SPE-169699-MS

Degradation (or Lack Thereof) and Drag Reduction of HPAM During

Transport in Pipelines
Stéphane Jouenne, SPE, Jérôme Anfray, SPE, Philippe R. Cordelier, SPE, Khalid Mateen, SPE, David Levitt,
SPE, TOTAL; Inès Souilem, Philippe Marchal, Lionel Choplin, GEMICO-LRGP-ENSIC-UL; Jonathon Nesvik,
SPE, Tom E. Waldman, SPE, NALCO TIORCO FABTECH

Copyright 2014, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE EOR Conference at Oil and Gas West Asia held in Muscat, Oman, 31 March–2 April 2014.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Rules of thumb that are used in the industry for polymer flooding projects, tend to limit the distance over which HPAM
polymers can be transported in pipelines without undergoing significant degradation. However, in sensitive environments
such as offshore facilities where footprint minimization is required, centralization of the polymer hydration process and long-
distance transport may be desirable. More reliable rules are required to design the pipe network and to estimate mechanical
degradation of polymers during transport in turbulent conditions.
In this work, we present evidence in the form of both empirical, large-scale pipeline experiments as well as theoretical
development refuting the claim that polymer pipeline transport is limited by mechanical degradation. Our work concludes
that mechanical degradation occurs at a critical velocity which increases as a function of pipe diameter. Provided the critical
velocity is not reached in a given pipe, there is no limit to the distance over which polymer solution can be transported.
In addition, the drag reduction of viscous polymer solutions was measured as a function of pipe length, pipe diameter,
fluid velocity and polymer concentration. An envelope was defined to fix the minimum and maximum drag reductions
expected for a given velocity on larger pipes. For pipes with diameter varying between 14” and 22’’ at a velocity higher than
1 m/s, the drag reduction percentage is anticipated to be between 45 and 80%. A more refined model was developed to
predict drag reduction with less uncertainty.
In conclusion, classical design rules applied for water transport (fluid velocity <3m/s) can be applied to the design of
a polymer network. Therefor, for tertiary polymer projects, the existing water injection network should be compatible with
the mechanical requirements of polymer transportation.

Introduction
When designing a polymer flooding project, the design of the piping network for polymer injection can be a serious
economic issue if the polymer dissolution process is far from injection wells. For tertiary polymer injection, a question likely
posed is the adaptability for polymer transport of the existing pipe network, initially designed for water. Rules of thumb have
been proposed (e.g. < 10 kilometers, maximum velocity of 1.5m/s) limiting the conditions of HPAM polymer transport
without undergoing significant degradation.
The objective of this study is to determine if a pipe network designed for a waterflood scenario is compatible with a
polymer injection, or whether, as previously suggested, additional precautions are necessary to prevent mechanical
degradation. We will not address precautions necessary to prevent oxidative degradation in this paper.
In an oilfield polymer distribution network, the transport will be turbulent on long distances. Two main questions are
addressed:
• Will the pressure drop while transporting the polymer solution be compatible with the pipe design and the pumping
capacity?
• Will it be necessary to over concentrate the polymer solution due to degradation of the polymer during its transport
in turbulent conditions?
To answer these questions, it is thus necessary to predict if drag reducing effect will occur during the transportation of
viscous polymer solutions. Although drag reduction of very dilute solutions have been extensively studied (Virk, 1975,
2 SPE-169699-MS

McComb and Rabie, 1982, Interthal, 1985, Bird et al., 1987, Kulicke, 1989) and can be as high as 80%, few papers deal with
semi dilute solutions. Moreover, a model is required for predicting the polymer degradation during the upscaling of the pipe
diameter and the transport distance.
A testing program was launched in 2012 in order to answer these questions by performing flow tests at two different
scales. A first series of tests was performed on 4 different pipes, 61 m long with internal diameter (ID) varying between ½
and 1’’. A second series of tests was performed at the pilot scale on a 7 km long, 6’’ diameter pipe.
For both test campaigns, polymer solutions at different concentrations were transported in turbulent conditions. The
strategy was to get trends on the small scale experiments (ID=0.5-1”, L=61 m) and to extrapolate to field conditions
(ID>14”), the extrapolation being validated through the large scale pilot tests (ID=6”, L=7 km).

