H045
Domain-Aware Machine Learning for Subsurface Applications

Session ID#: 85474

Session Description:
Machine learning (ML) has become a widely accepted tool in geosciences in general and subsurface transport specifically. In the same, time there are several recognized challenges in ML that must be addressed to improve robustness and reliability of ML methods and make them “domain aware”. These challenges include: 1) large, but  sparse data sets, 2) indirect and highly-uncertain measurements (e.g., permeability measurements), 3) uncertainty quantification in ML predictions, and 4) incorporating physics-based knowledge. These challenges are compounded by the extreme heterogeneity of subsurface environment and computationally expensive physical models with partially-known physics. This mini-symposium will explore applications of ML to subsurface flow and transport that addresses the above challenges and/or focus on Big Data and high-performance computing.
Index Terms:

1805 Computational hydrology [HYDROLOGY]
1829 Groundwater hydrology [HYDROLOGY]
1847 Modeling [HYDROLOGY]
1873 Uncertainty assessment [HYDROLOGY]

Primary Convener:  Alexandre M Tartakovsky, Pacific Northwest National Laboratory, Richland, WA, United States
Conveners:  Hari Selvi Viswanathan1, Bailian Chen1 and Jianchun Xu2, (1)Los Alamos National Laboratory, Los Alamos, NM, United States(2)China University of Petroleum (East China), Department of Petroleum Engineering, Qingdao, China
Primary Liaison:  Zhenyu Guo, Occidental Petroleum Corporation, Houston, United States
 
Machine learning emulators for reactive mixing (496798)
Bulbul Ahmmed1, Maruti Mudunuru2, Satish Karra2, Velimir V Vesselinov2 and Scott C James3, (1)Baylor University, Waco, TX, United States, (2)Los Alamos National Laboratory, Los Alamos, NM, United States, (3)Baylor University, Geosciences and Mechanical Engineering, Waco, TX, United States
 
A composite neural network that learns from multi-fidelity data: Application to function approximation and inverse PDE problems (497791)
Xuhui Meng, Brown University, Providence, RI, United States and George Em Karniadakis, The Division of Applied Mathematics, Brown University, Providence, RI, United States
 
Modeling flow and transport in fracture networks using machine learning and graphs (Invited) (506843)
Daniel O'Malley1, Satish Karra2, Jeffrey Hyman3, Hari Selvi Viswanathan2 and Gowri Srinivasan2, (1)Los Alamos National Laboratory, Computational Earth Sciences, Los Alamos, NM, United States, (2)Los Alamos National Laboratory, Los Alamos, NM, United States, (3)Los Alamos National Laboratory, Los Alamos, AZ, United States
 
Physics Informed Machine Learning for the Buckley-Leverett Problem. Application to Flow and Transport in Porous Media. (512626)
Cedric Fraces, Stanford, CA, United States, Olga Fuks, Stanford Earth Sciences, Stanford, United States and Hamdi Tchelepi, Stanford University, Energy Resources Engineering, Stanford, CA, United States
 
Multi-resolution neural networks for subsurface exploration with sparse labels. (522390)
Jennifer Fohring, Computational Geosciences Inc, Vancouver, BC, Canada and Eldad Haber, University of British Columbia, Earth, Ocean, Atmospheric Science, Vancouver, BC, Canada; Computational Geosciences Inc., Vancouver, BC, Canada
 
Developing a Machine Learning Framework to Reduce the Computational Burden of Reactive-Transport Simulations (526334)
Christopher A Holle1, Bulbul Ahmmed1 and Scott C James2, (1)Baylor University, Waco, TX, United States, (2)Baylor University, Geosciences and Mechanical Engineering, Waco, TX, United States
 
Gaussian process machine learning for groundwater system characterisation and modelling (562018)
Tao Cui1, Dan Pagendam2, Luk JM Peeters3, Adam J Siade4, Mat Gilfedder1 and Sreekanth Janardhanan1, (1)CSIRO, Land and Water, Brisbane, QLD, Australia, (2)CSIRO, Data 61, Brisbane, QLD, Australia, (3)CSIRO, Land and Water, Adelaide, SA, Australia, (4)University of Western Australia, Crawley, WA, Australia
 
