Development of Pavement Performance Prediction Models for

Development of Pavement Performance Prediction Models

for LADOTD PMS

Mohammad Jamal Khattak, Ph.D., P.E.,Associate Professor

Department of Civil Engineering,University of Louisiana at Lafayette

http://www.louisiana.edu/�

2

Layout

Introduction PMS Study: LTRC-04-2P Pavement Performance Prediction Models Network Condition Assessment Remaining Service Life Summary and Conclusions


3

PMS Study-LTRC 04-2P

Introduction Two Phase PMS study was initiated by

LADOTD Cost effective way to incorporate the PMS into

LA DOTD’s regular operation Make the information in the PMS usable for

engineers within the department


4

PMS Study-LTRC 04-2P

Objectives Identify the needs of the PMS users at LADOTD Establish a unified roadway identification system

acceptable to all PMS users Evaluate and update the existing pavement

condition assessment, and treatment selection models

Evaluate and update the existing pavement performance prediction models


5

Pavement Performance Prediction

Distress Index

Pavement Age

Preventive Maintenance

Minor Rehabilitation

Major Rehab/Reconstruction


6

Existing LADOTD Pavement Performance Models

Pavement Performance Prediction Models Based on Distress Index & Pavement Age Highway Classification

Interstate, Collectors and Arterials

Pavement Type Flexible, JCP, COM and CRC

Each Distress Type


7


Index Scale (I) 0 to 100 (100 being perfect) Calculated based on deduct points

Deduct Points (DP) Type of distress Extent of the distress Severity level

Low severity Medium severity High severity

I= 100- Σ(DP)


8


Regression based models for distress indices as function of Age

Transformation functions Roughness Index: Polynomial function All indices for CRCP: Power function Rutting Index: Exponential function All other: Linear function


Development of Pavement Performance Models


10

Development of Pavement Performance Models

DATA for Pavement Performance ModelingDistress Index Data for 1/10th mile for each

control section (1995-2005)Historical Data (TOPS)Last Resurface date (Surface Age)Construction/Reconstruction date (Age)


11

Pavement Performance Model

40

50

60

70

80

90

100

0 5 10 15 20

Years

Inde

x

Surface Age-1 Surface Age-2

Age

Threshold


12

Distribution of Element IDs (1/10th mile)

Pavement Type

Highway Classification (EL-ID)Total

IHS NHS SHS RHS

Flexible (Asphalt) 3,764 22,387 45,669 55,549 127,369

Composite 3,877 10,368 18,168 6,339 38,752

JPC 9,639 4,049 4,609 3,459 21,756

CRC 586 117 0 0 703

Total 17,866 36,921 68,446 65,347 188,580

5 x 188,580 = 942,900


13

Pavement Performance Models

Statistical Analyses Statistical Programming Using “R” Program

A program was developed Data Sorting Statistical Analysis and Modeling Program wasValidated and executed

Models for each Highways Classification Interstate Highway System (IHS) National Highway System (NHS) State Highway System (SHS) Regional Highway System (RHS)


14

Pavement Performance Models

Statistical Analyses (cont’d)

Pavement Type Asphalt (ASP) Jointed Concrete (JCP) Composite (COM) Continuously Reinforced Concrete (CRC)

Distress Type IRI, Fatigue, Patching, Rutting, Longitudinal Cracking, Transverse

Cracking, etc Four Transformation Functions (Model types)

Power function Two degree polynomial Logarithmic function Exponential Function


15

Asphalt Pavement (SHS)Control Section 056-30

(Unsorted Data)

50

55

60

65

70

75

80

85

90

95

100

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

Rut

ting

Inde

x


16


(Sorted Data)

50

55

60

65

70

75

80

85

90

95

100

0 2 4 6 8 10 12 14 16

Surface Age (tSA), years

Rut

ting

Inde

x

RTI= 100 - DP

Upper 1/3rd Zone

Middle 1/3rd

Lower 1/3rd Zone

Shifted Data


17

Four Models of “R” Program

DP= 100 - I

Exponential Logrithmic

Polynomial Power


18


(Sorted Data)

50

55

60

65

70

75

80

85

90

95

100

0 2 4 6 8 10 12 14 16

Surface Age (tSA), years

Rut

ting

Inde

x

Lower 1/3rd

Upper 1/3rd

Middle 1/3rd

( )aSAtbRTI .100−=


19


0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100

Actual Rutting Index

Pred

icte

d R

uttin

g In

dex

43% of the values are within (÷) 2.5% Error66% of the values are within (÷) 5.0% Error78% of the values are within (÷) 7.5% Error87% of the values are within (÷) 10.0% Error94% of the values are within (÷) 15.0% Error

Existing Model

Line of Equality


20

Asphalt Pavement (SHS)Control Section 056-30 (Predicted vs. Actual)

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100

Actual Rutting Index

Pred

icte

d R

uttin

g In

dex


New Model


21

Jointed Plain Concrete Pavement (NHS)Longitudinal CrackingControl Section 019-02

0

10

20

30

40

50

60

70

80

90

100

0 5 10 15 20 25 30

Surface Age

Inde

x

Sorted Data Upper Middle Lower

Upper 1/3rd

Middle 1/3rd

Lower 1/3rd


22


0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100

Actual Longitudinal Cracking Index

Pred

icte

d Lo

ngitu

dina

l Cra

ckin

g In

dex 16% of the values are within (÷) 2.5% Error

26% of the values are within (÷) 5.0% Error35% of the values are within (÷) 7.5% Error43% of the values are within (÷) 10.0% Error57% of the values are within (÷) 15.0% Error

Existing Model


23


0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100

Actual Longitudinal Cracking Index

Pre

dict

ed L

ongi

tudi

nal C

rack

ing

Inde

x


New Model


24

Generalized models were developed Based on Distress Index and Age Highway Classification Pavement Type Distress Type Example: IHS-ASP-Fatigue Model, etc….

