MEPDG Overview & National Perspective - Purdue … · Design Guide 1986; Update ... Rehabilitated...

MEPDG Overview & MEPDG Overview & National PerspectiveNational Perspective

(My Perspective)(My Perspective)

NorthNorth--Central MEPDG User GroupCentral MEPDG User GroupFebruary 19, 2008February 19, 2008

Harold L. Von Quintus, P.E.

Expanding the Realm of Possibility

2

OutlineOutline1.

The Beginning

2.

Local Implementation Efforts3.

Integration of MEPDG into Practice

4.

Enhancements5.

Summary


3

ItIt’’s Done; In Reality, s Done; In Reality, ItIt’’s the Beginning!!s the Beginning!!

April 2007 Irvine WorkshopApril 2007 Irvine Workshop


4

Should we wait until its Should we wait until its PERFECTPERFECT??

1958; Road Test initiated

1962; AASHO Road Test complete

1972; Interim Design Guide

1986; Update

1993; Update

2007; still not perfect.

1989; LTPP initiated

1998; MEPDG initiated

2007; MEPDG delivered

Time, yrs.

-4

10

24

31

45

AASHTO Guide MEPDG


5

Should we wait until its Should we wait until its PERFECTPERFECT?? If we wait until there are no more

changes, we will never use it. If we wait for perfection, it will be

impractical and cost will restrict its use.

There is NONO perfect procedure & it will never

be perfect!


6

The BeginningThe BeginningWe have decided to implement the

new MEPDG!

Where do we begin; regional versus agency issues?

What will it cost?

Will we get a better design?

What am I responsible

for?

Why, aren’t we doing a good job?


7

Remember where we Remember where we are coming from, as are coming from, as you use the MEPDG!you use the MEPDG!

Assumptions used in the Assumptions used in the Design Guides?Design Guides?

Calibration of both Design Calibration of both Design Guides?Guides?

Error in the service life Error in the service life predictions of both Design predictions of both Design Guides?Guides?


8

OutlineOutline1.

The Beginning

2. Local Implementation Efforts3.


4.

Enhancements5.

Summary


9

FHWA FHWA Summary Summary of Agency of Agency

PlansPlans

Efforts to Efforts to ImplementImplement

MEPDG MEPDG 20072007

0

20

40

60

80

100

Yes < 0.5 0.5 to 1.0 1.0 to 5.0 > 5.0

Range of Cost for Implementation Plan, $MPe

rcen

t of T

otal

R

espo

ndin

g (5

2)

Does Agency Have Implementation Plan?

010203040506070

Yes < Design Life Equals DesignLife

> Design Life

Perc

ent o

f Tot

al R

espo

ndin

g (5

2)Does Agency Track Perfomance for Use in Calibration?


10

MEPDG Global CalibrationMEPDG Global CalibrationLTPP LTPP GPSGPS Test Sections Used in Test Sections Used in Calibration of Distress Prediction Calibration of Distress Prediction

Models;Models;NCHRP Projects 1NCHRP Projects 1--37A & 137A & 1--40D40D

Many assumptions used.Many inputs estimated.


11

Local Implementation EffortsLocal Implementation Efforts—— DiversityDiversity

√√

√√

√√

√√

√√ √√

√√

√√

√√

√√

South-EastSouth-West

North-CentralNorth-EastNorth-West

1. Verify/confirm calibration factors & default values.

2. Establish input libraries.


12

Expansive SoilsExpansive Soils Maintenance/Preservation Maintenance/Preservation

StrategiesStrategies New Technology & Material New Technology & Material

DifferencesDifferences Other Design Policies & Other Design Policies &

FeaturesFeatures

Why Local Calibration?Why Local Calibration?

Consideration of factors Consideration of factors NOT NOT included in MEPDG global calibration included in MEPDG global calibration

process.process.


13

MEPDG MEPDG –– Local Local Validation/Calibration ToolsValidation/Calibration ToolsManual of Recommended Practice for Manual of Recommended Practice for

Calibration of MCalibration of M--E Based ModelsE Based Models1.

Confirming or adjusting the global calibration factors.

2.

Detailed and practical guide to complete local calibration.

