Further Calibration of the Distress Further Calibration of the Distress
Prediction Models & Reliability EffectsPrediction Models & Reliability Effects
ME PDG Rigid Pavement
Design Reliability Update
Expanding the Realm of Possibility
NCHRP 1NCHRP 1--40B & 140B & 1--40D Team40D Team
Applied Research Associates
� Michael Darter
� Jagannath Mallela
� Harold Von Quintus
� Gregg Larson
� Leslie Titus-Glover
� Chetana Rao
� Dulce Rufino
� Alex Gotlif
U of Minnesota
� Lev Khazanovich
� Graduate students
NCHRP
� Ed Harrigan & Panel
ASU
� Matt Witczak
� Mohamed El-Basyoumy
� Claudia Zapata
� Graduate students
Expanding the Realm of Possibility
PresentationPresentation
�Reliability Definition M-E PDG
�2004 Calibration
�2005 Independent validation
�2006 Recalibration
� Illustrations of project predictions
�Conclusions
Expanding the Realm of Possibility
MM--E PDG Design Reliability E PDG Design Reliability
DefinitionDefinition–– Rigid PavementRigid Pavement
� JPCP
� RF = P [Fault < Critical Fault]
� RC = P [Crack < Critical Crack]
� RIRI = P [IRI < Critical IRI]
� CRCP
� RPO = P [Punchout < Critical PO]
� RIRI = P [IRI < Critical IRI]
Expanding the Realm of Possibility
MM--E PDG Design Reliability E PDG Design Reliability
Output Output –– Rigid PavementRigid Pavement
0
40
80
120
160
200
240
280
320
360
400
0 2 4 6 8 10 12 14 16 18 20 22
Pavement age, years
IRI,
in
/mil
e
IRI at 50%
IRI at 95%
Reliability,
Residual
Critical IRI
Expanding the Realm of Possibility
MM--E PDG Design Reliability Output E PDG Design Reliability Output
–– Rigid PavementRigid Pavement
Distress
Target
Reliability
Target
Distress
Predicted
Reliability
Predicted Acceptable
172 95 130 83.15 Fail
15 95 1.6 99.98 Pass
0.15 95 0.089 93.13 Fail
Transverse Cracking (% slabs cracked)
Mean Joint Faulting (in)
Performance Criteria
Terminal IRI (in/mi)
Expanding the Realm of Possibility
Design Reliability ApproachDesign Reliability Approach
� IRI model: closed form variance solution:
Var {IRI} = Var {Initial IRI} + Var {Distress} +
Var {Site Factors} + Var {Residual}
� Distress models: used model residuals (predicted – measured) determined from calibration results
Expanding the Realm of Possibility
Residuals from Calibration Residuals from Calibration
� Residual = predicted –measured
� Causes� Random (replicate) variation
� Input selection error
� Distress/IRI measure error
� Difference between project mean and 500-ft section
� Model deficiency residual
� Relative magnitude of each?
0
0.05
0.1
0.15
0.2
0 3 6 9 12 15 18
Age, years
Mean tra
nsverse joint
faulting, in
LTPP data 1-37A model
Section Faulting History
Resid
Expanding the Realm of Possibility
Residuals from Calibration Residuals from Calibration
� Residuals represent the
knowledge that existsof the accuracy of the distress prediction
model.
� Large model deficiency residuals indicates we
are not explaining the
physical phenomenon
well.
Predicted Vs Measured Slab Crack
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
Measured percent slabs cracked
Pre
dic
ted
pe
rce
nt
sla
bs
cra
ck
ed
R2 = 0.86
SEE = 5.02 percent
N = 1585
Expanding the Realm of Possibility
y = 0.9698x
R2 = 0.8445
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
Measured percent slabs cracked
Pre
dic
ted
pe
rce
nt
sla
bs
cra
ck
ed
Derivation of Cracking Standard Deviation FunctionDerivation of Cracking Standard Deviation Function
21 3 4
Expanding the Realm of Possibility
Predicted vs. Standard Deviation of Measured Predicted vs. Standard Deviation of Measured
CrackingCracking
STDC = -0.00172*CRACK2 + 0.3447*CRACK + 4.6772
0
5
10
15
20
25
0 20 40 60 80 100
Predicted cracking, percent
Sta
nd
ard
dev
iati
on
s, p
erce
nt
cra
ckin
g
measured
predicted
Expanding the Realm of Possibility
JPCP Cracking ReliabilityJPCP Cracking Reliability
CRACKR = predicted cracking at reliability level R,
percent slabs
CRACK50= predicted cracking based on mean inputs
(50% reliability), percent slabs
STDC = standard deviation of cracking at predicted
level of mean cracking
ZR = standard normal deviate (one-tailed)
CRACKR
= CRACK50
+ STDC
• ZR
Expanding the Realm of Possibility
Residuals of PredictionResiduals of Prediction
� Variations/Uncertainties NOT included:� Future design traffic estimation error
� Calibration database limitations (short sections, missing some design and materials, etc.)