Drag reduction, coil-stretch transition and polymer degradation

The drag reduction phenomenon was first described by Toms (1948) and Mysels (1949). They showed that the addition
of minute quantities of long chain polymers decreases the pressure drop in a pipe at fixed flow rate. The experimental work
done by Toms led to drag reduction subsequently being referred to as “Toms’ effect”. Numerous studies have focused on
various parameters affecting drag reduction (Virk, 1975, McComb and Rabie, 1982, Bird et al. 1987, Tiu et al., 1995, 1996,
Interthal and Wilski, 1989, Kulicke et al., 1989, Choi et al., 2000, Jovanovic, 2006). Few of them are dealing with viscous
solutions.
Despite the great number of studies, the underlying mechanisms of drag reduction are not yet clearly defined, although
recent numerical simulations are on the right track to close the gap (White and Mungal, 2008). It seems clear that drag
reduction effectiveness correlates directly with the coupling between the local elongational field of the turbulent flow and
extensional properties of the drag reducing agent (Lumley, 1977, De Gennes, 1990). As a simple picture, flexible polymer
chains have the ability to be extended by flow and thus develop an elongational viscosity that limits the extent of turbulence
by suppressing the most energy-dissipating small scale eddies (Durst 1982, Jovanovic, 2006). However, stretching of the
polymer chains can lead to chain rupture, which in turn decreases the drag reduction effect. Hence, mechanisms responsible
for polymer degradation by chain rupture and loss of drag reduction in turbulent flow have been widely studied (Merrill and
Horn, 1984, Tiu et al., 1996, Brostow 2008, Vanapalli et al., 2005, 2006, Islam et al., 2004, Elbing et al., 2009).
As predicted by De Gennes (1974), when the characteristic time of the extensional flow, 1/𝜀, is lower than the relaxation
time of the polymer chains, 𝜏, polymer coils will experience a sudden coil-stretch transition (Keller and Odell, 1985). In
diluted regime, this transition will thus occur at a critical strain rate, 𝜀! such that the Deborah number 𝐷𝑒 = 𝜀! 𝜏 ≈ 1.
By using Stoke’s law to estimate the friction exerted by the solvent on a stretch chain in an elongational flow field, Odell
and Keller (1986) found that the distribution of stresses on the macromolecules is parabolic with its maximum at the center.
The tension, 𝑇, exerted at the chain midpoint is proportional to the solvent viscosity,  𝜂, the extensional rate,  𝜀, and the square
of the chain length,  𝐿, such that 𝑇 ∝ 𝜂  𝜀𝐿! . Hence, if the time in the elongational field is sufficient, it will lead to the
complete stretching of the chain since the force increases with the extended length of the chain. As a result of the maximum
stress exerted at the midpoint, one expects chain scission to occur at the middle of the chain if the strain rate is sufficient to
break the C-C bond of the macromolecule backbone. The strain rate for fracture,  𝜀  ! , should scale with 1/𝐿! ∼ 1/𝑀 ! where
M is the molecular weight of the chain. This scaling was confirmed in stagnation point flows experiments (Odell and Keller,
1986) for which the residence time in the flow field is high enough for a complete extension of the chain. It was confirmed by
size exclusion chromatography that chain scission occurs at the midpoint. In transient elongational flows such as flow
through an abrupt contraction, the strain rate for fracture was found to scale with 1/𝑀 (Nguyen and Kausch, 1988). In this
type of flow, the residence time is small and the chains do not have sufficient time to fully extend. However, the bond rupture
is still a non random process and occurs preferentially at the chain midpoint. Recently, Vanapalli proposed a model of chain
degradation in turbulent flow based on the Kolmogorov cascade that unifies stagnation points and transient experiments. At
the Kolmogorov scale, the maximum drag force on a chain scales with Re3/2 M2/ln(M), implying that the critical strain rate
for fracture scales universally with lnM/(M-2). This model was supported by degradation results in turbulent pipe flow at
Reynolds number up to 105 by Elbing et al. (2009).
In the semi dilute regime, where chain overlapping exists, chains will be still extended but the network structure will
strongly affect the degradation mechanism (Cates, 1993). If the chains have sufficient time to disentangle, high molecular
weight chains will be preferentially degraded with a cleavage at their center (Clay and Koelling, 1997). On the other hand, if
the chains have no time to disentangle during the stretching process, the degradation should occur between the junction
points corresponding to a random process (Muller et al., 1992). In that case, the final state of the chain is no longer dependent
on the molecular weight but rather on the characteristic of the network such as the entanglement density c/c* (Dupas, 2012).
Whatever the type of flow and the concentration regime, there is a kinetic aspect to degradation, for which models have
been presented (Nguyen and Kausch, 1989, Brostow, 2008). Indeed, passing once through a degrading geometry is not
sufficient to reach the final molecular weight distribution of the polymer. The elongational flow field is not uniform. Hence,
at each passage, there is a probability of chain scission depending on the position of the coil in the flow field. High molecular
weight chains will be preferentially broken. In a given geometry, the first passage will have the biggest impact on the
degradation level. Degradation on the following passes will continue but to a smaller extent until a steady state is reached
where all the chains have been subjected to the strain rate and are no more breakable by the flow field. The kinetics in
SPE-169699-MS 3

turbulent conditions often associated with a loss of drag reducing effect have been investigated through several devices,
including an orifice (Clay and Koelling, 1997), cross-slot (Islam et al., 2004), pipe or capillary (Hunston and Zakin, 1980,
Rho et al., 1996, D’Almeida et al., 1997, Bucholz et al., 2004, Dupas et al., 2012), rotating disk apparatus (Choi, 2000), high
shear concentric cylinders viscometer (Yu et al., 1979, Hénaut et al., 2012) or microfluidic channel (Nghe et al., 2010)…
From this very non exhaustive review, we anticipated that the prediction of the drag reduction effect in a transport
pipeline, along with the conditions at which mechanical degradation would occur, was a hard task. Moreover, several studies
aiming to measure the drag reduction and/or the degradation in a given geometry reported degradation at the upstream
contraction, which dominated the subsequent degradation in the geometry (Islam, 2004, Vanapalli, 2005, Elbing 2009).
Particular attention had to be paid to the experimental set up, in particular the entrance geometry of the piping, the type of
pump and any singularities on the piping (sharp elbows, restrictions) that could perturb the flow. Moreover, chemical
degradation had to be prevented in particular that one due to the oxidation of Fe2+ with oxygen.

Definitions

A solution is flowing in a pipe of diameter D (in m) at a flow rate Q (in m3/s).

The average velocity V (in m/s) is given by:

!!
Equation 1 𝑉 = !!!  

The Reynolds number is defined by:

!"#
Equation 2 𝑅𝑒 = !
3
where ρ is the fluid density in (kg/m ), η is the viscosity of the fluid (in Pa.s).

For non-Newtonian fluids, a generalized Reynolds number is defined taking into account the power law viscosity behavior of
the polymer η = Kγ!!! where K is the consistency index (viscosity at a shear rate of 1s-1) and n is the flow behavior index
(<1 for shear thinning fluid such as HPAM).
!!!!! !!
Equation 3 𝑅𝑒! = !"!! !  
!!!! !( )
!"
 
The frictional pressure drop (in Pa) for a length L (in m) of pipe is:
!!"#! !
Equation 4 Δ𝑃 = !
where 𝑓 is the Fanning friction coefficient factor.

In turbulent flow, for smooth pipes, 𝑓 can be expressed by the Blasius correlation:
Equation 5 𝑓 = 0.079𝑅𝑒 !!.!"

For rough pipes, 𝑓 can be expressed by the Churchill’s correlation:

!
! !" !"
!!.!
Equation 6 𝑓=8 !"
+ 𝐴+𝐵
!.! !! !"
! !
with 𝐴 = 2.457𝑙𝑛 + 0.27
!" !

!"#!$ !"
and 𝐵=
!"

