Colorado State University EcoCAR 3 Final Technical Report

2019-01-0360 Published 02 Apr 2019

Colorado State University EcoCAR 3 Final Technical ReportGabriel Christian DiDomenico, Jamison Bair, Vipin Kumar Kukkala, Jordan Tunnell, Marco Peyfuss, Michael Kraus, Joshua Ax, Jeremy Lazarri, Matthew Munin, Corey Cooke, Eric Christensen, Logan Peltz, Nathan Peterson, Logan Wolfe, Zach Vinski, Daniel Norris, Corrie Kaiser, Jacob Collier, Nick Schott, Yi Wang, and Thomas Bradley Colorado State University

Citation: DiDomenico, G.C., Bair, J., Kukkala, V.K., Tunnell, J. et al., “Colorado State University EcoCAR 3 Final Technical Report,” SAE Technical Paper 2019-01-0360, 2019, doi:10.4271/2019-01-0360.

Abstract

Driven by consumer demand and environmental regu-lations, market share for plug-in hybrid electric vehicles (PHEVs) continues to increase. An oppor-

tunity remains to develop PHEVs that also meet consumer demand for performance. As a participant in the EcoCAR 3 competition, Colorado State University’s Vehicle Innovation Team (CSU VIT) has converted a 2016 Chevy Camaro to a PHEV architecture with the aim of improving efficiency and emissions while maintaining drivability and performance. To verify the vehicle and its capabilities, the CSU Camaro is rigorously tested by means of repeatable circumstances of physical operation while Controller Area Network (CAN) loggers record various measurements from several sensors. This data is analyzed to determine consistent output and coordination between components of the electrical charge and discharge system, as well as the traditional powertrain.

The aim is to improve drivability and efficiency as measured by vehicle technical specifications (VTS) including accelera-tion, energy consumption, and emissions. In this interest, the team focused on the areas of mass reduction, efficient powertrain operation as well as optimal engine and motor use. While there is incomplete evidence showing that targets have been met in these areas, this study definitively shows improvement from year to year of the competition and specifically during Year 4 when the vehicle was tested exten-sively. Mass reduction resulted in more acceleration. Efficient powertrain operation resulted in better energy consumption and emissions. Optimal engine and motor use increased our EV range and further improved fuel economy and emissions. Our study reveals that our efforts have made drivability smoother and more responsive, lowered energy consumption while elongating range, and decreased emissions over previous iterations of our vehicle.

Introduction

The EcoCAR 3 competition is a collegiate automotive engineering competition organized by Argonne National Laboratory and sponsored by the United States

Department of Energy (DOE) and General Motors (GM). The competition takes place over four years, in which 16 North American universities compete to convert a 2016 Chevrolet Camaro into a hybrid electric vehicle (HEV). The teams’ aim is to maximize fuel economy without compromising vehicle performance or consumer appeal. Each team chooses its own vehicle architecture and sources then integrates the compo-nents and performs validation/verification testing. In meeting these challenges, teams compete against each other in the areas of vehicle performance, energy consumption, and overall quality and consumer appeal. The main purpose of the EcoCAR 3 competition is to train automotive engineers that will continue to drive innovation in the electrified vehicle industry.

As technology drives an increase in global awareness, the automotive industry strives to develop products that optimize performance in greenhouse gas and criteria emissions. In response to the damaging environmental impacts of

conventional automobiles, manufacturers are putting increas-ingly more HEVs on the market. [1] Hybrids use less fuel, emit less harmful pollutants, and maintain utility for most vehicle users [2]. HEV’s are already increasing in market share and further vehicle powertrain electrification is coming fast [1]. The students who work on the EcoCAR 3 program are in position to meet the increased demand for the development of high-voltage systems and testing methods as well as changes to traditional powertrain design.

Although not a new innovation, modern advances in light weighting vehicles are helping to reduce fuel consumption. Continued interest in carbon fiber reinforced polymers and other engineering composites has led to lighter, stiffer vehicle components that benefit performance and fuel economy. This is an area of the automotive industry that is poised to expand within the next 5 years [3].

For the EcoCAR 3 competition, CSU chose to build a P2 PHEV. The CSU VIT chose this architecture to minimize the need for complicated mechanical powertrains, while also allowing for high-speed all-electric driving. CSU’s design replaces the stock 3.6 L LGX V6 engine with a smaller, more efficient 2.4 L LEA I4 engine coupled with a Remy HVH250

Downloaded from SAE International by Colorado State Univ, Tuesday, October 22, 2019

COLORADO STATE UNIVERSITY ECOCAR 3 FINAL TECHNICAL REPORT 2

electric motor via a Tilton 5” carbon-carbon clutch. The motor is rigidly coupled to the input of the 8L45 8-speed automatic transmission. Because the P2 architecture allows for fully electric driving the vehicle is equipped with a 12.6 kWh LiFePO4 battery and complimented by a high voltage on-board charger in order to achieve a targeted 25 miles of EV range. The battery is managed by its manufacturer’s controller which is supervised by the vehicle’s Hybrid Supervisory Controller (HSC). The charger accepts an SAE J1772 charger and SAE Levels 1 and 2 charging powers, regulated for safe battery input. The vehicle’s belt-driven alternator was removed due to space constraints and so the vehicle is also fitted with a DC-DC converter to serve as an accessory power module, using the high voltage battery to keep the low voltage battery charged for auxiliary functions.

An overview of this architecture can be seen in Figure 1 below. All design work, simulation, analysis, mechanical/elec-trical integration and testing was executed by undergraduate and graduate student members of CSU VIT following industry best practices as prescribed by the sponsors GM and the DOE.

The target energy consumption, 0-60 mph time and other VTS were set by CSU’s VIT during the Year 1 of the competi-tion. Year 2 integrated the new components onto the vehicle; Year 3 developed baseline vehicle functionality; and Year 4 has focused on optimizing both vehicle performance and energy consumption. Key performance metrics are the vehicle 0-60 mph time and weight. Key energy consumption metrics are charge depleting energy consumption, charge sustaining fuel economy, and criteria emissions. In addition, drivability has been heavily tested and optimized throughout Year 4. Ultimately, there will always be a trade-off between vehicle drivability/performance and vehicle energy consumption;

however, the goal of all the testing was to find the optimal balance between the two in order to meet the CSU VTS.

CSU VIT’s vehicle testing methods focused on designing a series of test procedures in order to ensure vehicle function-ality and reliability as well as verify systems integration. For example, these test procedures were used to analyze vehicle effectiveness by comparing CSU’s target VTS to the actual vehicle’s performance acquired from drive tests. The drive cycles used were specifically chosen to test the above key performance metrics in a reliability-oriented iterative process. In short, the scope of CSU VIT’s vehicle testing prioritizes vehicle reliability using drive cycles designed around endurance.

Vehicle Testing Setup

Hardware, Software, and FacilitiesThe CSU VIT employs a variety of industry-standard and student-designed testing strategies for validating vehicle systems. For controls development, Years 1-2 focused on software-in-loop (SIL) and hardware-in-loop (HIL) testing to model and verify the control strategy before implementing on vehicle. Year 3 developed the vehicle-in-loop (VIL) model to achieve basic functionality at the vehicle level. A main goal for Year 4 was to improve the VIL model developed in Year 3 to achieve 99% vehicle readiness by the final competition in May. The team continued utilizing development in the SIL environment through Year 4 in order to quickly test control strategies and changes before implementing them in the vehicle. The SIL, HIL, and VIL models are all developed using Mathworks MATLAB Simulink software; the HIL model uses dSPACE real-time simulation hardware for real-time computations.

