Grant: Canada Research Chairs

This program intends to leverage advances in data analytics and machine learning to improve the modelling and analysis of subsurface flow systems that are widely encountered in many energy extraction and fluid storage applications. Accurate performance prediction and reliable uncertainty assessment are crucial elements in the optimal design of these processes. However, describing, modelling, predicting, and analyzing the response and performance of these systems is challenging, as it involves integrating many measurements and data of diverse types, scales, and sources while capturing the complex physics governing these processes. Dr. Leung aims to utilize her distinctive blend of expertise in numerical simulation of the subsurface flow processes, modelling of uncertain heterogeneous reservoir conditions and properties, engineering operations of subsurface geo-energy systems, and applied data analytics in this program. She will develop innovative ways for customizing data-driven models to support different modelling, data analysis, and design activities. She will lead a team of students and trainees to develop tailored data-driven models in these specific themes: (1) Providing more efficient workflows to predict system performance across multiple scales; (2) Developing adaptive models for improved real-time updating of model uncertainties and optimizing designs; (3) Tracking of environmental, social, and governance (ESG) performance in subsurface flow processes using surveillance data from multiple sources. A variety of datasets, including synthetic (simulated) and actual (field measurements from private and public data sources), will be used. Advances in data science for integrating multi-faceted data sets with varying characteristics, including dimensionality, resolution, sparsity, and uncertainty, will be leveraged. Domain expertise will be combined to extract features, identify patterns, and formulate models compatible with the complexity and limitation of the data and the underlying physics. Models will be used to make predictions or propose optimal operating strategies. Industrial partners from both private and public sectors will be engaged. The outcomes would improve the ability to forecast and design these subsurface processes for practitioners and researchers, but they could also inform additional monitoring and reporting requirements, while advancing the ongoing movement of data democratization and evidence-based policy-making.