2. A range of modelling studies were undertaken to assess and inform the
transport planning process:
Largely relied on the Main Roads Road Operation Model (ROM), a road-
based regional travel demand model focussing on daily traffic
Multimodal travel demand model - Strategic Transport Evaluation Model
(STEM) maintained by DoP/DoT was not used
Spreadsheet modelling and SIDRA was also used to inform the road
strategy and transport impact assessment
The proposed development would add significant pressure on the road
network, and would require significant investment in transport
infrastructure.
PREVIOUS MODELLING 2
“Insufficient time was allocated for transport modelling and the modellers
were under pressure from the Alliance to deliver something quickly”
3. An independent review was undertaken which included, reviewing past
reports and a series of interviews. This found:
The adopted mode share assumptions were based on aspirational targets
A review needed to be undertaken of the land uses and employment
growth (by type) adopted by the strategic models.
STEM is needed to estimate the future mode shares and to identify the
transport improvements that would achieve the mode share targets.
An effective transport solution had to be multimodal to address the needs
of both local movements and regional through movements.
Modelling going forward:
MODELLING REVIEW
4. LAND USE STRATEGIC MODELLING REVIEW
As a result of the modelling review, WSP | Parsons Brinckerhoff were
then engaged to:
1. Establish and agree future land uses.
2. Review strategic model forecasting and land use assumptions.
3. Implement recommendations from components 1 and 2.
6. Develop mesoscopic model of
Stirling City Centre study area
Assess alternative transport
network improvement options
Identify preferred transport network
improvement options
Undertake microsimulation
modelling of preferred transport
improvement options together with
“do minimum” case
SCOPE OF STUDY 6
20. POSSIBLE TRANSPORT NETWORK IMPROVEMENTS 20
Powis Street –
north facing ramps
McDonald / Main
Street signalised
Main Street to
Cape Street
westbound, right
turn removed
McDonald Street
bridge
Stephenson
Avenue extension
Drake Street
signalised
Liege Street 4 way
intersection
Bus Queue Jumps
Freeway ramp
modifications
Hutton Street
extension
Hector/Hutton
signals replaced by
Collingwood/Hutton
Scarborough Beach
Road bus lanes
27. Option Time Period
Network Performance Metrics
Travel Time
(sec/km)
Average
Speed (km/h)
Travel Time
(vehicle .
hours)
Travel
Distance
(vehicle
kms)
Option E
AM -10.5 +1.4 -1100 -6320
PM -19.3 +2.9 -1911 -5462
Option G
AM -3.6 +0.4 -501 -4988
PM -17.5 +2.2 -1478 +1956
NETWORK WIDE PERFORMANCE
DIFFERENCE TO OPTION C
27
Vehicle Hour reduction (h) Vehicle kilometre reduction (km)
Option E 3 011 11.6% 11 782 1.3%
Option G 1 979 7.6% 3 032 0.3%
COMBINED PEAK HOUR REDUCTIONS
28. Average
travel time
(all routes)
2014 Option C Option E Option G
AM (mm:ss) 9:46 10:54 8:45 9:09
PM (mm:ss) 10:10 11:18 10:02 10:05
PUBLIC TRANSPORT ASSESSMENT 28
Lower cycle times reduce bus journey times
Removal of freeway ramps at Cedric Street provides the majority of
the benefit to the public transport network
Option E is the superior option for both peaks
% Improvement E G
AM Bus journey times 25% 19%
PM Bus journey times 13% 12%
29. To cater for the additional demand in 2031 there will need to be the
following considerations made:
Parking cap enforced
Less attractive parking
More attractive public transport
Additional signalised intersections into residential areas
Better use of the freeway and arterial network
Less development to be considered
CONCLUSIONS 29
30. Option E performs the best, especially for public transport
Large improvements can be made to the study area by splitting the
freeway interchange and bus station
McDonald Street Bridge provides good access to the study area and
slightly alleviates congestion at Hutton Street Bridge
Hutton Street extension relieves Scarborough Beach Road of congestion,
however the interchange itself experiences severe delays due to
additional demand using Hutton Street
The bus lanes along Scarborough Beach Road offer limited benefit
Possible alternatives involve queue jump lanes or turning lanes and a
median to improve safety and pedestrian facilities
CONCLUSIONS 30
31. LESSONS LEARNED
Significant challenges calibrating the base network.