Drag reduction is defined as the ratio of reduction in the pressure drop using a drag reducer agent ∆P!"#$%&' to the pressure
drop of water ∆P!"#$% without any drag reducer at the same flow rate as given by:
∆!!"#$% !∆!!"#$%&'
Equation 7 𝐷𝑅(%) =
∆!!"#$%
×100  
4 SPE-169699-MS

The degradation of the polymer solution is calculated according to the formula:

!! !!!"#
Equation 8 𝐷𝑒𝑔(%) = ! ×100
! !!!!"

where 𝜂0 is the viscosity of the non degraded solution, 𝜂𝑑𝑒𝑔 is the viscosity of the degraded solution, 𝜂𝐻20 is the water
viscosity (0.63cP at 50°C, 1cP at 20°C).
Because of the shear thinning behavior of the polymer solution, the degradation value depends on the shear rate at which
viscosity is measured. For all the tests, degradation was calculated from viscosity measurements performed with a Brookfield
viscometer with ULA adapter at 60 rpm which is equivalent to a shear rate of 73 s-1.

Experimental

Small scale experiments (PERL TOTAL R&D center, Lacq, France): pipes with internal diameter between ½ and 1’’,
61 m long
Polymer tested and water composition
Flopaam 3630S (SNF Floerger, Andrezieux, France), a partially-hydrolyzed polyacrylamide (HPAM) with 30%
hydrolysis, Mw=18MDa in powder form, was used for all the experiments. It was dissolved in an aerated synthetic brine at
6g/L (4.7g/L NaCl, 0.75g/L CaCl2, 2H2O, 0.56g/L MgCl2,6H2O). A mother solution at 11000 ppm was first prepared by
incorporation of the polymer powder in a polymer slicing unit (PSU100, SNF Floerger). The maturation was then performed
by gentle stirring in a 500 L cylindrical tank with marine impellers. The mother solution was then diluted at the desired
concentration in a 4 m3 stirred tank. Concentration and viscosity were measured in order to check that the polymer was not
degraded compared to a polymer solution prepared in the lab.

General layout
The solution was pumped with a volumetric membrane high pressure pump through a 2’’ 316L stainless steel pipe.
Pressure and viscosity measurements were then performed on a 61 meter-long polyester reinforced PVC tubing with
intermediate pressure taps and sampling points. Four different PVC tubings (TRESS-NOBEL blue 40bar Tricoflex) were
tested with internal diameters given by the supplier 1/2, 3/5, 3/4 and 1’’ (12.5, 16, 19 and 25 mm).
At the entrance and the exit of the PVC tubing, it was necessary to do the connection with the 2’’ stainless steel pipe. In
order to limit the degradation at the loop entrance (Kulicke et al., 1989, Vanapalli, 2005, Elbing 2009), a “tapered”
contraction was machined by welding several cones (10° angle) in series (cf. Figure 1, left). However, this precaution was
inefficient since high degradation was measured at the entrance of the ½ and 3/5’’ tubings. Another identical piece was
installed at the exit of the loop in order to have a tapered enlargement that did not disturb the flow upstream the enlargement.
Obstacles to the flow, such as edges, sudden deviation or internal diameter variations, being a cause of chain breakage
and local pressure drop increase, specific cross fittings (Figure 1, right) were machined in order to have no internal diameter
variation relative to the internal diameter of the tubing at the locations of intermediate pressure taps and sampling points.

Sampling port

Pressure port

Figure 1: (Left) Specific convergent/divergent at the entrance / exit of the small scale pilot, (Right) Specific cross fitting with no
internal diameter variation relative to the pipe diameter.

The configuration of the flow loop is schematized on Figure 2. Points A, B, C and D are the locations of the cross fittings
for pressure measurements and solution sampling. They were spaced as follows: LA=0m (tubing entrance, the exit of the
tapered contraction corresponds to the beginning of the tubing), LB=1m, LC=1+10=11m, LD=11+50=61m.
Differential pressure drops were measured on each section as follows: first section of 1m DPAB=PA-PB, second section of
10m DPBC=PB-PC, third section of 50 m DPCD=PC-PD, and whole length of 61m DPAD=PA-PD.
High precision pressure transducers (Rosemount 3051S) in the range 0-130 bar, 0-20 bar and 0-2.5 bar were used
depending on the value of the measured pressure.
SPE-169699-MS 5

Figure 2: Drawing of the flow loop configuration. Points A, B, C and D are cross fittings on which pressure is measured and fluid is
sampled.

Pilot scale experiments (RMOTC flow loop, Casper, Wyoming, USA): pipe with internal diameter of 6’’, 7 km long
Polymer tested and water composition
For the field test, an uncertainty existed regarding the dissolved oxygen content in the available water. In case of
corrosion of the pipe due to salt and oxygen, the oxidation of Fe(II) is known to produce an intermediate hydroxyl radical that
attacks and disrupts the polymer chains (Levitt et al., 2011) which in turn, drastically reduces viscosity. Moreover, presence
of trivalent cations such as Fe(III) (dissolved in solution or precipitated in the hydroxide form) can reticulate the polymer
chains and thus suppress the drag reducing effect (Kulicke et al., 1989).
A lot of precautions were taken in order to prevent polymer oxidative degradation and interaction with rust. Before the
tests, the pipe was carefully cleaned with a rubber pig in order to remove residues, sediments and piece of rusts. Sodium
bicarbonate was added to the water to limit the corrosion and thus the polymer degradation (Levitt et al., 2011). It was
decided to use DL363 (SNF Floerger) in powder, which is a formulation including 3630S and a protection package (5% by
mass of oxygen scavenger and sacrificial agents) against oxidative degradation. Water was deoxygenated with Montbrite
1240 (Montgomerry Chemicals) and sodium bisulfite because of handling issues with the use of sodium dithionite as an
oxygen scavenger. Montbrite is a mix of sodium borohydride with sodium hydroxide whose reaction with sodium bisulfite
produces dithionite in solution. The mix was added to the water at the exit of the storage tank before the addition of the
polymer.
Deoxygenated water at 3 g/L (0.53g/L NaCl, 1.3g/L Na2SO4, 1g/L CaCl2, 2H2O, 0.25g/L MgCl2,6H2O), 90°C was
pumped from an aquifer near the facilities and stored in a 20000 bbl tank equipped with a floating roof in order to limit
exposure to oxygen in the atmosphere. However, it revealed to be ineffective and oxygen scavenger was needed. Water was
recirculated continuously in an air cooler to regulate the temperature to approximately 50°C.
A mother solution at 7000 ppm was prepared with a mobile polymer dissolution unit PU23 (TIORCO/FABTECH as seen
on Figure 3), and stored in 400 bbl (60m3) tanks. This unit includes a rotor/stator for powder incorporation and two batch
maturation tanks that work in alternating mode. The capacity of this unit was 48m3/d (310bbl/d) of mother solution at
7000 ppm, which was equivalent to 420m3/d (2700 bbl/d) of diluted polymer solution at 800ppm. The maximum flow rate
tested on the loop was 240m3/h (40000 bbl/d).

