Griess Thesis Defense PP

BENCHMARKING, CHARACTERIZATION AND TUNING OF SHELL ECOMARATHON

POWERTRAIN

Presented by Eric Griess

Masters of Science – Mechanical Engineering

California Polytechnic State University – San Luis Obispo

Outline

1. Project Summary

2. Test Setup

3. Calibration

4. Benchmark Testing

5. Simulation Development

6. Establishing AFR and Ignition Targets

7. Engine Tuning

8. Conclusions

9. Recommendations

10. Vehicle Performance Study

Competition Outline PROJECT SUMMARY

The Shell EcoMarathon is an annual national competition in which schools enter student-built and student-driven vehicles into various classes to achieve the best fuel efficiency possible.

Urban Concept Prototype

Competition Outline

Prototype class competition:

- Large circuit course on public roads

- Total distance of 6 miles

Only limits on operation:

- Average speed of 15 MPH

- Maximum time limit

PROJECT SUMMARY

Competition Outline

Even with minimal requirements, successful teams have

adapted a “burn and coast” method. This introduced unique

challenges such as:

- Engine temperature variation

- Clutch control for starts, restart

- Fuel for engine starting

PROJECT SUMMARY

Problem Definition

The team placed 6th in 2013 with 1300 MPG. However, some aspects

of vehicle design left room for improvement:

- Engine was not tuned

- Non-ideal gear ratio due to unknown engine performance

- Electronics issue limited engine speed

- Chain drive system introduced additional losses

- Lack of analysis tools to guide design direction

PROJECT SUMMARY

Project Scope

This study aims to:

1. Quantify engine performance

2. Develop vehicle/engine simulation tool

3. Use information from both to select tuning targets for air-

fuel-ratio (AFR) and ignition advance settings.

4. Perform engine tune

5. Quantify improvements over previous tune

PROJECT SUMMARY

Project Scope

Deliverables to the team:

- Fully tuned engine with engine parameter tables

- Vehicle simulation tool

- Suggestions for continuing project development

- Recommendations for simulation-based vehicle

design approach

PROJECT SUMMARY

Outline

1. Project Summary

2. Test Setup

3. Calibration




7. Engine Tuning

8. Conclusions

9. Recommendations


Table, Dynamometer

Dynamometer Table

Eddy Current Dyno

Controller

Power Supply

Torque Conditioner

TEST SETUP

Setup Requirements

In order to begin testing, the following items were necessary:

- Mounting system for engine, dynamometer

- Drive system

- Electronics integration (ECU, Power supplies, signal conditioners)

- Dedicated fuel system

- Temperature control system

- Safety cage

- User interface

TEST SETUP

Engine, Dyno Mounting TEST SETUP

Engine, Dyno Mounting TEST SETUP

One-piece aluminum engine mount was used during competition.

For extended testing, exhaust portion was replaced for steel runner with same diameter and length.

Drive System TEST SETUP

Direct drive system chosen after many iterations(discussed later)

Elastomer “spider” damper used to allow for small misalignment

Drive System TEST SETUP

• 2nd mount necessary

with new exhaust

• Placed as close to output

shaft as possible to

reduce local deflection

Drive System Alignment TEST SETUP

Parallel Alignment: .010” max

Axial Alignment: .010” max

Angular Alignment: 1º max

Electronics TEST SETUP

• Megasquirt II ECU

• Breakout Board

• Battery quick

disconnect for Cold

Cranking Amps

Requirement (40A)

Fuel System TEST SETUP

Temperature Control TEST SETUP

Team normally runs engine without coolant

due to relatively short burn times.

In order to control engine temperature

during testing, factory water-cooling system

was used without a thermostat.

Then centrifugal blower with carefully

designed ducting system was used to control

mass air flow over radiator, manipulated

with control lever (welding rod)

User Control TEST SETUP

Emergency Kill

ECU Power Fuel Pump Ignition

Engine Start Throttle Control

Safety TEST SETUP

Megasquirt Interface TEST SETUP

Dyno Control Interface TEST SETUP

Final Setup TEST SETUP

Outline

1. Project Summary

2. Test Setup

3. Calibration4. Benchmark Testing



7. Engine Tuning

8. Conclusions

9. Recommendations


Calibration Standards CALIBRATION

When possible, engine testing standard SAE J1349 was followed for

determining tolerances and test procedures. For data acquisition,

tolerances on sensors include:

Dynamometer Torque: ± 0.5%

Speed: ± 0.2%

Inlet Air Temperature: ± 1 ºC

Fuel Flow: ± 1%

Engine Temperature: ± 2 ºC

Dynamometer Calibration CALIBRATION

Sensor Calibration

Stein-Hart Equations for thermocouple behavior used for calibration

CALIBRATION

1

𝑇= 𝐴 + 𝐵 ln 𝑅 + 𝐶 ln 𝑅 3

Fuel Flow Calibration CALIBRATION

Fuel Flow Calibration CALIBRATION

Unfortunately, flow meter was not accurate at most of the flow rates

the engine was seeing. Instead, injector duty cycle was related to flow

rate through the previous calibration.

𝑚𝑓𝑢𝑒𝑙 = 0.474 ∗ 𝐷𝑢𝑡𝑦 𝐶𝑦𝑐𝑙𝑒 − 0.1114

Duty cycle output was captured for each steady state test and used for

fuel flow calculations.

Repeatability

In order to ensure repeatable results, repeatability testing was

conducted through 10 consecutive, identical tests.

𝑅 =𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑣𝑎𝑙𝑢𝑒 𝑜𝑣𝑒𝑟 1 𝑡𝑒𝑠𝑡

𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑣𝑎𝑙𝑢𝑒 𝑜𝑣𝑒𝑟 10 𝑡𝑒𝑠𝑡𝑠∗ 100

The importance of this is to assure valid results throughout a single

test, instead of repeating each test multiple times.

CALIBRATION

Repeatability Summary

Drivetrain iteration was necessary to meet acceptable repeatability

results ( < 1%)

CALIBRATION

Drivetrain Iterations

Iteration 1: Chain drive, no tensioner, no temperature control

CALIBRATION

- Chain vibration too violent for safe use


Iteration 2: Chain drive, torsion spring tension, no temperature control

CALIBRATION

- Chain ‘tight’ side still vibrating too much, repeatability unacceptable


Iteration 2: Chain drive, torsion spring tension, no temperature control

CALIBRATION

During tests, engine temperature was

varying from 120-190 ºF, even with

water cooling system.

Torque, power values were drifting

over time


Iteration 3: Chain drive, torsion tension, temperature control

CALIBRATION

Engine temperature was steady, but

torque was still varying excessively.

Next option was to address chain

vibration issue


Iteration 4: Chain drive, double idler tensioner, temperature control

CALIBRATION

- Still saw excessive vibration since lateral movement was allowed


Iteration 5: Chain drive, double idle tensioner, temperature control

CALIBRATION

• Chain overheated and caused

high stress condition.

• 4th crankshaft failure under chain

loading conditions

• Repeatability results were

unacceptable


Iteration 6: Direct Drive

Softer elastomers could not

withstand heating by friction.

Stiffer couplers used with

lubrication to alleviate issues.

CALIBRATION

Drivetrain Comparison

Figure shows engine speed comparison

over 60 seconds of testing.

Direct drive had ±20 RPM (±0.5%),

while chain drive saw ±60 RPM.

Best torque output repeatability with

chain was 3%, while direct drive was

0.2%.

CALIBRATION

Outline

1. Project Summary

2. Test Setup

3. Calibration

4. Benchmark Testing5. Simulation Development


7. Engine Tuning

8. Conclusions

9. Recommendations


Variable Definitions

BSFC = Brake Specific Fuel Consumption

= 𝑀𝑎𝑠𝑠 𝐹𝑙𝑜𝑤 𝑅𝑎𝑡𝑒 𝑜𝑓 𝐹𝑢𝑒𝑙

𝑃𝑜𝑤𝑒𝑟 𝑂𝑢𝑡𝑝𝑢𝑡 𝑜𝑓 𝐸𝑛𝑔𝑖𝑛𝑒

AFR = Air-fuel-ratio = 𝑀𝑎𝑠𝑠 𝑜𝑓 𝑎𝑖𝑟

𝑀𝑎𝑠𝑠 𝑜𝑓 𝑓𝑢𝑒𝑙

(λ) = Lambda = 𝐴𝐹𝑅𝐴𝑐𝑡𝑢𝑎𝑙

𝐴𝐹𝑅𝑠𝑡𝑜𝑖𝑐ℎ𝑖𝑜𝑚𝑒𝑡𝑟𝑖𝑐

VE = Volumetric Efficiency = 𝑚𝑎

𝜌𝑎,𝑟𝑒𝑓∗𝑉𝑑

MBT = Maximum Brake Torque

Φ = Spark Timing (relative to MBT advance)

BTDC = Before Top Dead Center

BENCHMARK TESTING

Baseline Testing Goals

1. Acquire performance data for engine used in

competition

2. Quantify relationships between fuel delivery, spark

timing, and temperature with torque output and

BSFC

3. Use these trends to identify minimum and maximum

points for torque output and BSFC.

