Volume 74-2 July- September 2013 ` 20.

00

JOURNAL
OF THE
INDIAN ROADS CONGRESS
 Journal Of The
Indian Roads Congress
Volume 74-2

CONTENTS
Page
Paper No. 595 Performance Evaluation of Stone Matrix Asphalt (sma) Using Low Cost Fibres 159
K.B. Raghuram and Venkaiah Chowdary
Paper No. 596 Longitudinal and Transverse Analysis of Multi-Cell Pre-stressed Concrete Box Girder 175
Rajesh Ranjan Rele and S.Talukdar
Paper No. 597 Well Foundation Construction in Bouldery Bed 188
R.K. Dhiman, VSM
Paper No. 598 Dynamic Vehicle Equivalent Factors for Characterisation of Mixed Traffic for 205
Multilane Metropolitan Arterials in India
Gaurang Joshi and Dinesh Vagadia
Paper No. 599 A Conceptual Strategy for Durability of Concrete Structures in the Arabian Gulf 220
and Similar Surrounds Afflicted by Humidity, High Temperatures, and Chlorides and Sulphates, etc.
V.K. Raina
List of Advertisers
Inside Front Cover - S.P. Singla Constructions Pvt. Ltd.
Inside Back Cover - Important Announcement
Outside Back Cover - Nehemiah Reinforced Soils (India)
Consulting Engineering Services (India) Pvt. Ltd. 236
TechFab India Industries Ltd. 237
Gloster Limited 238
Strata Geosystems (India) Pvt. Ltd. 239
Advertisement Tariff 240

The Indian Roads Congress Founded: December 1934

E-mail: secretarygen@irc.org.in/journal@irc.org.in
Jamnagar House, Shahjahan Road,
New Delhi 110 011
Tel: Secretary General: +91(11) 2338 6486
Sectt.: (11) 2338 5395, 2338 7140, 2338 4543, 2338 6274
Fax : +91 (11) 2338 1649
IRC Website: www.irc.org.in
Kama Koti Marg, Sector 6, R.K. Puram,
New Delhi 110 022
Tel : Secretary General : +91 (11) 2618 5303
Sectt. : (11) 2618 5273, 2617 1548, 2671 6778,
2618 5315, 2618 5319, Fax : +91 (11) 2618 3669

No part of this publication may be reproduced by any means without prior written permission from the Secretary General, IRC.

Published by the Indian Roads Congress

Copies can be had by sending advance payment to the Indian Roads Congress, Jamnagar House, Shahjahan
Road, New Delhi-110 011. Cheque/Demand Draft to be drawn in favour of Secretary General, IRC payable
at New Delhi.
New Delhi July-September 2013
 The Rights of Publication and Translation are reserved.

The Indian Roads Congress as a body does not hold

itself responsible for statements made, or for opinions
expressed in the Papers published in this Volume.

Edited and Published by Shri Vishnu Shankar Prasad, Secretary General, Indian Roads Congress, Jamnagar House,
Shahjahan Road, New Delhi on behalf of the Indian Roads Congress. Printed by Shri Madan Lal Goel on behalf of the
Indian Roads Congress at Aravali Printers & Publishers (P) Ltd., W-30, Okhla Industrial Area, Phase-II, New Delhi.
14,000 copies, July-September, 2013
 Paper No. 595

PERFORMANCE EVALUATION OF STONE MATRIX ASPHALT

(SMA) USING LOW COST FIBRES
K.B. Raghuram* And Venkaiah Chowdary**

Abstract
SMA is hot mix asphalt developed during 1960s in Germany and has been used successfully in various countries to provide better resistance to rutting.
SMA consists of coarse aggregate skeleton with higher binder content intended to increase the durability of the mix. Even though the use of higher binder
content improves the durability of the mix, it will also lead to asphalt draindown. Traditionally, cellulose fibre has been used to reduce the draindown
of asphalt from SMA mixes. This investigation considers the use of various low cost stabilizers and also high viscosity binders as draindown retarders.
The performance of the mix is evaluated in terms of stability and the resistance to rutting. The stability of SMA is evaluated using the standard Marshall
procedure. The rut depth is captured using wheel track test setup where the rut depth achieved after 20,000 repetitions is used to evaluate the performance
of SMA mixes. It is observed that the use of low cost fibres as stabilizer additives improved the performance of SMA mixes when compared to the control
mix. The stabilizers were able to retard the draindown of asphalt from the SMA mixes to a greater extent.

1 INTRODUCTION weights and gross weights have increased, as

The highway traffic in India is increasing at a faster
rate with the population and the road network has also
expanded in different regions of India. This has led
to an increase in the volume of heavy vehicles as the
travel time and vehicle operating costs have increased
tremendously. Apart from the increase in truck traffic,
there is also a huge difference in the maximum and
minimum temperatures of the country. The maximum
air temperature can reach even 50C in some parts of
the country and the resulting pavement temperature
can reach up to 60C. With the increase in loading and
temperatures, the pavements are subjected to various
types of distresses. Rutting has been observed to be
a major distress in flexible pavements and several
studies were carried out across the globe to quantify
the mechanisms of rutting and to reduce the effective
rut depth along the wheel path.
The increase in traffic volume has been observed
globally, especially the truck traffic. As truck axle
truck tyre pressures have increased, and as truck
contact areas have decreased in recent years, the
amount of rutting or permanent deformation that
has occurred in dense graded asphalt concrete
binder and surface course layers has also increased
dramatically (Scherocman, 1992). There are several
parameters contributing to rutting and the type
of mixes used in the construction of bituminous
layers plays a crucial role in the minimization of
the rut depth. One such mix is SMA. The 1990
AASHTO European Asphalt Study brought back
the German asphalt mix technology, known as
splittmastixasphalt as premium, long-lasting,
rut-resistant asphalt surface mix. The English
translation of splittmastixasphalt is stone
mastic asphalt. The Americanized version of this
mix technology is known as stone matrix asphalt
(Prowell et al., 2002). SMA was developed in
Germany during the mid 1960s and it has been
used in Europe for more than 20 years to provide

* Former M.Tech Student

** Assistant Professor } Transportation Division, Department of Civil Engineering, National Institute of Technology, Warangal-506 004,
E-mail: chowdaryazad@gmail.com
Written comments on this Paper are invited and will be received upto 30th October, 2013.

Journal of the Indian Roads Congress, July-September 2013

160 Raghuram And Chowdary on
better rutting resistance and to resist studded tyre
wear (Scherocman, 1991).
2 LITERATURE REVIEW
SMA is a gap graded aggregate-asphalt hot mix that
maximizes the asphalt cement content and coarse
aggregate fraction which provides a stable stone-on-
stone skeleton that is held together by a rich mixture of
asphalt cement, filler, and stabilizing additive (Brown
and Manglorkar, 1993). SMA has a much higher
percentage of large crushed coarse aggregate and
higher percentage of asphalt content when compared
to conventional asphalt mixtures. As a result, SMA
mixtures must contain the proper type and amount of
some stabilizing additive in order to retain the asphalt
binder during production and placement (Brown et al.,
1997a, Brown et al., 1997b). The use of higher binder
content enhances the durability of SMA. Because of its
higher rut resistance property and higher durability, it
is most preferred over the conventional dense graded
asphalt mixes.
In dense graded asphalt mixes, the mortar (fine
aggregate and asphalt cement) in the mix actually
carries the traffic load. The coarse aggregate particles
in dense asphalt mixes are not in close contact with
each other and there will be considerable amount of
space between the coarse aggregates which is filled
with fine aggregate and asphalt cement. The gradation
of the aggregate and optimum asphalt content for SMA
are considerably different from dense graded asphalt
mixes. The stone-on-stone contact is much higher in
SMA and the loads are carried by the coarse aggregate
particles instead of the mortar resulting in lower
permanent deformation in SMA (Scherocman, 1992).
Traditionally, the 30-20-10 rule has been followed
for SMA gradation which suggests that SMA should
have approximately 30 percent passing the 4.75 mm
sieve, 20 percent passing the 2.36 mm sieve, and 10
percent passing the 0.075 mm sieve. The first major use
of SMA in the United States of America was designed
following this rule (Scherocman, 1991). The percent
passing the 4.75 mm sieve is a critical factor in the
formation of stone-on-stone contact in SMA (Brown
Journal of the Indian Roads Congress, July-September 2013
and Mallick, 1994). As the percent passing 4.75 mm
sieve decreases, the VMA remains nearly constant, and
then begins to increase once the percent passing the
4.75 mm sieve reaches 30 to 40 percent. The point at
which the VMA begins to increase defines the condition
at which stone-on-stone contact begins to develop.
Below 30 percent, a lowering of percent passing the
4.75 mm sieve tends to increase the VMA by opening
up more space in the coarse aggregate structure. Hence,
the percent passing the 4.75 mm sieve must be lowered
below approximately 30 percent to ensure the formation
of stone-on-stone contact (Brown et al., 1997a; Brown
et al., 1997b).
Two methods are reported in the literature to establish
the stone-on-stone contact in SMA. In the first method,
the density of coarse aggregate skeleton in the total
SMA mixture sample is compared with the density of
the coarse aggregate only fraction. If the SMA coarse
aggregate skeleton density is greater than or equal to
the coarse aggregate only fraction density, the SMA
mixture has a stone-on-stone coarse aggregate skeleton
(Haddock et al., 1993). The second method is based on
the relationship between Voids in Mineral Aggregate
(VMA) and the percentage of fine aggregate (material
passing 4.75 mm) in the SMA mixture. If the SMA
coarse aggregate skeleton has VMA less than or equal
to the coarse aggregate only fraction VMA, the SMA
mixture is judged to have stone-on-stone contact
(Brown and Mallick, 1995). Five different compaction
methods are reported in the literature to determine
VMA of the coarse aggregate only fraction: Marshall
hammer, dry rodded method, vibrating table, Superpave
gyratory compactor, and the British vibrating hammer.
The Superpave gyratory compactor and dry rodded
methods produced best results (Brown and Haddock,
1997a; Brown and Haddock 1997b). Digital imaging
has also been used to establish stone-on-stone contact
where the results obtained from digital imaging agree
well with those obtained through the VMA method.
The advantage of digital imaging is that it can quantify
the number of contacts between aggregate particles
(Watson et al., 2004).
Since the aggregate gradation plays a crucial role in the
 Performance Evaluation of Stone Matrix Asphalt (Sma) Using Low Cost Fibres 161

Table 1 SMA Gradations Developed in Various Countries

IRC, India NAPA, USA NCHRP 425, USA German MORTH, India
Organization,
(IRC:SP:79, (Kandhal, (Brown and Specifications, (MORTH,
Country
2008) 2002) Cooley, 1999) Germany 2001)
Nominal maximum
13 19 12.5 19 12.5 13 10
aggregate size, mm
Sieve size, mm Cumulative percent by weight of total aggregate passing
26.5 - 100 - - - - -
25 - - - 100 - - -
19 100 90-100 100 90-100 100 - -
16 - - - - - 100 -
13.2 90-100 45-70 - - - 90-100 100
12.5 - - 85-95 50-74 90-100 - -
9.5 50-75 25-60 75 Max. 25-60 34-75 54-70 90-100
4.75 20-28 20-28 20-28 20-28 23-41 26-39 30-50
2.36 16-24 16-24 16-24 16-24 18-30 21-28 21-28
1.18 13-21 13-21 - 13-21 15-24 17-25 16-25
0.600 12-18 12-18 12-16 12-18 12-20 15-22 14-22
0.300 10-20 10-20 12-15 12-15 10-17 13-19 12-20
0.150 - - - - 9-14 9-15 10-17
0.075 8-12 8-12 8-10 8-10 8-13 8-13 9-13
0.020 - - 3 Max. - - - -

development of stone-on-stone contact in SMA mixes,
the SMA gradations developed in various countries are
summarized in Table 1. Recently, Indian Roads Congress
(IRC) introduced the tentative specifications for SMA
suited to Indian conditions (IRC:SP:79, 2008) where
the aggregate gradation closely matches with NCHRP
specifications. Previous to this IRC publication, SMA
mixes were designed in India using the Ministry of
Road Transport and Highways Manual for Construction
and Supervision of Bituminous Works (MORTH,
2001) where the aggregate gradation closely matches
with German specifications. The percentage material
passing the 4.75 mm sieve size is less than 30% for all
the standards reported in Table 1 except the German
specifications, and MORTH (2001) specifications. It is
worthwhile to highlight here that, according to Brown
et al. (1997a, 1997b), the percent passing the 4.75 mm
sieve must be lowered below approximately 30 percent
to ensure the formation of stone-on-stone contact.
Some of the studies on SMA mixes carried out in India
(Punith et al., 2004; Kamaraj et al., 2006) followed
the MORTH (2001) gradation. However, Kamaraj et
al. (2006) ensured stone-on-stone contact using the
method developed by Brown and Mallick (1995).
The use of higher binder content in SMA leads to
draindown when the mix is held at higher temperatures
during the production, storage, transport, and placement
of the mixture. Researchers across the globe have
tried using various stabilizers and modified binders in
SMA in an attempt to decrease the draindown of the
binder and to improve the performance characteristics.
Brown et al. (1997b) used SBS, polyolefin, cellulose
fibre, and rock wool fibre as stabilizers in SMA. Fibre
stabilizers are found to be more effective in reducing
draindown than polymer stabilizers. The draindown

Journal of the Indian Roads Congress, July-September 2013

162 Raghuram And Chowdary on
test was performed using NCAT method at three test
temperatures: 140C, 155C and 170C. However, for Construction and Supervision of Bituminous
mixtures modified with polymers showed better
resistance to rutting in laboratory wheel tracking tests.
Punith et al. (2004) used reclaimed polyethylene in
shredded form obtained from grocery bags as stabilizer
in SMA mixes. They used MORTH (2001) gradation
corresponding to NMAS 10 mm and reported a
decrease in draindown of the binder with the use of
reclaimed polyethylene. Kumar et al. (2004) used
cellulose fibres as stabilizer in SMA and reported higher
resistance of SMA to permanent deformation when
compared to asphalt concrete. Kamaraj et al. (2004)
used cellulose fibres as stabilizers and natural rubber
modified bitumen as the binder in SMA mixes. They
observed a decrease in draindown and improvement
in resistance to deformation in SMA mixes when
compared to dense graded bituminous macadam and
bituminous concrete. Muniandy and Huat (2006)
observed improved fatigue performance characteristics
of SMA using cellulose oil palm fibres. Kamaraj et al.
(2006) used three types of cellulose based additives:
Technocel, Topcel, and Genicel and observed that the
draindown values are in the range of 0.02 to 0.20%.
Apart from the decrease in draindown values, they also
reported improved performance characteristics of SMA
mixes using all the three cellulose based additives.
Kumar et al. (2007) used cellulose fibre, asphalt coated
jute fibre and CRMB as the stabilizers in SMA. The
draindown test was performed using NCAT method
at 170C. Cellulose fibre and asphalt coated jute fibre
are less effective in reducing draindown than CRMB
stabilizer. The mixtures modified with CRMB showed
better resistance to rutting when compared to cellulose
and asphalt coated jute fibres. Suchismita et al. (2010)
observed zero draindown by using coir fibre in SMA
and concluded that addition of fibre (coir) controls the
draindown effects of SMA mixtures. It is surprising
to note that SMA mixes without coir fibres using 60/70
and 80/100 grade asphalts resulted in draindown of
0.005% and 0.01%, respectively. These values are far
less than the permissible draindown content of 0.3%
and one can come to a conclusion based on these
results that the mix theoretically does not need any
fibres. They adopted the draindown test method given
Journal of the Indian Roads Congress, July-September 2013
in Ministry of Road Transport and Highways Manual
Works (MORTH, 2001) which was subsequently
adopted from the British Standards. Some of the
researchers (Kamaraj et al., 2006) used the draindown
test developed by Schellenberg Institute, Germany.
The Schellenberg draindown test is not identical to
NCAT draindown test (Brown and Cooley, 1999), but
the maximum permissible draindown limit of 0.3%
remains same as the NCAT specification. Prowell and
Hurley (2008) developed a new type of mixture termed
stone skeleton asphalt where the mixture would have
similar performance characteristics to typical SMA
mixtures, but without the elements contained in SMA
mixtures that increase its cost which include: modified
binders, fibres and mineral fillers. The first trial section
in India using SMA was constructed in 2006 at two
intersections (Kajuri Chowk and Brij Puri Chowk) in
New Delhi (Sharma and Kamaraj, 2008).
From the above studies one can observe that several
fibres have been used to reduce the draindown of the
binder from SMA mixes. It is important to note the
comment made by Prowell et al. (2009): fibres serve
no real purpose after the mix is compacted in-place.
Thus the main role of fibres in SMA mixes is to reduce
the draindown of the binder rather than improving the
mechanical properties of the SMA mixes. Cellulose fibre
has been the first choice till now and it is worthwhile
looking at other low cost fibres which are abundantly
available in India. Also, the use of various fibres, fillers
and modified binders in SMA mixes increases the cost
of the asphalt mix when compared to the traditional
mixes. In order to reduce the cost of SMA, some of
the ingredients are replaced in this study with low
cost ingredients. The objectives of this paper are: (i)
to evaluate the effect of stabilizers in SMA in terms of
draindown criteria, and (ii) to evaluate the performance
of the SMA through Marshall and wheel tracking tests.
Thus, the main focus of this paper is to consider the
use of various low cost fibres and modified binders
as draindown retarders in SMA mixes and to evaluate
its performance. The experimental investigation
considered aggregate from a single source and followed
a single gradation. The permanent deformation tests
 Performance Evaluation of Stone Matrix Asphalt (Sma) Using Low Cost Fibres 163

were performed only at one test temperature, that is, Table 2 Physical Properties of the Aggregate
60C. The degradation effects of natural fibres are not
considered in this paper. Since the natural fibres are Specification
Property Test Value (IRC:SP:79,
biodegradable, the life of these fibres is expected to be 2008)
higher as these fibres are coated with asphalt binder in
Cleanli- < 2% passing
the SMA. ness
Grain size analysis 0%
0.075 mm sieve

Particle Combined flakiness

3 MATERIALS 28% < 30%
shape and elongation index

3.1 Aggregate Hardness

The combined aggregate gradation (coarse aggregate,
fine aggregate and mineral filler) developed by Indian
Roads congress (IRC:SP:79, 2008) shown in Table 1
closely follows the SMA gradation of NCHRP 425. In
order to ensure stone-on-stone contact, the percentage
passing 4.75 mm sieve should be below 30% and the
gradation developed by IRC satisfies this condition.
Fig.1 shows the gradation of aggregate used in this
study which follows the IRC specifications for nominal
maximum aggregate size of 13 mm. The SMA mixes
prepared using this gradation can be used as a wearing
Fig. 1 SMA gradation used in this study [IRC:SP:79, 2008]
course with nominal layer thickness ranging from 40 to
50 mm. The entire aggregate was procured from a single
quarry and the physical properties of the aggregate are
shown in Table 2. Stone polishing value test was not
performed on the aggregates due to non-availability of
the equipment.
3.2 Mineral Fillers
Fly-ash, cement, and rock dust are used as fillers in
Los Angeles abrasion
22.2% < 25%
value
Aggregate impact
Toughness 14% < 18%
value
Water
Water absorption 0.12% < 2%
absorption
Coating and stripping
Retained
Stripping of bitumen aggregate 95%
coating > 95%
mixtures
this investigation. Even though the main focus of this
paper is on utilization of low cost ingredients in SMA,
cement is also considered as one of the alternate filler
to determine the relative draindown and performance
characteristics when used in SMA mixes. According
to IRC, mineral filler should consist of finely divided
mineral matter and fly-ash should not be used as filler.
The reason behind barring fly-ash as mineral filler is
not reported by IRC in its current specifications on
SMA. Some states in the US do not permit the use
of fly-ash because its particles are generally rounded
and, therefore, increase the rutting potential of the
bituminous mix when used on heavy traffic roads
(Kandhal and Chattaraj, 2009). Similar studies are
reported in India. Kumar et al. (2008) reported an
increase in permanent deformation with increase in
fly-ash content indicating an increase in rutting
tendency. However, they have also reported increased
resistance to low temperature cracking and least loss of
tensile strength due to moisture damage when fly-ash is
used as filler; other fillers being lime and lime-fly-ash.
The basic purpose of adding filler in SMA is to increase
the surface area which in turn can reduce draindown
of asphalt binder. In order to check the effectiveness
of fly-ash in retarding draindown of binder in SMA,
fly-ash is also used as mineral filler in this investigation.
Considering the fact that the main purpose of SMA

Journal of the Indian Roads Congress, July-September 2013

164 Raghuram And Chowdary on

is to reduce the effective rut depth in pavements, the Table 3 Physical Properties of Asphalt Binders
barring of fly-ash as mineral filler in SMA by IRC can 60/70 (VG-30) PMB-70 CRMB-60
be justified. The mineral filler of 2 percent by weight Property (IS:73, 2006) (IRC:SP:53, 2010)
of aggregate is used in this study by replacing the rock Value Spec. Value Spec. Value Spec.

dust. This will ensure that the rock dust including the Penetration at
25 C, 0.1 mm, 66 50-70 72 50-80 NA NA
mineral filler passing 75 mm sieve would be 10%. 100g, 5s
Softening point, C 50 47 58 55 60 60
3.3 Fibres (Min.) (Min.) (Min.)
Flash point, C 280 220 290 220 320 220
Six different types of fibres such as coconut fibres, oil (Min.) (Min.) (Min.)
Elastic recovery at NA NA 70 60 65 60
palm fibres (OPF), jute fibres, glass fibres, and fibres 15 C, %
extracted from refrigerator door panels (FERP), and
fibres extracted from old machinery belts (FEMB) are of this investigation, these tests are clearly out of the
used as stabilizers in this investigation. Coconut fibres, scope of this paper.
OPF, and jute fibres are the natural fibres, glass fibre
4 EXPERIMENTAL INVESTIGATION
is an artificial fibre, whereas FERP and FEMB are the
waste fibres. Except the glass fibre, remaining all are 4.1 Draindown Test
low cost fibres. The length of the fibres used in this
study is 10 mm. The standards developed in India and One of the important criteria in SMA mixes is to
abroad specify the dosage rate of cellulose fibres as control the draindown of the binder. In order to reduce
minimum 0.3% by weight of total mix. In this study, the the draindown of the binder, one can either increase the
fibre dosage rate is fixed as 0.3% by the weight of total surface area of the aggregate skeleton or can use high
mix. It is also possible to use different fibre content. In viscosity binders. The surface area can be increased
this study, the specified fibre content is kept constant by using either fillers or stabilizers in SMA. In order
and the properties of a particular fibre are assumed to to quantify the draindown characteristics of SMA, the
be homogeneous. procedure outlined in ASTM D 6390 is used. For this
purpose, a standard sieve size of 6.3 mm is used. The
3.4 Binder standard NCAT test procedure (later on adopted by
ASTM D 6390) outlines the use of forced draft oven,
Bitumen of 60/70 penetration grade (VG-30), and two
capable of maintaining the temperatures in the range of
different types of modified binders such as PMB-70
120 to 175C with 2C of the set temperature. The
and CRMB-60 are used in this investigation. The
draindown should be determined at two temperatures:
physical properties of asphalt binders are shown in
anticipated plant production temperature, and 10C
Table 3. The primary purpose for adopting modified
above the anticipated plant production temperature
binder is due to its higher viscosity when compared to
(Brown and Cooley, 1999). The SMA mix prepared
traditional unmodified asphalt which in turn can reduce
using the gradation shown in Fig.1 was kept in an oven
the draindown of the binder from SMA. Fibres are not
at 150C (expected plant production temperature) for
used when modified asphalt is used as the binder. The
60 minutes. The mixing and compaction temperature
minimum required asphalt content in SMA mixes as
varies with the type of binder and are usually determined
per IRC:SP:79 (2008) is 5.8%. In this study, 6% asphalt
from viscosity temperature relationship of the binder.
content is used to prepare the SMA mixes which is
In order to achieve proper workability of the mix,
the optimum asphalt content corresponding to 4% air
the modified binders were heated to relatively higher
voids and minimum 17% voids in mineral aggregate.
temperatures compared to the unmodified binder.
One should make a note that it is essential to use higher
However, in order to avoid using two variables (binder
bitumen content in SMA from durability point of view.
type and test temperature) the draindown test was
Even though durability tests are not performed as part

Journal of the Indian Roads Congress, July-September 2013

 Performance Evaluation of Stone Matrix Asphalt (Sma) Using Low Cost Fibres 165

conducted at the same temperature for mixes with both deformation of SMA mixes using a LVDT. All the
unmodified and modified binders so that the influence wheel tracking tests in this study were performed at
of modification could be captured. The mixing and 60C. The test temperature was maintained at 60C by
compaction temperatures for modified binders were placing a rectangular aluminium cover of dimensions
higher than the corresponding temperatures when 1500 mm x 750 mm x 760 mm over the wheel tracking
unmodified binder were used and depends mainly test setup as shown in Fig. 2(c).
on the type of modifier, and amount of modifier. The
draindown of binder from the mixture was calculated
using the Equation (1).
Initial sample massFinal sample mass
Draindown (percent)=
Initial sample mass
x 100 ...1

4.2 Marshall Test Method

The Marshall samples were prepared and tested using

the standard protocols outlined in ASTM D6926 and
D6927, respectively. The Marshall stability is the
peak resistance load obtained during a constant rate
of deformation loading which indirectly represents the (a)
strength of a paving mix at zero vertical stress which
is the critical. The samples were prepared using the
aggregate gradation shown in Fig. 1 by applying 50
blows on each face of the sample using 6% asphalt
content. It is important at this juncture to note the
observations made by Brown and Mallick (1995)
and Brown et al. (1997b). Brown and Mallick (1995)
reported that 75 blows of Marshall compaction would
breakdown the aggregate and recommended that SMA
mixes should be designed with 50 blows. Brown et al.
(1997b) observed that 50 blows of Marshall hammer
is approximately equal to 100 revolutions of the
(b)
Superpave gyratory compactor in terms of the resultant
SMA density. Following parameters were measured:
stability, flow, unit weight, percentage air voids in total
mix, voids in mineral aggregate (VMA) and voids filled
by binder (VFB).

