- Title
- Environmental impacts of coal seam gas activities on groundwater resources
- Creator
- Askarimarnani, Sara S.
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2017
- Description
- Research Doctorate - Doctor of Philosophy (PhD)
- Description
- Coal Seam Gas (CSG) (sometimes called Coal Seam Methane or Coal Bed Methane) is a form of natural gas found in some coal deposits. This gas is mainly methane with small amount of carbon dioxide. The methane gas is generated from the organic matter under biologic and/or thermal processes over geologic time. In many geologies hydraulic fracturing is used to increase the permeability and extract the gas more efficiently from the coal seam. This study aimed to understand the effects of hydraulic fracturing on a coal seam aquifer, performed in the HB02 gas well located in the north of the Sydney Basin at Broke, in the Hunter Valley, NSW, Australia. To do this, an observed pumping test response was used to compare the performance of three different groundwater models that simulate either single phase (water only) and multiphase (water and methane) conditions. This research posed four questions: (1) Are we able to realistically simulate the groundwater flow in a multiphase flow that includes a mixture of water and gas? (2) Can the extraction of both phases from an artificially fractured coal seam be simulated with enough accuracy that we can provide an adequate estimation of the coal and fracture properties? (3) What do the well functions for a fractured well look like? (4) How do these insights improve our understanding of a CSG well; particularly, how much water is produced? The methodology employed multiple stages of analysis. The key stages in this analysis provide insight into the development of the single phase groundwater simulation and they include a sensitivity analysis of the model to coal and fracture parameters using the GMS (Groundwater Modeling System) software. It was determined that a varying hydraulic conductivity with distance from the well (modeled by geology zonation with distance from the well) improved the realism of groundwater flow patterns and the match to the well drawdown curve. Following the confirmation of qualitatively correct flow patterns and drawdowns, subsequent calibration established that the conductivity and the specific storage of the unfractured coal seam were important parameters for which the model was highly sensitive. Despite obtaining reasonable agreement between the predicted and observed drawdown curves, some subtle but significant mismatches remained. To confirm the validity of the GMS simulations, an alternative single phase model was developed, using TOUGH2 (Transport Of Unsaturated Groundwater and Heat) with the single phase EOS1 module, with functionally equivalent to the GMS model. Both GMS and TOUGH2/EOS1 produced similar results, with the same subtle mismatches, though there were slight improvements using TOUGH2. This established that the two models, with slightly different numerical formulations, could produce similar results, but were similarly unable to capture some important behaviours observed in the multiphase flow data from the HB02 well. More refined simulations using the TOUGH2/EOS7C multiphase module, with the addition of gas modeling, revealed significant changes in model performance. For the same model parameters which gave the best simulations for the single phase flow simulations, the accuracy of the multiphase model was greatly reduced, requiring the model to be recalibrated. After recalibration, TOUGH2/EOS7 gave slightly better results than both TOUGH2/EOS1 and GMS, although the strong curvature evident in the observed drawdown curve could still not be adequately predicted. This implies that the application of single phase models to a multiphase problem may be associated with greater uncertainty than explicitly modeling gas flow. As such, in selecting a modeling approach for fracked gas wells, consideration should be given to the balance between computational efficiency and simplicity of the model, and ease of calibration, and ultimately the certainty (or uncertainty) constraints associated with the inferred coal seam properties. In answer to the second research question, the uncertainty and identifiability of the TOUGH2/EOS7C parameters was assessed by the use of the Generalized Likelihood Uncertainty Estimation (GLUE) method. This provided insight into the values of the parameters likely to best represent the coal seam where the HB02 well is located, and which parameters could not be reliably calibrated from the well test data. Most parameters were non-identifiable; however the model was sensitive to the half-length of the fracture and the permeability of the surrounding coal seam. Furthermore, it was discovered that there were high levels of interaction between these two parameters, suggesting a balance between the volume/conductivity associated with the fracture and surrounding coal seam was the key to providing a realistic simulation. This interaction was quantified and found to be robust against changes in the objective function used for model calibration. Furthermore it was found that information sometimes collected during fracking (total volume of sand injected and thus frack volume, permeability of the sand, length of the frack using microseismic) provided little incremental benefit in reducing overall parameter uncertainty other than for that characteristic directly measured. In answer to the third research question, the well test was best fit by a bilinear flow field, rather than a radial flow field, so that, at least for the first 11 days of the flow test, the well function of a bilinear flow is more appropriate than that of a radial flow field. In addition, there was strong evidence for a reduction with distance from the well of the unfracked coal permeability, which may reflect a non-trivial density of small subsidiary fracks around the well that are not modeled by using a single large frack. In answer to the fourth research question the aquifer water yield (i.e. as parameterised by the aquifer storativity or the coal matrix compressibility) was found to be unidentifiable from the well test meaning that the 11 day duration of the well test was insufficient to reliably assess the long term water yield of the well. Thus the well test was unable to improve on a priori estimates of the total long-term water yield for a production well field.
- Subject
- groundwater model; CSG; hydraulic fracturing; pumping test; GLUE
- Identifier
- http://hdl.handle.net/1959.13/1333518
- Identifier
- uon:27097
- Rights
- Copyright 2017 Sara S. Askarimarnani
- Language
- eng
- Full Text
- Hits: 1151
- Visitors: 2368
- Downloads: 1018
Thumbnail | File | Description | Size | Format | |||
---|---|---|---|---|---|---|---|
View Details Download | ATTACHMENT01 | Thesis | 23 MB | Adobe Acrobat PDF | View Details Download | ||
View Details Download | ATTACHMENT02 | Abstract | 302 KB | Adobe Acrobat PDF | View Details Download |