Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

D. A.

Storm Drag Reduction in Heavy Oil


R. J. McKeon Transporting heavy crude oil by pipeline requires special facilities because the viscosity
is so high at normal field temperatures. In some cases the oil is heated with special heaters
along the way, while in others the oil may be diluted by as much as 30 percent with
H. L. McKinzie kerosene. Commercial drag reducers have not been found to be effective because the
single-phase flow is usually laminar to only slightly turbulent. In this work we show the
effective viscosity of heavy oils in pipeline flow can be reduced by a factor of 3-4. It is
C. L Redus hypothesized that a liquid crystal microstructure can be formed so that thick oil layers slip
on thin water layers in the stress fleld generated by pipeline flow. Experiments in a 1 {-in.
Texaco Inc., flow loop with Kern River crude oil and a Venezuela crude oil BCF13 are comi.<ilent with
Exploration & Production Department, this hypothecs. The effect has also been demonstrated under field conditions in a 6-in.
3901 Briarpark, flow loop using a mixture of North Sea and Mississippi heavy crude oils containing 10
Houston, TX 77042-5301 percent brine.

Introduction it is difficult to restart the core annular flow after an emergency


shut-down.
Crude oils are usually transported from the producing field to a
Chemically heavy oil differs from light oil by containing a
market by pipeline. Heavy crude oils with an API gravity less than
group of molecules called the asphaltenes. The asphaltenes are
15° require special facilities. Their viscosities are too high to pump
really a mixture of a great many molecules. They are usually
at ambient temperatures. For example, the viscosity of Kern River
defined as those molecules that precipitate on addition of heptane.
crude oil is 9-10 Pa.s at room temperature, while 0.1 Pa.s is a
Although the asphaltenes are only a useful pseudo-component,
desired pipeline viscosity. It is being transported from the field in
viscosity correlates strongly with their concentration as if they
Bakersfleld, California to the San Francisco market in a heated
were a true component of the oil (Storm, 1991, 1993a, 199.5,
pipeline. Special heaters are used along the way to maintain the
1996).
temperature in the range of 130-160°F. Nevertheless, the pressure
At least part of the asphaltenes exist in heavy oil as colloidal
drop is still approximately 1000 psi over 40 mi.
particles (Storm, 1992). It is basically interactions between these
A more extreme case is crude oil from the Hamaca region in colloidal particles that causes higher viscosities (Storm, 1991,
Venezuela. It has a viscosity of approximately 400 Pa.s at room 1993a, 1996). Evidence also exists that supports the hypothesis
temperature. It must be moved approximately 100 mi from the that the asphaltenic colloidal particles are actually micelles which
Hamaca field to the Jose upgrader on the coast. One plan is to form in the oil by self-association at a fairly sharp asphaltene
dilute it by 30 percent with kerosene. concentration (Storm, 1995). It has been known for some time that
Not only are there extra capital costs for heaters, extra pumps, asphaltenes behave as surfactants at oil-water interfaces (Strassner,
and larger-diameter pipelines when producing heavy crude oils, 1968).
but operating costs are obviously higher. In the case of Hamaca, The presence of asphaltenic surfactants which organize into
the cutter will be separated from the oil at the upgrader, and micelles under certain conditions implies heavy oil is a complex
returned to the field in a second pipeline. These higher costs are a fluid, or a fluid that has a microstructure. It is well known that
significant consideration when new developments of heavy oil complex fluids such as shampoo have unusual rheological prop-
resources are being contemplated. erties (Herb et al., 1994). A shampoo must be thick and viscous in
Commercial drag reducers have been used successfully to re- the container, but flow through a small opening when the container
duce pumping costs in the Trans Alaska Pipeline (Burger et al., is squeezed. This is accomplished by creating a network of entan-
1980). However, these drag reducers are not effective in heavy oil gled cylindrical micelles which will break under the stresses gen-
pipelines. Drag reducers are thought to dampen turbulent fluctua- erated when the shampoo is squeezed out through the small open-
tions in the near wall region (Achia and Thompson, 1977), and ing.
since the flow in heavy oil pipelines is laminar, or only slightly The shampoo system is water based, as are most complex fluids
turbulent, at the fluid velocities that can be obtained practically, the that have received serious study over the years. Perhaps more
drag-reducing chemicals would not be expected to be effective. relevant to the case of heavy oil is the report by Roux et al. (1994)
Nevertheless, the possibility of being able to reduce operating that an oil-concentrated lyotropic lamellar phase can be made to
costs, or increase throughput on demand, by adding a chemical is exist in a mixture of dodecane (oil), water, sodium dodecylsulfate,
an extremely attractive option. and pentanol. This mixture also has unusual rheological properties.
Another solution is to create a lubricated flow in which water Roux et al. (1994) found that the viscosity drops by a factor of a
flows next to the pipe wall, while oil flows in the core annular 100 under a suitable high stress. Apparently, blocks of oil sepa-
space. The fundamentals and applications of two-phase lubricated rated by a thin water layer slip pass each other once a critical stress
flows have been discussed by Joseph and Renardy (1992). Com- is exceeded.
mercial and developmental experience has been summarized by Figure 1 shows the viscosity of Kern River crude oil as a
Amey et al. (1996). Apparently, there are two unresolved problems function of applied shear stress. There is a drop in viscosity over
for commercial applications. First, heavy components in the oil a narrow range of shear stresses that is very similar to that
tend to foul the pipe and eventually interrupt the flow, and second, observed by Roux et al. (1994).
Crude oils are usually produced along with water, and de-
emulsifiers are often employed to aid in the dewatering process.
Contributed by the Petroleum Division and presented at the Twentieth Annual Commercially available crude oils can contain up to 3 percent
Energy-Sources Technology Conference and Exhibition, Houston, Texas, February water. Since the Kern River crude oil contains 1-3 percent water,
1-3, 1999, of THE AMERICAN SOCIETY OH MECHANICAL ENGINEERS. Manuscript received
by the Petroleum Division, October 13, 1998; revised manuscript received May 10, natural asphaltenic surfactants and surfactants added during the
1999, Associate Technical Editor: W. P. Jepson. dewatering process, one could hypothesize that Fig. 1 implies