Figure 3: Picture of the PU23 unit from TIORCO/FABTECH to prepare the polymer mother solution.

General layout
The mother solution was pumped with a properly designed HP plunger piston triplex pump for in-line dilution with water.
In-line mixing was performed through a 8’’ KMV mixer from KENICS, oversized compared with the internal pipe diameter
of 6’’ in order to prevent the polymer solution from any mechanical degradation before its entrance in the flow loop. The pipe
6 SPE-169699-MS

connection between the static mixer and the loop inlet was with an internal diameter equal to that one of the RMOTC pipe
(6’’). In order to ensure a good mixing before entering in the loop, several elbows were installed on this portion in order to
increase the length of the upstream pipe to approximately 100 meters as seen on Figure 4. Numerous sampling points were
located on top of the pipe on this portion in order to ensure that mechanical degradation did not occur on the first 100 meters
downstream of the static mixer.

Figure 4: Pipe connection between the static mixer and the loop inlet.

Several sampling points (1’’ lines) were installed all along the loop in order to measure the viscosity of the polymer
solution as a function of the distance. Low shear sampling devices with portable capacities were used along with 8 high
pressure online viscometers (TOTAL intern development).
Pressure drops on each section of the loop were measured with absolute pressure transducers located at L=0m (P1, loop
inlet), L=1320 m (P2), L=3459 m (P3) and L=7076 m (P4, loop exit) as seen on Figure 5. Differential pressure drops were
calculated on each section as follows: first section of 1320 m DP12=P1-P2, second section of 2139 m DP23=P2-P3 and third
section of 3617 m DP34=P3-P4.

Figure 5: General layout of the pilot test with the location of the pressure measurements P1, P2, P3 and P4 all along the loop. P0 is a
sampling point 20 meters downstream the static mixer, at approximately 100 meters from the loop inlet P1.

At the exit of the loop, the solution was stored in discharge tanks. The solution was circulating through an aeration pump
for complete reoxygenation before discharging it into a 3-stages open pit.

Investigated pipe diameters, flow rates and velocities

For both test campaigns, there were some limitations on the maximum flow velocity because of constraints on maximum
allowing pressure and maximum flow rate delivered by the pumps.
On the small scale tests, the maximum flowrate and pressure limits were 7 m3/h and 40 bars, respectively. The diameters
of the tubing were determined by fitting the water pressure drops on the last section of 50 m (CD portion on Figure 2) with
Blasius correlation (Equation 5).
SPE-169699-MS 7

On the pilot tests, the maximum flowrate and pressure limits were 240 m3/h and 100 bars, respectively. The linear
pressure drop was the same on the three sections of the loop. Churchill’s correlation (Equation 6) was used to determine the
pipe diameter assuming a roughness of 100 microns.
Determined diameters, maximum velocities and water Reynolds numbers are summarized for each pipe in Table 1.
Pipe  Type   Pipe  ID   Flow  rate  max   Velocity  max   Re  water  
3
  mm   inch   m /h   m/s  
PVC  tubing  
 
12.5   0.49   4   9.1   113177  
PVC  tubing 14   0.55   5   9   126313  
PVC  tubing 18.9   0.74   7   6.9   130992  
PVC  tubing 25   0.98   7   4   99030  
Carbon  steel   152   5.98   242   3.7   561438  

Table 1: Investigated pipe diameters, flow rates, velocities and water Reynolds numbers during the two flow tests campaigns.

Results

Drag reduction versus pipe length, pipe diameter, fluid velocity and polymer concentration
Small scale experiments
There was no drag reduction on the first two sections (AB and BC portion of 1 m and 10 m respectively) of the flow loop.
This implies that some distance was neccesary to establish steady state flow. An additional experiment would have consisted
in increasing the length of the loop in order to check if the steady state flow is fully established after 11 m. Nevertheless, we
considered that the pressure measurements on the third section of 50 m were corresponding to steady state flow
measurements.
Pressure measurements on this third section and corresponding drag reduction percentage are presented for each polymer
concentration tested on the ¾’’ tubing on Figure 6. Results on other tubings are not presented there for the sake of not
surcharging this paper.
Depending on the concentration, velocity and pipe diameter, the drag reduction varies between 35 and 65%. The drag
reduction is increasing with the velocity and decreasing with the polymer concentration. The concentration effect vanishes at
high velocities. These observations and trends were the same for the 4 tubing tested (0.49, 0.55, 0.74 and 0.98’’). The
negative polymer dependence effect is contrary to what is observed on diluted solutions (Kulicke et al., 1989). Indeed, for
diluted solutions, the viscosity is very close to that for water. Hence, Reynolds numbers are identical for water and polymer
solutions. For viscous solutions, the Reynolds number decreases as a function of the viscosity. Hence, turbulence and drag
reduction are delayed because of the viscosity of the solution.

Third  sec@on  of  50m,  pipe  ID=3/4''   Third  sec@on  of  50m,  pipe  ID=3/4''  
14   80  
Blasius  fit  
12   Water   70  
Pressure  drop  (bar)  

Drag  reduc@on  (%)  

10   1200  ppm   60  
900  ppm  
50  
8   600  ppm  
100  ppm   40  
6   100  ppm  
30  
4   600  ppm  
20  
900  ppm  
2   10   1200  ppm  
0   0  
0   1   2   3   4   5   6   7   8   0   1   2   3   4   5   6   7   8  
Flow  rate  (m3/h)   Velocity  (m/s)  
Figure 6: Small scale experiments with a ¾’’ tubing. (Left) Pressure drops of water and polymer solutions on the third section of
50m as a function of the flow rate. The continuous line represents the fit of water pressure drops obtained with Blasius correlation.
(Right) Calculated drag reductions for each polymer concentration as a function of the fluid velocity in the tubing.

Pilot scale experiments

Pressure measurements on the first section of the loop (L=1320 m as seen on Figure 5) and corresponding drag reduction
percentages are presented for each polymer concentration tested in Figure 7.
8 SPE-169699-MS

Depending on the velocity, drag reduction percentage varies between 30 and 80% whatever the polymer concentration in
the range of 50-2000 ppm. Drag reduction is still increasing with velocity. The negative concentration effect is still apparent.
Indeed, at low velocity, drag reduction decreases with increasing polymer concentration.

