Mix Design Verification Process Variability Granite Construction Experience

1
Mix Design Verification Process
VariabilityGranite Construction Experience

• Adam J. Hand
• Jon A. Epps
• Rocky Mountain Asphalt User Producer Group
• Annual Meeting
• October 16-18, 2001
• Albuquerque, NM

2
Outline

• Introduction - Specification Evolution
• Variability
• Volumetric Example
• Case Histories
• Summary and Conclusions

3
When and Why are Verification Encountered?
Specification Types and Their Evolution
4
Types of Specifications

• Proprietary
• Method
• QC/QA
• End Result
• Performance-Related (PRS)

• Performance-Based (PBS)
• Performance (e.g., Warranty/Guarantees)
• Design/Build
• Design/Build/Operate

5
Evolution of Specifications

• Method Specifications
• 3 Generations
• 1st 3rd
• QC/QA
• 2 Generations
• 1st
• Current Generation

6
Evolution Terminology

• LMLC Lab Mixed Lab Compacted
• Mix Design and Volumetrics
• FMLC Field Mixed Lab Compacted
• Volumetrics
• FMFC Field Mixed Field Compacted
• Core Properties
• Green Changes
• Red Basis for Payment

7
Evolution of Method Specifications

• 1st Generation
• LMLC by Agency No Verification
• Field by Agency - AC, gradation
• 2nd Generation
• LMLC by Agency No Verification
• Field by Agency - AC, gradation, In-Place
  Density
• 3rd Generation
• LMLC by Agency No Verification
• FMLC by Agency
• Field by Agency - AC, gradation, In-Place
  Density
• Initial Recognition of LMLC vs. FMLC Issues

8
Evolution of QC/QA Specifications

• 1st Generation
• LMLC by Contractor Agency Verification
• Paper or Laboratory Verification
  
• FMLC Sporadic by Agency
• Field QA by Agency - AC, gradation, In-Pace
  Density, Smoothness
• No Attempt to Use Contractor Test Results for QA
  or Payment
• Current Generation
• LMLC by Contractor - Agency Verification
• Paper or Laboratory Verification
• FMLC by Agency and Contractor
• Field QC/QA by Contractor AC, gradation,
  Volumetrics, In-Place Density, Smoothness

9
What Has Evolved?

• Method Specifications
• Between Laboratory Differences NOT a Mix Design
  Verification Issue
• Field Management Initial Recognition of LMLC
  vs. FMLC Differences
• QC/QA Specifications
• Between Laboratory Differences ARE a Mix Design
  Issue
• Differences in QC and QA data ARE an Issue
• Volumetrics Are Playing A Significant Role in
  Specifications ? Within Between Lab
  Differences

10
What Has Evolved?

• The Central Issue
• Testing Variability Has Become Extremely
  Important
• Mix Design Verification
• Test Strip
• Production Testing
• Different Mix Design Methods Different Issues
• Superpave Volumetrics Only
• Hveem Volumetrics and Stability
• Marshall Volumetrics, Stability and Flow

11
Importance of Variability

• Mix Design Verification
• Test Strip
• Production
• QC and QA Data Comparisons (t-tests)
• Effect on HMA Performance
• PWLs ?Pay Factors

12
Modern QC/QA Specifications

• Quality Characteristics
• AC, Gradation, Voids, Density, Smoothness
• Target Values
• Tolerances (Limits)
• Percent Within Limits
• Pay Factors Performance

13
Modern QC/QA Specs Require

• Repeatable and Reproducible Test Methods
• Good Understanding of Impact of Variability on
  the Specification

14
Effect on Performance
15
Effect on PWLs - Payment(Equal Means but
Different Standard Deviations)
target
Lot 1
Lot 2
Upper limit
Lower limit
4.2 4.6 5.0
5.4 5.8
Asphalt Binder Content
16
Outline

• Introduction - Specification Evolution
• Variability
• Case Histories
• Summary and Conclusions

17
QC/QA and Variability Components
Variability variability variability
variability (QC/QA) (sampling)
(test method) (mat./const.)
S2QC/QA S2s S2t S2m/c
18
Sampling Variability
S2QC/QA S2s S2t S2m/c
19
Sampling Variability (s2s)

• Typically 10-30 of Total Variability
• Sample Location
• Sample Method
• Sample Size
• Sample Splitting and Reducing

20
HMA Sampling Location

• Plant Conveyor/Chute
• Truck at Plant (RTSD vs. Manual)
• Windrow
• Hopper in Paver
• Loose Mat (Behind Paver)

21
Sample Splitting/Reduction
22
Effect of Sampling Location on Gradation
Variability(Fine Mix)
23
Test Method Variability
24
Test Method Variability (s2t)

