PhD Dissertation Defense
- 1. Study of Transit Bus Duty Cycle and its Influence on Fuel Economy and Emissions of Diesel Electric-HybridsJairo A Sandoval LeónPhD Dissertation DefenseCenter for Alternative Fuels, Engines and EmissionsDepartment of Mechanical and Aerospace EngineeringWest Virginia University Tuesday, March 29, 2011
- 2. OutlineMotivation and Problem StatementResearch ApproachFE and Emissions PredictionResultsConclusionsDemo of IBIS Tools2
- 3. 1.1. MotivationEnvironmental and health effects of transportationVehicle duty cycle Fuel economy & Emissions Hybrid buses will not perform the same on all service routes3
- 4. 1.2. Problem StatementIntegrated Bus Information System (IBIS)http://ibis.wvu.edu/ - U.S. DOT / FTATransit Bus Emissions DatabaseTransit Fleet Emissions Model Emissions / FE -> f (Cycle Metrics/Statistics)Which cycle metrics correspond to a particular type of operation?4
- 5. 1.2. Problem StatementWhat set of metrics should be used as explanatory variables for emissions / FE prediction?What structure should be used for the predictive model?How to obtain an adequate set of data to train the predictive models?Which routes that take the most advantage of hybrid-electric buses?How much fuel / emissions can hybrid buses save under specific driving conditions?How to include uncertainty figures in the predictions?5
- 6. 1.3. ContributionAnalysis of in-use transit bus routesDatabase of bus Duty Cycles (~3000 miles) w/ gradeA robust FE / Emissions prediction methodologyTwo Tools for IBIS:Fuel and CO2 savings of hybrid busesPost-processing of GPS route logsPSAT model for Series-Hybrid transit bus6
- 7. 1.4. Previous Work7MOBILE 6 / MOVESContinuous Emissions / Fuel Consumption DataIBISLinear Cycle InterpolationCycle FE / Emissions Prediction Tools
- 8. 1.4. Previous Work8Vehicle Specific PowerContinuous Emissions / Fuel Consumption DataOn-Board Emissions MeasurementChassis-Dynamometer TestingVehicle Simulation EnvironmentVehicle Operating ModesCycle FE / Emissions Prediction Tool
- 9. 2. Research Approach9Extensive Duty Cycle DatabaseThe Ideal ScenarioWVUTransLab (*)Continuous Emissions / Fuel Consumption DataCycle FE / Emissions Prediction Tool(*) A.C. Nix, J.A. Sandoval, W.S. Wayne, N.N. Clark & D.L.McKain, Fuel economy and emissions analysis of conventional diesel, diesel-electric hybrid, biodiesel and natural gas powered transit buses, 3rd International Conference on Energy and Sustainability , Wessex Institute of Technology, Alicante, Spain, April 11-13, 2011
- 10. 2. Research ApproachHow can we predict FE / Emissions under a wide range or driving conditions?10Continuous Emissions / Fuel Consumption DataDevelop Vehicle Dynamic Model (Simulation) Develop Extensive Duty Cycle DatabaseCycle FE / Emissions Prediction Tool
- 11. 2. Research Approach2.1. Transit Bus Routes2.2. Analysis of Cycle Metrics2.3. Test Data for Vehicle Model2.4. Engine Model2.5. Vehicle Dynamic Model11
- 12. 2.1. Transit Bus RoutesWMATA, Spring 20092900 mi / 260 hrŪmin = 6.7 mph – Inner CityŪmax = 25.9 mph - Commuter12
- 13. 2.1. Transit Bus RoutesElevation / road grade from topographic map13Routes in the DC / Maryland AreaImage from: maps.google.com
- 14. 2.2. Analysis of Cycle Metrics14Correlations between pairs of metrics R2Strong: >80%Mild : 50% - 80%Low : < 50%Explore predictions with Ū, Idle, and Acceleration
- 15. Routes by Service TypeK-means Analysis15
- 16. 2.3. Test Data (*)MY 2006 40’ Orion diesel-electric hybrid busSeries architecture powered with the BAE Systems HybriDrive® propulsion system6 different drive cycles: Beeline, Manhattan, New York Bus, OCTA, UDDS, and WMATA16(*) Transit Resource Center and West Virginia University Center for Alternative Fuels Engines and Emissions. “Analysis of Tailpipe Emissions from Westchester County Transit Buses.” Submitted to Liberty Lines Transit and Westchester County Department of Transportation. May 31, 2007