MEPDG Software ItselfMEPDG Software ItselfNCHRP Project 1NCHRP Project 1--40B40B


14

Previous & OnPrevious & On--Going StudiesGoing Studies1. NCHRP 9-30 –

Experimental Plan for Calibration &

Validation of HMA Performance Models for Mix & Structural Design.

2. NCHRP 9-30(001) –

Conduct Pre-Implementation Studies & Database Enhancement.

3. NCHRP 1-40D –

A review of the M-E PDG software & prediction methodology; & Correcting errors/blunders in the software.

4. NCHRP 1-40B –

Local Calibration for the Recommended Guide for M-E Design of New and Rehabilitated Pavement Structures.


15

Previous & OnPrevious & On--Going StudiesGoing Studies

Calibration Documents:

NCHRP Digest 284, December 2003; Refining the Calibration & Validation of HMA Performance Models: An Experimental Plan and Database.

NCHRP Digest 283, December 2003; Jackknife Testing – An Experimental Approach to Refine Model Calibration and Validation.

FHWA: Use of PMS data for local calibration.

FHWA: Use of deflection basin data in the MEPDG.


16

OutlineOutline1.

The Beginning

2.

Local Implementation Efforts3. Integration of MEPDG into

Practice4.

Enhancements

5.

Summary


17

Integration into PracticeIntegration into PracticeHow do I get How do I get

this input level 1 this input level 1 or 2 for design?or 2 for design?

A Major Issue A Major Issue –– The The Unknowns!!Unknowns!!

Determination of properties Determination of properties & other inputs.& other inputs.

Factors affecting properties Factors affecting properties needed for design!!!!needed for design!!!!•• Source of MaterialsSource of Materials•• ContractorContractor•• Construction EquipmentConstruction Equipment 4-Day NHI Course

for MEPDG Software Training


18

Data Integration: Effective use of Data Integration: Effective use of available but limited local resources.available but limited local resources.

PMISDatabase

Project Project Level DataLevel Data

Data integration: an automated process?Probably NOT!


19

Structural Structural –– Mix Design Mix Design CompatibilityCompatibility

JMFJMFMixture design & material Mixture design & material specifications determine specifications determine

the inputs.the inputs.

11 22

33


20

Data Integration into PracticeData Integration into PracticeDevelop Designs

Low-bid; optimize on design features

Alternate bids; establish equivalent designs

Design-build & warranties; optimize on performance

Establish Specification Limits

Quality Assurance

Performance-

Related

End-Result


21

OutlineOutline1.

The Beginning

2.



4. Enhancements5.

Summary


22

Enhancements:Enhancements: Global & LocalGlobal & LocalGlobal√

Distress prediction equations or transfer functions

√

Revisions to the software; functionality

√

Add additional distresses

Local or Regional√

Revisions to the default values

√

Revisions to the transfer function and calibration coefficients

√

Build libraries of inputs

NCHRP Project 1-40B Local Calibration Guide


23

Previous & OnPrevious & On--Going StudiesGoing Studies1. NCHRP 1-40A –

Independent Review,

prioritize the modifications.2. NCHRP 9-30A –

Calibration of Rutting

Models for HMA Structural and Mix Design.3. NCHRP 9-42 –

Top-Down Cracking of HMA.

4. Reflection Cracking of HMA Overlays.5. NCHRP 9-44 –

Application of the Endurance

Limit for HMA Mixes.


24

New Construction Design New Construction Design StrategiesStrategiesIncluded:

Conventional flexible pavements

Deep strength

Full depth

Excluded:

Aggregate surfaced roadways

Semi-rigid pavements

Staged construction

Asphalt treated permeable base

Geogrids, fabrics, & other strengthening materials


25

Rehabilitation StrategiesRehabilitation StrategiesIncluded:

HMA overlays with & without milling

Excluded:

Full depth reclamation

Hot in place recycling

Cold in place recycling

Pavement preservation programs


26

Site Features Excluded from Site Features Excluded from MEPDGMEPDG

Super single tires or single tires.

Durability & mixture disintegration.

Volume change in soils (frost heave or expansive soils).


27

OutlineOutline1.

The Beginning

2.



4.

Enhancements5. Summary


28

SummarySummaryImplementation Considerations:

Regional design features not included in global calibration.

Regional defaults that are different from global defaults.

