You are on page 1of 5

PSIG 1418

A conceptual framework for predicting the effectiveness of a drag reducing


agent in liquid pipelines
Frank Vejahati, Enbridge Inc.

Copyright 2014, Pipeline Simulation Interest Group solely by using DRA. Nowadays, different types of drag
This paper was prepared for presentation at the PSIG Annual Meeting held in Baltimore, reducing products are commercially available and pipeline
Maryland, 6 May – 9 May 2014.
operators benefit from DRA to increase the pipeline capacity
This paper was selected for presentation by the PSIG Board of Directors following review of
information contained in an abstract submitted by the author(s). The material, as presented,
and/or reduce the cost of utility by lowering the pressure drop,
does not necessarily reflect any position of the Pipeline Simulation Interest Group, its officers, globally. In some cases the drag reduction is so dramatic, that
or members. Papers presented at PSIG meetings are subject to publication review by Editorial
Committees of the Pipeline Simulation Interest Group. Electronic reproduction, distribution, or pipeline operators are able to shutdown selected pump stations
storage of any part of this paper for commercial purposes without the written consent of PSIG
is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
without any noticeable drop in the overall throughput.
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, Pipeline
Simulation Interest Group, P.O. Box 22625, Houston, TX 77227, U.S.A., fax 01-713-586-5955. The mechanism of polymeric drag reduction is not clearly
known. Recent studies using Laser Doppler Anemometry
(LDA) have shown that the so far widely accepted hypothesis
ABSTRACT of turbulence suppression by polymeric molecule is not
completely true. By far the most credible theory on drag
Different types of polymeric Drag-Reducing-Agents (DRA) reduction mechanism is deemed to be a change in the
have been used in liquid pipelines to overcome the capacity turbulence structure, instead of suppression. This has been
limitations and/or reduce the cost of utility by lowering the supported by observed increase in the stream-wise turbulence
pressure drop. Optimum usage of DRA depends on how intensity and decrease in the turbulence intensity normal to the
accurately pipeline operators can predict its drag reduction wall1.
efficiency. The most widely used model in pipeline industry
for predicting the drag reduction efficiency correlates drag Drag reducing agents are more effective on low viscosity
reduction as a function of DRA concentration alone. However, fluids under turbulent flow regimes. The effectiveness of drag
in reality drag reduction can be significantly different from the reduction is dependent on several factors, such as polymer
predicted one depending on the variations in operating molecule size and its conformation, crude properties,
conditions and flow geometry. molecular interactions between polymer and crude, flow rate,
This work is an attempt to provide a conceptual framework temperature, polymer concentration, turbulent intensity and
wherein the effects of operating conditions and flow geometry, flow geometry. The longer and more flexible the polymer
expressed by Reynolds number, and polymer concentration on molecules are the more effective they become. On the other
drag reduction are quantified. The proposed model has been hand, larger polymer molecules are more costly and more
validated over a wide range of operating conditions, DRA susceptible to degradation by chain scission2. The
concentrations and pipe diameters for two types of effectiveness of DRA declines with decreasing Reynolds
commercially available DRA fluids using field test data. number and increasing viscosity of the crude. The
Overall, the proposed model gives an excellent reduction in effectiveness is also very sensitive to concentration.
the variability of drag reduction as a function of respective Concentrations as low as 5-10 ppm are enough to result in a
regressor variables. noticeable drag reduction. However, the effectiveness of DRA
diminishes as the concentration is increased and above a
certain concentration very minimal impact is observed.
INTRODUCTION
It has been experimentally proven that there is a so-called
The concept of polymeric drag reducing fluids was introduced maximum drag reduction asymptote which is the lowest
in the oil industry as early as 1946; however, its commercial friction factor attainable at a given Reynolds number
application was not practiced till about 30 years later in the regardless of polymer/crude system3. However, in practice this
Trans-Alaska pipeline system where the capacity of the maximum drag reduction is less likely to be reached due to the
pipeline was increased from 1.4 to 2.2 million barrels per day multiplicity of factors affecting the polymer/crude system.
2 FRANK VEJAHATI PSIG 1418