4.3 Wheel Tracking Test

The performance of SMA mixes with various

combinations of fibres, fillers and binders is evaluated by
measuring the permanent deformation in the laboratory
prepared SMA slabs. The SMA slabs were prepared
(c)
using the roller compactor shown in Fig. 2(a) to the
Fig. 2 Wheel tracking test setup: (a) roller compaction, (b) wheel
target density. The standard wheel tracking test setup
tracking equipment, and (c) aluminium
shown in Fig. 2(b) was used to measure the permanent chamber to maintain temperature

Journal of the Indian Roads Congress, July-September 2013

166 Raghuram And Chowdary on

In the wheel tracking test, the SMA slab prepared with
different combinations of fillers, fibres and binders
using roller compaction was placed on the test table
and was made to move in to and fro motion. While
a hard rubber wheel press against the specimen by
loading weight (705 N wheel load) through a lever and
a straight track wheel is indented in the specimen. The
complete table over which the slab was placed moves
such that there would be approximately 43 wheel passes
per minute. The rut depth was measured using a LVDT
with lapsed number of load repetitions. The SMA slab
was subjected 20,000 passes of a standard wheel load.
The number of passes was established after several
laboratory trials. In most of the trial tests, significant
rutting could be observed up to 20,000 wheel passes
after which the change in rut depth was negligible. The
permanent deformation caused after 20,000 cycles is
used to compare the performance of the SMA mixes
using various combinations of fillers, fibres and binders
with that of the control mix, i.e., mix prepared without
any filler, fibre and with unmodified VG-30 grade
asphalt.
when FERP is used as the stabilizer. Jute fibre showed
an entirely different trend which could not be explained
with the existing results. Using a particular type of
filler, the FERP fibre showed very less draindown when
compared to all other fibres. The mix with cement as
filler showed very less draindown for both natural
fibres: coconut and jute whereas the mix with fly-ash
as filler showed very less draindown for both glass and
FEMB fibres. This may be possibly due to fineness of
both cement and fly-ash. The usage of coconut fibres
in SMA resulted in higher draindown, which may be
due to less binder holding capacity of the fibres. It is
important to note that the draindown is less than 0.3%
when fillers and fibres are used as stabilizers, which are
very much less than the draindown of the control mix.

5 DISCUSSION OF THE RESULTS

5.1 Draindown Test Results

The results obtained from the draindown study using

Fig. 3 Draindown as a function of type of fibre and filler using
various fillers, fibres and binders are presented in unmodified asphalt
Figs. 3 and 4. The draindown for the control mix was
obtained as 1.33%, which is very much higher than
the mixes prepared with different fillers, fibres, and
binders. Fig. 3 shows the plot in which draindown is
plotted as a function of the type of fibre and the type of
filler using unmodified VG-30 grade asphalt. One can
observe that the mix without any filler showed a higher
draindown when compared to the mixes prepared with
fillers and using coconut, glass and FEMB fibres with
an exception to jute and FERP fibres. The reduction in
draindown with the utilization of fillers in the mix is
due to increased surface area. That is, higher asphalt
content being utilized to coat the increased surface
area resulting in reduced draindown. The draindown of
FERP fibre with and without fillers remained constant.
The possible reason may be ascribed to higher surface
area of the FERP fibres when compared to other fibres Fig. 4 Draindown as a function of type of filler and binder
and the addition of fillers has no effect on the draindown

Journal of the Indian Roads Congress, July-September 2013

 Performance Evaluation of Stone Matrix Asphalt (Sma) Using Low Cost Fibres 167

Fig. 4 shows the plot in which draindown is plotted as other fibres. In all other cases, higher stability values
a function of the type of filler and the type of binder are observed with the addition of fillers. That is, fillers
without using any fibres in the mix. SMA with CRMB-60 tend to fill micro voids within the matrix resulting in
showed less draindown when compared to other binders higher stability. From Fig. 6, it can be seen that stability
such as VG-30 asphalt and PMB-70. This shows that the values are higher for modified asphalt when compared
type of binder also significantly affects the draindown to VG-30 grade asphalt, which follow general trend of
of SMA. The main reason for reduced draindown with increase in stability of the mix with increase in hardness
modified asphalt is due to viscosity playing a key role of the asphalt. FERP and jute fibres resulted in higher
in controlling the draindown of asphalt from SMA. It stability values when compared to all other fibres.
is well known that viscosity is the resistance offered
to the flow of the material; with increased viscosity
one can expect a reduction in draindown of asphalt. In
sequence with the above statement, the SMA mixtures
with unmodified asphalt resulted in higher draindown
when compared to modified asphalts. Without the use
of any filler, all the above-mentioned three asphalts
resulted in higher draindown of the resulting mix. Of
all the three cases, the mix with unmodified asphalt
as the binder resulted in higher draindown. Addition
of fillers reduced the draindown drastically from the
SMA mixes. The SMA mixes with modified binders
and with fillers resulted in the lowest draindown values
Fig. 5 Marshall stability as a function of type of fibre and filler
when compared to the mix with unmodified asphalt as using unmodified asphalt
the binder. This shows that the modified asphalt also
has the potential to reduce the draindown. IRC:SP:79
(2008) recommends the use of PMB-40 when modified
asphalt is used as the binder in SMA mixes. Even though
PMB-40 is not used in this study, because of its higher
viscosity, SMA mixes manufactured with PMB-40 might
exhibit lesser draindown than PMB-70.

5.2 Marshall Test Results

5.2.1 Stability

With the incorporation of different types of fibres as

reinforcement within the SMA mix, maximum stability
is obtained with fillers compared to mixes prepared Fig. 6 Marshall stability as a function of type of filler and binder

without any filler as shown in Fig. 5. Stability is found 5.2.2 Flow

to be a function of type of filler and fibre for both VG-30
asphalt and modified asphalts as can be seen from Figs. 5
and 6. From Fig.5, one can observe that the stability
values are higher for jute fibre with cement as filler
and with VG-30 grade asphalt when compared to all
The flow value was measured as the deformation
between no load and maximum load carried by the
specimen during the test. From Figs. 7 and 8, one can
observe that, on an average the flow values are around
4 mm.

Journal of the Indian Roads Congress, July-September 2013

168 Raghuram And Chowdary on
Fig. 7 Flow as a function of type of fibre and filler using
unmodified asphalt
Fig. 8 Flow as a function of type of filler and binder
5.2.3 Air Voids in Total Mix
Air voids are necessary in asphalt mixes to allow
densification under traffic loads and to prevent bleeding
of asphalt during hot climates. The field performance
evaluation of SMA mixes in United States has shown
that lower air voids resulted in fat spots and rutting
(Brown et al., 1997c). Usually the laboratory compacted
SMA samples had air voids ranging between 2 and
3 percent. To minimize fat spots and rutting, the air
voids in warmer climates should be designed closed
to 4 percent (Brown et al., 1997b). Considering this
fact, IRC:SP:79 (2008) specified 4 percent air voids in
SMA mixes designed using the Marshall method. From
Fig. 9, it can be seen that SMA mixes manufactured
with (i) jute fibre and without filler, (ii) glass fibre with
rock dust filler satisfied this condition. Other SMA
Journal of the Indian Roads Congress, July-September 2013
mixes manufactured with different fillers, fibres and
binders seem to deviate from the prescribed air voids
content which can be observed from Figs. 9 and 10.
Fig. 9 Air voids in total mix as a function of type of fibre and filler
using unmodified asphalt
Fig. 10 Air voids in total mix as a function of type
of filler and binder
5.2.4 Voids in Mineral Aggregate
The primary purpose of the VMA is to ensure reasonably
high asphalt content. This can be accomplished by
specifying minimum asphalt content or a minimum
VMA. The best approach is to specify minimum VMA
because it is calculated on a volume basis and is not
affected by aggregate specific gravity (Brown et al.,
1997a). Based on the number of mixture designs and
field experiences in US, Brown et al. (1997a) specified
the reasonable minimum requirement of VMA as
17 percent. These guidelines were also adopted by the
IRC:SP:79 (2008). The variation of VMA with type of
 Performance Evaluation of Stone Matrix Asphalt (Sma) Using Low Cost Fibres
filler, type of fibre and type of asphalt is shown in Figs. 11
and 12. All the SMA mixtures manufactured with
combinations of fillers, fibres, and binders satisfied the
minimum required VMA.
Fig. 11 VMA as a function of type of fibre and filler using
unmodified asphalt
Fig. 12 VMA as a function of type of filler and binder
5.2.5 Air Voids Filled by Binder
Certain amounts of air voids are present in the mineral
aggregate and the asphalt is expected to fill 65 to 75
percent of these voids. The variation of air voids filled
by asphalt with type of filler, type of fibre and type of
asphalt is shown in Figs. 13 and 14. Since the asphalt
content used is higher in SMA mixes when compared
to the traditional asphalt mixes, one can expect higher
air voids in mineral aggregate being filled by the binder
and all the SMA mixes reported in Figs. 13 and 14
seems to satisfy this condition.
Journal of the Indian Roads Congress, July-September 2013
169
Fig. 13 Voids filled by binder as a function of type of fibre and
filler using unmodified asphalt
Fig. 14 Voids filled by binder as a function of type of filler and
binder
5.3 Wheel Track Test Results
Figure 15 shows the permanent deformation plotted
against the number of passes for the control mix
and SMA mixes with various natural fibres. One
can observe that the rate of increase of rut depth
(permanent deformation) decreased with increased
number of passes (cycles). The performance of the
mixes with coconut fibre up to 15000 passes is lower
than the control mix. This might be possibly due to
difference in internal structure evolution with number
of passes for the mix with and without coconut fibres.
It can be observed that all the permanent deformation
plots are falling in the primary zone of the three-stage
permanent deformation curve. That is, the increase in
deformation is essentially due to reduction in air voids.
The rate of air voids reduction is higher (steeper slope)
in the control mix when compared to rate of air voids
170 Raghuram And Chowdary on
reduction (flatter slope) in SMA with coconut fibre as
the stabilizer. After certain number of loading cycles,
the rate of change of rut depth reached an asymptotic
state, i.e., change in rut depth is almost negligible. The
rut depth at this point, that is, after 20,000 cycles for the
control mix is 3.91 mm. The corresponding rut depths
using various natural fibres such as coconut fibre, oil
palm fibre, and jute fibres are: 3.78 mm, 2.72 mm, and
2.25 mm respectively. It can be observed that all the
above mentioned SMA mixes are resistant to rutting
and the performance of SMA mixes stabilized with jute
fibres is slightly better than the SMA mixes stabilized
with other natural fibres.
Fig. 15 Permanent deformation as a function of natural fibres
Figure 16 shows the permanent deformation plotted
against the number of passes for the control mix and
Fig. 16 Permanent deformation as a function of waste fibres
Journal of the Indian Roads Congress, July-September 2013
SMA mixes with various waste fibres. The rut depths
using various waste fibres such as FERP, and FEMB
are: 2.35, and 2.70 mm respectively. It can be observed
that all the above mentioned SMA mixes are resistant
to rutting and the performance of SMA mixes stabilized
with FERP fibres is slightly better than the SMA mixes
stabilized with FEMB fibres.
Figure 17 shows the permanent deformation plotted
against the number of passes for control SMA mix and
SMA mix with glass fibre which is an artificial fibre.
The rut depth using this artificial fibre is 2.70 mm. It
can be observed that the performance of SMA mixes
stabilized with glass fibres is much better than the
control mix. From Figs. 15 to 17, higher performance
Fig. 17 Permanent deformation as a function of artificial fibres
Fig. 18 Permanent deformation as a function of type of fibres
 Performance Evaluation of Stone Matrix Asphalt (Sma) Using Low Cost Fibres 171

of SMA mixes can be expected with the use of jute fibre Table 4 Permanent Deformation of Various
as stabilizer when compared to other natural fibres. SMA Mixes
Similarly, FERP exhibited better performance under
Permanent
the waste fibres category and glass fibre seems to show Type of
Type of fibre deformation,
better performance characteristics when compared binder
to the control mix. Fig. 18 shows a plot in which the
mm
permanent deformation of SMA mixes prepared using - VG-30 3.91
natural, waste and artificial fibres including the control Coconut fibre VG-30 3.78
mix are plotted against the number of loading cycles.
It can be seen that jute fibre, which is a natural fibre is Oil palm
VG-30 2.72
effective in retarding the permanent deformation when fibre
compared to the waste and artificial fibres. The waste Jute fibre VG-30 2.25
fibre is the second choice of preference followed by the FERP VG-30 2.35
artificial fibre. Since the natural fibres are abundantly
available in India, jute can replace the costlier artificial FEMB VG-30 2.70
fibres which have been the primary choice of stabilizer Glass fibre VG-30 2.70
in SMA mixes. - CRMB-60 2.62
- PMB-70 3.20

It can be observed that all the above mentioned SMA

Fig. 19 Permanent deformation as a function of type of binder
As discussed earlier, fibres acts as draindown retarders
and also improves the performance of SMA mixes. In
order check whether the above objective can be fulfilled
by replacing the unmodified asphalt with modified
asphalt, without using any fibres, wheel tracking tests
were performed on SMA mixes prepared with fly-ash
as filler and modified asphalt as the binder. Fig. 19
shows the permanent deformation plotted against the
number of passes for SMA mixes with fly-ash as filler,
CRMB and PMB as binders without any fibres. The rut
depths using various modified binders such as CRMB-
60, and PMB-70 are: 2.62, and 3.20 mm respectively.
mixes with modified binders are resistant to rutting
and the performance of SMA mixes using CRMB-60
as the binder is slightly better than the SMA mixes
using PMB-70 as the binder. Fibres are observed to be
more effective in resisting the permanent deformation
of SMA mixes when compared to modified binders.
The permanent deformation observed after 20,000
passes in various SMA mixes prepared using different
types of fibres and binders are summarized in
Table 4. Even though the difference between the
maximum and minimum permanent deformations
observed is relatively low, this difference gives some
idea about the relative performance of mixes with
different fibre types.
The main purpose of using fly-ash is to observe the
relative effects of various fillers on the draindown
characteristics of asphalt binders. Minimum draindown
could be observed when fly-ash is used as filler in
combination with some of the fibres (coconut, OPF, and
FERP) and modified binders (PMB-70 and CRMB-60).
Having observed a reduction in draindown with the use
of fly-ash, studies were also performed to evaluate the
rutting resistance of asphalt mixes with fly-ash as filler.

Journal of the Indian Roads Congress, July-September 2013

172 Raghuram And Chowdary on
Even though there are restrictions on usage of fly-ash in
SMA by IRC, better rutting resistance could be observed
in this study by using fly-ash as filler in combination
with either fibres or modified binders when compared
to the control mix where fibres, modified binders,
and fillers were not used. Usage of fibres or modified
binders might have improved the rutting resistance of
SMA mixes. Considering the fact that there are many
advantages of using fly-ash as filler in asphalt mixes
as reported by Kumar et al. (2008), comprehensive
laboratory and field studies are necessary to relook into
the usage of fly-ash in SMA mixes.
6 CONCLUSIONS
The following conclusions are drawn from this study
i) There is a significant influence of the type of
filler, binder, and stabilizer on the draindown
of asphalt. The mixes with cement and fly-ash
fillers resulted in less draindown; this clearly
shows the importance of fillers in SMA mixes.
The smaller particles have more surface area
compared to larger aggregate which in turn can
hold higher asphalt in SMA mixes. SMA mixes
stabilized with FERP exhibited lower draindown
than the SMA mixes with other fibres. Modified
binders effectively reduced the draindown of
asphalt when compared to unmodified asphalt,
due to higher viscosity of the modified asphalt.
This shows that the modified asphalt also has
the potential to reduce the draindown. From the
experimental studies carried out in this study,
it could be observed that FERP effectively
reduces the draindown irrespective of the type
of filler.
ii) Marshall test parameters are function of the
type of filler, fibre and binder. Use of fillers,
fibres and modified asphalt resulted in higher
stability values when compared to the control
mix. FERP and jute fibres resulted in higher
stability values when compared to other fibres.
ii) Within the natural fibres, use of jute fibres as
stabilizer resulted in slightly better rut resistance.
Journal of the Indian Roads Congress, July-September 2013
Similarly, within the waste fibres, SMA mixes
with FERP as stabilizer resulted in slightly
better rut resistance. Natural fibres are more
effective in resisting permanent deformation
than waste fibres and artificial fibres. The use
of CRMB-60 as binder resulted in better rut
resistance when compared to PMB-70 and
VG-30 binders. Thus, it could be concluded
that jute fibres are effective in resisting the
permanent deformation when compared to all
other fibres used in this study.
iv) Jute which is a natural fibre and FERP which
is a waste fibre outperformed all other fibres
in terms of the draindown and performance
characteristics. These two low cost fibres might
effectively replace the costlier cellulose fibres
which has been the primary choice of stabilizer
in SMA mixes. This statement holds good only
after a comprehensive field evaluation of SMA
mixes manufactured with these two fibres as
stabilizers while subjecting the SMA mixes to
varied traffic and climatic conditions.
REFERENCES
1. 
ASTM D6390. (2005). Standard Test Method
for Determination of Draindown Characteristics
in Uncompacted Asphalt Mixtures, ASTM
International, formerly known as the American
Society for Testing and Materials, USA.
2. 
ASTM D6926. (2010). Standard Practice for
Preparation of Bituminous Specimens Using
Marshall Apparatus, ASTM International,
formerly known as the American Society for
Testing and Materials, USA.
3. 
ASTM D6927. (2006). Standard Test
Method for Marshall Stability and Flow of
Bituminous Mixtures, ASTM International,
formerly known as the American Society for
Testing and Materials, USA.
4. Brown, E.R., and Cooley, L.A. (1999).
Designing Stone Matrix Asphalt Mixtures
for Rut Resistant Pavements, NCHRP Report
 Performance Evaluation of Stone Matrix Asphalt (Sma) Using Low Cost Fibres
425, National Cooperative Highway Research
Program, Transportation Research Board,
Washington, D.C., USA.
5. 
Brown, E.R. and Haddock, J.E. (1997a). A
Method to Ensure Stone-on-Stone Contact in
Stone Matrix Asphalt Paving Mixtures NCAT
Report 97-02, National Centre for Asphalt
Technology, USA.
6. 
Brown, E.R. and Haddock, J.E. (1997b).
Method to Ensure Stone-on-Stone Contact
in Stone Matrix Asphalt Paving Mixtures
Transportation Research Record: Journal of
the Transportation Research Board, 1583,
pp. 11-18.
7. 
Brown, E.R. and Mallick, R.B. (1994). Stone
Matrix Asphalt Properties Related to Mixture
Design, NCAT Report 94-02, National Centre
for Asphalt Technology, USA.
8. Brown, E.R. and Mallick, R.B. (1995).
Evaluation of Stone-on-Stone Contact in Stone
Matrix Asphalt, Transportation Research
Record, 1492, Transportation Research Board,
Washington, D.C., pp. 208-219.
9. 
Brown, E.R. and Manglorkar, H. (1993).
Evaluation of Laboratory Properties of Sma
Mixtures, NCAT Report 93-05, National
Centre for Asphalt Technology, USA.
10. 
Brown, E.R. Haddock, J.E., Mallick, R.B.,
and Lynn, T.A. (1997a). Development of A
Mixture Design Procedure for Stone Matrix
Asphalt (Sma), NCAT Report 97-03, National
Centre for Asphalt Technology, USA.
11. 
Brown, E.R., Haddock, J.E., Mallick, R.B.,
and Lynn, T.A. (1997b). Development of A
Mixture Design Procedure For Stone Matrix
Asphalt (Sma), Journal of the Association of
Asphalt Paving Technologists, 66, pp. 1-24.
12. 
Brown, E.R. Mallick, R.B., Haddock, J.E. and
Bukowski, J. (1997c). Performance of Stone
Matrix Asphalt (Sma) Mixtures in the United
Journal of the Indian Roads Congress, July-September 2013
173
States, Journal of the Association of Asphalt
Paving Technologists, 66, pp. 426-449.
13. 
Haddock, J.E. Lilijedahl, B., Kriech, A.J., and
Huber, G.A. (1993). Stone Matrix Asphalt:
Application of European Design Concepts in
North America, Proceedings of the Canadian
Technical Asphalt Association, pp. 183-209.
14. 
IRC:SP:79. (2008). Tentative Specifications
For Stone Matrix Asphalt, Indian Roads
Congress, New Delhi, India.
15. Kamaraj, C. Jain, P.K., Sood, V.K. and
Sikdar, P.K. (2006). Design of Stone Matrix
Asphalt Using Different Stabilizing Additives,
Journal of the Indian Roads Congress, 67(1),
pp. 107-114.
16. 
Kamaraj, C. Kumar, G., Sharma, G., Jain, P.K.
and Babu, P.V. (2004). Laboratory Studies
on The Behaviour of Stone Matrix Asphalt
Vis--Vis Dense Graded Bituminous Mixes
Using Natural Rubber Powder (Wet Process),
Highway Research Bulletin, Indian Roads
Congress, 71, pp. 39-59.
17. Kandhal, P.S. and Chattaraj, R. (2009).
Discussion on Use of Mineral Filler in
Bituminous Mixtures in India, Indian
highways, Indian Roads Congress, 37(5),
pp. 65-68.
18. 
Kumar, P. Chandra, S., and Bose, S. (2004).
Permanent Deformation Characteristics of
Asphaltic Concrete Versus Stone Matrix
Asphalt, Highway Research Bulletin, Indian
Roads Congress, 71, pp. 77-86.
19. 
Kumar, P. Chandra, S., and Bose, S. (2007).
Laboratory Investigations on Sma Mixes With
Different Additives, International Journal of
Pavement Engineering, 8(1), pp. 11-18.
20. 
Kumar, P., Mehndiratta, H.C., and Singh, V.
(2008). Use of Fly Ash in Bituminous Layer
of Pavement, Indian Highways, Indian Roads
Congress, 36(8), pp. 41-50.
 Raghuram And Chowdary on
174 Performance Evaluation of Stone Matrix Asphalt (Sma) Using Low Cost Fibres
21. 
Ministry of Road Transport and Highways
(MORTH). (2001). Manual For Construction
and Supervision of Bituminous Works (Fourth
Revision), Indian Roads Congress, New
Delhi, India.
22. Muniandy, R. and Huat, B.B.K. (2006).
Laboratory Diameteral Fatigue Performance
of Stone Matrix Asphalt with Cellulose Oil
Palm Fiber, American Journal of Applied
Sciences, 3(9), 2005-2010.
23. 
Kandhal, P.S. (2002). Designing and
Constructing Sma Mixtures: State of the
Practice, Quality Improvement Series 122
(Revised Edition), National Asphalt Pavement
Association, Lanham, Marland, USA.
24. 
Prowell, B.D. and Hurley, G.C. (2008). Stone
Skeleton Asphalt: Concept Evaluation NCAT
Report 08-02, National Centre for Asphalt
Technology, USA.
25. 
Prowell, B.D. Cooley, L.A. and Schreck, R.J.
(2002). Virginias Experience With 9.5-mm
Nominal Maximum Aggregate Size Stone
Matrix Asphalt, Transportation Research
Record: Journal of the Transportation Research
Board, 1813, 133-141.
26. 
Prowell, B.D., Watson, D.E., Hurley, G.C.,
and Brown, E.R. (2009). Evaluation of Stone
Matrix Asphalt (Sma) for Airfield Pavements,
AAPTP Report 04-04, Airfield Asphalt
Pavement Technology Program (AAPTP),
Auburn University, USA.
The views expressed in the paper are personal views of the Authors. For any query, the author may be contacted at: E-mail: chowdaryazad@gmail.com
Journal of the Indian Roads Congress, July-September 2013
27. 
Punith, V.S., Sridhar, R., Bose, S., Kumar, K.K.
and Veeraragavan, A. (2004).Comparative
Studies on the Behaviour of Stone Matrix
Asphalt and Asphalt Concrete Mixtures
Utilizing Reclaimed Polyethylene, Highway
Research Bulletin, Indian Roads Congress, 71,
pp. 61-76.
28. Scherocman, J.A. (1991). Stone Mastic
Asphalt Reduces Rutting, Better Roads,
61(11), pp. 26-27.
29. 
Scherocman, J.A. (1992). Construction of
Stone Mastic Asphalt Test Sections in the
U.S., Journal of the Association of Asphalt
Paving Technologists, 61, pp. 642-664.
30. Sharma, B.M., and Kamaraj, C. (2008).
Construction and Performance Evaluation of
Experimental Sections Laid on Urban Roads
Using Stone Matrix Asphalt Mixtures, CDROM,
Proceedings of the 23rd Australian Road Research
Board Conference, Adelaide, Australia.
31. 
Suchismita, A., Panda, M., and Chattaraj, U.
(2010). Stone Matrix Asphalt for Paving
Applications, Proceedings of the International
Conference on Developments in Road
Transportation, Rourkela, India, pp. 392-401.
32.  atson, D.E., Masad, E., Moore, K.A., Williams,
W
K. and Cooley, L.A. (2004).Verification
of Voids in Coarse Aggregate Testing:
Determining Stone-on-Stone Cont Act of
Hot-Mix Asphalt Mixtures, Transportation
Research Record: Journal of the Transportation
Research Board, 1891, pp. 182-190.
 Paper No. 596

Longitudinal and Transverse Analysis of Multi-Cell

Pre-stressed Concrete Box Girder
Rajesh Ranjan Rele* and S.Talukdar**

Abstract
In present work, multi cell pre-stressed concrete box girder with simply supported ends has been analysed using finite element models. SAP2000 bridge
modeller has been employed for modelling box girder with pre-stressing cables. Pre-stressing force and losses have been included in the modelling.
An iterative procedure for time dependent losses such as due to creep and shrinkage has been developed and used in the bridge analysis using the afore
mentioned software. To perform transverse analysis of box girder, simplified models are developed in Finite Element framework and the results have been
compared with conventional moment distribution and effective width method used in design practice. It has been found that rigid link element model which
is connected between the transverse frames, is able to predict improved behaviour than simple frame element. However, for live load analysis of three-cell
pre-stressed concrete box girder, three dimensional Finite Element frame model gives closer estimate to the existing effective width method.

1 INTRODUCTION such as high torsional rigidity, high bending stiffness,

The spanning of bridges started with simple slabs.
As the span increased, the design depth of slab also
increased. It is known that the material near the centre
of gravity contributes very little for flexure and hence
could be removed. This led to beam and slab system.
As the width increased more longitudinal beams are
required leading to reduction in stiffness in transverse
direction. Under high transverse bending the webs do
no remain in their original position. To keep the webs
in original position, the bottom bulb of the webs is to be
tied together. This led to evolution of box girder.
A box girder bridge superstructure consists of a deck
slab, two or more vertical or inclined webs and a
bottom slab which results in a single or a multi-cell
rectangular or trapezoidal cross-section. Recently box
girder has got wide acceptance in flyover bridge system
due to their better stability, serviceability, pleasing
appearance and most importantly better structural
efficiency than T-beam or I-girder. The box girder can
also be found in portal frame bridges, arch, box string
girder bridges and cable-stayed and suspension bridges
of all kind. Moreover, it has got some other features
high transverse strength and economy due to hollow
section. Concrete box girders are usually cast-in-situ
or precast in segments erected on launching frame and
then prestressed. The decks could be of steel, reinforced
concrete, or prestressed concrete. The cross section
of box girder may take the form of a single cell (one
box), multiple spine (separate boxes) or multicell with
a common flange (cellular shape). Box girder bridges
are mainly chosen for span ranging from 20 m to
40 m for reinforced concrete bridges and 40 m to 100 m
for prestressed concrete bridges. For spans more than
100 m cable supported bridges are economical.
However spans up to 150 m could be covered with
box girder cross-section using cantilever or other
construction method. Multiple cellular box girder
becomes transversely stiff and accordingly warping
stresses will be minimum. The behaviour of box girder
is characterized by the deformability of the cross-
section i.e. it may be a deformable or non-deformable
section.
The cross-section of a box girder may be non-
deformable if it has relatively thick webs and slabs and
closely spaced rigid diaphragms. In such case box

* Post Graduate Student

** Professor } Department of Civil Engineering, Indian Institute of Technology, Guwahati-781 039,
E-mail-staluk@iitg.ernet.in
Written comments on this Paper are invited and will be received upto 30th October 2013.

