Best Value Granular Material

for Road Foundations

Mn/DOT FY09 H09PS07 Research Project TAP Meeting

PI: Erol Tutumluer

RA: Yuanjie Xiao

August 28, 2009

Research Need Statements
9 Aggregate base materials are becoming increasingly
expensive in many parts of Minnesota
9 gravel mines & rock quarries are being lost to other land uses
9 Aggregate specification, production & placement are based
on testing techniques & design procedures that are several
decades old
9 There is likely significant opportunity for better value to
be achieved
9 by implementing new mechanistic design procedures & testing
techniques, road construction can better optimize material use &
reduce waste
9 Mechanistic pavement design and the field & lab tests
required would be needed to implement
9 A granular material best value software tool to be added to
MnPAVE to further encourage implementation of mechanistic design
 Research Objective
9 Demonstrate that locally available materials can be
economically efficient in the implementation of the
available mechanistic based design procedures in
Minnesota through
MnPAVE Mechanistic-Empirical Pavement Design Method

9 Develop the components of a new granular material

best value software module to be added to the
MnPAVE program

9 Provide pavement designers with index aggregate

properties linked to modulus & strength characteristics
and include example pavement designs
Expected Benefits
(i) Proper material selection & utilization according to aggregate
properties

(ii) Aggregate layer thickness optimizations during the design

process based on cost and mechanistic material properties
related to performance, and as a result;

(iii) More economical use of the locally available aggregate materials

in Minnesota

The benefits & costs of implementing new mechanistic

design procedures & material testing techniques would
be demonstrated by these designs
 Project Schedule

2008 2009 2010

Task to Perform 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9

Tasks Report

1. Establish Aggregate
Index Properties

2. Collect Aggregate
Strength and Modulus Data
3. Establish Linkages
Between Aggregate Properties
& Design Inputs

4. Sensitivity Analyses
5. Development of Best Value
Granular Material Selection Tool
Components

6. Draft Final Report

7. Final Report & Implementation

Completed In Progress Future

Project Tasks

Task 1 - Overview:
Establish Index Properties of Minnesota Aggregates
Used for Aggregate Base/Subbase Courses

Work with Mn/DOT engineers to identify & categorize

the types, sources, & properties of locally available
aggregates in Minnesota and obtain typical costs.
This is an essential task for:
• Identifying types & qualities of aggregates to establish
mechanistic strength & resilient modulus (MR) properties
• Conducting a benefit/cost study to demonstrate life cycle
benefits & costs of these aggregate materials typically
used to construct road foundations throughout Minnesota
 Project Tasks
Task 1 – Data Sources:
Aggregate Source Information System (ASIS)
[Link]

Files received from Mn/DOT

Date File Name Description
MNagg pits [Link] UTM Coordinates for
Sep.-19-08
[Link] Prospect pits
Designation explanation
Oct.-8-08 ProspectedPits_LimsRawResnames.xls
for Agg tests
MnDOT agg
Oct.-16-08 pit_samid_gradations_ps_tname.xls Agg gradation for all pits
all_pit_lims111408.xls
Aggregate prices for
Jan.-30-09 MnDOT MAP Agg [Link] MnDOT owned & leased
Gravel pits
Project Tasks
Task 1 – Data Sources (Cont’d):
•Aggregate source data
for MN counties
• Limited agg properties
• Merged with reliable
data submitted

ASIS Database
Spreadsheets

ASIS Online Interface

Project Tasks

Task 1 – Data Sources (Cont’d):

[Link] MNagg pits [Link]

(Map of prospect pits) (UTM Coordinates of prospect pits)
Project Tasks

Task 1 – Data Sources (Cont’d):

ProspectedPits_LimsRawResnames.xls
(Designations of other aggregate tests
– only gradation and Proctor data were useful)
Project Tasks

Task 1 – Data Sources (Cont’d):

MnDOT agg pit_samid_gradations_ps_tname.xls

(Gradation data for prospect pits)
(* 27 prospect pits out of 114 have no gradation data in this file)
Project Tasks
Task 1 – Data Sources (Cont’d):

all pit [Link]

(Mainly gradation data for samples from all pits)
 Project Tasks

Task 1 – Data Sources

(Cont’d):

MnDOT MAP Agg [Link]

(Aggregate price reference
and hauling cost information
for cost estimation)
Project Tasks
Task 1 – Methodology:
9 Considering the fact that prospect pits have the most
reliable gradation, it was decided to use prospect pits to
demonstrate the research methodology
MnDOT agg pit_samid_gradations_ps_tname.xls for Gradation
MNagg pits [Link] for UTM Coordinates

