The gap in fire response isn't the trucks - it's where the stations sit. This study grids Greater Melbourne into candidate locations, then runs a coverage-maximising location–allocation model to place new stations. Adding 22 stations to the existing 47 lifts five-minute incident coverage from 81% to 95%. Explore the network below.
Green areas are within 3 minutes of a station, amber within the cutoff, and red is out of coverage. A live simulation overlays real-time demand: incidents ignite, the nearest station dispatches a unit, and the map tallies how many are reached in time. Add the 22 proposed stations to watch the red shrink.
Coverage is measured as the share of fire incidents a station can reach within the cutoff over the real road network.
With 47 existing stations, 81% of incidents are reached within five minutes. Twenty-two gap-filling stations (69 total) raise that to 95% - closing the outer-suburban red zones without over-serving the already-dense inner city.
The gain comes from placing few stations well. Candidate sites are drawn from a regular grid, then only the ones that add the most uncovered demand are kept - a maximise-coverage choice, not blanket construction.
New locations are generated first, then filtered by how much fire demand they cover within reach.
The Reduced-Gaussian grid spreads candidate sites evenly so the optimiser can trade them off objectively, rather than anchoring on where stations already happen to be.
The study found location–allocation over the road network a powerful tool for both adding stations and restructuring existing ones to lift coverage.
Free-flow travel times flatter the network. At the 6 pm peak, congestion shrinks every service area.
Re-running the proposed 69-station network with peak-hour travel times drops five-minute coverage from 95% to 83.7% - an 11.3-point loss precisely when the network is busiest.
Because the shortfall appears only under load, coverage assessed at free-flow speeds overstates real-world protection. Siting and staffing decisions should be stress-tested against peak-hour travel, not average conditions.
The paper also maps the temporal pattern of residential fires. The evening skew matters: it overlaps the traffic peak that erodes coverage.
Residential fires cluster in the late-afternoon and evening hours - cooking, heating and end-of-day activity - which is the same window when road congestion is worst. The two effects compound: demand rises just as the network's five-minute reach contracts. The schematic band below marks that overlap; it is illustrative of the timing, not a reproduction of the paper's hourly counts.
The inner city is already well served. The coverage payoff sits on the fringe, where existing stations leave five-minute holes - that is where new stations earn their keep.
Generate candidates systematically and let a maximise-coverage model choose. It surfaces locations intuition misses and makes the trade-offs explicit.
Coverage measured at free-flow speeds is optimistic. Validate against congested travel times so the five-minute standard holds when demand is highest.
Response time over the real road network - not straight-line distance - is what protects lives. Buffers should be built from travel time, as they are here.
A research showcase measuring response-time coverage of fire stations across Greater Melbourne.
Travel-time isochrones reveal which suburbs fall outside target response windows.
Coverage gaps cluster at the urban fringe - candidate sites close them.
