Vol. 227 No. 5

Understanding flowline-riser system operability

Slug flow in risers can be a vexing problem. Understanding and predicting it can improve production and safety for operators.

Saeid Mokhatab, Contributing Editor, Laramie, Wyoming, USA

In many oil and gas developments that incorporate multiphase flowlines, system instability is a major flow assurance concern. Excessive demands that large changes in oil and gas flowrates place on processing facilities are a major source of such instability. Multiphase instability can influence the selection of flowline route, flowline diameter and the requirement for gas lift for a subsea oil development concept. Consequently, they can potentially have a significant negative impact on the net present value of a system. Therefore, they should be addressed as early in the design phase as possible, and in sufficient detail to ensure a robust concept passes through to the later design phases.

Multiphase surges come in three forms: hydrodynamic slugs, operationally induced surges and riser-induced slugs. Hydrodynamic surges can usually be accommodated in a modestly sized slug catcher and are of less concern, which is fortunate, because accurate prediction of this phenomenon remains problematic. Operationally induced surges can be significant, but can also be controlled by suitable adjustment to operating procedures. However, riser-slugging phenomena (also called severe slugging) are considered the most important of the three surges, as it usually generates the most violent surging behavior.

Severe slugging flow can pose serious problems to the designer and operator of two-phase flow systems. Large and fluctuating gas/ liquid rates can severely reduce production and, in the worst case, shut down or damage topside equipment, such as separator vessels and compressors. As a result, prediction of severe slugging characteristics is essential for optimal, efficient, safe and economically feasible design and operation of two-phase gas-liquid slug-flow systems. The design (Fig. 1) of stable flowline-riser systems is particularly important in deepwater fields, since the propensity toward severe slugging is likely to be greater, and the associated surges become more pronounced with increasing water depth.

Fig. 1. Different riser shapes.

SEVERE SLUGGING

The severe slugging phenomenon occurring in multiphase transport system is illustrated in Fig. 2. Schematically, it is a periodic phenomenon that can be split into four steps. The first step, slug formation, corresponds to an increase of the pressure in bottom of the riser. The liquid level does not reach the top of the riser. During this period, the liquid is no longer supported by the gas and begins to fall, resulting in blockage to the riser entrance and pipeline pressure buildup, until the liquid level in the riser reaches to the top.

Fig. 2. Schematic of severe slugging in flowline-riser system.

During the second step, slug production, the liquid level reaches the riser outlet, and the liquid slug begins to be produced until gas reaches the riser base.

In the third step, bubble penetration, gas is again supplied to the riser, so the hydrostatic pressure decreases. As a result, the gas flowrate increases.

The fourth step corresponds to gas blowdown. When the gas produced at the riser bottom reaches the top, the pressure is minimal and the liquid is no longer gas-lifted. The liquid level falls and a new cycle begins.¹

For severe slugging to occur, the flow must be stratified in a downward inclined pipe before a low point. This is because the slug phenomenon is dependent on a large, compressible volume. Further, the pipeline must have downward flow at some location. It is also worth noticing that riser slugging occurs for relatively low flowrates of liquid and gas.²

Severe slugging types. Lunde classified flow in a flowline-riser system using four main categories: steady flow, oscillation flow, severe slugging and transitional flow (with occasional severe slugging).³ Oscillation is a wavy, pressure-cycling flow phenomenon, characterized by a continuous oscillation in liquid and gas production. Several classes of slug flow exist, depending on magnitude of pressure fluctuations and the size and number of alternating gas and liquid slugs. Tin described the different types of severe slugging present in an S-shaped riser as follows:⁴

1. Severe Slugging I (SS-I), or classical severe slugging, is characterized with a full column of liquid in the riser prior to gas blowdown. The gas blowdown is initiated by bubble penetration at the riser base. The liquid slug length is greater than one riser height.
2. Severe Slugging II (SS-II), where the liquid column is unstable, causing no constant slug production stage. There is no liquid backup along the pipeline; hence, the cycle time is generally shorter.
3. Severe Slugging III (SS-III) is characterized by a continuous bubble penetration in the riser, producing an unstable liquid column in the riser.

