2018-01-0413

Published 03 Apr 2018

Modeling and Controls Development of 48 V Mild

INTERNATIONAL. Hybrid Electric Vehicles
SoDuk Lee, Jef Cherry, Michael Safoutin, Anthony Neam, Joseph McDonald, and Kevin Newman 
U.S. Environmental Protection Agency

Citation: Lee, S., Cherry, J., Safoutin, M., Neam, A. et al., “Modeling and Controls Development of 48 V Mild Hybrid Electric Vehicles,”
SAE Technical Paper 2018-01-0413, 2018, doi:10.4271/2018-01-0413.

Abstract
which has been integrated into EPA’s ALPHA vehicle model

T
he Advanced Light-Duty Powertrain and Hybrid and was also used to model components within Gamma
Analysis tool (ALPHA) was created by EPA to evaluate Technology GT-DRIVE simulations. Te mild hybrid electric
the Greenhouse Gas (GHG) emissions of Light-Duty vehicle model was validated using vehicle data obtained from
(LD) vehicles. ALPHA is a physics-based, forward-looking, Argonne National Laboratory (ANL) chassis dynamometer
full vehicle computer simulator capable of analyzing various tests of a 2013 Chevrolet Malibu Eco 115 V 15 kW BISG mild
vehicle types combined with diferent powertrain technolo- hybrid electric vehicle. Te simulated fuel economy, engine
gies. Te ALPHA desktop application was developed using torque/speed, motor torque/speed, engine on-of controls,
MATLAB/Simulink. Te ALPHA tool was used to evaluate battery voltage, current, and State of Charge (SOC) were all
technology efectiveness and of-cycle technologies such as in good agreement with the vehicle test data on a number of
air-conditioning, electrical load reduction technology and drive schedules. Te developed 48  V mild hybrid electric
road load reduction technologies of conventional, non-hybrid vehicle model can be used to estimate the GHG emissions and
vehicles for the Midterm Evaluation of the 2017-2025 LD GHG fuel economy of 48 V mild hybrid electric vehicles over the
rule by the U.S. Environmental Protection Agency (EPA) EPA regulatory drive cycles and to estimate of-cycle GHG
Ofce of Transportation and Air Quality (OTAQ). Tis paper emissions. Te 48 V mild hybrid electric vehicle model will
presents controls development, modeling results, and model be further validated with additional 48 V mild hybrid electric
validation for simulations of a vehicle with a 48  V Belt vehicle test data in the future as more vehicle models become
Integrated Starter Generator (BISG) mild hybrid electric available. EPA has included 48 V BISG mild hybrid electric
vehicle and an initial model design for a 48 V inline on-axis vehicle technology in its assessment of CO2-reducing tech-
P2-confguration mild hybrid electric vehicle. Both confgura- nologies available for compliance with U.S. GHG standards.
tions were modeled with a MATLAB/Simulink/Statefow tool,

Introduction

T
he Advanced Light-Duty Powertrain and Hybrid
Analysis (ALPHA) tool was developed by EPA to model
vehicle performance, fuel economy, greenhouse gas
(GHG) emissions and battery pack performance for light-duty
-  FIGURE 1   Schematic representation of the relative electric
machine positioning (Pi) for diferent hybrid electric
vehicle architectures

conventional and hybrid electric vehicles (HEV). Te ALPHA

model can be used as a support tool for the development of
future GHG emissions regulations and as a research tool to
US Environmental Protection Agency

evaluate the efciency of new advanced technologies. Te

ALPHA [1, 2] hybrid model is related to the heavy-duty vehicle
Greenhouse Gas Emissions Model (GEM) currently used for ••
determining GHG emissions compliance for heavy-duty
vehicle applications in the U.S. [3]. Te basic model strategies
US Government /

and controls within GEM (with the exceptions of specifc

traction motor, generator, battery, regenerative braking
control, hybrid vehicle supervisory control, etc.) can be used
within ALPHA for modeling light-duty HEV applications.
Relative positioning of electric machines for HEVs are
shown schematically in Figure 1. Tis positioning includes
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 2 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES

P0 (direct coupling to the engine front accessory drive), P1

(direct coupling to engine crankshaf power output), P2 (posi- Vehicle Model
tioned between the engine and transmission or transaxle with
clutch isolation), P3 (coupled to a front diferential), and P4 In this section, the hybrid electric vehicle model is described
(coupled to a rear diferential). Mild hybrid electric vehicles in terms of its overall architecture/structure and each of the
(MHEVs) typically use working voltages at or below 150 V DC. component models. Te HEV model is a forward-looking
Limiting MHEV voltage to below 60 V DC (e.g., 48 V MHEV) vehicle model which represents light-duty (LD) HEVs. Te
can potentially reduce the cost, complexity, and weight of current version simulates vehicles with a fully warmed-up
systems necessary to comply with U.S. motor vehicle safety engine, with a base conventional powertrain, and with either
regulations [4] while potentially maintaining efectiveness P0 or P2 electric machine positioning.
comparable to higher voltage systems for reducing
GHG emissions.
Both 48 V Belt Integrated Starter Generator (BISG) P0
Model Architecture
and 48  V P2 MHEV models were developed to estimate Both the commercial GT-DRIVE and EPA ALPHA HEV
vehicle performance, fuel economy and GHG emissions. models consist of a user-friendly Graphical User Interface
Development of the P2 vehicle model shares many charac- (GUI) and the underlying component models. Te dynamic
teristics with development of the P0 model and details of linked libraries (DLLs) used within the GT-DRIVE vehicle
models for both MHEV types are presented in this paper. model are written within the MATLAB/Simulink/Statefow
Modeling and validation results are shown only for the P0 environment. Te GT-DRIVE and ALPHA generic GUIs can
MHEV in this paper due to the availability of P0 MHEV be used to setup light-duty and light-heavy-duty conventional
chassis dynamometer data from a production vehicle applica- and hybrid vehicle Design of Experiments (DOE) to estimate
tion. Modeling and validation results for the 48 V P2 MHEV drive cycle GHG emissions, are capable of generating response
will be presented separately in the near future pending the surface model inputs for specifc technology combinations,
availability of P2 HEV vehicle-level and component-level and also provide easy access for end-user input into the
chassis dynamometer data. models. Te compatibility of DLL’s between both models
Te vehicle supervisory controls (VSC), battery power allowed initial model and control system prototyping develop-
limits and battery state of charge (SOC) controls for both the ment to be conducted within the GT-DRIVE environment
P0 and P2 MHEV models were initially prototyped and devel- with eventual porting of modeling and control strategies into
oped using the Gamma Technologies, LLC (Westmont, IL) EPA’s ALPHA model.
GT-DRIVE™ vehicle model with the addition of dynamic Te base architecture of the HEV model consists of three
link libraries (DLLs) developed using Microsof Visual Studio. layers: Systems, Components, and Functions. Tere are six
Te MATLAB/Simulink- and Statefow-based control strate- plant model and control systems (Driver, Electric Motor,
gies and algorithms within the DLLs were also easily ported Battery, Engine, Transmission, and Vehicle) and VSC. Te
for eventual use within the EPA’s ALPHA vehicle ALPHA battery and engine modules were compiled using
electrifcation model. Microsoft Visual Studio 10 and the MATLAB/Simulink
Te battery and engine DLLs were created directly from (version 2016a) to create GT-DRIVE (version 2018) DLLs to
the same module used within the ALPHA vehicle model and precisely control and represent various engine states, battery
can be used either in ALPHA or GT-DRIVE. A 48 V VSC DLL power limits, battery SOC, battery charging efciencies, etc.
was developed to model P0 48 V MHEVs. Previously devel- Some of the systems (e.g. Electrical, Engine, Transmission,
oped P2 strong HEV control strategies [5] were used to model and Vehicle) consist of components, each of which represents
the 48 V P2 MHEVs since the hybrid electric vehicle systems, a physical entity that, when combined, make up the entire
motor, power coupling, engine plant models and controls for system. Functions are mathematical equations that represent
the inline P2 48 V MHEVs and P2 strong HEVs have similar the systems and/or components.
launch and regeneration power and also similar battery pack
energy (kWh) capacity. A lithium-ion (Li-ion) battery pack TABLE 1  overall Structure of 48 V MHEV System Model
model was used for simulation of both the P0 and P2 MHEVs.
Te battery pack model [6, 7, 8] contains a two-time constant Systems Components
equivalent circuit battery cell model, a lumped capacitance Driver and Ambient N/A
battery thermal model and battery management system (BMS) Engine internal Combustion Engine,
thermal control strategies. Initial model development and Mechanical Accessories
US Environmental Protection Agency

