driving next generation powertrain nvh refinement thru...

Driving the next generation of Powertrain NVH Refinement through Virtual Design

T. Abe1, M. J. Felice

2

1 Ford Motor Company

2400 Village Road, Dearborn, MI 48124 USA

email: [email protected]

2 Ford Motor Company

2400 Village Road, Dearborn, MI 48124 USA

email: [email protected]

Abstract

The drastic fluctuations of fuel prices as well as the stringent Government requirements in emissions are

driving a technology explosion in automotive powertrain design. These include diversification of

powertrain technologies consisting of hybrid electric motors, clean quiet Diesels and small efficient

gasoline Direct Injection Turbo Boosted engines to list a few. Furthermore, the traditional automatic

transmission is also being replaced with lighter and more fuel efficient DCT (Dual Clutch Transmissions).

To achieve these fuel efficiency requirements, the OEM’s need to drastically reduce weight and add a

great deal of technical content to satisfy customer expectations for NVH refinement while maintaining the

same cost structure.

The challenge to the Powertrain NVH engineer is GREAT! A deeper understanding of the NVH

phenomena is required, as well as an advanced virtual design process that includes “state-of-the-art”

multi-physics analytical methods that can evaluate the operating dynamics and vibro-acoustic response of

such high efficient powertrains. For example, one of Ford’s critical drivers for achieving high volume

fuel efficient powertrains are EcoBoost® engines with high efficiency transmissions. Such powertrains

present great challenges for NVH and require advanced CAE assessment methods. These include high

frequency impulsive noise from the engine due to high pressure fuel rail injector system, complex

valvetrains and cam-drive systems to name a few. While for transmissions, the introduction of DCT can

result in higher transmission gear rattle.

This technical paper will present the application of these advanced CAE methods used in the development

of our new small Gas Turbo Direct Injection Eco-Boost engines and DCT transmissions. These new

powertrains have achieved impressive levels of quietness and smoothness. The contents of the

presentation will detail analytical methods for powertrain structural NVH design, as well as Air Induction

& Exhaust system acoustics analysis for achieving best Sound Quality performance.

1 Introduction

The volatility of gasoline prices in recent years has led US consumers to an unprecedented shift in the

vehicles they buy and drive. Figure 1.1, illustrates the fluctuation of gasoline prices over the last few

years causing a significant change in customer vehicle buying preference. This fact, compounded with the

aggressive CO2 emission requirements imposed by US and EU governments are driving the automotive

OEMs to redefine their powertrain strategies and technologies for the next decade. Furthermore, these

new technologies have to be delivered at very competitive costs without sacrificing power performance,

vehicle comfort and NVH refinement.

4275

Figure 1.1: US Gasoline Prices compared to Vehicle Segment Sales over the last few years

As shown in Figure 1.2, there is a vast selection of power units being considered. These range from new

clean and efficient Diesel Engines, to Hydrogen fueled engines, to the use of Fuel Cells, to Battery driven

Electric Motors as well as Hybrid driven powertrains, and most importantly to more efficient direct

injected turbo boosted gasoline engines.

Figure 1.2: Power Units and Fuel Options considered to deliver fuel efficiency and emission requirements

The fuel options are also many as shown in Figure 1.2. From conventional Fossil fuels which feed

gasoline and diesel engines to a great range of electric alternatives as well as range of renewal bio-fuels

U.S. RETAIL SEGMENTATION

15.8%

24.6%

15.4%

16.8%

13.9%

12.2%

11.0%

9.0%7.0%

6.4%5.7%

4.7%3.8%

19.3%

16.9%16.3%

16.8%

21.1%

18.4%

15.0%

13.3%

14.2%

12.9% 13.3%12.3%

14.0%14.7%

13.2% 15.2%

8.6%

7.1%7.4%

Small C-Car

Medium Car

Full-Size Pickup

Medium/Large Utility

Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May*

ComVeRSE Estimates

Small C

ars

Small C

ars

Med CarsMed Cars

Med & Large Utility

Med & Large Utility VehVeh

Full Size Cars &

Full Size Cars & PickUpsPickUps

US

CUSTOMER

SHIFT

From Full Size

SUV’s/Pick-Up

to Small &

Medium Cars

Power Unit ChoicesPower Unit ChoicesPower Unit Choices

Hydrogen ICE

FuelCell

Battery

Electric

Hybrid

GasolineEngine

SterlingEngine

SteamEngine

Diesel

Atmospheric CO2

RenewablesRenewables

Solar

Hydro

Coal

Wind

Petroleum

Nuclear

FossilHydrocarbon

FossilHydrocarbon

ElectricityElectricity

Corn

Biomass

Biomass

Fuel OptionsFuel OptionsFuel Options

4276 PROCEEDINGS OF ISMA2010 INCLUDING USD2010

used in flex fuel engines. Obviously, each of these fuels resulting in different levels of CO2 emission

impact to the environment.

