Transient engine modeling at John Deere using GT- POWER · Transient engine modeling at John Deere using GT-POWER GT-SUITE NORTH AMERICAN CONFERENCE - 2009 ... (e.g. fuel, timing,

Transient enginemodeling at JohnDeere using GT-POWER

GT-SUITE NORTH AMERICANCONFERENCE - 2009

Transient engine modeling | GT-SUITE North American Conference | 7 December 20092 |

Agenda

• Definition of Transient Engine Model

• Components

• Engine Model• ECU Model• Interface Model

• Model Development Process

• Model Applications

• Sample Results A – Performance

• Sample Results B – Emissions

• Summary

Transient engine modeling | GT-SUITE North American Conference | 7 December 20093 |

Definition of Transient Engine Model

A co-simulation between an engine model and an Engine ControlUnit (ECU) model [Software-in-the-Loop (SiL)] or actual ECUhardware [Hardware-in-the-Loop (HiL)].

ECU Model orHardware

ECU toEngine

Engineto ECU

Transient Engine Model

Output toUser/Vehicle

Input fromUser/Vehicle

Engine ModelMay include

emissions andaftertreatment

prediction

Transient engine modeling | GT-SUITE North American Conference | 7 December 20094 |

Components – Engine Model

A model of engine performance that may include emissionsand/or aftertreatment. Model fidelity is determined by theintended use of the model.

High Fidelity (HiFi) Mean Value (MV)Real Time (RT)

PhysicalEngine

This step usuallyjust involves aproject type changeand possibly asolver change.

Transient engine modeling | GT-SUITE North American Conference | 7 December 20095 |

Components – ECU Model

A system level model of ECU software functionality, which canthen be used on a standard PC for virtual development. The ECUmodel is critical to generating accurate transient performancepredictions.

ECU Hardware ECU Model

In Out

ECU Functionality

Coded Modeled

Application Specific Calibration

a=1.234 d=4.567b=2.345 e=5.678c=3.456 f=6.789

Transient engine modeling | GT-SUITE North American Conference | 7 December 20096 |

Components – Interface Model

The model that defines the connections between all othermodels. It has a modular setup to make it easily configurable.

Throttle

Load

OtherInputs

EngineSpeed

OtherOutputs

Level #1Level #2

ECU Model orHiL Interface

Throttle

Actuators

Engine Model

Sensors

Load

EngineSpeed

Actuators

SensorProcessing

ActuatorProcessing

OtherOutputs

Sensors

L1L1

L1

L1

Level #3Engine

LoadEngineSpeed

ActuatorsOtherOutputs

Sensors

L2L2

L2

EmissionsIn Out

AftertreatmentIn1

In2Sensors L2

L2

Transient engine modeling | GT-SUITE North American Conference | 7 December 20097 |

Design Updates

Model Development Process

Rough Models

Final Models

Specificationsgeometry/functionality

Open Loop Simulationmodels not coupled

Closed Loop Simulationmodels are coupled

More Simulationboth open and closed loop

Component Test Datamodel validation/calibration

Engine Test Datamodel validation/calibration

Specification Refinement

Performance Verification

Algorithm Updates

Transient engine modeling | GT-SUITE North American Conference | 7 December 20098 |

Model Applications

Model based software development:• Control algorithm development

• Control algorithm testing and validation

• Auto-generation of ECU code

Engine system performance:• Optimization

• Non-standard conditions

• Gain tuning

Integration of engine into vehicle:• Controller interaction

• Performance verification and optimization

Transient engine modeling | GT-SUITE North American Conference | 7 December 20099 |

Sample Results A – Performance

High Fidelity transient engine model versus vehicle data:• Engine model calibrated to 41 steady state data points (see below).• Engine model coupled with ECU model for transient comparison.• Two inputs applied to transient engine model from vehicle test.

• Throttle• Load (drive shaft and auxiliaries)

Transient engine modeling | GT-SUITE North American Conference | 7 December 200910 |

Sample Results A – Performance (continued)

Engine Model Outputs

ECU Model Outputs

Transient engine modeling | GT-SUITE North American Conference | 7 December 200911 |

Sample Results B – Emissions

Engine models versus dynamometer data:• High fidelity engine model calibrated to 29 steady state data points.

• Mean value engine model created from high fidelity model.

Additional steady state and transient model validation without an ECUmodel, which requires additional inputs from the lab data:

• Engine speed• Actuator commands from ECU (e.g. fuel, timing, EGR valve, etc.)

• Steady state comparison to around 1400 lab data points. Includesranges of operation not included in the original calibration.• Wider range of actuator positions.• Swings in various system restrictions.• Swings in ambient temperature.

• Transient comparison to a set of NRTC (Nonroad Test Cycle) lab data.

Transient engine modeling | GT-SUITE North American Conference | 7 December 200912 |

Sample Results B – Emissions (continued)

High Fidelity

Steady state comparison to around 1400 lab data points.

Mean Value

Transient engine modeling | GT-SUITE North American Conference | 7 December 200913 |

Sample Results B – Emissions (continued)

Transient comparison to a set of NRTC (Nonroad Test Cycle) lab data.

Transient engine modeling | GT-SUITE North American Conference | 7 December 200914 |

Sample Results B – Emissions (continued)

Transient comparison to a set of NRTC (Nonroad Test Cycle) lab data.

Transient engine modeling | GT-SUITE North American Conference | 7 December 200915 |

Summary

• A transient engine model for today’s engine must include anengine model AND ECU model.• Engine model validation can be done without an ECU model but

engine performance prediction requires an ECU model.

• Model development is an integral part of hardwaredevelopment.

• GT-POWER is a very capable tool for developing enginemodels for each point in the process.