validation reported in this paper were conducted using Power-Coupling & Automatic Transmission, Clutches
GT-DRIVE, but the DLLs and modeling strategies have been Transmission
ported into ALPHA for use in future EPA MHEV analyses. Electric Machine Electric Machine, DC-DC Converter,
GT-DRIVE MHEV model validation was conducted by inverter, Electrical Accessories
comparison of vehicle and component simulation results with battery Li-ion battery and bMS
US Government /

vehicle-level and component-level results generated during Vehicle final Drive, Drive Axle, Tires, Chassis
chassis dynamometer testing of a P0 MHEV over EPA VSC Engine on/of, Motor Power, battery
regulatory cycles. Management System

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES 3

System Models Power-Coupling and Transmission Te P0 and P2

MHEVs have diferent power transfer mechanisms when
In this section, brief descriptions for each of the systems are combining engine power and motor electric power. Te regen-
provided. Tese system models remain consistent regardless erative brake energy recovery, belt-drive transmission losses,
of vehicle types and classes, in other words whether the vehicle and engine inertial power losses due to the direct, belt-drive
is a P0 or P2 HEV. Te ambient conditions, signifcant portions engine/motor coupling used by P0 MHEVs resulted in more
of driver interactions, engines, chassis, and conventional energy losses when recuperating the brake energy and during
transmissions all share common system models. Terefore, electric motor power transmission compared to the P2
these common system models will not be presented within MHEVs due to the inability of the P0 confguration to operate
this paper. the electric machine completely independently from the
engine. Terefore, diferent P0 and P2 hybrid power-coupling
Engine Most of the previously published engine system and energy fow controls and GM 6 T40 6-speed automatic
models in [1, 2] can be used for hybrid electric vehicle applica- transmission models are used by ALPHA and were incorpo-
tions by simply updating to include the engine operational rated into the hybrid electric vehicle model by using a variant
maps, such as from a gasoline direct injection (GDI) or an sub-system to represent P0 or P2 confgurations within the
Atkinson cycle port fuel injection (PFI) engine, and by elimi- MATLAB/Simulink environment.
nating some portion of the engine controls and operating
modes, e.g. idle speed control and idle operation. Po System Description Te modeled components of the
A 2.5 L GDI engine map was previously developed for P0 MHEV system are summarized in Figure 3. A small
ALPHA modeling based upon engine dynamometer testing 12/15 kW electric machine is directly mounted to the engine
of a 2013 GM Chevrolet Malibu 1LS [2]. For MHEV modeling, front accessory drive system by a seven-groove belt-pulley.
the 2.5 L GDI map was used in order to create a surrogate fuel Te tractive power free energy of the vehicle available during
map [9] for the engine used by 2013 GM Chevrolet Malibu deceleration is transferred to the electric machine and is used
Eco modeled within this study. Te surrogate BSFC map to charge the 48 V Li-ion battery even if the engine shuts of
shown in Figure 2 was then used to develop models of both during deceleration. However, there are energy losses during
P0 and P2 48 V MHEVs. the regenerative brake energy recovery process so that the
available energy has to be reduced by engine inertia, engine
friction, inverter losses, and belt-pulley system losses. Te

-  FIGURE 2   2013 GM Malibu EcoTec Engine brake Specifc

fuel Consumption (bSfC, top) and brake Thermal Efciency
(bTE, bottom) Maps
power transmission efciency of a seven-groove belt-pulley
drive is less than typical gear and clutch efciencies for P2
systems. Typical published efciencies for a multi-groove belt-
pulley [9] were used to model 48 V P0 MHEVs. Te power
available during P0 regenerative braking was calculated from
the motor/generator efciency and the regenerative charging
power as shown in the Eq. (3).
Te DC electric power used to charge the 48 V battery
.. .,. pack is converted by an inverter from the AC power generated
i
!- . 1,0
by the P0 electric machine. Te battery pack may enable
limited electric-only vehicle driving capability similar to a P2
MHEV system when the demanded vehicle tractive power is
less than the available battery discharge power limit at very

------------
0 ............... 1 .........'""- • · ·...• • ............. ...................................1.. .. . ......"

Sp.ed(RPM)
.............. ......... .... ."' ............ ............. ....
-  FIGURE 3   12/15 kW 48 V biSG P0 MHEV Model
Components
US Government / US Environmental Protection Agency
US Government / US Environmental Protection Agency

,0 .,.

-i
i
.

"
0

------------
••···· ............ _ ••••••. •• i. .. .......- ............... . . . . ... .-

5'»ed(RPMI
· •• - .....................i.... .. - .•••1...• • ...,. ............. .

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 4 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES

