WorksPhysics and Astronomy
A dual–continuum phase-field model for hydraulic fracturing: Viscosity-dominated regime and fluid lag
Abstract
The phase-field model regularizes sharp fractures into a diffuse representation, blurring the boundary between the fracture and the intact material. This blurring makes it difficult to capture distinct domain processes in hydraulic fracturing, where Reynolds flow governs the fracture and Darcy flow describes the surrounding porous matrix. Consequently, the blurred delineation artificially smears the pressure field across the fracture–matrix interface, which is acceptable in toughness-dominated hydraulic fracturing regimes where pressure drops within the fracture are negligible. However, in viscosity-dominated regimes, typically for actual subsurface injections due to high injection rates, the fluid pressure drops more drastically, and the fluid front may even lag behind the propagating fracture tip, a phenomenon that a smeared pressure field cannot capture. Despite its relevance, the viscosity-dominated regime has not been addressed by any existing phase-field models to date, likely due to its numerical instability. In this study, we propose a dual–continuum phase-field model based on double-porosity microporomechanics that explicitly separates mesoscale crack pressure from micropore pressure. The framework provides a variationally consistent formulation alongside phase-field–dependent poroelasticity. To ensure the numerical stability of the hydromechanical coupling, a fixed-stress split scheme is modified for two independent fluid pressures, while a variational inequality constraint is applied to reproduce fluid lag. Verified against the closed-form solutions in toughness-dominated, viscosity-dominated, and early-time transitional regimes, the model accurately captures complex fluid flow behavior and transient fluid lag within the fracture, and opens a new frontier for applying phase-field models to realistic viscosity-dominated hydraulic fracturing.
Introduction
Fracturing in porous media is central to numerous engineering applications, such as hydrogel failure [6, 10], biological tissue rupture [28], and both natural [61] and anthropogenic [43, 49] (Fu et al. 2017; Plúa et al. 2024) hydraulic fracturing of geological formations. The behavior of a hydraulic fracture is governed by the interplay between solid skeleton deformation, multi-scale (fracture and pore-space) fluid flow, and crack propagation. Beyond laboratory (Bunger et al., 2005, 2013; [24] (Ha et al. 2018) or field ([29, 49, 75]) investigations, dimensionless scaling enables a systematic analysis of fluid-driven fracture mechanics ([25, 62]). Garagash [37] identified three different regimes—the toughness-dominated, the viscosity-dominated, and the early-time regimes—based on a dimensionless toughness and a dimensionless time (Figure 1a). The toughness-dominated regime (K-vertex) is characterized by the dominant role of fracture toughness, while the viscosity-dominated regime (M-vertex) is characterized by the dominant role of fluid viscosity, high far-field stress , and prolonged fluid injection ([45]). The early-time regime (O-vertex) corresponds to the solution for small confining pressure and low toughness, in which a pronounced fluid lag zone develops near the fracture tip (Figure 1a). This lag zone is either filled by fluid vapor (in impermeable medium) or by pore fluid drawn from the surrounding matrix (in permeable medium).
Figure 1. (a) Hydraulic fracture propagation regimes, adapted from Garagash [37], where the parametric space of the three vertices of the limiting regimes: toughness-dominated (K), viscosity-dominated (M), and early-time (O). is the pressure in the fluid lag zone, and is the far-field confining pressure. (b) Pressure profiles across the hydraulic fracture. An actual pressure profile and a numerical smeared profile from a phase-field model. (c) The phase-field profile instability (broadening) driven by the high fluid pressure across the diffuse pressure profile and a negligibly small fracture toughness.
Despite recent advances, simulating hydraulic fractures remains challenging due to the inherent complexities of multiphysics coupling, moving fracture paths, and transient fluid fronts ([44, 45]). To address complex fracture paths, the variational phase-field model, regularization of the variational approach to brittle fracture ([33]) through a phase-field variable ([13]), offers distinct advantages, as it can capture crack branching, merging, and nucleation in a unified variational setting without explicit surface tracking. Subsequent algorithmic developments [11, 55, 72] have demonstrated the robustness and versatility of this framework. The approach has since been extended beyond brittle fracture to cohesive fracture [74] (Wu and Nguyen 2018), dynamic fracture [8, 68], thermal cracking [15], fatigue fracture [18, 67], and hydraulic fracturing – the main focus in this study. Bourdin et al. [12] first extended the variational fracture framework to hydraulic fracturing, and subsequent contributions by Lee et al. [47, 48]; Miehe and Mauthe [56]; Miehe et al. [57]; Mikelić et al. [58]; Wheeler et al. [77, 78]; Wilson and Landis [79]; Yoshioka and Bourdin [85] established diverse numerical methods and convergence analyses for phase-field hydraulic fracture in poroelastic media. You and Yoshioka [87] recently proposed a micromechanics-based approach to derive effective, damage-dependent poroelastic properties for fractured porous media within a single-continuum framework. They demonstrated that the proposed model reproduces crack opening displacements more accurately than existing phase-field hydraulic fracture models, irrespective of the pressure profiles or the values of the Biot coefficient. Beyond a single-fluid quasi-static hydraulic fracturing, further contributions include dynamic hydraulic fracturing [30, 31], heterogeneous and layered formations [81, 88, 92], and multi-fluid fracturing [46, 54] (Li et al. 2025).
In phase-field models, a diffuse fracture representation blurs the interface between the fracture and the porous media. Resolving this interface is critical in hydraulic fracturing because the domains are governed by different physical laws. Fluid flow within the fracture is typically idealized as flow between parallel plates, where the flow rate is proportional to the cube of the crack opening (a relation known as Reynolds flow). Meanwhile, fluid flow in the surrounding porous medium is normally described by Darcy flow, where the flow rate is proportional to the matrix permeability. To apply these two distinct hydraulic flows to each of the domains, various approaches have been proposed. One of the most popular, and intuitive, approaches is to use the phase-field value to delineate the domains. For instance, Mikelić et al. [58] used the phase-field variable to transition between the two flow regimes, while Yoshioka and Bourdin [85] applied two distinct permeabilities for the matrix and fracture based on the phase-field value. Along a similar line, a level-set function based on the phase-field variable have been adopted to average the two flow behaviors by many [48, 53, 78, 91] (Li and Zhou 2019; Li et al. 2021; Shahrozeisi et al. 2024; Xu et al. 2022). Santillán et al. [66] separated the fracture and reservoir domains based on the phase-field value, where the fracture domain is explicitly extracted and discretised using one-dimensional line elements to form a hybrid system with the surrounding two-dimensional reservoir elements. In their approach, however, this one-dimensional fracture domain must be updated dynamically as the phase-field profile evolves. Another hybrid method was proposed by Costa et al. [22], who used the phase-field model to capture fracture propagation; once a crack fully develops, the fractured domain transitions to a discrete contact formulation. Alternatively, Wilson and Landis [79] used the phase-field variable to scale fluid viscosity and recover Reynolds flow within the fracture, whereas Chukwudozie et al. (2019) leveraged phase-field calculus to homogenize the distinct domains. By contrast, the frameworks in Ehlers and Luo [30]; Heider and Markert [40] rely on the theory of porous media to govern the transition between Darcy flow in the reservoir and Navier-Stokes flow in the fracture. Furthermore, despite its simple implementation, modifying the permeability as a function of the phase-field and local deformation has proven to be an effective way to capture enhanced flow within fractures [57, 87].