First  sec@on,  pipe  ID=6''   First  sec@on,  pipe  ID=6''  

10   100  
Churchill  fit  
9   90  
Linear  Pressure  drop  (bar)  

Water    
8   80  

Drag  reduc@on  (%)  

50ppm  
7   100ppm   70  
6   300ppm   60  
800ppm   50ppm  
5   1200ppm   50  
100ppm  
4   2000ppm   40   300ppm  
3   30   800ppm  
2   20   1200ppm  
1   10   2000ppm  
0   0  
0   50   100   150   200   250   0   1   2   3   4  
Flow  rate  (m3/h)   Velocity  (m/s)  
Figure 7: Pilot scale experiments on the 6’’ internal diameter pipe. (Left) Pressure drop of water and polymer solutions on the first
section of the loop as a function of the flow rate. The continuous line represents the fit of water pressure drops obtained with
Churchill’s correlation. (Right) Calculated drag reductions for each polymer concentration as a function of the fluid velocity in the
pipe.

As seen on Figure 8, pressure measurements and drag reduction percentages were the same on the 3 sections of the loop.
This result, especially that one at 50 ppm, is a first indication that mechanical degradation did not occur during the transport
in turbulent flow. Indeed, at low concentration, polymer degradation should lead to a loss of drag reduction.

50ppm,  pipe  ID=6''   300ppm,  pipe  ID=6''   800ppm,  pipe  ID=6''  

100   100   100  
Drag  reduc@on  (%)  

80   80   80  
60   60   60  
40   SecDon  1   40   SecDon  1   40   SecDon  1  
SecDon  2   SecDon  2   SecDon  2  
20   20   20   SecDon  3  
SecDon  3   SecDon  3  
0   0   0  
0   1   2   3   4   0   1   2   3   4   0   1   2   3   4  
Velocity  (m/s)   Velocity  (m/s)   Velocity  (m/s)  

Figure 8 : Pilot scale experiments on the 6’’ internal diameter pipe. Drag reduction as a function of fluid velocity on each section of
the pipe. Each graph corresponds to the tests at a given polymer concentration (50, 300 and 800 ppm from left to right).

Mechanical degradation versus pipe length, pipe diameter, fluid velocity and polymer concentration
Small scale experiments
For each test, the stock solution was sampled in the tank where the diluted solution was stored in order to have the
viscosity of the non degraded solution. Then, solution was sampled while flowing at the entrance of the loop (sampling point
A, L=0m), after 1 meter (sampling point B), after 11 meters (sampling point C) and after 61 meters (sampling point D).
For each pipe, the degradation at each sampling location is plotted as a function of the fluid velocity on Figure 9. Only
results on the 1200 ppm polymer solution are presented.
- For the pipe ID=0.49’’, there is no degradation up to 2.3 m/s whatever the distance. For higher velocities, there is a
big degradation on the first meter and then, no further degradation over the next 60 meters except for the experiment
at 9 m/s for which there is a small additional degradation due to the transport in turbulent flow.
- For the pipe ID=0.55’’, the degradation at the entrance is quite low. It increases with the distance travelled over the
pipe. It means that there is an additional degradation due to the turbulent flow.
SPE-169699-MS 9

- For the pipe ID=0.74’’, there is a small degradation at the entrance especially at 6 m/s. It seems that there is a very
small additional degradation with the distance travelled over the pipe. However, measured degradations are very
small and their evaluation can be erroneous because of the sensitivity of the viscometer.
- For the pipe ID=0.98’’, there is no degradation whatever the distance at flow rates up to 4 m/s, which is the highest
flow rate that could be achieved with available pumps.

1200ppm,  pipe  ID=0.49''   1200ppm,  pipe  ID=0.55''  

45   45  
Degrada@on    /  ini@al  solu@on    (%)  

Degrada@on    /  ini@al  solu@on    (%)  

40   61m   40   61m  
35   11m   35   11m  
1m   1m  
30   30  
Entrance   Entrance  
25   25  
20   20  
15   15  
10   10  
5   5  
0   0  
0   1   2   3   4   5   6   7   8   9   10   0   1   2   3   4   5   6   7   8   9   10  
Velocity  (m/s)   Velocity  (m/s)  
1200ppm,  pipe  ID=0.74''   1200ppm,  pipe  ID=0.98''  
45   45  
Degrada@on    /  ini@al  solu@on    (%)  

Degrada@on    /  ini@al  solu@on    (%)  

40   61m   40   61m  
35   11m   35   11m  
1m   1m  
30   30  
Entrance   Entrance  
25   25  
20   20  
15   15  
10   10  
5   5  
0   0  
0   1   2   3   4   5   6   7   8   0   1   2   3   4   5  
Velocity  (m/s)   Velocity  (m/s)  
Figure 9 : Small scale experiments with 1200ppm polymer solution. Each graph represents the degradation of the sampled solution
at each sampling location as a function of the fluid velocity in the tubing. The degradation is related to the viscosity of the stock
-1
solution. Viscosity measurements were performed at 73s .

Pilot scale experiments

For each test, the polymer solution was diluted in line through a static mixer. Because of the variability on the mother
solution concentration and the impossibility to measure the concentration of each sampled solution, it was not possible to
compare with a reference solution from the lab. Thus, the viscosity of the solution 20 meters downstream the static mixer was
considered as the non-degraded reference solution. This sampling point was located at approximately 100 meters from the
loop entrance (as seen in Figure 5).
If mechanical degradation occurs under turbulent flow, it should increase with fluid velocity. For this reason, we only
present the results at the highest tested flow rate (242 m3/h), which corresponds to a velocity of 3.7 m/s. Viscosity as a
function of the distance travelled over the pipe at 240 m3/h is plotted for each polymer concentration in Figure 10. Whatever
the concentration, the viscosity is stable throughout the whole loop. Results were similar at lower flow rates. As a conclusion,
there was no degradation for velocities equal to or lower than 3.7 m/s in the 6’’ pipe whatever the polymer concentration in
the range 300-2000 ppm. For lower polymer concentrations (50 and 100 ppm), there was no confidence in the viscosity
measurements because of the poor sensitivity of the viscometer used.
10 SPE-169699-MS

240m3/h,  pipe  ID=6''  

downstream  the  sta@c  mixer    (%)  

12   2000ppm   1200ppm   800ppm   300ppm  

Degrada@on    /  viscosity  
10  
8  
6  
4  
2  
0  
-­‐200   500   1200  1900  2600  3300  4000  4700  5400  6100  6800  7500  
Distance  (m)  
Figure 10 : Pilot scale experiments on the 6’’ pipe. Viscosity as a function of distance travelled over the loop for each polymer
-1
concentration. Viscosity measurements were performed at 73 s at the same temperature as the flow experiment (~50°C).

