Development of a range extended electric vehicle powertrain for an integrated energy systems research printed utility vehicle Applied Energy 191 (2017) 99–110 Contents lists available at ScienceDirect[.]
Trang 1Development of a range-extended electric vehicle powertrain for an
Paul Chambon⇑, Scott Curran, Shean Huff, Lonnie Love, Brian Post, Robert Wagner, Roderick Jackson, Johney Green Jr.1
Oak Ridge National Laboratory, 1 Bethel Valley Rd, Oak Ridge, TN 37831, USA
h i g h l i g h t s
Additive manufacturing plus
hardware-in-the-loop speed up
vehicle development process
Additive manufacturing shown to
provide design flexibility and fast
turnaround time
Powertrain integrated and optimized
though HIL testing prior to vehicle
installation
Printed electric vehicle experiments
show range extension with
natural-gas genset
Vehicle experiments confirm
simulation + HIL + AM process to
accelerate vehicle design
g r a p h i c a l a b s t r a c t
a r t i c l e i n f o
Article history:
Received 2 August 2016
Received in revised form 3 January 2017
Accepted 15 January 2017
a b s t r a c t
Rapid vehicle and powertrain development has become essential to for the design and implementation of vehicles that meet and exceed the fuel efficiency, cost, and performance targets expected by today’s con-sumer while keeping pace with reduced development cycle and more frequent product releases Recently, advances in large-scale additive manufacturing have provided the means to bridge hardware-in-the-loop (HIL) experimentation and preproduction mule chassis evaluation This paper details the accelerated development of a printed range-extended electric vehicle (REEV) by Oak Ridge National Laboratory, by paralleling hardware-in-the-loop development of the powertrain with rapid
http://dx.doi.org/10.1016/j.apenergy.2017.01.045
0306-2619/Ó 2017 The Authors Published by Elsevier Ltd.
Abbreviations: ABS, acrylonitrile butadiene styrene; AM, additive manufacturing; AMIE, Additive Manufacturing Integrated Energy; APU, auxiliary power unit; BAAM, big area additive manufacturing; BEV, battery electric vehicles; BEVx, Range Extended Battery Electric Vehicle; CAD, computer-aided design; CAN, controller area network; CARB, California Air Resources Board; CNG, compressed natural gas; EPA, Environmental Protection Agency; EREV, extended-range electric vehicles; FEERC, Fuels, Engines and Emissions Research Center; HIL, hardware-in-the-loop; HOV, high occupancy vehicle; HWFET, Highway Fuel Economy Test; ORNL, Oak Ridge National Laboratory; PHEV, plug-in hybrid electric vehicles; PUV, Printed Utility Vehicle; REEV, range-extended electric vehicle; REx, range extender; RLF, road load force; SAE, Society of Automotive Engineers; UDDS, Urban Dynamometer Driving Schedule; VRL, Vehicle Research Laboratory; WOT, wide-open throttle.
q This manuscript has been authored by UT-Battelle, LLC, under Contract No DE-AC05-00OR22725 with the U.S Department of Energy The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan ( http://energy.gov/downloads/doe-public-access-plan ).
⇑ Corresponding author.
E-mail addresses: chambonph@ornl.gov (P Chambon), curransj@ornl.gov (S Curran), lovelj@ornl.gov (L Love), postbk@ornl.gov (B Post), wagnerrm@ornl.gov (R Wagner),
jacksonrk@ornl.gov (R Jackson), Johney.Green@nrel.gov (J Green Jr.).
