Modelling of Subsidence in Geothermal Fields

Abstract

Subsidence is an environmental issue at New Zealand geothermal fields, particularly Wairakei. It has resulted in submissions to geothermal resource consent hearings, subsequent operational consent conditions and long-term monitoring programs. Wairakei geothermal field has a complex subsidence history. It is an example of where withdrawal of geothermal fluid for generation of electricity and the presence of weak and soft highly compressible rocks have caused surface subsidence. For predicting subsidence, there was a need to develop coupled thermal-hydro-mechanical (THM) models. A hierarchy of 1-D, R-Z and 3-D THM models were set up using data and knowledge from the comprehensive investigation programme conducted by Contact Energy Ltd. This subsidence analysis was performed using a coupled approach where two codes (TOUGH2 and ABAQUS) were used to model subsidence. The TOUGH2 code is an established finite volume code that simulates complex multiphase, multi-component subsurface flows. It is used worldwide to model geothermal reservoirs. ABAQUS code is a special general purpose finite element code that can be used to provide solutions to geomechanical problems. It can handle 3-D problems for heterogeneous geological materials together with simple or complex constitutive modelling laws such as those applied in the study carried out here. Pressure and temperature information from the reservoir model over 50 years of production history are interpolated to provide input for the ABAQUS rock mechanics model where subsequent calculation of subsidence takes place. The subsidence model utilised both elastic and nonlinear elasto-plastic deformations and allowed for varied geological strata each characterised by distinct elastic constants and elasto-plastic parameters. Calibration of the reservoir model and geomechanics model was achieved by matching the models output to the measured pressure profiles and subsidence history at a few representative points. Automatic calibration with PEST was used in calibrating the R-Z model. PEST calibration software (Doherty, 2004) was integrated with the TOUGH2 and ABAQUS simulators to carry out a joint inversion of both reservoir and rock mechanics models together, after a reasonably good TOUGH2 model had been obtained. Using parameters determined from field and laboratory measurements together with calibration yielded good subsidence results. The match of the modelled subsidence to the data showed that these models are capable of well representing subsidence that has occurred at Wairakei geothermal field. The subsidence results were compared with measured subsidence in the Wairakei geothermal field and agreed very well for the 1-D and R-Z models. But for the 3-D model, pressures in the shallow zones did not satisfactorily match. This affected prediction of subsidence and therefore the match to the shape of the bowl is qualitatively satisfactory but does not match the magnitude. This study shows that the shape of the subsidence profile over a period of time can be predicted well by using non-linear elastoplastic constitutive model and realistic material properties.