Designing High Strength Aluminium Alloys for

Aerospace Applications

H.Aourag

Aluminium Alloys in Aerospace

Airbus A340

Despite competition from other materials, Al alloys still make up > 70% of structure of modern commercial airliner

Design Requirements

Components must beLightweightDamage tolerantDurable (corrosion resistant)Cost effective

Requires careful balance of material properties

Critical Material Properties

Aluminium Alloys

Pure aluminium has Low density (

relative Al=2.7, Fe=7.9)

Readily available (Al is 3rd most abundant element in Earth's crust)

Highly formable (FCC crystal structure)

Low strength and stiffness (EAl=70GPa,

EFe

=211GPa)

Low melting point (Tm=660oC)

Alloy with other elements to improve strength and stiffness - results in alloys with properties well matched to aerospace requirements

Aerospace Al-Alloys

Dominated by “high strength” wrought alloys Two main alloy series in particular

2xxx alloys (Al + Cu, Mg) UTS~500MPa 7xxx alloys (Al + Mg, Zn, (Cu)) UTS~600MPa

A) Slats - 2618B) D-Nose Skins - 2024C) Top Panel - 7150D) Bottom Panel - 2024

E) Spars / Ribs - 7010F) Flap Support - 7175G) Flap Track - 7075H) Landing Gear - 2024

A

B

C

E

D

FG

H

Alloys used in typical wing structure

Next Generation Aircraft

Bigger....

...FasterBoeing sonic cruiser > Mach.95

Airbus A380 > 950 seats

Goals

Next generation aircraft rely on advances in materials and assembly methods

Weight reduction is critical Alloy optimization

Increase strength and stiffness and/or reduce density whilst maintaining other properties

Assembly optimization Reduce weight associated with joints between

components

Alloy Design

Traditionally, alloy and process development largely by trial and error based on metallurgical experience

Recently, emphasis has changed to designing alloys and processes to meet specific property goals Improved understanding of relationships between

processing, microstructure and properties Development of models to predict alloy microstructure

and performance

Applications of Modelling

Models on a range of length scales Atomistic (nm)

Limited application as currently capable of dealing with only very small volumes of material

Microstructural (nm-m) Used to predict particle distributions, grain sizes etc.. as

function of alloy chemistry and processing conditions, often coupled to microstructure-property models

Macro-scale (>mm) Widely used to predict performance of components

during processing and service as a function of average material properties and stress, strain, temperature....

Modelling Examples

Finite element modelling to optimize extrusion processing of aerospace Al-alloys

Thermodynamic modelling for the development of weldable aerospace aluminium alloys

Precipitation kinetics modelling for optimization of dispersoid particles in 7xxx alloys

Ma

cro

Mic

ro

Finite Element Modelling of Extrusion

Direct

The Extrusion Process

Extrusion is widely used to produce aerospace components

Direct Indirect

Ram DieBillet

Extrusion

Extruded shapes are often complex - design of die is critical

(Al alloy)

Die Design

Die must be designed to ensure balanced metal flow to avoid bending of extrusion

Die shape influences metal temperature-aim to avoid cold or hot spots

Traditionally, die design based on past experience and modifications of existing dies

Alternative: Use finite element methods to model extrusion process and identify and test new die designs

The Finite Element Method

Divide billet/extrusion into small, connected elements

Relate displacements/temperature changes in one element to those in surrounding elements using well established physical laws

2D finite element mesh for an extrusion

Use of Finite Element Model

Use commercially available FE package to model metal flow and temperature during extrusion

New die design

Simulate extrusion process

Any problems?Unbalanced metal flowExcess temperature variation

Modify design

No

Yes

Make prototype die

FE Model - Example Simulations

Example Simulations in 2D and 3D

Weldable Aerospace Al-Alloys

Joining Aerospace Al-Alloys Mechanical fasteners (rivets) are still the most

widely used method of joining airframe components

Riveted joints have a number of disadvantages

Problem: Most high strength Al-alloys suitable for aerospace are considered “unweldable”

