Mechanical Properties of Polyurethane-Stabilized Ballast

Abstract

Ever increasing volume, tonnage, and speeds on rail systems are stressing rail substructure to levels never before evaluated or considered in depth. To improve maintenance techniques for problematic railway elements (e.g., bolted rail joints, intersections, bridge approaches), an in situ method involving ballast layer reinforcement with polyurethane is proposed. Ballast is a crucial material for structural support of the rail tracks. The structural integrity of highly fouled ballast (i.e., containing fine particles) can be compromised leading to track instability and ultimately train derailments. An application using polyurethane void filling and particle bonding technology has been developed and has the potential to mitigate impacts of ballast fouling, enhance rail freight capacity, and improve track-substructure maintenance efficiency. The purpose of this paper is to present the mechanical properties of Polyurethane-Stabilized Ballast (PSB) (e.g., compressive and flexural response), compare these properties to other materials commonly used in transportation infrastructure (e.g., natural aggregates, cement-stabilized soil), and address the suitability and compliance of PSB for use in track infrastructure. PSB has mechanical properties similar to cement-stabilized soil (i.e., displays flexural strength), but has much greater compressive strength than ballast, which is critical for stabilization of track substructure. Ease of injection and the negligible curing period for PSB makes it an attractive option for railway maintenance, especially for time-sensitive maintenance activities, such as intersections and bridge approaches.

Full text

Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 45 No.1 March 2014 ISSN 0046-5828
67
Mechanical Properties of Polyurethane-Stabilized Ballast
A. Keene1, J.M. Tinjum2 and T.B. Edil3
1Graduate Research Assistant, Dept. of Civil, Architectural, and Environmental Engineering,
University of Texas at Austin, USA
2Assistant Professor, Dept. of Engineering Professional Development,
University of Wisconsin-Madison, Madison, USA
3Professor Emeritus and Director, Recycled Materials Resource Center and Wisconsin Highway Research Program,
University of Wisconsin-Madison, Madison, USA
1E-mail: akkeene@utexas.edu
2E-mail: jmtinjum@wisc.edu
3E-mail: tbedil@wisc.edu
ABSTRACT: Ever increasing volume, tonnage, and speeds on rail systems are stressing rail substructure to levels never before evaluated or
considered in depth. To improve maintenance techniques for problematic railway elements (e.g., bolted rail joints, intersections, bridge
approaches), an in situ method involving ballast layer reinforcement with polyurethane is proposed. Ballast is a crucial material for structural
support of the rail tracks. The structural integrity of highly fouled ballast (i.e., containing fine particles) can be compromised leading to track
instability and ultimately train derailments. An application using polyurethane void filling and particle bonding technology has been
developed and has the potential to mitigate impacts of ballast fouling, enhance rail freight capacity, and improve track-substructure
maintenance efficiency. The purpose of this paper is to present the mechanical properties of Polyurethane-Stabilized Ballast (PSB) (e.g.,
compressive and flexural response), compare these properties to other materials commonly used in transportation infrastructure (e.g., natural
aggregates, cement-stabilized soil), and address the suitability and compliance of PSB for use in track infrastructure. PSB has mechanical
properties similar to cement-stabilized soil (i.e., displays flexural strength), but has much greater compressive strength than ballast, which is
critical for stabilization of track substructure. Ease of injection and the negligible curing period for PSB makes it an attractive option for
railway maintenance, especially for time-sensitive maintenance activities, such as intersections and bridge approaches.
1. INTRODUCTION
Rigid-polyurethane foam (RPF) applied to granular materials, after
injection and solidification, improves the strength by occupying the
pore space and cementing the particles together. Due to the
expansive properties of the foam, it has historically been applied in
foundation engineering to support footings and slabs. Due to these
advantages, there have been efforts to expand the applicability of
RPF to other infrastructure settings, including the rail industry. The
necessity for reinforcing railway substructure with strategic RPF
injections include: 1) reducing particle breakage and fines intrusion,
thus mitigating fouling generation, 2) correcting already fouled
ballast substructure and arresting permanent deformation in the
track, 3) improving substructure performance and preserving track
geometry thereby enhancing rail-freight capacity and rider-comfort,
and 4) providing a cost- and time-effective maintenance tool to
supplement rail maintenance capabilities.
For a railway embankment, the superstructure (i.e., rails, ties,
and fastening system) serves as a rigid structure that distributes the
loads over a large surface area to the substructure (i.e., ballast,
subballast, and subgrade) (Huang 2004). The superstructure
typically has much longer lifecycle than the substructure; however,
the superstructure lifecycle is dependent upon substructure
conditions and substructure maintenance intervals (Ebrahimi et al.
2012). Therefore, when considering polyurethane-stabilized ballast
(PSB) for use in constructing a stabilized substructure, the
mechanical behavior relative to other materials used in
transportation infrastructure needs to be evaluated.
Polyurethane interacts differently with the injected medium
depending on the nature of the medium. For instance, Buzzi et al.
(2012) injected RPF into expansive clay and found that the injection
created hydrofractures while forming into dendritic paths of foam.
Keene (2012) injected RPF into rail ballast (which has a more
favorable void structure for RPF injection) and formed a solid,
uniform geocomposite. The pore space in compacted ballast
conveniently allows injection of polyurethane, allowing space for
RPF expansion, and for target RPF volumes and densities to be met.
Keene (2012) proposed a set of criteria by which the mechanical
properties of the ballast layer were improved with polyurethane
injections. These criteria include: 1) extent to which the void space
of ballast was filled by the RPF, 2) strength and degree of bonding
that occurred between the ballast particles and RPF, and 3) limiting
volumetric expansion of ballast during RPF injection. Many
methods are available for the mechanical analysis of polymeric
cellular foams or for engineering properties of granular materials;
however, little is understood about the behavior of the combination
of an expanding polymer with ballast and effects on the mechanical
properties within track-substructure.
The objective during field injection of RPF into the ballast layer
is to strengthen the areas that transmit load down into the subballast
and subgrade layers. Moisture is prevented from infiltrating the
stabilized areas while drainage of surrounding untreated ballast is
left uninhibited. Ballast layer prototypes (i.e., boxes filled with
0.45-m-deep compacted ballast) were also created in Keene (2012),
where RPF was strategically injected into an unconfined layer of
ballast. In that physical model, methods for targeting the dimensions
of the stabilized areas were developed.
In this paper, specimen fabrication and experimental methods
for compressive and flexural testing and analysis of polyurethane-
stabilized ballast (PSB) are presented. Moduli and strength of each
of the constituents of PSB (i.e., ballast and RPF) are compared to
each other and to the PSB composite material. The mechanical
properties of PSB and PSB constituents (RFP and ballast) are
compared to other materials such as cement-stabilized materials
(CSM), natural base-course aggregates, and concrete for a broader
perspective. Compressive and flexural strength of PSB and RPF are
compared to CSM (at different cement-binder contents) to show the
similarity in relative strengths of these materials. Resilient modulus
of PSB, ballast, MN DOT Class 5 aggregate, and CSM are
compared to show the elastic behavior of the materials under cyclic
loading conditions. Flexural strength of CSM and PSB are compared
to demonstrate increase in strength with increase in binder content
(i.e., percent cement and percent RPF by weight). The strength-to-
bulk-density ratio (σ/ρ) of PSB, RPF, ballast, CSM, and concrete are
compared to show how each material possesses σ/ρ properties that
can be favorable depending on the application. The purpose of this

