Simulating the factory

It's not just components and products that are now simulated and tested in software, before any hardware is built - as E&T discovers, companies are simulating entire factories and supply chains as well.

When management at Honda's Swindon car plant was planning the installation of a new engine block machining line in July 2007, the company made a conscious decision to depart from previous practice.

Lee Beggan, an engineer involved in planning the line, travelled to the headquarters of Redditch-based simulation software vendor Lanner Group. There, he was trained in the operation of Lanner's 3D Witness simulation tool, which allows businesses to build 'virtual' production and supply chain processes, and then simulate their operation under various conditions.

"There was a lot going on in terms of new equipment, conveyors and overhead gantry robots, and we wanted to know if the layout and concept design was realistic in terms of the planned daily output rate," he explains. "We didn't want to rely on equipment suppliers' estimates in order to work out cycle times, buffer sizes and operational parameters."

The machining process in question wasn't especially complicated, says Beggan. Two machines operating in series performed two separate machining operations. These then fed three machine tools performing the same operation, in parallel. Overhead gantry robots were responsible for picking up the unmachined blocks at the start of the process, and moving them from machine to machine. "We didn't want to have bottlenecks in the process - or unproductive gaps, either," says Beggan.

The simulation didn't take long to create, and confirmed the 'gut feel' of the engine plant's management: only one overhead gantry robot was required to keep the process running - and not two, as had been suggested. The saving was substantial.

"Trying to get the same result without simulation would have involved massive Excel spreadsheets or complicated Gantt charts," sums up Beggan. "And if we'd made a mistake, it would have been difficult to spot."

Subsequent simulations have proved just as valuable - and even more difficult to perform using alternative techniques. More recently, for instance, a simulation of an extension and modification to the door welding line - using actual data, and not estimates from when the plant was commissioned, 15 years ago - showed that the line wouldn't be able to achieve the target output. Forewarned, management were able to take action.

Simulating products

The use of simulation during product development isn't new. Over the past 25 years, huge advances have been made in the use of techniques such as computational fluid dynamics and finite element analysis to maximise product performance at the design stage, rather than tweak it subsequently. Cheap computer horsepower; readily-available solutions; good interfaces with existing design tools - whereas simulation was once restricted to companies in sectors such as aerospace and defence, its use is now far more ubiquitous.

The result isn't just better products. Stourbridge, West Midlands-based Vee Bee Filtration, for example, uses simulation tools to develop cost-effective filtration solutions designed around the needs of specific customers. Previously, says Vee Bee research and development engineer Napoleon Motaban, the need for expensive and time-consuming analysis activities and laboratory tests meant that the company could only offer standard products.

Now, with computational fluid dynamics simulation software from Blue Ridge Numerics linked to Vee Bee's PTC Pro/ENGINEER CAD system, "we're able to run analyses and generate design models at the same time", says Motaban. "The CFdesign software reads the 3D CAD model directly, and you don't need to be an expert in computational fluid dynamics to use it."

The use of simulation to model - and improve - production and logistics processes is a much more recent phenomenon, but has been driven by many of the same factors. The objectives, though, are very different. 

"A huge focus of interest is improving manufacturability," says Bruce Klimpke, technical director of Winnipeg, Canada-based Integrated Engineering Software. "Companies find that they can design a device that works, but which then costs too much to manufacture."

And those costs can stem from a wide range of causes - machining time, reject rates, material use and assembly times. At Cambridge-based industrial printing manufacturer Domino Printing Sciences, the costs in question arose through the poor manufacturing yield of a recently introduced ink jet print head.

Discovering that the causes of the yield losses were impossible to determine through conventional means, as the print head was simply too small to permit the use of probes, the company turned to finite element analysis-based simulation.

It called in King's Lynn, Norfolk-based Fine R and D Ltd, which had been founded in 2002 by Dr John Clark and his wife Dr Jeannette Fine, who between them had 40 years' experience of simulating industrial products and manufacturing processes.

"An ink jet print head is a complex object to model," observes Dr Clark. "It contains piezoelectric actuators, solid parts, a reservoir of fluid, bolted joints, and contacting surfaces which might be out of flat by up to 5 microns - all of which in combination present formidable challenges to the modeller."

Nevertheless, Fine were able to eventually develop a working model, which duly highlighted where the sources of product failure were. The result, says Dr Clark, was that process yield went from a little over 50 per cent to around 90 per cent - a significant increase.

And yield losses aren't the only cost that has manufacturers turning to simulation tools for insights into how production processes can be improved. At Santa Ana, California-based MSC Software Corporation, which has been developing computer-based simulation tools for 40 years, senior vice-president of marketing Ted Pawela is seeing a renewed focus on the use of simulation to explore the detail of manufacturing processes.

"From soldering to forging, and stamping to injection moulding, manufacturing processes affect reliability and quality," says Pawela. "If you blow mould a tank for an automobile, the resulting thickness isn't the uniform thickness of the part as it appears on the CAD screen - it's affected by the manufacturing process itself. Likewise, in injection moulding, stamping or forging, you get 'thinning' at tight corner radii."

With processes such as welding and soldering, on the other hand, the issue is impact of the associated heat on related parts. "The thermal process actually affects the surroundings of the joint, and can add fatigue," Pawela explains. "By modelling the process, it's possible to understand how that heat is going to affect design reliability and performance."

