- Email: [email protected]
Numerical Modeling of the Aquistore CO2 Storage Project
Available online at www.sciencedirect.com
ScienceDirect Energy Procedia 114 (2017) 4886 – 4895
13th International Conference on Greenhouse Gas Control Technologies, GHGT-13, 14-18 November 2016, Lausanne, Switzerland
Numerical modeling of the Aquistore CO2 storage project Tao Jianga *, Lawrence J. Pekota, Lu Jina, Wesley D. Pecka, Charles D. Goreckia, and Kyle Worthb a
Energy & Environmental Research Center, 15 North 23rd Street, Stop 9018, Grand Forks, ND 58202-9018,USA b Petroleum Technology Research Centre, 6 Research Drive, Regina, SK S4S 7J7, Canada
Abstract Serving as a storage site for the world’s first commercial postcombustion carbon capture, utilization, and storage project from a coal-fired power generation facility, the Aquistore commenced injection operations in April 2015 at a site near SaskPower’s Boundary Dam Power Station. The Energy & Environmental Research Center through its Plains CO2 Reduction (PCOR) Partnership, is collaborating with the Petroleum Technology Research Centre and SaskPower to continue PCOR Partnership efforts to numerically model and interpret the performance of the project’s injection well and observation well to improve monitoring, verification, and accounting (MVA) strategies based on history matching and predictive simulation results. ©2017 2017T.The Authors. Published by Elsevier © Jiang Published by Elsevier Ltd. ThisLtd. is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of GHGT-13. Peer-review under responsibility of the organizing committee of GHGT-13. Keywords: Aquistore; CO2 storage; modeling; simulation; history match
* Corresponding author. Tel.:+1-701-777-5343; fax: +1-701-777-5181. E-mail address: [email protected]
1876-6102 © 2017 T. Jiang Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of GHGT-13. doi:10.1016/j.egypro.2017.03.1630
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
1. Introduction The Aquistore research and monitoring project serves as a storage site for the world’s first commercial postcombustion carbon capture, utilization, and storage project from a coal-fired power generation facility. The Aquistore project commenced CO2 injection and monitoring operations in April 2015 at a site located near SaskPower’s Boundary Dam Power Station. The Energy & Environmental Research Center through its Plains CO2 Reduction (PCOR) Partnership, is collaborating with the Petroleum Technology Research Centre (PTRC) and SaskPower to continue PCOR Partnership efforts to numerically model and interpret the performance of the project’s injection well and observation well in order to improve the MVA (monitoring, verification, and accounting) strategies based on the history matching and predictive simulation results. The Aquistore site is located 2.8 km from Boundary Dam Power Plant. The deep saline system targeted for the Aquistore project comprises the Black Island and Deadwood Formations, the deepest sedimentary units in the Williston Basin. A previous publication [1] described the creation of a static geologic model of the project area, construction of a dynamic simulation model, and initial forecasts of reservoir performance. However, the initial forecasts did not match the field injection performance. This appears to have been caused at least in part by complex behavior related to near-wellbore damage incurred during initial completion testing. Thus, a new attempt has been made to match site performance using a more simplified model, with averaged formation physical properties obtained from the previous static geologic model. This paper describes the performance and interpretation of the injection and observation wells since the start of injection, utilizing the simplified dynamic simulation model and GEM—a full compositional reservoir simulator from Computer Modelling Group Ltd. (CMG). At the end of July 2016, approximately 74,000 tonnes of CO2 had been injected during a series of operating periods. However, the injection rate has been periodically increased to a level of 350 to 550 tonnes per day since November 20, 2015, based on the quantity of CO2 made available from the pipeline. The majority of CO2 captured by the SaskPower carbon capture and storage facility is delivered for enhanced oil recovery operations. The current data set is well replicated by the simulation model with applied skin factor and adjusted local permeability reduction near the injection well bore, which provides insight into well performance and will improve the MVA strategies based on the predictive simulation results. 