Legacy Well Leakage Risk Analysis at the Farnsworth Unit Site

: This paper summarizes the results of the risk analysis and characterization of the CO 2 and brine leakage potential of Farnsworth Unit (FWU) site wells. The study is part of the U.S. DOE’s National Risk Assessment Partnership (NRAP) program, which aims to quantitatively evaluate long-term environmental risks under conditions of signiﬁcant geologic uncertainty and variability. To achieve this, NRAP utilizes risk assessment and computational tools speciﬁcally designed to quantify uncertainties and calculate the risk associated with geologic carbon dioxide (CO 2 ) sequestration. For this study, we have developed a workﬂow that utilizes physics-based reservoir simulation results as input to perform leakage calculations using NRAP Tools, speciﬁcally NRAP-IAM-CS and RROM-Gen. These tools enable us to conduct leakage risk analysis based on ECLIPSE reservoir simulation results and to characterize wellbore leakage at the Farnsworth Unit Site. We analyze the risk of leakage from both individual wells and the entire ﬁeld under various wellbore integrity distribution scenarios. The results of the risk analysis for the leakage potential of FWU wells indicate that, when compared to the total amount of CO 2 injected, the highest cemented well integrity distribution scenario (FutureGen high ﬂow rate) exhibits approximately 0.01% cumulative CO 2 leakage for a 25-year CO 2 injection duration at the end of a 50-year post-injection monitoring period. In contrast, the highest possible leakage scenario (open well) shows approximately 0.1% cumulative CO 2 leakage over the same time frame.


Introduction
Carbon capture, utilization, and storage (CCUS) in geological reservoirs is among the key strategies being employed to reach mid-century climate targets [1,2].Phase III of the Southwest Regional Partnership on Carbon Sequestration (SWP) has successfully injected and stored approximately 800,000 metric tons of CO 2 at the Farnsworth Unit (FWU) in northern Texas.A site of active CO 2 -enhanced oil recovery (CO 2 -EOR) since 2010 [3][4][5].The SWP project has demonstrated commercial CCUS field operations focusing on reservoir engineering, monitoring, simulations, and risk assessment to ensure safe and secure sequestration of CO 2 .The learnings from the project have contributed to best practices manuals for geologic carbon storage, providing insights and lessons for future CCUS projects.
It has become increasingly important that stakeholders require quantitative assessments of risks to help make decisions related to the effectiveness and management of geologic carbon sequestration sites [6].To better support the deployment of commercial-scale integrated carbon capture and storage (CCS) projects and provide guidance on strategic monitoring and post-injection site care (PISC) plans, many site-specific pre-feasibility and feasibility studies have been conducted to assess the risk of potential well leakage.White et al. [7] demonstrated this using NRAP tools for risk-based delineation of an Area of Review (AoR) that represents the region that may be affected by the injection of CO 2 and Energies 2023, 16, 6437 2 of 26 assessed site-specific leakage risks that might impact the quality of underground sources of drinking water (USDWs).Xiao et al. [8] reviewed leakage risk assessment studies, and pointed out that in most risk assessment studies, abandoned legacy wells with hypothetical open boreholes and/or wellbore failures are assumed as the most likely leakage pathways [9][10][11].
The NRAP-Open-IAM is a system-level integrated assessment model that was developed by the United States Department of Energy (DOE) for quantitative risk assessment of the geologic sequestration of CO 2 [12,13].It is built on the CO 2 -PENS model structure [14], which was developed with the GoldSim ® software framework [15].The model takes a stochastic approach with Monte-Carlo studies of multiple realizations.Examples of NRAP-IAM-CS applications can be found in Pawar et al., 2016 [13].For this study, a workflow has been developed to import physics-based reservoir simulator (ECLIPSE ® ) pressure and CO 2 saturation results into the NRAP integrated assessment toolset.
This manuscript describes the quantitative risk and uncertainty assessment applied to the Farnsworth Unit, a site in north Texas of ongoing CO 2 -EOR that has been under study by the Southwest Regional Partnership on Carbon Sequestration (SWP) since 2010 [3,[16][17][18][19][20][21].The focus of this study was the quantification of the CO 2 and brine leakage potential of legacy wells within the FWU.Geo-cellular models created with characterization data, along with advanced approaches, have been employed to improve the effectiveness of risk quantification [16,22,23].This manuscript reviews the key conclusions regarding leakage risk and uncertainty assessment and provides insights from the SWP project.This insight can be used to guide future risk assessment applications for other large-scale CCUS projects.

