Victorian Freight Network Intelligence is an independent analytical publication prepared by Traffic Analytics and published by Spotswood Trailers. It is not an official Victorian Transport Association, Department of Transport and Planning or Victorian Government publication.

The method describes the connected public evidence architecture through Stage 22.0.1 and Stage 23A.4.0, together with the Commercial Gold and West Gate question-register presentation layers. Later products consume governed upstream outputs rather than silently redefining them.

Where the reviewed project materials do not expose an exact threshold or component weight, the value is not guessed. The responsible governed stage artifact remains the controlling source.

TIRTL classified traffic evidence

TIRTL supplies classified passing records at monitored locations and headings. A TIRTL passing is one detector observation. It is not a unique truck, a unique load, an origin–destination movement or a complete journey. A vehicle may be observed at more than one monitored point.

Official source: Department of Transport and Planning, TIRTL Traffic Counts. The platform separately identifies the official source-availability window and the governed period used by a published analytical release.

SCATS movement evidence

SCATS contributes long-term signal-derived movement observations for general traffic, intersection and local-road context. A SCATS movement must not automatically be described as a unique vehicle count. Coverage and completeness vary by site and date. The detailed upstream SCATS ingestion and cleaning specification belongs to the established SCATS workflow and is not reconstructed from this freight page.

Coordinates and administrative geography

The foundational TIRTL workflow combines the analytical database with site coordinates and official Vicmap locality polygons. Each coordinate-bearing site-heading is treated as a point; the containing locality polygon supplies the locality name. In the documented foundational build, 562 of 562 site-heading rows were matched and zero remained unmatched.

1. Acquire source files and preserve source names, dates, schema and provenance.
2. Inspect fields, data types, temporal coverage, site identifiers and heading semantics.
3. Import source records into DuckDB and construct repeatable analytical tables.
4. Build site and site-heading masters using governed identifiers and coordinate sources.
5. Convert required spatial boundaries into analysis-ready geometry.
6. Run point-in-polygon and lookup joins for locality, council, electorate or other governed context.
7. Create nightly, hourly, corridor-direction, paired-site and speed-evidence tables.
8. Write CSV, JSON and GeoJSON products before rendering HTML, maps or graphs.

The primary implementation technologies are DuckDB for analytical storage and SQL aggregation, Python for ingestion, validation, spatial processing and export generation, and PowerShell for repeatable orchestration. The public site is static HTML, CSS and JavaScript with Leaflet maps, Google Charts and generated PNG/SVG graph assets.

Maps distinguish exact detector points from inferred analytical corridor lines. An inferred line connecting monitored locations is a presentation of monitored corridor evidence; it is not necessarily a surveyed or road-centreline-snapped engineering alignment.

Raw site-heading evidence

One monitored direction at one location. Used for exact sensor inspection, directional volume, truck share and raw overnight rankings.

Corrected corridor-direction exposure

Repeated monitoring is consolidated into governed corridor-direction components before community or political-geography interpretation. This prevents a corridor with more detector headings from automatically receiving proportionally more weight solely because it is monitored more densely.

Temporal freight pressure

Eligible records are organised across the complete eight-hour overnight window from 10 pm to 6 am. The analysis examines onset, peak, persistence and release, subject to complete-hour, outage, source-boundary and daylight-saving rules.

Paired whole-site evidence

Complementary directions are joined only under a governed paired-site identifier. This supports combined two-way freight scale, dominant-direction share, imbalance and balanced-night measures. Arbitrary headings must not be combined.

Point-speed pressure and reliability

Governed truck speed-bin evidence is reconstructed, validated and assigned an evidence tier before public temporal comparison. Relative point-speed measures are suitable for monitored pressure and reliability screening, but not for complete journey-time claims.

Truck share

Weighted truck share

Observed-to-reference speed

Observed speed deficit

The hourly P85 reference is a frozen observed upper-normal point-speed reference for a governed heading and hour. It is not the posted speed limit, an engineering free-flow speed or a pure pre-opening counterfactual.

