V&V Casebook · Per-case index

One row per Verification & Validation case. Reference dataset, acceptance metric, status, and links into the live coverage tables, the matching solver entry, and the case input file. Casebook PDFs are regenerated on every release; ◌ marks a PDF that is queued but not yet rebuilt.

Cases total
52
across 4 domains
Dispersion
12
19 / 20 arcs PASS
Radiation
15
15 / 27 ± 15 % wall flux
Fire / spray
10
Tier-1 only · Phase 3 pending
Explosion / BLEVE / toxic
10
Tier-1 PASS · Tier-2 pending
Verification
5
MMS · conservation · grid conv.

Acceptance criteria

Dispersion
Hanna FB ∈ [−0.3, +0.3] · NMSE ≤ 4 · FAC2 ≥ 0.5
Radiation
Wall flux Δ ≤ ±15 % vs published reference
Fire (Tier-1)
Receiver flux Δ ≤ ±20 % vs experimental data
Explosion (Tier-1)
Peak overpressure Δ ≤ ±25 % vs blast curve
Verification
Observed order matches design order; conservation Δ < 1e-9
00 · VERIFICATION MMS · conservation · grid convergence solvers/vv/cases/V-XXX/
V-001 Advection MMS (smooth Gaussian) PASS · order ~2.0
Method
Manufactured solution; Godunov PPM advection only
Reference
Closed-form Gaussian translation
Domain
1024 × 512 × 256 m · refined 16 / 8 / 4 m
coverage tier-2 advection PDF
V-002 Diffusion MMS (Gaussian decay) PASS · order ~2.0
Method
Pure diffusion; CN + MLMG
Reference
Closed-form heat-equation solution
Convergence
L₂ error halves per refinement
coverage tier-2 diffusion PDF
V-003 Mass conservation (closed box) PASS · Δ < 1e-9
Method
Tracer integral over time; periodic walls
Reference
Discrete divergence-free constraint
coverage PDF
V-005 Brinkman → EB consistency (cube) PASS · ≤ 20 % deviation
Method
Same cube; two geometry paths; centerline match
Reference
Cut-cell EB as ground truth
Notes
Closed only after Items 1–4 (IBM wall drag)
coverage tier-2 Brinkman PDF
01 · VALIDATION · DISPERSION Gas dispersion — passive & dense solvers/vv/cases/VC-XXX/ · 19 / 20 PASS
VC-002 Pasquill-D analytic centerline PASS · FAC2 = 0.84
Dataset
Briggs (1973) Pasquill-D rural
Substance
Passive tracer
Wind
4.0 m/s · class D
Notes
Reference baseline before HIT TurbInflow lands
coverage tier-2 disp. tier-1 ref. PDF
VC-003 PG no-AMR 8 m (smooth log-law) PASS · FAC2 = 0.81
Dataset
Prairie Grass (Barad 1958) PG-Neutral mean
Substance
Passive tracer
Notes
Requires amr.max_dt = 0.12 to avoid MLMG divergence
coverage tier-2 disp. PDF
VC-004 PG TurbInflow (HIT urms = 0.4 m/s) PASS · FAC2 = 0.79
Dataset
Prairie Grass · 10 % TI
Substance
Passive tracer
Notes
Bare turbinflows= namespace required (silent-fail trap)
coverage tier-2 turb inflow PDF
VC-005 PG-38 (Barad 1958, u = 4.28 m/s) PASS · FAC2 = 0.84
Dataset
Prairie Grass trial 38
Substance
SO₂ → tracer-equivalent
Domain
1024 × 512 × 256 m · 8 m uniform
Wall time
~3 hr on OMP=2
σy @ 104 m
16.5 m vs PG-D 16.7 m
coverage tier-2 disp. dashboard PDF
VC-005-multi PG-23 / 24 / 38 multi-trial sweep PASS · FAC2 = 0.62 (mean)
Dataset
Prairie Grass trials 23, 24, 38 (u = 4.3 — 6.1 m/s)
Required
D(x) wind scaling — patch 0002
Notes
Without scaling, PG-23 (u = 6.05) over-predicts at x = 800 m by 3.4×
coverage patches PDF
VC-006 AMR + TurbInflow stress FAIL · CFL collapse
Dataset
Prairie Grass · AMR 3-level
Failure
dt → 1e-12 around t ≈ 100–200 s
Workaround
Uniform 8 m grid for turb inflow; smooth log-law for AMR
Issue
Documented limitation; spectral-filter patch 0005 default-off
coverage limitations
VC-008 High-urms TurbInflow (0.6 m/s) PASS · FAC2 = 0.71
Dataset
Synthetic high-TI release
Required
Patch 0003 velocity clamp
Notes
Without clamp: step-1028 collapse even with Hypre bottom solver
coverage patch 0003 PDF
VC-009 / VC-010 Class F / A stable atmosphere WORKAROUND · FAC2 = 0.58
Dataset
Pasquill F (very stable) and A (very unstable)
Workaround
peleLM.gravity = 0; reduced Cs / D encode stability
Real fix
Anelastic / pseudo-incompressible reformulation (upcoming)
coverage tier-2 buoyancy
VC-014 Anisotropic dz = 2 m BLOCKED · solver assert
Failure
PeleLMeX Setup.cpp:48 hard-asserts dx = dy = dz
Estimate
~2 hr solver patch to lift the assertion
coverage limitations