When considering these various treatments, we must keep two factors in mind. The first is the physical dimensions of hydraulic fractures; while fracture length typically spans tens of meters, the aperture is merely 0.001 to 0.01 m1. Even if a fracture is discretized with an element size of 0.1 m in numerical simulations, which may already be too computationally expensive to resolve a fracture spanning tens of meters, the physical fracture opening (aperture) remains far smaller than the element size. Because the phase-field transition zone spans 5 to 10 elements (depending on the characteristic length and the specific phase-field model applied), delineating the fracture based on the phase-field variable or a level-set function would yield an “aperture” that is enormous compared to its physical size. Consequently, this nonphysical void space complicates the fluid mass balance. The second factor is the fluid pressure profile. Despite different approaches to the transition between the two distinct flow regimes (porous medium and fracture), all existing models treat pressure as a continuous field across the fracture-matrix interface2. Consequently, the pressure field becomes smeared (Figure 1b). This smeared pressure field appears to yield reasonable fluid pressure responses at least in the toughness-dominated hydraulic fracturing where the pressure drop within the fracture is negligible (Chukwudozie et al., 2013). Conversely, under a high injection rate, high fluid viscosity, or low fracture toughness, hydraulic fracturing transitions into the viscousity-dominated regime. In this regime, fracture toughness is negligibly small compared to the viscous dissipation. Given a negligibly small fracture toughness combined with a smeared continuous fluid pressure, local driving forces around the fluid injection becomes excessive. Consequently, damage tends to evolve in the transverse direction rather than out of the fracture tip (Figure 1c), causing numerical instability. Although this type of instability has been systematically studied in simulations of moving phase interface (e.g., Feyen and Moelans [32]; Zhang et al. [89]), it has not yet been addressed in variational phase-field models for fracture. Furthermore, dominant viscous dissipation causes a drastic pressure drop within the fracture (as shown as the actual profile in Figure 1b), and the fluid front may even lag behind the propagating fracture tip (Figure 1a). This distinct pressure behavior and fluid lag cannot be fundamentally captured by a smeared pressure representation.
Because of the high injection rates typically applied in subsurface operations, hydraulic fracturing field applications are predominantly governed by the viscous-dominated regime. Yet, thus far, no phase-field model has fully addressed the viscous-dominated regime (M-vertex). Wilson and Landis [79] investigated a regime close to the M-vertex with and Santillán et al. [66] investigated a state with . Despite their claims, however, neither of the values in these studies was sufficiently small to capture the M-vertex regime () [38], and the fluid lag was not resolved in their works.
To address the existing phase-field model’s limitations in the viscous-dominated regime, this study introduces a dual-continuum phase-field fracture model based on two-scale microporomechanics (also known as double-porosity media), which explicitly distinguishes the pressure within the mesoscale cracks from the pressure in the micropores. Natural rock formations commonly exhibit porosity at two distinct scales: a network of mesoscale cracks that constitute the primary conduits for rapid fluid flow, and a micropore space within the intact matrix that governs fluid storage and diffusive exchange between the two pore systems. The double-porosity concept was introduced by Barenblatt et al. [5] for fissured porous media and formalized as a dual-continuum flow model by Warren and Root [76]. Subsequent work incorporated the coupling between the two pore pressures and the solid deformation [7, 23], and the double-porosity model has found widespread application in reservoir geomechanics [9, 19]. Despite this extensive body of literature, a dual-continuum framework has not been extended to phase-field models for propagating fractures. In this paper, we propose a phase-field model for hydraulic fracturing in a two-scale double-porosity medium, building on the microporomechanics of [27, 63]. The principal contributions of this study are:
a variationally consistent phase-field model for hydraulic fracturing in two-scale double-porosity media in which the effective elastic tensor, the Biot coefficient tensors, and the Biot moduli are derived rationally from microporomechanics;
the relaxation of pressure continuity assumption across the fracture using two independent pressure variables (), thereby preventing artificially high local driving forces;
a variational inequality formulation that enforces non-negative fluid pressures in the fracture, providing a flexible treatment of the fluid lag without explicit tracking of the fluid front;
a fixed-stress split scheme for the two-pressure hydromechanical system that accounts for the coupling between the two pore networks; and
successful numerical verification against analytical solutions in the toughness-dominated (K-vertex), viscosity-dominated (M-vertex), and early-time transitional regimes (O-vertex).
The remainder of this paper is organized as follows. Section 2 presents the theoretical framework, including the microporomechanics-based state equations and the phase-field fracture formulation. Section 3 details the solution schemes, comprising the fixed-stress split and the variational inequality formulation for the fluid lag. Section 4 presents the numerical verification studies and double-porosity simulations. Concluding remarks are offered in Section 5.
Throughout the paper, the following notations are used. Tensorial product of any second-order tensors and and fourth-order tensor are and . The symbol is used to calculate the norm of any second-order tensor . The identity tensors for the second and fourth order are denoted by and , respectively. The fourth order projector and are expressed in the component form as and , respectively. The trace operator is defined as . is the gradient of , is the transpose of , and is the symmetric part of the gradient of defined as . The overhead dot symbol indicates the time differentiation of variable .
Theoretical frameworks
State equations
We consider a solid body () embedded with a lower-dimensional macroscopic fracture which is regularized by the well-known phase-field method into . As shown in Figure 2 (left), is the phase-field variable, , with representing the fractured state and representing the intact material. The body is subjected to a surface force on boundary and a flux on . In addition, and are the prescribed pressure and displacement on boundary and , respectively.

Figure 2. Schematic illustration of the scale separation of the cracked porous media where the meso-cracks and micro-pores are filled with fluid. Left: macroscopic domain containing a phase-field regularized macrocrack . The boundary is partitioned into complementary pairs: prescribed traction on and prescribed displacement on for the mechanical problem; prescribed fluid flux on and prescribed pressure on for the fluid problem, with . Middle: at the mesoscale, consisting of penny-shaped fluid-filled mesocrack inclusions with average pressure embedded in a porous matrix with stiffness . Right: at the microscale, consisting of spherical micropore inclusions with average pressure distributed in a solid matrix with stiffness .
Additionally, we consider the material at the mesoscale, i.e., in Figure 2 (middle), which is composed of porous matrix (index m, volume fraction ) and randomly distributed mesocracks (index c, volume fraction ), and , where is the volume of , is the volume of the porous matrix and is the total volume of mesocracks. At the lower microscale, i.e., in Figure 2 (right), the porous material is made up of solid matrix (index s, volume fraction ) and micropores (index p, volume fraction ), and , where and are the volume of solid matrix and the total volume of micropores. Note that the aforementioned definitions of , , , and are Eulerian, which represent the volume fraction of the corresponding microstructures in the current configuration. Using the reference volumes and , the Lagrangian porosities can be defined as , , , and .
The fluids in the mesocracks and micropores are interconnected through a permeable matrix, allowing fluid mass exchange between the two pore systems. However, the average fluid pressures in the mesocracks (denoted as ) and micropores (denoted as ) may not always be equilibrated i.e., if the matrix permeability is small, which is often the case. In the following, we will derive the macroscopic state equations of this two-scale double porosity medium. Unlike our previous work in You and Yoshioka [87], the consideration of allows a more general analysis of fluid flow and solid deformation at the two different scales. For example, when viscous stress dominates the action of the fluid, there will be a pressure jump between the fracture and the porous matrix. This pressure jump will enhance mass transfer between the two pore regions, thereby affecting their mechanical behavior.
