OTC 18691 Sand Transport Modeling in Multiphase Pipelines

Thomas J. Danielson, ConocoPhillips Co. increased risk of corrosion due to microbial attack under the sand bed, and increased risk of equipment failure due to sand. Sand transport in near-horizontal pipelines has four main regimes, depending on the fluid flow rate. Below a critical velocity, sand will drop out of the carrier fluid and form a stable, stationary sand bed. As the sand bed builds over time, the fluid above the bed is forced into a smaller cross-sectional area, causing the fluid velocity to increase. When the velocity reaches a critical value, sand is transported in a thin layer along the top of the sand bed. A steady-state is reached, such that the sand eroded from the top of the bed is replaced by new sand production from upstream. At higher velocities, the sand bed begins to break up into a series of slow-moving dunes, with sand particles transported from the upstream to the downstream side of the dune. As the flow velocity increases still further, the dunes break up entirely, and the sand forms a moving bed along the bottom of the pipe. At liquid velocities above the critical sand-carrying velocity, the sand is fully entrained in the fluid phase, and potentially entrained into the gas phase in multiphase flow. From an operational point of view, once sand enters a production system, it is critically important to design a system that keeps the sand moving. This requires that 1) the critical velocity below which sand settles out must be predicted, and 2) the production system should be designed such that the velocities stay above this minimum, or that some sort of periodic remediation (e.g., pigging) is done to remove the sand. Failure to do so will result in a mixture of sand and water settling to the bottom of the pipe, creating an environment for microbially-induced under-deposit corrosion. Description of Equipment Sand transport experiments upon which the models presented in this paper were based were conducted in SINTEFs inclinable, 6.9 cm flow loop1,2. Maximum inclinations were 1.35o to +4.0o. The total length of the loop is 215 m, and the length of the test section is 15 m. Nominal system pressure was 1.5 bara. General specifications for the sand flow loop are given in Table 1. A system for continuous sand injection was used; sand was injected at the upstream end of the 15 m test section using a sand slurry feeder. A slurry tank was prepared a homogeneous sand-liquid mixture at atmospheric conditions. This allowed for pre-wetting of sand with either water or Exxsol D80. Sand was injected into the flow as a dense slurry of sand in liquid (approximately 30% by volume) using a peristaltic pump. In the sand transport experiments, water and Exxsol D80 were used for the liquid phases, and air was used as the gas

Copyright 2007, Offshore Technology Conference This paper was prepared for presentation at the 2007 Offshore Technology Conference held in Houston, Texas, U.S.A., 30 April3 May 2007. This paper was selected for presentation by an OTC Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Offshore Technology Conference, its officers, or members. Papers presented at OTC are subject to publication review by Sponsor Society Committees of the Offshore Technology Conference. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Offshore Technology Conference 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 where and by whom the paper was presented. Write Librarian, OTC, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract Prediction of the critical flow rate that will result in sand bed formation in multiphase flow is a critical aspect of multiphase production. Many correlations have been developed for solids transportation in multiphase flow; however, all of them treat the multiphase flow in an ad hoc way that does not respect the complexity of the phenomenon. Further, many approaches rely on correlations that have been developed for much higher solids loading than would occur in oil and gas production. In this paper, first a correlation for liquid-solid transport is developed, based on data taken in the SINTEF STRONG JIP. A good fit to both sand bed height and measured pressure drop is obtained. A critical aspect of the model is the assumption that there is a critical slip velocity between the sand and liquid which remains relatively constant over a wide range of flow velocities. Second, the particle diameter is used to augment the surface roughness. A critical velocity correlation is developed, based on solid and fluid properties, and pipe diameter. An essential feature of the model is that the critical slip between the liquid and solid phases is unaffected by the presence of gas. Good fit to the data is obtained. A sand model is developed using OLGA2000 which does a good job in fitting sand hold-up against the experimental data. Such a model can be used to predict sand bed formation potential in field lines. Introduction Sand is often produced out of the reservoir in both onshore and offshore production systems, particularly in reservoirs that have a low formation strength. Sand production may be continuous, or sudden - as when a gravel pack fails. In the case that no downhole sand control is done, or that a sandcontainment strategy fails, a sand-management strategy must be employed, with operations designed to tolerate a certain amount of sand production. Deposition of sand beds poses several risks, including increased frictional pressure losses,

