Professional Documents
Culture Documents
Presented By:
Afzal Waseem
Supervisor:
Outline
Introduction What is the MEPDG ? What is the Problem ?
The Local Calibration Methodology Reconstructed Sections Rehabilitated Sections Conclusions & Recommendations
2
Introduction
To develop a Local Calibration Methodology Local calibration based on limited PMS database Multiple local optima during local calibration
Pavement Design
AASHTO (1993): Conventionally pavements are design based on AASHTO design guide. Empirical Design Method Only 9 Inputs (ESAL, Mr, Pt, a, m) Long term effects on pavements not considered.
INPUTS
MECHANISTIC
EMPIRICAL
RESULT
1. 2. 3. 4.
5. Permanent deformation
6. International Roughness Index
Inputs
Location: Latitude, Longitude & Elevation Climate Information Air Temperature Precipitation Moisture content Sunshine Wind speed Freezing Index
Wind speed
Sunshine
Vehicle Inputs Axle spacing Axle configuration Traffic Wander Axle width Dual tire spacing Tire Pressure Operational speed Vehicle class AC Layer Unit weight Air void Effective binder content Poissons Ratio Tensile Strength Thermal conductivity Heat capacity Sieve Passing
Traffic Information AADTT Growth rate Truck distribution Truck traffic classification No. of Lanes Monthly and Hourly Adjustment factor
W/T depth
(Waseem 2013)
Base, Sub-base and subgrade Layers Thickness Poissons Ratio Resilient Modulus Gradation
Permanent Deformation
Permanent deformations can be defined as the combination of transverse undulations, longitudinal depressions and distortions on the surface of the pavement.
a) Surface Rutting
b) Trench Study
2.
= 1
3.
= 1
(Jannat, 2012)
= = = = = 1 Default models are evaluated using pavement sections all over Ontario. Over-prediction of rutting for Ontario. 9
Local Calibration
Calibration is a systematic process to eliminate bias and minimize the residual error between observed and predicted values of a certain distress.
Observed Observed 8 8 7 7 Rutting (mm) 6 6 5 5 4 4 3 3 2 2 1 1 0 0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Before After Calibration
2007
=
=1
LONGITUDNAL CALIBRATION
Instead of only calibrating the end of cycle distress values, we calibrate the predicted distress curve over time for each pavement sections.
2008
10
Pavement Sections
12
Database
1. MEPDG Default values Guide for Ontario recently developed by the MTO staff.
2. MEPDG project inputs used in this research was collected from MTO Pavement Management System (PMS-2) Database.
AC Rutting
Granular Rutting
Subgrade Rutting
13
GB
AC AC1=0.3, AC2=1, AC3=1
14
8
6 4 2 0 1996
Observed (0.6, 1.0, 0.2) - 3.89 (0.3, 0.9, 0.3) - 3.46 (0.4, 0.3, 0.4) - 3.46 (0.35, 0.6, 0.35) - 3.45 (0.2, 0.1, 0.5) - 3.45
1998 2000 2002 Year 2004 2006 2008 2010
5 local optima showed similar rut predictions along time Calibration coefficients differ from each other Individual layer calibration is needed
15
16
32 %
58 %
9%
30
20 10 0 AC Base
AC Layer =50 %
Subgrade Layer = 0 %
Rehabilitated Sections
17
18
MATLAB
4. Termination Criteria
19
Local Calibration
1. Section-by-section Calibration
o Calibration section was individually calibrated for the available rut points throughout the service life of the pavement which is called the section-by-section longitudinal calibration.
2. Pooled Calibration
o The rutting RSS of all sections are pooled together to calculate the total RSS which is then minimized to get a single set of calibration parameters. o For comparison purpose, a calibration was also performed. pooled longitudinal
20
RSS
Granular 0.2378 0.2281 0.2128 1.0798 0.7121 0.3104 0.4038 0.2239 1.3401 2.8296 Subgrade 0.0074 0.0051 0.0047 0.0243 0.0163 0.007 0.0082 0.0055 0.0363 0.063
Total Bias
0.12 -0.075 -0.002 0.038 -0.018 0.055 -0.737 -0.303 -0.319 -0.158
Total RSS
0.6601 0.5649 0.4363 2.6932 1.9728 0.8067 2.9352 1.0645 3.4449 8.2359
21
RECONSTRUCTED SECTIONS
(a)
R2 = 0.8824 N = 68 Se = 0.583 10 9 8 7 6 I.D. # 9 I.D. # 43 I.D. # 191 I.D. # 376 I.D. # 1049 I.D. # 1053 I.D. # 1188 I.D. # 1189 I.D. # 1200 I.D. # 1311 0 1 2 3 4 5 6 7 8 Predicted Rutting [mm] 9 10 11 R2 = 0.016 N = 68 Se = 2.217
(b)
5
4 3 2 1 0
After-calibration prediction versus observation plots. (a) section-by-section calibration; (b) pooled calibration. 22
RECONSTRUCTED SECTIONS
2
Observed Section-by-section Pooled
3
2
0.5
2004
1997
1998
1999
2000
2001
2002
2003
2005
2006
2007
2001
1997
1998
1999
2000
2002
2003
2004
2005
2006
2007 2007
Pavement Service life [years] 0.6 0.5 Rutting [mm] 0.4 0.3 0.2 0.1 Rutting [mm] Observed Section-by-Section Pooled
2
1
2008
23
REHABILITATED SECTIONS
The pre-overlay rut is an important input parameter which has significant influence on the calculation of total surface permanent deformation of rehabilitated sections. 1. Frequency Distribution of the pre-overlay rut of 1175 flexible pavement sections
gave 7 mm average value.