Fig. 3 shows a flow-pattern map of the free-hanging catenary riser. The same pressure cycling region is observed for different riser configurations, and the same stability line is fitted into each flow-pattern map. Although the geometry of the riser does not influence the pressure cycling boundary, it does affect the severe slugging boundary and the various cycle characteristics.⁴

Fig. 3. Flow-pattern map of a free-hanging catenary riser.4

The stability problem. Severe slugging is very undesirable. Riser-base pressure and riser-outlet fluid-flow fluctuations result in unwanted flaring, which reduces the operating capacity of the separation and compression units. Figure 4 shows some typical predicted time traces accompanying this phenomenon. Clearly, such large transient variations will create severe feed disturbances for the topside separator, causing poor separation and, in some cases, overflow and shutdown of the separator.

Oscillations in gas production may cause operational and safety problems during flaring, and the high pressure oscillations can reduce the ultimate recovery from the field, reducing the amount of the recoverable reserves. Other adverse consequences are wear and tear on the equipment resulting in possible unplanned process shutdowns.^5,6,7

As Fig. 4 shows, pipeline-riser systems are susceptible to a form of instability that can lead to large surges in liquid and gas production rates, and, if the topside processing facilities are not adequately sized, this can result in equipment trip-outs and unplanned shutdowns. Indeed, in some circumstances, the magnitude of these surges can render a system inoperable, necessitating costly equipment modification, such as the retrofitting of a large slug catcher.

Fig. 4. Example time traces of surging during severe slugging.8

Liquid production characteristics. In terms of fluid production, the trend associated with pressure cycling is reflected in the liquid production characteristics. Typical examples of liquid mass production are shown in Figs. 5a and b. Severe Slugging I (Fig. 5a) is characterized by three main periods, the period of no production, of constant liquid production, and transient production. In the first period, liquid accumulates in the low point due to its low velocity and the liquid that falls down; this forms a slug until the pressure becomes sufficient to lift the liquid column. In the second period, the liquid slug starts upward along the riser. Gas begins to flow in the riser and accelerates the liquid. Finally, the gas arrives at the top of the riser and the pressure rapidly decreases, causing liquid flow down.

Fig. 5. a) Liquid production profile during SS I. b) Liquid production profile during SS II.8

As stated before, there is no constant production period during SS II; therefore, all fluid production in this flow regime is single-liquid transient production, which occurs during the gas blowdown period of severe slugging, Fig. 5b.

Last, during slug flow, there is continuous delivery of liquids, as slug flow is made up of a bubbly liquid and a long bubble/ film region. However, there are surges in liquid production that correspond to the arrival of gas bubbles.

Gas production characteristics. In terms of gas production, typical results are shown in Figs. 6a and b. As can be seen, during SS I, the gas production cycle is made of two parts, the period of no gas production – which occurs during liquid buildup, slug production and bubble penetration stages of the severe slugging cycle – and transient gas production. However, for SS II, gas production is transient, followed by a period of constant gas production. Finally, during slug flow, there is continuous delivery of gas, as slug flow is made up of a bubbly liquid and a long bubble/ film region, just as in the case of liquid production. However, there are surges in gas production, corresponding to the arrival of the slug body.

Fig. 6. a) Gas production time trace during SS I. b) Gas production time trace during SS II.9

MODELING

Identification of the three SS flow regimes was based on visual observation and the pressure profile at the riser base during the slugging process. However, to design a stable flowline-riser system, it is necessary to analyze the stability of these systems using computationally intensive parametric techniques. These techniques attempt to build a detailed map of regions of stable and unstable behavior. However, more comprehensive data analysis (used to build a stability map) is required, since considerable uncertainties exist in a number of areas. With the development of automatic data processing methods, it is now possible to carry out comprehensive sensitivity studies of system stability. These provide great insight into the behavior of flowline-riser systems, thus allowing the engineer to produce robust designs that avoid unstable and problematical conditions.

A model to study dynamic behavior of multiphase flow in pipeline-riser system was developed under the environment of OLGA2000 code, the market-leading simulator for transient multiphase flow of oil, water and gas in pipes, which was licensed from Scandpower (2000). The model consisted of three major parts: fluids PVT description, a pipeline-riser model and specifications of the boundary conditions. The PVT behavior is calculated from a model comprising a list of fluids, their fractions in the mixture, and the temperature and pressure range expected in the system.

Fluids consisted of air and water. Air was simulated using a mixture of nitrogen and oxygen. An equation of state (PR-EOS¹⁰) was used for calculating PVT behavior in all simulations, where PVT table limits of 0.5 to 10 bar and 5° to 50°C are used as input to the PVT calculations. Note that the pressure and temperature range of the simulation must not only cover the range of expected conditions within the pipeline-riser system, but also all conditions that the numerical scheme may encounter. Once the values are beyond the range of the PVT table, the thermodynamic properties cannot be associated with the calculation, causing a PVT table error, halting code execution.