The introduction of new more efficient powertrain hardware technologies (e.g., smaller Direct Injection

Turbo Boosted gasoline engines, Twin Clutch Transmissions, Hybrid Electric power units, etc.) also

introduce new NVH error states mainly having to do with aggressive combustion and high frequency

mechanical noise. These NVH error states are further compounded by significant efforts of weight

reduction to meet the aggressive fuel economy needs. These error states present great challenges to the

NVH engineer and need to be evaluated early on in the development process to insure design robustness

later in the hardware validation phase.

Next, we will be discussing some of the NVH challenges inherent to these new power unit technologies.

Ford's main strategy is to downsize its engines to deliver best fuel efficiency with equivalent power to the

larger older engines being replaced. This is done through the introduction of a family of "EcoBoost®"

gasoline twin turbo direct injection engines. However, these engines result in more aggressive

combustion excitation than conventional gasoline engines; impulsive noise such as injector tick, high

pressure pump and valvetrain tick; turbo related noise (e.g., moan, synchronous whine, tip-in/out noise,

and sub-synchronous noise); as well as torsional induced NVH issues (e.g., lugging boom, powertrain

moan, etc).

Figure 1.3: Conventional Auto Transmission compared to more fuel efficient Twin Clutch Transmission

Dual Clutch transmission technology greatly improves fuel economy and CO2 emissions from reduced

parasitic losses thru elimination of torque converter, and use of synchronizers instead of shift clutches.

Also, they are much lighter than conventional automatic transmission. However it presents a number of

NVH challenges (e.g., aggressive transients from elimination of torque converter resulting in potential

gear rattle, shuffle, and shift quality error states)

Hybrid Electric Vehicle also present NVH challenges such as: engine start/stop noise and vibration,

power-split system gear whine, power control unit high frequency switching noise, and high frequency

electric motor generator noise.

Finally, increased vehicle light weight material applications such as plastics and aluminum structures

along with the migration from Body-on-Frame to Uni-Body structures are key actions to further improve

fuel economy. These pose new NVH challenges such as road noise and powertrain noise as they become

more noticeable.

CAE simulation Methods for these new powertrain technologies are critical for understanding and

evaluating potential NVH Error States and Failure Modes prior to hardware build. The rest of this paper

will present the various analytical methods established at Ford to evaluate these error states to drive

upfront design and NVH refinement.

High Efficiency TransmissionsHigh Efficiency TransmissionsHigh Efficiency Transmissions

Conventional Automatic Transmission with Torque Converter

Dual Clutch Transmission

VEHICLE NOISE AND VIBRATION (NVH) 4277

2 Turbo DI engine NVH Challege

2.1 Engine down-sizing strategy with DI Turbo Charged Engines

As previously stated the introduction of new more efficient power units such as smaller Direct Injection

Turbo Boosted gasoline engines are normally inherent to high frequency fuel system noise. A simulation

process, as shown in Figure 2.1 is used at Ford Motor Company to evaluate the entire engine fuel system

excitation mechanism from a source, path, and receiver aspect. This process uses a multi-physics

approach starting with a 1-D fluid dynamic computation of the fuel system that includes the hydraulic

network, high pressure fuel pump and injector system. Fluid dynamic pressures along with the dynamic

excitations from the injectors are computed and applied to the engine structure Finite Element (FE) model.

Forced response FE analysis is performed to evaluate the vibrational response (i.e., surface velocities) of

the entire engine structure. Boundary Element Acoustic analysis is then conducted using the FE computed

surface velocities to determine the Sound Pressure levels of the engine due to fuel system excitation.

