A systematic methodology to calibrate wellbore failure models, estimate the in-situ stress tensor and evaluate wellbore cross-sectional geometry

Dahrabou, Asmae; Valley, Benoît; Meier, Peter; Brunner, Philip; Alcolea, Andrés

doi:10.1016/j.ijrmms.2021.104935

Cited by 7 publications

(15 citation statements)

References 68 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…4.1). Calibration of stress and strength parameters of the proposed failure model is made against existing data, typically collected along the shallower section of the well (Dahrabou et al, 2022b). The calibrated failure model is then used to predict failure in the subsequent well sections.…”

Section: Methodsmentioning

confidence: 99%

“…• Step 1: jointly estimate the stress state and rock strength parameters by 1 st order calibration (Dahrabou et al, 2022b). This fast calibration, involving a few parameters (those defining 1 st order parameter linear trends), is carried out for N different starting points in order to span the entire parameter space and account for non-uniqueness of the solution.…”

Section: Methodsmentioning

confidence: 99%

“…Step 5: perform 1 st order predictions using the extrapolated parameters and the failure model described in (Dahrabou et al, 2022b) in order to verify the consistency of the predictions. In this paper, we also carry out a verification analysis by comparing the predictions with available observations along the deepest section of the well, which allows us (1) to validate the methodology, and (2) to evaluate the accuracy of our predictions.…”

Section: Methodsmentioning

confidence: 99%

“…The vast volume of literature on borehole stability analysis and optimal borehole tra-jectory selection is mainly on deterministic analysis, specific to local conditions and fail at systematically integrate parameters uncertainties. In the first paper of this series, Dahrabou et al (2022b) propose an efficient methodology to estimate wellbore failure parameter sets that reproduce the observations of cross-sectional borehole geometry. In this paper, we extend the methodology to failure prediction.…”

Section: Introductionmentioning

confidence: 99%

“…For illustration purposes, we apply the methodology to the prediction of the deepest part of the BS-1 well (Häring et al, 2008) in order to verify the applicability and robustness of the developed method. First, the parameters of the failure model described in Dahrabou et al (2022b) are calibrated against data from the shallower part of the borehole in the crystalline rock (from 3 to 3.6 km depth). Second, a number of equally-likely stochastic simulations of the remaining 1.4 km, which include a severe change of diameter, is carried out.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Calibration methodology of near-field wellbore stress and strength to predict failure severity in deep geothermal boreholes: From theory to industrial applications

