Independent applicability review of the historical screen

Prepared 5 September 2026. Retain these results as provisional historical sensitivities. Arithmetic and archive checks pass; physical applicability to Woodfibre has not been established. The very large printed plume heights are separate AERSCREEN diagnostics, not demonstrated atmospheric trajectories. CDGF also carries an explicit AERMOD diameter warning. Neither a successful run nor an absence of warnings validates exposure predictions.

This review read the original run files, EPA source code, the April 2021 AERSCREEN guide, and July 2026 AERMOD formulation/user guides. It performed no dispersion runs and changed no model files.

The first WWGF water/summer/default-temperature case reports:

The large heat is not caused by the 1 g/s normalization. The October 2014 gas flow and heating value reproduce the April 2015 reported heat independently; the tracer emission is normalized while that full heat remains fixed. That establishes internal historical consistency, not the source’s present operation or the realism of holding maximum throughput continuously. Historical source audit, 9 April 2015 response, Table 1, p2.

The reporting calculation is identifiable in AERSCREEN’s PLUMEHGT routine. For positive buoyancy flux Fᵦ >55 m⁴/s³, it selects a final-rise distance xꜰ =119 Fᵦ^0.4 and evaluates a momentum/buoyancy rise expression there. Its additional stable-rise limits apply when the atmosphere is stable, or when the release already lies above the mixing height. In this case, the Monin–Obukhov length is negative and 183.588 <590 m, so those limits do not apply. The routine does not subsequently cap this convective reporting value when it exceeds the mixing height. It is therefore especially unsuitable as a literal plume-altitude prediction for this hot, light-wind case. EPA AERSCREEN source, lines 13775–13901; reporting call at lines 13055–13072.

The archived AERMOD SRCPARAM still contains 183.588 m, not 9,615.9 m. AERMOD’s actual convective concentration calculation separately treats direct, indirect and penetrated plume components, including their mass fractions and distance-dependent rise. Its treatment of penetration is not equivalent to placing all emissions at AERSCREEN’s printed final height. Thus, this diagnostic alone neither proves the concentration calculation erroneous nor validates it. The original concentration run did not export the component-specific rise and penetration diagnostics needed to assess which processes control this peak. EPA AERMOD formulation, §§5.2 and 5.6, printed pp73–78 and 93–96, AERMOD 26135 source, prise.f, DELTAH, CBLPRD, CBLPRN and CBLPR3.

There is a relevant documented formulation uncertainty. EPA’s July 2026 formulation describes an experimental Highly Buoyant Plume (HBP) option, introduced after studies found premature downward mixing and surface overprediction for some penetrated-plume conditions. This does not establish a correction factor or known bias for Woodfibre. HBP uses the change in mixing height between the current and following hour and remains an ALPHA option. Do not enable it casually on MAKEMET’s artificial ordering of meteorological combinations: those adjacent records do not describe the evolution of a real boundary layer. EPA formulation, §B.2.12, printed pp161–162/PDF pp163–164, EPA user guide, HBP entry, printed p3-6/PDF p73.

CDGF warrants an additional explicit flag. Its correctly reconstructed equivalent diameter is 31.335 m. In AERMOD 26135, a diameter above 20 m triggers warning W320 for DS; the program keeps the value. This is a parameter-range warning, not a documented hard flare-validity cutoff or an instruction to truncate diameter. WWGF and LPCF are below that threshold, but this does not establish their validity. No numerical upper heat-release validation limit was identified in the checked AERSCREEN guide/source; its listed heat-input check concerns negative/non-numeric entries. Absence of a ceiling is not evidence of validation at these approximately 0.896–9.353 GW historical heat rates. AERMOD source, soset.f, lines 2872–2881, AERSCREEN guide, Appendix A, printed pA-2/PDF p109.

The independently checked archive contains 40 runs. All stored refined peaks match the single asterisk-marked row in the original distance output. Selecting the earliest rounded maximum would give a different distance in 13 runs. For example, the selected WWGF maximum is at 4,715 m although several printed values tie, beginning at 4,675 m. Thirteen CDGF runs carry W320: the 12 original cases plus one extension. Their logs contain 26 occurrences, reflecting two modelling stages per run; the other 27 runs have no warning messages.

For the three selected cases, extending the search from 20 to 40 km preserves the printed refined peak and its distance:

Each extension has 215 shared distances whose printed concentrations match exactly; the complete grids differ. This establishes the observed extension result for those three cases, not general grid convergence or physical validation. Their printed final-height diagnostics are approximately 9.14, 9.27 and 39.52 km, respectively, reinforcing why those diagnostics must not become mapped plume heights. Independent machine-readable checks.

The next useful applicability check is to export AERMOD’s actual penetration fraction, direct/indirect/penetrated contributions and component heights for the controlling synthetic conditions, then assess the heat/diameter regime against suitable flare evidence and a defensible atmospheric profile. Do not cap diameter, suppress the warning, force the diagnostic to the mixing height, or tune heat to obtain a plausible-looking picture. Current-design source data, terrain, coastal transport, evolving inversions and event timing remain unresolved for a site prediction. These curves can show sensitivity within their stated assumptions; they cannot substantiate neighbourhood risk zones, safe distances, real plume altitudes or a change to the experimental valley plume map.