GeoConvention 2020 1

Geomechanics of underground gas storage: A case study from Molasse Basin South Germany Muhammad Zain Ul Abedin, Andreas Henk

Technical university Darmstadt, Germany

Summary

Underground gas storage is a multi-disciplinary approach which involves iterative simulations of fluid and geomechanical model as well as coupling of these simulations to evaluate the stress changes due to production and replenishment of working gas volume. The determination of maximum threshold pressure (maximum containing pressure of the storage reservoir) is essential to avoid fracturing and fault reactivation. History matching, material balance and production rate history of these reservoirs (porous media) are essential parameters to estimate the stress changes within and around the reservoir, well bore periphery as well as cap rock integrity. Coupled 3D geomechanical modeling can tackle these problems. This geomechanical assessment concentrates on a former gas field in the Bavarian Molasse Basin east of Munich for which a hypothetical transformation to underground gas storage (UGS) is considered. The workflow comprises 1D mechanical earth modeling (1D MEM) to calculate vital material properties and populate it on a 3D geomechanical model using Kriging interpolation method. This can then be used for geomechanical simulation to assess fault stability, cap rock integrity, stress perturbation and well bore stability analyses. The reservoir geometry was built up through seismic interpretation, thickness maps as well as through well data. The reservoir model is a faulted anticline structure located at a depth of about 1200 m with lateral dimensions of 4 km in N-S and 8 km in W-E direction, respectively. The pore pressure field has been derived from anfluid flow simulation through history matching for the production and subsequent shut-in phase.The 3D geomechanical model has been built by adding further stratigraphic sequences as over-and underburden rocks. This model is about 30 × 24 km2 reaching to a depth of about 5000 m.The material properties as well as a first estimate for the vertical and horizontal stresses at welllocations are derived using log data and 1D MEM, respectively. The calculated elasticproperties are calibrated with available core data and pore pressures for overburden andunderburden rocks were calculated by using Eaton’s method. The computed model provides astress state of the complete model and this data can be used for various applications such assurface subsidence analysis and stress state for a newly planned well.

Modeling workflow

The study applies simple 1D mechanical earth model to calculate initial minimum horizontal stress and important material parameters using well log data information. The second stage is to build up a 3D embedded model in order to incorporate 1D modeling results and couple it with geomechanical simulator (Visage) to simulate principal stresses (vertical and horizontal) in and outside of the reservoir. The stress orientations have also been calibrated with the stress data received from the World Stress Map. The complete modeling workflow has been illustrated in Figure 1.

GeoConvention 2020 2

Figure 1: Complete modeling workflow, starting with the construction of a geomechanical 3D model using a structural model (Petrel model) and a 1D-MEM (Techlog) and a coupling scenario for the hydro-mechanical coupling

The 1D MEM process starts with the overburden stress calculations by using density logs and creates a synthetic density curve that fills the gaps of bulk density curves. Both synthetic and bulk densities integrate the splice density to compute the overburden stress. The extrapolation density method has been used in order to fill air gap between bulk density logs in the computing process of overburden stress. The overburden stress is an essential parameter to calculate pore pressure throughout the log profile. The pore pressure and hydrostatic pressure were then calculated by using Eaton’s method. The method is standard in industry to estimate pore pressure using a semi-logarithmic or linear normal trend line. The fracture gradient was calculated by using two essential parameters: vertical stress and pore pressure. The Bradford correlation was used to compute dynamic material properties such as Shear modulus (dynamic), Bulk modulus (dynamic), Young’s modulus (dynamic) and Poisson ratio (dynamic). Then different available correlations allowed us to estimate static material properties calibrating the available core material properties. UCS was calculated by using Young’s modulus (static and dynamic), bulk modulus (dynamic), shear modulus (dynamic) as well as compressional slowness (DTCO). Tensile strength is a function of UCS and correspondingly calculated by using UCS. The workflow for building up the 1D-MEM is shown in Figure 2.

Figure 2: The workflow of building up the 1D-MEM for the calculation of material properties and horizontal stresses.

GeoConvention 2020 3

A 3D model (Figure 4) with dimension of 30×24×5 km3 with a total number of 368225 cells has been simulated successfully to get a first estimation of total as well as effective stresses within and outside reservoir section. The simulation run time on a local machine is about 1 hour for one complete geomechanical simulation. The results from (1D MEM) Techlog geomechanics have been used to do property modeling of 3D geomechanical model calibrating the available core data properties. The material properties (Young’s modulus, Poisson ratio, Biot’s coefficient, etc.) as well as a first estimate for the vertical and horizontal stresses at well locations are derived using log data. The krigging interpolation was used to populate the elastic properties on 3D geomechanical model. The calculated properties are calibrated with available core data and pore pressures for overburden and underburden rocks were calculated. The average elastic Young’s modulus and Poisson’s ratio are 4 GPa and 0.35, respectively, at C6 location at a depth of about 1786 m. The pore pressures for over and underburden formations were calculated. Based on these data an estimation of the vertical and horizontal stresses could be made. The workflow of the 3D modelling is shown in Figure 3.