Dahrabou¹

View full text Add to dashboard Cite

The economic viability of deep geothermal projects depends primarily on the extracted flow rate and thus its evaluation has to be made during its feasibility study stage. Borehole stability compromises sometimes the economic success of early stages of EGSs (Enhanced Geothermal Systems) and must therefore be comprehensively analysed. A stable borehole intersecting as many as possible hydraulically transmissive zones is the ideal target for a successful drilling. For instance, a highly irregular borehole cross-section involving deeply penetrating borehole breakouts or drilling induced tensile fractures (DITFs) may entangle the correct placing of packers used for a targeted stimulation of pre-existing fractures. As a consequence, the main challenge that drilling engineers usually encounter is the selection of an optimal borehole trajectory that will minimise borehole instabilities and maximise intersections with critically stressed fractures as they are favourably oriented to fail and exhibit very good fluid flow characteristics. For the aforementioned reasons, the accurate knowledge of the local state of stress and the rock strength is crucial to the success of deep geothermal projects. State-of-the-practice methods estimate stresses and strength properties separately. Several approaches neglect to account for the interplay between stress and strength, a shortcoming this thesis addresses. Indeed, the interplay between the in-situ stresses and rock strength parameters (cohesion and friction angle) mainly controls the borehole stability. Additionally, there is still considerable controversy concerning the selection of an appropriate failure criterion and its parametrisation to compute borehole failure and most studies have only focused on estimating stress and strength from breakout width, ignoring as such other relevant borehole failure indicators such as breakout extent, breakout orientation and presence/absence of DITFs. This dissertation aims at filling this gap. A systematic, generic and novel methodology to calibrate jointly wellbore stress and strength in the shallowest section of a deep borehole and to predict failure severity in the deepest is presented. Beyond the theoretical development, practical tools are provided in order to facilitate knowledge transfer to the practice. The calibration methodology was tested and developed with the extensive data set of the BS- 1 borehole (from the Basel Deep Heat Mining project). The methodology aims at estimating jointly depth profiles of the local stress tensor (magnitudes and orientations) and rock strength properties (cohesion and friction). Emphasis is placed on better understanding the relationship between stress and strength, handling solution non-uniqueness and considering both depth trends (1st order characterization) and variability (2nd order) of key parameters. Regularized pilot points method were used to estimate model parameters as implemented in PEST software (Parameter ESTimation) from borehole cross-sectional characteristics (breakout width, breakout extent/penetration depth and breakout orientation) and from the presence/absence of DITFs (including both axial and en-echelon tensile fractures, A-DITFs and E-DITFs). Results show consistency of the 1st order calibrated failure models with current knowledge of the stress in Basel. The reduction of parametric uncertainties was also investigated by including independent measurements of the minimum horizontal principal stress from hydraulic tests. It was concluded that this reduces drastically the range of calibrated parameters, which highlights the importance of collecting such data. Moreover, this methodology provides an effective and computationally efficient way of analysing extent of failure, which is potentially the most relevant parameter to assess packer sealing integrity. 2nd order calibrated models based on the pilot points technique allows reproducing the observed failure variability as well as most of the maxima/minima of breakout width and extent in addition to the small gaps without breakouts, which is critical to assess risk associated with packer placement. It provides also calibrated joint stress and strength profiles, including magnitude of all the stress components which gives a new insight into the source of stress variability in the earth crust. Our unique data set supports the idea that stress variability in the granitic basement of Basel arises primarily from fracture slip. After estimating the stress and strength in the shallowest section of the well, prediction of failure severity is necessary for the selection of an optimal drilling trajectory. A stochastic methodology to perform predictions accounting for parameters uncertainty and variability in depth is proposed in this thesis. In this methodology, three elements are needed: (1) a borehole trajectory and section along which failure should be predicted, (2) a 1st order calibrated stress and strength model that can be extrapolated to a greater depth and finally (3) a 2nd order stress and strength model that can be used to generate stochastic variability and add it to the predictions. A key feature of the proposed prediction methodology is the possibility to perform predictions in a stochastic manner and to account for existing dependencies between all the parameters of interest. To validate our methodology, BS-1 data set were used. Results show the accurate goodness of fit between predictions and observations in the deepest section of BS-1. In addition to that, correlations between the failure model inputs and outputs were satisfactorily reproduced by using multivariate simulations that are based on direct sampling multi-point statistical approach as implemented in DeeSse software. These findings illustrate the robustness, completeness and accuracy of the developed workflow. The applicability and limitation of the proposed workflow has been tested using synthetic cases covering a broad range of well trajectories, stress regimes and wellbore failure severity. In theory, the ability to predict failure in deviated boreholes with calibrated models based on data from a vertical well only is limited because the failure in the vertical hole is insensitive to the ratio between principal horizontal and vertical stresses which is important to assess stability of deviated wells. However, the proposed methodology in this work considers not only the data from borehole breakouts, but also from the drilling induced tensile fractures, DITFs. In addition, it accounts for depth trends of stress and strength and their variability in depth, which allows to mitigate these theoretical limitations. Nevertheless, with calibration performed only on a vertical borehole section, the predictions carried out for deviated section show some divergences. The degree of dispersion of the predictions differ from one stress regime to another and depends mainly on the stress anisotropy ratio and the difference between failure severity in the calibration and prediction sections. This section of the thesis highlights and quantifies these limitations, guides the proper application of the proposed workflow and helps evaluating uncertainty and anticipating potential stability problems. The significance of this thesis is that it enhances our theoretical understanding of borehole stability and optimal borehole trajectory selection by introducing a focus on the interplay between stress and strength and their stochastic variability in depth hitherto lacking, and informs our practical understanding of the importance of developing a set of versatile supporting software tool to quickly define the optimal borehole direction at a specific point during the drilling operations of deep geothermal wells. To that end, a graphical user interface was developed (’DG-WOW-app’, Deep Geothermal Well Optimization Workflow application) and is presented in this dissertation bringing together the calibration and prediction developed methodologies into practice. In addition, the proposed calibration and prediction methodologies as implemented in the DGWOW application were tested on two real case studies in order to assess their versatility and robustness. For illustration purposes, the CB1 (Bedretto) and GPK3 (Soults-Sous-Foret) boreholes were used. Data from both boreholes were pre-processed, formatted and compiled. Then 1st and 2nd order calibrations were carried out in order to estimate depth trends of stresses and strength and to characterize their variability around these trends. Results outline the consistency of the calibration outputs with the literature. The proposed methodology and associated tools call for their application on a wide range of new deep geothermal projects. This will allow them to be tested and improved so that they best meet the needs of the deep geothermal industry. Indeed, the research efforts deployed in this thesis are also aimed at transferring knowledge into practical and usable software tools with the ultimate objectives of making the extraction of deep geothermal energy possible and economical and thus assist with the energy transition.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations