6 August 2008

Sequential development, with manufacturing engineers waiting for design engineers, and process engineers feeding to controls engineers two weeks before launch, is old and gone

Robert Axtman, executive director at Delmia Knowledge of exactly what is happening on the plant floor can lead to better decision making and improved use of assets. But turning knowledge into action requires integrated technology, writes Mark Webb.

Most manufacturing companies have implemented the first generation of visibility. Plugged in to the robots, lines and machines is SCADA. For example, a palletiser at the end of a production line might have SCADA on it to provide information about how the unit is performing. “But this is a very narrow view,” says Nick Barnett, Microsoft’s EMEA director of manufacturing operations. “It would be helpful for the operator to know what is happening upstream and what affect his actions have on earlier parts of the production process. Certainly the line supervisor wants an overview of the whole line and the plant manager a view of the plant.”

Barnett explains that manufacturers want to know the reasons why a machine is running at only 77 per cent so that, by continuous improvement, they can get it up to 80-85 per cent. By learning what has gone wrong, engineering can get involved and get usage up. In a growing business, adding capacity to existing equipment means you don’t have to add new lines at considerable cost.

The manufacturing execution system (MES) works with SCADA and is like enterprise resource planning (ERP) for the plant floor, carrying out functions such as tracking where orders are up to and monitoring downtime. It also acts as an integrator when making plant floor data available to the wider business, notes Barnett.

Microsoft technology gets the parts to talk to each other. Typically in a large manufacturer, BizTalk Server enables communications with SAP ERP. In a smaller organisation, a partner application may work alongside Microsoft Dynamics NAV.

Integration is the key, as global manufacturer of commodity chemicals and polymers Lyondell has found. “Lyondell needs accurate and timely feedback loops and operational models for dozens or hundreds of people, working in diverse capacities and locations and dealing with unique and specific time horizons for dozens or hundreds of different activities,” says Eric Silva, vice president of information technology. “But until recently,” Silva adds, “they could not make those decisions as rapidly as they would have liked, because they lacked a timely and comprehensive method of collaborating, communicating and data sharing. Instead, they relied on conference calls, e-mails and distributing spreadsheets.”

And that, notes Kara Achtermann, manager of technical development, was a problem that Lyondell IT executives were determined to solve. So they set out to integrate the company’s manufacturing and process-control applications with enterprise resource planning and other lineof- business solutions throughout the company.

Lyondell launched the integration project by defining three primary goals: to simplify the management of non-compliant events, to dramatically reduce delays in decision making, and to implement an efficient and cost-effective way of mining the exponentially increasing data emerging from the plants.

Among the factors in the selection of Microsoft integration technologies was that those technologies also served as the foundation of the company’s manufacturing and process-control solutions, such as Aspen One and other applications from Aspen Technology.

For each of the Lyondell olefins plants, Aspen One runs in the manufacturing control room and is integrated, through Microsoft BizTalk Server, with upstream systems at headquarters, regional offices, and throughout the enterprise. The front end of the solution, based on Microsoft SharePoint Portal Server, delivers information.

Microsoft SharePoint is finding a leading role in getting useful information to individuals and teams on the plant floor, says Barnett: “SharePoint’s early take-up was around visualisation. Because it is Webbased, it avoids the difficulty of managing PCs out on the shop floor. You don’t have to install applications, which is difficult enough in an office, but causes unacceptable downtime in a factory.

“Now plant floor staff are beginning to exploit the collaboration features of SharePoint,” continues Barnett. “And often without the involvement of the IT department.” A group of engineers will install SharePoint for its own purposes and the installation builds up organically from the bottom up.

The culture of workarounds and getting things working against the odds means that plant floor engineers are happy to take responsibility for sharing information with colleagues, as they go on and off shifts for example, via a SharePoint Web site.

Microsoft partners integrate server technology with industry-specific applications and make sure that the solution is appropriate to the culture of the plant floor. Staff on the shop floor don’t want to bring complexity into an environment in which they are paid to produce things, not use software, says Romain LaVault, vice president at Intercim.

Workers are also mobile so Intercim’s Pertinence, built on .NET, works on a tablet PC with only Internet Explorer installed, communicating with the rest of the business via WiFi. The tablets replace a paper-based list of instructions with an all-digital workflow.

