Drag reduction induced by exible and rigid molecules in a turbulent

ow into a rotating cylindrical double gap device: Comparison between

Poly (ethylene oxide), Polyacrylamide, and Xanthan Gum
Anselmo S. Pereira, Rafhael M. Andrade, Edson J. Soares

LABREO, Department of Mechanical Engineering, Universidade Federal do Espirito Santo, Avenida Fernando Ferrari, 514, Goiabeiras, 29075-910 ES, Brazil
a r t i c l e i n f o
Article history:
Received 2 May 2013
Received in revised form25 September 2013
Accepted 28 September 2013
Available online 8 October 2013
Keywords:
Drag reduction
Polymer degradation
Development time
Flexible and rigid molecules
a b s t r a c t
Polymer-induced drag reducing ow has been investigated for over 60 years. One reason for this is that
the drag reducers in ow systems have been successfully applied and represent a great potential benet
to many industrial processes. However, the phenomenon is not completely understood and many aspects
of the problem remain unclear. Some important issues are related to the development of turbulent struc-
tures and to the breaking of the polymer molecules. These two phenomena impose a transient behavior
on the polymer efciency and the drag reduction, DR, can be clearly divided into three periods of time.
Over time, at the very beginning of the test, DR assumes a minimum value (sometimes negative) before
reaching its maximum efciency. When degradation becomes important, DR starts to decrease until it
achieves its asymptotic value, a time in which the polymer scission stops and the molecular weight dis-
tribution reaches a steady state. In the present paper, we study the drag reduction development from the
very beginning of a turbulent ow into a rotating cylindrical double gap device. DR is induced by three
different polymers: Poly (ethylene oxide) (PEO), Polyacrylamide (PAM) and Xanthan Gum (XG). The rst
two are known as exible molecules while the last one is considered rigid. The goal here is to compare the
effect of the different polymers on DR over time, paying particular attention to the difference between the
rigid and the exible molecules. The tests are conducted for a range of Reynolds numbers, concentrations
and temperatures, from the very start to the time when the drag reduction achieves its nal level of ef-
ciency. The time to achieve the maximum efciency is an increasing function of concentration and
decreases with Reynolds and temperature in PEO solutions. Such time seems to be very short for the
other polymers, less than 3 s. It is worth noting that no loss of DR was observed for high concentrations
of PAM, which suggests that PAM is more resistant than PEO. It is also shown that DR induced by XG is
qualitatively different from that of the other agents. XGs solution is highly inuenced by a pre-shearing,
which suggests the existence of polymer aggregates. In addition, it seems that degradation do not occurs
for solutions of XG. The observed loss of efciency in high concentrations is, possibly, caused by
de-aggregation during the test.
2013 Elsevier B.V. All rights reserved.
1. Introduction
Polymeric drag reducers have been successfully used in a num-
ber of applications for more than 60 years (see [17,6,21,42,20]).
Over the years, researchers have been successful in analysing this
phenomenon and many remarkable papers with practical interest
can be found (see [51,52,50,49,31]). Up to now, there has been
no generally accepted theory for the mechanism of drag reduction,
despite the fact that many researchers have contributed with some
very signicant papers (see [29,47,14,1]). White and Mungal [54] is
a good review of some recent progress in understanding the funda-
mentals of polymer drag reduction.
The drag reduction phenomenon is very dependent on the kind
of drag reducer used. Commonly, brous particles, surfactants and
polymers are used. These last are very far more efcient and have
been widely analyzed over the years. Concerning their mechanical
resistance, polymers can be divided into exible and rigid mole-
cules. The former are generally of hight molecular weight with a lin-
ear chain, among which Poly (ethylene oxide), Polyacrylamide and
Polyisobuthylene are examples which have been extensively tested
over the years (see [51,37,25,39,50,2,42,12,40,8,27,23,24,10,5]). A
huge disadvantage, which is a great obstacle to the practical use
of exible polymers is their mechanical degradation. A possibility
that deserves our attention is the use of polysaccharides such as
Hydroxypropylguar, Gar Gum and Xanthan Gum (see [26,12,7]).
0377-0257/$ - see front matter 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jnnfm.2013.09.008

Corresponding author. Tel.: +55 2740092162.