Recalling the microporomechanics model of Pichler and Dormieux [63] and considering the fluid saturated condition, the macroscopic state equations of this two-scale double-porosity media are defined as
where denotes the effective elastic tensor, is the macroscopic strain defined as , and are Biot coefficient tensors at different scales respectively.3
The effective stiffness tensor in Eq. (1) accounts for the presence of mesocracks and micropores, which can be defined using volume averaging as
where is the average strain concentration tensor of the porous media in . Similarly, with the volume fraction of micro-pore, , and the average strain concentration tensor of the solid matrix in , , we can define the homogenized stiffness tensor of , i.e., as
In this study, is considered to be an isotropic fourth-order tensor and thus can be written as
with and the bulk and shear moduli of the porous media.
Furthermore, the Biot coefficients and in Eq. (1) are given as [27]
where is the isotropic Biot coefficient tensor for the porous matrix given by
We note that as is defined throughout the domain taking into account for the state of the meso-cracks, while is defined in the domain excluding the mesocracks.
By denoting the compliance tensor of as (i.e., ) and multiplying Eq. (3) by , we obtain
Substituting Eq. (8) into Eq. (7) yields the Biot coefficient for porous media as
where denotes the bulk modulus of solid matrix and is the Biot coefficient for the porous media.
Similarly, combining Eq. (2) with Eqs. (5) and (6), we have
with being the Biot coefficient for the single continuum system.
where is the compliance tensor of defined as . Note that the effective stiffness may or may not be isotropic depending on the energy decomposition scheme used in the phase-field model. You and Yoshioka [87] applied the volumetric-deviatoric energy decomposition (Amor et al., [2]), which results in an isotropic and thus an isotropic Biot coefficient. In this study, however, we consider a more general case where may be anisotropic, and the Biot coefficients and may consequently be anisotropic tensors.
The fluid content changes in mesoscale (RVE) and microscale (RVE) can be defined as
Remark 1. and in Eqs. (13) and (14) are defined using the Lagrangian description of the porosities of the mesocracks () and micropores (), which describe the volume changes relative to the reference volumes. Under small deformations, the Lagrangian and Eulerian descriptions ( and ) can be assumed to be approximately equal, but their variations are not identical. For example, the rate of is defined as with being the crack strain rate (Pichler and Dormieux, [63]).
Additionally, the state equations for the mesocracks and micropores under a fully saturated condition are given by (Dormieux et al., [27])
with
where is the Biot modulus of the porous media defined as
Combining Eqs. (10), (11), (9) and (18), the Biot moduli (17) become
with defined as
Eq. (19) shows that the Biot moduli are symmetric, i.e., . Since the pressure within a given pore network must promote, not impede, its own volumetric expansion, the diagonal Biot moduli must be non-negative, i.e., and . These constraints yield , signifying a negative cross-coupling between the two pore pressures, i.e., an increase in diminishes , and vice versa. This behavior admits a clear physical interpretation. At a constant total porosity (i.e., in the absence of microstructural growth), an elevated drives the dilation of mesocracks, which must be accommodated by a commensurate compression of the micropore space.
A phase-field fracture model for dual-continuum porous media
Following the variational approach to fracture proposed by Francfort and Marigo [33], the total energy of the system is defined as a sum of the Helmholtz free energy, , and the surface energy:
where is the critical energy release rate.
The Helmholtz free energy, , considering the fluid-filled dual continuum, can be derived as (see Appendix § 9 for the details)
where denotes the strain energy density and accounts for the interaction between the mesocrack pressure and micropore pressure. The control variables and are implicit in the quantities and , according to Eqs. (13) and (14). Therefore, we can obtain the following constitutive relations as
Remark 2. Note that the derivation of Eq. (22) is based on the grand-canonical potential (Anand, [3]), following from the Legendre transform
With this, by substituting Eqs. (13) and (14), we have
This procedure was followed by several existing works ([26, 63, 82]) on microporoelasticity models. Zhu et al. [90] used a slightly different Legendre transform process to obtain the Gibbs free energy, which however leads to the same relations as Eq. (24).
Substituting Eq. (22) into Eq. (21) yields
Following the now-standard regularization proposed by Bourdin et al. [13], we define the regularized total energy of Eq. (26) as
where is the length scale parameter and is a normalization constant defined as .
With the energy decomposition (Amor et al., [2]; Freddi and Royer-Carfagni, [34]), the regularized strain energy is written as
where and are the positive and negative strain energies, and is the degradation function, defined as
with throughout this study.
Thus far, we have not specified the expression of , which accounts for the presence of mesocracks and micropores. Following our previous work (You and Yoshioka, [87]), we employ the phase-field model to capture fractures and the expression of can be specified through a phase-field degraded strain energy density:
The phase-field variable dependent captures the dependency of the Biot moduli on through Eqs. (19), and thus we simply have
The momentum balance and the phase-field evolution equations can be obtained by taking the first variation of the functional with respect to and :
with boundary conditions on and on .
The term in Eq. (32) is the phase-field driving force, which is defined as
where the phase-field derivatives of the Biot moduli, obtained by differentiating Eq. (19) with respect to , are
Eqs. (34)–(36) imply that is invariant with the damage (), which means the micropores do not evolve with damage, and the volume fraction of mesocracks is assumed to be negligible as we consider the penny-shaped cracks with zero volume, unlike -dependent porosity models [87] (Yi et al. 2020). This is consistent with the physical picture that the storage of the material is mainly due to micropores and will not be largely altered by the mesocracks.
For the energy decomposition in this study, we apply the no-tension model [34], where the positive strain () is a positive definite tensor and is coaxial with . Denoting the principal strains with with , we have
where is the eigenvector and parentheses around an index indicate that no sum is taken, i.e., the usual summation convention is suspended. Then the positive and the negative strains are given by
The positive definiteness of depends not only on the sign of but also on the material’s Poisson’s ratio. For the detailed construction of , we refer to Freddi and Royer-Carfagni [34] and the references therein. Using the decomposed strains, the positive and negative strain energies are given as
Note that with the no-tension strain energy decomposition, is anisotropic.
Mass balance equations
The fluid mass balances in mesocracks and micropores can be written as4
where is the fluid density. and are the source or sink terms, and are the fluid velocities in the mesocracks and micropores, respectively. We consider Darcy’s flow at both length scales that are given by
with the permeability for the fluid flow at the mesoscale, the permeability for the fluid flow at the microscale and the gravity. The permeability will be enhanced when the macroscopic fracture appears. We employ a volumetric flux equivalence, in which the permeability is defined as
with being the characteristic element size, the fracture aperture, and the normal direction of the fracture. The normal direction is estimated by the eigen vector of the maximum principal strain, and the fracture aperture is related to the maximum principal strain and the characteristic element size by as You and Yoshioka, [87]
Substituting Eqs. (13), (14), (42) and (43) into Eqs. (40), we have
Furthermore, and are the terms of mass interaction between the two porous scales and satisfy the closure condition . We can approximate the mass transfer as Kazemi et al., [41]
which automatically ensures the closure condition. Here, is a shape factor which depends on the geometry of the mesocracks and micropores.
Solution schemes
This section is devoted to numerical solution of the proposed coupling model, which involves four independent fields, i.e., the displacement , the fluid pressure in mesocracks , the fluid pressure in micropores and the phase-field . We adopt the staggered scheme to solve the phase-field problem, which is widely used in phase-field fracture modeling [1, 13, 55]. In each time step, we first solve the coupled problem with a fixed phase-field variable , and then update the phase-field variable by solving the phase-field equation with the updated displacement and pore pressures. The four governing equations are solved using the finite element method, and the time integration is performed using the backward Euler scheme.