• Within Laboratory Variability
• Between Laboratory Variability
• Defined by Interlaboratory (Round Robin) Studies
• Reported in Precision and Bias Statements

25
Precision and Bias
Precision poor good
good Bias low high
low
26
Example Asphalt Binder Content
27
Asphalt Binder Content
28
Test Method Precision and Bias

• Precision Statements Account for Inherent Test
  Method Variability (uncontrollable random error)
• Determined Through Interlaboratory Studies
• Single-operator, within lab, repeatability
• Multi-laboratory, between lab, reproducibility
• One-Sigma Limits (pooled standard deviation, ?p,
  1S)
• Difference Two-Sigma Limits ?p(2?2) or 1S(2?2)
  D2S

29
Interlaboratory Studies
ASTM 1S
ASTM D2S 1S(2?2)
30
Interlaboratory Studies
High Variability Lab 2
Low Variability Lab 1
Pooled Variability-All Labs
Precision Statements are Based on Pooled (all
labs) Variance (?p)
31
Precision and Bias Statements

• Examples
• National (AASHTO and ASTM)
• State DOT
• Regional (Western States)
• Local (County/City)
• Be Careful Understand Preparation, Materials,
• They are Basically Averages for the Materials
  Considered

32
Is it Possible for Slot to be gt Sprecision
statement
Absolutely!
33
Sources of Precision Information

• ASTM and AASHTO Test Methods
• AMRL
• http//patapsco.nist.gov/aashto/amrl/services/pspt
  oc.html
• Dave Savage [phone]
• Get the Details
• What Materials were Used to Develop Precision
  Statement?
• Absorptive Aggregates?
• Modified Asphalt Binder?
• How were Samples Prepared?

34
Outline

• Introduction - Specification Evolution
• Variability
• Volumetric Example
• Case Histories
• Summary and Conclusions

35
Within Laboratory Precision(Single Operator
Precision)
- Duplicate specific gravity results by the
same operator should not be considered suspect
unless they differ more than 0.02. ( ) -
supplemental procedure for mixtures containing
porous aggregate conditions (dryback procedure).
36
Between Laboratory Precision(Multilaboratory
Precision)
- Duplicate specific gravity results by the
same operator should not be considered suspect
unless they differ more than 0.02. ( ) -
supplemental procedure for mixtures containing
porous aggregate conditions (dryback procedure).
37
Air Voids Variability due to Gmb(Acceptable
Range of Two Results)
38
Air Voids Variability due to Gmm(Acceptable
Range of Two Results)
39
Test Method Variability Example
What is the Potential Difference in Air Voids Due
to Test Method Variability (Within and Between
Lab Acceptable Range of Two Test Results) Alone
40
Monte Carlo Simulation
Gmm Distribution
Gmb Distribution
Gmm - Gmb Gmm
AV 100 x
Same Gmm Distribution
Gmm and Gmb inputs AV output
AV Distribution
41
Between Laboratory Air Voids
VMA ? AV
1.4 AC
42
Sample Size

• ASTM and AASHTO Precision Statements
• 1 result per lab
• These Examples
• 1 result per lab
• As n Increases, S2 Decreases
• Differences Shown would decrease by ?n

43
Balancing Risk and Cost
Direct Cost ()
Agency and/or Contractor Risk
1 2 3 4 5 6 7
Number of Test Samples (n)
44
Interlaboratory Studies(Round Robins)

• Define Test Method Variability (Precision)
• Define Components of Test Method Variability
• What is Important?
• Total or Components of Total
• Situation Dependent

45
When Discrepancies Exist How Far Must We Go to
Understand the Cause?
46
Components of Gmb Variability
S2GmbTotal S2Gmb S2compaction S2curing
S2mixing S2batching
47
Components vs. Total VariabilityIsolating The
Issue!
S2Gmb
S2Gmb S2compaction
S2Gmb S2compaction S2curing
Total Variability
S2Gmb S2compaction S2curing S2mixing
S2Gmb S2compaction S2curing S2mixing
S2batching
48
QC/QA and Variability
S2QC/QA S2s S2t S2m/c
How can it be reduced?
49
Reduce Sampling and Testing Variability

• Technician Training and Certification
• Laboratory Accreditation (AMRL)
• Regionalize/Standardize Test Methods
• Regionalize/Standardize Test Method Options
• Proficiency Sample Programs (Round Robins)
• Increase Number of Samples (n)
• Pre-Construction Meeting Sample Exchange (NO
  RISK) Resolve it Before Hand

50
Outline

• Introduction - Specification Evolution
• Variability
• Volumetric Example
• Case Histories
• Summary and Conclusions

