Challenges in slug modeling and control for offshore oil and gas productions: A review study

International Journal of Multiphase Flow 88 (2017) 270–284

Contents lists available at ScienceDirect

International Journal of Multiphase Flow journal homepage: www.elsevier.com/locate/ijmulflow

Challenges in slug modeling and control for offshore oil and gas productions: A review study Simon Pedersen∗, Petar Durdevic, Zhenyu Yang Department of Energy Technology Aalborg University, Esbjerg Campus Niels Bohrs Vej 8, Esbjerg 6700, Denmark

a r t i c l e

i n f o

Article history: Received 15 February 2016 Revised 5 July 2016 Accepted 14 July 2016 Available online 14 September 2016 Keywords: Offshore Oil & gas Multi-phase flow Bifurcation Anti-slug Gas-lifting Riser slug Flow control Stabilization

a b s t r a c t The upstream offshore multi-phase well-pipeline-riser installations are facing huge challenges related to slugging flow: An unstable flow regime where the flow rates, pressures and temperatures oscillate in the multi-phase pipelines. One typical severe slug is induced by vertical wells or risers causing the pressure to build up and hence originates the oscillating pressure and flow. There exist many negative consequences related to the severe slugging flow and thus lots of investments and effort have been put into reducing or completely eliminating the severe slug. This paper reviews in details the state-of-the-art related to analysis, detection, dynamical modeling and elimination of the slug within the offshore oil & gas Exploration and Production (E&P) processes. Modeling of slugging flow has been used to investigate the slug characteristics and for design of anti-slug control as well, however most models require specific facility and operating data which, unfortunately, often is not available from most offshore installations. Anti-slug control have been investigated for several decades in oil & gas industry, but many of these existing methods suffer the consequent risk of simultaneously reducing the oil & gas production. This paper concludes that slug is a well defined phenomenon, but even though it has been investigated for several decades the current anti-slug control methods still have problems related to robustness. It is predicted that slug-induced challenges will be even more severe as a consequence of the longer vertical risers caused by deep-water E&P in the future. © 2016 Elsevier Ltd. All rights reserved.

1. Introduction In recent years the production optimization for offshore oil and gas Exploration & Production (E&P) facilities has been extensively investigated, as any potential enhanced fuel recovery can result in huge economic gains (Havre et al., 20 0 0). Fig. 1 illustrates a typical well-pipeline-riser section at a typical offshore oil & gas field. This specific system consists of three connected subsections: 1. The production well section; where liquids, gases and solid compounds from the reservoir flow through a vertical well. Some production wells use artificial lifting techniques to help keep reasonable production rate from the reservoir. In some constructions the well head goes above sea level to a manifold platform, where the flows from several wells can join into one stream and further move forward into a single pipeline. In most cases topside choke valves located at the top of each well are available to regulate the flow through the tubing of the production well. ∗

Corresponding author. E-mail addresses: [email protected] (S. Pedersen), [email protected] (P. Durdevic), [email protected] (Z. Yang). http://dx.doi.org/10.1016/j.ijmultiphaseflow.2016.07.018 0301-9322/© 2016 Elsevier Ltd. All rights reserved.

2. The subsea transport pipeline section, which consists of a transportation pipeline that follows the sea bed. This section consists of the majority of the complete pipeline length. 3. The vertical riser section; where the riser raises the well fluids from the subsea transport pipeline up to the topside platform above sea level, where a separation process separates the multi-phase (gas/oil/water) fluids. As with the well section, the riser sometimes uses artificial lifting at the riser base to improve the production rate (Hu, 2004). A topside choke valve is often placed before a separator to regulate the flow fed into the separator. It can be observed that nearly the entire pipeline section is placed subsea. As all maintenance of subsea equipment is both very expensive and time-consuming the number of pipelines, actuators, transmitters and separation equipments are very limited before the separation platform (Baardsen, 2003). This is the same reason the multi-phase fluids in most cases are not separated on the well manifold platforms. At deep oil & gas reservoirs the changes in temperatures and pressures to the wells become significant and cause several production challenges (Arnold et al., 1972), such as: Increased likelihood of wax and hydrate formation, internal mechanical loads, po-

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

271

Fig. 1. An illustration of a typical well-pipeline-riser system, which consists of: (i) the well, (ii) the transportation pipeline, and (iii) the riser.

