Austin, Texas, 24 April 2019

Geosynthetics in Roadways
over Expansive Clays
Jorge G. Zornberg, Ph.D., P.E., F. ASCE
The University of Texas at Austin, USA
Past-President, IGS

Expansive Clays

Unit contains abundant

clay having high swelling
potential
Part of unit (generally
less than 50%)
consists of clay having
high swelling potential

Source: USGS 1989

Examples of Detrimental Effect of
Expansive Soils
On Buildings: On Roadways:

Pavements on Expansive Clays
Location of
longitudinal cracks
Original ground profile

CL

Source: Zornberg and Gupta (2009)

The Badlands

1-year Expansive Clays

(adapted
FEMA 1997)
Expansive Clays

Smectites
Smectites

• Corresponds to the 2:1

type of clay minerals
• Montmorillonite is a
smectite where every sixth
Al3+ has been replaced by
a Mg2+
• n. H2O + cations between
layers
• Interlayer water can come
and go easily

TxDOT Test Procedure Tex-124-E

TxDOT Procedure Tex‐124‐E
TxDOT PDM Chapter 3 Section 2:
Tex-124-E, “Determining Potential Vertical
Rise,” is the recommended procedure for
determining PVR. A 15-foot soil column is
recommended for the analysis to determine
PVR. The least amount of PVR for design
is 1.5 inches for main lanes (2.0 inches for
frontage roads, when allowed), or as
established by the district SOP identifying
the requirements.
TxDOT Test Procedure Tex-124-E
TxDOT Procedure Tex‐124‐E
Pluses: Minuses:
• Good practical • Too many correlations:
implications: - To define linear swell from
– Outcome (i.e. PVR) easy volumetric swell
to grasp by designers - To define volumetric swell (at a
– Outcome can be related given load) from free volume swell
to performance - To define free volume swell from
• Accounts for the PI
relevant variables: - To correct for unit weight
– Soil characteristics - To correct for % binder
– Stratigraphy • Problematic experimental data:
– Initial moisture content - Too little
– Confining stresses - Too old
- Correlations extrapolated beyond
available data

McDowell’s 1956 Method

McDowell (1956) Tex-124-E

Tex‐124‐E Procedure
McDowell (1956) Tex-124-E

Typical Swell Test Results
9.0%

8.0%

7.0%

6.0%
Swelling (%)

5.0%
FS ‐ 1‐14
4.0%

3.0%

2.0% 1 Day 1 Week 20 Days

1.0%

0.0%
0 100 200 300 400 500 600

Time (hr)
6048 Centrifuge‐based Approach

Centrifuge axis

Ng

Small Centrifuges for Direct 
Measurement of Swelling
Centrifuge Device:
• Floor mounted
• Low cost
• Can achieve very high g‐
levels 
• In‐flight data acquisition 
system

Measurements:
• Six simultaneous 
specimens tested at once
• Vertical displacements
• G‐level
Source: Zornberg et al. (2017)
Typical Swell Test Results
9.0%

8.0%

7.0%

6.0%
Swelling (%)

5.0%
FS ‐ 1‐14
4.0%

3.0%

2.0% 1 Day 1 Week 20 Days

1.0%

0.0%
0 100 200 300 400 500 600

Time (hr)
Source: Zornberg et al. (2017)

Typical Swell Test Results
9.0%

8.0%

7.0%

6.0%
Swelling (%)

5.0%
FS ‐ 1‐14
4.0% Centrifuge ‐ Corrected

3.0%

2.0% 1 Day 1 Week 20 Days

1.0%

0.0%
0 100 200 300 400 500 600

Time (hr)
Source: Zornberg et al. (2017)
Swell‐stress Curve for Cook 
Mountain Clay
14%

12% Centrifuge ‐ 6048
Vertical Strain (%)

ASTM D4546 ‐ Method A
10%
Stress‐Swell Curve
8%

6%

4%

2%

0%
10 100 1000

Effective Stress (psf)

Swell‐stress Curve for Eagle Ford 
Clay
30%

Centrifuge ‐ 6048
25%
ASTM D4546 ‐ Method A
Vertical Strain (%)

Stress‐Swell Curve
20%

15%

10%

5%

0%
100 1000

Effective Stress (psf)
Benefits of the Centrifuge Testing 
Approach
• Expeditious
• Highly repeatable
• Generates swell data from multiple soil specimens in 
a single spin
• Requires comparatively small laboratory space
• Provides direct measurement of swelling:
– No need to use correlations with index properties
– Generates soil‐specific, project‐specific data
• Data suitable for generation of soil‐specific 
databases
• Results can be readily used for prediction of PVR