Journal of Energy Resources Technology Copyright © 1999 by ASME SEPTEMBER 1999, Vol. 121 / 145

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/05/2014 Terms of Use: http://asme.org/terms


9000,
8000
7000
H- ... . . ] . . „ , .

\'\ 1
6000 -1 \
£ 5000 LL .. ,\....
^ 4000
3000 - \ k
2000 - --- V V ;
1000 L , 7::^::-,..^
0
0 2 4 6 8 10 12 14 16 0 10 15 20 25 30

a(Pa) Wt. %

Fig. 1 Viscosity of Kern River crude oii as function of siiear stress Fig. 2 Viscosity of Kern River crude oil at 77°F (square i<erosene, circles
(circles at 104°F, diamonds at 113°F; 1 Pa = 1.45 x 10"'' psi) pentanoi; 1000 cP = 1 Pa.s)

commercially available Kern River crude oil also has a microstruc- North Sea and an asphaltenic 10° API from Mississippi. Flow rates
ture that becomes organized under stress into a liquid crystal with varied from 1000 barrels per day to 10,000 barrels per day. The
planes oriented in the direction of flow. watercut varied from 4 to 30 percent. After baseline runs with only
Since pentanoi is often added to stabilize a liquid crystal phase the oil, 5 percent pentanoi was added, and then after these runs,
(Bellocq and Roux, 1986), the effect of adding pentanoi and other 0.05 percent of a commercially available drag reducer was also
alcohols on the flow characteristics of heavy oil are described in added. The pressure drop was measured at several different tem-
this contribution. We find there is a significant drop in viscosity peratures for each condition within a narrow temperature range.
that depends on both the identity of the alcohol and the position of This data was used to develop an empirical relationship between
the alcohol functional group along the chain. The flow character- pressure drop and temperature that was used to correct for small
istics were evaluated in an 86-ft-l |-in. (i.d.) flow loop, and the temperature differences. The pressure drop is reported in Fig. 8 at
effect was tested at field conditions in an 1800-ft-6-in. flow loop at a common temperature of 110°F.
Texaco's pilot faciUties in Humble, Texas.

