Bayesian Optimization for Field Scale Geological Carbon Sequestration

We present a framework of the application of Bayesian Optimization (BO) to well management in geological carbon sequestration. The coupled compositional flow and poroelasticity simulator, IPARS, is utilized to accurately capture the underlying physical processes during CO2 sequestration. IPARS is coupled to IBM Bayesian Optimization (IBO) for parallel optimizations of CO2 injection strategies during field-scale CO2 sequestration. Bayesian optimization builds a probabilistic surrogate for the objective function using a Bayesian machine learning algorithm, Gaussian process regression, and then uses an acquisition function that leverages the uncertainty in the surrogate to decide where to sample. IBO addresses the three weak points of the standard BO in that it supports parallel (batch) executions, scales better for high-dimensional problems, and is more robust to initializations. We demonstrate these algorithmic merits by an application to the optimization of the CO2 injection schedule in the Cranfield site using field data. The performance is benchmarked with genetic algorithm (GA) and covariance matrix adaptation evolution strategy (CMA-ES). Results show that IBO achieves competitive objective function value with over 60% less number of forward model evaluations. Furthermore, the Bayesian framework that BO builds upon allows uncertainty quantification and naturally extends to optimization under uncertainty.

Potential Pb+2 mobilization, transport, and sequestration in shallow aquifers impacted by multiphase CO2 leakage: a natural analogue study from the Virgin River Basin in Southwest Utah

Geological carbon sequestration (GCS) is necessary to help meet emissions reduction goals, but groundwater contamination may occur if CO2 and/or brine were to leak out of deep storage formations into the shallow subsurface. For this study, a natural analogue was investigated: in the Virgin River Basin of southwest Utah, water with moderate salinity and high CO2 concentrations is leaking upward into shallow aquifers that contain heavy metal-bearing concretions. The aquifer system is comprised of the Navajo and Kayenta formations, which are pervasive across southern Utah and have been considered as a potential GCS injection unit where they are sufficiently deep. Numerical models of the site were constructed based on measured water chemistry and head distributions from previous studies. Simulations were used to improve understanding of the rate and distribution of the upwelling flow into the aquifers, and to assess the reactive transport processes that may occur if the upwelling fluids were to interact with a zone of iron oxide and other heavy metals, representing the concretions that are common in the area. Various mineralogies were tested, including one in which Pb+2 was adsorbed onto ferrihydrite, and another in which it was bound within a solid mixture of litharge (PbO) and hematite (Fe2O3). Results indicate that metal mobilization depends strongly on the source zone composition and that Pb+2 transport can be naturally attenuated by gas phase formation and carbonate mineral precipitation. These findings could be used to improve risk assessment and mitigation strategies at geological carbon sequestration sites.Thematic collection: This article is part of the Geoscience for CO2 storage collection available at: https://www.lyellcollection.org/cc/geoscience-for-co2-storage

Geological Carbon Sequestration by Reactive Infiltration Instability

We study carbon capture and sequestration (CCS) over time scales of 2000 years by implementing a numerical model of reactive infiltration instability caused by reactive porous flow. Our model focuses on the mineralization of CO2 dissolved in the pore water—the geological carbon sequestration phase of a CCS operation—starting 10–100 years after the injection of CO2 in the subsurface. We test the influence of three parameters: porosity, mass fraction of the Ca-rich feldspar mineral anorthite in the solid, and the chemical reaction rate, on the mode of fluid flow and efficiency of CaCO3 precipitation during geological carbon sequestration. We demonstrate that the mode of porous flow switches from propagation of a planar front at low porosities to propagation of channels at porosities exceeding 10%. The channels develop earlier for more porous aquifers. Both high anorthite mass fraction in the solid phase and high reaction rates aid greater amounts of carbonate precipitation, with the reaction rate exerting the stronger influence of the two. Our calculations indicate that an aquifer with dimensions 500 m × 2 km × 2 km can sequester over 350 Mt solid CaCO3 after 2000 years. To precipitate 50 Mt CaCO3 after 2000 years in this aquifer, we suggest selecting a target aquifer with more than 10 wt% of reactive minerals. We recommend that the aquifer porosity, abundance of reactive aluminosilicate minerals such as anorthite, and reaction rates are taken into consideration while selecting future CCS sites.