WorksPhysics and Astronomy
FESTIM v2.0: Upgraded framework for multi-species hydrogen transport and enhanced performance
Abstract
FESTIM is an open-source finite element framework for modelling the transport of hydrogen isotopes in materials. It provides a flexible and extensible tool for simulating diffusion, trapping, surface interactions, and other processes that govern hydrogen behaviour. This paper presents FESTIM v2.0, a major release that broadens both the physical scope and the software infrastructure of the framework. On the physics side, the formulation adopts a modular structure that supports multi-species transport, advanced trapping and reaction schemes, isotope exchange, decay, and advection. Interface and boundary conditions have been generalised, and interoperability with external solvers enables multiphysics workflows, including coupling with fluid dynamics and neutron transport codes. On the software side, FESTIM v2.0 has been migrated to DOLFINx, the next-generation FEniCS platform, providing improved performance, interoperability, and long-term sustainability. Taken together, these advances position FESTIM v2.0 as a versatile platform for investigating hydrogen transport in materials across scientific and engineering applications.
Introduction
Hydrogen isotope transport plays a central role in the development of fusion energy systems. Accurate modelling of diffusion, trapping, and retention processes is critical for predicting tritium inventories, ensuring regulatory compliance [1, 2], and supporting the design of fuel cycles and plasma-facing components [3]. Robust modelling is also essential for guiding near-term experimental campaigns and validation of transport properties in materials. Hydrogen transport solvers are routinely applied to laboratory-scale experiments such as gas-driven permeation [4, 5, 6, 7], which are used to determine transport properties of candidate materials and coatings, including permeation barriers [8, 9, 10]. They are also used to assess conceptual designs for fuel-cycle components such as breeder blankets [11, 12, 13, 14], plasma-facing components [15, 16, 17], and tritium extraction systems [18, 19, 20], as well as in broader hydrogen-handling technologies including hydrogen storage [21], nuclear fission systems [22], and related materials science studies [23, 24, 25, 26]. These diverse applications highlight the broad scientific importance of accurate, extensible, and efficient hydrogen transport solvers. FESTIM (Finite Element Simulation of Tritium In Materials) was developed to meet these needs in the context of fusion-relevant materials.
FESTIM was first released as an open-source finite element tool tailored to hydrogen transport modelling [27], built on the legacy version of the FEniCS Project (2019.1) [28]. Its Python interface and transparent development model enabled rapid uptake and early community contributions. However, FESTIM v1 faced two fundamental limitations. First, its core dependency (FEniCS 2019.1) has not been maintained since 2019, which has limited FESTIM’s long-term sustainability, scalability, and performance. Second, its treatment of material interfaces relied on a change-of-variable method [29] whose implementation limited efficiency and flexibility. While adequate for small-scale studies, it became increasingly unsuitable for engineering-focused simulations with fine meshes and multiple materials.
The release of FEniCSx, a modern and actively maintained successor optimised for high-performance computing, provided both the necessity and the opportunity for a complete rewrite of FESTIM. FEniCSx is organised as a collection of interoperable modules, including DOLFINx [30] as the problem-solving environment, FFCx for form compilation, Basix [31] for basis function definitions, and UFL [32] to describe finite element forms. Rather than attempting a direct port, FESTIM v2.0 was re-designed from the ground up with a focus on modularity, extensibility, and computational efficiency. This redesign removes the technical debt that had accumulated in v1 and provides a cleaner foundation for extending the physics that FESTIM can describe.
The most significant advance in FESTIM v2.0 is the introduction of a framework that enables fully coupled multi-species transport. This new capability allows users to model multiple hydrogen isotopes and their interactions, including multi-level occupancy trapping, isotopic exchange, decay, and advection. Such features were not possible in FESTIM v1 due to architectural limitations, which restricted extensibility and maintainability through ad-hoc implementations. By contrast, the restructured architecture in v2.0 provides a consistent and modular way to introduce new physics, opening the door to studies of more realistic material behaviour and isotope effects across fusion-relevant conditions. Alongside this development, FESTIM v2.0 introduces more efficient strategies for handling material interfaces. The original change-of-variable approach has been re-implemented to ensure continuity with legacy studies, but FESTIM v2.0 also integrates discontinuous Galerkin formulations, including penalty methods [33, 34] and Nitsche’s method [35, 36]. These approaches, well established in the numerical analysis community, substantially reduce the computational cost of multi-material simulations while preserving accuracy. In addition, work has been done to facilitate the coupling of external modelling tools (OpenMC [37] and OpenFOAM [38]) in FESTIM problems. Together, these advances establish FESTIM v2.0 as both a more capable and more efficient solver for hydrogen transport in materials.
This paper presents FESTIM v2.0 as the new reference for the code and its new capabilities. § 2 provides an overview of the software architecture and development philosophy. § 3–§ 6 describe the mathematical framework, including geometries, governing equations, interface conditions, and boundary condition treatments. § 7 benchmarks performance improvements relative to FESTIM v1, while § 8 demonstrates interoperability with external solvers in multiphysics workflows. Finally, § 9 summarises the advances and outlines perspectives for further development. All examples and data are openly available at https://github.com/festim-dev/FESTIM-v2-review.
FESTIM overview
Philosophy and design principles
FESTIM has been developed as a domain-specific framework for hydrogen isotope transport, rather than a generic finite element solver. Its architecture is guided by three principles: modularity, enabling new physics to be incorporated without altering the core; extensibility, allowing users to adapt or extend functionality for specialised studies; and sustainability, ensuring that FESTIM remains a long-term platform for its users. This philosophy reflects the lessons learnt during the development of FESTIM v1, where useful physics extensions were often introduced as ad-hoc workarounds that limited maintainability. In v2.0, these have been replaced with a clean and systematic structure, providing a robust foundation for scientific research and engineering-scale applications.
Core components
At its foundation, FESTIM builds on the familiar structure of finite element simulations: mesh, governing equations, boundary conditions, and numerical solvers. FESTIM v2.0 extends this foundation with a flexible, object-oriented architecture that assigns physics in a modular manner.
A key innovation is the introduction of Subdomain objects. Whereas v1 applied physics globally, v2.0 can now formulate solutions on a per-subdomain basis. This allows particular processes or governing equations to be restricted to specific regions of the domain. Materials are attached to subdomains rather than the other way around, which improves extensibility and avoids hard-coded links between geometry and physics. This makes multi-material problems more natural to construct and provides a consistent way to define material-dependent diffusivities, solubilities, or reaction networks.
Diffusivity is expressed by default through Arrhenius laws, but users may also supply custom dolfinx.fem.Function’s. This enables fully spatially or temporally varying fields, supporting advanced cases such as heterogeneous materials or turbulent diffusion without modifying the core.
Thermal coupling has also been restructured. In v1, the heat equation was embedded directly within the transport solver, whereas v2.0 separates the two. Heat transfer problems can be solved directly using the new HeatTransferProblem class. Users may still solve the heat equation internally with a dedicated solver, or instead provide external temperature fields and couple FESTIM with third-party thermal codes. This modularity simplifies the architecture while preserving the ability to run fully coupled thermo-diffusion simulations.
Additional flexibility is provided by classes such as ImplicitSpecies, which represent site balances without introducing extra PDEs. This supports efficient modelling of trap occupancies alongside explicit mobile and immobile species. Reactions and source terms are defined through extensible interfaces that accommodate trapping, isotope exchange, decay, surface recombination, and other phenomena relevant to hydrogen transport.