Staff working on the plant floor can receive instructions as 3D visuals, with forms to fill in to confirm that jobs have been completed. A scanner attached to the tablet is an option to scan barcodes to ensure the correct components are being used, as well as to confirm job completion.

Workers can hit a red button on the tablet to ask for help from the ‘control tower’. It’s not clear if Intercim got this name from their work with aerospace manufacturers, but it describes the situation well. The tower receives the worker’s alert, can view the status of the job under way and can then respond with a better explanation or more detailed instructions.

Intercim takes three months and two people to understand a manufacturer’s processes and to get its data into Pertinence. After that, says LaVault, the manufacturer can configure without needing any recoding.

At Teleste, a Finnish company that manufactures products for broadband video, plant floor supervisors can read online statistics about the yield from large screens on the shop floor. These real-time displays are created in SharePoint and they show statistics derived from the Predisys solution that analyses production data in real time. “We can build very intuitive user interfaces with SharePoint,” says Predisys’ Ari Pihlajavesi, who believes SharePoint is the best technology available for bridging various departments, including the plant floor, in a manufacturing business.

Based on the Microsoft BI stack, Predisys adds the manufacturing focus with its Sharepoint based statistical analysis solution that picks up early signals of problems, helping to identify their root causes and alerting staff before a small problem becomes a big one. “The production line reacts immediately to online alarms and generates corrective actions. Also R&D teams participate in solving product design related issues and co-operation between R&D and manufacturing has improved a lot. Now the supervisors truly monitor what goes on in the factories and in production,” says Jari Saali, director of the quality and process development department. “This creates considerable improvements in each production line as even the smallest changes are so easily visible.”

Line operators follow the yield of each product. “We use statistical methods to distinguish the signals from the normal variation of the processes,” says Pihlajavesi. “Using these early warnings means you don’t have to stop the processes, but can do corrective actions on the fly.”

Feedback from data analysis tools helps engineers improve the product manufacturability as well, points out Pihlajavesi. For example, a purchasing department may find the cheapest components to build the product but if you don’t know how that will impact the product manufacturability or yield, you can make very expensive mistakes. If you save ten per cent on the component price, what will be the impact in the final product?

“In IT terms we are talking about a vertical extension to traditional business intelligence,” says Pihlajavesi. “Moving into deeper analysis of the data and more into business activity monitoring type of applications. We are moving from reporting to proactive actions and control.”

Unusually, but perhaps showing the way forward, Teleste also uses these statistical tools in its R&D department to improve the next generation of products, by optimising the product design parameters and specifications, and especially to speed up the new product introduction (NPI) process through applying real-time statistical analysis to product test data in prototype and zero series production phases.

Manufacturers can ensure that machine programming and components are right first time by modelling the whole manufacturing process and sharing project data from the moment the first line is dragged out on a CAD workstation screen. “Using collaborative technologies, design engineers can share data with process and manufacturing engineers,” says Robert Axtman, executive director at Delmia.

“Sequential development, with manufacturing engineers waiting for design engineers, and process engineers feeding to controls engineers two weeks before launch, is old and gone,” says Axtman. Now everyone concerned with a product design can access the data. Manufacturing engineers can feed back to design engineers to optimise the build process. Automotive manufacturers are using collaboration to get models to market in nine or ten months, when it was recently up at 24 months.

Delmia is Dassault Systèmes’ virtual production technology and is designed to model the robot motion paths, machine tools, factory layouts; indeed everything necessary to replicate what will happen when manufacturing begins in the real world. Cost of change is very high, points out Axtman. If you can find errors early in the process then it’s easier and cheaper to change.

Using Delmia, engineers can verify and validate operations in simulations, detecting any failures in sensors or programmable logic controllers. Engineers can now take a laptop onto the plant floor and download instructions to the machines based on Delmia’s offline run-throughs. Equally, 3D images and video work instructions, based on the procedures devised in Delmia can be sent to workers in the real factories when they are built.

It seems that visibility is only the start when technology can analyse the data and deliver statistics, suggestions and instructions back to the plant floor workforce to improve current processes. And, through further analysis, positively influence the ability to manufacture the next generation of products or, indeed, to determine how to design the next production line or factory.

This article first appeared in the Summer 2008 issue of Prime.

Add a comment