Fixed stress split for the double-porosity problem
We adopt the fixed stress split scheme [42, 87] to decouple the mechanics from the flow problems, and solve them in a staggered manner. The key idea of the fixed stress split scheme is to fix the volumetric stress during the iteration in each time step, which has been shown to be unconditionally stable for single-porosity poromechanics problems Kim et al. [42] with lower-order elements. Here, this scheme needs to be extended to the double-porosity hydromechanical coupling problem.
At the iteration, is frozen with the staggered scheme. According to Eqs. (13) and (14), we have
Applying the backward Euler scheme, Eqs. (49) and (50) at time step and coupling iteration yield
In the fixed-stress scheme, the term, , is expressed with the strain at the previous iteration, by freezing the compliance weighted volumetric total stress during the coupling iteration5:
where is the effective compliance tensor defined as the inverse of . In our proposed double porosity system, we impose the following stress equalities for each scale individually:
which can be simplified with Eqs. (10) and (11) as
Eq. (55) fixes a linear combination of stress components associated with the mesocrack Biot tensor, while Eq. (56) independently fixes the mean stress . Note that for an isotropic , Eq. (56) is a consequence of Eq. (55); for an anisotropic , they are independent.
Substituting the constitutive relation into Eq. (53) and Eq. (54) respectively, we have
which can be further simplified using Eq. (10) and Eq. (11) as (see Appendix B for details)
Using Eqs. (59) and (60), the variations of (49) and (50) can be written as
As can be seen, this system of equations converges to the original system in Eqs. (13) and (14) when and .
Substituting Eqs. (61) and (62) into Eq. (40), we have the temporal discretized equation as
Dynamic update of the solution domain of fracture flow
In the double-porosity framework, the mesocrack pressure and the micropore pressure coexist throughout the domain . However, the fracture flow, Eq. (63), is physically meaningful only within the fractured zone, where the phase field has evolved sufficiently to open a macroscopic fracture. In the intact region, no fracture exists, thereby solving Eq. (63) would introduce spurious fluid transport. To address this, we introduce a dynamic phase-field-dependent indicator function to ensure that the fracture flow domain keep pace with the propagating crack without any user-defined geometric prescription of the fracture path:
where is a prescribed threshold. The condition identifies the fractured zone , while denotes the intact region. The fracture flow Eq. (63) is then replaced by its indicator-weighted counterpart:
where denotes the left-hand side of Eq. (63). Inside the fractured zone (), Eq. (66) reduces to the original fracture flow equation, so is governed by the full poromechanical coupling including Darcy flow and inter-scale fluid exchange. Outside the fractured zone (), Eq. (66) is trivially satisfied, and loses its governing equation. In this region, the mesocrack and micropore networks are in local pressure equilibrium, so we set
Note that this condition holds only for eliminating the inter-scale flux term in Eq. (64) outside the fracture (), recovering the behavior of a single-porosity medium in the intact region.
Variational inequality solution for fluid lag
In hydraulic fracturing, the advancing fracturing tip may outpace the fluid front, leaving a near-tip lag zone where the fracture is mechanically open but devoid of pressurized fluid, known as fluid lag (Figure 1a). Within this lag zone, the fluid pressure reduces to zero (or gas pressure), since fluid cannot sustain tensile stresses (cavitation). Without an explicit non-negativity constraint, the numerical solution of Eqs. (63) and (64) may yield negative pressures in the lag zone, which is physically inadmissible. To preclude such spurious states, the flow sub-problem is reformulated as a variational inequality, enforcing and throughout .
Let . We introduce the admissible convex cones:
which encode the physical requirement that both pressure fields remain non-negative. The variational inequality problem then reads: find such that
where is the left-hand side of Eq. (64).
The Karush–Kuhn–Tucker (KKT) conditions associated with Eqs. (69) and (70) yield the pointwise complementarity system
These conditions naturally partition the domain into two zones. In the fluid-filled zone, and the flow equation is satisfied as an equality (). In the lag zone, and the residual satisfies . The problems, Eqs. (71) and (72) are solved numerically by using the reduced space active set solvers of PETSc ([4]) for variational inequalities based on Newton’s method.
Numerical verification
In this section, we focus on the plane-strain hydraulic fracturing problem (KGD problem) and present three numerical examples to verify the proposed model across different regimes, as in Garagash [36]. The first example focuses on toughness-dominated hydraulic fracturing (K-vertex)6 without leak-off (i.e., ) to demonstrate that the proposed model recovers the solutions achieved by standard single-continuum phase-field hydraulic fracturing models. The second example illustrates the capability of the proposed framework to simulate viscosity-dominated hydraulic fracture (M-vertex), a regime that existing phase-field models struggle to capture. The final example demonstrates that the proposed model successfully captures fluid lag (O-vertex), a phenomenon that, to the best of our knowledge, has not yet been studied by any existing phase-field models.
Closed-form expression
We first present the closed-form expressions. By using viscosity scaling and considering the far-field confining pressure , the half-length of the fracture , the half-length of fluid-filled region , the net pressure and the fracture opening can be expressed in terms of the dimensionless terms as follows ([37]):
where the dimensionless spatial coordinate is and the viscosity scaling is defined as
where the parameters , and are defined as
and , and are the fluid dynamic viscosity, injection flow rate, and critical energy release rate, respectively. For simplicity, we consider zero vapor pressure, i.e., in this study.
The dimensionless terms , , , and in Eqs. (73)–(76) depend on the propagation regime. These self-similar asymptotic solutions for the plane-strain Kristianovich–Geertsma–de Klerk (KGD) problem at the M-vertex ([38]), the K-vertex ([36]), and the early-time O-vertex ([37]) not only have served as canonical benchmarks for numerical models but also have advanced our insight into hydraulic fracture mechanics.
Numerical verification cases
The geometry, boundary conditions, mesh, and initial settings of the computational model are illustrated in Figure 3a. The domain size is 450 m 600 m. The left and right edges are constrained in the normal direction. The top and bottom edges of the model are subjected to confining pressure . All the boundaries are drained except the left one. Fluid is injected at a constant rate of from the left side of the initial fracture which is specified by an initial phase-field value. The length of the initial fracture is 10 m. The elements are refined along the fracture propagation path, with a spacing of m (Figure 3b). Unless otherwise specified, the material properties for the toughness-dominated and viscosity-dominated numerical cases are listed in Table 1.
Figure 3. The geometry and boundary conditions of the plane-strain hydraulic fracture problem (a), and the initial phase-field of the discretized model (b).
| Parameter | Symbol | Unit | Toughness-dominated | Viscosity-dominated |
| Young’s modulus | GPa | 17 | 17 | |
| Poisson’s ratio | — | 0.2 | 0.2 | |
| Phase-field length scale | m | 1.0 | 1.0 | |
| Injection rate | m/s | |||
| Matrix permeability | m | |||
| Biot coefficient of the matrix | — | 0 | 0 | |
| Matrix porosity | — | 0 | 0 | |
| Fluid viscosity | Pa · s | |||
| Critical energy release rate | Pa · m | 120 | 60 | |
| Confining stress | MPa | 0 | 0.8 |
Table 1. Material parameters for the numerical examples.
