Dynamic Model and Report

Key Knowledge Document   |    NS051-SS-REP-000-00015

This technical report focuses on the Northern Endurance Partnership’s approach to building a subsurface dynamic model for CO2 sequestration in the Endurance structure in the southern North Sea.

What is a dynamic model and why is it important?

The static or geological model is a three-dimensional representation of the subsurface. The model contains physical characteristics which describe the subsurface. These characteristics include the thickness, porosity, and the permeability of the reservoir, rock composition and other properties depending on the available data. The geological model aims to show how these properties vary spatially within the subsurface. A dynamic model uses the static geological model as a foundation and introduces engineering parameters which allow flow, or movement of substances to be simulated in the model.

Reservoir dynamic modelling has been used widely in petroleum engineering for many decades and is important for solving complex subsurface engineering problems associated with extracting hydrocarbons. The modelling for injection of CO2 into the reservoir is built on similar theory and methodology. However, there is an important difference, CO2 is injected rather than extracted as in the case of hydrocarbons. The injection generally causes a pressure increase within the reservoir rather than a pressure decrease. The fluid properties in the model will be based on CO2 and water rather than oil-gas and water.

For CO2 sequestration projects such as Endurance, it is important to understand how the injected CO2 displaces the original pore water and moves from the injection well through the subsurface. This is one example of the use of a dynamic model. The size and extent of the CO2 plume can be modelled over time which is critical in determining the efficiency of the project and also for developing a monitoring plan to verify the location of the CO2 stored in the subsurface.

If there are sufficient data, modelling the dynamics of other properties can be done. These include heat flow, pressure, CO2 solubility and many other properties that change over time. For instance, injecting fluid CO2 into a reservoir already saturated with water will cause the pressure to increase in the reservoir. The magnitude of the pressure increase, how it changes away from the well bore and how quickly it changes can be modelled in the dynamic model. The models allow different scenarios to be tested within the computer simulation before injecting any CO2 in the field. The model allows different scenarios of CO2 injection and may help identify any problems that may arise from the injection. For example, we can model the injection of 1 MT of CO2 per year into one hypothetical well. Would the reservoir properties allow that volume of CO2 to be injected safely or would there be insufficient permeability causing a potential unsafe pressure build-up? If the pressure build-up was too high, could two wells be drilled and 0.5 MT/year be safely disposed of in each well?  There is a limit to how much pressure increase a reservoir can safely withstand which ultimately limits the amount of CO2 which can be injected. To further increase the CO2 storage capacity of a reservoir, it is possible to extract the original pore water thereby increasing space for CO2 storage. The original pore water is extracted using separate water extraction wells located away from the injector wells. However, the produced saline water must then be disposed of safely. This concept of pressure management or dewatering can be examined through the use of dynamic modelling.  

Dynamic models can be very useful for testing uncertainties as input variables can easily be changed with numerous scenarios for establishing a range of outcomes. These models can also be used to highlight which data need to be collected to decrease the risk and uncertainty of the models. As with the underlying static models, dynamic models should be continuously updated throughout the life of the project as new information becomes available.

This report relates to other key documents as reviewed here. The Primary Store Geological Model and Report presents a field scale geological model which is built incorporating the geophysical data (The Primary Store Geophysical Model and Report). Then, the geological model is integrated into the dynamic model with a series of engineering data, e.g., fluid properties, relative permeabilities, well design information. Finally, the geomechanical model (The Primary Store Geomechanical Model and Report) is run based on the results of the dynamic model to validate the containment of the stored CO2 and assess the possibilities of rock failure of the reservoir and overlying sealing formations during injection.

Technical summary

In this document BP summarizes the work and methodology for conducting the dynamic model of the Bunter sandstone reservoir in the Endurance anticline. After integrating the geological model, detailed fluid and rock properties and well completion information are entered to complete the initial setting of the dynamic reservoir model. Then, the model is run with a number of uncertainty factors, including: formation structure, segment transmissibility, porosity, permeability, aquifer connectivity, reservoir architecture, and displacement efficiency through relative permeability curves. This is done to evaluate the P10, P50, and P90 cases of CO2 storage capacity for the clustered and distributed development design. Lastly, the maximum CO2 storage capacity of Endurance has been evaluated with an increased number of CO2 injectors and surrounded brine producers. Integrating suitable pressure management through brine extraction, the amount of stored CO2 is estimated to be more than four times (approximately 450MT) than the base case (approximately 100MT) with no brine extraction.

