In part 1 of this blog series, we touched upon some of the potential of IoT for process industries. In this second part, we take a closer look at four fundamental use cases that IoT, combined with Artificial Intelligence (AI) technologies, can help accelerate the digital transformation of the process industries.
Use Case 1: Remote Monitoring and Diagnostics
Monitoring and diagnostics are the foundation use case for IoT in process industries. Data is streamed from sensors directly or ingested from process control systems to detect equipment failures and gas leaks, monitor pipe thickness, temperatures, and erosion in pipelines.
The data historian, can now be located in the cloud, acquire data from the instrumented systems, emergency shutdown systems, and fire and gas systems. The historian stores all relevant event data. All essential plant KPIs and statistics can then be monitored and available online or on the move.
Cloud Data Historian: Image Ref
Data does not have to come only from the process control systems – low-power wireless sensors can be incorporated into a worker’s clothing as a smart wearable when servicing equipment in a hazardous zone. Video analytics on thermal camera images can be used to detect loose contacts and increased friction in equipment without deploying personnel in hazardous zones.
Use Case 2: Operational Integrity
Downtime can cost process plants as much as $20,000 per minute. The Cloud Data Historian enables a vast amount of plant, asset, and operational data to be used in conjunction with advanced algorithms in order to simulate, predict, and prescribe maintenance and increase an asset’s availability, optimize uptime, improve operational performance, and extend their life. This would allow operators to identify high-risk assets and recommend corrective actions before failures occur. By combining operational and business data, an operational integrity platform can:
- Enhance operational safety through early warning in case of anomalies
- Reduce risk of production downtime
Enable real-time condition monitoring of assets.
As an illustration, in the case of predicting vessel and pipeline corrosion, the operating parameters, thickness measurements and ionization models can be combined to identify corrosion risks and predict wall-loss events with recommended interventions for assets at risk.
Operational Integrity through Predictive Machine Intelligence
Another real-life example is the case of a rolling milling, cobbling events are one of the most hazardous events on the milling floor and can lead to serious injuries, downtimes and asset damage. Data fused from Stock data, furnace and rolling mill sensors is fused together to create a prediction model that can provide early warning predictions of cobbling events.
Use Case 3: Process Optimization using Digital Twins
The control of nonlinear systems with real-time parameter uncertainty is one of the hardest problems within the control systems domain. Traditional Advanced Process Control attempts to determine the optimal process conditions based on statistical and linear optimization algorithms, but any non-linearities or model mismatch due to changes in equipment efficiency, environment conditions and abnormal operations may lead to suboptimal settings.
With IoT data, critical processes can be conceptualized as a Cyber-physical using Digital Twins. The Digital Twin can be used in the world of process automation:
- Run in parallel with the real process/equipment and traditional control algorithms, for process tuning
- To allow an operator to train on the digital twin without risk of damage to the real asset
- To identify potential issues, reduce waste, maximize value-added products through simulations
Using Digital Twins, equipment / process design and operation is no longer constrained by the physical constraints of the process or equipment. Companies can use these digital twins to understand their world better, by modelling different scenarios with a goal of making proactive instead of reactive decisions.
Process Tuning using Artificial Neural Networks
In the above example case, we use data from the process to learn a neural network that acts as a digital twin of the non-linear chemical mixing process. A model predictive optimizer then uses the digital twin model as a predictor of future plant output and the history of past control moves to determine the optimal control increments.
Use case 4: Cross Site Analytics
Finally, companies can use data for tracking plant performance not only for a single site but to perform Cross-site Analytics. Thus, one can compare a site’s performance against other sites e.g. a pharma company can use operation data lakes enabled by IoT to:
- Understand and identify differences in process behaviour where a product is manufactured at two different sites
- Remove impact of differences between site operations such as those relative to operational scale
- Focus on process variability
Stay tuned for Part 3 to learn how to address specific issues concerning retrieving data out of process automation silos.