A S Riser System With Viscous Liquids Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

In a pipeline riser system handling two phase flow, pressure and flow rate fluctuations may occur at relatively low liquid and gas flow rates. Such cyclical instabilities have been known as severe slugging. This investigation is an experimental study of severe slugging behaviour in pipelines - S riser system in a small-scale multiphase flow experimental facility. The experiments show the effect of the S-shaped riser on the severe slugging characteristics (period and pressure amplitude) as well as the effect of fluid viscosity (water and oil). The experiments results are compared with a slug tracking model.


Technology evolution has allowed advance in science development in many study fields, increasing not only the life quality of the human been but his capacity to face new and bigger challenges. A clear example of this enormous scientific and technological evolution is the oil industry field. Nowadays, the petroleum industry is provided with the highest technology that allowed produce oil from the most remote places and in the most complicated ways, from ultra-deep water production or extreme environmental conditions like North Sea or Sub Saharan Africa. Nevertheless, this quick progress has brought a faster descent of the hydrocarbon resources, pushing to the oil industry to find new source and new techniques that allow face the world energy demand.

Within this outlook, the interest and necessity of producing heavy and viscous oils have been increasing in the last years. However, transport of high viscosity oil may be one of the major challenges in the areas of flow assurance, especially for offshore fields (Wehunt, Burke et al. 2003). This means that the new developments should be focus in combine the common offshore troubles, with the consequence of handling viscous liquids.

A major subject in offshore production, from flow assurance point of view, is the multiphase flow phenomena, especially the slug flow due to its unstable behaviour. The severe slugging can be quite serious for offshore pipeline-riser systems because the slug sizes may be as large as the riser lengths. In this way, long slugs could be responsible of innumerable operation problems: overflow of downstream facilities, structural vibrations, pressure instabilities, production loss, among others (Nydal, Audibert et al. 2001)

In Norway, is considered heavy oils to oils with viscosities around 100 and 1000 cP, while viscosities between 1000 and 10000 cP are categorized as extra heavy.(Nossen 2012). In the literature is possible find experimental data for multiphase flow with viscosities around 50 cP, nevertheless, for higher viscosities, the accessible data is reduced and only a few authors have reported studies in this field (Nossen 2012).

Gokcal, Wang et al. (2008) studied the effects of high viscosity oil (0.181 Pa.s to 0.587 Pa.s) on flow pattern, pressure gradient and liquid holdup, for a horizontal pipeline. They found differences in flow behaviours between high and low viscosities oils. As remarkable conclusions they showed that the performances of some typical two phase models are not adequate for high viscosity flows.Gokcal, Al-Sarkhi et al. (2009) studied the effect on high viscosities (between 181cP and 589 cP) on slug frequency traveling in horizontal pipes. They found that the viscosity is a significant variable to consider in the slug frequency. Akhiyarov, Zhang et al. (2010) carried out several two phases experiments in order to study flow pattern, pressure gradient and oil hold up for a mineral oil with viscosities around 100 cP and 500 cP; they found significant effect of the oil viscosities in the gas dissolving phenomena. Gao, You et al. (2012) carried out a numerical simulation in order to study the flow behaviour in a pipeline riser system. The authors studied several combinations of gas - liquid; including water, crude oil and kerosene for the liquid phase, as well as air and methane for the gas phase. The research showed that the liquid properties like viscosity, density and surface tension had notable effect on the behaviour of severe slugging in the pipeline-riser system. Nossen (2012) studied the pressure drop in slug flow in pipelines under laminar flow conditions using heavy oil and gas. They proposed an unit cell model for steady state slug flow and compared it against published data for laminar slug flow with high viscosity mineral oil (178 -601 cP ) through a horizontal pipeline and found excellent agreement between both. Jeyachandra, Sarica et al. (2012) studied different viscosities between 181 cP and 585 cP in downward and upward pipe for +o- 2°inclination.

