Assigning road crossing delays

Once the attribution of the walking network was complete, the delays incurred when using the relevant links to cross the road were calculated for the entire network data set. The time to cross the road was calculated based on the delay incurred while waiting for a gap in the traffic, plus the time required to walk across the road. Table 14.5 shows the approach used to determine the average delay encountered before beginning to cross the road.

Table 14.5    Logic used to assign road crossing delays to the walking network

Crossing facility

Attribute code

Delay calculation method

Rationale

None
(not at intersection)

ROAD + hierarchy attribute

Tanners extended model applied to the full road width

Must cross multiple lanes of traffic at once in order to complete crossing.

None(at intersection crossing minor leg)

ROAD + hierarchy attribute

2 seconds

Confirmation delay to ensure no traffic is coming.

Pedestrian
refuge

REFUGE

2 * [Tanners extended model applied to half the road width]

Refuge reduces the crossing distance so a gap is only needed in one stream of traffic at a time.

Traffic
signals

LIGHTS

½ * (100% – [Percentage of traffic signal cycle available to pedestrians]) * [total traffic signal cycle time]

Pedestrians only allowed to cross for part of the traffic signal cycle. Assumes pedestrian arrives at signal half way through the unavailable period.

Pedestrian
crossing

ZEBRA

2 seconds

Confirmation delay to ensure traffic has stopped.

Roundabouts

(not coded)

 

Treated as none (at intersection crossing minor leg) or pedestrian refuge

 

Tanner’s (1962) extended model was used to model the road crossing delays when crossing arterial roads and roads with pedestrian refuges. Tanner’s extended model takes into account a host of variables including traffic volumes, crossing width, number of lanes to be crossed and bunching of vehicles. The traffic volumes and number of lanes were estimated from the road hierarchy information, while the crossing width was calculated directly in the GIS from the length of the network edges. The proportion of bunched vehicles was assumed to be constant.

The delays calculated using the methods listed in table 14.5 were capped at two minutes, on the basis that after this period of time a person was likely to use less safe methods to utilise any crossing opportunity for roads less than or equal to a minor arterial. For higher-order roads, pedestrians were modelled to divert to the nearest ‘safe’ crossing point.