Vehicle Choice Model

Presentation: vehicle_type_model_sensitivity_tests.pptx

Validation, compare to NHTS

• Scenarios/data:
  • MTC Base – choosing fuel type, vehicle type, etc at the beginning
  • MTC Base Probabilities – fuel type is chosen by probabilities (body type is a logit model)
  • NHTS for MTC
  • NHTS for US
• Results
  • Fuel type matches US under the Probabilities scenario because probabilities were taken from full US data. If you want to implement the probabilities for you region, you may want to update for your region instead of using the current US values.
  • You may want to calibrate for your region at the 20+ vehicle age category as there were mixed results there.

Sensitivity tests

• Income
  • Tests include -50% and +50% income
  • Income interacts with the new purchase price and vehicle age terms
  • Higher income results in fewer motorcycles, more cars and SUVs, more electric and hybrid vehicles
  • Changes to vehicle type and fuel type due to changes in income seem modest, but we don’t have anything to validate whether it’s reasonable or not.
  • More likely to purchase newer vehicles with higher income. Inflection point is around 11 years.
• Constants – MTC vs ATL
  • Logically differences between changes in constants for changes in body type, fuel type, and age.
  • RSG provided estimation results, so you can see estimated values for different regions.
    • NHTS_Vehicle_Type_Summary.xlsx
    • ActivitySimVehTypeEstimation10.xlsx
• Number of chargers per capita
  • Only one expression that has this term in it, and it only applies to electric vehicles (PEV and BEV).
  • Unclear whether the number of chargers are public or include private?
    • From Greg: The charger data is from the US Department of Energy Alternative Fuels Data Center. You can download the data here. We included electric chargers of types Level 2 or DC Fast (the default) with public access. The data are based on what we downloaded as of 10/16/2021. We overlaid these with the CBSA boundaries, as well as the number within each state. When joining to NHTS, we use the CBSA value if the NHTS household is in a specified CBSA, and we use the state value otherwise.
  • Test doubled number of chargers, which resulting in increases for PEVs and BEVs.
  • Changes to age of vehicle are a mixed, but % differences aren’t that high.

Auto operating cost assumptions

• Auto operating costs are the sum of fuel cost plus maintenance, repair, tires.
• Assumed $3.00 cost per gallon for fuel and calculated vehicle fuel cost by using the MPG field in vehicle type data.
• Used AAA’s 2017 driving report for maintenance, repair, tire cost, which varies by body type.
• BLS report included average maintenance cost by age of vehicle, but there was no clear trend from that data and couldn’t find anything else with more information, so maintenance does not vary by vehicle age.
• Average auto operating cost came out to be close to the previous default (18.4 cents and 18.3 cents). Default is still used for non-household vehicles (value used for households that don’t have a vehicle but could have access to a vehicle they don’t own).
• Documentation to be clear about where a user can update auto operating costs. For example, may want to update the $3.00 assumption to be higher in CA. To do this, there is a column in vehicle type data file, part of the configs for this model, which includes auto operating cost field.