1. STEPS Symposium December 8, 2018H2
Sustainable Transportation Energy Pathways (STEPS)
HD Truck Technology Transitions: Comparing an Energy Model with a Choice Model
Christopher Yang
Research Scientist
This presentation will weave together two different threads of research that we have been working on at ITS. The goal of this talk is to compare different approaches for modeling the transition of trucks, though this would also be relevant for other sectors like LDVs (which we’ve done already).
STEPS Symposium
December 8, 2018

2. Transition Models for Transport Decarbonization
Presentation Goal:
To explore the transition to a lower-carbon HDV sector, with emphasis on vehicle efficiency, advanced technologies, and alternative fuels
To understand how different modeling approaches can help us answer different questions about this transition
Methods:
Use different models of different types to explore the transition from conventional HDV vehicles to low-carbon vehicles and fuels
Truck Choice Model – Discrete choice simulation
CA-TIMES – Energy system optimization model 2

3. Truck Choice Model Researchers: Marshall Miller, Qian Wang, Lew Fulton
Core model: Nested multinomial logit (NMNL) discrete choice model
Brings in three important additional behavioral elements relative to optimization ( diversity in tech. adoption)
Consumer heterogeneity (i.e. segmentation)
Variation in preferences (probability distribution)
Adoption is driven by overall generalized cost
Capital and operating costs (over planning horizon)
Inclusion of non-monetary utility factors (besides costs)
Environmental perception
Uncertainty (Risk)
Model Availability
Vehicle Range
Refueling Time
Station Availability
First I want to talk about the choice model. Marshall miller and I have talked about this in previous symposia, so I wont rehash all of the details, but will review the overall approach.
Disutility due to lost driver time for finding stations and fueling vehicle 3

4. CA-TIMES Energy System Model
Students: Kalai Ramea, Saleh Zakerinia
Core model: Linear optimization (cost minimization) of investment and operating costs of entire energy system from
primary resources
conversion technologies (fuels and electricity production)
end-use technologies (vehicles, appliances)
Minimize cost of building and operating energy system to meet demand for energy services
Capital Cost
Operating costs (fuel use, maintenance)
Incentives
Carbon price/constraints
Global decision-maker
Constraints are critical to shaping technology adoption
Carbon caps
Policy constraints (CAFE, ZEV mandate, RPS) 4

5. Choice model vs Optimization
Attribute
Choice Model
Optimization Model
Focus of Analysis
Vehicle adoption behavior
Energy system linkages – vehicle adoption coupled with upstream supply infrastructure and resources
Representation of decision-maker(s)
Consumer heterogeneity – many individuals maximizing utility
Global decision-maker - a single decision-maker designing the system
Decision factors
Utility, including non-monetary factors
Hybrid, primarily economic cost factors (coupled with implications in other sectors) with high discount rates are used to approximate non-monetary factors*
Results
Probabilistic purchase behavior
Often see “winner-take-all” behavior (one technology is the lowest cost)
A little bit about the differences between the choice model and the optimization model in terms of approach
- Vehicles only vs an interconnected system
Many individuals random distribution vs a single deterministic decision maker optimizing with perfect foresight
Utility factors, vs hybrid
Yo ucan get a distribution of outcomes vs winner take all
* We have included non-monetary factors in the decision-making for LDV purchases (COCHIN) 5

6. Truck Choice Model Results
Scenarios: (L) BAU, (R) HDV ZEV mandate scenario (25% by 2050)
includes carbon tax ($150/tonne by 2050)
FCV
Diesel
Diesel
Here are some results in the truck choice model. Even though we have 8 heavy and medium duty truck/bus types (LH, SH, Heavy Vocational, Medium Vocational, Medium duty delivery, transit buses, other buses, hd pickups), I’m only showing LH and SH so as not to overwhelm you.
The scenario that was investigated was a BAU scenario and also a ZEV mandate scenario which required 25% ZEV sales in each of the truck categories by 2050.
The choice model cannot impose adoption on consumers so the only way to elicit adoption is to lower the cost (i.e. provide subsidies) and we don’t talk about which way they are subsidized (government or industry cross subsidies). We see FCVs at 25% in LH and a mix of BEV and FCV truck in SH. SH exceeds the 25% ZEV target in 2050.
LNG
BEV
LNG
FCV
Diesel
Diesel 6

7. CA-TIMES Results
Scenarios: (L) BAU, (R) Carbon cap (80% reduction by 2050)
FCV
Diesel
Diesel
These are the results for the same two truck categories in CA-TIMES. In this case, the differences between the BAU and Carbon cap scenario are driven by the need to reduce emissions state wide by 80%. SO this is not just trucks or transportation, but electricity, buildings, industry etc. . .
We see that we have significant ZEV adoption in the 80% cap scenario FCVs in LH and FCVs + BEVs in short haul. The contributions are much higher because the target is more stringent not just 25% ZEVs but really low carbon emissions even with growing activity.
FCV
Diesel
Diesel
BEV 7