Journal of the Indian Roads Congress, July-September 2013

176 Rele & Talukdar on
girder behaves like a simple beam and may be analyzed
using simple beams flexural theory and St. Venants
torsion theory. The various structural actions involved
in box girders are exure, shear, torsion, warping,
and distortion, in which the effects of distortion and
warping are particularly signicant in thin-walled box-
girder bridges. The behaviour of box girder Bridge
with deformable cross-section occurring in most of the
cases is more complex than the non-deformable case.
In this case the box girder analysis and design should
take into consideration stresses due to longitudinal
bending, shear force, torsion, and distortion of cross-
section, transverse bending and shear lag. Complex
interaction of these structural actions will occur under
eccentric loading conditions. Moreover, variation of
longitudinal flexural stress along the flange length,
known as shear lags is also a subject of concern to the
designer. Research on box girder bridges started long
back using Saint-Venant torsion theory and Vlasovs
thin walled beam theory as reported by Megson1 in one
his popular books. Initially, researchers2-4 started using
analytical methods, solving governing differential
equations for extension, flexure, torsion, distortional
warping and shear lag effect of thin walled section
beams with given boundary conditions. Benscoter5
derived differential equations and boundary conditions
relating warping displacements and rotations to applied
torsional loads for thin walled non-uniform multi-cell
beams. Li6 attempted to solve a thin walled box girder
bending-torsion problem including initial curvature.
Box girder bridge with diaphragm and openings have
been investigated by Sargious et.al7. Maisel8 extended
Vlasovs thin-walled beam theory to treat torsional,
distortional, and shear lag effects of straight, thin-walled
cellular box beams. Zhang9 established the governing
differential equation for distortion in the trapezoidal
single-cell girder with varying depth. A review paper
by Khaled and Kennedy10 revealed that for practical
applications in some countries orthotropic plate theory
has been recommended for a box girder bridge. In this
method stiffness of flanges and girders are lumped
into an orthotropic plate of equivalent stiffness and
stiffness of diaphragm is distributed over the girder
length. Canadian Highway Bridge Design Code11
has recommended this method for straight box girder
Journal of the Indian Roads Congress, July-September 2013
section. To overcome the restriction of orthotropic plate
theory for box girder section or voided slab, Hambly and
Pennells12 has idealized multicellular super structures
using grillage analogy. One main drawback of grillage
analogy is difficulty in correctly representing torsional
stiffness of a closed cell. An approximate method
by considering torsional stiffness of an equivalent I
beam has been tried in grillage analogy by Evans and
Shanmugam13.
Gradually, numerical techniques along with high speed
computations have gained popularity and were used for
practical design. Among these, finite element method
has been used in majority of cases for detail analysis.
Williums and Scordelis14 presented an elastic analysis
of cellular structures of constant depth with arbitrary
geometry in plan using quadrilateral elements. Fam
and Turkstra15 described a nite element scheme for
static and free-vibration analysis of box girders with
orthogonal boundaries and arbitrary combinations of
straight and horizontally curved sections using a four-
node plate bending annular element with two straight
radial boundaries, for the top and bottom anges,
and conical elements for the inclined web members.
Composite box Girder bridge has been analysed with
the help of finite element method by Moffat and Lim16
incorporating interaction with respect to distribution of
shear connector. Dritsos17 studied the distortion of thin-
walled concrete trapezoidal single- and double-cell box
beams subjected to eccentric loading at an arbitrary
transverse position on the top ange. The governing
equation for the distortional angle was developed, and
the mathematical analogy with the differential equation
for a beam-on-elastic-foundation was established.
A discrete finite difference energy method has been
developed by Mishra et.al18 as a suitable alternative to
finite element method which has been used for straight
box girder bridges. Shushkewich19 showed that the
actual 3D behaviour of straight box girder bridge, as
predicted by a folded plate, finite strip or finite element
analysis can be approximated by using some simple
membrane equations in conjunction with a plane
frame analysis. Kenneth20 used computer program for
transverse analysis of strutted box girder bridges. The
program allows deck prestressing with provision of
transverse flexure, shear and torsion using the principle
 Longitudinal and Transverse Analysis of Multi-Cell Pre-stressed Concrete Box Girder 177

of folded plate analysis. General criteria, not very are discretized into smaller segments. Discretization
specific to thin walled section prestressed concrete points are located at either end of each discretization
girder has been laid in IRC:11221. segment. When located within the bounding box of a
structural member, interpolation constraints connect
With increase in volume of traffic multilane flyover and these discretization points to all joints within that
bridges has gain a lot of importance. In congested cities member.
where space is very little, cast-in-situ construction
is advisable. In that case, multicell box girder of
prestressed concrete has become choice of most of
the designers from strength, economy and aesthetic
point of view. However, due to complex structural
action involved, codified procedure for the analysis
of box girder are not readily available to the designer.
Moreover, literature survey reveals that analysis of
prestreesed concrete multicell box girder has not been
adequately addressed, although substantial reports
exist on reinforced concrete girders. In what follows
Fig. 1 Finite Element model of Box girder created in SAP 2000
that present study focussed on the analysis of multicell
prestressed concrete box girder using finite element 2.1 Geometry and Material Property
package ( SAP2000 bridge modeller). Other softwares
do not have provision to model the cables and enable The bridge cross-section is of three-cell hollow pre-
prestressing them. In the present work, prestress force stressed box with two ends being simply supported as
in cables as well as the losses that would occur in shown in Fig. 2. Because of end conditions chosen,
the cables have been duly considered. Since the time there is no restraint against warping at the ends and
dependent losses due to creep and shrinkage would not therefore warping stress will be insignificant. Three
be directly implemented in the available commercial Daiphrgms are placed at an interval of 0.25 times
software, an iterative procedure has been developed the effective span. Provision of cross diaphragm also
and linked with the software used. The study compares reduces distortional effects. In the present analysis,
the results of different models with those obtained warping and distortional effects on stresses are ignored.
by conventional frame analysis22 and effective width
The bridge has span of 60 m and width of 15 m. The
method available in codes of practice23. A three cell
dead weight of the box girder is taken into account by
prestreeded concrete box girder section has been taken
SAP 2000 Bridge Modeller by assigning the material
to illustrate the method.
property of concrete of density 25kN/m3. The grade of
2 MODEL OF PRE-STRESSED CONCRETE concrete used is M45.
BOX GIRDER

The box girder has been analysed as an area object

using Bridge Modeller of SAP2000 as shown in
Fig. 1. The entire section of the box has been modelled
using shell elements. As compared to the spine model
the area model is preferable as it gives forces in the
individual girder which is not the case with the stick
model. Tendons are line objects which are embedded
within other objects (frames, shells, solids) to simulate
the effects of prestressing and post-tensioning. Tendons Fig. 2 Detail of Cross-section of three cell box girder

Journal of the Indian Roads Congress, July-September 2013

178
2.2 Layout of Cables
The box has 5 cables in each web, 12 long cables in
soffit and 20 short cables in soffit giving a total of 52
cables. The long cables will extend up to the support
into anchorage block whereas short cables are curtailed
at certain location by means of blisters provided in
soffit slab. The layout of cables is shown in Fig. 3.
Fig. 3 Model of Box girder showing cables in SAP 2000
2.3 Order of Stressing
The Figs. 4 and 5 show the order of stressing of webs
and soffit cables and short cables respectively. Since the
Rele & Talukdar on
major load is transferred through web, the web cables
are stressed first followed by long cables in soffit and
lastly the short cables in blister provided at suitable
interval.
2.4 Pre-Stressing and Losses
All pre-stressing strands have been modelled as
elements in SAP 2000 Bridge Modeller. At both ends
of pre-stressing strands a jacking force of 76.5% of
breaking load is applied. The software is able to estimate
only friction loss, slip loss and elastic shortening loss.
However, the estimation of time dependent losses such
as due to creep and shrinkage needs careful attention
and is not taken by SAP 2000 software in its usual way
of performing the analysis. In the present study, creep
and shrinkage loss values are given as input to the finite
element model by working out these values based on a
simple iterative process outlined in the present paper.
All the cables are stressed when concrete has attained
75% of specified cube strength (after 8 days). Losses
due to creep and shrinkage are calculated at the end of
28 days maturity. In a post tensioned girder, the total
losses may amount to 20-25 percent, out of which creep
and shrinkage loss remains usually in the range of 10-
13 percent24. The iterative process of finding creep and
shrinkage loss has been described below:

Step-I: Stress at mid-span is worked out on stretching

all cables due to dead load + pre-stressing force, taking
account of the loss due to slip and friction (P01), as
input to finite element model.

Step-II: A certain value of pre-stress loss due to creep

Fig. 4 Order of stressing for web and soffit cables
and shrinkage, p percent in the range of 10-13% may
be assumed at the onset. Initially at transfer creep and
shrinkage loss may be assumed as % (usually in the
range of 3-4%). Stresses due to dead load+(1-/100) of
prestrssing force (P01).

Step-III. The average stress of step-I and step-II is

calculated. Stresses due to creep loss and shrinkage
loss using the formula given in IRC:18-200023 are found
out. Total percentage loss due to creep and shrinkage
shall not exceed the assumed value. If exceeded,
step- II is to be repeated by assuming a new value, else
Fig. 5 Order of stressing for short cables
step-IV is followed.

Journal of the Indian Roads Congress, July-September 2013

 Longitudinal and Transverse Analysis of Multi-Cell Pre-stressed Concrete Box Girder
Step-IV: The balance loss, i,e =(p-)% is assumed.
Stresses for dead load +superimposed dead load
(railing, parapet and wearing surface load) +(1-/100)
of pre-stressing force (P01) are calculated. Average loss
in step-III and IV is taken. If average loss is more than
the assumed one, step-II is to be repeated by assuming
a new value of percentage loss, otherwise step-V is to
be followed.
Step-V: The creep loss after 28 days till three months
(before placing super imposed dead load) plus loss
after three months till infinite time is calculated as per
procedure laid down in IRC code21. Shrinkage is found
out by multiplying shrinkage strain (after 28 days in
table 3 of IRC 18) with modulus of elasticity of steel.
Step-VI: Knowing the creep and shrinkage losses, it is
enhanced by 20% for higher time dependent loss.
Step-VII: Pre-stressing steel with constant elongation
maintained over a period of time shows gradual
decrease of stress, which depends both on time duration
and amount of initial pre-stress force. This loss, known
as Relaxation loss is taken into account according to
IRC procedure21.
Step-VIII: Final losses are now calculated and given as
input to the software.
3 LONGITUDINAL ANALYSIS
The analysis for stresses along the span of the box girder
has been performed for self weight, superimposed
dead load and live load. Stresses at transfer and service
conditions have been evaluated and the results are
presented along the span at 0.1L interval (L=effective
span) of the girder. For the calculation of stresses,
dead load as per section detail and concrete of density
25kN/m3 has been taken. Superimposed dead load
has been taken as 45 kN/m. This includes wearing
coat, railing, cash barrier and service at cash barrier.
For live load analysis, Class 70R wheeled loading has
been considered as per IRC:6 25. Since it is a four lane
bridge two lanes of 70R is used for longitudinal live
load analysis. Multistep static analysis option is used in
which the program creates a different pattern of loading
Journal of the Indian Roads Congress, July-September 2013
179
for each time step. The load applied to the structure is
determined by longitudinal position of each vehicle in
its lane at the current time from its starting position.
For each individual concentrated load, consistent joint
loads are calculated at the corner of any loaded shell
element on the deck.
4 TRANSVERSE ANALYSIS
Transverse analysis of box girder has to be made
for designing the box section in order to determine
reinforcement required for taking care of flexural
stresses in the transverse direction. In the normal beam
and slab design only the slab is designed as a continuous
slab over the beam supports. But in box section, since
the webs are restrained at bottom also, the entire cross-
section becomes effective as a box in resisting the
transverse forces induced by wheel loads. The section
is designed as a frame. This is normally done using
a computer programme in view of the complications
involved in manual computations.
4.1 Modelling of Box Girder for Transverse
Analysis
In practical design offices the general trend is to model
the box girder as a simple two dimensional box frame.
Such an approach is very conservative. Modelling
the box girder for transverse forces becomes tedious
when three dimensional modelling of boxes with shell
elements is adopted, especially for moving load analysis.
In the present study, the frame element has been chosen
mainly to save computational time while reaching
acceptable accuracy for practical implementation. The
study also considers creation of different finite element
frame models to predict the box cell behaviour and to
compare with those obtained by conventional frame
analysis.
4.1.1 Model Discretization
The basic idea is to create the centre line model from
the actual cross-section of the box girder as shown in
Fig. 3. The deck, web and soffit are discretized at some
suitable interval depending on which part of box is
subjected to heavier stresses and moments. The deck
is discretized at 0.25 m interval throughout the width
 180 Rele & Talukdar on
transverse frames However, here top of the web
as well as bottom is connected to transverse
frame by rigid link.

A schematic diagram of a finite element model for

transverse analysis (A, B, C and D) is shown in Fig. 7.
Fig. 6 Centreline model with discretization of deck, web and
The models differ from each other in the properties and
soffit in SAP2000
location of rigid link as mentioned above.
of box. The web is discretized at 0.5 m and soffit at
1 m interval with bottom haunches at 0.225 m interval.
Fig. 6 shows the centerline model with discretization of
deck, web and soffit in CSI BRIDGE.

4.1.2 Types of Models

The conventional practice used by design engineer for

a box section was that of moment distribution. This
can be now easily performed on iterative computer
programme using simple frame model of the box.
However, live load analysis is still carried out using the
effective width concept as recommended in IRC 625. In
the present paper, authors have created four different
frame models (A, B, C and D) based on the connecting Fig.7 Finite element frame models with rigid link of different
elements of one frame to the adjacent frames by rigid properties (A, B, C and D)
links. The models differ from each other in the properties 5. RESULTS AND DISCUSSIONS
and location of rigid link as mentioned below. The
results of four different models are then compared with 5.1 Longitudinal Analysis
that of the traditional method to examine the reliability
of traditional methods as compared with that of simpler 5.1.1 Stresses at Transfer of Pre-stress Force
and faster three dimensional Finite Element Models.
The pre-stress force is applied when concrete has
i) Model A: In this model the links elements gained 70% of its cube strength that is after 8 days
etween the transverse frames have been of concreting. Table 1 shows the stresses due to dead
assigned weight per unit volume (). load and prestress force (without time dependent loss
which are creep and shrinkage). Tabulated values are
ii) Model B: Here, the link elements between flexural stress at 0.1L m interval up to the mid span
the transverse frames have been considered (L is effective span length taken as 60 m).
weightless as compared to the total weight of
Table 1 Stresses in Box Girder at Transfer of
the frame.
Prestress Force
iii) Model C: In this model the link elements (Negative Value Indicates Compression)
between the transverse frames have not been Section at Top (N/mm2) Bottom (N/mm2)
assigned weight per unit volume () and only 0.1L -3.34 -4.22
top of the web is connected to transverse frame 0.2L -5.6 -3.53
by rigid link. 0.3L -7.24 -4.82
0.4L -8.52 -3.52
iv) Model D: The modelling, again, has been done
0.5L -9.02 -2.94
with the weightless link elements between the

Journal of the Indian Roads Congress, July-September 2013

 Longitudinal and Transverse Analysis of Multi-Cell Pre-stressed Concrete Box Girder 181

5.1.2 Stresses at Service

The stresses at service are due to dead load,
superimposed dead load, live load and prestress force
(with time dependent loss which are due to creep and
shrinkage). It can be seen from Table 2 that the entire
box is under compression at service stage which is as
expected.
Fig. 9 Dead load bending moment (for model B (The number
Table 2 Stresses in Box Girder at Service shown in figure are Bending moment in kN.m/m)
(Negative Value Indicates Compression)
In model C, the transverse frames are connected by
Section at Top (N/mm2) Bottom (N/mm2) rigid link at top of all four webs of three cell box girder.
0.1L -4.30 -3.39 This makes the structure to behave uniformly as one
unit which is precisely expected for dead load case.
0.2L -7.06 -1.93
The result has been shown in Fig. 10.
0.3L -9.01 -2.81
0.4L -10.51 -1.23
0.5L -11.10 -0.54

5.2 Transverse Analysis

5.2.1 Dead Load Analysis

The models A , B, C and D have been analysed for dead Fig. 10 Dead load bending moment for model C (The number
load and the bending moment (B.M) curve at mid-span shown in figure are Bending moment in kN.m/m)
section is shown in Figs. 8,9,10 and 11. In model A,
The model D has rigid links at top and bottom of the
the link elements between the transverse frames have
webs of box girder. This model is said to be refinement
been assigned weight per unit volume and hence during
of above other models and expected to predict close
dead load analysis they contribute more moments at the
behaviour to actual structure. The result is presented in
nodes as observed in Fig. 8.
Fig. 11. Since rigid links are also provided at bottom,
the soffit moments are improved which was not the
case with model C.

Fig. 8 Dead load bending moment for model A (The number

shown in figure are Bending moment in kN.m/m)
Fig. 11 Dead load bending moment for model D (The numbers
shown in figure are Bending moment in kN.m/m)
In model B, the link elements are simple frame elements
with mass participation factor being zero. Therefore,
5.2.2 Superimposed Dead Load Analysis
magnitude of the bending moments have been
appeared (Fig. 9) less compared to the dead load results The models A, B, C and D have been analysed
of model A. for superimposed dead load (railing loads and wearing

Journal of the Indian Roads Congress, July-September 2013

182
coat loads).The models A and B both seem to
predict incorrect behaviour when superimposed
dead load case is analysed. The result of model
A for superimposed dead load has been shown in
Fig. 12. When box girder is analysed for general
items of loading, it should have more moment
at support A than at B considering the physical
behaviour of box.
Rele & Talukdar on
Fig.15) from IRC 6:200025 has been adopted.
For 400kN bogie load the edge distance is
1.2 m. The vehicles are moved in the transverse
direction using moving load analysis option provided
in the software.

Fig. 15 400kN bogie load

For maximum moment at the junction of exterior web

Fig.12 Superimposed Dead Load bending moment for model A
(The numbers shown in Figure are Bending moment in kN.m/m)
In model C, the transverse frames are connected by
rigid link at top of all four webs of three cell box girder.
This makes the structure to behave integrally as one
unit which seems to reflect the true structural action.
The result of model C and model of D are shown in
Fig.13 and Fig. 14 respectively.
and deck (support A), and also at the junction of interior
web and deck (support B), placement of live load has
been shown in Fig. 16 and Fig. 17 respectively. The
distributions of moment along the transverse section
for model A and B have been represented in Fig. 18
and Fig. 19 respectively.

Fig. 16 Position of 400kN bogie load for maximum support

bending moment at A (all linear dimensions are in metre)

Fig. 13 Superimposed Dead Load bending moment for model C

(The number shown in figure are Bending moment in kN.m/m)

Fig 17 Position of 400kN bogie load for maximum support

bending moment at B (all linear dimensions are in metre)

Fig. 14 Superimposed Dead Load bending moment for model D

(The numbers shown in figure are Bending moment in kN.m/m)

5.2.3 Live Load Analysis

Fig. 18 Live load Bending moments at exterior web-flange
To analyse the box girder for span moments junction for model A and B (The number shown in figure are
standard vehicle of 400kN bogie (shown in Bending moment in kN.m/m)

Journal of the Indian Roads Congress, July-September 2013

 Longitudinal and Transverse Analysis of Multi-Cell Pre-stressed Concrete Box Girder
Fig. 19 Live load bending moment at interior web-flange junction
for model A and B (The number shown in figure are Bending
moment in kN.m/m)
The distribution of bending moment along the
transverse section for the maximum effect at the
middle of exterior box span (AB) and interior box
span (BC) has been considered for the position of live
load as illustrated in Fig. 20 and Fig. 21. Fig. 22 and
Fig. 23 represent the moment for the live load position
in the middle of exterior box span for model A and
B respectively. It may be noted that only difference
in model B was the consideration of weightless rigid
links between the transverse frames, hence as expected
the results of live load analysis remain unaffected.
Fig. 20 Position of 400kN bogie load for maximum bending
moment at mid-span AB (All linear dimensions are in metre)
Fig. 21 Position of 400kN bogie load for maximum bending
moment at midspan BC (All linear dimensions are in metre)
Fig. 22 Live load bending moment for maximum effect of live
load in exterior box span for model A and B (The number shown
in figure are Bending moment in kN.m/m)
Journal of the Indian Roads Congress, July-September 2013
183
Fig. 23. Live load bending moment for maximum effect of live
load in interior box span for model A and B (The number shown
in figure are Bending moment in kN.m/m)
Further, model C and D have been analysed for live
load effect in order to produce maximum moment at
the junction of exterior web and flange as well as at the
junction of interior web and flange as per the placement
of live load shown in Fig. 16 and Fig. 17 respectively.
In addition to this, the span moment in exterior and
interior boxes has been investigated for the live load
placing shown in Fig. 20 and Fig. 21 respectively as
stated earlier in case of model A and B. The models C
and D exhibit similar behaviour under the action of live
load. The results are presented in Fig. 24 and Fig. 25
for exterior and interior box junctions respectively.
Fig. 24 Live load bending moment (kN-m) at exterior web-flange
junction for model C and D (The numbers shown in figure are
Bending moment in kN.m/m)
Fig. 25 Live load bending moment (kN-m) at interior web-
flange junction for model C and D (The number shown in
figure are Bending moment in kN.m/m)
The analysis has been again performed to examine
the effect of live load for maximum span moment in
exterior and interior boxes of the model C and D, as
per the load position shown in Fig. 20 and Fig. 21
mentioned earlier. Live load effect for models C and
D do not exhibit changing behaviour. The results for
184 Rele & Talukdar on
exterior and interior box span for both the models C and width). Total bending moment per metre width of the
D are presented in Fig. 26 and Fig. 27 respectively. flange at kth location is then calculated by summing up
the product of wheel load and corresponding ordinate
of influence line diagram for bending moment at kth
location and then dividing the result of summation by
the effective width of dispersion.

Fig. 26 Live load bending moment for maximum effect of live

load in interior box span for model C and D (The numbers shown
in figure are bending moment in kN.m/m)

Fig. 27 Live load bending moment for maximum effect of live

load in interior box span for model C and D (The numbers shown
in figure are bending moment in kN.m/m)

5.3 Analysis by Conventional Method

Fig. 28 Influence line for moment at support A
To evaluate the transverse forces due to dead and
superimposed dead load, moment distribution method22
has been followed. Live load effect due to 400 kN
boggie load moving in transverse direction has been
evaluated from influence line diagrams. Influence
line diagram for the moment at exterior web-deck
junctions (support A), interior web-deck junctions
(support B), middle of exterior and interior boxes have
been presented in Fig. 28 to Fig. 31. Use of influence
line ordinate has been made to calculate bending
moment at different locations of centre of wheel as
Fig. 29 Influence line for moment at support B
given earlier in Figs. 17, 18, 20 and 21. For flange of
the box girder supported on two edges and carrying
concentrated loads, the maximum live load bending
moment is calculated by considering the effective
width of the slab. This effective width, also called the
effective width of dispersion is measured parallel to
the supporting edges of the slab and can be estimated
by using the following formula in IRC 112:201121. The
extent of effective width depends on the location of the
wheel load with reference to support and the dimension
of the slab. Thus concentrated load virtually transforms
into a uniformly distributed load, distributed along some Fig. 30 Influence line for moment in mid span AB
length (dispersed length along span) and width (effective

Journal of the Indian Roads Congress, July-September 2013

 Longitudinal and Transverse Analysis of Multi-Cell Pre-stressed Concrete Box Girder 185

load cases, the model B predicts a slightly closer

behaviour to the simple frame analysis as observed in
Table 3. For dead load and superimposed load, model
B follows similar trend with model A and closer to the
values obtained by conventional method. The model
B may only be used for analysis of dead load and
superimposed dead load.

Table 3 Comparison of Bending moment (kN.m/m)

for Dead Load and Superimposed Dead Load
Location of Finite Element Method Moment
point Distribution
Fig. 31 Influence line for moment in mid span BC Model Model Model Model
method
A B C D
5.4 Comparison of Results Junction of 33.9 24.9 27.5 25.7 26.8
exterior web
The results obtained by three dimensional frame and flange
model (Model A, B, C and D) and by conventional Junction of 26.4 17.4 10.5 12.5 14.8
method have been compared in Table 3 and Table 4. interior web
Table 3 gives the Bending Moment for dead load and and flange
superimposed dead load whereas Table 4 presents Mid span of 6.8 3.8 4.3 3.7 3.9
the Bending Moment for live load cases. It may be exterior box
mentioned that longitudinal and transverse grid lines Mid span of 6.7 3.3 5.0 4.9 3.6
have been kept in three dimensional frame model at interior box
closer spacing to allow the moving load to get dispersed
The transverse frames are connected by rigid link at
in the direction perpendicular to the movement of the
top of all four webs of three cell box girder in model
load. This would result in more accurate results when
C. This makes the structure to behave uniformly as
compared to conventional method. For various trial
one unit which is precisely expected for dead load and
position of live load, the present case yields maximum
superimposed load cases. Model C reasonably agrees
bending moment at the junction of interior web and
well with conventional method as far as dead load and
flange which will govern the design and recommended superimposed dead load analysis are concerned, but
practice is to use haunches at the junctions, which has deviation is more compared to model B. For live load
been considered in the present case. In model A, the case, the model C yields improved result if compared
link elements between the transverse frames are been to the effective width method. In model D, the rigid
assigned weight per unit volume and hence during dead links are assigned at top and bottom of the webs of
load analysis they contribute more moments as reflected box girder. The model D predicts more closely to the
in Table 3. For live load case as seen from Table 4, the results obtained by effective width method. Since
model A does not agree well with the manual method rigid links are also provided at bottom, the soffit
of transverse deck analysis. The result proves that moments are improved which was not the case with
model A is not a suitable option for three dimensional model C. For the different cases considered in the
transverse modeling of Box girder. present study, it is apparent that model B and model
D are better option for three dimensional transverse
In model B, the link elements are simple frame elements modeling of Box girder for dead load and live load
without mass. For dead load and super imposed dead analysis respectively.