The data acquisition system of the CSU EcoCAR is based on a CAN backbone. CAN communication is managed by an on-board MotoHawk hybrid supervisory controller (HSC) that executes control code, commands components, and processes CAN signals from various components and sensors. CSU maintains 3 different CAN channels; one for the majority of stock GM components excluding the engine, another channel just for the engine and the final channel that is specifi-cally used for CSU-integrated components such as the battery and motor. These signals are recorded to an SD card using a Vector GL1000 external data logger. A Vector VN1630A Network Interface is used along with Vector CANoe software to monitor and record signals in real time. The use of two logging methods allows for redundant logging in case of failure. CANoe is also used to export data to be post-processed and analyzed. Mathworks MATLAB is used to create plots, verify VTS testing results and analyze data further.

The post processing and analysis process has been improved in Year 4 with the development of a custom MATLAB application which automatically imports data from any given log with the most common signals such as system speeds, torques and temperatures. This application also quickly generates graphs and shows useful information such

FIGURE 1 CSU PHEV Chevrolet Camaro vehicle architecture

as minimum and maximum values. In addition to stock sensors, the vehicle has multiple current clamps attached to high power drawing components within the 12V system as well as a VBOX installed that records acceleration using a GPS signal.

NX Laminate, a Siemens PLM software, was utilized in the design for composite parts. In Year 4, this software was used to design the carbon fiber hood that was fabricated and installed on the vehicle. The software allows for detailed part analysis on a ply failure basis using Tsai Wu failure criterion. The advanced software has allowed students to design the material in parallel with the part, which is a key element to creating impactful composite components. In order to verify material strength parameters, LabVIEW was used to capture testing data on two alternative matrices. The data was used to determine which matrix could be used for the final part prior to manufacturing.

Colorado State University has provided many opportuni-ties and resources for its students to gain essential hands-on experience. For the past several years the Powerhouse Energy Institute Campus has given students a safe, ample work space and many of the tools required to service, test and trouble-shoot the vehicle. Additionally, the CSU Christman Airfield has provided a 1 mile closed course runway for vehicle driving tests including all endurance drives and 0-60mph time trials. Another notable facility is the CSU Factory, a base for many materials engineers as well as the location that has provided the equipment important for manufacturing and testing composite parts and materials. On top of CSU’s facilities, the vehicle has taken part in dynamic testing at GM’s Milford Proving Grounds as well as emissions testing at the Transportation Research Center. Honda R&D in Denver, CO donated testing time on their chassis dynamometer for further emissions testing in Year 4.

Testing Methods

Initial Vehicle Movement (IVM)-60mph TestingThis test evaluates the basic maximum acceleration of the vehicle from a stationary position to 60 mph. The objective is to ensure the car can reliably accelerate from zero vehicle speed and at idle to 60 mph in under 7.9 seconds as per EcoCAR requirements. For team VTS testing the target will be 5.9 seconds. Using a pass/fail criteria, the car must be able to reliably achieve the required acceleration in under 7.9 seconds to pass.

The test is conducted at the Christman Airstrip and a VBOX is used to measure vehicle speed accurately. The testing procedure is laid out below:

1. Move car into position at start line, apply brake and shift PRNDL lever into Park until co-driver signals the beginning of the test

2. When Co-driver gives the signal, shift PRNDL lever into drive, release the brake and actuate the accelerator pedal to 100%

3. Steer the vehicle in a straight line at full acceleration until the vehicle reaches 65 mph according to the VBOX display

4. Apply brake until car reaches maneuvering speed and return to start line

5. Analyze time and vehicle speed results logged via external logger and VBOX

Drivability TestingPowertrain Vibration Observations This test’s purpose is to help determine which driveline component was the source of a powertrain vibration issue. The powertrain was operated in different states to determine possible sources of noise, vibration, and harshness (NVH). The states tested are listed below:

1. Motor on, engine off, clutch disengaged 2. Motor on, engine off, clutch engaged 3. Motor on, engine on, clutch engaged 4. Motor off, engine on, clutch disengaged

The primary method of evaluation was binary, whether the metallic rattle was clearly present or not while idling unloaded and in park. This evaluation method allowed the team to identify the component or interface associated with the noise (and eliminate others).

Powertrain Alignment and Concentricity Checks The purpose of this test was to check for any misalignment or non-concentricity that could cause the powertrain vibration. The procedure for checking the align-ment and concentricity of the driveline is outlined below:

1. Disassemble the powertrain to expose the clutch assembly mounted to the flywheel; remove the clutch assembly to expose the flywheel; remove spark plugs to make manual cranking of the engine easier

2. Fix a magnetic dial indicator to a square steel beam and clamp the beam to the engine carriage such that the dial indicator runs along the rotating face of the flywheel that is parallel to the engine along the traditional y-axis to test for runout along the face of the engine at the engine-flywheel interface

3. Using a ratchet and socket, one tester should crank the engine in the positive direction while a second tester observes the dial indicator

4. The component shall be considered sufficiently concentric and aligned when the second tester measures a change of no more than ±0.002” during rotation

5. Rotating components outside of this specification should be adjusted

6. In order to make adjustments, the tester should loosen the respective mounting hardware in a star-shaped pattern and no more than a half-turn at a time; use a soft blow mallet on the orthogonal face to tune the location of the component; repeat steps 1-4 after each adjustment as well as after re-torquing the mounting hardware to specification

7. Repeat steps 1-4 on each component as the powertrain is reassembled (clutch assembly, inertia ring, electric motor throughshaft), repeat steps 1-4 on all applicable rotating faces (orthogonal face of the clutch, both exposed faces of inertia ring)

8. Additionally, if the electric motor shaft is suspected to be bent, it should be removed and placed in a lathe to be measured using a dial indicator along a non-splined portion of its longitudinal axis

These checks should be executed any time the vehicle’s powertrain is removed. Successfully completing these checks allows the team to eliminate alignment problem as a potential root cause of the NVH issue.

Charge Depleting Shift Calibration The purpose of this test was to calibrate shifts while in electric vehicle (EV) mode until they were acceptable and wouldn’t inhibit driv-ability. The procedure for calibrating the inertia compensation strategy for EV mode is outlined below:

1. With the vehicle in park, depress the brake pedal and shift the PRNDL lever into drive

2. Drive with light pedal application (<10%) and calibrate the inertia value using MotoTune until driver and passenger agree that shifting feels adequate

3. Maneuver the vehicle back to its starting position, stop and shift the PRNDL lever into park

4. Repeat steps 1-3 with medium pedal application (>10% and <50%) and high pedal application (>50%)

5. With calibrations at their expected final value, repeat steps 1-4 one more time to ensure that changes at each pedal application did not affect previous results

6. Review CAN data logs to verify shifting performance; analyze measured engine, turbine and transmission output speed traces.

This procedure was first done at the Powerhouse Energy Campus at limited vehicle speeds. It was also performed at the Christman Airfield where higher speed shifting could be calibrated. The evaluation of this test is subjective, however, the team’s GM mentor, Anthony Heap, helped verify the results of this calibration exercise based on his own experience driving the vehicle and looking at the resulting data.