Discussion

On the possibility to predict the drag reduction on larger pipes

Drag reduction data are plotted for both test campaigns as a function of the generalized Reynolds number ReG (Equation
3) in Figure 11. A master curve is not obtained but there is a general trend. Drag reduction in the 6’’ pipe is the highest
whatever the generalized Reynolds number. An envelope can be defined that will fix the minimum and maximum drag
reduction expected at a given generalized Reynolds number.

100  
90  
Drag  reduc@on  (%)  

80  
70  
60  
50   ID=6''  
40   ID=1"  
30   ID=0.74''  
20   ID=0.55''  
10   ID=0.49''  
0  
0   100000   200000   300000   400000   500000  
Generalized  Reynolds  number  ReG  
Figure 11 : Drag reduction of all the experiments on small pipes (1/2-1’’) and 6’’ pipe as a function of the generalized Reynolds
number. Black lines define an envelope for predicting drag reduction on larger pipes.

For larger pipes, the anticipated generalized Reynolds number is calculated in Figure 12 as a function of the velocity and
the pipe diameter. The fluid is assumed to be a solution at 800ppm whose viscosity obeys the power law behavior η(in  cP) =
10γ!!.!! . Drag reduction is then determined from the envelope in Figure 11.
ð For pipes with diameter varying between 14 and 22’’ at a velocity of 1 m/s, the generalized Reynolds number is
between 70000 and 90000. Being conservative, we anticipate the drag reduction to be between 45 and 80%.
ð For velocity higher than 1m/s, the generalized Reynolds number will be higher than 100000, a drag reduction superior
to 55% is anticipated.
SPE-169699-MS 11

600000  
22''  
500000   20''  
18''  
400000   16''  
14''  
ReG   300000  

200000  

100000  

0  
0   1   2   3   4  
Fluid  velocity  in  the  pipe  (m/s)  
Figure 12: Generalized Reynolds number as a function of the fluid velocity for different pipe diameters. Calculation made with a
polymer solution at 800 ppm.

A more refined model was developed on the basis of Virk’s phenomenology (Virk, 1975) by GEMICO laboratory
(LRGP, ENSIC, Lorraine University, France). Predictions are compared with measured drag reductions for both test
campaigns on Figure 13. Predictions are within an error of +/- 15% for drag reductions lower than 60%, within an error of +/-
10% for drag reductions higher than 60%.

100  
DR  experimental  (%)  

80  

60  
ID=0.49''  
ID=0.55''  
40   ID=0.75''  
ID=1''  
20   ID=6''  
+/-­‐15%  
+/-­‐10%  
0  
0   20   40   60   80   100  
DR  model  (%)  
Figure 13: Comparison of the measured drag reduction level as a function of the predicted drag reduction for both test campaigns.
Dash lines are error curves that help define the uncertainty of the model.

Based on this model (not shown on Figure 13), the drag reduction is predicted to be even higher than 55% for velocities
higher than 1m/s on pipes with an internal diameter higher than 14’’.

Mechanical degradation
Vanapalli (2006) formulated a simple scaling theory for chain scission in turbulent flow. He considered the interaction of
extended polymer chains with turbulent fluctuations from the scaling perspective of the Kolmogorov cascade. He obtained
the expression of the drag force experienced by the extended polymer chains:
!!! !"!/! !!
Equation 9 F!"# = A
!!!! !"(! !)

where A is a constant, a is the diameter of the chain, η is the viscosity, L is the length of the chain, D is the pipe diameter.
The force exerted on the chain is scaling with  Re!/! D! . From this equation, one can calculate the critical velocity Vc at
which mechanical degradation will occur for a given pipe diameter. It is found that the critical velocity will increase with the
pipe diameter (our calculations for an HPAM Mw=18 MDa gave Vc=3 m/s for a ID=1/2’’ pipe, Vc=5 m/s for a ID=1’’ pipe).
12 SPE-169699-MS

From these results, the following strategy was defined: if a trend on critical velocities was obtained from the small scale
experiments on ½-1’’ pipes, it would be thus possible to extrapolate on bigger diameter pipes.

In the small scale experiments, an entrance degradation was experienced for the pipes with ID=0.49 and 0.55’’.
Depedending on the experiment, this entrance degradation occurred at the exit of the tapered contraction as would be
anticipated but also after 1 meter. The degradation after one meter is unexpected. From our experience and the analysis that
will follow, the degradation in turbulent flow is progressive and not brutal. In other words, it takes some time to reach a high
degradation. For these reasons, we thought that the polymer was elongated in the tapered contraction and then degraded by
the turbulent flow on the first meter because of the pre-elongation. In the absence of pre-elongation, the elongational rate of
the turbulent flow should not have been sufficient to break the chains. The degradation after 1 meter was thus considered as
an entrance effect due to the contraction.
Subsequently, for all experiments on each pipe diameter, the critical velocity at which mechanical degradation occurs due
to the turbulent flow, was determined. For this, only the experiments for which there was an additional degradation after the
first meter were used. In the 0.49’’ pipe, the critical velocity was found to be less than or equal to 2 m/s. By extrapolating the
results of the 0.55’’ pipe, the critical velocity was found to be 2 m/s. For the 0.74’’ pipe, we found 6 m/s with some reserve
(due to the small measured degradation). At last, for the 1’’ pipe, no degradation was observed at rates up to and including
4 m/s. The critical velocity was thus higher than 4 m/s.