Overall, the report outlines the industry-standard workflow of dynamic modelling, resulting in a very thorough and exemplary CO2 storage simulation. The uncertainty analysis is within reasonable engineering ranges and gives valuable insights of the lower and upper limits of the storage volumes, which lays out the foundation of the project feasibility.

Landmark Graphics reservoir simulator ‘Nexus’ is used for the dynamic modelling of CO2 storage. Other simulators for fluid properties, CO2 solubilities, thermal fracturing effects, and statistical uncertainty analysis are also used. For the dynamic modelling physical and chemical properties of the gas injected is assumed to be pure CO2 with no impurities. An equation of state PR78 (Peng Robinson EOS) and water PVT table with CO2 solubilities at different water salinities (Henry’s Law) were developed using the CMG WinProp module. The equation of state is needed to estimate the solubility of CO2 in water. Since the CO2 solubility in brine for Endurance is low due to the hypersaline condition in the reservoir, immiscible CO2 without solubility into brine has been mainly used for the dynamic modelling. Brine properties were obtained from samples from Well 42/25d-3 during drill stem testing. These samples indicate that there is a potential increase of brine salinity with depth, which could explain the relatively large pressure difference observed between Well 42/25d-3 and Well 42-25-1.

Reservoir energy is studied by incorporating the following properties: water compressibility (generated by REToolkit), rock compressibility, aquifer connectivity to the broader Bunter basin, and permeability contrast for areas of the Bunter sandstone where seismic phase reversal indicates lower permeability. How much of the Bunter sandstone outside the actual Endurance structure should be incorporated into a model and the methodology used to do this is discussed in the report. The broader Bunter basin aquifer is modelled both numerically through a pore volume multiplier at the edge and analytically through the Carter-Tracy model. The reservoir pressure responses for 4 MTPA CO2 injection in various extended aquifer models are compared using both methodologies. This sensitivity study can give flexibilities and options in the later uncertainty analysis with intensive simulation runs.

Displacement efficiency is studied through a series of CO2 – water relative permeability models based on measured values from Special Core Analysis (SCAL) tests on core plugs from well 42/25d-3. The ultimate residual water saturation (Swrg) has been corrected to capture the slowly changed CO2 and water saturations during post-injection due to gravity drainage and capillary pressure effects over geological time in comparison to the relatively short laboratory-based measurements.

Reservoir architecture is modelled by applying the different ratios of vertical and horizontal permeabilities, which are Kv/Kh = ~0.1 for good quality sandstone and lower values of Kv/Kh <0.01 which represent possible vertical baffles interpreted from drill stem test measurements.

Average injection rate is assumed to be 1 MTPA per well and a limitation for reservoir pressure not to exceed approximately 200 bars fracture pressure at the crest over the 25-year life of the project. This is based on benchmarking against analogous offshore CCS projects such as Sleipner, Snohvit, and Northern Lights. It is found that no brine production would be required for Phase 1 of the project as CO2 injected volumes are not expected to exceed 100 MT over 25 years. Salt precipitation, mostly halite, is considered a significant risk to well injectivity over time for Endurance due to the high salinity of the brine and the creation of a dry-out zone in the near wellbore region. The water vaporization phenomenon during continuous CO2 injection is modelled using CMG GEM (Primary Store Geochemical Model and Report LINK). The injectivity loss becomes prominent at low injection rates. It is recommended to perform a pre-injection initial flush with fresh water to dilute high-salinity reservoir brine near the injectors followed by a one-to-two-day long freshwater flush per well per year to keep the integrity of well injectivity.

After test runs and calibrations, the fine-scale geological model is upscaled to a coarse-scale reservoir model. Numerous simulation runs are conducted through a Monte Carlo probabilistic workflow to understand effects of reservoir uncertainties on CO2 storage capacity for Endurance in the downside, base, and upside scenarios. A list of selective subsurface uncertainties has been reviewed:

  • Structural uncertainty: three distinct grids have been generated to represent different brine volumes above spill point.
  • Fault transmissibility: faults in the overburden and extended into the Bunter reservoir are modelled in segments with varied transmissibility multipliers.
  • Petrophysical uncertainty: global permeability and porosity multipliers are used to account for uncertainty in permeability prediction.
  • Aquifer connectivity: pore volume multipliers at the edge of the model are used to numerically represent different volume of the broader Bunter basin aquifer.
  • Reservoir architecture: three geologic models have been utilised to account for uncertainty in the extent and severity of the heterolithic-rich intervals.
  • Displacement efficiency: as reflected by the end points of gas (CO2)-water relative permeability curves.