Journal of the Indian Roads Congress, July-September 2013

 186 Rele & Talukdar on
Table 4 Comparison of Bending Moment 2. Kristek, V, Tapered Box Girders of
(kN.m/m) for Live Load Deformable Cross Section, Journal of
Structural Engineering, ASCE, 96(8),
Location of Finite Element Method Effective
pp. 1761-1793.
point width
Model Model Model Model
A B C D
method 3. Maisel, B.I., Review of Literatures Related
to the Analysis and Design of Thin Walled
Junction of 29.9 29.9 56.1 56.1 58.8
Beams, Tech. Report No. 42440, Cement and
exterior web
Concrete Association, London, 1970.
and flange

Junction of 48.3 48.3 64.2 64.2
interior web
and flange
Mid span of 31.5 31.5 28.5 28.5
exterior box
Mid span of 41.5 41.5 34.0 34.0
interior box
6 CONCLUDING REMARKS
In the present paper, three-cell prestressed concrete box
girder has been analysed using finite element software
for longitudinal and transverse effect considering dead
load, super imposed dead load and live load as per
Indian standard. The time dependent prestressing loss
has been incorporated in the software using an iterative
process. Four different types of models have been
considered for transverse analysis and compared with
conventional method adopted in design practice. From
the different finite element models, it was found that
the model having mass assigned to the link elements
between the transverse frames gives higher moments
at the node points and thus overestimates the forces
in transverse direction. With weight less rigid links
connected at top and bottom at all four webs the box
girder behaves uniformly under all loading conditions
and compares well with the conventional methods.
The result obtained by effective width method shows
reasonably close behaviour with transverse frame
element model with rigid links connected at top and
bottom of the web for live load condition.
REFERENCES
1. Megson, T.H.G., Linear Analysis of Thin
Walled Elastic Structures, John Willey and
Asons, Newyork, 1974.
67.5 4. Maisel, B.I., Concrete Box Girder Bridges,
Journal of Structural Engineering,51,
pp. 363-376, 1973.
32.8
5. Benscoter, S,U., A Theory of Torsion Bending
for Multicell Beams, Journal of Applied
36.4 Mechanics, 21, pp. 25-34, 1954.
6. Li, G, Torsion and Bending of Thin Walled
Box Girder with Initial Curvature, China Civil
Engineering Journal, 20 (10), pp. 65-70, 1987.
7. Sargious, M.A., Dilger, W.H., and Hwak,
H., Box Girder Bridge Diaphragms with
Openings, Journal of Structural Engineering,
ASCE, 105 (1), pp. 53-65, 1979.
8. Maisel, B.I., Analysis of Concrete Box
Beams Using Small-Computer Capacity,
Canadian Journal of Civil Engineering., 12 (2),
pp. 265-278, 1985.
9. Zhang, S., Distortional Stresses in Trapezoidal
Box Section Girder with Single Cell and
Varying Depth, China Civil Engineering
Journal, 20, pp. 62-73,1987.
10. Khaled, M.S. and Kennedy, J.B., Literature
Review in Analysis of Box Girder Bridges,
Journal of Bridge Engineering, ASCE, 7(2),
pp. 134-143, 2002.
11. Canadian Highway Bridge Design Code
(CHBDC), Ontario Ministry of Transportation
and Communications, Downsview, Ontario,
Canada, 2000.
12. Hambly, E.C. and Pennells, E., Grillage
Analysis Applied to Cellular Bridge Decks,
Structural Engineering, 53 (7), pp. 267-275,
1975.

Journal of the Indian Roads Congress, July-September 2013

 Longitudinal and Transverse Analysis of Multi-Cell Pre-stressed Concrete Box Girder
13. Evans, H. R and Shanmugam, N. E., Simplified
Analysis for Cellular Structures, Journal
of Structural Engineering, ASCE, 110 (3),
pp. 531-543, 1984.
14. William, K.J., and Scordelis, A.C., Cellular
Structures of Arbitrary Plan Geometry,
Journal of Structural Engineering, ASCE, 98
(7), pp. 1377-1394, 1972.
15. Fam, A.R. and Turkstra, C. J., A Finite Element
Scheme for Box Bridge Analysis Computers
and Structures,5, pp.179-176,1975.
16. Moffiat, K. R. and Lim, P. T. K., Finite Element
Analysis of Composite Box Girder Bridges
Having Complete or Incomplete Interaction,
Proceeding of Institution of Civil Engineers,
Part-2, 63(3), pp. 1-22, 1976.
17. Dritsos, S.E., Distortion of Concrete Box
Beams Due to Eccentric Transverse Loads.
Journal of Structural Engineering,117,
pp. 29-47,1991.
18. Mishra, P.K., Das, S. and Dey, S.S., Discrete
Energy Method for the Analysis of Right Box-
Girder Bridges, Computers and Structures,
43(2), pp. 223-235, 1992.
19. Shushkewich, K.W., Approximate Analysis
of Concrete Box Girder Bridge, Journal
The views expressed in the paper are personal views of the Authors. For any query, the author may be contacted at: E-mail: staluk@iitg.ernet.in
Journal of the Indian Roads Congress, July-September 2013
187
of Structural Engineering ASCE, 114(7),
pp. 1644-1657, 1988.
20. Kenneth, W.S., Transverse Analysis of
Strutted Box Girder Bridges, Journal of
Bridge Engineering ASCE,11(1), pp. 33-47,
2006.
21. IRC:112-2011, Code of Practice for Concrete
Road Bridge, Indian Roads Congress, New
Delhi, 2011.
22. Junnarkar, S.B., Mechanics of Structures-
Vol.II, Vivek Publication Pvt. Ltd, Bombay,
1974.
23. Raina, V.K., Concrete Bridge Practice-
Analysis, Design and Economics, Tata
McGraw Hill Publication, Third Edition,
1999.
24. Lin, T.Y. and Burns, N.H., Design of
Prestressed Concrete Structures, John Willey
and Sons, NewYork, 1981.
25. IRC:6-2000, Standard Specifications and
Codes of Practice for Road Bridges, Section-II-
Loads and Stresses, Indian Roads Congress,
New Delhi, 2000.
 Paper No. 597

WELL FOUNDATION CONSTRUCTION IN BOULDERY BED

R.K. Dhiman, VSM*

Abstract
Well foundation has been used for number of bridge piers in various parts of the world. Construction of well foundation in case of sandy soil strata is easy
task, whereas well foundation in case of bouldery bed is really a challenge for bridge engineers. There are number of cases where this type of foundation
has been used at bouldery bed sites but it has posed a challenge to bridge engineers due to heterogeneous character of the soil strata underneath. A case
study of a well foundation construction for bridges in bouldery bed strata has been discussed in this paper.

1 INTRODUCTION guidelines. In case the soil strata is comprises of rock

Foundation design and construction in bouldery bed
has its own challenges firstly due to non availability of
any reliable formula for bouldery bed scour and sinking
problem there on due to tough strata. Normally depth
is initially finalized as per Indian Roads Congress code
formula and revision is planned based on tough strata.
As the revision involved sizable financial effect on either
side, need has been felt to finalize the depth in such
strata based on greater examination. Foundation depth
be finalized keeping in view the construction technique
available viz, safety and stability requirement. Infact
in case variation in strata is encountered it should be
considered accordingly and revision on either side
be considered. Also the tough strata be considered
as engineering friendly. Whenever there is change
in properties of the soil, the design be amended
accordingly for completion of foundation. Based on
the experience of the Border Roads Organisation well
foundation construction on bouldery bed strata has
been discussed in this paper.
2 WELL FOUNDATION AN OVERVIEW
Correct assessment of scour is important task for the
over all safety of the bridge. Normally no tension is
permitted at the base of well foundation as per codal
underneath, the well is sunk into rock and rock level
is assumed as scour level, design is finalized as per
the codal provisions. However, in case of rock 20%
tension is permitted if need arises and well is suitably
anchored. Well foundations are used to ensure design
and hydraulic requirement. Shape of well can be square,
circular, double-D depending upon the site requirement
and construction methodology available, circular well
foundation is preferred. Foundations are taken to a
sufficient level below the maximum scour level to satisfy
design and stability requirement. However, during the
process of sinking if change in strata is observed, an
action is taken according to engineering consideration.
There are cases where based on the difficulties faced
in sinking the well, pneumatic sinking technique has
been used to expedite sinking. Pneumatic sinking is
used when the soil strata is tougher and is not possible
to excavate with conventional sinking arrangement.
Design parameters, to be considered for foundation,
are hydraulic and soil parameters. Stratification of soil
underneath plays key role in finalizing the design and
expediting the construction of foundation. Hydraulic
parameters are discharge, velocity, bed slope and soil
properties silt factor, bulk density, angle of internal
friction, angle of wall friction and cohesion which are
assessed based on geological condition of the strata.

* Director (Works & Budget), HQ (P) Vijayak (GREF), C/O 56 APO, E-mail: d_himan12@yahoo.com
Written comments on this Paper are invited and will be received upto 30th October, 2013

Journal of the Indian Roads Congress, July-September 2013

 Well Foundation Construction In Bouldery Bed 189

On finalization of this data, foundation levels are
decided. Correct assessment of scour in soil is very
important and the same is decided with the help of
available formula followed by judicious judgment and
if required model study is also carried out to assess the
likely scour during the service life of the bridge. This
will have vital bearing on completion and serviceability
of the bridge at later date.
3 CONSTRUCTION OF BRIDGE
BOULDERY BED
A flow chart of various step being followed
for well foundation construction is indicated in
Figs. 3 to 7.
(c) Generally cutting edges are made very strong
due to likely blasting anticipated at the site.
This is done as per IRC Code based on past
experience of foundation construction in similar
strata.
IN

a) Bouldery bed river strata generally encountered

is shown in Fig.1 & Fig. 2 Bouldery bed
generally located and near foothills. The soil
strata is generally heterogeneous in character.
Based on size and shape of the soil strata further
action on construction of bridge foundation is
taken.

Fig.1 Bouldery Bed River Fig. 3 Flow chart of activities for well foundation construction

Fig. 2 Soil Strata generally encountered in bouldery bed

b) Number of bridges has been constructed/under

construction on bouldery bed by Border Roads
Organisation (BRO) Viz.

Ans (J&K) Dalai (AP) Jaiswal (J&K) Passighat (AP)

Teesta (Sikkim) Lai (AP) Dimwe (AP) Lohit (AP)
Ishnu (Bhutan) Bandy Simekorong (AP) Fig. 4 Cutting edge being used for
(Bhutan) well foundation on bouldery bed

Journal of the Indian Roads Congress, July-September 2013

190 Dhiman on
DETALIS OF CUTTING EDGE
Fig. 5 Cross section of cutting edge generally being used where
large size of boulder are anticipated during process of sinking
Fig. 6 Well curb being used for well foundation
in bouldery bed
Fig. 7 Process of manual sinking during
initial sinking of wells
Journal of the Indian Roads Congress, July-September 2013
4 WELL FOUNDATION CONSTRUCTION
Based on the construction experience of the
organization the various case studies of well foundation
construction has been discussed in this paper. Case
studies includes Dimwe bridge, conventional sinking,
pneumatic sinking at Dalai bridge, well launching in
water, conventional sinking and pneumatic sinking at
Pasighat bridge. Well foundation at all the locations
are single circular well.
4.1 Dimwe Bridge
a) Dimwe bridge is located in Arunachal Pradesh
and the span arrangement of this bridge is two
spans of length 65.5 m and 22.8 m respectively.
Central pier is on well foundation in the main
channel of the river and both the abutments are
on open foundation. Construction of the bridge
took more time than the planned period which
include delay due to logistic reasons. Working
period was also very less due to heavy rainfall
and unprecedented flood in this area. Salient
feature of the bridge are as under:-
i) Total length of bridge : 88.30 m
(65.50 + 22.80)
ii) Type of structure : Simply supported
iii) Velocity : 5 m/sec
iv) Design HFL : RL 238 m
v) Vertical clearance : 8.5 m
vi) Maximum scour level : RL 222.0 m
vii) River Bed Level : RL 232.00 m
viii) Foundation Level : RL 207.00 m
ix) Silt Factor : 1.4
x) Design Parameters
a) Coefficient of lateral friction : 30
b) Angle of wall friction : 20
c) Bulk density of soil : 1.8t/cum
d) Submerged weight of soil : 0.8 t/cum
e) Buoyancy : 100%
 Well Foundation Construction In Bouldery Bed 191

b) Initial work on this bridge started smoothly.

Sinking of well from RL 237.45 m to
RL 224.00 m was through loose boulders
embedded in silty/sandy matrix. Below RL
224.00 to RL 222.59 m rock was moderately
hard closely spaced lightly jointed and
represented by feebly weathered strained
banded quartz, grandiorite, gneisses. Fresh
rock occurred below RL 222.59 to RL 219.00
m. Rock was traversed by falling prominent
Fig. 9 Anchoring well foundations in rock
sets of joints (Fig. 8).
5 DALAI BRIDGE
a) N 30 E S 53 W Dipping vertical
b) N 70 E S 57 E Dipping 60 on NE a) The total length of Dalai bridge is
129.65 mtrs prestressed concrete bridge and
c) Transverse joints N 20 E, S 20 W
is also located in Arunachal Pradesh. Both
sub vertical.
the piers are on well foundations in the main
GEOLOGICAL SECTION OF FOUNDATION channel of the river and both the abutments are
on open foundations. Salient features of the
bridge are as under:-

i) Span - 129.65 m
(69.52 m + 45.12 m + 15.01m)
ii) Deck Level - RL 35.60 m
iii) HFL - RL 29.60 m
iv) LWL - RL 23.20 m

Fig. 8 Soil strata encountered v) Velocity - RL 8.83 M/Sec

during sinking process Dimwe bridge vi) Discharge - 5714 M/Sec

c) Besides the two parallel gougy seam of 8 to
9 cm thick striking NE-SW dipping SE were
recorded. The intersection of different joints
were developing cubical block which were
blocked by crushed rock but there were no
evidence of crushing. Geologist recommended
that anchoring should be provided for the depth
of 3 to 5 m in the direction perpendicular to
the bending plane (Fig. 9). The foundation
RL 219.00 m consisting of fresh hard quartz,
granodiorite, gneiss lies within the permissible
limit. Due to tough strata sinking of foundation
became difficult.
vii) Silt Factor (taken for design) - 9
viii) Maximum Scour Level - RL 11.90 m
ix) Foundation Level
A1 - RL 29.00 m
P1 - RL (-) 3.400 m
P2 - RL 6.000 m
A2 - RL 29.00 m
(x) Diameter of circular well
P1 - 12.00 m
P2 - 9.00 m

Journal of the Indian Roads Congress, July-September 2013

192 Dhiman on

b) Work was started as per approved bridge Table1 Details of Foundation Level
scheme. Both the abutments A1 & A2 are open
foundations and piers P1 & P2 are on well Sl. Description A1 P1 (RL) P2 (RL) A2
No. (RL) (RL)
foundation. Initial sinking was started on both
the well locations with conventional sinking 1 *NIT FL 29.00 (-) 3.40 6.00 29.00

only. There was no problem in construction of 2 Design FL 29.00 (-) 12.00 2.10 --
both the abutments and both were completed
3 First Revision -- 6.00 (+) 8.00 --
up to designed RLs. However, in case of well
foundation, sinking difficulties were faced. 4 Final FL 29.00 7.49 11.50 29.00
Accordingly after reaching RL 18.300 m in
*NIT : Notice Inviting Tender
case of P1, difficulty in sinking was faced due FL : Foundation level
to tough strata. However, case was examined
in greater details in terms of cost aspects and Table 2 Strata Report of Excavated Materials :
further efforts were made with open grabbing. P1 Well Foundation
On further examining, difficulties faced the Sl. Date Sample Distribution by percentage Remarks
No. collected
pneumatic sinking was used and work was at RL

restarted accordingly. The scope of work for 1 04 19.56 (a) Boulder :-

sinking was also examined in terms of tough Feb (i) size 1000 to 2000 mm-10%
98 (ii) size 500 to 1000 mm-25%
strata which is considered to be less scourable
(iii) size up to 500 mm - 25%
and revised RL of foundation were also (b) Sand & Gravel - 40%
given to utilize the available strata in a better 2 07 16.90 (a) Boulder :-
way (tough strata considered as technically Mar (i) size 1000 to 3000 mm-15%
98 (ii) size 500 to 1000 mm -30%
stronger). Pneumatic sinking at this location (iii) size upto 500 mm - 30%
was started from RL 18.300 m. Soil strata (b) Sand & Gravel - 25%

details encountered were recorded. After 3 31 15.47 (a) Boulder :-

reaching at RL 12 m, there was problem in Mar (i) size 1 m x 2 m - 25% Width
98 (ii) size 0.5 m x 1 m - 25% 20 to
sinking the foundation specially for excavating (iii) size upto 0.5 m -30% 50 cm
the strata even in pneumatic condition. A close (b) Sand & shingle -20%

watch was kept for soil strata and SBC got 4 21 14.50 (a) Boulder :- Width
Apr (i) size 1 m x 2 m - 25% varying
checked at RL 7.490 m which was found out in 98 (ii) size 0.5 m x 1 m -23% from
order and plugging was done at this RL. Since (iii) size upto 0.5 m -32% 20 to
(b) Sand & shingle -20% 50 cm
the soil strata was tough, blasting was resorted
5 19 13.50 (a) Boulder :-
to and approximately 1.50 tones gelatine May (i) size 1 m x 1.5 m - 20%
was used while sinking. Details of various 98 (ii) size 1 m x 0.8 m -22%
(iii) size 0.8 m x 0.5 m -23%
RLs for foundation and strata report are as in (iv) size below 0.5 m -12%
Table 1 & 2. (b) Sand & shingle -23%

Journal of the Indian Roads Congress, July-September 2013

 Well Foundation Construction In Bouldery Bed 193

Efforts made during pneumatic sinking viz-a-viz d) Well Curb was anchored with the ground with
sinking achieved is as under as Table 3. anchor bar as indicated in Fig. 11.
Table 3
Man Power Explosive
consumed
Sinking
Month achieved
Sup Labour Gel E/D (Mtrs)
Kg Nos
Man-days
May 98 114 2007 44.59 334 1.035
Jun 98 65 1109 13.13 58 0.100
Jul 98 102 1561 58.98 266 0.680
Aug 98 57 1918 46.44 255 0.785
Sep 98 102 1578 29.42 154 0
Oct 98 66 768 11.46 60 0.220 Fig.11 Anchoring of well foundation
Nov 98 96 1234 23.72 170 0.150
after reaching foundation level
Dec 98 116 1884 43.12 424 0.160 Well foundation construction at both the location was
Jan 99 98 1850 21.14 130 0.280 difficult due to tough strata and it involved lot of efforts
to achieve desired level for plugging well foundation.
*Feb 99 69 1330 14.17 102 0.400

Total: 5.1 Caisson Launching

1030 17665 391.95 2540 5.120

After reaching RL 7.50 m plugging of well considered
in order.
c) The soil strata plan at RL 7.50 at foundation
level is indicated in Fig. 10
Major bridge having eight well foundations including
two abutment wells where all the activities of well
sinking, caisson launching, pneumatic sinking and
tilt rectification has been discussed. Out of eight
foundations, two are to be constructed in that portion
of the river where water depth remains about 7 to 8
meters even at lowest water level. Maximum flow in
this river is near the planned location of pier P3 for
which launching was proposed. Owing to the river bed
condition and considerable water depth at this location,
construction of foundation by method of Floating
Steel Caisson was considered to be a viable solution.
Salient features of bridge are as under :

a) Length of bridge - 703.50 m

Fig.10 Soil Strata at RL 7.50 m
b) Maximum Discharge - 46000 cu
c) Lowest Water Level - 153.50 m
d) Highest Water Level - 164.000 m
e) Velocity of water at HFL - 8 m/sec
f) Top Water Width at LWL - 400 m
g) Design maximum scour - 133.00 m

Journal of the Indian Roads Congress, July-September 2013

 194
h) Foundation level - 125.00 m
i) Foundations
i) Well Foundation - Circular
ii) Outer diameter - 11.70 m
iii) Inner diameter - 6.64 m
iv) Height of Well Curb - 4.50 m
The river bed consists of boulders-sand-clay-matrix
and large size boulders at lower depths. Possibility
of sheet piles, cofferdam and sand island method, for
construction of well foundation was ruled out due to
large sized boulders. Caisson launching, pneumatic
sinking and tilt rectification of this bridge has been
discussed in succeeding paragraphs.
5.2 Design Details
Caisson was designed for following :
a) Bed Level - 147.50 m
b) Max water depth while grounding - 7.00 m
c) Max scour while grounding - 1.00 m
d) Free Board - 1.50 m
e) Draught for caisson - 2.86 m
Based on above river details caisson was planned to be
floated at lowest water level during November onward
up to mid March. Also the discharge and velocity 1.0
mtr/sec to 1.2 mtr/sec and had not much effect on the
structure during this period. Draught for floating the
caisson in the river for movement up to desired location
was 2.86 mtrs.
5.3 Steel Caisson Launching
a) Resources used for above event are as follows:
i) Crane -1
ii) Power Generator 55 KVA -1
iii) Power Generator 10 KVA -1
iv) Concrete Mixer -2
v) Vibrator -4
vi) Power Tug
a) Power boat with single 95 -1HP engine
b) Mallard type 90 HP engine -2
Journal of the Indian Roads Congress, July-September 2013
Dhiman on
vii) Barges
a) For crane mounting -1 (200 Ton capacity)
b) For Mixer 12 x14 m - 1 (150 Ton capacity)
c) For aggregate 12x12 m - 1 (130 Ton
capacity)
d) For feeding aggregates/- 1(100 Ton cement/
HSD at 8 x 4 m capacity)
viii) BUOYS
Main - 5 Nos
Ordinary - 10 Nos
Anchor Blocks RCC Blocks - 15 Nos
PSC Blocks - 2 Nos
ix) D shackles 5 ton to 20 - 100 Nos
ton capacity
x) Hand Winches (5 ton capacity)
For caisson - 5 Nos
For Pontoons - 10 Nos
xi) Wire Ropes (Minimum)
3/4 wire rope - 3000 RM
1 wire rope - 600 RM
1.25 wire rope - 200 RM
xii) Pully Blocks
Single sleave - 5 Nos
Double sleave - 5 Nos
Three sleave - 5 Nos
xiii) Welding machine - 3 Nos
5.4 Sequence of Activities Followed
Sequence of activities that followed were as under :
a) Area along the U/S and D/S side of the river with
reference to proposed bridge axis was surveyed
to collect information required for proposed
launching. Suitable place for fabrication of
caisson and for base line to locate the proposed
caisson position was finalized .
b) Cutting edge of the caisson was fabricated
in parts at workshop and shifted to proposed
location of launching. Caisson was fabricated
on the right bank at about 500 m on U/S of
 Well Foundation Construction In Bouldery Bed
proposed bridge location. Fabricated caisson
was as under :
Well Curb and Steining upto guage height
9 m (structural steel)
i) 20 mm PL - 6.50 MT
ii) 12 mm PL - 8.50 MT
iii) 10 mm PL - 11.00 MT
iv) 6 mm PL - 17.60 MT
v) 3.5 mm PL - 2.20 MT
vi) 50 x 50 x 6 angle - 2.20 MT
vii) 90 x 90 x 8 angle - 7.00 MT
viii) 65 x 65 x 6 angle - 13.50 MT
Well Curb Reinforcement
(i) 16 mm dia MS - 0.60 MT
(ii) 25 mm dia MS - 17.0 MT
(iii) 28 mm dia MS - 2.00 MT
(c) Height of caisson was 9 m and outer diameter
was 11.70 m. Total weight of the caisson was
87.50 MT. Draught required for the caisson was
arrived at 2.86 m. After fabrication structure
was anchored with the winches placed on right
bank. Fabrication of caisson was done at 40 m
from river bank, from where the caisson was
hauled towards water, with the help of winches
and rail track laid on the ground up to 1.5 m
from the water edge. (Fig.12).
Fig. 12 Fabrication place for caisson
d) After placing caisson near to water edge,
jetty was extended which was 20 m wide and
15 m long towards water. Rail track was laid
up to jetty for movement of the caisson towards
water edge with the help of winches placed on
the bank. Caisson was further moved on to the
Journal of the Indian Roads Congress, July-September 2013
195
jetty where water depth was about 3.5 m all
around. Caisson was kept under control at jetty
location.
e) Open grabbing was done on the jetty location
so as to render the caisson under floating
condition. During the open grabbing cutting
edge got punched into the existing bed on one
side and the caisson movement got stuck. To
overcome this difficulty, further open grabbing
was resorted, to bring the caisson, into floating
condition. This was achieved with the help
of divers and experienced persons. Once the
caisson was made to float, it was planned to
detach its control from the bank winches and the
same was shifted to buoys in floating condition.
This was done with the help of steel boat.
f) On floating the caisson, the movement of the
caisson towards the proposed location was
controlled. Path of travel was kept parabolic
towards D/S to avoid the drag of water. Five
sets of identical anchor blocks consisting of
hexagonal shape were dropped at predetermined
locations on U/S and D/S of the proposed
location of well P3 attached to the buoys. One
set of anchor block was having four blocks of
4 to 5 tonnes each. These blocks were dropped
keeping in view the requirement of reaction
required so that sufficient resistance is provided
during the movement of caisson. Movement
was controlled from the winches on the bank
till the control was transferred to the buoys in
floating conditions. Caisson was moved to D/S
in 3 hours and on reaching near the proposed
location the movement of caisson was slowed
down.
g) At proposed location the caisson was positioned
and movement of winches was locked.
Theodolites were placed on the left bank on
two stations already marked location which is
206 m apart and 103 m on either side of proposed
line of bridge. The location of stations were so
 196 Dhiman on

Fig.13 Final placing of caisson

selected that they should form well conditioned

triangle. The position of well was checked with
two theodolites and cross checked with third
theodolite placed along centre line of bridge
(Fig.13).

h) After alignment positioning of caisson,

concreting of curb/steining was started. Concrete
was shifted from the left bank with the help of
concrete buckets on a boat (Fig.14 and Fig.15).

For this purpose two cranes were used, one was

at bank and the other on the floating pontoon.
During the process of concreting the self-
weight of the structure gradually increased
and correspondingly lowering of the caisson
was done, till grounding of the caisson. To
keep the well in alignment during the course
Fig. 14 Shifting of concrete from banks to the
proposed bridge location.
of concreting/sinking the winch anchoring
arrangements were retained undisturbed until
well was sunk by 3.50 m.

