Research

Our research

GeoCquest research has focused on enhancing our understanding of the impact of heterogeneity in reservoir rocks on the storage of CO2, and the extent to which we can use heterogeneity to enhance the value of a commercial-scale geological storage site.

GeoCquest has used a combined approach of core and outcrop studies, analogue laboratory experiments, detailed numerical simulations, and reduced mathematical models to provide confidence in predictions of trapping, which bridges the critical scales between core and field or commercial scale.

In order to take advantage of the beneficial effects of heterogeneity, a new set of storage strategies and injection schemes are needed. A series of coordinated process studies to investigate this are now progressing, using experimental, analytical and numeric approaches.

The current focus of GeoCquest research is:

Optimization of injection strategies in heterogeneous reservoirs that increase pore scale utilization and

Accelerating trapping in large scale CO2 storage projects in those heterogeneous reservoirs

Why heterogeneity?

In contrast to reservoir engineering for oil and gas recovery, where geological heterogeneity is an obstacle to overcome, it can beneficially increase storage efficiency, solubility and residual gas trapping of CO2. GeoCquest research into the heterogeneity or variability of reservoir rocks, has provided new understanding of how even small-scale changes in sediments can profoundly influence CO2 trapping rates.

Heterogenity core

Development of reservoir engineering strategies

To take advantage of the beneficial effects of geological heterogeneity on storage efficiency, solubility and residual gas trapping of CO2, a new set of storage strategies and injection schemes are needed. The approach we are taking in GeoCquest builds on the scientific insights regarding the effects of geological heterogeneity on flow and trapping of CO2 and the advanced reservoir simulation techniques we have developed.

Lead Investigators – Sally Benson & Hamdi Tchelepi at Stanford University

Heterogeneous outcrop

Three different strategies for enhanced reservoir engineering are being pursued:

Optimization of injection wells (including long reach horizontal wells), injection rates (e.g. time varying flow rates) and locations to maximize pore-space utilization and trapping in reservoirs with different degrees of heterogeneity;

Co-injection of brine as a method to increase pore-volume utilization and accelerate trapping. Various injection schemes will be evaluated, including WAG (water-alternating-gas), co-injection, and injection in others well (e.g. in horizontal wells above the CO2 injectors);

Investigating novel schemes for increasing pore volume utilization and trapping, for example foam injection to create local flow barriers, creation of barriers with silica gels, microbubbles, and biobarriers.

Site selection strategies to assess and increase pore scale utilization

Geological models are typically based on the distribution of facies, with each facies having a statistical probability distribution of rock and flow properties. Lithological boundaries within each facies are controlled by grid cell boundaries, where the number of grid cells of a model is limited by what is deemed to be computationally practical. But this approach underestimates the abundance of lithological boundaries and respective permeability off-sets. Consequently, the CO2 plume geometry and pressure distribution are poorly predicted. This translates into uncertainties when estimating the CO2 storage capacity, injectivity and CO2 trapping by different mechanisms. GeoCquest is applying a machine-based learning approach to overcome this and determine effective rock type categorical values of upscaled reservoir grid cells. The aim is to develop a generalized workflow for upscaling categorical properties, such as rock types, in reservoir models. Upscaling of rock properties is an efficient method for building computationally feasible reservoir models while incorporating sub-grid scale heterogeneity. A new workflow for deriving effective categorical parameters, specifically rock types, for each grid cell of an upscaled reservoir model has been developed utilising a k-means clustering algorithm and outcomes are validated using multiphase flow simulations.

Lithological heterogeneity is believed to lead to greater pore space utilization efficiency and greater CO2 storage capacity, but it is also known to reduce injectivity due to pressure build up. A reservoir quality screening tool to estimate injectivity and pore space utilization efficiency by accounting for reservoir geometry (height) and lithological heterogeneity is being developed by GeoCquest.

GeoCquest is also assessing the feasibility of enhanced injectivity through accelerated mineral dissolution within the CO2 plume. Multi-phase core flood experiments are planned, where CO2 enriched with water and its reagents will be flushed through core over time. The accumulation rate of the reagent in the water will be measured by means of its concentration at the end of the flow-through period. The dissolution rate of (clay) minerals induced by the reagent and the respective change in permeability will be measured separately in long-term static experiments.

Lead Investigator – Ralf Haese at University of Melbourne

Upscaling rock types

Upscaling rock types from a resolution of 1m x 5cm (characterized with 15 lithological units) to 14m x 3m (characterized with 7 composite lithological units) (Figures from Mishra et al. (2020) and Mishra et al. (submitted).

Generalization of a full physics model for advanced reservoir engineering

To improve speed and accuracy of the simulation of highly heterogeneous geomodels, GeoCquest has developed an asynchronous time-stepping approach that focuses computational effort on rapidly evolving parts of models, rather than solving each cell in every timestep. This approach and an improved field-scale simulator is applied across the GeoCquest research and will be enabled by:

  • Moving from proof-of-concept curve-fits at the cm (core-plug) scale to the parametric grid-block scale (1-100m-scale); and
  • Developing and demonstrating field-scale simulation with lower-dimensional representations of high-permeability streaks using a reduced-order approach for them.

Lead Investigator – Stephan Matthai at University of Melbourne

Full Phy model

Reduced physics models for advanced reservoir engineering

GeoCquest research is focused on the development of fast, efficient simulators of CO2 migration in heterogeneous reservoirs. This will be a prerequisite for the design of injection strategies to maximize storage security through enhanced trapping rates, and for de-risking injection at future storage sites. The approach taken here builds on previous work in which fast, reservoir scale simulators were developed that exploit the large aspect ratio (vertically confined and laterally extensive) of many CO2 storage reservoirs.  In this archetypal geometry, the flow of CO2 is primarily in the direction of the dominant bedding surfaces. A reduced model of the CO2 plume can be constructed by locally modelling the flux through the CO2 plume via appropriate vertical averages of the flow properties.  While this flow is often driven by buoyancy, it is principally horizontal. Consequently, simplified, depth-averaged models of the pressure propagation and flux (of CO2 and water) can be derived.  Incorporating sub-grid scale heterogeneities within these depth-averaged models provides the groundwork for future GeoCquest modelling.

GeoCquest research is carrying out a series of reduced models and analogue laboratory experiments to provide a fast, efficient framework for assessing storage risks and optimizing trapping potential.

Lead Investigator – Jerome Neufeld at University of Cambridge

    The focus is on four key settings;

    the quantification of uncertainty during site permitting, storage and post-injection site closure,

    the optimization of injection strategies to enhance trapping rates,

    the benchmarking of flow simulations against existing field data sets,

    the benchmarking of parameterizations of trapping rates against analogue laboratory experiments. In each of these four key settings we also assess the simulation speed that can be achieved with the reduced-order approaches.