Characterization of a carbonate geothermal reservoir using rock physics-guided deep neural networks

Deep neural networks are used to characterize the porosity and permeability of the Dogger formation north-east of Paris, France, that already hosts a number of geothermal plants and is set to become even more important with the transition toward renewable energies. Due to the limited availability of well data, the networks are trained on synthetic well data generated through a combination of theoretical rock physics models and statistical simulations. The networks are applied to 5 seismic lines from the 1980s and the output reservoir properties highlight several highly porous and permeable layers.