Verification and comparisons¶
Morana’s verification evidence combines data-invariant tests, analytic problems, independently assembled references, and file-reader checks. These activities test whether the code implements its documented equations consistently. They do not validate the model against experimental reactor measurements; verification and experimental validation are distinct activities in computational science (Oberkampf and Trucano, 2007).
Not validated for reactor decisions
Morana is intended for research, teaching, and software development. It has not been validated against experimental reactor measurements or qualified for design, operational, licensing, or regulatory decisions. Results must be independently reviewed and validated for their intended use. The example cross sections are illustrative and are not reactor-design data.
Verification cases¶
The detailed derivations, acceptance criteria, and reproducible artifacts for the maintained numerical cases are on separate pages:
| Case | Evidence | Details |
|---|---|---|
| One-dimensional axial core-reflector | Analytic fundamental eigenvalue and source-normalized axial cell averages | One-dimensional axial core-reflector case |
| Fixed-source MMS | Three-group, variable-height refinement with independently evaluated source and Dirichlet data | Fixed-source manufactured-solution case |
| k-effective MMS | One-group, three-dimensional eigenpair refinement with manufactured fission and Robin data | k-effective manufactured-solution case |
The theory and numerical conventions page defines the equations and conventions these cases exercise.
Comparative evidence¶
Comparison cases exercise a complete modeling workflow against an independent calculation without claiming an exact solution, formal benchmark, or experimental validation.
| Case | Comparison | Details |
|---|---|---|
| OpenMC–Morana SRE-derived mini-core | Heterogeneous OpenMC continuous-energy reference against homogenized multigroup Morana diffusion, including energy and axial convergence | OpenMC–Morana SRE-derived comparison |
The OpenMC case reports agreement and remaining systematic differences. It appears in a separate comparison table because neither code supplies an exact reference solution for that model.
Reproducing the checks¶
The contributor workflow gives the exact commands
for pytest and the documentation checks. The test suite exercises the Python
implementation; the documentation checks build the public site, verify links
and generated reference pages, and check spelling and terminology. Maintained
examples provide numerical evidence; the browser-based smoke check verifies
that representative documentation pages and equations render offline.
What automated checks cover¶
Problem definition and geometry¶
- Material, source, boundary, and solver-setting inputs are checked for shape, type, finite values, and physical admissibility at construction or assembly, according to each input’s contract. The public value objects retain owned read-only numerical data.
- Planar and material-mesh checks cover hexagonal geometry, OpenMC ordering, neighbors, variable-height axial stacks, excluded regions, and boundary-face classification.
- Configuration and boundary checks cover complete material coverage, immutable snapshots, group-resolved boundary data, and deterministic boundary-selection precedence.
- OpenMC runtime-MGXS reader tests use an OpenMC-written synthetic fixture and dynamic malformed files. They cover file identity, exact record and temperature selection, both diffusion conventions, compact scattering, optional multiplicity, nonfissionable, separable, and general-transfer fission data, recoverable energy, warnings, and rejected ambiguity or inconsistent metadata without making OpenMC a test dependency.
Operators and solutions¶
- Hand-calculated operator references check removal, scattering, fission, radial and axial leakage, and source and boundary contributions. They cover ragged axial layouts as well as group-major data and packed solver ordering.
- Fixed-source tests compare analytic and direct linear-reference solutions, including downscatter, upscatter, subcritical multiplication, and the supported boundary conditions. They also reject singular, negative-flux, and over-multiplying same-group-scatter responses.
- Criticality tests compare analytic homogeneous cases and a heterogeneous dense-eigenvalue reference. They check normalization, convergence, residuals, balance closure, and supported acceleration behavior.
- Fission-data tests check separable and event-oriented transfer inputs, transfer orientation and production/emission conservation in assembly, equivalent fixed-source and criticality responses, and snapshot and tagged archive preservation of the selected representation.
- Scattering-multiplicity tests establish the unit-multiplicity compatibility limit, off-diagonal and same-group effects, archive provenance, and the assembled fixed-source and criticality identities.
Results and output¶
- Result and archive checks cover immutable layered flux, balance diagnostics, solve provenance, normalization, and corrupted or incompatible archive data.
- Plot checks verify selected group and axial slices, labels, hover data, and excluded-cell handling. Python VTK readers reopen VTU and VTM output to confirm hexagonal-prism geometry, material layout, and flux fields.
Limits of the evidence¶
The automated suite does not constitute experimental validation, qualification for a particular application, or a performance and scalability benchmark.
For exact test and documentation commands, continue with the contributor workflow. For numerical derivations and reproducible artifacts, use the linked verification and comparison pages above.
References¶
Oberkampf and Trucano (2007). W. L. Oberkampf and T. G. Trucano, Verification and Validation Benchmarks, SAND2007-0853, Sandia National Laboratories, 2007. OSTI bibliographic record and open full text.