Towards Autonomous Operations of Hydrocarbon Processing – Part 2

/

In the previous part of the blog (Towards Autonomous Operations of Hydrocarbon Processing – Part 1), I touched upon few points that how Process Industry is evolving with time, the gradual shift from Manual/Semi Automated to Automated and finally Autonomous Operation. In this blog I will discuss Digital Transformation at early stages of Asset Lifecycle, which makes this shift smoother and faster, and how Artificial Intelligence (AI) is utilizing vast data available with the industry and would lead operations toward Autonomous Operations.

Autonomous operation requires the novel combination of several mature and novel technologies including:

  • Rigorous simulation

  • Multivariable predictive control

  • Modular procedure automation

  • Knowledge graphs

  • Machine learning

  • Symbolic AI

Initially these technologies will complement the operations team but, in the future, autonomous operation may be possible.

Figure 1

Figure 1

Most of the successful use cases in the processing industry have been related to asset management. Using IIoT sensors and predictive analytics to forecast required maintenance.  In addition, new decision support systems and augmented reality are being applied to support operators.

All companies have standard operating procedures (SOPs) and there are industry standards and best practices. However, these are often in the form of documents and drawings which are difficult to find and use to their full potential. Supplementing these with enterprise knowledge graphs can dramatically improve the integration of design, operating and maintenance data. This can improve every business process including operations and maintenance.

It is not possible for humans to process the huge amounts of plant data that are collected every minute.  However, machine learning models can be built to identify patterns and anomalies in the data and alert the users.  Knowledge graphs can then be used to propose the likely causes of the anomalies and to recommend corrective actions.

Figure 2

Figure 2

IIoT allows the addition of cheap wireless sensors. When combined with machine learning they can identify maintenance issues months in advance of failure improving reliability and reducing maintenance costs. For example, Sushi sensor on pump motor performs two basic functions (Figure-3): -

  1. Early detection of abnormality (3 months before bearing failure)

  2. Faster decisions from actionable insights

Figure 3

Figure 3

Yokogawa and KBC always work with their clients with the approach of Co-innovation and connecting the dots. Corrosion is a major hazard in the hydrocarbon industry.  First principles and machine learning can be combined to predict corrosion and advise change to operating parameters to minimize it.

This article on Corrosion Prediction with Digital Twin explains that, now a days it is not very hard to save assets from getting corroded. ( https://lnkd.in/gNysZw8)

The ultimate digital twin for asset management combines first principle equipment models such as detailed compressor models with probabilistic machine learning and a rigorous process simulator to forecast not only the future performance of the equipment but also its impact on production.

Figure 4

Figure 4

Machine learning can be combined with dynamic process simulation to provide operators with real time decision support by identifying the current state of the plant and predicting the future state of the plant and identifying potential excursions from the safe operating window well in advance.

Figure 5

Figure 5

Many of our clients are now deploying mobile devices in the field, most are using tablets but there are also industrial heads up displays that can provide hands free augmented reality.  They include audio and video.

Figure 6

Figure 6

Cognitive and autonomous operations will require the use of edge and cloud computing to be able to process vast amounts of data in near real-time coming from IIoT devices using on-line rigorous models and the latest artificial intelligence.

The Cloud also allows the aggregation of data from multiple plants to support value chain optimization and analyze equipment and plant performance and connect the right expertise to trouble shoot problems.

And finally, this allows new outcome-oriented business models.

Figure 7

Figure 7

In summary autonomous operations require the combination of the latest digital technologies and traditional chemical engineering applications to:

  • Manage knowledge

  • Automate and optimize assets

  • Improve safety

  • Reduce emissions

 

*This blog is based on webinar presented by Simon Rogers, Vice President, Digital Solutions at Yokogawa