TABLE 1. Pipe Properties

The pipeline model consisted of a pipe model and a geometry model, which provides generic information to simulate pipeline-riser system. The pipe is made of 4-in. nominal bore, stainless steel pipe (SS 304). Table 1 gives some pipe properties that were used in the simulation. The value of pipe roughness was estimated from carrying out single-phase water-flow experiments in the pipeline-riser system.

Successful simulations require accurate pipeline geometry, where the flowrate at which the riser base begins to accumulate liquid depends on the pipeline topography.¹¹ Fig. 7 shows the pipeline-riser geometry. The geometry model was taken from the tabulated x-y values for the pipeline-riser profile. The x-y coordinates form a series of pipe sections of short length, which are again divided into individual computational cells. In fact, for having a good model of the flow in the riser region, short section lengths should be used for “discretization” of the riser and the pipe joining the riser. The structure of the grid is specified such that at least two cells are in each pipe section and that the ratio of cell length from section to section is about two. The section length before the riser should be small; large sections will increase the time for liquid blockage in the bend.

Fig. 7 Elevation profile for the pipeline-riser system.

As seen in Fig. 7, the entire simulation length also included the 3-m horizontal pipe section at the riser top, entering the topside separator. In fact, recent simulations studies have confirmed that without such a section, simulation results will give unphysical flows exiting the riser, particularly excess liquid fall-back into the riser post blowdown.^8,12

To set up pipeline simulations, one must also specify boundary conditions at the pipeline inlet. These boundary conditions should closely approximate the behavior of the actual upstream and downstream systems, if instabilities are to be modeled correctly. Inappropriate boundary conditions can dampen or amplify the fluctuations in the unstable region, and will even affect the position of stability boundary. Hence, careful thought must be given to the level of detail required from the study. The boundary conditions are usually either defined as fixed pressure or fixed flowrate, often fixed flowrate at the inlet or fixed pressure at the outlet. This is common practice for simulation studies of multiphase pipelines using the OLGA software.¹³

Note that transient simulations require initial conditions to begin the simulation. These conditions are then updated as time progresses, subject to the boundary conditions and defining equations of the system. To assess the stability of the system during normal operation, it is essential to ensure “fully developed flow.” Consequently, it is important to run the simulation for a significant length of time to ensure that start-up transients, arising from the selection of initial conditions, have decayed, leaving the underlying behavior. Once fully developed flow has been achieved, the model must be simulated for a sufficient time in order to capture the full periodicity of any surges and to make sure that the system is fully developed.

It is important, however, to select the limits of the time-step size carefully. The initial value is set to assure convergence of the simulation in the initial stages of computation. The maximum integration time-step, which should be below the Courant-Friedrich-Levy (CFL) criterion for the flow conditions encountered in the simulation, is set to prevent the PVT table errors, while the minimum integration time step is set to ensure practical simulation times.^8,14

SIMULATION RESULTS

The predicted results of the OLGA model of pipeline-riser system were compared to the experimental data. The experimental campaign covered two sets of experiments for different inlet air-water flowrates. These experiments were conducted at constant topside separator pressure (1 barg). For the first experiment, the inlet air flowrate was fixed at 10 Sm³/hr, and the inlet water flowrate was changed as shown in Fig. 8. Continuing from the final step of the first experiment, Test 2 was conducted with the fixed water flowrate at 0.5 lit/sec and variable air flowrates as given in Fig. 9. The fluctuations in the measured inlet air and water flowrates for both Tests 1 and 2 are most likely caused by variations in the topside separator pressure.

Fig. 8. Inlet water and air flowrate profiles investigated in the first test of experimental campaign.

Fig. 9. Inlet water and air flowrate profiles investigated in the second test of experimental campaign.

Results of the OLGA model predictions compared to the experimental data for pressure cycling characteristics are shown for Tests 1 and 2. Figs. 10 and 11 show the OLGA simulation results and experimental data for the riser-base pressure in Tests 1 and 2 of the experimental campaign. The OLGA simulations have been carried out with a constant pressure boundary equal to the average topside separator pressure (1 barg). For these cases, the pressure difference between maximum and minimum riser-base pressure for the model compared to the experimental data does not agree very well. The OLGA model also shows a severe slugging frequency that is slightly higher than that of the experimental data. However, comparisons between experimental data and computational results for varying water flowrates at a constant air flowrate (Fig. 10) show good correspondence at low water flowrates (about 1.0 lit/sec). There is still a slight time shift between experimental and simulation results at lower water flowrates (0.5 lit/sec). 

