Static Traffic Assignment

We can also perform static traffic assignment using Open-Source libraries for back-of-the-envelop analysis during debugging efforts

Exporting matrices

sphinx_gallery_thumbnail_path = ‘../../examples/modelling_like_the_old_days/sta_results.png’

from pathlib import Path

from polaris.analyze.trip_metrics import TripMetrics
from polaris.runs.convergence.convergence_iteration import ConvergenceIteration
from polaris.runs.static_skimmer.static_assign import static_assignment
from polaris.runs.static_skimmer.static_graph import build_graph

Matrix

We get the demand matrices for the AM peak for one iteration

project_dir = Path("/tmp/Bloomington")
supply_pth = project_dir / "Bloomington-Supply.sqlite"

iteration_3 = ConvergenceIteration.from_dir(project_dir / "Bloomington_iteration_3")
tm3 = TripMetrics(supply_pth, iteration_3.files.demand_db)

# Let's say that one afternoon peak hour is from 16:45AM to 17:45AM, so we egt trips starting during that time
matrix = tm3.vehicle_trip_matrix(from_start_time=16.75 * 3600, to_start_time=17.75 * 3600)

# This matrix has multiple vehicle types, and we could separate them to make sure we observe
# link type constraints, but that shouldn't be needed in a back-of-the-envelope exercise
# Instead, we will just multiply the PCEs for each matrix to the matrices themselves

pces = {"SOV_0": 1.0, "TAXI_9": 1.0, "MD_TRUCK_17": 2.5, "HD_TRUCK_18": 4.0, "BPLATE_19": 2.0, "LD_TRUCK_20": 1.8}
for i, mat in enumerate(matrix.names):
    matrix.matrices[:, :, i] *= pces[mat]

Traceback (most recent call last):
  File "/home/jamie/git/anl/polarislib/docs/examples/modelling_like_the_old_days/plot_sta.py", line 31, in <module>
    iteration_3 = ConvergenceIteration.from_dir(project_dir / "Bloomington_iteration_3")
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jamie/git/anl/polarislib/polaris/runs/convergence/convergence_iteration.py", line 57, in from_dir
    iteration.files = PolarisInputs.from_dir(Path(dir), db_name)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jamie/git/anl/polarislib/polaris/runs/polaris_inputs.py", line 24, in from_dir
    raise FileNotFoundError(f"No such directory {dir}")
FileNotFoundError: No such directory /tmp/Bloomington/Bloomington_iteration_3

Graph

We build an AequilibraE graph using the underlying Polaris supply model

This procedure asserts some things about the links so we can get everything we need for a static traffic assignment Assumptions are made about (HOURLY) capacities and centroid connector placements

graph = build_graph(supply_pth)

One can see the results of these assumptions in the graph object

graph.network.head()

Assignment

We perform traffic assignment and skimming using the AequilibraE library

We can load the assignment parameters from the default values and change them

from polaris.runs.static_skimmer.static_skimmer_inputs import STAInputs

sta_pars = STAInputs()

# Not sure why somewhat would want msa over bi-conjugate Frank-Wolfe, but...
sta_pars.assignment_algorithm = "msa"
sta_pars.max_iterations = 10
sta_pars.rgap = 0.01

# By default assignment uses BPR, but we can change the parameters
sta_pars.bpr_alpha = 0.14
sta_pars.bpr_beta = 3.9

No turning constraints are observed in this assignment

assig = static_assignment(graph, matrix, sta_pars)

Then we can see the results

link_loads = assig.results()
link_loads.head()

And get the skims

assig_class = assig.classes[0]
skim = assig_class.results.skims
# And show some elements
skim.time[:10, :10]