Since in the literature very little is reported on two-phase viscous flows and almost nothing regarding severe slugging phenomena with the viscous effect, the present research will be focus in severe slugging behaviour in pipelines S- riser considering the effect of fluid viscosity. The S shaped arrangement is often used to connect subsea pipelines and the surface facilities in order to reduce structural stresses, although this configuration may promote slug flow (Nydal, Audibert et al. 2001). According to Nydal, Audibert et al. (2001), it is possible identify three kind of flow in a pipeline S-shaped riser system: stable flow characterised by inlet pressure almost constant and two types of unstable flow, which can be classified as terrain slugging I, where the riser base is full blocked by liquid, and terrain slugging II, where the riser base is just partially blocked.

The main goal of this research work at NTNU is carry out a preliminary evaluation of the effect of the high viscous liquids in the severe slugging behaviour, especially in a S-shaped riser systems. The following report include some gas-liquid experimental data for two different liquid viscosities (water and oil) and the comparison of the experiments results with a slug tracking model developed at NTNU.


SDP Standard Deviation of Inlet Pressure

P Pressure

Usl Liquid Superficial Velocity

Usg Air superficial velocity


max Maximum

min Minimum

O Oil

W Water

Experimental Facility

The experiment was carried out at Multiphase laboratory of the NTNU, in a S-shaped riser system (Figure 1) of 14 m length and around of 6.43 m of total height, where include a first riser of 4 m height and a second riser of 3.7 m height like is shown in Figure 2. The line has 50 mm of internal diameter and it is constructed in an acrylic material for better visualization. The first section of the line has a downwards slope towards the first riser base in order to promote the terrain slugging generation.

Figure 1: S-shaped Riser at Multiphase laboratory of the NTNU

Figure 2: S-riser geometry (units in meters)

The system can be fed with air, water and oil through separated lines. Therefore, it is possible working with two or three phases. For this study only two-phase experiments were performed: air-water and air-oil. The single phase lines are connected to a manifold, where the phases are mixed previously to the downwards section inlet.

The inlet of the line, before the gas-liquid mixing, is connected to a buffer tank that simulates a large pipe volume upstream. In this experiment was used a volume of 0.255 m3, which is equivalent to around 130m of pipe length.

The whole system is equipped with different kind of instruments, control valves and pumps, which are connected to a DAQ system and controlled through a dedicated program written in LabVIEW®. The instrumentation in the S-riser include a pressure sensor in the buffer tank that measure the inlet pressure in the S-riser, three conductance probes located in the base and top of the first riser, and in the top of the second one, allow measure the liquid fraction for the air-water system (Figure 2).

Air Line

It is supplied for the central line of the laboratory to 7 bara and is reduced first to 4 bara to go into its feed line, and then is reduced again by a control valve before flow into the buffer tank. The air velocity is measured by means of a coriolis flow meter before the control valve.

Water Line

The water is stored in a tank (the oil-water separator) and supplied to the water line through centrifugal pumps or through screw pumps. In this experiment, the screw pumps were used in order to get more stable liquid flow since this kind of pump are less sensible to the pressure variations at the discharge produced by the severe slugging cycle. The liquid flow is controlled through a control valve and the velocity is measured by means of electromagnetic flow meters before the control valve.

Oil Line

The oil line arrangement is similar to the water line, the oil is stored in the same separator tank, and it is fed through similar centrifugal pumps. The default arrangement includes controls valve for regulate the amount of flow and coriolis meters for the velocity measurements. However for this study, since the flowmeters were out of the range for the operating conditions of the experiments, the velocity measurements were taken by means of the calibration of the screw pump frequency and the liquid flow rate supplied by the pump.

Flow test

Since the main goal of this study is the viscosity influence in the two-phase flow in a S-shaped riser, two cases were evaluated: air-water and air-oil. The water is tap water with a green colorant for better visualization. The oil is NEXBASE supplied by SINTEF laboratory. The properties and characterization of both liquids are showed in the Table 1

Table 1: Liquids properties


Density kg/m3 849 999 1.21

Viscosity cP 89 0.96 0.02

The experiment was developed under ambient conditions: outlet pressure 1.01bara and temperature around 21°C.