Engine Bump Start Smoothing Calibration The purpose of this test was to calibrate engine bump start logic while the vehicle was in motion until engine bump starts did not adversely affect drivability. In order to counteract the expected vehicle deceleration during bump start events the motor was calibrated to produce a brief torque boost throughout the event. The procedure for calibrating the bump start smoothing logic is outlined below:

1. With the vehicle in park, depress the brake pedal and shift the PRNDL lever into drive

2. Drive with light pedal application (<10%) and calibrate the bump start torque boost value using MotoTune until bump starts feel smooth.

3. Maneuver the vehicle back to its starting position, stop and shift the PRNDL lever into park

4. Repeat steps 1-3 with medium pedal application (>10% and <50%) and high pedal application (>50%)

5. With calibrations at their expected final value, repeat steps 1-4 one more time to ensure that changes at each pedal application did not affect previous results

6. Review CAN data logs to verify bump start performance; measure length of each bump start and transmission output speeds to evaluate performance (this process is described as a separate test below)

This calibration routine was performed primarily at the Powerhouse Energy Campus at low speeds. High speed testing was also done at the Christman Airstrip. The evaluation method for this test is also subjective although further data analysis allowed the team to define quantifiable evaluation criteria, defined below.

Engine Bump Start Smoothing Testing The efficacy of the bump start smoothing calibration described above was tested by analyzing vehicle CAN logs in MATLAB. Data sets of 23 bump start events before smoothing and 19 bump start events after smoothing were analyzed.

1. Each bump start event was defined as the period from 200 ms before initial engine movement until the engine and motor speeds were within 30 rpm of one another

2. The difference in transmission output shaft speed from the beginning of the bump start event to the end was measured

3. This difference was normalized by dividing it by the average transmission output shaft speed over the interval

4. The average of all of the resulting normalized transmission output shaft speed differences was taken for each condition, before and after calibration

5. The bump start calibration shall be considered sufficiently smooth when the change in average normalized transmission output shaft speed is 0.5% or less

Charge Depleting Energy Consumption TestingThe purpose of this test is to evaluate EV energy consumption while operating in charge depleting (CD) mode. The CSU VTS targets 100 mpgge. There is no competition requirement value for this test.

This test was performed at Honda R&D on their chassis dynamometer. The HWYFET drive cycle was used for the test. The test procedure is outlined below:

1. Align vehicle on chassis dynamometer 2. Secure vehicle on chassis dynamometer 3. Verify that starting state of charge is above 80%. 4. Verify that ABS fuse is pulled so that an error doesn’t

occur during testing 5. Verify external data logger is functioning properly

before starting the vehicle for the test

6. Allow the trained driver to start the vehicle, shift the PRNDL lever into drive and follow the prescribed drive trace

7. Monitor the battery state of charge as reported by the BCM

8. Allow the trained driver to bring the vehicle’s wheels to a halt and shutdown before approaching the vehicle after the drive trace is completed

9. Extract the logged data from the external data logger, export data from CANoe to MATLAB

10. Use MATLAB to calculate EV fuel economy in units of mpgge

The objective of this test is to understand the AC electrical consumption over a standard drive cycle in order to minimize the CD energy consumption. This test does not include pass/fail criteria.

Charge Sustaining Fuel Consumption TestingThe purpose of this test is to evaluate the fuel consumption while operating in charge sustaining (CS) mode as well as to analyze thermal conditions in the engine bay during CS opera-tion. The objective is to achieve 35 mpgge to meet the CSU VTS target and to ensure that no critical components overheat during normal operation. The fuel consumption will be continuously tested and improved; the test will be consid-ered a failure if the engine, electric motor (EM) or Energy Storage System (ESS) reach a maximum temperature greater than their rated specification at any point during the test or if a component fails during the test, causing premature termi-nation of the test.

The test is performed on a chassis dynamometer. The Transportation Research Center and Honda R&D have been used by the team to perform this test. The EcoCAR 3 E&EC drive cycle is run three times, back to back, in order to evaluate the CS fuel consumption. The test procedure is outlined below:

1. Align vehicle on chassis dynamometer 2. Secure vehicle on chassis dynamometer 3. Verify that the vehicle will start into CS mode 4. Verify that the ABS fuse is pulled so that an error

doesn’t occur during testing 5. Verify external data logger is functioning properly

before starting the vehicle for the test 6. Allow the trained driver to start the vehicle, shift the

PRNDL lever into drive and follow the prescribed drive trace

7. Extract the logged data from the external data logger; export data from CANoe to MATLAB

8. Use vehicle data to calculate average fuel economy in units of mpgge; compare the maximum recorded value of the engine coolant, inverter core, electric motor and ESS cells to their respective maximum specifications

Criteria Emissions TestingThe purpose of this test is to evaluate the utility factor weighted tailpipe criteria emissions. The objective is to meet federal regulations for tailpipe emissions and to minimize criteria emissions.

The test is performed on a chassis dynamometer. The Transportation Research Center and Honda R&D have been used by the team to perform this test. The EcoCAR 3 E&EC drive cycle is run three times, back to back. Modal and bag emissions are collected and then processed. The test procedure is outlined below:

1. Align vehicle on chassis dynamometer 2. Secure vehicle on chassis dynamometer 3. Verify that the facility’s emissions measurement

system is appropriately calibrated and functional 4. Verify that the vehicle will start into CS mode 5. Verify that the ABS fuse is pulled so that an error

doesn’t occur during testing 6. Verify external data logger is functioning properly

before starting the vehicle for the test 7. Allow the trained driver to start the vehicle, shift the

PRNDL lever into drive and follow the prescribed drive trace

8. Export the emissions data as a tabulated set that can be further processed in Excel or MATLAB

9. Use given WTW data and measured emissions to calculate emissions score per EcoCAR 3 Emissions Testing Event analysis procedures

It should be noted that it is not a requirement to begin this test in CD (full EV) mode since the range is already known and it can reasonably be assumed that there are no tailpipe emissions in CD mode.

Results

Initial Vehicle Movement (IVM)-60mph TestingThroughout Year 3, the team struggled with transmission shifting because of an intermittent issue causing critical messages to the Transmission Control Module (TCM) to be dropped. The team was not able to solve the issue before the end of Year 3 and thus started Year 4 stuck in 4th gear. This issue severely affected 0-60 mph time. The 0-60 mph time at the beginning of Year 4 was 22.4 seconds. This does not meet the passing criteria for this test; as such improving the transmission shifting was a priority in Year 4.

The TCM limp mode can be identified by the existence a U0100 trouble code reported by the TCM and observed through the vehicle’s OBDII port. This trouble code indicates lost communications with the Engine Control Module (ECM) and forces the vehicle into a “limp home” mode where the transmission will remain in 4th gear. Many steps were taken to attempt to get out of this limp mode including altering

message send rates from the HSC, adjusting the terminating resistances on the CAN bus and altering the supervisory CAN message pass-through logic. Unfortunately, none of this changed the results; the TCM was still not getting the correct number of messages. Eventually it was discovered that messages were being dropped more frequently than expected due to the team’s definition of the CAN bus transmit and receive queues. Once the message queue sizes were changed to be appropriate for the application, the transmission was shifting and the TCM’s U0100 trouble code was resolved.