Welded jointNo extra material (less weight)Process readily automated

Riveted jointExtra material requiredLabour intensive

Difficulties with welding

One of the major metallurgical problems preventing the widespread application of welding to aerospace Al-alloys is solidification cracking

250 m

7075 TIG Weld

Cracks arise when the thermal stresses generated during cooling exceed the strength of the almost solidified metal

Factors Influencing Solidification Cracking

1) Level of Thermal Stresses

3) Absolute Freezing Range

- alloys with a wide freezing range are susceptible to cracking

5) Volume Fraction of Low Melting Point Eutectic Phases

- if there is sufficient liquid at the end of solidification to flow around the dendrites, then any cracks might be healed

2) Grain Structure of Fusion Zone

- columnar grains vs equiaxed grains

4) Freezing Range for Dendrite Cohesion

- thought to occur at about 50-60% Solid (depend on grain structure)

?

?

?

Thermodynamic Modelling

Thermodynamic Modelling For any alloy system, set of conditions and

configuration of the components there will be an associated free energy

Use computer models to calculate the free energy for complex systems (lots of elements) from data for simple systems (1,2 or 3 elements)

Calculate the equilibrium (minimum free energy) configuration and hence phase diagrams for complex systems Can be useful in the interpretation of real microstructures

Calculate phase fractions and compositions for certain other well defined non-equilibrium problems

Simple Phase Diagrams

300

350

400

450

500

550

600

650

700

0 10 20 30 40 50 60

wt.% Mg

Tem

pe

ratu

re (

C)

Liquid

Liquid + -Al

-Al

-Al + -AlMg

-A

lMg

-AlMg

-A

lMg

Al-Mg System (Al-Rich)

300

400

500

600

700

800

900

1000

1100

1200

0 20 40 60 80 100

wt.% Mg

Tem

per

atu

re (

C)

Liquid

Liquid + Mg

CuMg2 + Mg

CuM

g2

L + CuMg2

L + -Cu

Laves - C15

-Cu +

Laves - C15

Liquid + Laves - C15

Cu-Mg System

Ternary Phases S - Al2CuMg, T - Mg32(Al,Cu)49, V - Al5Cu6Mg2, Q - Al7Cu3Mg6

Even for simple 2xxx alloy (Al-Cu-Mg), need data for 3 binaries and information about ternary phases

400

450

500

550

600

650

700

0 10 20 30 40 50 60wt.% Cu

Tem

per

atu

re (

C)

Liquid

Liquid + -Al

-Al

Liquid + -Al2Cu

-Al + -Al2Cu-Al2Cu

Al-Cu System (Al-Rich)

MTDATA predicted phase diagrams

Real, commercial Al-alloys may contain > 10 alloying elements!

Success of thermodynamic models relies on availability of sufficient, high quality, thermodynamic data

Solidification Microstructures

Solidification occurs rapidly under non-equilibrium conditions

However, given certain assumptions, thermodynamic calculations and the equilibrium phase diagram can still be used to predict solidification microstructure

Scheil Solidication Model - Assumptions:

(i) Local equilibrium exists at the solid/liquid interface

(ii) No diffusion in the solid phases

(iii) Uniform liquid composition

(iv) No density difference between

solid and liquid

% Solute

T

Cliq1

Csol1Cliq2

Csol2Cliq3

Csol3

C0

Csol0

Liquid

Solid

Microstructure

Predictions for Binary Al-Cu Alloy

Freezing Range

520 540 560 580 600 620 640 660 680 700

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Mas

s P

has

e F

ract

ion

Liquidfcc -Al

- Al2Cu

EutecticReaction

Temperature (C) fcc -Al dendrites

- Al2Cu eutectic fcc -Al

eutectic

Predictions for Ternary Al-Cu-Mg alloy

470 490 510 530 550 570 590 610 630 650

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0M

ass

Ph

ase

Fra

ctio

n

Temperature (C)

