Intelligent Soil Compaction:Research Update
By
David J. White, Ph.D.Mark Thompson, Pavana Vennapusa, Heath Gieselman,
Mike Kruse, Amy Heurung, and Eddy Blahut
Intelligent Compaction Open HouseAkeley, MN
July 20, 2006
Acknowledgements - UPDATE
FHWA Technology Deployment ProgramMnDOTIowa Highway Research BoardCaterpillar, Inc.Center for Transportation Research and Education (CTRE) at Iowa State University
Overview - UPDATE
IntroductionField Project Results:
February 2005; Peoria, ILAugust 2005; Peoria, ILJuly and October/November 2005; TH 14, Janesville, MNJune 2006; Peoria, ILJune, July 2006; TH 64, Akeley, MNJuly 2006; MnROAD, Albertville, MN
Upcoming Work
Phase 2 Methodology
Construct uniform test strips, varying soil type, moisture content, and lift thickness (23 strips)Compare power to engineering properties of compacted soil over entire compaction curve (e.g. 1, 2, 4, 8 passes)
DensityClegg Impact ValueDCP indexGeoGauge modulusPFWD modulusPlate load test (PLT)
Nuclear Moisture-Density Drive Core Dynamic Cone Penetrometer
Clegg Impact & Portable FWD GeoGauge Plate Load Test
Raw Data Along Strip Length
5 10
Dry
Uni
t Wei
ght (
kN/m
3 )
10
12
14
165 105 10
Net
Pow
er
0
5
10
15
20
25
30
NPpDD
5 10
Location Along Strip (m)
5 10
Pass1
Pass2
Pass4
Pass8
Pass0
Kickapoo Silt (Strip 1)
Average Clegg Impact Value
2 3 4 5 6Av
erag
e N
et P
ower
0
5
10
15
20
25NPp = -12.96*CIV + 64.72
R2 = 0.845
Average DCP Index (mm/blow)
0 50 100 150 200
Aver
age
Net
Pow
er
0
5
10
15
20
25NPp = 0.15*DCPI - 6.55R2 = 0.975
Average Dry Unit Weight (kN/m3)
8 10 12 14 16 18
Aver
age
Net
Pow
er
0
5
10
15
20
25NPp = -3.62*γd + 63.10
R2 = 0.996
CleggDCP Index
Kickapoo Silt (Strip 1)
Dry Density
Relationships for single test strip using averaged data (1 point per roller pass)
High R2 values, but generally only 4 points and identical soil conditions
Linear Regression Analysis
So how about combining data from test strips of varying moisture content, lift thickness, and cohesive soil type?
Average Net Power
0 5 10 15 20 25Av
erag
e E G
G (M
Pa)
0
20
40
60
80
R2 = 0.72
Average Net Power
0 5 10 15 20 25
Aver
age
DC
P In
dex
(mm
/blo
w)
0
50
100
150
200
R2 = 0.67
Average Net Power
0 5 10 15 20 25
Aver
age
Cle
gg Im
pact
Val
ue
0
3
6
9
12
15
R2 = 0.66
Average Net Power
0 5 10 15 20 25Ave
rage
Dry
Uni
t Wei
ght (
kN/m
3 )
10
12
14
16
18
20
R2 = 0.73
Clegg
DCP Index
GeoGauge
Dry Density
Kickapoo Silt (30-cm lift) – data from 3 strips
Multiple Linear Regression Analysis
Based on laboratory Proctor data, quadratic “compaction model” developed for relating net power to density, strength, and stiffness
wEb wEb wb wb Eb Eb b 265
