OpenImpala is a high-performance framework for computing effective transport properties (diffusivity, conductivity, tortuosity) directly on 3D voxel images of porous microstructures.

Architecture

The codebase is organized into three layers:

Layer	Location	Purpose
I/O Readers	`src/io/`	Read TIFF, HDF5, RAW, DAT images into `amrex::iMultiFab`
Physics Solvers	`src/props/`	Solve transport equations, compute effective properties
Fortran Kernels	`src/props/*.F90`	Performance-critical matrix fill and flux calculations

Key Classes

I/O Layer

OpenImpala::TiffReader — Single/multi-directory TIFF and TIFF sequence reader
OpenImpala::HDF5Reader — HDF5 dataset reader with parallel I/O
OpenImpala::RawReader — Flat binary file reader (UINT8, INT16, FLOAT32, etc.)
OpenImpala::DatReader — Legacy DAT binary format reader

Physics Solvers

OpenImpala::Tortuosity — Abstract base class defining the solver interface
OpenImpala::TortuosityHypre — Primary solver: HYPRE-based tortuosity via Laplace equation
OpenImpala::TortuosityDirect — Legacy Forward Euler iterative solver
OpenImpala::EffectiveDiffusivityHypre — Effective diffusivity tensor via homogenization
OpenImpala::VolumeFraction — Phase volume fraction calculator
OpenImpala::PercolationCheck — Flood-fill percolation connectivity check

Configuration & Output

OpenImpala::PhysicsConfig — Maps solver output to physical quantities
OpenImpala::ResultsJSON — Structured JSON output (BPX/BattINFO compatible)

Data Flow

TIFF/HDF5/RAW file
  -> Reader.threshold() -> iMultiFab (phase IDs)
    -> PercolationCheck (connectivity?)
    -> VolumeFraction (phase fraction?)
    -> TortuosityHypre or EffDiffusivityHypre
      -> Fortran kernel fills HYPRE matrix
      -> HYPRE solve
      -> Flux integration -> D_eff, tortuosity
        -> ResultsJSON -> results.json

Dependency Graph

Use the Include Dependency graphs on each file page to trace how modules depend on AMReX (amrex::iMultiFab, amrex::Geometry, etc.) and HYPRE.