stu
d
p
r
o
tra
n
for
2.
Ba
l
qu
a
ra
n
we
i
et
a
co
r
we
r
ch
a
cle
a
us
e
W
a
ad
d
an
d
wi
t
tw
o
res
i
Ba
y
dif
f
sy
n
co
m
p
o
l
cat
a
b
u
b
kn
o
the
the
p
r
o
clo
mo
M
a
a c
l
b
al
cu
r
the
ge
o
sta
b
de
n
fur
t
co
n
wi
t
du
r
to
r
for
m
mi
n
co
n
for
m
p
r
e
tra
c
ap
p
3.
3.1
A
m
sta
n
str
e
fle
x
an
d
ad
o
d
y was to mea
s
o
pe
r
ties to pr
o
n
sportation inf
r
use in track in
f
MATERI
A
l
last (Figure 1)
a
rry near Che
y
n
ged between 2
5
i
ght was obtai
n
a
l. (2012), res
u
r
responding cl
e
r
e 15.8 kN/m
3
a
racteristics w
e
a
n ballast injec
t
For character
i
e
d are fouling
i
a
ters 1994, Ebr
a
d
ing percent o
f
d
percent of p
a
t
h FI between 2
The 486STA
R
o
-component,
h
i
n system. Th
e
y
er Material S
f
erent applica
t
n
thesis of t
h
m
ponents (p
o
l
yisocyanate)
a
a
lyst (Szycher
,
b
ble formation
o
wn as blowin
g
blowing agen
t
cellular struc
t
o
viding RPF st
r
se
d
-cell struct
u
dulus of RPF.
a
terial Science
(
l
ose
d
-cell cont
e
During injecti
o
last pore spac
e
r
ing phases, bo
n
reacting RP
F
o
composite is
b
ilized ballast
(
n
sities within
t
t
her details r
n
stituent densit
i
t
h the ballast
p
r
ing the polyur
e
r
ough surfaces
m
ed during
t
n
eralogy (Kee
n
n
trol strength
m
ation is an i
n
e
mixing with a
g
c
k shutdown,
a
p
lication.
METHOD
Flexural L
m
ethod for te
s
n
dards for tes
t
e
ngth and fati
g
x
ural strength
w
d
the method u
s
o
pted from an
A
Geo
t
s
ure the mecha
n
o
perties of ot
h
r
astructure, an
d
f
rastructure.
A
LS
was provided
b
y
enne, Wyom
i
5
and 63 mm (
n
ed following t
h
u
lting in a clea
n
e
an ballast dry
and 1,611 kg/
m
e
re targeted i
n
t
ed with RPF i
n
i
zing fouled b
a
i
ndex (FI) and
a
himi et al. 20
1
f
particles pas
s
a
rticles passing
0% and 39% i
s
R
-4 BD, a RPF
h
igh-density, e
x
e
486STAR-4
cience in part
n
t
ions includin
g
h
ermoset pol
y
o
lyester or
a
re proportion
a
,
M. 1999). T
h
during the po
g
. Gas bubble f
o
t
(Szycher, M.
1
t
ure of the R
P
r
ength and mo
d
u
re (ASTM D
6
In the techni
(
2010), the 48
6
e
nt of 90%.
o
n, RPF flows
e
. While the R
P
n
ds are establi
s
F
. When the
formed and
(
PSB -Figure 1
t
he PSB com
p
egarding pha
s
i
es are given in
p
articles is a
c
e
thane foamin
g
of the ballast
t
he polyureth
a
n
e 2012), whi
c
in asphalt a
n
n
situ stabiliz
a
g
gregates, soil
,
a
nd reaches 90
%
S
oading of PS
B
s
ting PSB bea
m
t
ing concrete
a
g
ue analysis.
T
w
as adopted fr
o
s
ed for ascertai
n
A
USROADS
p
t
echnical
E
ngine
n
ical propertie
s
h
er materials
d
to evaluate t
h
b
y BNSF Rail
w
i
ng. The parti
c
ASTM D6913
)
h
e procedure d
e
n
ballast void r
a
unit weight
(
m
3
, respectivel
y
n
fabrication o
n
this study.
a
llast, two co
n
moisture cont
e
1
2). The foulin
g
s
ing through a
a 0.075-mm
(
s
considered hi
g
supplied by
U
x
panding, ther
m
BD (Figure 1
)
n
ership with
U
g
void fillin
g
y
urethane-resin
polyether po
a
tely mixed i
n
h
e foam struct
u
lyurethane pol
y
o
rmation is the
1
999). As deta
i
P
F is an impo
r
d
ulus. The gre
a
6
226) the gre
a
cal data sheet
6
STAR-4 BD i
s
as a liquid and
P
F transitions
t
s
hed with mat
e
reaction is c
o
referred here
i
). The average
p
osite specime
n
s
e calculation
s
Keene (2012).
c
ritical interac
t
g
process. The
b
particles and
i
a
ne reaction
w
c
h are commo
n
n
d concrete.
A
a
tion method t
h
,
or with wate
r
%
of full stren
g
B
Beams
m
specimens
w
a
nd soil-cemen
t
T
he method
u
o
m ASTM C7
8
n
ing flexural f
a
p
rocedure desc
r
ering Journal of
s
of PSB, relat
e
commonly u
s
h
e suitability o
f
w
ay Company
f
c
le size distri
b
)
. Maximum d
r
e
veloped in Eb
r
a
tio, e
b
, of 0.6
2
(
γ
d
) and densit
y
y
. These comp
f each speci
m
n
ventions com
m
e
nt (MC) (Sel
i
g
index compr
i
4.75-mm (P4)
(
P200) sieve.
B
g
hly fouled.
U
retek USA In
c
m
oset, polyure
t
)
was formula
t
U
retek USA In
c
g
and sealin
g
foams, the
lyol and o
r
n
the presence
u
re results fro
m
y
merization p
r
result of intro
d
i
led in Keene (
2
r
tant compone
a
ter the extent
a
ter the streng
t
produced by
s
indicated to p
o
expands throu
g
t
hrough the p
o
e
rials in contac
o
mpleted, a b
i
n as polyure
t
resulting RPF
n
s were 200
k
s
and typical
The bonding
o
t
ion that takes
b
onding is attr
i
i
ntermolecular
w
ith the agg
r
n
characteristi
c
A
pplication o
f
h
at does not r
e
r
, would not r
e
g
th within 15
m
w
as develope
d
t
beams for fl
e
u
sed for deter
m
8
and ASTM
D
a
tigue properti
e
r
ibed in Midgl
e
the SEAG
S
& A
G
e
these
s
ed in
f
PSB
f
rom a
b
ution
r
y unit
r
ahimi
2
. The
y
(ρ
d
)
action
m
en of
m
only
i
g and
i
ses of
sieve
B
allast
c
., is a
t
hane-
t
ed by
c
., for
g
. For
two
r
ganic
of a
m
gas
r
ocess,
d
ucing
2
012),
nt for
of the
t
h and
Bayer
o
ssess
g
h the
o
lymer
t with
onded
t
hane-
phase
k
g/m
3
,
PSB
o
f RPF
place
i
buted
bonds
r
egate
c
s that
f
PSB
e
quire
e
quire
m
in of
d
from
e
xural
m
ining
D
1635
e
s was
e
y and
Yeo
flex
u
trans
p
occu
r
setu
p
redu
c
impr
o
dime
D16
3
p
arti
c
mol
d
spec
i
data
a
Figu
r
and
P
I
n
distr
i
uppe
r
oute
r
stren
calc
u
w
rupt
u
setu
p
dime
whic
h
T
flex
u
corr
e
and
Y
strai
n
as