E-mail address: edson@ct.ufes.br (E.J. Soares).
Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287
Contents lists available at ScienceDirect
Journal of Non-Newtonian Fluid Mechanics
j our nal homepage: ht t p: / / www. el sevi er . com/ l ocat e/ j nnf m
Such molecules are rigid and much less susceptible to mechanical
scission. The main problem in this case is the biological degrada-
tion, though it is much less accelerated when compared to the
mechanical scission.
Among the different rigid polymers known in the literature,
Xanthan Gum seems to be of particular practical interest in food,
pharmaceutics, cosmetics, and the oil industry, in which it is
widely used as a drag reducer in drilling well operations. From
the molecular point of view, XG possesses a linear main chain of
(14)-b-D-glucose, similar to cellulose, with a trisaccharide side
chain on every second D-glucose (see [4] for details). In fact, such
a complex structure is responsible for its rigidity and stability. Its
structure is highly dependent on the temperature and salinity. At
moderate temperatures and low ionic forces, XG presents a stable
organized helical conformation resulting in a rigid molecular struc-
ture, as reported by Morris [30] and Norton et al. [36]. Such an or-
ganized structure can be modied by increasing temperature or
salinity. In such conditions, the ionic forces are altered and the
helical conguration changes to a coiled one. In this new congu-
ration, the Xanthan Gums capability to reduce drag drops dramat-
ically. In fact, temperature plays a complex role in the structural
conguration of XG. Below a certain value of T, known as the tran-
sition-midpoint temperature, an increase in T causes an increase in
the mean molecule length, keeping the helical structure stable as
suggested by Sohn et al. [45]. The salinity also plays a very impor-
tant role in the structural conguration of XG. For more details of
the mechanism and dynamics of XGs structure, see Morris [30],
Norton et al. [36] and Muller et al. [32].
As with exible polymers, the drag reduction obtained with the
use of Xanthan Gum is an increasing function of its concentration
and molecular weight (see [4,3,45]). Also similar to other polymers,
the level of DR achieves an asymptotic value with increasing c and
M
v
. However, the drag reduction mechanism for rigid polymers
seems to be considerably different. Such a difference can be easily
noticed observing the Fanning friction factor in Prandtlvon
Krmn coordinates, for example. After the onset, the curves for
XG are parallel for different concentrations, which suggests a weak
dependence on the Reynolds number, as reported by many authors
who investigated rigid polymers [49,4,3,45,22]. Such an effect is
known as retro-onset and was discussed by Virk [49]. Retro-onset
is generally associated with rigid molecules. According to Virk
et al. [53], drag reduction can be divided into two very distinct
mechanism: Type A and Type B. The former is associated with
polymers that stay coiled at rest. Such materials need a certain le-
vel of turbulence to stretch and start to reduce drag. In contrast,
the second mechanism is related to polymers that stay extended
at rest. Consequently, the onset of the drag reduction is expected
to occur early when the Type B mechanism is dominant. As sug-
gested by Gasljevic et al. [19], in the case of Type B mechanism
of drag reduction, polymer molecules may be fully stretched after
the retro-onset, and, consequently, a further increase in the level of
turbulence could not result in a substantial change of molecular
conformation. The mechanism of drag reduction for Xantham
Gum solutions is clearly of Type B.
A huge obstacle to attempts to obtain an accepted theory of the
phenomenon of drag reduction is the mechanical molecular degra-
dation. This issue involves a strong interdisciplinary connection
between chemistry and uid mechanics. This issue has received
deserved attention over the years and many aspects of the problem
have been studied, such as the effect of concentration, molecular
weight, Reynolds number and temperature on the efciency of
drag the reduction (see [37,33,57,46,31,40,38]). Using an experi-
mental turbulent pipe ow apparatus, Vanapalli et al. [48] per-
formed some careful analyses to show that DR decreases as a
consequence of polymer degradation but reaches a steady state
after a certain number of passes through the pipe ow apparatus.
In other words, the molecular scission stops after a long enough
time. This tendency is supported by many other results, such as
those reported by Nakken et al. [34], Choi et al. [10], Kalashnikov
[24] and Pereira and Soares [38].
The dependence of drag reduction on time is not exclusively re-
lated to molecular degradation. As reported by Dimitropoulos et al.
[13], the turbulent structures take some time to rearrange follow-
ing a polymer deformation and DR does not achieve its ultimate le-
vel instantaneously. In fact, DR is a complicated function of time.
Fig. 1 shows schematically the development of a polymer induced
near-wall drag reduction, dened as follows:
DR 1 f
p
=f
0
; 1
where f
p
is the friction factor of the polymeric solution and f
0
that of
the solvent. This kind of gure can be constructed by monitoring
the drag reduction along a pipe or channel after the polymer injec-
tion or by using any rotating apparatus. The last strategy is evi-
dently easier. As sketched in Fig. 1, the available results
concerning exible polymers suggest that at the very start of the
test, DR decreases from DR
0
to DR
min
before achieving its top level
of efciency at DR
max
. Since polymers extract energy from the vor-
tices and release energy to the mean ow in a coil-stretch cycle, we
presume that the maximumdrag reduction occurs when a sufcient
number of the molecules are in this coil-stretch cycle [14] and a
state of equilibrium with the turbulent structures has been
achieved. We will refer to the time to achieve DR
max
as the develop-
ing time, denoted t
d
. The increasing friction factor at the beginning
of the process is related to an instantaneous increment of the local
extensional viscosity after a high polymer stretching. Following t
d
,
we observe a constant value of DR for a period of time, which is de-
noted by t
r
, the resistance time. Finally, after this period, DR begins to
fall, reaching a minimum level after a long enough time, when the
degradation process has reached its steady state and DR assumes
an asymptotic value, DR
asy
. The time to reach DR
asy
, t
a
, is relatively
large compared with the stretching time of a single molecule, be-
cause the molecules are stretched and degraded step-by-step (see
[15]). Thus, we could presume that during t
r
the increasing number
of molecules in the coil-stretch cycle is balanced by the molecular
degradation, and the ultimate level of drag reduction is sustained.
Following that, with a continuous degradation, the turbulent struc-
tures depart from their equilibrium and start to increase until
achieving the nal steady state in which the level of drag reduction
assumes a constant value, DR
asy
. This theory seems to be quite
Fig. 1. Sketch of the evolution over time of the polymer-induced drag reduction.
A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287 73
reasonable for exible polymers but does not explain the behavior
of DR(t) for rigid polymers. The coil-stretch process, for example,
probably does not play an important role in this kind of material,
once the rigid molecules stay almost well-extended at rest.
There have been a number of papers treating DR as a function of
time. Recently, Pereira and Soares [38] showed a great number of
data in an attempt to understand the effect of the temperature,
Reynolds number, concentration, and molecular weight on t
d
, t
r
,
and t
a
, but most of the data is related to PEO solutions. In the pres-
ent paper, the experiments were conducted in an attempt to pro-
vide a direct comparison between PEO, PAM, and XG. The last
polymer is a rigid molecule and the results related to it is particu-
larly interesting and it is the main goal of this research. In our tests,
we are not showing DR decreasing over the time. Such measure-
ments are really difcult to be accomplished, since the increasing
drag occurs during the rst seconds of the test and an especial
attention to the control of the elapsed time in order to achieve a
constant rotor speed of rotation is needed.
2. Experimental apparatus and procedure
The majority of the experimental results on drag reduction by
polymer additives available in the literature have been obtained
for pipe ow systems, obviously, because they are widely used in
many industrial transport processes. However, the use of pipe sys-
tems to analyze degradation is extremely difcult and time con-
suming. A way to overcome this difculty is by using a rotational
apparatus, such as coaxial cylinders [23,24], a rotating disk
[39,9,10,28], and a double-gap cylindrical geometry [34,5]. This
last geometry has a large contact area, which provides measure-
ments with a quite good accuracy, even for small values of the Rey-
nolds number. As in our previous work [38], we use here this kind
of apparatus, shown in Fig. 2, to obtain our results.
The tests were carried out using a commercial rheometer, mod-
el HAAKE MARS II, manufactured by Thermo Scientic, Germany.
The sample was located between the two rigidly interconnected
coaxial and stationary surfaces, which have an axial symmetry.
The rotor is a thin-walled coaxial tube located between these
two xed cylindrical surfaces which can rotate over the sample
holders axis of symmetry at a given angular velocity.
For a given angular velocity, the mean shear rate is determined
as a function of the rotor speed of rotation and the nominal shear
stress as a function of the applied torque. The rotor reaches its nal
angular velocity in less than 3 s. Only after such period of time we
calculate the Fanning friction factor based on the characteristic ra-
dius, which is given by the mean radius R
R
2
R
3
2
:
f
2s
qu
2

2s
q xR

2
: 2
The Reynolds number is dened by Eq. (3):
Re
q

hu
g

q

h

xR

g
; 3
where g is the solutions viscosity, xR is a characteristic velocity
and

h is the average gap given by ((R
2
R
1
) + (R
4
R
3
))/2. Since
our solutions of PEO and PAMare Newtonian, g is directly measured
at low rotor speed of rotation, in which the ow is laminar. For the
Xanthan Gum solutions, which are shear-thinning even for low con-
centrations, g is the Newtonian plateau at high shear-rate (g
1
tted
by Eq. (5)). It is worth noticing that our tests were carried out at
xed Reynolds number. Hence, the values of g used to determine
Re are that for the same temperature of the test. After that, the ro-
tors speed is adjusted for xing the desirable Reynolds number.
In order to distinguish the distinct ows in the double-gap
geometry, for a range of analyzed Reynolds numbers, we used
the Taylor number given by Eq. (4)
Ta
R

h
3
x
2
m
2
; 4
where m is the cinematic viscosity.
We tested solutions of Poly (ethylene oxide), Polyacrylamide,
and Xanthan Gun. The molecular weight of the rst and second
materials is M
v
= 5.0 10
6
g/mol, whereas that of the last one is
M
v
= 2.0 10
6
g/mol. All our chemical supplies were provided by
SigmaAldrich. We obtained the molecular weight by calculating
the intrinsic viscosity using the Huggins equation (for details see
Flory [18]) and our measurements were very close to the values
quoted by SigmaAldrich. The measured intrinsic viscosity, [g],
was also used to estimate the overlap concentration for PEO and
PAM by means of the relation c

[g] = 1. For the PEO solutions, the

calculated value was c

= 3125. For the PAM, the overlap concen-

tration was around c

= 100 ppm. This procedure was not used

for the XG solutions, as it is highly shear-thinning. An attempt to
Fig. 2. The axial symmetric double gap geometry.
Fig. 3. The zero-shear viscosity as a function of polymer concentration. The test was
conducted in an attempt to determine the overlap-concentration of Xanthan Gum.
74 A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287
obtain the overlap concentration of this polymer was conducted by
measuring its zero-shear viscosity at different concentrations.
There is a value of concentration for which the variation of g
0
is
considerably increased, and we suppose that the overlap concen-
tration, c

, is below such a value. We conducted here the same pro-

cedure used by Jaafar et al. [22]. By means of this technique, we
estimated the overlap concentration of the Xanthan Gum to be
c

= 940 ppm (see Fig. 3). In fact, this technique is not very precise
and c

could be smaller. By means of the same technique Wyatt

et al. [55] found c

around 70 ppm using a Xanthan Gum with

the same molecular weight used by this work. The maximum poly-
mer concentration used in this work was 100 ppm which suggests
we are working with diluted solutions. Using deionized water as a
solvent, the polymer powders were gently deposited on the solvent
surfaces. Each test was carried out after 24 h, time for complete
natural diffusion. This procedure was adopted to avoid any poly-
mer degradation before the beginning of the test.
The solution viscosity was measured with the HAAKE MARS II
for a range of rotations in which the ow was viscometric. The val-
ues measured were compared with that obtained from a capillary
viscometer and quite a good agreement was observed. For concen-
trations smaller than 100 ppm (for both PEO and PAM) no signi-
cant shear-thinning behavior was noticed. Such behaviour was
only observed for very hight concentrations, larger than
1000 ppm (see Fig. 4A and B). Differently, the Xanthan Gum viscos-
ity has a signicant shear thinning behaviour, even for very small
concentrations (Fig. 4C). The shear-thinning behaviour of PEO,
PAM (at high concentrations) and XG is quite good tted by a Car-
reau-Yasuda like equation
g g
1
g
0
g
1