For a constant velocity of liquid, several air velocities were registered (Table 2), in order to determine the flow behaviour reported by Nydal, Audibert et al. (2001). The inlet pressure was monitored by means of the pressure transducer, and logged in LabVIEW, this allowed define the type of terrain slugging I, II or stable by simple observation of the pressure cycle (e.g. Appendix A and B). The sample rate was set to 100 ms.

Table 2: Case definition based in nominal superficial liquid velocity

Cases Name Usl (m/s)

CASE1 0.18

CASE2 0.30

CASE3 0.36

Numerical Method

Dynamic two-phase simulations are carried out using the dynamic slug tracking model SLUGGIT developed at NTNU. This model represents any elongated bubble as a region of stratified flow and slugs/plugs liquid blockages through which gas may not pass freely. A standard two-fluid model is applied to all stratified/bubble regions, while a mixture momentum balance governs fluid acceleration in slugs.

Model structure and slug tracking

At the heart of the SLUGGIT scheme lays the object oriented programme structure. The central notion is that of a non-structured numerical grid which adopts in-time to the solution development. Such a dynamic grid is achieved by constructing the numerical grid as interacting computational objects that will move, appear and disappear according to predetermined rules. Essentially, the numerical grid, or rather computational objects, is programmed to propagate in sync with slugs in a Lagrangian fashion. Slug front and tail propagation is again determined by the empirical correlations of Bendiksen (1984) the problem of numerical slug diffusion. Such a strategy for avoiding numerical slug diffusion is known as slug tracking

Figure 3 shows a schematic example of the SLUGGIT scheme in the case of terrain slugging. Here, momentum equations solved on the liquid phase will be gravity dominated, and so liquid will tend to trickle towards the pipeline dip/trough/low point/sag where it will accumulate. Once a liquid blockage of the pipe cross section has taken place, the offending stratified flow object will be replaced with a slug object, blocking the continued gas flow and causing upstream gas accumulation. This in turn causes pressure build-up, and finally the expulsion of liquid through the discharge of the slug object.

Figure 3 Schematic of the dynamic, numerical grid

Governing Equations

For briefness, the governing equations of the SLUGGIT method are presented in

Table 3

Table 3 Governing equations

A) Mass and pressure equations control volume

B) Momentum equation control volume

Mass conservation equation

Pressure equation

Momentum equation

Sub- and superscripts:

σ represents the borders, i.e., σ∈ {w,e}, where w is the "western" border and e the "eastern".

κ represents the phases, i.e., κ∈{g,l},where g is "gas" and l is "liquid". ↑κ↓ indicates "the other" phase

n the time step increment.

s denotes a source term.

((∙)) ̂ (hat) indicates an upwind value.

Symbols: M is mass, L the section length, V the section volume, u the phase velocity, v the grid border velocity, ρ the phase density, p the pressure, T the phase temperature, A phase or pipe area, ψ is a volume error correction term, S is a periphery length, λ is the Darcy friction factor, g is the constant of gravitational acceleration, ϕ is the pipeline inclination and n is the longitudinal component of the outward control volume unit normal vector.

The momentum equation for a slug takes the same form as in a stratified region, only for a single incompressible mixture phase.

Slug propagation relies on the empirical correlations for bubble nose propagation of Bendiksen (1984) Given that a slug-bubble border is identified as a bubble nose , border propagation is given by:

v_n=C_0 u_(s,l)+U_0

where v_n is the bubble nose border velocity, u_(s,l) is the liquid velocity in the slug and the factor term and drift velocity are defined by:

â- (â- (C_0=1.05 + 0.15 sin^2⁡2Ï• @U_0= 0.35√gD sin⁡ϕ + 0.54√gD cos⁡ϕ )&|u_(s,l)≤3.6 √gD/cos⁡ϕ [email protected](C_0=1.2 @U_0= 0.35√gD sin⁡ϕ )&|u_(s,l)>3.6 √gD/cos⁡ϕ ┤ )

Should the local pressure gradient be in the direction of the slug border outward normal the border will be a slug front, whose border velocity is determined by a mass balance on the liquid:


where α_(b,l) and α_(b,g)are the liquid and gas volume fractions in the bubble, respectively.

