What-If Studio · Module 2 — Closure / Detour Shift

Where does traffic exposure move when a road closes? Fixed-demand reassignment on the Grampians study network — exposure & kinetic-energy shifts only, not an outcome prediction. crashstats.com.au family POC.

Closure scenario

Share of daily traffic exposed to closure: 100%

Scales the delta only (e.g. overnight works ≈ 15–25%, peak-only ≈ 40%, full day = 100%). It does not re-estimate demand.

Demand fixedDelta-onlyNo OD inferencePrecomputed scenarios Method: Fixed-demand, all-or-nothing reassignment on a pre-extracted subnetwork (OSM-derived geometry; Vicmap section volumes under the ckpt-63/64 band convention: personal-exposure figures use band-LOW or the measured value, collective-exposure figures use band-HIGH; results shown as ranges). When a link closes, its through volume is re-routed by travel-time shortest path computed offline. Section volumes are two-way AADT; closures are modelled as full-carriageway closures with a 50/50 directional split (single-direction scenarios carry half the two-way volume). Demand is held constant — no OD matrix estimation, no elasticity, no induced demand. Outputs are scenario deltas only.

Re-routed volume

–

Exposure shift (Δ veh-km/day)

–

Kinetic-energy shift proxy

–

Travel-time cost

–

Boundary leakage

–

volume with no in-network detour

closed link gains exposure loses exposure unchanged line weight ∝ |Δ exposure|

Link-level shifts (top gainers / losers)

Road	Length	Base volume (band)	Δ veh/day	Δ vs base

Honesty box. Demand is fixed: this is a full-reassignment stress test, not a guaranteed bound — real closures also shed, retime or divert trips beyond the study area, and drivers may queue or choose other paths. This module reports exposure and kinetic-energy movement only — it makes no claims about harm outcomes; converting exposure shifts into any outcome claim requires a separate, gated evaluation design. Base volumes come from the statewide banded layer (ckpt-63/64): measured sections carry point values in the source layer; banded sections deliberately bracket personal (band-LOW) vs collective (band-HIGH) readings, and 2024 purchased count admissions are still pending, so some corridor volumes will tighten later. A few short town connector / stitch links (kept only to preserve real detour connectivity) have no base exposure band in the source layer; they are shown as “no band” and excluded from base-exposure denominators. Detour roads may differ in design standard, roadside hazard and intersection density in ways exposure and speed proxies do not fully capture. Scenarios are precomputed offline; this page is not a live network model. Western Highway full-closure scenarios were tested and excluded by the leakage kill-rule (>25% of closed volume has no in-network detour at this study-area size — the honest answer is "traffic leaves the area", not a fabricated detour).