Prepared 5 September 2026. This audit concerns the October 2014 assessment design and its April–June 2015 clarifications. The public records support a physically linked, explicitly conditional historical screening case. They do not yet support exact reproduction of the original assessment or a prediction for the present facility. No current simulation defaults were changed.
The reported flare heat and equivalent source dimensions can be reconstructed together. The October 2014 inventory supplies flow and heating value; the April 2015 response supplies heat and equations; the October 2014 modelling appendix supplies the resulting source parameters. These documents agree at their printed precision. Inventory, Tables 22–23, printed pp17–18/PDF pp21–22, 9 April 2015 response, Tables 1–2, pp2–3.
| Historic source | Reported “Exhaust Rate”, kg/h | HHV, Btu/lb | Total heat, cal/s | Recomputed height, m | Recomputed diameter, m |
|---|---|---|---|---|---|
| WWGF — depressuring feed gas | 256,000 | 21,039 | 829,441,536 | 183.5881 | 19.09442 |
| LPCF — shut down Tank RG | 80,000 | 17,382 | 214,146,240 | 143.7567 | 9.70217 |
| CDGF — MR compressor maintenance | 760,000 | 19,099 | 2,235,346,960 | 234.2616 | 31.34628 |
The reconstructed equations are:
The 252 conversion is printed in the April response. The rounded 2.2 conversion is inferred, because it reproduces all three published heat totals; using the exact international-pound conversion instead increases heat by 0.2101%. All heights round to 183.6, 143.8 and 234.3 m, and diameters to 19.094, 9.702 and 31.346 m. The underlying physical height is 100 m above a base at 34.6 m ASL. All three equivalent sources use 20 m/s and 1,273 K. Equivalent height is above that base, not ASL and not the final plume centreline. Modelling appendix, Table 6, printed pp12–13/PDF pp16–17.
The diameter coefficient already represents 45% sensible heat and 55% radiant loss. Applying another 55% loss to its heat input would count that loss twice. The April follow-up explicitly explains this convention. EPA’s AERSCREEN equation uses 9.88×10⁻⁴√((1−loss)×heat), which at loss 0.55 gives coefficient 0.00066277055. Its diameters are consequently 0.0346% smaller than those from the historic rounded coefficient. Preserve this small implementation difference. 20 April 2015 response, pp1–2, EPA AERSCREEN guide, §2.1.2, printed p14/PDF p24.
For an AERSCREEN FLARE input, use physical height 100 m, total heat and loss fraction; do not enter the already adjusted height as physical height. For an equivalent point source, use the adjusted height, diameter, velocity and temperature together. Ordinary plume-rise calculations then act from that equivalent source. EPA SCREEN3 derives the reference buoyancy flux as Fᵦ = 1.66×10⁻⁵ × total heat, with heat in cal/s and flux in m⁴/s³, under its stated thermal assumptions. This verifies the convention, not the facility’s actual radiation fraction or combustion efficiency. EPA SCREEN3 guide, §§2.5 and 3.3, printed pp15 and 38–39.
Emission timing remains conditional. Dividing each Table 23 daily mass by the Table 22 operating duration gives the following uniform active-period rates. These conserve the published daily masses but are not directly recovered original hourly model inputs.
| Source | Assumed active duration | NOx, g/s | CO, g/s | PM₂.₅, g/s | VOC, g/s |
|---|---|---|---|---|---|
| WWGF | 0.5 h | 51.000 | 276.500 | 1.222 | 46.556 |
| LPCF | 14 h | 26.200 | 142.800 | 55.700 | 5.278 |
| CDGF | 0.5 h | 137.111 | 745.944 | 263.000 | 1,555.833 |
The inventory’s §6 says daily emissions were distributed across every hour. However, the modelling appendix’s printed p5/PDF p9 explicitly distinguishes maximum hourly inputs for 1-hour/3-hour criteria from maximum daily inputs for 8-hour/24-hour/annual criteria. It therefore does not support a claim that all original short-term runs were daily-smoothed. The April follow-up says original model inputs, outputs and a readme.docx would be supplied to BC MOE; recovering those files is the direct way to resolve the ambiguity. Modelling appendix, p5, 20 April response, §3, p2.
A steady screen using an active rate represents a sustained-source sensitivity; it does not simulate arrival and cessation of a 30-minute plume. Halving a 30-minute event’s active rate gives its contained one-hour source-mass average, not automatically a one-hour receptor concentration. Likewise, the assessment’s simultaneous three-flare case differs from the inventory’s statement that only one stack would flare at a time. These choices must be named in any run.
The June 2015 preliminary plan introduces different event descriptions: full flaring up to five minutes for unplanned shutdown and ten minutes for emergency depressurization, 20–30-minute loading episodes, and a derrick around 130 m ASL. It supplies no matched revised pollutant/heat time profiles or clear crosswalk to the three 2014 source IDs. Do not combine those durations or height with 2014 event masses as though they formed one documented design case. June 2015 supplement, pp1–3/PDF pp4–6.
Remaining fields include the original hourly inputs and schedules, within-event flow/heat profiles, Linde calculation sheets, exact historical source coordinates and building/downwash inputs, gas composition and species fractions. The inventory already includes a 20% mass contingency; this audit adds none. NOx is expressed on an NO₂ mass basis, which does not establish the emitted NO₂ fraction. Equal TSP/PM₁₀/PM₂.₅ categories are nested and must not be added; unquantified sulphur entries are not zeros. The molecular-weight column’s printed kg/mol unit also requires clarification before deriving volume flow. Inventory, §5.9 and Tables 22–23, Air Quality chapter, §5.2.1.2, printed p5.2-5.
The accompanying calculation record retains unrounded results, separate reported and reconstructed values, daily and conditional active rates, assumptions, source links and local PDF hashes. All three heat, rounded-geometry and mass-conservation checks pass.