1 National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, CO 80401, USA.
Contents lists available atScienceDirect
Applied Energy
j o u r n a l h o m e p a g e : w w w e l s e v i e r c o m / l o c a t e / a p e n e r g y
Trang 2Printed vehicle
Range extender
Additive manufacturing
Rapid prototyping
Hybrid vehicles
Natural gas
chassis prototyping using big area additive manufacturing (BAAM) BAAM’s ability to accelerate the mule vehicle development from computer-aided design to vehicle build is explored The use of a hardware-in-the-loop laboratory is described as it is applied to the design of a range-extended electric powertrain to
be installed in a printed prototype vehicle The integration of the powertrain and the opportunities and challenges it presents are described in this work A comparison of offline simulation, HIL and chassis rolls results is presented to validate the development process Chassis dynamometer results for battery elec-tric and range extender operation are analyzed to show the benefits of the architecture
Ó 2017 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
1 Introduction
The use of additive manufacturing for rapid vehicle prototyping
is an emerging field Along with hardware-in-the-loop (HIL),
large-scale additive manufacturing (AM), or Big Area Additive
Manufac-turing (BAAM), has recently been shown to have potential use for
rapid prototyping of vehicles for evaluation and development
pur-poses AM creates components directly from a computer model
using an additive process in which material is added to build a
com-ponent, as opposed to a subtractive process such as machining, in
which material is removed from a billet to manufacture a
compo-nent AM is well suited for rapid prototyping, as it is extremely
flexible and enables the rapid creation of very complex geometries
with minimal waste This technology could be transformative in
many areas, including the automotive sector
A recent paper by Curran et al.[1]documents how the use of
polymer-composite AM with a BAAM system and a HIL test facility
can facilitate the rapid manufacture of lightweight, complex
com-ponents and impact a broad spectrum of manufacturing industries
In that paper, the process to develop a functional battery electric
vehicle 3D-printed as a Shelby Cobra replica was described,
includ-ing vehicle systems simulations, HIL development, and integration
of the powertrain into the printed vehicle platform
Range-extended electric vehicles (also referred to as
extended-range electric vehicles (EREV) or Range Extended Battery Electric
Vehicle (BEVx)) are defined by CARB as ‘‘a vehicle powered
pre-dominantly by a zero emission energy storage device, able to drive
the vehicle for more than 75 all-electric miles, and also equipped
with a backup auxiliary power unit (APU), which does not operate
until the energy storage device is fully depleted” It is a
sub-category of plug-in hybrid electric vehicles (PHEV) and consists
most of the time of a series hybrid architecture Simply stated, they
are electric vehicles with an auxiliary power unit (APU) (also
known as range extender (REx)) used solely to recharge the ESS
and extend the vehicle electric range; the APU internal combustion
engine cannot propel the vehicle directly as it does not have a
con-nection to the wheels REx can help alleviate ‘‘range anxiety”
com-pared to battery electric vehicles (BEV) as they can significantly
increase the range of the vehicle should a longer trip be required,
without burdening the vehicle with extra weight and cost of an
oversized ESS In zero-emission vehicle regulation, these
defini-tions become important in the design and marketing of such
pow-ertrain architectures For example, recent polices from the
California Air Resource Board are driving the design of
range-extended electric vehicles as they can grant their owners access
to high occupancy vehicle (HOV) lanes, and provide manufacturers
the same credits as BEVs towards California’s zero emissions
vehi-cles (ZEV) mandate[2]
To provide a flexible research platform for investigating range
extender options for electric powertrains and for integrated energy
research, Oak Ridge National Laboratory (ORNL) has partnered
with many collaborators through the Additive Manufacturing
Inte-grated Energy (AMIE) project [3] AMIE collaborators have
designed and built a functioning printed utility vehicle and printed house, as shown inFig 1
The underlying key innovative solutions used in the develop-ment of AMIE include the following
Advanced Manufacturing: The vehicle and building were 3D printed using ORNL’s BAAM system, demonstrating how 3D printing can get products to market faster than traditional man-ufacturing techniques
Vehicle Technologies: The vehicle has a hybrid electric power-train with onboard power generation from natural gas A single engine extends the vehicle’s range and produces power for both vehicle and building Energy flows between the two using fast, efficient level-2 bidirectional wireless power transfer—the world’s first wireless charging technology of its kind[4]
Building Technologies: A team of researchers and architects designed a single-room building to demonstrate new manufac-turing and building technologies It incorporates low-cost vacuum-insulated panels into a 3D printed shell that was assembled at Clayton Homes, the nation’s largest manufactured home builder, in a partnership with the University of Tennessee (UT)
Sustainable Electricity: A 3.2 kW solar panel system paired with electric vehicle batteries generates and stores renewable power Advanced building control and power management strategies integrate the various energy systems and enable the building
to function as a virtual battery The building charges the vehicle, and the vehicle can provide power to the building
The HIL concept is deployed to develop the powertrain and its controls within the same accelerated time frame that is achieved
by BAAM to create the vehicle chassis In a larger context, HIL plat-forms are used to analyze any component or subsystem hardware
in a virtual vehicle environment These platforms are valuable for developing controls in a safe, controlled environment where the regular surroundings of the unit under development are simulated using physics-based models[5–7] Typically, a HIL platform con-sists of a real-time run-time computer running a model of the com-plete vehicle, except for the component being evaluated in the research test facility Analog and digital signal emulation hardware boards are used to interact with sensors and actuators, and handle the communications the component would experience in a real vehicle With this environment in place, the component being eval-uated responds as though it is installed on a vehicle driving under anticipate usage (either real-world conditions or certification drive cycles) In the remainder of this paper, -in-the-loop may be appended to the name of any component characterized on the HIL platform, such as motor, powertrain, or controller The exten-sion of vehicle systems simulations to HIL to a working prototype marks a step forward in rapid controls prototyping[8]
The vehicle design, build and validation process can be summa-rized as shown in Fig 2 Controls were developed offline and refined in the HIL environment In parallel, the chassis was
Trang 3designed in computer-aided design (CAD) and printed using BAAM.