243
2210d ⋅+⋅⋅+⋅+⋅+⋅+⋅+=γ
Moisture Content (%)
5 8 11 14 17 20
Dry
Uni
t Wei
ght (
kN/m
3 )
16
17
18
19
20
21
Iowa tillR2 = 0.96
Moisture Content (%)
5 8 11 14 17 20
Dry
Uni
t Wei
ght (
kN/m
3 )
16
17
18
19
20
21Edwards tillR2 = 0.94
Moisture Content (%)
10 13 16 19 22 25
Dry
Uni
t Wei
ght (
kN/m
3 )
14
15
16
17
18
19
LoessR2 = 0.93
Moisture Content (%)
8 11 14 17 20 23D
ry U
nit W
eigh
t (kN
/m3 )
15
16
17
18
19
20
ShaleR2 = 0.95
Model verification: laboratory data (points) and model predictions (lines)
Average Net Power
0 5 10 15 20 25Av
erag
e E G
G (M
Pa)
0
20
40
60
80EGG = b0 + b1*NPp + b2*NPp2
R2 = 0.96
R2 = 0.72
Average Net Power
0 5 10 15 20 25
Aver
age
Cle
gg Im
pact
Val
ue
0
3
6
9
12
15CIV = b0 + b1*NPp + b2*w + b3*NPp*wR2 = 0.98
R2 = 0.66
Average Net Power
0 5 10 15 20 25
Aver
age
DC
P In
dex
(mm
/blo
w)
0
50
100
150
200DCPI = b0 + b1*NPp + b2*wR2 = 0.93
R2 = 0.67
Average Net Power
0 5 10 15 20 25Aver
age
Dry
Uni
t Wei
ght (
kN/m
3 )
10
12
14
16
18
20γd = b0 + b1*NPp + b2*wR2 = 0.93
R2 = 0.73
Clegg
DCP Index
GeoGauge
Dry Density
Kickapoo Silt (30-cm lift) – data from 3 strips
RAPCA6-C
CA5FA6 CA6-G
Five base materials (IL DOT classifications)
Roller-Generated Data on Base MaterialsN
et P
ower
(kJ/
s)
0
10
20
30
0 4 8 12
CM
V
0
10
20
30
0 4 8 12 0 4 8 12 0 4 8 12
Roller Pass
0 4 8 12
RAP CA5-CCA6-C FA6 CA6-G
w = 8% w = 4% w = 4% w = 6% w = 8%
Power and CMV compaction curves for 5 stripsAll passes @ amplitude = 2.0 mm
Average Net Power (kJ/s)
10A
vera
ge C
MV
0
5
10
15
20
25
CMV = 34.08 - 12.10 log Pn
R2 = 0.97
305
Average Net Power (kJ/s)
10
Ave
rage
CM
V
0
5
10
15
20
25
CMV = 33.89 - 9.97 log Pn
R2 = 0.92
305
Average Net Power (kJ/s)
10
Ave
rage
CM
V
0
5
10
15
20
25
R2 = 0.84CMV = 65.84 - 21.37 log Pn
305Average Net Power (kJ/s)
10
Ave
rage
CM
V
0
5
10
15
20
25CMV = 47.20 - 15.37 log Pn
R2 = 0.96
305
FA6
CA5-C
CA6-G
CA6-C
CMV and Machine Power for Base Materials
Average CMV
0 5 10 15 20 25
Ave
rage
Dry
Uni
t Wei
ght (
kN/m
3 )
12
14
16
18
20
Average Net Power (kJ/s)
10
Ave
rage
EPF
WD (M
Pa)
10
20
30
40
50
5 30
Average Net Power (kJ/s)
10
Ave
rage
Dry
Uni
t Wei
ght (
kN/m
3 )
12
14
16
18
20
RAPCA6-CCA5-CFA6CA6-G
5 30
Mod
ulus
Power
Den
sity
Density and Modulus from Power and CMV
CMV
Average CMV
0 5 10 15 20 25A
vera
ge E
PFW
D (M
Pa)
10
20
30
40
50
1 2
5
3
CBR
0 5 10 15 20
Dep
th (m
m)
0
200
400
600
800
12345
Pt
0-75mm: w = 12%
0-75mm: w = 23%
(Rutting at Pts 1 and 2)
CBR
0 5 10 15 20 25
Dep
th (m
m)
0
200
400
600
800
1234
1
2 43
Location Along Strip (m)
0 5 10 15 20 25
k B
0
10
20
30
40
50
E LW
D (M
Pa)
-20
180
DC
P In
dex
(mm
/blo
w)
-30
70
Cle
