Woodfibre LNG flare emissions: evidence ledger

Research checked 5 September 2026. This is a starting inventory of primary public documents, not an exhaustive record search. Public PDF dates take precedence over search-engine dates and website update labels.

Main finding: public 2014–2015 environmental-assessment documents contain usable historic flare-specific mass estimates, source geometry, and event assumptions. The current permit describes a materially different design. These historic inputs can support a clearly labelled EA reproduction or sensitivity study; they do not establish the facility's present expected event emissions.

1. Current permit and design version

BCER consolidated LNG facility permit 100105360, read 5 September 2026. First amendment in this 20-page PDF is 8 April 2026; underlying permit is 2 July 2019. The landing page displays an older update date.

The 2025 IAAC project-change analysis, Annex I repeats a 140 m maximum flare stack height in the designated-project description. This is a project envelope, not a confirmed installed or effective plume height. It also allows up to two trains at the EA level, whereas the facility permit allows one: approvals and engineering inputs must be versioned separately.

The currently linked PE-111578 is a construction effluent permit. It is not an operational flare air-emission limit. No current flare-specific air-discharge rate schedule, accepted Flaring Management Report, or final installed flare drawing was found in this initial search. That absence is a retrieval gap, not a claim that such records do not exist.

2. Historic source mass inventory

Appendix 5.2-4, Air Emission Details, Golder report 1314220006-108-Appendix B, 17 October 2014, filed January 2015. Tables 22–23, report pp.17–18 (PDF pp.21–22), visually checked against rendered pages.

Table 23 values are tonnes per maximum activity day, not annual totals and not directly quoted instantaneous g/s.

Each pilot row lists NOₓ and CO <0.0001 t/day. SO₂, PAHs, metals, and the H₂S/HAP column contain dashes for these flare rows: retain these as unquantified in this table, rather than assert zero emissions. TSP, PM₁₀ and PM₂.₅ are nested size categories and must not be added together. NOₓ is not NO₂; total VOC is not a named toxic compound.

Table 22 and §5.9 give the following historic activity inputs:

Linde engineering estimates included a 20% mass contingency. Pilot fuel consumption is stated as 12.5 kg/h. LPWF was not modelled in the upset case. Actual operation was described as one flare at a time, while the assessment model assumed three simultaneously. Section 6, report p.20, describes applying maximum daily emissions throughout the day. However, the modelling appendix’s printed p.5/PDF p.9 separately states that maximum hourly inputs were used for 1-hour/3-hour criteria and daily inputs for 8-hour/24-hour/annual criteria. The documents do not establish that all short-term runs were daily-smoothed. Original hourly source-input files and schedules are needed for exact reproduction. The subsequent historical source consistency audit verifies linked heat and geometry while preserving this timing ambiguity.

Conversion discipline: daily-average g/s = tonnes/day × 1,000,000 / 86,400. A conditional uniform active-event rate would instead divide daily mass by listed active hours × 3,600. The latter is a derived assumption, not a directly reported measurement. For WWGF NOₓ these are 1.0625 g/s and 51.0 g/s respectively. Mixing these averaging conventions would change a short-event exposure estimate by a factor of 48.

3. Historic flare geometry and heat

Appendix 5.2-3, Dispersion Modeling Methodology, Golder report 1314220006-108-Appendix 5.2A, 17 October 2014, Table 6, report pp.12–13. All flare bases have elevation 34.6 m; physical pilot stacks are 100 m above base. The upset model uses the following pseudo-source parameters:

The effective height is a flare-model input; it must not be confused with physical stack height, altitude above sea level, or the final plume centreline after additional modelled rise. Pseudo diameter is not steel pipe diameter. Avoid double-counting plume rise.

Golder response to MOE information requests 53A/53B, 9 April 2015, report 1416711-002-TM-Rev1, Tables 1–2, pp.2–3, confirms the pseudo parameters (diameters rounded) and heat-release inputs:

The response uses the 2008 BC flare-parameter method, a conversion of 252 cal/Btu, fixed 20 m/s and 1,273 K. The subsequent 20 April 2015 response, §2 p.2, argues that the 55% radiant heat loss is already embedded in the pseudo-source equations, so no extra heat-loss multiplier is required. It leaves the parameters unchanged. Section 3 states the original and winter-test model input/output files would be supplied to BC MOE on a portable hard drive: a concrete record set to request. The filename contains a year typo; the document itself is dated 2015.

4. Historic operating-event descriptions

Planned and Emergency Flaring Supplemental Report, Woodfibre LNG, June 2015, §2.1–2.2 report pp.2–3 (PDF pp.5–6). Preliminary design only:

The described derrick reaches approximately 130 m above sea level, north of Mill Creek at the site's northwestern end (Appendix 1 layout). This differs from some other historic/current height conventions; do not combine dimensions without reconciling reference levels and design versions. The report expressly says design information is preliminary. Its frequency assumptions are planning estimates, not measured occurrence probabilities. Report p.13 refers back to the EA air modelling rather than supplying updated event rates.

5. Chemistry and additional public records

Appendix 5.3-1, Greenhouse Gas Methodology, 6 January 2015, Table 9, separately reports flare-event CO₂ and CH₄ rates in g/s: WWGF 193,380 / 1,208; LPCF 49,410 / 174; CDGF 580,278 / 850. These are historic GHG-inventory inputs; check contingency and time weighting before combining with the air-contaminant inventory. CO₂ and methane mass do not substitute for CO, NO₂, particulates, or speciated VOC exposure.

Appendix 5.2-5, VOC Model Predictions, 7 January 2015, Table 1, includes normal-operation predictions for named compounds including benzene, formaldehyde, acetaldehyde and acrolein. These whole-project ambient predictions are not flare-specific emission factors and cannot establish flare VOC composition.

BCER FOI release 2025-014, response letter 27 October 2025, is a 496-page public redacted package concerning gas inlet composition, isolation and emergency planning. Downloaded for subsequent targeted review; its mostly scanned content has not been fully extracted or validated. No unredacted gas composition is asserted here.

FortisBC environmental protection notice, 3 September 2026, application PA 112277, concerns the separate Fortis Squamish compressor station at the Woodfibre site. It belongs in a cumulative-source inventory, not in Woodfibre flare rates.

6. Inputs still needed for a current-event model

1. Accepted Flaring Management Report, final flare basis of design and source-coordinate drawing, operational air-discharge application/permit and technical air assessment.
2. Per-event flare-gas flow time series; fuel composition including sulfur and refrigerants; pilot/purge flows; combustion/destruction efficiency and soot assumptions across wind/load conditions.
3. Per-source hourly NOₓ, NO/NO₂ split, CO, PM₂.₅, SO₂ and individual VOC rates, with units, uncertainty, design version, and event averaging period.
4. Validation of whether the original EA short-event inventory was smeared over 24 hours; retrieve CALPUFF input files and the full 2015 technical-review exchange.
5. Distinct records for thermal oxidizer, Fortis compressor station, vessels, traffic and construction when estimating total ambient concentration. A flare-only increment cannot be presented as total exposure.

Until these gaps are resolved, an interactive model should default to a unit-release tracer or explicitly hypothetical scenario, with historic EA cases separately labelled. It should not show a confident chemical-specific current health-impact footprint.