NRAP Tools and Integrated Assessment Modeling Approach
Among the NRAP toolset used for this study, the Integrated Assessment Model for Carbon Storage (NRAP-IAM-CS) adopts a stochastic approach in which predictions address uncertainties in storage reservoirs, leakage scenarios, and shallow groundwater impacts.It is derived from detailed physics and chemistry simulation results that are used to train more computationally efficient models, referred to here as reduced-order models (ROMs), for each component of the system.The Reservoir Reduced-Order Model Generator (RROM-Gen) [24] is a utility program for creating reservoir ROM lookup tables to be fed into the NRAP-IAM-CS model.These tools are used in the workflow to import storage reservoir simulator results of pressure and CO 2 saturation as input for the NRAP integrated assessment model (IAM) via RROM-Gen, carry out the CO 2 and brine leakage calculations, and perform risk analysis for the FWU site.FWU site descriptions can be found in Czoski 2014 [16], Ross-Coss et al., 2016 [17], and Xiao et al., 2020 [8].
Figures 1 and 2 show the NRAP-IAM-CS model schematic diagram and domain.Detailed instructions on how to use the model can be found in the NRAP Integrated Assessment Model-Carbon Storage (NRAP-IAM-CS) Tool User's Manual [25].
Figure 3 shows NRAP's approach for rapid prediction of whole-system risk performance.The approach includes the following steps: A.
Divide the system into discrete components; B.
Develop detailed component models that are validated against lab/field data; C.
Develop reduced-order models (ROM) that rapidly reproduce component model predictions; D.
Link ROMs via integrated assessment models (IAMs) to predict system performance and risk; calibrate using lab/field data from NRAP and other sources; E.
Develop strategic monitoring protocols that allow verification of predicted system performance.C. Develop reduced-order models (ROM) that rapidly reproduce component model predictions; D. Link ROMs via integrated assessment models (IAMs) to predict system performance and risk; calibrate using lab/field data from NRAP and other sources; E. Develop strategic monitoring protocols that allow verification of predicted system performance.Applying the NRAP tools (NRAP-IAM-CS, RROM-Gen) to CO2-EOR operations at the FWU, we conducted a leakage risk analysis using the ECLIPSE reservoir simulation results (for a detailed discussion, refer to Sections 2.2 and 2.3).We selected the pressure and CO2 saturation results from a comprehensive range of CO2-EOR reservoir simulation scenarios.Each reservoir scenario represented distinct reservoir conditions resulting from different assignments of relative permeability and capillary pressure aimed at assessing the impact of CO2 injection volume and fluid mobility.These fields were then converted and input into the NRAP's Integrated Assessment Model to perform risk quantification for CO2 and brine leakage calculations.
There are three main components of NRAP-IAM-CS used for this study: the storage reservoir, legacy wells, and the shallow groundwater aquifer.The storage reservoir component predicts pressure and saturation changes due to CO2 injection.For this study, these fields are imported as lookup tables from the ECLIPSE reservoir simulation results.We use the cemented wellbore component and the open wellbore component for legacy well CO2 and brine leakage calculations.
The NRAP-IAM-CS model is developed using FWU site-specific well locations, stratigraphy, and characterization data following the guidelines described in the NRAP Integrated Assessment Model-Carbon Storage (NRAP-IAM-CS) Tool User's Manual [25].Multi-realization probabilistic simulations are carried out, sampling various reservoir scenarios and different wellbore integrity distributions.Detailed analysis and quantification of wellbore CO2 and brine leakage are performed.Applying the NRAP tools (NRAP-IAM-CS, RROM-Gen) to CO 2 -EOR operations at the FWU, we conducted a leakage risk analysis using the ECLIPSE reservoir simulation results (for a detailed discussion, refer to Sections 2.2 and 2.3).We selected the pressure and CO 2 saturation results from a comprehensive range of CO 2 -EOR reservoir simulation scenarios.Each reservoir scenario represented distinct reservoir conditions resulting from different assignments of relative permeability and capillary pressure aimed at assessing the impact of CO 2 injection volume and fluid mobility.These fields were then converted and input into the NRAP's Integrated Assessment Model to perform risk quantification for CO 2 and brine leakage calculations.
There are three main components of NRAP-IAM-CS used for this study: the storage reservoir, legacy wells, and the shallow groundwater aquifer.The storage reservoir component predicts pressure and saturation changes due to CO 2 injection.For this study, these fields are imported as lookup tables from the ECLIPSE reservoir simulation results.We use the cemented wellbore component and the open wellbore component for legacy well CO 2 and brine leakage calculations.
The NRAP-IAM-CS model is developed using FWU site-specific well locations, stratigraphy, and characterization data following the guidelines described in the NRAP Integrated Assessment Model-Carbon Storage (NRAP-IAM-CS) Tool User's Manual [25].Multirealization probabilistic simulations are carried out, sampling various reservoir scenarios and different wellbore integrity distributions.Detailed analysis and quantification of wellbore CO 2 and brine leakage are performed.
To address the uncertainties regarding wellbore leakage and the limited availability of wellbore permeability data at FWU, an investigation into the influence of wellbore integrity on leakage was undertaken.NRAP's Integrated Assessment Model was configured for simulating and assessing CO 2 and brine leakage across various scenarios of wellbore integrity and permeability.Numerous probability distribution models for wellbore effective permeability were examined, drawing from Carey's 2017 study [26] on wellbore effective permeability in areas like Alberta, the Gulf of Mexico, and FutureGen's low/high flow rate conditions.These effective permeability distributions have been consolidated in Table 1.It is important to note that the last two categories-a cemented well with an extremely high fixed permeability of 5 × 10 −11 (which is the upper limit allowed in the NRAP-IAM-CS model) and an open well-are hypothetical situations included here to facilitate a conservative worst-case scenario assessment.As the first step in this study, a preliminary IAM simulation with a fixed wellbore permeability of 5 × 10 −11 m 2 is performed to test out the workflow.The simulation duration is set up to be 75 years, with CO 2 injection lasting 25 years and a post-injection monitoring period of 50 years.Figure 4 shows the mean cumulative CO 2 and brine leakage for a combination of storage reservoir scenarios.The CO 2 leak rate increases with injection; once injection stopped, the leak rate plateaued, resulting in a curved cumulative CO 2 leakage time series, while the brine leak rate did not vary significantly over time, corresponding to a linear cumulative leakage.Detailed analysis of each scenario and quantification of wellbore leakage uncertainties are described in Section 3, with several aspects of analysis performed: quantification of CO 2 and brine leakage with various reservoir scenarios (Section 3.1), impact of different wellbore integrity conditions (Section 3.2), open well scenario (Section 3.3), and contribution from individual wells to overall leakage (Sections 3.4 and 3.5).