� Others???
Expanding the Realm of Possibility
2004 Sections Used 12004 Sections Used 1--37A37A
22 States55CRCP LTPP GPS-5
20-years, US 40, IL6CRCP Vandalia, IL
Chicago, Illinois11CRCP Other
23 States196JPCP LTPP GPS-3 & SPS-2
14 States
10 States
26
42
Rehab JPCP CPR, OL
Rehab CRCP OL
9 States36JPCP FHWA RPPR
LocationNo. of SectionsProjects
Expanding the Realm of Possibility
2005 Validation: NCHRP 12005 Validation: NCHRP 1--40B40B
�Objective was to evaluate adequacy of models developed under NCHRP 1-37A for ME PDG system.
�Criteria
� Pavement sections NOT used in original calibration under NCHRP 1-37A.
� Pavement sections had mostly level 1 and 2 data available for analysis.
Expanding the Realm of Possibility
2005 Test Sections NCHRP 12005 Test Sections NCHRP 1--40B40B
Minnesota9MnRoad
CT, IA, MI, TX4LTPP GPS-5 (CRCP)
South Carolina12I-77 (CRCP)
Illinois25
Extended AASHO
(Interstate 80)
Virginia4“Smart” Road (CRCP)
Iowa12LTPP SPS-2
Arkansas12LTPP SPS-2
Arizona8
LTPP SPS-2 (Supplemental)
LocationNo. of SectionsProjects
Expanding the Realm of Possibility
2006 Recalibration: NCHRP 12006 Recalibration: NCHRP 1--40D40D
� Objective was to recalibrate models included in the original M-E PDG.
� DATA: Major increase in quantity & quality of performance data.
� Original sections: Updated observed distress/IRI from 2000 to 2005.
� Added new JPCP, CRCP, and Rehab sections.
� Combined database: 2004, 2005, & 2006
LTPP GPS-3 JPCP Sections LTPP SPS-2, MnROAD, & AASHO JPCP Sections
Expanding the Realm of Possibility
LTPP GPS-5 CRCP Sections
Illinois, South Carolina, & Virginia CRCP Sections
Expanding the Realm of PossibilityLTPP, NCHRP 10-41, ACPA Diamond Grinding Sections SPS-6 Sections
Expanding the Realm of Possibility
2006 Recalibration: JPCP2006 Recalibration: JPCP
Minnesota9MN Road
Illinois25Extended AASHO
(AASHO + I-80)
25 States260LTPP GPS-3, SPS-2
LocationNo. of SectionsProjects
Note: Most sections have multiple observations over time
Expanding the Realm of Possibility
2006 Recalibration: CRCP2006 Recalibration: CRCP
22 States59LTPP GPS-5 (CRCP)
South Carolina12I-77 (CRCP)
Virginia4 (crack spacing)“Smart” Road (CRCP)
Chicago, Illinois10 (heavy traffic)Illinois Interstate
1947-67 US 40
Vandalia, IL 6Vandalia, IL
LocationNo. of SectionsProjects
Note: Most sections have multiple observations over time
Expanding the Realm of Possibility
2006 Recalibration: Rehab2006 Recalibration: Rehab
15 States30
JPCP CPR and OL
(SPS-6, GPS-9, NCHRP
10-41, Others)
To be completedLTPP SPS-6 (HMA
overlays)
10 states42CRCP Overlays (GPS-9,
NCHRP 10-40, others)
LocationNo. of Data Points
Projects
Note: Most sections have multiple observations over time
Expanding the Realm of Possibility
2006 Recalibration: NCHRP 12006 Recalibration: NCHRP 1--40D40D
�Calibration process
� Preparation of expanded database (huge effort)
� Improvement of models and algorithms & software modification
� Calibration process, models
� Preparation of final calibrated software
Expanding the Realm of Possibility
2006 Recalibration: NCHRP 12006 Recalibration: NCHRP 1--40D40D
�Calibration process documentation
�NCHRP Research Results Digest 308: >300 software and engineering improvements Version 0.900
�Many additional improvements for Version 1.000 (Feb 2007)
Expanding the Realm of Possibility
JPCP Cracking Model 2006JPCP Cracking Model 2006
2
11
1(%)
CFDC
Cracking+
=
Model Coefficients
1-37A: C1 = 1.0 ; C2 = -1.68
1-40D: C1 = 1.0; C2 = -2.00
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
Measured percent slabs cracked
Pre
dic
ted
pe
rce
nt sla
bs c
racke
dR2 = 0.86
SEE = 5.02 percent
N = 1585
Expanding the Realm of Possibility