9 87 prospect pits from 34 counties were primarily

considered

9 Merged gradation data of prospect pits (reliable) & other

pits with ASIS database spreadsheets, and included
other aggregate index properties
Project Tasks
Task 1 – Methodology (Cont’d):
9 Of those 87 prospect pits, 56 had no cost information
recorded in ASIS database; default costs were thus
estimated using the known costs of the closest pits
MnDOT MAP Agg [Link] for default cost information

9 Aggregate location, property and cost data for 87

prospect pits were all collected and built into a GIS
based database for further analysis

9 The typical functions of this GIS based database include

searching, storing, retrieving, and displaying data
Task 1 – Selected Prospect Pits

• 87 prospect pits with most reliable gradation

selected for demonstrating the methodology
Task 1 – Collected Cost Information

Identify default
cost info

1
$ Without Cost Information
Approach:
Estimate using the known
costs of the closest pits
Task 1 – Collected Aggregate Data
• CLASS • MCLASS1 • MCLASS2
• QUAN1 • QUAN2
• COSTCYM1 • COSTCYM2
• YRPRICECL1 • YRPRICECL2
UTM Coordinates
From / to
ASIS
From From
MNagg pits [Link] MnDOT agg
pit_samid_gradations
_ps_tname.xls
From
MnDOT MAP Reliable Gradations
Agg [Link]

Default Costs
Aggregate Location, Property & Cost Data
Task 1 – Aggregate Database Illustration
ArcGIS based Database Management System (DBMS) was
developed for storing, retrieving and displaying aggregate
index properties

ArcGIS Functions:
• Search
• Store
• Retrieve
• Display

GIS based Aggregate Index Property Database

Task 1 – Aggregate Database Illustration
(Cont’d)

• Search for features

Task 1 – Aggregate Database Illustration
(Cont’d)

Graphical representation of
Percent Passing #200
property distribution
Passnum200
1
• Example feature
2.5 display
5
7.5
10
Task 1 – Aggregate Database Illustration
(Cont’d)

Graphical representation of
Cost per Cubic Yard
property distribution

Costcym1
0.1
• Example feature
0.25
display
0.5
0.75
1
Project Tasks

Task 1 - Summary:
Details of the techniques used to establish aggregate
index property database will be given in final report

Deliverables that have been sent include:

• Spreadsheet & Map of 87 selected prospect pits
• ASIS database spreadsheets for 87 Minnesota Counties
merged with reliable prospect pit gradations
Project Tasks
Task 2 - Overview:
Collect mechanistic pavement analysis & design inputs
as the strength & MR for unbound aggregate pavement
base/subbase applications, together with corresponding
aggregate index properties

LRBB Investigation 828 report (Chadbourn, 2007),

Davich et al. (2004) study, Kim & Labuz (2007) report,
other related research studies; a large database of
previous MR test results by the PI & data from the
current Illinois DOT research project on three different
types & qualities of aggregate materials
• This task does not include conducting new laboratory or field
aggregate tests
 Project Tasks
Task 2 – Data Sources:
Files received from Mn/DOT
Date File Name Description
Names & listings of
Oct.-10-08 MnDOT MRR agg [Link]
Available aggregate tests
MnDOT Mr Peaks last [Link] Mr load-time history, and
Nov.-21-08
MnDOT Mr [Link] Mr model k1 k2 k3 data
MnDOT Mr Agg Lab LIMS Column Gradation & Proctor data
Dec.-4-08
[Link] of Mr test samples
Resilient Modulus [Link] Mr equations, and Mr test
Dec.-5-08
[Link] status codes
Explanations of LIMS
Dec.-19-08 Material Types 12-19-08_1.xls
Material Types
Project Tasks
Task 2 – Data Sources (Cont’d):

Material Types 12-19-08_1.xls

(Material type explanations)

Resilient Modulus [Link]

Project Tasks
Task 2 – Data Sources (Cont’d):

[Link]
(Mr test status codes)
Project Tasks
Task 2 – Data Sources (Cont’d):

MnDOT Mr Agg Lab LIMS Column [Link]

(Gradation and Proctor data of Mr test samples)
Project Tasks
Task 2 – Data Sources (Cont’d):

MnDOT Mr Peaks last [Link]

(Load-time history data of last 4 or 5 cycles)
Project Tasks
Task 2 – Data Sources (Cont’d):

MnDOT Mr [Link]
(k values of Mr equations)
Project Tasks
Task 2 – Data Sources (Cont’d):

Actual Moisture Content & Dry Density for Mr Samples

Project Tasks
Task 2 – Data Sources (Cont’d):

The template for reporting Mr load-time history data

Project Tasks
Task 2 – Data Sources (Cont’d):

The template for reporting index properties of Mr samples

Project Tasks

Task 2 – Mn/DOT Database Analysis:

9 Grouped index property data by Aggregate Maximum

Sieve Size, Gradation Type (dense or gap), Percent
Passing No. 200 sieve, Density, etc.