Finally, FESTIM includes a configurable export system. Users can track both raw fields and derived quantities such as fluxes, inventories, or averages over subdomains, ensuring that simulations can be compared directly with experiments or embedded in broader workflows.
Extensibility and integration
The modular class-based design of FESTIM v2.0 ensures that all major components, species, traps, sources, reactions, boundary conditions, and exports are implemented as extensible classes. Users can create custom subclasses to introduce new physics, define non-standard boundary conditions, or extend export routines, without modifying the core code. This resolves one of the main limitations of FESTIM v1, where extensions often relied on modifying internal routines, leading to fragile and inconsistent workflows. The v2.0 design enables FESTIM to integrate into broader workflows, such as parametric optimisation, multiphysics coupling, or system models.
Community and sustainability
FESTIM is openly developed on GitHub using Git version control [39]. The repository supports issue tracking, pull requests, and collaborative review, ensuring contributions are transparent and reproducible. This open model has enabled FESTIM to grow into a multi-institutional project, with contributions from over a dozen researchers across academia and national laboratories.
Modern software practices further reinforce sustainability. An extensive test suite is integrated into FESTIM’s Continuous Integration (CI) pipeline, automatically verifying new contributions at both unit and system levels. Documentation is openly available at https://festim.readthedocs.io, including theoretical background, a user guide, an API reference, and a contributor’s guide. Complementary hands-on tutorials are maintained at https://festim-workshop.readthedocs.io, helping new users explore FESTIM’s capabilities. Stable releases are distributed via PyPI under the permissive Apache-2.0 licence, while development versions can be installed directly from source for early access.
Verification and validation
FESTIM v2.0 includes a dedicated and continuously updated verification and validation (V&V) resource, hosted as a dynamic Jupyter Book at https://festim-vv-report.readthedocs.io. Unlike static publications, this format evolves alongside the code: new benchmark cases can be added as capabilities grow, corrections can be issued immediately, and results remain reproducible and citable. This living V&V record strengthens confidence in FESTIM as a research platform and ensures that its development is directly aligned with experimental validation.
Geometry, meshes and subdomains
In FESTIM, geometries are represented by a mesh that defines the discretised spatial domain together with mesh tags that identify distinct regions and boundaries. These tags enable users to assign materials, sources, or boundary conditions with flexibility. All cell types available in DOLFINx are supported, including higher-order triangles, tetrahedra, quadrilaterals, and hexahedra, enabling both simple structured meshes and curved geometries to be modelled with high fidelity (see Figure 1). Depending on the problem, users may choose between lightweight built-in options for 1D cases, native DOLFINx constructors for regular 1D-3D domains, or external tools for complex geometries.

Figure 1. Illustration of how external meshes can be generated and integrated in FESTIM. A user can interact with FESTIM through any of the supported input formats in DOLFINx. Additionally, meshio can be used to convert files from other formats to XDMF, which can then be read into FESTIM.
For simple 1D studies, FESTIM provides a dedicated Mesh1D class. The user specifies vertex coordinates and optional locators for boundaries, and FESTIM constructs the mesh and associated tags automatically:
import festim as F
import numpy as np
vertices = np.linspace(0, 1, num=100)
my_mesh = F.Mesh1D(vertices=vertices)
More general meshes can be created directly with DOLFINx constructors such as create_interval, create_rectangle, or create_box. Subdomains and boundaries may then be defined using locators or geometric conditions. This provides an efficient workflow for simple problems:
import festim as F
import numpy as np
left_boundary = F.SurfaceSubdomain(id=1, locator=lambda x: np.isclose(x[0], 0.0))
right_boundary = F.SurfaceSubdomain(id=2, locator=lambda x: np.isclose(x[0], 1.0))
bottom_boundary = F.SurfaceSubdomain(id=3, locator=lambda x: np.isclose(x[1], 0.0))
top_boundary = F.SurfaceSubdomain(id=4, locator=lambda x: np.isclose(x[1], 1.0))
For more complex geometries, FESTIM integrates smoothly with meshing tools such as GMSH [40], fTetWild [41], or SALOME [42], which provide advanced capabilities for generating structured or unstructured meshes and support the import of CAD models. Meshes created in SALOME can be exported as .med files and converted to .xdmf with meshio [43], then loaded directly into FESTIM through the MeshFromXDMF class. Alternatively, GMSH meshes in .msh format can be converted natively within DOLFINx into dolfinx.mesh.Mesh and dolfinx.mesh.MeshTags objects, which FESTIM can use via the Mesh class. Within these frameworks, users can either construct meshes manually or generate them from CAD models, enabling the direct simulation of realistic engineering designs with FESTIM.
Governing equations
This section presents the mathematical formulation used by FESTIM to simulate hydrogen transport. As in any standard finite element method (FEM) problem, the solution process begins with the definition of the spatial domain, followed by governing equations, boundary and interface conditions, and the specification of desired outputs. What distinguishes FESTIM v2.0 is its ability to flexibly combine these elements across multiple materials and species, enabling efficient and accurate treatment of complex, multi-physics scenarios.
In hydrogen transport problems in FESTIM v2.0, one or more species can be defined. The following generalised partial differential equation governs the concentration of each species:
where is the concentration of species , and is its diffusion coefficient, which may vary in space and time and can depend on the temperature field . The volumetric source term, , represents external injection or removal of the species, independent of its interaction with other species. The coupling term, , accounts for all local interactions between species, such as trapping and detrapping, isotope exchange, and radioactive decay. This can act as either a sink or a source, depending on the direction of the interaction, and may exhibit non-linear behaviour in species concentrations. The final term, , represents advection of the species by a prescribed velocity field . This formulation allows FESTIM to represent independent diffusion processes, externally imposed sources, and complex coupled reaction networks within a unified framework.
It should be noted that this formulation can be further extended to account for additional phenomena, such as thermophoresis (Soret effect) [44] or stress-assisted diffusion [45].
Multi-species
One of the new features of FESTIM v2.0 is the support of multiple mobile species. Species can be classified into two categories: mobile and immobile. Mobile species can diffuse through the material, while immobile species cannot, i.e. is zero.
In specific cases, it is possible to define a species as implicit, meaning that its concentration can be directly obtained from other species’ concentrations. For instance, for immobile empty trapping sites, for which the concentration can be expressed as:
where is the total concentration of trapping sites, and is the concentration of trapped particles.
trapped_H = F.Species("ct", mobile=False)
n_total = 1e10
empty_traps = F.ImplicitSpecies(n=n_total, others=[trapped_H])
Here, we assume that a single trapping site can only retain one atom; however, this assumption can be extended to multi-occupancy trapping. This new paradigm, in combination with the reaction framework, enables users to simulate the transport of multiple hydrogen isotopes and non-hydrogenic species (e.g., vacancies, vacancy clusters, interstitials).
Reactions
One of the key additions to FESTIM v2.0 is the introduction of reactions. Many physical processes in hydrogen transport can be described as reactions. A reaction is defined by specifying reactants and products, along with forward and backwards rate constants:
Taking into account only the reaction described in equation (3), the temporal evolution of species concentrations is given by:
This framework can be used to model phenomena such as trapping/detrapping, decay, and isotopic exchange. Reactions can also be combined to create complex reaction schemes.
