Impact of Accelerated Pavement Testing (APT) on Pavement ...€¦ · Impact of Accelerated Pavement Testing (APT) on Pavement Engineering Jean-Maurice Balay Jean-Michel Piau Pierre

Impact of Accelerated Pavement Testing (APT)

on Pavement Engineering

Jean-Maurice BalayJean-Michel Piau

Pierre HornychChantal de La Roche

LCPC - FranceTRB 2008 - Worshop 110 - Validation of advanced flexible pavement modeling

with accelerated Pavement testing data - Sunday 13 January 2008.

Contents

1 LCPC - January 13, 2008

• Introduction• LCPC activities on pavement design and modeling

• Presentation of LCPC APT facilities

• Use of APT for pavement engineering• Advantages, Limits and important rules

• Examples of APT experiments and their impacts– Fatigue experiments

– Airbus Pavement Experimental Program

– Use of APT for pavement structure innovation

• Conclusion

Introduction

LCPC activities on pavement design and modeling


• Contribution to the development and evaluation of French pavement design methodology

• Development of models for :• Low trafic pavements : Non linear behaviour of soils and unbound

granular materials, prediction of rutting

• Bituminous pavements : visco-elastic behaviour, fatigue, visco-plastic behaviour (rutting)

• Rigid pavements : study and modelling of reflective cracking

• Airfield pavements ; impact of multi peak loading

• Design of innovative pavement structures

Presentation of LCPC APT facilities


Fatigue test track – large circular test track (pavement length 120 m)

“FABAC” Machines : mobile machines – simulation of full wheel loading on small length (2m)

LCPC Fatigue test track


• Circular outdoor facility, built in 1984• Track width up to 6 m - length 122 m (radius 19.5m)• Radius of rotation : adjustable from 15 m to 20 m• Maximum speed 13.5 rpm (100 km/h at radius 19.5 m)• Simulation Transverse wandering of real traffic can be

reproduced• Various types of loads from 8-ton single-axle to 30-ton

multiple-axle (only a half-axle per arm)• 3 test sites - one including a water table

control system (pumping station)

Experiments performed in partnership withRoad authorities, public or private companies,European research program

Use of APT for pavement engineering


Performance models – Two main different approaches, almost disconnected today ?

• Extensive survey campaigns of road networks (ex: at national scale) & global statistical approach

• Tool: statistical analysis of broad data base, including equivalent pavement structures at different ages and supporting different traffic

• Efficient way to derive “evolution laws” for PMS (maintenance & reinforcement planification) on standard pavement techniques

• Local mechanistic approach on test sections, looking more deeply to the nature of pavement materials and structures and to their behaviors

• Tools: Real loading test, APT, laboratory testing, constitutive laws, structural models

• More “introspective” approach ; specially recommended for innovation testing (materials, structures)

Advantages and limits of APT


Advantages• Well controlled pavement construction and experimentation conditions

(load, traffic, temperature,…)

• Internal instrumentation and detailed monitoring of pavements

• Response in relatively short time owing to the acceleration of traffic

• Possibility to make comparative tests

• Good quantitative knowledge of the resilient behavior of the structures and damaging mechanisms

Limits

• Not representative of real traffic & climatic variations

• No long term ageing of materials

• Obtained results more comparative than intrinsic

Important rule for APT success


• Build a pavement dedicated to the distress to be studied and trynot to mix pavement distresses

• Examples :• permanent deformation of bituminous layers

use hydraulically bound base and sub base• fatigue of bituminous materials

avoid using surface layer that could hide bottom to top cracks• ….

