PROPRIETARY DAHER SOCATA 21/03/20141 MASTER DEGREE DISSERTATION IN MECHANICAL, AERONAUTICAL...

PROPRIETARY DAHER SOCATA21/03/2014 1

MASTER DEGREE DISSERTATION IN MECHANICAL, MASTER DEGREE DISSERTATION IN MECHANICAL, AERONAUTICAL ENGINEERINGAERONAUTICAL ENGINEERING

Development of an

automatic shape

optimization platform

for a laminar profile

March - September 2013

Relatori :Prof. Jan Pralits

Ing. Thomas Michon

Studente :Marcello Tobia Benvenuto

PROPRIETARY DAHER SOCATA 2

Daher Socata produces the world’s fastest

single turboprop aircraft: TBM 850.

As each aeronautic company, Reduce the consumptionit works every day to improve

the aircraft performance. Increase the max. speed

Fluid mechanicsReduce the drag on the surfaces:

WING

Introduction

21/03/2014


• When a body is in motion in a flow, the flow adhere to it because of the viscosity.

A thin layer arises close to the shape, called boundary layer.

Physical phenomenon

21/03/2014


External disturbances can enter the boundary layer and generate a turbulent flow through a Transition process.

• Laminar boundary layer:

Thin with regular streamlines;

low skin friction.

• Turbulent boundary layer:

Thick with irregular fluctuations;

high skin friction.

The transition phenomenon is very sensitive to the shape variations

Physical phenomenon

Skin Friction

X/C

21/03/2014


Reduce the friction drag on an airfoil by keeping the flow laminar over the largest possible portion of the surface.

Automatic Shape Optimization

Advantages:

1) Save time during a process

2) Run multiple repetitive simulations

3) Analyze automatically the good results, finding the optimum

Objective

21/03/2014


• Optimization platform for 2D Geometry

• 2D optimization High and High/Low speed- results- discussion

• Creation wing- results- discussion

• Conclusions

• Future works

Contents

21/03/2014


The wing’s behaviors are given by its profiles.

Relative Thickness: 16%

Chord: 1.675 m

Why a 2D geometry?

21/03/2014

PROPRIETARY DAHER SOCATA

Create the 2D geometry

Create the domain and the mesh

Flow Solver

Boundary layer and its stability

8

Catia V 5

ANSYS: Design Modeler and Mesh

ANSYS: Fluent

bl3D and Nolot code

Optimization platform

Mode Frontier

Optimization steps and tools

21/03/2014




Flow Solver


9

Catia V 5


ANSYS: Fluent

bl3D and Nolot code


Mode Frontier


21/03/2014


To limit the number of the geometric design variables

10

Describing the shape with a small set of inputs

9 Polynomial approximations of curves CAD Software: Catia V 5


21/03/2014


Design Parameters Constraints

• Radius of the circle

• Position of point 2 and 9 inside square

• Thickness of trailing edge

• Tension of points 2,3,8,9

• Chord = 1 meter

• Thickness at 25% and 75% of the chord fixed.


21/03/2014




Flow Solver


12

Catia V 5


ANSYS: Fluent

bl3D and Nolot code


Mode Frontier


21/03/2014


•O-type domain•Radius = 90 meters

Different domains and meshes have been investigated to find the best grid in terms of time and quality

Grid close to the profile:

Profile

Grid


21/03/2014




Flow Solver


14

Catia V 5


ANSYS: Fluent

bl3D and Nolot code


Mode Frontier


21/03/2014


Numerical solution of the Navier-Stokes’s equations

Velocity and pressure distribution

FLUENT

Pressure Coefficient distribution on the root airfoil of TBM 850. Cruise conditions.

Key point for the stability analysis

• Smoothness• Good quality

Flow solver

X/C

Cp

21/03/2014




Flow Solver


16

Catia V 5


ANSYS: Fluent

bl3D and Nolot code


Mode Frontier


21/03/2014


bl3D codeIt calculates the parameters of the boundary layer from the Cp distribution

Laminar Boundary Layer's Equations

Boundary layer and its stability: bl3D

21/03/2014


NOLOT is based on the Linear Stability: Flow decomposed in mean flow and unsteady disturbances

u = U + u'

The unsteady disturbance is represented by a wave with infinitesimal amplitude

Boundary layer and its stability: NOLOT

Streamwise Wave number

Spanmwise Wave number

Frequency

21/03/2014


Semi-empirical eN method

Mack’s Law:

N = - 8.43 – 2.4 ln(Ti) 0.0007 < Ti < 0.0298

N factor Turbulence intensity

Boundary layer and its stability: NOLOT

21/03/2014


1. To maximize the position of transition

1. To minimize ∆Cl = |Cl – ClTBM|

1. To minimize ∆Cm = |Cm – CmTBM|

A change of the shape of a profile can lead to different value of Cl and Cm

Changes of global repartition of lift

• Stability problems• Stalling problems

Objective functions

21/03/2014




Flow Solver


21

Catia V 5


ANSYS: Fluent

bl3D and Nolot code


Mode Frontier


21/03/2014


Optimization platform: Mode Frontier

21/03/2014


Lift and Mom. coeff

∆Cl ∆Cm

Optimization platform: Mode Frontier

21/03/2014



• 2D optimization High and High/Low speed- results

- discussion


• Conclusions

• Future works

Contents

21/03/2014


• high speed (cruise): M=0.51; h=26000 feet; aoa=0 degrees

• Strategy optimization

- explore all the domain of input parameters DOE

- optimize the best profiles found by DOE with genetic algorithm

Optimization 2D High speed

21/03/2014


• 399 profiles have been explored in 8 days

Transition location

∆Cl

Max ∆Cl 3%

TBM (trans. 26% of the chord)

Max trans. 47% of the chord

Pareto front opt. 2D high speed

21/03/2014


Best solution opt. 2D high speed

BLACK = TBM RED = BEST

21/03/2014


Solution not robust

0.07% of 1765 mm = 1.19 mm

c A big influence of the leading edge on the transition

Robustness solution for manufacturing?

21/03/2014


To evaluate the difference of drag, the SST-transition model is used in Fluent to study the natural transition:

Drag evaluation with transition model

21/03/2014


- High speed (cruise): M=0.51; h=26000 feet; aoa=0 degrees- Low speed (take-off): M=0.18; h=0; aoa= > 15 degrees

To analyze stall characteristics at low speed, the profile has been optimized also at take-off conditions

Optimization 2D High/Low speed

21/03/2014


Cruise condition:

1. To maximize the transition location

2. To minimize ∆Cl and ∆Cm

Take-off condition:

1. Maximize the max Lift coefficient

Objective functions

21/03/2014


Pareto front

Transition high speed

Cl low speed

The objective functions are in opposition one with the other

The same optimization has been done for the tip profile of the wing

Pareto front 2D opt. High/low speed

21/03/2014


High speed• Big sensibility of the phenomenon by the shape variations

• Transition moved from 26% to 47% of the chord

• Viscous drag reduced of 14.26%

• Improvements limited by the constraints of the shape: transition occurs close to the maximum thickness

High/low speed• Each flight condition requires a different optimal shape

• The presence of a new O.F. has not penalized the transition (42%)

• Improvements limited by the constraints of the shape

Discussion optimization 2D

21/03/2014





• Conclusions

• Future works

Contents

21/03/2014


Creation of a wing with the optimal root and tip profile obtained previously

Wing parameters:

The same of the wing of TBM 850- span: 12161.3 mm- dihedral: 6.5 degree

Creation wing

21/03/2014


To compare the wing of the TBM 850 with the wing using the optimal profiles.

Skin Friction

TBM NEW

CFD Simulation 3D

21/03/2014


Wing Visc. drag Press. drag Total drag Lift coeff

TBM 0.00273 0.00754 0.01027 0.1919

New 0.00279 0.00755 0.01035 0.1903

Skin friction on profile at 50% of the span

Results 3D

Skin Friction

Chord

New

TBM

21/03/2014


• The validation on the wing has given unexpected results in terms of drag:

The effects of the flows on 2D and 3D geometry are different

- trailing vortex

- cross flow disturbances

X - Wall shear stress

Discussion

21/03/2014


• The validation on the wing has given unexpected results in terms of drag:

The effects of the flows on 2D and 3D geometry are different

- trailing vortex

- cross flow disturbances

Discussion

21/03/2014





• Conclusions

• Future works

Contents

21/03/2014


I am familiar with software like Catia V 5, Fluent (2D and 3D), Fortran, Python, modeFRONTIER

I created an automatic shape optimization for 2D geometry

• The strategy used, has allowed to obtain good results for 2D geometry

- transition phenomenon delayed from 26% to 47% of the chord

- Viscous drag reduced more than 14%

Conclusions

21/03/2014


Optimization 2D:

1. New parameterization (CST) with other constraints can be tested

2. More time for the iterations can lead a better results

3D Validation:

1. To consider 3D effects we can run the following loop:

Study the flow around the wingTake Cp distribution of three profiles of the wing (root, middle, tip)Run optimization platform for the three profilesTo rebuild the wing with the three new profiles and study the flow on the

wing

Future work and suggestions

21/03/2014


Thank you for your attention

21/03/2014

PROPRIETARY DAHER SOCATA 21/03/20141 MASTER DEGREE DISSERTATION IN MECHANICAL, AERONAUTICAL...

Documents

Transcript of PROPRIETARY DAHER SOCATA 21/03/20141 MASTER DEGREE DISSERTATION IN MECHANICAL, AERONAUTICAL...