1
1 k
CY
_ c
a

n=a
; 5
where g
0
and g
1
are the viscosities in the zero-shear and innite
shear plateaus while 1/k
CY
and n are, respectively, a characteristic
shear rate and the power-law index. The parameter a were intro-
duced by Yasuda et al. [56]. It appears an abrupt increase in the
shear viscosity in the PEO solutions at shear rate around 1000 s
1
,
which seems rather strange. Shetty [43] also observed such behav-
ior of PEO, although no explanation was reported. The curve tting
parameters is displayed in Table 1. We also have measured the ratio
of storage and loss modulus, G
0
/G
00
, for solutions of PEO, PAM and XG
at different concentrations (Fig. 5). As in the previous test, the stor-
age and loss modulus was only computed for high concentrations of
PEO and PAM solutions, but for XG, even in relative small concen-
tration, G
0
/G
00
could be measured. Clearly, for each polymer, the rel-
ative importance of the elastic modulus increase with concentration
but what is worth noting here is the effect of the frequency on each
material. In PEO and PAM solutions G
0
/G
00
considerably increases
when the frequency is incremented, whereas in XG solutions the
parameter is almost constant in high concentrations. As reported
by Rochefort and Middleman [41], such behavior is an indication
that a gel-like structure exists. In fact, at high concentration the
solution of XG is essentially gel-like. An increasing importance of
the elastic effect with frequency is a very known characteristic of
viscoelastic materials. Whereas, obviously, the opposed trend is
an indication of rigidity. Using a Capillary Breakup Extensional Rhe-
ometer (CaBER) manufactured by Thermo Scientic, Germany, in
which a sample is elongated with a controlled upper-plate velocity,
we measured the dimensionless sample diameter variation over the
time and it is showed in Fig. 6. Clearly, the sample break up occurs
earlier in XG solutions for all concentrations tested and it is an evi-
dent indication that such kind of polymer is a rigid-like material.
We also added curve ttings to take into account a measure of
the solutions characteristic relaxation time, using the equation
A
B
C
Fig. 4. Dynamic viscosity as a function of shear rate. The measurements were
carried out for PEO, PAM and XG for a range of concentrations, and with the
temperature xed at 25 C. The molecular weight for PEO and PAM was maintained
at 5.0 10
6
g/mol whereas there was used a XG with M
v
= 2.0 10
6
g/mol.
A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287 75
proposed by Entov and Hinch [16].
1
As showed by the authors, the
midpoint radius of the lament evolves according to the following
equation:
Dt
D
0

g
0
g
S
D
0
2kr

1=3
expt=3k: 6
Here D
0
is the plate diameter in which the sample is xed, r is the
surface tension, g
0
is solutions viscosity at zero shear rate, g
S
is the
solvents viscosity and k is the characteristic relaxation time, which
is displayed in the Fig. 6. The values of k are clearly more pro-
nounced on PEO and PAM solutions. Hence, measures showed in
Figs. 5 and 6 conrm, what was just expected, that XG is, in fact,
a rigid kind of polymer.
The maximum rotational speed of the rotor used was
n = 3000 rpm (revolution per minute). The ow eld becomes
unstable in Ta, Eq. (4), close to 1700. This value of Ta is achieved
when n is close to 500 rpm. This corresponds to Re
q

hu
g
P350.
Drag reduction is only observed for values of Ta beyond this critical
value. In the main tests, the rotational speed was kept constant to
display the drag reduction as a function of time which was ex-
tended over 7000 s and around 5400 shear stress values were
measured.
The kind of geometry used here can, eventually, exhibit some
laminar instabilities, such as Taylor-vortices, before reaching to
fully turbulent ow. Thus, someone could question whether our re-
sults are related to Taylor instabilities or to turbulence. In an at-
tempt to quantify the real importance of laminar instabilities on
the drag reduction and degradation, Pereira and Soares [38] per-
formed a sequence of tests using the double gap and a standard
Taylor-Couette geometry for a range of Taylor numbers. The
authorss analysis conducted for exible polymers showed that,
the drag reduction and, principally, the degradation, in the double
gap are predominantly related to turbulence instead of any kind of
laminar instability.
3. Results and discussion
The tests were conducted in attempt to highlight some quanti-
tative and qualitative differences between two types of drag reduc-
ers: exible (PAM and PEO) and rigid (XG) polymers. The results
are displayed considering variations in the Reynolds number, con-
centration, and temperature. We present our results in three parts.
Table 1
Carreau-Yasuda parameters for PEO, PAM and XG solutions.
Polymer M
v
(g/mol) c (ppm) g
0
(Pa s) g
1
(Pa s) k
CY
(s) n a
PEO 5.0 10
6
1000 0.2000 0.0025 3.00 0.69 0.21
PEO 5.0 10
6
2500 1.0000 0.0040 3.40 0.58 0.15
PEO 5.0 10
6
5000 2.0000 0.0050 4.00 0.58 0.26
PEO 5.0 10
6
10000 5.0000 0.0075 4.80 0.58 1.50
PAM 5.0 10
6
2500 0.4500 0.0032 9.00 0.51 0.15
PAM 5.0 10
6
5000 0.7000 0.0040 10.00 0.48 0.14
PAM 5.0 10
6
10000 1.1500 0.0050 11.00 0.46 0.25
XG 2.0 10
6
37.5 0.0920 0.0010 3.80 0.87 0.35
XG 2.0 10
6
50 0.1050 0.0011 4.00 0.80 0.39
XG 2.0 10
6
100 0.1800 0.0012 6.00 0.70 0.32
XG 2.0 10
6
250 0.3300 0.0012 3.70 0.73 0.55
XG 2.0 10
6
500 0.6000 0.0015 3.60 0.72 0.38
XG 2.0 10
6
1000 1.6000 0.0020 9.10 0.73 1.80
XG 2.0 10
6
2500 6.0000 0.0030 9.50 0.80 1.50
XG 2.0 10
6
5000 40.0000 0.0040 25.00 0.83 3.00
XG 2.0 10
6
10000 450.0000 0.0050 70.00 0.88 2.00
A
B
C
Fig. 5. G
0
/G
00
as a function of oscillation frequency. The measurements were carried
out for PEO, PAM and XG for a range of concentration and temperature xed at
25 C. The molecular weight for PEO and PAM was maintained at M
v
5.0 10
6
g/mol
whereas it was used a XG with M
v
= 2.0 10
6
g/mol.
1
Clasen et al. [11] reported that an additional factor of 2
1/3
is missing in the
prefactor of Eq. (6), though this does not change the value of the relaxation time.
76 A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287
In subSection 3.1, we show the Fanning friction factor in Prandtl
von Krmn coordinates for a range of concentrations of each poly-
mer. SubSection 3.2 presents DR(t) over time from the very begin-
ning of the test until reaching its asymptotic value after a long
enough time. As the transient response of the rheometer lasts
around 2 s, at the highest rotors speed used here, we disregarded
data before such a period of time. For this reason, we were not able
to capture the drag increase at the very start of the test. Finally, in
subSection 3.3, we present the time dependent relative drag reduc-
tion, DR
0
(t), where the loss of efciency, eventually caused by the
degradation mechanism, is more clearly discussed.
3.1. Fanning friction factor in Prandtlvon Krmn coordinates
Fig. 7 shows the Fanning friction factor in the Prandtlvon
Krmn coordinates. The dashed line is the MDR asymptote for
the double gap device. Such asymptote was built using an expres-
sion for DR
max
as a function of Reynolds number, concentration,
temperature and molecular weight proposed by Pereira and Soares
[38]. Maximizing each parameter, we could nd an alternative
Virks law, which is given by
1