whe
r
(kPa
)
widt
h
G
SSEA Vol. 45 N
(2008). The t
w
u
ral strength a
n
p
ortation infr
a
r
s. From these
p
p
was selecte
d
c
ing the effects
o
ved analysis
o
nsions of the
3
5) were incre
a
c
le sizes in b
a
d
s were 200 m
m
i
men fabricatio
n
a
nalysis are gi
v
r
e 1 Pictures o
f
P
SB specimen
c
n
the flexural
b
i
butes the load
r
-inner “third-
p
r
“thir
d
-
p
oints
”
g
th (kPa), ot
h
u
lated as

·
·

w
here P is the
u
re, L is the sp
a
p
, w is the base
nsion of the b
e
h
are shown in
T
he flexural mo
u
ral stress an
d
e
sponding defl
e
Y
eo (2008) an
d
n
, and flexural






·10



r
e S
max
is the
fl
)
, ε
t
is the flex
u
h
of the beam (
m
o.1 March 2014
I
w
o methods ar
e
n
d fatigue char
a
a
structure bec
a
p
rocedures and
d
. A “thir
d
-
po
of shear stress
o
f flexural str
e
beams from
p
a
sed by a ratio
a
llast. The di
m
m
x 200 mm x
n
, flexural test
i
v
en in Keene (
2
f
granitic balla
s
c
ut in half with
b
eam tests, th
e
evenly throu
g
p
oints” of the
b
”
of the beam
h
erwise kno
w
peak load (k
N
a
n length (m) b
e
width (m) of t
h
e
am between t
h
Figure 2.
dulus is derive
d
d
strain are i
n
e
ction at the
m
d
this study, th
e
stress were ca
l
·
·
·


··
·
·



·
fl
exural modul
u
u
ral strain (m/m
m
), h is the hei
I
SSN 0046-5828
e
commonly us
a
cteristics of
m
a
use of the c
y
d
standards, a “t
h
o
int” loading
s
during flexur
a
e
ngth and fati
g
p
revious stan
d
of 2.63:1 to a
c
m
ensions used
f
763 mm. Furt
h
i
ng procedures
,
2
012).
s
t (top-left), R
P
concrete maso
n
e
load is appli
e
g
h two loading
b
eam. Rollers
support the
b
w
n as rupture
N
) during the
t
e
tween the bot
t
h
e beam, and h
h
e top and bott
o
d
using elastic
b
n
ferred from
a
m
i
d
-span of the
e
beam flexura
l
l
culated using
e
··

·

·
u
s (MPa), σ
t
is
), P is the peak
ght of the bea
m
ed for determi
n
m
aterials for u
s
y
clic loading
h
ir
d
-
p
oint” loa
d
s
etup is ideal
a
l testing and f
o
g
ue properties.
d
ards (e.g., A
S
c
count for the l
a
f
or the PSB b
h
er detail on b
,
methodology,
P
F foam (top-ri
g
n
ry saw (botto
m
e
d to a fixture
rollers at the
on the two lo
w
b
eam. The fle
x
modulus (R)
(1)
t
est or load b
e
t
om supports o
f
is the depth (
m
o
m supports, a
l
b
eam theory w
h
a
pplied loads
beam. In Mid
g
l
modulus, fle
x
e
lastic beam th
e
(2)
the flexural s
t
load (kN), w i
s
m
(m), L is the
s
68
n
ing
s
e in
that
d
ing
for
o
r an
The
S
TM
a
rge
eam
eam
and
g
ht),
m
)
that
two
w
e
r
-
x
ural
, is
e
fore
f
the
m
) or
l
l of
h
ere
and
g
ley
x
ural
e
ory
t
ress
s
the
s
pan

