Reliability in the MEPDG

One States Perspective

Linda Pierce, PE

State Pavements EngineerWSDOT

Transportation Research Board 86th Annual Meeting

January 21, 2207

Introduction

• What level of reliability to use?

• NCHRP 1-40 results

• Reliability and pavement thickness

• Thoughts on the reliability concept in MEPDG

What values to use?

• 1993 AASHTO Guide definition

– The actual number of ESAL’s to a terminal

serviceability level is less than the predicted ESAL’s

• MEPDG definition

– Each key distresses will be less than the critical

distress level over the design period

• The more important the project (consequences

of failure) the higher the reliability

MEPDG Flow Diagram

Image courtesy of Scott Wilson Pavement Engineering LTD, NCHRP 1-40A(03) Final Report

MEPDG Reliability Process

1. Input design parameters (traffic, climate, structure, material properties)

2. User defines target distress conditions at the end of the design period

3. User defines reliability level

4. MEPDG estimates distress at the end of the design year and determines “pass” or “fail” condition

MEPDG Recommended Values

50-7550-75Local

70-8075-85Collectors

75-9080-95Principal Arterial

80-9585-97Interstate

RuralUrbanFunctional Class

Distress Types

Fatigue cracking (chemically

stabilized layer)

Minimum and maximum

crack spacing

Rutting – HMA onlyLoad transfer (cracks)

Rutting – total pavementCrack width (CRCP)

Thermal crackingPunchouts

Bottom up crackingFaulting

Top down crackingTransverse cracking

Terminal IRITerminal IRI

Flexible PavementRigid Pavement

WSDOT Recommendation

• Until local calibration/validation can be

completed

• New Construction

75%Minor Arterial and Collector

85%Principal Arterial

95%Interstate

ReliabilityFunctional Class

• Rehabilitation

– Reliability = 50 percent

NCHRP 1-40 Results

• Rehabilitation design may be conduced at a greater level

– Traffic, material properties, subgrade

moisture conditions, pavement distress can

be measured in the lab and field

Global Calibration Hypothesis

Image courtesy of ARA

NCHRP 1-40 Results

• For some deterioration modes there is more

calibration data at low levels of distress

• Variability for higher distress levels is likely to

be less reliable (based on fewer data points)

• Until more calibration data (at varying distress

levels) is available, it is difficult to asses the

effects of reduced input variability and

therefore, reliability of the output parameters

NCHRP 1-40 Results

• The same standard deviations are used for all design levels (except for thermal cracking)

– No improvement in reliability when more

accurate input parameters are used

• Investigate the use of

– Monte Carlo simulation techniques

– Other alternative methods

NCHRP 1-40 Results

• Insufficient data in the LTPP database to determine the effect of input level on the calibration error

– Except for thermal cracking properties

• Need to have accurate measure of past traffic loadings

• Local calibration needed to confirm the national calibration models

NCHRP 9-30 Results

• Use of performance data from carefully controlled experiments (WesTrack, MnRoad, NCAT, FHWA-ALF) are more useful for model validation/calibration

• Recalibrate MEPDG

Reliability and Thickness Design

• HMA Example

– Level 1 and 3 analysis

– 40 year design

– Reliability levels: 50 and 90

Reliability and Thickness Design

Fail19.770.340.25Permanent Deformation (AC Only) (in):903

Fail02112100AC Thermal Fracture (ft/mi):903

Pass99.99940.9100AC Bottom Up Cracking (%):903

Fail13.23226.5172Terminal IRI (in/mi)903

Fail19.770.340.25Permanent Deformation (AC Only) (in):503

Fail02112100AC Thermal Fracture (ft/mi):503

Pass99.99940.9100AC Bottom Up Cracking (%)503

Fail13.23226.5172Terminal IRI (in/mi)503

CommentReliability Predicted

Distress Predicted

Distress TargetPerformance CriteriaR

Input Level

Thoughts on the Reliability

Concept

• Process used in the MEPDG is a starting point

• Local calibration is essential