9 Performed statistical analyses on MR data with an aim

to get typical MR ranges for different index property
groups
Project Tasks
Task 2 – Mn/DOT Database Analysis
(Cont’d):

63mm & 50mm

Project Tasks
Task 2 – Mn/DOT Database Analysis
(Cont’d):

37.5 mm
Project Tasks
Task 2 – Mn/DOT Database Analysis
(Cont’d):

31.5 mm
Project Tasks
Task 2 – Mn/DOT Database Analysis
(Cont’d):

25 mm
Project Tasks
Task 2 – UI Database Brief Introduction:
Databases collected from U of I previous research studies
• Recently-completed ICT R27-1 Project:
Characterization of Illinois Aggregates for Subgrade Replacement and Subbase
• NCHRP 4-23 Project:
Performance Related Tests of Aggregates for Use in Unbound Pavement Layers
• FAA P209 & P154 Granular Base Materials Study
Investigation of the Behavior of the FAA NAPTF P209/P154 Base/Subbase
Materials
• IDOT CA-6 & CA-11 Materials & Effect of Fines Content
Studies
Characterization of Anisotropic Granular Layer Behavior in Flexible Pavements
Project Tasks
Task 2 – UI Database Brief Introduction:

ICT R27-1 Project

Studied a Laboratory Aggregate Test Matrix with typical
midrange IDOT CA-6 gradations for constructing aggregate
layers as subgrade replacement & subbase

• Aggregate type: (1) dolomite, (2) limestone, (3) gravel

• Gradation: Midrange CA-6
• Fines content: 4%, 8%, 12%, & 16% passing No. 200 sieve size
• PI or plasticity of fines: 0% (non-plastic mineral filler) & 10% –
should be conducted on material passing the No. 40 sieve to be consistent with
IDOT procedure
• Moisture-density (compaction) condition: At optimum moisture
content (OMC), 2% dry of OMC, and 2% wet of OMC
ICT R27-1 Project: Engineered Gradations
(1) dolomite
4%, 8%, 12%, & 16%
passing No. 200 sieve size

(3) gravel

(2) limestone
Project Tasks

Task 2 - Summary:
Collecting MR and strength data for establishing a
comprehensive database will be continued in the
subsequent tasks

Task 2 Deliverables include:

• Aggregate strength and modulus database spreadsheets
• Corresponding index property database spreadsheets
Project Tasks
Task 3 - Overview:
Establish linkages between collected field and
laboratory aggregate strength and MR data and
aggregate physical properties for identifying mechanistic
design moduli ranges

• gradation
• shape, texture and angularity
• fines content
• PI of fines, and
• moisture state in relation to optimum moisture (OMC)
OR density achieved in relation to maximum Proctor
density (MDD)
 Project Tasks

Task 3 – Data Sources:

File received from Mn/DOT

Date File Name Description

Explanation of Mn/DOT
Feb.-06-09 MnDOT samples image testing [Link] samples used for image
analysis
Project Tasks
Task 3 – Data Sources Received:

MnDOT samples image testing [Link]

(Information about Mn/DOT samples for Image Analysis)
Project Tasks
Task 3 – Data Sources Used:

Gradation and Proctor data of Mr test samples

(Independent Variables)
Project Tasks
Task 3 – Data Sources Used (Cont’d):

Actual Moisture Content & Dry Density for Mr Samples

(Independent Variables)
Project Tasks
Task 3 – Data Sources Used (Cont’d):

k parameters of MR modulus models

(Dependent Variables)
Project Tasks
Task 3 – Methodology:
9 Organized Mr data and aggregate property data
• Mr tests without aggregate properties excluded
• Multiple aggregate index properties averaged

9 Related selected physical properties of MR samples to k

parameters of MR models
• 376 data sets (80% for model development and 20% for
validation)

K1, K2 & K3:

= f(index properties)
K4, K5 & K6:
Project Tasks
Task 3 – Methodology (Cont’d):

9 Both stepwise regression analysis and Artificial Neural

Network (ANN) modeling techniques were applied

9 Results from both techniques were compared

Project Tasks
Task 3 – Stepwise Regression:
Dependent Variable Independent Variables Note
K1 X1 - OMC Optimum Moisture Content
K2 X2 - MDD Maximum Dry Density

K3 X3 - Cu Coefficient of uniformity
X4 - Cc Coefficient of curvature
X5 – PP_3" Percent passing 75mm sieve

