Connected Eco-Bus: An Innovative Vehicle-Powertrain Eco-Operation - PowerPoint PPT Presentation

Fed. funding: $3.15M Length 36 mo. Connected Eco-Bus: An Innovative Vehicle-Powertrain Eco-Operation System for Efficient Plug-in Hybrid Electric Buses PI: Matthew Barth, University of California-Riverside Partners: Oak Ridge National Lab, US Hybrid Project Goal Develop an innovative vehicle-powertrain eco-operation system for plug-in hybrid electric buses through co-optimization of vehicle dynamics and powertrain controls, achieving 20+% energy efficiency increase Current Technical Status • Integrated control strategies have been designed and implemented for PHEB. • Simulation shows up to 24% energy improvements for its specific target corridor. • Actual bus is being exhaustively tested using a Dyno-in-the-Loop (DiL) approach.

Team Information Matthew Barth: faculty, electrical and computer engineering Kanok Boriboonsomsin: research faculty, transportation engineering Guoyuan Wu: research faculty, mechanical engineering Peng Hao: research faculty, transportation engineering Mike Todd: development engineer, environmental engineering Fei Ye: Ph.D. student, electrical and computer engineering Ziran Wang: Ph.D. student, mechanical engineering Zhiming Gao: R&D Staff, hybrid powertrain simulation & analysis Tim LaClair: R&D Staff, hybrid powertrain testing & analysis Abas Goodarzi: president; hybrid powertrain design, manufacturer & integration Christophe Salgues: on-board vehicle controls Transit Partner: Riverside Transit Agency 2

Technical Accomplishments Vehicle: A unique plug-in hybrid bus platform has been built by US Hybrid. Traffic and Road Infrastructure Sensing SPaT Transmission Analysis Boundary Location Destination node Speed Stop line Scenario 2 Scenario 3 Scenario Information Gap keeping Preceding vehicle Scenario 1 Scenario 4 Identification Define reachable Integration Distance region Intersection of Interest Accelerating Cruising Cruising Time Source node Identify target state PHEB operation mode selection Input: Wheel spd/trq demand, SOC control data, & bus data/CEED data 3s Calculate PHEB tractive power, wheel torque and speed, optimal 2s gears Braking Parking Propelling, Braking, or Parking 1s Simplified Propelling SOC monitoring Braking mode Engine charging mode at parking at SOC<SOC parking chg Trajectory YES NO 4s Powertrain SOC>SOC ub 0s DDC CDC DDC or CDC Simplified model as PEV mode (or motor Planning propelling mode) NO NO 𝜐 𝑥ℎ𝑚 > 𝜐 𝑓𝑜𝑕−𝑥ℎ𝑚 , 𝑛𝑏𝑦 𝜐 𝑥ℎ𝑚 > 𝜐 𝑓𝑜𝑕−𝑥ℎ𝑚 , 𝑛𝑏𝑦 Model Cost in the graph NO YES YES 𝜃 𝑓𝑜𝑕−𝑞𝑥𝑢 > 𝜃 𝑛𝑝𝑢−𝑞𝑥𝑢 Engine & motor propelling mode Engine & motor propelling mode Engine propelling & charging mode YES Engine propelling mode Motor propelling mode Integrated Eco-Operation System: Filter and modulate the operating states of engine, motor and transmission as functions of engine/motor operation envelop, minimum engine on/off time, minimum wheel power demand activating engine on etc. Outputs: engine/motor load demand Key strategy as Optimal trajectory Vehicle dynamics control (in traffic) simplified model as driving cycle has been optimally integrated with Powertrain Model in powertrain control to maximize Simulink overall energy efficiency.

Technical Accomplishments Traffic and Road Infrastructure Sensing SPaT Transmission Analysis Boundary Location Destination node Speed Stop line Scenario 2 Scenario 3 Scenario Information Gap keeping Preceding vehicle Scenario 1 Scenario 4 Identification Define reachable Integration Distance region Intersection of Interest Accelerating Cruising Cruising Time Source node Identify target state

Technical Accomplishments Integrated traffic and Identified Target State road information PHEB operation mode selection Input: Wheel spd/trq demand, SOC control data, & bus data/CEED data 3s Calculate PHEB tractive power, wheel torque and speed, optimal gears 2s Braking Parking Propelling, Braking, or Parking 1s Simplified Propelling SOC monitoring Braking mode Engine charging mode at parking at SOC<SOC parking chg Trajectory YES NO 4s SOC>SOC ub Powertrain 0s DDC CDC DDC or CDC Simplified model as Planning PEV mode (or motor propelling mode) NO NO 𝜐 𝑥ℎ𝑚 > 𝜐 𝑓𝑜𝑕−𝑥ℎ𝑚 , 𝑛𝑏𝑦 𝜐 𝑥ℎ𝑚 > 𝜐 𝑓𝑜𝑕−𝑥ℎ𝑚 , 𝑛𝑏𝑦 Model Cost in the graph NO YES YES 𝜃 𝑓𝑜𝑕−𝑞𝑥𝑢 > 𝜃 𝑛𝑝𝑢−𝑞𝑥𝑢 Engine & motor propelling mode Engine propelling & charging mode Engine & motor propelling mode YES Engine propelling mode Motor propelling mode Filter and modulate the operating states of engine, motor and transmission as functions of engine/motor operation envelop, minimum engine on/off time, minimum wheel power demand activating engine on etc. Outputs: engine/motor load demand Key strategy as Optimal trajectory simplified model as driving cycle Powertrain Model in Simulink