Toughness-dominated regime
Garagash and Detournay [38] introduced a dimensionless viscosity (defined in Eq. (77)) to distinguish different hydraulic fracture regimes. Note that the value in this study is computed using the effective critical energy release rate with element size m and phase-field length scale m to account for the discretization error in the phase-field approximation of the fracture energy [14, 86].
The toughness- and viscosity-dominated regimes can then be classified according to
under which the self-similar analytical solutions for the fracture length, aperture, and pressure can be derived [38].
Using the parameters in Table 1 for the toughness-dominated case, we have , which is located in the K-vertex. In this simulation, the initial time increment is set to 5 s, followed by uniform increments of 1 s for the remainder of the analysis. This time setting allows fluid to rapidly fill the pre-existing fracture, thereby capturing subsequent fracture propagation more accurately.
Figure 4 shows comparisons between the numerical and analytical solutions for the pressure at the injection point, fracture aperture at the injection point, fracture length, and pressure profile along the fracture at time 100. The numerical results agree well with the analytical solutions, demonstrating that the proposed model can capture the behavior of toughness-dominated hydraulic fracturing.
Figure 4. The analytical and numerical comparisons of the pressure at the injection (a), fracture aperture at the injection point (b), fracture length (c), and pressure profile along the fracture at time 100 s (d) in the toughness-dominated regime.
The small discrepancies in the pressure and aperture at the injection point is attributed to a slight delay of the initiation of the hydraulic fracture as present in the analytical solution. The fracture length matches almost perfectly with the analytical solution, and the pressure distribution along the fracture is uniform, indicating that the overall fracture propagation driven by pressurization is well captured by the numerical model.
Viscosity-dominated regime
For this case, , placing it in the viscosity-dominated regime according to Eq. (80). Unlike the viscosity-dominated solution presented in Garagash and Detournay [38], which assumes no fluid lag, our numerical model does not prohibit the fluid lag development. Thus, for a fair comparison with the analytical solution, we applied a confining stress of 0.8 MPa to the top and bottom boundaries to prevent the fracture from propagating too rapidly, which minimizes the fluid lag.
The simulation employed an initial time increment of 10 s, followed by uniform increments of 1 s. Figure 5 compares the numerical and analytical solutions for the injection pressure, fracture aperture at the injection point, fracture length evolution, and the spatial pressure profile along the fracture at 100 s. The close agreement demonstrates that the proposed model captures the behavior of viscosity-dominated hydraulic fracturing. Notably, the breakdown pressure matches the analytical solution more closely than in the toughness-dominated case (Figure 5a). This is because the energy dissipation in this viscosity-dominated regime is dominated by fluid viscous dissipation, which is less sensitive to the phase-field regularization of the fracture tip. Furthermore, the pressure distribution along the fracture is non-uniform, the highest at the injection point with a steady decrease towards the fracture tip due to the viscous resistance (Figure 5d). This pressure gradient is absent in the toughness-dominated case, where the primary energy dissipation is fracture surface creation rather than viscous fluid flow.
Figure 5. The analytical and numerical comparisons of the pressure at the injection (a), fracture aperture at the injection point (b), fracture length (c), and pressure profile along the fracture at time 100 s (d) in the viscosity-dominated domain.
The pressure distribution along the fracture ()7 agrees well with the analytical solution (Figure 5d), except near the fracture tip (). The analytical solution yields a negative net pressure near the fracture tip because it assumes the absence of a fluid lag zone; consequently, it is only applicable in the region away from the fracture tip [38]. In contrast, the numerical model inherently accommodates the lag effect, and therefore prevents the unphysical negative net pressure near the fracture tip, which explains the deviation from the analytical solution.
In addition to the comparing the fracture aperture evolution at the injection point, the spatial aperture profiles are compared with the analytical solutions from 20 s to 200 s in Figure 6. Our model closely captures increasing evolution of the aperture profiles predicted by the analytical solutions of viscosity-dominated hydraulic fracturing. These spatial and temporal comparisons confirm that both fluid flow and fracture propagation are well captured by the proposed model.
Figure 6. The fracture aperture profiles along the fracture at different time frames from 20 s to 200 s in the viscosity-dominated regime without fluid lag.
Hydraulic fracture with fluid lag
This example examines the behavior of hydraulic fractures with fluid lag. A fluid front may lag behind the fracture tip when the fluid viscosity is high and the confining stresses are low, such as in near-surface hydraulic fracturing ([17]) or magma dykes ([65]). The parameters for this simulation are the same as those in the viscosity-dominated case in Table 1, except that no confining stress is applied to allow fluid lag development (). To investigate how the size of the fluid lag zone varies with fracture toughness, we consider four values of the critical energy release rate, Pa m. The corresponding dimensionless toughness are listed in Table 2. These four cases are chosen to match the values presented by Garagash [37] (Table 2 in their manuscript) in order to evaluate the numerical model against the analytical solutions and explore the influence of toughness on the extent of the fluid lag zone.
| Case | (Pa m) | (Pa m) | |
| 1 | 60 | 71.25 | 0.593 |
| 2 | 42 | 49.88 | 0.496 |
| 3 | 28 | 33.25 | 0.405 |
| 4 | 17 | 20.19 | 0.316 |
Table 2. Critical energy release rate and corresponding dimensionless toughness for the four fluid-lag cases ( m, m). All other parameters are identical to those of the viscosity-dominated case in Table 1.
The initial time increment is 10 s, followed by a uniform increment of 1 s for the remainder of the analysis. For Case 1, Figure 7 presents the evolution of the fracture aperture and the fluid pressure distribution over the time interval [15, 140] s (Figure 7a and b), together with the phase-field and pressure contours at the final time of 140 s (Figure 7c and d). The aperture profiles in Figure 7a exhibit the smooth, concave shape characteristic of viscosity-dominated fracture growth. The opening is greatest at the injection point and decreases monotonically toward the fracture tip. The aperture at increases from approximately m at 15 s to m at 140 s, and the fracture half-length grows from roughly 15 m to 110 m over the same period.
Figure 7. The fracture aperture profile (a) and pressure profile (b) along the hydraulic fractures at different times for Case 1 as listed in Table 2, and the phase-field (b) and pressure (c) distribution at 140 s.
The pressure profiles in Figure 7b reveal two distinct zones along the fracture: a pressurized fluid-filled zone and a near-tip lag zone where the fluid pressure is zero despite the fracture being mechanically open (). The relative length of the lag zone compared with the fracture length remains the same, illustrating the self-similar nature of the analytical solution ([37]). Figure 7c shows the phase-field distribution at 140 s, where the smeared crack band is well resolved without any numerical instability. Figure 7d presents the corresponding pressure field in fracture at 140 s, displaying a pronounced gradient from the injection point towards the fracture tip. The fluid lag can be clearly observed by comparing Figure 7c against Figure 7d as the near-zero pressure region adjacent to the crack tip, corroborating the one-dimensional profiles in Figure 7b. The sharp transition from a pressurized fracture to an unpressurized, open fracture is captured by the proposed model, confirming that the dynamic domain update (§ 3.2) correctly confines the fracture flow equation to the fractured zone () while the variational inequality formulation (§ 3.3) enforces zero pressure in the lag zone without explicitly prescribing the fluid front.