Figure 2.1: Multi-physics Fuel System Noise analysis process

LMS/AMESim is used to model the complete fuel injector hydraulic network to be able to compute

internal pressure pulsation. The model consists of the fuel high pressure pump, engine control unit, high

pressure lines, fuel rails and all injectors. The outputs from the Fuel System Model include:

A schematic of the hydraulic network model is included in Figure 2.2 as well the correlation data for Fuel

Rail hydraulic resonances between test data in (Blue Curve lines) and simulation (Green Curve lines).

The computed resonant frequency peaks correlate fairly well, however the amplitudes levels are higher

due to the complex fluid damping characteristics which the model still can not capture accurately.

AMEsimAMEsim ModelsModels

FEA ModelsFEA Models

Coupled to FSI of Rails

11111111

22222222

33333333 44444444

55555555

Fuel System Noise SimulationFuel System Noise SimulationFuel System Noise SimulationFuel System Noise SimulationFuel System Noise SimulationFuel System Noise SimulationFuel System Noise SimulationFuel System Noise Simulation

Input

Input

FEA Forced Response

Overall Engine

Vibro-Acoustic Response

Jemai Missaoui

D 3 5 G D I G e n 2 P P - I d l e R a d i a t e d N o i s e

1 m 4 M i c A v g

- 1 0

- 8

- 6

- 4

- 2

0

2

4

6

8

1 0

1 0 0 0 2 0 0 0 3 0 0 0 4 0 0 0 5 0 0 0 6 0 0 0 7 0 0 0 8 0 0 0 9 0 0 0 1 0 0 0 0

F r e q u e n c y H zC

rite

ria

- S

tatu

s L

ev

el

dB

(A)

Sound: PP Top MicCAE Process combinesCAE Process combines

• 1D Fluid Dynamics (AMEsim)

• 3D Fluid Structure Interaction

• 3D FEA (MSC/Nastran)

• BEM Acoustics (Virtual Lab)1D Injection System

Injector Pump

LMS Imagine.Lab(AMEsim)

Hydraulic Network


Figure 2.2: 1D Fuel System Hydraulic Network Model with Correlation of Hydraulic Resonances

As stated above, these computed fuel system hydraulic excitations along with the injector opening impact

excitations are applied to FE models to compute the engine surface vibration velocities. Acoustic analysis

is then performed by dividing the engine into multiple panels to pin point which area of the engine is most

sensitive to fuel system excitation and consequently radiated noise. Figure 2.3 shows this process

indicating panels 10 and 11, which are the valve covers and engine valley with fuel rails being the most

active. The analysis data is displayed using panel contribution charts as well as narrow band and 3rd

octave band plots to further understand the data. This analysis is critical to determine counter-measures

for added structural ribs and/or isolation to better attenuate fuel system noise.

2.2 Fuel System Impulsive Noise

Injector Tick Source

There are several noise paths generated from the direct injection engine fuel system. These include; the

high pressure pump interface with its mounting structure which is typically the cylinder head, the fuel rail

to cylinder head connection, and the injector contact with the cylinder head as it cycles from opening to

closing. These interfaces need to be analyzed carefully to understand the high impact energy transmitted

to the engine that ultimately results in high frequency tick noise. Direct injection engines typically result

in much higher fuel pressures than traditional port fuel injection systems since the injectors are directly

mounted to the cylinder heads. Test data shows that significant ticking noise occurs at injector opening

and closing due to impacts between the armature and stopper and the injector's needle and seat. The often

objectionable tick noise results from the structure-born and fluid-born excitations that transmit to the

cylinder heads and travel through other engine components, e.g. the engine block, oil pan, cam covers,

front cover, and the intake and exhaust manifolds. Figure 2.4 illustrates how higher hydraulic fuel

pressure and higher mechanical forces in the injector can lead to high frequency impact noise. As the

Complete Injection System

HP

Pump

Injectors

HP Line

Fuel Rails

Cross over

Line

ECU

Fuel Rail Hydraulic

Resonances

Red = CAE Blue=Test

6

Compare AMESim model to Spin Rig

Confidential | GS-FI/ENG2-NA | 11/29/2007 | © 2007 Robert Bosch LLC and affiliates. All rights reserved.