Technical Accomplishments Dynamometer-in-the-Loop (DiL) Testing: High fidelity simulation drives the bus on the dyno while actual bus capabilities are fed back to govern the bus in simulation. HD Chassis Dyno and Test Vehicle: PHEB Simulation Environment: VISSIM HDCD Control PC Power Dyno PC (UDP) Socket (UDP) Socket CAN bus DriverModel.dll (coded in C++) CAN bus Interfacing Tool (UDP) Socket (UDP) Socket Powertrain Control: Matlab & Simulink Vehicle Dynamics Optimization: Matlab

Updated Efficiency Breakdown Table Efficiency Improvements Due to Control Strategies: Strategy Description Savings Source Traffic Eco-Approach and Determines energy-efficient speed profile based on Signal Phase 5% - 20% simulation & Departure and Timing information field studies Eco-Stop and Determines energy-efficient speed profile for decelerating to and Numerical Vehicle 3% - 17% Launch accelerating from bus stops and stop signs simulation Dynamics (VD) Control Determines cruising speed profile based on look-ahead traffic and Numerical Eco-Cruise Up to 10% terrain conditions simulation Traffic Integrated VD Combined vehicle dynamics control strategies on target corridor 8% - 14%; simulation; Efficiency-Based PT Optimizes both the engine and motor/generator operation by 13 - 15% Simulation Control managing transmission and battery state-of-charge Powertrain (PT) Control Intelligent Energy Optimizes power split between ICE and electric motor for the 3 - 8% Simulation Management vehicle speed and power demand profiles Integration of above strategies with VD&PT co-optimization 18% - 24% Simulation Integrated VD&PT Control on target corridor 16% (DiL) DiL

Efficiency Improvements on Target Corridor Target “Innovation Corridor”: University Avenue between UCR and downtown City of Riverside Traffic Level Duration Distance Powertrain control strategy % Savings V/C = 0.0 2.4 hours 36 km Charge/Discharge Dominant Control 18.7 V/C = ~ 0.2 2.4 hours 36 km Charge/Discharge Dominant Control 20.6 V/C = ~ 0.4 2.4 hours 36 km Charge/Discharge Dominant Control 19.9 V/C = ~ 0.6 2.4 hours 36 km Charge/Discharge Dominant Control 21.7 2.4 hours 36 km Charge/Discharge Dominant Control 24.0 V/C = ~ 0.8

Tech-to-Market Strategy • Participating and following Phase I and Phase II NSF I-Corps through UC Riverside’s Office of Technology Partnerships • Managing IP and licensing through UC Riverside’s Office of Technology Commercialization • Initial testing, evaluation, and deployment of Driver’s Aid with transition to OEM integrated longitudinal powertrain and vehicle dynamics control. • Development expanding beyond fixed route transit to include heavy duty goods movement and medium duty delivery applications • Extensive Dyno in Loop (DiL) evaluation to advance technology development

Techno-Economic-Analysis Activities • Comparative analysis between driver’s aid (Level 1) to automated longitudinal control (Level 2+)  off-the-shelf HMI vs. Integrated System • Identification of deployment requirements: infrastructure, on-board vehicle requirements, regulatory, liability, standards, IP • Estimation of production process costs • Estimation of scalability • Cost-performance model www.truck.man.eu

Key Lessons Learned • Traditional on-board control strategies can be greatly enhanced based on (partial) outside information: knowledge of location and stop locations, traffic information, signal information, past/current/future state of the bus operation (route, etc.). It is non-trivial to balance between the solution optimality and real-time performance. Location Destination node Location Destination node Location Destination node Stop line Stop line Stop line Gap keeping Preceding vehicle Time Time Time Source node Source node Source node No Constraints Constraints from signal and stop line Constraints from preceding vehicle • Tight integration between vehicle dynamics and powertrain controls is critical, requiring feedback in both directions: VD  PT and PT  VD. It is challenging to integrate simulation and dynamometer operation (time synchronization is critical). • Eco-Approach & Departure: Deep Learning-based calculation is computationally efficient.