The simultaneous nature of both the powertrain and chassis
devel-opment activities allows significant reduction of the overall vehicle
development duration
The resulting prototype vehicle, called Printed Utility Vehicle
(PUV), is intended to be a flexible research platform[9,10] This
means that the vehicle architecture is designed to be modular such
that most components can be replaced to investigate different
technologies For instance, the APU and battery pack could be
sub-stituted with alternative specification units in order to provide
researchers with experimental vehicle results after evaluating
these units in a laboratory This flexibility comes at the cost of
packaging which is not optimized so that different shapes and
forms can be accommodated Yet additive manufacturing allows
quick iterative designs should packaging of new units require a
major chassis modification
This paper describes the use of a hardware-in-the-loop
labora-tory for the design of a range-extended electric powertrain to be
implemented in a printed prototype mule BAAM’s ability to
accel-erate the mule development from computer-aided design to
vehi-cle build is explored The integration of the powertrain and the
opportunities and challenges it presents are detailed in this work
A comparison of offline simulation, HIL and chassis rolls results is presented to validate the development process Chassis dynamometer results for battery electric operation and range extender operation are analyzed to show the benefits of the architecture
2 Mule vehicle chassis rapid prototyping with BAAM The following section describes the use of BAAM to print the body and frame of the research vehicle (Fig 3) Details of the use
of BAAM for rapid vehicle prototyping were presented in the printed Cobra study by Curran et al.[1] The BAAM system used for this project was upgraded from the system used for the Cobra The new Cincinnati Inc BAAM used a new higher capacity (45 kg/ h) extruder on a larger (2.4 6.1 1.8 m) gantry system Other key features include the use of a 7.6 mm diameter nozzle, resulting
in a 0.76 mm surface variation.Fig 4shows the BAAM printing the PUV frame The BAAM extruder shown inFig 3uses conventional carbon fiber–reinforced pellets from the injection molding indus-try; these are typically under $5/lb.Table 1gives the characteris-tics of the extruder shown inFig 3
Similar to the printed Cobra[1], an acrylonitrile butadiene styr-ene (ABS) plastic reinforced with carbon fiber was used for the frame and body of the PUV In this relatively new concept, chopped carbon fiber is blended with a thermoplastic Previous experiments had determined that blending carbon fiber with the ABS at higher than 15–20% led to a significant reduction in warping without use
of an oven This is described further in a recent study on the effects
of adding carbon fiber to ABS[11] Elimination of the oven signifi-cantly reduces the energy intensity (i.e., the energy required to manufacture parts) of the manufacturing process Conventional polymer extrusion systems with heated chambers have an energy intensity in excess of 100 kW h/kg This behavior makes the addi-tion of carbon fiber an enabling technology for large printed work-pieces and can eliminate the need for additional ovens typically used to prevent warping BAAM’s energy intensity is 1.1 kW h/kg, approximately two orders of magnitude below other polymer
Fig 1 Completed AMIE demo with 3D printed house and 3D printed utility vehicle.