2. Simulation model development The simplified simulation model is based on the previous simulation model of 34 square kilometers around the observation and injection wells (Fig. 1) [2]. The new simulation model has smaller areal size of about 11 square kilometers, and an attached aquifer boundary with leakage to eliminate pressure buildup during simulation (Fig. 2). Using a tartan grid system judiciously for the area where the injection and observation wells are located reduced the total number of cells. Those efforts maintain the infinite-acting behavior of the reservoir, enable computational efficiency, and keep good resolution of the area of interest. The fluid model includes two components, CO2 and brine. The CO2 is allowed to dissolve into brine to mimic the nature of the saline system undergoing CO2 injection. Correlations from Rowe and Chow [3] and Kestin and others [4] were used for the density and viscosity, respectively, of the aqueous fluids. Solubility of CO2 in water is modeled via Harvey’s correlation for Henry’s Law constants [5]. The rock–fluid models used in the simulation model were based on the lithology of the previous static geologic model. Figure 3 shows the three sets of relative permeability curves that were used in the simulations, including RPT 1 and RPT 2 obtained by Bennion and Bachu [6] and RPT 3 measured by Schlumberger Reservoir Laboratories [7]. RPT 1 and RPT 2 were used for perforated zones, and RPT 3 was used for the zones between perforations. For the near-real-time simulation, improper numerical settings would result in computational inefficiency and error and cause small time steps. Thus, several numerical parameters, including maximum pressure change, convergence tolerance, and maximum Newton iterations, were tuned utilizing optimization workflow software CMOST from CMG, to produce the lowest optimized critical point, which is a function of total central processing unit (CPU) run time, material balance error, and solver failure percentage. The global objective function and total CPU run time were reduced up to 40% after the numerical tuning and optimized for simulation runs with eight cores; increasing the core
4887
4888
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 1. Previous static model with local grid refinement [2].
Fig. 2. Close-up image of the simulation model with a tartan grid system constructed for history matching. Small grid cells near the wells are 7.62 square meters; the largest grid cells are 228.6 square meters. 120 layers are used to simulate the 232-meter-thick interval of interest.
number would not improve the elapsed time significantly. In addition, another sensitivity analysis workflow was adopted to determine the key parameters that affect the results of history matching. Parameters such as skin factor, transmissibility, and ratio of horizontal to vertical permeability were chosen using a response surface proxy model to determine the parameters that significantly affect the simulation results. Sensitivity analysis results are shown in Figure 4. Skin factor and transmissibility both greatly affect the injector bottomhole pressure response and injectivity [8].
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 3. Three sets of relative permeability curves used for the simulation [6 and 7].
Fig. 4. Sensitivity analysis of injection well pressure response to chosen parameters [8].
3. Simulation results and discussion The Aquistore field pressure and rate data are provided by SaskPower and PTRC. The field pressure, rate, and downhole temperature data are shown in Figure 5. Because of a short-term data-reporting system failure and scheduled power plant maintenance shutdowns, there are periods of no pressure data or when CO2 injection was stopped. History matching was conducted using average daily rate as the input constraint, with a maximum downhole pressure constraint set as 42.75 MPa. The CO2 travels an approximate 7-km pipeline route from the capture facility at the power
4889
4890
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 5. Field pressure, rate, and temperature data provided by SaskPower.
