Energy-Controlled PFF Solvers#
This section describes energy-controlled phase-field fracture (PFF) solvers presented in Castillón et al.[1].
Note
Please view the examples related to phase-field fracture Energy-Controlled PFF Solvers.
Traditional PFF strategies, such as displacement or force control, often fail to capture the complete equilibrium path during instabilities like snap-back or snap-through behavior. Displacement-controlled solvers may skip stable equilibrium points, while force-controlled schemes may not trace the path entirely.
To address these issues, Castillón et al.[1] proposes two energy-controlled schemes that robustly trace the equilibrium path during crack propagation by using a monotonically increasing energy-like parameter. These schemes ensure stable progress through instabilities and accurately characterize the system’s response throughout the fracture process.
Variational scheme#
The first approach is a variational scheme driven by an energetic control function, \(\tau(t)\). This is achieved by introducing a constraint that equates \(\tau(t)\) to a weighted combination of crack surface energy and external work. A Lagrange multiplier, \(\lambda\), enforces this constraint:
Here, \(c_1\) and \(c_2\) are numerical parameters for solver convergence and dimensional consistency.
The augmented functional for this constrained system, \(V\), is defined as:
The equilibrium equations are obtained by enforcing stationarity, \(\delta V = 0\). The resulting weak form is:
From the solution \((\boldsymbol{u}, \phi, \lambda)\), the Lagrange multiplier provides key insights. A correction factor \(\alpha\) allows recovering the physical response for a constant \(G_c\):
Non-variational scheme#
Alternatively, a non-variational energy-controlled scheme computes the physical equilibrium path for an effective constant \(G_c\) without post-processing. The governing equations are:
The scalar \(\lambda\) in this scheme is not strictly a Lagrange multiplier but serves a similar role, ensuring that the material toughness \(G_c\) remains constant and allowing the solver to trace the physical force-displacement curve.
Both energy-controlled solvers trace the same physical equilibrium path. The choice between variational and non-variational schemes depends on computational preference, while numerical parameters \(c_1\) and \(c_2\) are used solely for convergence and do not affect physical results.