Crate routee_compass_powertrain 

Compass Powertrain Crate

This crate provides energy-aware route planning models that integrate RouteE Powertrain into RouteE Compass.

Energy Estimation in RouteE Compass

A RouteE Powertrain is an energy estimation model trained on NREL’s FASTSim model. RouteE Powertrain is an ML model which makes it sufficiently fast to run within the loop of a route planner cost model. RouteE Powertrain models are trained and exported via the RouteE Powertrain utility and then loaded into a runtime at the core of this crate.

Usage

The TraversalModel in this crate is integrated into compass-app and can be installed by running compass-app with a TraversalModel that uses energy. An example traversal model configuration that uses this crate may look like this:

[traversal]
type = "energy_model"
speed_table_input_file = "edges-posted-speed-enumerated.txt.gz"
speed_table_speed_unit = "kph"
output_time_unit = "minutes"
output_distance_unit = "miles"

[[traversal.vehicles]]
name = "2012_Ford_Focus"
type = "single_fuel"
model_input_file = "models/2012_Ford_Focus.bin"
model_type = "smartcore"
speed_unit = "mph"
grade_unit = "decimal"
energy_rate_unit = "gallons gasoline/mile"
ideal_energy_rate = 0.02857143
real_world_energy_adjustment = 1.166

This TOML section is deserialized into JSON and passed as arguments to the EnergyModelBuilder which in turn loads the [EnergyModelService] in this crate. This in turn builds the [EnergyTraversalModel].

Search

TraversalModels in this crate will add energy estimation to road network search, and will differ in their dependencies and evaluation procedures.

EnergyTraversalModel

Dependencies

• speeds per graph edge as a lookup table
• grade per graph edge as a lookup table (optional)
• an energy_cost_coefficient value, (optional with default of 1.0)
• a real_world_adjustment_factor value, (optional with default of 1.0)

Evaluation

1. lookup speed for the edge in table
2. compute travel time as distance / speed for this edge
3. lookup grade for this edge in table; if missing, use 0.0
4. perform inference to retrieve energy rate from speed and grade values
5. compute energy as energy_rate * distance * real_world_adjustment_factor for this edge
6. compute link cost as (energy * energy_cost_coefficient) + (time * (1 - energy_cost_coefficient))

Real-World Adjustment Factors

In addition to calculating the energy based on a RouteE Powertrain output, an adjustment factor should be applied to capture real-world effects of running a powertrain in an environment. As a result of NREL research, some recommended values for this adjustment are:

powertrain type	factor
combustion vehicle (CV)	1.1660
hybrid vehicle (HV)	1.1252
electric vehicle (EV)	1.3958

A factor of 1.0 equates to 100% of the original energy value.

Modules

model