Initially the engine was allowed to control shifting torque cuts without interference from the HSC. However, it was known that this was not the best strategy since the inertia of the new powertrain is different from that in the stock vehicle powertrain. The shifts took longer than desired with this strategy; nevertheless, it was important to verify that shifting would help decrease the 0-60 mph time to be acceptable. The 0-60 mph time with the engine taking care of shifting torque cuts was reduced to 8.5 seconds. This was a significant improvement; however, it was still above the maximum time of 7.9 seconds set by the EcoCAR 3 competition rules. This was mainly due to the shifts taking too long as well as a mechanical issue with misfire in the engine’s first cylinder.

The next step in improving 0-60 mph time was to take over the shifting torque cuts with the HSC. In order to do this, the HSC listened to the transmission torque request message during a shift, applied an inertia compensation factor as calcu-lated using Equation 1, and commanded the torque cut through the torque split strategy.

stock enginestock=

Once this strategy was implemented, the inertia compen-sation factor had to be calibrated. This was done by calibrating until the shifts felt smooth and looking at specific CAN signals to understand how it could be tuned to be even better. These changes reduced the 0-60mph time to 7.6 seconds, which meets EcoCAR 3 competition requirements but is not quite as quick as the CSU VTS. At this point, the engine still had the misfire issue, so it can be seen that taking over the shifting torque cut reduced the 0-60mph time by 10.6%.

All three times can be seen compared in Figure 2. It is important to note that more tuning and the new engine that

is not misfiring are expected to help reduce the 0-60mph time before final competition in May.

DrivabilityPowertrain Vibration After the new powertrain was integrated in the car during Year 2, there was significant NVH observed in the system, especially in the form of an audible metallic rattle near the front of the vehicle. During Year 3, this issue was researched further when the powertrain failed at the clutch interface and was disassembled. The issue was incorrectly determined to be caused by lateral motion of the driveline. A damping material was introduced at the interface of the motor shaft and torque converter and the electric motor shaft was further constrained laterally using a steel collar; neither solution was successful in eliminating the noise.

The team identified developing a solution to the NVH issue as a priority in Year 4. The solution to this problem was necessary to improve vehicle reliability as well as to meet the criteria for the consumer appeal event of the competition. The initial research into the root cause of the NVH problem was focused around testing described in the Methods section of this report. From the observations made, it was determined that torque fluctuations from the engine were causing NVH, but it was unclear whether NVH was coming from the clutch or from the transmission. In order to verify that the noise was not due to misalignment or non-concentric shafts, concen-tricity and misalignment were measured during each powertrain reassembly according to the described procedure. There were no resulting changes observed in the problem or in powertrain performance and so alignment was eliminated as a possible root cause.

After eliminating alignment as a root cause the team looked into the frequency response of the powertrain. Rough vibration data was collected using accelerometers on the engine, clutch housing and motor housing in addition to a professional grade microphone placed roughly 8” directly below the torque converter. This data was collected with the clutch coupled, engine producing torque, motor producing no torque, and idling around 750 rpm. The acoustic data shown in Figure 3 proved to be most useful in implementing a solution.

FIGURE 3 Audio analysis results of powertrain NVH problem

FIGURE 2 IVM-60 mph acceleration testing progress and results

The issue is most prevalent at lower frequencies which agrees with the observation of the issue presenting more aggressively when the vehicle is idling. Overall, two options were considered to mitigate the NVH. One solution was to increase the inertia of the driveline by engineering a custom high-inertia flywheel, thus reducing its natural frequency and shifting the system’s frequency response further away from its excitation at idle. The second option was to increase the idle speed of the engine and motor. Both of these solutions have been implemented and the NVH has been significantly reduced, although not entirely eliminated. The team is currently working further with the clutch manufacturer to reduce backlash, and eliminate any remaining NVH.

Charge Depleting Shifting An issue encountered early on was that the HSC would not listen to shift commands from the transmission while in CD mode and the vehicle would shift without any modification to the torque commanded of the motor. This caused the shifts to take a long time to complete and jerk the vehicle. To mitigate this an inertia compensation algorithm was developed. During shift events in CD mode, the HSC reads the torque commanded by the transmission, applies an inertia compensation factor to that value, and commands the motor appropriately. Equation 2, below, shows how the inertia compensation factor was used.

JTmotor

stock enginestock=

Because the stock engine inertia was not known, testing was required in order to calibrate the stock engine inertia value and get high quality shifts. In addition to this compensa-tion logic, CAN signals sent to the transmission, event-based logic and controller structure reorganization were all required. Figure 4 shows the shifting events from before and after these changes were made. The top graph shows that transmission flaring was occurring, the driver had to adjust the pedal in order to finish a shift, shifting events were taking a long time, and that the transmission output speed was not very smooth. This is not desirable for a vehicle as the driver would be unhappy to have long and lurching shifts. The lower graph shows that the torque cuts eliminated flaring, shifts were much quicker, and transmission output speed was smooth instead of jagged.

Engine Bump Start Smoothing To achieve optimal CS fuel economy, a strategy of cycling the engine on and off was developed. The architecture of the vehicle does not include a starter motor; therefore, the engine is “bump started” by closing the clutch between the motor and engine while the motor is turning. The clutch on the vehicle is a carbon racing clutch actuated by a throw-out bearing pressurized by an electric linear actuator. The clutch closure event is brief and the torque across the clutch is significant. The torque produced forward of the transmission is decreased and so is the trans-mission input shaft speed. As a consequence, the transmission output shaft speed is briefly reduced, and the driver feels a slight but noticeable vehicle deceleration.

To mitigate this undesirable effect on drivability, the commanded motor torque was increased for the duration of the bump start in order to compensate for the negative torque

applied by the engine inertia during startup. Equation 3, below, shows the equation used to get an estimate for initial bump start torque increase needed. Data from bump starts was analyzed in order to get an initial boost torque estimate. This initial value was then calibrated further using the proce-dure discussed above. The scaling factor is a constant.

T Jboost engine clutch engine clutch= *µ+ + (3)

The efficacy of applying the calibrated torque boost was evaluated by comparing the average percent change in trans-mission output shaft speed for samples of bump starts from before and after the change was implemented. The results are shown in Table 1.

The above data shows that the drop in transmission output shaft speed was significant before implementing the algorithm. After smoothing, the average change is nearly zero; this is translated to the driver as an almost non-existent inter-ruption when the engine is bump started while the vehicle is in motion. This analysis validates the subjective assessments

FIGURE 4 Visualization of shifting events before and after inertia compensation logic was implemented

TABLE 1 Effects of bump start smoothing control algorithm

Condition SamplesAverage Change in Transmission Output Speed

Before Smoothing 23 -2.65%

of team members who calibrated the bump starts until they felt acceptable.

Mass ReductionThe importance of this is to determine that the curb mass of the vehicle throughout development remains below 1938 kg (4273 lb). This requirement was set as a rule by the EcoCAR 3 Non-Year Specific Rules. The current curb mass as of March 29th, 2018 was 1800 kg, well below the EcoCAR 3 rule limit. The lower curb weight was in part due to the implementation of engineered composite parts in place of stock components, which saved 12.64 kg. This is a 0.7% weight reduction from the curb weight of 1812.64 kg without the use of composites. Figure 5, below, shows compares the weight reduction due to each composite component that has been implemented in the vehicle.