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
4891
plant to the injection well. At the wellhead, CO2 arrives at approximately ambient temperature but is slowly warmed by the geothermal gradient as it travels down the well. Reservoir temperature is 112°C, but depending on the injection rate, corresponding residence time in the well, and duration of the injection period, CO2 arriving at the storage formation depth may be at a considerably lower temperature. For a given injection rate, the well may take several days to reach thermal equilibrium; thus, the bottomhole injection temperature is rarely stable, as indicated in Figure 5. Other important data regarding the injection flow profile were provided by spinner log surveys, such as those shown in Figure 6, conducted by Schlumberger Carbon Services on April 20, 2015 [9]. Figure 6 indicates that ~40% to ~50% of total flow was observed at both Perforation Interval 2 and Interval 4, while Perforation Interval 1 took only ~10% of the flow, and there was no injection into Perforation Interval 3. The spinner log data were further used to adjust local permeability in order to match both the pressure response and the injection flow profile. The injection well pressure response between the simulation and history data would be mismatched without adding an additional skin factor into the simulation model for designated dates. The effect of this is demonstrated in Figure 7. Elimination of the skin value (red solid line) resulted in a significant reduction of injection pressure response during the CO2 injection period compared with history-matched simulation results (blue solid line). This work notably improved the history match result not only with respect to injection well pressure but also to flow distribution (Figs. 8 and 9.) The overall simulation results of pressure response match the field data well when an additional skin factor value was applied to the injector at selected time steps. This dynamic skin change was required mainly at the time when the injection rate had a sudden increase or decrease or simply when the well resumed injection after a shut-in. The mechanism causing this behavior remains unproven. A cross section and a 3-D image of CO2 plume evolution based on the history-matched simulation are shown in Figure 10. The results match the spinner log data described earlier, and it can be seen that the simulated CO2 plume
Fig. 6. Production log data with CO2 injection rate of 10 tonnes/hr [9].
4892
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 7. Injection pressure response, with or without additional skin value for the first 4-month simulation. BHP refers to bottomhole pressure.
Fig. 8. History match of injection well pressure response.
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 9. Cumulative CO2 injected into each perforated interval (unit in thousands of standard cubic meters). SCTR means at standard condition.
Fig. 10. CO2 plume evolution in February 2016 [10].
was forecast to approach the observation well in February 2016 [10]. New pulsed-neutron log data collected in February confirmed the CO2 breakthrough in the second perforation interval, upper Deadwood Formation. However, the log data did not show CO2 saturation in the fourth perforation interval. Similarly, a geophysical image processed from 4-D seismic data collected in February shows a clear CO2 plume anomaly at the level of the second perforation interval (but no comparable anomaly for the fourth interval) [11]. The simulated plume is circular around the well. This is because the simulation model used the average physical properties obtained from the geocellular model and distributed them homogeneously, while geologic conditions may exist in the field that cause permeability or other properties to be distributed heterogeneously, which can have a significant impact on CO2 plume evolution. Construction of a new geologic and simulation model using seismic data inversion is ongoing. The new model will be
4893
4894
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
better able to evaluate the CO2 plume evolution and geological uncertainty. The reporting of project performance based on the new geologic model may be presented in the future. 