Predicting Valve Train Dynamics using Simulation … American GT Conference 2017 Predicting Valve...

North American GT Conference 2017

Predicting Valve Train Dynamics using Simulation with Model

Validation

Brice Willis, EngineerKevin Ireland, Engineer

Computational Engine ResearchHonda R&D Americas, Inc.


Overview• Objectives

• Validation Test Setup

• Model Build

• Model Validation

• Other Systems

• Achievements

• Q & A

2


Objective

Develop and validate 2D valve train modeling methodology using GT-SUITE and motoring bench testing

Develop a modeling methodology that can be used consistently for different valve train configurations and engine operating conditions

Minimize model calibration and maintain accuracy

A validated model can help; Reduce overall development cost

Lead design Reduce required testing

Predict, evaluate, and improve valve train system performance Design of experiments Evaluate cam profiles Evaluate valve spring stiffness Capture dynamic valve lift

3


Testing-Simulation: One Approach

XThe test plan does not always consider what measurements are needed to validate the simulation. This leads to high levels of model calibration and non-repeatable model correlation.

4


Testing-Simulation: Another Approach

Planning test measurements for model validation will reduce model calibration and increase the repeatability of model correlation.

5


Crank Encoder Wheel

Painted Valves with Camera

Gap Sensor Rocker Arm Strain Gauges

Valve Spring Strain Gauges

Motoring RPM Valve Lift via calibrated voltage

Valve Lift ≤ 1mm Measured strain tappet force

Measured strain spring stress

Validation Test Setup Validation testing was completed using motoring bench testing

Constant engine temperature (25°C to maintain build condition) Speed sweeps (0-6800 RPM) High and low valve timing/lift configurations

*Exhaust rocker shown

Validation testing was intended to capture tappet force and valve lift over a range of engine speeds while maintaining the build condition.

6


Test Instrumentation Layout

Due to the engine configuration, instrumentation of multiple cylinders allows six different valve train configurations to be evaluated (Front & Rear / Intake & Exhaust / High & Low).

R = Rocker Arm strain gauge

S = Spring Strain Gauge

G = Gap sensor

V = Painted Valve

Total of 40 channels

T = ThermocoupleT

7


Raw Information

Intermediate Model

Direct GT Input

Optional GT Input

2D Valve Train Model

Model Build Overview

Material Properties

Valve Train

Design & Layout

Part/Assy. Geometry (3D CAD)

𝜇𝜇𝑃𝑃 = 𝜇𝜇𝑜𝑜𝑒𝑒𝛼𝛼𝑃𝑃

8


Raw Information

Intermediate Model

Direct GT Input

Optional GT Input



Kinematic VT Layout

Material Properties

Valve Train

Design & Layout


Mass Properties


9


Raw Information

Intermediate Model

Direct GT Input

Optional GT Input



Kinematic VT Layout

Material Properties

Valve Train

Design & Layout


Mass Properties

Measured


10


Raw Information

Intermediate Model

Direct GT Input

Optional GT Input



Kinematic VT Layout

Material Properties

Part Stiffness

Valve Train

Design & Layout


Mass Properties

3D Finite Element Analysis

Measured


11


The rocker stiffness was predicted independent from the valve train assembly and supporting structure. Using 3D finite element analysis (FEA) the rocker was defined as a flexible body while the cam and shaft were defined as rigid bodies.

FEA Rocker Stiffness Prediction

Frictionless contact between roller, cam, rocker shaft, and rocker

Rocker shaft and cam are defined as rigid bodies

Force applied to tappet through valve stem contact along valve axis

𝐹𝐹𝐹𝐹Deflection measured

from nodes at the center of the contact radius

Fixed condition applied to rigid bodies

12


FEA Rocker Stiffness Prediction

𝐹𝐹

𝛿𝛿

𝜑𝜑

𝐾𝐾𝑟𝑟𝑟𝑟𝑟𝑟 =𝐹𝐹𝛿𝛿

cos𝜑𝜑

𝐾𝐾𝑟𝑟𝑟𝑟𝑟𝑟 = rocker stiffness𝐹𝐹 = force along valve axis (load axis)𝛿𝛿 = deflection along valve axis𝜑𝜑 = angle between valve axis and rocker perpendicular

roller

rocker shaft

contact center

• The red geometry annotations represent the 2D rocker element (as defined by GT)

• Actual rocker geometry is represented by the shaded green image.

