CSG water reinjection impacts: Modelling, uncertainty and risk analysis. Groundwater flow and transport modelling and uncertainty analysis to quantify the water quantity and quality impacts of a coal seam gas produced water reinjection scheme in Surat Basin, Queensland

[摘要] This study developed and applied an integrated multi-scale groundwater modelling methodology to assess the risks of water quantity and quality changes resulting from a large scale reinjection scheme proposed by Australia Pacific LNG (APLNG) in the Surat Basin, Queensland. The proposed scheme, in the full scaleoperation phase, would inject 30 - 40 ML/day of reverse osmosis-treated coal seam gas (CSG) produced water into the Precipice sandstone at the Reedy Creek site in Surat Basin. Similar injection schemes are planned at other APLNG CSG production sites including Spring Gully, Condabri and Talinga. The quantification of groundwater quantity and quality changes resulting from such large scale injection schemes is challenged by a number of factors. Simulating the impacts of operational scale injection scenarios required upscaling of the information from injection trial and tracer tests data to larger spatial and temporal scales in the numerical groundwater models used to assess cumulative impacts. Prediction ofthe risk of impacts using a stochastic modelling framework necessitated development of the models with least possible computational burden. Specifically this study;- Developed a multi-scale integrated groundwater modelling approach and used it to predict groundwater head and quality changes in the Precipice sandstone and other neighbouring aquifers resulting from the reinjection of CSG produced water at the operational scale.- Developed an inversion based upscaling approach, to efficiently upscale information from the wellscale injection trials, pump tests and tracer tests conducted by APLNG at the Reedy Creek site anduse this information to constrain regional scale groundwater models.- Applied state-of-the-art uncertainty analysis techniques to quantify the prediction uncertainty in the simulated groundwater quality and head changes.- Explored the potential impacts of the presence of faults in the injection well field on the predicted groundwater quality changes.- Explored a hypothesis testing framework to test whether the occurrence of an undesirable level of water quality could be rejected with high confidence at a risk receptor.- Investigated the relative worth of different data types and their spatial and temporal disposition to optimise injection trial design for the Spring Gully site as well as for the long-term monitoring ofgroundwater quality at the Reedy Creek site.- Employed stochastic particle track analysis to determine the probability distribution of groundwater age in different formations in the study area.- Developed a novel simulation-optimisation modelling approach for the stochastic optimisation of injection well field design and tested it using a proof-of-concept case study.The groundwater models used in this study were developed based on the USGS modelling codes MODFLOW and MT3D. Model calibration and uncertainty analysis were based on PEST and its utility software. Also, new utilities were built for specific purposes as required.The integrated modelling approach developed in this study built and applied groundwater flow and transport models for four different but overlapping spatial and temporal domains. Each model wasused to simulate the groundwater flow and transport process that was best represented in one spatiotemporal scale. For example, the coarse scale regional groundwater model developed by the Office ofgroundwater impact assessment, Queensland (OGIA model) for the Surat Basin was used to characterise the regional flow system and was informed by the regional groundwater monitoring data.Similarly a very fine scale 2-dimensional radial transport model was built and calibrated to the APLNG injection tracer test data and was used to simulate near-well break through curve. The modelssimulating groundwater flow and transport processes in different scales were then used in an integrated modelling framework using inversion based upscaling. The integrated models were then used...

[发布日期] 2015-03-15 [发布机构] CSIRO

[学科分类] 地球科学(综合)