Trang 4extrusion AM systems ABS reinforced with carbon fiber shows
increased strength and a significant reduction in distortion
Addi-tional details on carbon fiber–filled ABS for large-scale printing
can be found in Refs.[12,13]
2.1 Mechanical design
The mechanical design of the vehicle body and frame was
cre-ated to accommodate the drivetrain, suspension, and battery
com-ponents already selected during the simulation study described
later on in this paper,Fig 5shows the vehicle chassis and main
component layout
The design was also driven by the special requirements of 3D printing with the carbon fiber–reinforced ABS polymer For instance, the vehicle frame was designed specifically to ensure that stresses remain below 69 Bar, which is a factor of safety on the inter-laminar strength, (i.e the minimum stress that would cause layer separation and frame failure) Printed structures generally have anisotropic material properties, with the weakest direction being a result of the layer-to-layer adhesion The design incorpo-rated threaded rods that could pass through the layers serving two purposes: providing attachment points for drivetrain compo-nents and putting the printed material under compression, thus improving the integrity of the chassis (Fig 6)
SolidWorks design software was used for all mechanical model-ing As with conventional additive manufacturing, CAD files were transformed to stereolithography (STL) files and entered into a slic-ing program that transformed the 3D geometry to machine tool path commands The frame was printed in a single process taking approximately 12 h, with the final frame weighing approximately
386 kg This rate is 2500 times above that normally associated with the polymer AM processes
In addition to the frame, the skins took approximately 8 h to print, and support structures took approximately 4 h to print Rather than making attempts to light-weight for this case study, components were printed for robustness and workability 2.2 Vehicle build
While the entire frame and body of this vehicle were printed, other components such as the suspension, and powertrain were conventional A similar process was used for the printed Shelby Cobra, in which the front suspension was modified from a com-mercially available aftermarket suspension kit The rear suspen-sion was modified from a rear-wheel-drive passenger vehicle Modifications to the rear suspension included the addition of shock mounting points and a cradle for the traction motor and gearbox which integrated into space originally occupied by the rear differ-ential All structural modifications to both the front and rear sus-pensions were made by welding new redesigned carbon steel attachment points An aftermarket brake system was used The frame was designed so that the front suspension would bolt on with the modified suspension pieces There were three general types of printed body parts: (1) the frame, which is the load-bearing structure that integrated the drivetrain, (2) thin skins which were printed for body panels, and (3) skeletal beam struc-tures, which serve as the interface between the skins and the frame The skins were bonded using Valvoline Pliogrip to the skele-tal structures, which were then mechanically attached to the frame for ease of removal (Fig 7)
3 HIL development of a hybrid electric vehicle powertrain This section describes the development of the PUV powertrain
It relied heavily on the HIL process and was conducted in parallel with the chassis printing and mechanical assembly activities This phase started with simulations for component sizing, performance prediction and controls development, before implementing a HIL environment for the complete powertrain to validate its controls and coordination as if it were already installed in a vehicle 3.1 Simulation phase
The simulation study is the first phase of the HIL process Dur-ing simulation, all components are simulated and represented by a mathematical model The same model is used throughout develop-ment, but some components will progressively be replaced with
Fig 4 ORNL BAAM printing PUV frame.
Fig 3 BAAM system installed at the ORNL manufacturing demonstration facility.
Table 1
BAAM system extruder characteristics.
Specification Extruder type Single screw
Fill rate 45 kg/h
Extruder temperature 350 °C
Bed temperature 120 °C
Trang 5real parts as hardware becomes available: they then become
‘‘in-the-loop” with the rest of the simulated environment Autonomie,
the system-level modeling environment from Argonne National
Laboratory, was selected for this simulation study, whereas a
dSPACE platform ran the HIL real-time model Both are based on
MathWorks Simulink and therefore provide a seamless transition
from pure offline simulation to HIL experiments since the same
model is used in both phases without the need for any conversion
A series hybrid electric architecture[14]was created and
pop-ulated with each component’s actual specifications, as shown in
Fig 8
3.1.1 Powertrain component sizing study
This model was first used to size components based on vehicle
performance requirements determined by the engineering team:
Range-extended electric vehicle architecture,
30–40 miles (48-64 km) of all electric range,
over 60 mph top speed, and
Under 20 s for 0–60 mph
Even though in-house AM provides complete control over chas-sis design and lead time, component selection is restricted by whether products are available off the shelf This is especially true for batteries, because procuring one-off battery packs for prototype vehicles can be very difficult Fortunately, Johnson Controls, Inc offered their 14.2 kW h pack for this project This battery capacity will allow 30 miles or more of all-electric range if the vehicle con-sumes 300 W h/mile and 70% depletion out of the battery is per-mitted The pack operates at a nominal voltage of 345 V with a capacity of 41 A h and peak power of 100 kW (continuous power rating of 41.4 kW)
