Articles | Volume 16, issue 23
Methods for assessment of models
05 Dec 2023
Methods for assessment of models |  | 05 Dec 2023

An emulation-based approach for interrogating reactive transport models

Angus Fotherby, Harold J. Bradbury, Jennifer L. Druhan, and Alexandra V. Turchyn

Data sets

a-fotherby/GMD_2022: Archival release Angus Fotherby

Model code and software

a-fotherby/Omphalos: Initial release (v0.9.0) Angus Fotherby and Harold Bradbury

a-fotherby/dissertation_xgboost: Initial release (v1.0.0) Angus Fotherby

Short summary
We demonstrate how, given a simulation of fluid and rock interacting, we can emulate the system using machine learning. This means that, for a given initial condition, we can predict the final state, avoiding the simulation step once the model has been trained. We present a workflow for applying this approach to any fluidrock simulation and showcase two applications to different fluidrock simulations. This approach has applications for improving model development and sensitivity analyses.