Trapping
The following reversible reaction can represent hydrogen trapping:
where H is mobile hydrogen, is an empty trap site, and is a trapped hydrogen particle. Considering a system with a single mobile species, and a single trapped species, , once the user has defined the reactions, the equations are assembled into rate equations. The temporal evolution of this system is then given by:
where is the number of trapping sites. This is commonly known as the McNabb and Foster model [46]. The rates of trapping and detrapping, and respectively, can be expressed by Arrhenius laws [47]:
where is the trapping pre-exponential factor in , is the trapping energy in eV. is the detrapping pre-exponential factor in , is the detrapping energy in eV, is the Boltzmann constant in .
mobile_H = F.Species("H")
trapped_H = F.Species("trapped_H", mobile=False)
empty_traps = F.ImplicitSpecies(n=n, others=[trapped_H])
trapping_reac = F.Reaction(
reactant=[mobile_H, empty_traps],
product=[trapped_H],
k_0=k_0, E_k=E_k,
p_0=p_0, E_p=E_p,
)
Multilevel trapping
FESTIM supports multilevel trapping, allowing multiple individuals of the same or different species to be trapped at a single site [48]. For instance, mobile hydrogen interacting with a trap with three levels can be described by the following set of reactions:
mobile_H = F.Species("H")
trapped_1H = F.Species("H1", mobile=False)
trapped_2H = F.Species("H2", mobile=False)
trapped_3H = F.Species("H3", mobile=False)
empty_traps = F.ImplicitSpecies(n=10, others=[trapped_1H, trapped_2H, trapped_3H])
reac1 = F.Reaction(
reactant=[mobile_H, empty_traps],
product=[trapped_1H],
k_0=k1,
p_0=p1,
...
)
reac2 = F.Reaction(
reactant=[mobile_H, trapped_1H],
product=[trapped_2H],
k_0=k2,
p_0=p2,
...
)
reac3 = F.Reaction(
reactant=[mobile_H, trapped_2H],
product=[trapped_3H],
k_0=k3,
p_0=p3,
...
)
This can also be extended to multiple species (eg, different hydrogen isotopes):
Isotope swapping
Isotope swapping refers to the exchange of a mobile hydrogen isotope with another isotope that is already trapped:
mobile_H = F.Species("H")
mobile_T = F.Species("T")
trapped_H = F.Species("trapped_H", mobile=False)
trapped_T = F.Species("trapped_T", mobile=False)
swapping_reaction = F.Reaction(
reactant=[mobile_H, trapped_T],
product=[mobile_T, trapped_H],
k_0=k_swap,
p_0=p_swap,
...
)
Other reactions
The generalised FESTIM reaction framework also allows the inclusion of additional chemical reactions in the domain. This capability is particularly relevant in reactive environments, such as molten salt systems (e.g. FLiBe), where hydrogen may partake in chemical transformations.
Radioactive decay can be formulated as a one-way reaction with no products:
with a decay constant in .
T = F.Species("T")
decay_reaction = F.Reaction(
reactant=[T],
product=[],
k_0=decay_constant,
...
)
Advection
To model the transport of hydrogen within a fluid, FESTIM allows the inclusion of an advection term:
where is the velocity field in units of . This formulation captures the influence of a convecting fluid or moving lattice on the spatial distribution of hydrogen species.
Advection can be included directly using the AdvectionTerm class:
F.AdvectionTerm(
velocity=u,
...
)
where is a user-defined dolfinx.fem.Function. Further details on coupling velocity fields from CFD solvers are provided in § 8.1.
To illustrate the impact of advection relative to diffusion, consider a 2D domain with a constant velocity field . The initial condition consists of a species concentration within a circular region of radius 0.1 centred at , and elsewhere. Two diffusion regimes are examined:
: diffusion-dominated (Pe ),
: advection-dominated (Pe ).
Figure 2 shows the concentration fields at and . In the diffusion-dominated regime, the species spreads isotropically. In contrast, in the advection-dominated regime, the concentration distribution is transported downstream by the flow, retaining its shape. Some artificial dissipation of the solution is visible in this case, which is expected when solving the advection-diffusion equation with continuous Galerkin elements in a finite element framework [49].
Figure 2. (a) Initial condition, s; (b) Diffusion dominated, s; (c) Diffusion dominated, s; (d) Advection dominated, s; (e) Advection dominated, s. Influence of advection on species transport in diffusion- and advection-dominated regimes.
Interface conditions
When hydrogen moves through a component composed of multiple materials, it encounters interfaces where transport behaviour can change abruptly. The appropriate interface condition depends on the materials involved, reflecting differences in their solubility laws. For example, at an interface between two Sieverts materials, continuity of hydrogen partial pressure gives:
similarly for two Henry materials:
where is the solubility of hydrogen in a material. If the solubility differs across the interface (), the hydrogen concentration may exhibit a discontinuity.
In FESTIM v1, such conditions were handled by a change-of-variable method, solving for rather than . Although straightforward, this approach had limitations: it required costly projections during post-processing and prevented species from being restricted to specific subsets of the computational domain.
FESTIM v2.0 introduces several alternative strategies for handling interface conditions, providing users with the flexibility to select the method most suitable for their application. Options include discontinuous Galerkin/Nitsche formulations and penalty-based approaches. The following subsections outline these approaches.
Discontinuous Galerkin/Nitsche method
FESTIM v2.0 leverages the submesh functionality provided by DOLFINx [50, 51]. For interfaces between materials of the same type (either Sievert or Henry), a discontinuous Galerkin/Nitsche method is available [52]. To illustrate, consider two species, , , where denotes their interface. The variational formulation of § 4 can then be written as
Find such that:
Here, represents the average operator at the interface, is the jump operator, and is the normal vector on pointing out of . is the stabilisation of penalty parameter and is the local mesh size. Note that on each subdomain , is discretised with a continuous Lagrange space, minimising the number of degrees of freedom needed to represent the discontinuity at . This formulation allows the weak enforcement of continuity conditions across the interface while still accommodating possible discontinuities in when material properties differ.
Penalty methods
For more complex cases, such as Sievert-Henry interfaces, the relation between concentrations becomes nonlinear:
Handling such nonlinear interface conditions requires alternative strategies. FESTIM v2.0 offers penalty-based methods that enhance the weak formulation to enforce the interface constraint efficiently. Two variants are available: an energy-based formulation and a Robin-type boundary condition.
Energy perspective
We consider the same domains and interfaces as above. The strong formulation of the problem is
where .
The corresponding energy minimisation problem can be expressed as:
where
To enforce the non-linear interface condition, the functional is augmented with a quadratic penalty term:
where is a penalty parameter controlling the strength of the constraint. This leads to the following weak formulation by finding the optimality conditions: Find , such that:
This energy-based approach incorporates nonlinear interface conditions into a minimisation problem, which can be solved efficiently in the finite element framework. The penalty term enforces the interface constraint weakly, allowing a controlled approximation while maintaining computational stability.
Change of boundary condition
An alternative penalty formulation is to impose the condition directly as a Robin-type boundary condition on both sides of the interface:
on ,
on
This formulation enforces the interface condition in a weak sense. Starting from the strong form, multiplication by the corresponding test functions and integration by parts yields the weak formulation
Boundary conditions
FESTIM v2.0 provides a comprehensive set of boundary condition treatments. These range from simple fixed concentrations and prescribed fluxes to a general framework for surface reactions, which derives boundary fluxes directly from chemical processes at material interfaces. This breadth of functionality allows users to model both imposed concentrations and surface kinetics within the same simulation environment.