Reservoir Model of FWU
We performed multiple numerical simulations of CO2-EOR operations to analyze the impact of various reservoir permutation scenarios.A history-matched numerical model was developed by the SWP from the latest characterization data, covering an area of 31 km 2 centered on the active western half of the field and encompassing the entire Morrow

Reservoir Model of FWU
We performed multiple numerical simulations of CO 2 -EOR operations to analyze the impact of various reservoir permutation scenarios.A history-matched numerical model was developed by the SWP from the latest characterization data, covering an area of 31 km 2 centered on the active western half of the field and encompassing the entire Morrow Formation [17,22,23].
The simulation model is 5.52 km east-west and 5.63 km north-south, discretized into 121 × 123 cells.Each cell has an average area of 2082 m 2 .The vertical domain (z) is discretized into 19 layers, including six layers in the Morrow Shale, six layers in the Morrow B Sandstone, four layers in the underlying Morrow D unit, and three layers at the bottom of the Morrow Formation.The model includes both the overlying Morrow Shale and the underlying section of the Morrow Formation, resulting in a total of 282,777 active cells in the model.The overlying Thirteen Finger Limestone was not included in the model as it is assumed to have negligible dynamic behavior due to FWU operations [22,23,27].
All simulation permutations were developed using the same underlying numerical model, including porosity and permeability distribution, fluid model, injection schedule, and initial conditions.The model incorporates the latest characterization data from the FWU, including updated well logs and 3D porosity and permeability maps.Porosity from 25 wells and permeability from 14 wells were upscaled to the numerical model grid.Stochastic methods are then employed to distribute the porosity and permeability across the simulation grid.For a detailed description of methods and assumptions used in the numerical model, see Moodie et al., 2019 [23] and Moodie, Ampomah, Heath, et al., 2021 [27].
The CO 2 injection schedule for the numerical simulations was consistent across all the presented simulation cases.The simulation duration spans 75 years, with CO 2 injection taking place during the first 25 years, followed by a post-injection monitoring period of 50 years.The injection schedule has been designed to mimic the potential future CO 2 -EOR operations at the FWU, with the initial period receiving an unlimited amount of CO 2 , including any recycled CO 2 .Post-2016, the availability of "new" CO 2 in the field is gradually reduced, and more reliance is placed on recycled CO 2 to meet targets.CO 2 injection ceases by 2035, which aligns with the anticipated end of the field's lifetime.For specific details of the injection schedule, see Table 2.

Reservoir Model Scenarios
To assess potential well leakage at the FWU, we conducted 17 simulation permutations categorized into two main groups: (1) ten simulation cases with a single relative permeability and capillary pressure relationship assigned homogeneously across the model domain, and (2) seven simulation cases with multiple relative permeability and capillary pressure relationships assigned heterogeneously by hydrostratigraphic unit (Table 3).For a comprehensive explanation of the method and assumptions used to assign relative permeability and capillary pressure by hydrostratigraphic unit, see Moodie et al., 2019 [23] and Moodie, Ampomah, Heath, et al., 2021 [27].The CO 2 saturation and reservoir pressure results were extracted and converted to the required format using RROM-Gen before being applied to the NRAP-IAM-CS tool.It can be seen from Table 4 that the assigned relative permeability and capillary pressure can significantly influence the volume of CO 2 that can be injected, affecting its distribution in the reservoir and the potential for leakage.In Figure 5, we show the locations of the wells used in the NRAP-IAM-CS model.This includes 21 CO 2 injection wells (represented by red dots) and 31 legacy wells considered potential pathways for leakage (shown as blue dots).An example of the pressure distribution and CO 2 saturation after processing the numerical simulation results through RROM-Gen is illustrated in Figure 6.