BENCHMARK TESTING

Baseline Performance

Test procedure for baseline testing:

• Each test conducted at maximum 500 RPM intervals

• Data only recorded during steady state (T = 167 ± 3 ºF)

• Sampling at 1 Hz for one minute

• Engine returned to idle, then next speed tested immediately

BENCHMARK TESTING

Baseline Performance

Max Torque: 2.71 ftlb @ 4800 RPM

Max Power: 2.48 HP @ 4800 RPM

Minimum BSFC: .473 lb/hphr @ 3700RPM

BSFC is abnormally high for small spark

ignition gasoline engine. (0.35 - 0.45 exp.)

Torque output changes by 8% across entire

speed range

Power continues to increase before cutoff

BENCHMARK TESTING

Baseline Performance BENCHMARK TESTING

• At this point, cannot tell how far

ignition values are off (MBT not

found at each point)

• Engine was running up to 14%

rich at points, and an average of

8% rich across operating range.

• In order to find ideal operating

points for engine, variation

studies were performed.

Lambda Variation BENCHMARK TESTING

Testing Method:

• Engine Speed: 4000 RPM

• Engine Temperature: 167 ± 3 ºF

• Ignition: MBT Advance (ϕ = 0º)

By changing fuel delivery parameter in Engine Control Unit (ECU), Lambda was varied from λ = 0.8 (20% rich) to λ = 1.2 (20% lean). Each test was taken for 1 minute at 1 Hz.

Lambda Variation BENCHMARK TESTING

Maximum torque: = 0.95

Minimum BSFC: = 1.10

Minimum Torque: = 1.2 (12% loss)

Maximum BSFC: = 0.8 (32% loss)

This trend is widely accepted. The dotted lines show the trend observed in Heywood.

Compression ratio, engine size, and flow/combustion characteristics can shift these curves

Ignition Variation BENCHMARK TESTING

Testing Method:


• Engine Temperature: 167 ± 3 ºF

• Lambda: = 1.0 (Stoichiometric)

Ignition advance was changed from 18º - 26º BTDC. Since 22º was MBT timing, then these values mean was varied from -4º to 4º.

Ignition Variation BENCHMARK TESTING

Maximum Torque: = 0º

Minimum BSFC: = 0º

Minimum Torque: = ±4º (3.8% loss)

Maximum BSFC: = ±4º (2.9% loss)

• This was also an expected trend, as seen from Heywood prediction lines.

• Dip is thought to be caused by a hardware issue during testing.

Temperature Variation BENCHMARK TESTING

Testing Method:


• Ignition: MBT Advance (ϕ = 0º)

• Lambda: = 1.0 (Stoichiometric)

Temperature was varied from 130 – 210 ºF in 20 ± 2 ºF increments. Tests were performed separately at steady state for one minute.


Maximum Torque: T = 120 ºF

Minimum BSFC: = 170 ºF

Minimum Torque: = 210 ºF (5.4% loss)

Maximum BSFC: = 210 ºF (1.2% loss)

BSFC = 𝑚𝑓

𝑃=

1

𝜔𝑒

𝑚𝑓

𝑇

Hypothesis: Engine torque decreases, but slower rate due to closing tolerances (up to 170 ºF), then drops faster past that point due to increasing friction losses


Temperature variation study revealed several important conclusions:

- Higher temperature is not always better

- Actual BSFC losses relatively small

- In order to minimize these losses, fueling needs to adapt to

temperature changes. If mass flow rate of fuel is constant, then

running = 1.0 at stoichiometric at 120 ºF will result in running

nearly 10% rich at 210 ºF, causing a 15% increase in BSFC.


Temperature variation study revealed several important conclusions:

- Higher temperature is not always better

- Actual BSFC losses relatively small

- In order to minimize these losses, fueling needs to adapt to

temperature changes. If mass flow rate of fuel is constant, then

running = 1.0 at stoichiometric at 120 ºF will result in running

nearly 10% rich at 210 ºF, causing a 15% increase in BSFC.