f
p 17:00logRe

f
p
11:55: 7
Our results were displayed for a range of concentrations
(1 ppm6 c 6 50 ppm) of PEO, PAM, and XG with the temperature
xed at 25 C. The rotation was gradually increased from 0 to
3000 rpm over 5 min. The values of f addressed in Fig. 7 are related
to DR
max
regime, sketched in Fig. 1. As widely reported by a number
of researchers, the friction factor falls and the onset of drag reduc-
tion occurs at smaller values of Reynolds numbers with increasing
concentration. This is clearly observed for the exible polymers
(PEO and PAM) and also for the rigid one (XG), Fig. 7AC. It is also
clear that the values of the coefcient 1=

f
p
are more pronounced
in the PEO solutions. The smallest values of this parameter are ob-
served in the XG solutions. However, what is worth noting here is
the fact that the phenomenon for XG solutions seems to be consid-
erably different. For PEO and PAM solutions, it is clearly observed
that the curves at distinct concentrations are moving away from
each other with increasing Reynolds number. In other words, the
slopes of the curves are an increasing function of concentration.
In contrast, the lines displayed for each concentration of XG, after
the onset, seem to be parallel. The slopes of the curves for different
concentrations are quite the same. It is also noticeable that the on-
set of drag reduction occurs at a much smaller Reynolds number in
XG solutions. This is easily perceived for c = 50 ppm, in which the
onset is coincident with the MDR asymptote. This observation sug-
gests that the Reynolds number plays a weak role in Xanthan Gum
solutions. Results reported by Bewersdorff and Singh [4] and Be-
wersdorff and Berman [3], using Xanthan Gum, and more recently
by Jaafar et al. (2009) using Scleroglucan (another well-known ri-
gid-chain polymer), also indicate such an effect of the Reynolds
number on the friction factor. The reason for these observations
is related to the fact that XG is a rigid polymer and is already ex-
tended on its equilibrium state at rest. Hence, our observation is
in agreement with the idea of a Type B mechanism of drag reduc-
tion reported by Virk et al. [53].
3.2. Drag reduction decay
In this subsection, we display the drag reduction over time to
take into account the phenomenon from the very start of the test
until the asymptotic value of DR is achieved. The tests were carried
out for each polymer at a range of Reynolds numbers, concentra-
tions and temperatures. The dependence of the drag reduction on
time can be, at least, divided into three distinct periods. A sketch
of DR(t) is shown in Fig. 1. From the very beginning of the test,
A
B
C
Fig. 6. Evolution of the dimensionless diameter over time. The measurements were
carried out for PEO, PAM and XG for a range of concentrations, and with the
temperature xed at 25 C. The molecular weight for PEO and PAM was maintained
at M
v
5.0 10
6
g/mol whereas there was used a XG with M
v
= 2.0 10
6
g/mol.
A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287 77
the DR increases before reaching its maximum efciency. DR
max
is
not achieved instantaneously because a period of time, the develop-
ing time t
d
, is required for the turbulent structures to arrange after
the high degree of polymer deformation at the very start of the
test. It is worth noting that t
d
is far larger than the relaxation time
of a single molecule. Such a mechanism was for the rst time
numerically computed by Dimitropoulos et al. [13], who reported
negative values of DR, and, as far as we know, experimentally ob-
served by Pereira and Soares (2012). The maximum level of ef-
ciency is sustained for a while, the resistance time t
r
. Supposedly,
when the degradation becomes important, DR starts to decrease
until achieving its asymptotic value, a time during which the poly-
mer scission stops and the molecular weight distribution reaches a
steady state. A discussion concerning the effect of the Reynolds
number, concentration, molecular weight, and temperature on
the developing time and the resistance time was reported by Pere-
ira and Soares [38]. The authors analysis considers turbulent ows
of PEO and PAM solutions. The explanation for the complex behav-
iour of DR(t) is related to the delay of the rearrangements of the
turbulent structures after the polymer stretching. Here our main
interest is to highlight the main difference between the molecules
PEO, PAM, and XG from such a point of view. As mentioned previ-
ously, the rst and the second are exible molecules whereas the
last is considered a rigid one.
Fig. 8 displays DR against time for each polymer (PEO, PAM and
XG) for a range of Reynolds numbers at two different concentra-
tions. The temperature was maintained at 25 C. From the top to
the bottom, the results are displayed, respectively, for PEO, PAM
and XG. We can observe, comparing the gures A and B that the
Reynolds number plays a more important role for the less concen-
trated solution of PEO. We can clearly see that the curves for the
less concentrated solutions are more separated from each other.
This is expected considering that the drag reduction by polymers
is bounded by a maximum drag reduction asymptote (MDR) and
100 ppm of PEO is very close to the MDR asymptote. Thus, an in-
crease of the Reynolds number value at the less concentrated solu-
tion could provide a more signicant change of DR. Moreover, the
developing time, t
d
, seems to be delayed when the concentration is
increased. Such an effect is apparent for PEO solutions, as could be
observed comparing the black symbols in Fig. 8A and B, in which
the Reynolds number is exactly the same. Such an effect was also
reported by Pereira and Soares [38]. The authors argue that an
increasing concentration causes a more intense ow disturbance
at the beginning of the process and consequently the time to reach
a steady state turbulent structure is also increased. In PAM and XG
solutions the effect of concentration on t
d
was not noticeable. The
developing time seems to be very related to the polymer confor-
mation at the very start of the test. It is known that PEO is coiled
and XG has an extended conformation in a sample at rest. Hence,
after the test start-up, the solution of PEO requires more energy
from the mean ow to have their molecules stretched and this
causes more disturbances in the mean ow than that caused by
the XG solution, in which the molecules are already stretched.
We suppose the molecules of PAM are less coiled than the mole-
cules of PEO and consequently t
d
is shorter for such polymer. In
fact, the aggregates in PEO solutions, possibly, increase the level
of molecule entanglement, which, eventually, play an important
role on the developing time. However, more experiments focused
on the developing time are necessary to achieve a nal conclusion.
For example, both PAM and XG seems to have a very fast develop-
ing time and, to take into account any difference in t
d
for those
polymers, we should include measurements in a period of time be-
low three seconds.
Fig. 8B, D and F are displayed to provide a direct comparison be-
tween each polymer. The tests were carried out for the same range
of Reynolds number and the same concentration and temperature.
A
B
C
Fig. 7. Effects of concentration on Fanning friction factor, f, as a function of
Reynolds number, Re. The rotation was gradually increased from 0 to 3000 rpm. The
measurements were carried out for PEO, PAM and XG, respectively, A, B and C, at
different concentrations, and with the temperature xed at 25 C. The molecular
weight for PEO and PAM was maintained at 5.0 10
6
g/mol whereas there was
used a XG with M
v
= 2.0 10
6
g/mol.
78 A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287
Considering PEO and PAM, (B) and (D), we can see that DR
max
is
slightly larger in PEO solutions, but DR
asy
is considerably more pro-
nounced in PAM solutions. This is clearly shown in Fig. 10, in which
DR
max
is displayed together with DR
asy
for each polymer as a func-
tion of concentration. This is evidently related to the fact that the
degradation is more intense in PEO solutions. Even though higher
levels of DR are achieved with use of PEO, the mechanical molecu-
lar scissions act more intensely in such a polymer and its nal level
of drag reduction is greatly reduced. Such decrease in the efciency
of PEO can also be related to molecule de-aggregation, as sup-
ported by Shetty and Solomon [44]. It can also be concluded that
the highest level of drag reduction is achieved more quickly in
PAM solutions with increasing Reynolds numbers. This fact can