Examples of APT experiments and their impacts

Experimenton fatigue behaviour

Objectives of the experiments


• Comparison of fatigue behaviour of different bituminous materials, with different binders

- in the laboratory, using different fatigue tests- in pavements

• Evaluation of the French fatigue design approach for bituminous pavements

• Determination of shift factor to be applied to High Modulus Mixtures (French EME)

• 3 full scale experiments on the LCPC test track (between 1990 and 1994)

Third “fatigue” experiment


4 structures : 8 or 10 cm thick bituminous layer40 cm thick granular baseclayey subgrade : E = 30 to 40 MPa

3 bituminous materials:BBB : Bituminous concrete with 50/70 grade bitumen (reference)BBS : Bituminous concrete with 50/70 polymer modified bitumenEME : high modulus bituminous mix, with 10/20 grade bitumen

2 structures – 8 and 10 cm thick

Loading conditions :65 kN dual wheel load, 72 km/h3.2 million loads applied

In situ performance of the 4 structures


Cracking extent

010

2030

4050

6070

8090

100

0 500000 1000000 1500000 2000000 2500000 3000000 3500000

Number of loads

Exte

nt o

f cra

ckin

g (%

)

Struct. 1 - BBB

Struct. 2 - BBS

Struct. 3 - EMEStruct. 4 - EME

Laboratory fatigue tests


5 different test procedures

Procedure Type of loading

Fatigue design method


Criterion on maximum tensile strain :

( ) ( ) k10NEf,b6

6adt ⋅⋅θε=ε

ε6(θ,f) : Strain leading to failure for 106 cycles,depending on temperature θ and frequency f

NE : number of standard axle loadsb : slope of the fatigue linek : shift factor, taking into account the risk of failure, the bearing capacity of the soil, the difference between the model and observed pavement behaviour

Predicted pavement life: ( )b1

6

t6f,k

10NE ⎟⎟⎠

⎞⎜⎜⎝

⎛θε⋅

ε⋅=

Comparison of design predictions and field performance


Calculation of shift factor k (k = 1 exact prediction of pavement life) ( )

b1

6

t6f,k

10NE ⎟⎟⎠

⎞⎜⎜⎝

⎛θε⋅

ε⋅=

k

Fatigue test procedure S1BBB

S2BBs

S3EME 8 cm

S4EME 10cm

3 – strain control Continuous

1,58 1,28 1,14

0,8

1,10

1,20

5 – strain control, with rest periods

1,04 1,03 0,8

6 – stress control, continuous

3,21 2,57

10 – stress control, with rest periods

1,73 1,49 1,08

Conclusions – “fatigue experiments”


• Large differences in fatigue life predictions from different fatigue test procedures,

• Fatigue tests with rest periods seem more representative of in situ behaviour

• For the high modulus material (EME), the stress controlled fatigue test seems more representative of in situ behaviour.

more research needed on intrinsic characterisation of fatigue in laboratory

• Correction between in situ and lab behaviour dependent on the type of material (BB/EME)

Shift factor for EME = 1,0

[De la Roche et al, TRB 94, ISAP 98] [Rivière et al, ISAP 98

Airbus Pavement Experimental Program

Airbus experimental program on flexible pavements – 1998-2003


30m 30m 30m 30m

Section DAC 8 cm

BAC24 cmUGA 1.40 m

CBR 3

Section CAC 8 cm

BAC 24 cmUGA 0.60 m

CBR 6

Section BBBSG 8 cmGB3 24 cmGNT 20 cmSol CBR 10

Section ABBSG 8 cmGB3 24 cmSol CBR 15

Objectives of the Airbus experimental program


Tests on 4 instrumented pavement structures with soils of different bearing capacity (CBR 3 to 15)

Simulation of loads of different aircrafts , using a load simulator

Objectives :

Study of the behaviour of flexible airfield pavements under heavy aircraft loading conditions

Evaluation of the possibility of applying the French road pavement design method to flexible airfield pavements

Observed pavement performance


Main mode of distress of the flexible pavements = ruttingNo fatigue cracking - densification of the bituminous layers under heavy loading

-1.80

-1.60

-1.40

-1.20

-1.00

-0.80

-0.60

-0.40

-0.20

0.00

1000/2000 1000/3000 1000/4000 1000/5000

Max

rut d

epth

(cm

)Load 1

Load 2

Load 3

Load 4

Evolution of rutting Section C – CBR 6

Number of loads

Linear elastic calculations (ALIZE Software)


