Reservoir-scale transdimensional fracture network inversion

Márk Somogyvári, Michael Kühn, Sebastian ReichAdvances 2019 · DOI

TL;DR

This paper introduces a novel method for characterizing complex underground fracture networks using borehole temperature data. By employing a "transdimensional" inversion technique, the authors can infer the number and location of fractures without prior assumptions, offering a more detailed understanding of geothermal reservoirs than traditional methods.

What Problem Does This Paper Address?

Characterizing the intricate network of fractures within geological formations, particularly geothermal reservoirs, is crucial for understanding groundwater flow and heat transport. Traditional methods often rely on continuous porous media models that average out fracture details, failing to capture the precise locations and geometries of these critical features. Alternatively, discrete fracture network (DFN) models offer a more realistic representation but face a significant challenge: classic inversion techniques require knowing the exact number of fractures beforehand, which is rarely the case in real-world scenarios. This leaves a gap in our ability to accurately map and interpret fractured subsurface systems, especially when data is limited.

Methods

The core of this research lies in the application of a transdimensional Discrete Fracture Network (DFN) inversion methodology. This approach utilizes a reversible-jump Markov chain Monte Carlo (rjMCMC) algorithm. Unlike standard inversion methods, this technique allows the number of fractures in the model to vary dynamically during the inversion process.

The algorithm works in iterative cycles:

  1. Update Phase: The current DFN model is randomly modified by adding, deleting, or moving fractures. The probabilities of these operations are carefully chosen to ensure the reversibility of the chain.
  2. Evaluation Phase: The modified DFN model is used to simulate the steady-state temperature distribution within the aquifer using a forward model. This simulation is based on a simplified 2D DFN model and a tracer transport simulator.
  3. Acceptance/Rejection: The Metropolis-Hastings-Green acceptance criterion is used to decide whether to accept the updated model based on how well it fits the observed borehole temperature data.

A key modification for this study was adapting the methodology for reservoir-scale scenarios with limited data, leading to a simplified DFN model that incorporates geological information and focuses on inferring fracture geometry rather than detailed hydraulic properties.

Datasets

The study utilizes two main types of data:

  • Synthetic Dataset: A synthetic DFN model was created to test the viability and robustness of the transdimensional inversion methodology.
  • Field Data: Steady-state borehole temperature profiles from the Waiwera geothermal reservoir in New Zealand were used for the first field application of the method. These profiles were treated as virtual boreholes derived from existing temperature data.

Key Insights and Ideas

  • Flexible Fracture Count: The transdimensional approach is a game-changer for DFN inversion, as it removes the need to pre-define the number of fractures, a major hurdle in previous studies. This allows for a more data-driven and less assumption-laden interpretation.
  • Inferring Structure from Temperature: The research demonstrates that even a limited number of steady-state temperature logs can provide sufficient information to resolve key structural elements of a fractured aquifer. This is particularly valuable in scenarios where other geophysical data might be scarce.
  • DFN vs. Continuum Models: While continuum models can provide a good overall fit, they struggle to pinpoint fracture locations. DFN models, even simplified ones, can highlight areas where fracturing plays a significant role, offering insights that continuum models miss.
  • Geological Simplification for Data Scarcity: The adaptation of the DFN model to incorporate existing geological knowledge (e.g., known fault sets, stratification) and simplifying fracture parameters (e.g., aligning centers, calculating lengths from position) is crucial for making the inversion tractable with limited observational data.

Key Results

  • Synthetic Example Success: The synthetic test case showed that the algorithm could accurately reconstruct the general features of the fracture network, particularly the stratified sections, demonstrating the method's proof-of-concept.
  • Waiwera Reservoir Characterization: For the Waiwera geothermal reservoir, the inversion successfully identified distinct features:
  • Strong stratification near the bedrock, suggesting pathways for heat spread.
  • A less fractured zone in the center of the domain.
  • Fractures in the upper sections likely contributing to the mixing with colder waters.
  • Fit Quality: While the reconstructed temperature fields from the simplified DFN model did not achieve the same level of fit as complex 3D continuum models, they successfully indicated areas where fracturing was dominant, a feat not achievable with continuum models alone. The fit quality of borehole temperature profiles was deemed comparable to other methods, though limitations in the 2D simulation and complex geometry were noted.

Limitations and Future Work

  • Simplified DFN Model: The current DFN model is a simplification. It does not account for flow and transport within the rock matrix, complex hydraulic/thermohaline conditions, or 3D effects. This limits the accuracy of the reconstructed anomaly shape and temperature profile fits compared to more complex models.
  • Geometry-Focused Inversion: The current methodology primarily inverts for fracture geometry and does not estimate physical fracture properties (like hydraulic conductivity or aperture). Including these as free parameters would significantly increase model freedom and require more data to avoid ill-posed solutions.
  • 2D Limitation: The simulation is limited to 2D, which restricts the ability to capture 3D geological complexities.
  • Future Work: The authors suggest that future work could involve:
  • Testing more complex fracture patterns.
  • Involving additional data to enhance the inversion.
  • Exploring the stochastic behavior of the ensemble of models for uncertainty quantification.
  • Investigating the impact of different borehole locations and numbers.
  • Combining DFN models with 3D continuous models for a more complete aquifer characterization.
  • Future studies will also focus on selecting aquifer sections for more detailed exploration and characterizing hydraulic properties.

Summary updated

Abstract

The Waiwera aquifer hosts a structurally complex geothermal groundwater system, where a localized thermal anomaly feeds the thermal reservoir. The temperature anomaly is formed by the mixing of waters from three different sources: fresh cold groundwater, cold seawater and warm geothermal water. The stratified reservoir rock has been tilted, folded, faulted, and fractured by tectonic movement, providing the pathways for the groundwater. Characterization of such systems is challenging, due to the resulting complex hydraulic and thermal conditions which cannot be represented by a continuous porous matrix. By using discrete fracture network models (DFNs) the discrete aquifer features can be modelled, and the main geological structures can be identified. A major limitation of this modelling approach is that the results are strongly dependent on the parametrization of the chosen initial solution. Classic inversion techniques require to define the number of fractures before any interpretation is done. In this research we apply the transdimensional DFN inversion methodology that overcome this limitation by keeping fracture numbers flexible and gives a good estimation on fracture locations. This stochastic inversion method uses the reversible-jump Markov chain Monte Carlo algorithm and was originally developed for tomographic experiments. In contrast to such applications, this study is limited to the use of steady-state borehole temperature profiles – with significantly less data. This is mitigated by using a strongly simplified DFN model of the reservoir, constructed according to available geological information. We present a synthetic example to prove the viability of the concept, then use the algorithm on field observations for the first time. The fit of the reconstructed temperature fields cannot compete yet with complex three-dimensional continuum models, but indicate areas of the aquifer where fracturing plays a big role. This could not be resolved before with continuum modelling. It is for the first time that the transdimensional DFN inversion was used on field data and on borehole temperature logs as input.
BibTeX
@inproceedings{somogyvari2019reservoir,
  author = {Márk Somogyvári and Michael Kühn and Sebastian Reich},
  title = {{Reservoir-scale transdimensional fracture network inversion}},
  booktitle = {Advances},
  year = {2019},
  doi = {10.5194/adgeo-49-207-2019},
}
Figure