Fig. 15 Shifting of concrete for steining casting

Journal of the Indian Roads Congress, July-September 2013

 Well Foundation Construction In Bouldery Bed 197

5.5 Problems Encountered 6 PNEUMATIC SINKING

a) Due to uneven bed profile, it was very difficult
to move caisson after fabrication in workshop to
the bank from where launching was proposed.
Initial fabrication of cutting edge was done in
the workshop and remaining portion of caisson
was fixed at proposed location from where the
launching was monitored. This was done, as it
was difficult to move the caisson after erection
due to uneven bed surface.
b) Draught calculated for the steel caisson was
2.86 m and accordingly the jetty was prepared
where the required draught is available after
breaking the jetty. After caisson was placed
at jetty location, it got stuck in the sandy bed
and was difficult to bring in normal alignment
as the soil on which it was placed filled one.
Innovative efforts were applied for this,
simultaneously horizontal forces and divers
were deployed at particular location to make
the caisson free for movements.
Pneumatic sinking is used when soil can not be
excavated through the open shaft in the caisson and
soil to be taken out is below water level and include
boulders buried timber or unusual properties of the soil
itself. Pneumatic sinking is used in bouldery strata
which can not be excavated by normal conventional
methods. However, with the introduction of pneumatic
sinking at site the work can be carried at faster rate
than the rate before starting of pneumatic sinking. The
technique has following advantages:
i) All the work is done in near to dry condition,
therefore control over the work and foundation
preparation are better.
ii) Plumbness of the caisson is easy to control as
compared with open grab method.
iii) Boulders and logs encountered during sinking
can be easily removed. (Fig 16)

c) After final placing of caisson at the

proposed location, well curb concrete
was done from the pantoons only where
cement and coarse aggregates were
placed. However, this led to slow down the
out put as the concrete was poured manually
and the space available on the pantoon was
less. However, this problem was over come
by shifting the arrangement to bank the
Fig. 16 Large size boulder in river bed at 10 to 15 m deep where
remaining concrete for the caisson was done sinking of well is difficult
from the bank which was shifted through steel
bucket of 1.2 cum capacity by flat pantoon ESSENTIAL PARTS AND TECHNICAL
and two cranes were used for this, one at bank SPECIFICATION OF PNEUMATIC CAISSON
location and the other near to the well location
6.1 Working Chamber
on floating pantoons.
The working chamber is the space at the bottom of
d) Due to 9 m height of caisson the metacentric
pneumatic caisson surrounded by the beveled wall of
height was more and there were every likelihood
cutting edge and roof by steel diaphragm or concrete
of toppling the caisson. However, this problem
plug (corbel slab). Since it is the space for workmen
was overcome by proper control. Better
to excavate the soil, it should be at least 2.5 m high
solution could have been part fabrication at
from cutting edge. The side wall and roof are designed
bank and remaining at the final location before
to withstand the maximum anticipated air pressure.
casting the well curb.

Journal of the Indian Roads Congress, July-September 2013

198 Dhiman on
This air pressure in general is greater than the pressure
due to the head of water above the bottom of caisson.
(Fig.17 & 18).
Fig.17 Pneumatic Sinking Arrangement
Fig. 18 Air Lock
6.2 Air Shaft period for compression and
Air shaft or shafts are the vertical passage-ways for
workmen and materials. In smaller caisson, one shaft
may be sufficient with a ladder. One side of shaft is
for workmen and the other side for material (removal
of soil and placement of concrete). In large caisson,
two or three shafts are provided, one for passage
of workmen and the other for the materials. Shafts
in concrete caisson may be made of steel or may be
simply a cylindrical hole. In latter case the concrete
must be reinforced to withstand the internal pressure
due to the compressed air and the air lock of the shaft
must be anchored down to the concrete. Steel shaft
are fabricated in short sections of 1.5 to 3.0 m lengths.
The points between the section are made air tight with
rubber gaskets. As the sinking progresses, additional
length of the shaft is added. The shaft must be
designed to withstand the internal pressure. Therefore,
the common cross-sections are circular, elliptical, or
figure of eight. It is ensured that the height of steining
is always higher than the LWL.
Journal of the Indian Roads Congress, July-September 2013
6.3 Air Lock
An air lock is required for each shaft. It is mounted
on top of the shafts and extends above water level.
The function of air lock is to permit the workmen
and materials to go in and out of the caisson without
releasing pressure in the caisson. Air-lock consists of
steel chamber with two air-tight doors one of which
opens to shaft and other opens to atmosphere. When
a men enters the air lock through the outside door, the
pressure in the air lock is equal to that of atmosphere
and then the door is closed and air pressure is allowed to
rise slowly. When the pressure in the air lock becomes
equal to that of inside of caisson, the door to the shaft
is opened and men descend into the working chamber.
The procedure is reversed when a men comes out of
caisson.
6.4 Hours of Work in Compressed Air
Hours of work for workmen who are subject to
compression and decompression shall not be more
than specified below in any consecutive 24 hours
(IS : 4138 : 1977).
Pressure kg/cm Number of hours excluding the
decompression
Min. Max.
0 1.25 8 (Normal Working)
1.25 1.25 6
2.20 3.40 4
6.5 Physiological Effects of Compressed Air
Usually no harmful effects are felt on entering
in pneumatic caisson or while remaining inside,
but occasionally eardrums are broken and blood
vessels ruptured. Problems usually come during
decompression. The attack may be light or severe. A
light attack is characterized by severe pains mainly
in the joints and closely resembles rheumatism and
its effects. This disease is known as BENDS, when
attack is severe, it usually PARALYSIS, its victim
and is often FATAL. The sensations felt on entering
compressed air are slight giddiness, inability to whisper
 Well Foundation Construction In Bouldery Bed
and a feeling of resistance to movement owing to
density of the air. Pain may be felt in the ears which
may be relieved by closing the mouth and holding the
nose and at the same time trying to expel air from the
lungs. On leaving the air pressure, one feels cold, the
sensation being keenest during the passage through air
lock. It is due to expansion and liberation of gases in
the body. To counteract the effects of this cold, air lock
should be heated, the men should be given hot coffee
to drink on coming out and they should dress warmly.
Another sensation often manifested on emerging is
itching, pricking feeling under the skin on all parts
of the body and this may disappears in few minutes.
Even experienced person find it difficult to enter when
trouble with a bad cold. In India there had not been any
major catastrophes during pneumatic sinking.
6.6 Preventions and Cure for Caisson Sickness
If the cause of caisson disease is a mechanical action due
to the development of bubbles in the bloods and fluid
tissues, which is in turn due to too rapid decompression,
then cure is decompression at a slow rate. The length
of time will depend upon the amount of the gas in
the fluid tissues and upon physical characteristics of
person being decompressed. The amount of gas in
the fluid tissues will in turn depend upon degree of the
pressure in the working chamber and length of time
body is under pressure. The length of time taken to
saturate the body fluid at any fatness of the subjects
depend upon the amount of bodily work done heat
and moisture present. Experiment shows that the fatty
tissues absorb about five times as much gas as does
the blood and the rate of absorption is much slower,
corresponding to the rate of de-saturation will be slow.
For this reason the men inclined towards fatness should
never be employed for compressed air work. Better the
circulation of the blood, the more quickly and easily
will the gases be thrown out from the system. For this
reason only men in good physical condition should be
employed. Old men, those who have abused themselves
by excessive drinking should never be allowed in
the working chamber. The theory upon which stages
decompression based is that, the gas in the blood will
not effervesce until a marked decrease of pressure. To
Journal of the Indian Roads Congress, July-September 2013
199
the point of effervesce, the gases are discharged at rate
varying with some function of change of pressure. The
more rapid and lowering of pressure, the more quickly
will the blood vessels be freed of the gases contained
therein. Apart from the matter of slow decompression,
other precautions if taken, will minimize the occurrence
of caisson disease. Anything, which tends to lower
resistance of the human system tends to promote
caisson illness. The only cure for caisson disease is
recompression with slow decompression. If the patient
can be put into air before the gas bubbles have had a
chance to tear the blood vessels and fluid tissues, a cure
can usually be effective, but otherwise not. From this
reason a medical air lock, large and well ventilated
should always be maintained in readiness and the men
should be housed nearby, so that in case of delayed
attacks they may be immediately recompressed.
6.7 General Precautions to Avoid Caisson
Sickness
 a) Work under compressed air should
be executed under the supervision of
responsible engineer.
b) Age of workers under compressed air
should be 20 years to 40 years.
c) All employees must be medically examined
carefully before they are selected for work
under compressed air. They must also be
periodically examined (Say 15 days) to
certify their fitnes. When first employed
under pressure above 2.10 Kg/Cm
recommendation letter to be taken from
authorised doctor after 1st shift.
d) Never go on shift with empty stomach.
Also eat moderately.
e) Sleep at least 7 to 8 hours. Keep bowels
regular.
f) Liquor should not be given to man suffering
from compressed air sickness.
g) Report for medical checkup for any ailment
no matter how slight.
 200 Dhiman on
h) Take over clothing to avoid becoming
chilled during decompression
i) Move limbs freely during decompression
to stimulate circulation
j) Stay at job site for at least half hour after
coming out of working chamber. One man
should not work more than one shift.
k) For accurate control of air pressure, a
gauge attendant should watch the pressure
gauge constantly and the gauge should be
accurate and in good working condition.
There should be a master pressure gauge at
site for periodical check up of the pressure
gauges in the lock.
l) To avoid the air in the working chamber as
well as in main chamber becoming stale,
fresh air must be circulated in the lock
constantly. This can be done by opening a
valve in the air lock. In granular soil where
certain amount of leakage takes place
through the cutting edge and the soil, the
air is automatically circulated.
m) Man working under compressed air must
be decompressed slowly. If coming out too
fast, they are subjected to caisson disease.
This disease is due to air bubble formed
in the blood and body tissues, which are
compressed while working under pressure.
n) The air lock to be protected against
temperature rise by means of the sun.
o) The air lock, shaft and the interior of the
well are to be provided with sufficient
electric lighting. During working, every
shift in-charge in the working chamber and
every attendant must have sufficient 3 to 4
cell for lighting purpose.
p) The IN and OUT of person should at LWL condition to a distance of 350 M across the
only be monitored by a responsible and
Journal of the Indian Roads Congress, July-September 2013
experience of lock attendant who is fully
conversant with instructions.
q) Unauthorised manipulation of valves of the
compressed air pipe is very dangerous for
the health of all employees.
r) During locking in, any body felt pain in
the ear shall try to remove, the same by
continually swallowing as the air pressure
is increased or by holding the nose and
blowing as this action tends to increase
the freedom with which the air can pass
through the Eustachain tube into the middle
ear there by equalizing the pressure on the
inner and outer surface of the ear drum.
s) In case a person feels indisposed during
working, he should immediately inform the
lock attendant in order to be locked out at
once. Before locking out extra cloths are
to be put on. Also person who becomes
indisposed during locking out must
immediately contact the lock attendant
who can if necessary again increase the air
pressure.
t) Men should not be over crowded to avoid
discomfort while entering in the chamber.
u) Persons should not be allowed to carry
any match box in the air lock nor any body
should be allowed to smoke.
7 TILT RECTIFICATION
Caisson as per Para 5.1 was launched from Nov 1990
onwards. Due to flash flood of April 91 the work was
severely disturbed and no major event was carried out
till monsoon was over after April 1991. The well got
over tilted during flood of 1991 monsoon for period
of June to August to I in 2.34 to u/s on left quadrant.
On critical examination it was difficult to make any
rectification scheme due to wide spread of water even
channel. There was no possibility of diversion at
 Well Foundation Construction In Bouldery Bed 201

this location. However, taking the advantage of bank

condition a scheme was made to rectify the tilt.

Fig. 20 Well in tilted position

Fig.19 Well in Tilted Position
7.1 Design and Layout of Rectification Scheme
The position of the well in tilted condition and soil
strata already encountered was examined. Based on
present position of well, it was planned to pull the
well with horizontal forces to bring the tilt to normal
position. For this, pulling arrangement with the aid of
prestressing was finalized. Stressing was planned to
be done from both ends from D/S. Cable was planned
to be taken around a anchor end block to be a part of
steining a later date. Important details of the scheme
are as under (Fig.20).
7.3 Left bank anchor block location was located on
a bed having deposits of wooden logs subsequent the
flood and also the area was low lying compared to same
on right bank which was full of boulders at higher level.
Concrete block was made to resist design force of 501
Tons against requirement of 330 Tones only. Space in
front of the blocks were filled with boulders to avoid
resistance in horizontal direction. Arrangement for
tilt rectification was implemented systematically on
ground (Fig 21).

a) Tilt of well - 1 in 2.34

b) Weight of well - 2407 MT

(after adding block)

c) Weight of well - 1407 MT

(submerged in water)

d) CG of well above base - 7.76 M

e) Force required to over come tilt - 330 MT

f) Design pull of six cables. - 520 MT

7.2 Based on the analysis of the site condition

and force required to pull the well, two anchor blocks
were made on both the banks at downstream of the
bridge as per following details :
(a) Left Bank at 260 Mtrs D/S Fig. 21 Addition of block on the top of the
steining for tieing around
(b) Right Bank at 315 Mtrs D/S

Journal of the Indian Roads Congress, July-September 2013

 202 Dhiman on

7.4 Rectification of Tilt 7.5 Problems Encountered

On completion of ground work for rectifications a) Due to high rainfall in the area the time period
scheme, action was taken to improve the tilt. available was very less. There was urgent need
to add additional steining after rectification of
a) Sag of cable was removed and grabbing the tilt.
was on downstream of the well.
b) There was a problem of logistic due to remote
b) Stressing of cable was done simultaneous location of site.
in stages to have the loads upto 301. This
c) Since distance between two banks was large,
was done in following stages.
coordination was not possible without Walkie-
Stage Force applied Improvement in tilt talkie. However, two boats were kept busy to
1st Stage 50 T 1 in 3.0 ensure better communication across two banks.
2nd Stage 100 T 1 in 7.0 Also signaling system between banks for
3rd Stage 150 T 1 in 10.0
indication was ensured with the help of flags
(Fig. 22).
4th Stage 200 T 1 in 12.0
5th Stage 250 T 1 in 14.0 8 RECOMMENDATIONS AND
CONCLUSION
c) After three days efforts the tilt which
was 1 in 2.34 was reduced to 1 in 15. All the bridge schemes be finalised based on the
Subsequently one lift of concrete was construction method planned or equipment/plant to
added to increase the height of steining. be deployed. However the following points be given
After this tilt rectification was done to the due consideration while planning well foundation in
desired limit. bouldery bed.

Fig. 22 Cable stressing from end for rectification of over tilted well

Journal of the Indian Roads Congress, July-September 2013

 Well Foundation Construction In Bouldery Bed
a) Bridge scheme be finalized based on
examination of the pros and con of the data
and guidelines exists. Numbers of foundation
be so decided that there is less sinking efforts
involved considering all factors. Construction
system/methodology available should also be
kept in view while planning a major bridge.
b) Scour depth in bouldery bed, should be finalized
scan pattern data of existing bridge and same
should be analyzed. The existing bridges on
bouldery beds should be considered as live
model for studying scour pattern.
c) Change in soil strata properties be considered
or examined with due consideration. All the soil
strata actually encountered should be recorded
per meter wise and same should be compared
with soil strata anticipated during investigation
stage. Tough strata should be considered as
engineering friendly,
d) Construction difficulties be given greater
attention from execution point of view.
e) Sinking should be progressed in a systematic
way to avoid the over tilting of any well
foundation during the sinking process. Tilt and
shift should be checked regularly.
f) Pneumatic sinking technique be used
judiciously wherever applicable as this is very
costly technique and require lot of precautions.
g) Soil strata likely to encountered in bouldery
beds are large size boulder which will delay
the progress of well foundation construction.
Special attention should be given to the initial
studies being carried out for all projects on such
strata.
h) Overall depth of well foundation should be
so designed that it is safe, economical and be
feasible on ground as per available construction
methodology. Experienced gained at one site
be dove tailed for other location to tackle the
Journal of the Indian Roads Congress, July-September 2013
203
similar problems in a better way. However,
the certain construction difficulties are likely to
occur in way which can only be solved with
systematic planning and skilful execution
coupled innovative efforts.
ACKNOWLEDGEMENTS
Author expresses sincere thanks to the Director
General Border Roads and Additional Director
General Border Roads for their guidance for
preparation of this Paper.
REFERENCE
1. Dhiman R.K., Bouldery Bed Scour- Proposed
Formula Indian Roads Congress (IRC) Journal
65 Vol.3 Paper No. 508, Nov. 2004.
2. Dhiman R.K., Pneumatic Sinking - A Case
Study Indian Roads Congress (IRC) Indian
Highways, Feb. 1996.
3. Dhiman R.K., Caisson Launching A- Case
Study Civil Engineering and Construction
Review (CE&CR) - 1996.
4. Dhiman R.K., Foundation Level for Bridges
A Programmatic Approach, 1996 New
Building Material and Construction World
(NBM & CW).
5. Dhiman R.K., Essence of Silt Factor for Scour
Around Bridge Pier, International Conference
on Scour of foundation (ICSF-I) held at Texas
USA (17-20 Nov. 2002).
6. Dhiman R.K., Construction Problem of
Bridges in Hilly Region A Review-1997
International Association of Bridges and
Structural Engineers (IABSE).
7. Dhiman R.K., Dimwe Bridge Foundation
A Case Study 4th International Seminar on
Bridge and Aqueduct 1998.
8. Dhiman R.K., Well Foundation Construction
in Bouldery Bed- A Case study International
Association of Bridges and Structural Engineers
(IABSE) Colloquium Feb. 1999.
 Dhiman on
204 Well Foundation Construction In Bouldery Bed
9. Dhiman R.K., Affect of Flash Flood-A
Case Study- Disaster Mangement NERIST.
Itanagar-1999.
10. Dhiman R.K., Caisson Sickness During
Pneumatic Sinking International Symposium
at University of Dundee Scotland (UK)- Sep.
2003.
11. Dhiman R.K., Bridge Construction
Problems and Solutions- A Review 17th
National Convention of Civil Engineering at
Bhubneshwar-2001.
12. Dhiman R.K., Extension of Span Ranga
II Bridge A Case Study- 17th National
Convention of Civil Engineering & Seminar on
Modern Trend in Construction and maintenance
of Roads, Flyovers and bridges- Bhubaneshwar-
Nov 2001.
13. Dhiman R.K. Tilt Rectification of Well
Foundation A Case Study - Indian Roads
Congress (IRC) Indian Highways May 2002.
14. Dhiman R.K., Essence of Training of
Manpower for Concrete Technology,
Sixth International Confrence on Concrete
The views expressed in the paper are personal views of the Author. For any query, the author may be contacted at: E-mail: d_himan12@yahoo.com
Journal of the Indian Roads Congress, July-September 2013
Technology (61CCT) at Aman (JORDAN)
Oct. 2003.
15. Dhiman R.K., Construction Challenges for
Bridges in Hilly Area- An Over View- Indian
Roads Congress (IRC) Indian Highways, Jan.
2004.
16. Dhiman R.K., Damages to Bridges due to
Flash Flood- A Case Study Indian Road
Congress (IRC) Indian Highways Oct. 2004.
17. Dhiman R.K., Effective Construction
Management of Bridges International
Association of Bridges and Structural Engineers
(IABSE) Dec. 1996.
18. Special Issue on Round Table Conference
on Scour Around Bridge Pier 1993 Indian
Institute of Bridge Engineering (IIBE), New
Delhi.
19. Model Study Report of Pasighat Bridge by
UPIRI, Roorkee-1984.
20. Indian Road Congress (IRC) 78:2000.
21. Onnchi Sadaken Border Roads Organization
(BRO) Annual Technical Magazine (1990 to
2004).
 Paper No. 598

Dynamic Vehicle Equivalent Factors for

Characterisation of Mixed Traffic for
Multilane Metropolitan Arterials IN INDIa
Gaurang Joshi* And Dinesh Vagadia**

Abstract
Traffic volume which is key input for road system design is represented in terms of standard passenger car unit. IRC: 106-1990; Guidelines for Capacity
of Urban Roads in Plain Area, recommends two sets of PCU values based on 5% and 10% and more composition of various vehicle types in total
traffic volume. It is also established that the vehicle type with major proportion in the traffic stream dominates behaviour of other vehicles and therefore
conversion of mixed traffic with predominance of two wheelers in to passenger cars is also questionable. In the present paper, Dynamic Vehicle Equivalent
Factors (DVEF) namely Dynamic Car Unit (DCU) and Dynamic Two-wheeler Unit (DTU) are derived using modified homogenization coefficient approach
based on comprehensive field traffic data collected by videographic technique on mid-block road sections of multilane arterials in seven metropolitan cities
of India. A simple method to derive traffic volume in DCU/DTU for different traffic composition and traffic volume for a standard 10.5 m wide three lane
arterial road is discussed in the present study. Significance of DVEF and its variation with influencing parameters in mixed traffic is also discussed in this
paper.

1 INTRODUCTION car for traffic analysis and facility design by applying

During last two decades, India has witnessed high GDP
growth rate indicating faster economic development
coupled with simultaneous growth in urban population
with major thrust on metropolitan cities having better
infrastructure for growth and scope of livelihood. This
phenomenon has created huge mobility requirement
in terms of efficient movement of high volume of
vehicular traffic on metropolitan arterials which
needs to be satisfied through scientific approach of
planning & design of road system based on realistic
characterization of heterogeneous traffic condition
prevailing on urban roads in India. The information
on traffic volume is an important input required for
planning, analysis, design, and operation of road
systems. Traffic in developing countries such as India
is highly heterogeneous, comprising vehicles of
wide ranging physical and operational characteristics
occupying any convenient lateral position on the road
based on space availability. It is universal convention
to represent the traffic in terms of standard passenger
conversion factors based on the effect on the speed
of all car stream or equivalent volume at capacity
of other vehicles compared to all car condition.
IRC: 106-1990; Guidelines for Capacity of Urban Roads
in Plain Area, recommends two sets of PCU values
based on 5% and 10% and more composition of
various vehicle types in total traffic volume. In many
field observations, use of IRC recommended PCU
values are found to result into unrealistic traffic volume
in PCU/hr; even more than that of the prescribed
capacity; for the free flowing traffic condition. Research
efforts in last decade have shown that the PCUs of
various vehicles depend on traffic composition, traffic
volume and road geometry besides their size and
speed characteristics and hence is dynamic in nature.
It is also established that the vehicle type with major
proportion in the traffic stream dominates behaviour of
other vehicles and therefore conversion of mixed traffic
with predominance of two wheelers in to passenger car
units (PCU) is also deserves some pondering. In view
of this, the need was felt to critically examine the issue

* Associate Professor, Civil Engineering Department, S.V. N.I.T. Surat., E-mail: gjsvnit92@gmail.com
** Lecturer, Civil Engineering Department, Government Polytechnic, Rajkot.
Written comments on this Paper are invited and will be received upto 30th October, 2013.

Journal of the Indian Roads Congress, July-September 2013

206 Joshi & Vagadia on
of traffic characterisation; particularly for multilane
metropolitan arterials encompassing the different road
geometry, volume and composition on dynamic scale
considering two wheeler and car as reference vehicle.
In the present study, Dynamic Vehicle Equivalent
Factors (DVEF) called as Dynamic Car Unit (DCU)
and Dynamic Two-wheeler Unit (DTU) are derived
using modified homogenization coefficient approach
based on the field traffic data collected by videographic
technique covering only the multilane arterial mid-
block road sections in seven metropolitan cities
namely Lucknow, Kanpur, Jaipur, Patna, Surat,
Pune and Thiruvananthapuram. The selected cities
have significant difference in the socio-economic
characteristics, city size and structure, vehicle types,
predominant transportation modes, traffic volumes,
traffic control conditions and road width. Difference in
carriageway width and side friction effect are adjusted
by modifying the observed flow rate to equivalent flow
rate for standard three lane one way road width of 10.5 m.
Traffic volume and composition variation effect on
DCU and DTU values are captured by partitioning
traffic data into sub-groups based on proportion of
2W (two-wheeler) and slow moving non-motorised
vehicles like bicycle and pedal rickshaw.
DCU/DTU variations with flow rate and composition
are studied with the help of scatter plots using flow
and composition grouping approach based on 1-minute
traffic data. Flow groups of 250 veh/hr class interval and
composition group of 5% class interval are used to draw
scatter plots. DCU/DTU values are estimated for each
traffic composition based subgroup of all cities as well
as for whole data with the help of computer program
prepared in Scilab. DCU/DTU values obtained are
matching with the expected variation pattern. DCU and
DTU values are suggested for various levels of traffic
volume. A simple method to convert mixed traffic in
terms of DCU/DTU for different traffic composition
and traffic volume for a standard 10.5 m wide three
lane arterial road is also proposed.
2 LITERATURE REVIEW
Earlier studies regarding Passenger Car Unit (PCU) or
Passenger Car Equivalent (PCE) carried out in western
countries mostly in USA and UK are based on delay,
speed, density, headway and queue discharge. Also PCU
Journal of the Indian Roads Congress, July-September 2013
or PCE estimated are mainly for heavy vehicles under
homogeneous traffic condition which is totally different
from Indian mix traffic condition and requirement. In
India, studies related to PCU were initiated in eighties.
Justo and Tuladhar (1984) developed mathematical
models to derive PCU values for vehicles on urban
roads based on empirical data under mixed traffic
flow. Ramanayya (1988) estimated the PCU factors for
different vehicle types at different levels of services
taking the Western car as the Design Vehicle Unit DVU.
In 1990, Indian Road Congress prescribed static PCU
values for different vehicle categories based on field
observed data for urban areas (IRC: 106-1990) and rural
areas (IRC: 64-1990). As mentioned earlier, the PCU
value for each vehicle category in the case of urban
areas is differentiated based on vehicle composition of
5% and 10% or more in IRC: 106-1990. Chandra and
Sikdar (2000) through an empirical study found that
for a given road width, an increase in volume level of
mixed traffic causes more density on the road resulting
in reduced uniform speed of vehicles. The lower speed
difference between cars and subject vehicles yield
smaller PCU value for the vehicle type. Chandra and
Kumar (2003) observed that the PCU for a vehicle type
increases linearly with the width of carriageway which
is due to the greater freedom of movement on wider
roads and therefore a greater speed differential between
a car and a vehicle type. Chandra (2004) reported that
PCU value decreases with decrease in gradient. Dey,
et. al. (2007) observed that for traffic on two-lane road,
PCU for a vehicle decreases with increase in V/C ratio
because speed differentials decrease as traffic volume
increases. They observed that the speed of car is not
affected much by the mix of traffic stream at low volume
level but reduces considerably at high volume level
and with rise in proportion of second category vehicle.
Arasan and Koshy (2004) through simulation for four-
lane divided road found that PCU values decrease
with increase in their percentage composition in the
traffic stream. However in case of auto-rickshaws, the
PCU value initially increases and thereafter decreases
with increase in their percentage composition in the
traffic stream. Arasan and Krishnamurthy (2008) used
simulation technique for four-lane divided road and
found that the PCU value of a vehicle significantly
changes with change in traffic volume. At low volume
levels, the PCU value of vehicles increases with
increase in traffic volume, whereas under higher
 Dynamic Vehicle Equivalent Factors for Characterisation of Mixed Traffic for Multilane
Metropolitan Arterials in India
volume conditions the PCU values decrease with
increase in traffic volume which is obviously due to
reduced speed. They also found that for any vehicle
type in heterogeneous traffic, the PCU value increases
with increase in the width of road space. Arasan and
Arkatkar (2010) observed in simulated traffic on four-
lane as well as six-lane divided road that; in the case
of vehicles that are larger than passenger cars, at low
volume levels, the PCU value decreases with increase
in traffic volume and at high traffic volume levels, the
PCU value increases with increase in traffic volume.
In the case of vehicles that are smaller than passenger
cars, at low volume levels, the PCU value increases
with increase in traffic volume and at high volume
levels, the PCU value decreases with increase in traffic
volume. Cao, et. al. (2010) has estimated Motor Cycle
Unit (MCU) for mixed traffic and demonstrated its
utility in planning for separate two wheeler lanes where
two-wheeler traffic is dominant. Joshi, et. al. (2011)
have developed speed-flow and density relationship
between speed density, density-flow and flow-speed
in terms of vehicles, static passenger car units (SPCU),
dynamic two wheeler units (DTU) and dynamic car
units (DCU) for free flow regime for a mid block section
of access controlled six lane divided arterial road. They
have introduced Dynamic Vehicle Equivalent Factors
(DVEF) on the premise that the equivalent factors vary
dynamically as the speed of reference vehicle in the
mixed flow changes. The study infers that the speed
and maneuverability of the vehicles in the stream
is generally governed by the vehicle which has high
proportion in the stream. They found that the values of
dynamic PCU for two wheeler and auto rickshaw are
much less compared to those recommended by IRC. In
summary, the review of literature on the subject matter
reveals that the studies conducted abroad are mostly
related to fairly homogeneous traffic conditions. Limited
studies conducted in India under heterogeneous traffic
conditions indicate dynamic nature of equivalent factor
(like PCU) which needs to be studied in comprehensive
manner to replicate the diverse field conditions
accurately and hence appropriateness of this study is
established.
3 DATA COLLECTION
The source of data for this study is the database of
traffic surveys carried out under the IRC-SVNIT
Journal of the Indian Roads Congress, July-September 2013
207
collaborative research study during 2008-2012.
During field surveys, 16 hours (6:00 am to 10:00 pm)
classified vehicle volume count survey is carried out
manually using tally sheets on normal working day in
normal climatic conditions. Vehicles are classified in
nine categories i.e. Two-wheeler (2W), Three-wheeler
(3W), car (4W), Mini Bus (MB), Standard Bus (SB),
Light Commercial Vehicle (LCV), Truck, Bicycle and
Pedal Rickshaw (PEDRXW). Vehicles not included in
this classification are recorded as other non-motorized
vehicles and other motorized vehicles. Based on the
hourly traffic volume profile; various traffic condition
hours like morning peak, evening peak, off peak,
pre-peak are identified and video recording is carried
out using high resolution video camera during these
durations.
Traffic survey data of seven selected cities
Lucknow, Kanpur, Jaipur, Patna, Surat, Pune and
Thiruvananthapuram is used for this study. Recorded
videos are played on computer screen, and classified
traffic volume is counted manually for 1 minute time
interval. Stop watch as well as subtitle time display in
video are used to achieve the best possible accuracy
during specific observation time interval. Spot speed
of each vehicle category is calculated by measuring
time required by the vehicle to traverse a defined trap
length marked on the carriageway with white paints.
As traffic volume is high, a maximum six or actual
number of vehicles; whichever is less, are taken as
samples for spot speed of each vehicle category in one
minute time interval. Samples are selected randomly
in such a manner that they are from whole one minute
time interval. Traffic volume observed in each minute
is converted into flow rate by multiplying with 60.
All selected road sections are six lane divided, albeit
with varying width and level of side friction. Road
sections in Surat and Jaipur are 10.5 m wide with
service lane and hence are free from roadside friction.
Flow rate on road sections of other cities is modified
for standard road width of 10.5 m considering linear
variation of flow rate with road width. For example,
Thiruvananthapuram road section is 9.0 m wide, hence
observed flow rate is modified by a multiplying factor
1.17 (=10.5/9.0).
As study is carried out in urban areas, traffic has major
208 Joshi & Vagadia on
share of two-wheelers, cars and three-wheelers. Cities
of north India have significant share of bicycle (9.5 to
22.5%) and pedal rickshaw (6.5 to 13.5%) in the traffic.
To arrive at a reasonably accurate and representative
vehicle dimensions for each vehicle category, domestic
sales figures published by SIAM during the last five
years (2006-2011) are used. From the database, figures
of domestic car sell in each segment i.e. mini, compact,
mid size, executive and premium are considered.
Average representative dimension is calculated by
weighted average of each car segment dimension.
Similarly for two-wheelers, three wheelers, mini
bus, standard bus and LCV; average representative
dimensions are determined. Due to unavailability of
reference data and as other vehicle categories have little
share in urban traffic, their representative dimensions
are directly taken from similar earlier studies. Table-1
shows representative vehicle dimensions considered in
the study.
4 DYNAMIC VEHICLE EQUIVALENT
FACTOR (DVEF): CONCEPT
Basically the derivation of passenger car equivalent
is carried out by comparing the influence of a vehicle
entering into a stream of homogenous traffic on the
speed of the standard or reference vehicle. It is evident
that the effect of interference from other vehicles to
the movements of the reference vehicle is significantly
affected by the flow rate which in turn affects the speed
of the vehicles. Therefore, for the given composition
and roadway conditions, the vehicle equivalent factor
is most likely to vary as the flow rate and speed of the
stream, which changes from time to time. This is more
relevant in urban areas where hourly variations of flow
rate are quite significant. Under such circumstances,
it is desirable to derive the vehicle equivalent factors
which take care of the fluctuations in stream speed
and also the speed of the individual vehicles for the
urban roadway. Such vehicle equivalent factors vary
dynamically as the speed of the reference vehicle in
the mixed flow changes and hence are called Dynamic
Vehicle Equivalent Factors (DVEF).
The dynamic vehicle equivalent factor includes effect
of the traffic interaction and interference on vehicular
speed. The speed and maneuverability of the vehicles
in stream is generally governed by the vehicles which
has high proportion in the stream. In the present study,
dynamic vehicle equivalent factors are estimated
considering two-wheeler as well as car as reference
vehicle. The terrain is flat and road stretches are straight
and free from effect of intersections, hence vehicles are
assumed to move with constant speed in the trap area.