- 17. 2.4. Engine ModelTarget: MY 2007-2009 Cummins ISB 260H (1.2-1.5 g/bhp∙hr NOx)Test data: MY 2006 ISB 260H (2.5 g/bhp∙hr NOx)Correction factor for NOx emissions: 0.85 (-15%)2D Maps + ANNs17
- 18. 18and moreLug CurveRegion Covered By Test DataNOxEmissions(MY 2006)Fuel Consumption
- 19. 2.4. Engine Model19ANN ArchitectureSpeed and Torque:tt - 0.1 sect - 1 sec t - 5 sec% Errorfor NOx
- 20. 2.5. Vehicle Dynamic Model20Vehicle Architecture in PSATChar / Disch160 / 230 hp600V, NiMHHybrid Controller8 hp(6 kW)(*)(*)(*)320 hp Peak(238 kW)AC Induction2,700 ft-lb(3,660 Nm)260 hp Cont(194 kW)Perm. Magn.1:1260 hp Peak(194 kW)5.9 L Diesel600 ft-lb Peak(815 Nm)8 hp(6 kW)Ratio4.9 80 ft2Af0.79 Cdφ38.5”8.6∙10-3 Cr(*) Bell, C. “An Investigation of Road Load Effects on Fuel Economy and NOx Emissions of Hybrid and Conventional Transit Buses.” Thesis (MS). West Virginia University, 2011
- 21. 2.5. Vehicle Dynamic ModelLoad-Following Control + ImprovedMotor / Generator / Battery: approximated based on BAE specs21
- 22. OutlineMotivation and Problem StatementResearch ApproachFE and Emissions PredictionResultsConclusionsDemo of IBIS Tools22
- 23. 3. FE and Emissions PredictionDriving + Idle Contributions:Idle FC/NOx rates:F.C.Driving = f ( Ūno idle, ãno grade)Fuel savings: gallons per hour23
- 24. 24Flow Chart
- 25. OutlineMotivation and Problem StatementResearch ApproachFE and Emissions PredictionResultsConclusionsDemo of IBIS Tools25
- 26. 4. ResultsSimulation Vs. Test:26Fuel EconomyNOx Emissions
- 27. 4. ResultsRegressions for Driving Contributions:27Fuel ConsumptionNOx Emissions
- 28. 28Fuel EconomyNOx EmissionsCO2 Emissions
- 29. Hybrid Advantage and Fuel Savings29Hybrid AdvantageFuel Savings
- 30. Service Categories30Conventional vs. HybridFuel Savings
- 31. OutlineMotivation and Problem StatementResearch ApproachFE and Emissions PredictionResultsConclusionsDemo of IBIS Tools31
- 32. 5. ConclusionsPredictor metrics: Average speed, Idle fraction, and Characteristic accelerationThe highest fuel saving rates (gallons per hour) were observed at middle to high speed serviceThe highest fuel savings correspond to routes with the highest values of characteristic acceleration Not all routes in a particular service category get the same benefit from hybridizationThe data did not reveal a benefit in NOx emissions from hybridization32
- 33. Future WorkStudy the effect of road gradeBuilt into IBIS road grade calculationExplore laboratory bias uncertainty and detection limits33
- 34. 6. Demo of IBIS ToolsGPS Data Cleaning ToolHybrid Savings Calculation ToolQuestions ?34
- 36. Fleet Route Speed Distributions36WMATADART∴ Operation differs between transit agencies… Importance of present work
- 37. Vehicle Level Validation37Fuel EconomyNOx Emissions
- 38. Limited User InputsUse approximations for Idle and Acceleration:38Idleãno grade
- 39. Conventional Vehicle39FE vs. ŪPredicted vs. Target
Editor's Notes
- #2: My StoryIt was five years ago when my wife and I moved to West Virginia. Here we had the most wonderful welcome by Dr Wayne who guided us through all the process of adjusting to a new country and culture. For that and all of his support through my time at WVU I want to express my gratitude.I thank all the people who have given us a hand: Dr Clark, my professors, MAE and CAFEE staff, the list is endless. I am also thankful for all the friendships we’ve made and which we’ll never forget.Thank you for listening and now, let’s get to what we’re here for.
