Page 282 - Machine Learning for Subsurface Characterization

244 Machine learning for subsurface characterization

1 Introduction
1.1 Mechanical discontinuities

Mechanical discontinuity in the material is generally referred as crack or frac-
ture. Predicting and monitoring the geometry, distribution, and condition of
mechanical discontinuities are critical for structural health monitoring, rock
mechanics, geotechnical projects, geothermal reservoir development, seques-
tration, and hydraulic fracturing. For example, in the oil and gas industry, geom-
etry and direction of induced fracture systems are critical to the hydrocarbon
production rate. Monitoring, description, and prediction of the state and behav-
ior of discontinuities are an important research topic.
Mechanical discontinuity is created when the stress inside the material
exceeds the strength of the material, causing the material to lose cohesion along
its weakest plane. Discontinuity can be formed due to compression, tension, or
shear stress. Mode I (opening) discontinuity develops when tensile stress acts
normal to the plane of the discontinuity; the discontinuity planes move away
from each other. Mode II discontinuity develops due to sliding, where in-plane
shear stresses act parallel to the plane of the discontinuity and perpendicular to
the discontinuity front. Mode III discontinuity develops due to tearing, where
out-of-plane shear stress acts parallel to the plane of the discontinuity and par-
allel to the discontinuity front. Brittleness is an important characteristic that
affects the rate of development of discontinuity. Brittle material breaks without
significant plastic deformation. At low temperature, materials are brittle due to
the constrained molecular motion. High confining pressure hinders the gener-
ation of discontinuities and thus makes materials ductile.

1.2 Characterization of discontinuities

Different characterization techniques have been used to evaluate the location,
orientation, and density of discontinuities in various materials. Ultrasonic mea-
surement and CT scanning are the most commonly used nondestructive testing
(NDT) methods applied to the static characterization of discontinuities in mate-
rials. A popular NDT method for dynamic characterization of discontinuities
is the acoustic emission (AE) testing to determine locations and modes when
discontinuities initiate or propagate in material [1]. NDT methods are widely
used in the oil and gas industry, geological characterization, and civil engineer-
ing to characterize the discontinuities. Common NDT methods can be divided
into two categories: mechanical methods and electromagnetic methods. Both
methods utilize wave propagation to investigate the interior structure of mate-
rial. Mechanical methods include seismic, ultrasonic, hammer, and acoustic
emission methods. Electromagnetic methods include radar, galvanic resistivity,
induction, and dielectric methods. These characterization techniques differ
in the spatial scale of investigation, such that finer structures can be assessed
by decreasing the wavelength of the propagating wave. In geology and