Research Associate On Enabling Co2 Storage Using Artificial Intelligence

Contract: Full-time (35 hours per week), Fixed Term for 14 months  The Lyell Centre at Heriot-Watt university (HWU), Edinburgh, Scotland has an opening for one PDRA position to work on the project ECO-AI (Enabling CO2 storage using Artificial Intelligence techniques). This post is funded through EPSRC. Further details about ECO-AI are available at the project webpage https://ai4netzero.github.io/ecoai_project/  Detailed Description  The successful candidate is expected to develop cutting edge deep learning models for storage site characterization to determine optimal CO2 injection and leakage risk assessment. Two aspects are of special interests: (a) Development of a framework for fast capacity estimation using AI-solver including geological uncertainties across scales, and (b) development of optimal CO2 injection and monitoring strategies using model-based reinforced learning algorithms.   In addition, the successful candidates will contribute to a wide range of AI applications in subsurface flow modelling including (a) stochastic generation of porous media realizations using deep generative models (b) deep learning based property prediction using various architectures (c) Deep learning based proxy modelling with physics based losses and built-in model constrains (e) Effective optimization techniques for physics constrained implicit neural models (f) Efficient coupling of deep learning models to numerical solvers for hybrid CO2 flow modelling. The developed machine learning techniques will be open-sourced and be validated across a wide range of applications and on experimental data and direct numerical simulations generated by the project team.  The successful candidates will be part of a large multidisciplinary research project on maximising CO2 storage in deep geological formations. The candidates will benefit from interactions with the project team across Heriot-Watt university and Imperial College London   Institute of GeoEnergy Engineering (IGE) at Heriot-Watt University  Lyell Centre at Heriot-Watt University  Institute of Mechanical, Process and Energy Engineering (IMPEE) at Heriot-Watt University  School of Mathematical and Computer Sciences (MACS) at Heriot-Watt University  Department of Earth Science and Engineering (ESE) at Imperial College London  Department of Chemical Engineering at Imperial College London  Key Duties & Responsibilities   The successful candidate will be expected to undertake the following:  Develop clear scenarios relevant for CO2 storage capacity and leakage risk assessment where AI can support uncertainty prediction and re-risking  Develop deep learning models for storage capacity estimation and/or leakage risk assessment from the metre to the reservoir scale.  Disseminate research results in peer reviewed journals and interdisciplinary conferences.  Publish open-source code repositories demonstrating all developed techniques and associated computational notebooks, blogs and presentation materials.  Organize and lead Hackathons as a part of ECO-AI project activities.   Participate in regular project meetings with team members and project sponsors.  Education, Qualifications and Experience  Qualifications  Essential Criteria   A PhD degree in subsurface geosciences or engineering, with a strong focus on computational science & engineering, applied mathematics, physics or in a related computational field.  A solid understanding of subsurface carbon storage in porous reservoirs  Prior experience in developing deep learning models using open-source libraries (e.g., pytorch).  Prior experience in computational fluid dynamics or pore-network models using open-source software packages (e.g., MRST).  Strong track record of publications in high impact scientific journals.  Working experience in modern software development techniques (version control, continuous integration, software testing, etc).  Excellent verbal and written communication skills, and ability to write professional reports.  How to Apply  Applications can be submitted until midnight on the 31st of January 2024.