As a conclusion, the trend is not perfect but the critical velocity, at which mechanical degradation occurred, clearly
increases with pipe diameter. From this trend, we anticipated that the critical velocity would be higher than 4 m/s on a pipe
with an internal diameter higher than 1’’. It was confirmed by the pilot test on the 6’’ pipe that mechanical degradation does
not occur at a velocity lower than or equal to 3.7 m/s. This result validated our approach. It is thus anticipated (quite
conservatively) that the critical velocity will be higher than 3.7 m/s on pipes with internal diameter equal to or greater
than 6’’.
Finally, for the experiments in which turbulence in the pipe led to polymer degradation, it seems that degradation evolves
linearly with the logarithm of the distance until reaching a plateau corresponding to the steady state degradation, as seen in
Figure 14. This result is coherent with numerous lab degradation experiments (publication pending) in which the same trend
was observed. The degradation varies linearly with the logarithm of the number of degrading events whatever the degrading
geometry. A parallel can be done with the correlation proposed by Henaut et al. (2012) for which the loss of drag reduction
(equivalent to an increasing degradation) in a rheometer varies with the logarithm of the dissipated energy. These results are
also very coherent with the universal scaling law from Vanapalli, which is quite insensitive to the flow geometry since it has
been validated on experiments with various degrading geometries, such as contraction, crosslot, and rotational turbulent
devices. All these results indicate that degradation will vary linearly with the logarithm of distance. Hence, if
degradation does not occur over a distance of 100 m, it can be anticipated that similarly, no degradation will occur
over a distance of 1000 m.

1200ppm,  pipe  ID=0.55''  

35  
Degrada@on    /  entrance    (%)  

30   V=9.1m/s  
V=6.8m/s  
25   V=4.5m/s  
V=2.3m/s  
20  
15  
10  
5  
0  
1   10   100  
Distance  (m)  

Figure 14 : Small scale experiment with 1200 ppm polymer solution on a 0.55’’ pipe. Evolution of the degradation with the distance
traveled over the pipe. The degradation is related to the viscosity at the entrance of the loop. Viscosity measurements were
-1
performed at 73 s .
SPE-169699-MS 13

Conclusions

A testing program was launched in 2012 in order to define design rules for polymer solution transportation. A first series
of tests was performed on 4 different pipes, each 61 m long with internal diameter (ID) varying between ½ and 1’’. A second
series of tests was performed at a pilot scale on a 7 km long, 6’’ diameter pipe.
For both test campaigns, polymer solutions at different concentrations were transported in turbulent conditions. The
strategy was to get trends on the small scale experiments (ID=0.5-1”, L=61 m) and to extrapolate to field conditions
(ID>14”), the extrapolation being validated through the large scale pilot tests (ID=6”, L= 7 km). Tests were carefully
prepared in order to limit the mechanical degradation at the entrance of the pipe and the chemical degradation during the
flow.

Drag reduction:
At all pipe diameters, a high level of drag reduction was measured (as high as 70-80% on a 6’’ pipe for velocities higher
than 2 m/s). This level increased with the velocity and decreased with polymer concentration. The concentration effect
vanished at high velocities. The negative concentration dependence effect is contrary to what is observed for diluted solutions
because of the more viscous character of the solutions tested.
A general trend was obtained when the drag reduction measurements of both test campaigns were plotted as a function of
the generalized Reynolds number. An envelope was defined to fix the minimum and maximum drag reduction expected for a
given velocity on larger pipes. For pipes with diameter varying between 14” and 22”, at a velocity higher than 1 m/s, the drag
reduction is anticipated to be between 45 and 80%. A refined model was developed to predict drag reduction. Predictions are
within an error of +/- 15% for drag reduction lower than 60%, within an error of +/- 10% for drag reductions higher than
60%.

Mechanical degradation
On the basis of Vanapalli’s model for degradation in turbulent flow (Vanapalli, 2006), the critical velocity at which
mechanical degradation occurs in turbulent flow increases with the pipe diameter. This trend was confirmed on small scale
diameter tests with pipe diameters varying between ½ and 1’’. From this trend, the critical velocity was anticipated to be
higher than 4 m/s on a pipe with an internal diameter higher than 1’’. This was confirmed by the pilot tests campaign on the
6’’ pipe for which mechanical degradation did not occur for velocities lower or equal to 3.7 m/s. This result validated our
approach. It is thus anticipated that the critical velocity will be higher than 3.7m/s (being very conservative) on pipes with
internal diameter higher than 6’’. At last, it was observed that the degradation in a pipe varies linearly with the logarithm of
the distance travelled over the pipe. As a consequence, there is no distance limit for transport of polymer solutions, provided
the critical velocity is not exceeded and other degradation mechanisms (for example, oxidative) are avoided.

ACKNOWLEDGMENTS
To all the people involved in these two test campaigns:
- TOTAL E&P Houston: A. Mitchell
- TOTAL E&P France: G. Salabert, V. Loustalot, J. Lafourcade, J.-L. Daudrix, J.-C. Aka, P. Fortane, S. Nowé, M.
Questel, G.Heurteux, M. Joly, N. Passade-Boupat, M. Bourrel, A. Goulois, O. Garnier, S. Venes, L. Badaire, E. Lasailly
- GEMICO laboratory, ENSIC, Lorraine University, Nancy, France: C. Lemaitre
- NALCO/TIORCO FABTECH: Josh Foy, PJ Whittle, Cole Hor, Jesus Martinez, Adam Simanton, Andrew Sinclair, Jim
Holder, Seth Tubbs, Robbie Marnell, Scott Lair, Zach Walsh, Ben Turner, Rob Kevwitch, Holly Selchert, Crystal
Samples, Christy Van Campen, James Fuqua

To TOTAL E&P management for the permission to publish this work.

