Why is this research needed?

The significance of subsurface storage of CO2 in deep saline and hypersaline aquifers in achieving the UK’s net-zero target cannot be overstated. Estimates indicate that saline aquifers beneath the UK Continental Shelf (UKCS) have a huge potential to be utilised for long-term storage of CO2, owing to their capacity, containment, and availability. Maintaining CO2 injection into these storage sites is crucial for the success of carbon capture and storage (CCS) projects. However, injecting dry supercritical CO2 (scCO2) can lead to formation of salt aggregates within pore space, particularly in the vicinity of injection points. Salt precipitates may narrow or even block the pores, resulting in decreased permeability, injectivity loss, and adverse impact on the long-term storage potential.

Capillary pressure gradient due to water evaporation and precipitation of mono and polycrystalline salt crystals with micro porosity can cause suction of brine from farther regions toward the dry-out front. Brine suction promotes further water evaporation and salt precipitation, thereby posing a significant challenge in maintaining reservoir injectivity. This complex pore-scale phenomenon, which depends upon the interplay of thermodynamic conditions, hydrodynamics, and pore structure, herein referred to as THR, is yet to be characterised and incorporated sufficiently in available models for accurately predicting permeability and porosity changes induced by salt formation. To the best of our knowledge, there is no pore-scale experimental data, detailing the spatiotemporal evolution of dissolved salt concentration and velocity fields within the aqueous phase, and capillary pressure map, despite their significant role in controlling the salt precipitation dynamics.

This project aims to address the above gaps by utilising Micro-Particle Image Velocimetry (μPIV) coupled with laser-induced fluoroscopy (LIF) for simultaneous pore-level quantification of brine evaporation rate, capillary suction, and salt precipitation together with characterisation of spatiotemporal evolution of dissolved salt concentration induced by brine evaporation and salt precipitation during scCO2 injection in porous media and evaluate their dependence on the THR factors. The findings will provide fundamental understanding of the synergistic effect of the THR factors on the capillary suction and salt precipitation, providing required experimental data to develop a predictive Machine Learning (ML) based model. The research outcomes assist with proposing mitigation strategies to avoid or minimise salt precipitation during CO2 injection, thereby facilitating the development of predictive tools and mitigation strategies to optimise injection scenarios and prevent/minimise injectivity loss for efficient long-term storage of CO2 in saline and hypersaline aquifers.

What is this research investigating?

The proposed research involves injection of scCO2 into brine-saturated glass micromodels under high-pressure high-temperature (HPHT) conditions, visualised using advanced microfluidic kits at The University of Manchester (UoM) and Stuttgart University (SU), benefiting an extended field of view. The aqueous phase will be laden with fluorescent latex beads and a passive tracer, enabling (i) μPIV analysis to visualise the velocity field and quantify capillary suction and (ii) LIF analysis to visualise the spatiotemporal evolution of the salt concentration field. Two micromodels with distinct heterogeneity patterns will be used under typical HPHT conditions of saline and hypersaline aquifers in North Sea (P: 10-15 MPa, T: 60 °C, salinity: 100-200kppm), and scCO2 will be injected at typical capillary numbers (Ca) of 10-7-10-5. Of particular importance, capillary suction reinforcement and evaporation enhancement due to formation of mono and polycrystalline salt structures will be investigated by comparing the results with the case at which pure deionised water is used instead of brine. The images will be processed using ImageJ, MATLAB Image Processing Toolbox, and PIVlab.

Expected results for each experiment include (1) capillary pressure evolution map (determined by scCO2-brine interface curvature analysis), (2) real-time velocity map within the aqueous phase, (3) real-time differential pressure data (indicative of blockage), (4) the pore space, porosity, and permeability temporal evolution due to salt precipitation, (5) real-time dissolved salt concentration map, (6) salt precipitates map, and (7) pore-level and upscaled evaporation and salt precipitation rates. The duration of each experiment can vary from one day to up to four days, depending on the system parameters, particularly Ca. The visualised velocity, dissolved salt concentration, and salt precipitates maps together with geometrical details of the porous structures and parameter space (PT conditions, Ca, salinity) will be used to train a Convolutional Neural Network (CNN) based model developed in Python to predict dry-out and salt precipitation induced capillary suction in two-dimensional (2D) porous media.

The advanced HPHT microfluidic facilities at SU will be used to conduct preliminary μPIV experiments and receiving training from SU partner. The setup at UoM will be equipped with LIF capabilities to capture dual fields (velocity and concentration) and used to conduct the proposed experiments. The analysis of the image data will be performed by both UoM and SU partners. It is worth mentioning that both teams have several years of experience in conducting pore-scale microfluidic studies with publications in prestigious journals within the context of solute transport and CCS.

What does the research hope to achieve?

The analysis of the experimental results will advance our pore-level understanding of the physical mechanisms governing and salt precipitation in porous media for subsurface CO2 storage, with a focus on the role of capillary suction in exacerbating the salt precipitation around injection wells, and impact of THR factors during dry-out and salt precipitation in porous media. The experimental data also provide a rich databank to benchmark a physical model being developed at UoM using lattice Boltzmann method (LBM) for pore-level simulation of this Multiphysics phenomenon. Development of this physics-based model will be used to provide more detailed information to enhance the accuracy of the CNN-based model, which is not only valuable from scientific standpoint, but also essential in enhancing short and long-term predictive capabilities from operational perspective.

We anticipate several key outcomes from this research. These include publications in esteemed journals such as ES&T, PNAS, and Nature Communications, and presentations at prestigious conferences such as InterPore2025 and GHGT-18. We will also target the JPT to disseminate our research findings to a broader academic and industrial audience, engaging industrial partners and senior stakeholders. Moreover, we will share the experimental databank with the CCS community via open access repositories, providing researchers with the necessary experimental datasets to test and advance their predictive modelling tools. A workshop event will be also organised to (i) share the project outcomes with the CCS community and (ii) provide training to PGRs and ECRs to assist them with their research development. The event will be open to all researchers interested in CCS and will be hosted by UoM. Lastly, we aim to engage young students through STEM events to raise awareness and promote interest in scientific research and its environmental impact.