• The rocker shaft is a fixed joint free to rotate and the roller is free to rotate and slide.

• The deflection values predicted by FEA are used to calculate the stiffness

The method used to predict rocker stiffness is consistent with GT’s rocker element definition.

13


FEA Rocker Support Stiffness Prediction

𝐹𝐹𝐹𝐹

Rocker components are modeled as rigid bodies.

All applicable contacts are enforced

Forces are applied to rigid body reference nodes in space

All relevant boundary conditions are applied

The rocker support stiffness was predicted independent of the rocker arm and includes all relevant supporting structure and boundary conditions. Rigid body rockers were used to load the assembly.

14


The boundary conditions include; bolt pretension, shaft clearances, all contact, fixed conditions at head bolts, and symmetry definitions.

FEA Rocker Support Stiffness PredictionSymmetry boundary condition to cut faces

Fixed boundary condition to faces of head bolt sleeves

Apply clearance contact to rocker shaft, cam, and journals

Apply pretension loads to all relevant bolts

15


The support stiffness prediction considered the contact footprint between the rocker to the shaft and was calculated in X & Y directions consistent with GT’s coordinate system.

FEA Rocker Support Stiffness Prediction

• Loading direction must be consistent with X and Y directions in the respective GT model.

• Predict stiffness in each direction separately.

• Apply load along an axis that is coincident with the rocker shaft (pivot) center.

• Using the rocker to load the support structure captures the contact footprint

FEA Vertical

FEA Horizontal

GT X-axis

GT Y-axis

𝐹𝐹𝑥𝑥 𝐹𝐹𝑦𝑦

𝛿𝛿𝑥𝑥𝐾𝐾𝑖𝑖 =

𝐹𝐹𝑖𝑖𝛿𝛿𝑖𝑖

𝐾𝐾𝑖𝑖 = pivot stiffness𝐹𝐹𝑖𝑖 = force along local axis𝛿𝛿𝑖𝑖 = deflection along local axis𝑖𝑖 = 𝑥𝑥, 𝑦𝑦

𝛿𝛿𝑦𝑦Rigid body reference nodes

16


An average stiffness for the rocker and its support was calculated by measuring respective deflections at predetermined force intervals.

Calculating Stiffness for GT

𝐹𝐹𝑖𝑖′

𝛿𝛿𝑖𝑖′

𝑦𝑦 = 𝑚𝑚𝑥𝑥 + 𝑏𝑏(linear fit)

𝑚𝑚 = 𝐾𝐾

𝑖𝑖 = 1,2,3, … ,𝑁𝑁

𝐹𝐹𝑖𝑖 = 𝑖𝑖 ∗𝐹𝐹𝑚𝑚𝑚𝑚𝑥𝑥𝑁𝑁

𝐹𝐹𝑖𝑖′ = 𝐹𝐹𝑖𝑖 − 𝐹𝐹1

𝛿𝛿𝑖𝑖′ = 𝛿𝛿𝑖𝑖 − 𝛿𝛿1

𝛿𝛿𝑖𝑖 = deflection predicted by FEA𝛿𝛿𝑖𝑖′ = shifted deflection predicted by FEA𝐹𝐹𝑖𝑖 = interval force𝐹𝐹𝑖𝑖′ = shifted interval force𝐹𝐹𝑚𝑚𝑚𝑚𝑥𝑥 = maximum force𝑖𝑖 = interval number𝐾𝐾 = calculated stiffness linear fit slope𝑁𝑁 = number of loading intervals integer

FEA Linear fit 𝐾𝐾

𝑖𝑖 = 1

𝑖𝑖 = 2

𝑖𝑖 = 𝑁𝑁

17


Low Valve Timing

High Valve Timing

Basic (w/o VTEC)

Key Model Input RequirementsRequirement VT Configuration

DependenciesMust Have

Should Have

CouldHave

Rocker Stiffness Basic (w/o VTEC) ●

Rocker Stiffness High Valve Timing ●

Rocker Stiffness Low Valve Timing ●

Support Stiffness (X&Y) Base (w/o VTEC) ●

Support Stiffness (X&Y) High Valve Timing ●

Support Stiffness (X&Y) Low Valve Timing ●

Cam Journal Stiffness (X&Y) Independent ●

Valve Spring Stiffness Independent ●

Valve Stiffness Independent ●

Valve Seat Stiffness Independent ●

Spring Retainer Stiffness Independent ●

Cam Journal Clearance Independent ●

Valve Guide Clearance Independent ●

Valve Lash Independent ●

The rocker and support stiffness are dependent on operating valve train configuration and are key model inputs. Additional key model inputs include; clearances, journal stiffness, valve lash, and component mass.