Geosynthetics in 
Roadways
Introduction: Paved Roads
• Very Strong
• Strong • Durable
• Free-draining • Impermeable
• Manufactured • Manufactured
• Still expensive • Expensive

• Moderate strength
• Free-draining
• Natural material
• Less expensive

• Weak
• Moisture sensitive
• In-situ soil
Source: Christopher et al. (2006)

Geosynthetic Materials in Roadways
Geotextiles:
Woven or nonwoven
PP, PET

Geogrids:
Uniaxial, biaxial, multiaxial 
HDPE, PP, PET, PVA

Geocells:
Smooth, perforated
HDPE
Source: Geosynthetic Institute (GSI)
Geosynthetic Functions in Roadways

ASPHALT OVERLAY
ORIGINAL PAVEMENT

BASE

SUBBASE

SUBGRADE

Source: Zornberg (2017)

Geosynthetic Functions in Roadways
ASPHALT OVERLAY
Separation

ORIGINAL PAVEMENT

BASE
Barrier
Reinforcement

Filtration
SUBBASE Stiffening

Drainage

SUBGRADE

Source: Zornberg (2017)

Stabilization of Road Bases
ASPHALT OVERLAY
ORIGINAL PAVEMENT

BASE
Stiffening

SUBBASE

SUBGRADE

TxDOT Experience
Roadways over Expansive Clay 
Subgrades
Roadway over expansive clay subgrade – Dry Season:
C.L.
Original
ground profile

Roadway over expansive clay subgrade – Wet Season:
C.L.

Original
ground profile

Distance from C.L. (mm) Distance from C.L. (mm)
‐4000 ‐3000 ‐2000 ‐1000 0 1000 2000 3000 4000
Elevation from C.L. (mm)

‐20

‐70
Sec #8 8/11/2015 Sec #8 8/25/2015 Sec #8 9/27/2015
Sec #8 1/28/2016 Sec #8 6/16/2016 Sec #8 2/21/2017
Sec #8 6/20/2017 Sec #8 8/9/2017 Sec #8 10/28/2017
Sec #8 1/30/2018 Sec #8 3/25/2018
Effect of Geosynthetic Stabilization
Geogrid Section 1: No longitudinal FM 1915 (Milam County)
cracks
Control Section: Longitudinal cracks

Geogrid Section 2: No longitudinal

cracks

Lesson: Geosynthetic reinforcement prevented the

development of longitudinal cracks
Source: Zornberg and Gupta (2009)

To Be or Not to Be?
SH7 (Bryan District)

Geogrid is there, but 1.2 m short…

What are they laughing at?
No Geogrid!
Geogrid Section: Longitudinal
cracks soon after construction
Lesson: Geogrids appear to work … if in place.
To Spec or not to Spec?
FM 1774 (Grimes County)
Geogrid Section 1 (Product A): No
longitudinal cracks

Geogrid Section 2 (Product B):

Longitudinal cracks

Lesson: Current geogrid specifications have not led

to consistent system performance Source: Zornberg and Gupta (2009)

FM2 Evaluation
Level
FM2: 
Distress 

Avg Control

Avg Lime

Avg GS07+LM

Avg GS05+LM

Avg GS06+LM

Avg GS07

Avg GS06
GS06

FM2: Distress Level
Avg GS05
GS05
GS07
Performance  Avg GS07 Avg GS06 Avg GS05 Avg Control   

over time

Control and
Geosynthetic-
stabilized
Sections
(Surveys #14
to #18)

Performance  Avg Control    Avg Lime     Avg GS06

over Time Avg GS07 Avg GS05

Control,
Lime-treated,
and
Geosynthetic-
reinforced
Sections
(Surveys #14 to
#18)
The Quest for an Adequate
Performance Property

Stabilization Mechanisms

Membrane tension support Bearing capacity increase

Lateral restraint

Source: Haliburton et al. (1981)

A New Property: Why?
- The ultimate goal is to minimize the
risk of collapse
Earth - Design based on the performance
Retention of systems under failure condition