Figure 3: The workflow for 3D geomechanical modeling, including input data sources, validation and calibration. Modified after Henk et al (2019).

GeoConvention 2020 4

Figure 4: Embedded geomechanical 3D model with reservoir and the overburden and underburden rock zones. (a) Top view (b) Oblique view from south-west.

Results, Observations, Conclusions

The orientations of the stresses were calibrated from the regional stress map i.e. SHmax in N-S

direction and subsequently Shmin in E-W direction. The stress magnitudes have been modified by adopting the boundary conditions from the 1D MEM model. The imposed boundary

conditions are modified strain value of minimum horizontal strain and is εh = 0.009 which was

0.0015 prior to calibrating. Consequently, the maximum horizontal strain value is εH = 0.0012 which was 0.0021 before adjusting.

Well C6 has been used for calibration of the simulation results, since it is the best available well for which maximum data is available. Additionally, further data for model calibration such as core, laboratory and field measurement data is available for calibration. The best fit model is considered as validated model in the given circumstances. The pore pressure field has been derived from a fluid flow (Eclipse) simulation through history matching for the production and subsequent shut-in phase. The coupling involves the pore pressure distribution throughout the production history of the reservoir, taken as input for only the reservoir section of the 3D geomechanical model from the fluid flow model.

The predicted vertical stress σv in well C6 at a depth of 1786 m is 42.8 MPa; the value is quite close to the observed value of 42.2 MPa. In the same manner estimated maximum horizontal SHmax stress at well C6 is about 38.7 MPa matching the observed value of 38.2 MPa at the

same depth. The minimum horizontal stress Shmin achieved from the simulation is about 36.7 MPa that is also vindicating the observed value of 36.7 MPa (Figure 5). It is quite clear that the

GeoConvention 2020 5

reservoir lies in the normal stress regime by analyzing the stress perturbations since 𝝈𝒗 (42.8

MPa) > SHmax (38.7 MPa) > Shmin (36.7 MPa). The maximum subsidence at the top of reservoir is observed to be -12 mm (negative sign represents negative subsidence) during depletion period which reduces towards the edge of the reservoir to about -8 mm on each side. During replenishment period reverse surface behavior has been observed which resulted in slight upheaval of 9 mm (relative to subsidence at t1) at the surface above reservoir (Figure 6).

Figure 5: Map view of the computed principal stresses in the reservoir. a=mmagnitude of the maximum horizontal stress (SHmax), b= orientation of SHmax, c= magnitude of the minimum horizontal stress (Shmin) and d= magnitude of the vertical

stress (𝝈𝒗).

Figure 6: Subsidence predicted at the surface above reservoir during depletion (t1) and replenishment (t2) relative to pre-production time (which was 0 mm displacement). The area in the box represents the reservoir area and rest of it is

sideburden. Color scale is in metres.

GeoConvention 2020 6

Novel/Additive Information

The coupled 3D model for the case study can be used in many different ways and not only allows testing of arbitrary injection and withdrawal scenarios. Rather, it also allows the prediction of all components of the stress tensor (total and effective) in the entire model space, i.e. not only in the reservoir, but also in the overburden and underburden rocks. The validated3D geomechanical model provides stress state at any location and well path in order to plan anew well for well planning stage. Furthermore the model is an open platform for short term(weekly) as well as long term (seasonal) gas storage scenario testing in future. This type ofstudy is unique in Molasse Basin in South Germany and would be applied on a real depletedreservoir in the area.

Acknowledgements

The authors acknowledge Federal Ministry of Education and Research (BMBF) for providing funding and financial support for the SUBI research project. We also thank Uniper SE for providing data for this research. The first author is especially grateful to Schlumberger Geomechanics Centre of Excellence at United Kingdom for technical support during one week software training.

References

Guerra C, Fischer K, Henk A, "Stress prediction using 1D and 3D geomechanical models of a tight gas reservoir-A case study from the Lower Magdalena Valley Basin, Colombia," Geomechanics for Energy and Environment, vol. 19, no. HTTP://doi.org/10.1016/j.gete.2019.01.002, 2019.

Henk A. , "Perspectives of geomechanical reservoir models-why stress is important," Oil Gas-Eur. Mag., vol. 35, pp. 18-22, 2009.

Henk A., "Geomechanical reservoir models- a case study from the Sirte Basin, Libya," Oil Gas-Eur. Mag., pp. 18-22, 2010.