Stochastic inversion of tracer test data with seismicity constraint for permeability imaging in enhanced geothermal reservoirs

Abstract

Inference of permeability distribution in an enhanced geothermal reservoir is crucial for the sustainable management of geothermal energy. However, interpreting the uneven permeability distribution in a deep geothermal reservoir remains a challenging task because in practice there often are merely one or two wells available for hydrogeophysical tests. Considering that the induced seismicity data are widely captured during reservoir stimulation in enhanced geothermal systems, a framework of tracer test data inversion with the constraint of induced seismic data is developed for permeability imaging. Hydraulic diffusivity, representing the prior estimation of permeability, is inferred from the occurrence time of seismic events. This is followed by the determination of a petrophysical model, which relates hydraulic diffusivity to permeability, by tracer test data inversion based on the Markov chain Monte Carlo algorithm. Implementation of the seismicity-constraint tracer data inversion algorithm in the Habanero enhanced geothermal system, Australia, demonstrates that our inversion model allows uneven permeability estimation at field scale in shorter burn-in period and lower uncertainty than the traditional inversion model without seismicity constraint. Using the estimated permeability in the hydrothermal model enables accurate prediction of thermal performance in a 150 day trial-production test. Results indicate that this algorithm can reliably characterize the spatial distribution of permeability in deep enhanced reservoirs, based on the tracer test via doublet wells.