Class 8 Truck External Aerodynamicsmdx2.plm.automation.siemens.com/sites/default/files/...Class 8 Truck External Aerodynamics Choice of Numerical Methods Security Classification Line1

Class 8 Truck External AerodynamicsChoice of Numerical Methods

1Security Classification Line

PVE Vehicle Analysis

Portland, March, 19th 2013

Dinesh Madugundi, Anna Garrison

Product Validation Engineering

Vehicle Analysis

Agenda

� Motivation

� Review of existing literature

� Understanding Truck aerodynamics

� Choice of numerical methods for Truck aerodynamics

Daimler Trucks North AmericaPVE Vehicle Analysis 2

Source:

� Numerical methods comparison study

� References


Vehicle Analysis

MotivationClass-8 Truck with Standard 53‘ Trailer

� Accurately predicting complex flow phenomenon in Truck aerodynamics using

numerical methods can be challenging.

Daimler Trucks North America

Wake interaction between the drive tires and trailer bogies

PVE Vehicle Analysis 3

Source:

Tractor trailer gap

Trailer back face


Vehicle Analysis

MotivationWhy CFD to Evaluate Class 8 Truck Aerodynamics?

� Standard Class 8 trucks are ~2.8m wide and ~22m long including the trailer.

� Very few full scale wind tunnels can accommodate full tractor trailer

configuration, while simulating real road wind conditions.

� Advantages in using CFD


Source:

— Full tractor trailer configuration

— Simulating real road conditions

— Predicting performance of an aero component before modifying or installing

— Wide range of design validations

— Detailed flow visualization


Vehicle Analysis

CFD ModellingStandard Numerical Methods

� Reynolds Averaged Navier Stokes (RANS)

� Unsteady RANS

� Large Eddy Simulation (LES)

� Detached Eddy Simulation (DES)


Source:

� Detached Eddy Simulation (DES)


Vehicle Analysis

� Bruce L. Storms AerospaceComputing Inc. “A Summary of the Experimental Results for a Generic

Tractor-Trailer in the Ames Research Center 7- by 10-Foot and 12-Foot Wind Tunnels”.

� Generic Conventional model (GCM) of class-8 tractor-trailer 1/8th scale results available for

validation study.

Experimental Setup (Reference)NASA/TM – 2006-213489

� Simplified model of standard class-8 tractor-trailer,

no grille opening, no underhood components.

� Experiments were performed at Re* = 1.1e6 –

Daimler Trucks North America 6PVE Vehicle Analysis 09.04.2013

� Experiments were performed at Re* = 1.1e6 –

6.2e6.

� For this study, results from Re = 6.0e6 are

compared, no T-T gap aero treatment, no trailer

aero treatment.

*Re was calculated based on Truck width.


Vehicle Analysis

CFD SetupGeometry and CFD Mesh Overview� GCM 1:1 scale, closed grille, flat underbody.

� No T-T gap aero treatment, no trailer aero devices

� Computational domain with far field domain, moving ground, and

spinning tires.

� Mesh settings

— Base size = 40mm


— Base size = 40mm

— Trim mesh, surface size 5mm – 40mm

— Wake refinement

— Low Re prism mesh

— Total number of volume cells ~15M


Vehicle Analysis

CFD Results of GCMTransient vs Steady� Flow conditions, constant density, Re = 6.0e6, 0yaw

and 6yaw. No side extenders.

� Solvers RANS, URANS and DES with Spalart-Allmaras

turbulence model, are compared.

� The three solvers predicted Cd that matched relatively

close to the measurements.


� RANS and URANS results matched well with each

other.

� DES results are closer to the measurements at yaw,

compared to RANS. More accurate wake predictions?

80mph @0yaw: DES

80mph @6yaw: DES

80mph @0yaw: URANS

80mph @6yaw: URANS

Plane View

Time Avg Ptotal Plots Time Avg Ptotal Plots


Vehicle Analysis

CFD ResultsUnderstanding Truck Aerodynamics 1/2� Production Cascadia sleeper, 45” T-T gap, and 53’

standard trailer.

� CFD Methods: Current DTNA best practices.

� About 50% of total drag is from tractor.

� Yaw effects are predominant on trailer bogies and

trailer back face.

