How many articles have you read that tout the promise of "shop floor to top floor" integration? Unless you just fell off the apple cart, that number is probably high enough to make you think that promise is not likely to be fulfilled -- at least not in your lifetime. Since the concept of linking production systems to enterprise IT systems first appeared a decade or two ago, it has held much allure for manufacturers, who need visibility into factory floor data in order to make better decisions. But the obstacles -- chiefly disparate data types -- have proved tough to surmount.
Without standard definitions and data formats, production-to-enterprise integration has been a matter of point-to-point projects, which, due to their expense, are mainly the province of the largest companies in the world.
Groups like the Instrumentation, Systems and Automation Society (ISA), the Open Applications Group Inc. (OAGi), and RosettaNet have been working for years to formulate standards to help ease the task of plant-floor to enterprise-data integration, putting it within reach for small and medium-sized companies. Recent developments show potential. OAGi (which concentrates on discrete manufacturing) and ISA (which focuses chiefly on the process sector) announced last December that they would work together under the oversight of the Open Standards for Operations and Maintenance (OpenO&MTM) organization to converge their respective integration standards, with the aim of simplifying the standards picture.
"Alignment of the OAGi and ISA-95 teams is a good thing. It's not just throwing things on the table and saying, 'You take this and I'll take that,'" says Bob Mick, vice president of emerging technologies for ARC Advisory Group. "They will develop common models and implementation strategies," though the goal of the joint effort is not to develop a single standard, since this would not serve the needs of different verticals. This development is particularly welcome news for hybrid industries that have both discrete and process aspects in their manufacturing, Mick adds.
It is also, he says, a positive sign that the groups are ready to work together. "All standards bodies have their own agendas and issues. There is no natural convergence path for them. Their natural inclination would be to keep going on their individual path. But their leaders are taking this on," Mick says approvingly. "Rather than have to deal with two completely different standards infrastructures, there is hope that some of this might be common."
Of course, no single standard will serve every company in every vertical, no matter how much the different bodies work together. "No one standard can cover everyone's needs. ISA-95 was built from the ground up for process manufacturing. It's not a good fit for our customers, who are in the complex discrete space," such as aerospace, automotive, and high tech, says Carter Johnson, vice president of strategy for Visiprise Inc. (Atlanta), a purveyor of manufacturing execution system (MES) software. While OAGi's long-standing OAGIS standard (Open Applications Group Integration Specification) is a much better fit for these sectors, he says, "any standard will lack the granularity of the real-world processes our software needs to handle."
Visiprise, like other vendors, plans to extend the production-to-enterprise integration standards such as OAGIS for specific verticals. "The 'verticalization' of the standard is very important," Johnson says, hinting at the specter of non-standard standards, which has bedeviled manufacturers for time immemorial.
Slow But Steady?
ISA-95 parts 1, 2, and 3 define vocabulary and a model that users from different parts of the manufacturing enterprise can employ as the basis for conversation. This is important, since traditionally the way a business system defines "work order" and the way the production system defines "work order" have been worlds apart. ISA-95 and OAGIS in large part exist to provide a common definition for terms between business and production systems.
"The standard is the vehicle that allows for a common language for what you are integrating. It is also valuable from a project implementation point of view. It helps you define expectations -- who is responsible for what," says Claus Abildgren, program manager for production and performance management solutions for Wonderware (Lake Forest, CA), a unit of Invensys plc of London.
And yet beyond these organizations' encouraging move to work together, the fact remains that the standards are not of much practical use to manufacturers today. At the definitional level, for instance, the standard does not provide a framework to which vendors or manufacturers can code. ISA-95 part 5, currently in draft form, addresses business-to-manufacturing transactions, and will incorporate a subset of language used in the OAGIS standard. Part 5 is expected to incorporate XML schemas, which will go far toward creating interoperability.
No one knows how long it will take for ISA-95 to get to that point, however. AMR Research analyst Alison Smith points out in a recent report that it took 10 years for the ISA-SP95 working group to finalize the first three parts of the standard. Thus it is anyone's guess how long it might take before useable data flows are defined -- even with the help of OAGi. After that, vendors will have to build their products to incorporate meaningful support for the standard, as opposed to just marketing hype. For end-user companies, the standards story often boils down to "hurry up and wait." That is still largely the case with ISA-95.
"In our experience, this is still somewhat academic. Our customers are not clamoring for standards," says Johnson of Visiprise. "We don't have a lot of customers specifically asking for products that are compliant with [ISA-95 or OAGIS]."
Arla Foods amba (Viby, Denmark), a $7.8 billion dairy producer, was one of the first companies in the world to embrace ISA-95, according to Abildgren. To help Arla achieve standards-based plant-to-business integration, Wonderware began working with SAP in 2004, using SAP NetWeaver XI on the ERP side and Wonderware's ArchestrA architecture on the plant side. This collaboration leveraged the latest plant-to-business XML standards defined by ISA-95 in combination with open integration architectures such as business-to-business manufacturing-markup-language (B2BMML) schemas.
"During the last year and a half, Arla, Wonderware, and SAP have taken those B2BMML schemas and put them to use in real-world implementations," Abildgren says.
Arla needed to leverage standards to reduce the amount of custom integration required to rationalize disparate systems. "Arla has become a very big company through mergers and acquisitions. It has around 70 manufacturing sites that all run different systems," Abildgren says. "It had every kind of processing and SCADA system known to mankind. Arla realized three or four years ago it needed a common definition of the business processes, independent of the systems and equipment," he says. "The cost of doing it piece by piece was too high."
So, while most manufacturers are not yet asking for standards-based products to help ease integration -- or for help in figuring out how to apply newly minted B2BMML schemas -- giants like Arla are working ahead of the pack, paving the way of the future.
Full Interoperability
When the discussion of integration standards threatens to become overwhelming, it is helpful to remember why manufacturers are interested in connecting shop floor systems to enterprise information systems in the first place.
At the most basic level, production-to-enterprise integration is about making data that is available on one side (the shop floor) available on the other side (the corporate IT network) and vice versa. The goal is fully interoperable systems, but these systems are often based on disparate hardware, software, and operating systems, and share no agreement on the meaning of the terms (such as "production order"). Viewed in that light, it is easy to see why the goal of plant-to-enterprise integration remains frustratingly out of reach.
Should this sort of integration become possible, however, the benefits would be legion. "I not only get to know what is the schedule for the day from the scheduling system, I also get to see that my sales guy took a rush order from our best customer last night at 8 p.m. and how that order will affect the next shift. The sooner and more efficiently I get that information, the sooner I can act on it," says Krishnakumar Nagarajan, head of plant automation solutions for the Engineering and Industrial Services practice of Tata Consultancy Services Ltd. (Mumbai, India), a $2.25 billion IT services provider.
There are three components to meaningful shop-floor to top-floor integration, according to Nagarajan: data exchange between the systems, useful information extraction from that data, and enablement of real-time management by exception. "This enables you to have plant-level dashboards that show [key performance indicators] and allow you to drill down for more data," he says.
The ultimate vision is to be able to have analytical engines creating predictive analytics from the plethora of information emanating from the shop floor. "They will be able to predict a quality problem coming down the line, a raw materials shortage, and so on. We're a little ahead of that stage," Nagarajan concedes. But the work being done now to define standards that can be used on the shop floor as well as in the boardroom is good news for manufacturing companies today, even if they have to wait until tomorrow to see the benefit for themselves.