X6 – PP_2-1/2" Percent passing 63mm sieve

X7 – PP_2" Percent passing 50mm sieve

X8 – PP_1-1/2" Percent passing 37.5mm sieve

X9 – PP_1-1/4" Percent passing 31.5mm sieve
Selected
X10 – PP_1" Percent passing 25mm sieve
X11 – PP_3/4" Percent passing 19mm sieve Independent
X12 – PP_5/8" Percent passing 16mm sieve Variables
X13 – PP_1/2" Percent passing 12.5mm sieve

X14 – PP_3/8" Percent passing 9.5mm sieve

X15 – PP_#4 Percent passing 4.75mm sieve

X16 – PP_#8 Percent passing 2.36mm sieve
X17 – PP_#10 Percent passing 2mm sieve
X18 – PP_#16 Percent passing 1.18mm sieve

X19 – PP_#30 Percent passing 600um sieve

X20 – PP_#40 Percent passing 425um sieve

X21 – PP_#50 Percent passing 300um sieve

X22 – PP_#100 Percent passing 150um sieve

X23 – PP_#200 Percent passing 75um sieve

Project Tasks
Task 3 – Stepwise Regression (Cont’d):
Dependent Variable The Most Significant Variables
K1 X2 X13 X14 X15 X20 X23
K1=-2.2854+0.0260*X2+0.1085*X13-0.1550*X14+0.0511*X15+0.0128*X20-0.0354*X23
RMSE=0.591538
2
R =0.152309
2
Adj R =0.133 Regression
F=8.05543; p=5.189e-008
Results
K2 X2 X3 X8 X10 X18
K2= 9.7667-0.0180*X2+0.0077*X3-0.2059*X8+0.126*X10-0.0049*X18
RMSE=0.363049
2
R =0.374385
2
Adj R =0.363
F=32.315; p=0
K3 X2 X3 X7 X9 X11 X13 X15 X20
K3= -10.8816+0.0133*X2-0.0035*X3+0.2253*X7-0.141849*X9+0.0424*X11-0.0398*X13+0.0131*X15+0.0021*X20
RMSE=0.140868
2
R =0.326842
2
Adj R =0.307
F=16.2047; p=0

Possible reasons for low R2:

• Different test setups • Colinearity between index properties
• Data inconsistency • Critical index properties not available
Project Tasks
Task 3 – Stepwise Regression vs. ANN:
ANN Models:
ANN Models (27 input Variables):

• Slightly better R2 Values

• Different covariates identified
Dependent Variable The Most Significant Variables

R2 K1 X1 X2 X3 X5 X7 X9 X10 X11 X12 X14 X16 X18 X24

K1=-12.5519+0.7116*X1+0.1226*X2-0.0049*X3-1.6067*X5-0.2152*X7+26.4846*X9-13.1223*X10+0.6112*X11-0.0186*X12-0.2993*X14+0.0905*X16-
X25 X26

0.1093*X18+0.0496*X24-0.0770*X25+0.1913*X26
RMSE=0.533932
R2=0.332
Adj R2=0.294
F=8.63339; p=4.44089e-016
K2 X2 X5 X13 X17
K2= -0.9477+0.0141*X2-0.2363*X5+1.2853*X13-0.0095*X17
RMSE=0.146541
R2=0.260615
Adj R2=0.2497
F=23.8802; p=1.11022e-016
K3 X3 X4 X5 X6 X8 X11 X17 X18 x19 X21 X23
K3= 0.4168+0.0005*X3-0.0047*X4+0.7884*X5-0.0519*X6-2.1530*X8+0.1303*X11-0.1126*X17+0.1962*X18-0.1297*X19+0.0639*X21-0.0152*X23
RMSE=0.302675
R2=0.574822
Adj R2=0.557
F=32.4469; p=0
Project Tasks
Task 3 – Important Aggregate Physical Properties:

Previous research at UIUC indicated shape properties to

have a significant impact on aggregate modulus & strength

γd & %Crushed AI & ST

(Tutumluer, E. & Seyhan, U., 1998) (Tutumluer, E. & Pan, T., 2007)
Project Tasks
Task 3 – Imaging based Shape Indices:

The University of Illinois Aggregate Image Analyzer (UIAIA) System

Project Tasks
Task 3 – Image based Shape Indices (Cont’d):

Angularity Index (AI) Surface Texture (ST) Index

Aggregate Type
Range Mean Range Mean

Uncrushed Gravel 250-350 300 0.5-1.20 0.90

Crushed Gravel 300-450 400 1.00-1.50 1.20

Crushed
400-550 475 1.20-1.80 1.60
Limestone

Crushed Granite 500-650 550 1.80-2.90 2.20

Typical Ranges and Mean Values of AI and ST

(Pan and Tutumluer, 2005)
Project Tasks
Task 3 – Image based Shape Indices:

12 samples received
from Mn/DOT for
Image Analysis

All analyzed to develop

imaging shape indices
except for
Dark colored
TH 52 Taconite Tailings

Very fine-graded
(< 2mm)
TH 47 SGB
Project Tasks
Task 3 – Image based Shape Indices (Cont’d):

Image Analysis Results of Mn/DOT Samples Processed using UIAIA

 Project Tasks X24 – F&E Ratio
X25 – AI
X26 – ST
Task 3 – Regression with Shape Indices: X27 – SA

Dependent Variable Covariates Goodness of Regression

X1, X2, X23 R2 = 0.898368; Adj R2 = 0.837389 (base line)

X1, X2, X23, X24 R2 = 0.898443; Adj R2 = 0.796887 (Ð)

Add FE_Ratio
K1 X1, X2, X23, X25 R2 = 0.935412; Adj R2 = 0.870823 (Ï)
Add AI
X1, X2, X23, X26 R2 = 0.983525; Adj R2 = 0.967051 (Ï) Add ST
X1, X2, X23, X24, X25, X26 R2 = 0.989273; Adj R2 = 0.957092 (Ï) Add all three
X1, X2, X10, X23 R2 = 0.910647; Adj R2 = 0.821293 (base line)

X1, X2, X10, X23, X24 R2 = 0.914783; Adj R2 = 0.772754 (Ð) Add FE_Ratio
K2 X1, X2, X10, X23, X25 R2 = 0.937404; Adj R2 = 0.833077 (Ï) Add AI
X1, X2, X10, X23, X26 R2 = 0.913328; Adj R2 = 0.768874 (Ð) Add ST
X1, X2, X10, X23, X24, X25, X26 R2 = 0.999937; Adj R2 = 0.999498 (Ï) Add all three
X10, X13, X20 R2 = 0.745918; Adj R2 = 0.593468 (base line)

X10, X13, X20, X24 R2 = 0.749059; Adj R2 = 0.498117 (Ð)

Add FE_Ratio
K3 X10, X13, X20, X25 R2 = 0.889244; Adj R2 = 0.778487 (Ï) Add AI
X10, X13, X20, X26 R2 = 0.873318; Adj R2 = 0.746635 (Ï) Add ST
X10, X13, X20, X24, X25, X26 R2 = 0.932507; Adj R2 = 0.730028 (Ï) Add all three
• ST is the most important for K1; while AI is the most important for K2 and K3
Project Tasks
Task 3 – Methodology (Cont’d):
9 Stepwise regression equations for K1 K2 and K3 without
shape properties have low R2 values
9 Predictive equations have high R2 values only when
shape properties are included

9 Typical trends in K values can be estimated for different

group MR data groups for the conditions:
‰ Dense(Achieved density>=125 pcf)/ loose(Achieved density<125 pcf)
‰ Crushed(Angularity Index AI>=400)/ uncrushed(AI<400)
‰ Coarse(Max. size>=#4)/ fine(Max. size<#4)
‰ Clean(Percent passing #200<8%)/ dirty(Percent passing #200>=8%)
 Project Tasks
Task 3 – Grouping results (Cont’d):
20 .0 0

5.00

10 .0 0
2.50

$
D D$ $
D D$
$ D$ $
D$
D D$
$ D
$
$
D D
$
$
D $$ $ $
D $
D$$
$
D$
D$
D $
D$
D D$
D D $
D
$
$
D
D$
D$
$
D
D
$
$
D
D$
D$
D
$
D $
D D
$ D
$D D
D
$
$D
D
$
$ DD
$ D
$
$D$ D
$ $
D $
D $$ $
D $
D $
D$
$
D
D D$
$
D D$ D
D $
D$
D
$
$
D$
D$
D$
D$
D $
D$
DD$
$
D$
D
$
D $
D
$$
$
D
D $
D $
D$
DD$
$ D
$ D
$
D
$D
$
$D
$ D
$ $ $
D $

K2
D
$D $D
D $D
$D$D D $ D
D $
D$
DD $$
D D DD D$D$
$
D
D
$
DD
$
D $ $
D $
D $
D$
$
$
D
D
D$
D$
D
$
D$
D$
D
$
$
D
D$
D$
D
D$
D $
D$
D$
$
D$
D$
D$
D
D$
D
D$
D$
D$
D
$
D$
D D D
$
$D
$
D D
$ D$
$
K1