be perceived by analysing the change of the Reynolds number from
its smallest value (Re = 733) to the intermediate one (Re = 1047).
The blue and black curves are closer to each other for PAM. A direct
A B
C D
E F
Fig. 8. Effect of Reynolds number on DR as a function of time. The measurements were carried out for PEO, PAM and XG at xed temperature (25 C) in two different
concentrations. The molecular weight for PEO and PAM was maintained at M
v
5.0 10
6
g/mol whereas there was used a XG with M
v
= 2.0 10
6
g/mol.
A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287 79
comparison between XG and the other molecules are not evident
because the results for this polymer were obtained with a smaller
molecular weight. However some comparisons can be highlighted.
The rst one is concerned with the loss of efciency, which is more
pronounced in PEO solutions, which have higher molecular weight.
The second conclusion is related to the Reynolds number effect. As
with PAM, the increase in DR with increasing Re is less signicant
with XG than with PEO. The results displayed in (E) and (F) conrm
that the Reynolds number is, in fact, much less effective in XG solu-
tions, as reported by Bewersdorff and Singh [4] and Bewersdorff
and Berman [3]. It seems that Re = 733 is very close to the onset
and an increase in this parameter moves DR to a value very close
to its maximum, for the specic concentration. This is consistent
with the idea that XGs structure at rest is already extended and
can work even at a low level of turbulence, the Type B mechanism
of drag reduction reported by Virk et al. [53].
Fixing again the temperature (T = 25 C) and Reynolds number
(Re = 1360), Fig. 9 shows the effect of the concentration on DR(t)
for our three drag reducers (PEO, PAM and XG). As expected and
widely reported by a number of researchers, DR increases with
an increasing concentration. This is evident in all drag reducers
used here. Concerning PEO and PAM, (A) and (B), the concentration
of 100 ppm seems to be close to the MDR asymptote. Thus, DR
max
for both materials is very close to each other (approximately
0.23). For any other value of the concentration, the ultimate level
of drag reduction is higher in PEO solutions. Fig. 10 shows clearly
such a thing. It is worth noting that a very small amount of PEO
(2 ppm) causes a signicant reduction in the drag (maximum value
of 0.14). The same quantity of PAM provides the maximum drag
reduction of 0.07, just one-half of the previous one. When the con-
centration of both polymers is continuously increased, we observe
clearly that the growth of DR is more pronounced in PAM solutions.
In other words, the curves of different concentrations of PEO are
closer to each other. With respect to the XG solutions, Fig. 9(C),
the changes in DR with concentration are gradual, as in PAM solu-
tions. However, as the molecular weight in this case is smaller, this
is not a conclusive comparison. An interesting point that deserves
our attention is the fact that, with 100 ppm of XG, DR
max
achieves a
value around 0.27. For PEO and PAM, the maximum value of DR
achieved is 0.23, which is, supposedly, close to the MDR. Thus, pos-
sibly, the XG solution of 100 ppm is providing a drag reduction be-
yond the MDR asymptote. Concerning this fact, a more careful
A
B
C
Fig. 9. Effect of concentration on DR as a function of time. The measurements were
carried out for PEO, PAM and XG at xed temperature (25 C) and Reynolds number
(Re = 1360). The molecular weight for PEO and PAM was maintained at M
v
5.0 10
6
g/mol whereas there was used a XG with M
v
= 2.0 10
6
g/mol.
Fig. 10. Effect of concentration on the maximum and asymptotic drag reduction for
each polymer. The curve ts were obtained using an expression for DR
max
proposed
by Pereira and Soares [38] (see Eq. (7) of the referred paper).
80 A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287
analysis is strongly recommended once only few points were mea-
sured. Hence, it is safer to consider as DR
max
the rst asymptotic le-
vel of drag reduction, which shows DR around 0.22.
It is interesting to note that t
d
seems to have a rather weak rela-
tion to the concentration of XG and PAM, whereas the relation be-
tween t
d
and c is very strong in PEO solutions, as reported by
Pereira and Soares [38]. For PAM and XG solutions, DR
max
is
achieved in less than 3 s, even for the highest concentration of each
polymer. As mentioned previously, this is probably related to mol-
ecule conformation at the test start-up and deserves attention. As
reported numerically by Dimitropoulos et al. [13], the turbulent
structures take time to rearrange and achieve their nal form after
a large polymer deformation. In fact, we suppose that the steady
state ow is achieved when the necessary number of molecules
A
B
C D
E F
Fig. 11. Effect of temperature on drag reduction, DR, as a function of time, t. The measurements were carried out for PEO, PAM and XG at xed Reynods number (Re = 1360) in
two different concentrations. The molecular weight for PEO and PAM was maintained at 5.0 10
6
g/mol whereas there was used a XG with M
v
= 2.0 10
6
g/mol.
A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287 81
is working in a state of equilibrium with the turbulent structures.
Denitely, this state of equilibrium is sensitive to any change in the
molecular structural conguration and the developing time in-
creases when changes in the molecular conguration play a impor-
tant role in the process. Hence, smaller values of t
d
are to be
expected for XG solution as its molecular conguration does not
change signicantly at the beginning of the test. This very fast
developing time observed for XG is consistent with the idea of a
Type B drag reduction mechanism. We suppose the molecules of
PAM are not so coiled as the molecules of PEO at the very start
of the test and, consequently, t
d
should be shorter. However, this
point deserves more attention. Tests in a period of time below
three seconds could clarify such point.
A practical aspect of the problem that is worth attention is con-
cerned with the resistance of the highly concentrated solution of
PAM. In Fig. 9B we observe that the maximum level of efciency
is maintained over all the test time for the solution of 100 ppm.
This observation contributes to the hypothesis that PAM solutions
are stronger than PEO solutions (see [15]). Regarding the solutions
resistance, the results for XG are quite curious. In contrast to the
other polymers, XG solutions with very small concentrations
(c 6 5 ppm) showed no loss of efciency, suggesting that molecular
degradation does not play an important role in these solutions. In
fact, what is more curious is that the loss of efciency appears
when the concentration is increased. Clearly, the period of time
in which the maximum level of DR is maintained falls with increas-
ing concentration. Such a fall is abrupt at very high values of con-
centration, c P75 ppm. It is also worth noting that the curve of
c = 100 ppm is quite different from all the other results. After the
abrupt DR decrease at the very beginning of the test, the drag
reduction achieves a constant level before it starts to decrease
again. Such observations concerned with the effect of concentra-
tion in XG solutions are considerably distinct from those noticed
in the other polymers, PEO and PAM. The particular way in which
the XG decay curve appears is worth attention. There are clearly
two power-law regions between DR
max
and DR
asy
, whereas in PEO
and PAM only one power-law region is seen. We will return to this
point again.
The effect of temperature on drag reduction, DR, as a function of
time, t, is displayed in Fig. 11. The measurements were carried out
for PEO, PAM and XG at a xed Reynolds number (Re = 1360) in
two different concentrations. For each polymer, the temperature
seems to play a more important role for the less concentrated solu-
tion. In other words, an increase in the temperature causes an in-
crease in DR more pronounced at smaller concentrations. Such
evidence can be observed when comparing the black and blue
curves in Fig. 11C and D. When T is changed from 25 C to 35 C,
the growth of DR in the PAM solution of 10 ppm is signicantly lar-
ger. The same conclusion can also be seen by observing the black
and red curves of the PEO solutions, (A) and (B). The evidence is
not clear for XG. Observing the data for each material, Figures
(B), (D) and (F), we have an impression that the drag reduction is
more affected by temperature in XG solutions. Regarding the data