51
Granite Case Histories in 1 Year

• Compactor Calibration
• Multiple Locations
• LMLC vs. FMLC
• Multiple Locations Within and Between Labs
• Influence of Modified Asphalt
• Ignition Oven Calibration/Bias
• Sample Preparation - Batching

52
Compactor Calibration

• Location A
• Large ? in AC?AV in Mix Design Verification
• Actually ? Large ? Gmb
• Procedures Reviewed
• 3rd Party Compactor Calibration
• Design and Agency Labs At Extremes of Allowable
  Operating Ranges
• ?2 consistent difference in AV between labs
• Both Compactors Calibrated by One Service to
  Middle of Allowable Ranges
• Additional Test Strip Cost 35 Lost Days Saved

53
Compactor Calibration

• Location B
• Large ? in AC?AV in Mix Design Verification
• Actually ? Large ? Gmb
• Procedures Reviewed
• Design Lab Compactor Calibration
• Verified by two independent Calibration Services
• 0 for 20 on Mix Verifications!!!
• The Cost???

54
FMLC vs. LMLC Within Lab

• Loss of Voids Through Plant
• FHWA, NCAT, NAPA work have Documented
• Most Specifications Allow for
• Range of AV
• Reduction in VMA
• Well Documented, but Not Always Easily Handled
• 13.0 vs. 12.9 VMA and Performance

55
FMLC vs. LMLC Within and Between Labs

• Differences due to Gyratory Compactor
  Manufacturers
• FHWA Addressing See Harman Tomorrow
• Specific Gravity Test Methods
• NCAT
• Fine Aggregate Specific Gravity
• Vacuum Sealed (Corelock) for Gmb Gmm?
• Balloted ASTM Test Methods
• Vacuum Sealed Gmb
• Field VMA Moving Source Gsb during Production

56
FMLC vs. LMLC Between Lab

• QC/QA Project
• Marshall Stability and Flow Pay Items
• Paper Mix Design Verification
• During Production Large ? in Flow
• All other Properties in Spec/Similar to LMLC
• Shut down with RR Order
• Engineering Analysis
• Modified Binder
• Antiquated Test Method (for MBs)
• Still Cost Time, Money, Scheduling, Perception,

57
Stability and Flow Test
Modified Binder
Neat Binder
Load or Stability
Time
Difference in Flow
Flow
Time
Lab A
Lab B
58
Ignition Oven Calibration

• Bias can Exist in Calibration
• Positive and Negative
• Typically Positive
• Can Lead to Inappropriate Changes
• Reducing AC
• Why?
• Low Specific Gravity and Higher Absorption
  Aggregates?
• Its simple, the oven knows the mixture has been
  through the plant

59
Ignition Oven Calibration
Range of ? 0.75
60
Sample Preparation - Batching

• Several Methods Available
• Split Bulk Stockpile Samples and Combine
• Breakdown (dry sieving) into fractions
• top size to - 8 (AI, MS-2 AASHTO)
• All sizes (top size to 200 x pan)
• Which Method Is Best?
• Does it Matter?

61
Sample Preparation - Batching

• Early WRSC
• Several Verification Problems Attributed to
  Different SGCs
• Several States
• Sample Preparation (Not Correcting for Adhesion)
  Problem in Every Case
• It Does Matter!

62
Sample Preparation - Batching

• Must Correct for Adhesion of p200 to Coarser
  Particles if Breaking Down into Individual
  Fractions and Recombining
• Perform Washed Sieve
• Analyses to Define
• Difference in Target
• And Batched Gradings
• Make Adjustments

63
Example of Sensitivity to p200
64
Sample Preparation - Batching
65
Sample Preparation - Batching
66
Consequences of Non-Verification

• Agency
• Time User Costs
• Poor Public Perception
• Breakdown of Working Relationships
• Contractor
• Lost Revenue
• Over 0.5M at One Granite Branch this Year
• Lost Time
• Lost Opportunity
• Public Perception
• Breakdown of Working Relationships

67
Solution

• Requires
• Cooperative Engineering Analyses
• Time
• Money
• We Must Work the Problems Together to Develop
  Understanding of and Solutions to Differences
• Open Doors to Each other

68
Outline

• Introduction - Specification Evolution
• Variability
• Volumetric Example
• Case Histories
• Summary and Conclusions

69
Summary

• Specification Changes
• Reliance on Volumetrics
• Awareness of LMLC vs. FMLC Issue
• Awareness of Importance of Testing Variability
• Awareness of Need for Refined New Test Methods
• Important to Understand Total and Components of
  Variability
• Must Strive to Reduce

S2QC/QA S2s S2t S2m/c
70
Summary

• Work Together - Cooperative Engineering Analyses
• Save
• Foster Positive Working Relationships
• Time Money
• Protect Public Perception of Both
• Solutions are Attainable!

71
Questions/Comments