Companies who are pioneering the use of IT on their production lines will steal a march on those still without an ounce of silicon in their manufacturing operations.
The factory employee of the future could be an electronic creature with six intelligent legs each capable of operating independently. If one is removed, the remaining legs continue to function effectively, compensating for the one that is missing whether the giant insect is building cars, assembling aircraft engines or filling cans with baked beans.
The Hexapod, as it is known, has been developed by the cybernetics department at Reading University to demonstrate how a series of computer processors (located in the legs) can communicate with each other and tailor their own actions in response to the information received.
Professor Kevin Warwick, the department's head, says: 'The advantage of such a technique in manufacturing production is that it is fairly cheap, very reliable, very simple to set up, and fault tolerant.' The Hexapod technology is already being used in the manufacture of printed circuit boards at Quad Europe's factory in Reading. However, most UK factories have nowhere near this level of computerisation. They are certainly spending money - last year the manufacturing sector forked out some £2.6 billion on computers, equivalent to 24% of the total IT market and second only to financial services. But the overall picture is still primitive.
Numerically controlled machines such as lathes and mills are in fairly wide use, says Paul Watts, research manager at Benchmark, a market research consultancy specialising in factory automation. But relatively few shop-floor machines collect information and send it back to managers. In an ideal situation, production-line machines would use computer technology to gather information about the processes they are involved in and transmit it back to managers in a simple, intelligible format - perhaps with coloured graphics to highlight areas where production is falling behind schedule or where variations in loading are causing a machine to function inefficiently. Computer technology can also be used by machines to adapt themselves to the task in hand - speeding up to handle particularly heavy workloads or shutting down to save energy during quiet times and re-starting when volumes reach a high enough level. Yet, despite the high levels of expenditure, few factories have reached such a level of sophistication in their use of IT.
'Among factories with more than 50 employees, it is still the case that only a minority have any form of shop-floor automation,' says Watts. His view is shared by Warwick: 'If you walked in to your average large manufacturing production company you would be hard put to find an awareness of any very new developments in engineering, let alone IT.' One problem, Warwick reckons, is that manufacturers are not hiring people who know about computers; another is the pressure for return on investment. 'If you invest in IT you'll still be in the picture in five years' time, but most manufacturers in the UK have much shorter timescales than that.' These problems have been exacerbated by the unsuitability of many products developed by IT suppliers in the past. The classic offering is the Manufacturing Resource Planning (MRP) package which became widely available during the late 1980s. It was inflexible and ill-suited to many factory requirements. A survey carried out by Benchmark in 1993 among UK manufacturers found that overall satisfaction with MRP systems scored only a marginally favourable 3.6 out of 5. A significant minority of participants, 16%, gave worryingly low ratings of 1 or 2. 'Lots of early MRP systems were designed to match a stereotyped view of what a manufacturing operation is, whereas in practice factories have a wide variety of different requirements,' says Watts. Companies that wanted to computerise had either to change their procedures or adapt shrink-wrapped software to suit their needs - often leading to huge overheads in software maintenance.
The bad experiences of manufacturers who adopted IT in the 1980s are still taking their toll, even among quite sophisticated users. For instance, Pedigree Petfoods wants to develop reporting systems to track the performance of its Melton Mowbray canning plant and print out daily shift reports. It recently looked at what could be done by adapting off-the-shelf software. 'Many IT suppliers said they could do it,' says Steve Allanson, the factory's information systems manager. 'But we were worried that there would be high, up-front investment costs and that the system would then take a long time to get going.' The fear was that the business needs would have changed by the time the system was implemented.
During the early 1990s, the recession did little to encourage manufacturers to be more adventurous with IT. But now that the economy is picking up, they are looking again at the power of computing as a competitive weapon. Most factories want a better way of controlling and monitoring activity on the factory floor. They want to 70e gather information such as employee time and attendance statistics, production quality, material location, and job status. They also want IT to improve their versatility and speed to market. This means drawing together data from all sides of the business including sales, finance, distribution, shop-floor data collection, time and attendance, asset and maintenance management and office automation. Not only must it be collated and presented in a consistent format, it must also be constantly updated.