4. Conclusions History matching of the Aquistore injection monitoring data has been performed using a simplified simulation model. The current data set, including unusual injectivity performance, is well replicated by the simulation. The bottomhole temperature has a significant effect on the injectivity index, as a higher index was correlated with a lower bottomhole temperature range. The thermal effect should be further investigated to clearly identify the mechanism causing this correlation and confirm how it will affect the well performance during long-term CO2 sequestration operations. The geophysical imaging of 4-D seismic data shows a clear CO2 plume anomaly. The simulation results of plume size and depth compare well with the geophysical interpretation, but because of geologic structure and apparent heterogeneous property distribution, the shape and position of the simulated plume is different. Construction of a new geologic model is ongoing using seismic data inversion techniques. Additional performance confirmations will be made as the Aquistore project continues, which will greatly assist future history-matching efforts and refine our understanding of the injection site performance. Acknowledgements This work was performed under DOE NETL Cooperative Agreement No. DE-FC26-05NT42592. The authors would like to thank Schlumberger and Computer Modelling Group, Inc. (CMG) for allowing the use of their software packages in support of this work. This paper was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government, nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. References [1] Peck W, Bailey T, Liu G, Klenner R, Gorecki C, Ayash S, Steadman E, Harju J. Model development of the Aquistore CO2 storage project. Energy Procedia 2014;63:3723–34. [2] Liu G, Gorecki C, Bailey T, Peck W, Steadman E. Geologic modeling and simulation report for the Aquistore project: Plains CO2 Reduction (PCOR) Partnership Phase III, Task 1 – Deliverable D93 Update. Technical report to U.S. Department of Energy National Energy Technology Laboratory; September 2014. [3] Rowe A, Chou J. Pressure–volume–temperature-concentration relation of aqueous NaCl solutions. Journal of Chemical Engineering Data 1970;15(1):61–6. [4] Kestin J, Khalifa H, Correia R. Tables of dynamic and kinematic viscosity of aqueous NaCI solutions in the temperature range 20°–150°C and the pressure range 0.1–35 MPa. Journal of Physical and Chemical Reference Data 1981;10:71–87. [5] Harvey A. Semiempirical correlation for Henry’s constants over large temperature ranges. American Institute of Chemical Engineers Journal 1996;42:1491. [6] Bennion B, Bachu S. Relative permeability characteristics for supercritical CO2 displacing water in a variety of potential sequestration zones. SPE Annual Technical Conference and Exhibition, Dallas, Texas; October 9–12, 2005. [7] Schlumberger Reservoir Laboratories. Relative permeability by unsteady-state method. Prepared for Petroleum Technology Research Centre, Well: 5-6-2-8, Regina, Saskatchewan; 2013. [8] Jiang T, Pekot L, Jin L, Peck W, Gorecki C. Geologic modeling and simulation report for the Aquistore project: Plains CO2 Reduction (PCOR) Partnership Phase III, Task 1 – Deliverable D93 Update 2. Technical report to U.S. Department of Energy National Energy Technology Laboratory; February 2016.
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895 [9] Schlumberger Carbon Services. PTRC INJ 05-06-02-08 production log. Technical memorandum to SaskPower; May 2015. [10] Pekot L, Jiang T. An update of Aquistore CO2 storage simulation. Presentation at CMG’s 37th Technical Symposium, Calgary, Alberta; June 13–14, 2016. [11] White D, Roach L. First 3D time-lapse seismic results from the Aquistore CO2 storage project. Presentation at Aquistore Annual General Meeting in Ottawa, Ontario; August 16–17, 2016.
4895
ScienceDirect Energy Procedia 114 (2017) 4886 – 4895
13th International Conference on Greenhouse Gas Control Technologies, GHGT-13, 14-18 November 2016, Lausanne, Switzerland
Numerical modeling of the Aquistore CO2 storage project Tao Jianga *, Lawrence J. Pekota, Lu Jina, Wesley D. Pecka, Charles D. Goreckia, and Kyle Worthb a
Energy & Environmental Research Center, 15 North 23rd Street, Stop 9018, Grand Forks, ND 58202-9018,USA b Petroleum Technology Research Centre, 6 Research Drive, Regina, SK S4S 7J7, Canada
Abstract Serving as a storage site for the world’s first commercial postcombustion carbon capture, utilization, and storage project from a coal-fired power generation facility, the Aquistore commenced injection operations in April 2015 at a site near SaskPower’s Boundary Dam Power Station. The Energy & Environmental Research Center through its Plains CO2 Reduction (PCOR) Partnership, is collaborating with the Petroleum Technology Research Centre and SaskPower to continue PCOR Partnership efforts to numerically model and interpret the performance of the project’s injection well and observation well to improve monitoring, verification, and accounting (MVA) strategies based on history matching and predictive simulation results. ©2017 2017T.The Authors. Published by Elsevier © Jiang Published by Elsevier Ltd. ThisLtd. is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of GHGT-13. Peer-review under responsibility of the organizing committee of GHGT-13. Keywords: Aquistore; CO2 storage; modeling; simulation; history match
* Corresponding author. Tel.:+1-701-777-5343; fax: +1-701-777-5181. E-mail address: [email protected]
1876-6102 © 2017 T. Jiang Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of GHGT-13. doi:10.1016/j.egypro.2017.03.1630
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