Managing stakeholder perception of congestion and the best solutions.
The performance difference between the mesoscopic and the hybrid
model was significant at congested locations.
Notas del editor
The development of the Aimsun model involved:
Building a calibrated and validated 2014 base mesoscopic model
A 3 step process in testing 8 mesoscopic scenarios for 2031
Comparing and discussing the results of each mesoscopic scenario
Converting the base 2014 model to a hybrid model with 3 microsimulation pockets
Developing 3 hybrid 2031 models consisting of a base model and 2 options
The study area is a 4km by 4km area with the centre being approximately 8km from the Perth CBD. It’s a very diverse area with a mix of both low density and high density residential around the fringes of the city centre. There is a large Westfield retail complex within the activity centre as well as various big box retailers such as Ikea which generate a significant amount of trips. There are commercial offices in the south within the Herdsman business park. In the east of the study area there is a large amount of light industrial businesses that attract a significant amount of heavy vehicles.
The key roads include the Mitchell Freeway. It’s the most important and busiest road in the city as it links with the Kwinana Freeway to provide the main passage over the Swan River as well as access from the northern suburbs to the CBD. Within the median of the Mitchell Freeway is the Joondalup train line and there are 2 stations inside the study area, Stirling Station and Glendalough Station.
The main east-west link within the study area is Scarborough Beach Road. It has a large amount of bus services and a high volume of vehicles. It suffers from severe congestion due to the amount of intersections, business driveways and turns taking place on it.
The TransPriority map for the study area was updated in consultation with the City of Stirling and agreed by the Stakeholder Working Group. The key aspects of the TransPriority map are:
- Private motor vehicle movements will be encouraged to go around the activity centre and not through it
- The activity centre, west of the Mitchell Freeway will be a pedestrian zone
- Stephenson Avenue will provide priority to cycling and walking
- Scarborough Beach Road (between Mitchell Freeway and Odin Road) will provide priority to cycling and public transport
- Ellen Stirling Boulevard will provide priority to public transport and pedestrians.
Scarborough Beach Road is a key distributor for residents and businesses within the study area and surrounding suburbs. Due to the orientation of the Mitchell Freeway and the limited amount of access points crossing the freeway, Scarborough Beach Road is also used as a route to the north of the study area.
Because none of the roads within the study area provide a full north-south route, vehicles are forced to drive on to Scarborough Beach Road to make turns. This contributes greatly to congestion.
In its present form Scarborough Beach Road carries large volumes of traffic. Although there are a numerous shops and food outlets along the route, it remains undesirable for pedestrians. This is largely due to the lack of a median in the road and long distances between traffic signals in some areas. Where there are traffic signals, the cycles are at 140 seconds.
The proposed cross section in the structure plan provides a median for pedestrians, however will involve significant land purchases and will extend the road to 42m.
Part of the reason that the study area has such a low public transport mode share is due to the type of industry that the area currently supports (i.e. retail and light industrial), the perceived ease by which it can be accessed by car and the abundance of parking. The scale on the graph is from dark green, excellent, to black which is below minimum service.
The City of Perth study also assessed a future scenario based on projects and services that are likely to be implemented by 2031. As shown public transport within the study area would be mostly very good. However, the south-western section of the study area is average and the eastern section of the Osborne Park industrial area still does not meet minimum services requirements. This highlights the need to consider improving services in these parts of the network.