low vehicle speeds (e.g., parking maneuvers and vehicle creep Battery Pack An A123 Systems 0.4kWh 48 V 14S1P Li-ion
in heavy traffic). The P0 system can be mathematically battery pack was tested at the U.S. EPA National Vehicle and
described in the following manner: Fuel Emissions Laboratory (NVFEL) battery testing labora-
Tcoupler = Tengine +TBISG . rBISG . b eff tory using battery pulse tests to characterize the ohmic short/
(1)
long-time resistance and capacitance. Te tests were conducted
dw using 10 second discharging and charging pulse currents to
Tloss = TFMEP +Tinertia = TFMEP + I (2) measure the parameters necessary to develop an initial model
dt
of the 48 V Li-ion battery pack’s electrical characteristics. A
Peng _ on = Tcoupler .w ,Peng _ off _ rgen = j .Ptractive +Tloss w (3) detailed description of development of the 48 V battery pack
model has been published within a related paper [14]. Te
where, rBISG is the belt-pulley ratios of the BISG electric tested and modeled pack design uses a proprietary Li-ion
dw pouch cell design. Tis 48 V battery pack design was the basis
machine. βef, w , I , and are the belt power/torque transmis-
dt for all of the 48 V MHEV simulations in this study. Te 2013
sion efciency, engine speed, the moment of engine inertia, Malibu Eco mid-size MHEV was originally equipped with a
and engine angular acceleration, respectively. Tengine, TBISG, somewhat higher capacity and higher voltage battery pack.
and TFMEP are torque from the engine, from the BISG and Te vehicle used a 0.5kWh, 4.4 Ah, 32 cell, 115 V Li-ion battery
engine friction torque, respectively, Peng_on is the power pack. Te recently introduced 2018 Buick Lacrosse eAssist
during the engine on-state, Peng_of_rgen is the regenerative mid-size car is equipped with a 0.45kWh, 24 cell, 86 V Li-ion
power during the engine-of state. An empirical regenerative battery pack that has been repackaged more compactly and
brake energy recuperation efciency, φ, was used to validate yet can store a level of regenerative braking energy comparable
the regenerative braking energy recuperation of both the P0 to the older Malibu Eco pack. Te compact packaging of the
and P2 48 V MHEVs. Te negative portion of vehicle tractive newer GM 0.45kWh battery pack design has also allowed it
power (Ptractive) multiplied by the regenerative braking energy to be located under the center console in the 2016 GM
efciency and the positive friction and engine inertia torque Silverado 1500 eAssist light-duty pickup truck application.
losses (T loss) were added to estimate the electric power Te battery model contains an equivalent circuit cell
applied to the BISG motor for P0 applications. Te regenera- model, a lumped capacitance battery thermal model, and
tive brake power, Prgen_brake, [11, 12] of the BISG is calculated Battery Management System (BMS) controls. Te MATLAB/
by using 48 V motor/generator efciency maps as shown in Simulink-based equivalent circuit battery model was imple-
Eq. (4): mented in both EPA’s ALPHA vehicle model and Gamma
Technology GT-DRIVE vehicle model using Microsof Visual
Prgen _ brake = Peng _ off _ rgen .y (4) Studio DLLs [14].
where, ψ is the BISG efciency. Development of the motor As shown in Figure 4, both the discharge and charge
efficiency maps is described in a subsequent section of power limits were reduced to zero when the battery pack
this paper. temperature rises to above 65 °C or falls below −30 °C, which
represents the upper and lower operating temperature limits,
P2 System Description Te P2 MHEV has a combination respectively, for this particular Li-ion cell chemistry. Te
of a single traction motor/generator, a gear box and a clutch. desired operating temperature of the modeled 48 V Li-ion

_ L________
The clutch allows independent operation of the electric battery is between 20 and 55 °C although a limited battery
machine by allowing it to completely decouple from the engine operating mode can be extended between −30 and 65 °C.
and transmission.
Te engine and motor torque in a parallel HEV system
can be estimated by equation (5). Te driveline shaf torque  FIGURE 4   The Power Limits of a 48 V Li-ion battery Pack
is calculated by multiplying the fnal drive gear ratio to the
clutch output shaf torque shown in equation (6). Battery Pack Power Limit

20
Tcoupler = Tengine + Tmotor (5)
15
Tdriveline = Tcoupler *rGear ( gear position ) * g fdr (6)
US Government / US Environmental Protection Agency

[ 10

where, Tcoupler, Tengine, and Tmotor are torque from the torque aj 5

coupler, engine, and traction motor, respectively. Te rGear !
[I.

term is the gear ratio of the selected gear position, and g fdr is -"'
u
the fnal drive gear ratio. 8:.
~
Te approximately 80% efciency for regenerative brake QJ
ig
energy recovery of the P2 system is signifcantly higher than cc -10
the approximately 50% efciency of the P0 system since the
K0 clutch (Figure 1) between the engine and electric machine
of the P2 system can be disengaged during vehicle deceleration -20 -+-,r-,-?---r"T""T--r-,r,--,-....-r..,.---.-,,-,_.....,...,.--,---,---r-, I

to isolate the losses such as those caused by engine friction -40 -20 0 20 40 60100
and, engine inertial forces [13].
Pack Temperature ["C]

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES 5

Te maximum allowable charging and discharging power

limits of the 48 V battery pack are 16 kW and 15 kW respec-
tively at 50% SOC, 25 °C battery pack temperature, and near
the beginning-of-life of the battery pack. Te BMS has self-
balancing SOC control functionality. At −30 °C temperature,
-  FIGURE 6   SoC Simulations of a 48 V Li-ion battery
at UDDS

the modeled 48 V battery pack can still discharge approxi-

mately 30A for 10 seconds at 50% SOC, which is sufcient to

US Government / US Environmental Protection Agency

crank the engine during cold start. Te 10 second discharging
and charging current limits are approximately 370A at
between 30% SOC and 60% SOC, and between 30 and 60 °C
0 200 400 600 800 1000 1200 1400
battery pack temperature.
Te battery charge and discharge power was derived from
ANL chassis dynamometer data of the 2013 Chevrolet Malibu
Eco 115 V MHEV over regulatory drive cycles [15, 16] and
was applied to the A123 Systems 48 V Li-ion battery pack for u
laboratory testing using a hardware-in-the-loop confguration 0

[14]. As shown in the frst plot of Figure 5, the battery pack

C/J 40
1- EPATestl
- - - Model
can maintain acceptable voltage levels between 40 and 50 V 30 ~ - ~ - ~ - ~ - ~ - ~ - ~ ~
at a root-mean-square (RMS) current of 47.2A over the Urban 0 200 400 600 800 1000 1200 1400
Elapsed Time [Second]
Dynamometer Driving Schedule (UDDS). Te modeled 46.2 V
RMS battery pack voltages were in excellent agreement with
the 46.01  V RMS battery pack voltages measured during National Laboratory (ORNL) [17]. EPA scaled this motor ef-
hardware-in-the-loop (HIL) testing over simulated UDDS ciency map to estimate both 12 kW launch assist and 15 kW
operation [14]. regenerative energy charging. Te scaled efciency map was
Figure 6 shows that the modeled and measured SOC then compared to a proprietary 12/15  kW 48  V electric
during HIL testing were in good agreement. Te “tooth- machine efciency map provided to EPA by a Tier 1 automo-
shaped” measured SOC was due to 20 Hz/50 ms battery BMS tive supplier and initial GT-DRIVE simulations were
CAN transmission rate updates. Te 48 V Li-ion battery pack conducted to compare estimated UDDS and Highway Fuel
model calculated reasonable battery pack voltages, SOC, and Economy Test (HwFET) fuel economy and GHG emissions
pack temperatures while satisfying the requirements of between the scaled and proprietary maps. Te simulated fuel
discharging and charging power and current limits at the economy and GHG emissions showed negligible diferences
estimated battery pack temperatures and SOC levels. A more between using the scaled versus the proprietary efciency
detailed description of the 48 V Li-ion battery pack testing maps. Terefore, the scaled electric machine efciency map
and modeling along with battery model validation results is based upon the published data from ORNL was used to model
presented elsewhere [14]. the 48 V electric machines included in this study. Te resulting
scaled 4-quadrant motor power and efciency map for the P0
Electric Machine Torque and Efciency Maps Efciency 48 V MHEV model is shown in Figure 7.
relative to speed and torque of the 8.5 kW electric machine

-
used by the 2011 Sonata HEV was evaluated by Oak Ridge

 FIGURE 5   Voltage and Temperatures of a 48 V Li-ion

battery Pack
-  FIGURE 7   A four-Quadrant 48 V biSG Motor Efciency
Map Derived via Scaling of Publicly Available Data (an enlarged
version of this fgure is also shown in Appendix figure 1)