1. Introduction The Aquistore research and monitoring project serves as a storage site for the world’s first commercial postcombustion carbon capture, utilization, and storage project from a coal-fired power generation facility. The Aquistore project commenced CO2 injection and monitoring operations in April 2015 at a site located near SaskPower’s Boundary Dam Power Station. The Energy & Environmental Research Center through its Plains CO2 Reduction (PCOR) Partnership, is collaborating with the Petroleum Technology Research Centre (PTRC) and SaskPower to continue PCOR Partnership efforts to numerically model and interpret the performance of the project’s injection well and observation well in order to improve the MVA (monitoring, verification, and accounting) strategies based on the history matching and predictive simulation results. The Aquistore site is located 2.8 km from Boundary Dam Power Plant. The deep saline system targeted for the Aquistore project comprises the Black Island and Deadwood Formations, the deepest sedimentary units in the Williston Basin. A previous publication [1] described the creation of a static geologic model of the project area, construction of a dynamic simulation model, and initial forecasts of reservoir performance. However, the initial forecasts did not match the field injection performance. This appears to have been caused at least in part by complex behavior related to near-wellbore damage incurred during initial completion testing. Thus, a new attempt has been made to match site performance using a more simplified model, with averaged formation physical properties obtained from the previous static geologic model. This paper describes the performance and interpretation of the injection and observation wells since the start of injection, utilizing the simplified dynamic simulation model and GEM—a full compositional reservoir simulator from Computer Modelling Group Ltd. (CMG). At the end of July 2016, approximately 74,000 tonnes of CO2 had been injected during a series of operating periods. However, the injection rate has been periodically increased to a level of 350 to 550 tonnes per day since November 20, 2015, based on the quantity of CO2 made available from the pipeline. The majority of CO2 captured by the SaskPower carbon capture and storage facility is delivered for enhanced oil recovery operations. The current data set is well replicated by the simulation model with applied skin factor and adjusted local permeability reduction near the injection well bore, which provides insight into well performance and will improve the MVA strategies based on the predictive simulation results. 2. Simulation model development The simplified simulation model is based on the previous simulation model of 34 square kilometers around the observation and injection wells (Fig. 1) [2]. The new simulation model has smaller areal size of about 11 square kilometers, and an attached aquifer boundary with leakage to eliminate pressure buildup during simulation (Fig. 2). Using a tartan grid system judiciously for the area where the injection and observation wells are located reduced the total number of cells. Those efforts maintain the infinite-acting behavior of the reservoir, enable computational efficiency, and keep good resolution of the area of interest. The fluid model includes two components, CO2 and brine. The CO2 is allowed to dissolve into brine to mimic the nature of the saline system undergoing CO2 injection. Correlations from Rowe and Chow [3] and Kestin and others [4] were used for the density and viscosity, respectively, of the aqueous fluids. Solubility of CO2 in water is modeled via Harvey’s correlation for Henry’s Law constants [5]. The rock–fluid models used in the simulation model were based on the lithology of the previous static geologic model. Figure 3 shows the three sets of relative permeability curves that were used in the simulations, including RPT 1 and RPT 2 obtained by Bennion and Bachu [6] and RPT 3 measured by Schlumberger Reservoir Laboratories [7]. RPT 1 and RPT 2 were used for perforated zones, and RPT 3 was used for the zones between perforations. For the near-real-time simulation, improper numerical settings would result in computational inefficiency and error and cause small time steps. Thus, several numerical parameters, including maximum pressure change, convergence tolerance, and maximum Newton iterations, were tuned utilizing optimization workflow software CMOST from CMG, to produce the lowest optimized critical point, which is a function of total central processing unit (CPU) run time, material balance error, and solver failure percentage. The global objective function and total CPU run time were reduced up to 40% after the numerical tuning and optimized for simulation runs with eight cores; increasing the core
4887
4888
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 1. Previous static model with local grid refinement [2].