NRAP-IAM-CS Quantification of CO 2 and Brine Leakage
NRAP tools are applied for risk quantification of wellbore leakage that covers the full parameter range of reservoir simulations at the Farnsworth Unit Site.Multi-realization probabilistic simulations of CO 2 and brine leakage were carried out, and the impact of various reservoir scenarios and wellbore integrity distribution scenarios is analyzed.Simulations are set up for a CO 2 injection duration of 25 years with a 50-year post-injection monitoring period.Figures 7 and 8 show the cumulative CO 2 and brine leakage to the groundwater aquifer for different reservoir scenarios with a fixed wellbore permeability of 5 × 10 −11 m 2 .
Figure 7a,b shows simulations with six reservoir category scenarios picked from Table 3 to illustrate the impact of assigned relative permeability and capillary pressure.Some scenarios allow greater fluid movement across the widest saturation range and more leakage (e.g., linear scenario, green line in Figure 7a), while other scenarios with capillary pressure added show greatly reduced fluid mobility (e.g., HS2wPc scenario, yellow line in Figure 7a).
The 11 categories shown in Figure 8a,b are chosen over various parameter scenarios to represent the more complex site condition and account for uncertainties associated with FWU site data.Note that all simulation results presented in this study from Figure 8 on are all based on these 11-category reservoir scenarios.
Analysis shows that relative permeability and capillary pressure assignment are the largest controls on fluid flow; CO 2 leakages are more sensitive to the variation in the assignment, while brine leakages are less affected.
Linear (green line): Relative permeability curve homogeneously assigned with no capillary pressure; allow greatest fluid mobility across the widest saturation range.HS2wPc (yellow line): Five relative permeability curves and five associated capillary pressure curves are heterogeneously assigned; capillary pressure reduces fluid mobility.
parameter range of reservoir simulations at the Farnsworth Unit Site.Multi-realization probabilistic simulations of CO2 and brine leakage were carried out, and the impact of various reservoir scenarios and wellbore integrity distribution scenarios is analyzed.Simulations are set up for a CO2 injection duration of 25 years with a 50-year post-injection monitoring period.Figures 7 and 8 show the cumulative CO2 and brine leakage to the groundwater aquifer for different reservoir scenarios with a fixed wellbore permeability of 5 × 10 −11 m 2 .

Impact of Various Wellbore Integrity Distribution on Leakage
For each wellbore permeability distribution scenario, 1000-realization probabilistic simulations are carried out, sampling among 11 reservoir permutations.Among the four wellbore integrity distributions listed in Table 1, the FutureGen high flow rate model shows the highest permeability distribution overall and therefore has the fastest leaking potential, while Alberta shows the lowest permeability distribution overall, which may lead to the slowest leaking potential for CO 2 and brine leakage.Compared to the total amount of CO 2 injected, the highest well permeability distribution scenario (FutureGen high flow rate) shows ~0.01%cumulative CO 2 leakage at the end of the 50-year post-injection monitoring period.

Open Well Leakage Scenario
To check on the largest possible CO 2 and brine leakage in the hypothetically open well condition, we performed CO 2 and brine leakage calculations for the open well scenario.Comparing with cemented wellbore permeability distribution scenarios (Table 4), the open well case (Figures 11 and 12) shows several orders of magnitude increases for CO 2 and brine leakage, starting early, a few years after CO 2 injection started, and then flattening out at around year 12. CO 2 and brine leak very fast through an open well, and leak all CO 2 and brine at the well location quickly.Compared to the total amount of CO2 injected, the highest well permeability distribution scenario (FutureGen high flow rate) shows ~0.01%cumulative CO2 leakage at the end of the 50-year post-injection monitoring period.

Open Well Leakage Scenario
To check on the largest possible CO2 and brine leakage in the hypothetically open well condition, we performed CO2 and brine leakage calculations for the open well scenario.Comparing with cemented wellbore permeability distribution scenarios (Table 4), the open well case (Figures 11 and 12) shows several orders of magnitude increases for CO2 and brine leakage, starting early, a few years after CO2 injection started, and then flattening out at around year 12. CO2 and brine leak very fast through an open well, and leak all CO2 and brine at the well location quickly.Compared to the total amount of CO2 injected (about 10 million tons for each permutation scenario), the hypothetical highest possible leakage scenario (open well) shows ~0.1% cumulative CO2 leakage at the end of the 50-year post-injection monitoring period.

Individual Well Leakage Analysis-Open Well Scenario
CO2 and brine leakage calculations are performed in hypothetically open well conditions for each individual well of 31 potential leaking legacy wells to check on each well's contribution towards overall leakage.The simulations were carried out for each legacy well shown in Figure 13 (blue dot).Compared to the total amount of CO 2 injected (about 10 million tons for each permutation scenario), the hypothetical highest possible leakage scenario (open well) shows ~0.1% cumulative CO 2 leakage at the end of the 50-year post-injection monitoring period.