To match the battery power, the 90 kW Remy traction motor and 100 kW Reinhart inverter were selected With a top motor speed of 8000 rpm, the motor requires a final drive ratio smaller than 12:1 to achieve a top vehicle speed faster than 60 mph The Borgwarner eGearDrive axle with a final drive ratio of 8.28:1 was chosen
APU sizing has been reported for other series hybrid vehicle configurations[15,16] A similar study was conducted for this vehi-cle The model was run to determine average electrical power con-sumption on key drive cycles (Urban Dynamometer Driving Schedule (UDDS), Highway Fuel Economy Test (HWFET) and US06 cycle), as well as for steady state speeds for several road grades.Fig 9shows the simulation results and was used to deter-mine the APU size: the target electrical power for the APU was set
to 20 kW in order to be able to sustain a highway cycle and cruise speeds of 50 mph on level ground
Due to procurement delays with the component originally selected, the team reverted to a commercially off-the-shelf APU with a power limitation of 5.5 kW This sub-optimal initial compo-nent selection will be corrected in the future thanks to nature of the PUV’s flexible research platform concept which provides a modular architecture where most components can be replaced to investigate different technologies at the vehicle level
3.1.2 Performance simulation study The sizing study was followed by a performance prediction study The same model was populated with specifications for the selected components and ran over a few predetermined cycles and maneuvers:
0–60 mph wide open throttle acceleration,
UDDS drive cycle, representative of city driving,
HWFET drive cycle, representative of highway driving, and
US06 drive cycle, a more aggressive cycle
Fig 5 Powertrain configuration for PUV.
Trang 6Table 2shows the simulated energy efficiency for the three key
drive cycles when operating in all-electric mode Note that the
highway cycle consumes significantly more energy than the city
cycle due to the large aerodynamic drag coefficient of this utility
vehicle causing large high speed losses
Acceleration simulations yielded slower-than-expected perfor-mance of 25 s to accelerate from 0 to 60 mph due to the limited power available from the battery due to its conservative fusing of
120 A; note that the fuse is capable of larger currents for shorter durations but that feature was not modelled and instead a hard limit was used in the model That limitation was accepted, as it allowed the project to proceed on time thanks to the battery pack being readily available off the shelf
The final task performed during the simulation phase was the development, debugging, and optimization of the hybrid electric powertrain control algorithms The off-line environment is ideal
to create strategies They can be evaluated quickly over an extended set of experimental conditions because they do not have
to run in real time This is also a safe debugging environment, as no hardware is involved
3.2 HIL experimentation phase During the HIL experimentation phase, some hardware is being evaluated in the research cell, and some is simulated on a real-time computer so that the real components respond as though they are
Fig 7 PUV powertrain components before final assembly.
Fig 8 PUV series hybrid electric Autonomie architecture.
Fig 9 Vehicle average electrical traction power chart used for APU sizing study.
Trang 7installed in their normal environment, as a vehicle driving on a
road The principle is shown inFig 10
In this case, the unit being evaluated is the whole powertrain
(electric traction motor, inverter, battery pack, DC/DC converter,
etc.) It is installed in ORNL’s Vehicle System Integration laboratory
research cell and subjected to road load conditions generated by a
dynamometer The dynamometer is controlled by a real-time
com-puter which runs a model of the vehicle components not physically
present in the research cell (e.g., chassis, wheels, final drive axle,
driver, speed profiles) The unit being evaluated and the
real-time computer interact with each other to create realistic
condi-tions: the powertrain behavior affects the vehicle model, and the
vehicle model dictates the control commands out to the
powertrain
The HIL phase is instrumental in implementing and debugging
vehicle communications, which are often transparent in the
simu-lation phase, as model blocks seamlessly exchange floating point
variables with consistent units, whereas in the real world, different
manufacturers use different units, resolution, and protocols
Once interfaces are debugged, control algorithms can be
modi-fied and fine-tuned to account for the real-time implementation
and limitations that might not have been accounted for by the
model
Because of this project’s accelerated time frame, once the
pow-ertrain achieved full functionality on the HIL system, it was
decommissioned for installation in the vehicle Therefore, drive cycle performance was not evaluated in that configuration Only wide-open throttle (WOT) experiments were performed to check powertrain operation under high speeds and loads These tests were completed successfully providing the team with the confir-mation that the powertrain was mechanically and electrically ready to be safely installed in the vehicle
3.3 Vehicle integration phase The powertrain system was then installed in the actual vehicle (seeFig 7), allowing for immediate operation, because most inte-gration and controls issues had been identified and resolved during HIL evaluations This reduced the vehicle integration phase to
Fig 11 PUV being evaluated on ORNL chassis dynamometer.