Fig. 10. Comparing riser-base pressure time traces (Test 1).

At higher water flowrates (about 2.0 lit/sec), full blocking occurs, trapping gas in the pipeline and causing build-up of liquid in the riser, and an increasing slugging period. Also for Test 2 (as shown in Fig. 11), at high air flowrates (20 Sm³/hr), the slugging periods become difficult to estimate as the flow changes to SS II, where there is no slug production period (indicating that no liquid is backed up the pipeline when the severe slugging is fully formed), and the cycle time of severe slugging is also lower than that of Severe Slugging I, increasing the slugging frequency. 

Fig. 11. Comparing riser-base pressure time traces (Test 2).

The continuous air penetration at the riser-base indicates that there is a lack of gas compression in the pipeline that is not allowing the pure liquid formation in the riser. This then may suggest the OLGA model is not predicting pipeline behavior accurately in terms of gas accumulation. Finally, at higher air flowrates (about 40 Sm³/hr), flow changes to slug flow, where the maximum pressure difference during slug flow is less than that experienced during SS I, indicating that the riser is not completely filled with liquid.

During slug flow, the frequency of the pressure fluctuation is also dependent on the frequency at which slugs enter/ leave the riser, and the fluctuation magnitude is dependent on the slug size. As it can been seen, the OLGA model is also unable to predict slug flow in the pipeline-riser system, where it predicted bubbly flow in the riser rather than slug flow (characterized by stable riser-base pressure profile). The prediction of a bubble flow regime indicates that the flow regime transition mechanism in the OLGA code is not functioning in the region close to the riser-base. This is again attributable to inaccurate prediction of the pipeline gas accumulation behavior, corresponding to over-prediction of the inlet gas superficial velocities.

The OLGA model is moderately successful in predicting the flow regime in the pipeline-riser system during Severe Slugging I. However, in terms of the severe slugging pressure cycling characteristics, the model globally under-predicts the cycle time. In each case, the predicted liquid build-up and slug production times are different to those observed. The liquid slug build-up period is longer and the slug production period is shorter than experimental values.

In terms of fluid production, the trend associated with the pressure cycling is reflected in the liquid and gas production characteristics. Examining the predicted liquid production profiles during SS I for both Tests 1 and 2, there is a marked difference between the predicted and experimental liquid production profiles. Shorter cycle times in simulation results lead to higher instantaneous outlet liquid flowrate and smaller slug size than observed during experiments. It is also found that the simulations over-predict the peak liquid production rate during Severe Slugging I, which may be due to incorrect prediction of the rate of bubble propagation through the riser and the gas/ liquid interface position in the pipeline.

The OLGA model is also unable to predict the Severe Slugging II and slug flows, due to over-prediction of the bubble penetration at the riser-base. Similar to the liquid production profile, the model predictions of peak flow in gas production, for both cases, are vastly different from the experimental results. This is again attributable to inaccurate prediction of the riser-base bubble penetration, pipeline gas accumulation and the gas blowdown. Details of the simulation results can be found in a report by the author.⁹

CONCLUSIONS

Considering external factors such as calibration, and boundary condition effects in the experiments, the OLGA results compare rather well with the experiments.

OLGA modeling of pipeline-riser systems is able to predict the occurrence of classical SS1 in the test series. However, the detailed results show that the model does not give acceptable prediction of the characteristics of the observed classical severe slugging.

The OLGA model has difficulty in predicting the SS2 with respect to the bubble penetration at the riser base and the blowdown process. The model is also unable to predict slug flow in the pipeline-riser system, due to the presence of bubble flow at the riser-base. This is attributable to inaccurate prediction of the pipeline gas compression behavior, corresponding to over-prediction of inlet gas superficial velocities.WO