$D
D $
D D D$
D$ D$
D$
D$D$D D
$$
D $
D $
D$ $ $
$$
DD
D
$$
$
D$
D$
$$$$$
$
$
D
D
D$
D $
D$
D D
$ D D$
$ D
D$D
D $$
D
$ D
$D $D $
D $
$$ D$ $ $D $ $
D $
D$ DD
D D D
DDD$
D $
D$
D$
D $$
$
DD
D
D $$
D D
$ $
D D D
D D $
D
$ $
D$D$D
$ $ $D $D
D $ DD
$
$ $
D D D$
D D
$
D$
D $
D $
D D$ D D$
$ D
D
$$
D D
$ D
$D
$D
$D
$ D
$D
$ D$ D
$D
$D
$
D
$ D
$$
D
D$
$
DD D
$
D
$
$
D$
D
D
$$
D
D
$
$
D
$
DD
$
D
$
DD
$
D
$
D
D
$
$D
$ $D
$
D
$D
D
$$D
D
$
$ DD
$D
$DD
$
$
D
$
$
D
$D
$
$
D
D$
D$D
$D
$
D
$ D
$
D$ $
D $
D$
DD
$
$D
$
D
$
D$
D D$D
$$D
$D $ $D
D
$
D $ D
D$ $ DD $
$ $ D$
D$
D $
D$
D
$
D$ $
D $
$
D $
D
$
D $
D $
D$
D D
$ $
D
$
D$
D
$
D$
D$ $
D
D
$ D
D
$
$
D$
$ DD
$ $
D$
D D
$D
$
$ $
D D$$
D
D$D
D
D
$$DD
$ D $ D$D
$D D
$
$$D
D$ D$ D
$
$
D
$D
$$
D$D
D$D
D
$ $ D
D $
D
$
$D D$
$D$
D
$
D$D
D$
$
$
D
DD
$D
D
$
$D D
$ D
$D
D
$ D
$D
$
D$
DD
D
$
$
D
$
$D
$
$$
D$
D$D
D$D
$D
D
$D $
D
$D$$$D
D D $
$D $ $ D $
D $ D D
$ D
$
D$ $D
$ $ D
D $ D $ D $D
$ D DD
$D
$
$ $D
$D$
$D
$
$
D
D$D
$
D
D
$
D
$
$
D$D
D $
$
D$
D
$
D$
DD
$
D
$D
$$D
$DD
D
$
$D
$
$
DD
$
D
$
D
$
D
$D
$
$D
D
D
$
$
D $
D
$$D
D $$D
D D
$
$
D$
$D
$
D
$$
D
D
$
D$
D
$D
$
D
D
$$D
$
D
D$
D$
$D$ $
D
$$
D DD D
$D
$$$
$
D
D
D
$D
DD
$
D
$$
$D
D$
$D
$D
$D$
$
$
D$
D
D$D
$D$ D $
D
$
D
$
D$D
D$
D
0.00
$
D D
$ $
D
0.00

-2.50

-10.00

-5.00

11 0.00 12 0.00 13 0.00 14 0.00 11 0.00 12 0.00 13 0.00 14 0.00

De nsity 10 .0 0 De nsity

5.00

Select 125 pcf $

D
$
D
as density threshold
$
D
D$
$
D $
D
$
D $
$
D
D
K3

$ D
D $
D $ D
D $ D
D $ $ D
D
DD
$$ $
D $
D $
D
$ $
D$
D$
D$
$
D
D $
D
$D $
D
$ $
D$
D $
$
D
$ $
$
D
D
$ $$D
$ $$
D$
D
$
D$
D
$D$
D
$
D$
D
D$
$D $
$
D$D
$
D
D $D
$$D
D$ $ $$
$DD$
$ D
D $
0.00 $$ D
D $D
$D $D
$D$ D
D $D
D D
$D D
$
D
$D
$D$D $D
$
$D $
$
D
D $
D
D$
D
$
$
D$
D
$
DD
$DD
$
D$
D $D
D D$D
$
D $ D $
D$D
D $
$D DD
D
$ $
D$
D
DD$
$ D
$
D$ $
D DD$
D
$
D$
D$
D $
$D $$
D $D
D$
$D$DD
$D$D
$
D $D
$
$
D D$
D
$
D
$
D
$
D
$$
D
D
$$
$
$
$
$
$
D
D
$
D
$
D
$D
D
D $
$D
$
D$
$
D
D$
$D
D $
$D$
DD
$
D
$
D
$
DD
D
D
$
$
$
D
$
$
$
$D
D
D
$
$$
$
D
$D
D
D$
$$
D$
$DD
$
D
$
$D $D
D $$
D$
D$D
D $
$
D
D
$
$
D
$D
D$
$
$D
D
$
D $
$D $
D
D
$$
D $D
D $D
D $
D
$
$ D
D
D$
$D $ $ $
D$
D$
D$
$
D
D
$D
$D
D $$
D
D
D$
$
D
$
D$
D
$
$D
$D$
$
D
D$
D
$D
D
D
$
D
$
$D
D
D
$D
$
$DD$$
$
DD$
D D
DD
$
D
D$
D
$
D
$
DD
$
D
$ DDD
D
$ $ $$
D $
$
D $
D
$D$
$
DD
$
DDD
$
$
D $
D $D
$$
DD$D
$
D
D $ D DD D
$ $D
D $ D $
D
$ D$