Gasoline Systems

FFT Ford Cyclone

620 RPM 20 bar 1 ms PW

XO mod Hardware

0.0001

0.001

0.01

0.1

1

0 1000 2000 3000 4000 5000 6000

Frequency (Hz)

Pre

ssu

re (

ba

r)

Model - Rail 123 Rail 123 -3 Revised model VK Rig - Rail 123-3

Rail 1-2-31660 Hz1660 Hz 2200 Hz2200 Hz

Fuel Rail Hydraulic Frequencies

Freq (Hz)

Fuel Rail Hydraulic Resonances (TestTest vs. CAECAE)


injector opens and closes, internal dynamic forces (Fdynamic) are transferred to and amplified by the cylinder

head structure.

Figure 2.3: Typical Analysis Results for full Fuel System Engine Model

FDynamic

Teflon

Seal

Isolator

FHydraulic

Injector

Injector

Motion

Cylinder

Head

Figure 2.4: Injector Tick Source

Minimizing Injector Tick – Injector Isolation CAE Lead Design

One method for minimizing injector tick is to provide isolation between the fuel injector and the cylinder

as shown in Figure 2.4. The design of the isolator can be very changeling; some of the obstacles in

designing and implementing a suitable isolator are listed below:

Panel Contribution Analysis� Top microphone (0.5m)

11109

87

6

54

321

Total

11109

87

6

54

321

Total1

23

78

10

1

23

78

10

45

69

11

45

69

11

Panel 11 engine valley

+ fuel rails

Panel 10 valve covers

Panel Contribution to Acoustic ResponsePanel Contribution to Acoustic Response

Fro

nt

of

En

gin

eF

ron

t o

f E

ng

ine

Re

ar

of

En

gin

eR

ea

r o

f E

ng

ine

Acoustic mesh divided into a total of 11 panels

Frequency (Hz)Frequency (Hz)

Most of the Injector Fuel System Noise comes out of the

valve covers and fuel rails in the engine valley


• Limited package space

• Assembly concerns

• Hostile environment – high temperatures

• Conflicting NVH and durability requirements

The focus of our discussion is determining an isolator design that meets the NVH and durability

requirements. Simply stated, NVH desires a soft isolator to absorb the transmitted force while durability

requires a stiff isolator to limit injector operational movement. Limited injector motion is needed to

guarantee the life of the Teflon seal between the injector and cylinder head. The generation of injector

isolator designs to satisfy the NVH needs and durability requirements relied heavily on CAE capabilities.

Using non-linear FEA, the team developed numerous designs that meet NVH needs and durability

requirement while satisfying the package space and assembly concerns. Figure 2.5 displays a few

examples of the developed isolators.

(a) Conical Isolator (b) Flat Isolator

Figure 2.5: Sample Isolator Geometries

CAE was not only used to develop nominal designs, but was heavily leveraged in determining the injector

isolator manufacturing process. The parametric modeling allowed the use of Monte Carlo techniques to

determine the effects of dimensional variation in proposed injector isolator designs. In this case, limits on

various isolator dimensions were determined in CAE prior to prototype manufacturing. The required

precision was found and an appropriate manufacturing process was selected to achieve robust

performance.

Injector Isolation – Physical Results

Figure 2.6 demonstrates the reduction in cylinder head impact force (directly related to fuel system

impulsive noise) when incorporating injector isolators. The results in Figure 2.6 are based on by injector

isolators designed using the aforementioned CAE methodology. Again, CAE was extensively used in

prototype design synthesis, baseline design iteration, manufacturing process selection and reliability and

robustness assessment.

2.3 Torsional Vibration Induced Vehicle NVH

The engine torque fluctuation (AC torque) is a prime source of excitation in an internal combustion engine

(ICE) that greatly influences the vehicle NVH responses. Vehicle NVH concerns due to the engine torque

signature has become even more challenging with smaller high boosted direct injection engines. This

technology for example allows downsizing an engine from a V6 to an I4 with the use of turbo charging

and direct fuel injection to achieve much better fuel economy without sacrificing the power of the larger

engine that is replacing.


Figure 2.6: Injector Isolator Results

Engine torque signature is the resultant of combustion pulses and dynamics of cranktrain, valvetrain, and

cam drive assemblies. The output cyclic variation from combustion and engine dynamic events introduces

torque fluctuation which can excite the transmission and driveline torsional vibrations. It is critical to

analytically accurately predict engine brake torque upfront in the design process to optimize the

transmission clutch damper such that these torsional excitations are greatly reduced.