Pressure and recurrence

Below-90%, below-80% and below-70% occurrence measures report the share of eligible governed hours below each relative threshold. The governed pressure index is a 0–100 composite point-speed measure derived from relative speed and recurrence evidence.

Paired-site measures

Average normalised eight-hour combined trucks represents governed two-direction passings normalised to the full overnight window. Median directional imbalance is the typical absolute difference between paired directional shares. Balanced nights report the share of eligible paired nights satisfying the governed Stage 22 rule.

Quality gates are applied before evidence enters a public ranking or temporal comparison. Depending on the stage, gates address schema validity, coordinate availability, duplicate handling, complete hours, eligible nights, source boundaries, outages, daylight-saving transitions, evidence density, governed pairing and speed-bin completeness.

The Stage 23A.4 governed population includes 2,152 reference-ready heading-hours, 374,900 complete speed hours, 46,697 eligible heading-nights and 372,586 eligible pressure observations. Heading-nights are accumulated across headings; they are not unique network-wide calendar nights.

Stage 22 contains 119 governed paired sites. Pair construction must remain fixed. A visually plausible pair is not sufficient unless it exists in the governed paired-site register.

The workflow distinguishes a measured zero from absent, incomplete, excluded or non-comparable evidence. These states have different meanings and must remain different in analytical tables and public presentation.

• Measured zero: a valid eligible observation exists and the measured quantity is zero.
• No record: no source row is present for the expected scope.
• Incomplete hour: the hour fails the governing completeness rule.
• Detector outage or source boundary: evidence is unavailable because the monitored source is not valid for that interval.
• Excluded evidence: the record or heading fails an evidence gate.
• Not applicable: the metric does not logically apply to the selected evidence object.
• Not derivable: the available datasets cannot support the requested measure.

Major stages write explicit analytical files before web integration. These may include source manifests, CSV tables, JSON data products, GeoJSON layers, method notes, QA summaries, graphs and stage packages. The public HTML is therefore not the only evidence object.

Web integrations use explicit START/END markers so a later script can replace its own governed block without rewriting unrelated sections. Integration scripts create timestamped backups, update both the framework and published roots, run structural audits and compare generated asset hashes.

Static QA checks include marker uniqueness, required assets, duplicate IDs, internal-anchor integrity, expected source references, parseable JSON, required downloads and framework-to-published file equality. Browser QA should additionally examine console errors, failed requests, map/chart rendering, keyboard behaviour, mobile overflow and performance.

The connected evidence can describe monitored passings, movements, truck share, directional concentration, corrected exposure, temporal profiles, point-speed pressure, threshold recurrence, persistence and relative reliability.

Stage 23B remains blocked until governed segment distances, linked travel times and complete journey definitions are available. Point-speed evidence must not be converted into journey delay, truck-hours or dollars without those inputs.

The Stage 23A.4 frozen reference runs from 1 November 2025 to 21 May 2026. The West Gate Tunnel opened to traffic on 14 December 2025. The reference therefore contains 43 pre-opening nights and 159 nights on or after opening.

The correct classification is a mixed pre-opening, opening-transition and early-operation reference. It is suitable as a frozen reference for later comparisons, but it cannot be described as a pure pre-opening baseline.

A clean opening-effect study requires Stage 23A.5: separately governed pre-opening, transition and stable operating windows, with fresh coverage, density and completeness gates applied to each segment.

This integration publishes every method detail supported by the reviewed project materials. It deliberately does not invent the following:

• exact Stage 17 evidence-class thresholds;
• the exact Stage 20.1 corridor-density correction formula;
• the exact Stage 21 context thresholds and full excluded-night schedule;
• the exact Stage 22 pairing criteria and balanced-night threshold;
• the exact Stage 23A.3 Tier A/B/C threshold values;
• the exact Stage 23A.4 pressure-index weights and sustained-pressure rule;
• the complete upstream SCATS ingestion and cleaning specification;
• the exact construction of the older combined freight-exposure score in the analytics workspace.