Systems - Relevant material properties

correspond to large displacement
conditions
Geosynthetic-
reinforced
- The ultimate goal is to minimize
Systems excessive distress
- Design is based on the
Pavement performance of roads under
Systems service conditions
- Relevant material properties
correspond to small displacement
conditions

A New Property: Why?
Back to the basics:

Membrane tension support Bearing capacity increase

Lateral restraint

Source: Haliburton et al. 1981

Stabilization of Road Bases

Tendency for aggregate

to displace laterally

Asphalt layer

Base
Geosynthetic-
induced lateral
restraint
Stress distribution Geosynthetic Stress distribution
Stress
distribution Subgrade

Non-stabilized Road Base Stabilized Road Base

Source: Zornberg (2017)

A New Property
An evaluation was conducted:
 To identify:
- a single, yet
- relevant parameter,
that quantifies the:
Confined Stiffness of the Soil-Geosynthetic
Composite under Small Displacements
 To develop a practical experimental approach to
obtain it using:
- Monotonic loading
- Conventional load frame
- Comparatively expeditious procedures
Analytical Framework: Solution
T(x)2 = KSGC  u(x)
• KSGC = Stiffness of the soil-geosynthetic composite
• T(x) = Unit tension at location x
• u(x) = Geosynthetic displacement at location x

with:
KSGC = 4 y  Jc
• y = Yield shear stress
• Jc = Confined stiffness of the geosynthetic

61
Source: Zornberg et al. (2017)

Experimental Setup (1st Generation):
Large Soil‐geosynthetic Interaction Device


u0 Frontal 
Tension (T0)
u

u0

u5 Frontal 
u4 Tension (T0)
u3
u2
u1
• Large volumes of soil
• Significant effort and time
• Comparatively complex testing procedures
Experimental Setup: Determination 
of Model Parameters
Determination of KSGC from T2 vs. u curves:
y Displacement (measured directly at multiple
ui locations)
T0 Frontal unit tension (measured directly)
T0
Unit tension (estimated using T0 and interface
Ti
shear relationship)

T2
LP 3
T0
LP 2
Ti
LP 1

ui

Li u

2
T
K SGC  i
ui

Experimental Setup: Typical 
Results
Approach to define KSGC from T2 vs. u curves:

T vs. u T2 vs. u
16 120
y = 40.795x
Unit Tension Square (kN/m)2

LVDT1
14
100 R² = 0.9915
Unit Tension (kN/m)

12
80 Linear (LVDT1)
10

8 60

6 40
4
LVDT1 20
2 Parabolic Linear
0 0
0 2 4 6 8 10 0.1 0.6 1.1 1.6 2.1 2.6
Displacement (mm) Displacement (mm)

(kN / m) 2
K SGC AVG  41.5
mm
 n = 3 psi
Source: Zornberg et al. (2017)
Experimental Setup: Typical 
Results
Approach to define KSGC from T2 vs. u curves:

T vs. u T2 vs. u
16 120
LVDT1 y = 40.795x

Unit Tension Square (kN/m)2

LVDT2
14 100 R² = 0.9915
LVDT3
Linear (LVDT1)
Unit Tension (kN/m)

12
80 Linear (LVDT2)
10 Linear (LVDT3) y = 39.763x
R² = 0.9811
8 60
y = 43.326x
R² = 0.9834
6 40
LVDT1
4 LVDT2
LVDT3 20
2 Parabolic LVDT4 Linear
LVDT5
0 0
0 2 4 6 8 10 0.1 0.6 1.1 1.6 2.1 2.6
Displacement (mm) Displacement (mm)

(kN / m) 2
K SGC AVG  41.5
mm

 n = 3 psi
Source: Zornberg et al. (2017)

Experimental Setup (2nd Generation):
Small Soil‐geosynthetic Interaction Device
• The proposed parameter was obtained using a
Large Soil-geosynthetic Interaction Device

Yet,
• A new apparatus was developed: Small Soil-
geosynthetic Interaction Device

Benefits of this comparatively smaller box include:

o Significantly smaller soil volume (0.0113 m3
vs. 0.27 m3)
o Smaller geosynthetic specimens
o Use of conventional loading frame
o Use of conventional grips for tensile testing
o Expeditious procedure
Small Soil‐geosynthetic Interaction Device
Connection to 
Load cell
compressed air supply
Universal joint

Data acquisition 
Specimen  system
(geosynthetic)

Soil‐geosynthetic 
interaction device

LPs Load frame 
Normal pressure
control system
(compressed air)

