Bayesian Inversion of Logging-While-Drilling Extra-Deep Directional Resistivity Measurements Using Parallel Tempering Markov Chain Monte Carlo Sampling

Wang, Lei; Hu, Li; Fan, Yiren

doi:10.1109/tgrs.2019.2917839

Cited by 25 publications

(7 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the inversion of borehole resistivity measurements, an alternative is to apply statistics-based methods. [24][25][26] The statistical methods perform forward simulations hundreds of times, which also require large computation times. 27 Both gradient and statistics-based methods only evaluate the inverse operator.…”

Section: Introductionmentioning

confidence: 99%

“…To alleviate the high computational costs associated with this inversion method, simplified 1.5‐dimensional (1.5D) methods are common (e.g., References 14,15,23). For the inversion of borehole resistivity measurements, an alternative is to apply statistics‐based methods 24‐26 . The statistical methods perform forward simulations hundreds of times, which also require large computation times 27 .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Error control and loss functions for the deep learning inversion of borehole resistivity measurements

Shahriari

Pardo

Rivera

et al. 2021

Numerical Meth Engineering

View full text Add to dashboard Cite

Deep learning (DL) is a numerical method that approximates functions. Recently, its use has become attractive for the simulation and inversion of multiple problems in computational mechanics, including the inversion of borehole logging measurements for oil and gas applications. In this context, DL methods exhibit two key attractive features: (a) once trained, they enable to solve an inverse problem in a fraction of a second, which is convenient for borehole geosteering operations as well as in other real-time inversion applications. (b) DL methods exhibit a superior capability for approximating highly complex functions across different areas of knowledge. Nevertheless, as it occurs with most numerical methods, DL also relies on expert design decisions that are problem specific to achieve reliable and robust results. Herein, we investigate two key aspects of deep neural networks (DNNs) when applied to the inversion of borehole resistivity measurements: error control and adequate selection of the loss function. As we illustrate via theoretical considerations and extensive numerical experiments, these interrelated aspects are critical to recover accurate inversion results. K E Y W O R D S deep learning, deep neural networks, error estimation, geophysical applications, real-time inversion

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Error control and loss functions for the deep learning inversion of borehole resistivity measurements

Shahriari

Pardo

Rivera

et al. 2021

Numerical Meth Engineering

View full text Add to dashboard Cite

show abstract

“…Inversion of large-scale geophysical data often requires the use of high performance computing systems, especially when measurements at multiple source-receiver positions are considered (Puzyrev et al, 2018). Probabilistic inversion methods have been applied to various EM problems as well (Gunning et al, 2010;Minsley, 2011;Ray et al, 2013;Wang et al, 2019) though being computationally demanding even in low-dimensional setting.…”

Section: Introductionmentioning

confidence: 99%

Inversion of 1D frequency- and time-domain electromagnetic data with convolutional neural networks

Puzyrev

Swidinsky

2021

Computers & Geosciences

View full text Add to dashboard Cite

“…Electromagnetic (EM) logging while drilling (LWD) has been widely used in high-angle and horizontal (HA/HZ) wells for real-time geosteering and formation evaluation (Netto et al 2013;Hu et al 2017;Yuan et al 2018;Lai et al 2018). A fast and accurate EM LWD modeling algorithm is required for the pre-drilling forward simulation, the real-time inversion and the post-drilling evaluation in directional wells (Tan et al 2012;Shao et al 2013;Wang et al 2019). The simulation of EM LWD in HA/HZ wells is essentially a 3D modeling problem (Deng et al 2010(Deng et al , 2015Zhang et al 2019a, b).…”

Section: Introductionmentioning

confidence: 99%

Numerical simulation and dimension reduction analysis of electromagnetic logging while drilling of horizontal wells in complex structures

Deng

et al. 2020

Pet. Sci.

View full text Add to dashboard Cite

Electromagnetic logging while drilling (LWD) is one of the key technologies of the geosteering and formation evaluation for high-angle and horizontal wells. In this paper, we solve the dipole source-generated magnetic/electric fields in 2D formations efficiently by the 2.5D finite difference method. Particularly, by leveraging the field's rapid attenuation in spectral domain, we propose truncated Gauss-Hermite quadrature, which is several tens of times faster than traditional inverse fast Fourier transform. By applying the algorithm to the LWD modeling under complex formations, e.g., folds, fault and sandstone pinch-outs, we analyze the feasibility of the dimension reduction from 2D to 1D. For the formations with smooth lateral changes, like folds, the simplified 1D model's results agree well with the true responses, which indicate that the 1D simplification with sliding window is feasible. However, for the formation structures with drastic rock properties changes and sharp boundaries, for instance, faults and sandstone pinch-outs, the simplified 1D model will lead to large errors and, therefore, 2.5D algorithms should be applied to ensure the accuracy.

show abstract

Bayesian Inversion of Logging-While-Drilling Extra-Deep Directional Resistivity Measurements Using Parallel Tempering Markov Chain Monte Carlo Sampling

Cited by 25 publications

References 33 publications

Error control and loss functions for the deep learning inversion of borehole resistivity measurements

Error control and loss functions for the deep learning inversion of borehole resistivity measurements

Inversion of 1D frequency- and time-domain electromagnetic data with convolutional neural networks

Numerical simulation and dimension reduction analysis of electromagnetic logging while drilling of horizontal wells in complex structures

Contact Info

Product

Resources

About