(m
)
an
d
loa
Fi
g
3.2
Pri
s
for
p
ri
s
p
ri
s
loa
loa
U
C
in
K
Un
wh
wh
p
h
e
(ε
e
)
wh
lo
n
Yo
u
b
y
wh
ra
n
)
between bott
o
d
δ is the defle
c
d.
g
ure 1 Typical
Unconfine
d
Prisms
s
matic specim
e
unconfine
d
-c
o
s
m dimension
s
s
ms were subj
e
ding machine
d and displace
m
C
S testing proc
e
K
eene (2012).
confined comp
r





ere F is peak a
ich it is appl
i
e
nomenon in e
l
)
increases line
a
ere plastic stra
i
n
ger linear.
u
ng’s modulus
 



ere σ is the a
x
n
ge of the mate
r
Geo
t
o
m two suppor
t
c
tion (m) at th
e
“thir
d
-
p
oint” l
o
for P
S
d
Compressiv
e
e
ns with a 2:1
(
o
mpression str
e
s
were 200 m
m
e
cted to UCS t
e
that compress
e
m
ent. Further d
e
dures, method
o
r
essive strengt
h
xial load (kN)
i
ed, both are
d
l
astic-
p
lastic
m
a
rly with axial
i
n (ε
e
) increase
s
, E, occurs wit
h
x
ial stress (kP
a
r
ial.
t
echnical
E
ngine
t
s of the “thir
d
-
p
e
mi
d
-span un
d
o
ading setup a
n
S
B beams
e
Strength (U
C
(
height-to-widt
h
e
ngth (UCS) t
e
m
x 200 mm
x
e
sts, which inv
o
e
d the specim
e
etail on prism
s
o
logy, and dat
a
h
(σ
c
) was dete
r
applied and A
d
iagramed in
m
aterial is that
i
stress until rea
c
s
and the relati
o
h
in the linear
r
a
) being appli
e
ering Journal of
p
oint” loading
d
er the corresp
o
n
d chosen dime
n
C
S) Testing o
f
h
) ratio were c
r
e
sting. Nomina
l
x
400 mm. Th
e
o
lved placeme
n
e
ns while mea
s
s
pecimen fabri
c
a
analysis are c
o
r
mined by
(3)
is the area (m
2
Figure 3. A t
y
i
nitially elastic
c
hing the yield
o
nship to stres
s
r
egion and is d
e
4
e
d within the
e
the SEAG
S
& A
G
setup,
o
nding
n
sions
f
PSB
r
eated
l
PSB
e
PSB
n
t in a
s
u
r
ing
c
ation,
o
vered
) over
y
pical
strain
point,
s
is no
e
fined
e
lastic
3.3
Cyli
n
p
rov
i
simi
l
triax
i
teste
d
with
cyli
n
p
roc
e
Figu
r
T
cons
t
with
i
p
isto
n
appl
i
have
r
(Ebr
a
duri
n

whe
r
(A)
o
cycl
e
mea
s
foun
d
each

whe
r
The
r

whe
r
strai
n
T
the s
t
of th
stiff
n
cons
t
p
rov
i
Addi
load
i
G
SSEA Vol. 45 N
Cyclic Tria
x
n
drical speci
m
i
des an approp
r
ar to the one u
s
i
al testing of
d
in a triaxial
c
the data give
n
n
drical specime
n
e
dures, and dat
a
r
e 2 Picture of
of co
m
T
he cyclic tri
a
t
ant confining
i
n a membrane
n
that extends
t
es the cyclic
r
sine, bell-sha
p
a
himi et al. 20
n
g the rest (17.
6







r
e the load (F)
f
o
f the plate o
n
e
, non-recover
a
s
ured and reco
v
d
by subtractin
g
load pulse, an
d






r
e L is specime
n
r
esilient modul
u









r
e σ
P
is
p
eak st
r
n
(m/m).
T
he resilient m
o
t
iffness of a m
a
e resilient mo
d
n
ess changes o
t
ant value afte
r
id
es the ope
r
tionally, accu
m
ng cycles are a
l
o.1 March 2014
I
x
ial Compressi
o
m
ens with a
m
r
iate particle di
a
s
ed in the stud
y
railway ballas
t
c
ell and the co
r
n
in Ebrahimi
n
fabrication,
c
a
analysis are g
compression t
e
m
pression testi
n
a
xial compress
i
pressure (σ
3
)
and sealed in
t
hrough a seal
load. The cy
p
e
d
loading
p
12). The devi
a
6
kPa)
p
eriod is
f
rom the pisto
n
n
top of the s
p
a
ble deformat
i
v
erable defor
m
g
δ
P
from the
m
d
the elastic st
r
a
n
length.
u
s is denoted a
s
r
ess (kPa), σ
R
i
s
o
dulus is cycli
c
a
terial. In the c
d
ulus over ma
n
ver the life c
y
r
numerous lo
a
r
ational resili
m
ulated plasti
l
so recorded in
I
SSN 0046-5828
o
n Testing of
P
m
inimum dia
m
a
meter to spec
i
y
by Anderson
t
. Clean balla
s
r
responding
r
es
u
et al. (2012).
c
yclic-triaxial
c
g
iven in Keene
(
e
sting apparatu
s
n
g parameters (
r
i
on test consi
to a specime
n
a triaxial cha
m
in the top plat
e
clic load is
a
p
ulse with pe
a
a
tor stress at p
e
given by
n
(kN) is appli
e
p
ecimen (m
2
).
A
i
on (plastic d
e
m
ation (elastic
d
m
easured total
d
a
in ε
E
is calcul
a
s
M
R
(kPa) and
s
rest stress (k
P
c
Young’s mod
u
yclic triaxial t
e
n
y loading cycl
y
cle of the ma
t
ading cycles.
I
ent propertie
s
c strains (ε
p
)
this test.
P
SB Cylinders
m
eter of 254
i
men diameter
r
and Fair (200
8
s
t specimens
w
u
lts were vali
d
Further detail
s
c
ompression te
s
(
2012)
s
(left) and dia
g
r
ight)
sts of applyi
n
n
that is conta
i
m
ber. A plung
e
e
of the triaxial
a
pplied as a
5
a
k and rest l
o
e
ak (300 kPa)
(5)
e
d through the
A
fter each loa
d
e
formation, δ
P
d
eformation, δ
E
d
eformation (δ
T
a
ted as
(6)
is calculated f
r
7
P
a), and ε
E
is el
a
u
lus and quant
i
e
sts, the calcul
a
es reveals ho
w
t
erial or reach
e
I
n either case,
s
of a mat
e
with numbe
r
69
mm
r
atio
8
) on
w
ere
d
ated
s
on
s
ting
g
ram
n
g a
i
ned
e
r or
cell
5
-Hz
o
ads
and
area
d
ing
P
) is
E
) is
T
) in
r
om
a
stic
i
fies
a
tion
w
the
e
s a
M
R
e
rial.
r
of

Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 45 No.1 March 2014 ISSN 0046-5828
70
Figure 4 Triaxial chamber used for testing specimens with nominal
dimensions of 254-mm diameter x 506-mm height (left) and PSB
cylindrical specimen (right)
4. RESULTS
4.1 Mechanical Properties of PSB and Infrastructure
Materials
4.1.1 Mechanical Properties of PSB Constituents
RPF strengths are similar in each mode of testing, as seen in
Figure 5. When comparing flexural test results, when the average
(AVG) RPF density, ρRPF, is 200 kg/m3, the AVG PSB flexural
modulus (274 MPa) is greater than the AVG RPF flexural modulus
(124 MPa); however, the AVG PSB flexural strength of 938 kPa is
less than the AVG RPF flexural strength 3,652 kPa as shown in
Figure 5. Greater flexural stiffness of PSB compared to RPF can be
attributed to the stiffness of the ballast particles. The lower flexural
strength of PSB relative to RPF can be attributed to weakness in the
bonding interface between the ballast particles and RPF. As
described in Akçaoğlu et al. (2003), the surface texture and the
bonding area between cement binder and aggregates are critical to
concrete strength and stiffness. Akçaoğlu et al. (2003) defined this
bonding area as the interfacial transition zone (ITZ) and described
this zone as the weakest component in concrete mechanical
behavior. When focusing on an ITZ for PSB, the strength of the
composite can be attributed to two likely factors: 1) RPF-ballast
bond interface strength and 2) PSB composite/matrix strength based
on cell orientation/geometry within the ballast pore space and
around ballast particles. The flexural strength and tensile strength of
RPF are greater than the flexural strength of PSB; consequently, the
ITZ for PSB likely controls flexural strength. The cellular structure
of RPF (as characterized by closed-cell content, cell-wall thickness,
cell elongation, and cell aperture) likely plays a large role in RPF
strength and elastic modulus.
In Salim (2004), the compressive strength of ballast particles
was used for determining the characteristic tensile strength of the
ballast. Salim (2004) cites Jaeger (1967) for explaining that the
fracture of rock grains occurs due to tensile failure and that fracture
strength in tension can be measured indirectly through compression
tests conducted on rock particles using “diametral compression
between flat platens.” Trends establishing RPF tensile strength
versus density (Keene 2012) indicate that at a characteristic RPF
density of 200 kg/m3, the corresponding tensile strength is 3,912
kPa, which is far less than the characteristic particle tensile strength
(5,400 kPa to 22,300 kPa) of the granite ballast used in this study as
reported by Ebrahimi et al. (2012). Since the characteristic tensile
strength of ballast is higher than the tensile strength of RPF, RPF
may govern the rupture strength of PSB in monotonic flexural
loading tests. However, failure likely occurs at the ITZ since the
AVG PSB flexural microstrain (με) at rupture (8.94 με) is less than
RPF (ρRPF = 200 kg/m3) flexural microstrain at rupture (28.7 με).
Characteristic tensile strength of ballast particles is the only
instance where ballast would contribute to the overall strength of
PSB instead of RPF. Higher tensile strength of ballast particles (i.e.,
higher stiffness) must also contribute to the flexural stiffness of PSB
being higher than the compressive stiffness of PSB. However, it is
likely that in fatigue testing conducted in Keene (2012), fatigue of
ballast particles contributed to fatigue failure since fracture of
ballast particles was observed after the fatigue testing. Therefore,
ballast particles may fatigue under flexural/tensile loading before
RPF fatigue occurs.
Figure 5 Shown are the mechanical strengths (a) and moduli (b)
for PSB and PSB constituent materials. Representative mechanical
properties of RPF are at a 200 kg/m3-density. Representative ballast
compressive modulus and strength are at a 100-kPa confining stress
(Ebrahimi et al. 2012). Error bars indicate maximum and minimum
mechanical property values (i.e., range) for materials with varying
confining stresses (ballast) or densities (RPF and PSB).
0
1,000
2,000
3,000
4,000
5,000
6,000
7,000
Compressive
Flexural
Tensile
Compressive
Flexural
Tensile
Compressive
Flexural
Tensile
Mechanical Strengths (kPa)
a)
0
50
100
150
200
250
300
350
400
450
500
Compressive
Flexural
Tensile
Compressive
Flexural
Tensile
Compressive
Flexural
Tensile
Elastic Moduli (MPa)
b)

Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 45 No.1 March 2014 ISSN 0046-5828
71
Similar to the PSB constituent flexural strengths, the RPF
(ρRPF = 200 kg/m3) compressive strength (3,752 kPa) is higher than
the AVG PSB compressive strength (2,607 kPa). Ballast
compressive strength at 100 kPa confining pressure is 594 kPa,
which is 77% less than PSB. A limitation in PSB compressive
strength relative to RPF compressive strength is likely due to limits
in bonding strength (i.e., weakness of the ITZ), as was identified for
PSB flexural strength. However, an increase in PSB compressive
strength relative to ballast compressive strength is attributed to both
the predominant strength of RPF and the high characteristic ballast
tensile strength. Similar interactions take place with aggregates and
asphalt binder resulting in the superior behavior of asphalt mixes
(Tia 2003).
Marginal differences were observed between PSB compressive
modulus (95 MPa) in monotonic loading tests and resilient modulus
(100 MPa) in cyclic triaxial tests. Therefore, monotonic testing on
PSB can be a useful alternative for predicting PSB resilient modulus
generally determined under cyclic compressive loading. Regarding
deformational behavior of PSB, minimal accumulation of plastic
strain, εp, was observed over 200,000 loading repetitions at a
representative state of stress (Ebrahimi et al. 2012) in PSB cylinders
in cyclic triaxial testing. Specimens tested up to 500,000 loading
repetitions had a marginal increase in plastic strain. Over the first
200,000 loading repetitions, PSB plastic strain (εp = 0.22%) was far
less than clean ballast (εp = 0.96%) or fouled ballast (εp = 3%) with a
fouling index of 5% and moisture content of 15%. Cumulative
plastic strain is the main limiting performance parameter for clean or
fouled ballast. However, the cumulative plastic strain under cyclic
loading conditions in PSB specimens was significantly reduced
making PSB elastic properties a more important performance
parameter for design of PSB in rail substructure. Since PSB and
RPF compressive strengths are far greater than the clean ballast
compressive strength (at the representative confining stress), the
functionality of PSB in rail infrastructure would likely be driven by
PSB compressive modulus.
4.1.2 PSB Mechanical Properties Compared to Bounded and
Unbounded Aggregates
When comparing the compressive strength of PSB to cement-
stabilized materials (CSM), PSB has 2.5 times less compressive
strength (see Figure 6). RPF with a 200-kg/m3 density has a
compressive strength of 3,752 kPa and ballast (tested at 100 kPa
confining pressure) has a compressive strength of 594 kPa;
therefore, both materials possess lower compressive strengths than
CSM. From RPF test results compiled from literature (Keene 2012),
RPF with a density ranging from 26 to 417 kg/m3 has compressive
strength ranging from 2,774 kPa to 6,167 kPa (Figure 6); therefore,
CSM still possesses higher strength than RPF formed at high
densities.
When comparing the flexural strength of PSB and RPF to CSM,
as shown in Figure 6, PSB has a flexural strength similar to that of
CSM flexural strength tested at 28-day curing time given by
Midgley and Yeo (2008). RPF (ρRPF = 200 kg/m3) possesses a much
higher flexural strength (3,652 kPa) than PSB and CSM. With RPF
density ranging from 26 to 417 kg/m3, the range of RPF flexural
strength is 2,774 kPa to 6,167 kPa, which is greater than both PSB
and CSM (Figure 6). In Midgley and Yeo (2008), the flexural
modulus increased as the relative density increased, similar to how
modulus of PSB increases as PSB density increases (Keene 2012).
Unlike the materials being compared to RPF, RPF has similar
strengths in each mode of load application (i.e., compressive,
flexural, and tensile) and, as indicated later in this paper, RPF has
superior strength-to-bulk-density ratio.
Since PSB and CSM have similar AVG flexural strength
properties, a comparison is also made between flexural strength and
the percentage of binder content. A study by Zhang and Wei (2011)
is used for comparison where the flexural strength of CSM (at 28-
day curing time) was marginally higher than CSM strength given in
Midgley and Yeo (2008) that was used in earlier comparisons. With
a range of binder content (percent cement) from 4 to 7%, the
flexural strength of CSM reported in Zhang and Wei (2011) ranged
from 1,150 kPa to 1,895 kPa, corresponding to a 39% increase in
flexural strength with 3% increase in binder content. Over the same
range of binder content in PSB (percent RPF by weight), PSB
flexural strength ranged from approximately 682 kPa to 1,290 kPa,
corresponding to a 28% increase. PSB and CSM flexural strength
versus binder contents are shown in Figure 7. Thus, an increase in
cement binder content is more effective in increasing the flexural
strength of CSM in comparison to an increase in RPF binder content
on the flexural strength increase in PSB. In addition, an increase in
volume of RPF in PSB is much higher than an increase in volume of
cement needed to obtain the same proportional increase in flexural
strength.
Figure 6 Comparison of RPF and CSM compressive strengths (a)
and comparison of RPF and CSM flexural strengths (b). Hornfels
and Siltstone data are from Midgley and Yeo (2008) and TNZ M4
are from Arnold (2009).
The flexural strength of RPF is greater compared with other
materials (e.g., CSM and PSB); however, CSM has greater
compressive strength than PSB or RPF at a density of 200 kg/m3. In
addition, CSM has far greater flexural modulus (AVG 13,800 MPa)
0
1,000
2,000
3,000
4,000
5,000
6,000
7,000
PSB RPF Ballast Hornfels 3
% CSM
Siltstone 4
% CSM
Compressive Strength (kPa)
a)
0
1,000
2,000
3,000
4,000
5,000
6,000
7,000
PSB RPF TNZ M4
2 %
CSM
Hornfels
3 %
CSM
Siltstone
4 %
CSM
TNZ M4
4 %
CSM
Flexural Strength (kPa)
b)