Fixed concentrations and surface fluxes
Dirichlet boundary conditions specify the value of a species concentration at a boundary. This imposes a constant concentration throughout the simulation at the selected location, which may be used to model perfect sinks, imposed gas-side conditions, or fixed inventory states. FESTIM offers several convenience classes to help users define common fixed-value conditions. These include options for deriving boundary concentrations from pressure-based relationships such as Sieverts’ or Henry’s law, where the surface concentration depends on a given gas pressure and material solubility.
More generally, users can prescribe any arbitrary function of space, time, and temperature to define a fixed boundary value. This is particularly useful in simulations with time-dependent values or temperature-evolving conditions:
my_custom_value = lambda x, t, T: 2 * x[0] + 3 * t**2 + 4 * T
my_fixed_bc = F.FixedConcentrationBC(
value=my_custom_value,
...
)
Similarly, flux boundary conditions can be defined using the ParticleFluxBC class. Here, the boundary condition enforces a prescribed particle flux across the surface. Flux values may depend not only on position, time, and temperature, but also on the concentrations of one or more species:
H = F.Species("H")
my_custom_value = lambda x, t, T, c_H: x[0] / 2 + t ** 2 + 2 * T + 10 * c_H
my_flux_bc = F.ParticleFluxBC(
value=my_custom_value,
species=H,
species_dependent_value={"c_H": H}
...
)
This flexibility enables a wide range of scenarios to be represented, including time-varying injections, decaying sources, and feedback-driven fluxes that dynamically depend on local concentrations.
Surface reactions
One of the key additions in FESTIM v2.0 is the introduction of surface reactions, a generalised framework for defining species fluxes at boundaries. This extends the reaction-based approach described in § 4.2 to surfaces, allowing users to describe a wide range of physical processes through chemical kinetics. A surface reaction is defined by specifying the reactants and products, along with forward and backwards rate constants. For example, a general surface reaction may be written as:
Users define such reactions using the SurfaceReaction class:
A = F.Species("A")
B = F.Species("B")
C = F.Species("C")
react1 = F.SurfaceReaction(
surface=1,
reactant=[A, B],
product=[C],
k_r0=k,
k_d0=p,
...
)
The resulting flux is evaluated automatically and applied as a Neumann-type boundary condition:
This framework can be used to describe recombination and dissociation processes, such as:
Forward and backwards rates can be expressed with Arrhenius-type laws. The flux is therefore expressed as:
Where is the partial pressure of in units Pa.
Surface reactions can involve multiple isotopes, enabling more complex pathways such as:
The corresponding fluxes for H and T can then be expressed as:
Each reaction is declared explicitly by the user, and FESTIM automatically applies the corresponding flux conditions:
H = F.Species("H")
T = F.Species("T")
reac1 = F.SurfaceReaction(
surface=1,
reactant=[H, H],
k_r0=K_r1,
k_d0=K_d1,
...
)
reac2 = F.SurfaceReaction(
surface=1,
reactant=[T, T],
k_r0=K_r2,
k_d0=K_d2,
...
)
reac3 = F.SurfaceReaction(
surface=1,
reactant=[H, T],
k_r0=K_r3,
k_d0=K_d3,
...
)
Finally, isotopic exchange can also be modelled. For example:
In cases where one species dominates (e.g. ), the reverse reaction may be neglected, and the process reduces to a first-order flux in T. This can be implemented using the ParticleFluxBC class, with a user-defined expression based on local tritium concentration and a fixed H background.
Performance benchmark
To illustrate the benefits of the new implementation in FESTIM v2.0, we consider the Diffusion: multi-material benchmark case from the FESTIM verification and validation (V&V) book [53], available at festim-vv-report.readthedocs.io/en/latest/verification/mms/discontinuity.html. This case was selected because it highlights two essential aspects: (i) it directly probes the treatment of material interfaces, which represented a bottleneck in FESTIM v1, and (ii) it has been deliberately configured to run for long physical times on a sufficiently large mesh, thereby exposing performance differences between the available interface treatments. The same example has also been shown to exhibit good numerical scaling, as documented on the example’s page. All input files are openly available.
The benchmark simulates 2D hydrogen diffusion across a multi-material system in the transient mode. The simulation was run with a uniform 300 300 mesh of a unit square, a time step of 0.1 s, and a total simulation time of 200 s. All simulations were executed on a single core of a workstation equipped with an Intel(R) Core(TM) i9-285 K CPU (3.70 GHz) and 191 GB of RAM. This choice of hardware (a Windows workstation rather than a high-performance computing cluster) was deliberate, as it is representative of the environment in which most FESTIM users currently operate. Each configuration was repeated ten times, and the mean runtime was recorded. In FESTIM v1, the only available interface treatment was the change-of-variable method. FESTIM v2.0 retains this method for continuity with legacy studies and introduces alternative methods (see § 5).

Figure 3. Multi-material test case hydrogen field solution
FESTIM v1 required an average runtime of 1256.4 s to complete the case (see Figure 4). With the re-implemented change-of-variable method in FESTIM v2.0, the runtime was reduced to 113.5 s, corresponding to a speed-up of about . The discontinuous Galerkin approaches offered further improvements, achieving average runtimes of 81.1 s with the penalty method and 88.1 s with the Nitsche method, both around faster than FESTIM v1.
Figure 4. Comparison of average runtimes for the Diffusion: multi-material benchmark case [53] using FESTIM v1 and v2.0 with different interface treatments. Results are averaged over ten runs.
These results demonstrate that FESTIM v2.0 delivers both greater flexibility and substantial improvements in computational performance. The change-of-variable method remains useful for continuity with legacy work but becomes less practical in large-scale, engineering-focused studies. By contrast, the discontinuous Galerkin formulations provide robust enforcement of interface conditions while reducing runtimes by almost an order of magnitude relative to FESTIM v1. As part of the ongoing development of FESTIM v2.0, a performance benchmark test suite will be established to systematically verify both numerical scaling and parallel scaling, in line with the design principles of DOLFINx [54]. For the present work, however, such parallel performance studies were outside the scope of this paper. Nevertheless, this step change in efficiency makes high-fidelity simulations of engineering-relevant components feasible, paving the way for multiphysics workflows, uncertainty quantification, and reactor-scale modelling with FESTIM v2.0.
Multi-physics coupling
FESTIM’s ethos in multi-physics modelling is to provide a robust and efficient hydrogen transport solver that integrates seamlessly with specialised external solvers, rather than reproducing their functionality. This approach prioritises interoperability, allowing FESTIM to be combined with established tools for heat transfer, fluid dynamics, and neutron transport.
To facilitate this interoperability, FESTIM v2.0 is accompanied by new companion packages developed within the FESTIM-dev ecosystem: openmc2dolfinx (https://github.com/festim-dev/openmc2dolfinx) and foam2dolfinx (https://github.com/festim-dev/foam2dolfinx). These libraries provide dedicated workflows for importing fields from OpenMC and OpenFOAM into the DOLFINx framework, enabling users to couple FESTIM directly with neutronics and CFD solvers.
Figure Figure 5 illustrates this philosophy: external codes generate fields such as temperature, velocity, or tritium generation, which are passed through the dedicated coupling tools into FESTIM. By focusing on extensibility and interoperability, FESTIM allows users to build multi-physics workflows tailored to their applications without compromising computational efficiency.