50

1=~ ]
40

30
US Government / US Environmental Protection Agency

US Government / US Environmental Protection Agency

40 20
0 200 400 600 800 1000 1200 1400
E
6 10

i.::~~W~]
<ll
:::,
g 0
t--
.8
0
-10
~
0 200 400 600 800 1000 1200 1400 -20

F : : :1
~~~.1l
-30

-40
E i---
l :0 200 400 600 800 1000 1200 1400
-50
-1 .5 -1 -0.5 0 0.5 1.5
Elapsed Time [Second] Motor Speed [RPM] x 10 4

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 6 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES

Electric machine torque from model simulations was etc.). Te electric motor discharge power, EMpwr_dmd, is depen-
compared with ANL chassis dynamometer vehicle test data dent on the battery discharge power limits (Pbatt):
since discharge current is heavily dependent upon traction
EM pwr _ dmd = min ( Pbatt , EM peak _ pwr ) .Pedalaccel / threshold (12)
motor torque demand. Te simulated motor torque was in
good agreement with motor torque calculated from the vehicle Te battery discharge power limits depend on the pack
test data [16]. Te motor currents were calculated using the temperature, SOC level, battery aging, etc. Te battery SOC
following relationship: was controlled by the overall energy fows, and therefore total
energy fows of the model simulations and the ANL vehicle
Im =
(T motor wm )
heff =
Pmotor test data were in good agreement, although the modeled SOC
(9)
VBatt VBatt does not mimic the tested SOC profles on a second-by-second
basis (Figure 8). Rule-based controls [18, 19] were then used
where the subscript m is the P0 or P2 motor, the subscript Batt to further refne the engine speeds, engine power and motor/
represents battery, ω is motor angular speed, ηef is motor generator power to operate the engine closer to high efciency
efficiency, P is power, and T is torque. By supplying the points of operation while satisfying the demanded vehicle
demanded motor torque and speed generated by the vehicle tractive power requirements. Te same engine power can be
supervisory controller, the motor current was calculated by obtained by controlling engine speeds and transmission gear
dividing the battery pack voltage from the battery power esti- selection around the area of the high efciency (sometimes
mated using a two-dimensional look-up table. referred to as the “sweet spot”) of engine speed vs. torque.
Detailed development of the rule-based VSC systems is
beyond the scope of this paper, and will be presented sepa-
Hybrid Vehicle Supervisory Controls 
rately in the future.
Engine ON and Torque Control A 10 second discharge As shown in Figure 9, the engine is turned on if the
power limit for the Li-ion battery pack was used to estimate demanded road-load power is greater than the available
the available Discharge Power Limit (DPL) by subtracting the battery pack discharge power limit. However, the engine is
discharge power at the minimum SOC from the discharge operated near a high-efciency region of engine speed and
power at the current SOC as shown in equation (10). torque to minimize fuel consumption and CO2 emissions as
shown in Figures 8 and 9. Any excess engine power is used to
DPLavail = DPL ( SOC ( t ) ) - DPL ( minimum SOC ) (10) charge the battery pack.
Figure 9 shows a total of 1071 seconds of engine-on time
A 30% SOC was used as a typical default minimum SOC.
from the model simulation of the UDDS, which is in an excel-
Te engine can be turned of when the sum of the demanded
lent agreement with the 1070 seconds of engine-on time from
road load power and the accessory electric power is less than
ANL chassis dynamometer test data of the 2013 Chevrolet
the available discharge power limit since the traction motor
Malibu Eco over the UDDS. Tus, the simulated engine-on
can be designed to provide sufcient electric-only propulsion
time was comparable to vehicle test data when the available
at low vehicle speeds for a P2 confguration or under more
battery discharge power limits were maintained.
limited conditions for a P0.

-
Engine power required is calculated by subtracting the
battery pack power from the sum of the road-load power and
the accessory electric power as shown in equation (11):  FIGURE 8   Engine Power Near Minimum bSfC Modeled for
2.5 L GM Ecotec Engine Compared to Chassis Dynamometer
Test Data of Engine Power and road Load (top) Along with the
Pengine = Proad _ load + Pacc - Pbatt
resulting Measured (red) and Modeled (blue) SoC (bottom)
Pengine for Phase 1 (initial 505 s) of the UDDS.
Tengine = (11)
(we )
1UU

where ωe is engine speed in radians per second. Te road-load ~

50
power contains the driveline system losses, gear efciency c!:.
cii 0
losses, aerodynamic drag, tire friction, etc.
US Government / US Environmental Protection Agency

~
0

Te VSC was developed primarily to calculate electric 0.. -50

- - - Engine Power
motor activation timing, the required torques from the motor/ -100 - - Engine Power@ min BSFC
generator, and the demanded engine torque. Te battery power 0 50 150 200 250 300 350 400 450 500

is discharged to provide the demanded electric motor torque

when the pedal acceleration is greater than a defned pedal 60

acceleration or engine load threshold while the vehicle tractive

power is greater than zero. Various threshold values were lso
u
stored in two-dimensional lookup tables as functions of 0
rn 40
engine speed and vehicle speed since an engine might be oper-
ating more efciently at somewhat higher vehicle speed,
30
depending on the transmission gear selected and road load 0 50 100 150 200 250 300 350 400 450 500
forces encountered (aerodynamic drag, rolling resistance, Elapsed Time (s]

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES 7

-  FIGURE 9  M easured (red) and Modeled (blue) Engine

Speed and Engine-oN State of 2013 GM Malibu Eco for the
UDDS (1 is Engine-oN, 0 is Engine-off).
- F IGURE 10  Vehicle Tractive Power, Engine Torque and SoC
Simulations (blue) Compared to Chassis Dynamometer Test
Data (red) over the UDDS.

j':;~ j

US Government / US Environmental Protection Agency

US Government / US Environmental Protection Agency

- 0 200 400 600 800 1000 1200 1400

o ~-~~-~~~~~~~-~--~~~-~~~

0 200 400 600 800 1000 1200 1400 i~O~

{=
Cl)
C
-1~ - . .
~SI~ . ~
1-- Engine ON I ·o,
ill 0 200 400 600 800 1000 1200 1400
z1 ~
0 60 ~ - - ~ - - ~ - - - ~ - - ~ - - ~ ~ - - - - ~ ~
Cl)
C
·o,
C
WO
u
~ 40

0 200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400
Elapsed Time [s] Elapsed Time [s]

∆PowerBatt,is estimated by subtracting the power limit at the

Te engine torque error, e(t) in equation (13), is obtained SOC target from the power limit at a current SOC level. Te
by subtracting actual vehicle speed from the specifed drive PowerBatt for compensating ∆SOC(t) must be limited to opera-