Fig. 2. Close-up image of the simulation model with a tartan grid system constructed for history matching. Small grid cells near the wells are 7.62 square meters; the largest grid cells are 228.6 square meters. 120 layers are used to simulate the 232-meter-thick interval of interest.
number would not improve the elapsed time significantly. In addition, another sensitivity analysis workflow was adopted to determine the key parameters that affect the results of history matching. Parameters such as skin factor, transmissibility, and ratio of horizontal to vertical permeability were chosen using a response surface proxy model to determine the parameters that significantly affect the simulation results. Sensitivity analysis results are shown in Figure 4. Skin factor and transmissibility both greatly affect the injector bottomhole pressure response and injectivity [8].
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 3. Three sets of relative permeability curves used for the simulation [6 and 7].
Fig. 4. Sensitivity analysis of injection well pressure response to chosen parameters [8].
3. Simulation results and discussion The Aquistore field pressure and rate data are provided by SaskPower and PTRC. The field pressure, rate, and downhole temperature data are shown in Figure 5. Because of a short-term data-reporting system failure and scheduled power plant maintenance shutdowns, there are periods of no pressure data or when CO2 injection was stopped. History matching was conducted using average daily rate as the input constraint, with a maximum downhole pressure constraint set as 42.75 MPa. The CO2 travels an approximate 7-km pipeline route from the capture facility at the power
4889
4890
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 5. Field pressure, rate, and temperature data provided by SaskPower.
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
4891
plant to the injection well. At the wellhead, CO2 arrives at approximately ambient temperature but is slowly warmed by the geothermal gradient as it travels down the well. Reservoir temperature is 112°C, but depending on the injection rate, corresponding residence time in the well, and duration of the injection period, CO2 arriving at the storage formation depth may be at a considerably lower temperature. For a given injection rate, the well may take several days to reach thermal equilibrium; thus, the bottomhole injection temperature is rarely stable, as indicated in Figure 5. Other important data regarding the injection flow profile were provided by spinner log surveys, such as those shown in Figure 6, conducted by Schlumberger Carbon Services on April 20, 2015 [9]. Figure 6 indicates that ~40% to ~50% of total flow was observed at both Perforation Interval 2 and Interval 4, while Perforation Interval 1 took only ~10% of the flow, and there was no injection into Perforation Interval 3. The spinner log data were further used to adjust local permeability in order to match both the pressure response and the injection flow profile. The injection well pressure response between the simulation and history data would be mismatched without adding an additional skin factor into the simulation model for designated dates. The effect of this is demonstrated in Figure 7. Elimination of the skin value (red solid line) resulted in a significant reduction of injection pressure response during the CO2 injection period compared with history-matched simulation results (blue solid line). This work notably improved the history match result not only with respect to injection well pressure but also to flow distribution (Figs. 8 and 9.) The overall simulation results of pressure response match the field data well when an additional skin factor value was applied to the injector at selected time steps. This dynamic skin change was required mainly at the time when the injection rate had a sudden increase or decrease or simply when the well resumed injection after a shut-in. The mechanism causing this behavior remains unproven. A cross section and a 3-D image of CO2 plume evolution based on the history-matched simulation are shown in Figure 10. The results match the spinner log data described earlier, and it can be seen that the simulated CO2 plume
Fig. 6. Production log data with CO2 injection rate of 10 tonnes/hr [9].
4892
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 7. Injection pressure response, with or without additional skin value for the first 4-month simulation. BHP refers to bottomhole pressure.
Fig. 8. History match of injection well pressure response.
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
Fig. 9. Cumulative CO2 injected into each perforated interval (unit in thousands of standard cubic meters). SCTR means at standard condition.
Fig. 10. CO2 plume evolution in February 2016 [10].
