2 Model development
3 Demand development
- Zoning refinement
- Static OD adjustment – Sanity check
- PM peak-related assignments
- Static departure adjustment
- Dynamic OD adjustment
- Findings summary
Given the level of detail of a mesoscopic/microscopic model compared to a strategic model, it was expected that a refinement of the zoning system within the model would have been necessary to better replicate the behavior of trip generators of the study area. It should also be noted that the resulting trip length distribution of the new trip is slightly different from that of the existing trip from the model. Due to an issue noted in the defined public transport routes of the PM scenario, it was not possible to run the model's PM peak adjustment tools.
Inspection of the existing tools defined in the model indicated that a warm-up period of 60 minutes was specified for the "static launch adjustment" tool in both the AM and PM periods. The "static departure adjustment" tool considers the simulated network travel time for demand profiling for each origin-destination pair passing an observation point. Due to the extension of the model, 60 minutes is an insufficient warm-up duration for a static OD launch adjustment, and it is recommended to increase this interval.
A general overview of the implemented demand adjustment procedure is provided in Figure 3-3 of the PRCUTS Base Model Development Report. This figure indicates that Aimsun's 'dynamic OD adjustment' tool was used for demand calibration, but no evidence of this was found in the model.
4 Key modelling parameters and assumptions
Variation from software defaults
- Reaction times
- Attractiveness weight
- Give-way time
- Queue discharge acceleration
- Additional reaction time at stop and traffic light
- Lane changing cooperation
- Acceleration factor
Additionally, the "stop reaction time" and "signal response time" values used for turns in Stochastic Route Choice (SRC) assignments are inconsistent with similar factors for Dynamic User Balance assignments ( DUE) for AM and PM periods. The "attraction weight" factor is an import parameter in dynamic route selection calculations, as it takes into account the importance of the route hierarchy to users. The Parramatta Road Corridor model has used an attractiveness factor of zero, meaning the model does not.
It is not clear from the model results whether wheel running is a problem in the base year model. Therefore, it is recommended that the modeler consider using the attractiveness weighting factor in the dynamic assignment. Additional reaction time at stop” and “additional reaction time at traffic light” are two local features that are sometimes used to calibrate the simulated traffic flow and right of way of intersections.
It has been observed in the model that negative values were assigned for "Additional reaction time at stop" and "Additional reaction time at traffic lights" for the approaching section of "City Rd" to "Parramatta Rd", while negative values were assigned for "Additional reaction time at stop" and "Further reaction time at traffic lights". The increase in "lane change cooperation" attributes for a section means that vehicles will support each other for easier lane changes.
As shown in Figure 1, different speedup factors have been applied to sections within the network. Increasing an acceleration factor means a faster response from users, which indirectly increases the capacity of a route. It is recommended to increase this weight (from zero) to account for the impact of the road hierarchy in vehicles' route selection decisions.
A value of zero means that there is no difference between local and trunk roads in user route selection. The main findings regarding these modeling parameters and plant-related assumptions are summarized in Table 6.
5 Intersections and signalisation
The Tebbutt St' intersection appears to be incorrect - only one movement phase is defined (small movements do not have any phase). The phasing of external ID 17437 appears to be incorrect. The phasing of the intersection “N4528 SITE 86 - Parramatta Rd & Renwick St & Railway St” appears to be incorrect: only one movement phase is defined (minor movements do not have any phase).
6 Public transport inputs
This is due to a new section that was added to the model but was not updated in the routes of these PT tracks.
7 Active transport and pedestrians
8 Assignment and convergence
- Static assignment convergence
- Dynamic User Equilibrium (DUE) convergence criteria Dynamic User Equilibrium (DUE) convergence criteria is relatively loose
- Calculate additional path for hybrid DUE
- SRC assignments - Seed numbers
- SRC assignments - Route choice variability
- SRC assignments – Median or average
- Findings summary
Because additional path computation is disabled in the DUE assignment, a high level of confidence in the resulting path file from the static assignment is desirable. This means that the DUE assignment only uses routes shortlisted in the static assignment results (maximum 3 routes for each origin-destination) and cannot add (or check) new routes. Since the static allocation of the AM period was not converged, disabling the calculation of additional paths for DUE could lead to significant shortcomings in the resulting allocation results.