XX XX

A380 Load configuration

Prediction of vertical strains at top of subgrade (structure C)

0

200400

600

8001000

1200

-14 -12 -10 -8 -6 -4 -2 0 2Transversal profile XX (m)

µdé

f

4 wheels 4 wheels6 wheels 6 wheelsMeasurement Alizé software

Results of linear elastic calculations


Good prediction of vertical strains in granular layers and subgrade

Poor prediction of maximum tensile strains at bottom of bituminous layers :• Maximum values poorly simulated

• Discordance in directions of maximum tensile strains εt:

Measurements : εt transversal > εt longitudinal

Calculations : εt longitudinal > εt transversal

⇒ Attempt to take into account viscoelastic behaviour

ACεtrans.

εlong.

Visco-elastic model for bituminous materials


HUET - SAYEGH MODEL (1965)

10

100000

060phase angle, (°)

E0

E∞δ, k

h E MPa E MPak h

∞ = == = =

25000 452 7 0 27 0 7

0; ;, ; , ; ,δ

Black space

E E E Ei ik h*( , )

( ( )) ( ( ))ω θ

δ ωτ θ ωτ θ= +

−

+ +∞− −0

01

Huet-Sayegh’s model measurements

E*

, MP

a

3000

0 5000 25000

Cole & Cole axes

Real(E*), MPa

Im(E

*), M

Pa

Finite element model : module CVCR of CESAR-LCPC


Modelling of a pavement under moving wheel load (in 3D)

Hypotheses : Constant speed V

Constant properties along x

V zy

O' O x

σ( , y, z) ε(X, y, z)

X’

V zy

O' O x

σ( , y, z) ε(X, y, z)

Calculation in the referential of the moving load (O’, X’, y ,z) X’ = x +VtStatic mechanical problem - no time stepsModification of the visco -elastic law : becomes a non - local law

Models available in CVCR :Linear and non linear elasticity ( Boyce model, k-θ model)Huet-Sayegh visco-elastic model


- 2 0 0

- 1 5 0- 1 0 0

- 5 0

05 0

1 0 0

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

measurementmeasurement

calculations

εtlongitudinalbase AC

µµstrainstrain

time (s)time (s)

-6 0 0

-4 0 0

-2 0 0

0

2 0 0

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

measurement

calculations

εttransversalbase AC

µµstrainstrain

time (s)time (s)

Viscoelastic modelling of strains at bottom of bituminous layers

Load : 6 wheel bogie 240 kN per wheel

First conclusions of the Airbus experiment


Main mode of distress of the flexible pavements = ruttingNo detectable fatigue of bituminous layers, despite high tensile strains

Modelling of resilient behaviour

- Reasonable prediction of vertical strains in subgrade with linear elastic calculations.

- Visco-elastic modelling necessary to predict correctly strains in bituminous layers

Modelling of fatigue- Need to adapt the fatigue tests to aeronautical loading conditions

(high strain levels ≈ 400 µstrain, lower number of cycles (104))

Rutting :

Need to develop suitable design criteria and specifications for resistance to rutting of materials (bituminous and unbound)

Use of APT for pavement structure innovation

Grave-Mousse test at LCPC’s APT facility


Grave-Mousse ®- new material for treated road bases developed by EJL Contractor (now EUROVIA) since about 10 years. - bitumen-foam treated aggregates, used for new pavement as for overlay construction

- New pavement structure and innovative concept using Grave-Mousse in the middle part of a three-layer structure design model to be defined and validated- In overlay, functioning and endurance of this new material under heavy traffic to be checked.

Grave-Mousse test at LCPC’s APT facility


Experiment with LCPC’s APT facility :

• Partners : Private contractor EJL• Fatigue test performed from July 1995 to March 1996• 3 overlays and 1 new pavement, 28 m long x 3.5 m width• 2.87 millions single axle load from 45 kN to 85 kN /twinned

wheels, equivalent to 4.3 millions of the French standard load (130 kN/single axle) ie about 15 years of service on average trafic national network

• Load speed: - 44 km/h until 70 000 loadings (consolidation stage)- 68 km/h for all the other loadings.