A key issue within the study area is that the Stirling Bus Station is running at capacity and the freeway interchange is highly congested. There are 16 bus routes that use the station and a limited amount of layover. The reliability of buses is poor, as there is no bus priority provided on approach to the station. This in part explains the poor public transport accessibility rating in the previous slide as the connecting buses to the station can take a significant amount of time to reach the station despite their proximity.
Onto building the base model. We largely used SCATS at signalised intersections to calibrate the traffic flows within the network. However to determine the demand profile and heavy vehicle proportions we used the significant amount of tube counts to determine where and when certain vehicles were travelling.
The travel time data was collected from TomTom travel time data. This provided us with a distribution of how long trips were taking within the study area. It highlighted the unreliability of certain routes as well as a few congestion hot spots within the network. This was used to validate the model.
A challenge we faced was that we were unable to have access to the strategic model ourselves and just the outputs provided. There was significant discussion with the department of planning about exactly what assumptions were made not only within the study area but in adjacent areas that could generate significant traffic through the model.
Of significant difficulty was that the 2011 sub area network that the original matrix was based on only included 5 internal zones. To overcome the lack of detail of trips within the study area we disaggregated the matrix into 120 zones. Some of the original zones were combined of different land uses so we used first principles to allocate the existing strategic model trips into each new zone.
To replicate shorter trips within the model we seeded a small amount of trips from the 2031 denser sub-area matrix into the 2011 model. When we ran static adjustment within Aimsun, we gave the program a larger degree of flexibility to manipulate the matrix.
Challenges of static adjustment and calibration:
- logical routes not necessarily being used by both the model and drivers, so we had to spend significant time applying static costs to turns and links.
- The model has a large amount of route choice and many local streets that do carry significant flow. Had to iterate several times to ensure the local network was handling a realistic demand
- The usual situation with these data collection periods. Road works and network changes occurred during the collection. Including the implementation of new signals and road works on a key route.
- Had problems with the multiple connector to centroid method and essentially decided to remove that situation from the network.
Example of the multiple connector to centroids which we did use in future models when internal access roads were upgraded. However in the base network it would cause static adjustment to add or remove vehicles to the network, only to further throw out calibration.
Stakeholder perception of congestion was a difficulty. Telling someone that the median journey time along their route is 2.5 minutes seems to anger people when they frequently take over 5 minutes to make the trip. We aimed for the model to always have journey times slightly above the median for each route and would adjust demand accordingly. Because the network is quite variable with the amount of vehicles passing through, queues into the network could also vary from a few vehicles to 100s of meters on different days.
As I mentioned earlier we did a lot of manual adjustment to the network as well as seed small amounts of internal trips. To check that we did not change the trip length distribution we compared the profile of trips from the strategic model and our final trip matrix. This showed that we had a similar trip length within the network and hadn’t reached calibration simply by adding in copious amounts of small trips to reach the SCATS flows.
For the hybrid model we also did a visual validation check. We made several site visits to the area as well as had queue length surveys so we had a good idea of what the network should look like.
Talk about Do nothing, Option C (and it’s progression to the do minimum), Option E and Option G.
Talk about the problem with the Powis Street north facing ramps.
Talk about how the extended network provides better north-south connection and doesn’t force traffic onto Scarborough Beach Road
Talk about the density of the network changes and how it’ll provide better access. Talk about the 100 second cycle times and slower speed limits within the activity centre.
Talk about the circle route and how it’s an important bus for the network as it services universities, schools and train stations.
Reduction in journey times for circle route
- Additional off-road walking paths are implemented across the activity centre
- Improved walking connections are provided to Stirling Station.
25% reduction in bus journey times for Option E in the AM and 13% in the PM
Talk about the demand reduction.
Talk about how funding for PTA will need to supply a “Stirling CAT” or similar service
Talk about how the mesoscopic scenarios over estimated the capacity of unprotected right turns, of which there are many in this study area, including at freeway interchanges.