tential creep problems and increased corrosion (Bai and Bai, 2014; Hassanein and Fairhurst, 1998; Montgomery, 2002; TTA-Group, 2008). These more complicated and expensive conditions for new explorations cause the industry to pay more focus on maximizing the oil & gas production rate by increasing the recovery on existing well-pipeline-riser systems (Rassenfoss, 2014). Implementation of automatic control solutions have proved to be able to improve the oil recovery significantly (Bailey et al., 2014; Foss, 2012). This paper will examine the problems that arise when severe slug flow is present in the well-pipeline-riser process. This study will investigate the physical properties defining the severe slug and the corresponding negative consequences in Section 2, examine the mathematical slug criteria in Section 3, present control-oriented models in Section 4, and the methods developed to avoid or eliminate the slug in Section 5. Other review studies examining the slug flow can be found in Foss (2012); Meglio et al. (2012a); Mokhatab (2010); Mokhatab and Poe (2012); Mokhatab and Towler (2007); Pedersen et al. (2015a). 2. Properties of severe slugs Gas-liquid multi-phase flow in a pipeline can take a large number of possible shapes with subject to the specific flow conditions. However, these forms can be classified according to different types of fluid distributions; commonly called flow patterns or regimes. The work in Taitel et al. (1980) described the most typical multi-phase gas-liquid flow patterns in vertical pipelines: Bubble flow where there exist a dispersion of gas bubbles within the liquid, slug flow where larger bubbles fills the diameter of the tube causing gas pockets, churn flow where the slug bubbles have broken down to give a churn pattern, and annular flow where the liquid flows on the tube wall and the gas in the middle of the tube. The flow pattern is determined by a number of parameters, such as the ratio of gas-to-liquid, the flow rates, the liquid and gas material characteristics, and the materials, and diameters and shapes of the pipelines. In reality the multi-phase fluids might also consist of solid components, such as sand, however the dominant phases are generally gases and liquids. At installations where the solid components are causing significant negative impact to the daily production, de-sanding separation of the solid components are taking place early in the process, e.g. at the wellhead FMC Technologies (2010); Whitney and Larnholm (2015). The slug flow is a common flow pattern in the oil & gas production process. The gas (natural gas) and liquid (mainly consisting of water and oil) may not be evenly distributed throughout the

wells, transportation pipelines and risers under certain operating conditions. Sometimes the gas travels in large plugs through the pipeline. This phenomenon is referred to as slug (Biltoft et al., 2013). The slug regime can be caused by the physical construction in both the horizontal and vertical pipelines, due to transient response related to pigging, start-up, shut-down, or changes in the set-points of pressures or flow rates (Sivertsen et al., 2010). An example is a hilly terrain on the seabed which can cause terraininduced slugs, see Al-safran et al. (2004). Severe slug can occur when multiphase flow passes through a vertical well or riser. It has to be noted that severe slug is a loose definition as there is no strict definition differing severe slugs from ordinary slugs. However in general process engineers define the severe slug as an increased amplitude in the pressure and flow oscillations, hence minimizing the pressure or flow amplitude oscillations will reduce the severe negative impact from the slug (Pedersen et al., 2014b). An example of severe riser-induced slug from gas-holdup can be observed in Fig. 2a, where the periodic severe slugging behavior is induced at the low-point connection between the inclined horizontal pipeline and the vertical riser. This slugging behavior can be divided in 4 sequential phases: (1) Liquid accumulates at the riser base due to the lack of capability to lift the dense liquid the entire riser length immediately. (2) The gas is blocked by the liquid at the riser base. When more fluids enter the pipeline, the bottom pressure will increase and the riser section will be filled with liquid. (3) When the blocked gas has accumulated to overcome the hybar drostatic pressure (0.06 − 0.11 meter for liquids only), the gas blows the liquid out of the riser. (4) After the liquid blow-out, a gas surge flows through the riser while the remaining liquid in the riser falls back and begins to build-up in the riser base once again. This cycle repeats. Fig. 2b shows the riser bottom pressure during the 4 steps of one severe riser-induced slug cycle. The data is obtained from Aalborg University’s first generation laboratory test rig. As the gas pressure in the riser base needs to overcome the hydrostatic pressure of the liquid in the riser, the riser length is a key parameter to determine the slug’s magnitude. As the offshore risers can extend 200 m from subsea to a separation platform (Dansk Undergrunds Consortium, DUC, 2012), the slug magnitudes can be significant. Similar slug behavior can be observed in the gas-lifting production well due to the “Casing-heading” mechanism. Here the gaslifting causes pressure build-up in the casing with no production, until a blow-out phase occurs due to intermittent gas injection rate from the casing to the tubing (Sinègre et al., 2005). Fig. 3 shows an illustration of the gas-lifting production well. Two choke

272

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

(a) Illustration from Biltoft et al. (2013) of the cyclic behavior in a pipeline-riser when severe riser-induced slugs are present. The cycle consist of 4 steps: Slug formation, Slug production, Liquid blowout, and Gas surge.