tha
n
CS
M
co
m
loa
co
n
b
e
dis
t
F
b
ag
g
lo
w
(2
0
str
e
ti
m
ti
m
kg
/
b
al
p
e
r
re
q
an
a
the
mo
F
i
b
m
o
Flexural Stren
g
th
(
kPa
)
n
PSB and RP
F
M
would perf
o
m
pliance (i.e.,
e
ding condition
s
n
struction. For
favorable due
t
t
ributed from t
h
F
igure 7 Comp
b
inder content (
When evalua
t
g
regate, and
C
w
est M
R,
while
0
09), has the hi
e
ss of 208 kPa
h
m
es greater tha
n
m
es greater tha
n
/
m
3
, M
R
range
d
last. As is the
r
form more fa
v
q
uired for desi
g
a
lysis determin
e
track was inc
o
dulus was incl
u
i
gure 8 Compa
r
b
ulk stress of 6
0
o
dulus of MN
D
CSM with
0
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
01
g( )
Geo
t
F
(274 and 12
4
o
rm more favo
r
e
lastic strain) i
s
s
, hence the ty
p
rail infrastruct
u
t
o the strains t
h
h
e superstructu
r
arison of PSB
a
percent RPF a
n
t
ing the M
R
o
f
C
SM (Figure 8
CSM with 2
%
ghest modulus
.
h
ad a M
R
that
w
n
PSB. The M
R
n
PSB. When
P
d
63–181 MPa
,
case for flexu
r
v
orably in ap
p
g
n under com
p
e
d that the effe
o
nsequential w
h
u
ded in the trac
r
ison of resilie
n
0
0 kPa (Ebrahi
m
D
OT Class 5 te
s
2% cement bi
n
23
Bi
n
Polyurethane
-
(Zhang and
W
t
echnical
E
ngine
4
MPa, respect
i
r
ably in applic
a
s
allowed unde
r
p
ical applicatio
n
u
re, higher co
m
h
at can be tole
r
r
e down throu
g
a
nd CSM flexu
r
n
d percent cere
m
f
PSB, ballast
,
); PSB (AV
G
%
cement bin
d
.
MN DOT Cl
a
w
as 18% less t
h
R
of ballast (27
5
P
SB density r
a
,
which is still
r
al properties o
p
lications wher
e
p
ressive loadi
n
ct on the over
a
h
en PSB with
k substructure
(
n
t modulus of
P
m
i et al. 2012),
s
ted at a bulk s
t
n
der tested by
A
456
n
der Content (
%
-
Stabilized Ball
W
ei 2011) Cem
e
ering Journal of
i
vely). Conseq
u
a
tions where m
i
r
operational fl
e
n
of CSM in ro
a
m
pliance of PS
B
r
ated under the
g
h the substruct
u
r
al strength ve
r
m
ent, respectiv
,
MN DOT C
l
G
100 MPa) h
a
d
er tested by
A
a
ss 5 tested at
a
h
an ballast and
o
5
MPa) was o
v
a
nged 1,536 to
less than the
f CSM, CSM
w
e
higher stiff
n
n
g. A finite el
a
ll elastic respo
n
a lower compr
(
Keene et al. 2
0
P
SB, clean ball
a
summary resil
i
t
ress of 208 kP
a
A
rnold (2009).
y = 455e0.1
R² = 0.95
789
%
)
ast
e
nt-Stabilized
S
the SEAG
S
& A
G
u
ently,
i
nimal
e
xural
a
dway
B
may
loads
u
re.
r
sus
ely)
l
ass 5
a
s the
A
rnold
a
bulk
o
ver 2
v
er 2.5
1,683
M
R
of
w
ould
n
ess is
ement
n
se of
essive
0
13).
a
st at
i
ent
a
, an
d
T
com
p
kPa/
k
the
d
has
a
kPa
c
also
com
p
cond
18.3
)
PSB
and
Y
0.53,
GPa
com
p
flex
u
CS
M
havi
n
Fi
sc
a
4%
c
5.
In t
h
b
alla
mate
r
com
p
less
t
with
is lo
w
aggr
e
defo
r
p
erh
a
PSB
foul
e
B
PSB
b
alla
laye
r
accu
m
muc
h
1x
10
S
oil
1
Strength-to-Bulk-Density Ratio (σ/ρ)
G
SSEA Vol. 45 N
T
he compress
i
p
ared using
k
g/m
3
) in com
p
d
ensity or weig
h
a
σ/ρ far great
e
c
onfining stres
s
c
ompared usin
g
p
ares to the
w
i
tions (Figure
9
)
has the highe
s
has a flexural
σ
Y
eo (2008) (
A
and concrete
compressive
p
aring flexural
u
re. As was se
e
M
, both materi
a
n
g a marginall
y
gure 9 Compa
r
a
le) of PSB co
n
c
ement siltston
e
so that ea
c
CONCLUSI
h
is study, the
m
st (PSB) are c
o
r
ials typicall
y
p
ressive streng
t
t
han CSM. The
PSB stiffness
b
w
er than clean
e
gate; however
r
mational resp
o
a
ps may be
b
e
n
plastic strains
e
d ballast.
B
ased on the
m
has at least t
w
st from the be
r
, the comp
r
m
ulation of p
l
h
longer life
0.1
1.0
10.0
1
00.0
Compressive
Fl l
o.1 March 2014
I
i
ve properties
average stre
n
p
ression to sh
o
h
t of the mate
r
e
r than that of
b
s
. The mechani
c
g
average flex
u
w
eight of the
m
9
). As was see
n
s
t flexural σ/ρ
σ
/ρ of 0.57, CS
M
A
VG bulk dens
(2,403-kg/m
3
strength has
σ/ρ, RPF has t
h
e
n with flexur
a
a
ls have a ver
y
y
lower σ/ρ tha
n
r
ison of the stre
n
n
stituents, PSB
,
e
from Midgle
y
c
h ratio can be
ONS
m
echanical pro
p
o
mpared to tha
t
y
used in tr
a
t
h of PSB is m
u
flexural stren
g
b
eing a little lo
w
ballast, CSM,
, contribution
o
o
nse is not c
n
eficial. Most
s
under cyclic l
m
echanical
p
ro
p
w
o potential ap
p
aring surface
o
r
essive stren
g
l
astic strain in
d
cycle than
u
Fl
exura
l
Compressive
Flexural
I
SSN 0046-5828
of PSB, RP
F
n
gth-to-bulk-de
o
w how the st
r
r
ials (Figure 9)
ballast (σ/ρ =
0
c
al properties
o
u
ral σ/ρ to sho
w
m
aterials und
e
n
for compressi
of the materia
l
M
with 4% ce
m
ity of 2,146 k
g
bulk density)
a flexural σ/
ρ
h
e same ratio i
n
a
l strength pro
p
y
similar flexu
r
n
PSB.
n
gth-to-bulk-d
e
,
typical 20.7
G
y
and Yeo (200
visualized in t
h
p
erties of pol
y
t
of its constit
u
a
nsportation i
n
u
ch greater tha
n
g
th of PSB is v
e
w
er. The resili
e
and typical hi
g
o
f the lower PS
onsidered to
b
s
ignificantly, t
h
oading is far l
p
erties of PS
B
p
lications. Fir
s
o
f the tie dow
n
g
th results a
n
d
icate that the
u
ntreated balla
Compressive
Flexural
Compressive
F
, and ballast
nsity ratio
(
r
ength compar
e
. RPF (σ/ρ = 1
0
.31) tested at
o
f PSB and RP
F
w
how the stre
n
e
r flexural loa
d
ve σ/ρ, RPF (
σ
l
s. For compar
i
m
ent, from Mid
g
g
/m
3
) has a σ/
ρ
designed for
2
ρ
of 1.31.
W
n
compression
a
p
erties of PSB
r
al σ/ρ, with
C
e
nsity ratio (lo
g
G
Pa concrete, a
n
8). Log-scale
u
h
e figure.
y
urethane-stabil
u
ent parts and
o
n
frastructure.
n
clean ballast
,
e
ry similar to C
e
nt modulus of
P
g
hway base-co
B stiffness to t
r
b
e significant
h
e accumulatio
ess than clean
B
presented he
r
s
t, when stabili
z
n
to the subb
a
n
d resistance
se areas can
h
st. Second,
w
Flexural
Compressive
Flexural
72
are
(
σ/ρ,
e
s to
8.8)
100
F
are
n
gth
d
ing
σ
/ρ =
i
son,
g
ley
ρ
of
2
0.7
W
hen
a
s in
and
C
SM
g
-
n
d
u
sed
ized
o
ther
The
,
but
SM,
P
SB
urse
r
ack
and
n of
and
r
ein,
z
ing
a
llast
to
h
ave
w
hen

Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 45 No.1 March 2014 ISSN 0046-5828
73
stabilizing ballast at the base of the ballast layer (i.e., as an
underlayment), the flexural strength results indicate that PSB can
withstand loading while serving to prevent intrusion of fines and
water from the subballast and subgrade layers. Data and
performance from actual field installation of PSB is still needed for
validating the laboratory results.
PSB is found to have suitable mechanical properties for use as a
material in track-substructure. The ease of injections and negligible
curing period for implementation of PSB makes it an attractive
alternative for railway maintenance. PSB may find appropriate
application in areas that cannot afford track shutdown or where
traditional maintenance capabilities are impeded or unachievable.
6. ACKNOWLEDGEMENTS
This research was funded by the National Center for Freight and
Infrastructure Research and Education. The contents of this report
reflect the views of the authors, who are responsible for the facts
and the accuracy of the information presented herein. The U.S.
Government assumes no liability for the contents or use thereof. The
contents do not necessarily reflect the official views of the National
Center for Freight and Infrastructure Research and Education, the
University of Wisconsin, the Wisconsin Department of
Transportation, or the USDOT’s RITA at the time of publication.
Contributions of Mr. Steven Reed and Dr. Randall Brown of Uretek
USA, Mr. Henry Lees of BNSF Railway Company, Dr. Ali
Ebrahimi of Geosyntec, Inc., and Ben Warren of the University of
Wisconsin-Madison are gratefully acknowledged.
7. REFERENCES
Akçaoğlu, T., Mustafa, T., and Çelik, T. (2003). "Effect of Course
Aggregate Size and Matrix Quality on ITZ and Failure
Behaviour of Concrete Under Uniaxial Compression."
Cement & Concrete Composites, 26, 633–638. Elsevier Ltd.
Anderson, W.F. & Fair, P. (2008). "Behavior of Railroad Ballast
under Monotonic and Cyclic Loading." Journal of
Geotechnical and Geoenvironmental Engineering, ASCE,
134(3), 316–328.
Arnold, G. (2009). "Reducing the risk of pavement failure and
utilisation of local materials in New Zealand through
Repeated Load Triaxial and Beam Fatigue Tests." AAPA
Thirteenth International Flexible Pavements Conference,
Gold Coast, Queensland, Australia, October 11-13, 2009.
Bayer Material Science. (2010). 486STAR Polyurethane Foam
Grout. Technical Data Sheet, Spring, Texas.
Buzzi, O., Fityus, S., & Sloan, S. (2010). "Use of Expanding
Polyurethane Resin to Remediate Expansive Soil
Formations." Canadian Geotechnical Journal, 47, 623–634.
Ebrahimi, A., Tinjum, J. M. and Edil, T. B. (2012) “Protocol for
Testing Fouled Railway Ballast in Large-Scale Cyclic
Triaxial Equipment,” Geotechnical Testing Journal, ASTM,
Vol. 35, No. 5, pp. 796-804.
Huang, Y.H. (2004). Pavement analysis and design – second
addition. Pearson Prentice Hall, Upper Saddle River, NJ.
Jaeger, J.C. (1967). "Failure of Rocks Under Tensile Conditions."
International Journal of Rock Mineral Science, 4, 219–227.
Keene, A.K. (2012). "Mitigating ballast fouling and enhancing rail-
freight capacity." Master's thesis, Department of Civil and
Environmental Engineering, University of Wisconsin–
Madison.
Keene, A., Edil, T.B., Fratta, D., and Tinjum, J.M. 2013. “Modeling
Effect of Polyurethane Stabilization on Track Response.”
ASCE Geotechnical Special Publication No. 231,
Proceedings Geo-Congress 2013, San Diego, CA.
Midgley, L. & Yeo, R. (2008). The Development and Evaluation of
Protocols for the Laboratory Characterisation of Cemented
Materials. Austroads Publication No. AP–T101/08. Austroads
Ltd. Sydney, Australia.
Salim, W.M. (2004). "Deformation and Degradation Aspects of
Ballast and Constitutive Modeling Under Cyclic Loading."
PhD thesis, School of Civil, Mining, and Environmental
Engineering, University of Wollongong.
Selig, E.T. & Waters, J.M. (1994). Track Geotechnology and
Substructure Management. Thomas Telford, New York, NY.
Szycher, M. (1999). Szycher’s Handbook of Polyurethanes. CRC
Press, Boca Raton, FL.
Tia, M. (2003). Bituminous Materials and Mixtures Civil
Engineering Handbook, Second Edition. CRC Press, Inc.,
Boca Raton, Florida, 45.1-45.36.
Zhang, P., & Wei, X. (2011). "Study on Flexural Strength and
Flexural Modulus of Elasticity of Cement." Advanced
Materials Research, 287-290, 990–993.