Table1 Vehicle Categories and Representative Dimensions

No. Category Vehicles Included Average Dimensions (m) Projected Rectangular Area (m2)

1 2W Scooter, Motorbike, Mopeds 2.00 x 0.74 1.48

3-wheeled goods/passenger
2 3W 2.73 x 1.20 3.28
vehicles

3 Car Car, Jeep, Small Van 3.69 x 1.55 5.72

4 Mini Bus Mini Bus, Large vans 6.90 x 2.20 15.18

5 Std. Bus Standard Bus 10.50 x 2.45 25.73

6 LCV Small 4-wheeled goods vehicles 5.00 x 1.50 7.50

7 Truck* All large trucks 7.50 x 2.35 17.63

8 Cycle* Bicycles 1.90 x 0.45 0.86

9 Pedal Rickshaw* Pedal rickshaw, Carts 2.70 x 0.95 2.57

* Source: Chandra (2004)

Journal of the Indian Roads Congress, July-September 2013

 Dynamic Vehicle Equivalent Factors for Characterisation of Mixed Traffic for Multilane
Metropolitan Arterials in India
The traffic survey data revealed the predominance of
two wheelers with nearly 50% to 65% share followed
by car with 20% to 30% and three-wheelers with
10% to 20% share. In north Indian cities, bicycle and
pedal rickshaws have significant share in total traffic.
Bicycle share observed in Lucknow, Kanpur and Patna
is respectively 9.5%, 22.5% and 19.9% while that of
pedal ricksaw is 6.5%, 17.7% and 13.1% respectively.
Chandras (2000) modified homogenization co-efficient
approach is used in which effect of static characteristics
is incorporated by comparing projected area of reference
vehicle and the other vehicle in terms of area ratio ().
Speed ratio described as ratio of speed of reference
vehicle to the other vehicle is adopted to include effect
of speed differential as dynamic characteristics.
Mathematically,
DVEFX = (/)
Where, DVEFX = Dynamic Vehicle Equivalent Factor
considering x reference vehicle
x-y = Speed Ratio = Vx/Vy, x-y = Area Ratio = Ax/Ay, Vx
= Spot speed of x reference vehicle, Vy = Spot speed
of y reference vehicle, Ax = Projected area of x
reference vehicle, Ay= Projected area of y reference
vehicle
Extracted data from recorded videos is entered in Excel
sheets and DVEF are calculated using car and two-
wheeler as reference vehicle for each observation of
all selected cities. The term DCU (Dynamic Car Unit)
is the vehicle equivalency with reference to Car [For
example 2W DCU = (Vcar/V2W)/(Acar/A2W)]. Similarly,
the term DTU (Dynamic Two-wheeler Unit) indicates
the vehicle equivalency with reference to two-wheeler.
Scattered plots of DCU (/DTU) versus flow rate are
drawn to know the trend and range of DCU/DTU
values for each category of vehicles.
5 GROUPING APPROACH FOR FLOW
RATE AND COMPOSITION
To have conclusive relationship, observed flow rates and
the respective DCU and DTU values are grouped with
250 veh/hr class intervals. For each group, minimum,
maximum, mean value and standard deviation in DCU
and DTU values are calculated. Likely variation in
DCU and DTU values at 5% significance level is also
Journal of the Indian Roads Congress, July-September 2013
209
calculated for all the vehicle types. Variation is found
in standard deviation of DCU and DTU values for each
vehicle category. Standard deviation of DCU and DTU
values indicates driver behaviour influence on vehicular
speed choice and also non-uniformity of vehicular flow
throughout the observation interval of 1 minute. It is
possible that interaction may not be same throughout
the observation interval as well as in all lanes. Outer
lane is generally influenced by slow moving vehicles
and pedestrians. These factors ultimately get reflected
in the standard deviation. As large amount of data is
used for the analysis, normal distribution is expected;
hence mean of each flow group can be considered
representing the behaviour of corresponding traffic
parameters. Upper and lower limit has been defined for
each flow group to filter outlier data points using 5%
significance level. As data is realistic and limited to
certain hours for each city, many times it happens that
only two-three data points are available in some flow
group giving result mismatching with the observed
trend. Such points are eliminated judiciously.
The effect of traffic composition on DCU and DTU
values are captured by partitioning traffic data into
sub-groups based on proportion of 2W (predominant
mode) and slow moving non-motorised vehicles (major
influencing modes) of bicycle and pedal rickshaw.
Initially groups are made based on proportion of 2W in
traffic. Three groups are formed as (a) 2W proportion
less than 40% of total traffic volume (a2w4), (b) 2W
proportion between 40% to 60 % of traffic volume
(a2w46) and (c) 2W proportion more than 60% of
traffic volume (a2W6). Each group is subdivided in 4
subgroups based on slow moving vehicles i.e. bicycle,
pedal rickshaw and other non motorized vehicles if
any. These subgroups are represented as (a) Proportion
of slow moving vehicles less than 5% of total traffic
volume (s5), (b) Proportion of slow moving vehicles
between 5% and 15% of traffic volume (s515), (c)
Proportion of slow moving vehicles between 15 %
and 25% of traffic (s1525) and (d) Proportion of slow
moving vehicles more than 25% of traffic volume (s25).
Thus, flow group code a2w46s25 represents subgroup
having two-wheeler share between 40% to 60% and
slow moving vehicle share more than 25%.
210 Joshi & Vagadia on
6 DVEF DERIVATION ALGORITHM
As repetitive work consumes lot of time, a program is
prepared in Scilab software to calculate DCU and DTU
values. Flow rate, traffic composition and classified
spot speeds are used as input. In output, program
gives composition based data partition and processed
Fig.1 Steps for Derivation for DCU/DTU
Fig. 2 DCU for 2W & DTU for Car v/s Flow Rate (Trvm46s5)
Journal of the Indian Roads Congress, July-September 2013
DCU-DTU values for each category of vehicles. Flow
chart in Fig. 1 highlights the main steps in DCU/DTU
calculation and scatter plot preparation process.
For all selected seven cities, DCU and DTU values are
estimated and plotted with respect to flow rate. Scatter
diagrams are also prepared for all flow group DCU and
DTU values (ungrouped data) versus flow rate and DCU
and DTU values versus composition of corresponding
vehicle category. For example, scatter plots for some
selected vehicle categories for subgroup Trvm46s5
(Thiruvananthapuram traffic having 2W proportion
between 40% to 60% and slow moving vehicle
less than 5% of total traffic) are shown in Fig.2 and
Fig. 3. DCU/DTU variation trends are on expected
lines as reported in the literature. It is observed that
DCU value for 2W ranges from 0.25 to 0.30. DTU
value for Car ranges from 3.4 to 3.8. DCU for 2W
decreases while DTU for Car increases with increase
in flow rate.
It is observed that for 3W, DCU value ranges from 0.6
to 0.8 and DTU value ranges from 2.3 to 2.7. Both DCU
and DTU values decrease with increase in flow rate.
7 ESTIMATED DCU AND DTU VALUES
OF ALL CITIES
Table 2, 3 and 4 show range of DCU and DTU values
obtained for various traffic composition groups for
each vehicle category for all selected cities. Values are
shown for only prime traffic composition groups and
all groups combined. Second row in Table 2 to 4 shows
flow rate variation observed in each city. Also it should
be noted that as data is based on field observations, DCU/
 Dynamic Vehicle Equivalent Factors for Characterisation of Mixed Traffic for Multilane
Metropolitan Arterials in India 211

Fig. 3 DCU & DTU for 3W v/s Flow Rate (Trvm46s5)

DTU values cannot be derived due to unavailability of
sufficient data points of some vehicle category in few
traffic composition groups.
8 DVEF: DERIVATION METHODOLOGY
Estimated values in Table 2 to Table 4 are informative
but at application level, it is difficult to implement
because it does not consider both influencing
parameters i.e. traffic volume (flow rate) and traffic
composition simultaneously. (Road geometry influence
is considered already by modifying observed flow
rates). Therefore, tables should be prepared for each
traffic composition subgroup in which DCU and DTU
values are available for different flow rate range. But
this may be cumbersome at application level. A new

Table 2 DCU and DTU Values for Thiruvananthapuram and Jaipur City
City Thiruvananthapuram Jaipur
Carriageway width 9.0 m 10.5 m

Modified Flow Rate (veh/hr) Modified by 1.17 for width 1000-3500 1500-6750

Composition Group 6S5 46S5 All 6S5 46S5 All

Vehicle Type DCU
2W 0.26-0.30 0.25-0.30 0.25-0.30 0.26-0.30 0.24-0.28 0.25-0.30
3W 0.6-0.8 0.6-.8 0.6-0.8 0.7-0.9 0.7-0.8 0.7-0.8
MB - 2.5-3.0 2.5-3.0 3.0-5.0 3.0-6.0 3.0-6.0
SB 4.0-6.0 4.0-6.0 4.5-5.5 - - -
LCV - 1.3-1.6 1.3-1.6 1.5-2.0 1.5-2.0 1.5-2.0
CYCL - 0.35-0.55 0.35-0.55 0.40-0.60 0.40-0.50 0.45-.055
TRK - 3.0-4.0 3.0-4.0 - - -
Vehicle Type DTU
Car 3.4-3.8 3.4-3.8 3.4-3.8 3.0-4.0 3.5-4.0 3.2-4.0
3W 2.3-2.7 2.3-2.7 2.3-2.7 2.5-3.0 2.6-3.2 2.6-3.2
MB - 9.0-10.5 9-10.5 12.0-17.0 10.0-20.0 10.0-20.0
SB 14.0-20.0 14.0-20.0 15.0-20.0 - - -
LCV - 4.5-5.5 4.5-6.0 5.0-7.0 6.0-7.5 5.5-7.0
CYCL - 1.2-1.8 1.3-1.7 1.7-2.0 1.5-2.0 1.5-2.0
TRK - 11.0-15.0 12.0-16.0 - - -

Journal of the Indian Roads Congress, July-September 2013

212 Joshi & Vagadia on
Table 3 DCU and DTU Values for Pune and Surat City

City Pune Surat

Carriageway width 11.6 m 10.6 m

Modified Flow Rate (veh/hr) Modified by 0.91 for width 2750-12250 1000-4500

Composition Group 6S5 46S5 All 4S5 46S5 All

Vehicle Type DCU
2W 0.20-0.26 0.22-0.26 0.20-0.25 0.35-0.45 0.40-0.45 0.40-0.42
3W 0.5-0.7 0.6-0.7 0.5-0.9 0.85-1.05 0.9-1.0 0.9-1.0
MB 2.0-3.0 2.4-3.0 2.0-4.0 - 4.2-4.8 4.2-4.8
SB 3.5-5.5 4.5-6.0 4.0-6.0 - - -
LCV 1.0-1.6 1.1-1.6 1.0-2.0 2.1-2.5 2.1-2.5 2.1-2.5
CYCL 0.20-0.40 1.2-0.5 0.15-0.50 0.50-0.55 0.5-0.6 0.5-0.6
TRK 3.0-4.5 - 3.0-5.0 - 5.2-5.6 5.0-5.0
Vehicle Type DTU
Car 4.0-5.0 3.8-4.6 3.0-5.0 2.2-2.8 2.3-2.6 2.4-2.6
3W 2.4-3.0 2.5-3.0 2.4-2.8 2.1-2.7 2.3-2.6 2.2-2.5
MB 9.0-12.0 10.0-12.0 10.0-15.0 - 10.0-12.0 9.0-12.0
SB 15.0-24.0 19.0-23.0 15.0-25.0 - - 17.0-19.0
LCV 4.5-7.0 5.0-6.0 5.0-7.0 5.4-5.8 5.0-6.0 5.0-6.0
CYCL 0.6-1.6 0.8-1.9 1.1-1.7 1.1-1.4 1.1-1.5 1.3-1.5
TRK 14.0-18.0 14.0-18.0 12.0-18.0 - 12.0-16.0 12.0-16.0

Table 4 DCU and DTU Values for Kanpur, Lucknow and Patna City
City Kanpur Lucknow Patna
Carriageway width 10.9 m 10.5 m 10.5 m
Modified Flow Rate Modified by 1.4 for width and Modified by 1.4 for side
1750-8000
(veh/hr) friction 3250-6250 friction 500-5550
Composition Group 46S25 All 46S515 46S1525 All 4s25 All
Vehicle Type DCU
2W 0.18-0.25 0.17-0.25 0.23-0.25 0.22-0.25 0.22-0.25 0.23-0.30 0.23-0.31
3W - - 0.6-0.7 06-0.7 0.6-0.7 0.6-0.8 0.6-0.8
MB 2.0-4.0 2.0-4.0 3.0-4.0 3.0-3.7 3.0-3.5 - 2.5-4.0
SB - - 5.0-6.0 4.0-6.0 4.5-6.0 - -
LCV - - 1.4-1.8 1.4-1.8 1.4-1.8 - -
CYCL 0.22-0.30 0.20-0.30 0.3-0.4 0.25-0.35 0.25-0.40 0.30-0.50 0.25-0.50

Journal of the Indian Roads Congress, July-September 2013

 Dynamic Vehicle Equivalent Factors for Characterisation of Mixed Traffic for Multilane
Metropolitan Arterials in India 213

PEDRXW 0.65-0.90 0.65-0.90 1.2-1.6 1.2-1.6 1.2-1.6 1.0-2.0 1.0-2.2

TRK - - - - - - -
Vehicle Type DTU
Car 4.0-5.5 4.0-5.5 4.0 -4.5 3.5-4.5 4.0-4.5 3.0-4.5 3.0-4.5
3W - - 2.5-3.0 2.5-3.0 2.5-3.0 24-3.0 2.4-3.0
MB 10.0-15.0 10.0-16.0 11.0-15.0 12.0-16.0 12.0-16.0 - 10.0-14.0
SB - - 20.0-25.0 18.0-25.0 18.0-25.0 - -
LCV - - 6.0-7.0 6.0-7.0 5.5-7.0 - -
CYCL 1.1-1.2 1.1-1.2 1.4-1.6 1.0-1.5 1.2-1.6 1.0-1.8 1.0-1.7
PEDRXW 3.4-3.8 3.4-3.8 4.5-7.0 5.0-7.0 5.0-7.0 4.0-7.0 4.0-7.0
TRK - - - - - - -

approach is adopted in which first traffic volume (flow
rate) in veh/hr is converted in either DCU/hr or DTU/
hr by predefined reference DCU/DTU values based on
composition. This converted equivalent traffic volume
can be used to estimate refined DCU/DTU values from
tables based on traffic volume in DCU or DTU for
various range of flow.
DCU and DTU values for various level of composition
of major vehicle categories i.e. 2W, 3W, Car, Bicycle
and Pedal Rickshaw which represent typical Indian
urban traffic are estimated using grouping approach and
considering whole data of all cities with composition
groups of 5% class interval (0 to 5%, 5 to 10 % etc.).
Modal share of other vehicle categories like Mini
Bus, Std. Bus and LCV is very less; hence accuracy
of their DCU/DTU values for conversion purpose is
insignificant. Therefore, reference values for these
categories are taken as average value obtained for
whole data. Table 5 shows reference DCU and DTU
values to convert traffic volume in veh/hr to DCU/hr
and DTU/hr.
DCU/hr or DTU/hr at initial stage and then refined
DCU and DTU values can be obtained from Chart/
Table having relationship between traffic volume in
DCU/DTU and DCU/DTU values for each vehicle
category. To prepare such chart, the traffic volume
(flow rate) for each data point for each city is converted
in DCU/hr and DTU/hr by using corresponding DCU
and DTU values in each cities. After converting
traffic volume in DCU/hr and DTU/hr for all data
points, scatter plots are drawn using flow grouping
approach.
Fig. 4 shows variation in DCU for 2W with flow rate in
DCU/hr. 2W DCU value decreases from 0.32 to 0.22
with increase in flow rate. Strong linear relationship is
reflected by high value of R2.

Table 5 is required to convert traffic volume either in

Table 5 Reference DCU/DTU Values for Fig. 4 DCU for 2W v/s Flow Rate in DCU/hr

Traffic Volume Conversion

Fig. 5 shows 3W DCU variation with flow rate in
To convert in 2W 3W Car MB SB TRK LCV CYCL PEDRXW
DCU/hr. DCU for 3W decreases from 0.85 to 0.5 with
DCU 0.3 0.8 - 3.5 5.4 4.3 1.7 0.3 1.1 increase in flow rate. R2 value of the linear relationship
DTU - 2.6 3.5 12.5 20.0 15.0 6.0 1.3 4.5 is also very good.

Journal of the Indian Roads Congress, July-September 2013

214
Fig. 5 DCU for 3W v/s Flow Rate in DCU/hr
Fig. 6 shows variation in DCU for Bicycle with flow
rate in DCU/hr. Bicycle DCU value decreases from 0.5
to 0.25 with increase in flow rate in linear fashion and
is explained very well with significant R2.
Fig. 6 DCU for Bicycle v/s Flow Rate in DCU/hr
Fig. 7 shows variation in DCU of LCV with flow rate in
DCU/hr. DCU for LCV also decreases from 2.0 to 1.4
with increase in flow rate.
Fig. 7 DCU for LCV v/s Flow Rate in DCU/hr
Fig. 8 shows DCU variation in DCU for Mini Bus with
flow rate in DCU/hr. DCU for Mini Bus decreases from
4.0 to 2.5 with increase in flow rate.
Fig. 8 DCU for Mini Bus v/s Flow Rate in DCU/hr
Journal of the Indian Roads Congress, July-September 2013
Joshi & Vagadia on
Fig. 9 shows variation in DCU for Standard Bus
with flow rate in DCU/hr. DCU value for SB is
found to decrease from 6.0 to 4.0 with increase in
flow rate.
Fig. 9 DCU for Std. Bus v/s Flow Rate in DCU/hr
Fig. 10 shows variation in DCU for Pedal Rickshaw
(PEDRXW) with flow rate in DCU/hr. PEDRXW DCU
value decreases from 2.4 to 1.0 up to flow rate of 2000
DTU/hr and then it fluctuates in the range of 1.0 to 1.5.
Here, speed of reference vehicle car reduces rapidly
with increase in traffic volume while speed of pedal
rickshaw remains largely unaffected. Hence, at high
traffic volume levels, difference between speed of car
and pedal rickshaw reduces considerably and it varies
over a narrow range due to higher level of mutual
interaction resulting into stable DCU value.
Fig.10 DCU for Pedal Rickshaw v/s Flow Rate in DCU/hr
Fig. 11 shows variation in DTU for Car with flow rate
in DTU/hr. Car DTU value ranges from 3.0 to 4.5.
It increases beyond flow rate of 6500 DTU/hr with
increase in flow rate. This happens because of ability
of two wheelers to maintain relatively high speed at
high traffic volume due to their less width and better
maneuverability. Fig. 12 shows 3W DTU variation
with flow rate in DTU/hr. DTU of 3W ranges from 2.5
to 2.8 and increases with the flow rate. Fig. 13 shows
Bicycle DTU variation with flow rate in DTU/hr.
 Dynamic Vehicle Equivalent Factors for Characterisation of Mixed Traffic for Multilane
Metropolitan Arterials in India
DTU of Bicycles ranges from 1.7 to 1.3. It decreases
marginally with increase in flow rate.
Fig. 11 DTU for Car v/s Flow Rate in DTU/hr
Fig. 12 DTU for 3W v/s Flow Rate in DTU/hr
Fig. 13 DTU for Bicycle v/s Flow Rate in DTU/hr
Fig. 14 shows variation in LCV DTU with flow rate
in DTU/hr. DTU of LCV ranges from 5.2 to 6.5 and
increases linearly with increase in flow rate. Fig.15
shows variation in DTU for Mini Bus with flow rate
in DTU/hr. DTU for MB value ranges from 10.0 to
14.0. DTU for MB increases linearly with increase in
flow rate. Fig. 16 shows variation in DTU for Standard
Bus with flow rate in DTU/hr. DTU of SB ranges from
15.5 to 21.5. DTU for SB also increases linearly with
increase in flow rate.
Fig. 14 DTU for LCV v/s Flow Rate in DTU/hr
Journal of the Indian Roads Congress, July-September 2013
215
Fig. 15 DTU for Mini Bus v/s Flow Rate in DTU/hr
Fig. 16 DTU for Std. Bus v/s Flow Rate in DTU/hr
Fig. 17 shows variation in DTUs for Pedal Rickshaw
with flow rate in DTU/hr. DTU for PEDRXW decreases
linearly from 7.0 to 4.0 with increase in flow rate. Thus,
with increase in flow rate, DTU values for all motorised
vehicles which are bigger in size than the two wheeler
increases, whereas DTU for NMT modes of bicycle and
pedal rickshaw decreases with increase in flow rate.
Fig. 17 DTU for Pedal Rickshaw v/s Flow Rate in DTU/hr
To find the DCU/DTU value for a vehicle category,
first traffic volume is to be converted either in
DCU/hr or DTU/hr using reference values given in
Table 5. DTU value should be adopted in traffic where
two wheelers have dominant modal share. Based on
charts in Fig. 4 to 17; DCU/DTU values are suggested
for various traffic volume level as shown in Table 6.
Three values are given for low, moderate and near
capacity traffic volume. For intermediate values of
traffic volume, DCU/DTU can be calculated by linear
interpolation. Flow rate in DCU/hr or DTU/hr is to be
used for calculating refined DCU and DTU values from
Table 6.
216 Joshi & Vagadia on
Table 6 Suggested DCU/DTU Values for Different Traffic Volumes
DCU value DTU Value
Vehicle
Flow Rate in DCU/hr Flow Rate in DTU/hr
Category
1000 and Less 3500 6000 and More 1000 and Less 5000 12000 and More
Two wheeler 0.31 0.28 0.26 1.0 1.0 1.0
Three wheeler 0.8 0.7 0.6 2.5 2.6 2.8
Car 1.0 1.0 1.0 3.0 3.0 4.0
Mini Bus 4.0 3.4 2.7 10.0 11.5 14.0
Standard Bus 5.8 5.2 4.6 15.0 17.5 21.5
Truck 4.0 4.5 4.0 13.0 14.0 16.0
LCV 2.0 1.6 1.3 5.0 5.7 6.5
Bicycle 0.45 0.35 0.25 1.6 1.5 1.3
Pedal Rickshaw 2.0 1.2 1.3 7.6 6.5 4.4

9 Comparison OF DVEF VALUES with is contradictory to the present IRC recommendation.

Existing Guidelines
Table 7 below shows comparison of estimated DCU
values in the study and PCU values recommended by
IRC: 106-1990: Guidelines for Capacity of Urban
Roads in Plain Areas. In IRC-106, PCU factors are
given only based on composition of concerned vehicle
category in total traffic volume. PCU values for 2W
and 3W are more than the suggested values of DCU
in the present study. As 2W are dominating in Indian
urban traffic, PCU value of 0.75 leads to unrealistic
traffic volume in PCU/hr more than the prescribed
capacity for roads with high traffic volume in peak
hours. DCU values are found decreasing with increase
in composition for most of the vehicle categories which
The difference in the DCU values found in the study
and PCU recommended in IRC: 106-1990 may be
also due to the definition or the method of estimation
adopted. IRC: 106-1990 doesnt throw light on the
approach used for derivation of PCU values. In this
study, Chandras approach which uses both speed and
projected area of the vehicle is used. Therefore the
vehicles like two-wheeler and three-wheeler which
can maintain relatively higher speed than other vehicle
categories at high traffic volume have less DCU values
than PCU suggested in IRC: 106-1990. The truck and
bus have large area compared to new technology cars
having high degree of maneuverability which results
in high DCU values compared to recommended PCU
values.