Charge Depleting Mode Energy ConsumptionLow Voltage Electrical Consumption In order to minimize CD energy consumption and maximum CD range many controls strategies were considered. A particular area of focus was the low voltage (LV) electrical system because the 12V battery is charged using the ESS and a DC/DC converter rather than by an alternator, as found in a traditional vehicle. In order to analyze the LV system, four current clamps were placed on what was thought to be the highest drawing electrically powered components that were integrated by the team (ie not stock parts on the 2016 Camaro) as well as one on the main 12V line. The chosen components to attach current clamps to were the electric motor cooling pump, inverter cooling pump, and rear radiator fan. These compo-nents are part of the student design cooling system. The data taken from testing at Honda R&D was used to produce Figure 6, which shows the energy consumption breakdown of the LV system.

This breakdown shows that the pumps are the highest consuming components consuming 31% of the entire low voltage system’s power consumption. The “other” section that

is 65.38% includes stock components such as headlights, interior lighting and power to various stock controllers, sensors and actuators that are not modified in the vehicle. The rear radiator fan consumed 3.37% of the power when operating in CD mode.

When the vehicle operated in CS mode, the front radiator fan (not shown in Figure 6) consumes on average 3.5% of the power. This component is not powered when in CD mode, however measuring its consumption in CS mode is important in order to predict its effect on the ESS during CS operation. Because the inverter has a built in heat exchanger, sending a duty cycle or cutting back on the power to the pump could cause harm to the inverter, as well as harm the electric motor if the same was done to its pump. The rear radiator fan control was chosen to be altered as a way to save some energy while driving. This was possible because data showed that the inverter and motor were not near their respective temperature limits due to the fan kicking on. This new control strategy has not been tested yet, however, it is expected that this will help increase CD mode range.

Data collected at Honda R&D indicated that the maximum operating temperatures during CD operation of the inverter and motor were 72.4C and 94C, respectively. These temperatures are comfortably under their specified maximums; as a result, a control strategy was created to reduce the on-time of the rear radiator fan in order to raise the maximum temperature by 20C each (still maintaining safe operating conditions). The expected result is an overall low voltage energy consumption savings. Table 2 shows that before implementing this control strategy the rear radiator fan consumed on average 21.7 Watts of power over a HWFET drive cycle. When the actuation temperature for the rear radiator fan was increased, simulations have shown that the rear radiator fan would only turn on in the last thirty seconds of a HWFET drive cycle and consume 1 watt of power over the drive cycle. This is a 7.6KJ savings of energy when running a HWFET. Further simulations in MATLAB based on the on-time for the rear radiator fan estimate for city & highway driving (on an average drive of 15 minutes) estimate a savings of 2.96% overall low voltage consumption. It’s important to note these savings are simulated and are planned on being tested and verified at Honda R&D later this month.

FIGURE 5 Weight Reduction by Individual Composite Components

FIGURE 6 Energy consumption of low voltage system components (Total power consumption)

Charge Depleting Shift Schedule The 8L45 transmis-sion that is on the vehicle has a shift pattern that keeps the engine operating around 1500 rpm when driving on drive cycles. This speed is acceptable for engines since it will operate the engine at a more efficient state; however, the motor is not operating as efficiently as it could when at 1500 rpm. The motor has a much larger efficiency band between 2000 and 4000 rpm, as can be seen in Figure 7 below.

With the transmission keeping the motor speed around 1500 rpm, the charge depleting fuel economy was determined to be 93.5 mpgge. In order to force the transmission to shift at higher input shaft speeds, a control strategy with the accel-erator pedal signal sent to the transmission artificially increased by 25%, was developed for EV mode. This has yet to be tested, however simulations indicate that the increased motor efficiency will allow the CD fuel economy to meet the CSU VTS target of 100 mpgge.

Charge Sustaining Fuel ConsumptionDuring Year 3, the vehicle was operated only under the propulsive power of the engine. Subsequently, the fuel economy of the vehicle was not as high as desired; however, this data serves as a good baseline for future work. The fuel economy calculated from CAN data collected at competition in Year 3 was approximately 8.3 mpgge. This number is not calculated from drive cycles, but rather from driving that was done at low speeds and aggressive pedal application. It is not a comprehensive representation of vehicle operation but does

indicate that the vehicle control strategy was not sufficiently meeting the team’s VTS.

At the beginning of Year 4, a control algorithm was devel-oped for CS mode, which included a complete controller redesign. There were three charge sustaining control strategies proposed and tested: a load following strategy, an engine optimal strategy, and what is called an optimal path controller. The optimal path controller switches between load following and engine optimal operation whenever it is more efficient to operate the entire vehicle in the respective mode. The optimi-zation code for this controller uses both engine and motor data to determine when it is more efficient to run in a load following versus engine optimal routine. SIL simulations showed that the optimal path controller was more efficient across numerous drive cycles, and thus was chosen as the controller to be implemented in the vehicle.

A map of the engine torque as a function of desired powertrain torque and instantaneous speed is illustrated in Figure 8. This figure shows that in general, when the desired powertrain torque is above about 120Nm, the engine operates along its ideal operating line, however, below that, it is controlled through a load following algorithm. This map was created based on generic engine data available to the team; the map that is integrated in the vehicle controller has been calibrated to account for unavailable parameters. The optimal path controller follows a set of rules that approximate the optimization code created the map below. This was done so that it is easy to calibrate and so that it does not rely on accurate engine data, as which has not been available. In addition, the optimal path controller turns the engine on and off when it is more efficient to use the electric propulsion system, such as at low speed or low requested torque values.

The Emissions Testing Event (ETE) served as the first time the vehicle has been tested for fuel economy after the

TABLE 2 Low voltage power savings simulated on a HWFET cycle

Component

Initial Average Power (W)

Final Average Power (W)

Δ Power (W) % Change

Rear radiator fan 21.7 1.03 20.7 95.4%

Total Power Usage

643 623 20.7 3.11%

FIGURE 7 HVH250-115 efficiency plot [4]

FIGURE 8 Optimal engine torque map as a function of powertrain torque and speed

optimal path algorithm had been implemented. The vehicle achieved an average of 29.67 mpgge in CS operation over the course of three E&EC drive cycles. The testing revealed that the charge sustaining algorithm was not charge neutral, which is not desired. This can is demonstrated in Figure 9. This figure shows the first of three drive cycles done in charge sustaining mode. It is evident that there was a significant depletion of the battery state of charge over this drive cycle. The state of charge was expected to wander slightly with the control imple-mented, however, it was not expected to end at a much lower state of charge, as this figure shows.

The first two cycles were driven with engine on/off func-tioning but the engine was on all the time for the third cycle, as the state of charge had decreased too much. During the first cycle, the vehicle achieved 38.2 mpgge, 34.1mpgge over the second drive cycle, and 22.7 mpgge over the third drive cycle. This shows that engine on/off functionality impacts fuel economy significantly, as the fuel economy decreased by 33.5% when the engine was forced on at all times. Operating more charge neutral will allow the engine on/off functionality to last longer, with the idea that that would decrease overall fuel consumption.

Again, simulations in SIL were used to adjust the control strategy in order to make the control strategy more charge neutral and to optimize fuel economy. The control algorithm was adapted to provide more energy to the low voltage system, provide more aggressive regenerative braking, and turn the engine off at lower speeds so that the motor did not have to propel the vehicle as much, thus reducing the amount of power needed from the battery. While this strategy uses the engine more, it will still allow for high fuel economy and allows for the system to be charge neutral instead of slightly charge depleting. This was done in hopes that engine on/off would never be disabled and that fuel economy would overall increase.

Overall, the strategy has been implemented in vehicle and tested at Christman Airfield but has yet to be tested at Honda R&D, however, that will occur before competition in May. It is expected, from testing at the airstrip, that the new strategy

should be more charge neutral and should result in higher fuel economy on the E&EC drive cycle.