gg Im
pact
Val
ue
-5
55
E PLT
(MPa
)
-20
340
ELWD
DCPCIVEPLT
Strip 1
Strip 1 (subgrade and median)
EPLT (MPa)
0 50 100 150 200
k B
0
10
20
30
40
50
ELWD (MPa)
0 50 100 150 200
k B
0
10
20
30
40
50
kB = 14.94*ln ELWD - 34.93R2 = 0.883
DCP Index (mm/blow)
0 20 40 60 80 100
k B
0
10
20
30
40
50
Clegg Impact Value
0 10 20 30 40
k B
0
10
20
30
40
50
kB = 0.156 EPLT + 2.341R2 = 1
kB = 1.379 CIV+ 1.110R2 = 0.861
kB = 91.16 DCPI -0.503
R2 = 0.614
CIV
DCP Index
Modulus
Modulus
Regressions for field measurements and kB
(Slide courtesy of )
Variable Feedback Control for UniformityC
ompa
ctio
n/So
ilSt
iffne
ss
Number of Passes/Time
Com
pact
ion
Dep
th
kB
20 25 30 35 40
Freq
uenc
y
0
20
40
60
80
100
Pass 3
μ = 30.5σ = 2.87n = 287
Evaluation of Variable Feedback Control
20 25 30 35 40
Freq
uenc
y
0
20
40
60
80
100
Pass 2
μ = 30.3σ = 2.14n = 280
20 25 30 35 40
Freq
uenc
y
0
20
40
60
80
100
Pass 1
μ = 31.7σ = 1.74n = 295
Strip 2 (relatively uniform Class 5)
AMMANN kB Comparison to Test Rolling
Location (m)
0 20 40 60
k B (M
N/m
)
5
10
15
20
25
30
Rut
Dep
th (c
m)
0
2
4
6
8
10
kB
Rut
Strip 6, Track 1
X Distance (m)
Y D
ista
nce
(m)
0
5
10
15
20
25
30
4.3 8.5 12.8 17.10.0
200-mm lift
510-mm lift
Test point
192 points
Design of spatial testing
Iowa State crew testing with nuke, Clegg, PFWD, DCP
CMVPower
Machine power and CMV distribution plots
Effect of lift thickness on…
CMV
0 5 10 15 200
40
80
120
160
Machine Power (kJ/s)
0 10 20 30 40
Freq
uenc
y
0
40
80
120
160
200 mm lift510 mm
Power
Machine power and CMV – raw data
CMV
0
5
10
15
20
25
30
0 3 6 9 12
0.0 4.3 8.5 12.8 17.1
X Distance (m)
CMV
Y D
ista
nce
(m)
0
5
10
15
20
25
300 5 10 15 20
0.0 4.3 8.5 12.8 17.1
X Distance (m)
Net Power (kJ/s)
Dry Unit Weight (kN/m3)
18 19 20 21 22
Freq
uenc
y
0
10
20
30
40
EPFWD (MPa)
0 10 20 30 400
10
20
30
40
Mean DCP Index (mm/blow)
10 20 30 40 50
Freq
uenc
y
0
10
20
30
40
Clegg Impact Value
0 2 4 6 80
10
20
30
40
ModulusDensity
CIVDCP Index
Distribution plots of field measurements
Moisture
Moisture and density – kriged surfaces (192 points)
Density
7.0 7.5 8.0 8.5 9.0
X Distance (m)
Y D
ista
nce
(m)
0
5
10
15
20
25
30
4.3 8.5 12.8 17.10.0
19.0 19.5 20.0 20.5 21.0
0
5
10
15
20
25
30
4.3 8.5 12.8 17.10.0
(a)
Dry Density(kN/m3)
Moisture
Moisture and CIV – kriged surfaces (192 points)
Clegg Impact Value
7.0 7.5 8.0 8.5 9.0
X Distance (m)
Y D
ista
nce
(m)
0
5
10
15
20
25
30
4.3 8.5 12.8 17.10.0
2 4 6 7 8
0
5
10
15
20
25
30
4.3 8.5 12.8 17.10.0
CleggImpact Value
Moisture
Moisture and modulus – kriged surfaces (192 points)
Modulus
7.0 7.5 8.0 8.5 9.0
X Distance (m)
Y D
ista
nce
(m)
0
5
10
15
20
25
30
