Another fine MES we’ve gotten into

Manufacturers, consultants and system and software vendors have been talking for the best part of 25 years about the need to achieve closer integration between the plant floor and corporate level IT. Talk of computer integrated manufacturing (CIM) and of General Motors' Manufacturing Application Protocol (MAP) in the 1980s may have given way in the '90s to catchier tags, like 'shopfloor to top floor' and 'sensor to boardroom', but the message was still one of solving physical and data incompatibility problems. Of connecting disparate business IT and automation systems. Most vendors had their versions of logical IT diagrams showing the manufacturing enterprise structured as a series of stratified layers, each with its own horizontal communications network and each linked by various, often unspecified, means to their immediately adjacent higher and lower layers. Quite what was being transferred up and down between these layers, and in quite what form, was often rather less clearly defined. But the overall impression was one of total integration between plant or factory floor-based sensors, switches and actuators, and the machine on the chairman's – or, more likely, his secretary's – desk. Today that hierarchical structure has collapsed both physically and conceptually. Physically, the structure had recognised the received wisdom that the hardware, system software and communications media and protocols of the IT world were unsuited to the hurly-burly of the plant floor. Similarly, peer-to-peer comms at the cell level was seen as posing different problems from that between controllers and field devices. Those assumptions have now been overturned. The Windows platform is almost ubiquitous, and Ethernet extends right down to the controller and I/O level. Indeed it is only constrained from reaching out to field devices themselves by cost considerations which will inevitably succumb in time to economies of scale. The result has been to remove the physical justification for a hierarchical segregation at a time when that same hierarchy is under challenge conceptually, as manufacturing is increasingly seen as but one element in a supply chain extending seamlessly from raw material or component supplier to customer or consumer. But the hierarchical pyramid isn't just being eroded from the top down; it's also being undermined from below. The automation world has been talking about digital fieldbus plant comms for so long that it comes as something of a surprise to realise that adoption of the technology is now expanding exponentially. Profibus International, the body responsible for the most widely used of all fieldbus protocols, claimed 10 million nodes installed world-wide by the end of 2003 and predicted that figure doubling to 20 million by 2008. There's no reason to suppose that the other protocols such as US-market leader DeviceNet or the oil and gas and petrochemical industries' preferred solution, Foundation fieldbus, are not showing similar rates of growth. Fieldbuses are accelerating Couple all that with a complementary expansion of the inherent intelligence and capability of the latest generation of 'smart' plant level devices, and the result is an explosion in the information that field devices are making available to the enterprise. For example, today's measurement devices don't just measure one parameter and transmit a signal back to a controller. They can handle multiple parameters, perform complex calculations and transmit the values of higher level derived functions, while at the same time monitoring and reporting on their own condition and advising of their need for maintenance or calibration. The onward march of Industrial Ethernet will do nothing to curb this trend. Quite the reverse. All of the major fieldbus and plant level communications protocols – Profibus, DeviceNet, Foundation fieldbus, Modbus – now have their own compatible Ethernet implementations. And while some observers predict a return to the 'fieldbus wars' of the past, the implications for end users are less disturbing since multiple protocols will happily coexist on the same Ethernet network. That suggests that Ethernet can only further accelerate fieldbus adoption. Indeed Profibus International predicts that its own Profinet Ethernet implementation will overtake conventional Profibus in terms of the number of nodes installed by 2010, by which time the total number of Profibus and Profinet nodes together could be well in excess of 40 million. Moreover, the extra bandwidth offered by high speed Ethernet opens up the potential for even greater quantities of information to be made available by ever smarter field devices. It also means those same devices can become their own web servers, allowing their information to be accessed directly by the browsers of authorised users throughout, or external to, the enterprise. From which it might be concluded that the integration problems addressed by those hierarchical models of the '80s and '90s have now been largely eliminated. Well, not really. In fact, removing the problem of how to communicate has served only to reveal that the real problem all along was what to communicate and in what form. Fortunately, while those in discrete manufacturing and in the continuous process industries have, until recently, largely ignored this broader issue, the hybrid or batch industries – pharmaceuticals, fine chemicals, food and beverage – have recognised its fundamental importance and come up with some solutions. That's at least partly because legislation, such as the US Food and Drug Administration (FDA)'s 21 CFR part 11 regulations on electronic records keeping in the pharmaceuticals industry, and the new traceability regulations being imposed on the food and beverage industries in both Europe and North America, demand that manufacturers adopt a highly structured approach to defining their manufacturing