References

Bird, R.B., Armstrong, R.C., Hassager, O., 1987, Dynamics of Polymeric Liquids, 2nd ed., vol.1:Fluid Mechanics, John
Wiley & Sons, NY.
Brostow, W., 2008, Drag reduction in flow: review of applications, mechanism and prediction, Journal of Industrial and
Engineering Chemistry (Amsterdam, Netherlands), 14(4), 409-416.
Buchholz, B. A., Zahn, J. M., Kenward, M., Slater, G. W., Barron, A. E., 2004, Flow-induced chain scission as a physical
route to narrowly distributed, high molar mass polymers, Polymer, 45(4), 1223-1234.
Cates, M., McLeish, T., Marrucci, G., 1993, The rheology of entangled polymers at very high shear rates, Europhysics
Letters, 21(4), 451-456.
Choi, Hyoung J.; Kim, Chul A.; Sohn, Jeong-In; Jhon, Myung S., 2000, An exponential decay function for polymer
degradation in turbulent drag reduction, Polymer Degradation and Stability, 69(3), 341-346.
14 SPE-169699-MS

Clay, J. D.; Koelling, K. W., 1997, Molecular degradation of concentrated polystyrene solutions in a fast transient
extensional flow, Polymer Engineering and Science, 37(5), 789-800.
D'almeida, A. R., Dias, M. L., 1997, Comparative study of shear degradation of carboxymethyl cellulose and poly(ethylene
oxide) in aqueous solution, Polymer Degradation and Stability, 56(3), 331-337.
De Gennes, P.G., 1990, Introduction to polymer dynamics, Cambridge University press, Cambridge.
De Gennes, P.G., 1974, Coil-­‐stretch transition of dilute flexible polymers under ultrahigh velocity gradients, Journal of
chemical physics, 60, 5030
Dupas, A., Henaut, I., Argillier, J.-F., Aubry, T., 2012, Mechanical degradation onset of polyethylene oxide used as a
hydrosoluble model polymer for enhanced oil recovery, Oil & Gas Science and Technology, 67(6), 931-940.
Elbing, B., R., Winkel, E.S., Solomon, M.J., Ceccio, S.L., 2009, Degradation of homogeneous polymer solutions in high
shear turbulent pipe flow, Exp Fluids, 47, 1033-1044
Durst, F., Haas, R., Interthal, W., 1982, Laminar and turbulent flows of dilute polymer solutions: a physical model,
Rheologica Acta, 21, 572-577.
Hénaut, I., Cassar, C., Gainville, M., Hamdi, K., Pagnier, P., Glénat, P., 2012, Mechanical Degradation Kinetics of Polymeric
DRAs, BHR Group.
Hunston, D. L.; Zakin, J. L., 1980, Flow-assisted degradation in dilute polystyrene solutions, Polymer Engineering and
Science, 20(7), 517-23.
Interthal, W., Wilski, H., 1985, Drag reduction experiments with very large pipes, Colloid and Polymer Science, 263(3), 217-
29.
Islam, Mohammad T.; Vanapalli, Siva A.; Solomon, Michael J., 2004, Inertial Effects on Polymer Chain Scission in Planar
Elongational Cross-Slot Flow, Macromolecules, 37(3), 1023-1030.
Keller, A.; Odell, J. A., 1985, The extensibility of macromolecules in solution; a new focus for macromolecular science,
Colloid and Polymer Science, 263(3), 181-201
Kulicke, W. M., Koetter, M., Graeger, H., 1989, Drag reduction phenomenon with special emphasis on homogeneous
polymer solutions, Advances in Polymer Science, 89(Polym. Charact./Polym. Solutions), 1-68.
Levitt, David B.; Slaughter, Will; Pope, Gary A.; Jouenne, Stephane, 2011, The effect of redox potential and metal solubility
on oxidative polymer degradation, SPE Reservoir Evaluation & Engineering, 14(3), 287-298.
Lumley, J.L., 1977, Drag reduction in two phase and polymer flows, The Physics of fluids, 20(10), S64-S71.
Jovanovic, J., Pashtrapanska, M., Frohnapfel, B., Durst, F., Koskinen, J., Koskinen, K., 2006, On the Mechanism
Responsible for Turbulent Drag Reduction by Dilute Addition of High Polymers: Theory, Experiments, Simulations, and
Predictions, Journal of Fluids Engineering, 128(1), 118-130
McComb, W.D., Rabie, L.H., 1982, Local drag reduction due to injection of polymer solutions into turbulent flow in a pipe,
AIChE J., 28, 547-565
Merrill, E. W., Horn, A. F., 1984, Scission of macromolecules in dilute solution: extensional and turbulent flows, Polymer
Communications, 25(5), 144-6
Muller, A. J., Odell, J. A., Carrington, S., 1992, Degradation of semidilute polymer solutions in elongational flows,
Polymer, 33(12), 2598-604.
Nghe, P., Tabeling, P., Ajdari, A., 2010, Flow induced polymer degradation probed by a high throughput microfluidic set-up,
Journal of Non-Newtonian Fluid Mechanics, 165(7-8), 313-322.
Nguyen, T. Q., Kausch, H. H., 1989, Kinetics of polymer degradation in transient elongational flow, Makromolekulare
Chemie, 190(6), 1389-406.
Nguyen, T. Q., Kausch, H. H., 1988, Chain scission in transient extensional flow kinetics and molecular weight dependence,
Journal of Non-Newtonian Fluid Mechanics, 30(2-3), 125-40.
Odell, J. A., Keller, A., 1986, Flow-induced chain fracture of isolated linear macromolecules in solution, Journal of Polymer
Science, Part B: Polymer Physics, 24(9), 1889-916.
Rho, T., Park, J., Kim, C., Yoon, H.-K., Suh, H.-S., 1996, Degradation of polyacrylamide in dilute solution, Polymer
Degradation and Stability, 51(3), 287-293.
Tiu, C., Moussa, T., Carreau, P. J., 1995, Steady and dynamic shear properties of nonaqueous drag-reducing polymer
solutions, Rheologica Acta, 34(6), 586-600.
Tiu, C., Moussa, T., Tam, K. C., 1996, Degradation in turbulent pipe flow, Plastics Engineering (New York), 31(Handbook
of Applied Polymer Processing Technology), 189-215.
Vanapalli, S. A., Ceccio, S. L., Solomon, M. J., 2006, Universal scaling for polymer chain scission in turbulence,
Proceedings of the National Academy of Sciences of the United States of America , 103(45), 16660-16665.
Vanapalli, S. A., Islam, M. T., Solomon, M. J., 2005, Scission-induced bounds on maximum polymer drag reduction in
turbulent flow, Physics of Fluids, 17(9), 095108/1-095108/11.
Virk, P. S., 1975, Drag reduction fundamentals, AIChE Journal, 21(4), 625-56.
White, C. M., Mungal, M.G., 2008, Mechanics and prediction of turbulent drag reduction with polymer additives, Annual
Review of fluid mechanics, 40, 235-256.
Yu, J. F. S., Zakin, J. L., Patterson, G. K., 1979, Mechanical degradation of high molecular weight polymers in dilute solution,
Journal of Applied Polymer Science, 23(8), 2493-512.