Torque Prediction

Predicting transient behavior due to torque fluctuation requires not only accurate measured combustion

pressures of each cylinder, but the ability to capture instantaneous friction losses in the system as well.

The calculated instantaneous friction losses presented below are based on Rezeka and Henein [1] model.

The instantaneous friction, inertia, gas pressure and load torques are solved through series of equations of

motion. By calculating the loads with measured cylinder pressures, crank inertia and speed, the total

friction torque can then be calculated, see the Figure 2.7.

Fig: 2.7: Predicted Instantaneous Friction Torque (N.m)

0

50

100

150

200

250

300

0 90 180 270 360 450 540 630 720

Crank Angle (deg)

Fri

cti

on

To

rqu

e (

Nm

)

Piston Skirt

Valve train

Crank Main/ ConRod

Big End/Piston Pin

Bearings

Fuel injection

Cam drive

Oil pump

Turbo

Auxiliaries

Viscous/Mixed Ring Lubrication

∑

Fuel pump

Gear tooth rolling and sliding

contact

Instantaneous Friction Torque

0

50

100

150

200

250

300

0 90 180 270 360 450 540 630 720

Crank Angle (deg)

Fri

cti

on

To

rqu

e (

Nm

)

Piston Skirt

Valve train

Crank Main/ ConRod

Big End/Piston Pin

Bearings

Fuel injection

Cam drive

Oil pump

Turbo

Auxiliaries

Viscous/Mixed Ring Lubrication

∑

Fuel pump

Gear tooth rolling and sliding

contact

Instantaneous Friction Torque

Cylinder Head Impact Force (No Isolator) Cylinder Head Impact Force (With Isolator)


Fig: 2.8 Predicted Instantaneous Friction Torque (N.m)

Torsional system model

A powertrain torsional CAE model is fundamental for understanding the root causes for torsional

vibration induced NVH error states. In general, a powertrain torsional CAE model includes the engine,

transmission and driveline as a complete systems, see the Figure 2.9. The excitations are calculated using

AVL/Excite from firing cylinder pressures and cranktrain mass inertias that include crankshaft, piston

assemblies, and connecting rods. These excitations are then combined with the calculated instantaneous

friction torque shown in Figure 2.7 to compute the final torque at the transmission flywheel. Figure 2.8

shows correlation of simulated flywheel torque to measurement. Accurate prediction of the engine torque

fluctuation is a critical input to the analysis of transmission and/or driveline rattle.

Fig: 2.9 Powertrain Torsional Model

3 Transmission NVH Challenges

As already mentioned, downsizing the engine coupled with high demanding for torque capability provides

additional challenges for the transmission due to the high AC engine torque fluctuation. This is even more

critical for Manual Transmissions (MT) and Dual Clutch Transmissions (DCT) because it can result in

Sim

ula

ted D

C L

eve

l

AC Torque simulated 4th order


transmission gear rattle, which is especially the case for smaller engines. Transmission clutch design as

well dual mass flywheels are sometimes used to reduce these engine torque fluctuations.

Gear rattle noise is one of the main NVH challenges in the automotive transmissions, along with gear

whine and pump noise, and occurs in a broad range of frequencies. Its dynamic behavior is strong non-

linear due to the existence of multi-gear and spline lashes, as well as non-linear characteristics of the

clutch. Since gear backlash is designed to compensate for manufacturing tolerance and operating heat

distortion, it can cause the unloaded (or loose) gears are free to bounce in the backlash zone with the

constraint of oil drag torque only. High engine torque fluctuation could result in loss of contact between

the transmission driving and driven (loose) gears, thus teeth impact may occur and can range from

intermittent or continuous with single or doubled sided teeth impact. The impact forces propagates

through the shafts, to the bearings and finally to the casing structure. Although small level of noise is

directly radiated from the impact of gear teeth themselves, the majority of rattle noise is radiated from the

transmission case structure. Upfront CAE simulation is important to evaluate gear rattle. Simulation

usually includes the full transmission model where all gears, shafting and drag torque distribution are

included.

3.1 Simulation Approach

There are various simulation techniques ranging from 1-dimemsional model (only rotational properties are

involved such as inertia, mesh stiffness, angular velocity and acceleration) to more complicated 3-

dimensional MBD model, where the explicit gear teeth are represented by detailed gear design parameters

including profile and lead modification, along with flexible shafts with bending and torsional dynamic

characteristics.