The RMS Traffic Modeling Guidelines provide a list of random seed values to be used in SR commands and produce small levels of variability to reflect behavior exhibited in the real world. There are three different sets of SR runs in the model for the AM rush hour. In figure 3-3 of the PRCUTS Base Model Development Report it is stated that the median seed for each peak period was used for model calibration and reporting, but no evidence of this approach was found in the report.
Static Key Issue In static allocation, convergence is not achieved after 50 iterations for the AM period. It is recommended to allow for some degree of variability in the route selection of SRC runs, but this is not considered for PM peak SRC runs.
9 Calibration and validation
- Core area calibration criteria
- Adjusted criteria was not met
- Validation criteria was not met
- Stability check
- Findings summary
Based on the calibration results presented in Table 4-3 of the PRCUTS base year model development report, the developed model did not meet the adjusted criteria for one hour of the simulation period PM. The statement provided on this in the report was that "this is attributed to the impact of peak dispersion for the PM peak hour and the dispersion of traffic from Parramatta Road to the surrounding road network due to congestion along the corridor." According to Table 11.1 of the RMS Traffic Modeling Guideline, GEH =5 tolerance limits should be included in the plots of observed vs modeled hourly flow.
These lines were not listed in the corresponding figures in the reporting (Figures 4-1 through 4-8 of the Base Year PRCUTS Model Development Report). Insufficient information was provided for this model to justify validating travel times. Based on the RMS guidelines, a cumulative travel time graph for each route should be created and observed, and the modeled results should be compared for each segment separately.
AM Peak period: The review of the modeled travel time outputs showed that the travel time criteria were not met for the WB direction of route #4 (Booth Street and Moore Street between Paramatta Road and Catherine Street). PM Peak: Validation criteria were not met for the EB direction of route #5 (Collins Street and Marion Street between Johnston and Ramsay Streets) and the EB direction of the route.
10 Conclusion and recommendations
Land use assumptions
Public Transport Project Model
Identification of network capacity
- Identifying network traffic capacity
- Matching network capacity between models
The practical capacity of the road network can be characterized as the point at which the delay experienced by vehicles making a trip during the peak period exceeds the utility that the vehicle derives from making the trip. For large-area simulation models in urban environments, the primary failure pattern when demand exceeds capacity is usually the result of network congestion where vehicles are stuck in stationary queues and block throughput in the rest of the model. This blocked condition can be identified by plotting the total vehicle lap in the model during the modeled period, as shown in Figure 1.
Ideally, network traffic capacity should match the peak traffic demand that can be accommodated on the network during peak times before the network reaches "blocked". Once the practical peak period capacity is determined, a metric is needed to match the assigned traffic volume from the PTPM to the determined network capacity from the simulation model. Traffic volumes on specific links within the study area: can be problematic due to differences in network details and allocation between models.
General traffic demand or traffic growth within the study area: Can be problematic due to differences in base travel demand and travel zone resolution between models. Cumulative VKT over the entire study area: may be problematic to compare VKT directly, but equivalent proportional change in VKT may be a good benchmark. To best account for the differences in model coverage, zonal resolution, base matrices and volume delay response, option 3 (cumulative VKT) is considered the best candidate for matching.
Proposed strategic modelling interface
Subject: Parramatta Road Corridor Urban Transformation Strategy (PRCUTS) Transport Model Update Action Plan Delivery Technical Note. PwC was commissioned by the Department of Planning, Industry and Environment (“DPIE”) to undertake a review of the Parramatta Road Corridor Urban Transformation Strategy (“PRCUTS”) Inner West AIMSUN hybrid model and the methodology applied to estimate predicted traffic growth. present recommendations and develop an action plan to deliver them. New Canterbury Road / Stanmore Road / Crystal Street / Shaw Street Parramatta Road / Catherine Street / Phillip Street.