Simulation procedure

The simulation sequence takes the following form

Border velocities are determined form the Bendiksen (1984)and slug front criteria.

An implicit pressure-momentum system is set up across the entire pipeline wherein both phases, in bubbles and slugs, are coupled through interfacial friction and pressure.

This system is solved using a sparse linear solve. Pressures, velocities and positions are updated.

The mass equations are solved for each phase. This happens implicitly in bubble regions. Masses are updated.

A management procedure is run which takes care of the grid (object sequence) maintenance. Here fluid grid objects which have grown too long are split into smaller objects, while those shrinking below a given criterion will be merged together with their neighbouring objects. The nature of an object may here also be changed according the pre-imposed phenomenological models. For example, in the eventuality of hydrodynamic slugging a bubble object may be substituted with a slug object should it fulfil some given criterion, say the well-known Kelvin-Helmholts criteria for linear wave instability. Another example is a bubble being turned into a slug because its liquid fraction is approaching unity. This will be the dominating mechanism for slug generation in terrain slugging.

For more information on the SLUGGIT model see e.g., (Kjølaas 2007)

Results and Discussion

Experimental Results

The operating points studied are reported in Figure 4 for both cases air-oil and air - water.

Figure 4: Operating Conditions for Oil and Water for the three cases studied

The case W1 was run using a centrifugal pump, which is highly sensible to the discharge pressure variations. This means that the liquid flow rate is affected by the large pressure variations during the severe slugging cycle. Therefore, the slight variation of W1 for low air velocities in Figure 4 corresponds to instabilities in the pump during the cycle. The other cases were run using the screw pump that permitted to keep a more stable flow rate during the development of the experiments.

The comparison of the two fluids was based on a combination of visual inspection of the flow in the acrylic pipe and the inlet pressure recorded. Figure 5 to Figure 7 show the standard deviation of the inlet pressure (plot A) and the global maximum, and minimum inlet pressure (plot B) for the different cases studied. The pressure standard deviation was considered as indicative of the magnitude of the pressure fluctuation between the maximum and minimum pressures reached in the system.

Taking as reference the type of severe slugging defined by Nydal, Audibert et al. (2001), the stage for terrain slugging type I, type II and stable was easily identifiable in the water case, while for the oil cases, the three type of flow were not always present. For water cases, the amplitude attenuation with the increasing of air velocity was gentler that in the oil cases, because the gradually diminution of the liquid column in the riser. This would be the transition from slug type I to slug type II.

Standard Deviation of the Inlet Pressure Global Maximum and Minimum Inlet Pressure

Figure 5: Pressure Standard Deviation for Oil-Air and Water-Air at CASE1

For CASE1 was possible identify for both liquids the severe slugging effect. The SDP showed a decreasing tendency in direction of the increasing air flow velocity. Thus the larger the SDP, the larger the amplitude will be.

In oil cases, the slug type II was almost absent, showing deep changes between the unstable flow to stable flow. For higher liquid velocities (CASE3), the flow was stable for all air velocities and was not possible reduce the air velocity below 1.8m/s because the liquid penetrated completely the upstream section until to reach into the buffer tank.