For drive rattle, the driving gears are loaded. Therefore the prescribed static and geometric backlash will

be altered because of the shafts deflection caused by the applied torque in the operating condition.

Moreover, in a speed sweep case, the system could go through certain shafting system resonances,

especially in lateral direction which would significantly change the teeth impact mechanism due to the

"dynamic backlash". The 3-D MBD approach offers advantages in these areas. Figure 3.1 shows

examples of Manual and DCT transmission models.

Figure 3.1: Manual Transmission Modeling

MT

DCT


3.2 Rattle Mechanism Diagnosis

It is well known that when the inertia torque is larger than the drag torque Tdrag on a loose gear, i.e., index

1≥β

Where J is inertia and looseθ&& is the angular acceleration of the loose gear. The rattle could occur in the form

of single or doubled impacts, as shown in Figure 3.2.

Figure 3.2: Single and double impacts

To validate the model, the measured engine torque on flywheel is inputted to the clutch and the predicted

transmission shaft angular velocity is compared with the measured data. Figure 3.3 shows a good

correlation between the simulation and the measurement. The clutch spring rate and hysteresis were

included in the analysis.

-600.0

-400.0

-200.0

0.0

200.0

400.0

600.0

800.0

0.00 0.50 1.00 1.50 2.00

time (s)

flyw

heel

Test

(rad

/s^

2)

Figure 3.3: Measured and predicted transmission input shaft velocity

The transition from a single to double impact depends on the drag torque level and engine excitation

magnitude as well as the system dynamics. The idea is to avoid the onset of double impacting, although

one should not guarantee that the single impact would not be a concern. Figure 3.4 indicates that with the

engine dynamic angular acceleration increasing, the impact will gradually become double impact, which

could be regarded as the rattle threshold. The predicted "threshold" accurately matched with the field test.

drag

loose

T

J

TqDrag

TqInertia θβ

&&∗==

Time (sec)

Rel

ativ

e d

ispla

cem

ent

w/i

n c

lear

ance

Single impact Double impact


Engine Excitation Level Engine Excitation Level

Figure 3.4: (A) Impact transition from single impact to double impact with increasing engine excitation

(B) Impact forces for single mass flywheel (SMF) and dual mass flywheel (DMF)

4 AIS / Exhaust System NVH

4.1 Turbo AIS Development

Turbo-charged systems are a key enabler in delivering significant fuel economy improvement while

maintaining performance that customers have grown to expect. Such systems come with significant

challenges including scope of unique NVH phenomena as well as application of traditional CAE tools.

Figure 4.1 illustrates various turbo related NVH phenomena such as whine, whoosh, blade passing noise,

and tonal resonances related to the low and high pressure ducting. As with naturally aspirate engines, the

order based engine harmonic content is present at the lower frequencies.

Inlet Orifice Noise / Sound Quality

Albeit more challenging in boosted systems, the same 1-D engine simulation tools used for naturally

aspirated engines are used to predict ordered orifice noise content. Typical industry standard applications

are Ricardo Wave and GT-Power.

Figure 4.1: Turbo Related NVH Phenomena

Impac

t F

orc

e

-----SMF

------DMF

Single impact Double impact

(A) (B)

Impac

t F

orc

e


Figure 4.2 shows test data and predictions for inlet orifice noise a the engine combustion order. One is a

direct injected turbo boosted version, while the other is a naturally aspirated direct injected system. Both

systems use the same dirty side duct and air cleaner box configurations, but have different clean side

ducting owing to their routing to the unique engines. Immediately evident is the much lower inlet noise

levels of the boosted system. This is fairly typical of turbo charged systems since the turbo itself tends to

"filter" some of the acoustic content from the engine harmonics as they work their way towards the inlet.

Although the inlet noise levels in boosted applications can be lower relative to similar naturally aspirated

systems, there is still effort to avoid lower frequency booms or moans. Robust upfront engineering

involves proper design of the ducting and air box or the addition of resonators when needed.

Depending on the sound quality strategy, lower inlet noise levels can make it challenging to deliver

targeted experience to the customer. To that end, some in the industry have utilized other means to

provide desired engine presence such as added sound generator devices and active noise control.