(b) The bottom pressure during one severe riser-induced slug cycle is illustrated and divided into 4 steps. Based on laboratory data from Aalborg University Esbjerg’s first generation slug test rig. Fig. 2. The 4 steps of each severe riser-induced slug cycle.

Production valve wm,out wg,in

Gas lift valve

Annulus : Multi-phase : Gas-phase

Tubing

Check valve

w m,in Reservoir Fig. 3. Illustration of the gas-lifting well where the gas injection into the annulus is controlled by a gas lift choke. The annulus is connected to the tubing through a mechanical check valve, and a production choke valve is placed downstream the tubing.

valves are used for control: One for the gas injection into the annulus, and one for the production control located downstream the tubing. Besides that a check valve is located on the connection between the annulus and the tubing. The work examined in Xu and Golan (1989) described the gas-lift casing-heading repeating cycle, similar to what can be observed on each cycle on Fig. 4: (1) At the annulus connection to the well tube a sudden reduction of flowing tubing pressure (indicating the static pressure in the bottomhole of the tubing) results in more gas injected through the downhole orifice to the tubing. (2) More gas discharge will further reduce the flowing tubing pressure, promoting more gas flowing through the downhole orifice. (3) As the gas supply through the surface choke cannot deliver the increased gas rate injection into the production tube, the casing pressure and the upstream pressure at the downhole orifice will be reduced. (4) The flowing tubing pressure starts to increase. This causes the reduction of gas injection into the tubing to continue and results in a decrease of gas flow into the tubing. (5) The high flowing tubing pressure and low annulus pressure cause the downhole orifice to inject less gas than the gas injection can supply. This results in the casing pressure to build-up. (6) As the casing pressure builds up the gas injection rate into the production tube increases as well and reduces the tubing flowing pressure. These steps repeat for each slug cycle. Fig. 4 from Jepsen et al. (2012) shows slug data from an offshore well in the North Sea. The top graph shows the topside temperature in the well, the picture in the middle shows the well bottom and topside tube pressure, and the bottom picture shows the gas injected into the annulus. The topside control valve is fully open under the entire test and the downstream back pressure is almost constant during the testing period. The pressure and temperature data show significant oscillations and thus it is clear that the actuators are not controlled to handle the slug. Hence in this case the gas-lifting controller aims for a constant gas inflow rate and not considering the possible change in flow regimes, and the topside choke valve controller is only used for start-up, shut-down and safety control. The operator has specified safety boundaries for both maximum and minimum allowed pressure and temperature, at which he has to shut-down the process by choking some safety valves. In this case both the temperature and pressures’ peaks are on the edge of the safety process boundaries. It is important to notice that the “unstable flow” term is not an unstable system although the linearized model poles can indicate this Pedersen et al. (2014a), but is actually more linked to limit cycles due to the system’s closed trajectories. However the transition to slug flow from another flow regime can be characterized as an unstable system transition as the system does not stay in the original flow pattern during the influence of disturbances. The multiphase instability transitions can be rather unpredictable for the flow assurance operator as the flow regime is not constant during varying running conditions. It should also be noted that slug flow is not the same as flow instability because some flow instability scenarios can occur even when the slug flow is not present. The study examined in Hu and Golan (2003) concluded that flow instability (limit cycles) can occur in an oil & gas well even if the casing-heading slugs are eliminated. An example is the “density wave instability” which can occur even if the gas injection rate is constant at the bottom of the well (Plucenio et al., 2012; Sinègre et al., 2006a). It occurs when the reservoir pressure is not high enough to overcome the hydrostatic head in the riser and the gas rate is too low to lift the liquid. Then the gas phase will flow through the liquid in the tube and release at the top of the well. The liquid level may be oscillating within the tubing, but no production is observed at the wellhead. By increasing the gas flow rate to a certain level, a burst-like liquid production will occur out of the well.

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

273

Fig. 4. Data from one offshore well slugging in the North Sea. The top graph shows the topside temperature in the well, the picture in the middle shows the lowpoint and topside pressure, and the bottom picture shows the gas injected into the annulus. The topside control valve is fully open under the entire test. Modified from Jepsen et al. (2012).