Standard Deviation of the Inlet Pressure Global Maximum and Minimum Inlet Pressure

Figure 6: Pressure Standard Deviation for Oil-Air and Water-Air at CASE2

Standard Deviation of the Inlet Pressure Global Maximum and Minimum Inlet Pressure

Figure 7: Pressure Standard Deviation for Oil-Air and Water-Air at CASE3

In Figure 5, Figure 6, Figure 7 is also possible to note that for the same type of flow, the maximum pressure for the oil was a little higher than in the water cases. Moreover, the minimum pressure parameter was always higher for the oil than for the water. This could be related with the different levels of liquids in the pipe cross section that for the same air and liquid velocity, the oil showed some higher liquid level in comparison with the water. The liquid level in the pipe for both liquids is showed in Figure 8.

Water Oil

Figure 8: Liquid Level at Usl=0.18 m/s and Usg = 2.43 m/s. Picture taken close to the mix section.

Figure 9 shows the slugging period registered at conditions of CASE1 for both water and oil. For low air velocities large time periods were observed, which correspond to the time that take the liquid to reach the top level of the riser and for the air get enough pressure to raise the liquid. Experiment results did not show a conclusive relation with the viscosity increment and the cycle period. The oil behaviour was disordered and did not follow a consistent pattern.

Figure 9: Period of the Inlet Pressure for the Severe Slugging Cycle for Oil-Air and Water-Air at CASE1

Figure 10, shows the liquid level inside the pipe at the operation condition Usl = 0.30 m/s and Usg = 4.68 m/s that corresponds to stable flow for both liquids. In the water case was observed a continuous flow without any perturbation in the entrance, hydrodynamic slug flow only was observer throughout the riser. Instead, for the oil (Figure 10 B), the hydrodynamic slug was not only present all the way through the risers but also in the entrance pipeline, close to the mix section.

Water: stable flow at the entrance Oil: Hydrodynamic slug at the entrance

Figure 10: Entrance Flow at Usl=0.30 m/s and Usg = 4.68 m/s. Picture taken close to the mix section

Simulation Results

Air -Water

The Figures from Figure 24.A to Figure 26.A show time trace plots of the pressure development at the inlet in simulations and experiments for air and water. From Figure 24.A, Figure 25.A and Figure 26.A, it is apparent that the simulations are in good agreement with experimental results for low and intermediate superficial gas velocities (Cases 1-10.) Simulation pressure time traces for intermediate to high gas velocities appear to give multiple-frequency signals, or signals with random switching between terrain slugging and bubbly flow pressure profiles. The low frequency part of these signals may not be completely contained within the time span of the experimental data, as can be seen e.g. in Cases 6-7 and 10-13 in Figure 25and Cases 14-17 in Figure 26.A.

Figure 11 to Figure 13 show the standard deviation of the pressure signals of experimental and simulated data. A 50 second delay has been set on the simulation data in order to dismiss start-up influence. Again we see good correspondence in the low and intermediate superficial gas velocity range. The same can be said for the dominant pressure oscillation periods presented in Figure 15and Figure 16. These data are compiled applying a standard discrete Fourier transform on the pressure signal. Correspondence between experiments and simulations are striking.

At higher gas rates simulated pressure profiles tend to include both the terrain slugging character of low gas rates and the high frequency noise of bubbly gas penetration in the dip. Experimental data appear to only contain the latter at high gas rates. Whether the terrain slugging part of the simulation profile is purely sporadic or has itself a low-frequency periodic character cannot be said from the limited time interval presented here.

The strange peak in Figure 15 at Usg 4.11m/s is caused by one of these sudden returns to terrain slugging in the simulation, in this gas at the very end of the simulation interval (see Figure 25.A.) This one burst is here enough to make the slugging frequency dominant because the pressure signal is otherwise near-constant in amplitude.