OTC 18691

phase. Sand used in the experiments had a median diameter of 280 and 550 microns, with a specific gravity of 2.7. Table 2 gives the properties of Exxsol D80 and water at experimental conditions. Sand Transportation Theory As mentioned above, the most critical aspect of a sandresistant production system is that the liquid velocity stay high enough to keep a stationary sand bed from forming. By definition, the velocity of liquid that results in sand settlement and incipient sand bed formation is the critical velocity UC. Liquid-Sand Modeling In single-phase flow, predictions of the critical sand suspension velocity based on a force balance, encompassing lift and drag forces on a single particle, e.g., the analysis of Wicks3. One drawback to the Wicks approach is that particle size is not explicitly accounted for. A second drawback is that Wicks type correlations are developed from solids transport situations where the solids loading is very high. Typical sand loading in offshore applications is much smaller than most industrial solid-liquid slurry transport, on the order of 5-40 lb of sand per 1000 bbl of produced liquid. Under these circumstances, the sand is transported in a layer not much thicker than a single particle diameter4. Another approach utilizes turbulence theory, and equates the strength of turbulent eddies to entrain particles into the fluid against gravity forces, which act to settle the particles out. When the condition of critical velocity is attained, the energy required for the particles to remain in suspension must be equal to fraction of the turbulent energy effective in suspending them5. This type of model can be written in the form: UC2CD = 4/3gD(s-1) where UC = critical suspension velocity CD = drag coefficient of particles g = acceleration due to gravity D = pipe diameter s = ratio of particle to carrier fluid density The drag coefficient CD can be estimated by the following expression: CD Re-n where Re = LUCd/L and L = fluid density UC = critical velocity d = particle diameter (2) (1)

L = fluid viscosity This gives an equation of the form: UC = K -n/(2-n)dn/(2-n) (gD(s -1))1/(2-n) (3)

For n = 1/5 (a value that gives a good fit to the SINTEF data for incipient sand bed formation), the following expression is obtained: UC = K-1/9d1/9 (gD(s-1))5/9 (4)

where K is an experimentally determined constant, equal to approximately 0.23 based on the SINTEF data. A similar equation is given by both Stevenson and Salama6. Equation 4 matches the experimentally-determined exponents from the Salama correlation fairly closely (with the caveat that Equation 4 is dimensionally consistent, whereas the model of Salama is not), and reasonably approximates the data upon which Salama based his model. Comparison of Liquid-Sand Model to Data Figure 1 gives a sample data set for sand hold-up as a function of superficial water velocity, at inclination angles of -1.35o, 1.00o, and 4.00o from horizontal. The sand is 280 microns, and sand injection rate is 2.2 g/s. Measurements of both the sand bed height and the pressure drop were taken. Note that the sand hold-up is a very insensitive function of angle, over the range of angles tested. Sand hold-ups were backed out of the sand bed height as follows: Figure 2 gives a schematic of a pipe partially filled with sand. The angle is defined as indicated on the figure, and the sand bed height h is measured along the pipe center line. The sand hold-up is calculated from the measured bed height h has follows: = acos(1-2h) HS = ( sin(2))/ (5a) (5b)

where must be measured in radians. From Figure 1, the sand bed begins to form at the critical carrying velocity of the liquid, which appears, from the figure, to be at approximately 0.47 m/s. It is apparent from the figure that once the sand bed starts to form, the increase in sand hold-up as a function of decreasing water velocity is quite linear. Let us treat the sand as a pseudo-phase, with its own welldefined velocity. Then the slip between the water and sand can be backed out of the data in the following way: USLIP = UW US = USW/(1-HS) USS/HS where USLIP = characteristic slip between sand and water UW = water velocity (6)