18


Intake High Tappet Force

Good correlation of magnitude and frequency was observed for the tappet force of the intake high rocker.

19

Crank Angle (deg)

Forc

e (N

)

Intake High Tappet Force at 4000 RPM

CH27:CYL4-A

CH28:CYL4-B

CH29:CYL6-A

GT

Crank Angle (deg)

Forc

e (N

)


Crank Angle (deg)

Forc

e (N

)


Crank Angle (deg)

Forc

e (N

)



Intake High Valve Lift

Good correlation of magnitude and closing response was observed for the valve lift of the intake high rocker.

20

Crank Angle (deg)

Lift

(mm

)

Intake High Valve Lift at 4000 RPM

CH99:CYL4-B

CH101:CYL6-A

GT

Crank Angle (deg)

Lift

(mm

)


Crank Angle (deg)

Lift

(mm

)


Crank Angle (deg)

Lift

(mm

)



Intake High Frequency Map

Frequency spectrum shows resonance correlation for the intake high rocker.

21


Intake Low Tappet Force

Good correlation of magnitude and frequency was observed for the tappet force of the intake low rocker.

22

Crank Angle (deg)

Forc

e (N

)

Intake Low Tappet Force at 4000 RPM

CH27:CYL4-A

CH28:CYL4-B

CH29:CYL6-A

GT

Crank Angle (deg)

Forc

e (N

)


Crank Angle (deg)

Forc

e (N

)


Crank Angle (deg)

Forc

e (N

)



Intake Low Valve Lift

Good correlation of magnitude and closing response was observed for the valve lift of the intake low rocker.

23

Crank Angle (deg)

Lift

(mm

)

Intake Low Valve Lift at 4000 RPM

CH99:CYL4-B

CH101:CYL6-A

GT

Crank Angle (deg)

Lift

(mm

)


Crank Angle (deg)

Lift

(mm

)


Crank Angle (deg)

Lift

(mm

)



Intake Low Frequency Map

Frequency spectrum shows resonance correlation for the intake low rocker.

24


Valve Train with Hydraulic Lash Adjuster

Fixed Rocker Pivot HLA Rocker Pivot

The methodology used for the previously explain fixed pivot system was expanded to include a system using a hydraulic lash adjuster (HLA) at the pivot.

25


Engine Oil Viscosity


𝜇𝜇𝑃𝑃 = dynamic viscosity (kg/m-s) 𝜇𝜇0 = dynamic viscosity at atmospheric pressure (kg/m-s)𝛼𝛼 = viscosity-pressure coefficient (Pa-1)𝑃𝑃 = pressure (Pa)

Barus viscosity-pressure formula: • Barus formula was used to calculate dynamic viscosity of oil under high pressure

• Barus formula is valid for pressures under 0.5 GPa

• α is unknown and can be a function of pressure and temperature

Pressure effects on oil viscosity must be considered. For this study, the Barus formula, with a calibrated viscosity-pressure coefficient, was applied. Along with small calibration of oil aeration and leak down pressure.

26


Key Model Inputs for HLA

Requirement Must Have

Should Have

CouldHave

Oil Temperature ●

Oil Bulk Modulus as a function of pressure and

temperature●

Oil Viscosity as a function of pressure and temperature ●

HLA Part Leak Down Time (per spec.) ●

HLA Part Leak Down Time asa function of applied

load/chamber pressure●

HLA Starting Height ●

• Oil Bulk Modulus as a function of pressure and temperature is known

• Oil Viscosity is known for temperatures between 40 and 150°C at atmospheric pressure

• Applied Barus Formula

• 𝛼𝛼 is unknown

• Calibration of 𝛼𝛼 provides HLA model accuracy for all engine speeds and operating temperatures

*All applicable inputs from a fixed pivot system must also be applied

Oil properties become a key model input for valve train systems that include a HLA. Other key inputs include; the leak down time and starting height of the HLA

27


HLA Cyclic Test: LDT & Temp. at 1000 RPM (Cam)

MeasuredGT

Cyclic test results were used calibrate oil pressure-viscosity coefficient. HLA behavior correlates well for multiple leak down times (LDT) and oil temperatures (60 and 120°C). Verifying the model for LDT and temperature.

