Chemical manufacturers depend upon control systems to maximize plant performance and economics. These control systems increasingly use open technologies and digital communications such as fieldbus to supply vast amounts of data. However, all these data provide nothing more than a "data storm" or overload if infrastructure and standards aren't in place to manage and convert them into information.
The International Society of Automation (ISA) through its ISA-95 committee and other organizations such as MESA and MIMOSA are working together to define the tools and methods for transferring data between the various layers of today's integrated enterprise. Figure 1 depicts the ISA-95/Purdue model of these layers except for the business local area network (LAN) proper, which would be Level 5 and above. Fortunately, standards are coming into play that facilitate transfer of data between and through all these layers.
The OPC Foundation continues to develop the common interface for the field levels (1–3) of the model. These layers, because they involve process and real-time control requirements, must respond more quickly than others. OPC, which has been designed for easy connectivity between memory registers of devices, well suits the timelines required for data acquisition. The latest version, UA (Unified Architecture), is platform independent and thus applies to a wider range of products. OPC is rapidly becoming the de facto standard for transferring data in Levels 1–3.
Meanwhile, some variant of SQL likely will emerge as the de facto standard for Level 3 and above. At these levels the time constant often is in the range of minutes, hours or days -- so the definition of "real time" is relative. What's important, however, is seamless bidirectional integration of data. Production orders and business data must flow down to the control system for optimizing manufacturing operations for maximum profit; actual output rates, inventory data and equipment health/status information must go up into the business logistics systems for scheduling work, process orders and minimizing operational expenses in the real time of business transactions.
Now that data have become more open and portable across multiple platforms and systems, the hardware problem -- sensing and gathering these data -- must be addressed. The probable solution will be "Ethernet everywhere" using the Internet Protocol (IP), which now is on version 6 (IPv6). A number of Ethernet appliances and most operating systems already support IPv6.
What makes IPv6 important? Its most consequential feature is a much larger address space (the sets of three numbers between 0 and 225 you enter for LAN addresses) than that of IPv4. Addresses in IPv6 are 128 bits long compared to 32-bit addresses in IPv4. This means you will use a five number address mask such as 255:255:255:192:068. IPv6 supports a total of 2128 (about 3.4×1038) addresses -- or approximately 295 addresses for each of the roughly 6.5 billion people alive in 2006 or 252 addresses for every observable star in the known universe. Simply put, we shouldn't have a problem putting as many sensors as we need anywhere we need them, especially considering we can use the same tricks employed today to extend the capability of IPv4.