The 4 structures tested


BFTA 10 cmCracked AC 7cm

UGA 40 cm

Spoilt micaschist 2 m

Overlay structure 3

AC 8 cm (reference)Cracked AC 7cmUGA 40 cm


Overlay structure 1

HMAC 6 cmBFTA 10 cmHMAC 6 cm


New pavement structure 4

ESA 10 cmCracked AC 7cmUGA 40 cm


Overlay structure 2

Rotation radius of loads : 18 m

3.5 m

AC : asphat concrete BFTA : bitumen-foam treated aggregatesESA : emulsion stabilised aggregates HMAC : high modulus asphalt concreteUGA : untreated graded aggregates

Cracking and rutting vs traffic


0

20

40

60

80

100

0.00 0.50 1.00 1.50 2.00 2.50 3.00

Equivalent traffic - 130 kN standard loads (millions)

Crac

king

exte

nt (%

)

Overlay structure 1 (AC) Overlay structure 2 (ESA) Overlay structure 3 (BTFA) New structure 4 (BFTA)

05

1015202530

0.00 1.00 2.00 3.00 4.00Equivalent traffic - 130 kN standard loads (millions)

Ruttin

g de

pth

(mm)

New structure 4: dammage mechanism


Typical horizontal strains profiles measured under the 65 kN standard load

HMAC

BFTA

HMAC

-25

-20

-15

-10

-5

0-150 -100 -50 0 50 100

Horizontal strains (µstrains)

Dept

h (c

m)

Untrafficked areas

Longitudinal strain Transversal strain

Trafficked areas

-25

-20

-15

-10

-5

0-150 -100 -50 0 50 100

Horizontal strains (µstrains)

Dept

h (c

m)(tensile strains are positive)

failure area (?)

New structure 4: dammage mechanism


Trench in the new structure 4 at the end of the experiment

HMAC 6 cm

BFTA 10 cm

HMAC 6 cm

Subgrade (spoilt micaschist)

Failure areahorizontal shear crack

Wheel path

Numerical modeling: main results


Theoritical modeling:• multilayer elastic model (Burmister). Young moduli are back-

calculated from deflections, curvature radii and strains measured.Young modulus (15°C, 10Hz)

- reference AC : 10 000 Mpa- BFTA : 4 000- ESA : 2 500- Cracked AC : 2 000

• Fatigue behavior evaluated from fatigue laboratory tests, and calibrated by ajustement with experimental results.

• Scattering of the fatigue parameters are taken into account, leading to a probabilist determination of the pavement dammagedue to the traffic.

• The theoretical risk of failure is assimilated to the cracking extent.• New structure 4 : shear failure plane is modelized by a low

modulus thin layer (2 cm thick, at 2 cm above the BFTA/HMAC interface, Young modulus 750 MPa).



New structure 4 : internal shear en tensile stresses

Shear stresses at the midlepart of the BFTA layer

Vertical distribution of stresses in the BFTA

-16

-14

-12

-10

-8

-6

0 0.05 0.1 0.15 0.2 0.25 0.3Stresses (MPa)

Depth

(cm)

Major stress S1Maximal shear stress

(tensile stresses are positive)

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

-1.5 -1 -0.5 0 0.5 1 1.5Distance to the load axle (m)

Shea

r stre

ss σ

xz (M

Pa)

Standard twinned wheel load (65kN)



Measured and calculated evolution of cracked area

0

20

40

60

80

100

0.00 0.50 1.00 1.50 2.00 2.50 3.00Equivalent traffic - 130 kN standard loads (millions)

Crac

king

exte

nt (

%)

