For many of us, purchasing or receiving the gift of a fitness tracker ushered us into the Internet of Things era. These devices not only collect and report on the number of steps we take each day, but also depending on their sophistication, the quality of our sleep, resting heart rates and much more. For the Industrial Internet of Things (IIoT), it means moving beyond descriptive analytics that aim to answer the question, “What happened?” in order to be forward-looking and answer the question, “What will happen?” Predictive analytics enable us to do that, and move the decision-making process from sense and respond to sense, predict, and act.
The IIoT architecture involves sensors, “Big Data” storage and analytical tools that provide automated analysis and creation of new information, the ability to distribute graphical representations of the information anywhere given the proper levels of security clearance. The ultimate goal is to create applications that produce actionable information and add value.
As Peter outlined in the title of his talk, this journey has been one traveled by process manufacturers and automation suppliers for decades. Examples of these advancements over time include inferential measurements that perform quality predictions of key operating parameters. As technology has advanced, more of these inferential measurement applications have become available. Some example applications include distillation boiling point, product composition and impurities concentration predictions. We highlighted one example in a post, Distillation Column Control Basics – Part 2.
The need to maximize the utilization of plant assets, improve quality & yield, and identify problems before they create abnormal situations is driving the need for these sensors and multivariate fault detection applications. Through fault-detection diagnostics, problems with measurement devices (drift, offset, frozen value, etc.), process conditions, and equipment status can be identified in time to take action. Other examples are instrument and signal validation for online analyzers and laboratory-entered data.
Today, many of these solutions to interpret this information are complex and engineering intensive as well as difficult to implement and maintain. The key is to expand on-line analytics solutions that are easier to implement and maintain. Some examples of where field trials and pilots that IIoT technologies were being put into practice include:
- Salt buildup on evaporators
- Flare gas recovery and flare system management
- Mining SAG mill optimization
- Inferred measurements in divided wall distillation columns
- Boiler feedwater quality
- Wellhead integrity monitoring
Many of these applications, such as flare management, offer big returns in downtime avoidance, regulatory compliance and operations that are more reliable. The scope of solutions extends beyond monitoring and control to include energy management, reliability and safety and the ability to connect collaborating experts no matter where they are in the world.