Design criteria as compared to measured values included in calibration.


29

SummarySummary

Response Parameter & Calibration Factors

Transfer Function & Calibration Factors

Standard Error orStandard Deviation

Agency/User has options readily Agency/User has options readily available for design!available for design!


30

Value of Increased Costs & Time?Value of Increased Costs & Time?

If improved performance/longer lasting pavements & reduced life cycle costs;

Then it is worth the effort, time, and cost!! Then it is worth the effort, time, and cost!! Assuming enforcement of specifications.Assuming enforcement of specifications.

Thank you.Thank you. Any Questions?Any Questions?


32

Questions!Questions!


33

Advanced Pavement DesignAdvanced Pavement Design

Distress specific—defines what controls the design.

Design features—considers many more design features and what effect they have on distress.

Smoothness

Construction defect effects.


34

Current Pavement Design Current Pavement Design ProceduresProcedures

01020304050607080

AgencySpecific

72-AASHTO

81-AASHTO

93-AASHTO

98-AASHTO

Pavement Design Procedures

Perc

ent o

f Tot

al

Res

pond

ing

(52)

Flexible Pavements Rigid Pavements


35

Agency Road Maps for:Agency Road Maps for: Implementing the MEPDGImplementing the MEPDG

••

To streamline the design process To streamline the design process using the MEPDG, & to verify and using the MEPDG, & to verify and calibrate the distress prediction calibrate the distress prediction models or transfer function included models or transfer function included in the MEPDG.in the MEPDG.


36

ReliabilityReliability1993 AASHTO

Based on traffic level; standard deviation of the design process & applied to traffic level.

MEPDG

Standard error for each distress prediction model & defined from calibration.

Carefully review and consider the standard errors of the distress prediction equations for application to low volume roadways.


37

SummarySummary

Remember:The hierarchical input procedure allows a user to use the MEPDG with NO major investment.


38

Deciding on Design InputsDeciding on Design Inputs

Communications within & between departments

TrafficMaterialsConstruction


39

Decision FactorsDecision Factors1.

Cost of validation/calibration.

2.

Value of potential improvement in prediction accuracy.

3.

Cost/Value of design & construction.

Decision: Yes, validate & calibrate MEPDG to our local

conditions!


40

Preparation of a MPreparation of a M--E Design in E Design in a Lowa Low--Bid ProcessBid ProcessFoundation

AnalysisClimateEICM

MaterialsMaterialsLibraryLibrary

Traffic Library

Design Criteria Structural DesignStructural Designby Agencyby Agency

Layer Thickness,Layer Thickness,Type & PropertiesType & Properties

Mixture Design & Mixture Design & Confirm PropertiesConfirm Properties


41

Implementation Areas & Implementation Areas & Technology TransferTechnology Transfer

Training & communications within & between departments

Traffic

Materials

Construction

Calibration


42

1. Project Selection

2. Planning

3.Structural Design

4. Plan Preparation

5. Bid Letting

6.HMA Mixture Design

7. Construction

8. As-Built Plans & Performance Confirmation

Library of MixLibrary of MixProperties:Properties:

Dynamic ModulusDynamic ModulusFlow NumberFlow NumberIDT StrengthIDT StrengthIDT CreepIDT Creep

VMAVMA

DesignDesignValues:Values:E*>?E*>?SSIDTIDT >?>?D(tD(t)<?)<?

Structural Structural –– Mix Design IntegrationMix Design Integration

Volumetric& SPT

Design, more Design, more than just than just volumetricsvolumetrics..


43

Technology Transfer & Technology Transfer & Implementation ProductsImplementation ProductsRemember Products:

Management video

Interactive CD for software

Implementation notes

Training course

Guide text & appendices.

User’s Manual in support of software.


44

Important Activities Important Activities for Implementationfor Implementation

Training Courses:

Determining inputs & using software

Communication:

Departments need to know what information is needed & how it is used.

Establish sensitivity of inputs to distress

Identify problem areas to reduce frustration with software use


45

Global Calibration of Distress Global Calibration of Distress Transfer FunctionsTransfer Functions

Pavement Pavement ResponseResponse DistressDistress

TRANSFERTRANSFERFUNCTIONFUNCTION

●●

StressesStresses●●

StrainsStrains

●●

DeflectionsDeflections

●●

CrackingCracking●●

DistortionDistortion

●●

RoughnessRoughnessCalibrat

ion

Calibrat

ion

is a k

ey.

is a k

ey.