-5.00

-10.00

Dense/ Loose 11 0.00 12 0.00

De nsity
13 0.00 14 0.00
Project Tasks
Ndense = 136
Task 3 – Grouping results (Cont’d): Nloose = 55
Total = 191
(for 15 Sequences)

Dense/ Loose 1 psi = 6.89 kPa

 Project Tasks
Task 3 – Grouping results:
Crushed/ Uncrushed
1.00

0.00
3.00 $
D
$
D D
$
$
D
D
$ D
$ $
D
$
D
0.80
D
$
D
$
D
$ -0.50
2.00 D
$ $
D
$
D
K1

K2
D
$

K3
$
D

D
$
D
$
$
D D
$
0.60
1.00 $
D -1.00 $
D
$
D
$
D
D
$
D
$
$
D

0.00
0.40 -1.50

300.00 350.00 400.00 450.00 500.00 300.00 350.00 400.00 450.00 500.00 300.00 350.00 400.00 450.00 500.00

AI AI AI

K1, k2, k3 vs. AI •43 Mr tests with AI < 400;

(AI values from imaging of Mn/DOT samples) •72 Mr tests with AI >= 400;
•Total = 115
Project Tasks
Task 3 – Grouping results (Cont’d):
N=376
Coarse/ Fine Mean=1.70
Std. Dev. = 0.675 Total
Normal Distribution Line

N=363
Mean=1.70
Std. Dev. = 0.684 Coarse

N=13
Mean=1.55
Std. Dev. = 0.283 Fine

Histogram of K1
Coarse/ Fine K2 and K3 similar
 Project Tasks
Task 3 – Grouping results (Cont’d):
Clean/ Dirty
4.00

8.00

2.00
2.00

$
D
4.00 $
D
1.00 $
D$ D$
D $
D
$
D $ D $$
D
$
D$
D $
D
$
D$
D
$
D$ D$
D $
D D D$
$
$
D D$ $
D
$
D$ D $
D
D $
D $
$D $ $D
$ $
D $
D
DD
D $ D
D $
DD $$
$
D
D $
D DD $
$ D $
D
D$ D$
$ $ $
D$
D
$
D $
D $
$
D
D 0.00 $
D $
D$
D
$
DD
$$
D$
D $
DD
$ D
$D $DD$D
D D$ $
D $
D $ $ D
D
$
$ D
D $ D
$D $D
$D
$D
$
$$ D
$
D
$ D
$D
D $$ D
$D $$
D D
$ $
D$
D
$
D $
D
$
D
$
D $
D
$
D D$
$ D
$
D$
D
D$
$ D $ DD $
$
$
D D $
$
D$
D
$
D$$
DD$
D
D
$D D$DD
$D
$ D D
$ $
D D
$D D
$$D
$D
$ D $$D $ D $
D$ D $ D
D
$ $D
$D $D
$ D $ $ D $ D
$$
$
D D $
D$ $
D
$
$
D
D D
$
$
D $
D$D$D$
D $ D$
D $
D
$
D
$
D D
0.00 $
D$D $
D
0.00

-2.00

-1.00

-4.00
-4.00

-2.00
5.00 10 .0 0 15 .0 0 20 .0 0 5.00 10 .0 0 15 .0 0 20 .0 0
5.00 10.00 15.00 20.00

PP#200 (%) PP#200 (%) PP#200 (%)

• 257 Mr tests with PP#200 < 8%;

K1, K2, K3 vs. PP#200 (%)
• 119 Mr tests with PP#200 >= 8%;
•Total = 376
Task 3 – ICT R27-1 SASR ANOVA Results:

9 For ICT R27-1 project results, two primary procedures

were conducted using SASR (Statistical Analysis Software)
to develop the best models for predicting K-parameters:
9 ANOVA (Analysis of Variance)
9 Stepwise regression

9 The ANOVA is a common method to study the effect of

treatments or independent variables (aggregate properties
in the test matrix herein) on some dependent variables
(K1, K2, …, K6 herein)
Task 3 – ICT R27-1 SASR ANOVA Results:
From the plot, it can be
seen that the mean K1
values are different for
different AI values.

The p-value (<0.0001)

from ANOVA is less than
0.05 (alpha level
selected at this stage)

Therefore, AI value has

a significant effect on K1

α = 0.05 probability of
Type –I error
Task 3 – ICT R27-1 SASR ANOVA Results:

From the plot, it can be

seen that the mean K1
values are different for
different ST values.