REFERENCES

   Manning, F. S. and R. E. Thompson, “Oil field processing, Vol. 2: Crude oil,” Pennwell Publishing Co., Tulsa, Oklahoma, 1995.
   Patel, M. H., and F. B Seyed, “Review of flexible riser modeling and analysis techniques,” Engineering Structures, Vol. 17, No. 4, pp. 293 – 304, 1995.
   Hatton, S. A., and H. Howells, “Catenary and hybrid risers for deepwater locations worldwide,” paper presented at: Advances in Riser Technologies Conference, Aberdeen, UK, June 1996.
   Sertã, et al., “Steel catenary riser for the Marlim field FPS P-XVIII,” presented at the Offshore Technology Conference, OTC 8069, Houston, May 6 – 9, 1996.
   API RP 17B, “Recommended practice for flexible pipe,” 3rd Edition, American Petroleum Institute, Washington, USA, March 2002.
   Sertã, O.B., “Riser concepts for mexican deepwater production systems,” World Oil, Vol. 225, No. 3, March 2004.

LITERATURE CITED

¹ Fabre, J., et al., “Severe slugging in pipeline/ riser systems,” SPE Production Engineering, Vol. 5, No. 3, pp. 299 – 305, Aug. 1990.
² Schmidt, Z., et al., “Severe slugging in offshore pipeline riser-pipe system,” SPE Journal, pp. 27 – 38, Feb. 1985.
³ Lunde, O., “Experimental study and modeling of slug stability in horizontal flow,” PhD Thesis, The Norwegian Institute of Technology, Trondheim, Norway, 1989.
⁴ Tin, V., “Severe slugging in flexible risers,” Proceeding of the 5th International Conference on Multiphase Technology, pp. 507 – 525, Cannes, France, 1991
⁵ Sarica, C., and J. Ø. Tengesdal, , “A new technique to eliminate severe slugging in pipeline/ riser systems,” Proc. SPE Annual Technical Conference & Exhibition, SPE 63185, 633 – 641, Dallas, Texas, Oct. 2000.
⁶ Yocum, B. T., “Offshore Riser Slug Flow Avoidance: Mathematical Model for Design and Optimization,” paper presented at SPE London Meeting, SPE 4312, London, UK, 1973.
⁷ Hill, T. H., “Gas injection at riser base solves slugging, flow problems,” Oil & Gas Journal, Vol. 88, No. 9, pp. 88 – 92, Feb. 26, 1990.
⁸ Montgomery, J. A., “Severe slugging and unstable flows in an s-shaped riser,” PhD Thesis, Cranfield University, Bedfordshire, England, Feb. 2002.
⁹ Mokhatab, S., “Interaction between multiphase pipelines and downstream processing plants,” Report No. 3, TMF3 Sub-Project VII: Flexible Risers, Transient Multiphase Flow (TMF) Program, Cranfield University, Bedfordshire, England, May 2005.
¹⁰ Peng, D. Y., and D. B. Robinson, “A new two-constant equation of state,” Ind. Eng. Chem. Fundam., Vol. 15, pp. 58 – 64, 1976.
¹¹ FEESA, “Life of field stability of a North Sea oil development,” FEESA Ltd Case Study, Surrey, UK, 2003.
¹² Nydal, O. J., M. Audibert and M. Johansen, “Experiments and modeling of gas-liquid flow in an s-shaped riser,” Proceeding of the 10th International Conference on Multiphase Flow (Multiphase ’01), Cannes, France, June 13 – 15, 2001.
¹³ Ek et al., “The field simulator-combined multiphase flow and process simulation,” paper presented at the 3rd North American Conference on Multiphase Technology, Banff, Alberta, Canada, June 6-7, 2002.
¹⁴ Bendiksen, K. H., Malnes, D., Moe, R., and S. Nuland, “The dynamic two-fluid model OLGA: Theory and application,” SPE Production Engineering, pp. 171 – 180, May 1991.

THE AUTHOR
	Saeid Mokhatab is a member of the International Advisory Board for EMERTEC Research and Development Ltd, in Halifax, Nova Scotia, Canada, as well as an advisor of natural gas engineering research projects in the Chemical and Petroleum Engineering Department of the University of Wyoming. Previously, Mr. Mokhatab worked as a researcher at Cranfield University, England, working on TMF3 Consortium Sub-Project VII: Flexible Risers. He has participated as a flow assurance consultant in various projects and has published more than 50 academic and industrial oriented papers and reports. Mr. Mokhatab served on the Board of SPE’s London Section during 2003 – 2006, and is currently a member of the ASME/ PSD Offshore Technical Committee, ASME/ OCT General Program Committee, ASCE Pipeline Research Committee, SPE and Sigma Xi.