cycle vehicle speed. A generic PID controller shown in tion within the available battery discharging and charging
equation 13 can be used to estimate an optimum engine torque power limits.
to follow the demanded drive cycle vehicle speed during
engine-ON driving. Motor Torque and Running-Time Control Te traction
motor torque in the 48 V parallel MHEV simulations was
estimated by subtracting engine torque from the demanded
d
Tengine = K p e ( t ) + K I ò e ( t ) dt + K D e ( t ) dt driver torque:
dt
K P = 5.23, K I = 0.01, K D = 0.021
2 (13) ( ( ) )
Tmotor _ est = min Tmot _ max , Ttrac / g fdr -Tengine .rg / rg (15)

where T is torque, g fdr is the fnal drive gear ratio, rg is the

For the initial and fnal values of SOC balancing, the transmission gear ratio, Ttrac is the driver demanded torque/
following PID controller was implemented to compensate vehicle tractive torque, and Tmot_max is the maximum allowable
battery power in order to rapidly control the fnal SOC values motor torque estimated by the minimum values of the
to be closer to the initial SOC value as shown in the third plot discharge/charge power limits of the battery pack and motor
in Figure 10. Te SOC “swing windows” of the 0.4kWh 48 V peak power. Finally, the motor torque, Tmotor, is calculated by
battery pack were greater than the SOC “swing windows” of using the pulley ratios of the P0 electric machine or the speed
the production 0.5kWh 115 V 2013 Malibu Eco battery pack gear reduction ratios in the case of a P2 electric machine.
since SOC varies more quickly from the somewhat smaller As shown in Figure 11, motor power is provided in
capacity 48 V battery pack when charging and discharging addition to the engine power required to propel the vehicle to
using a similar magnitude of electric power. SOC is repre- assist with vehicle launching. Te peak motor power is limited
sented in the model in the following manner: for the frst 10 seconds, and reduced to normal motor power
levels afterwards. The motor runs for approximately
(
DPowerBatt = PL ( SOC ( t ) ) - PL SOCtarget ) 3~5 seconds to meet the demanded driver torque quickly
before reaching approximately 1200~1500 RPM, which is near
Powercomp = DPowerBatt + k P DSOC ( t ) a high efciency region of engine speed vs. torque. Te motor
+ k I ò DSOC ( t ) d ( t ) torque is gradually reduced to conserve battery power afer
the engine provides sufcient torque to propel the vehicle.
d
+ kD DSOC ( t ) d ( t ) Also, the battery is charged by excessive engine power while
dt running the engine more efciently. Te smaller negative
DSOC ( t ) = SOC ( t ) - SOCtarget , charging power to the battery pack is reduced by engine
k P = 15.7,k I = 3.5, k D = 0.018 (14) friction power, engine inertia and motor efciency converted
from mechanical power to electrical power. To rapidly balance
the fnal SOC values to the initial SOC values, the electric
where PL is the discharge and charge power limit for machine run time at launch is controlled adaptively as
positive and for negative current. Te battery power diference, described in the next section.
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 8 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES

-
 FIGURE 11   battery and Engine Power Simulation of a P0
48 V MHEV over the initial 200 Seconds of the UDDS

§" 25
- - Engine Power
- - Battery Power
-
 FIGURE 12   Measured (red) and Modeled (blue) SoC
Trajectory of 2013 Malibu Eco over the UDDS

60 ~ - ~ - - ~ - - ~ - - ~ - - ~ - ~ - - - - ,

-"' ~50
.,,:
~12 .5
8 40

US Government / US Environmental Protection Agency

US Government / US Environmental Protection Agency

0 0
a.. C/l

-15
30 L . _ - - ~ - - ~ - - ~ - - - ~ - - ~ - - ~ - - ~
0 20 40 60 80 100 120 140 160 180 20(
0 200 400 600 800 1000 1200 1400

f 70 I ~-~--~-~--~--:=.-=--=--=--=--=--=.,=._-=.-:=_-=,,
~
!- - Vehicle Speed
it 90
~
1- - Veticle Speed I
~ 50
.,
a.
.,
~70
~50
~ 30 (/)
0
.,
:i: ~ 30
.s::::
> QJ
0 >
0 20 40 60 80 100 120 140 160 180 20C
200 400 600 800 1000 1200 1400
Elapsed Time [s]
Elapsed Time [s]

SOC Trajectory and Balancing Control As shown in

TABLE 2  Chassis Dynamometer A, b, C coefcients for the
Figure 12, the SOC trace trajectories from the model and ANL standard and MHEV “Eco” version of the 2013
vehicle test data over the UDDS are similar even though the Chevrolet Malibu.
initial SOC values are diferent. Te SOC trajectory surface
can be constructed by various optimization schemes, but the Chassis Dynamometer Road-Load

US Government / US Environmental
optimization processes are computationally intensive. Vehicle Model Year Coefcients
Terefore, a delta SOC compensation from vehicle test data 2013 Chevrolet Malibu A 169.3883 N
and a PID controller were implemented to emulate typical b 2.3595 N/ (m/s)
hybrid electric vehicle SOC traces for the vehicle

Protection Agency
C 0.4092 N/(m/s)^2
model simulations. 2013 Chevrolet Malibu Eco A 135.24146 N
b 2.6976488 N/ (m/s)
C 0.3213237 N/(m/s)^2
Validation and Simulation
Table 2 shows the target road load coefcients used for chassis
dynamometer testing of the 2013 GM Chevrolet Malibu and
the 2013 Malibu Eco MHEV. Road load coefcients for a
conventional (non-MHEV/Eco) version of the 2013 Malibu
-
 FIGURE 13   Schematic representation of GT-DriVE Engine
State and ALPHA Engine Model DLLs (an enlarged version of
this fgure is also shown in Appendix figure 2)

~ _IEngllle-­

are shown for comparison. Te same 2013 Malibu Eco road-

cir._tng_Clrl
tf'l; _dft_a.1__
tng_Oft_lcieel_lnlll,_...

US Government / US Environmental Protection Agency

load coefcients were also used when modeling the Malibu
fflL_(Q9P_. . ._INOlt

Eco-based 48  V MHEV. Te ANL test results were used to

..._.,__
. . ._.,.._.._prog
.--_e111_-v.._ _tp0_rpa

......,____._,..
kC_ICIC. ._tlllCI_OOOI

Ar.'S..,.,__naq
.,_rq_...,.:
develop models of 48 V P0 and P2 MHEVs and to validate the
P0 MHEV model. Other test and model parameters were
obtained from ANL as well as from the published specifca-
tions of the vehicle’s manufacturer.
Te GT-DRIVE vehicle model was used by EPA for 48 V
MHEV model development at an early conceptual stage. A
visual schematic of the GT-DRIVE model is shown in fgure
13. Te EPA’s engine and battery sub-models were compiled
by Microsof Visual Studio 10 using the 2016a version of the
MATLAB/Simulink/Statefow toolbox to create DLLs repre-
senting these components and were used as components
within the GT-DRIVE model during model development. Te
ALPHA battery DLL was especially useful to precisely control represent any vehicle electrifcation and resultant changes in
the battery discharge and charge power limits, battery fuel consumption and GHG emissions. Te MHEV DLLs and
charging efciency, pack temperature, etc., within GT-DRIVE. control strategies developed in GT-DRIVER will eventually