Figure 4.2: Inlet Noise: Test Data vs CAE Predictions

Counter-Measures for Turbo Related Error States

In addition to traditional AIS design consideration other counter-measures are utilized to avoid some of

the error states as circled in red in Figure 4.1. One such counter-measure is an in-line resonator place just

upstream of the turbo compressor in the hot charged duct to reduce the synchronous pressure pulsations

created by the turbo. Reducing the in-duct spectral content in the frequency range of about 1500 to 3500

Hz is helpful in reducing the risk of radiated noise from the components downstream of the turbo. Figure

4.3 shows the test data and prediction of the acoustic performance of an integrated resonator design that

provides attenuation in the targeted frequency band.

Figure 4.3: Transmission Loss of Integrated Hot Charged Duct Turbo Resonator

Inlet Noise at engine combustion harmonic order

10 dB


4.2 Low Exhaust Back Pressure Challenge – Rasp / Weak Shock Formation

While the trend is to reduce engine displacement for fuel economy, the need for power remains.

Subsequently the power density (power per liter of displacement) of newer advanced engines is on the

rise. Additionally, aggressive fuel economy improvements are resulting in significant reduction of exhaust

back pressure [6]. One of the interesting challenges that this presents is the increased risk of exhaust rasp.

Rasp is a sharp, impulsive and potentially metallic sound that emanates from the tail pipe (orifice noise) or

exhaust structure (mechanical excitation) or both. In the absence of mechanical excitations, the sharp and

impulsive orifice noise, which has been described as similar to the 'blatty' sound of a trombone played

double forte, is due to the presence of higher order harmonics. These higher order harmonics are caused

by the onset of weak shock waves in the exhaust pipe(s) while the exhaust pulses travel from the cylinder

head to the tailpipe.

The shock wave forms due to wave steepening phenomena. Wave steepening occurs due to nonlinear

terms in the governing equations that cause the wave propagation speed to vary with pressure amplitude.

The propagation velocity differences mean that a pressure peak will travel faster than the preceding

pressure trough up until the wave steepens into a shock (see Figure 4.4). If the wave steepens into a shock

before encountering a large expansion (muffler or tailpipe exit), rasp is likely.

Figure 4.4: Rasp Root Cause: Wave Steepening

For a simplified case, nonlinear wave steepening can be treated analytically. Considering the leading order

nonlinear terms, an proportionality relationship for the distance a sine wave requires to transform into a

weak shock (∆x ) may be expressed as

pf

pcx oo

s∆

∆ ~ (1)

where c is the speed of sound, p is the pressure, ∆p is the crest-to-trough pressure difference, f is the

frequency and the subscript o refers to the mean value. For the steep, non-sinusoidal pulsations in an

exhaust system, the frequency in Equation 1 may be taken to represent initial steepness of the pulse, rather

than the engine firing frequency. There is a risky balance between rasp and higher power density engines.

In order to mitigate this risk, when it cannot be avoided through upfront design, known counter-measures

such as expansion volumes, resonators, and bank-to-bank communication can be employed when

appropriate.

CAE methods have been utilized to help assess the risk of weak shock formation. Figure 4.5 shows a

correlation between test and prediction of an in-duct pressure transfer downstream in the exhaust system

after a weak shock has had a chance to form.


Figure 4.5: Downstream Pressure Transducer: Test vs. Prediction

5 Required CAE Capability Enhancement

Automotive powertrain NVH CAE capabilities have mainly been limited to a frequency range of up to 3K

Hz. Because of the newer lighter, more efficient and power packed powertrain technologies it is crucial to

evaluate beyond the frequency range of 3K Hz to properly capture the NVH error states previously

discussed. Hence, acoustic CAE capabilities must address the low, mid and high frequency ranges within

a practical solution time to be able to drive design robustness for NVH upfront in the development

process.

As presented in section 2 of this paper, one of the major NVH challenges is high frequency impulsive

noise from smaller direct injected engines. High frequency acoustic predictive capabilities are still

evolving, but one that is most promising is the fast multipole BEM method. This method is available in

the LMS/Acoustics software package. The fast multipole BEM implements high-speed iterative

techniques to solve the BEM equations with sophisticated algorithms based on multipole expansion and

multi-level hierarchical cell sub-structuring. This method reduces the computational complexity and the

memory requirements compared to standard BEM. It drastically accelerates an iterative solution of large

scale linear system without the dense influence coefficient matrices used for the conventional BEM

[2,3,4,5]. Furthermore, efficiency is further improved by using this method in combination with acoustic

transfer vector technology when running powerplant simulation models with multiple rpm speeds and load

conditions up to 10,000 Hz.

