When factories first installed electric motors as a replacement for steam engines and waterwheels, something unexpected happened. Electrified factories turned out to be no more productive than the old.
The reason was that electric motors depended on existing layouts of shafts and drive belts to transfer their energy to where it was needed. While electricity made one part of the operation more efficient, the manufacturing process as a whole didn’t change. Machine tools were still laid out for access to motive power rather than for optimum manufacturing efficiency. And electric motors had a high capital cost, which needed to be repaid.
Electricity achieved its potential only when machine tools were fitted with individual motors, allowing production lines to be redesigned from scratch. In some industries the process took decades.
Today, at last, something similar is happening with information technology. IT is being applied on the factory floor not just to collect information or to control individual tools, but also to redesign whole processes.
In the 27 years since the microprocessor was invented, a revolution has swept through data-handling efficiency. In 1975, the hardware to process a million instructions a second cost $1m 20 years later, that cost had fallen to $20.
The generation of managers who grew up with this revolution are now in their forties and in the driving seat. ‘Things are starting to happen,’ says Leon Opit, founder of LocSoft, a new UK firm specialising in software for embedded microchips. ‘People are aware of technology outside their own particular industry sector and don’t see it as a threat.’
The IT revolution has had obvious cost benefits, but direct payoffs remain hard to pin down. ‘I’m not sure what the benefits are in nails, but I have yet to meet a manager who seriously takes the view that not to have information is a good thing,’ says Opit.
Information systems don’t themselves generate money. They are like lathes in a machine-tool factory: tools for making tools. To justify the cost of systems, manufacturers must apply the tool in a way that radically reshapes their process.
UK firms tend to invest in IT systems for three reasons: when starting up from scratch, typically as the result of an inward investment; when forced to by the need to coordinate complex high-tech processes with other manufacturers; and when government regulation demands extraordinary control of processes.
Honda’s Swindon plant, which exchanges data with suppliers via an advanced intranet, is an example of the first category. The Typhoon European fighter aircraft project, test-flown on screen and designed with tightly integrated product data management (PDM) systems, is an example of the second.
Pharmaceutical companies are an example of the third. Drug producers all over the world are re-engineering their production lines with manufacturing execution systems (MES). These fill the gap between high-level management information and direct process controllers. In the pharmaceutical business, the driver is the US Food and Drugs Administration (FDA) which, since the 1980s, has regulated the quality of the manufacturing process as well as that of the finished product. This new wave of regulation is putting on the pressure in a highly competitive business.
Until recently, says Alan Shields of MES supplier Base Ten Systems, drug makers did not concern themselves too much with manufacturing costs. In an industry which takes an average of 10 years and $250m (£148m) to get a new product on the market, the manufacturing process was almost an afterthought.
The FDA’s rules have changed that attitude. Failure to comply can be fatal. The story of Fisons’ pharmaceutical business is a grim warning. It was forced out of the business when a plant failed an FDA inspection.
Technically, MES fills the gap between MRP (manufacturing resource planning) and the actual plant-control systems. In the most integrated plants, MES is linked to ERP (enterprise resource planning) and PDM. The idea is to provide ‘real time visibility and control of manufacturing processes’.
It creates an audit trail of what each part of the process is doing at any time. This is crucial for pharmaceutical plants, where chasing paperwork can take longer than the manufacturing process itself. Producing one well-known drug involves a batch cycle of 24 weeks, yet the production itself takes only a few days. Each batch is worth £2m £3m an inventory-planning nightmare.
MES not only cuts that delay but enforces quality assurance by ensuring that data from different parts of the processes are constantly reconciled. With manual recording on paper, says Shields, there is always the risk of mistakes or omissions.
This is of interest to other industries. For example, in the US, GE Plastics uses an MES as part of its strategy to cut the number of mistakes to lower than one in a million.
Another industry being driven to use IT systems better is microelectronics. Philips Semiconductors is using an ERP system from Netherlands-based software house Baan as part of a project called DEM, for dynamic enterprise modelling.
The term, coined by Baan, is claimed to be a step forward from ERP. It is a framework for adapting an organisation’s software in real time to fit changing management and manufacturing processes. The capability is crucial if an organisation is not to be held to its software infrastructure as it was once held to a steam engine’s drive chain.
Is this all leading inexorably to that 1980s’ dream, the fully automated factory? Probably not. There are still basic issues of technology to address. One is the apparently arcane choice of operating systems. Microsoft’s Windows CE system, a stripped-down version of the dominant PC operating system, is spreading into embedded chips. Not everyone thinks this is a good thing.
LocSoft’s Opit favours the TCP/IP protocol, which is not dependent on a single supplier’s product strategy. ‘Most of our customers are coming for long-term solutions in a market where long-term thinking seems to be disappearing,’ he says.
Another battle is over database software, between Microsoft and Oracle (CIMfocus, September 1998).
Contrary to predictions of a decade or so ago, IT has not ushered in the age of the ‘dark factory’, run entirely by robots. Humans are still cheap and, if not cheerful, are unbeaten when it comes to improvisation and handling imperfect parts and processes.