The p-value (<0.0001)

from ANOVA is less
than 0.05 (alpha level
selected at this stage)

Therefore, ST value
does have a significant
effect on K1

Source DF Type I SS Mean F Value Pr > F

Square
α = 0.05 probability of
ST 2 1.33096 0.6654799 18.29 <.0001 Type –I error
Task 3 – ICT R27-1 IDOT Project Results
ST for Limestone ST for Gravel

AI for Limestone AI for Gravel

Task 3 – ICT R27-1 SASR ANOVA Results:
From the plot, it can be seen
that the mean K1 values
corresponding to each optimum
moisture content, demonstrate a
roughly increasing pattern

The p-value (=0.0001) from

ANOVA is less than 0.05 (alpha
level selected at this stage)

Therefore, Optimum Moisture

Content (OMC) has a
significant effect on K1

α = 0.05 probability of
Type –I error
Task 3 – ICT R27-1 SASR ANOVA Results:
From the plot, it can be seen
that the mean K1 values
corresponding to each Cu,
demonstrate a roughly
decreasing pattern

The p-value (=0.0009) from

ANOVA is less than 0.05 (alpha
level selected at this stage)

Therefore, coefficient of
uniformity (Cu) has a
significant effect on K1

α = 0.05 probability of
Type –I error
Task 3 – ICT R27-1 SASR ANOVA Results:
From the plot, it can be seen
that the mean K1 values
corresponding to each fines
content, demonstrate a
roughly decreasing pattern

The p-value (=0.0009) from

ANOVA is less than 0.05
(alpha level selected at this
stage)

Therefore, fines content

(passing #200 sieve) has a
significant effect on K1

α = 0.05 probability of
Type –I error
Task 3 – ICT R27-1 SASR ANOVA Results:
From the plot, it can be seen that
the mean K1 values
corresponding to Plastic/Non-
Plastic, do not demonstrate any
particular pattern.

The p-value (=0.2773) from

ANOVA is greater than 0.05

Therefore, Plastic/Non-Plastic
fines categories do not have a
significant effect on the value
of K1

α = 0.05 probability of
Type –I error
Project Tasks
Task 3 - Summary:
Correlations developed between collected laboratory
MR data and aggregate index properties. Certain trends
were verified with ICT R27-1 project findings.

Task 3 Deliverables include:

• Developed stepwise regression equations for predicting
K model parameters from aggregate physical properties
• Image analysis results and improved correlations for 12
samples received from Mn/DOT
• Comparisons between stepwise regression equations and
LTPP equations
Project Tasks
Task 4 - Overview:
Conduct sensitivity analyses for mechanistic design
moduli inputs & seasonal pore suction resistance
factors for different Mn/DOT aggregate classes using
the MnPAVE program to generate pavement life
expectancies

Relative importance of all pavement design input

parameters will be better understood
• guidelines will be established to choose a range of design
moduli for different Mn/DOT aggregate classes
• target values for strength, modulus & thickness will be
recommended for different design scenarios involving
various types & qualities of locally available aggregate
materials
Project Tasks
Task 4 - Methodology:
9 Aggregate properties that may significantly affect MR
were identified from both Mn/DOT & ICT R27-1 data:
• AI &/ or ST
• Optimum moisture content (OMC)
• Maximum Dry Density (MDD)
• Coefficient of Uniformity (Cu)
• Percent Passing #200 (% fines)
9 Different design scenarios considered for Mn/DOT:
• Dense & Loose (Achieved γd)
• AI/ ST or Crushed & Uncrushed (find K1,2,3 trends)
• Clean & Dirty (PP#200)
• Coarse & Fine
• Dry, Opt. and Wet of OMC
Project Tasks
Task 4 – Methodology (Cont’d):
9 For those cases developed, effects of changing
aggregate properties on K1, K2 , K3 and hence MR
mechanistic design inputs will be investigated

9 Default values of K1, K2 , and K3 will be recommended;

MR values at typical stress states suggested by NCHRP
1-28A will be predicted

9 Comprehensive matrix of design moduli and seasonal

pore suction resistance factors will be used to conduct
MnPAVE analyses and identify the sensitivity of the
design inputs to pavement life expectancies
Project Tasks

Task 5 - Overview:
Develop a best value software tool to incorporate into
the MnPAVE program and implement mechanistic
pavement design concepts in aggregate
selection/utilization

Aggregate Material
Resource Map similar to
Soil Class Map
• source
• type
• quality
• impact on M-E design
Project Deliverables

9 A Final Report will be prepared at the end of the

2-year study to include all research findings

9 Revised MnPAVE Manual pages will be prepared

for The Best Value Software Tool by giving
examples on how to use the developed correlations
and mechanistic design moduli inputs

The ultimate benefit: More economical use

of the locally available aggregate materials in
Minnesota
Thank you!..