46

Local CalibrationLocal Calibration

A difficult & costly issue to resolve!

How close is close enough?


47

Outline of Recommended Practice Outline of Recommended Practice for Local Calibrationfor Local Calibration

Introduction

Scope of Document

Referenced Documents

Definition of Terms

Significant of Use

Defining Accuracy

Standard Error Components

Step-by-Step Procedure

Revisions to calibration coefficients

Demonstration Studies


48

Validate/Calibrate MEPDGValidate/Calibrate MEPDG

How many sites?Length of roadway segments?How many distress observations?

Multiple points within segment at any one point in time?

Multiple points over time?


49

Distress Data SourcesDistress Data SourcesLTPPSpecial Agency Test SectionsPMIS Sections –

A quick check:

Do the distress predictions match our Do the distress predictions match our local experience for traditional designs?local experience for traditional designs?

What is the primary failure mode What is the primary failure mode triggering rehabilitation?triggering rehabilitation?

NCHRP 9NCHRP 9--30 Database for 30 Database for MM--E Based ModelsE Based Models


50

Calibration: Ease of UseCalibration: Ease of Use

Calibration Options:Calibration Options:

Special AnalysisSpecial Analysis

National CalibrationNational Calibration

State CalibrationState Calibration

Typical Agency Typical Agency ValuesValues

National/Global National/Global Calibration;Calibration;

kk--

or Cor C--ValuesValues

State/Regional State/Regional Calibration;Calibration;ββ--

or Cor C--ValuesValues


51


Transfer Transfer FunctionFunction

CC-- ValuesValues

Standard Standard DeviationDeviation

HMA Alligator HMA Alligator CrackingCracking


52


Response Parameter Response Parameter & Calibration Factors& Calibration Factors

Transfer Function & Transfer Function & Calibration FactorsCalibration Factors

Standard DeviationStandard Deviation

JPCP CrackingJPCP Cracking


53

MEPDG Unique FeatureMEPDG Unique Feature

Response Parameter & Calibration Factors

Transfer Function & Calibration Factors

Standard Deviation

Agency/User has options readily Agency/User has options readily available for design!available for design!

Batch runs can be completed!Batch runs can be completed!


54

Local Validation/Calibration Local Validation/Calibration HypothesesHypothesesMathematical models –

assumed to be

correct.Pavement response modelsClimatic model –

ICM

HMA aging/PCC strength time dependent model

Statistical or empirical models (transfer functions) may result in bias.Revision of model coefficients to remove bias.


55

General Approach to Validation & General Approach to Validation & Calibration of MCalibration of M--E Based ModelsE Based Models

1.

Traditional Split-Sample Approach2.

Jack-Knife Testing Approach


56

Validate/Calibrate MEPDGValidate/Calibrate MEPDG

How many sites?

Length of roadway segments?

How many distress observations?

Multiple points within segment at any one point in time?

Multiple points over time?

What data do we use?LTPP/PMIS & other databases!!


57

Data Quality ReviewData Quality Review

0.0

0.1

0.2

0.3

0.4

0 20 40 60 80 100Time (Months)

Rut

ting

(in/m

ile)

Measured

Predicted

What caused this increase in measured rutting?

Data measurement issues!Data measurement issues!


58

Data Quality ReviewData Quality Review

0.0

1000.0

2000.0

3000.0

4000.0

0 20 40 60 80 100 120 140Time (Months)

Tran

sver

se C

rack

ing

(ft/m

ile)

Measured

Predicted

Why are transverse cracking values so diverse?

Pavement preservation issues!Pavement preservation issues!


59

Distress Data, Example:Distress Data, Example:

Variability; A key data

issue.

Example; Fatigue

Cracking

0100200

300400

500600

700

47 48 49 50 51 52 53

Mile Post

Fatig

ue C

rack

ing,

sq.

ft./0

.1m

i.

Neat Mix, Region 6, Route 36B, Both Directions, 2004

0200400600800

1000120014001600

85 90 95 100 105

Mile Post

Fatig

ue C

rack

ing,

sq.

ft. p

er0.