Detailed modeling of battery power is critical to properly be ported back into the EPA ALPHA model.

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES 9

UDDS Simulations of 48 V P0 speed. Te transmission gear selection during the model simu-
lations was in an excellent agreement with the transmission
MHEV Model gears selected during chassis dynamometer testing.
As shown in Figure 14, the simulated engine torque was Te traction motor speed was determined by using vehicle
operated near a region of high efciency for the engine to speed, fnal drive ratio, tire radius, and speed reduction gear
minimize fuel consumption and CO2 emissions. Te trends ratio (including torque convertor lock/unlock condition). Te
of simulated engine torque and speed were in good agreement motor speed can be estimated by simple algebraic equations.
with the trends of the ANL chassis dynamometer test data. An accurate estimation of traction motor torque is critical since
Te simulated 57.1 Nm RMS engine torque for the 48 V P0 the demanded motor torque and speed are used as inputs to
MHEV is within 4.5% of the 54.7 Nm RMS torque for the the motor power maps to provide an estimate of battery current.
ANL vehicle test data [16]. Overall, the simulated engine Te 1.99 kW RMS battery charge power modeled at 48 V was
torque and speed (Figure 14) were in good agreement with the comparable to 1.99 kW RMS battery charge power from the
chassis dynamometer test data. 115 V Malibu Eco test data [16]. Overall, the simulated battery
Te gear position shown in Figure 15 was estimated based power and motor speed of 48 V P0 MHEV shown in Figure 16
upon driver pedal acceleration and transmission/vehicle agreed well with the battery power and motor speed measured
during chassis dynamometer testing of the 115 V Malibu Eco.

-  FIGURE 14   Comparison of Engine Torque, Engine Speed,

and fuel flow for the modeled (blue) 48 V P0 MHEV and
tested (red) 115 V P0 MHEV (Malibu Eco) over the UDDS
Te estimated motor current and accessory current were
used as inputs into the model of the battery pack to estimate
the resulting battery pack SOC and voltage. Charge/discharge
efciencies and battery pack temperature were also taken into
account when estimating the battery pack SOC. Te simulated
42.7A RMS current for the 48 V P0 MHEV was signifcantly

~ ~M~
higher than the 17.9A RMS from the 115 V MHEV vehicle test
~~
e- -1s8 ~ .

'1 data over the UDDS due to the lower battery pack voltage and
system voltage for the modeled 48 V system relative to the
US Government / US Environmental Protection Agency

~ tested 115 V system. As shown in Figure 17, the simulated fnal

0 200 400 600 800 1000 1200 1400
SOC of the 48 V battery pack had lower discharged battery
~3000 - - - ~ - - ~ - - ~ - - ~ - - ~ - - I power and higher fnal SOC over the UDDS compared to the
D.
~2000
~---~
production 115 V Malibu Eco battery pack used during vehicle
~ 1000 testing, and thus the modeled fuel economy results for the
48 V P0 MHEV represent a conservative estimation over the
Q)
a.
(/)
200 400 600 800
Elapsed Time (s]
1000 1200 1400
UDDS as shown in Table 3.
~3---~------~------~---l~----
~
0
u::
2
HwFET Simulations of 48  V
P0 MHEV
1
ai
tr O0
200 400 600 800 1000 1200 1400
As shown in Figure 18, engine torque and speed were operated

-
Elapsed Time (s]

near a high efciency region to minimize fuel consumption

 FIGURE 15   Comparison of Transmission Gear Selection
Data for the Modeled 48 V P0 MHEV and Measured P0 115 V
MHEV (top) relative to Pedal Acceleration and Vehicle Speed
(bottom) over the UDDS -  FIGURE 16   Comparison of battery Power and Motor Speed
for the Modeled 48 V P0 MHEV (blue) and Measured 115 V P0
MHEV (red) over the UDDS
US Government / US Environmental Protection Agency

US Government / US Environmental Protection Agency

2
1 u__ _,__,_.__ _ ,_,__.,_,_, _L.L-1-'-'-U-J..UUL~ - -LL-c.L-l-W.-'-'U-~_._,_._,

0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400

,. . . . , 100 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _--.,
40
:r:
D.
~
-0
Q)

~ 50 20
(/)
Q)
TI
:c
Q)
> 0 Lill:.ELJIJl..__JLlll1LlULl-1Y.-11JlC!LlllllliL"-lllllllJLUW.UWIL-1illLllll4liJ_fil_J 0 LL_ cLL~ _JJ_.Lil-L-l.1......U.JL- ' - ~ - - ' -...LI-'-_lllJLLJl--1-1.J_J

0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200 1400
Elapsed Time [s] Elapsed Time [s]
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 10 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES

TABLE 3  - Comparison of UDDS and HwfET fuel Economy for the 48 V and 115 V MHEVs

MHEVs Test Cycle Initial/Final SOC (%) CO2 (g/km) Fuel Economy (mpg) Test/Model

Protection Agency
115 V P0 MHEV UDDS 42/43.3 162.4 34.0 ANL Test

US Environmental
US Government /
48 V P0 MHEV UDDS 42/45.1 162.0 34.1 Model
115 V P0 MHEV HWfET 43/48.3 112.9 48.9 ANL Test
48 V P0 MHEV HWfET 43/46.7 115.4 47.9 Model

-
 FIGURE 17   Comparison of battery Voltage & SoC for the
Modeled 48 V P0 MHEV (blue) and Measured P0 115 V MHEV
(red) over the UDDS
RMS engine torque and engine power (respectively) for the
115 V MHEV chassis dynamometer test data [16]. Overall, the
simulated engine torque and speed shown in Figure 18 were
in good agreement with engine torque and speed from the
chassis dynamometer test data.

(':c:;·~:--·"t=SJI Figure 19 shows that the transmission gear was engaged

in the 6-speed position more frequently over the HwFET than
for lower speed drive cycles like the UDDS. Te gear selection

US Government / US Environmental Protection Agency

a. during the 48 V model simulations was in excellent agreement
0 200 400 600 800 1000 1200 1400
with gear selection during chassis dynamometer testing of

~~ 200 ~ Ii .I I I .I ~ . I :
~
- - ANL Test
Model~
the 2013 Malibu Eco 115 V P0 MHEV.
Te simulated 1.88KW RMS battery power for the 48 V
~ ~ , t t·r,~
0 1 P0 MHEV was within 6.0% of the 2.0 kW RMS battery power
a. -200 for the 115 V MHEV test data over the HwFET [16]. Overall,
0 200 400 600 800 1000 1200 1400
the simulated battery regenerative charging power and the
60 ~--~--~--~--~---.--r::::::::~M~ode
-:--:-1--,..,
motor speed of the 48 V MHEV shown in Figure 20 agreed
- - - ANL Test
~50 well with the regenerative charging power and motor speed
observed during chassis dynamometer testing of the 115 V
MHEV over the HwFET.
200 400 600 800 1000 1200 1400 As shown in Figure 21, the 39.6A RMS current for the
48  V MHEV was signifcantly higher than the 17.7A RMS

-
Elapsed Time (s]

current for the 115 V MHEV test data over the HwFET due to
 FIGURE 18   Comparison of Engine fuel flow for the the battery pack and system voltage change from 115 V to
Modeled 48 V P0 MHEV (blue) and Measured 115 V P0 MHEV 48 V. Te fnal SOC of the modeled 48 V battery pack was
(red) over the HwfET closer to the initial SOC and was lower than the fnal SOC of