Liquid

fcc -Al

TLTST

S - Al2CuMg

- Al2Cu

Ternary Eutectic Predicted at ~ 500ºC

Predictions for 2xxx (Al-4.5Cu-1.5wt%) Mg alloy

Prediction of Freezing Range

To reduce tendency for solidification cracking, need to minimize absolute freezing range

Use thermodynamic model to predict freezing range for different alloy compositions

Effect of Mg content on freezing range of eutectic in Al-4.5Cu-x Mg alloy

0

5

10

15

20

25

30

35

40

45

50

0 0.5 1 1.5 2 2.5 3

wt.% Mg

T

(F

ree

zin

g R

an

ge

of

Eu

tec

tic

)

Binary Eutectic[ + ]

Ternary Eutectic[ + S]

Saddle Point[ + S]

Optimum composition range

Value of Calculations

Thermodynamic calculations suggest modifications to current alloy compositions to improve weldability

Focus experimental investigation on promising compositions Save both development time and cost

New weld filler wires have been developed on the basis of these calculations and are now being tested

Modelling Dispersoid Precipitation in 7xxx Aerospace Al Alloys

Prediction of Microstructure Thermodynamic calculations give an indication of

likely phases but give no information about How phase is distributed

Particle size, spacing and location

How microstructure changes as function of time Transformation of metastable phases Evolution of volume fraction of phase and particle

size distribution

These factors depend on phase transformation kinetics and are critical in determining microstructure and hence properties

Kinetic Modelling

Aim to predict key microstructural parameters as a function of alloy composition, temperature and time

Difficult problem for aerospace Al-alloys due to complex microstructures and processing routes Large number of possible phases evolving

simultaneously Metal subjected to thermal cycling and

complex deformation during processing

7050 Plate

Components machined from 7050 alloy thick plate are widely used in load bearing applications e.g. wing spars

7050 composition specification

Focus on one alloy (7050) and product (thick hot rolled plate)

Processing Sequence - 7050 Plate

CastDirect chill

Homogenize~475oC, 24h

Hot roll~350-450oC20+ passesreduction~70%

Solution treat475oC, 1hspray quenched

Age

Cast SolutionizedHomogenized Rolled

RD

Microstructural ChangesT

em

pe

ratu

re

Aged

Time

50nm

Dispersoids

Fine Al3Zr dispersoid particles precipitate during homogenization of 7050

Dispersoid particles are important for the control of grain structure during processing Act to “pin” grain boundaries

Al3Zr dispersoid particles in 7050 after homogenization

Modelling Dispersoid Precipitation

Effectiveness of dispersoids depends on their size, spacing and distribution

Develop model for dispersoid precipitation and use to optimize homogenization treatment to give best dispersoid distribution

To model dispersoid precipitation must account for both non-uniform distribution of Zr due to microsegregation during casting and Al3Zr precipitation kinetics

Start

Average zirconium

concentration (depends on

position in slab)

Microsegregation Model

(MTDATA Scheil Model)

Local zirconium concentration (as a function of position

within grain)

Homogenization temperature/time

profile

PrecipitationKinetics Model

Dispersoid size, number density, spacing and size

distribution

Schematic of Model

Precipitation Kinetics

The precipitation of Al3Zr dispersoids is a diffusion controlled phase transformationClassically, precipitation of particles considered as 2-step process of nucleation and growth, followed by coarsening

Clusters of Al, Zr atoms form by random in matrix. Stable clusters become particle nuclei

Particles grow, controlled by diffusion of Zr

Small particles dissolve at the expense of large particles to reduce total interfacial area

Nucleation Coarsening

Time = t1 t2 t3

Nucleation+growth

Kinetics Model

Time is divided into a large number of small steps

Growth, nucleation and coarsening allowed to occur concurrently governed by driving force and concentration gradients

At each step new particles nucleate and existing particles grow (or shrink) depending on local interfacial compositions

After each step, solute supersaturation in the matrix is recalculated and used for next step

