “They use simulation when changing their product mix to make a different type of cake. They can simulate it to ensure they can achieve the desired throughput and production quantity.” Indeed, some software vendors offer different options for smaller companies who wish to exploit simulation. Delmia will make a model for them, and they can then use Quest Express to perform what-if tests on this model. AutoLogic offers entry-level products, which can also be used by larger manufacturers to model a small part of their facility to allow them to start working with simulation in a small way. “You also don’t need to go through weeks of training to use it,” says Carter. “We can provide on-line tutorials, and we offer an optional one-day training class.” Then again, according to Jones, “the entry level for the type of software Lanner is involved with is around £15,000. At that level of investment that is not a barrier to companies that could be saving tens or hundreds of thousands of investment on capital equipment or resources. We can very quickly simulate a plant – it’s not rocket science.” So what’s the next step? Once companies of any size have got to grips with simulation, Carter believes the next step is emulation – something that AutoLogic’s products can already do. “With emulation, instead of modelling what your PLCs do, for example, we actually connect the model up to the control systems that are going to control the real hardware.” For example, when a photo-eye in the simulation model trips, the message can be sent to the actual PLC, which then reads it and makes the decision just as it would in the real world – whether to stop a motor, machine or conveyor. That decision is then fed back to the model. “You can test your PLCs and other systems in a very low-risk environment,” says Carter. “This also allows you to test your plant at higher rates, to allow for future growth. You may want to commission a plant at 60 jobs per hour, but need to plan for 90 per hour in two years. In real-life commissioning you may only be able to feed in 60 jobs per hour, but with emulation you can test it at 90.”