Essential Dynamics for Developing Models for Control of Connected and Automated Electried Vehicles: Part A - Powertrain Sadra Hemmati ( [email protected] ) Michigan Tech https://orcid.org/0000-0003-0793-4677 Rajeshwar yadav GKN Driveline Kaushik Surresh Michigan Technological University Darrell Robinette Michigan Technological University Mahdi Shahbakhti University of Alberta Research Article Keywords: Connected and Automated Vehicles, Automotive Control, Energy Eciency Posted Date: May 18th, 2021 DOI: https://doi.org/10.21203/rs.3.rs-536651/v1 License: This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License ESSENTIAL DYNAMICS FOR DEVELOPING MODELS FOR CONTROL OF CONNECTED AND AUTOMATED ELECTRIFIED VEHICLES: PART A - POWERTRAIN Sadra Hemmati Rajeshwar Yadav Kaushik Surresh Michigan Technological University GKN Driveline Michigan Technological University Houghton, Michigan 49931 Email: [email protected] Email: [email protected] Email: [email protected] Darrell Robinette Mahdi Shahbakhti Michigan Technological University University of Alberta Email: [email protected] Edmonton, Alberta, Canada Email: [email protected] ABSTRACT model considers different operating modes and associated en- ergy penalty terms for mode switching. Thus, the vehicle con- Connected and Automated Vehicles (CAV) technology troller can determine the optimum powertrain mode, torque, and presents significant opportunities for energy saving in the trans- speed for forecasted vehicle operation via utilizing connectivity portation sector. CAV technology forecasts vehicle and power- data. The powertrain model is validated against the experimen- train power needs under various terrain, ambient, and traffic tal data and shows prediction error of less than 5% for predicting conditions. Even though the CAV technology is applicable to vehicle energy consumption. both conventional and electrified powertrains, the energy sav- ing opportunities are more apparent when the CAVs are Hybrid Electric Vehicles (HEVs). This is because of the flexibility in the 1 Introduction vehicle powertrain and possibility of choosing optimum power- With increased penetration of electrification and connectiv- train modes based on the predicted traction power needs. In this ity technologies in the market, the potential for intelligent and paper, the powertrain dynamics essential for developing power- energy-efficient transportation becomes more salient [1]. Ac- train controllers for a class of connected HEVs is presented. To cording to the U.S. Energy Information Administration (EIA) this end, control-oriented powertrain dynamic models for a test 2020 outlook for the transportation industry, light-duty hybrid vehicle consisting of full electric, hybrid, and conventional en- electric vehicle sales in the U.S. are predicted to increase 3.1% gine operating modes are developed. The resulting powertrain per year, rising to a projected sales of more than 900,000 vehi- model can forecast vehicle traction torque and energy consump- cles in 2050, while battery electric vehicle (BEV) sales will in- tion for the specified prediction horizon of the test vehicle. The crease by 6% per year on average [2]. HEV/ BEV powertrain is a 1 Copyright © by ASME complex system consisting of numerous sub-systems. To ensure locity profiling in CAVs’ control that highly depends on power- good fuel economy and drivability, it is imperative to model and train capability and vehicle dynamics, but also affects cabin ther- characterize the dynamic interactions among the components. To mal management due to the effect of vehicle speed on convective establish and understand these interactions, physical prototyping heat transfer between ambient and the cabin [29]. The three ar- and testing prove to be too expensive [3], whereas modeling and eas in Fig.1 include a large number of important dynamics for simulation is considered cost-effective and time-saving for mod- CAV control. This paper (part A) focuses on powertrain dynam- eling and control of connected electrified powertrains [4]. Con- ics, while our subsequent paper (part B) focuses on thermal dy- nectivity facilitates forecasting future tractive and thermal loads namics and vehicle dynamics. In particular, part A is centered and power demands to the vehicle. This can be utilized for intel- on the dynamics essential for developing control-oriented and ligent control and energy saving [1]. Even though CAV informa- computationally-efficient powertrain models, and covers mode tion is helpful for the energy-efficient operation of conventional switching dynamics (clutch and power split mechanisms), inter- vehicles, EVs, and HEVs, the energy saving opportunities are nal combustion engine (ICE) transient dynamics, and e-motor more apparent when the