Individual Well Leakage Analysis-Open Well Scenario
CO 2 and brine leakage calculations are performed in hypothetically open well conditions for each individual well of 31 potential leaking legacy wells to check on each well's contribution towards overall leakage.The simulations were carried out for each legacy well shown in Figure 13 (blue dot).
The simulation results for all 31 legacy wells with a hypothetical open well scenario show that only two legacy wells (well #7 and well #15) have CO 2 leakage to the groundwater aquifer.All legacy wells show brine leakage, but only the two wells with CO 2 leakage (#7 and #15) show a higher level of brine leakage.Well #7 shows the highest cumulative CO 2 and brine leakage.
The analysis for a list of representative legacy wells is summarized in Table 5.Table 5 lists the representative legacy well location, mean cumulative CO 2 and brine leakages to the groundwater aquifer at the end of the 50-year post-injection monitoring period, and also the relative distance of the legacy well from a nearby injection well (i.e., closer or far away from the nearby injection well).
Figure 14 shows the wellbore pressure for the selected legacy wells.It shows wells #7 and #15 have higher wellbore pressure during the time from year 10 to year 25, with well #7 showing the generally highest pressure during this time period.Figure 15 shows that well #15 displays much higher CO 2 saturation, wells #6 and #7 show lower CO 2 saturation, and wells #11 and #22 show an order of magnitude lower amount of CO 2 saturation compared to well #15.The analysis for a list of representative legacy wells is summarized in Table 5.Table 5 lists the representative legacy well location, mean cumulative CO2 and brine leakages to the groundwater aquifer at the end of the 50-year post-injection monitoring period, and also the relative distance of the legacy well from a nearby injection well (i.e., closer or far away from the nearby injection well).Figures 16 and 17 show the cumulative CO 2 and brine leakage to the groundwater aquifer from the selected wells.Well #7 displays the highest cumulative CO 2 and brine leakage to the groundwater aquifer.For brine leakage, well #7 displays the highest amount of cumulative leakage, followed by well #15, then well #6, and then wells #11 and #22.CO 2 and brine leak very fast through an open well, and leak all CO 2 and brine at the well location quickly.
Figure 14 shows the wellbore pressure for the selected legacy wells.It shows wells #7 and #15 have higher wellbore pressure during the time from year 10 to year 25, with well #7 showing the generally highest pressure during this time period.Figure 15 shows that well #15 displays much higher CO2 saturation, wells #6 and #7 show lower CO2 saturation, and wells #11 and #22 show an order of magnitude lower amount of CO2 saturation compared to well #15.Figures 16 and 17 show the cumulative CO2 and brine leakage to the groundwater aquifer from the selected wells.Well #7 displays the highest cumulative CO2 and brine leakage to the groundwater aquifer.For brine leakage, well #7 displays the highest amount of cumulative leakage, followed by well #15, then well #6, and then wells #11 and #22.CO2 and brine leak very fast through an open well, and leak all CO2 and brine at the well location quickly.Figures 16 and 17 show the cumulative CO2 and brine leakage to the groundwater aquifer from the selected wells.Well #7 displays the highest cumulative CO2 and brine leakage to the groundwater aquifer.For brine leakage, well #7 displays the highest amount of cumulative leakage, followed by well #15, then well #6, and then wells #11 and #22.CO2 and brine leak very fast through an open well, and leak all CO2 and brine at the well location quickly.For open well scenario individual well analysis, the situation with multiple injection wells and legacy wells mingled together seems to complicate the analysis in terms of the impact of wellbore pressure, CO2 saturation, and the distance between a legacy well and a nearby injection well on the potential legacy well CO2 and brine leakage.For example, well #6 is farther away from a nearby injection well compared to well #7, but it shows about the same level of CO2 saturation compared to well #7, and it does not show any CO2 leakage.Well #11 is a legacy well that is the closest to a nearby injection well; however, it does not show CO2 leakage either.Well #15 shows the highest level of CO2 saturation, but it has less CO2 leakage compared to well #7, which has a lower level of CO2 saturation but generally higher wellbore pressure during the time period of year 10 to year 25.For open well scenario individual well analysis, the situation with multiple injection wells and legacy wells mingled together seems to complicate the analysis in terms of the impact of wellbore pressure, CO 2 saturation, and the distance between a legacy well and a nearby injection well on the potential legacy well CO 2 and brine leakage.For example, well #6 is farther away from a nearby injection well compared to well #7, but it shows about the same level of CO 2 saturation compared to well #7, and it does not show any CO 2 leakage.Well #11 is a legacy well that is the closest to a nearby injection well; however, it does not show CO 2 leakage either.Well #15 shows the highest level of CO 2 saturation, but it has less CO 2 leakage compared to well #7, which has a lower level of CO 2 saturation but generally higher wellbore pressure during the time period of year 10 to year 25.