However, to achieve accurate high frequency acoustic prediction, it is important to have accurate FEA

structural vibrations up to 15 KHz. Hence, at Ford we have spent a significant effort to improve our FEA

modeling capabilities to these frequency levels. Powertrain Modal and Frequency Response (FRF) CAE

is correlated up to 15 KHz on a component, sub-system and system level. Figure 5.1, shows correlation

for the engine fuel rail as an example of reasonably good FRF and Sound Pressure correlation between

CAE and test data up to 15 KHz.


(a) Vibro-acoustic model setup (b) FRF correlation up to 15 KHz (c) SPL correlation up to 10 KHz

Figure 5.1

Further ongoing refinement of this new advanced virtual design process includes the Transient BEM

method (TBEM). The main benefit of this method is that it solves directly in the time domain which is

more suitable for impact/transient phenomena. Figure 5.2, shows an example of TBEM application in

automotive powertrain for high frequency impulsive noise of high pressure fuel system rail.

Figure 5.2: Transient Sound Pressure color map

In summary, the new virtual design process enables Ford to efficiently optimize automotive powertrain for

NVH in the low, mid, and high frequency ranges using advanced CAE structural methods, conventional

BEM, transient BEM and Fast multipole BEM methods.

6 Summary and Conclusions

Due to global requirements for stringent emissions and better fuel efficiencies for automobiles,

significantly diversified power units need to be developed and implemented in a very short period of time.

Up-front design optimization of power train to simultaneously meet requirements of fuel economy and

customer comfort such as NVH becomes the most critical capability for development. Full utilization of

the newest CAE methodologies is a key enabler of up-front powertrain design optimization. This paper

summarized major examples of CAE-led NVH design optimization for the new generation of Ford

powertrains, such as turbo-charged DI gasoline engines (air induction/exhaust systems, fuel systems), new

transmission (DCT) and torsional system NVH. This facilitates NVH development much faster with

significant improvement of initial design quality of hardware systems.

Although these CAE methodologies are powerful tools to enable the powertrain design more up-front and

efficient, significant advancement of CAE capabilities are desired to cope with new NVH challenges of


new powertrain systems, specifically significant expansion of high frequency analyses limit, non-

stationary (transient vibration/impulsive noise) analyses capabilities, and multi-physics analyses

capabilities such as Fluid-Structure Interaction analyses.

Acknowledgements

The authors would like to acknowledge the team at Ford for their great contribution to this paper: The

team includes: Giueseppe DeRose, Hassan Nehme, Bin Juang, Yuping Cheng, Salah Hanim, Fumin Pan,

Aamir Marvi, Ed Hernandez, Nolan Dickey, Brian Schabel and Lloyd Bozzi.

References

[1] Reseka, S. F. and Henein, N. A. A New Approach to Evaluate Instantaneous Friction and its

Components in Internal Combustion Engines. SAE paper 840179, 1984, pp. 1932-1943

[2] Nishimura, N., "Fast Multipole Accelerated Boundary Integral Equation Methods," Applied

Mechanics Reviews, Vol. 55, 2002, No. 4, pp. 299-324.

[3] Wolf, W. R., and Lele, S. K., "Fast Multipole Boundary Element Method for Sound Scattering from

Aerodynamic Bodies," Proceedings of the 14th AIAA/CEAS Aeroacoustics Conference, AIAA Paper

2008-2872, 2008, pp. 1-22.

[4] Rokhlin, V., "Rapid Solution of Integral Equations of Scattering Theory in Two Dimensions," Journal

of Computational Physics, Vol. 86, 1990, pp. 414-439.

[5] Darve, E., "The Fast Multipole Method: Numerical Implementation," Journal of Computational

Physics, Vol. 160, 2000, pp. 195-240

[6] Abe, T. and Okada, M., "Study of Generation Mechanism for Abnormal Exhaust Noise" SAE paper

871924 (1987)


driving next generation powertrain nvh refinement thru...

Documents