OTC 18691

USW = superficial water velocity US = sand velocity USS = superficial sand velocity HS = sand hold-up Here, the superficial velocities are defined as USW = QW/A USS = QS/A Where QW = volumetric rate of water QS = volumetric rate of sand A = pipeline cross-sectional area Figure 3 gives a plot of USLIP vs. USW. Note that the calculated slip between the sand and the water is almost constant at a value very close to the critical carrying velocity UC (indicated by the solid line). In fact, let us equate USLIP to UC. Equation 6 can be written: UCHS2 + (USW + USS UC)HS USS = 0 (7)

A few alterations are required to this expression to obtain a good fit to the experimental data. First, the Reynolds number must be based on the hydraulic diameter: DH = 4(area/wetted-perimeter) = (1HS)D/( + sin()) (10)

Second, an additional term for particle size must be added, in a manner analogous to a surface roughness. This gives: f = 0.0055*(1 + ((-)/(/DH)/50 + sin()(d/DH)/40 + 106/Re)1/3) (11)

Note that Equation 9 is recovered as 0, as required. A plot of the measured vs. predicted pressure gradient is given in Figure 4; an extremely good fit to data can be obtained with expression above. The gravity component of pressure loss is given by: dP/dzgravity = Lgsin() (12)

This quadratic equation can easily be solved to give the hold-up based on the superficial liquid and sand velocities and the critical carrying velocity UC. The model results are plotted in Figure one as a line through the experimental data. The fit to data is extremely good, indicating that the assumption of a constant slip between the water and sand is a very good approximation. The physical picture is as follows: the sand trails the carrier liquid by a well-defined slip or critical velocity UC. When the carrier fluid velocity drops below this value, a sand bed begins to form. The bed height increases until the cross sectional area available to fluid flow is reduced enough to restore the fluid velocity over the bed back to the critical velocity UC. Pressure Gradient Calculation The SINTEF solid-liquid experiments also included pressure gradient information. For a flow without any significant sand bed formation, the following equation would apply for the determination of the frictional pressure gradient: dP/dzfriction = fUL2/(2D) (8)

where is the angle from horizontal. The sand bed does not contribute to the gravity pressure loss, as the sand bed is not fluid, but static, and therefore modeled as a solid. Figure 4 has an interesting feature, which is that the slope pressure gradient with respect to flow rate is negative from USL = 0 to roughly USL = UC. When USL increases above UC, then the slope pressure gradient becomes positive. A negative slope of pressure gradient with respect to flow rate is known to cause unstable behavior in multiphase flow, resulting in what is called terrain slugging the periodic accrual of liquid which is then pushed out of the system as a large slug. This has obvious analogy to the formation of moving dunes in sandliquid flow. At very low rates of USL (about USL = 0.2 m/s in the SINTEF data), the sand bed stabilizes, perhaps due to the transition from turbulent to laminar flow above the bed. Gas-Liquid-Sand Modeling There have been many attempts to generalize liquid-sand models to gas-liquid-sand models. Examples include Salama6, Angelsen et al.7, and Oudeman8. The Wicks model has been extended by Angelsen et al. to 2-phase flow, using the calculated liquid velocity and the hydraulic diameter with reasonable results. Differences between the experimental data and the model were attributed to inaccuracies in the multiphase model, rather than in the sand settling model. This general approach was also used by Oudeman, with some success. Angelsen and Oudeman generalize a Wicks-type formulation, whereas Salama generalizes the formulation of Oroskar et al.. Salamas equation has the form: UM,C = USL + USG = (USL/UM)0.04d0.17-0.09(s1)0.55D0.47 (13)

where f is the friction factor, which can be obtained from the following expression: f = 0.0055*(1 + (/D/50 + 106/Re)1/3) where = pipe roughness (in microns) Re = L UL D/ L (9)

OTC 18691

Where UM,C is the critical mixture velocity of the multiphase flow below which sand settling occurs. The problem with all of these approaches is the attempt to create a multiphase sand model without first developing a multiphase gas-liquid model. Gas-liquid multiphase flow is extremely complex, exhibiting different flow regimes (slug, stratified, bubble) depending on the local conditions in the pipe. It is known that flow regime has a large impact on the sand carrying capacity of the flow. Therefore, it is essential that some sort of treatment be given to the multiphase flow first; only then can the liquid-solid problem be reasonably approached. The Drift-Flux Model An excellent and simple model for multiphase slug flow (which also works quite well for other regimes as well, particularly at high pressure) is the drift-flux model. This model was first put forward by Nicklin et al.9. The drift flux model exploits the fact that, over a wide range of conditions, the gas velocity is a linear function of the mixture velocity: UG = CUM + UO (14)