To assess the accuracy of the proposed model, Figure 8 compares the numerically computed aperture and pressure profiles for all four cases against the self-similar analytical solutions of Garagash [36] in dimensionless form. The vertical axes and denote the dimensionless aperture and pressure, normalized by the characteristic scales of the viscosity-dominated (M-vertex) solution. The numerical results of and are post-processed using Eqs. (76) and (75). The time step for this post-processing is 18 s, which is selected to match the analytical solutions of as closely as possible. The analytical family corresponds to dimensionless fluid-front position , where is the ratio of the fluid front to the crack tip location, where corresponds to a fully fluid-filled fracture (zero lag), and decreasing corresponds to a progressively larger lag zone.
Figure 8. The comparisons of the scaled pressure (left column) and fracture aperture (right column) along the fracture between the numerical at s and analytical solutions in the 4 cases, where the analytical solutions for 4 different in each row ranging from 0.3 to 0.6 are extracted from Figure 5 and Figure 6 of Garagash [36].
Figure 8a, Figure 8c, Figure 8e, and Figure 8g show that the numerically calculated aperture profiles for Cases 1–4 fall consistently within the analytical family, ranging from the (Figure 8a) to (Figure 8g). All aperture profiles exhibit the expected smooth decay from a near-unit value at the injection point to zero at the crack tip. The size of the lag zone increases with decreasing , in agreement with the analytical predictions. Figure 8b, Figure 8d, Figure 8f, and Figure 8h reinforce these observations through the pressure profile comparisons. The trend is physically consistent, i.e., lower fracture toughness allows the crack tip to advance more readily relative to the fluid front, enlarging the lag zone.
Although the current model can capture hydraulic fracturing with fluid lag, it underestimates the size of the fluid lag zone, as seen in Figure 8b, Figure 8d, Figure 8f, and Figure 8h, where the simulated fluid front () is more advanced than the corresponding analytical solution, and the pressure profile along the fracture is slightly more diffuse. This discrepancy may be attributed to non-strict enforcement of the mass conservation condition, such as zero flux from the pressurized zone to the lag zone. Additionally, numerical diffusion may also contribute to this discrepancy. Future work could incorporate advanced cavity detection methods, such as Elrod–Adams model (Schultz et al., [69]), or adaptive mesh refinement near the fracture tip to mitigate these effects and improve the accuracy of lag zone simulation.
Conclusions
This paper proposes a variationally consistent dual-continuum phase-field model for hydraulic fracturing in two-scale (double-porosity) porous media, accommodating two independent fluid pressures and whose governing parameters are derived rationally from micromechanics. The fixed-stress split scheme is extended to solve this two-pressure system, while a variational inequality combined with a dynamic domain update enforces non-negative pressures in the fluid lag zone. This formulation effectively describes fluid flow in the fracture zone and accurately reproduces the fluid lag behavior without explicit front tracking. The proposed model was verified against three plane-strain (KGD) solutions across the toughness-dominated (K-vertex), viscosity-dominated (M-vertex), and early-time (O-vertex) regimes. The results demonstrate that the model agrees well with the analytical solutions, correctly captures the parametric dependence of the lag zone on fracture toughness. Furthermore, accounting for the pressure discontinuity across the fracture and the porous media prevents phase-field profile instability under the viscosity-dominated regime (M-vertex). This double-porosity framework provides a promising tool for understanding and predicting
hydraulic fracturing under realistic stress conditions and varying fluid properties, where the viscous dissipation and/or the fluid lag effect may be significant.
For fluid lag cases, the tendency to overestimate the pressurized zone size suggests that incorporating a more rigorous cavitation detection method is a promising direction for future improvement. Beyond these numerical refinements, two physically motivated extensions are of particular interest. First, activating fluid leak-off between the mesocrack network and the micropore space will reveal how the pressure equilibration between the two scales influences fracture propagation and lag-zone evolution in double-porosity media. Second, extending the fracture flow to multiphase conditions, such as accounting for a gas phase in the lag zone or an immiscible displacing fluid, would broaden the applicability of the framework to scenarios such as supercritical CO injection and magma-driven fracturing.
Acknowledgments
This research was funded in whole or in part by the Austrian Science Fund (FWF) 10.55776/PIN9246524. The first author (TY) would like to thank Dr. Weihong Yuan for identifying the anisotropic feature of the Biot coefficient tensor.
Conflict of interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Appendices
Derivation of the Helmholtz free energy for the fluid-filled dual continuum
Considering randomly distributed mesocracks (Figure 2) and fully saturated medium, the potential energy of the two-scale double-porosity material derived by Pichler and Dormieux [63] simplifies to
According to the Legendre transform, the free energy functional as a function of the control variables , and can be written as
Substituting Eqs. (15), (16), and (A.1) into Eq. (A.2), we obtain Eq. (22).
Simplification of the fixed-stress stabilization coefficients
This appendix presents the detailed derivation leading from Eq. (57) to Eq. (59). The goal is to express the operator appearing in Eq. (57) and in Eq. (58) solely in terms of Biot coefficient tensors defined in Eqs. (10)–(11).
We first establish that
In index notation, the left-hand side reads
Invoking the major symmetry of , i.e., , and the inversion relation , i.e., , gives
hence .
Applying to both sides of Eq. (10) yields
Since the major symmetry of implies as second-order tensors, the above relation is equivalently
Substituting Eq. (B.1) into the two contraction terms of Eq. (57) and Eq. (58) gives
Inserting these results into Eqs. (57) and (58) directly yields Eqs. (59) and (60).
References
- [1]Ambati, M., Gerasimov, T., and De Lorenzis, L. (2015). Phase-field modeling of ductile fracture. Computational Mechanics, 55(5):1017–1040.DOI
- [2]Amor, H., Marigo, J.-J., and Maurini, C. (2009). Regularized formulation of the variational brittle fracture with unilateral contact: Numerical experiments. Journal of the Mechanics and Physics of Solids, 57(8):1209 – 1229.DOI
- [3]Anand, L. (2015). 2014 drucker medal paper: A derivation of the theory of linear poroelasticity from chemoelasticity. Journal of Applied Mechanics, 82(11):111005.DOI
- [4]Balay, S., Abhyankar, S., Adams, M. F., Brown, J., Brune, P., Buschelman, K., Dalcin, L., Dener, A., Eijkhout, V., Gropp, W. D., Karpeyev, D., Kaushik, D., Knepley, M. G., May, D. A., McInnes, L. C., Mills, R. T., Munson, T., Rupp, K., Sanan, P., Smith, B. F., Zampini, S., Zhang, H., and Zhang, H. (2019). PETSc users manual. Technical Report ANL-95/11 - Revision 3.11, Argonne National Laboratory.