Figure 5. Fields passed between OpenMC, OpenFOAM, and FESTIM using the openmc2dolfinx and foam2dolfinx packages. This setup demonstrates FESTIM’s extensibility and its ability to interoperate with specialised solvers in multi-physics workflows.
Coupling CFD codes to FESTIM
FESTIM can incorporate velocity and temperature fields obtained from external CFD solvers to model advection-dominated hydrogen transport. However, CFD solvers such as OpenFOAM and finite element frameworks such as DOLFINx use different mesh and data formats, making direct interoperability non-trivial. To bridge this gap, the foam2dolfinx package was developed as part of the FESTIM-dev ecosystem. It enables FESTIM to use CFD fields directly by providing a lightweight interface between OpenFOAM and DOLFINx.
foam2dolfinx
The foam2dolfinx package streamlines FESTIM-OpenFOAM coupling by converting OpenFOAM fields into DOLFINx functions. It reads case files (.foam) and converts selected fields (e.g. velocity, temperature, turbulent viscosity) into dolfinx.fem.Function objects that can be used directly in FESTIM solvers.
Internally, foam2dolfinx uses PyVista’s POpenFOAMReader [55] to parse OpenFOAM outputs and reconstruct an equivalent DOLFINx mesh by matching OpenFOAM cell connectivity to DOLFINx element topology. Supported cell types include tetrahedra (VTK type 10) and hexahedra (VTK type 12). This direct mapping of cell and vertex connectivity is significantly more performant than interpolation, while still preserving fidelity to the OpenFOAM solution. Fields can be taken at specified timesteps, and subdomains can also be targeted if required, allowing both steady-state and transient simulations to be handled naturally.
A minimal example is shown below:
from foam2dolfinx import OpenFOAMReader
import festim as F
my_reader = OpenFOAMReader(filename="my_case.foam", cell_type=12)
velocity_field = my_reader.create_dolfinx_function(t=2.5, name="U")
...
F.AdvectionTerm(
u=velocity_field
...
)
Here, the OpenFOAM reader loads the velocity field U at simulation time s and converts it into a dolfinx.fem.Function. This field is then passed directly into FESTIM’s advection term, allowing seamless coupling between the two codes. In practice, this workflow is efficient and straightforward: OpenFOAM performs the fluid simulation, foam2dolfinx converts the relevant fields, and FESTIM natively uses them in hydrogen transport calculations.
Lid-driven cavity example
The lid-driven cavity (LDC) setup consists of a square domain with a side length of m. No-slip boundary conditions are applied on the left, right, and bottom walls, while the top wall moves with a constant horizontal velocity m s. The problem is governed by the incompressible Navier–Stokes equations for a viscous fluid, solved in transient mode up to s. The velocity boundary conditions are:
where is the velocity vector. Once the velocity field is computed using OpenFOAM (see Figure 6b), it is exported and passed to FESTIM via foam2dolfinx.
Figure 6. LDC benchmark: geometry, velocity field, and comparison of species concentration fields with and without advection.
For the FESTIM model, the domain is assumed to be filled with a fluid with a diffusion coefficient of m s. A constant temperature of 500 K is imposed, together with a uniform volumetric source term of m s. The outer boundaries are set to a fixed concentration of zero, and the problem is solved in steady state. The mesh used for the hydrogen transport calculation is refined compared with that of the OpenFOAM simulation. Interpolation of the velocity field onto the FESTIM mesh is achieved using native functionality in DOLFINx, which efficiently supports non-matching meshes.
The concentration fields in Figure 6 highlight the influence of advection on the transport process. Without advection, diffusion dominates, producing a symmetric profile with peak concentration at the centre and gradual decay to the zero-concentration boundaries (see Figure 6c). With advection, species are transported more rapidly to the boundaries, and the concentration distribution follows the recirculating cavity flow (see Figure 6d).
This example demonstrates the FESTIM-OpenFOAM workflow. While the LDC problem is a simplified benchmark, it clearly illustrates how velocity fields can be incorporated into hydrogen transport calculations. With foam2dolfinx, the same approach extends naturally to more complex geometries and multi-physics scenarios, reinforcing FESTIM’s extensibility and interoperability.
Coupling neutronics codes to FESTIM
In nuclear fusion systems, neutron and photon transport determine necessary source fields for hydrogen transport, most notably tritium generation and nuclear heating. Such fields are typically computed with Monte Carlo transport codes such as OpenMC [37], which can tally spatial distributions of reaction rates and export the results in VTK format. The openmc2dolfinx package, developed within the FESTIM-dev ecosystem, provides a direct pathway for importing these tallies into FESTIM simulations. By converting OpenMC outputs into dolfinx.fem.Function objects enable nuclear source terms to be mapped seamlessly into hydrogen transport calculations without manual preprocessing.
openmc2dolfinx
The openmc2dolfinx package was developed to facilitate the transfer of OpenMC results into FESTIM. It reads .vtk files produced by OpenMC tallies and converts selected quantities (e.g. tritium generation rates, nuclear heating, or reaction scores) into dolfinx.fem.Function objects. These fields can then be used directly in FESTIM simulations.
Internally, the package relies on PyVista’s VTKDataSetReader to parse OpenMC outputs. Because OpenMC tally values are constant across each mesh cell, the conversion process is more straightforward than for CFD fields: values are mapped cell by cell onto an equivalent DOLFINx mesh without requiring interpolation. Both structured (hexahedral) and unstructured (tetrahedral) tally meshes are supported via dedicated reader classes. This ensures that OpenMC results can be imported into FESTIM efficiently while preserving the fidelity of the transport solution.
A minimal example is shown below:
from openmc2dolfinx import StructuredGridReader
import festim as F
reader = StructuredGridReader("tally_results.vtk")
t_gen_field = reader.create_dolfinx_function(data="(n,Xt)")
...
Tritium = F.Species("T")
my_source = F.Source(
species=Tritium,
value=t_gen_field
Here, the tritium generation tally is read from an OpenMC .vtk file and converted into a dolfinx.fem.Function. This field is then passed directly to FESTIM as a source term for tritium transport.
Lithium cube example
To demonstrate the use of openmc2dolfinx, a benchmark was carried out on a simplified geometry: the irradiation of a lithium cube. This example highlights how neutron transport calculations from OpenMC can be imported into FESTIM and used for tritium transport.
Figure 7: Coupling of OpenMC and FESTIM in the lithium cube example: tritium generation tallied in OpenMC (left) and resulting concentration field computed by FESTIM (right).

Figure 7. (a) Tritium generation field from OpenMC (b) Steady-state tritium concentration from FESTIM
The OpenMC model consists of a lithium cube of side length 120 cm surrounded by a vacuum boundary. A monoenergetic point source of 14.1 MeV neutrons is placed 10 cm outside the cube along the axis, emitting isotropically. The FENDL 3.1d nuclear data library is employed, and tritium generation is tallied using the reaction score on a regular mesh. The resulting 3D tritium generation field is exported in VTK format.
In FESTIM, this field is imported using openmc2dolfinx and converted into a source term for tritium transport. The OpenMC tallies are produced on a structured hexahedral mesh. An equivalent hexahedral mesh is generated in dolfinx, which enables direct mapping of the tally values. In this example, the tritium transport is solved on a refined tetrahedral mesh that is denser than the original tally grid. dolfinx performs the interpolation between the hexahedral and tetrahedral discretisations, illustrating how fields can be transferred between meshes of different cell types (see Figure 7). This interpolated source term is then used in the hydrogen transport simulation. The cube boundaries are set to zero concentration, the temperature is fixed at 300 K, and the problem is solved in steady state.
