(413f) On the Feasibility of Geophysical Methods for CO2 Monitoring in the North Dakota Carbonsafe Project

Geophysical methods are key for characterizing the geologic formations to store CO₂ and monitor the injected CO₂ over time to ensure containment. Before applying these methods, feasibility studies are required to assess the detectability of the reservoir and the geophysical response of the CO₂ injected into the reservoir. These studies are the basis for successful CO₂-monitoring surveys. We conducted seismic and controlled-source electromagnetic (CSEM) feasibility studies as part of North Dakota CarbonSAFE (Carbon Storage Assurance Facility Enterprise), a multidisciplinary project that assesses safe, permanent, commercial-scale geologic storage of CO₂ generated by the Milton R. Young coal-fired power plant. The Energy & Environmental Research Center leads the project in partnership with the U.S. Department of Energy National Energy Technology Laboratory, Minnkota Power Cooperative, and BNI Energy. The study area is located near Center, North Dakota, with storage expected to be in the Broom Creek and Deadwood Formations at depths of 1700 and 3000 m, respectively. In the integrated multimeasurement geophysical approach considered for North Dakota CarbonSAFE, it is expected that the CSEM method is a strong contributor to mapping the CO₂ movement. This paper focuses on seismic and CSEM feasibility-monitoring studies for the Broom Creek Formation.

Seismic is the standard geophysical method for reservoir characterization and CO₂ monitoring in carbon capture, utilization, and storage (CCUS) projects. However, electromagnetic (EM) methods such as the CSEM method are also well situated for monitoring CO₂ injected into a reservoir because of the strong conductivity contrast generated from CO₂ replacing brine or oil. In the case of CO₂ replacing brine, the reservoir’s resistivity increases.

We conducted a seismic feasibility study based on our analysis of the rock physics and seismic modeling of the Broom Creek Formation using core data and well logs from four wells available for this study. In the case of the CSEM feasibility study, a 3D modeling of the CSEM response and a field noise test were considered using well logs and seismic interpretation information. The goal was to define the expected level of surface EM field response caused by an increase in CO₂ saturation and determine if signals of that magnitude could be detected in the field in the presence of observed noise levels.

Seismic Feasibility

The seismic feasibility study consisted of three parts: quality control of the well logs and core data, crossplot and diagnostic modeling of rock properties for the Broom Creek Formation, and fluid substitution and seismic modeling of the expected seismic response. The log data were checked for consistency between the wells by comparing the logs and their histograms. The log data were checked for accuracy by comparing log measurements with core lithology, porosity, and density measurements.

Rock physics crossplot analysis of elastic properties and density with porosity and lithology were used to identify data trends. The trends suggest that the seismic data may be sensitive to changes in lithology and porosity in the Broom Creek Formation. Rock physics diagnostic modeling was used to explain the observed trends. A contact cement model explains the initial properties of the sand at high porosity (33%) and a small volume (0.5%) of dolomite cement. A constant cement model is consistent with the change in elastic properties as sorting of the sand decays and porosity is reduced. This trend is complicated by the presence of an increasing volume of dolomite as porosity and sand volume decrease.

The CO₂ injected into the Broom Creek and Deadwood Formations will be supercritical with the density of a liquid and the bulk modulus of a gas. The CO₂ will mix with and displace the reservoir brine which has a higher density and bulk modulus. The way the fluids mix also affects the elastic properties of the mixture. The change in fluid properties will change Broom Creek Formation properties, potentially allowing the CO₂ plume to be monitored with time-lapse seismic. Rock physics models were calculated using Gassmann’s equation with uniform and patchy mixtures of CO₂ and brine to simulate reservoir properties of the plume during a seismic monitor survey.

The baseline (wet) and monitor (fluid substituted) properties were used to model expected seismic reservoir response before and during CO₂ injection. The baseline model consists of clean wet sands. The monitor models consist of a 2D wedge model of CO₂ with constant saturation and a model with constant plume thickness and decreasing CO₂ saturation where the fluid has either a uniform or a patchy mixture. The models predict a Class 4 amplitude versus offset (AVO) response where the intercept amplitude increases up to 25% when CO₂ is present. The plume may be difficult to detect when it is thin (<10 m), has low CO₂ saturation, or has a patchy fluid mixture.

CSEM Feasibility

The CSEM workflow’s input data are interpreted 3D seismic horizons and well logs extending from shallow through the Broom Creek Formation to the base of the Deadwood Formation. A 1D anisotropic model was built based on algorithms for cumulative total electrical conductance and total cumulative resistance. The integration of surface and borehole data is an essential requirement derived from 3D CSEM modeling. The data integration is achieved by measuring between surface-to-borehole and calibrating the information with conventional logs and considering resistivity anisotropy. Borehole logs are taken as ground truth as they are on a reservoir scale and their limitations are well understood. A fluid substitution (using Archie’s equation) was carried out by replacing the brine with CO₂. This 1D anisotropic model is used to build the 3D model considered in this workflow.

3D modeling is the core of the feasibility analysis; one transmitter and eight receivers along three-receiver lines were used to build various 3D models for the 3D feasibility study. The receivers are at the injection well and represent the most distant receivers from a transmitter location (north location from the receivers) and serve as a reference. The horizontal electric field response (Ex and Ey) and the vertical magnetic field as it changes with time (dBz/dt) were modeled.

The 3D model was benchmarked against the 1D solutions for the same anisotropic model to understand EM field behavior regarding numerical noise and the noise caused by the modeling grid’s approximation errors. These models are also used to estimate the optimum parameter range. This verification process allows separation of numerical accuracy limits, modeling artifacts, and anomalies caused by the model’s features (e.g., very strong resistivity contrasts). The benchmark models covering most of the field scenarios are based on petrophysical analysis. Then the equipment/sensor choice is added to minimize the 3D modeling effort. The result is a set of models including the expected anomaly within the measurable time window.

First, 21 months of CO₂ injection in the Broom Creek Formation is simulated using a 60% average fluid saturation and an injection radius of 500 m. The simulation results demonstrated that 15–18 months of injection produced a sufficiently strong anomaly. Next, an injection radius of 150 m was used, and the required receiver spacing was estimated. the 3D modeling results for the Broom Creek Formation for Ex, Ey, and dBz/dt showed that the CO₂ anomaly can be reconstructed up to 300-m receiver spacing.

Conclusions

An analysis of rock physics crossplots and diagnostic models suggests that the baseline seismic data may be sensitive to changes in lithology and porosity of the Broom Creek Formation and that these changes may be detectable by prestack seismic inversion. AVO seismic modeling of the clean Broom Creek Formation sand indicates Class 4 AVO anomalies where the amplitude of the seismic intercept increases by 25% for the monitor survey when the reservoir is saturated with a mixture of 79% CO₂ and 21% brine. 2D seismic modeling also suggests that when the plume is thin (<10 m) or has low CO₂ saturation, it will be difficult to detect. It will also be more difficult to detect when the fluid mixture is patchy. The results above show that a rock physics feasibility analysis contributes important details to reservoir characterization and monitoring of CO₂ sequestration in the Broom Creek Formation. Seismic modeling suggests that CO₂ saturation changes will be visible in the middle Broom Creek Formation sand.

The feasibility study results demonstrate that the CSEM method can monitor injected CO₂ in the study area; based on the noise analysis, changes in the resistivity data due to CO₂ injection can be resolved. It is feasible to use CSEM to image CO₂ injection in the Broom Creek Formation. The optimum receiver spacing for the CSEM data acquisition is 200 to 300 m. The CSEM survey should be repeated after 18 months to monitor the reservoir fluid changes from CO₂ injection.