Experimental. The viscosity of Kern River crude oil shown in Results and Discussion
Fig. 1 was measured as a function of shear stress with a Bohlin The behavior of the viscosity of Kern River crude oil as a
constant stress rheometer. All other viscosities reported in this function of shear stress shown in Fig. I is quite suggestive, but not
work were measured using a Brookfield viscometer. In some cases, completely reproducible. Such behavior has been observed fre-
an effective viscosity was calculated from flow loop data by quently in our experiments. The critical stresses are in the range of
assuming Poiseuille flow. 5-15 Pa and the temperatures in the range of 77-113°F. If the
The effect of alcohols on the flow characteristics of heavy oils asphaltenes are treated chemically, the reduction in viscosity can
was evaluated in an Ij-in. circulating oil flow loop. The loop be made to occur both more consistently, and at lower critical
consisted of a 120-gal storage tank, a pulseless progressive cavity stresses. However, the chemical treatment is too difficult to control
Moyno pump, with a maximum discharge pressure of 200 psi, a on a large scale to be practical.
frame and plate heat exchanger and a 120-ft loop of 1 |-in. Sched- The analogy between the viscosity behavior shown in Fig. I and
ule 40 pipe. The test section was 86 ft long, and contained four that reported by Roux et al. (1994) suggest a liquid crystal phase
90-deg bends. A Rosemount differential pressure transducer was could be forming in Kern River crude oil, and this phase could be
used to measure the difference between inlet and outlet pressures stabilized by adding pentanoi. Pentanoi (1-pentanoi) is thought to
across the test section of the loop (86 ft). The tank and pipe were get between surfactant molecules and reduce head group repulsion
insulated, and the outside pipe temperatures were controlled using at the interface. Apparently, the five-carbon chain is just long
heating tape in order to maintain isothermal conditions. In a typical enough to be incorporated in the hydrocarbon environment, but not
test, an oil temperature and flow rate was chosen, the oil was too long as to cause disorder. It is important for the alcohol group
allowed to circulate through the loop for 2-4 h, and then the to be at the end of the chain in order for the alcohol chain not to
differential pressure drop across the test section was measured. spoil the packing of the surfactant molecules.
A three-day test was done at Texaco's fully automated three- Figure 2 shows the effect of pentanoi on the viscosity of Kern
phase Humble flow facility using the 1800-ft-6-in. flowline. Oil River crude at 77°F. Also shown is the effect of an equivalent
and brine was stored in the three-phase separator and pumped amount of kerosene. Kerosene is expected to be a simple dilluent.
separately into the loop. The test section of the loop was 1666 ft Figure 2 illustrates that 1-pentanoi is a least twice as effective in
long and contained two 90-deg bends. There were five temperature reducing the viscosity as kerosene, suggesting that it is not a
and pressure transducers equally spaced along the loop. There simple dilluent.
were two differential pressure transducers, one across the complete Figure 3 further illustrates that the effect is not simple dilution.
1666-ft test section, and one across one-half of the test section. The Butanol and pentanoi are not only more effective than kerosene,
crude oil in this test was a mixture of an 11° API crude from the but also more effective than other similar alcohols. Figure 3 also

Nomenclature
DR = drag reducer ri'„ = dynamic viscosity at zero shear applied shear stress
BOPD = barrels of oil per day stress measure in Bohlin rheometer viscosity measure in Brookfield vis-
AP/L = pressure gradient in flow loop cometer
experiments

146 / Vol. 121, SEPTEMBER 1999 Transactions of the ASIUIE

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/05/2014 Terms of Use: http://asme.org/terms


3500 0.4
»
0.35 L II -
3000 - \
i 0.3 r- - ••- - •

„ 2500 0.25

a. 0.2
•t 2000 -.-- - .- - •-.- - -
41
8 I <
4» -
0.15
- •. -.. I .*
> 1500 0.1
i>
i it
i 0.05
1000
0
02 0.4 0.6 1.2 1.4
500

Carbon Number FLOWRATE(gpm)