Criteria EmissionsIn order to first determine some base level emissions data, the vehicle was tested at the Emissions Testing Event during Year 4 of the competition. The results from the event summarized in Table 3. Overall, the vehicle performed well, but was high in CO emissions.

After the Emissions Testing Event, it was evident that a form of emissions control would need to be implemented in order to help reduce CO emissions and to adhere to federal standards. The developed strategy uses energy based logic to warm up the engine. There are three phases to the control strategy. During phase 1, the engine is limited to 30 Nm of torque until it has expelled 0.5MJ of propulsive energy. After the engine has reach 0.5MJ of energy, phase 2 begins, where the hybrid supervisory controller (HSC) employs a load following strategy and the engine is limited to a maximum torque that is based on the energy produced by the engine. The maximum torque varies linearly as a function of energy until the engine has produced 3MJ. Once the engine has reached 3MJ of energy, it is fully warm, which is phase 3, and the HSC uses the optimal path control algorithm discussed previously to split torque between the engine and motor. Inspiration for this warm-up strategy was taken from the theory tested in the Smith et al. paper [5] and in the Walsh and Nelson paper [6]. This warm-up strategy was first tested in SIL for logic validity and then implemented in vehicle. The strategy operation can be seen below in Figure 10. The first

FIGURE 9 Battery SOC during the EcoCAR 3 E&EC drive cycle

TABLE 3 Utility factor weighted emission results after vehicle testing at TRC

UF-Weighted Emission ValueCO (g/mi) 13.9

THC (mg/mi) 4.08

FIGURE 10 SIL demonstration of emissions control algorithm

phase operates from 0 seconds to about 180 seconds. The second phase lasts from 180 seconds to about 475 seconds. The final phase starts at 475 seconds and lasts until the end of the drive cycle.

This emissions control strategy has been implemented on vehicle and will be tested at Honda R&D before going to competition in order to validate its performance. It is expected that the strategy will follow similar results found in the Smith et al. paper [5] and in the Walsh and Nelson paper [6].

Innovation Topics

Year 1CO2 Reduction Strategy Year 1 Innovation team focused on reducing CO2 output from the vehicle to meet future US Environmental Protection Agency (EPA) require-ments. The EPA’s goal is to reduce car emissions CO2 level by 143 grams/mile or an overall reduction of 54% by 2025. The design and science behind this was to introduce different materials between the engine and ambient air to create this reduction. The team researched possible designs but one seemed to be promising in this specific application which was the Series Bosch Reactor designed by NASA for purifying respiratory CO2 manned missions. Using this design, the team simplified it down to a single pass, single chamber and dual reactor with an introduction of Hydrogen to reverse water gas shift and carbon formation.

Testing this theory, a 950cc motor was used with two separate reactors to mitigate any variables except for RPM and hydrogen flow rate. This test revealed many truths about this type of technology in which it did reduce the CO2 emissions only by 3.88% which was far from the goal of 50%. The factors that played a critical role in this situation was: there was very little insulation on the reactor so for very high temperatures the reverse water gas shift would create CO but would also create methane that in turn would still need to be converted back to CO2. This would also reduce the overall reduction as well.

Year 2Advanced Materials This team designed a composite constant velocity axle system incorporated to avoid problems due to engine and configuration changes. The carbon fiber shafts were manufactured via filament winding and were bonded with epoxy to the stock metal splined ends. Full scale testing of the shafts showed they were able to withstand up to 1000 ft-lb of torque without being damaged: enough to handle the increased strength requirements of the engine.

Keysight APM To better understand the electric power conversion system within the hybrid electric car, the EcoCAR Keysight sub-team was formed. The goal was to create a test system to characterize power conversion to better understand the performance, control and integration of two main power components, the TDK-Lambda Auxiliary Power Module (APM) and the Rinehart Motion Systems Electric Motor Inverter.

Year 3Hot Temperature Engine This year’s EcoCAR 3 Innovation team project hypothesized increased engine coolant temperature correlates with increased vehicle fuel economy. The engine coolant was raised from the conven-tional 90°C to an elevated 120°C steady-state fluid tempera-ture. The engine, LEA Ecotec 2.4L, combusted E85 fuel, and the engine was loaded using a Land & Sea water-brake dyna-mometer. Measurements of the output work and fuel efficiency were conducted at regulated coolant temperatures of 90°C, 100°C, 110°C, and 120°C. At each coolant temperature the engine was dynamically tested at specific engine speed and torque values corresponding to estimated driving velocities of 25, 35, 40, 55, and 65 MPH.

Calculations on data collected from an engine dynamom-eter determined the fuel economy of the vehicle will improve by 9.8% as a result of high-temperature coolant operation. Aerodynamic drag at a specific wind velocity will decrease proportionally to decreases in the coefficient of drag (shape) and frontal area (size) of the vehicle facing wind flow. Elevating coolant temperature increases the amount of heat that can be transferred by the radiator; so, increasing the coolant temperature increases the heat transfer rate per radiator area efficiency of the radiator.

Brake specific fuel consumption (BSFC) was expected to decrease as a result of high-temperature coolant opera-tion however test data decisively concluded that the BSFC of the engine increased with increased coolant tempera-tures. Fuel efficiency benefits peaked at 13.2% for a vehicle speed of 40 mph and the beneficial effects dwindled signifi-cantly at vehicle speeds greater than 55 mph. The data trends suggest that the high-temperature modifications improve engine performance at city speeds, but the effects become negligible at highway speeds. Testing results are summa-rized in Table 4.

Although the team could not include WHR systems into the scope of the project, the innovation research can surely be expanded by future research teams to WHR systems. Thermoelectric generator (TEG) systems are scalable and operate well in the coolant temperature range by converting

TABLE 4 BSFC values at varying engine coolant temperatures

Test Velocity [MPH]

Coolant Temp. [°C] BSFC [g/kWhr]

Percent Change from 90 °C

25 90 775 -

120 705 -9.08

35 90 620 -

120 541 -12.76

40 90 553 -

120 480 -13.16

55 90 407 -

120 384 -5.73

65 90 382 -

120 375 -1.72

Average 90 547 -

the heat from the engine coolant directly into electricity simi-larly to photovoltaics. Increased engine coolant temperatures allow for greater electric generation, and the energy generated could be used to power auxiliary systems without the need for mechanical power losses to an alternator.

Keysight Electric Motor Dynamometer This year’s EcoCAR3 Keysight team are related to electric motor effi-ciency and improvements in high voltage testing hardware. The project sponsor, Keysight Technologies, is looking to expand its role in the automotive industry as a supplier of electronic testing components and solutions. As such, in collaboration with Colorado State University’s EcoCAR program, the EcoCAR3 Keysight team was tasked with creating an electric motor dynamometer which will work in collaboration with Keysight Technologies’ existing high-voltage testing equipment. Repurposing parts from a dyna-mometer previously designed by a CSU grad student, the team has designed a new dynamometer capable of testing a Remy HVH250 high-voltage electric motor, the same motor that is used in the EcoCAR3 Camaro.