This example illustrates the workflow for coupling OpenMC and FESTIM using openmc2dolfinx. While the lithium cube setup is simplified, it shows how nuclear source fields can be imported directly into FESTIM and used in hydrogen transport calculations. The same workflow extends to reactor-relevant cases, such as breeding blankets, where neutron fluxes govern tritium production and nuclear heating.
Conclusions
FESTIM v2.0, an open-source finite element framework for modelling hydrogen isotope transport in fusion-relevant materials, has been presented. The new release builds on earlier versions while introducing major extensions in scope and usability. The revised formulation adopts a modular structure, supporting multi-species transport and a wide range of physical processes, including multi-level trapping, isotope exchange, decay, and advection. Flexible interface and boundary condition treatments further extend the range of systems that can be represented within a single framework. Beyond the physics, FESTIM v2.0 incorporates a refactored architecture designed for transparency, reproducibility, and ease of adoption. Improved documentation, version control, and continuous integration practices provide a sustainable foundation for community-driven development. Comprehensive export and post-processing options facilitate direct comparison with experiments and integration with multiphysics workflows.
Together, these advances establish FESTIM v2.0 as a versatile and sustainable platform for hydrogen transport studies in fusion materials. By lowering barriers to entry and providing a reliable computational backbone, the framework enables investigations ranging from fundamental science to engineering-scale tritium management. Future developments will focus on extending physical capabilities, broadening validation, and strengthening collaboration across the growing user community.
Acknowledgments
The design of FESTIM v2.0 was guided by input from the user community. A public discussion forum and monthly development meetings were established to facilitate collaboration, solicit feature requests, and support users. Community feedback highlighted the need for new capabilities, including multi-isotope transport, multi-level trapping, and streamlined coupling to external solvers. These requirements informed the development roadmap and design principles of the new version.
This material is based upon work supported by the National Science Foundation under Grant No. 2449339.
The authors would like to acknowledge funding from the Advanced Research Projects Agency-Energy (ARPA-E) of the U.S. Department of Energy. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
References
- [1]M. R. Gilbert, A. Zacharauskas, P. Almond, N. Scott-Mearns, S. Reynolds, M. Y. Lavrentiev, Fusion waste requirements for tritium control: Perspectives and current research, Fusion Engineering and Design 202 (2024) 114296. doi:10.1016/j.fusengdes.2024.114296. URL https://www.sciencedirect.com/science/article/pii/S0920379624001492DOI
- [2]D. H. Dorman, OPTIONS FOR LICENSING AND REGULATING FUSION ENERGY SYSTEMS, Tech. Rep. SECY-23-0001, U.S. Nuclear Regulatory Commission, Rockville, MD (2023). URL https://www.nrc.gov/docs/ML2227/ML22273A163.pdf
- [3]M. Abdou, M. Riva, A. Ying, C. Day, A. Loarte, L. Baylor, P. Humrickhouse, T. F. Fuerst, S. Cho, Physics and technology considerations for the deuteriumâĂŠtritium fuel cycle and conditions for tritium fuel self sufficiency, Nuclear Fusion 61 (1) (2020) 013001, publisher: IOP Publishing. doi:10.1088/1741-4326/abbf35. URL https://dx.doi.org/10.1088/1741-4326/abbf35
- [4]F. Montupet-Leblond, L. Corso, M. Payet, R. Delaporte-Mathurin, E. Bernard, Y. Charles, J. Mougenot, S. Vartanian, E. A. Hodille, C. Grisolia, Permeation and trapping of hydrogen in Eurofer97, Nuclear Materials and Energy 29 (2021) 101062. doi:10.1016/j.nme.2021.101062. URL https://www.sciencedirect.com/science/article/pii/S2352179121001290
- [5]G. A. Esteban, A. Perujo, L. A. Sedano, K. Douglas, Hydrogen isotope diffusive transport parameters in pure polycrystalline tungsten, Journal of Nuclear Materials 295 (1) (2001) 49–56. doi:10.1016/S0022-3115(01)00486-X. URL https://www.sciencedirect.com/science/article/pii/S002231150100486XDOI
- [6]K. Uehara, K. Katayama, H. Date, S. Fukada, Hydrogen gas driven permeation through tungsten deposition layer formed by hydrogen plasma sputtering, Fusion Engineering and Design 98-99 (2015) 1341–1344. doi:10.1016/j.fusengdes.2015.01.027. URL https://www.sciencedirect.com/science/article/pii/S0920379615000460
- [7]P. Calderoni, P. Sharpe, M. Hara, Y. Oya, Measurement of tritium permeation in hibe (2LiFâÊBeF²), Fusion Engineering and Design 83 (7) (2008) 1331–1334. doi:10.1016/j.fusengdes.2008.05.016. URL https://www.sciencedirect.com/science/article/pii/S0920379608000926
- [8]A. Aiello, A. Ciampichetti, G. Benamati, An overview on tritium permeation barrier development for WCLL blanket concept, Journal of Nuclear Materials 329 (2004) 1398–1402. patent Number: B Place: Amsterdam Publisher: Elsevier Science Bv WOS:000223505200101. doi:10.1016/j.jnucmat.2004.04.205. URL http://www.webofscience.com/wos/alldb/full-record/WOS:000223505200101DOI
- [9]V. NemaniÃN, Hydrogen permeation barriers: Basic requirements, materials selection, deposition methods, and quality evaluation, Nuclear Materials and Energy 19 (2019) 451–457. doi:10.1016/j.nme.2019.04.001. URL https://www.sciencedirect.com/science/article/pii/S2352179119300109
- [10]M. Utili, C. Alberghi, L. Candido, F. Di Fonzo, F. Papa, A. Venturini, Design of the Test Section for the Experimental Validation of Antipermeation and Corrosion Barriers for WCLL BB, Applied Sciences 12 (3) (2022) 1624, number: 3 Publisher: Multidisciplinary Digital Publishing Institute. doi:10.3390/app12031624. URL https://www.mdpi.com/2076-3417/12/3/1624DOI
- [11]J. Dark, R. Delaporte-Mathurin, Y. Charles, E. A. Hodille, C. Grisolia, J. Mougenot, Influence of hydrogen trapping on WCLL breeding blanket performances, Nuclear Fusion (2021). doi:10.1088/1741-4326/ac28b0. URL http://iopscience.iop.org/article/10.1088/1741-4326/ac28b0DOI