Fig. 3 Viscosity of Kern River crude oii witli 5 percent alcohol as Fig. 5 Pressure gradient in 1 f i n . flow loop (circles, Kern River; squares
function of alcohol chain length (1000 cP = 1 Pa.s) with 0.1 percent DR; triangles with 0.1 percent DR + 5 percent pentanoi)

illustrates an optimum chain length for the alcohol. Butanol and since it is basically a high molecular weight polymer. The role that
pentanoi are clearly more effective than methanol or ethanol, and it plays when pentanoi is present is not understood at present. It is
also more effective than octanol. Figure 4 shows the position of the possible that the drag reducer also contains surfactants, and these
alcohol group on the C, chain has an effect; the best position for could help stabilize the liquid crystal microphase. Another con-
the alcohol group is at the end of the chain. jecture is that the high-polymer drag reducer increases the stability
The observations illustrated in Figs. 2-3 are in accord with what of the membrane interface to thermal fluctuations.
one would expect if a liquid crystal phase plays a role in deter- Figure 6 illustrates another interesting observation. The effec-
mining rheological properties. tive viscosity calculated from the flow loop data is relatively
Table 1 shows the addition of a commercial drag reducer in temperature independent when pentanoi and the drag reducer are
combination with 1-pentanoi lowers the viscosity even more. The present. This can be interpreted as also supporting the liquid
optimum concentration depends on both the oil and the concen- crystal hypothesis. The liquid crystal phase is thermodynamically
tration of alcohol, and Table 1 is shown for illustrative purposes stable within a region in the temperature and composition diagram.
only. As long as the fluid is in this phase, the effective viscosity should
Figure 5 shows that the reduction in viscosities observed in the not depend strongly on temperature. Effective viscosity in the
viscometer measurements translate into a decrease in pressure flowing system is calculated from the pressure drop and pump flow
gradient in the Ij-in. flow loop experiments for Kern River crude rate by assuming Poiseuille flow.
oil with 5 percent pentanoi and 0.1 percent of the drag reducer. As Figure 7 illustrates that similar results can be obtained with the
expected, the drag reducer by itself increases the pressure gradient. heavy Venezuelan crude oil BCF13. The maximum flow rate that
can be obtained in the Ij-in. flow loop can be increased by more
than a factor of two when 5 percent pentanoi and the drag reducer
are present.
1150
Figure 8 illustrates the results obtained in the 6-in. flow loop at
1100 Texaco's Humble facility. In this case the oil was a mixture of a
heavy oil from the North Sea and from Mississippi. The watercut
„ 1050 (brine) in the runs without pentanoi were in the range of 4-9
a.
o percent, while those in runs with pentanoi were in the range of 10
•t 1000 percent. Apparently, water contents up to 10 percent can be toler-
o
u ated, and salt does not have a deleterious effect. Addition of 0.01
tf)

> 950 percent of the drag reducer on the third day did not produce a
beneficial effect, however.
900

850 2500

2000
Alcohol Position Number On Chain

Fig. 4 Viscosity of Kern River crude oil with 5 percent pentanoi as 1500
function of aicohoi position on chain (1000 cP = 1 Pa.s)
m
O
O 1000
Table 1 Viscosity (Pa.s) of Kern River crude oil
500
1 percent 3 percent 5 percent
Percent DR pentanoi pentanoi pentanoi
0 0.825 0.554 0.402 130
0.1 0.918 0.544 0.361
0.3 0.868 0.544 0.374
0.5 0.838 0.508 0.407 TEMPERATURE{OF)
1.0 0.962 0.551 0.416
1.5 0.972 0.571 1.440 Fig. 6 Effective viscosity in 1 f i n . flow loop (triangles, Kern River;
squares with 5 percent pentanoi; 1000 cP = 1 Pa.s)

Journal of Energy Resources Technology SEPTEMBER 1999, Vol. 121 / 147

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/05/2014 Terms of Use: http://asme.org/terms


4

3.5 h
r'""
3 7- l\
2.5 ^ V
2
2 1.5 i- --
1
0.5
1:;.=.= -—= =•-
J
0
92 94 96 98 100 102 104 106 108 / \ .

TEMPERATURE(OF)
Time(s)
Fig. 7 Maximum fiow rate in 1 J-in. fiow loop with BCF13 (circies, BCF13;
Fig. 9 Pressure transient In 6-ln. flow line after restart at 10,000 BOPD
diamonds with 5 percent pentanol)