Table 2
Simulated energy efficiency modeling results.
Drive
cycle
Distance
(miles)
Energy (W h)
Model energy consumption (W h/mile)
UDDS 7.47 2201.6 294.8
HWFET 10.21 4089.5 400.5
US06 7.78 4087.6 525.2
Trang 8mostly wiring and mechanical installation tasks This phase
com-pleted the vehicle build The vehicle is now operational and ready
for evaluation
4 Chassis dynamometer setup
Chassis rolls are an ideal environment to conduct reliable and
repeatable vehicle evaluations because of the controlled indoors
test conditions and the facility’s ability to emulate the road load
experienced by the vehicle when driving outdoors
ORNL’s Fuels, Engines and Emissions Research Center (FEERC)
Vehicle Research Laboratory (VRL) is equipped with a Burke E
Por-ter 300 hp motor-in-the-middle, two-wheel drive, 48-inch, single
roll AC motoring chassis dynamometer The dynamometer meets
the requirements of the US Environmental Protection Agency
(EPA) specifications for large roll chassis dynamometers The
flex-ible driver’s visual aid and control system facilitate performance of
all standard federal drive cycle tests, as well as European, Japanese,
or custom cycles The laboratory has been cross checked against
independent certification labs, and results are in excellent
agree-ment for fuel economy and vehicle emissions
4.1 Road load determination
The forces exerted on a moving vehicle can be broken into three
components: (1) forces due to the vehicle’s inertia, (2) gravitational
forces due to traversing up or down a grade, and (3) the road load
force (RLF) The RLF accounts for the vehicle’s mechanical drag, i.e.,
its rolling resistance and its aerodynamic drag This force must be
derived so that it can be duplicated by the dynamometer, which is
a two-step process The Society of Automotive Engineers (SAE)
J2263 Standard is used to derive the RLF that a given vehicle
experi-ences under real world conditions Once that is completed, the SAE
J2264 Standard is used to replicate the RLF on the dynamometer
The range extender PUV was subjected to the road load
deriva-tion J2264 coast-down test in the VRL The vehicle was confirmed
to be in good working order, with tires filled to 40 psi Normally,
per the SAE J2264 Standard, the vehicle is preconditioned by
run-ning a double HWFET In order to avoid discharging the vehicle’s
battery, a motoring warmup procedure was implemented that
approximates the average speed and the same time as a double
HWFET test to warm-up the tires, dyno and powertrain This cycle
ramped up to 45 mph by 3 mph/s, and then using an alternating speed from 45–55 mph
The same RLF coefficients that were used in the offline simula-tion were used to develop the set coefficients on the motoring dynamometer for performing model validation These coefficients are not expected to match what the vehicle would experience dur-ing SAE Standard J2263 coast downs on the road but instead are part of the rolling laboratory procedure of going from vehicle sys-tems simulations and HIL experiments to chassis dynamometer experiments (Fig 11)
4.2 Instrumentation The vehicle was instrumented with a Hioki power analyzer that measured battery and electric machine current The vehicle con-troller area network (CAN) bus was accessed to record information reported by key powertrain components such as the battery pack, the electric machine inverter, and the on-board charger Power and energy were calculated using a combination of external instru-ments (such as the Hioki power analyzer) and self-reported values
on the CAN bus
5 Vehicle chassis dynamometer experiments First, the chassis laboratory was used to refine the regenerative braking strategy and further develop controls Then the PUV per-formed certification drive cycles to measure energy consumption,
Fig 12 Brake regeneration event on prototype vehicle on chassis rolls (blue line) and in simulation (dotted red line) (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig 13 REx APU efficiency as a function of load.