The overall hold-up is calculated from Equation 17, and the sand hold-up HS is derived from: UC = UL US = USL/HL USS/HS = USL/(HO-HS) USS/HS which gives: UCHS2 + (USL+USS-UCHO)HS USSHO = 0 (21) (20)

Where UG is the gas velocity, C has a value of roughly 1.2 for air/water, and UO is the bubble rise velocity UO is given by: UO = 0.4((L-G)/L)1/2(gD)1/2 (15)

The relationship between USG, UG, and HL is as follows: UG = USG/(1-HL) (16)

When the sand superficial velocity USS = 0 (i.e., no sand production), the original liquid hold-up equation is recovered. At typical sand production rates, the ratio of USL:USS is on the order of 5000:1. Using this ratio, and a ratio of USG/USL ratio of 1 (slug flow), the drift-flux model is used to determine the overall hold-up (Equation 17), and the sand hold-up is determined from the solution of Equation 23. Figure 5 gives the gas, liquid, and sand velocities employing these models as a function of mixture velocity UM. Note that the liquid velocity drops steadily until the critical carrying velocity UC (here equal to 0.4 m/s) is reached. At this point, the sand bed begins to form in such a way that the liquid velocity is maintained at UC, as required by the models. Figure 6 gives the sand and water hold-ups as a function of UM. Note that as the liquid velocity reaches the critical velocity, the liquid hold-up begins to drop and the sand holdup increases. This occurs in order to maintain the difference between the sand and liquid velocities at the critical velocity UC. Test of Model against SINTEF Air-Liquid-Sand Data The gas-liquid-sand data taken at SINTEF is extremely complex and, in some instances, difficult to interpret1. However, a few general conclusions can be drawn. First, when the flow regime is stratified, a USG of around 8 m/s is required to keep the sand entrained and moving when the USL is 0.05 m/s, resulting in a measured liquid velocity UL of 0.45 m/s. Note that this is extremely close to the critical velocity UC calculated for the single-phase experiments. Figure 7 was produced by calculating the combination of USG and USL which led to a liquid velocity UL equal to the critical velocity UC, which leads to incipient sand bed formation. There is excellent agreement between this figure and that produced by Angelsen in his air-water-sand rig7. It is clear from the data, and observations made during the experimental runs, that the sand carrying capacity of the liquid is not directly influenced by the presence of gas. The key to understanding and predicting the sand-carrying capacity of a multiphase flow lies in the liquid velocity, which is only indirectly influenced by the gas velocity. Thus, it is the behavior of the multiphase flow, and the slip between the gas and liquid phases, which is the key to understanding and predicting the sand-carrying capacity of a multiphase flow. Slug flow does a better job of carrying sand because it tends to have a higher liquid loading, and therefore a higher liquid velocity. At higher pressures, stratified flow could also potentially do a good job of sweeping solids from the line, due

The liquid hold-up can then be obtained from the following expression: HL = 1 USG/UG = 1 USG/(CUM + UO) (17)

Once the liquid hold-up is obtained, the liquid velocity can be obtained from: UL = USL/HL (18)

It is this liquid velocity that is used to determine the sand hold-up. It has been observed that the gas velocity has little direct influence on critical carrying velocity of the liquid1. The main impact of increasing the gas rate seems to be indirect; an increased gas velocity results in a higher liquid velocity, which in turn results in a lower sand hold-up. Now the overall hold-up, HO, defined as: H O = HL + HS (19)