The downside scenario does offer some degree of compartmentalization and limited connected aquifer. The upside scenario offers greater connectivity to the Bunter aquifer as well as improved vertical connectivity. A limited aquifer associated with some sub-seismic baffling will lead to rapid compartmentalization and therefore a requirement for active pressure management through brine production. On the other end of the spectrum, excellent rock properties for an extensive aquifer alongside favourable reservoir architecture (i.e. high Kv/Kh) would enhance the pressure dissipation and allow for longer injection periods without brine production. The pressurization of the structure at Endurance might also lead to the release of brine into the sea through the underwater Bunter outcrop 20 km east of Endurance. This is dependent on hydraulic communication with the Endurance structure. It is thought that the salinity of the water in the shallow depths of the outcrop will be similar to seawater assuming it is in static equilibrium with the seawater above the outcrop.

Ultimate storage capacity is impacted by well placement as a subsea development scheme can allow for the wells to be better distributed across the structure providing robust mitigation against any compartmentalization. Storage capacity for Endurance without brine management is at least 104 MT of CO2 for a distributed well layout for the 25 year-long project.

The injected CO2 is expected to reach 10-12 degrees centigrade at the bottom hole location, which indicates that injectors might be subject to thermal fracturing as the temperature of the reservoir is approximately 57 degrees centigrade. However, the cooled region is expected to be restricted to the near-wellbore region as shown by the results of CMG GEM with the thermal option turned on. The evaluation in REVEAL (geomechanics simulator coupled fluid flow) shows that the risk of vertical fracture growth is low with no cases presenting fracturing reaching the top of Bunter Formation by the end of CO2 injection. On the other hand, skin build-up associated with salt precipitation can be offset by thermal fractures.

The technical limit for Endurance is studied with the increased number of CO2 injectors and a suitable number of brine producers for pressure management. The maximum CO2 storage capacity of Endurance is estimated to be approximately 450 MT (25 years at 18 MTPA or 30 years at 15 MTPA). Storage tipping point is around 18 MTPA, above which injection rates cannot be maintained to 2050 (25 years) without CO2 breakthrough into brine producers. Before significant investment for brine production, the development case of 10 MTPA (14 CO2 injectors + 10 brine producers) is recommended to achieve the CO2 storage capacity of 400 – 450 MT. This may fall into Phase 2 of the project development.

The dynamic modelling is a vital part of a Monitoring, Measurement, and Verification (MMV) plan for the CO2 storage project at Endurance. The MMV plan is created to explain how the CO2 plume will be monitored and outline potential risks and mitigation strategies associated with CO2 disposal.

Technical comments for possible future work on the dynamic model

This study presents reasonable end members in terms of the overall system connectivity and its associated response when CO2 volumes are injected. It gives a solid guideline for the preliminary front-end engineering and design (pre-FEED). However, uncertainties can be reduced if more data can be obtained as the development continues.

The structural uncertainty can be reduced when more wells are drilled in the eastern side of the structure. Similarly, the petrophysical uncertainty can be reduced with more core data collection with any additional wells drilled. The acquisition of new, more detailed seismic data will allow better identification and quantification of the presence of any faults in the structure. More well tests can be conducted to have a narrower range of Kv/Kh ratio for reservoir architecture.

If the run time allows, it would be beneficial to incorporate near-wellbore permeability changes due to water vaporization in the field-scale model, which could affect the well injectivity and therefore storage capacity.

Although the expectation is not to see the thermal induced fractures in the vicinity of CO2 injectors reach the top of the Bunter Formation, it would be valuable to understand how the CO2 plume grows along the fractures. This work is recommended to be done in conjunction with further geomechanical work.

Document Information

Document name: Primary Store Dynamic Model & Report

Reference number: NS051-SS-REP-000-00015

Document length: 71 pages

Topic area: Dynamic model for CO2 sequestration

Project: Net Zero Teesside / Northern Endurance Partnership

Original report date: August 2021

Original Author: BP Exploration Operating Company

Link to all original reports

This website uses cookies to ensure you get the best experience on our website.