- [5]Barenblatt, G., Zheltov, I., and Kochina, I. (1960). Basic concepts in the theory of seepage of homogeneous liquids in fissured rocks. Journal of Applied Mathematics and Mechanics, 24(5):1286–1303.DOI
- [6]Baumberger, T., Caroli, C., and Martina, D. (2006). Solvent control of crack dynamics in a reversible hydrogel. Nature materials, 5(7):552–555.pubmed.ncbi.nlm.nih.gov/16751765
- [7]Biot, M. (1941). General theory of three-dimensional consolidation. Journal of Applied Physics, 12(2):155–164.DOI
- [8]Borden, M., Verhoosel, C., Scott, M., Hughes, T., and Landis, C. (2012). A phase-field description of dynamic brittle fracture. Computer Methods in Applied Mechanics and Engineering, 217-220:77–95.DOI
- [9]Borja, R. I. and Koliji, A. (2009). On the effective stress in unsaturated porous continua with double porosity. Journal of the Mechanics and Physics of Solids, 57(8):1182–1193.DOI
- [10]Bouklas, N., Landis, C. M., and Huang, R. (2015). Effect of solvent diffusion on crack-tip fields and driving force for fracture of hydrogels. Journal of Applied Mechanics, 82(8):081007.DOI
- [11]Bourdin, B. (2007). Numerical implementation of the variational formulation for quasi-static brittle fracture. Interfaces and free boundaries, 9(3):411–430.DOI
- [12]Bourdin, B., Chukwudozie, C., and Yoshioka, K. (2012). A variational approach to the numerical simulation of hydraulic fracturing. In SPE annual technical conference and exhibition?, pages SPE–159154. SPE.DOI
- [13]Bourdin, B., Francfort, G., and Marigo, J.-J. (2000). Numerical experiments in revisited brittle fracture. J. Mech. and Phys. of Solids, 48(4):797–826.DOI
- [14]Bourdin, B., Francfort, G., and Marigo, J.-J. (2008). The variational approach to fracture. Journal of Elasticity, 91(1-3):5–148.
- [15]Bourdin, B., Marigo, J. J., Maurini, C., and Sicsic, P. (2014). Morphogenesis and propagation of complex cracks induced by thermal shocks. Physical Review Letters, 112(1):1–5.pubmed.ncbi.nlm.nih.gov/24483901
- [17]Bunger, A. P., Gordeliy, E., and Detournay, E. (2013). Comparison between laboratory experiments and coupled simulations of saucer-shaped hydraulic fractures in homogeneous brittle-elastic solids. Journal of the Mechanics and Physics of Solids, 61(7):1636–1654.DOI
- [18]Carrara, P., Ambati, M., Alessi, R., and De Lorenzis, L. (2020). A framework to model the fatigue behavior of brittle materials based on a variational phase-field approach. Computer Methods in Applied Mechanics and Engineering, 361:112731.DOI
- [19]Choo, J., White, J. A., and Borja, R. I. (2016). Hydromechanical modeling of unsaturated flow in double porosity media. International Journal of Geomechanics, 16(LLNL-JRNL–757002).DOI
- [22]Costa, A., Cusini, M., Jin, T., Settgast, R., and Dolbow, J. E. (2022). A multi-resolution approach to hydraulic fracture simulation. International Journal of Fracture, 237(1-2):165–188.arxiv.org/abs/2207.00661
- [23]Coussy, O. (2004). Poromechanics. John Wiley & Sons.
- [24]De Pater, C., Weijers, L., Savic, M., Wolf, K., Van den Hoek, P., and Barr, D. (1994). Laboratory hydraulic fracturing experiments on rock cores. SPE Production & Facilities, 9(02):116–124.
- [25]Detournay, E. and Garagash, D. (2003). The near-tip region of a fluid-driven fracture propagating in a permeable elastic solid. Journal of Fluid Mechanics, 494:1–32.DOI
- [26]Dormieux, L., Kondo, D., and Ulm, F.-J. (2006a). A micromechanical analysis of damage propagation in fluid-saturated cracked media. Comptes rendus. Mécanique, 334(7):440–446.DOI
- [27]Dormieux, L., Kondo, D., and Ulm, F.-J. (2006b). Microporomechanics. John Wiley & Sons.DOI
- [28]Dougan, C. E., Song, Z., Fu, H., Crosby, A. J., Cai, S., and Peyton, S. R. (2022). Cavitation induced fracture of intact brain tissue. Biophysical Journal, 121(14):2721–2729.
- [29]Economides, M. J. and Nolte, Kenneth G, E. (2000). Reservoir stimulation, volume 2. Wiley, New York.
- [30]Ehlers, W. and Luo, C. (2017). A phase-field approach embedded in the Theory of Porous Media for the description of dynamic hydraulic fracturing. Computer Methods in Applied Mechanics and Engineering, 315:348–368.DOI
- [31]Feng, Y., Haugen, K., and Firouzabadi, A. (2021). Phase-field simulation of hydraulic fracturing by CO₂, water and nitrogen in 2d and comparison with laboratory data. Journal of Geophysical Research: Solid Earth, 126(11):e2021JB022509.DOI
- [32]Feyen, V. and Moelans, N. (2023). Quantitative high driving force phase-field model for multi-grain structures. Acta Materialia, 256:119087.DOI
- [33]Francfort, G. and Marigo, J.-J. (1998). Revisiting brittle fracture as an energy minimization problem. J. Mech. and Phys. of Solids, 46(8):1319–1342.DOI
- [34]Freddi, F. and Royer-Carfagni, G. (2010). Regularized variational theories of fracture: A unified approach. Journal of the Mechanics and Physics of Solids, 58(8):1154–1174.DOI
- [36]Garagash, D. I. (2006a). Plane-strain propagation of a fluid-driven fracture during injection and shut-in: Asymptotics of large toughness. Engineering fracture mechanics, 73(4):456–481.DOI
- [37]Garagash, D. I. (2006b). Propagation of a plane-strain hydraulic fracture with a fluid lag: Early-time solution. International journal of solids and structures, 43(18-19):5811–5835.DOI
- [38]Garagash, D. I. and Detournay, E. (2005). Plane-strain propagation of a fluid-driven fracture: Small toughness solution. Journal of Applied Mechanics, 72(6):916–928.DOI
- [40]Heider, Y. and Markert, B. (2017). A phase-field modeling approach of hydraulic fracture in saturated porous media. Mechanics Research Communications, 80:38–46.DOI
- [41]Kazemi, H., Merrill Jr, L., Porterfield, K., and Zeman, P. (1976). Numerical simulation of water-oil flow in naturally fractured reservoirs. Society of Petroleum Engineers Journal, 16(06):317–326.DOI
- [42]Kim, J., Tchelepi, H. A., and Juanes, R. (2011). Stability and convergence of sequential methods for coupled flow and geomechanics: Fixed-stress and fixed-strain splits. Computer Methods in Applied Mechanics and Engineering, 200(13-16):1591–1606.DOI
- [43]King, G. E. (2010). Thirty years of gas shale fracturing: what have we learned? In SPE Annual Technical Conference and Exhibition?, pages SPE–133456. SPE.DOI
- [44]Lecampion, B., Bunger, A., and Zhang, X. (2018). Numerical methods for hydraulic fracture propagation: A review of recent trends. Journal of natural gas science and engineering, 49:66–83.DOI
- [45]Lecampion, B. and Detournay, E. (2007). An implicit algorithm for the propagation of a hydraulic fracture with a fluid lag. Computer Methods in Applied Mechanics and Engineering, 196(49-52):4863–4880.DOI
- [46]Lee, S., Mikelic, A., Wheeler, M. F., and Wick, T. (2018). Phase-field modeling of two phase fluid filled fractures in a poroelastic medium. Multiscale Modeling & Simulation, 16(4):1542–1580.DOI
- [47]Lee, S., Wheeler, M., and Wick, T. (2016). Pressure and fluid-driven fracture propagation in porous media using an adaptive finite element phase field model. Computer Methods in Applied Mechanics and Engineering, 312:509–541.DOI
- [48]Lee, S., Wheeler, M., and Wick, T. (2017). Iterative coupling of flow, geomechanics and adaptive phase-field fracture including level-set crack width approaches. Journal of Computational and Applied Mathematics, 314:40–60.DOI
- [49]Legarth, B., Huenges, E., and Zimmermann, G. (2005). Hydraulic fracturing in a sedimentary geothermal reservoir: Results and implications. International Journal of Rock Mechanics and Mining Sciences, 42(7-8):1028–1041.DOI
- [53]Liu, J., Liang, X., Xue, Y., Fu, Y., Yao, K., and Dou, F. (2020). Investigation on crack initiation and propagation in hydraulic fracturing of bedded shale by hybrid phase-field modeling. Theoretical and Applied Fracture Mechanics, 108:102651.DOI
- [54]Liu, Y., Yoshioka, K., You, T., Li, H., and Zhang, F. (2025). Variational phase-field fracture approach for non-isothermal co2-water two-phase flow in deformable porous media. Computers and Geotechnics, 188:107596.DOI
- [55]Miehe, C., Hofacker, M., and Welschinger, F. (2010). A phase field model for rate-independent crack propagation: Robust algorithmic implementation based on operator splits. Computer Methods in Applied Mechanics and Engineering, 199(45):2765–2778.DOI
- [56]Miehe, C. and Mauthe, S. (2016). Phase field modeling of fracture in multi-physics problems. part iii. crack driving forces in hydro-poro-elasticity and hydraulic fracturing of fluid-saturated porous media. Computer Methods in Applied Mechanics and Engineering, 304:619–655.DOI
- [57]Miehe, C., Mauthe, S., and Teichtmeister, S. (2015). Minimization principles for the coupled problem of darcy-biոտ-type fluid transport in porous media linked to phase field modeling of fracture. Journal of the Mechanics and Physics of Solids, 82:186–217.DOI
- [58]Mikelić, A., Wheeler, M., and Wick, T. (2015a). A Phase-Field method for propagating fluid-filled fractures coupled to a surrounding porous medium. Multiscale Modeling & Simulation, 48(1):162–186.DOI
- [61]Mōri, A. and Lecampion, B. (2022). Three-dimensional buoyant hydraulic fractures: constant release from a point source. Journal of Fluid Mechanics, 950:A12.DOI
- [62]Nilson, R. H. (1981). Gas-driven fracture propagation. Journal of Applied Mechanics, 48(4):757–762.