EB-001 → EB-004 CODASC canyon (cut-cell EB) PASS · FAC2 = 0.74
Dataset
CODASC wind tunnel (KIT) urban canyon
Geometry
EB cut-cell + state redistribution
coverage tier-2 EB geometry PDF
DG-CL2 Dense gas — Cl₂ ground release PASS · FAC2 = 0.66
Dataset
Maplin Sands · Cl₂
Buoyancy
Boussinesq + MW; |Δρ/ρ| ≤ 0.03 budget
coverage tier-1 BM tier-2 buoyancy PDF
DG-LNG Dense gas — LNG vapour cloud PASS · FAC2 = 0.69
Dataset
Burro & Coyote (DOE) LNG
Cross-check
HEGADAS-S Tier-1
coverage tier-1 HEGADAS PDF
02 · VALIDATION · RADIATION Radiative transfer — gas + soot solvers/radiation/standalone/inputs.vcr* · 15 / 27 ± 15 % wall flux
MC is the engine of record. P1 is reported alongside as a fast comparator and fails systematically at τ < 0.5. Failures below are P1-only; the MC pass rate is materially higher. See tier2 → MC radiation.
VCR-001 Thynell & Lin (planar slab) PASS · Δ = +4 %
τ regime
~ 1
Method
MC Mode 3 hemisphere wall flux
coverage MC modes PDF
VCR-002 Hsu & Farmer (case A) PASS · Δ = +11 %
τ regime
~ 0.5
Notes
P1 borderline; MC closes
coverage P1 PDF
VCR-003 Selçuk planar furnace PASS · Δ = −6 %
τ regime
~ 2
coverage PDF
VCR-004 Crosbie 3-D box PASS · Δ = −8 %
τ regime
~ 3
coverage PDF
VCR-005 Hsu & Farmer C2 (optically thin) P1 FAIL · Δ = +22 %
τ regime
~ 0.3 (thin)
Notes
P1 known to fail at τ < 0.5; MC passes
coverage switch to MC
VCR-006 Chui cylinder (axisymmetric) PASS · Δ = +9 %
τ regime
~ 1
Geometry
Cylinder primitive
coverage primitives PDF
VCR-007 (k = 0.5 / 1.0) Henson L-shape benchmark PASS · Δ ± 13 %
τ regime
~ 0.5 — 1
Geometry
Reentrant L-shape · primitive scene
coverage primitives PDF
VCR-008 Anisotropic equilibrium (scattering) PASS · Δ = −7 %
Method
Henyey-Greenstein phase function on
coverage PDF
VCR-009 (k = 2 / 5 / 10) Soot-laden gas (Rayleigh κ) PASS · Δ ± 9 %
Coupling
Rayleigh small-particle κ + WSGGM gas
coverage properties PDF
VCR-010 (1800 / 2200 / 2500 K) High-T gas mixture PARTIAL · Δ ± 18 %
τ regime
~ 1 across temperatures
Notes
WSGGM accumulation bug fixed 2026-04-28; remaining error from gas-property fit
coverage WSGGM
VCR-011 (k = 0.5, ε = 1.0) Flux benchmark (low-τ) P1 FAIL · Δ = +24 %
τ regime
~ 0.5
Notes
Same P1 failure mode as VCR-005
coverage switch to MC
AR-Confluence H₂O + CO₂ (Bressloff thesis) PASS · Δ ± 12 %
Dataset
Bressloff (1996) thesis benchmarks
τ regime
1 — 3
coverage WSGGM mix PDF
03 · VALIDATION · FIRE Pool fire · jet fire · flare Tier-1 PASS · Tier-2 Phase 3 pending
F-001 Pool fire — 10 m propane (solid flame) Tier-1 PASS
Reference
Mudan / Rew-Hulbert
Substance
Propane
tier-1 pool fire PDF
F-002 LNG pool fire (Maplin Sands) Tier-1 PASS
Reference
Maplin Sands LNG (HSE)
Tier-2
Pending PeleMP cryogenic spray (Phase 4)
tier-1 pool fire tier-2 cryo
F-003 Jet fire — 50 mm propane Tier-1 PASS
Reference
Chamberlain (1987)
Method
Tilted-shell line source
tier-1 jet fire PDF
F-004 Multi-burner flare Tier-1 PASS
Reference
Cook (1990) multi-burner flare
Method
JetFire superposition
tier-1 multi-flare
F-005 → F-010 NIST firemodels/exp ingest CURATE · Phase 3
Reference
NIST firemodels/exp datasets
Status
To curate to solvers/vv/datasets/firemodels_exp/
coverage
04 · VALIDATION · EXPLOSION · BLEVE · TOXIC Compressible blast + cloud probit Tier-1 PASS · Tier-2 Phase 4–5 pending
E-001 VCE — propane congested (TNO MEM) Tier-1 PASS
Reference
TNO Yellow Book CPR 14E
Class
7 (typical process plant)
tier-1 VCE PDF
E-002 VCE — BST high reactivity Tier-1 PASS
Reference
Baker, Strehlow, Tang (1994)
Cross-input
Same CAM front-door as MEM
tier-1 BST PDF
E-003 BLEVE — propane 10 t fireball Tier-1 PASS
Reference
Roberts (1982); Hymes / Martinsen fits
tier-1 BLEVE PDF
T-001 Toxic — Cl₂ AEGL-2 distance Tier-1 PASS
Reference
NRC AEGL tables · ten-Berge time scaling
tier-1 toxic PDF
T-002 Toxic mixture — sour gas Tier-1 PASS
Reference
TNO Green Book probit · multi-species HI
tier-1 mix PDF
E-004 → E-006 · T-003 → T-005 Tier-2 PeleC compressible PENDING · Phase 4–5
Solver
PeleC compressible reactive · HLLC Riemann
Geometry
PDR required at coarse grid for industrial congestion
coverage tier-2 explosion