∑=

−−

=

l

i

xxx iil

CdCdCumulative

1

1

Tracto

r d

rag

~5

0%

Traile

r d

rag

~5

0%


� Sections of drag effects

— A-surface

� Stagnation pressure on the grille

� Flow around the bumper and hood

� Stagnation pressure on wind shield

� Flow over the roof cap

� Effectiveness of roof deflector

� Effectiveness of side extenders and chassis fairings

Cumulative Cd[-] plot over the length of the Truck, normalized by

total vehicle Cd0yaw.

Tracto

r d

rag


Vehicle Analysis

CFD ResultsUnderstanding Truck Aerodynamics 2/2� Sections of drag effects, cntd…

— Underhood pressure: flow below the bumper determines

underhood pressure.

— Underbody flow: chassis components and drive tires are

exposed to high speed flow.

— T-T gap: Pressure in T-T gap influences effectiveness of

side-extenders and roof deflector.

∑=

−−

=

l

i

xxx iil

CdCdCumulative

1

1

T-T gap Pressure

Daimler Trucks North America



10PVE Vehicle Analysis 09.04.2013

side-extenders and roof deflector.

— Trailer bottom and trailer back face.

Plan view: Z-section along tire center: 55mph, 0yaw

Plan view: Z-section along tire center: 55mph, 6yaw

55mph, 0yaw 55mph, 6yaw

Time Avg Velocity Magnitudes


Vehicle Analysis

CFD ModelingChoice of Numerical Methods for Truck Aero� Choice of numerical methods is critical to capture complex flow phenomenon of Truck

aerodynamics.

— Surface bounded flow (Current industry standards, RANS, k-e, or SA).

— Under the cab wake interaction (accurate prediction of vortex shedding).

— Tractor trailer wake interaction.

� RANS methodology with Low-Re mesh can achieve accurate boundary flow.


Source:

� LES methodology to capture vortex shedding.

— Highly mesh dependant in BL.

— Can be computationally expensive.

� For Truck aero applications, hybrid model DES (Detached Eddy Simulation) can deliver best aspects

of RANS and LES methodologies.

— Less sensitive to boundary layer mesh with RANS methodology.

— Low Re mesh to accurately predict flow separation.

— LES methodology to predict wakes; sensitive to mesh wake refinements.

— Computationally less expensive than LES


Vehicle Analysis

CFD ResultsRANS vs DES 1/6� The CFD models are created using STAR-CCM+ v6.06.017. The following numerical methods are

compared for this study

— RANS

� Turbulence model – Spalart Allmaras

� Time dependency – Steady

� Segregated Flow

Daimler Trucks North America12PVE Vehicle Analysis 09.04.2013

� Segregated Flow

� Wall Treatment – All y+

— DES (As per DTNA’s best practices)

� Turbulence model– Spalart Allmaras Detached Eddy

� Time dependency – Implicit Unsteady

� Segregated Flow

� Wall Treatment – All y+


Vehicle Analysis

CFD ResultsRANS vs DES 2/6� The difference in total vehicle drag between RANS vs

DES is about 15% - 20% depending on yaw condition.

� Delta Cd on tractor is about 5% - 8%; resultant

difference is on trailer bogies and back face.

� Flow Comparison,

— Flow separation over the hood.

∆C

d ~

15

% -

20

%

∑=

−−

=

l

i

xxx iil

CdCdCumulative

1

1

∆Cd ~5% - 8%


— More diffusion under the bumper.

� Effects underhood pressure.

� Higher drag on chassis and drive tires.

— Difference in T-T gap pressure (influences roof deflector

and side extenders’ performance).



DES RANSTime Avg Velocity Magnitudes


Vehicle Analysis

CFD ResultsRANS vs DES 3/6� RANS predicted drag on the trailer is ~12% - 15%

lower.

— Difference in drag is higher at 0yaw; can be accounted

to wake interaction.

— At 6yaw, the wake interaction is reduced due to free

stream effects; shift in wake direction.

— Similar differences on trailer back face at 0yaw and

∑=

−−

=

l

i

xxx iil

CdCdCumulative

1

1

∆C

d ~

12

% -

15

%


— Similar differences on trailer back face at 0yaw and

6yaw.