Figure 11 Standard deviation of inlet pressure, air - water. CASE1, cf. Figure 24.A

Figure 12 Standard deviation of inlet pressure, air - water. CASE2, cf. Figure 25.A

Figure 13 Standard deviation of inlet pressure, air - water. CASE3 , cf. Figure 26.A

Figure 14 Dominant pressure oscillation period air - water, CASE1 cf. Figure 24.A

Figure 15 Dominant pressure oscillation period air - water, CASE2 cf. Figure 25

Figure 16 Dominant pressure oscillation period air - water, CASE3 cf. Figure 26.A

Interpretation of occasional terrain slugging simulation behavior at high gas rates

Figure 17 presents animation snapshots of the simulation point [Usl=0.36 m/s,Usg=4.30 m/s, air-water] (case 16, Figure 26.A.) Animations are generated using PLOTIT (Novak 2012). Image pipe diameter has been scaled by a factor 10 for visibility. The top left image shows the usual mode of liquid production through the generation of short slugs being pushed through the S. However, at the time t=193 s (second image from the left), a stratified section is pushed partially through the dip without forming a slug object. This liquid the falls back to merge with the succeeding liquid being accumulation (3rd image.) Given a suitable match in frequency, this produces a liquid blockage sufficient for a break-down of the "bubbly" mode of gas penetration and a return to the terrain slugging pattern characterized by blow-out events.

Figure 17 Animation snapshots of case 16, Usl 0.36 m/s. Times of snapshot (from top left to bottom right): t∈{171.0,193.0,194.5,198.5} seconds

Tests with increased grid resolution and a lower critical holdup criterion all reduce to amount of terrain slugging patterns at high gas velocities, indicating that this is a numerical phenomenon related to the computational grid and tuning parameters. At the highest gas rates, the effect of refining the grid resolution on numerical fluctuations is diminishing. The most reasonable explanation for this is the quick, small scale bubbly flow present in the dip at high gas rates is poorly represented by the core model approximation of purely stratified and plugging fluid elements. Incorporating gas entrainment features into the slug objects, or some elements of unit cell modelling, may go some way in improving the model response to bubbly flow situations.

Air - Oil

Figure 27.B to Figure 29.B show time trace plots of oil and air for the superficial oil velocities for CASE1, CASE2 and CASE3, respectively. It is apparent that the simulations do manage to replicate the sudden drop in pressure oscillation amplitude reasonably, though the oil rate point of decrease is not always matched perfectly. Again we also see the tendency for the simulations to predict minor high-frequency oscillations at high gas rates

Standard deviation plots similar to those reviewed earlier in Figure 14- Figure 16 are presented in Figure 18- Figure 20. The main sources of discrepancy are the slight mismatches in the critical gas rate at which oscillations suddenly die out, and the numerical noise at high gas velocities.

Figure 20 - Figure 23 display the dominant oscillation periods of air and oil similar to that seen for air and water in Figure 14- Figure 16. Compared with the plot's respective time traces (Figure 27- Figure 29) these Fourier analyses are obviously flawed as the dominant periods of oscillation are far greater than the signal noise. The simple reason for this is that the noise, particularly in the experimental data, is so low in amplitude that the wave mode associated with some other random frequency is picked up instead.

Figure 18 Standard deviation of inlet pressure, air and oil. CASE1, cf. Figure 27.B

Figure 19 Standard deviation of inlet pressure, air and oil. CASE2, cf. Figure 28.B

Figure 20 Standard deviation of inlet pressure, air and oil. CASE3 , cf. Figure 29.B.

Figure 21 Dominant pressure oscillation period oil - air, CASE1 cf. Figure 27.B

Figure 22 Dominant pressure oscillation period oil - air, CASE2 cf. Figure 28.B

Figure 23 Dominant pressure oscillation period oil - air, CASE3 cf. Figure 29.B.


Some experimental data were reported and compared with slug tracking models for two different viscosities: 89cP (NEXBASE Oil) and 0.96 cP (Water) in a S-shaped riser system.

The experimental results suggest that the stability map for severe slugging (applied to S-risers) could present notables variations for high viscous flow, especially in the unstable flow boundaries, however further investigations would be necessary in order to confirm this observation.

Comparison between the experimental data and the slug tracking model SLUGGIT showed a fairly good approximation of the model with the physical phenomena registered in the experiments.