Physics-Informed Deep Neural Networks for Multiphysics Data Assimilation in Subsurface Transport Problems (563617)
Qizhi He, Guzel Tartakovsky, David A Barajas-Solano and Alexandre M Tartakovsky, Pacific Northwest National Laboratory, Richland, WA, United States
 
Machine Learning for Estimating Methane Distribution in 2-D Nano-pores (570420)
Hao Wu1, Animesh Agarwal2, Javier E. Santos3, Mohamed Mehana4, Hari Selvi Viswanathan5 and Nicolas Lubbers2, (1)Blacksburg, VA, United States, (2)Los Alamos National Lab, Los Alamos, United States, (3)University of Texas at Austin, PGE, Austin, TX, United States, (4)University of Oklahoma Norman Campus, Norman, OK, United States, (5)Los Alamos National Laboratory, Los Alamos, NM, United States
 
Active-learning for upscaling nano-confinement effects (571072)
Javier E. Santos1, Nicolas Lubbers2, Mohamed Mehana3, Yu Chen4, Masa Prodanovic5, Michael Pyrcz6, Matt Andrew7, Qinjun Kang8 and Hari Selvi Viswanathan4, (1)University of Texas at Austin, PGE, Austin, TX, United States, (2)Los Alamos National Lab, Los Alamos, United States, (3)University of Oklahoma Norman Campus, Norman, OK, United States, (4)Los Alamos National Laboratory, Los Alamos, NM, United States, (5)The University of Texas at Austin, Hildebrand Department of Petroleum and Geosystems Engineering, Austin, TX, United States, (6)University of Texas at Austin, Austin, TX, United States, (7)Zeiss (formally Xradia, Inc.), Pleasanton, CA, United States, (8)Los Alamos National Lab, Los Alamos, NM, United States
 
Determination of Infill Well Placement at a Petroleum Reservoir Using a Data-driven Multi-modal Convolutional Neural Network with Sequential Training (574250)
Baehyun MIN1, Min-gon Chu2, Seoyoon Kwon3, Gayoung Park3 and Nguyen Xuan Huy4, (1)Ewha Womans University, Department of Climate and Energy Systems Engineering, Seoul, Korea, Republic of (South), (2)Ewha Womans University, Center for Climate/Environment Change Prediction Research, Seoul, South Korea, (3)Ewha Womans University, Department of Climate and Energy Systems Engineering, Seoul, South Korea, (4)Ho Chi Minh City University of Technology, Faculty of Geology and Petroleum Engineering, Ho Chi Minh City, Vietnam
 
Optimal Structure of a Data-driven Multi-modal Convolutional Neural Network for Determination of Infill Well Placement at a Petroleum Reservoir (575123)
Seoyoon Kwon1, Min-gon Chu2, Gayoung Park1, Baehyun MIN3 and Jungtek Lim4, (1)Ewha Womans University, Department of Climate and Energy Systems Engineering, Seoul, South Korea, (2)Ewha Womans University, Center for Climate/Environment Change Prediction Research, Seoul, South Korea, (3)Ewha Womans University, Department of Climate and Energy Systems Engineering, Seoul, Korea, Republic of (South), (4)SmartMind, Seoul, South Korea
 
Model inversion via conditioned Karhunen-Loève expansions (621500)
David A Barajas-Solano and Alexandre M Tartakovsky, Pacific Northwest National Laboratory, Richland, WA, United States
 
Parameter Estimation using Gaussian Process Regression and Conditional Polynomial Chaos Surrogate Model (623132)
Alexandre M Tartakovsky and Jing Li, Pacific Northwest National Laboratory, Richland, WA, United States
See more of: Hydrology
<< Previous Session Proposal | Next Session Proposal >>