ST1 reference AC - measures

ST2 ESA - measures

ST3 BFSA - measures

ST4 HMAC/BFSA/HMAC measures

ST1 reference AC - model

ST2 ESA - model

ST3 BFSA - model

ST4 HMAC/BFSA/HMAC model

General Conclusions


APT is a useful tool for :• Identification of pavement deterioration mechanisms, and suitable

models • Validation of models• Experimentation of innovative structural design• Improvement of laboratory test proceduresBut validation on real pavement sections is also necessary

Recent research at LCPC on design /performance models focuses in particular on :

• Prediction of rutting

• Design of airfield pavements or special pavements (ex: industrial platforms) subject to heavy, complex loads

• Viscoelastic behavior of bituminous pavements and development softwares based on analytical models such as Viscoroute-LCPC.

References – Research on pavement modelling


•Bitumen Cracking modeling (Nguyen, 2006) (Chailleux, ICAP2006)

•Damage modeling (Bodin, 2002) (Bodin et al., ASCE2004), ….

•Visco-elastic structural model (Duhamel et al., BLCPC2005) (Chabot et al, ICAP2006), …

•Cracking structural model (Tran, 2004) (Chabot et al., BLPC2005), ….

•Visco-plastic modeling (Nguyen, 2006) (Nguyen et al., ICAP2006), …

•Non linearity modeling for GNT and soil material (El Abd, 2006) (Hornych et al., RMPD2006 or 2007) …

Additionnal not presented

Model for the prediction of rutting in unbound pavement layers

Developed in European project SAMARIS (2002-2006)

SAMARIS


• SAMARIS : European project of the 5th PCRD• 2002-March 2006• Pavement and Structure Streams

• Task 5 : Development of a performance-based approach for the prediction of rutting of unbound pavement materials

• Selection of permanent deformation models for unbound granular materials

• Development of a structural method of calculation of rutting of unbound pavement layers

• Comparison with results of ALT full scale pavement experiment

Laboratory study of permanent deformations


Test method : cyclic triaxial test Advantages : realistic simulation of « stress paths » due to traffic loading

Test proceduresTest equipment

p = (σ1+2σ3)/3 q = (σ1-σ3)

0

100

200

300

400

500

600

0 100 200 300 400

q/p = 1

q/p = 1,5

q/p = 2

q/p = 2,5

P (kPa)

q (kPa)

0

20

40

60

80

100

120

0 50000 100000 150000 200000

Number of cyclesPe

rman

ent

axia

l str

ain

(10-

4 )

Selected permanent deformation models


2 modelling approaches :

• Routine level : utilizable for designEmpirical permanent deformation model

• Advanced level : for research or analysisElasto-plastic model with isotropic and kinematic hardening Chazallon (2000)

)q,p(g).N(f)N( maxmaxp =ε 1 Gidel (2001)


Example of calibration of empirical permanent deformation model

[ ])(

1.1.)(

max

max

max

011

pq

psmp

LNNn

a

Bpp

−+⎥⎦

⎤⎢⎣

⎡⋅−= −εε Gidel (2001)

0

50

100

150

200

250

300

0 50000 100000 150000 200000 Number of cycles

ε 1p (1

0-4)

∆q/∆p = 1

∆q/∆p = 2

∆q/∆p = 2,5

Empirical model

Maraichère UGM

Cyclic Triaxial Tests atw = 5 %

Structural modeling approach


Main hypothesis : For one cycle

Separate modelling of resilient behaviour and permanent deformations

Three steps :

1. 3D Finite Element calculation of the stress fields in the pavement structure using the resilient behaviour (non linear elastic, visco-elastic models)

2. Use of the stress fields and stress path to calculate permanent strains at the different points of the pavement in the vertical transversal plane

3. Calculation of the displacement field (rutting)

ep ε<<εδ

- FEM method (program ORNI) : 3D structural calculation.- Simplified method: integration of ε1

p in the vertical direction

Modelling of a full scale experiment


5 low traffic pavement structures (each 25 m long)Full scale loading conditions : 65 kN dual wheel load, 72 km/h1.5 million loads appliedLow water table level ( –2.6 m)