in (E), we can conclude that DR is, in fact, highly inuenced by tem-
perature. From 20 C to 40 C, DR changes signicantly its decay
function. At 20 C, the green symbols, it is easy to perceive three
power-law regimes in DR(t) between its maximum and asymptotic
level. The curve for 25 C, black symbols, is characterized by two
power-lawregimes, whereas in temperatures larger that 30 C only
one power-law regime is perceived. Up from this level of temper-
Fig. 12. Drag reduction as a function of time for XG solutions previously sheared at
different shear rate.
Fig. 13. Effect of concentration on drag reduction, DR, as a function of time, t, for
Xanthan Gum solutions previously sheared at a xed _ c.
Fig. 14. DR as a function of time, t, for Xanthan Gum solutions previously shaken.
82 A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287
ature, the XG decay function is quite similar to that observed for
PEO solutions. It is worth noting that DR is considerably increased
in PAM and XG solutions. Increasing the temperature from 25 C to
45 C, the value of DR
max
in PEO, PAM and XG solution was in-
creased by, respectively, 20%, 26% and 35%. An increasing temper-
ature provides an increase in the mean length of the molecule,
which probably causes such gain in efciency. Nakken et al. [35]
also observed an increasing efciency on DR when temperature
rises and such effect is attributed to changes in the solvent condi-
tions. In fact, when temperature is decreased a single polymer as-
sumes a more contracted conformation, which causes a reduction
in the efciency of the drag reduction.
A B
C D
E
F
Fig. 15. Effect of Reynolds number on DR
0
as a function of time. The measurements were carried out for PEO, PAM and XG at xed temperature (25 C) in two different
concentrations. The molecular weight for PEO and PAM was maintained at 5.0 10
6
g/mol whereas there was used a XG with M
v
= 2.0 10
6
g/mol.
A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287 83
After a close examination of the drag reduction decay functions
for each polymer, it is clear that each variables, Re,T and c, plays a
quite different role in the Xanthan Gum mechanism of drag reduc-
tion. In an attempt to understand the main reason for such differ-
ences, we carried out some other specic tests for XG. The rst one
is shown in Fig. 12, in which DR(t) is displayed for three different
previously sheared samples at xed Reynolds number, tempera-
ture and concentration (Re = 1360, T = 25 C and c = 50 ppm). The
shear rates used to mix the solutions for 900 s was kept below
the minimal necessary to start the ow instabilities. In other
words, the samples were mixed in laminar ows, in which, suppos-
edly, degradation does not play an important role. Despite this fact,
DR
max
clearly falls from 0.20 to 0.15 (25% loss of efciency). The for-
mer value (red balls) was obtained using the original solution (not
previously sheared), whereas the last was for a sample mixed for
900 s at _ c 3995 s
1
(purple squares). Some could argue that the
samples were not totally mixed and the fall in DR
max
could have
been caused, simply, by an increasing viscosity. Obviously, we
should measure different g _ c using a sample not totally homoge-
nized. However, this seems not to be the case. The apparent viscos-
ity was measured as a function of the shear-rate, we veried that
the viscosity does not change at all. We tested three samples of
50 ppm of XG at 25 C. The rst one was diluted in 72 h and the
other two in 576 h. Moreover, one of these older samples was pre-
viously mixed at _ c 3995 s
1
. Hence, neither the dilution period of
time nor the mixing shear rate are the cause of a signicant change
in g. Hence, from the viscosity point of view, the solutions are quite
well mixed. The mixing effect is highly dependent on the concen-
tration, as can be seen in Fig. 13. In this gure DR, is displayed
against time for a range of XG concentrations, as in Fig. 9C. But un-
like this gure, all the samples were previously mixed at
_ c 3995 s
1
. We can see that for concentrations below 10 ppm,
both results are quite similar. These similarities vanish for
c P25 ppm. The maximum drag reduction level falls considerably
when it is compared with the sample not previously sheared. As
widely reported, the main cause of an efciency decrease is molec-
ular scission. However, molecular scission, supposedly, does not
play an important role in shear ows, since in such ows the mol-
ecules predominantly spin instead of stretching. Hence, an exten-
sional ow is necessary to mechanically break a signicant
number of molecules and cause a decrease in the efciency. Based
on this hypothesis, we suppose this fall in DR
max
is not related, at
least predominantly, to molecular degradation. Moreover, another
reason to eliminate degradation as the main cause of this effect is
the fact that the maximum drag reduction loss is more pronounced
in high concentrations and it is known that an increasing concen-
tration supposedly makes the solution stronger. Our results for PEO
and PAM solutions, besides others from the literature, conrm this
statement. Thus, what could be happening is something related to
a microstructural change in the polymer. Shetty and Solomon [44],
using light scattering, give evidence of formation of aggregates in
aqueous solutions of PEO, even in ultra-low concentration. This
polymer structure comprised of more than one molecular chain in-
creases the drag reduction efciency. We suppose the Xanthan
Gums solution also forms aggregates which are partially or totally
destroyed when the solution is sheared and in this new condition
DR
max
falls. However, unlike PEO, we believe the aggregates are
more probable at high concentrations. Thus, there would not be ex-
pected any difference of behavior for low concentrations of pre-
sheared solutions. In fact, at very small values of c, there is not per-
ceived any great difference between the original solutions (dis-
played in Fig. 9C) and the previously mixed one (Fig. 12). Such a
difference is only noticed at high concentrations. Hence, Fig. 9C
shows an increase in DR with concentration, but at the same time
an increasing c can favor the formation of aggregates. During the
test, a number of the aggregates can break and this, eventually,
A
B
C
Fig. 16. Effect of concentration on DR
0
as a function of time. The measurements
were carried out for PEO, PAM and XG at xed temperature (25 C) and Reynolds
number (Re = 1360). The molecular weight for PEO and PAM was maintained at
5.0 10
6
g/mol whereas there was used a XG with M
v
= 2.0 10
6
g/mol.
84 A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287
causes a decrease in the efciency of the drag reduction. Thus, we
believe that the decrease of DR in XG solutions can be related, pre-
dominantly, to a change of structure instead of to degradation.
Such a structural change could, eventually, be partially reversible.
An attempt to investigate this hypothesis is displayed in Fig. 14,
in which DR(t) is shown for xed temperature and Reynolds num-
ber. The red balls are our original sample of 50 ppm. The blue tri-
angles represent DR after a highly shaken 50 ppm of XG. Clearly,
DR
max
falls and the level of DR stays constant over the test. The va-
lue of DR
max
is very close to that obtained for a pre-sheared solu-
tion. After 100 h, a new test was conducted. DR(t), represented
by the green diamond, shows a intermediate behavior. This sug-
A B
C
D
E F
Fig. 17. Effect of temperature on the relative drag reduction, DR
0
, as a function of time, t. The measurements were carried out for PEO, PAM and XG at xed Reynods number
(Re = 1360) in two different concentrations. The molecular weight for PEO and PAM was maintained at 5.0 10
6
g/mol whereas there was used a XG with M
v
= 2.0 10
6
g/
mol.
A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287 85
gests that there is some regeneration. In other words, after some
time at rest, the aggregates could be recovered. Evidently, more
tests must be conducted in order to verify such a theory.
3.3. The relative drag reduction decay
In this subsection we display the relative drag reduction against
the time from the maximum level of efciency to its asymptotic
value after a long enough time. Hence, DR
0
varies between 0 and
1. This style of data exhibition is an attempt to look directly at
the loss of efciency, which is related to the mechanical molecular
scission or, eventually, to de-aggregation.
Using data from Fig. 8 and 15 shows DR
0
against time to clarify
the effect of the Reynolds number on the loss of efciency, L
ef
= 1 -
DR
0
asy