- [12]L. Candido, C. Alberghi, F. Moro, S. Noce, R. Testoni, M. Utili, M. Zucchetti, A novel approach to the study of magnetohydrodynamic effect on tritium transport in WCLL breeding blanket of DEMO, Fusion Engineering and Design 167 (2021) 112334. doi:10.1016/j.fusengdes.2021.112334. URL https://www.sciencedirect.com/science/article/pii/S0920379621001101DOI
- [13]X. Zhao, M. Ni, B. Nie, B. Zhang, L. Chen, K. Huang, S. Liu, 3D tritium transport analysis for WCCB blanket based on COMSOL, Fusion Engineering and Design 151 (2020) 114105. doi:10.1016/j.fusengdes.2019.111405. URL https://www.sciencedirect.com/science/article/pii/S0920379619309019DOI
- [14]G. Ferrero, S. Meschini, R. Testoni, A Preliminary CFD and Tritium Transport Analysis for ARC Blanket, Fusion Science and Technology 78 (8) (2022) 617–630, publisher: American Nuclear Society _eprint: https://doi.org/10.1080/15361055.2022.2096365. doi:10.1080/15361055.2022.2096365. URL https://doi.org/10.1080/15361055.2022.2096365
- [15]R. Delaporte-Mathurin, H. Yang, J. Denis, J. Dark, E. A. Hodille, G. D. Temmerman, X. Bonnin, J. Mougenot, Y. Charles, H. Bufferand, G. Ciraolo, C. Grisolia, Fuel retention in WEST and ITER divertors based on CFEETM monoblock simulations, Nuclear Fusion 61 (12) (2021) 126001, publisher: IOP Publishing. doi:10.1088/1741-4326/ac2bbd. URL https://doi.org/10.1088/1741-4326/ac2bbd
- [16]E. A. Hodille, F. Ghiorghiu, Y. Addab, A. Zaloÿnik, M. Minissale, Z. Piazza, C. Martin, T. Angot, L. Gallais, M.-F. Barthe, C. S. Becquart, S. Markelj, J. Mougenot, C. Grisolia, R. Bisson, Retention and release of hydrogen isotopes in tungsten plasma-facing components: the role of grain boundaries and the native oxide layer from a joint experiment-simulation integrated approach, Nuclear Fusion 57 (7) (2017) 076019, publisher: IOP Publishing. doi:10.1088/1741-4326/aa6d24. URL https://dx.doi.org/10.1088/1741-4326/aa6d24
- [17]J. Guterl, Modeling of hydrogen isotopes retention in plasma-facing components for fusion applications, Ph.D. thesis, UC San Diego (2015). URL https://escholarship.org/uc/item/1gt6p37w
- [18]M. Utili, C. Alberghi, L. Candido, F. Papa, M. Tarantino, A. Venturini, TRIEX-II: an experimental facility for the characterization of the tritium extraction unit of the WCLL blanket of ITER and DEMO fusion reactors, Nuclear Fusion 62 (6) (2022) 066036, publisher: IOP Publishing. doi: 10.1088/1741-4326/ac5c74. URL https://doi.org/10.1088/1741-4326/ac5c74
- [19]A. Venturini, F. Papa, C. Alberghi, D. Martelli, M. Utili, Testing of PAV-ONE, a Permeator Against Vacuum mock-up with niobium membrane, in lithium-lead eutectic at 350 °C, Fusion Engineering and Design 200 (2024) 114215. doi:10.1016/j.fusengdes.2024.114215. URL https://www.sciencedirect.com/science/article/pii/S0920379624000681DOI
- [20]T. F. Fuerst, C. N. Taylor, P. W. Humrickhouse, The Source Permeator System and Tritium Transport in the TEX PbLi Loop, Fusion Science and Technology 79 (1) (2023) 77–94, publisher: American Nuclear Society. _eprint: https://doi.org/10.1080/15361055.2022.2090784. doi:10.1080/15361055.2022.2090784. URL https://doi.org/10.1080/15361055.2022.2090784
- [21]A. Zájittel, Hydrogen storage methods, Naturwissenschaften 91 (4) (2004) 157–172. doi:10.1007/s00114-004-0516-x. URL https://doi.org/10.1007/s00114-004-0516-x
- [22]C. Fiorina, I. Clifford, S. Kelm, S. Lorenzi, On the development of multi-physics tools for nuclear reactor analysis based on OpenFOAMÃ: state of the art, lessons learned and perspectives, Nuclear Engineering and Design 387 (2022) 111604. doi:10.1016/j.nucengdes.2021.111604. URL https://www.sciencedirect.com/science/article/pii/S0029549321005562DOI
- [23]T. Franklin, C. Icenhour, P.-C. A. Simon, P. Humrickhouse, F. Kong, L. B. Carasik, New developments and verification of fusion blanket simulation capabilities in the MOOSE framework, Fusion Engineering and Design 218 (2025) 115128. doi:10.1016/j.fusengdes.2025.115128. URL https://www.sciencedirect.com/science/article/pii/S0920379625003254
- [24]N. Marrani, T. Hageman, E. MartÃnez-Pañeda, A neural network machine-learning approach for characterising hydrogen trapping parameters from TDS experiments, International Journal of Hydrogen Energy 167 (2025) 150874. doi:10.1016/j.ijhydene.2025.150874. URL https://www.sciencedirect.com/science/article/pii/S0360319925038741DOI
- [25]A. Dñaz, J. M. Alegre, I. I. Cuesta, E. MartÃnez-Pañeda, A COMSOL framework for predicting hydrogen embrittlement, Part I: Coupled hydrogen transport, Engineering Fracture Mechanics 319 (2025) 111007. doi:10.1016/j.engfracmech.2025.111007. URL https://www.sciencedirect.com/science/article/pii/S0013794425002085
- [26]E. GarcÃa-MacÃas, Z. D. Harris, E. MartÃnez-Pañeda, TDS Simulator: A MATLAB App to model temperature-programmed hydrogen desorption, International Journal of Hydrogen Energy 94 (2024) 510–524. doi:10.1016/j.ijhydene.2024.11.014. URL https://www.sciencedirect.com/science/article/pii/S0360319924046688
- [27]R. Delaporte-Mathurin, J. Dark, G. Ferrero, E. A. Hodille, V. Kulagin, S. Meschini, FESTIM: An open-source code for hydrogen transport simulations, International Journal of Hydrogen Energy 63 (2024) 786–802. doi:10.1016/j.ijhydene.2024.03.184. URL https://www.sciencedirect.com/science/article/pii/S0360319924010218
- [28]M. Alnæs, J. Blechta, J. Hake, A. Johansson, B. Kehlet, A. Logg, C. Richardson, J. Ring, M. E. Rognes, G. N. Wells, The FEniCS Project Version 1.5, Archive of Numerical Software 3 (100) (Dec. 2015). doi:10.11588/ans.2015.100.20553. URL https://journals.ub.uni-heidelberg.de/index.php/ans/article/view/20553
- [29]R. Delaporte-Mathurin, E. Hodille, J. Mougenot, Y. Charles, G. D. Temmerman, F. Leblond, C. Grisolia, Influence of interface conditions on hydrogen transport studies, Nuclear Fusion 61 (3) (2021) 036038. doi:10.1088/1741-4326/abd95f. URL http://iopscience.iop.org/article/10.1088/1741-4326/abd95f
- [30]I. A. Baratta, J. P. Dean, J. S. Dokken, M. Habera, J. S. Hale, C. N. Richardson, M. E. Rognes, M. W. Scroggs, N. Sime, G. N. Wells, DOLFINx: The next generation FEniCS problem solving environment (Dec. 2023). URL https://zenodo.org/records/10447666
- [31]M. W. Scroggs, I. A. Baratta, C. N. Richardson, G. N. Wells, Basix: a runtime finite element basis evaluation library, Journal of Open Source Software 7 (73) (2022) 3982. doi:10.21105/joss.03982. URL https://joss.theoj.org/papers/10.21105/joss.03982DOI