1 m

i.

Modified Mix, Region 5, Route 285B, Direction 2, 2004

PMA Mixes

Neat Mixes


60

Distress Data, Example:Distress Data, Example:

0

0.1

0.2

0.3

0.4

0.5

47 48 49 50 51 52 53

Mile Post

Rut

Dep

th, i

nche

s

Neat Mix, Region 6, Route 36B, Both Directions, 2004

0

0.050.1

0.150.2

0.250.3

0.35

86 88 90 92 94 96 98 100 102

Mile Post

Rut

Dep

th, i

nche

s

Modified Mix, Region 5, Route 285B, Direction 2, 2004

Variability; A key data

issue.

Example; Rut Depth

Neat Mixes

PMA Mixes


61

Distress Data Analyses:Distress Data Analyses: Within Project Variation; Within Project Variation; OutliersOutliers

0

0.2

0.4

0.6

0.8

1

278 278.5 279 279.5 280 280.5 281

Mile Post

Rut

Dep

th, i

nche

s

PMA Mix, Region 6, Route 6G, Both Directions, 2005

Limited area with significant different distress value within PMIS segment.

Outliers, removed from calibration set.


62

Distress Data Analyses:Distress Data Analyses: Within Project Variation; Within Project Variation; Abrupt ChangeAbrupt Change

0

0.1

0.2

0.3

0.4

0.5

0.6

211 211.5 212 212.5 213 213.5

Mile Post

Rut

Dep

th, i

nche

s

Neat Mix, Region 6, Route 25A, Both Directions, 2005

Sudden increase or decrease in distress value within PMIS segment.

Identify reasons for abrupt changes.


63

Distress Data Analyses:Distress Data Analyses: Within Project Variation; Within Project Variation; DriftDrift

00.05

0.10.15

0.20.25

0.30.35

0.40.45

0.5

258 260 262 264 266 268

Mile Post

Rut

Dep

th, i

nche

s

Modified Mix, Region 6, Route 70-A, Both Directions, 2005

Consistent change in distress over project length, within PMIS segment.

Use segments w/consistent averages.


64

Data Analyses: Bias & Residual Data Analyses: Bias & Residual Error from Validation/CalibrationError from Validation/Calibration

02468

1012141618

0 2 4 6 8 10 12

Measured Value

Pred

icte

d Va

lue

Line of Equality

02468

1012141618

0 2 4 6 8 10 12

Measured Value

Pred

icte

d Va

lue

Line of Equality

02468

1012141618

0 2 4 6 8 10 12

Measured Value

Pred

icte

d Va

lue

Line of Equality

02468

1012141618

0 2 4 6 8 10 12

Measured Value

Pred

icte

d Va

lue

Line of Equality

BiasBias

Residual Residual ErrorError

Suggestion: Determine bias and error for Suggestion: Determine bias and error for local conditions & materials.local conditions & materials.


65

Global Calibration HypothesisGlobal Calibration Hypothesis

Level 3DefaultValue

Level 2Calculated

Value

CalibrationError

Level 1Measured

Value

Reduce uncertainty, use level 1


66

How Close is Close Enough?How Close is Close Enough?

0

20

40

60

80

100

120

140

5 10 15 20 25 30

Standard Error

Res

earc

h/C

alib

ratio

nC

osts

222

22

PureInputFitofLack

tMeasuremenTotal

Quantify the total error to answer this question!Quantify the total error to answer this question!

60

80

100

120

140

160

180

5 10 15 20 25 30

Standard Error

Con

stru

ctio

n C

osts

Impact of Reliability

Lack-of-fit & Input

Pure & measurement


67

Uses of PMIS Data for Uses of PMIS Data for Validation/CalibrationValidation/Calibration

Identify factors/site features deviations or anomalies between performance observations.

Identify bias between observations & predicted distresses for different design features & strategies.

Determination of the standard error – Probably NOT!


68

Pavement Management Pavement Management Systems Data Systems Data -- A Resource:A Resource:Large time-series database on

distress & pavement performance.Many dollars expended to develop

PMIS and collect data.Use of PMIS data for

validation/calibration not a simple & quick process!!!!


