Petroinformatics: a digital platform for optimizing oil refining

Learn how artificial intelligence can help make predictions that provide insights into more energy-efficient and cost-effective processes and technologies for bitumen upgrading and partial upgrading, and petroleum refining to reduce life-cycle greenhouse gas emissions associated with the production of transportation fuels.

Project objectives

This project will create a digital platform built on artificial intelligence (AI) algorithms that integrates pilot plant and published commercial data, petroleomics (the chemical characterization of petroleum) and engineering process models.

The platform will support technology assessment and decision-making for bitumen upgraders and petroleum refiners under variable scenarios of feedstock composition, process scheme and configuration, operating conditions, and product slates.

Expected results

The petroinformatics platform will be used to develop predictive models and tools for selected physical and chemical properties of conventional and unconventional oil streams, based on CanmetENERGY data and published data. It will also develop AI and machine learning (ML) models using one of CanmetENERGY‘s hydroprocessing pilot plant units and its associated data to serve as a test case for applications to other pilot plant units at CanmetENERGY.

Digital and artificial intelligence techniques

Machine learning: principal component analysis (PCA), partial least square regression (PLS), support vector machines (SVM), neural networks (NN), decision trees (DT)
Optimization techniques: simulating annealing (SA), genetic algorithms (GA), particle swarm optimization (PSO)
Data-driven models based on molecular-level kinetic models