OTC 18691

to the lower slip velocity between the gas and liquid phases at higher pressure. Use of OLGA2000 to Predict Incipient Sand Bed Formation The OLGA2000 multiphase flow code can be used very effectively to assess sand bed formation tendency for a multiphase flow. In the OLGA gas-oil-water 3-phase model, it is possible to set the slip velocity between the oil and water phases to a constant value in the WATEROPTIONS. This slip velocity can be set equal to the critical mixture velocity calculated by Equation 4. OLGA uses a static look-up table for fluid properties; water density can be set equal to the particle density by editing the gas-oil-water table. Thus, the water phase in OLGA is replaced with a pseudo-solid phase, with a slip velocity equal to the critical velocity UC. Recall that the SINTEF experiments showed that the slip between the water and sand is not directly affected by the gas velocity. Therefore, the OLGA model, using sand as a pseudo-phase, can be used to assess the sand-bed forming potential of any multiphase flow, including all multiphase flow regimes. Figure 8 shows the behavior of the 3-phase OLGA model, where the gas rate was set to zero, and the slip between the oil and sand phases was set at the experimentally determined critical velocity, UC = 0.47 m/s. Note that the sand hold-up calculated by OLGA is in very good agreement with the experimental values. If one had access to the OLGA source code, Equation 9 (which is used in OLGA to determine the friction coefficient) could be replaced with Equation 11, and an equivalent fit to the frictional pressure drop data (shown in Figure 4) could also be achieved. Figure 9 shows the sand-forming possibilities for a long, arctic wet gas pipeline (about 5 bbl/mmscf liquid loading), with a gas rate of 2 bcfd. The critical velocity for sand accumulation is assumed to be 1 m/s. Note that sand beds are formed in selected portions of the line, as a function of the local liquid velocity. In uphill sections, the liquid velocity drops, resulting in an accumulation of sand. Near the shore, the gas velocity increases, due to a drop in gas density. The higher gas rate results in a higher liquid velocity, which in turn results in a drop in the bed-forming capacity of the pipeline. Note that this line is predicted to have significant sand settling in the first half of the line. Figure 10 gives the sand hold-up as a function of angle, based on the model. An interesting result is that while the sand hold-up is rather insensitive to angle for liquid-solid flow, it is quite sensitive for gas-liquid-solid flow. This is because the liquid velocity in multiphase flow can be a very sensitive function of angle particularly for wet gases. An additional effect that multiphase flow brings is that of flow regime. Pipeline leaks tend to occur in areas where there is a dip in the pipeline. At low flow rates, a multiphase flow is often stratified going down into the dip, and slug in the inclined section flowing out of the dip thus, slug flow is initiated in the pipeline dip. Formation of slugs is a very energetic process (likened to a washing machine), and any sand entrained in the pipeline acts to scour the pipeline wall at this point, resulting in pipeline wall loss, and ultimately pipeline failure.

Conclusions A model for critical solid-carrying velocity UC was developed which gives a good fit to experimental data. The critical velocity in liquid-solid flow has no angle dependence over the range of angles investigated. This model can be generalized to predict sand bed height by assuming that the slip velocity between the sand and carrier liquid is maintained at UC for all liquid rates. The model shows excellent fit to data. One the sand hold-up is known, the frictional and gravity pressure gradient can be predicted with a modified friction factor equation which takes account of the reduced area for flow, and the increased pipe roughness created by the sand bed surface. A key observation in the gas-liquid-solid experimental data was that the gas rate had no direct influence on the critical slip velocity between the sand and the carrier liquid. The liquid velocity is calculated from a multiphase correlation, and this velocity is used to determine sand bed formation tendency. Agreement with data, and with previously published data is excellent. The OLGA2000 code can be used to determine the liquid velocity UL, and an estimate of the sand hold-up by modeling sand as a pseudo-phase with a slip velocity equal to UC. Good fit to data was obtained for both liquid-solid and gas-liquidsolid experiments using this approach. A key observation of the multiphase model was that, because the liquid velocity can be a strong function of angle in multiphase flow (particularly for wet gases), sand bed formation is also strongly correlated to pipe angle. This is a critical difference between liquid-solid (where pipe angle had negligible influence on sand bed formation) and gas-liquidsolid systems. Acknowledgements The author would like to thank ConocoPhillips for their permission to publish this paper. The author would also like to thank SINTEF, which provided in large part the data upon which this analysis is based. Nomenclature A pipeline cross-sectional area, (m2); C drift-flux model constant, (-); d sand particle median diameter, (m, m); D pipe inner diameter, (m); DH hydraulic diameter, (m); g acceleration due to gravity, (9.8 m/s2); h sand bed height, (m); HL liquid hold-up, (-); HO overall hold-up, (-); HS sand hold-up, (-); P pressure, (Pa); QS volumetric sand rate, (m3/s); QW volumetric water rate, (m3/s); s ratio of particle to carrier fluid density, (-); UC critical carrying velocity, (m/s); USG = superficial gas velocity, (m/s); USL superficial gas velocity, (m/s); USS superficial sand velocity, (m/s); UG gas velocity, (m/s);