8. Truck Choice Analysis
Identifying how to achieve a specific adoption target
Target of 25% ZEV adoption in each truck category by 2050
Carbon tax of $150/tonne CO2 (~$1.90/Diesel Gal) in 2050.
LH incentives: cost total $2.8B for 29k extra ZEVs  $95k/veh
SH incentives: cost total $0.2B for 10k extra ZEVs  $18k/veh
Economic costs diff. (capital and fuel) vs BAU is $820 million for LH & SH
<$20,000 per vehicle (lifetime cost)
Difference of about $2.2 Billion used to overcome non-monetary utility factors
$300k/veh
$140k/veh
$50k/veh
$0/veh
Now I’m going to talk about the analysis of the truck choice scenarios.
We want to identify what it takes to achieve the ZEV target of 25%. Some targets are easier to meet than others.
Don’t get totally fixated on the exact number, but just that these incentives are high – to counteract the barriers to adoption - higher prices, infrastructure costs and the non-monetary issues: the risk, model availability, and refueling inconvenience. This incentive level calculated is the one needed to get 25% of the purchases to be a ZEV. It is high and adds to the consumer surplus as many of the people buying the ZEV would have bought it at a lower incentive price.
We calculate just the economic cost difference between BAU and ZEV scenario (as only the difference in vehicle capital and fuel costs) and the cost differential is much lower as it doesn’t include any of these non-monetary factors. 8

9. CA-TIMES Transportation Decarbonization
CA-TIMES GHG reduction scenario is driven primarily by the goal of meeting carbon reduction goal of 80% by 2050
Requires decarbonization across supply and end-use sectors
Non-energy emissions are challenging to reduce
Transport > 90% reduction
Energy System Cost and Emissions ( )
Cost ↑ vs BAU: $137B
Emissions ↓ vs BAU: 2804 MMT
$49/tonne CO2e abatement
Relatively low average cost of
emissions reduction
But this only includes economic
costs and ignores other factors
Transportation
Commercial
Industrial
Residential
Non-Energy
Looking at CA-TIMES, it is important to again note that the model results are driven by the carbon target. This requires substantial decarbonization across all sectors.
If you look at the graph here, you’ll see that the energy sector actually reduces emission by more than 80% because there is this wedge of non-energy emissions, that is difficult to reduce (methane leaks, cement, enteric fermentation, land emissions, etc.).
Non-energy emissions (high GWP gases)
431 MMT 258 in  86 in 2050 (target is 86 MMT in 2050) for entire state. 1.7tonnes/person (equivalent to ~6000 miles/year in a 40mpg car).
The broader point is that given this requirement, we need to have lots of adoption of ZEVs in heavy duty as we saw a couple of slides ago.
In calculating costs, looking at the entire energy system we see that if you look at the difference in costs and emissions between the two scenarios we can calculate the abatement cost of $49/tonne, which is relatively low. Of course this is an average across all years. Generally it is lower in early years (or even negative) and the grows as thigns get more stringent and you have to do more and more things (think of a carbon reduction supply curve, like a McKinsey curve).
But again this cost ignores things other than capital and operating costs. 9

10. Discussion of modeling insights
We have two models that have very different costs results:
Choice model - Very high levels of subsidies are required for carbon abatement, even though extra capital and fuel costs are significantly lower
Energy system model – Modest costs associated with carbon reductions across all sectors
Carbon tax/price can have two meanings:
a calculation of the estimated extra economic costs associated with policies (may not include non-monetary factors)
as an policy instrument that can help to change behavior in the adoption of low-carbon technologies and fuels (should account for non-monetary factors)
Highlights a tension in modeling between results that are societally optimal and behaviorally realistic
It comes down to how we view costs.
1st approach is what we do in TIMES and just compares the amount that people paid for capital and fuel in the two scenarios and ignores any utility differences (what people would have preferred even if the costs were the same).
2nd approach is a differnet question, which is important for policy and regulation, what do we need to do to change behavior. 10

11. Conclusions and Takeaways
In either BAU scenario, we don’t see adoption of ZEVs and low-carbon technologies
Choice Model Results
A ZEV mandate can achieve ZEV adoption but requires significant additional incentives (subsidies and carbon tax in our approach)
CA-TIMES Results
Decarbonization of the energy system is a huge undertaking
Requires significant adoption of ZEVs in transportation
Fuel cell vehicles are chosen in the long-haul sector while FCVs and BEVs are chosen in short-haul
Cost of emissions reduction is relatively modest mainly due to cost savings from efficiency (highlighting the efficiency gap)
Can be thought of as difference between simulation and optimization (forward looking vs back-casting). 11

12. Conclusions and Takeaways (2)
Adding behavioral elements to a model makes the results more relevant to the real-world
These real-world, behavioral elements are barriers to adoption
Capital and fuel costs
Refueling/charging inconvenience and costs
Model availability and risk/uncertainty
Technology readiness
The goal is to understand what is needed to drive adoption
Monetary incentives (subsidies, carbon tax)
Infrastructure deployment
Technology maturation and perception 12

13. Thank You 13