Table 7 Comparison of DCU Values with IRC Guidelines

Equivalent PCU factors Suggested DCU values
Sr. No. Vehicle type % Composition Flow Rate in DCU/hr
5% 10% and above 1000 and Less 3500 6000 and More
1. Two wheelers 0.5 0.75 0.31 0.28 0.26
2. Car 1.0 1.0 1.0 1.0 1.0
3. Auto-rickshaw 1.2 2.0 0.8 0.7 0.6
4. LCV 1.4 2.0 2.0 1.6 1.3
MB 4.0 3.4 2.7
5. Truck or Bus 2.2 3.7 SB 5.8 5.2 4.6
TRK 4.0 4.5 4.0
6. Cycle 0.4 0.5 0.45 0.35 0.25
7. Cycle Rickshaw 1.5 2.0 2.0 1.2 1.3

Journal of the Indian Roads Congress, July-September 2013

 Dynamic Vehicle Equivalent Factors for Characterisation of Mixed Traffic for Multilane
Metropolitan Arterials in India 217

10 VALIDATION OF SUGGESTED VEHICLE considered as 5143 pcu/hr (=3600/0.7) as per IRC:

EQUIVALENT FACTORS
To verify the significance of suggested vehicle
equivalent factors, traffic volume data of selected cities
is used. Peak hour traffic volume which is generally
of prime importance in urban areas is converted from
veh/hr to DCU/hr as shown in Table 8. Traffic volume
is also converted in PCU/hr as per IRC guidelines for
urban areas. Capacity of 3-lane one-way road can be
106-1990, Guidelines for Capacity of Urban Roads in
Plain Areas. Here it is observed that by using suggested
dynamic vehicle equivalent factors, observed traffic
volume which is very high (particularly in case of
Pune) than the capacity as per IRC guidelines, reduces
drastically near to capacity as per IRC guidelines.
Thus suggested DCU values seem to be more realistic.
Traffic volumes converted in DTU/hr are indicating the
carrying capacity of the road vis-a-vis of cars.

Table 8 Validation of Suggested DCU Values

Total
City & Mini Std. Bi- Pedal Total Total Total
2W 3W 4W LCV Truck PCU/
Road Bus Bus cycle Rxw veh/hr DCU/hr DTU/hr
hr

Pune,
6949 975 1999 94 146 121 4 67 0 10355 9113 5506 23068
Karve Road

Jaipur,
JL Nehru 3575 200 2455 22 0 42 4 177 12 6487 5581 3855 14883
Road

Lucknow,
Vidhan-sabha 3777 845 694 14 58 88 5 641 239 6361 6188 3366 12899
Road

Surat, Gaurav
1938 460 1039 8 17 38 7 84 0 3591 3201 2210 7554
Path

11 Observations AND CONCLUSIONS behaviour reduces with increase in traffic

volume due to traffic flow constraints as shown
Following are the observations made during the process by dark lines in Fig. 18.
of DVEF derivation.
v) Deviation in vehicle dimensions plays vital role
i) Each city has one or two dominant typical traffic particularly for LCV and Mini bus. Loading
flow groups based on composition. condition also affects DCU/DTU values for
LCV, Truck and Pedal Rickshaw reflecting on
ii) Traffic being stochastic phenomenon, same speed variation.
interaction between the vehicles is rare.
However, with the use of extensive data,
representative effect can be captured.

iii) As study is based on the real world observations,

results may not exactly match simulation results.
Deviation in estimated DCU/DTU value is
observed due to certain reasons as illustrated in
Fig. 18 and 19.

iv) Driver behaviour and local conditions (side

friction and traffic control measures etc.) lead
Fig. 18 Deviation in DCU/DTU Due to Driver Behaviour and
to higher deviation in results. Effect of driver Vehicle Dimension

Journal of the Indian Roads Congress, July-September 2013

 218 Joshi & Vagadia on
Fig. 19 Deviation in DCU/DTU Due to Intersections & Non-
Uniformity of Flow
vi) Traffic is not uniformly mixed. Within stream,
there are vehicle groups of 2W, Car, Bicycle etc.
Therefore type of interaction may be different.
i.e. 2W-2W, 2W-Car, Car-Car, Car-Bicycle etc.
vii) For observation interval of 1 minute also,
traffic is not uniform throughout the whole
time period of one minute. This results in less
accurate estimate of flow rate for particular
observation. At low traffic volume, it may be
possible that there is no interaction between
vehicles or different type of interaction for same
composition and same traffic volume giving
different results. This deviation is shown in Fig.
19.
viii) In urban areas, when distance between
intersections is 500m or less, the traffic flow
at mid-blocks is always under the effect of
intersection control particularly during peak
hours i.e. at high traffic volume. Effective flow
rate may be 20-25% more than the observed
flow rate because 10-15 seconds of observation
interval may have very less volume and
remaining 45-50 seconds may carry very high
traffic volume. Vehicles in the beginning of
the broken traffic flow segment move with
high speed and hence display entirely different
type of interaction. In such condition, speed
observation of vehicle categories with less
share in traffic has large standard deviation for
Journal of the Indian Roads Congress, July-September 2013
speeds and hence variation their DCU/DTU
values. This can happen for bus, LCV, truck
etc. This deviation is small at low volume and
large at high volume as shown in Fig.19.
ix) Considering unavoidable possible deviations in
DCU and DTU values and ease at application
level, suggested methodology to derive refined
DCU/DTU values can become very helpful to
urban traffic and transportation planners.
ACKNOWLEDGEMENT
Authors acknowledge the financial support received
from the Indian Roads Congress, New Delhi and S.V.
National Institute of Technology, Surat for carrying out
the study.
REFERENCES
1. Arasan, V.T. and Arkatkar, S.S. (2010),
Microsimulation Study of Effect of Volume
and Road Width on PCU of vehicles
under Heterogeneous Traffic, Journal of
Transportation Engineering, Vol.136(12),
pp. 1110-1119.
2. Arasan, V.T. and Koshy R.Z. (2004), Simulation
of heterogeneous traffic to derive capacity and
service volume Standards for urban roads,
Journal of Indian Road Congress, Vol.65-1,
pp. 219-242.
3. Arasan, V.T. and Krishnamurthy K. (2008),
Effect of Traffic Volume on PCU of Vehicles
under Heterogeneous Traffic Conditions, Road
& Transport Research, Vol.17-1, pp. 32-48.
4. Chandra, S. (2004), Capacity Estimation
Procedure for Two-Lane Roads under Mixed
Traffic Conditions, Journal of Indian Road
Congress, Vol.65-1, pp. 139-170.
5. Chandra, S. and Kumar, U. (2003), Effect of
Lane Width on Capacity under Mixed Traffic
Conditions in India, Journal of Transportation
Engineering, Vol. 129(2), pp. 155-160.
6. Chandra, S. and Sikdar, P.K. (2000), Factors
affecting PCU in mixed traffic situations on
 Dynamic Vehicle Equivalent Factors for Characterisation of Mixed Traffic for Multilane
Metropolitan Arterials in India
urban roads, Roads &Transport Research, Vol.
9(3), pp. 40-50.
7. Dey P.P., Chandra S. and Gangopadhyay
S. (2007), PCU Factors for Two Lane
Roads, HRB, Indian Road Congress No.77,
pp. 111-119.
8. Dey, P. P., Chandra, S. and Gangopadhyay, S.
(2008), Simulation of Mixed Traffic Flow on
Two-Lane Roads, Journal of Transportation
Engineering, Vol. 134 (9), pp. 361-369.
9. Indian Roads Congress. (1990), Guidelines for
capacity of roads in rural area, IRC code of
Practice, IRC: 64-1990, New Delhi.
10. Joshi, G., Sinha, V. and Patel, J. (2011),
Heterogeneous Traffic Characterisation and
Flow Behaviour Modeling for Metropolitan
Arterial in India, Journal of the Eastern Asia
Society for Transportation Studies, Vol. 9,
pp. 1684-1699
11. Justo, C.E.G. and Tuladhar S.B.S. (1984),
Passenger Car Unit Values for Urban
Roads, Journal of Indian Roads Congress,
pp. 188-238.
The views expressed in the paper are personal views of the Authors. For any query, the author may be contacted at E-mail: gjsvnit92@gmail.com
Journal of the Indian Roads Congress, July-September 2013
219
12. Nguyen, C.Y., Sano, K. and Minh, C.C. (2007),
Dynamic Motorcycle Unit and Mean Stream
Speed under Mixed Traffic Conditions on Urban
Roads, Journal of the Eastern Asia Society for
Transportation Studies, Vol. 7.
13. Nguyen, C.Y., Sano, K., Tu, T.V. and Tan,
D.T. (2010), Methodology for Motorcycle
Equivalent Unit at Road-Segments in Urban
Roads, TRB 2010 Annual Meeting CD-ROM,
89th Annual Meeting of the Transportation
Research Board, January 10-14, Washington D.C.
14. Ramanayya T.V. (1988), Highway capacity
under mixed traffic conditions, Traffic
Engineering and Control, U.K., 29(5),
pp. 284-287.
15. Vagadia, D.H. (2012), Development of
Dynamic Vehicle Equivalent Factors (DVEF)
Under Mix Traffic Condition in Context of
Multilane Metropolitan Arterials in India,
unpublished M.Tech. Thesis, SVNIT, Surat,
India.

A Conceptual STRATEGY for DURABILITY
of Concrete Structures in the Arabian Gulf and
similar surrounds afflicted by Humidity, high
Temperatures, and Chlorides and Sulphates, etc.
1 INTRODUCTION AND BACKGROUND
The Challenge:
The Arabian Gulf waters are one of the most highly
charged with the attacking chlorides, sulphates and
molluscs the tripple killers of structuredurability.
Indeed same is the case with almost all coastal waters
in the tropics.
Evolvement and fixing of the various technical
requirements and parameters for design of the
structures in such waters that are laiden with chlorides,
sulphates and molluscs, are a class of their
own that call for unstinted clinical attention
of the best minds there are in bridge/
structure design and construction in such testing
circumstances.
2 OVERALL REQUIREMENTS
DURABILITY
2.1 General
The projects here are located in most severe marine
enviromment with extreme exposure in terms of:
high water salinity and water temperatures; airborne
salt spray and salt-laden dust; high temperatures
and temperature gradients; the potential occurrence
of limestone-devouring marine borers (also called:
molluscs or piddocks).
This most extreme environmental exposure, combined
with the agreed project service lifetime of over 100
* Principal Advisor, ICT Pvt. Ltd., Professor Emeritus, CoEP. E-mail: rainavk@gmail.com
Written comments on this Paper are invited and will be received upto 30th October, 2013
Journal of the Indian Roads Congress, July-September 2013
Paper No. 599
V.K. Raina*
years, calls for the highest attention during design,
construction, operation and maintenance.
To achieve the more than 100 years service life in this
extreme environment is a very demanding task that can
not be fulfilled with traditional and simple materials and
procedures. This makes it important that state-of-the-art
concrete durability technology be incorporated in the
project. The basic methodology in this respect is to
define exposure conditions for a number of relevant
exposure classes, and then use these classes as basis
for a strategy to be consistently incorporated
in the design and specification. For each class the
relevant deterioration mechanisms be defined, and
the strategy be developed accordingly. Wherever
possible, a multi-stage protection strategy should be
FOR
applied so that the structures durability is ensured with
some redundancies.
A multi-stage protection strategy comprises two or
more selected barriers that each:
prevents
l  or slows down transportation of
aggressive chemicals/pollutants from the
environment into the concrete structure,
or/and
l passivates
 the subsequent potential
deterioration mechanism.
 A Conceptual Strategy for Durability of Concrete Structures in the Arabian Gulf and similar
surrounds afflicted by Humidity, high Temperatures, and Chlorides and Sulphates, etc.
2.2 Exposure Conditions and Deterioration
Mechanisms
2.2.1 Exposure Conditions
2.2.1.1 Sea water
The exposure condition exerted from sea water in the
shallow areas of the Arabian Gulf between Qatar and
Bahrain originates from:
l Sea water and ground water pressure
This zone comprises all foundations and
piers up to lower level for splash and
tidal zone. Sea water with salts enters the
concrete by hydraulic pressure and salt
diffusion.
l Water splash
The splash zone varies and extends from
almost 0 m near land and in certain
other parts up to about +4 m above high
tide level (HTL). The concrete surface
is subjected to alternate wetting (which
causes capillary suction) and drying,
resulting in repetitive salt concentration
build up again and again.
l Water spray
Likewise the spray zone also varies and
extends from a few metres near land and
in certain other parts up to about 6 to 7 m
above HTL. The concrete surface is thus
subjected to alternate wetting (capillary
suction) and drying, again causing salt
concentration build up.
l Sea water composition, chlorides,
sulphates, other salts, pH
The Arabian Gulfs shallow areas are
characterised by very saline water with up
to about 4 4.5 % salts (40,000 45,000
ppm). It is noted that for instance further
south in the Bay of Salwa (between Saudi
Arabia and Bahrain) the salinity reaches
about 6%. Out of the salinity measured,
more than half the salts are chlorides.
Journal of the Indian Roads Congress, July-September 2013
221
l Temperature
 The sea water temperature varies between
o
about 17 C and 37oC with a ground water
temperature of about 27oC.
2.2.2 Molluscs (Marine Borers)
One group of molluscs (also called piddocks or marine
borers) of the family: Pholadidae has an ability to
excavate for itself depressions in calcareous rocks and
some types of concrete. Hence concrete with limestone
(calcareous) aggregate can be the victim. Concrete with
e.g. granite and gabbro stones can only be destroyed
by molluscs which rasp the material off which is
unusual.
Molluscs have been identified in the Arabian Gulf.
2.2.3 Air
The exposure to air-borne actions, in the sea areas of
the Arabian Gulf, originates from:
l Airborne chlorides, sulphates, sea water
spray
causing Salt concentration same or higher
than in the sea water.
l Windborne dust and sand
Dust and sand storms occur, bringing
material which is contaminated with
chlorides (NaCl and Ca Cl2) and sulphates
(specially gypsum).
l Humidity
This can be very high, even 100% during
many months in a year, and this eases
movment of chloride and sulphate ions that
use mositure as the vehicle in travelling
through the pores in concrete in reaching
steel and cement.
l Temperature
The air temperature in the shade can very
o o
between close to 0 C and 47 C (Qatar
Meteorological Office for Al Arish and
Bahrain Meteorolgical Directorate at
Bahrain International Airport). The
 222 Raina on

maximum temperature on irradiated - Cement type

surfaces exposed to direct sun can reach
- Cement content
about 80O C.
- Additions, such as silica fume and fly ash
l Oxygen and carbon dioxide
- Water/cement ratio
The ambient atmosphere contains oxygen
and carbon dioxide, former enhances - Availability of oxygen.
the corrosion by chlorides and the latter - Chloride contamination (from surrounding
enhances carbonation (reacting with water and soil, and aggregates, mix-water,
calcium Hydroxide, forming calcium and curing-water)
carbonate and lowering pH, thus increasing
acidity). - denseness of concrete

2.3 Deterioration Mechanisms The main factor deciding whether or not chloride
induced corrosion takes place is the chloride
2.3.1 General
concentration around the steel (reinforcement and
Description of transport and deterioration mechanisms prestressing cables). The following general guidelines
has been described in many texts. apply:
2.3.2 Steel Reinforcement Corrosion
Chloride Risk of corrosion
Deterioration of concrete structures caused by corrosion concentration,
% by weight of
of reinorcement is caused by two different actions:
concrete
- Chloride-induced corrosion due to High
Below 0.05 % Low
concentration of chlorides within concrete Corrosion does not normally take place,
matrix, and but there are severe exemptions to this.

- Corrosion due to inward penetation of 0.05 0.10 % Medium

Carbonation-front from concrete surface.
l Chloride induced corrosion :
Above 0.10 %
For marine structures the main reason for
steel reinforcement corrosion is a high
chloride content which normally causes
dissolution of the protective passive
ironoxide layer on the steel, resulting
in active corrosion, typically as pitting
corrosion. Pit corrosion is usually a
localised corrosion which reduces the
rebar cross-section significantly over small
areas.
The chloride concentration which is
required to cause corrosion, the so-called
threshold value, is not a constant, but
varies with many parameters, of which the
most important are:
- Relative humidity in the concrete pores
Active corrosion is normally decided by rest
of the above mentioned parameters
High
Corrosion will normally take place
l Corrosion caused by carbonation-front
penetrating inwards from suface to
steel:
For the superstructures, corrosion may
be initiated when the concrete cover gets
completely carbonated. The penetration
of carbonation from the concrete surface
depends mainly on:
l Relative humidity in the concrete pores
l Cement type
l Cement content
l Additions such as silica fume and fly ash

Journal of the Indian Roads Congress, July-September 2013

 A Conceptual Strategy for Durability of Concrete Structures in the Arabian Gulf and similar
surrounds afflicted by Humidity, high Temperatures, and Chlorides and Sulphates, etc.
l Water/cement ratio
l Availability of carbon dioxide.
l Denseness of concrete.
The maximum rate of carbonation is
usually found when the relative humidity
is in the range 6080%. Carbonation does
not take place in saturated concrete and
very slowly in near-saturated concrete
because water-filled concrete pores
prevent ingress of carbon dioxide.
Cementiteous combinations which form
a high calcium hydroxide content during
hydration will cause a slower rate of
carbonation e.g. use of CEM I (ASTM
cement Type I) without any pozzolanas.
Combinations with ASTM-CEM III or
pozzolanas will give denser concete and
develop a lower penetrability of carbon
dioxide and thus a lower rate of carbonation
although the amount of calcium hydroxide
formed during the hydration is lower.
Corrosion, caused by loss of the
protective passive alkaline layer (high
pH) around steel due to carbonation
of the concrete cover, is usually the so-
called general corrosion which results
in a relatively uniform reduction of the
rebar cross section in large areas.
2.3.3 Sulphate Attack
Expansive sulphate reactions are seen in presence of
sulphate ions when CEM I with a moderate to high C3A-
content is used in concrete. The expansive reactions
will result in cracking and ultimately disintegration
of the concrete structure (due to formation of sulpho
Aluminate crystals which occupy more volume and
hence the expansive pressure).
Sulphate attacks are usually not seen in concrete with
ASTM-CEM-V (because it has low C3A content).
2.3.4 Seawater Attack
The combination of chloride and sulphate salts in
seawater usually will not cause much deleterious effects
Journal of the Indian Roads Congress, July-September 2013
223
of chemical reactions in the concrete if the concrete is
dense, has low water/cement ratio, and C3A-content is
below 8%.
For concrete made with cement with a C3A-content
above about 8%, there is a risk of expansive sulphate
reaction leading to cracking and disintegration of the
concrete matrix (sulpho aluminate expansion).
2.3.5 Salt Weathering
Alternate wetting and drying of concrete in the splash
zone and spray zone causes a build-up of salts within
the concrete pores in consequence of ingress of
seawater, Evaporation of the water leaves precipitated
(damaging) salts in the concrete pores.
During subsequent wetting the salts are re-hydrated,
generating a physical expansive pressure in the
concrete pores. Such pressure may lead to cracking and
spalling of the cement skin and later the surface layer
of the cement mortar. This effect reduces the thickness
of the concrete cover and thereby the protection of the
rebars.
This deterioration mechanism is specially effective in
the Arabian Gulf where the salt content in the seawater
is very high and the drying effect between wettings is
rapid due to high temperatures.
2.3.6 Alkali Aggregate Reactions
Alkali-silica and alkali-carbonate reactions are chemical
reactions between alkalis (sodium and potassium ions)
in cement and reactive silica or carbonate aggregate
particles (e.g. Opal). The reactions take place in a moist
alkaline environment only. The reaction products upon
drying, occupy a larger volume than the dissolved parts
of the aggregate particles and thereby create bursting
stresses within concrete, leading to cracking and
disintegration of the concrete matrix.
Alkali-silica reactions and alkali-carbonate reactions
are rarely found in the Arabian Gulf (due to non-
reactive C A used), but care must be taken to ensure
that the new aggregate sources are adequately surveyed
and tested.
2.3.7 Leaching
Concrete exposed to water with a low carbonate
content may be subjected to gradual dissolution of first
 224
the calcium hydroxide and later the calcium silicated
hydrates thus exposing the matrix of concrete. This
effect is called leaching and is normally not active in
the Arabian Gulf due to the high salt contents which
also includes carbonates.
2.3.8 Temperature
Difference between coefficients of thermal expansion
of coarse aggregate and cement mortar may lead to
internal cracking in the concrete when the concrete is
exposed to large temperature and moisture variations
(differential thermal movements).
As an example, we note that a concete mix with 370 kg
cement per m3 made with a dense limestone aggregate
like the Ras al Khaimah aggregate has been determined
to have a coefficient of thermal expansion of about
6.2 x10-6 per oC, whereas the same concrete made with
e.g. gabbro coarse aggregate has been determined to
have a coefficient of thermal expansion of 9.3 x 10-6 per
o
C. The temperature sensitivity of a concrete structure
is therefore reduced significantly by using limestone
instead of gabbro coarse aggregate.
2.3.9 Weathering from Abrasion (Mechanical Action)
of Airborne Dust and Sand
Weathering of concrete surfaces from abrasion with
airborne dust and sand is normally only observed on
inland structures in open landscape where the wind can
carry dust and sand unhindered to the lower surfaces
of concrete structures. Structures with high surface
strength are normally not significantly affected.
2.3.10 Non-Structural Cracks
Non-structural cracks in concrete comprise:
l Plastic settlement cracks
.. caused by badly compacted concrete,
badly designed concrete composition or
a giving form work. Concrete settles
downwards around rebars, thus breaks its
back over the rebars (creating cracks in
cover); also creates water bellies under the
rebars which causes loss of bond when this
water dries up !
l Plastic shrinkage cracks
. caused by lack of evaporation-
Journal of the Indian Roads Congress, July-September 2013
Raina on
protection during the first few hours after
placing the concrete. If rate of surface
evaporation is faster than rate of bleeding,
the concrete surface tends to dry up and
shrink and hence cracks at surface and the
crack traverses downward all the way until
its propagation stops due to initial setting
of concrete.
l Thermal cracks from heat of hydration
.caused by large differential temperatures
between newly and previously cast
concrete or between different locations
in newly cast concrete.
l Drying shrinkage cracks
. caused by drying out of the concrete
matrix when effective curing is stopped
too early. Drying out does not take place
in fully submerged structures. A structure
in contact with air will gradually obtain
a moisture content in equilibrium with
the airs relative humidity.
2.3.11 Structural Cracks
The most important structural cracks are:
Cracks
l  caused by restrained thermal
movement
Such cracks should be hindered by (1)
providing a sufficient quantity of small
diameter reinforcement close to the
concrete surface and (2) by dividing the
structure into units which can contract and
expand relatively freely.
l Cracks caused by structural tensile
stresses
Reinforced concrete structures are designed
to crack in tensile zones. The design
should, however, ensure that the spacing of
the cracks and the associated crack widths
are both sufficiently small (crack width
control).
 A Conceptual Strategy for Durability of Concrete Structures in the Arabian Gulf and similar
surrounds afflicted by Humidity, high Temperatures, and Chlorides and Sulphates, etc. 225

Cracks
l  caused by mechanical damage vessels passing through the bridges.
and impact
2.4 Exposure Classes
Minimise the risk and the consequences of The structures are divided into following different
impact from vehicles on the bridges and exposure classes:

Exposure Class Structures Exposure Condtiotons Deterioration mechanisms

SUBM-A (buried, Foundations and other Water temperature Chloride induced steel
submerged) structures buried in the 17-37oC reinforcement corrosion
ground, Up to 20-30 m water pressure by sea- Expansive sulphate attack
Submerged, marine structures water/groundwater Abrasion
Cl and SO4 in sea water/soil Mollusc attack
Waterborne sand
Molluscs
SUBM-B Portions of Structures in Air temperature in shade: 0-47oC Chloride induced steel
(in tidal zone) tidal zone, up to Design High Max. temperature on irradiated surface reinforcement corrosion
Water DHW-level exposed to direct sun: 80oC Salt weathering
Alternating wetting and drying with Carbonation induced steel
seawater reinforcement corrosion
Cl and SO4 in seawater Abrasion
Atmospheric O2 and CO2 Mollusc attack
Waterborne sand
Molluscs
SPLA Portions of Structures in Air temperature in shade: 0-47oC Chloride induced steel
(in splash and spray splash and spray zone Max. temperature on irradiated surface reinforcement corrosion
zone) exposed to direct sun: 80oC Salt weathering
Occasional wetting with seawater spray Carbonation induced steel
Cl and SO4 in seawater reinforcement corrosion
Atmospheric O2 and CO2 Abrasion
Waterborne sand
SUP (above splash and Portions of structures above Air temperature in shade: 0-47oC Chloride induced steel
spray zone) splash and spray zone Max. temperature on irradiated surface reinforcement corrosion
exposed to direct sun: 80oC Salt weathering
Airborne Cl and SO4 from seawater Carbonation induced steel
Atmospheric O2 and CO2 reinforcement corrosion
Waterborne sand Abrasion

3 ELEMENTS OF MULTISTAGE l Concrete execution including curing

PROTECTION STRATEGIES l Reinforcement
3.1 General l Coating
 of concrte surfaces up to certain
A multistage protection strategy for a severely level
exposed concrete structure of a long service life has to Experience
 with performance of similar structures in
consider the following with great care: similar exposure conditions should be utilised as much
l Design as possible.
3.2 Design
l Concrete materials and composition
For all structures it would be a general requirement
l Concrete Production that good run-off is facilitated, accumulation of dust