OTC 18691

UL liquid velocity, (m/s); UM mixture velocity, (m/s); UM,C critical mixture velocity, (m/s); UO drift flux model constant, (m/s); US slip velocity, (m/s); UW water velocity, (m/s); z pipeline axial distance, (m); L liquid viscosity, (kg/m-s); G gas viscosity, (kg/m-s); = pipe angle from horizontal, (degrees); L liquid density, (kg/m3); G gas density, (kg/m3); S sand density, (kg/m3). References
1. Dahl, A.M., Ladam, Y., Unander, T.E., Onsrud, G., SINTEF-IFE Sand Transport 2001-2003, SINTEF Internal Report, 2003. 2. Yang, Z.L., Ladam, Y., Laux, H., Danielson, T.J., Leporcher, E., and Martins, A.L., Dynamic Simulation of Sand Transport in a Pipeline, 5th North American Conference on Multiphase Technology, Banff, Canada, 2006. 3. Wicks, M., Transport of Solids at Low Concentrations in Horizontal Pipes, Advances in Solid-Liquid Flow in Pipes & Its Application, pp. 101-124, Pergammon Press, 1971. 4. Stevenson, P., Thorpe, R.B., Kennedy, J.E., and McDermott, C., The Transport of Particles at Low Loading in Near-Horizontal Pipes by Intermittent Flow, Chemical Engineering Science, Vol. 56, pp. 2149-2159, 2001. 5. Oroskar, A.R., and Turian, R.M., The Critical Velocity in Pipeline Flow of Slurries, AIChE Journal, Vol. 26, No. 4, pp. 550-558, 1980. 6. Salama, M.M., Sand Production Management, ASME Transactions Energy Resources Technology Journal, Vol. 122, No. 1, pp. 29-33, 2000. 7. Angelson, S., Kvernvold, O., Lingelm, M., and Olsen, S., Long Distance Transport of Unprocessed HC Sand Settling in Multiphase Flow Lines, Proc. Of 4th International Conference on Multiphase Flow, BHRA, Nice, pp. 149-170, 1989. 8. Oudeman, P., Sand Transport and Deposition in Horizontal Multiphase Trunklines of Subsea Satellite Developments, SPE Production & Facilities, pp. 237-241, Nov 1993. 9. Nicklin, D.J., Wilkes, J.O., and Davidson, J.F., Two-Phase Flow in Vertical Tubes, Trans. Instn. Chem. Engrs., Vol. 40, pp. 6168, 1962.

Table 2: Measured Fluid Properties for Exxsol D80 and water Density (kg/m3) Viscosity (cP) Surface tension Exxsol water Exxsol Water (mN/m) D80 D80 795.3 999.8 2.17 1.33 34.4

Tables Table 1: General Specifications of the Flow Loop Gas Phases Air, nitrogen, SF6 Liquid Phases Exxsol D80, water Maximum Pressure 8 bara Pipe diameter 0.069 m Loop length 215 m Superficial oil velocity 0.01-2 m/s Superficial water velocity 0.01-2 m/s Superficial gas velocity 0.1-8 m/s

OTC 18691

Figures
0.9

0.8

0.7

0.6 sand hold-up (-)

0.5

0.4

0.3

0.2

0.1

0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 superficial water velocity (m/s) angle = -1.35 angle = 1.00 angle = 4.00 model

Figure 1. Plot of liquid-sand model against data taken in the SINTEF medium scale flow loop. Note that there is very little influence of pipe angle on sand bed hold-up. An excellent fit to data can be obtained by setting the slip between the liquid and sand phases equal to a constant.

sand bed h

Figure 2. Schematic of a sand bed in a pipeline. Here is the angle between a vertical line through the center of the pipe, and a second line from the center of the pipe to the point where the top of the sand bed intersects the pipe wall. The sand bed height h is defined as the height of the bed at the centerline of the pipe.