CAVs are HEV due to flexibility in se- energy conversion efficiency. lecting the vehicle powertrain operating modes. In this paper, Vehicle powertrain models can be classified as steady state, the powertrain dynamics that should be modeled for developing quasi-static, and dynamic. Steady state models (e.g., Autonomie powertrain controller for connected vehicles is presented, with [40] and ADVISOR [36]) and quasi-static models (e.g., pow- emphasis on connected HEVs (CHEVs). ertrain system analysis toolkit (PSAT) [37]) typically use map- based models of vehicle sub-systems. Their main advantage is quick computation time; however, since they do not consider sys- Important Dynamics for Vehicle Controls to Enable tem dynamics, they become inaccurate for transient operations. Energy Saving in Hybrid Electric CAVs Dynamic modeling approach typically uses dynamic physics- based models for vehicle sub-components, to ensure better ac- Powertrain Thermal Vehicle curacy in transient conditions, compared to static/ quasi-static Dynamics Dynamics Dynamics models. Vehicle models are further classified as forward-looking - Vehicle cold-start - Vehicle velocity - Mode switching (ICE coolant & profiling including clutch (driver driven) or backward-looking (vehicle driven) models, de- aftertreatment and power split systems) - Road gradient mechanisms pending on the direction of power flow calculation [38]. The cal- effects and - Battery thermal predictive energy - ICE transients culation proceeds in the forward direction of powertrain power management consumption - E-motor and ICE flow using transmitted torque and reflected torque and driver - Cabin thermal air - Vehicle efficiency maps conditioning platooning needs for speed tracking. On the other hand, backward models initiates from the traction force request at the wheels to the pri- FIGURE 1: Essential dynamics affecting control of connected hybrid electric mary energy sources and is generally made of quasi-static mod- vehicles (CHEVs) els. ADVISOR is one example of backward model [36]. A list of prior CAV studies, including powertrain models, is The overview of the dynamics essential for energy-efficient presented in Fig.2. Three classes of connected vehicles, grouped CHEV controls is provided in Fig.1. These dynamics include by powertrain type, can be seen in Fig. 2: powertrain dynamics, thermal dynamics, and vehicle dynamics. i) Conventional engine-based connected vehicles (ECVs): As shown in Fig.1, these three dynamics are not decoupled: each In [10], the authors implemented Eco-Approach, Eco-Departure, one has implications on the other two. For instance, the opera- and Eco-Cruise control algorithms on a 2018 Cadillac CT6 tion of engine during cold-start includes thermal dynamics and testbed, in a high-fidelity dynamic model, and reported that the may also include inefficient powertrain operation due to incom- scenario in which driver was informed of the preview data had plete engine combustion. Another example includes vehicle ve- an 11% fuel saving compared the baseline. In [11], the authors 2 Copyright © by ASME Powertrain Modeling Approaches For Connected Vehicles Conventional Vehicles Electric Vehicles Hybrid Electric Vehicles Predictive Energy Gear Shifting Predictive Energy Electric Motor and Predictive Energy Drive Unit Management Optimization Management Battery Control Management Mode Selection • 2019 • 2020 • 2020 • 2019 Shao et al. : Optimal • 2019 • 2018 Gupta et al. :Estimation of Vehicle Speed and Gear Oncken et al.: Real-Time Zhao et al. : “InfoRich” Lu et al. : Energy-Efficient Batra et al. : Real-time Fuel Economy on Real- Model Predictive Position Control for World Routes for Next- Eco-Driving Control Adaptive Cruise Control model predictive control Powertrain Control for a Connected and Generation Connected Strategy for Connected for Electric Connected of connected electric Connected Plug-In Hybrid Autonomous Vehicles and Automated Hybrid and Automated Vehicles and Autonomous vehicles Electric Vehicle Vehicles Powertrains Nikzadfar et al. :An Han et al. : Fundamentals • 2019 Optimal Gear Shifting • 2017 of energy efficient Han et al. : Fundamentals Azad et al. : Non-Linear Zhao et al. : Optimal driving for combustion Strategy for Minimizing of energy efficient Model Predictive Anti- Ma et al. : Integrated Fuel Consumption Based Powertrain Energy Vehicle Dynamics and engine and electric driving for combustion Jerk Cruise Control for Powertrain Control for vehicles: An optimal on Engine Optimum engine and electric Electric Vehicles with Management and Vehicle Operation Line Coordination for Multiple Connected and control perspective vehicles: An optimal Slip-Based Constraints Automated Vehicles control perspective Connected Hybrid Electric • 2017