28

LDT = 2.20s

LDT = 0.45s

LDT = 0.20s

HLA Hysteresis Overlay at 60C HLA Hysteresis Overlay at 120C




HLA Cyclic Test: 1000-8000 RPM (Cam)

Simulation and test results diverge as engine speed increases, but correlation remains good between 1000 and 4000 RPM. This verifies the HLA model over a range of engine speeds.

29


Tappet Force – Rocker Strain w/HLA Pivot

Oil Temperature = 120˚C

Using a limited data set, the frequency of the predicted tappet force was correlated to measured strain.

30

Crank Angle (deg)

Forc

e (N

)

Mic

rost

rain

Tappet Force - Rocker Strain 5000 RPM

GT

Measured


Valve Lift w/HLA Pivot

Oil Temperature = 120˚C

Good correlation was observed for valve lift magnitude and closing response.

31

Crank Angle (deg)

Lift

(mm

)

Valve Lift at 5000 RPM

Measured

GT

Crank Angle (deg)

Lift

(mm

)


Crank Angle (deg)

Lift

(mm

)


Crank Angle (deg)

Lift

(mm

)



Achievement GT-SUITE accurately predicts dynamic responses for modeled valve train systems

Modeling methodology validated via correlation of predicted and measured valve train system responses Tappet force Valve lift Valve spring response

Modeling methods are valid for multiple valve train configurations and engine operating conditions High and low valve timing With and without HLA High and low temperature High and low engine speed

32


Questions 33


Appendix 34


Calculating Tappet Force

IntakeSlope = 0.002361969

• Measured strain from rocker arms was used to calculate tappet force

• Slope was used to scale measured strain to calculate force

Component level testing of rocker arms was used to determine a calibration factor to convert measured strain to tappet force.

35

ExhaustSlope = 0.001843369


Raw Information

Intermediate Model

Direct GT Input

Optional GT Input



Kinematic VT Layout

Material Properties

Part Stiffness

Valve Train

Design & Layout


Mass Properties

3D Finite Element Analysis

Measured


36


Exhaust Spring Stress

The frequency of measured spring strain and predicted spring shear stress shows good correlation in the exhaust springs for multiple engine speeds.

37


Intake Spring Stress

The frequency of measured spring strain and predicted spring shear stress shows good correlation in the intake springs for multiple engine speeds.

38


Exhaust Tappet Force

Good correlation of magnitude and frequency was observed for the tappet force of the basic exhaust rocker.

39

Crank Angle (deg)

Forc

e (N

)

Exhaust Tappet Force at 4000 RPM

CH20:CYL4-A

CH21:CYL6-A

CH22:CYL6-B

GT

Crank Angle (deg)

Forc

e (N

)


Crank Angle (deg)

Forc

e (N

)


Crank Angle (deg)

Forc

e (N

)



Exhaust Valve Lift

Good correlation of magnitude and closing response was observed for the valve lift of the basic exhaust rocker.

40

Crank Angle (deg)

Lift

(mm

)

Exhaust Valve Lift at 4000 RPM

CH97:CYL4-A

CH100:CYL6-B

GT

Crank Angle (deg)

Lift

(mm

)


Crank Angle (deg)

Lift

(mm

)


Crank Angle (deg)

Lift

(mm

)



Exhaust Frequency Map

Frequency spectrum shows resonance correlation for the basic exhaust rocker.

41


Next Steps 42

Use current models to support and drive design

Increase development support

Build confidence in model predictions

Expand modeling capabilities

Preform further validation

Predicting Valve Train Dynamics using Simulation … American GT Conference 2017 Predicting Valve...

Documents

Transcript of Predicting Valve Train Dynamics using Simulation … American GT Conference 2017 Predicting Valve...