Carbon Capture/CO2 Mitigation: Evaluation of the joint effect of uncertain parameters in CO2 storage in the Sleipner project
The Sleipner carbon capture and storage (CCS) project in the North Sea off the coast of Norway was the first storage project on a commercial scale.
The Sleipner carbon capture and storage (CCS) project in the North Sea off the coast of Norway was the first storage project on a commercial scale. It is necessary to quantify any uncertainties in the storage site to manage the risk of leakage through all stages of the storage process. Also, a reliable estimation of storage capacity and plume dynamic behavior is required to make better decisions.
Several researchers have studied the Sleipner model to better understand the inherent flow physics, with various sources of uncertainty in the geological model and after investigating the fluid. Most previous Sleipner studies considered one factor at a time (OFAT). However, it has been proven that the effect of a parameter on the carbon dioxide (CO2) plume outline can be different in the presence of another parameter.
The author’s company was part of a research group—comprising Masoud Ahmadinia and Seyed M. Shariatipour, Centre for Fluid and Complex Systems, Coventry University, Coventry, England; Odd Andersen, SINTEF Digital, Mathematics and Cybernetics Department, Oslo, Norway; and the author—which, for the first time, investigated the joint effect of six important parameters: temperature, pressure, injection rate, porosity, permeability and caprock elevation. The study reveals these parameters’ impact on the overall CO2 migration and trapping in Sleipner and identifies which parameters should be prioritized and calibrated more carefully.
Temperature and pressure
One source of uncertainty addressed in previous studies of the Sleipner model is CO2 density, a function of pressure and temperature, which was considered as one of the uncertain parameters in this most recent study.
The Sleipner aquifer is characterized with high porosity, permeability and lateral extension. As this is a suitable combination for CCS, there has been negligible pressure build-up since the beginning of the injection phase.
In this study, changing temperature and pressure was shown to have an impact on CO2 dissolution. When the temperature increased by 4°C in the base case model (at constant pressure condition), CO2 solubility was reduced by ~1.1%. Conversely, decreasing pressure by 4 bars at isothermal conditions dropped solubility by ~0.2%.
Injection rate
The original Sleipner model is made up of nine layers, each separated with a thin shale layer (FIG. 1), and the plume is injected at a depth of 1,010.5 m into Layer 1 below sea level. In a real case storage process, once injected, the plume encounters and passes through eight intra-formational shale layers before reaching Layer 9. Of course, its flow behavior is still subject to uncertainties and the mechanisms of vertical migration (diffusion, migration points or both).
FIG. 1. Cross-sectional model for the Sleipner saline aquifer. The purple line is the bottom of the reservoir; the blue lines are intra-formational shale layers; the red zone is the only sand layer considered in this study; the dotted line is the injection well in the original model; and 15/9-A-16 is the injection well.
Porosity, permeability and caprock elevation
To investigate the importance of the topography variations below the seismic detection range, 10,000 realizations of top surface elevations within the range of ± 5 m were considered in this study, using Gaussian random fields.
Simulation approach
The Sleipner condition is close to the critical point (30.4°C and 73.8 bar), and CO2 has a gas-like behavior in a supercritical condition. Therefore, increasing the temperature results in a significantly lower density and, consequently, a higher buoyancy force. Moreover, a higher temperature at pressures close to the average pressure of 83 bar in Layer 9 results in lower viscosity and, consequently, higher mobility. In high-temperature conditions, the CO2 plume conforms more accurately to the caprock morphology. Previous research has suggested that increasing the temperature would improve the match between simulation and seismic surveys results. However, this study shows that a rate multiplier of 0.86 results in the best average plume match.
Note: The results presented here are just one of many possible "acceptable" results. Since the parameters are not entirely independent, a separate set of input parameters might potentially lead to the same (if not better) results. The problem when dealing with the Sleipner is complex, so a separate set of parameters might account for the best match in each time step. Therefore, a data-driven modeling approach was used to identify the contribution of each parameter in CO2 plume migration more precisely.
The joint effect of parameters
Results clearly show that the impact of each parameter might change throughout the simulation. Meanwhile, the importance of injection rate seems to increase with time and its percentage predictor importance changes from 17.83% in 2001 to 32.40% in 2010. The injection rate is the only parameter that impacts the mass flowrate in the aquifer directly.
The study’s results indicate the caprock elevation as the most important parameter in controling the plume outline in the Sleipner model. This study also shows that permeability and porosity contribute to changing the shape of the plume outline. Pressure and temperature both have an impact on viscosity and density. As expected, a larger and positive caprock elevation change and porosity increase the structural trapping.
Takeaways
For this study, the upwards migration of CO2 through internal layers was implicitly modeled by rate multipliers. A more detailed study would involve applying a vertical equilibrium model to each internal layer, which the author’s company considers worthy of future work.
The results showed that CO2 density values of ~390 kg/m3 improve the plume match in the Sleipner model. The caprock morphology was also shown to be the most critical parameter in controling the plume migration, with an overall importance of 26%, followed by injection rate (24%), temperature (22%), heterogeneity in permeability (13%), pressure (9%) and porosity (6%).
As was shown in the results of previous studies using the OFAT approach, the effect of a parameter on the plume outline can be different in the presence of another parameter, which could be considered as one of the limitations of the OFAT approach. For example, while previous studies showed that increasing temperature would result in a better match in the Sleipner model, the author’s company’s results showed that this statement is not always valid and depends on the realizations used for caprock, porosity and permeability. However, no fixed correct sets of realizations exist for these data. Any distribution of porosity and permeability within the reported range, and any elevation variations within the ranges lower than the seismic detection limit, can be considered as a valid answer. Therefore, this study cannot make a general statement on the impact of a parameter on the plume match in Sleipner, based on the results from the OFAT approach.
This current work also helps us understand which of the addressed uncertain parameters for the Sleipner model in literature should be prioritized and calibrated more carefully to improve the match. A similar study could be performed on any CO2 storage or oil and gas site to find the importance of uncertain parameters before performing a history matching. After, it is possible to minimize the mismatch between simulation and observed data more efficiently by improving the geological, operational and fluid properties.
CCS can play a significant role in reducing global CO2 emissions and reaching net-zero targets (FIG. 2). Most saline aquifers are sparsely drilled with minimal dynamic data and are subjected to substantial uncertainty. The author’s company has used data-driven models to comprehensively analyze the joint effect of fluid and model uncertainties on CO2 plume migration and trapping mechanisms. The current research improves understanding of the CO2 storage process in large storage sites, which helps to foster a more secure long-term storage plan. HP
FIG. 2. CCS has the potential to reduce greenhouse emissions by up to 32% by 2050.
The Author
Nobakht, B. - TÜV SÜD National Engineering Laboratory, Glasgow, Scotland
Behzad Nobakht is a Data Scientist at TÜV SÜD National Engineering Laboratory and presently works on data-driven strategies to develop condition-based monitoring (CBM) solutions for use in industry. He is also responsible for analyzing large, complex datasets and identifying meaningful patterns that lead to actionable recommendations for the UK’s national standard for fluid flow and density measurement.
Dr. Nobakht earned a BS degree in chemical engineering from Sharif University of Technology in Tehran, Iran, and an MS degree in petroleum engineering from Polytechnic University of Turin in Italy. He achieved his PhD in petroleum engineering at the Uncertainty Quantification group led by Professor Mike Christie at Heriot-Watt University.
Related Articles
From the Archive