Individual Well Leakage Analysis-Cemented Well Scenario
To check on each well's contribution towards overall leakage CO 2 and brine leakage, analysis is also carried out for cemented wellbore settings.This section describes calculations for individual wells of 31 potential leaking legacy wells at the FWU site with the highest possible effective wellbore permeability (5 × 10 −11 m 2 ) allowed in the NRAP integrated assessment model (NRAP-IAM-CS) as a conservative assessment for cemented wellbore potential leakage risk.According to wellbore permeability distribution data from Carey 2017 [26], the effective wellbore permeability value used in this analysis is many orders of magnitude higher compared to the most likely wellbore failure/degradation scenario (outside casing completion failure or degradation of the annulus, approximately 10 −24 to 10 −16 m 2 ).
Table 6 lists all 31 legacy wells locations, the mean cumulative CO 2 and brine leakages to the groundwater aquifer calculated at the end of the 50-year post-injection monitoring period, and their impact on the overlaying groundwater aquifer in terms of pH and TDS plume volume.The table lists CO 2 leakage amounts ranked from high to low.Analyzing leakage results of 31 legacy wells shows that Well #15 (see Figure 18 for well # labeled) has the highest cumulative CO 2 leakage amount at the end of the 50-year post-injection monitoring period, while three wells that are located at the site domain edge (Wells #2, 25, and 31) show zero CO 2 leakage.The CO 2 leakage amount extends in two orders of magnitude, from the lowest 5.38 × 10 −6 MT to the highest 6.30 × 10 −4 MT for the non-zero CO 2 leakage legacy wells.A list of representative legacy wells is summarized in Table 7. Table 7 lists the representative legacy well's location, cumulative CO2, and brine leakages to the groundwater aquifer at the end of the 50-year post-injection monitoring period, and also A list of representative legacy wells is summarized in Table 7. Table 7 lists the representative legacy well's location, cumulative CO 2 , and brine leakages to the groundwater aquifer at the end of the 50-year post-injection monitoring period, and also the relative distance of the legacy well from a nearby injection well (i.e., closer or far away from the nearby injection well).Figure 14 shows the wellbore pressure time series for selected legacy wells.It shows Well #7 and Well #15 have higher wellbore pressure during the time period from year 10 to year 25, with Well #7 showing generally the highest pressure during this time period.Well #2 generally shows the lowest pressure.
Figure 15 shows that Well #15 displays much higher CO 2 saturation, Wells #6 and #7 show lower CO 2 saturation, Wells #11 and #22 show an order of magnitude lower CO 2 saturation compared to Well #15, and Well #2 shows zero CO 2 saturation.
Figure 19a,b show the temporal cumulative CO 2 and brine leakage to the groundwater aquifer from 31 legacy wells, respectively.Well #15 (grey colored line) displays the highest cumulative CO 2 leakage to the groundwater aquifer.Cumulative CO 2 leakage to the groundwater aquifer followed the same sequence in amount as CO 2 saturation.
Figures 20 and 21 show each legacy well's cumulative leakage amount relative to each other quantitatively in bubble plots (bigger bubble size corresponding to larger leakage).We can see that the wells with a higher CO 2 leakage amount (Figure 20) are located between 318,000 m and 320,000 m in the east-west direction and between 4,013,000 and 4,016,000 m in the north-south direction.There are more legacy wells in the FWU site located towards the eastern side of the domain, but the CO 2 leakage amount from these wells is relatively smaller.
Brine leakage amounts (Figure 21) are distributed relatively evenly among all legacy wells, with similar cumulative leakage amounts in the same order of magnitude for each well.Well #20 displays the highest brine leakage amount of 9.21 × 10 −5 MT, and three near-site domain edge wells (#2, 25, and 31) display the lowest leakage amount of 6.71 × 10 −5 MT.
saturation compared to Well #15, and Well #2 shows zero CO2 saturation.
Figure 19a,b show the temporal cumulative CO2 and brine leakage to the groundwater aquifer from 31 legacy wells, respectively.Well #15 (grey colored line) displays the highest cumulative CO2 leakage to the groundwater aquifer.Cumulative CO2 leakage to the groundwater aquifer followed the same sequence in amount as CO2 saturation.Figures 20 and 21 show each legacy well's cumulative leakage amount relative to each other quantitatively in bubble plots (bigger bubble size corresponding to larger leakage).We can see that the wells with a higher CO2 leakage amount (Figure 20) are located between 318,000 m and 320,000 m in the east-west direction and between 4,013,000 and 4,016,000 m in the north-south direction.There are more legacy wells in the FWU site located towards the eastern side of the domain, but the CO2 leakage amount from these wells is relatively smaller.On individual well leakage analysis for a cemented wellbore scenario, multiple factors could affect the potential CO 2 and brine leakage risk, such as wellbore pressure, CO 2 saturation, the distance between a legacy well and a nearby injection well, and wellbore integrity condition.For example, Well #11 is a legacy well that is the closest to a nearby injection well, but it shows lower CO 2 leakage following the CO 2 saturation pattern.Well #15 shows the highest level of CO 2 saturation and leakage, but it has a lower wellbore pressure compared to Well #7 during the time period from years 10 to 25 (last 15 years of the injection period) and is not the closest legacy well to the nearby injection well.It seems that under the conservative assumption that all legacy wells have the same high effective permeability of 5 × 10 −11 m 2 , CO 2 saturation plumes impact wellbore CO 2 leakage the most, while reservoir pressure impacts brine leakage the most.All legacy wells leak brine in the same order of magnitude.Brine leakage amounts (Figure 21) are distributed relatively evenly among all legacy wells, with similar cumulative leakage amounts in the same order of magnitude for each well.Well #20 displays the highest brine leakage amount of 9.21 × 10 −5 MT, and three nearsite domain edge wells (#2, 25, and 31) display the lowest leakage amount of 6.71 × 10 −5 MT.
On individual well leakage analysis for a cemented wellbore scenario, multiple factors could affect the potential CO2 and brine leakage risk, such as wellbore pressure, CO2 saturation, the distance between a legacy well and a nearby injection well, and wellbore integrity condition.For example, Well #11 is a legacy well that is the closest to a nearby injection well, but it shows lower CO2 leakage following the CO2 saturation pattern.Well #15 shows the highest level of CO2 saturation and leakage, but it has a lower