- [63]Pichler, B. and Dormieux, L. (2010). Cracking risk of partially saturated porous media—part i: Microporoelasticity model. International journal for numerical and analytical methods in geomechanics, 34(2):135–157.DOI
- [65]Rubin, A. (1995). Propagation of magma-filled cracks. Annual Review Of Earth And Planetary Sciences, Volume 23, pp. 287-336., 23:287–336.DOI
- [66]Santillán, D., Juanes, R., and Cueto-Felgueroso, L. (2017). Phase field model of fluid-driven fracture in elastic media: Immersed-fracture formulation and validation with analytical solutions. Journal of Geophysical Research: Solid Earth, 122(4):2565–2589.DOI
- [67]Sarem, M., Deresse, N. E., Verstrynge, E., and François, S. (2025). Micromechanics-based variational phase-field modeling of fatigue fracture. Journal of the Mechanics and Physics of Solids, 194:105932.DOI
- [68]Schlüter, A., Willenbücher, A., Kuhn, C., and Müller, R. (2014). Phase field approximation of dynamic brittle fracture. Computational Mechanics, 54(5):1141–1161.
- [69]Schultz, M., Orgassa, H., Rom, M., and Müller, S. (2025). New interpretations of the elrod-adams cavitation model. Tribology International, page 110967.
- [71]Sneddon, I. and Lowengrub, M. (1969). Crack problems in the classical theory of elasticity. The SIAM series in Applied Mathematics. John Wiley & Sons.
- [72]Tanné, E., Li, T., Bourdin, B., Marigo, J.-J., and Maurini, C. (2018). Crack nucleation in variational phase-field models of brittle fracture. J. Mech. Phys. Solids, 110:80–99.DOI
- [73]Ulloa, J., Noii, N., Alessi, R., Aldakheel, F., Degrande, G., and François, S. (2022). Variational modeling of hydromechanical fracture in saturated porous media: A micromechanics-based phase-field approach. Computer Methods in Applied Mechanics and Engineering, 396:115084.DOI
- [74]Verhoosel, C. V. and de Borst, R. (2013). A phase-field model for cohesive fracture. International Journal for Numerical Methods in Engineering, 96:43–62.DOI
- [75]Warpinski, N. and Teufel, L. (1987). Influence of Geologic Discontinuities on Hydraulic Fracture Propagation. Journal of Petroleum Technology, 39(02):209–220.
- [76]Warren, J. and Root, P. (1963). The behavior of naturally fractured reservoirs. Society of Petroleum Engineers Journal, 3(3):245–255.DOI
- [77]Wheeler, M., Wick, T., and Wollner, W. (2014). An augmented-Lagrangian method for the phase-field approach for pressurized fractures. Computer Methods in Applied Mechanics and Engineering, 271:69–85.DOI
- [78]Wheeler, M. F., Wick, T., and Lee, S. (2020). IPACS: Integrated phase-field advanced crack propagation simulator. an adaptive, parallel, physics-based-discretization phase-field framework for fracture propagation in porous media. Computer Methods in Applied Mechanics and Engineering, 367:113124.DOI
- [79]Wilson, Z. and Landis, C. (2016). Phase-field modeling of hydraulic fracture. Journal of the Mechanics and Physics of Solids, 96:264–290.DOI
- [81]Xia, L., Yvonnet, J., and Ghabezloo, S. (2017). Phase field modeling of hydraulic fracturing with interfacial damage in highly heterogeneous fluid-saturated porous media. Engineering Fracture Mechanics, 186(October):158–180.
- [82]Xie, N., Zhu, Q.-Z., Shao, J.-F., and Xu, L.-H. (2012). Micromechanical analysis of damage in saturated quasi brittle materials. International Journal of Solids and Structures, 49(6):919–928.DOI
- [85]Yoshioka, K. and Bourdin, B. (2016). A variational hydraulic fracturing model coupled to a reservoir simulator. International Journal of Rock Mechanics and Mining Sciences, 88:137–150.DOI
- [86]Yoshioka, K., Naumov, D., and Kolditz, O. (2020). On crack opening computation in variational phase-field models for fracture. Computer Methods in Applied Mechanics and Engineering, 369:113210.DOI
- [87]You, T. and Yoshioka, K. (2023). On poroelastic strain energy degradation in the variational phase-field models for hydraulic fracture. Computer Methods in Applied Mechanics and Engineering, 416:116305.DOI
- [88]You, T. and Yoshioka, K. (2025). Remote hydraulic fracturing at weak interfaces. Computers and Geotechnics, 177:106830.
- [89]Zhang, J., Chadwick, A. F., Chopp, D. L., and Voorhees, P. W. (2023). Phase field modeling with large driving forces. Npj Computational Materials, 9(1):166.DOI
- [90]Zhu, Q.-Z., Yuan, S.-S., and Shao, J.-F. (2023). Microporomechanics of quasi-brittle rocks: Theoretical formulations and analytical simulations. International Journal of Plasticity, 171:103789.DOI
- [91]Zhuang, X., Li, X., and Zhou, S. (2022). Three-dimensional phase field feature of longitudinal hydraulic fracture propagation in naturally layered rocks under stress boundaries. Engineering with Computers, pages 1–24.DOI
- [92]Zhuang, X., Zhou, S., Sheng, M., and Li, G. (2020). On the hydraulic fracturing in naturally-layered porous media using the phase field method. Engineering Geology, 266:105306.DOI






