Characterization of a Geothermal Reservoir in Denmark based on Seismic Inversion Results

9 Oct 2019, 11:55
KIT - AudiMax (Karlsruhe)

KIT - AudiMax


Karlsruhe Institute of Technology South Campus Forum-Hörsaal AudiMax, Blg. 30.95 Strasse am Forum 1 76131 Karlsruhe Germany
Oral Topic 2: Exploration of Geothermal Reservoirs Session 2: Exploration of Geothermal Reservoirs


Dr. Runhai Feng (Department of Geoscience, Aarhus University)


The geophysical characterization of reservoir properties, such as lithofacies, porosity and other petrophysical variables, is essential for the exploitation of subsurface reservoirs. Indeed, reservoir lithofacies classification provides significant information about the petrophysical behavior of reservoir rocks and their degree of compartmentalization (Bosch et al., 2002). Furthermore, the porosity of reservoir rocks determines the storage capacity of energy resources such as hydrocarbon and water (Keelan, 1982). Here, we present a methodology to infer lithofacies and porosity in an onshore geothermal reservoir located at the north of Copenhagen, Denmark, based on seismic attributes obtained from seismic inversion. As for the majority of Danish deep basins, Lower Jurassic – Upper Triassic potential geothermal reservoirs are found (Mathiesen et al., 2010; Poulsen et al., 2017; Røgen et al., 2015).
The inversion of seismic data provides acoustic impedance (AI) and the ratio between P-wave and S-wave velocity (Vp/Vs). The goal of this method is to predict the lithofacies distribution from these seismic attributes. To account for the dependency in the lithofacies along the vertical direction, a first-order stationary Markov chain is introduced. The relationship between seismic variables (AI and Vp/Vs) and reservoir parameters (lithofacies and porosity) is modeled using an Artificial Neural Networks (ANN), to overcome the non-linearity of the relation and the non-Gaussian distributions. ANN is a machine learning approach based on weights and biases in the synaptic nodes to mimic the behaviors of neurons and is suitable to model complex unknown physical relations.
Three different lithofacies are defined based on observations in nearby wells: sand, shaly-sand and shale. Sand is characterized by high AI and low Vp/Vs, whereas shale exhibits the opposite behavior. First lithofacies are classified based on elastic attributes; then, porosity is predicted using the classified lithofacies as constraints. The classification is based on an integrated method combining ANN and HMM (ANN-HMM) using the seismic inversion results as input. The ANN-HMM algorithm is trained using data from well logs and core samples. The integrated ANN-HMM approach provides better results compared to ANN and HMM, applied separately. In particular, the classification results show an improved continuity in the sand layer between the lower Jurassic reservoir unit and the Gassum Formation. The integrated approach also reduces the non-uniqueness. The results have been compared to those obtained from a standard approach where porosity is estimated without the lithofacies constrained; in the proposed approach, the predicted porosity is generally more accurate. However, the correct prediction of porosity strongly depends on the accuracy of the lithofacies prediction. Because the Markov model only accounts for transitions in the vertical direction, a future research direction will focus on the horizontal correlations between lithofacies (Feng et al., 2018).

Bosch, M., M. Zamora, and W. Utama, 2002, Lithology discrimination from physical rock properties: Geophysics, 67, no. 2, P573-P581, doi: 10.1190/1.1468618.
Feng, R., S. M. Luthi, D. Gisolf, and E. Angerer, 2018, Reservoir Lithology Determination by Hidden Markov Random Fields Based on a Gaussian Mixture Model, IEEE Transactions on Geoscience and Remote Sensing, 56, no. 11, 6663 – 6673.
Keelan, D. K., 1982, Core analysis for aid in reservoir description, Journal of Petroleum Technology, 34, 2483 – 2491, doi: 10.2118/10011-PA.
Mathiesen, A., L. H. Nielsen, and T. Bidstrup, 2010, Identifying potential geothermal reservoirs in Denmark, Geological Survey of Denmark and Greenland Bulletin, 20, pp. 19 – 22.
Poulsen, S. E., N. Balling, T. S. Bording, A. Mathiesen, and S. B. Nielsen, 2017, Inverse geothermal modelling applied to Danish sedimentary basins, Geophysical Journal International, 211, no. 1, 188 – 206.
Røgen, B., Ditlefsen, C., Vangkilde-Pedersen, T., Nielsen, L.H. and Mahler, A., 2015, Geothermal energy use, 2015 country update for Denmark. Proceedings, World Geothermal Congress 2015, Melbourne, Australia, 9 pp.

Primary author

Dr. Runhai Feng (Department of Geoscience, Aarhus University)


Prof. Niels Balling (Department of Geoscience, Aarhus University) Prof. Dario Grana (Department of Geology and Geophysics, University of Wyoming)

Presentation Materials

There are no materials yet.
Your browser is out of date!

Update your browser to view this website correctly. Update my browser now