InterPore2025

Validating a Massively Parallel Dusty Gas Model for Field-scale Modeling of Underground Hydrogen Storage

20 May 2025, 11:35

15m

Oral Presentation (MS23) Advances in Experimental, Computational, and Analytical Approaches for Underground Hydrogen Storage MS23

Speaker

Dr Heeho Park (Sandia National Laboratories)

Description

Geological H2 storage enables long-term (days to months) energy storage and grid services for both grid and off-grid energy feedstock (e.g., solar and wind). Scalable, cost-effective H2 storage can dramatically increase the efficiency of carbon-free energy and further promote usage. Economic storage requires minimizing losses by diffusion through the caprock and contamination by methane and hydrogen sulfide. However, designing and planning realistic porous media storage sites (e.g., depleted gas reservoirs) via simulations has significant challenges due to hydrogen’s exceptionally small molecular cross-section, light weight, and low viscosity.
Subsurface transport simulators typically predict the behavior of gas transport using a combination of Darcy’s Law for advection and Fick’s Law for diffusion and dispersion. This ad hoc advection-diffusion model (ADM) is a simple and effective model for dilute solutions, which is typical in aqueous systems, but ADM is inaccurate for multicomponent gases as it assumes a common solvent velocity. To accurately model transport of gases in a non-dilute multicomponent solution requires not only modeling the concentration gradient driving transport, but also accounting for the relative velocity and cross section of all gases in the solution.
While H2 can migrate upgradient under some circumstances potentially common in a storage environment, H2 flow in porous media can also be dominated by Klinkenberg slip or Knudsen molecular flow while larger and more massive gases (e.g., methane and CO2) behave more conventionally. Slip flow results in larger apparent permeabilities for H2 in the reservoir formation, which is desirable during injection and withdrawal but can lead to unwanted caprock permeation causing losses and contamination.
The dusty gas model is one approach that uses the Maxwell-Stefan approach for multicomponent gas diffusion and extends its applicability to transition and rarefied flow by additionally accounting for collisions with the solid, porous media. This model seamlessly spans multicomponent continuum, transition, and rarefied flow. However, implementation of the dusty gas model is challenging as there are no tractable explicit solutions for the flux of each gas species for non-dilute multicomponent systems. Here, direct linear solvers are utilized to solve for the flux of each species in a multicomponent system. Under the assumption of creeping flow, these equations are time-independent and are be embedded into fully implicit time, finite volume models for mass and energy. This model has been implemented using the finite volume framework of the subsurface flow and transport code PFLOTRAN and massively parallel non-linear solvers of PETSc.
To validate the dusty gas model implementation, three test cases will be presented. The first is comparison is against steady-state experimental data of countercurrent helium and argon transport through low-permeability graphite (3.1⨉10-18 m2) [1][2]. The second comparison is against transient diffusion of a ternary mixture of H2-N2-CO2 [3], demonstrating it is capabile of modeling “reverse” diffusion due to the entrainment. Finally, the scalability to larger systems, such as permeation through caprock in an underground hydrogen storage system, is demonstrated to assess the feasibility of applying this model to geologic systems.
SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525. SAND2024-17021A.