Experiment performed in 2003 on the LCPC fatigue test track

Modelling of structure 4


14

6

10

1821

18

10

64

10

5

10

15

20

25

-50 -40 -30 -20 -10 0 10 20 30 40 50position (cm)

perc

enta

geof

load

s(%

) Lateral load distribution

0

5

10

15

20

25

30

0 500000 1000000 1500000 2000000Number of loads

rut

dept

h(m

m)

0

20

40

60

80

100

120

perc

enta

geof

sur

face

cr

acke

d(%

)

mean rut depthmin. rut depthmax. rut depthextent of cracking (%)

UGM

Soil

Structure 46.5 cm

50 cm

2.24 m

Evolution of rutting and cracking

Modelling hypotheses


Modelling of rutting of UGM layer and subgrade (empirical model)Simulation of load wandering and variations of temperature with traffic

Temperature distribution used for calculations

Per

cen

tage

of a

pplie

dlo

ads

(%)

0

6

18

31 29

16

0 00 0

5

17

60

19

0 00 0

42

24

16 16

200

4

42

26

8

2 20 1

34 33

29

2 1 0

11

21

52

13

30 0 0

16

0

10

20

30

40

50

60

70

7-10 10-15 15-20 20-25 25-30 30-35 35-40 40-45

0-50.000 Loads

50.000-100.000 Loads

100.000-200.000 Loads

200.000-500.000 Loads

500.000-1.000.000 Loads

1.000.000-2.000.000 Loads

Temperature (°C)

Examples of rut depth calculations


220cm

• Bituminous concrete : linear elastic• UGM : non linear elastic+ empirical permanent deformation model• Soil : linear leastic E = 100 MPa, n = 0,35

Pavement structure Loading : 65 kN load (single or dual wheel)

1.5 million loads - Constant temperature8cm

50cmUGM

SOIL

Stress paths in UGM layer

0

100

200

300

400

500

600

0 50 100 150 200 250p (kPa)

q (k

Pa)

z = -0.18 m

z = -0.43 m

3D mesh modelling of resilient behaviour

Calculation : influence of lateral load wandering


0

0.5

1

1.5

2

2.5

3

0 400000 800000 1200000 1600000

Number of loads

Max

imu

m r

ut

dept

h(m

m)

With wanderingWithout wandering

Low influence of lateral wandering

No lateral Lateral wandering wandering (11 positions )

Evolution of maximum rut depth

Rut profiles after 1.5 million loads (single wheel) :


0123456789

10

0 400000 800000 1200000 1600000Number of loads

Max

imu

m r

ut

dept

h(m

m)

T=27,5°

T=23°

T=35°

T=15°

-1400

-1200

-1000

-800

-600

-400

-200

0

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

Lateral position y (m)

w (

mm

/10

0)

Rut profiles for different Temperatures (8.5 to 42.5°C)

Large influence of temperature

Evolution of maximum rut depth for different temperatures :

Influence of wearing course temperature

Comparison of model with experiment


0

2

4

6

8

10

12

0 500000 1000000 1500000 2000000Number of Loads

max

imum

rut d

epth

(mm

)

ORNI - w = 5 %ORNI - w = 4 %Min. Mean experimentMax.

0

2

4

6

8

10

12

0 500000 1000000 1500000 2000000Number of loads

Max

imum

rut d

epth

(mm

)

Max. Mean experimentMin. ORNI w = 4 % + soilRoutine method

Rutting of UGM only

Rutting of UGM + soil

Conclusions – Modelling of rutting


• First results encouraging but difficulty to simulatereal in situ conditions (temperature and moisture variations…)

• Models predict a too fast stabilisation of permanent deformations

Perspectives :

- more detailed evaluation of ORNI

- Improvement of the models for unbound materials

- Modelling of permanent deformations of bituminous materials

Documents

Impact of Accelerated Pavement Testing (APT) on Pavement ...€¦ · Impact of Accelerated Pavement Testing (APT) on Pavement Engineering Jean-Maurice Balay Jean-Michel Piau Pierre