Conclusions
To characterize the leakage potential of FWU wells, a workflow has been developed to import physics-based reservoir simulator (PETREL-ECLIPSE) pressure and CO 2 saturation results into the NRAP Integrated Assessment Model.NRAP tools, including NRAP-IAM-CS and RROM-Gen, are used to perform quantitative risk assessments of CO 2 and brine leakage.We applied NRAP tools for risk quantification of wellbore leakage that cover the full parameter range of ECLIPSE reservoir simulations at the Farnsworth Unit (FWU) Site and for various wellbore integrity distribution scenarios.Risk analysis of the leakage potential of FWU wells shows: Comparing the total amount of CO 2 injected, the highest cemented well integrity distribution scenario (FutureGen high flow rate) shows ~0.01%cumulative CO 2 leakage at the end of the 50-year post-injection monitoring period for a 25-year CO 2 injection duration; the highest possible leakage scenario (Open Well) shows ~0.1% cumulative CO 2 leakage.
To further analyze the leakage contribution from individual wells at the FWU site, two sets of NRAP simulations are carried out for leakage under different wellbore integrity conditions.For the open well scenario, compared to the total amount of CO 2 injected (about 10 million tons for each reservoir permutation scenario), the highest possible leakage scenario (open well) shows that Well #7 leaks the most with ~0.1% cumulative CO 2 leakage, and Well #15 has ~0.01%cumulative CO 2 leakage.All other legacy wells show no CO 2 leakage.For the cemented wellbore scenario, compared to the total amount of CO 2 injected (about 10 million tons for each reservoir permutation scenario), a conservative assessment with the highest effective wellbore permeability (5 × 10 −11 m 2 ) shows that Well #15 leaks the most with ~0.0063% cumulative CO 2 leakage, and all 31 legacy wells show total ~0.04% cumulative CO 2 leakage at the end of the 50-year post-injection monitoring period.There is small brine leakage from each legacy well, with a total leakage amount of 2.34 × 10 −3 MT for all 31 legacy wells.There is no impact on the groundwater aquifer in terms of pH and TDS.
This analysis is a highly conservative assessment, assuming all legacy wellbores fail/degrade at the same high effective permeability of 5 × 10 −11 m 2 .For the real situation at the FWU site, individual well leakage situations will be different based on each well's degradation status; more complete wellbore observational data will help facilitate the site-specific leakage risk assessment accurately.

Figure 1 .
Figure 1.The NRAP-IAM-CS model's schematic diagram shown on top level dashboard.

Figure 3 .
Figure 3. NRAP's approach for rapid prediction of whole-system risk performance.

Figure 3 .
Figure 3. NRAP's approach for rapid prediction of whole-system risk performance.

Figure 4 .
Figure 4. Mean cumulative CO2 and brine leakage for a combination of reservoir scenarios.
Brine Leakage CO2 leakage to Aquifer (MT) brine leakage to Aquifer (MT)

Figure 4 .
Figure 4. Mean cumulative CO 2 and brine leakage for a combination of reservoir scenarios.

Figure 5 .
Figure 5. NRAP-IAM-CS model simulation well location for the FWU site.Red dots represent CO2 injection wells, and blue dots represent potential leakage legacy wells.Numerical values (1 to 31) correspond to successive legacy wells (not marked on this figure).

Figure 5 .
Figure 5. NRAP-IAM-CS model simulation well location for the FWU site.Red dots represent CO 2 injection wells, and blue dots represent potential leakage legacy wells.Numerical values (1 to 31) correspond to successive legacy wells (not marked on this figure).

Figure 5 .Figure 6 .
Figure 5. NRAP-IAM-CS model simulation well location for the FWU site.Red dots represent CO2 injection wells, and blue dots represent potential leakage legacy wells.Numerical values (1 to 31) correspond to successive legacy wells (not marked on this figure).

Figure 6 .
Figure 6.RROM-Gen Response Surface from PETREL-ECLIPSE Reservoir Process Model, as shown in (a) (reservoir pressure, left panel) and (b) (CO 2 saturation, right panel).The color bar legend shows reservoir pressure ranging from around 18 to 27 MPa and CO 2 saturation ranging from 0 to 80%.

Figure 7 .
Figure 7. (a) Cumulative CO 2 leakage to the groundwater aquifer for a selected group of six reservoir category scenarios shown in Table4.(b) Cumulative brine leakage to the groundwater aquifer for a selected group of six reservoir category scenarios shown in Table4.

Figure 7 .Figure 8 .Figure 8 .
Figure 7. (a) Cumulative CO2 leakage to the groundwater aquifer for a selected group of six reservoir category scenarios shown in Table4.(b) Cumulative brine leakage to the groundwater aquifer for a selected group of six reservoir category scenarios shown in Table4.

Figures 9 27 Figure 9 .
Figure 9. Mean cumulative CO2 leakages to the groundwater aquifer for different wellbore permeability scenarios.Figure 9. Mean cumulative CO 2 leakages to the groundwater aquifer for different wellbore permeability scenarios.

Figure 9 .
Figure 9. Mean cumulative CO2 leakages to the groundwater aquifer for different wellbore permeability scenarios.Figure 9. Mean cumulative CO 2 leakages to the groundwater aquifer for different wellbore permeability scenarios.