the 115 V battery pack from the HwFET chassis dynamometer
test results, and thus the modeled 48 V MHEV modeling
results are conservative from a standpoint of modeled SOC

-
US Government / US Environmental Protection Agency

 FIGURE 19   Comparison of Transmission Gear Selection

1:~E+:-~ 0 100 200 300 400 500 600 700 800

Data for the Modeled 48 V P0 MHEV and Measured P0 115 V
MHEV (top) relative to Pedal Acceleration and Vehicle Speed
(bottom) over the HwfET

1~- ~ - -~ - - ~ - -~ - ~ - -~ - -~ - ~
~, ~ h 6

i;~ ~ I
US Government / US Environmental Protection Agency

ro
Q) 4
(!)

u. 0 100 200 300 400 500 600 700 800

2
Elapsed Time [s]
,~-~--~--~--~-~--~--~-~
0 100 200 300 400 500 600 700 800

and CO2 emissions. Excessive engine power was used to charge

the battery pack when demanded engine power was low.
However, demanded engine torque can be reduced to be closer
to regions of high efciency by supplying additional motor 20
torque when the demanded driver torque exceeded the
optimum engine torque. Te 79.2  Nm and 11.7  kW RMS
engine torque and engine power for the modeled 48 V P0 ~-~~-~-~---~-~--~~-~~~ o

100 200 300 400 500 600 700 800

MHEV was within 2% and 0.9% of the 77.7 Nm and 11.8 kW Elapsed Time [s]

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES 11

-  FIGURE 20   Comparison of battery Power and Motor

Speed for the Modeled 48 V P0 MHEV (blue) and Measured
115 V P0 MHEV (red) over the HwfET
Conclusion
In this paper, development and validation of a 48 V MHEV
model and co-simulations using 48 V Li-ion battery test data,
a scaled 48  V BISG motor efficiency map, the Gamma
[
iii;: 0
Technology GT-DRIVE vehicle model, and vehicle component
0 DLLs developed using EPA’s ALPHA vehicle model were
US Government / US Environmental Protection Agency

Cl..
i'.:' presented. Te MHEV simulation modeled using GT-DRIVE
~ - 12
includes a mathematical and rule-based VSC. Te VSC identi-
ID "' - ANLTest
- Model

fes overall energy fows by controlling key parameters such

0 100 200 300 400 500 600 700 800 as SOC, pedal acceleration/deceleration, vehicle speed, battery
power limits, and driver torque demand and allows the simu-
6000
~ lation to model 48 V MHEV GHG and fuel economy on a
second-by-second basis.
Cl..
~4000
.,.,
"O
Te 48 V MHEV P0 vehicle model was validated using
a.
~ 2000
2013 Chevrolet Malibu Eco 115 V P0 MHEV chassis dyna-
_g mometer test data provided by ANL. Te same motor power/
0
~
0
l= :t'~es,I torque and belt pulley ratios from the 2013 Malibu Eco 115 V
0 100 200 300 400 500 600 700 800 P0 MHEV were applied to the modeled 48 V P0 MHEV while
Elapsed Time [sl
using a 48 V/0.4kWh battery pack approximately equivalent

-  FIGURE 21   Comparison of battery Voltage, Current & SoC

for the Modeled 48 V P0 MHEV (blue) and Measured 115 V P0
MHEV (red) over the HwfET
in energy storage capacity to a 2018 Buick Lacrosse 86  V
MHEV battery. Hence, the major diferences between the
modeled and tested vehicle confgurations are the battery pack
and system voltage changes. Tere is also an approximately
50-pound weight reduction from the 115 V pack to a smaller
48 V Li-ion battery pack, but the resultant reduction in vehicle
inertia was not taken into consideration for this modeling in
order to facilitate MHEV model validation with vehicle test
data using the inertia of the production vehicle. Including the
weight reduction may result in slightly lower GHG and
US Government / US Environmental Protection Agency

improved fuel economy for the 48 V MHEV system relative

to the original 115 V MHEV system. Te modeled 48 V MHEV
simulations showed good correlation to MHEV test data. Te
48 V MHEV strategies developed for the GT-DRIVE model
simulations have subsequently been integrated into EPA’s
ALPHA vehicle model.
so ,-----~-~-~---;====:::;--'--~-7 Development of an HEV/MHEV model enables EPA to
u
1=-=--=~~~estl " determine the GHG efectiveness of new advanced technolo-
5l 40 ).... - ...,,,, I - ~ ..., - - ,., - - - -- ,,,... - .,.. - - .,.. - .J \.. r-. - - I gies. Te model also allows estimation of the resulting fuel
30 ~-~-~-~--~-~-~--~-~ economy and GHG emissions obtainable via optimization of
0 100 200 300 400 500 600 700 800 various vehicle system design variables such as vehicle weight,
Elapsed Time fsl drag coefcient, tire friction, HEV power-coupling systems,
regenerative braking efciency, engine fuel mapping, motor
recovery over the HwFET relative to the chassis efciency, battery power and energy densities, battery SOC
dynamometer tests. operating windows, battery discharge and charge power
During the HwFET, the SOC swing windows for both the limits, BMS thermal control strategies, and other vehicle
modeled 48 V and the tested 115 V battery pack were smaller design parameters.
than over the UDDS since the engine was already operating Modeling results for the 48 V P2 MHEV will be presented
relatively efciently. Te SOC values for both the modeled separately in the near future. High voltage P2 HEV and low
48 V P0 MHEV and the chassis dynamometer tested 115 V voltage P2 MHEV applications share many similarities except
P0 MHEV were within an acceptable range of the initial SOC that the 48  V MHEVs use smaller, lower torque electric
values for valid regulatory test cycle results. machines and a lower voltage battery pack for cost-savings
As shown in Table 3, the CO2 and fuel economy difer- and weight reduction at a penalty of lower GHG efectiveness.
ences between 48 V P0 MHEV model simulations and 115 V Conserving vehicle mass to provide additional available
P0 MHEV test data [16] were both within 0.3% over the payload is critical for the large light-duty trucks, and thus
UDDS, and were within 2.2% and 2.0%, respectively, over 48  V P0 and P2 MHEV systems may be well suited for
the HwFET. such applications.

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 12 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES

References Pack,” SAE Technical Paper 2018-01-0433, 2018,

doi:10.4271/2018-01-0433.
15. U.S. Department of Energy - Ofce of Energy Efciency and
1. Lee, B., Lee, S., Cherry, J., Neam, A. et al., “Development of Renewable Energy - Vehicle Technology Ofce, “AVTA: 2013
Advanced Light-Duty Powertrain and Hybrid Analysis Tool Chevrolet Malibu HEV Testing Results,” https://energy.gov/
(ALPHA),” SAE Technical Paper 2013-01-0808, 2013, eere/vehicles/downloads/avta-2013-chevrolet-malibu-hev-
doi:10.4271/2013-01-0808. testing-results, Dec 2017.
2. Newman, K., Kargul, J., and Barba, D., “Benchmarking and 16. Argonne National Laboratory, “2013 Chevrolet Malibu Eco,”
Modeling of a Conventional Mid-Size Car Using ALPHA,” from the Downloadable Dynamometer Database, http://
SAE Technical Paper 2015-01-1140, 2015, doi:10.4271/2015- www.anl.gov/energy-systems/group/downloadable-