Trang 9and acceleration experiments to characterize 0–60 mph
performance
5.1 Regenerative braking tuning
Some noteworthy drivability considerations were found going
to full chassis dynamometer experiments on the vehicle, in
partic-ular with regenerative braking
The model implemented some phase-out strategies so that
foundation brakes will progressively take over from regenerative
braking at low speeds However, regenerative algorithms were
aggressively calibrated to capture as much braking energy back
to the battery as possible and would indicate that close to 100%
regenerative braking was available However, this ignores
drivabil-ity issues around the feel of the braking in the actual vehicle
Dur-ing vehicle experiments, some refinement of the regenerative
braking strategy was completed to eliminate shudder and improve
brake feel This implied limiting motor braking therefore reducing
amount of energy recovery to the battery and lowering vehicle
energy efficiency This issue was not detected during the
simula-tion phase because the driveline model is not detailed enough to
capture resonance phenomenon induced by foundation brakes, for instance gear backlash and shaft stiffness and inertia are not modelled The HIL phase did not help with this issue either because the driveline from the motor to the wheels was not ‘‘in-the-loop”, and it was modelled the same way as in simulation HIL could the-oretically capture this issue but it would require a dual dynamometer set-up to have the powertrain and rear differential under test, with each dynamometer emulating a wheel
Fig 12shows the motor torque during a deceleration event at the end of the highway cycle for both the simulation (dotted red line) which tried to maximize brake regeneration, and for the prop-totype vehicle tested on chassis rolls (blue line) The experimental torque is much lower than the simulated one; the real vehicle does not regenerate as much braking energy in order to prevent shudder
in the driveline
5.2 Range extender characterization For the initial range extender experiments with the PUV rolling laboratory, a Honda EU7000 APU capable of 7.5 kW peak (5.5 kW
Fig 14 PUV electrical power consumption as a function of vehicle speed.
Fig 15 Comparison of instantaneous chassis rolls (blue line) results with simulation results (red line) on HWFET cycle (For interpretation of the references to colour in this
Trang 10continuous) was used The APU was converted to natural gas using
an aftermarket retrofit kit[17]
For the chassis dynamometer experiments, residential natural
gas was supplied to the generator instead of using the compressed
natural gas (CNG) tanks on board The fuel consumption was
calcu-lated using the standard EPA carbon balance method which
employs a constant volume sampling system with bagged exhaust
emissions over the drive cycle[18] The lower heating value of the
residential natural gas was calculated using the component volume
percent from the gas chromatograph posted online for the dates
used and was calculated to be 49,537 kJ/kg The supplied natural
gas was reported to be composed of 92% methane, 7.2% ethane,
0.45% N2, 0.31% propane, and 0.16% CO2, with other minor species
Fig 13shows the generator efficiency as a function of load
5.3 Steady state battery electric vehicle results
The vehicle was subjected to steady-state road loads to
charac-terize its electric energy consumption as a function of speed, with
the REx disabled The results are shown inFig 14 Experimental
results match well with simulation results obtained during the
siz-ing study and confirm that less than 20 kW is required to cruise at
50 mph, and the 5.5 kW provided by the APU is adequate to
main-tain the steady-state speed of 25 mph
5.4 Battery electric vehicle model validation
The vehicle was subjected to the same drive cycles and
acceler-ation experiments performed in the offline model environment:
UDDS, HWFET, and US06 drive cycles and 0–60 mph wide-open
throttle acceleration
5.4.1 BEV drive cycle results
The UDDS and HWFET cycles were repeated three times on the
chassis rolls setup with the vehicle in its BEV configuration
Exper-imental cycles showed very good repeatability: the coefficient of
variance of energy consumption was less than 0.7% on the three
experiment replicates
Experimental results were then compared with simulation
results An example of instantaneous trace comparison is shown
inFig 15: blue traces are simulation results whereas red traces
are experimental results collected on the chassis rolls with the
pro-totype vehicle: traces are matching very closely for all variables:
vehicle speed, motor torque and battery current
Battery energy was also calculated over complete drive cycles
for both methods The difference is consistently within 1.5% on
all three cycles, as shown inFig 16
The good correlation between the chassis experiment and sim-ulation is a clear indicator of the quality of the model, but it also benefits from the fact that the chassis rolls road load coefficients were calculated from the theoretical model values instead of from experimental coast-down results, which would have most likely yielded different rolling resistance and drag coefficients, which in turn, would have resulted in different vehicle level efficiencies 5.4.2 BEV acceleration experimental results
HIL experiments were performed before the prototype vehicle was completed and used an anticipated vehicle weight of
1587 kg (3500 lb) which was lighter than the actual prototype vehicle ended up being (1824 kg (4000 lb)) Therefore HIL WOT
Fig 16 Comparison of experimental BEV results with modeling results.
Fig 17 Comparison of experimental chassis rolls BEV acceleration time with modeling and HIL results.
Fig 18 Comparison of HIL results of BEV acceleration time with modeling results.