. A fall of L
ef
is expected with increasing concentration. This
is clearly observed in the PEO and PAM solutions, but not in the XG
solutions. The black and blue curves of the more concentrated
solution of PEO and PAM, respectively, (A) and (D), are closer to
1. In contrast, observing (E) and (F), it seems that the two curves
are very similar to each other. In fact, DR
0
asy
for the two concentra-
tions of XG, at the same Reynolds number, are practically identical.
If an increasing concentration should turn a polymer solution into
a stronger one, we could suppose the mechanism of loss of ef-
ciency here is really different and, as mentioned before, could be
related to de-aggregation, instead to mechanical scission. We will
return to this point later when the effect of the concentration of
XG is reported. Comparing the different polymers, it is very clear
that the PAM solutions present the smallest loss of efciency. At
the maximum Reynolds number (Re = 1360), L
ef
is only 20% in
PAM solutions, whereas it is 50% and 55% for XG and PEO solutions,
respectively. As the maximum efciency is expected for the small-
est Reynolds number, DR
0
at Re = 733 behaves strangely, at least at
rst glance, in the PEO and XG solutions. We suppose this value of
Re is very close to the onset of the drag reduction and, after a per-
iod of time, a great amount of degraded molecules, or de-aggre-
gates in the case of XG, stop being stretched and interact with
the turbulent structures. In other words, the Re = 733 is below
the onset Reynolds number for a signicant number of molecules.
We suppose this is the cause of the abrupt fall of efciency and the
observed intersection of the curves.
Using data from Figs. 9, 16 displays DR
0
(t). Figures AC show
that the efciency of drag reduction is strongly dependent on the
concentration c. For PEO and PAM solutions, L
ef
is considerably re-
duced with increasing c. Supposedly, the loss of efciency for ex-
ible polymers is related to polymer degradation and the solution
resistance increases with increasing concentration. Thus, such an
effect on DR
0
is expected. It is interesting to note here that the
PAM solution of c = 100 ppm does not lose efciency. In other
words, DR
0
is kept at its maximum level over all the test time,
around 2 h. Comparing (A) and (B), we can conclude that the ef-
ciency of PAM solutions is notably larger than that of the PEO solu-
tions. The data for the XG solutions, depicted in Fig. 16C, show an
effect of concentration considerably different from the other poly-
mers and quite curious, at rst glance. For the range of concentra-
tion 2 6 c 6 37.5 ppm, DR
0
falls with increasing c. After
c = 37.5 ppm, the effect of concentration is similar to that observed
for PEO and PAM, and DR
0
rises with an increase in this parameter.
As mentioned before, we believe this effect is related to molecule
aggregations, which is here a polymer structure comprised of more
than one molecular chain. We are not talking about a gel-like
structure. Results reported by Shetty and Solomon [44] give some
support for this hypothesis. The authors give evidence of such
structure for ultra diluted solutions of PEO. In contrast to PEO,
we argue that the aggregates of XG only appear for c P5 ppm. In
addition, we suppose that the loss of efciency is only related to
de-aggregation. Hence, at very small values of c, in this case
c 6 5 ppm, there are not aggregates and, during the test L
ef
is equal
to 1. Increasing the concentration, structures comprised of two or
more molecules start to appear and could, eventually, de-aggre-
gate. We suppose that the de-aggregation is intensied by the tur-
bulence and a loss of efciency is perceived. Hence, possibly, a
great part of the loss of efciency observed could be related to
de-aggregation of the Xanthan Gum. In other words, we could
say that polymer scission in this case do not play an important role.
Fig. 17 shows the effect of temperature on the relative drag
reduction using the data from Fig. 11. The effect of temperature
on the drag reduction efciency is really complex. For PEO solu-
tions, whose results are displayed in (A) and (B), we can observe
that an increasing temperature accelerates the loss of efciency,
but the asymptotic value of DR
0
increases with temperature, at
least for the range of temperature below 45 C. It seems that DR
0
asy
starts to fall for increments of T above this value. The dependence
of DR
0
on temperature is quite similar in PAM solutions. DR
0
asy
also
increases with an increasing temperature. The role of temperature
at the smallest concentrated PAM solution (C), is more pronounced
than in the solution of 37.5 ppm (D). Such a difference is not ob-
served in the PEO solutions. Temperature also plays a very signi-
cant role in the XG solutions, especially in the less concentrated
solution, where the difference in DR
0
between 25 C and 45 C is
clearly more pronounced. The shape of DR
0
(t) changes considerably
from 20 C to 25 C in the more concentrated solution, (E). As in
other polymers, DR
0
asy
increases with an increasing temperature.
4. Final remarks
We have analyzed the evolution of the drag reduction over time
for three polymers, PEO, PAM and XG. The rst and second are ex-
ible polymers and the third is a rigid polymer. A wide range of Rey-
nolds numbers, concentrations, and temperatures were studied
from the very beginning of the test until the asymptotic value of
DR was achieved. The main results were displayed in an attempt
to provide a direct comparison between all three polymers.
As reported by many researchers, our results show that the Rey-
nolds number plays a weak role in DR and DR
0
for Xanthan Gum
solutions. This is very different from what is observed for PEO
and PAM, where Re is quite important. In fact, the mechanism of
drag reduction in rigid polymers is very different from that in ex-
ible ones and it is related to the moleculess microstructure, as re-
ported by Virk et al. [53]. Concerning the concentration, it is worth
noting that no loss of DR was observed for high values of c in the
PAM solutions. This suggests that PAM is more resistant than
PEO, though the last one provides higher values of DR
max
. As ex-
pected, the level of drag reduction, DR, and the its efciency, DR
0
,
increases with an increasing concentration for PEO and PAM. How-
ever, the relative drag reduction dependency on concentration for
Xanthan Gum solutions showed a considerably different behavior.
With an increasing concentration, DR
0
falls before starting to in-
crease again. We believe such an effect is mostly related to struc-
tures formed by polymer aggregates of the XG, which can change
from a structure comprised of two or more molecules to the single
molecule conguration. Such a change is helped by the ow, even
in the laminar regime. In other words, we argue that the loss of
efciency is mainly caused by de-aggregation during the test.
Our results also suggest that there is a kind of regeneration for
XG solutions. In other words, DR
max
of a pre-shaken solution can in-
crease after some time at rest. Probably, some molecules can, even-
tually, return to aggregated form after a long enough time.
Temperature plays a very important role in DR, specially in XG
solutions. An increase in T below the transition-midpoint tempera-
ture causes a signicant gain in the efciency of Xanthan Gum.
At 45 C, the loss of efciency is only slightly perceived.
86 A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287
Acknowledgements
This research was partially funded by grants from the CNPq
(Brazilian Research Council), the ANP (Brazilian Petroleum
Agency), and Petrobras.
References
[1] R. Benzi, A short review on drag reduction by polymers in wall bounded
turbulence, Physica D 239 (2010) 13381345.
[2] N.S. Berman, Drag reduction by polymers, Annual Reviews of Fluid Mechanics
10 (1978) 4764.
[3] H.W. Bewersdorff, N.S. Berman, The inuence of ow-induced non-Newtonian
uid proprerties on turbulent drag reduction, Rheologica Acta 27 (1988) 130
136.
[4] H.W. Bewersdorff, R.P. Singh, Rheological and drag reduction characteristics of
xanthan gum solutions, Rheologica Acta 27 (1988) 617627.
[5] V.C. Bizotto, E. Sabadini, Poly(ethylene oxide) polyacrylamide which one is
more efcient to promote drag reduction in aqueous solution and less
degradable, Journal of Applied Polymer Science 25 (2008) 18441850.
[6] E.D. Burger, L.G. Chorn, Studies of drag reduction conducted over a broad range
of pipeline conditions when owing prudhoe bay crude oil, J. Rheology 24
(1980) 603.