� Transient phenomenon with controlled wake under the

trailer? For example, trailer skirts.Cumulative Cd[-] plot over the length of the Truck, normalized by


DES

RANS

DES

RANS



Vehicle Analysis

CFD ResultsRANS vs DES 4/6� Trailer skirts shield high speed flow impinging the

trailer bogies.

� With controlled wake under the trailer, we expect less

transient phenomenon under the trailer.

� The difference in total vehicle drag from RANS vs DES

increased to 25% .

∑=

−−

=

l

i

xxx iil

CdCdCumulative

1

1

∆Cd ~5% - 8%

∆C

d ~

15

% -

25

%


— Tractor drag difference remained at 5% – 8%; significant

differences on trailer drag.

� With larger wake, transient phenomenon becomes

more prominent.DES (0yaw)

RANS (0yaw)

DES (6yaw)

RANS (6yaw)





Vehicle Analysis

Degree of AccuracyRANS vs DES 5/6� Drag predictions using DES methods are

compared to WT testing for validation; the

results are within 2% accurate at a given

Re.

� Drag predictions using RANS are off by 15%

- 20%.

� Flow characteristics with DES methods

W/o Trailer Skirts With Trailer Skirts

[DES - RANS] [DES – Exp] [DES - RANS] [DES – Exp]

∆Cd 0yaw 20.84% 1.4% 25.49% 1.5%

Chassis 3.66% 3.89%

Tractor Tires 0.91% 1.36%

Trailer 14.81% 18.87%

∆Cd 6yaw 14.47% Not Avail 16.10% Not Avail

Chassis 2.52% 1.79%

Tractor Tires 0.79% 1.25%

Trailer 10.38% 11.81%


Source:

� Flow characteristics with DES methods

— The amplitude of Cd oscillations are

about10% - 20% of average Cd.

— Requires longer physical time to achieve

converged solution (say, 10ms of TS, 25s

total physical time with 5s/10s running

avg); computationally expensive.

— Not appropriate for design optimization

study when the Cd resolution per design

iteration is ∆Cd<1%.

� RANS applicability in Truck aerodynamics?


Vehicle Analysis

Degree of AccuracyRANS vs DES 6/6� RANS methods applicability in Truck aerodynamics?

— Qualitative analysis of aero performance of design

variants., Eg., mirrors.

— Possible to obtain general drag trend due to the

variants.

— The drag trends can be misleading depending on the

location of aero device.


Source:

location of aero device.

� For example, validation of aerodynamic performance of

multiple design variants of a roof cap.

— Evaluating all the design variants using DES methods

can be very expensive.

— RANS methods to get preliminary understanding of the

performance of each variant.

— Best design was re-evaluated using DES methods for

final confirmation.

— Drag performance on the trailer showed inverse trend.

� In most cases, evaluating an aero component using DES methods becomes necessary!!!!

Truck image is only for reference. Actual

Truck used for this study is not shown.


Vehicle Analysis

Conclusions

� To achieve accurate aerodynamic drag evaluation of Class-8 trucks, numerical methods capable of

predicting vortex shedding can be influential in design evolution.

� DES methods are

— Proved to be accurate during validation of CFD methods.

— Accurate evaluation of aero components.

— Computationally expensive.


Source:

— Sensitive to mesh refinement.

— Capturing aero performance of minor changes in design can be questionable.

� RANS methods are

— Good for qualitative analysis of aero performance of design variants.

— Capable of generating general drag trend of a given design modification.

— Trends can be misleading depending on the type of aero application.


Vehicle Analysis

References

1. Bruce L. Storms Aerospace Computing Inc. “A Summary of the Experimental

Results for a Generic Tractor-Trailer in the Ames Research Center 7- by 10-Foot

and 12-Foot Wind Tunnels”

2. Product Validation Engineering – Analysis, DTNA LLC. “Computational Fluid

Dynamics Certification”


Source:

3. SAE J2966, “Guidelines for Aerodynamic Assessment of Medium and Heavy

Commercial Ground Vehicles Using Computational Fluid Dynamics”.

Class 8 Truck External Aerodynamicsmdx2.plm.automation.siemens.com/sites/default/files/...Class 8 Truck External Aerodynamics Choice of Numerical Methods Security Classification Line1

Documents