Advanced Materials, Composite Anti-Roll Bar The Anti-Roll bar consists of three pieces: the fiberglass-carbon fiber composite shaft, and two 7075 Aluminum arms bolted to the bonded Aluminum inserts. The Anti-Roll Bar laminate and bonded aluminum inserts were verified through the testing a sample composite shaft given a torque of 1000 in-lbs and 8 degrees of rotation. Data showed a linear torque vs displacement curve, validating the laminate, and the bonded inserts held throughout the test. The full-scale Anti-Roll Bar was tested by applying a load to one arm, while keeping the other arm stationary through use of a test rig. A load cell was used to record the force applied to the arm and a linear poten-tiometer was used to calculate displacement. Stiffness was calculated by dividing force by distance.

Advanced Materials, Carbon Fiber Muffler The muffler was manufactured using a carbon fiber shell with stainless-steel end caps. Pyrogel XT-E insulation was utilized within the shell in order to prevent the composite from failing. Noise, temperature, and leak testing was performed to ensure that the decibel reading fell within the range of 60-90dB, and that it could hold 1 psi without leaking. Thermal performance was predicted using Fluent for both the end caps and the Pyrogel insulation and was verified on vehicle using thermal cameras. A predicted attenuation profile for the sound was calculated from the proposed expansion chamber design, with the goal of attenuating 10dB across all engine frequencies from stock. On vehicle testing with a microphone was performed to measure the frequency spectrum of the muffler. Some frequencies were attenuated as designed while others did not. The composite muffler assembly reduced the weight to 19lbs from the stock 40lbs.

ADAS, Development of a PC and Embedded System As a part of the EcoCAR 3 competition the ADAS team developed a PC ADAS system that included vehicle, pedestrian and sign identification as well as S32V lane iden-tification. The May 2017 year-end competition specified

certain requirements to be tested during a live driving demon-stration driven on a closed course complete with signs and dummy vehicles. Accomplishing this task meant combining lane, vehicle, pedestrian, and sign detection into a single unified application capable of doing each of these things in real-time. The competition specified a minimum of 5 frames per second. To accomplish these tasks the following steps were completed.

The PC system included an MSI laptop and a custom built stereo vision camera. To mount the camera on the windshield a high-grade suction cup was purchased. A mount was also 3D printed to fit on that suction cup. Calibration was then applied to the stereo vison camera to perform distance estimation.

The provided MATLAB ground truth labeler app was used to create ground truth data for the recorded footage. There were 10 labeled sessions containing the path to the footage that is labelled. These labelling sessions contain the required vehicle bounding boxes, along with at minimum two stop signs, two speed limit signs, and two pedestrians labelled. This ground truth information was used to determine the accuracy of object classifiers developed by comparing the bounding box locations of the objects.

A vehicle classifier was trained using a provided MATLAB function and a custom collection of input data. Positive and negative vehicle images were gathered from two major sources; publicly available online datasets and data gathered during live driving in Fort Collins, CO. The Stanford Cars Dataset provided the team with 16,185 positive vehicle images in a wide variety of orientations complete with annotated ground truth information. It was determined early on that the clas-sifier should focus on vehicle rears since classifiers trained on mixed vehicle orientat ions produced seemingly random classifications.

A Convoluted Neural Net (CNN) classifier called YOLO9000 was also used which utilizes Dark Flow and tensor flow to support the Nvidia GPU. This classified multiple objects into one detector with a given confidence score.

Bounding boxes are colored green by default and yellow if the vehicle is determined to be in the same lane as the ADAS vehicle. Boxes are colored red if the vehicle is both in the same lane and less than 25m ahead. Generally, a vehicle can be determined to be in the lane using the self-designed method described in Figure 11, below. Vehicles have been bound and labeled with tracking numbers using a custom tracking algorithm based on the centroid of the picture.

For lane detection, serval methods were used to detect the lines in the road. The first step was to remove the portions of the video that do not show the road. The next step was to prepare each image for line detection. The peaks of the lane lines were then found using the Hough peak detection method. For depth estimation semi-global block-matching was used and a 9x9 pixel average of the centroid generated by depth. These steps were combined into a single system using a python application with 15 fps.

The S32V234 board is the embedded system to be inte-grated into the vehicle. The goal for this embedded system is to implement the described functions in the vehicle in real time.

Year 4Vehicle to Grid (V2G) The Year 4 Innovation project focused on bringing bi-directional charging infrastructure, DC fast charging, and 120V power options to the 2016 Chevy Camaro. Implementation of V2G capabilities requires a bi-directional charger, a reworked high voltage junction box, redesigned high and low electrical, and a thermal management system. The function of the bi-directional charger is to bring bi-directional power flow to the vehicle and the ability to both Level 2 and DC fast charge. The reworked electrical, high voltage junction box and thermal management system are vital components for the new charger integration to function successfully.

At the heart of the design is the Current Ways Bi-directional charger. This charger was selected for its bi-directional capabilities. The charger can charge up to 6.6kW at 240V AC and has a 300-450V DC output. The charger will work with the current system because it commu-nicates with CAN messages. The implementation of this charger requires a rework of the high voltage wiring in the vehicle.

A large portion of the project consisted of the ability to control the bi-directional power transmission to and from the vehicle’s onboard 12.6kWh Energy Storage System (ESS). To charge the vehicle faster, the team’s objective also included incorporating a CCS charge port with the ability to DC fast charge. These DC fast charging lines would be controlled separately from the charger.

The team was also tasked with packaging the improved charging system for Y4. The team successfully installed the CCS Combo charge port and 120V outlet. The team awaits installation of the Current Ways bi-directional charger, a high voltage junction box, a contactor housing as of April 2018. Future work includes improving control strategy that main-tains safe charging and discharging rates while meeting functional requirements.

Advanced Materials, Carbon Fiber Hood A carbon fiber hood was developed to decrease the weight of the vehicle and increase functionality. A. hood scoop was incorporated

to increase the airflow through the engine bay to assist in cooling. The layup was 0-90 woven carbon fiber / +-45 woven carbon fiber / Aeromat Core / +-45 woven carbon fiber / 0-90 woven carbon fiber with Kevlar placed in key areas to prevent part fracture in case of collision. Hard points for stock mounting locations were designed using Garolite and balsa wood. A break in the Aeromat core was placed in front of the hood scoop to provide a fracture zone. Tensile testing of small coupons was done to determine the ideal layup. A pendulum impact tester was used to physically test coupons to verify the use of Kevlar and verify the purpose of the break in Aeromat. A mold was made for the hood using the stock Camaro hood as the master. Siemens NX 10 Laminate FEA was first used to provide evidence that the hood would fail according to plan: not violating FMVSS and failing at the break in Aeromat prior to manufacturing. Figure 11 below illustrates the laminate analysis conducted. Vacuum resin infusion was used to create the final part from the mold. The pendulum impact tester was used to verify that the hood satisfied FMVSS 212 and 219, not shattering, shearing off its mounts, or impeding the drivers view over the specified amount in a 30-mph head on collision. A test fixture was designed to match the stock mounting position and simulate the hood’s mounting rotation as well as experienced lateral crumple zone when subjected to the pendulum impact tester.

Keysight Electric Motor Dyno Previous work on this team project showed that reliable data was difficult to collect from this system without some additional modifications. This year, CSU’s Keysight team eliminated a large vibration issue within the torque transducer subsystem that was being caused both by a slight alignment issue between the two motors and issues with the mounting of the motor. The test stand was redesigned to include dampened motor mounts; a new spline adapter was machined as well to address the alignment issue previously discovered. Finally, the team replaced the load motor to match the driving motor exactly in order to capture

FIGURE 11 ADAS vision training image example

FIGURE 12 Siemens NX laminate analysis of carbon fiber hood

efficiency data over the whole range of torque/speed of the driving motor. With this design currently in testing the team expects to utilize the motor efficiency knowledge gained through the Keysight team to improve the motor torque control strategy and Charge Depleting shift schedule.

