CFD Online Logo CFD Online URL
Home > Jobs > Job Record #15718

CFD Jobs Database - Job Record #15718

Job Record #15718
TitleMachine Learning-Enhanced Modeling and Simulation of subsurface
CategoryPhD Studentship
EmployerHeriot-Watt University, UK
LocationUnited Kingdom, Edinburgh
InternationalYes, international applications are welcome
Closure DateMonday, April 15, 2019
Machine Learning-Enhanced Modeling and Simulation of subsurface reservoirs

The School of Energy, Geoscience, Infrastructure and Society at Heriot-Watt
University (HWU), is looking for a PhD candidate to work on an industrially
funded project at the interface of machine learning and subsurface flow
modeling. Multi-phase flow simulation codes tries to model complex physical
phenomena with huge variations in scales within a single reservoir. Adding to
that the uncertainties associated with the spatial subsurface properties (e.g.
permeability and porosity), one is then faced with a statistical uncertainty
quantification problem requiring a large number of simulation runs using an
expensive computer model. In realistic settings, this problem is usually
intractable due to the massive computational costs.

This project builds on two research themes developed at HWU: supervised machine
learning of efficient emulators of physical models [1, 2] and unsupervised
representation learning of spatial models [3] for compact representation of
complex and high-dimensional spaces. The combined use of these two approaches
provides a robust pathway to quantifying uncertainties in large-scale subsurface
flow models. This project will investigate the statistical consistency and
physical plausibility of the machine learning models. The aims is to address
problems associated with learning using limited training data and handling
spatial non-stationary fields.

Main references:
[1] J. Nagoor Kani, Ahmed H. Elsheikh, Reduced-Order Modeling of Subsurface
Multi-phase Flow Models Using Deep Residual Recurrent Neural Networks, Transport
in Porous Media, (2019) 126: 713.
[2] Shing Chan, Ahmed H. Elsheikh, A machine learning approach for efficient
uncertainty quantification using multiscale methods, Journal of Computational
Physics, (2018) Volume 354, Pages 493-511.
[3] Shing Chan, Ahmed H. Elsheikh, "Parametric generation of conditional
geological realizations using generative neural networks",

Essential skills:
-- Master’s degree in computational mathematics, physics or in a relevant
engineering discipline with strong computational skills
-- Programming skills preferably in Python and/or C++
-- Ability to write reports, collate information and present it in a clear and
engaging manner
-- Excellent communication skills

Desirable skills:
-- Machine learning techniques (theory and applications)
-- Background in computational statistics (Spatial Geo-statistics, Bayesian
-- Numerical optimization and nonlinear partial differential equations solvers
(FEM, FVM, etc.)

Funding is available to UK/EU/Overseas candidates. It includes tuition fees and
an appropriate stipend for 3.5 years at the EPSRC recommended level.
Contact Information:
Please mention the CFD Jobs Database, record #15718 when responding to this ad.
NameAhmed Elsheikh
Email ApplicationYes
Record Data:
Last Modified13:07:36, Monday, March 11, 2019

[Tell a Friend About this Job Advertisement]

Go to top Go to top