4.3 8.5 12.8 17.10.0
10 13 16 19 22
0
5
10
15
20
25
30
4.3 8.5 12.8 17.10.0
Modulus(MPa)
Moisture
Moisture and DCP index – kriged surfaces (192 points)
Mean DCP Index
7.0 7.5 8.0 8.5 9.0
X Distance (m)
Y D
ista
nce
(m)
0
5
10
15
20
25
30
4.3 8.5 12.8 17.10.0
10 20 30 40 50
0
5
10
15
20
25
30
4.3 8.5 12.8 17.10.0
DCP Index(mm/blow)
Effect of confining pressure on sand density
110 120
Dep
th (m
m)
0
100
200
300
STA 250
110 120
STA 252
110 120
STA 254
Dry Density (pcf)
110 120
STA 256
110 120
STA 258
110 120
STA 260
250 252 254 256 258 260
Cle
gg Im
pact
Val
ue
0
5
10
15
20
Surface175 mm
Station
250 252 254 256 258 260
E PFW
D (M
Pa)
0
20
40
60
80
Effect of confining pressure on sand stability
Overcompaction?
0 3 6 9 12 15 18
Dry
Uni
t Wei
ght (
kN/m
3 )
16
17
18
19
0 3 6 9 12 15 18
DC
P In
dex
(mm
/blo
w)
10
20
30
40
Roller Pass
0 3 6 9 12 15 18
EP
FWD (M
Pa)
10
20
30
40
Overcompaction?
Who’s actually compacting the sand?
Upcoming Analysis with TH 64 Data
Regressions using proof testing dataEvaluation of IC machine use (e.g. calibration procedure)Spatial analysis
Spatial relation of various measurementsMultiple scales
Implementation of GIS for IC data management
Roller Pass
0 2 4 6 8E
PFW
D (M
Pa)
0
10
20
30
40
Roller Pass
0 2 4 6 8
Dry
Uni
t Wei
ght (
kN/m
3 )
8
11
14
17
20
Roller Pass
0 2 4 6 8
CM
V
0
5
10
15
20
Roller Pass
0 2 4 6 8
Net
Pow
er (k
J/s)
3
6
9
12
15
Density
CMV
Modulus
Power
Compaction curves for two strips at two amplitudes
2.0 mm0.8 mm
Net Power (kJ/s)
3 6 9 12 15
Dry
Uni
t Wei
ght (
kN/m
3 )
10
12
14
16
18
γd = b0 + b1*Pn+b2*ampR2 = 0.98
Mod
ulus
Power
Den
sity
Density and Modulus from Power and CMV
CMV
Net Power (kJ/s)
3 6 9 12 15
EP
FWD (M
Pa)
0
10
20
30
40E = b0 + b1*Pn+b2*amp
R2 = 0.93
CMV
3 6 9 12 1510
12
14
16
18
γd = b0 + b1*CMV+b2*wR2 = 0.85
CMV
3 6 9 12 150
10
20
30
40E = b0 + b1*CMV+b2*wR2 = 0.95
Upcoming Analysis with MnROAD Data
Amplitude studies (mapping and test strips)Variable soil conditions
Lift thicknessMoisture contentSoil type (subgrade and base materials)
Comparison of machines and IC outputCaterpillarBOMAGAMMANN
Resilient Modulus Testing
• TASK 1:– Relationship between Moisture-Density and
Mr• Task 2:
– Relationship between In-situ Modulus (from LWD and Roller predicted values) and Mr
• Task 3:– Relationship between laboratory compaction
methods and Mr
Task 1 – Test ResultsBorrow Material - Std Proctor
1500
1550
1600
1650
1700
1750
0 5 10 15 20 25
Moisture Content (%)
Dry
Dens
ity (k
g/m
^3)
0
20
40
60
80
100
120
140
Resi
lient
Mod
ulus
(Mpa
)
εp = 0.016%
εp = 0.3%
εp = 1.26%
εp = 0.79%
εp = 8.0%
SampleFailure
Task 1 – Test Results contd..