ADAS, Sensor Fusion and System IntegrationHardware Integration The NXP S32V234 board mount was 3D printed with PLA filament. This material has adequate mechanical properties to withstand vibrations and minor impacts during normal vehicle operation while being easier to print in large volumes than ABS. The design itself is a simple rectangular box to surround the board with holes designed for ventilation and wiring.

The camera module is composed of two small circuit boards and a removable lens. Two mounting holes are available on one circuit board as well as a coaxial connector that is accessible through the enclosure. This camera will be mounted securely inside of the EcoCAR3 windshield, right below the rear-view mirror.

One RADAR and one LiDAR sensor are each mounted in the front of the vehicle within the front fascia. The mount is manufactured with aluminum and 3D printed ABS for increased structural support.

Bosch MMR Corner Radar sensors are mounted at the rear of the vehicle to assist with rear object detection. These are mounted inside the rear bumper, one on the driver side and one on the passenger side. An overview of sensor locations is shown in Figure 13.

Sensor Fusion and Software Development Since the team utilizes multiple types of sensors for range estimation and object detection, sensor fusion is necessary after enough data has been acquired from each sensor individually. Sensor fusion is the development of software that incorporates each sensor into one cohesive system. The flowchart of the team’s software development process is shown in Figure 14.

While Year 3 developed the baseline algorithms on a PC, this year all algorithms are run on the NXP S32V234

embedded system. CSU ADAS team developed algorithms transfer coordinates between camera generated coordinates and RADAR generated coordinates and uses this fusion to draw bounding boxes on the vehicles identified in the video. The range and range rate measurements are directly from the front RADAR sensor.

Summary and ConclusionsCSU’s participation in the EcoCAR 3 competition has led its VIT to design, build and test a PHEV 2016 Chevrolet Camaro. The final year of this competition has been primarily focused on in-vehicle dynamic testing aimed at the verification and validation of VTS that were developed as design objectives in the first year of the competition. In that regard, CSU’s team has been successful in both executing tests and using the testing data to show that the vehicle meet’s the performance objectives and constraints set by the competition organizers. The team overcame numerous obstacles along the way including various mechanical, electrical and control systems challenges. The result is a 2016 PHEV Camaro that consistently achieves 30+ mpgge, maintains regulation consistent emissions and achieves a 0-60 mph acceleration in under 8 seconds. CSU’s VIT has built a fuel efficient and environmentally friendly hybrid sports car that is fun to drive while being respectful of market trends towards intelligent energy consumption. Furthermore, the entire system has been designed by college students who will go on to contribute to the electrified vehicle industry in numerous capacities. CSU’s team considers this multiyear project to be a success in terms of both demonstrated performance results and education of the future automotive engineering workforce.

FIGURE 14 ADAS software development and sensor fusion flowchart

FIGURE 13 ADAS in-vehicle sensor locations

© 2019 SAE International. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of SAE International.

Positions and opinions advanced in this work are those of the author(s) and not necessarily those of SAE International. Responsibility for the content of the work lies solely with the author(s).

ISSN 0148-7191

References 1. “Plug-In Hybrid Electric Vehicle Market Introduction Study:

Final Report,” Oak Ridge, TN: Oak Ridge National Laboratory, 2010.

2. Gao, Y., Ehsani, M., and Miller, J.M., “Hybrid Electric Vehicle: Overview and State of the Art,” in Proceedings of the IEEE International Symposium on Industrial Electronics, 2005. ISIE 2005, 2005, doi:10.1109/isie.2005.1528929.

3. Joost, W.J., “Reducing Vehicle Weight and Improving U.S. Energy Efficiency Using Integrated Computational Materials Engineering,” Jom 64(9):1032-1038, 2012, doi:10.1007/s11837-012-0424-z.

4. Remy HVH250-115 Representative Efficiency Plot, Digital image, 2016. Accessed April 19, 2018. https://cdn.borgwarner.com/docs/default-source/default-document-library/remy-pds---hvh250-115-sheet-euro-pr-3-16.pdf?sfvrsn=ad42cd3c_9.

5. Smith, D., Lohse-Busch, H., and Irick, D., “A Preliminary Investigation into the Mitigation of Plug-in Hybrid Electric Vehicle Tailpipe Emissions Through Supervisory Control Methods,” SAE Int. J. Engines 3(1):996-1011, 2010, doi:10.4271/2010-01-1266.

6. Walsh, P. and Nelson, D., “Impact of Supervisory Control on Criteria Tailpipe Emissions for an Extended-Range Electric Vehicle,” SAE Technical Paper 2012-01-1193, 2012, doi:10.4271/2012-01-1193.

Contact InformationGabe DiDomenicoColorado State Universitygabriel.di_domenico@rams.colostate.edu

Jamison BairColorado State Universityjbair@rams.colostate.edu

AcknowledgmentsDr. Thomas Bradley - ME Faculty AdvisorDr. Sudeep Pasricha - ECE Faculty AdvisorAnthony Heap - General Motors student mentorMarc Henderson - Senior Engineer, Honda R&D AmericasRandy Fabrizio - Senior Engineer, Honda R&D Americas

Definitions and AbbreviationsAbbreviation - Description

AC - Alternating CurrentADAS - Advanced Driver Assistance SystemBSFC - Brake Specific Fuel ConsumptionCAN - Controller Area NetworkCD - Charge DepletingCS - Charge SustainingCSU - Colorado State UniversityDC - Direct CurrentDOE - Department of EnergyE&EC - Emissions and Energy ConsumptionECM - Engine Control ModuleEM - Electric MotorEPA - Environmental Protection AgencyESS - Energy Storage SystemEV - Electric VehicleGM - General MotorsHEV - Hybrid Electric VehicleHIL - Hardware-in-the-LoopHSC - Hybrid Supervisory ControllerHWFET - Highway Fuel Economy Testmpg - miles per galonmpgge - miles per galon- gasoline equivalentNVH - Noise, Vibration and HarshnessPHEV - Plug-in Hybrid Electric VehiclePRNDL - Park, Reverse, Neutral, Drive and LowSIL - Software-in-the-LoopTCM - Transmission Control ModuleV2G - Vehicle to GridVIL - Vehicle-in-the-LoopVIT - Vehicle Innovation TeamVTS - Vehicle Technical SpecificationsWTW - Well to Wheels

Colorado State University EcoCAR 3 Final Technical Report

Documents

EcoCar Challenge

Ecocar 10 electric

EcoCar Hybrid

University of Victoria EcoCAR Team -...

EcoCAR Challenge – Team 2 - MWFTR · 2013. 12. 6. ·...

Colorado Association for Career and Technical Educators

Education Technical Assistance - Colorado

EcoCAR 3 Request for Proposal - UVic.camech459/EcoCAR 3...

EcoCAR Mobility Challenge Communications Plan

ECOcar Magazine Issue 10

Ecocar 10 bifuel

Ecocar Luglio-agosto 2013

2010 EcoCAR Competition Finals Program

Colorado Technical College - November 2011

TECHNICAL SERVICE CENTER Denver, Colorado · TECHNICAL...

Colorado Career Technical Education Futuring Panels ·...