Dual Porosity Numerical Models of Geothermal Reservoirs

Abstract

The aim of the research described in this thesis was to determine when dual porosity models should be preferred ahead of single porosity models for modelling geothermal systems. In a single porosity model the reservoir rock is represented as a uniform porous material whereas for a dual porosity model the reservoir rock is represented as a dual continuum with high permeability, low volume fractures running through a low permeability matrix. Both dual porosity and single porosity models were applied for investigating the following geothermal reservoir engineering problems: (i) constant rate pressure drawdown/build-up test in a two-phase reservoir (ii) constant rate cold water injection into a well in a two-phase reservoir; (iii) discharge test at Fushime; (iv) injection tests at Bacman; and (v) large scale 3D model of a geothermal system at Mt. Apo in the Philippines. For drawdown, the dual porosity models with its small volume fracture blocks induced more vigorous boiling, higher steam saturations, and generated more excess enthalpy than the single porosity model. For injection, the dual porosity models showed faster thermal front propagation characterised by t1/4 behaviour in the pressure vs. time plot compared to t1/2 behaviour for single porosity. For the discharge test at Fushime and the injection test at Bacman, the dual porosity models gave better matches to the data than the single porosity model. Inverse modelling was used to obtain the best-fit model to the field data. For the Mt. Apo model, the combined inversion of the single porosity natural-state model (NS) and single porosity production history (SP) model gave the lowest value for the objective function. Certainly a DP model can be calibrated to produce a better match to production history data than a SP model but only at the expense of the fit to the NS data. The two sets of sequentially calibrated models (NS+SP vs. NS+DP) give a similar fit to the data with the NS+DP combination only slightly better than NS+SP. Although the joint calibration of the NS+DP combination was not successful it is likely that it would not greatly improve the results from the jointly calibrated NS+SP