[7] S. Chakrabarti, B. Seidl, J. Vorkwerk, P.O. Brunn, Correlations between porous
medium ow data and turbulent ow results for aqueous hydroxypropylguar
solutions (part 2), Rheologica Acta 30 (1991) 124130.
[8] H.J. Choi, M.S. Jhon, Polymer-induced turbulent drag reduction, Industrial and
Engineering Chemistry Research 35 (1996) 29932998.
[9] H.J. Choi, C.A. Kim, M.S. Jhon, Universal drag reduction characteristics of
polyisobutylene in a rotating disk apparatus, Polymer 40 (1999) 45274530.
[10] H.J. Choi, C.A. Kim, J. Sohn, M.S. Jhon, An exponential decay function for
polymer degradation in turbulent drag reduction, Carbohydrate Polymers 45
(2001) 6168.
[11] C. Clasen, J.P. Plog, W. Kulicke, M. Owens, M. C, L. Scriven, M. Verani, G.H.
McKinley, How dilute are dilute solutions in extensional ows?, Journal of
Rheology 50 (6) (2006) 849881
[12] S.R. Deshmukh, R.P. Singh, Drag reduction characteristics of graft copolymers
of xanthan gum and polyacrylamide, Journal of Applied Polymer Science 32
(1986) 61636176.
[13] C.D. Dimitropoulos, Y. Dubief, E.S.G. Shaqfeh, P. Moin, S.K. Lele, Direct
numerical simulation of polymer-induced drag reduction in turbulent
boundary layer ow, Physics of Fluids 17 (2005) 14.
[14] Y. Dubief, C.M. White, V.E. Terrapon, E.S.G. Shaqfeh, P. Moin, S.K. Lele, On the
coherent drag-reducing and turbulence-enhancing behaviour of polymers in
wall ows, Journal of Fluid Mechanics 514 (2004) 271280.
[15] B.R. Elbing, M.J. Solomon, M. Perlin, D.R. Dowling, S.L. Ceccio, Flow-induced
degradation of drag-reducing polymer solutions within a high-Reynolds-
number turbulent boundary layer, Journal of Fluid Mechanics 670 (2011) 337
364.
[16] V. Entov, E. Hinch, Effect of a spectrum of relaxation times on the capillary-
thinning of a lament of elastic liquid, Journal of Non-Newtonian Fluid
Mechanics 72 (1997) 3153.
[17] A.G. Fabula, Fire-ghting benets of polymeric friction reduction, Transactions
of the ASME Journal of Basic Engineering (1971) 93453.
[18] P.J. Flory, Principles of Polymer Chemistry, Cornell University Press, Ithaca, NY,
1971.
[19] K. Gasljevic, G. Aguilar, E.F. Matthys, On two distinct types of drag-reducing
uids, diameter scaling, and turbulent proles, Journal of Non-Newtonian
Fluid Mechanics 96 (2001) 405425.
[20] J. Golda, Hydraulic transport of coal in pipes with drag reducing additives,
Chem Engng Commun 45 (1986) 5367.
[21] H.L. Greene, R.F. Mostardi, R.F. Nokes, Effects of drag reducing polymers on
initiation of atherosclerosis, Polym Engng Sci (1980) 20449.
[22] A.J. Jaafar, M.P. Escudier, R.J. Poole, Turbulent pipe ow of a drag-reducing rigid
rod-like polymer solution, Journal of Non-Newtonian Fluid Mechanics 161
(2009) 8693.
[23] V.N. Kalashnikov, Dynamical similarity and dimensionless relations for
turbulent drag reduction by polymer additives, Journal of Non-Newtonian
Fluid Mechanics 75 (1998) 209230.
[24] V.N. Kalashnikov, Degradation accompanying turbulent drag reduction by
polymer additives, Journal of Non-Newtonian Fluid Mechanics 103 (2002)
105121.
[25] P.R. Kenis, Turbulent ow friction reduction effectiveness and hydrodynamic
degradation of polysaccharides and synthetic polymers, Journal of Applied
Polymer Science 15 (1971) 607.
[26] C.A. Kim, K. Lee, H.J. Choi, C.B. Kim, K.Y. Kim, M.S. Jhon, The turbulent drag
reduction by graft copolymers of guar gum and polyacrylamide, Journal of
Applied Polymer Science 31 (1985) 40134018.
[27] C.A. Kim, K. Lee, H.J. Choi, C.B. Kim, K.Y. Kim, M.S. Jhon, Universal
characteristics of drag reducing polyisobutylene in kerosene, Journal of
Macromolecular Science Pure and Applied Chemistry A34 (1997) 705711.
[28] K. Lee, C.A. Kim, S.T. Lim, D.H. Kwon, H.J. Choi, M.S. Jhon, Mechanical
degradation of polyisobutylene under turbulent ow, Colloid Polym Sci 280
(2002) 779782.
[29] J.L. Lumley, Drag reduction by additives, Annual Review of Fluid Mechanics 11
(1969) 367384.
[30] E.R. Morris, Molecular origin of xanthan solution properties, Extracellular
Microbial Polysaccharides, ACS Symposium Series 45 (1977) 8189.
[31] T. Moussa, C. Tiu, Factors aafecting polymer degradation in turbulent pipe
ow, Chemical Engineering Science 49 (1994) 16811692.
[32] G. Muller, M. Anrhourrache, J. Lecourtier, G. Chauveteau, Salt dependence of
the conformation of a single-stranded xanthan, International Journal of
Biological Macromolecules 8 (1986) 167172.
[33] A. Nakano, Y. Minoura, Relationship between hydrodynamic volume and the
scission of polymer chains by high-speed stirring in several solvents,
Macromolecules 8 (1975) 677680.
[34] T. Nakken, M. Tande, A. Elgsaeter, Measurements of polymer induced drag
reduction and polymer scission in taylor ow using standard double-gap
sample holders with axial symmetry, Journal of Non-Newtonian Fluid
Mechanics 97 (2001) 112.
[35] T. Nakken, M. Tande, B. Nystrom, Effects of molar mass, concentration and
thermodynamic conditions on polymer-induced ow drag reduction,
European Polymer Journal 40 (2004) 181186.
[36] I.T. Norton, D.M. Goodall, S.A. Frangou, E.R. Morris, D.A. Ress, Mechanism and
dynamics of conformational ordering in xanthan polysaccharide, Journal of
Molecular Biology 175 (1984) 371394.
[37] R.W. Paterson, F. Abernathy, Turbulent ow drag reduction and degradation
with dilute polymer solutions, Journal of Fluid Mechanics 43 (1970) 689710.
[38] A.S. Pereira, E.J. Soares, Polymer degradation of dilute solutions in turbulent
drag reducing ows in a cylindrical double gap rheometer device, Journal of
Non-Newtonian Fluid Mechanics 179 (2012) 922.
[39] P. Peyser, R.C. Little, The drag reduction of dilute polymer solutions as a
function of solvent power, viscosity, and temperature, Journal of Applied
Polymer Science 15 (1971) 26232637.
[40] T. Rho, J. Park, C. Kim, H. Yoonb, H. Suhb, Degradation of polyacrylamide in
dilute solution, Polymer Degradation and Stability 52 (1996) 287293.
[41] W. Rochefort, S. Middleman, Rheology of xanthan gum: salt, temperature, and
strain effects in oscillatory and steady shear experiments, Journal of Rheology
31 (1987) 337369.
[42] R.H.J. Sellin, J.W. Hoyt, J. Poliert, O. Scrivener, The effect of drag reducing
additives on uid ows and there industrial applications part ii: present
applications and futures proposals, Journal of Hydraulic Research 20 (1982)
235292.
[43] A.M. Shetty, Connecting Structure and Dynamics to Rheological Performance
of Complex Fluids, PhD Thesis, The University of Michigan, Ann Arbor,
Michigan, 2010.
[44] A.M. Shetty, M.J. Solomon, Aggregation in dilute solutions of high molar mass
poly(ethylene) oxide and its effect on polymer turbulent drag reduction,
Polymer 50 (2009) 261270.
[45] J.I. Sohn, C.A. Kim, H.J. Choi, M.S. Jhon, Drag-reduction effectiveness of xanthan
gum in a rotating disk apparatus, Polymer Degradation and Stability 69 (2001)
341346.
[46] M. Tabata, Y. Hosokawa, O. Watanabe, J. Sohma, Direct evidence for main chain
scissions of polymers in solution caused by high speed stirring, Polymer
Journal 18 (1986) 699712.
[47] M. Tabor, P.G. de Gennes, A cascade theory of drag reduction, Europhysics
Letter 7 (1986) 519522.
[48] S.A. Vanapalli, T.M. Islam, J.M. Solomon, Scission-induced bounds on
maximum polymer drag reduction in turbulent ow, Physics of Fluids
(2005) 17.
[49] P.S. Virk, Drag reduction by collapsed and extended polyelectrolytes, Nature
253 (1975) 109110.
[50] P.S. Virk, Drag reduction fundamentals, AIChE Journal 21 (1975) 625656.
[51] P.S. Virk, H.S. Mickley, K.A. Smith, The toms phenomenom: turbulent pipe ow
of dilute polymer solutions, Journal of Fluid Mechanics 22 (1967) 2230.
[52] P.S. Virk, H.S. Mickley, K.A. Smith, The ultimate asymptote and mean ow
structure in Toms phenomenon, ASME Journal of Applied Mechanics 37
(1970) 488493.
[53] P.S. Virk, D.C. Sherman, D.L. Wagger, Additive equivalence during turbulent
drag reduction, AIChE Journal 43 (1997) 32573259.
[54] C.M. White, M.G. Mungal, Mechanics and prediction of turbulent drag
reduction whit polymer additives, Annual Review of Fluid Mechanics 40
(2008) 235256.
[55] N.B. Wyatt, C.M. Gunther, M.W. Liberatore, Drag reduction effectiveness of
dilute and entangled xanthan in turbulent pipe ow, Journal of Non-
Newtonian Fluid Mechanics 166 (2011) 2531.
[56] K. Yasuda, R.C. Armstrong, R.E. Cohen, Shear ow properties of concentrated
solutions of linear and star branched polystyrenes, Rheologica Acta 20 (1981)
163178.
[57] J.F.S. Yu, J.L. Zakin, G.K. Patterson, Mechanical degradation of high molecular
weight polymer in dilute solution, Journal of Applied Polymer Science 23
(1979) 24932512.
A.S. Pereira et al. / Journal of Non-Newtonian Fluid Mechanics 202 (2013) 7287 87