01-1140. dynamometer-database/hybrid-electric-vehicles/2013-
3. Lee, S., Lee, B., Zhang, H., Sze, C. et al., “Development of chevrolet, Dec 2017.
Greenhouse Gas Emissions Model for 2014-2017 Heavy- and 17. Burress, T., “Benchmarking of Competitive Technologies,”
Medium-Duty Vehicle Compliance,” SAE Technical Paper Oak Ridge National Laboratory, Project ID: APE006, U.S.
2011-01-2188, 2011, doi:10.4271/2011-01-2188. DOE Hydrogen and Fuel Cells Program and Vehicle
4. U.S. Code of Federal Regulations, Title 49, Part 571, Subpart Technologies Program Annual Merit Review and Peer
B, §571.305 “Electric-Powered Vehicles: Electrolyte Spillage Evaluation Meeting, Washington, DC, May 2012.
and Electrical Shock Protection,” https://www.ecfr.gov/, 18. Johnson, V., Wipke, K., and Rausen, D., “HEV Control
accessed 13 July 2017t. Strategy for Real-Time Optimization of Fuel Economy and
5. Lee, S., Lee, B., McDonald, J., Sanchez, J. et al., “Modeling Emissions,” SAE Technical Paper 2000-01-1543, 2000,
and Validation of Power-Split and P2 Parallel Hybrid doi:10.4271/2000-01-1543.
Electric Vehicles,” SAE Technical Paper 2013-01-1470, 2013, 19. Kum, D., Peng, H., and Bucknor, N. K., “Supervisory Control
doi:10.4271/2013-01-1470. of Parallel Hybrid Electric Vehicles for Fuel and Emission
6. Chen, Min and Rincon-Mora, G.A., “Accurate Electrical Reduction,” ASME Journal of Dynamic Systems,
Battery Model Capable of Predicting Runtime and I-V Measurement and Control, Apr 2010 (DS-09-1340).
Performance," IEEE Transaction on Energy Conversion
21|(2:504-511, June 2006.
7. Lee, S., Lee, B., McDonald, J., and Nam, E., “Modeling and Contact Information
validation of Li-ion automotive battery packs,” SAE
Technical Paper 2013-01-1539, 2013, doi:10.4271/2013-01- SoDuk Lee, Ph.D.
1539. Assessment & Standards Division
8. Lee, S., Lee, B., McDonald, J., and Safoutin, M., “HIL US EPA - Ofce of Transportation & Air Quality
Development and Validation of Lithium-Ion Battery Packs,” 734-214-4373
SAE Technical Paper 2014-01-1863, 2013, doi:10.4271/2014- lee.soduk@epa.gov
01-1863.
9. Dekraker, P., Kargul, J., Moskalik, A., Newman, K. et al.,
“Fleet-Level Modeling of Real World Factors Infuencing Disclaimer
Greenhouse Gas Emission Simulation in ALPHA,” SAE Int. Tis document has been reviewed in accordance with U.S.
J. Fuels Lubr. 10(1):e0187101, 2017, doi:10.4271/2017-01-0899. Environmental Protection Agency policy and approved for
10. Rodriguez, J., Keribar, R., and Wang, J., “A Comprehensive publication. The views expressed by individual authors,
and Efcient Model of Belt-Drive Systems,” SAE Technical however, are their own, and do not necessarily refect those
Paper 2010-01-1058, 2010, doi:10.4271/2010-01-1058. of the U.S. Environmental Protection Agency.
11. Kim, N., Rousseau, A., and Rask, E., “Autonomie Model
Validation with Test Data for 2010 Toyota Prius,” SAE
Technical Paper 2012-01-1040, 2012, doi:10.4271/2012-01- Acknowledgments
1040.
Te authors would like to acknowledge the following persons
12. Kuypers, M., “Application of 48 Volt for Mild Hybrid for their cooperation to this model development and validation.
Vehicles and High-Power Loads,” SAE Technical Paper 2014- Kevin Stutenberg and Henning Lohse-Busch, Eric Rask,
01-1790, 2014, doi:10.4271/2014-01-1790. Michael J. Duoba of Argonne National Laboratory for
13. Kok, D., “Power of Choice: Te Role of Hybrid Vehicle providing chassis dynamometer test data.
Technology in Meeting Customer Expectations,” Te Battery Tim Burress at Oak Ridge National Laboratory for
Show and Electric & Hybrid Vehicle Technology providing motor efciency test data.
Conference, 2014. Mr. Joseph Wimmer and Ryan Tooley at Gamma
14. Lee, S., Cherry, J., Safoutin, M., McDonald, J. et al., Technologies for their extensive technical support with
“Modeling and Validation of 48 V Lithium-Ion Battery GT-DRIVE.

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES 13

Defnitions/Abbreviations GEM - Greenhouse Gas Emissions Model

GHG - Greenhouse gas
ALPHA - Advanced Light-Duty Powertrain and HEV - Hybrid electric vehicle
Hybrid Analysis
HIL - Hardware-in-the-loop
ANL - Argonne National Laboratory
HwFET - Highway fuel economy test procedure
ATDC - Afer top dead center
LD - Light duty
BISG - Belt integrated starter generator
MHEV - Mild hybrid electric vehicle
BMS - Battery management system
MTE - Midterm Evaluation
BSFC - Brake specifc fuel consumption
ORNL - Oak Ridge National Laboratory
BTE - Brake thermal efciency
PFI - Port fuel injection
CAD - Crankshaf angle degrees (°)
PID - Proportional-integral-derivative
DLL - Dynamic link library
SOC - State of charge
DOE - Design of experiments
TDC - Top dead center fring
EPA - Environmental Protection Agency
UDDS - Urban dynamometer driving schedule
GDI - Gasoline direct injection
VSC - Vehicle supervisory control

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 14 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES

Appendix

- ~---------------
 APPENDIX FIGURE 1   A 4-Quadrant 48 V biSG Motor Efciency Map Derived via Scaling of Publicly Available Data

80 .---------.---.-----~-

40

30

20
E'
~ 10
(lJ
::::::,
rr .
0 0
I-
L

.9
0
-10
~
-20

US Government / US Environmental Protection Agency

-30

-40

-1 .5 -1 -0.5 0 0.5 1.5

Motor Sp di [RPMQ x 10

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 MoDELiNG AND CoNTroLS DEVELoPMENT of 48 V MiLD HybriD ELECTriC VEHiCLES 15

- ~---------------
 APPENDIX FIGURE 2   Schematic representation of GT-DriVE Engine State and ALPHA Engine Model DLLs.

ALPHA_Engine Inputs:
ctrl_eng_on
eng_des_ac l_norm
eng_des_load_limit_ Nm
eng_ comp_iner_enable
trans_shifl_in_prog
trans_ctrl_eng_des_spd_rpm
tcc_lockup_cmd_bool
veh_Spd_kph
REVS_phase_fi ag
Ti acc_trq_Nm
acc_iner_kgm2
eng_spd_rpm

ALPHA_Engine Outputs:
eng_ spd _in_rpm
eng_des_pwr_kW
eng_max_pwr_kW
IK 1' - - - - - ---411
l ll
EN Gs pd -te st - Engin eS p ee d - eng_max_trq_Nm
1 1 eng_load_norm
Ga in-7- 1 Ve h S pd_ACC- 1 Re cv Ve h
Accel -3 eng_load_Nm
target_idle_ spd _rpm
ISC_state
eng_trq_Nm
US Government / US Environmental Protection Agency

eng_iner_kgm2
engine_fuel_rate_gps
engine_luel_consumed_g

ctrl_e ng_ o n-
1

Pe da lACC­ Ti me-6
t est- 2

This is a declared work of the U.S. Government and is not subject to U.S. copyright protection. foreign copyrights may apply. The U.S. Government assumes no
liability or responsibility for the contents of this paper or the use of this paper, nor is it endorsing any manufacturers, products, or services cited herein and any
trade name that may appear in the paper has been included only because it is essential to the contents of the paper.

Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE international. The author is solely responsible for the
content of the paper.

iSSN 0148-7191