Journal of the Indian Roads Congress, July-September 2013

226 Raina on
and water is prevented, and steps taken to slow down
sulphate, carbonation and chloride ingress. This will
require proper geometric shapes such as rounded
corners, sloping self draining surfaces, etc.
The structures shall be designed with adequate
consideration to minimising the formation of cracks
from temperature effects.
The risk of formation of cracks from tensile stresses
exceeding the acceptable tensile strength of the concrete
shall be minimised by adequate amount and distribution
of the reinforcement close to the concrete surface,
as well as avoidance of large member thicknesses,
abrupt changes of cross sections leading to large stress
gradients, etc.
3.3 Concrete Materials and Composition
3.3.1 Concrete Materials
The concrete shall be made of materials and with a mix-
design which ensure a potentially dense and durable
concrete without built-in weaknesses (e.g. high built-in
chloride and sulphate content) and an efficient barrier
between the surface and the reinforcement.
l Cement and pozzolanas
Several types of cement and combinations
of cements and pozzolanas have been
used in the Arabian Gulf during the past
30 years or so.
Concrete with CEM I (Type I ASTM
Cement) with a low C3A- content unless
made very dense, should only be used for
un-reinforced structures.
CEM I with a moderate C3A- content,
alone or in combination with silica fume
or fly ash, may be used for dense concrete
in marine structures.
Another option is e.g. CEM III- B,
blastfurnace-slag cement.
Evaluation of the adequacy of a binder
or binder combination should consider
some important parameters e.g. rate of
and total heat development, risk of non-
Journal of the Indian Roads Congress, July-September 2013
structural crack formation, resistance
against chloride penetration, thresh-hold
value for steel corrosion, etc.
l Coarse aggregate
High quality coarse aggregates are
available as limestone aggregates as well
as gabbro aggregates. The limestone
aggregate has sufficient strength, and
wear on equipment in contact with
the aggregate. Thermal movement is
lower for limestone aggregate than for
the high strength gabbro. Gabbro has a
specific gravity 5 -10 % higher than that
of the limestone aggregate. Both types
have a low absorption of 0.3 0.6 %.
Gabbro is safer against molluscs. Unit
weight of Gabbro-made concrete is
slightly higher.
l Fine aggregate
Fine aggregates are usually reclaimed
from the sea or from inland deposits and
require very thorough washing.
In both cases, the grading of a source tends
to be single sized, thus requiring blending
of two or more sources to provide a
suitable grading.
Usage of marine sand requires careful
washing and draining to remove most
of the chlorides. Many of the marine
sand sources consist to a large extent of
weak limestone or broken sea shells. It is
generally difficult to wash out the chlorides
from the calcium carbonate sands because
some of the chlorides are apparently
trapped inside the particles, requiring a
long diffusion time for removal.
Fines from the coarse aggregates can be
used to blend with natural fine aggregate
to obtain a better sand grading.
l Water
Water for all usage in and in contact with
concrete shall be of high quality with low
 A Conceptual Strategy for Durability of Concrete Structures in the Arabian Gulf and similar
surrounds afflicted by Humidity, high Temperatures, and Chlorides and Sulphates, etc.
sulphate and chloride contents (not just
potable water because humans can drink
even slightly salty water !).
l Admixtures
Water reducing admixtures and high
range water reducing admixtures are
normally required to make workable
concrete mixes with low water/cement
ratios.
A retarder may be required to ensure
sufficient time for placing, compacting and
finishing the concrete as well as for being
able to compact newly and previously cast
fresh concrete together without creating a
cold joint.
l Corrosion inhibitors
Corrosion inhibitorsare available for
mixing into the concrete with the aim
of passivating the chloride ions which
penentate the concrete cover on the way
to the rebars. It is critical that the inhibitor
is still active after many years (e.g. 50
years), when the need for it arises. There
is at present some uncertainty about
the long term performance of corrosion
inhibitors.
3.3.2 Concrete Composition (w/c Ratio, Grading,
Admixtures)
A critical parameter for a low permeability concrete
mix is a low water/cement ratio and/or low water/
binder ratio where additions such as pozzolanas are
used. Capillary continuity (non-denseness) is normally
unavoidable even in well cured concrete if the water/
cement ratio is above about 0.50.
For high performance concrete a maximum water/
cement ratio between 0.42 and 0.35 is normally
specified. The lower the water/cement ratio the lower
will be the permeability if the concrete can be placed,
compacted and finished well!
The structural design and reinforcement should allow
for usage of upto 40 mm maximum coarse aggregate
Journal of the Indian Roads Congress, July-September 2013
227
size in most of the structural elements if possible in
order to increase the coarse aggregate content (and low
shinkage, etc.) and reduce the cement content.
A high bulk density of the combined coarse aggregate
should be obtained by suitable combination of the
coarse aggregate fractions. Similarly should the high
bulk density of the combined aggregate be obtained by
suitable combination of coarse and fine aggregates.
Due consideration should be given to obtain a cohesive
mix with a suitable workability with a high aggregate
content (Gradation).
Admixtures and Admixture dosages should be so
selected as to give a robust fresh concrete with low
tendency of segregation and bleeding, which has a
reliable workability loss, slow and not abrupt, giving
sufficient time for placing, compaction and casting
together with previously placed fresh concrete.
3.4 Concrete Production
All storage and manufacturing facilities for concrete
constituents as welll as batching and mixing plants
shall be of such design, types and quality as to ensure
continuous production of very uniform concrete
mixes.
Hot weather concreting precautions shall be taken
to meet the fresh concrete temperature requirements.
Cooling with chilled water, flaky ice or liquid nitrogen
(in water) are methods of reducing the fresh concrete
temperature.
During the hot weather season, scheduled secondary
dosage of water reducing admixture and/or high
range water reducing admixture to be added on
site immediately before discharge may be required to
maintain a workable concrete mix for a longer time.
3.5 Concrete Execution, Including Curing
The main aim during execution of concrete for a
concrete structure is to ensure development of the
required potential properties of the concrete.
Cracking should be kept to a minimum or avoided
(using partial prestressing if required).
Blow holes and other surface defects should be
minimised by e.g. using self compacting concrete
228 Raina on
or controlled permeability formliner. Application
of controlled permeability formliner will result in a
densification of the outer few mm of the cover which
will be significantly less permeable.
The concrete cover shall be sufficient to prevent
chloride concentrations exceeding the thresh-hold value
at the reinforcement level within the specified service
life. The minimum cover should be at least 70 mm, but
should not be more than 80-90 mm (risk of cracking of
a thick un-reinforced cover layer). The cover could be
reduced if stainless steel or non-metallic reinforcement
is used.
Another element in providing an impermeable concrete
cover is to minimise or eliminate the risk of formation
of thermal cracks. This can be done by application
of binder combinations with a low heat development
combined with conformity with temperature difference
limits obtained from experience.
3.6 Reinforcement
One of the main effects of the requirement of a dense,
impermeable concrete and its cover is to protect steel
reinforcement from corrosion and inhibit ingress and
movement of pollutants through any capillaries in the
concrete matrix.
If the concrete cover protection is not considered
sufficient, the steel reinforcement could be cathodically
protected right from the beginning or the structure
prepared for cathodic protection by ensuring rebar
continuity and as yet another option to build-in the
anodes for future-use when required. (But large scale
cathodic protection in the field is neither easy nor
dependable.)
An element in a multistage protection strategy could
be to use reinforcement which does not corrode:
Epoxy coated steel has been used for many
l 
severely exposed concrete structures over
the past more than 40 years with varying
degrees of success. In some cases (e.g.
many road bridges in U.S.A.) corrosion
has started long before expected. As a
consequence several states in U.S.A.
have prohibited use of epoxy coated
steel.
Journal of the Indian Roads Congress, July-September 2013
The
l  main cause for premature corrosion of
epoxy coated steel is practical defects in the
coating, either pin holes or holes formed
during handling and construction, and/
or defective bond to the steel. In both cases
chloride ions can get access to small areas
of the steel causing severe pit corrosion
and a quick, dramatic reduction of the
rebar cross section.
Usage
l  of epoxy coated steel is not
recommended for concrete structures
in the severely aggressive environment
under discussion here.
l Stainless steel
Stainless steel is available in grades that
will not corrode in a chloride contaminated
environment. Usage of stainless steel
reinforcement may eliminate any durability
related measures needed against chloride
ingress and steel corrosion.
The long term performance of stainless
steel in concrete structures is well
documented with examples of more
than 60 years excellent performance.
Stainless steel reinforcement can be used
as replacement for all reinforcement or
in selected, specially severely exposed
structural elements or parts of structural
elements; stainless steel can be used in
combination with ordinary black steel.
As an example, space restrictions may call
for a smaller concrete cover in certain parts
of the concrete girders or the edge beams.
Such a reduction could be accepted if the
rebars in such locations are of stainless
steel.
Non-metallic
l  reinforcement: bars,
sheets, fibres
Many types of non-metallic materials are
available as alternative or supplementary
reinforcement for concrete structures.
Some types e.g. fibres used as
supplementary reinforcement, result in
 A Conceptual Strategy for Durability of Concrete Structures in the Arabian Gulf and similar
surrounds afflicted by Humidity, high Temperatures, and Chlorides and Sulphates, etc.
more evenly distributed stresses in the
structure with a reduced crack formation.
In addition to relatively high prices,
unfortunately few of these materials have
a long track record.
3.7 Coating the Concrete Surfaces up to a
Certain Level
Coatings may be applied on concrete surfaces as an
element (one barrier) in a multistage protection strategy
to provide protection against:
l Water absorption
Ingress
l  of chlorides, sulphates and other
salts in seawater
l Ingress of carbon dioxide and oxygen
All surface protection systems should, as a minimum,
be water repellent and prevent ingress of chlorides,
sulphates and other salts. Impermeable systems will
also protect against ingress of carbon dioxide and
oxygen, whereas water vapour diffusion open systems
may allow penetration of these gases.
Each systems ability to bridge across cracks in the
concrete, its bond strength, flexibility, mechanical
resistance and UV resistance should be assessed.
Impermeable epoxy systems can be used for high
performance concretes and these systems dry out due
to self desiccation during the hydration.
When applied correctly, impermeable systems, e.g.
epoxy coatings, have good long-term records from
marine structures in the Arabian Gulf.
4 SELECTED DURABILITY STRATEGY
4.1 General
In the preparation of the durability strategy for the
structures under reference here, it is logical to search
for and study major structures subjected to similar
exposure conditons. The obvious example is the
King Fahd Causeway i.e. Sea-link (KFC) which was
constructed in the first half of the 1980s.
The strict specification for the concrete structures of
KFC has resulted in an excellent performance with one
minor exception:
Scaling
l  to a depth of 7-8 mm on the
vertical faces on the north side of the
Journal of the Indian Roads Congress, July-September 2013
229
southern girders. This scaling is generally
only seen on the low bridges, but extends
to a height of about (levels) +6 to +7 m
in some locations on the high bridges. The
total extent is estimated to be about 1 % of
the total concrete surface area.
A test pile constructed similarly and placed close to the
causeway prior to construction of the causeway has been
inspected and tested regularly. The pile is not epoxy
coated. From measurements of chloride penetration in
the test pile in 1994 and 1999 it has been calculated that
the time to initiation of corrosion of the reinforcement
is about 600 years.
Although this figure connot be accurate, it no doubt
gives an indication of the magnitude of slow timeframe
with which this important deterioration mechanism
is progressing. The bridge structures were all epoxy
coated in the precast yard up to +4 m level, and the
chloride penetration behind this coating is nil. During
construction the epoxy coating was extended in-situ up
to +8 m level on all piles. The girders are not epoxy
coated.
KFC is the oldest major bridge structure in this part
of the Arabian Gulf and, with its excellent durability
performance, it constitutes an important example.
Based on the current condition of KFC concrete
structures and the estimated remaining service life from
chloride penetration tests, it is obvious that KFC should
form a significant part of the basis for the durability
strategy for the subject concrete structures.
Despite the fact that there has been significant
development within concrete construction of durable
strucures in aggressive exposure conditions over the
last 30 years, only changes which represent significant
and very well long-term documented improvements
should be considered as deviations to the materials,
methods and procedures applied for KFC.
4.2 Overall Durability Strategy
The most important deterioration mechanism for
the subject concrete structures is chloride induced
steel reinforcement corrosion.
When normal steel is used, the main worry is to
prevent chlorides from penetrating the concrete
230 Raina on
cover to the location of the steel rebars, as also
the corrosion of steel from chlorides within the
concrete matrix.
The selected strategy is therefore summarised as
follows:
l Provide
 a dense, impermeable concrete
and concrete cover of minimum 70 mm
thickness to steel rebars.
l Alternatively,
 use stainless steel locally
(in areas with the most severe exposure,
such as the splash zone, or areas where
the minimum 70 mm cover can not be
obtained).
The general use of stainless steel could
also be considered as the most effective
and safe method of securing a long service
life. The concrete cover can be reduced to
a minimum of 35 mm where stainless steel
is used.
l Apply
 measures for prevention/reduction
of crack formation in fresh, hardening and
hardened concrete.
l Use a low/slow heat cement; optimise mix

design through trial mixes and full scale
castings.
l Ensure
 compliance with temperature
requirements during hardening.
l Apply
 of Controlled Permeability
Formliner (CPF) on all formed surfaces
which gives a highly densified surface
layer of concrete (to 35 mm).
l Apply
 epoxy surface coating on all
concrete surfaces upto 2 m above splash
zone.
l Use
 gabbro aggregate in all foundations
and substructures to a level of 2 m above
HTL to prevent mollusc (marine borer)
attack.
The selected strategy has multistage barriers
against chloride ingress and carbonation for the
substructure from:
Journal of the Indian Roads Congress, July-September 2013
Epoxy coating of concrete surfaces
l  upto
a certain level.
Densified
l  surface layer, 3-5 mm, by
application of CPF.
Dense
l  and low permeability concrete
and concrete cover with minimum
cracks.
The selected strategy has multistage barriers
against chloride ingress and carbonation for the
superstructure from:
l Densified
 surface layer, 3-5 mm, by
application of CPF.
l Dense
 and low permeablility concrete
and concrete cover with minimum
cracks.
Protection against subsequent steel reinforcement
corrosion should chlorides or carbonation eventually
reach the rebars could comprise use of stainless steel
rebars or establishment of rebar continuity to allow
for a future cathodic protection sysem (although the
efficacy of the latter is more dependable in Laboratory
work than in a large project in the field!).
The protection against mollusc attack is provided by
using gabbro coarse aggregate which is considered a
sufficient barrier against these marine borers (_ they can
eat into limestone, not into gabbro).
4.3 Design
For all structures it would be a general requirement
that
l good drainage run-off is facilitated,
l accumulation of dust and water prevented,
and
steps
l  are taken to slow down carbonation,
sulphate and chloride ingress.
This will require proper:
- geometric shapes such as rounded corners,
sloping surfaces, and
- highly dense concrete of low w/c ratio
(High Performance Concrete).
 A Conceptual Strategy for Durability of Concrete Structures in the Arabian Gulf and similar
surrounds afflicted by Humidity, high Temperatures, and Chlorides and Sulphates, etc. 231

The structures shall be designed with adequate l Fine aggregate

consideration to mimimise the formation of cracks The fine aggregate will probably have to be
from temperature and other effects. formed out of two or more raw sands from
inland and marine sources, thoroughly
The risk of formation of cracks from tensile strain washed.
exceeding the tensile strength of the concrete shall
l Water
be minimised by:
Water for all usage in and in contact with
adequate
l  amount and distribution of the concrete shall be of high quality and with
small diameter reinforcement close to the low chloride and sulphate contents (not
concrete surface, and just the potable water).
by
l  avoidance of large thicknesses and l Admixtures
abrupt changes of cross sections that lead Water reducing and high range water
to steep stress gradients. reducing admixtures are expected to
be used for very low w/c concrete with
4.4 Concrete Materials and Composition
a normal workability and also if a self
4.4.1 Concrete Materials compacting concrete is sought. (There is
insufficient experience available with
Cement and Pozzolanas
the latter.)
- 
Blastfurnace slag cement with a slag No other admixtures are expected to be
content of 66 80% by weigh of cement, required, except for a retarder to allow
CEM-III/B 52.5 SR, shall be used for all enough time for concrete placement and
reinforced and prestreseed concrete. (This compaction .
cement type, but a lower strength class, 4.4.2 Concrete Composition
was used for KFC.)
Following seven classes of concrete for reinforced and
- No pozzolanas will be required. pre-stressed concrete and for blinding can be adopted
- Silica fume shall not be allowed, except (ref. section 2.3 earlier):
for underwater concrete where it may
Exposure Concrete Maximum Minimum Coarse Maximum Minimum
improve the cohesiveness of the concrete class class water/ cement aggregate size of characteristic
to prevent washing out of fine material. cement content type coarse compressive
ratio Kg/m3 aggregate cube
(by wt.) mm strength
l Coarse aggregate (28 days)
Mpa
Gabbro shall be used for all substructures
A40 380 40
to prevent attack from molluscs (marine SUBM-A Non
A20 400 20
borers). SUBM-B
S40
0.40
400
calcareous
40
60
(Gabro)
Limestone aggregate shall be used for all S20 420 20

superstructure in order to obtain lower B40 380 Calcareous 40

SPLA B20 400 (Lime- 20
thermal movements. SUP stone)
C - - - 20 or 10 12
The lower specific gravity of the limestone (Blinding
aggregate will reduce the concrete density layer)

by 3-5%. The wear on all equipment in

The above seven concrete classes for the suggested
contact with concrete with this coarse
exposure classes for the indicated structure-portions,
aggregate is significantly reduced.

Journal of the Indian Roads Congress, July-September 2013

 232
shall be as follows:
l Class A40 concrete
shall be used for all structures in exposure
classes SUBM-A and SUBM-B, except
for concrete deposited underwater and
when the usage of 38 mm nominal coarse
aggregate size is not permitted
l Class A20 concrete
shall be used for other structures in
exposure classes SUBM-A and SUBM-B,
except for concrete deposited underwater.
l Class S40 concrete
shall be used for concrete deposited
underwater, except when the usage of 38
mm nominal coarse aggregate size is not
permitted.
l Class S20 concrete
shall be used for all other concrete
deposited underwater.
l Class B40 concrete
shall be used for all structures in exposure
classes SPL and SUP, except when the
usage of 38 mm nominal coarse aggregate
size is not permitted.
l Class B20 concrete
shall be used for all other structures in
exposure classes SPL and SUP.
l Class C concrete
shall be used for blinding for all structures.
Coarse aggregate size shall be selected in
accordance with blinding thickness.
~ Other concrete composition
requirements:
- Concrete workability shall be
determined by the contractor taking into
consideration complications in placing
and compacting the concrete in different
Journal of the Indian Roads Congress, July-September 2013
Raina on
structural elements. Self compacting
concrete may be a useful alternative for
certain structural elements, but there is
insufficient experience available in this
as yet.
- The total acid-soluble chloride content
in the concrete mix, from all sources,
shall not exceed 0.10% by weight of
cement for all reinforced and prestressed
concrete. This limit is difficult to meet,
but not un-realistic.
- The total acid-soluble sulphate
content as SO3 in the concrete mix,
from all sources, shall not exceed
4. 0 % by weight of cement.
- To eliminate any risk of alkali-silica
or alkali-carbonate reactions, the
maximum content of equivalent Na2O
shall be 3.0 kg/m3 concrete with a
mortar content of 60%.
- The concrete temperature shall not
exceed 32OC at the point and time of
placing.
- A high bulk density of the combined
coarse aggregate shall be obtained
by suitable combination of the coarse
aggregate fractions. Similarly, shall
the high bulk density of the combined
aggregate be obtained by suitable
combination of coarse and fine
aggregates. Due consideration shall
be given to obtain a cohesive mix with
a suitable workability using a high
aggregate content and low w/c ratio.
- Admixtures and Admixture-dosages
shall be so selected as to give a robust
fresh concrete with low tendency of
segregation and bleeding, which has a
reliable workability loss - slow and not
abrupt, giving sufficient time for placing,
compaction and casting together with
previously placed fresh concrete.
 A Conceptual Strategy for Durability of Concrete Structures in the Arabian Gulf and similar
surrounds afflicted by Humidity, high Temperatures, and Chlorides and Sulphates, etc. 233

4.5 Concrete Production differential temperatures during casting

and curing;
All elements of multistage protection strategies,
mentioned in section 3.3 and repeated below, shall - restrictions would apply to ensure
be considered by the Contractor and implemented as careful handling during transport and
required. erections of elements;
- strict QA procedures would be
l All storage and manufacturing facilities

required.
for concrete constituents as well as
batching and mixing plants shall be 4.6.1 Concrete Cover
of such design, types and quality as to The concrete cover to any steel also including stirrups
ensure continuous production of very and links shall comply with the following:
uniform quality concrete mixes.
l Hot
 Weather Concreting Precautions Minimum Minimum Nominal
cover thickness of cover
shall be taken to meet the fresh concrete
to any steel spacer mm
temperature requirements. Cooling with mm mm
chilled water, flaky ice or liquid nitrogen
Ordinary,
(in water) are methods of reducing the black steel 70 75-80 75-85
fresh concrete temperature.
l During
 the hot season, scheduled Stainless
secondary dosage of water reducing steel 35 40-45 40-50
admixture and/ or high range water
reducing admixture, to be added on Spacers shall be of cement mortar of similar quality as
site immediately before discharge, may the mortar of the concrete. Stainless steel spacer may
be required to maintain a workable be used if the contractor can demonstrate complete
concrete mix for a longer time. bond with the surrounding concrete.
4.6 Workmanship and Construction Methods
4.7 Reinforcement
The specification shall define limits for methods and
workmanship, with due concern for the variation Reinforcement shall be ordinary steel (HYDS) or
in exposure classes. This would include basic stainless steel.
requirements, including:
Ordinary and Stainless steel may not be mixed in the
- the use of slip-forming would not be same structural element and need not be electrically
prefered; isolated.

- Controlled Permeability Formliner
(CPF) shall be used on all formed
surfaces in order to minimise blow holes
and other surface defects and to further
densify the outermost 3-5 mm of the
concrete cover;
Epoxy coated steel reinforcment shall not be used.
Application of non-metallic reinforcement, whether
bars, sheets or fibres would not be allowed for a
concrete structure in this aggressive environment
with a 100 years service life requirement.

- the curing requirements would be strict 4.8 Coating of Concrte Surfaces upto a Certain
and comprehensive, steam curing would Level
not be allowed; The 3-layer (primer is in 2 coats) solvent free, two
- restrictions would apply to total and component, epoxy coating on King Fahad Causeway

Journal of the Indian Roads Congress, July-September 2013

 234 Raina on
(Sea-link) has proven to perform excellently in
preventing chloride ingress and already has been
durable for about 30 years so far.
The spray zone on KFC extends up to 6 to 7 m above
MSL. In order to ensure that the coating protects all
surfaces in risk of being subjected to salt spray, it is
recommended that all reinforced structures up to a
certain level be epoxy coated. No need to epoxy-coat
the supertructructure.
4.9 Quality System
To achieve the set Durability target in the structures in
the harsh environment, it is very important that the
selected contractor establishes and strictly follows a
quality assurance system.
It is assumed that the selected contractor will operate
such quality system and will be ISO 9000 certified.
Additionally, all aspects of the project, right from
the preliminary and detailed design, through method
statements, work procedures, inspection-and-test-plans,
etc., will be covered by and implemented under strict
compliance of written quality assurance and quality
control plans.
4.10 Testing of Finished Concrete Structure
Critical properties of the finished concrete (in the
structure) shall be documented as part of the required
quality documentation prior to contractual handover.
Such documentation shall include measurement of:
l Concrete cover
l Chloride diffusion
and may also include:
l Accelerated carbonation depth
l Electric resistivity
Capillary
l  absorption or ISAT (initial
surface absorption test).
These sets of test results will provide the base-line
quality of the truly achieved performance-related
qualities in the finished structure from which, later,
residual- service-life updating can be made.
Journal of the Indian Roads Congress, July-September 2013
5 OPERATION AND MAINTENANCE
REQUIREMENTS TO BE CONSIDERED
IN THE DESIGN
5.1 General
a) The structure shall be designed and built
with due regard for future inspection and
maintenance, based on the strategies and
decisions adopted in design, specifications and
construction methods.
b) Advanced structural monitoring systems for
early warning of risk of chloride induced steel
corrosion shall be incorporated at strategic
locations as automatically monitored corrosion
and reference cells.
c) An inspection and maintenance programme
for the life-time of the structures should be
established at the design stage itself as an
integrated part of the strategy for achieving the
required durability.
Also organisational aspects of the future
operation and maintenance should be taken
into account at an early stage, ensuring that
all parts of all causeway structures will be
maintained to the same standard, following the
same pre-planned procedures and providing
consistent records in the same formats.
d) The planning should facilitate easy access
and easy repair/replacement of special items
like Expansion Joints, Bearings, wearing
courses and surface protection, crash barriers,
mechanical and electrical installations, and
other items where the recurring or potential
needs for maintenance, repairs and replacement
can be foreseen.
5.2 Planning for Access and Inspection
Platforms
The planning for access will comprise the design of the
structure (working space, stairs, inspection openings)
and also the provision of the necessary permanent
 A Conceptual Strategy for Durability of Concrete Structures in the Arabian Gulf and similar
surrounds afflicted by Humidity, high Temperatures, and Chlorides and Sulphates, etc. 235

inspection platforms (gantries), which must be tailor- REFERENCES

made to suit the inspection and maintenance of piers
and girders in the various bridge structures (viaduct
and high bridge portions).
The inspection gantry should be easy to transport and
install at any location along the structure, with the
ability to let the inspection crew safely reach all parts
of the piers and girders.
Permanent rails under the bridge superstructure
should be provided for easy movement of the
gantry, and the gantry should be able to pass the
piers when travellling on such rails along one
bridge in its entire length.
The gantry should, for safety reasons, be operated
directly by a properly trained and fully self-reliant crew
placed on the travelling gantry platform itself, not by
persons on the bridge deck above.
During periods between inspections, the gantry should
be stored and safely maintained in the maintenance
yard.
1. AASHTO LRFD Bridge Design Specifications
S.I Units, 2nd Edition, 2001, Interim
Revisions, May 2001.
2. AASHTO LRFD Bridge Construction
Specifications, First Edition 1998.
3. AASHTO for Concrete Durability Design.
4. Guide to the construction of reinforced concrete
in the Arabian Peninsula, CIRIA and The
Concrete Society, 2002.
5. Adam Neville: Properties of Concrete, Fourth
Edition, Addison Wesley Longman Limited,
England 1995.
6. Durable Concrete Structures, CEB Design
Guide, Second Edition, Bulletin No. 182,
Lausanne, 1989.
7. V.K. Raina: Concrete for Construction,
Facts and Practice, Shroff Publishers and
Distributors, Mumbai.

The views expressed in the paper are personal views of the Author. For any query, the author may be contacted at: E-mail: rainavk@gmail.com

Journal of the Indian Roads Congress, July-September 2013

Copyright
Regd. No. 17549/57 with the Registrar of Newspapers

ISSN 0258-0500