Borrow Material - Mod Proctor
1500
1550
1600
1650
1700
1750
1800
1850
1900
1950
2000
0 5 10 15 20 25
Moisture Content (%)
Dry
Den
sity
(kg/
m^3
)
0
50
100
150
200
250
Res
ilien
t Mod
ulus
(MP
a)
εp = 0.02%
εp = 0.006%
εp = 0.034%
εp = 0.73%εp = 13.3%
SampleFailure
Moisture Content Vs. Mr
y = -11.217x + 261.81R2 = 0.7277
y = -7.3779x + 158.31R2 = 0.7718
0
50
100
150
200
250
0 5 10 15 20 25
Moisture Content (%)
Res
ilien
t Mod
ulus
(MPa
)
High MrLow MrLinear (High Mr)Linear (Low Mr)
``
Strong Influence on Mr !!!
Relationships
• Moisture Content - highest influence on Resilient Modulus
• Non-linear regression model between Moisture Content – Density – Soil Index Properties and Resilient Modulus for a range of Subgrade Soils and Base Materials.
Task 2Prima/Zorno LWD In-situ Modulus and Roller Data
VsLaboratory Resilient Modulus
- Perform LWD tests on subgrade materials- Obtain Shelby Tube samples at the location of LWD
tests and perform Lab Mr.- Prepare re-constituted samples at target In-situ
Moisture and Density at the locations of LWD tests
Need more data !!!
Comparison - Zorno LWD Vs. Mr
y = 0.7644x - 42.377R2 = 0.8169
y = 1.6804x - 49.036R2 = 0.7375
0
20
40
60
80
100
120
140
160
180
200
0 20 40 60 80 100 120 140 160 180 200
Resilient Modulus, Mr (MPa)
LWD
Mod
ulus
, E (M
Pa)
High Mr
Low Mr
Linear (High Mr)
Linear (Low Mr)
Need More Data for Better Relationships !!!
Comparison - Prima LWD Vs. Mr
y = 0.563x - 19.033R2 = 0.6915
y = 1.49x - 36.696R2 = 0.922
0
20
40
60
80
100
120
140
160
180
200
0 20 40 60 80 100 120 140 160 180 200
Resilient Modulus, Mr (MPa)
LWD
Mod
ulus
, E (M
Pa)
High Mr
Low MrLinear (High Mr)
Linear (Low Mr)
Need More Data for Better Relationships !!!
Task 3
Laboratory Compaction Methods.. And its Influence on:
• Roller Compaction Data• Resilient Modulus
GIS Database
• Task 1:– Create spatial maps using Roller Data
• Task 2:– Results of all spot tests as a geographic
database• Task 3:
– Spatial analysis of spot tests from TH 64 over Class 6 base layer – Comparison to Roller Spatial Data
IC Research Projects at Iowa State University
FHWA/IHRB (Phase 1 completed, Phase 2 just completed)
http://www.ctre.iastate.edu/reports/tr495.pdfCaterpillar IC Evaluation (ending Fall 2006)Minnesota DOT IC Evaluation (ending Spring 2006)NCHRP 21-09 (just started, ending 2008)
Our plans for 2006 and 2007...Our plans for 2006 and 2007...
……on the roadon the road