Rick Kephart rounds out the 2019 Ovation Users’ Group conference with a look into the future around architectures and algorithms. Rick opened with a discussion of Ovation Droplets. It is a lightweight controller that’s part of the 3.7 release. Any Compact Controller can be configured as a Droplet in the Dev Studio configuration software. These compact controllers, configured as standalone, can be adopted as Droplets into an Ovation system.
Some capabilities include that they do not consume a drop on the Ovation network and uses wide area network (WAN) or datalink communication. All points can be historized. They do not communicate alarms in the standard Ovation Alarm Screen but on a private alarm screen to better manage widely distributed applications. Points in a droplet do not have to be unique from one another to simplify configuration. This helps reduce complexity and support scalability over time. One area of engineering effort for a coming release is to add secure VPN from a Droplet back to the Ovation system.
The next area Rick discussed was application-specific controllers geared to high-repeatability fixed applications such as wind turbine controllers or for substation control and load shedding. These controllers would use standard hardware and I/O with application specific software.
Next topic area was in simulation. The Ovation Digital Twin model library is being enhanced to support large-scale hydraulic networks for water and wastewater processing applications. Live Digital Twin is another area of focus. This synchronizes the real control system settings and control logic with the live simulation. The opportunity for live digital twin technology is that the models can provide measurements like temperature gradients that are not measurable in the actual process. This can help identify failure conditions well before they manifest themselves in the live process.
The next topic area was about Intelligent Machines. Intelligence recognizes situations, have historical knowledge about how to react, and use machinery to interpret or react. Humans use this assimilation of diverse facts in everyday activities—to date machines do not do this. For machine intelligence, it must be able to draw from wells of knowledge in control technology, performance monitoring, simulation technology, data analytics, and architecture. However, from work done by different industries to date, the more complex and sophisticated the control systems the more humans need to be part of the control process (think automated automobile driving crashes.)
Adding the commonsense notion into semi-autonomous operations is at the heart of the Intelligence Framework Architecture which includes sequence advisor, analytics & machine learning, process optimizer, performance advisor, advanced calculations and model validation. These feed a programmable expert system and configurable data sources to deliver monitoring & diagnostics, process control, and triggered workflows.
Local automated pattern recognition can run at multiple levels—workstation, controller, and I/O modules using high-resolution data. Preconfigured APR and prognostics available for gas turbines, coal plants, hydropower, nuclear and renewable sources of energy.
I’ll save some of the longer-term research areas that Rick shared for you to hear for yourself next year at the 2020 Ovation Users’ Group conference.