This type of highly-integrated system is now in operation at Shropshire-based Fullwood, a manufacturer of milking machines with 290 staff and a turnover of around £16 million. Since Fullwood's system went live in November 1993, stock turns have more than doubled from four to eight-and-a-half a year. The percentage of machines ready by their target date has also doubled - from 40% to 80%. Many of the business processes have been streamlined: sales staff can monitor the progress of work on the shop floor; production staff know when required materials will be available; and customers can be given firm delivery dates rather than vague estimates. One bonus of the system, which cost around £200,000, is there are fewer complaints from dealers, says Meirion Davies, the company's materials director. Previously, they had to compensate for Fullwood's inefficiency by ordering early and holding more stock. 'Now dealers are starting to believe in our ability to deliver on time and we're hoping this will eventually mean more orders,' says Davies.
Industries such as motor, aircraft and semiconductor manufacture tend to have much of the most advanced technology. They are increasingly using computerised production histories, for example, to track the performance of goods after they have left the factory. Subsequent failures in air-bag activators or engine components can then be traced back via chassis numbers to the precise machine tools used and the amount of torque applied during assembly.
As for fancy robotics and cybernetics, the enthusiasm of most manufacturers depends on the perceived business benefits. Semiconductor manufacturers such as US-based Quad are among the world's most advanced users of robotics. 'The high level of investment required is worth it for companies in wafer fabrication,' says George Mackison, industry consultancy manager at Digital Equipment, the US computer manufacturer. 'If their yields go down they can lose millions of pounds.' The pharmaceuticals and food-processing industries also tend to be technical trail-blazers. 'The business driver is there because everyone wants safer foods and drugs with more information about contents and sell-by dates,' says Mackison. Supervisory systems at Pedigree Petfoods, for example, can set up the machines according to the recipe or process required. They can spot when a line has not got enough cans available, and instruct it to wait until it has. And they can slow down or stop the process when they identify a breakdown further along the line.
Some manufacturers are trying out production machines that use artificial intelligence and neural networks - programming techniques that mimic the operation of the human brain and can radically improve on it. Such systems have no difficulty thinking in nine dimensions, whereas humans find it hard enough to think in two or three. This makes them potentially better at machine minding than human operators.
Information such as motor-speed, product temperature, moisture content and flow can be measured and fed into a computer system which is quite capable of dealing with a high number of related variables. The computer then works out how the variables are related, says Warwick. 'It learns what are good and bad modes of process operation, and which should be avoided due to a high probability of failure or at least poor product quality.' These techniques are being adopted for high-speed production lines at Molins, which makes production machinery, and Unilever, the Anglo-Dutch group.
For manufacturers, as for other large-scale IT users, the biggest change in adopting new technology is cultural. 'In the past, manufacturers had walls erected between departments,' says Dave Anderson, manager for manufacturing at Bull Information Systems, suppliers of the Fullwood system. Now these walls are being demolished. MRP, or ERP (Enterprise Resource Planning) as some suppliers have rechristened their packages, requires manufacturing teams to liaise with many other departments. The wider the system expands across the organisation, the harder it is to plan, justify and implement.
Success hinges on users having a clear idea of the information they need to gather and why it is necessary. There's no point in having technology for its own sake but there are many tasks that IT can perform, often more efficiently than humans. As the pioneers are demonstrating, IT can gather a wide range of shop-floor data and transmit it in an intelligible form to managers. It can also adjust production machinery to the constraints and opportunities of the task in hand, ensuring that processes run at optimum efficiency.
Computer-based production systems are not as expensive as many manufacturers appear to think, says Warwick. 'The cost will also be offset by the more widespread and important role that computers will be able to take in operational activities as they become more intelligent,' he reckons. And as they become more intelligent, they will be able to learn from the results achieved, their subsequent action being influenced and enhanced by what has been learnt. 'This should increase their cost-effectiveness for many users in the manufacturing sector.' In a market where some companies already have multi-dimensional information-processing and electronic Hexapods, those without an ounce of silicon on the shop floor had better be worried. Companies that implement advanced IT now will steal a march on those that don't, warns Warwick. 'The others may continue making a profit for a year or two, but in the end, they'll fall behind.'.