Figure 9 .
Figure 9. Mean cumulative CO2 leakages to the groundwater aquifer for different wellbore permeability scenarios.

Figure 10 .
Figure 10.Mean cumulative brine leakages to the groundwater aquifer for different wellbore permeability scenarios.

Figure 10 .
Figure 10.Mean cumulative brine leakages to the groundwater aquifer for different wellbore permeability scenarios.

Figure 11 .
Figure 11.Mean cumulative CO2 leakage to the groundwater aquifer for different wellbore permeability distributions.

Figure 11 .
Figure 11.Mean cumulative CO 2 leakage to the groundwater aquifer for different wellbore permeability distributions.

Figure 12 .
Figure 12.Mean cumulative brine leakage to the groundwater aquifer for different wellbore permeability distributions.

Figure 12 .
Figure 12.Mean cumulative brine leakage to the groundwater aquifer for different wellbore permeability distributions.

Figure 13 .
Figure 13.31 potential leakage legacy well locations (blue dots) and 21 CO2 injection well locations (red dots).A few representative legacy wells are marked with well numbers.The simulation results for all 31 legacy wells with a hypothetical open well scenario show that only two legacy wells (well #7 and well #15) have CO2 leakage to the groundwater aquifer.All legacy wells show brine leakage, but only the two wells with CO2 leakage (#7 and #15) show a higher level of brine leakage.Well #7 shows the highest cumulative CO2 and brine leakage.The analysis for a list of representative legacy wells is summarized in Table5.Table5lists the representative legacy well location, mean cumulative CO2 and brine leakages to the groundwater aquifer at the end of the 50-year post-injection monitoring period, and also the relative distance of the legacy well from a nearby injection well (i.e., closer or far away from the nearby injection well).

Figure 13 .
Figure 13.31 potential leakage legacy well locations (blue dots) and 21 CO 2 injection well locations (red dots).A few representative legacy wells are marked with well numbers.

Figure 14 .
Figure 14.Bottom-hole pressure for the selected legacy wells.

Total CO 2
Leakage to Aquifer (MT)

Figure 16 .Figure 16 . 27 Figure 17 .
Figure 16.Total cumulative CO2 leakage to the groundwater aquifer for selected legacy wells in a hypothetical open-well scenario.Leakage from wells #6, 11 and 22 are zero.

Figure 17 .
Figure 17.Mean cumulative brine leakage to the groundwater aquifer for selected legacy wells in a hypothetical open-well scenario.Leakage from wells # 11 and 22 are an order of magnitude smaller.

Energies 2023 , 27 Figure 18 .
Figure 18.A total of 31 potential leakage legacy well locations (blue dots) and 21 CO2 injection well locations (red dots).A few representative legacy wells are marked with well numbers.

Figure 18 .
Figure 18.A total of 31 potential leakage legacy well locations (blue dots) and 21 CO 2 injection well locations (red dots).A few representative legacy wells are marked with well numbers.

Figure 19 .
Figure 19.(a) Temporal cumulative CO2 leakage to the groundwater aquifer for 31 legacy wells.(b) Temporal cumulative brine leakage to the groundwater aquifer for 31 legacy wells.

Figure 19 .
Figure 19.(a) Temporal cumulative CO 2 leakage to the groundwater aquifer for 31 legacy wells.(b) Temporal cumulative brine leakage to the groundwater aquifer for 31 legacy wells.

Figure 20 .
Figure 20.Relative view of CO2 cumulative leakage amount for each legacy well at the end of the 50-year post-injection monitoring period.Bigger circle denotes higher leakage amount.(Note: background small orange dots shown are injection wells location).

Figure 20 .
Figure 20.Relative view of CO 2 cumulative leakage amount for each legacy well at the end of the 50-year post-injection monitoring period.Bigger circle denotes higher leakage amount.(Note: background small orange dots shown are injection wells location).

Figure 21 .
Figure 21.Relative view of brine cumulative leakage amount for each legacy well at the end of the 50-year post-injection monitoring period.Bigger circle denotes higher leakage amount.(Note: background small orange dots shown are injection wells location).

Figure 21 .
Figure 21.Relative view of brine cumulative leakage amount for each legacy well at the end of the 50-year post-injection monitoring period.Bigger circle denotes higher leakage amount.(Note: background small orange dots shown are injection wells location).

Table 2 .
Injection schedule for numerical simulations.

Table 3 .
The numerical simulation models used in the analysis.

Table 4 .
Total CO 2 injected in each reservoir simulation permutation.

Table 5 .
Representative legacy well locations, relative distance to nearby injection wells, and cumulative CO2 and brine leakage to the groundwater aquifer at the end of the 50-year post-injection monitoring period.

Table 5 .
Representative legacy well locations, relative distance to nearby injection wells, and cumulative CO 2 and brine leakage to the groundwater aquifer at the end of the 50-year post-injection monitoring period.

Table 6 .
List of all 31 legacy well leakage and impact results to the groundwater aquifer at the end of the 50-year post-injection monitoring period.

Table 7 .
Representative legacy well's location, relative distance to nearby injection well, and cumulative CO 2 and brine leakage to groundwater aquifer at the end of the 50-year post-injection monitoring period.