Fundamental concept of rate of deterioration was applied

Generalized Model (Family Curves)


25

Roughness Index Generalized Model (NHS)

( )tfdtdp

=

y = 0.27x0.35

R2 = 0.92

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

0 5 10 15 20 25 30 35 40 45 50 55 60

Pavement Age (Years)

Rat

e of

Det

erio

ratio

n, (d

p/dt

)

26

Roughness Index Generalized Model (NHS)

The generalized model based on the age of the pavement can be used to determine the deterioration of a new road with possible

successive rehabilitation actions during its life span.

[ ]Cttb Aa

ARI +−= )log()( 1110100

( ) avgaAavg tbRI .100−=

OR

27

National Highway System (NHS)Flexible Pavement

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100

Actual Roughness Index

Pred

icte

d R

ough

ness

Inde

x


No data points = 14000

Existing Model


28

National Highway System (NHS)Flexible Pavement

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100

Actual Roughness Index

Pred

icte

d R

ough

ness

Inde

x


No data points = 14000

New Model


Network Condition Assessment


30

Existing (2005) and Predicted (2010) Condition

0

500

1000

1500

2000

2500

3000

3500

4000

0 10 20 30 40 50 60 70 80 90 100

Roughness Index

Freq

uenc

y

Predicted Condition 2010 Existing Condition 2005

NHS Condition-Existing Model

2005 35% @ Maintenance 2010 91% @ Maintenance

85


31

Existing (2005) and Predicted (2010) Condition

0

500

1000

1500

2000

2500

3000

3500

4000

0 10 20 30 40 50 60 70 80 90 100

Roughness Index

Freq

uenc

y

Predicted Condition 2010 Existing Condition 2005

2005 35% @ Maintenance 2010 68% @ Maintenance

85

NHS Condition-New Model


32

Model Comparison-NHS Condition

Maintenance

35

65 64

91

0

20

40

60

80

100

Network Condition(2005)

ExistingGeneralized Model

(2010)

New GeneralizedModel (2010)

New ControlSection Model

(2010)

Perc

ent o

f NH

S N

etw

ork


33

Model Comparison-NHS Condition

5

14 17

36

0

20

40

60

80

100

Network Condition(2005)

ExistingGeneralized Model

(2010)

New GeneralizedModel (2010)

New ControlSection Model

(2010)

Perc

ent o

f NH

S N

etw

ork

Major Rehab/Reconstruction


Remaining Service Life(RSL)


35

Remaining Service Life (RSL)

RSL Concept

Index

Year1995 2005

RSL1

RSL2

Same Index Value

Threshold

2000 2010

Surface Age

Section 1Section 2

Good Indicator of Prediction

1. Index Value

2. Rate of Deterioration


36

National Highway System(Flexible Pavement)

0

50

100

150

200

250

0 5 10 15 20 25 30/More


Freq

uenc

y

1534 Sections >= 30 RSL

633 Sections < 30 RSL


37

National Highway System(Flexible Pavement)

0

500

1000

1500

2000

2500

<2 3-5 6-10 11-15 16-20 >20


Freq

uenc

y

0

10

20

30

40

50

60

70

80

90

100

Cum

ulat

ive

(%)

2000 Sections >20 RSL

25% Immediate Action

45% Immediate Action


38

0

40

80

120

160

200

<2 3-5 6-10 11-15 16-20 >20


Freq

uenc

y

Flexible Composite Rigid

RSL Distribution - Pavement Type

39

Summary and Conclusions

Most existing pavement performance models were developed using the initial few years of distress data Tends to either under predict or over predict the

pavement condition Calibrating/ Developing models based on available

10 years data was required to enhance the predicting capabilities of LADOTD PMS


40

Summary and Conclusions

Index based pavement performance models were developed Control sections that exhibited good performance

data and historical records. Three categories; upper, middle and lower 1/3rd

percentile The results indicated that predicted values

exhibited good agreement with the actual values.


41

Summary and Conclusions Fundamental relationship between the

pavement rate of deterioration and pavement age was evaluated The increase in age of the pavement the rate of

deterioration increases. The results of the Generalized models analyses

showed good agreement of predicted index values Furthermore, approximately, 70-95% of data

exhibited within 7.5% error between the predicted and observed values.


42

Summary and Conclusions The developed models can be used to:

Determine the health of the pavement network Index Predictions Remaining service life

Enhances the PMS capabilities in predicting Pavement conditions Maintenance activities Major Rehab/Reconstruction activities Treatment selection


Thanks!


Documents

Development of Pavement Performance Prediction Models for