Layering in shales controls primary migration in source rocks

The process of primary migration, which controls the transfer of hydrocarbons from source to reservoir rocks, necessitates the existence of fluid pathways in formations with inherently low permeability. Primary migration starts with the maturation of organic matter that produces fluids which increase the effective stress locally. The interactions between local fluid production, microfracturing, stress conditions, and transport remain difficult to apprehend in shale source rocks. Here, we analyze these interactions using a coupled hydro-mechanical model based on the discrete element method. The model is used to simulate the effects of fluid production emanating from kerogen patches contained within a shale rock alternating kerogen-poor and kerogen-rich layers. We identify two microfracturing mechanisms that control fluid migration: i) propagation of hydraulically driven fractures induced by kerogen maturation in kerogen-rich layers, and ii) compression induced fracturing in kerogenpoor layers caused by fluid overpressurization of the surrounding kerogen-rich layers. The relative importance of these two mechanisms is discussed considering different elastic properties contrasts between the rock layers, as well as various stress conditions encountered in sedimentary basins, from normal to reverse faulting regimes. Results show that the layering causes local stress redistribution that controls the prevalence of each mechanism over the other. When the vertical stress is higher than the horizontal stress in kerogen-rich layers, microfractures propagate from kerogen patches and rotate toward a direction perpendicular to the layers. Microfracturing in kerogen-poor layers is more pronounced when the confinement in these layers is higher. Those mechanisms were shown to be representative of Draupne formation.

is used to simulate the effects of fluid production emanating from kerogen patches con-27 tained within a shale rock alternating kerogen-poor and kerogen-rich layers. We iden-28 tify two microfracturing mechanisms that control fluid migration: i) propagation of hy-29 draulically driven fractures induced by kerogen maturation in kerogen-rich layers, and 30 ii) compression induced fracturing in kerogen-poor layers caused by fluid overpressur-31 ization of the surrounding kerogen-rich layers. The relative importance of these two mech-32 anisms is discussed considering different elastic properties contrasts between the rock lay-33 ers, as well as various stress conditions encountered in sedimentary basins, from normal 34 to reverse faulting regimes. Results show that the layering causes local stress redistri-35 bution that controls the prevalence of each mechanism over the other. When the ver-36 tical stress is higher than the horizontal stress in kerogen-rich layers, microfractures prop-37 agate from kerogen patches and rotate toward a direction perpendicular to the layers. 38 Microfracturing in kerogen-poor layers is more pronounced when the confinement in these 39 layers is higher. Those mechanisms were shown to be representative of Draupne forma-   The simulated rock specimen is made of 50, 000 spherical particles packed in a vol-117 ume of dimension 14 × 10 × 10 cm 3 . The medium is divided into 7 horizontal layers, 118 alternating material with high organic content (#1, red) and material with low organic 119 content (#2, blue), as shown in figure 3). The layers with high organic content contain 120 kerogen patches, whereas the layers with low organic content do not. The interfaces be-121 tween the layers have the same cohesion as between particles withing the layers.

122
The kerogen patches within high organic content layers are modeled as penny shaped 123 cracks with a radius of 1 cm. Each crack extends over 6.5 discrete elements in the bed-124 ding direction of the shale material to reproduce the preferential orientation of kerogen 125 patches observed using synchrotron X-ray tomography imaging (Kobchenko et al., 2011;126 Panahi et al., 2013126 Panahi et al., , 2018. The number of elements has been set as a compromise be-127 tween a fine discretization of the layers, with 6.5 discrete elements across the layer thick-128 ness, and a reasonable simulation time. Kerogen maturation is simulated by injecting 129 fluid within all penny shaped kerogen-filled cracks at constant flow rate. The fluid has 130 a bulk modulus of 2.2 × 10 9 P a, a viscosity of 10 −3 P a.s.

131
The injection rate and the simulation time cannot be directly linked to the geo- in red correspond to material #1, a layer with high organic content. The elements colored in blue correspond to material #2, a layer with low organic content. In dark are shown the initial kerogen patches where fluid pressurization occurs.   The state of stress in a stratified material formed by two series of isotropic layers 155 with different elastic properties can be obtained from Hooke's law:

Stress boundary conditions
We consider here a shale formed by two kinds of layers: kerogen-rich layers con-     To conclude, microfracturing in low organic content layers is controlled by the con-250 fining pressure in this layer (i.e., the first stress invariant I 1,#2 ). Figure 7. Amount of microfractures created in a homogeneous layer with low organic content rock #2 subjected to a compressive loading (kerogen maturation is not considered here). The horizontal stress is fixed for all simulations σ h = −11 M P a while the vertical stress σv is increased. Five cases are shown for five different Young's moduli ranging from 6.8 GP a to 9.9 GP a (table 1).

Summary of the results for a non-layered rock 252
In this section, we discuss the two microfracturing mechanisms described in figure 1 253 taking place in a non-layered rock. The results show that: far field stress: high deviatoric stresses tend to favor a propagation of the hydrauli-256 cally driven fracture in the direction of the major principal stress.
the Young's modulus of the host medium as well as with the first stress invariant 259 I 1,#1 and the stress perpendicular to the kerogen patch.  (b) Scenario with E1 = 9.9 GP a in high organic content layers, and E2 = 14.9 GP a in low organic content layers.
-13-manuscript submitted to JGR: Solid Earth on this damage are the initial confinement of the layer with low organic content (I 1,#2 , 325 (iii)) and the Young's modulus of the layer with high organic content E 1 .  Here, we specifically study the effect of structural layering in shales on the devel-  Scenario with E1 = 9.9 GP a in the high organic content layers, and E2 = 14.9 GP a in the low organic content layers. damaging process is the initial confinement of the layer with low organic content, (I 1,#2 ).

408
The higher the initial confinement is, the more efficient this damaging process will be.

409
The initial confinement is controlled by the elastic contrast between the layers and the tent is caused by the compression of those layers induced by the pressure buildup 483 of the high organic content layers (mechanism 2, white microfractures in figure 1.

484
The two main parameters controlling this fracturing process are the initial con-485 finement of the layers with low organic content (I 2,#2 ), which results from the elas-486 tic contrast between the layers, and the Young's modulus of the layers with high 487 organic content.

488
Appendix A State of stress in a layered linear elastic material 489 A layered rock made of two linear elastic isotropic materials is considered. Both 490 materials are characterized by a Young's modulus E i , a Poisson's ratio ν i and a volume 491 fraction q i (with q 1 + q 2 = 1), as shown in figure A1. The applied loading is assumed 492 to be transverse isotropic, with (σ v , σ h ) respectively perpendicular and parallel to the 493 layers (see Figure A1). In this configuration the vertical stress, along z-axis, is the same 494 in both layer and equal to σ v . 495 Figure A1. Schematic representation of a layered rock with imposed elastic parameters and loading conditions.
Applying Hooke's law on an elastic isotropic material (equation 1), we can project 496 the strain tensor for both material on the x axis. The strain along the x axis is geomet-497 rically constrained to be the same for both materials and is noted ε t .