Material science

And such 'material science' simulations are becoming increasingly straightforward, as the major developers of CAD technology - like Dassault Systèmes, Autodesk and Siemens PLM Software - make it easy to link the 3D design containing a product's geometry and material parameters to programs that will perform simulations based on that data.

Take Runcorn-based APPH, for instance. Part of aerospace first-tier supplier BBA Aviation, the company uses Dassault Systèmes' Simulia suite of material science simulation tools to optimise the trade-off between weight and service life on the undercarriage gear it manufactures for military aircraft.

"It's about complex geometries where hand calculations are difficult to carry out," says Geoff Haines, managing director of Oxfordshire-based Desktop Engineering, a Dassault business partner, and himself a former stress analyst. "You can go to a much higher level of accuracy than you can with hand calculations - optimising the weight without the need for such a big 'fudge factor'."

But if the growth of materials science-based simulation has been facilitated by easy links to CAD platforms, one of the hottest areas in manufacturing simulation is prospering with few such obvious synergies.

Variously described - terms such as 'kinematic simulation' and '3D discrete event simulation' are employed - it's about exploiting the power of simulation tools to model entire manufacturing or logistics processes. And the recession, if anything, is bolstering manufacturers' enthusiasm.

"We're seeing a lot of enquiries - and from industrial sectors that we don't usually hear from," says David Jones, chief executive of Lanner Group, which - as a former part of British Leyland owned Istel Ltd - traces its roots back to the 1970s, and which developed the Witness simulation tool employed by Honda's Swindon plant. "As the global economic slowdown continues, it seems that process simulation is increasingly recognised as an approach which can deliver significant productivity gains and efficiency savings."

On one level, explains Haines, the attraction is fairly simplistic - digitally 'modelling' pieces of equipment, in much the same way as one might with physical mock-ups, in order to perform basic validations.

"Take robots welding body panels on an assembly line, for instance: you want to make sure that as they operate, they don't collide with each other," he explains. "You can model an entire flow line, and know on a second-by-second basis exactly what it is doing."

Human interaction can also be modelled, via digital mannequins. The design of Boeing's 777 - the first to be designed using digital assembly simulation tools - had to be changed when it emerged that a digital mechanic 'discovered' that a human mechanic wouldn't be tall enough to change the bulb in the red navigation light on top of the aircraft. "Think of it as traditional time-and-motion studies - but without the people," says Haines.

In the UK operations of Airbus, for example, Dassault Systèmes DELMIA simulation tool is used to develop optimum production procedures for major wing assemblies, as well as for large component machining.

"3D design data, developed using Dassault Systèmes CATIA, is re-used in DELMIA with models of production facilities and workforce mannequins," explains Geoff Tantum, simulation group leader at Airbus' Broughton plant. "The software also simulates jigs and tools so that assembly sequences, reachability and accessibility can be accurately investigated."

But - as at Honda's Swindon plant - it's in the optimisation of complex manufacturing and logistics processes that simulation really comes into its own.

Mirko Bäcker, European marketing director for digital manufacturing at Siemens PLM Software, is fond of using a particular example when explaining the benefits of simulation in this context.

"Take a simple line with five machines operating in series, producing a part a minute, and with each machine having a 90 per cent level of availability - how big should the inter-machine buffers be?" he asks. "Most people assume that the line output is going to be 90 per cent of 1,440, which is 1,296. In fact, it's only 850, an output drop that often stuns them. Through simulation, we demonstrate that an inter-machine buffer of 20 is required - and that a buffer that's any greater than 20 doesn't increase output further."

At Nissan's Sunderland manufacturing plant, Lanner's simulation tool performs a similar task, optimising the use of purpose-designed stillages and containers used in materials handling processes.

These are expensive, explains Nissan industrial engineer Anthony Timmiss, so the idea is to keep their numbers low - but there's a trade-off involved. Should the demand for the stillages and containers exceed supply at any given time, ordinary wire baskets can be used. But parts can't actually be transported in wire baskets, merely stored in them, so their use necessitates a costly transfer operation, involving both operatives and forklift trucks.

 "We run the simulation with set numbers of each of the nine types of containers we use, together with a theoretically unlimited number of wire baskets, and the simulation calculates the number of wire baskets required to keep production going, and the labour cost of the associated transfer," explains Timmiss.

Airbus has gone even further, with manufacturing simulations feeding back into the component design stage, using simulation-driven insights to guide design-for-manufacture and design-for-assembly decisions. "Complex operation procedures can be clearly seen, leanness achieved, problems avoided, and decisions made in the certainty that they will be the right ones," says Tantum. 

So far, simulations in such painstaking detail are a relative rarity in manufacturing industry - despite confident assertions of the benefits from those who have taken the plunge. "The cost savings we've generated are huge," stresses Nissan's Timmiss, for example.

But longer term, that will happen - with 'automotive like' industries with high volumes and high levels of repetition leading the way, says Prabhu Bendre, the Pune, India-based senior vice-president of manufacturing business at specialist consultants KPIT Cummins.

"Design for manufacturability is becoming increasingly important - there's a growing awareness of what the automotive and aerospace sectors have achieved," he notes.

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them