Capacity of large-scale CO2 storage in North Sea sloping aquifers from numerical simulation

The overall objective of the project has been to gain experience with simulation of carbon dioxide in aquifers and prepare a recommendation for the use of simulation tools for this purpose. Besides articles that shed light on specific themes in the work packages, the emphasis has been on preparing reports for field simulation studies and a note summarizing the recommendations. The effort in the project has been in the work packages for comparison of simulation tools, sensitivity studies and uncertainties arising from the equation of state. The challenges in these activities are quantification of uncertainties as well as recommendations for increased reliability by simulation studies. In the work packages for comparison of simulation tools and sensitivity studies, we have developed tools for fast simulations, efficient sensitivity studies and efficient analysis of large amounts of data. Furthermore, there are developed conversion tools for transforming models between simulation tools and tools for extraction of data for comparison of results from different simulators. Finally, there has been developed a framework for running various simulators on the same reservoir model. Based on a model for one simulator one can use this tool to create models for other simulators. This makes the effort to compare simulation tools significantly easier and time-consuming error search can be avoided. Based on a comprehensive effort from industry partner DEA, the project has received a geological model of a sloping aquifer associated with a gas field. Because of circumstances for the original storage site, the injection now takes place closer to the gas layer than initially intended. There are plans to move parts of the injection to a zone of the field which only contains water, but which may have poorer flow properties. Since one wants to avoid contamination of the gas layer in case it proves viable in the future, it is therefore interesting to study migration time and direction for the carbon dioxide plume, both in the original zone and in the new, planned zone. The project goal has been to compare the sensitivities for different parameters and make assessments with several simulation tools. Therefore, both a compositional and a thermal simulation model for the original zone have been made using conventional simulation tools, and a model for the aqueous zone has been made where in addition a simpler model based on vertical equilibrium may be applied. The latter may, because of its speed, be used to run a large number of simulations, and it is therefore interesting to see to what extent the results of these are comparable with the results of the conventional reference models. The studies have shown that there is large sensitivity to the relationship between permeability in horizontal and vertical direction. Within a range that is feasible, there exists both solutions that provide expansion limited to the original layer, and solutions that provide migration to other layers. Affiliated with this, there has also been developed a statistical model which gives an indication of how likely it is that there is a leak point in the domain where such a migration may take place. A simulation model for Skade is developed for Eclipse and GSC's simulator with vertical equilibrium. The Skade field covers a large area with high potential storage capacity for carbon dioxide. Simulation results have indicated that maximum pressure buildup can be the most limiting factor for the capacity of Skade. Maximum pressure buildup for the caprock of Skade was estimated from literature data and a reported leakoff test in this caprock. Simulations show that limitations in injection quantity may be due to pressure buildup in the shallower parts of the caprock, away from the injector, due to lower initial stress there. The sensitivity of the storage capacity is investigated with respect to six quantities which affect the pressure build. To estimate the contribution to the variation of the solution, statistical methods are used which reduce the computational requirements from weeks to a few days. The runs show that the main contributors to the variance is permeability and formation thickness. The rock compressibility and reservoir boundary conditions also have considerable impact on how much carbon dioxide that can be injected into Skade. In the work package for uncertainties resulting from the equation of state there has been developed a generalized cubic state equation of state which gives a significant improvement of the state description for carbon dioxide while being as fast as conventional cubic equations of state. Furthermore, an explicit function for water solubility in dense phase carbon dioxide is derived. Using this function, the undersaturation of injected carbon dioxide at bottomhole conditions can be determined and the dehydration area in the near-well zone estimated.

The key aspects of successful large-scale CO2 storage are evaluating the capacity of suitable storage sites and understanding the risk of injecting CO2 into large open systems. In the North Sea, many suitable saline aquifers have been identified in the ea stern provinces, ranging from the relatively shallow Utsira and Skade aquifer to the deeper Johansen and Cook Formations, each with enormous volumes of potential storage capacity for CO2. Estimating the practical or realistic storage capacity requires a f undamental understanding of how CO2 migrates and is eventually trapped in these aquifers over long timescales. By investigating the uncertainty in storage capacity and identifying important factors that affect this uncertainty, we can determine the optima l injection strategies to attain the maximum capacity possible for a given storage site. Primary objective: Development of a best practice manual for assessment of CO2 storage in sloping aquifers by numerical simulation. Secondary objectives: 1. Compar e simulation tools on representative data sets for CO2 storage on the Norwegian Shelf. 2. Quantify the uncertainty associated with different equations of state for relevant storage conditions. 3. Quantify the effect of caprock topography for relevant CO2 storage data sets. 4. Make a broad sensitivity study on the effects of boundary conditions and parameterization for CO2 storages. 5. Include vertical cross-flow and pressure-induced geomechanical deformation in the vertical-equilibrium model and implement these effects in the in-house simulator. Challenges: 1. Implementation of robust and reliable simulation code for large-scale and longtime CO2 storage in sloping aquifers. 2. Understanding of impact of uncertainties on storage capacities. 3. Applicatio n of the simulation tools on relevant storage sites on the Norwegian Shelf Application potential: Evaluation of capacity of relevant CO2 storage sites and open software for such evaluations.