- #4: PM: Respiratory, Haze, acid in water reservesNOx: Respiratory and heart, ozone and smog, acid rain, GHGHC: oxone, GHGCO2: GHG
- #5: Operation: IBIS target users in the transit bus industry. Except for average route speed, They have limited information about the cycle metrics for their operation.
- #6: Metrics: There was some preliminary work done by the IBIS development group. But I wanted to analyze this with the dataset I collected.Structure: Again, preliminary work by IBIS group. Make further developments, so to speak.Dataset: The original dataset was from standard transit bus cycles. As we wanted the model to predict over a wide range of operation, we needed to expand the cycle dataset.Routes: We were interested in allowing transit agencies to evaluate their routes in order to place their hybrids where they could be best exploited. We also were interested in finding if there was a specific driving condition which was best suited for hybrids: e.g. inner-city vs. urban.Fuel: Of course one of the criteria to decide which route may be best is fuel savings which translates to dollars.Uncertainty: Of course the predictions cannot be absolute, they have uncertainty. Previous models included only absolute values of their predictions. I wanted to include confidence intervals.
- #8: MOBILE 6/MOVES: EPADid not do what I needed
- #9: What about the implications of the hybrid system?Regenerative brakingBattery State of Charge
- #10: Perform chassis dynamometer tests over an extensive Duty Cycle database ExpensiveExtensive Cycle database was not available
- #11: This is how the project flowedDuty cyclesDynamic modelSimulationsPrediction Tool
- #13: GPS Speed / Position + Internal AccelerometersECU: J1939 interfaceBarometric pressure ~ Elevation
- #15: This procedure is standard practice, nothing fancy here.Correlated. The predictive power of the second parameter may not be significant.Uncorrelated. The predictive power of both parameters should be significant.Ū_no_idle is known from Ū and Idle
- #17: Capacity: 39 seated + 20 standingPM, CO, and HC were very low due to the DOC/DPF aftertreatmentCurb / GVWR / Test weights: 33,440 lb / 42,540 lb / 38,540 lb 15,170 kg / 19,295 kg / 17,480 kgULSD#1 fuelDiesel oxidation catalyst (DOC) and particulate trap (DPF)
- #18: % Torque was translated to Nm with the engine performance curve supplied by Cummins.The data were not corrected for dispersion / diffusion.Time alignment and CFR Part 86 Subpart N reductionECU: Speed, % Torque, and Fuel RateFuel consumption and NOx (CO2: carbon balance)
- #19: Cell counts= 10 rpm x 5 Nm2D Map Grid: 50 rpm x 10 Nm for Fuel, R^2 = 0.98210 rpm x 5 Nm for NOx, R^2 = 0.902Max Efficiency: 43%
- #20: rarbas = radial basistansig = hyperbolic tangent sigmoidThe figure at the bottom shows an example of the improvement with ANN
- #21: Batteries in test vehicle were lead-acid. In the model: NiMH
- #22: * SOC recharge if below threshold* Better monitoring of SOC* Limit performance if SOC is low* Mechanical accessory loads at braking/stops
- #28: Plot
- #33: the higher the ã, the more room there is for the hybrid system to recover regenerative braking energy and to balance engine operation.NOx: This fact can be explained by the lower loads experienced by the engine on the hybrid bus, as lower loads are associated with higher brake-specific NOx.
- #37: Nothing below 5 mph or above 30 mph
- #38: Overall, the model predictions have good agreement with test data.Results are in the ballpark