Therefore, for the commercially available drag reducing Table 1 Crudes properties
agents, the efficiency of drag reduction falls somewhere Specific gravity @ Viscosity @ 60F
between the asymptotic limit and the untreated turbulent pipe Fluid DRA
60F (cP)
flow and primarily dependent on the polymer type. A concise
L1 0.669 0.3
review of existing literature on polymeric drag reduction and
factors affecting the effectiveness of DRA has been provided L2 0.692 0.56
by Toonder2. L3 0.810 3.9
L4 0.817 2.77
The dearth of a generic correlation which is capable of
explaining the drag reduction irrespective of polymer/crude L5 0.839 7
system has compelled the pipeline operators and pipeline L6 0.840 5.74
simulation software developers to employ a polymer/crude L7 0.841 4.07
specific model which expresses the drag reduction in terms of
DRA concentration alone4: L8 LDRA 0.850 9.75
L9 0.859 5.73
L10 0.861 6.33
L11 0.861 11.6
L12 0.865 10.4
Where DR is drag reduction, ppm is polymer concentration,
and A & B which are specific polymer/crude constants. L13 0.870 19.7
M1 0.871 7.63
However, several factors make the drag reduction predictions M2 0.878 14.4
by equation (1) less accurate including, variations in pipe
diameter in a single pipeline system, line rate variations, M3 0.884 12.1
seasonal and regional temperature fluctuations and crude H1 0.921 104
property variations in a batched pipeline. H2 0.922 158
H3 0.923 214
To this respect, the underlying objective of this work is to H4 0.926 273
develop a model where the effects of operating conditions and
H5 0.928 166
flow geometry expressed by Reynolds number and agent
concentration on drag reduction are taken into account using H6 0.930 267
the available data on existing pipeline systems. H7 0.931 238
H8 0.932 210
HDRA
DATA H9
H10
0.932
0.932
456
102
H11 0.935 252
Pipeline system
H12 0.935 349
Drag reduction data were collected on a wide range of pipe H13 0.936 345
diameters on existing pipelines ranging from 18”-36” NPS.
H14 0.936 277
The range of axial velocity and Reynolds number were 2.13-
18.63 ft/sec and , respectively. H15 0.937 318
H16 0.937 93
Crudes and drag reducing agents H17 0.939 280

A wide range of petroleum crudes were examined for the drag


reduction efficiency including 17 heavy crudes, 13 light Table 2 Drag reducing agent properties
crudes and 3 medium crudes. Table 1 shows the properties of Density Viscosity concentration range
crudes and the type of polymeric agent used. DRA type
[lbs/gal] [cP] [ppm(wt)]
LDRA 8.0 350 5 to 50
Two types of commercial polymeric drag reducing products HDRA 8.51 95 5 to 50
are typically used on existing pipelines for light/medium
(LDRA) and heavy crudes (HDRA). Their properties are listed
in Table 2.
PSIG 1418 A conceptual framework for predicting the effectiveness of drag reducing agent in liquid pipeline 3

FIELD TESTS the existing pipeline fitted to equation (3) as a function


concentration for two different Reynolds numbers on crude
M1. The ranges of concentration and Reynolds number are the
The drag reduction was determined by measuring pressure practical operation windows on the respective pipeline. This
drop across a selected section of pipe and it is defined as: Figure also provides the outstanding goodness of fit in terms
of coefficient of determination (R2) and averaged absolute
errors defined as:

| |

Where is the pressure loss per unit length of the pipe for
crudes without drag reducing agent, and is the same
quantity for crudes seeded with polymeric DRA. Figure 1 also clearly show the significance of Reynolds
number in predicting the drag reduction and furthermore
The selection of sections for drag reduction study was based indicates to what extent Equation (1) could underestimate the
on the proximity and distribution of pressure transmitters and drag reduction based on the day to day operation of the
shear points. The selected sections were long enough to ensure existing pipeline.
that the pressure drop measurements were statistically
significant compared to the uncertainties introduced by
Re= 9.19E+04 Re= 2.08E+05
instrumentation, and the SCADA system. On the other hand, 80
the selected sections were short enough to ensure polymer
degradation would not develop significantly. Also, sections
were located where no shear points (e.g. Tees, and flow 60
restricting devices) were available to minimize polymer chain
scission.
DR%
40
SP.GR @ 60F: 0.871
Viscosity @ 60F: 7.63 cp
The Model
AAE %: 2.06
20 R2: 0.99
A series of preliminary tests revealed that in addition to
concentration, some other factors also have significant effect
on drag reduction, such as density, flowrate, viscosity and pipe 0
diameter. Obviously, designing experiments to study all 0 20 40 60
DRA Concentration (ppm)
individual factors was not practical. So, using Buckingham
theorem the factors were reduced to Reynolds number and
concentration per se. It can be easily shown that for a total 6 Figure 1 Drag reduction performance curve for crude M1 and
variables (i.e. DR, density, viscosity, flowrate, pipe diameter polymer LDRA
and dimensionless concentration) with 3 repeating variables
(i.e. density, viscosity and pipe diameter) the three terms Similar trends were observed for other types of crudes and the
become: DRA polymers. Table 3 summarizes the parameter estimations
and the goodness of fit for all the crudes and two drag
reducing agents. Overall, Equation (3) gives an excellent
reduction in the variability of drag reduction as a function of
Reynolds number and polymer concentration. A closer look at
Table 3 reveals that drag reduction is strongly dependent on
the crude/polymer system. The sensitivity of drag reduction to
Where Reynolds number even for a given drag reducing agent is
clearly a function of crude type. Figure 2 illustrates this fact
Different forms of functions were considered for for crudes L12 and M3 on a same drag reducing agent, LDRA.
and the best reduction in variability of as a function of
Reynolds number and polymer concentration was obtained by One of the implications of equation (3) is that the drag
following correlation: reduction increases with increasing Reynolds number;
however, similar to concentration, above a certain limit very
minimal gain in drag reduction is observed with increasing
Reynolds number. This has been illustrated in Figure 3 for
crude L1.
Figure 1 show the typical DRA performance measurements on
4 FRANK VEJAHATI PSIG 1418

Table 3 Parameter estimation and goodness of fit 80


2 70
L12
Fluid DRA n a b AAD% R
L1 0.028 0.022 0.085 0.5 0.98 60
L2 0.131 0.081 0.335 0.83 0.98 50

DR%
L3 0.024 0.016 0.088 1.76 0.98 40
L4 0.133 0.060 0.437 1.33 0.97 30
20 Re= 4.36E+04
L5 0.154 0.080 0.552 1.6 0.97
Re= 2.87E+05
L6 0.066 0.031 0.266 2.29 0.97 10
L7 0.191 0.107 1.248 2.58 0.96 0
0 10 20 30 40 50 60
L8 LDRA 0.079 0.035 0.364 2.62 0.96
L9 0.166 0.086 0.796 1.61 0.98
L10 0.169 0.089 0.819 2.05 0.97
80
L11 0.087 0.037 0.451 2.75 0.97 M3
70
L12 0.222 0.135 2.211 2.65 0.97
60
L13 0.106 0.044 0.677 3.23 0.97
50
M1 0.172 0.090 0.874 2.06 0.98