JPCP Unbonded Overlay Cracking Model 2006JPCP Unbonded Overlay Cracking Model 2006
2
11
1(%)
CFDC
Cracking+
=
Model Coefficients
1-37A: C1 = 1.0 ; C2 = -1.68
1-40D: C1 = 1.0; C2 = -2.00
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
Measured percent slabs cracked
Pre
dic
ted p
erc
en
t sla
bs c
racke
dR
2 = 0.72
SEE = 3.95 percent
N = 60
Expanding the Realm of Possibility
JPCP CPR Cracking Model 2006JPCP CPR Cracking Model 2006
2
11
1(%)
CFDC
Cracking+
=
Model Coefficients
1-37A: C1 = 1.0 ; C2 = -1.68
1-40D: C1 = 1.0; C2 = -2.00
0
10
20
30
40
50
60
70
80
90
100
0 20 40 60 80 100
Measured percent slabs cracked
Pre
dic
ted
perc
en
t sla
bs c
racked
R2 = 0.90
N = 94
SEE = 6.5 percent
Expanding the Realm of Possibility
JPCP Joint Faulting Model 2006JPCP Joint Faulting Model 2006
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.05 0.1 0.15 0.2 0.25 0.3
Measured mean transverse joint faulting, in
Pre
dic
ted
me
an
tra
nsve
rse
jo
int fa
ultin
g, in
R2 = 0.62
SEE = 0.0276 in
N = 1260
Expanding the Realm of Possibility
JPCP CPR Faulting Model 2006JPCP CPR Faulting Model 2006
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Measured transverse joint faulting, in
Pre
dic
ted
tra
nsv
ers
e j
oin
t fa
ult
ing
, in
R2 = 0.61
N = 40
SEE = 0.02 in
Expanding the Realm of Possibility
CRCP Punchout Model 2006CRCP Punchout Model 2006
3
2
1
1C
FDC
CPO
+
=
Model Coefficients
1-37A: C1 = 105 ; C2 = 4; C3 = -0.38
1-40D: C1 = 196; C2 = 20; C3 = -0.50
y = 0.9993x
R2 = 0.7395
0
10
20
30
40
50
60
0 10 20 30 40 50 60
Predicted punchouts, #/mile
Mea
sure
d p
un
cho
uts
, #
/mil
e
R2 = 0.74
SEE= 3.6
N = 345
Expanding the Realm of Possibility
Comparison of Measured and Comparison of Measured and
Predicted JPCP IRIPredicted JPCP IRI
0
50
100
150
200
250
0 50 100 150 200 250
Measured IRI, in/mile
Pre
dic
ted
IR
I, i
n/m
ile N = 1148
R2 = 0.64
SEE = 13.7 in/mi
Expanding the Realm of Possibility
LTPP 6LTPP 6--3030, CA3030, CA
0
20
40
60
80
100
0 5 10 15 20 25 30 35
Pavement Age, years
Pe
rce
nt
Sla
bs
Cra
ck
ed
Measured (12-ft) Predicted (12-ft)
Measured (19-ft) Predicted (19-ft)
Expanding the Realm of Possibility
LTPP 12LTPP 12--3811, FL3811, FL
0
20
40
60
80
100
0 5 10 15 20 25 30 35
Pavement Age, years
Perc
en
t S
lab
s C
racked
Measured Predicted
Expanding the Realm of Possibility
LTPP 5LTPP 5--0214, AR0214, AR
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0 2 4 6 8 10
Pavement Age, years
Mean
Jo
int
Fau
ltin
g,
in
Measured Predicted
Expanding the Realm of Possibility
LTPP 53LTPP 53--3013, WA3013, WA
0
0.05
0.1
0.15
0.2
0.25
0 5 10 15 20 25 30 35
Pavement Age, years
Mean
Jo
int
Fau
ltin
g,
in
Measured Predicted
Expanding the Realm of Possibility
Crack Spacing in VA Smart RoadCrack Spacing in VA Smart Road
0
20
40
60
80
100
1 2 3 4
Cra
ck
Sp
ac
ing
, in
ch
M-E PDG prediction - friction limit and set temperature
Four year field data
ATB 12 CTB 14 ATB 12 CTB 14
Expanding the Realm of Possibility
Crack Width Crack Width at Steel Depthat Steel Depth
VA Smart RoadVA Smart Road
0
5
10
15
20
25
1 2 3 4
Cra
ck W
idth
, m
ils
ME PDG - at friction limit, set temperature
ME PDG - field crack spacing
Measured at surface
ATB 12 CTB 14 ATB 12 CTB 14
Expanding the Realm of Possibility
Punchout Prediction for LTPPPunchout Prediction for LTPP Texas 48Texas 48__53105310
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20 25 30
Pavement age, years
Pu
nch
ou
t p
er
mil
e Measured punchout
Predicted punchout
Expanding the Realm of Possibility
Punchout Prediction for LTPPPunchout Prediction for LTPP Connecticut 9Connecticut 9__05010501
0
20
40
60
80
100
0 5 10 15 20 25 30
Pavement age, years
Pu
nch
ou
t p
er
mil
e
Measured punchout
Predicted punchout
