Hydrogen and CFD: Fluid Mechanics Modeling for the Hydrogen Industry
Presentation of Principia
Principia is an engineering consultancy with 40 years of existence, specializing in numerical simulation of structures and fluids. The company employs approximately 140 people and generates €16 million in annual revenue. French sites include the headquarters in La Seyne-sur-Mer, with offices in Nantes and Lorient. Principia also has a presence in Kuala Lumpur and a representative office in Tokyo. For the past three to four years, Principia has been part of the Artelia Group, with Naval Group and NMDC as additional shareholders.
Application domains cover offshore (marine renewable energies and conventional), naval (defense and civil), port engineering, nuclear, hydrogen and e-fuels.
Principia is also a software publisher and developer, with around ten scientific software packages developed in-house, as well as a naval architecture team capable of designing floating solutions (barges, wind turbine foundations, dry docks).
CFD at Principia
Computational Fluid Dynamics (CFD) has been a core activity at Principia for over thirty years. The originality of the approach rests on two complementary pillars: the development of in-house models through R&D projects, which consolidates internal expertise, and expert use of leading commercial CFD codes such as Fluent and Star-CCM+.
CFD Applications to Hydrogen
Hydrogen exists in different phases depending on conditions — gaseous, liquid or two-phase — generating distinct modeling challenges.
Risk Analysis: Leaks and Dispersion
The main focus of Principia's hydrogen CFD work is risk analysis: modeling liquid or gaseous hydrogen leaks in systems that transport it, whether in confined environments or open atmospheric conditions.
In open environments, CFD determines the extent of the gas cloud and maps concentrations that could trigger fire or explosion risks — defining ATEX zones. In confined environments, jet velocities can be very high, locally supersonic, with shock zones that require precise modeling.
Liquid Hydrogen Leaks
Liquid hydrogen presents specific thermodynamic characteristics. Upon leaking, contact with the ambient environment causes sudden vaporization (flash). The liquid core of the jet forms a droplet cloud that continues to vaporize. Larger droplets that do not have time to vaporize fall onto surrounding structures, generating extreme thermal stresses that can cause deformation or fracture. CFD enables the definition of appropriate insulation zones by anticipating these leak scenarios.
For this specific challenge, Principia primarily relies on its in-house calculation code, developed and validated on cryogenic fluids over many years, complementing Fluent.
Leak Consequences: Explosion and Fire Propagation
When a leak generates an explosion, a pressure wave is produced and propagates through the system, impacting structures and walls. CFD simulates this wave and couples results with a structural model to assess the mechanical integrity of the system.
Mitigation studies can also be conducted numerically: for example, testing the effectiveness of a blast wall in attenuating the pressure wave before it reaches critical zones.
Fire propagation generates thermal loading on structures, quantifiable through fluid-solid thermal coupling. Dedicated post-processing also evaluates smoke concentration and visibility in areas frequented by personnel.
High-Pressure Tank Filling
A very different challenge concerns the filling of high-pressure gaseous hydrogen tanks. Compressing the gas generates heat that must be controlled to prevent thermal stratification, which could cause wall overheating and deformation risks.
CFD reveals the direct link between filling speed and the onset of stratification: beyond approximately 300 bar, significant fluctuations develop, creating hot zones in the upper part of the tank. The challenge is to find the best compromise between filling speed (typically around ten to fifteen minutes for industrial reasons) and thermal control.
Two specific modeling points are noteworthy for this application: the use of real gas equations of state (ideal gas laws are no longer valid at high pressures), and the multi-layer nature of tank walls (including insulation systems), with thermal coupling between the fluid and the different wall layers.
1D Network Modeling
Alongside 3D simulations, Principia carries out component-level modeling (1D approach) to optimize complex networks: pipes, bends, valves, compressors, tanks. This type of model allows each network element to be optimized to meet process objectives.
What 3D CFD Delivers in Practice
3D modeling discretizes complex geometries into mesh cells, in each of which the Navier-Stokes equations are solved. Calculated fields include velocity, pressure, temperature, species concentrations (hydrogen, combustion products), mass transfers and turbulence effects.
Results take the form of local 3D fields identifying poorly ventilated zones, overheating risk areas or critical concentration zones, as well as transient tracking videos that characterize how phenomena evolve over time.
Once the geometry is meshed and the model configured (initial and boundary conditions), sensitivity studies on input parameters can be run rapidly: leak flow rate, filling speed, tank geometry, presence of internal baffles, and so on.
CFD and Physical Testing: A Complementary Relationship
CFD and physical testing are not opposed but complementary. Preliminary calculations help guide the design of test rigs, reduce the number of tests needed by optimizing the experimental plan, and improve interpretation of results. In return, physical tests provide validation data that consolidate CFD models.
Increasingly, CFD is taking a growing role relative to physical testing, thanks to the gains in flexibility and cost it enables.
Q&A
On Fluent solvers used — All applications presented were carried out with Fluent. The density-based solver appears best suited to problems with strong pressure gradients. For non-standard configurations, user-defined modules allow specific equations of state to be integrated directly into Fluent.
On explosion modeling — The full physics leading to the explosion is modeled: chemical reaction, ignition, generation and propagation of the pressure wave. Pressure field characteristics are calculated throughout the fluid domain and at all walls in contact with it.
On dispersion simulation — The leak is simulated from the source, with a leak flow rate as the boundary condition. The Navier-Stokes model then handles concentration dispersion, accounting for turbulent diffusion.
On the Fluent vs FLACS benchmark — Benchmarks have been conducted on dispersion problems (LNG), comparing Fluent with Principia's in-house code. Fluent is more general-purpose than FLACS and offers better fine geometric representativeness of systems. FLACS offers a practical porosity-based meshing approach for complex geometries, but with less geometric precision.
On meshing tools — Principia has progressively shifted toward Fluent Meshing in recent years, due to its significantly lower mesh generation times compared to ICEM CFD, while retaining ICEM for simulations using the in-house code.
On water-H₂/vapor mixture separation — These complex two-phase modeling challenges are currently under active development at Principia, at the forefront of advances in CFD modeling.
