Calibration of a Mechanistic-Empirical

Fatigue Model under Different Moisture

Conditions

Eliécer Arias-Barrantes
José P Aguiar-Moya, Ph.D.
Luis G Loría-Salazar, Ph.D.
Edgar Camacho-Garita
Presentation Outline
Introduction

APT program

Test sections

Instrumentation

Results

Conclusions
 NANO

MACRO
MICRO
Introduction

• There is a great need to address pavement

performance issues in tropical regions.

• There is limited data for this regions to allow adequate

long term performance of pavement structures

APT is considered an adequate alternative to develop

.pavement performance models
APT Program

• A Costa Rican APT program was implemented

(PaveLab) in 2012, relying on a Heavy Vehicle
Simulator (HVS) since it was considered the best option
for the local and regional needs.
• Specifically, the PaveLab had to meet the following
.requirements:
1. Mobility.
2. Accelerated pavement evaluation.
3. Application of real loads.
4. Comparable results to similar equipment
APT Program
PaveLab´s goals:

• Mechanistic-empirical pavement design methodology and

software based on material conditions, weather, traffic and actual
construction practices.

• Development of new material specifications based on actual

performance and contribution of structural materials in the field.

• Optimization of pavement structures in use at the national level,

based on structural, materials, traffic, and climatic conditions
specific to the area where the structure is planned to be built.
APT Program
APT Program
• The initial set of experiments performed at PaveLab
corresponds to four structures detailed in the next figure

This paper
 Test Sections
• The models are based on 2 test sections with your replicas on
wet conditions.

Section 001 003 008 007

Properties AC1 AC2 AC1 AC2

AC Thickness (H1), cm 6.1 6.3 6.1 6.3

Base Thickness (H2), cm 21.9 21.2 21.9 21.2

Subbase Thickness (H3), cm 30.1 30.1 30.1 30.1

AC Modulus (E1), MPa
3500 3500 3500 3500
[@ 25°C, 1.5 Hz]
Base Modulus (E2), MPa 1200 115 1750 300

Subbase Modulus (E3), MPa 142 75 500 100

Instrumentation
• The measurements were performed using the HVS integrated
instrumentation and embedded sensors in all four test sections.

Embedded sensors included:

• Asphalt strain gauges
• Pressure cells
• Multi-depth deflectometers
(MDDs)
• Moisture and temperature
Instrumentation
• The HVS was equipped with a laser profiler that can be used to
recreate a three-dimensional profile of the section.
• A road surface deflectometer (RSD) is also used to obtain
deflection basins.
Results
PROPOSED FATIGUE MODEL
β γ
 resp   E  δT
Damage = A × MN x α  x  xe (1)
 resp  E 
 ref   ref 
Where
MN = the number of load repetitions (ESAL) in million
resp = the response (stress or strain)
respref = a reference response (can be related to strength)
E = the modulus of the material (adjusted for climate and damage in MPa)
Eref = a reference modulus (MPa)
T = temperature in °C
A, α, β, and γ are model constants.
Results
Four Point Bending Beam (4PBB) Fatigue Tests

Damage of the asphalt mixture was defined as the relative decrease in modulus dE
relative to the initial modulus Ei
For each scenario; this was used to calibrate the following model:

 ε 
1.07 0.535
 E  (2)
ω = 0.189 × ( MN )0.271 ×   ×  × e(0.035×T )
 200   3000 
Where:
MN = the number of load repetitions (ESAL) in millions
ԑ = tensile microstrain
E = material modulus (MPa)
T = temperature (°C).
Results
Back calculated Layer Moduli

(3)

Where:
k1, k2, k3 and k4 = calibration coefficients corrected with HVS data.
 Results
Damage on AC2 optimal moisture condition
1,6

1,4
Damage
1,2 labaratory
regresion
1
Damage

Calibration with
0,8 HVS data

0,6
AC2 dry Damage
0,4

0,2

0
0 1 2 3 4 5 6 7 8
Mesal
 Results
Damage on AC2 wet condition
1
0,9
0,8
0,7
0,6 Damage laboratory
Damage

0,5 regresion
0,4
0,3
Calibration with
HVS data
0,2
0,1
AC2 layer damage
0
0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4 0,45 0,5
Mesal
 Results
Damage on AC1 optimal moisture condition
0,8

0,7

0,6

0,5
Damage

0,4 AC1 layer damage

0,3
Damage laboratory
0,2 regresion

0,1 Calibration with

HVs data
0
0 2 4 6 8 10 12
Mesal
 Results
Damage on AC1 wet condition
0,7

0,6

0,5

0,4
Damage

AC1 layer damage

0,3
Damage laboratory
regresion
0,2
Calibration with HVS
0,1
data

0
0 0,1 0,2 0,3 0,4 0,5 0,6
Mesal
Results
Results
 Results

Section AC1 Dry conditions Section AC1 Wet conditions

Damage level Damage level
Coefficients Coefficients
60% 70% 90% 60% 70% 90%
k1 0.040 0.061 0.124 K1 0.004 0.006 0.012
k2 -3.006 K2 -2.784
k3 -1.503 K3 -1.392
k4 -0.099 K4 -0.092
 Results

Section AC2 Dry conditions Section AC2 Wet conditions

Damage level Damage level
Coefficients Coefficients
60% 70% 90% 60% 70% 90%
k1 276.89 425.62 857.89 K1 10.16 15.78 32.36
k2 -2.983 K2 -3.056
k3 -1.492 K3 -1.528
k4 -0.098 K4 -0.100
Conclusions

• It was verified that the 4PBB regression equation does

not adequately match with the full-accelerated
pavement test performed on HMA thin layer sections.

• The calibration of the fatigue model coefficients

between 4PBB and APT sections are satisfactory for
each of the test tracks.

• (The fatigue models were calibrated for strain levels

corresponding to the pavements AC1 - AC2)
Conclusions

• Current tests will allow the analysis of thick HMA layer

behavior under different humidity conditions.

• These efforts will allow the calibration of a more robust

model based on additional data that considers:
• Damage as a function of load applications, tensile
deformation, initial stiffness and moisture.
• Different types of base layers (CTBs vs granular).
• Different layer thicknesses.
Thank You!

Thank you!