was forecast to approach the observation well in February 2016 [10]. New pulsed-neutron log data collected in February confirmed the CO2 breakthrough in the second perforation interval, upper Deadwood Formation. However, the log data did not show CO2 saturation in the fourth perforation interval. Similarly, a geophysical image processed from 4-D seismic data collected in February shows a clear CO2 plume anomaly at the level of the second perforation interval (but no comparable anomaly for the fourth interval) [11]. The simulated plume is circular around the well. This is because the simulation model used the average physical properties obtained from the geocellular model and distributed them homogeneously, while geologic conditions may exist in the field that cause permeability or other properties to be distributed heterogeneously, which can have a significant impact on CO2 plume evolution. Construction of a new geologic and simulation model using seismic data inversion is ongoing. The new model will be
4893
4894
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895
better able to evaluate the CO2 plume evolution and geological uncertainty. The reporting of project performance based on the new geologic model may be presented in the future. 4. Conclusions History matching of the Aquistore injection monitoring data has been performed using a simplified simulation model. The current data set, including unusual injectivity performance, is well replicated by the simulation. The bottomhole temperature has a significant effect on the injectivity index, as a higher index was correlated with a lower bottomhole temperature range. The thermal effect should be further investigated to clearly identify the mechanism causing this correlation and confirm how it will affect the well performance during long-term CO2 sequestration operations. The geophysical imaging of 4-D seismic data shows a clear CO2 plume anomaly. The simulation results of plume size and depth compare well with the geophysical interpretation, but because of geologic structure and apparent heterogeneous property distribution, the shape and position of the simulated plume is different. Construction of a new geologic model is ongoing using seismic data inversion techniques. Additional performance confirmations will be made as the Aquistore project continues, which will greatly assist future history-matching efforts and refine our understanding of the injection site performance. Acknowledgements This work was performed under DOE NETL Cooperative Agreement No. DE-FC26-05NT42592. The authors would like to thank Schlumberger and Computer Modelling Group, Inc. (CMG) for allowing the use of their software packages in support of this work. This paper was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government, nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. References [1] Peck W, Bailey T, Liu G, Klenner R, Gorecki C, Ayash S, Steadman E, Harju J. Model development of the Aquistore CO2 storage project. Energy Procedia 2014;63:3723–34. [2] Liu G, Gorecki C, Bailey T, Peck W, Steadman E. Geologic modeling and simulation report for the Aquistore project: Plains CO2 Reduction (PCOR) Partnership Phase III, Task 1 – Deliverable D93 Update. Technical report to U.S. Department of Energy National Energy Technology Laboratory; September 2014. [3] Rowe A, Chou J. Pressure–volume–temperature-concentration relation of aqueous NaCl solutions. Journal of Chemical Engineering Data 1970;15(1):61–6. [4] Kestin J, Khalifa H, Correia R. Tables of dynamic and kinematic viscosity of aqueous NaCI solutions in the temperature range 20°–150°C and the pressure range 0.1–35 MPa. Journal of Physical and Chemical Reference Data 1981;10:71–87. [5] Harvey A. Semiempirical correlation for Henry’s constants over large temperature ranges. American Institute of Chemical Engineers Journal 1996;42:1491. [6] Bennion B, Bachu S. Relative permeability characteristics for supercritical CO2 displacing water in a variety of potential sequestration zones. SPE Annual Technical Conference and Exhibition, Dallas, Texas; October 9–12, 2005. [7] Schlumberger Reservoir Laboratories. Relative permeability by unsteady-state method. Prepared for Petroleum Technology Research Centre, Well: 5-6-2-8, Regina, Saskatchewan; 2013. [8] Jiang T, Pekot L, Jin L, Peck W, Gorecki C. Geologic modeling and simulation report for the Aquistore project: Plains CO2 Reduction (PCOR) Partnership Phase III, Task 1 – Deliverable D93 Update 2. Technical report to U.S. Department of Energy National Energy Technology Laboratory; February 2016.
Tao Jiang et al. / Energy Procedia 114 (2017) 4886 – 4895 [9] Schlumberger Carbon Services. PTRC INJ 05-06-02-08 production log. Technical memorandum to SaskPower; May 2015. [10] Pekot L, Jiang T. An update of Aquistore CO2 storage simulation. Presentation at CMG’s 37th Technical Symposium, Calgary, Alberta; June 13–14, 2016. [11] White D, Roach L. First 3D time-lapse seismic results from the Aquistore CO2 storage project. Presentation at Aquistore Annual General Meeting in Ottawa, Ontario; August 16–17, 2016.
4895