OTC 18691

0.6

0.5

calculated slip velocity (m/s)

0.4

0.3

0.2

0.1

0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 superficial water velocity (m/s)

Figure 3. Plot of the experimentally measured slip velocity between the liquid and sand as a function of the superficial water velocity. Note that this velocity is almost constant over a wide range of superficial water velocities, and is equal to the calculated critical slip velocity UC beween the liquid and the sand.
140

120

100 pressure gradient (Pa/m)

80

60

40

20

0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 superficial water velocity (m/s) experimental dP model dP

Figure 4. Plot of the experimental and model pressure gradient as a function of superficial water velocity for liquid-sand flow. Note that there is a point where the pressure gradient with respect to USW is negative. In multiphase flow, operation in this region results in terrain slugging; in sand-liquid flow, this region results in the formation of moving sand dunes.

OTC 18691

2.5

2.0 gas, liquid, and sand velocity (m/s)

1.5

1.0

0.5

0.0 0.0 0.2 0.4 0.6 mixture velocity (m/s) gas velocity liquid velocity sand velocity 0.8 1.0 1.2

Figure 5. Plot of gas, liquid, and sand velocity predicted by the multiphase sand model. Note that the difference between the sand and liquid velocity is maintained at a contant value, equal to the critical velocity UC.
0.90

0.80

0.70 liquid and sand hold-up (-)

0.60

0.50

0.40

0.30

0.20

0.10

0.00 0.0 0.2 0.4 0.6 mixture velocity (m/s) liquid hold-up sand hold-up 0.8 1.0 1.2

Figure 6. Liquid and sand hold-ups corresponding to the velocities in Figure 5. Note that the liquid hold-up drops as the sand hold-up increases; the overall hold-up is well-described by a drift-flux type model formulation.

10

OTC 18691

0.25

0.2

superficial liquid velocity (m/s)

0.15

Solids Transport

0.1

No Solids Transport
0.05

0 0 1 2 3 4 5 6 7 8 9 10 superficial gas velocity (m/s)

Figure 7. Prediction of the border beween sand bed formation and no sand bed formation for air-water-sand flow. This theoretical result quantitiatively matches the experimental data of Angelsen.
1.0 0.9 0.8 0.7 sand hold-up (-) 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 superficial water velocity (m/s) experiment olga model

Figure 8. Test of the OLGA, modeling sand as a pseudo-phase with a slip velocity equal to the critical velocity UC measured in the SINFEF liquid-sand experiments. This shows that OLGA can be used to predict sand bed formation in single-phase flow. Since the slip between the sand and liquid is not a function of the gas velocity, but only the liquid velocity, OLGA can be used to predict multiphase sand bed formation tendency.

OTC 18691

11

0.50 0.45 0.40 0.35 total and sand hold-up (-) 0.30 0.25 0.20 0.15 0.10 0.05 0.00 0 100 200 300 distance (km) total sand 400 500 600

Figure 9. Sand bed formation tendency in an Arctic pipeline, as predicted by OLGA. Note that sand bed tendency is high from 0to 300 km, due to the lower gas velocities there. The lower gas velocity gives a lower liquid velocity, resulting in sand bed formation in the inclined sections of the pipeline.
0.40

0.35

0.30

0.25 sand hold-up (-)

0.20

0.15

0.10

0.05

0.00 -5 -4 -3 -2 -1 0 1 2 3 4 5 angle from horizontal (degree)

Figure 10. Plot of sand bed height as a function of pipeline angle for the Arctic field line case. Note that while sand bed formation tendency is not a function of angle in liquid-sand flow, it is a very strong function of angle in gas-liquid-sand flow. This is due to the fact that the liquid velocity in a multiphase flow is a very strong function of angle, particularly in stratified flow.