By Geline Canayon
When building multi-resolution models, the available strategic planning models are often used as a starting point for importing the roadway geometries, centroids/connectors, and traffic demand. Following the importation of the strategic model’s network, the geometry refinement is the first step in preparing the model for dynamic simulation. There might be some minor roads and turning movements that need to be added or modified to provide sufficient detail for the dynamic simulation. As part of this effort, it’s good to investigate how the centroids/zones from the strategic model load traffic onto the network and evaluate whether to split large zones, especially those that have more than four connections.
The large TAZ (Traffic Analysis Zones) that are typical in strategic models can be potentially problematic with the distribution of traffic in a dynamic model. It’s important to consider the usage of each entrance and exit based on land use, proximity to major roads and freeways, available traffic counts and of course, local knowledge of the area. Splitting zones where there is a mixed use of residential, academic, commercial or industrial space is critical to make sure that cars and trucks are entering and exiting the road network at the right places.
For example, in Figure 1, the centroid is connected to five pairs of entrances and exits and the land use is distinguished by color. The centroid can be split into four centroids – one centroid each for gates A, B and C, one centroid shared by gates D and E.
To split a centroid, simply right click on the parent centroid and select Split Centroid.
In the Split Centroid dialogue, set the number of new centroids to create. The Automatic Factor can be used to evenly divide the trips between the new centroids or specify proportions for each new centroid.
Once the command is executed, the following will happen:
1 – New centroids will be created and connected to the network.
The new centroids generated will inherit the name of the parent centroid with the numbers from the Split Centroid dialogue. However, each one of the centroids will also inherit all the connections of the parent centroid and some manual work is necessary to make sure that the new centroids are connected to the correct entrances and exits as shown in the following images.
2 – The OD matrices will be updated to reflect the new centroid configuration with the new proportions.
All the OD matrices in the centroid configuration will be updated with the new centroids and the values will reflect the proportions set for each new origin and destination. Note that the original matrices will be updated, not duplicated.
For example, the 80 trips from Zaidin to Uni South Campus are assigned to the new centroids 20, 50, 20 and 10 percent, respectively. The total number of trips in the matrix is preserved.
3 – A grouping of the new centroids will be generated.
The grouping generated can be used to show the original matrix structure with the original centroid. More importantly, this allows matrices with the original structure to be pasted with the same split automatically applied. For example, a future scenario matrix from the same strategic model imported can be copied directly into the updated Aimsun Next model.
To paste a matrix with the original structure, first create a copy of the matrix object itself (not a new empty matrix) and then activate the grouping category for original centroids. Simply paste the new matrix and the values will be split according to the underlying values of the grouped cells. When the grouping category is removed from the display, it will show the correct split values.
4 – If there are existing skim matrices such as cost and distance, the same values are kept. Skim matrices for trips will split accordingly.
Example of a cost skim matrix:
Example of a trip skim matrix:
5 – If the centroid has generation/attraction data (for example, land use data such as residents, employees, etc.), they are distributed among the split centroids according to the proportions defined in “Split Generation/Attraction Attributes into…”. However, while the Split option distributes the attributes according to user-defined proportions as a starting point, there may be some additional work necessary to make sure that the value for each attribute makes sense for each new centroid. If there’s a distinct difference in land use for each new centroid, the values should be revisited and modified.
To join centroids, select the centroids that must be merged, right-click and select Join Centroids. Note that only two centroids can be joined at a time.
Once the command is executed the following will happen:
1 – The centroid from which the “Join Centroid” command was executed will be kept. All centroid connections will be inherited as shown in the figure above.
2 – The OD matrices will be updated to reflect the new centroid configuration with the joint centroids.
3 – If there are trip skim matrices, the sum of the cells of the joined centroids is taken and for cost skim matrices, the average of the cells is taken. The arithmetic average is only an approximation of the cost, so it is recommended to rerun the macro static assignment or simulation that generated the original skim matrices after the zones are modified to get the exact values in new skim matrices.
4– If the centroids have generation/attraction data (for example, land use data such as residents, employees, etc.), they are added in the joined centroid.