One point of special interest in these tests was what would flow can be completely re-established within moments after an
happen after an emergency shut down. In core annular flow, one emergency shutdown.
would expect that it would be difficult to re-establish the flow.
However, if a liquid crystal phase is involved, one might imagine
that the pipeline could be restarted with only a momentary pressure References
increase corresponding to the effort required to reorient the liquid Achia, B. U., and Thompson, D. W„ 1977, "Structure of the Turbulent Boundary
crystal planes in the direction of flow. Figure 9 shows this is indeed in Drag-Reducing Pipe Flow," Journal of Fluid Mechanics, Vol. 81, pp. 439-464.
the case. The line was restarted after a 25-s shutdown, and after a Arney, M. S., Ribeiro, G. S., Bai, R., and Joseph, D. D., 1996, "Cement-Lined
brief pressure transient, the pressure in the line returned to its Pipes For Water Lubricated Transport of Heavy Oil," International Journal of
previous value. Multiphase Flow, Vol, 22, pp. 207-221.
Bellocq, A.-M., and Roux, D., 1986, "Phase Diagram and Critical Behavior of a
Quaternary Microemulsion System," Microemulsions: Structure and Dynamics, S.
Conclusions Friberg and P. Bothorel, eds., CRC Press, Boca Raton, LA, pp. 33-77.
Burger, E. D., Munk, W. R., and Wahl, H. A„ 1980, "Flow Increase in the Trans
The pressure drop associated with pipeline flow of heavy crude Alaska Pipeline Using a Polymeric Drag Reducing Agent," Society of Petroleum
oil can be reduced significantly by adding butanol or pentanol. Engineers 55th Annual Fall Technical Conference, Dallas, TX, September 21-24,
This result has been demonstrated with Kern River crude oil and a SPE Paper 9419.
Herb, C. A., Chen, L. B., and Sun, W, M., 1994, "Correlation of Viscoelastic
Venezuelan crude oil called BCF 13 in a Ij-in. flow loop that was
Properties With Critical Packing Parameter for Mixed Surfactant Solutions in the i |
86 ft long. It was also demonstrated using a mixture of North Sea Region," Structure and Flow in Surfactant Solutions, ACS Symposium Series 578,
and Mississippi heavy oils in a 6-in. flow loop that was 1666 ft pp. 153-166.
long. Up to 10 percent brine did not have a deleterious effect. The Joseph, D. D., and Renardy, Y. Y., 1991, Fundamentals of Two-Fluid Dynamics,
Part II: Lubricated Transport, Drops and Misible Liquids, Springer-Verlag, New
York, NY.
Roux, D., Nallet, F., and Dial, O., 1994, "Relation Between Rheology and Micro-
structure of Lytropic Lamellar Phases," Structure and Flow in Surfactant Solutions,
ACS Symposium Series 578, pp. 300-305.
Storm, D. A., Barresi, R. J., and DeCanio, S. J., 1991, "Colloidal Nature of Vacuum
Residue," FUEL, Vol. 70, pp. 779-782.
Storm, D, A., and Sheu, E. Y., 1993a, "Rheological Studies of Ratawi Vacuum
Residue at 366 K," FUEL, Vol. 72, pp. 233-237.
Storm, D. A., Barresi, R, J., and Sheu, E. Y., 1995, "Rheological Study of Ratawi
Vacuum Residue in the 298-673 K Temperature Range," Energy & Fuels, Vol. 9, pp.
168-176.
Storm, D. A., Barresi, R. J., and Sheu, E. Y„ 1996, "Development of Solid
Properties and Thermochemistry of Asphalt Binders in the 25-65°C Temperature
Range," Energy & Fuels, Vol. 10, pp. 855-864.
Storm, D. A., Sheu, E. Y., and DeTar, M. M., 1993b, "Macrostructure of Asphalt-
enes in Vacuum Residue by Small Angle X-Ray Scattering," FUEL, Vol. 72, pp.
977-981.
Storm, D. A., BaiTesi, R. J., and Sheu, E. Y., 1995, "Evidence for the Micellization
of Asphaltenic Molecules in Vacuum Residue," 210th National Meeting of American
Flowrate{BOPD) Chemical Society, Chicago, IL, August 20-25, Division of Petroleum Chemistry
Preprints, Vol. 40, pp. 776-779.
Fig. 8 Pressure drop in 6 in. flow loop at 316 K (IIO^FH*, as Is; • with Strassner, J. E., 1968, "Effect of pH on Interfacial Films and Stability of Crude
5 percent pentanol) Oil-Water Emulsions," Journal of Petroleum Technology, Mar., pp. 303-312.

148 / Vol. 121, SEPTEMBER 1999 Transactions of the ASME

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/05/2014 Terms of Use: http://asme.org/terms

You might also like