A4.2 Configuration and calculation

The Christchurch road centreline network GIS layer was imported into Accession. Accession is ‘hard wired’ to classify roads using the UK Ordnance Survey road classification system, so Christchurch road hierarchy values had to be assigned to one of the Ordnance Survey road classes. The matrix used to transform the Christchurch road hierarchy classification to an appropriate Ordnance Survey class is shown in table A.2.


Table A.2      Accession road hierarchy classifications

Christchurch road hierarchy

Ordnance Survey class

Motorway

Motorway

Major arterial

A road

Minor arterial

B road

Collector

Minor roads

Local road

Local street

Private road

Private roads (public)

 

Road speeds were assigned to the road network. A road speed value is assigned to vehicle, cycle and walk modes. In the case of a mode type not being able to use a particular road link, for example cyclists and pedestrians not being permitted to use motorways; road speed values for these modes are assigned as 0, making the link non-traversable by that mode.

One-way street and turn restriction data was not included with the road centreline dataset as this required additional data capture. Manually coding each one-way link and turn restrictions was not considered beneficial at this stage of the project. A total of 9853 road links were imported into Accession to form the Christchurch road network.

For the purposes of demonstration and performance assessment of Accession, the accessibility of general practitioners from meshblock centroids through the road network was calculated.

The centroids of Christchurch city meshblocks were used as the origin dataset. Similarly, a point dataset of general practitioners in Christchurch was imported into Accession for use as destinations. Importing point data into Accession to represent origins and destinations is a simple process. A screenshot of Accession, showing the Christchurch road network, meshblock centroids (blue square) as origins, and general practitioners (orange diamond) as destinations is shown in figure A.9.


Figure A.9     Christchurch city GIS data loaded in Accession

 

The analysis was undertaken for trips taken across the road network in a private motor vehicle. As public transport data is not currently available in New Zealand in a suitable format for import into Accession the ‘road only’ network was the only available option for this trial.

The default values currently used in the UK were selected for the analysis. Figure A.1 and section A3.1 outline the default values used by Accession.

The road node spacing option is a method Accession uses to break up road sections so connections between origins and destinations can be made at road nodes that connect two links. More frequent road nodes, and therefore smaller distances between nodes, provide better accuracy for connecting OD points to the road network. Conversely this option also increases processing times for modelling. A value of 1000m was selected to reduce the processing time of the model.

At this stage of the modelling dialogue, Accession asks the user to specify network speeds for walk, car and cycle. This is limiting because cars and the other transport modes move at different speeds depending on the specific location of congestion, road type and topography, although Accession does not differentiate been these variables. It is also confusing because network speeds are already set in Accession when the road network is imported as the user specifies network speed based on road type. The Accession help file fails to explain which of the two road speed values is used for the calculation of accessibility. Regardless, in this trial the two have been set to the same value.

Accessibility can be assessed in a number of ways. For the purposes of this trial, the calculation of accessibility was based on the shortest travel time between the origins and destinations.

Catchments are used in Accession to optimise the calculation time by stopping calculations once the accumulated value (time, distance or cost) exceeds the catchment unit maximum. Selecting a value of 30, means once trips from an origin to a destination exceed 30 minutes, the routing algorithm will be aborted and the accessibility value for the particular OD pair will be recorded as 0.

The next stage of the dialogue presents a screen to the user, summarising the calculation that is about to be undertaken and an estimation of the size of the calculation. The combination of 2710 meshblocks (origins) and 113 general practitioners (destinations) will generate 306,230 OD values and Accession estimates this as a ‘large calculation’.

Despite this estimation the processing time to calculate the OD routes was only about 10 minutes. The computer that ran the calculations was a standard desktop computer with a 2.6GHz Core 2 Duo processor and 3.5GB of RAM. The results are stored in a database with a maximum size of 2GB. The size of the database generated by this test was 240MB.

The output from the calculations was a matrix including travel time, cost and distance. This matrix can be output to Microsoft Excel and assessed in more detail as required. This matrix result is the ‘OD matrix’ pictured in the output information flowchart in figure A.2. It was noted it took longer than 10 minutes to export this dataset to an Excel spreadsheet. Once the network results had been exported they were then available to be analysed for specific accessibility assessments.