Idle vs. Cut Test BENCHMARK TESTING

In order to verify the burn/coast method as desirable to begin with, the

fuel flow during engine starting and idle were studied.

Crossover time: 0.28 seconds

Meaning if car is coasting for

more than 0.28 seconds, cutoff

is more efficient than idling.

Benchmark Testing Conclusions BENCHMARK TESTING

• BSFC is vastly more sensitive to AFR than spark timing

and engine temperature.

• Temperature control is only important due to varying

volumetric efficiency causing changes to AFR.

• Burn and coast method verified

Variable Maximum Observed

BSFC Loss

AFR/Lambda () 32 %

Spark Timing 2.9 %

Temperature 1.2 %

Outline

1. Project Summary

2. Test Setup

3. Calibration


5. Simulation Development6. Establishing AFR and Ignition Targets

7. Engine Tuning

8. Conclusions

9. Recommendations


Simulation Goals SIMULATION DEVELOPMENT

1. Introduce relationships only seen during transient engine

operation to replicate competition conditions

2. Vary both AFR and ignition under competition conditions

to define theoretical targets

3. Compare these targets to steady state targets (is

minimizing BSFC best for fuel economy?)

4. Suggest simulation-based design techniques

Disclaimer SIMULATION DEVELOPMENT

Simulation is NOT verified. Although the vehicle model was carefully

developed and tested, the team needs to perform their own verification

of this model to improve confidence. Some large assumptions in this

model:

- Straight line acceleration/deceleration

- Clutch engagement speed is constant

- Constant average rolling resistance

- Constant environment for both vehicle and engine

Model Overview SIMULATION DEVELOPMENT

1. Engine / clutch speed control

2. Calculates output torque based on speed, where lambda, ignition, and temperature effects are integrated

3. Tractive forces calculated after drag, rolling resistance, and inertial losses calculated.

4. Vehicle acceleration, velocity, distance calculated then velocity used to calculate closed loop engine speed

Model Overview SIMULATION DEVELOPMENT

Simulation features:

- Constant modification of torque and BSFC with lambda, ignition

and temperature tracking.

- Centrifugal clutch engagement model

- Accounts for environmental effects (engine bay temp, local altitude)

- Energy path analysis

- 2D , 3D trend studies between variables

Outline

1. Project Summary

2. Test Setup

3. Calibration



6. Establishing AFR and Ignition Targets7. Engine Tuning

8. Conclusions

9. Recommendations


Goals AFR, IGNITION TARGETS

• Use simulation to establish theoretical AFR,

ignition targets

• Investigate differences, if any

• Select target values

Acquisition Method AFR, IGNITION TARGETS

For 3D trend studies, the simulation was run through

an iteration matrix of lambda and ignition values.

• Lambda from 0.8 – 1.2 (.05 step)

• Ignition from = -4º to + 4º (1º step)

• 9 x 9 combinations

Results AFR, IGNITION TARGETS

Results AFR, IGNITION TARGETS

Component Simulation Steady State

Lambda () 1.05 1.1

Ignition Advance () 0 (MBT) 0 (MBT)

Suggested Improvement 19.5%

Despite having steady state

performance data in simulation,

vehicle analysis suggests running

an AFR 5% richer than minimum

BSFC.

Hypothesis AFR, IGNITION TARGETS

• Suggested = 1.05 is between minimum BSFC

(λ = 1.1) and maximum torque (λ = 0.95),

• Hypothesized that the best fuel economy comes from

a balance between the two

To investigate this, the lambda variation trend

was revisited.


Here, BSFC and torque trends are normalized to % loss from ideal value (maximum torque / minimum BSFC)

Solid line is both of the losses added together. The smallest represents the point where total loss is smallest. This coincides with = 1.05.


Identical plot, only with total vehicle simulation results normalized and superimposed.

This shows that fuel economy trend follows ‘total loss’ curve, instead of BSFC or torque.

Target Selection AFR, IGNITION TARGETS

1. The lambda variation trend observed at 4000 RPM is expected vary

slightly at different speeds. Aiming for = 1.05 should represent a

good “average” value that sacrifices only 1.4% BSFC at 4000 RPM.