Expanding the Realm of Possibility
IRI Prediction for LTPP CRCP IRI Prediction for LTPP CRCP Iowa 19_5046Iowa 19_5046
25
50
75
100
125
150
175
200
225
250
0 5 10 15 20 25 30
Pavement age, years
IRI,
in
/mi
Predicted IRI
Measured IRI
Expanding the Realm of Possibility
Comparison 1Comparison 1--37A with 137A with 1--40D40D
N,
R2
Punchouts
Faulting
Cracking
Distress
345
0.74
993
0.64
1132
0.81
1-40D
512
0.69
220
0.67
564
0.74
1-37A
Expanding the Realm of Possibility
Local Calibration PotentialLocal Calibration Potential
� All models can be adjusted (Tools, Calibration, Coefs.)
� Key effect: Eliminate “bias”of prediction (significant over
prediction or under prediction
of distress).
� Possible effect: Reduce
residual of prediction
(depends on quality of data).
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Measured mean transverse joint faulting, in
Pre
dic
ted
mean tra
nsvers
e
join
t fa
ultin
g, in
R2 = 0.74
SEE = 0.025 inN = 43
Expanding the Realm of Possibility
ConclusionsConclusions
� The design reliability methodology used in the
ME-PDG is radically different than that used in
the 1986-93 version.
� Overcomes the major problem that exists with the old
AASHTO version for heavy traffic where the AASHTO model must be entered with extremely large ESALs
which extrapolates far, far, far beyond the accuracy of
the model.
� The result is extreme conservatism in thickness which results in large over design (and excessive thickness
often doesn’t extend life).
Expanding the Realm of Possibility
ConclusionsConclusions
� Reliability design methodology in ME-PDG is practical to use (not complicated). But, requires consideration of more than thickness.
� JPCP: thickness, dowel diameter, & joint spacing must be considered together. Other factors may be varied as well.
� CRCP: thickness and reinforcement content must be considered together. Other factors may be varied as well.
� Methodology appears to produce reasonable results over range of reliability
Expanding the Realm of Possibility
Reliability Level Impact for JPCP ProjectReliability Level Impact for JPCP Project
Design Reliability Vs Slab Thickness
7
8
9
10
11
12
13
14
15
60 80 90 95 99 99.9
Design Reliability
Sla
b T
hic
kn
ess, in
Expanding the Realm of Possibility
Reliability Level Impact for JPCP ProjectReliability Level Impact for JPCP Project
Design Reliability Vs Dowel Diameter
0.7
0.9
1.1
1.3
1.5
1.7
1.9
60 80 90 95 99 99.9
Design Reliability
Do
wel D
ia., in
Expanding the Realm of Possibility
ConclusionsConclusions
� Concepts applied uniformly between pavement
types and overlays and thus should produce
equitable designs.
� The value of “Reliability” (e.g., 90% or 95%)
should not be considered as exact probability, but
relative probability only.
Expanding the Realm of Possibility
ConclusionsConclusions
� Selection of Design Reliability values for
designs of low to heavy traffic requires local
consideration and should be a policy issue. It
should be based on consequences of success
or failure of the pavement to perform as
designed & on cost implications. Do NOT use
the old AASHTO recommendations for the new
ME-PDG reliability.
Expanding the Realm of Possibility
ConclusionsConclusions
� At the end of the day, the prediction models gain validity for:
�predicting in situ pavement performance, &
�showing reasonable practical sensitivity (see CA, TX, AR, IA, MN, & KS studies).
� This validation activity is far larger and more comprehensive than any other ever conducted to validate predictive engineering models.
� This widespread validation adds to improvement in design reliability.
Expanding the Realm of Possibility
Thank You!
Any Questions?