69

In SummaryIn Summary

0

1

2

3

4

0 1 2 3 4

Measured IRI, m/km

Pred

icte

d IR

I, m

/km

Local validation/calibration:It’s an important decision but it’s your to make.AccuracyCosts

Manual of Manual of Recommended Recommended Practice for Calibrating Practice for Calibrating MM--E Based Models.E Based Models.


70

Material Testing Material Testing –– Equipment Equipment

Purchased for ImplementationPurchased for ImplementationUnbound Mtls.Montana –

No

Missouri –

NoUtah -

Yes

HMAMontana –

No

Missouri –

YesUtah -

Yes

PCCMost agencies have equip.


71

Testing Testing –– Lab TestingLab Testing

Build library of material properties

Build library of default values


72

Questions for Establishing a Questions for Establishing a Data Collection PlanData Collection Plan1.

Is input parameter important for predicting distress?

2.

Is the input parameter site/facility specific?

3.

Can input parameter be easily/adequately measured?


73

TimeTime--History Data for CalibrationHistory Data for Calibration

00.10.20.30.40.5

0 2 4 6 8 10 12

Age, years

Rut

Dep

th, i

nche

sMnRoad Measured Mississippi Measured Arizona MeasuredMnRoad Predicted Mississippi Predicted Arizona Predicted


74

Local Calibration HypothesesLocal Calibration HypothesesMathematical models –

assumed to be

correct.Response modelsClimatic model –

ICM

HMA aging/PCC strength time dependent model

Statistical or empirical models (transfer functions) may result in bias.Revision of model coefficients to remove bias.


75

Steps for Local CalibrationSteps for Local Calibration1.

Select hierarchical input level

2.

Develop experimental design & matrix

3.

Determine sample size4.

Identify roadway segments

5.

Collect & evaluate data for anomalies

6.

Conduct field investigations

Level of confidence.

1.

LTPP sections2.

Research sections

3.

PMIS segments

Testing & distress definitions

Policy decision.


76

Steps for Local CalibrationSteps for Local Calibration7.

Assess bias

8.

Eliminate bias9.

Assess standard error

10.Improve model precision

11.Interpretation of results & decide on adequacy of calibration factors

Execute MEPDG & evaluate residual

errors.

Dispersion around line of equality


77

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8

Observed Distress Value

Pred

icte

d D

istr

ess

Valu

e

eDtD edictedCI Pr

0

0.2

0.4

0.6

0.8

0 10 20

Age, year

Pred

icte

d D

istr

ess

Reliability = 50%

Reliability = 75%

Reliability = 95%

Design Confidence Interval ConceptDesign Confidence Interval Concept

Predicted Distress ValuePredicted Distress Value

50%50%

75%75%

95%95%

BUT, where does the BUT, where does the error come from?error come from?


78

Quantifying the Prediction Quantifying the Prediction ErrorError

PureInput

tMeasuremenFitofLackTotal

ee

eee

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8

Predicted Values

Obs

erve

d V

alue

s


79

Objective:Objective:

•

To streamline the design process using the MEPDG, & to verify and calibrate the distress prediction models or transfer function included in the MEPDG.


80

Rutting in Unbound Layers & Rutting in Unbound Layers & SubgradeSubgrade..

Trenches not completed for the

LTPP test sections!


81

Based on Volumetric Based on Volumetric PropertiesProperties

HMA RuttingHMA Rutting

Wear vs. plastic deformation


82

Based on Based on Volumetric Volumetric PropertiesProperties

Fatigue Fatigue CrackingCracking


83

Fatigue CrackingFatigue Cracking

Cores not taken to

confirm crack direction for LTPP Test Sections!


84

Transverse/Thermal CrackingTransverse/Thermal Cracking


85

Smoothness or IRISmoothness or IRI

Regression equations developed from LTPP

test sections,

higher speed

roadways!


86

Calibration Data OptionsCalibration Data Options

3. Using PMIS 3. Using PMIS segments for the segments for the local calibrationlocal calibration

1. Using existing test 1. Using existing test sections (LTPP & other sections (LTPP & other projects) for calibrationprojects) for calibration

2. Using PMIS segments with detailed distress

surveys & LTPP sections.


87

Date post:	09-Sep-2018
Category:	Documents
Upload:	doanthu
View:	235 times
Download:	0 times