- [32]M. S. Alnæs, A. Logg, K. B. Øylgaard, M. E. Rognes, G. N. Wells, Unified form language: A domain-specific language for weak formulations of partial differential equations, ACM Trans. Math. Softw. 40 (2) (2014) 9:1–9:37. doi:10.1145/2566630. URL https://dl.acm.org/doi/10.1145/2566630DOI
- [33]I. BabuÚka, The finite element method with penalty, Mathematics of Computation 27 (122) (1973) 221–228. doi:10.1090/S0025-5718-1973-0351118-5. URL https://www.ams.org/mcom/1973-27-122/S0025-5718-1973-0351118-5/DOI
- [34]J. W. Barrett, C. M. Elliott, Finite element approximation of the Dirichlet problem using the boundary penalty method, Numerische Mathematik 49 (4) (1986) 343–366. doi:10.1007/BF01389536. URL https://doi.org/10.1007/BF01389536
- [35]P. Hansbo, Nitsche’s method for interface problems in computa-tional mechanics, GAMM-Mitteilungen 28 (2) (2005) 183–206, _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/gamm.201490018. doi:10.1002/gamm.201490018. URL https://onlinelibrary.wiley.com/doi/abs/10.1002/gamm.201490018DOI
- [36]M. Juntunen, R. Stenberg, NitscheÊŽs method for general boundary conditions, Mathematics of Computation 78 (267) (2009) 1353–1374. doi:10.1090/S0025-5718-08-02183-2. URL http://www.ams.org/jourcgi/jour-getitem?pii=S0025-5718-08-02183-2DOI
- [37]P. K. Romano, N. E. Horelik, B. R. Herman, A. G. Nelson, B. Forget, K. Smith, OpenMC: A state-of-the-art Monte Carlo code for research and development, Annals of Nuclear Energy 82 (2015) 90–97, publisher: Elsevier BV. doi:10.1016/j.anucene.2014.07.048. URL https://linkinghub.elsevier.com/retrieve/pii/S030645491400379XDOI
- [38]OpenFOAM. URL www.openfoam.org
- [39]S. Chacon, Pro Git, Apress, Berkeley, CA, 2009. doi:10.1007/978-1-4302-1834-0. URL http://link.springer.com/10.1007/978-1-4302-1834-0DOI
- [40]C. Geuzaine, J.-F. Remacle, Gmsh: A 3-D finite element mesh generator with built-in pre- and post-processing facilities, International Journal for Numerical Methods in Engineering 79 (11) (2009) 1309–1331, _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/nme.2579. doi:10.1002/nme.2579. URL https://onlinelibrary.wiley.com/doi/abs/10.1002/nme.2579DOI
- [41]Y. Hu, T. Schneider, B. Wang, D. Zorin, D. Panozzo, Fast tetrahedral meshing in the wild, ACM Trans. Graph. 39 (4) (2020) 117:1–117:18. doi:10.1145/3386569.3392385. URL https://dl.acm.org/doi/10.1145/3386569.3392385DOI
- [42]Salome (Jul. 2023).
- [43]N. Schlümer, meshio: Tools for mesh files (Jan. 2024). doi:10.5281/zenodo.1288334. URL https://zenodo.org/records/1288334
- [44]D. Dasgupta, S. Blondel, E. MartÃnez, D. Maroudas, B. D. Wirth, Impact of Soret effect on hydrogen and helium retention in PFC tungsten under ELM-like conditions, Nuclear Fusion 63 (7) (2023) 076029, publisher: IOP Publishing. doi:10.1088/1741-4326/acda4a. URL https://doi.org/10.1088/1741-4326/acda4a
- [45]K. S. Reddy, A. Rajagopal, T. Rabczuk, A phase-field model for stress-assisted hydrogen diffusion and fracture: An Abaqus implementation, Theoretical and Applied Fracture Mechanics 139 (2025) 104988. doi:10.1016/j.tafmec.2025.104988. URL https://www.sciencedirect.com/science/article/pii/S0167844225001466
- [46]A. McNabb, P. K. Foster, A New Analysis of the Diffusion of Hydrogen in Iron and Ferritic Steels, Transactions of The Society for Modeling and Simulation International 227 (1963) 618–627.
- [47]G. Alefeld, J. VölkI, S. Amelinckx, V. P. Chebotayev, R. Gomer, H. Ibach, V. S. Letokhov, H. K. V. Lotsch, H. J. Queisser, F. P. Schäfer, A. Seeger, K. Shimoda, T. Tamir, W. T. Welford, H. P. J. Wijn (Eds.), Hydrogen in Metals I, Vol. 28 of Topics in Applied Physics, Springer, Berlin, Heidelberg, 1978. doi:10.1007/3-540-08705-2. URL https://link.springer.com/10.1007/3-540-08705-2DOI
- [48]E. A. Hodille, Y. Ferro, N. Fernandez, C. S. Becquart, T. Angot, J. M. Layet, R. Bisson, C. Grisolia, Study of hydrogen isotopes behavior in tungsten by a multi trapping macroscopic rate equation model, Physica Scripta 2016 (T167) (2016) 014011, publisher: IOP Publishing. doi:10.1088/0031-8949/2016/T167/014011. URL https://dx.doi.org/10.1088/0031-8949/2016/T167/014011
- [49]P. Siegel, R. MosáI, P. Ackerer, J. Jaffre, Solution of the Advection⊎Diffusion Equation Using a Combination of Discontinuous and Mixed Finite Elements, International Journal for Numerical Methods in Fluids 24 (6) (1997) 595–613, eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/%28SICI%291097-0363%2819970330%2924%3A6%3C595%3A%3AAID-FLD512%3E3.0.CO%3B2-I. doi:10.1002/(SICI)1097-0363(19970330)24:6<595::AID-FLD512>3.0.CO;2-I. URL https://onlinelibrary.wiley.com/doi/abs/10.1002/%28SICI%291097-0363%2819970330%2924%3A6%3C595%3A%3AAID-FLD512%3E3.0.CO%3B2-I
- [50]J. Dean, Mathematical and computational aspects of solving mixed-domain problems using the finite element method, Ph.D. thesis, University of Cambridge, Cambridge, UK (2023). URL https://www.repository.cam.ac.uk/handle/1810/367855DOI
- [51]C. Richardson, I. Baratta, J. Dean, A. Jackson, G. Wells, An efficient multigrid solver for finite element methods on multi-GPU systems Publisher: Elsevier (Jul. 2025). doi:10.17863/CAM.120313. URL https://www.repository.cam.ac.uk/handle/1810/387576DOI
- [52]D. N. Arnold, F. Brezzi, B. Cockburn, L. D. Marini, Unified Analysis of Discontinuous Galerkin Methods for Elliptic Problems, SIAM Journal on Numerical Analysis 39 (5) (2002) 1749–1779, publisher: Society for Industrial and Applied Mathematics. doi:10.1137/S0036142901384162. URL https://epubs.siam.org/doi/10.1137/S0036142901384162DOI
- [53]R. Delaporte-Mathurin, J. Santana, FESTIM V&V book, 2024, accepted: 2024-09-09T16:40:43Z. URL https://dspace.mit.edu/handle/1721.1/156690
- [54]R. Delaporte-Mathurin, J. Dark, J. S. Dokken, S. Meschini, Upgrading FESTIM for High-Performance Computing with FEniCSx (Nov. 2025). doi:10.5281/zenodo.17184892. URL https://zenodo.org/records/17184892
- [55]C. B. Sullivan, A. A. Kaszynski, PyVista: 3D plotting and mesh analysis through a streamlined interface for the Visualization Toolkit (VTK), Journal of Open Source Software 4 (37) (2019) 1450. doi:10.21105/joss.01450. URL https://joss.theoj.org/papers/10.21105/joss.01450DOI


