Digital transformation of the refining and chemical industry is now playing a crucial role in energy efficiency improvement and associated GHG emission reduction. Digitalisation in view of remote performance monitoring is a major aspect of today’s unit operation management and optimisation.

The main advantages of digital tools are first to reduce the delay between the client’s request for unit monitoring or troubleshooting and the technology provider’s answers, and to open up access to customised unit optimisation tools.

Implementing Software as a Service, accessible to our customers 24/365 and fed by process and lab data through an automatic and near real-time transfer, has been the first step of our digital transformation and paved the way for a new paradigm for technical services.

For such fast-track implementation, supported by licensor and catalyst experts, operator involvement to automate data transfer while addressing all potential concerns related to cybersecurity, data ownership, and lifecycle is minimum.

For operators, access to appropriate alerts, data analyses, and optimisation tools are success factors in fast decision-making in increasing overall unit profitability. Typical examples of client expectations that can be addressed through remote performance monitoring are:

• unit performance prediction while changing operating conditions;
• comparing actual performances with normalised ones;
• catalyst end-of-cycle prediction to anticipate turnaround or get the most from the catalyst;
• or even proposing continuously the best set of operating conditions for a given set of performance targets, such as yields optimisation, utilities reduction, and cycle length extension.

Reliability of the projections is ensured through advanced data analysis algorithms with a preliminary data reconciliation step using, for instance, principal component analysis (PCA) and robust regression methodologies such as partial least squares (PLS) or Theil Sen estimator.

Other available tools are very accurate and continuously update shift vector generation, which can become a major help for the planning department.The generation of synthetic data thanks to machine learning is another example of either densifying laboratory data or generating new models to foresee a product’s properties based on existing process data and unit performances and acting as an on-line analyser. Improving operation survey quality at no extra cost becomes a key adoption factor.