DR%
40
M2 0.125 0.052 0.823 3.74 0.97
30
M3 0.194 0.109 1.333 1.88 0.98 Re= 7.74E+04
20
H1 0.155 0.046 2.803 6.71 0.96
Re= 1.30E+05
H2 0.141 0.042 2.461 6.4 0.97 10
H3 0.215 0.085 5.029 6.75 0.96 0
H4 0.159 0.049 2.900 6.26 0.96 0 10 20 30 40 50 60
H5 0.153 0.045 2.678 7.5 0.96 DRA Concentration (ppm)
H6 0.210 0.076 4.666 7.46 0.96
H7 0.255 0.111 7.304 9.87 0.95 Figure 2 Effect of crude properties on drag reduction for LDRA
H8 0.178 0.058 3.569 7.5 0.96 drag reducing agent
HDRA
H9 0.234 0.104 6.111 6.29 0.96
H10 0.238 0.111 6.520 5.52 0.96
H11 0.148 0.044 2.592 6.66 0.96 70
L1
H12 0.244 0.090 7.027 8.94 0.95 60
H13 0.241 0.087 6.989 8.9 0.95
H14 0.150 0.040 2.780 7.46 0.96 50

H15 0.152 0.047 1.261 6.46 0.97 40


DR%

H16 0.212 0.098 3.790 2.78 0.98


30
H17 0.187 0.071 1.697 3.44 0.98
Re= 1.07E+06
20
Re= 2.24E+06
10

0
0 10 20 30 40 50 60
DRA Concentration (ppm)

Figure 3 Diminishing DRA effectiveness with increasing


Reynolds number
PSIG 1418 A conceptual framework for predicting the effectiveness of drag reducing agent in liquid pipeline 5

CONCLUSIONS AUTHOR BIOGRAPHY


Accurate prediction of drag reduction resulted from the Frank Vejahati, PhD, P.Eng. is a leak detection specialist at
application of polymeric flow improvers over the whole range Enbridge Inc. He holds a doctoral degree in Chemical
of pipeline operating window is of prime importance to both Engineering from University of Alberta, Canada. Over the
operators and hydraulic models. The most widely used model past 15+ years he has been acting on several positions. He
for predicting the drag reduction in the context of pipeline started his career as a process design engineer in
industry overlooks the effects of operating conditions, flow petrochemical industry, followed by a few years of
geometry and crude properties by simply correlating drag postdoctoral fellowship in the area of computational fluid
reduction to polymer concentration alone. The available field dynamic (CFD) simulation of multiphase flow related to clean
data corroborates that ignoring the effect of aforementioned energy technology development. For the past two years, he has
factors could potentially result in a sizable underestimation or been working in the area of real time transient hydraulic
overestimation of drag reduction efficiency. To this respect, a modeling of liquid pipelines and its application on leak
simple correlation has been developed that includes the effect detection systems.
of all discounted factors in terms of Reynolds number and
polymer concentration. The proposed correlation has been
validated over a wide range of operating conditions, pipe
diameters and crude properties. Overall, an excellent reduction
in the variability of drag reduction as a function of polymer
concentration and Reynolds was achieved using the proposed
model.

REFERENCES
1. J. M.J. den Toonder (1996) “Drag reduction by polymer
additives in turbulent pipe flow: Laboratory and
numerical experiment,” Thesis Technische Universiteit
Delft, 1995.
2. Horn, A.F. & Merrill, E.W. (1984) “Midpoint scission of
macromolecules in dilute solution in turbulent flow,”
Nature, 312, 140-141.
3. Virk, P.S., Merrill, E.W., Mickley, H.S., Smith, K.A., &
Mollo-Christensen, E.L. 1967. The Toms-phenomenon -
turbulent pipe flow of dilute polymer solutions. J. Fluid
Mech., 30, 305.
4. Burger, E.D., Munk, W.R., Wahl, H.A. (1982) “Flow
Increase in the Trans-Alaska Pipeline through Use of a
polymeric Drag-Reducing Additive”. Journal of
Petroleum Technology, February 1982.

NOMENCLATURE
SCADA – supervisory control and data acquisition
DRA – drag reducing agent
Re – Reynolds number

ACKNOWLEDGEMENTS
The author would like to thank Ron Threlfall, Ray
Philipenko, Nitesh Goyal, Noorallah Rostamy and other
colleagues in Pipeline Control System and Leak Detection
(PCSLD) Department of Enbridge Pipeline Inc. for the
support and insightful comments.

You might also like