2. Hypothesis from simulation data, supported by experimental trends

Component Selected Target

Lambda () 1.05

Ignition Advance () 0 (MBT)

Outline

1. Project Summary

2. Test Setup

3. Calibration




7. Engine Tuning8. Conclusions

9. Recommendations

Tuning Method ENGINE TUNING

Test Procedure:

1. Reach steady state at each operating point

2. Change ECU fuel table to result in = 1.05

3. Change ignition values for = 0º (MBT)

4. Verify both targets, acquire steady state

5. Record data for 1 minute, 1 Hz

6. Repeat for each point

Lambda Values ENGINE TUNING

Ignition Values ENGINE TUNING

Results ENGINE TUNING

BSFC Improvement ENGINE TUNING

Projected Competition Improvement ENGINE TUNING

New engine tune replaced

baseline tune in engine, with all

other variables unchanged.

Estimated 17.5% improvement,

which is similar to the 18.8 %

average BSFC improvement

from engine tune.

Outline

1. Project Summary

2. Test Setup

3. Calibration




7. Engine Tuning

8. Conclusions9. Recommendations


Summary CONCLUSIONS

Experimental Conclusions:

- BSFC is much more sensitive to AFR than spark timing or

engine temperature.

- MBT timing is always optimal for maximizing torque and

minimizing BSFC

- Engine temperature does not play a large role, as long as

fueling can adapt to changing volumetric efficiency.

Summary CONCLUSIONS

Simulation-based hypotheses:

- Vehicle fuel economy trends are governed more by the total loss curve between torque and BSFC, instead of individual curves.

- The lambda value that minimizes total loss should result in best overall fuel economy.

- Average BSFC improvement across operating range in steady state should be expected to result in similar overall fuel economy improvement.

Outline

1. Project Summary

2. Test Setup

3. Calibration




7. Engine Tuning

8. Conclusions

9. Recommendations


Engine Tuning RECOMMENDATIONS

Primary recommendations:

1. Consider using narrow-band oxygen sensor for more accuracy

2. Acquire lambda variation trends at all testing speeds

3. Develop accurate temperature compensation curves for fuel

4. Alleviate artificial engine speed limit caused by sensor

Drivetrain Efficiency RECOMMENDATIONS

Switching from chain

drive to direct drive

allowed direct comparison

of drive systems.

Average drivetrain

efficiency with centrifugal

clutch and single chain

was 85%.

Crankshaft Failure RECOMMENDATIONS

Believed that the transverse loading

condition from chain tension

caused the multiple crankshaft

failures.

Suggestions for prevention:

- Decrease cantilever loading

- Minimize potentially

unbalanced weight

Simulation Development RECOMMENDATIONS

In order to improve accuracy of the simulation, the following variables

should be studied further:

- Average rolling resistance

- Coefficient of drag

- Clutch control behavior

- Overall drivetrain efficiency

- Engine bay transient behavior

- Environmental conditions (average temperature, wind speed, etc)

Simulation Development RECOMMENDATIONS

Suggested verification methods:

1. Simulate fraction of drive cycles on engine or chassis dynamometer, and record

all relevant data. Then change a variable such as AFR or gear ratio, then

compare recorded data with simulation data.

2. Perform the same in a very controlled environment, but necessary to take more

samples in order to address outlying data.

3. Implement an electronic drive motor with well-known characteristics to

measure vehicle losses

Vehicle Design RECOMMENDATIONS

Once simulation is further verified, then perform sensitivity analysis on

variables of interest.

𝑆𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 =% 𝐶ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝑂𝑢𝑡𝑝𝑢𝑡 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒

% 𝐶ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝐼𝑛𝑝𝑢𝑡 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒

If listed in descending order, it may provide design direction for the

team in order to maximize results.

For example, the gear ratio trend study saw a 0.83 % change in MPG

for every % change in gear ratio, while vehicle mass only saw 0.14%.

Outline

1. Project Summary

2. Test Setup

3. Calibration




7. Engine Tuning

8. Conclusions

9. Recommendations

10.Vehicle Performance Study

Goals VEHICLE PERFORMANCE STUDY

Quick additional exercise, changed some variables that

team has control over:

1. Gear Ratio

2. Clutch Pad Mass

3. Drivetrain Efficiency

Improvement Summary VEHICLE PERFORMANCE STUDY

NOTE: These are only projected improvements from simulation

Acknowledgements

Special thanks to:

Patrick Lemieux

Jim Gerhardt

Dorian Capps

Sean Michel

QUESTIONS?

CONCERNS?

INQUIRIES?

Documents

Griess Thesis Defense PP