Nucleation

Nucleation rate (number of new particles formed/s) depends on Thermodynamic driving force for formation

of new phase Diffusion rate (temperature) Interfacial energy between nucleus and

matrix

Nucleation rate

Te

mp

era

ture

I/f e

nerg

yNucleation rate

Driving forceincreasing but diffusion ratedecreasing

Growth

Growth rate for each particle depends on Concentration gradient ahead of particle

Equilibrium compositions from phase diagram Particle size

Diffusion rateZ

r co

ncen

trat

ion

distance

Concentration profiles

Small particle

Large particle

Zr in particle

Zr in matrix at interface(depends on particles size)

Coarsening

Coarsening does not need to be modelled separately but arises naturally from growth model in later stages of precipitation

Early stages

All particles growing

c

Co

nce

ntr

atio

n Z

r

Late stages

Large particles growing, small particles shrinking

c

Co

nce

ntr

atio

n Z

r shrinkinggrowing

Testing the Model

First test model against experiment for a single initial Zr concentration

Size

Num

ber

Evolution of size distribution with time

Comparison of model prediction and experiment at 500oC

Effect of Zirconium Segregation

In practice, Zr concentration varies across a grain due to segregation during casting

Leads to non-uniform dispersoid precipitation during homogenization

Zr concentration after castingObserved dispersoid distribution after homogenization

Centre

Disperso

id free

zone

Edge Low Zr

High Zr

EDGE CENTRE

Including Effect of Segregation

To model Al3Zr distribution across a grain Divide the distance from grain edge to

centre into large number of elements Model dispersoid evolution in each element Allow zirconium redistribution by diffusion

between elements

Edge Centre

Zr

conc

entr

atio

n

Zr removed into Al3Zr dispersoids

Zr diffusing into elementZr diffusing out of element

Predicting Across a Grain

Volume Fraction Zr in solution Mean radius

Centre Edge Centre Edge Centre Edge

Centre

Edge Can the model reproduce the observed behaviour?

Effect of Dispersoid Distribution Inhomogeneously distributed dispersoids

are not best for control of grain structure In regions where there are few

dispersoids, new grains can form (recrystallization) - this is undesirable

Structure after processingNew grains have formed and partially consumed original grains - this structure does not give best properties

Optimizing Dispersoid Distribution

Use model to determine optimum homogenization conditions to promote dispersoid precipitation in low Zr regions

Aim is to reduce the formation of new (recrystallized) grains during processing

For best recrystallization resistance, want a large number of small dispersoid particles, as uniformly distributed as possible

Model Predictions

Nucleation

Growth

Use model to investigate kinetics in detail

Temperature /oC Time /hT

em

per

atu

re /o

C

To promote dispersoid nucleation in low Zr regions need to hold at ~425oC

Optimizing Homogenization

BUT Homogenization temperature for 7050 is restricted

Model suggests that best temperature for precipitating dispersoids in low Zr regions lies below this range

Hom

ogen

izat

ion

ran

ge

AA7050

Need to dissolve these phases during homogenization

Must avoid onset of melting

Two Step Practice

Two step homogenization practice may be of benefit Step 1: Hold at a temperature to precipitate

optimum dispersoid distribution Step 2: Hold at final homogenization

temperature

Model used to determine best conditions for step 1 5h Hold time at 430oC

Test 2 step homogenization practice

Effect on Dispersoids

Standard Homogenization Two step treatment

Comparison of Recrystallization

Standard PracticeRecrystallized Fraction = 30.4%

Hold + Homogenize PracticeRecrystallized Fraction = 14.0%

Two step homogenization practice, developed entirely by computer modelling, is effective in significantly reducing the fraction of recrystallization

Summary

Aerospace aluminium alloys are complex materials, developed over a long period of time by empirical experiment to meet industrial needsIn recent years, the understanding of the metallurgical processes governing the microstructure and properties of these alloys has greatly increased

This has led to the development of models that have practical application in the design of new alloys and processes

Acknowledgements

For provision of data and examples of FE and thermodynamic modelling Dr Qiang Li, Birmingham University Dr Andy Norman, Manchester Materials

Science Centre

Luxfer and Alcoa for funding some of this research