procedures. They must also manage the data which describes them. Food and beverage manufacturers which have not already faced up to these challenges are likely to find that paper-based records, or the relatively coarse data held in existing ERP systems, prove inadequate when having to respond to demands such as that in the newly enacted US Bioterrorism Act. For that, they must provide details, possibly within as little as eight hours, of what ingredients went into a particular batch, where they came from and where the product subsequently went. Similar 'one up, one down' traceability is also a feature of new European legislation which came into force on January 1 2005. Under the auspices of the US Instrumentation, Systems and Automation Society (ISA), two international standards addressing these issues have been developed and adopted by the International Electrotechnical Commission (IEC), although both are still more widely known by their ISA references. S88 (IEC61512) is primarily concerned with batch control and was initially developed to enable operators of batch processes to define them in a manner that is independent of the specific plant and equipment on which a product is produced in a particular location. That enables the same recipe to be used to produce the same product in different locations and on different combinations of plant and equipment. However, S88 has a much wider significance because it also forms the basis of the second standard, S95 (IEC62264), which addresses the broader issue of enterprise and control system integration. S88 defines a physical model which progressively breaks down the manufacturing enterprise into, successively, sites, areas, process cells, units, equipment modules and control modules. But it focuses specifically on the cell and its sub categories and on batch process cells in particular. S95 adopts the same hierarchical structure for a manufacturer's physical assets and the same terminology, but focuses on the site and the area, only taking an interest in the cell and the unit to the extent that it needs to in order to exchange information between the enterprise and the control system. For example, it might be necessary for a logistic system at the enterprise level to know about the cell in order to be able to schedule longer-term activities in it. What shall we talk about? Both S88 and S95 continue to be extended. Parts 1 and 2 of S95 in effect now define the interfaces between business and manufacturing while the shortly to be released Part 3 extends the standard to define what takes place within manufacturing. S95's focus on the site and the area means that, despite its lineage, it's not specific to batch processes but is equally applicable to continuous production units at one extreme and to discrete production lines at the other. It's for that reason that it is fast becoming the accepted standard for all communication and information exchange between manufacturing systems and enterprise level IT. But it isn't the only contribution the batch community has made to the wider world of manufacturing. S95 defines a framework within which information exchange between manufacturing and the business can be structured – although it still needs a common language which both parties can understand. That requirement has been met by the World Batch Forum (WBF) which has developed a Business to Manufacturing Mark-up Language (B2MML), based on XML. S95 and B2MML provide a rigorous structured framework in an area where fragmented and bespoke solutions have hitherto held sway. Under the widely discredited umbrella term manufacturing execution system (MES), vendors have sought to colonise the no man's land between control and enterprise systems. They've provided a variety of packages to address specific issues ranging from statistical process control (SPC) to downtime analysis, and from asset management to overall equipment effectiveness (OEE). Most, if not all, of these packages generate information that is required at the enterprise level but, in the absence of accepted standards, their integration with ERP has involved custom interfaces, with all the development and ongoing maintenance costs those imply. Now S95 provides the means and the incentive to standardise the integration of the enterprise with the plant and hence to open up that integration to shrink-wrapped solutions. Already major manufacturers such as Nestlé and Proctor & Gamble are applying the principles of S95 within their operations, while software vendors such as Invensys Wonderware, AspenTech, Siemens and GE Fanuc are adapting, or in some cases rewriting, their packages to conform to the standard. Most significant, perhaps, SAP is now supporting the use of S95 to enable integration of MES solutions with the Exchange Infrastructure (XI) integration server element of its NetWeaver service oriented architecture (SOA) based integration platform. That's particularly significant because control and MES software vendors have until recently been concerned that ERP vendors such as SAP might seek to extend their reach into, and even beyond, the MES arena. Adoption of S95 seems to represent a tacit acceptance that the way forward is through co-operation and integration with MES vendors rather than by trying to steal their clothes. One early beneficiary of this new approach is Danish dairy co-operative Arla Foods whose major combined ERP and MES integration project is being held up as a model of what manufacturers could now be achieving. Arla is upgrading its SAP installation, replacing its custom MES systems with shrink-wrapped Wonderware packages, and integrating the whole using S95/B2MML-based mappings between SAP's NetWeaver and Invensys' ArchestrA plant communication architecture. Arla's experience suggests that, after 25 years, S95 has the potential finally to turn the mirage of the wholly integrated manufacturing enterprise into a reality.