2. Regression Analysis of pre-overlay rut showed poor correlation with DMI and RCI
Pre-overlay rut value of 7mm was used for rehabilitated pavement sections whose pre-overlay rut value was unavailable.
24
REHABILITATED SECTIONS
I.D.
139 217 347 348 349 350 353 356 357 358 361 377 378 379 386 803 811 951 981 1.01 0.9 1.01 0.276 1.05 1.1 0.49 0.27 0.98 0.225 0.366 0.44 0.3 1.01 0.295 0.356 0.4 0.287 1.25
Rutting Model AC
1.23 1.23 1.23 1.601 1.255 1.155 1.235 0.935 1.055 1.447 1.408 1.175 1.08 1.23 1.166 1.115 1 1.157 1.17
Granular
1.09 1.274 1.86 1.124 1.297 1.308 1.303 0.713 1.337 0.726 0.614 0.853 0.985 1.336 0.804 1.14 2.245 1.237 0.641
Layers RSS AC
0.982 0.299 0.444 0.22 0.238 0.467 0.242 0.324 0.109 0.257 0.181 1.067 1.071 0.571 0.296 0.245 0.085 0.567 1.878
0.77 0.77 0.77 0.581 0.745 0.85 0.865 1.185 0.945 0.774 0.721 0.925 1.13 0.77 0.989 0.935 1 0.983 0.82
Granular
0.52 0.215 0.241 0.362 0.419 0.962 1.463 0.757 0.525 0.538 0.277 2.064 6.596 1.078 0.762 0.112 0.112 0.391 2.731
25
REHABILITATED SECTIONS
I.D. # 139 I.D. # 217 I.D. # 347 I.D. # 348 I.D. # 349 I.D. # 350 I.D. # 353 I.D. # 356 I.D. # 357 I.D. # 358 I.D. # 361 I.D. # 377 I.D. # 378 I.D. # 379 I.D. # 386 I.D. # 803 I.D. # 811 I.D. # 951 I.D. # 981 13 12 11 10 Observed Rutting (mm) 9 8 7
6
5 4 3 2 1 0 0 1 2 3 4 5 6 7 8 9 Predicted Rutting (mm) 10 11 12
I.D. # 139 I.D. # 217 I.D. # 347 I.D. # 348 I.D. # 349 I.D. # 350 I.D. # 353 I.D. # 356 I.D. # 357 I.D. # 358 I.D. # 361 I.D. # 377 I.D. # 378 I.D. # 379 I.D. # 386 I.D. # 803 I.D. # 811 I.D. # 951 I.D. # 981
26
Summary
1. Developed two stage layer-by-layer, section-by-section local calibration methodology. o Multiple Local Optimums o Layer Contribution studies
2. Automated methodology using DARWin-METM coupled with Excel Macros. 3. Local Calibration of 9 Reconstructed Sections. 4. Local Calibration of 19 Rehabilitated Sections. o Pre-overlay study
27
Conclusions
1. For local calibration with PMS data, layer-by-layer calibration has to be done, otherwise there would be multiple local optima. 2. Based on the layer contribution of rutting from AASHO study, section-by-section calibration has higher accuracy than pooled calibration but pooled calibration is still better than the results from global calibration. 3. The pooled calibration parameters ( = . , = . , = . , = . and = ) of rehabilitated pavement sections has better R2 (0.464) and standard error (1.608 mm) than that obtained during calibration of global rutting models. 4. The level-3 pre-overlay rut input value of 7 mm is proposed.
28
Recommendations
1. Layer contribution study for flexible pavement sections in Ontario and its sensitivity to major input parameters is desperately needed. 2. New pavement sections should be included in this database before recalibration of MEPDG rutting model for Ontario. Similarly information for more reconstructed sections should be collected to enhance the local calibration study. 3. It is recommended that MTO should collect more rehabilitated pavement sections whose performance history and pre-overlay rut information is available.
29
References
Waseem, A. (2013). "Methodology Development and Local Calibration of MEPDG Permanent Deformation Models for Ontario's Flexible Pavements." MASc Civil Engineering, Ryerson University, Toronto. Waseem, A., Yuan, X.-X., and Shehata, M. (2013). "Evaluation of Pre-overlay Rut with MEPDG Rutting Model for Flexible Pavements using Ontario's Long Term PMS Data." CSCE 3rd Specialty Conference on Engineering Mechanics and Materials, Montreal, Quebec Waseem, A., and Yuan, A. (2013). "Longitudinal Local Calibration of MEPDG Permanent Deformation Models for Reconstructed Flexible Pavements Using PMS Data." 8th International Conference on Road and Airfield Pavement Technology Taipei, Taiwan. http://www.scribd.com/doc/146792015/Met hodology-Development-and-LocalCalibration-of-MEPDG-PermanentDeformation-Models-for-Ontario-s-FlexiblePavements-Thesis-by-Afzal-Waseem http://www.scribd.com/doc/149970728/Ev aluation-of-Pre-overlay-Rut-Depth-forLocal-Calibration-of-the-MEPDG-RuttingModel-in-Ontario-Pavements
http://www.scribd.com/doc/149984444/Lo ngitudinal-Local-Calibration-of-MEPDGPermanent-Deformation-Models-forReconstructed-Flexible-Pavements-UsingPMS-Data
30
Questions
32