Small Soil‐geosynthetic 
Interaction Device
Frontal Unit Tension

12”

1
5 LPs

1 3
2
3 9.8” 4
4
5
5

LP1 LP2 LP3 LP4 LP5

Draft Test Procedure

Results from GG Testing Program
35 32.7

30

25 24.0 23.9
(kN/m)2/mm

20 18.8
17.2 16.9
15.6
14.5
15 13.7
12.2
KSGC

9.6 9.6
10

4.7
5

0.0
0
GG 10

GG 11

GG 12

GG 13

No GS
GG 2

GG 9

GG 7

GG 6

GG 5

GG 1

GG 3

GG 4

Sheet
Mylar

Source:
Roodi et al. (2018)
A New Property
A single, yet relevant parameter that quantifies 
the “Confined Stiffness of the Soil‐
Geosynthetic Composite under Small 
Displacements” was identified, and validated 
experimentally using a practical experimental 
approach.

“Essentially, all models are wrong,

but some are useful”
George E. P. Box

Validation against Field 

Performance
Consistency between Experimental and 
Field Results
FM2
GG1:   KSGC = 13 
GG5:   KSGC = 11 
GT2:   KSGC = 10 
Control: 

Consistency between Experimental and 
Field Results
SH21
GG1:   KSGC = 13 
GG7:   KSGC = 24 
GG10:   KSGC = 32 
Control: 
Consistency between Experimental and 
Field Results
FM1644
GG6:   KSGC = 14 
Control: 
Control GG6

Consistency between Experimental and 
Field Results
Cabeza Road
GG7:   KSGC = 19 
GG9:   KSGC = 24 
Control: 

GG7 Control GG9

Consistency between Experimental and 
Field Results
Geogrid Section 1 (Product A):
No longitudinal cracks
FM1774
Geogrid Section 2 (Product B):
Longitudinal cracks

GG1:   KSGC = 13 

GG4: KSGC =  8  Source: Zornberg and Gupta (2009)

Consistency between Experimental and 
Field Results
Geogrid Section 1: No longitudinal
cracks
Control Section: Longitudinal cracks FM1915
Geogrid Section 2: No longitudinal
cracks

GG1:         KSGC = 13 
Control:  Source: Zornberg and Gupta (2009)
Results from GG Testing Program

Type III

Type II

Pass  Type I

Threshold
KSGC = 10
Fail 

Source:
Roodi et al. (2018)

Recommended GG Specification
https://library.ctr.utexas.edu/ctr-publications/5-4829-03-1.pdf

US82 in Paris, TX
Project Includes:   Widening non Freeway, Grading, Paving, 
Adding two lanes
Geosynthetic‐Stabilized Base with 6” 
Cement‐treated Base
Total Project Length:   19.8 miles
Rehabilitation started in: 2019

US82
Reconstruction

19.8 miles
US82 in Paris, TX

Geogrid Type III

IH10 in San Antonio, TX
Geosynthetic

ASPHALT 8.5
”
8.5 ASPHALT
”

BASE 13” BASE

11”

6” 6”
SUBBASE SUBBASE

SUBGRADE SUBGRADE

Original Design Reduced Base Course 88
IH10 in San Antonio, TX

Test Sections on EBFR  Test Sections on EBFR 
Location 1 (1,000’) Location 2 (2,000’) in 3 Parts

Conclusions
Conclusions
• The use of geosynthetics was found to effectively minimize the
detrimental effects of expansive soil subgrades on flexible
pavements
• Geosynthetic-stabilized pavement sections on expansive clay
subgrades showed significantly better field performance than
control (non-reinforced) sections
• Compared to unreinforced sections, lime treatment was found not
to minimize the development of longitudinal cracks
• The Confined Stiffness of the Soil-geosynthetic Composite under
Small Displacements was identified as a relevant property
• An analytical framework was developed: Simple, realistic, and practical. It
involves a single parameter: The Stiffness of the Soil-geosynthetic
Composite (KSGC)
• The relative values of KSGC were found to be consistent with the
relative field performance of pavement sections subjected to
environmental loads

Final Remarks
• Overall, geosynthetics have been
demonstrated to improve, often significantly,
the performance of roadways
• The state of practice is rapidly improving as
new research is identifying the properties
governing the effect of geosynthetics in
roadway stabilization
Thank You
Jorge G. Zornberg, Ph.D., P.E., F. ASCE
[email protected]