Back in the 1980s, as a junior process engineer, I spent a great part of my first job working with DCS and scada systems. At the time, I was working on a synfuels plant that made use of a Honeywell TDC3000 DCS. Most of the field instruments were connected to the proprietary Honeywell LCN (local control network) and process engineers were not allowed to touch that part of the system. The DCS was connected through an interface to a local scada system, called Cygnus (later Adriot), running on a bulky PDP-11 minicomputer that required its own room. This was my playground. The 10 Mb removable hard disks were also bulky devices coming in a special foam-lined case, which at the time was our version of Big Data. This system was invaluable for monitoring and optimising the plant and a process engineer’s dream.
These early experiences with the PDP-11, using a real-time language from ICI called RTL/2, taught me a great deal about analysing large volumes of plant data. My view at the time was that fully automated manufacturing was ultimately possible, but only if instrumentation data was first consolidated in a DCS/scada system, and only then processed by external systems.
I soon found that this hierarchal layered model of integrating process and business systems, was shared by most of the instrumentation and control community, and also by the big software vendors providing ERP. Later, as I learned more about business processes, I realised that in practice, other federated data integration models were viable and there are many (and often better) ways of configuring data flows relating to a manufacturing process.
A new generation of sensors
Recently, I was interested to learn of several case studies coming out of the oil and gas industry where the IIoT is being used to make significant improvements to energy efficiency and equipment reliability on plants. These stories made me curious, after all, back in the 1980s we had implemented some of the early versions of what evolved into very powerful platforms for process control and automation. So why is it that nearly three decades later there still seems to be so such room for process improvement, particularly in a mature industry like oil and gas?
On closer investigation I learned that many of these recent success stories involved IIoT sensors that completely bypass the DCS and scada systems. These sensors were never included in the original design. To solve specific problems new temporary sensors were located in places that previously had no instrumentation; such as the interior of furnaces, or monitors attached to the surface of moving equipment. In other situations additional performance data is now obtained from existing actuators; information that was previously never used by the DCS. This data stream accessed using a protocol called WirelessHART, connects wirelessly to an IoT gateway, which in turn streams the data to a third-party, cloud-service platform. The result of all these different techniques augments existing plant data with additional data streams accessed from the cloud. This then allows engineers to do a more powerful analysis of equipment performance than was ever possible using the existing scada data.
It occurs to me that the engineers and designers of industrial scale plants have always been focused on getting the basics right, such as traditional safety and operability, but hardly focus at all on optimisation. Optimisation requires a different mind-set, different instruments, different data analysis and different modelling systems. The large expensive DCS/PLC and scada systems implemented in a typical project are not always suited for optimisation, the full requirement for which only manifests itself once the plant has been commissioned.
Cloud-based toolkits
The good news for engineers today is that no longer is it necessary to reconfigure and change already complex proprietary DCS systems to solve every operational problem. There are already a number of cloud-based ‘toolkits for innovation’ that will allow them to build specific diagnostic solutions that use state-of-the-art machine learning, modelling and advanced visualisation capabilities, without disrupting any existing critical control system.
The platforms on which these specialised IIoT solutions are built are evolving fast. For example, recently SAP positioned its Leonardo IoT cloud platform as a ‘digital innovation system’. Leonardo promises to enable exactly the optimisation scenario I have described. The solutions can be small and specialised, such as predictive maintenance on a specific machine, or much more complex such as energy optimisation or logistics management across an entire plant or supply chain.
Leonardo is a ‘container’ of complex interrelated technologies that are still evolving; it is a work in progress. A significant element of the platform is advanced analytics and machine learning. For those manufacturing companies that have a long-term vision, SAP’s cloud ecosystem platform is worth a closer look. There are other alternatives as well, as cloud vendors bring their own competing technologies to market. Ecosystems of partners and developers are also converging around industry standards and starting to package real-world solutions as templates. Owing to the relative newness of the technology, expect some vendor churn, fallout and consolidation ahead, but this should not prevent you from getting started.
We have come a long way from the DCS systems of the 1980s, which still serve a useful purpose. However, the pressing need for new rapid innovation has in many instances meant that these older proprietary approaches are no longer suited to the changing needs of manufacturers. There are now many simpler, more elegant and quicker to deploy tools for optimising plant efficiency. It is a wonderful time as vendors and manufacturers innovate together creating new opportunities to push efficiency and productivity boundaries even further.
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version