The negative consequences of severe slugging flow on the production facilities are quite extensively studied in Hill and Wood (1994); Yang et al. (2013). The studies conclude that the main negative impacts of the slug are as listed below: Liquid overflow and high pressure in the separators. The liquid blowout results in varying flow inputs to the separator. Even though the separator acts as a buffer tank, large slug blowouts can result in bigger liquid volumes than the buffer tank can handle. The study examined in Yang et al. (2010) observed that a varying input flow can affect a controller’s ability to reduce the separator output flow and hence reduces the separation efficiency. Furthermore the study from Husveg et al. (2007) proved that a poor separation in the separator will affect the performance in the rest of the separation process. Thus handling the slugs upstream the separator would be preferable as the slug reduces the efficiency of the oil-water-gas separation ultimately resulting in a limited production rate, reduced production quality, as well as difficulties and challenges for the produced water treatment. The study in Wilhelmsen (2013) proposed control methods for 3-phase separator outlet valves to handle the large slug disturbances to the separation process. Overload on gas compressors. The equipment’s handling capacity can overload as the slugs often produce a much larger pressure and flow rate than the equipment is designed for. This can especially be a problem for the compressors which are not designed to handle that high pressure rates.

Fatigue caused by repeating impact. The oscillating pressure can shorten the pipelines lifetime due to the extra fatigue load (Hill and Wood, 1994). Increased corrosion. It is well-known that the slug can accelerate the corrosion in the pipelines (Kang et al., 1996; Sun and Jepson, 1992; Zhou and Jepson, 1994). The high flow rate oscillations cause increased friction in the pipelines. The high frictional pressure drop results in increased shear stressing to the wall (Hill and Wood, 1994), which ultimately increases the corrosion rate. Low production. The average daily production rate is significantly reduced, both due to the increased friction in the liquid blowout stage and in the liquid fallback stage where the production rate is transitory negative. Many studies such as Isaac et al. (2011) have shown that there is potential to increase the production by eliminating the slugging flow using feedback control. The production loss can also be caused by emergent shut-off of production. Production slop. The slugs can cause the produced natural gas to be flared as waste by a gas combustion device to secure a safe pressure. This problem arises when the gas surges result in increased pressure over the safety level. As a consequence flaring of the gas is applied to handle the big amounts of gas in the separator. Thus it is clear that avoiding the slug is necessary for safety and economic interests. The slugging flow regime can be avoided by

274

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

changing the operating conditions, hence it is important to know which flow pattern occurs under which running conditions. Traditionally, flow maps have been used for this purpose. Flow maps have been used for many decades, see Hewitt and Roberts (1969), and is still being applied in research today (Li et al., 2013). The maps normally indicate the relationship between the superficial velocities of gas and liquid, and the corresponding flow pattern. The maps can be created for each concerned system based on experiments. The flow maps have proved that the severe slugs occur when both the superficial fluid velocities are low. This was also concluded in Xiaoming et al. (2011) where it also was experimentally observed that by increasing the gas or liquid superficial velocity the riser bottom pressure’s fluctuating amplitude will initially increase due to the associated increased friction. However, by the end this pressure will decrease causing a steady non-slugging flow. Even though flow maps can determine at which conditions the slugs will occur, the acquisition of these maps may require exhaustive tests. Besides the studies on flow maps, some studies have focused on the analysis of the slug properties; the work described in Sagatun (2005) developed a simple model to determine the slug build-up time as well as the entire slug period, and the work in Malekzadeh et al. (2012) made experimental, theoretical and numerical tests to analyze the physics behind the slug cycles. 3. Slug criteria and detection Slug criteria have traditionally been used for the pipeline and equipment construction designs to handle the slug problems already in the design stage. The main advantage of using slug criteria is that they can be computed fast. The work described in Taitel and Dukler (1976) gave a general criterion for stratified flow in horizontal pipelines, based on the maximum allowed superficial gas velocity for stratified flow. The industry generally used this method in combination with the studies of Schmidt et al. (1980), who described the occurrence of stratified flow in a horizontal pipeline as a requirement before severe slugging in a pipeline-riser system could take place. The Bøe criterion, presented in Bøe (1981), extended the criterion from Taitel and Dukler (1976) further with the assertions from Schmidt et al. (1980) based on ideal gas law: The rate of gas accumulation at the riser base must be greater than the rate of pipeline gas pressure increase for a severe slug to form in the riser. The pressure increase is caused by the liquid column build-up in the riser during slug and consequently increasing the hydrostatic pressure too. A similar criterion was studied by Pots et al. (1987), which was also based on the buildup of gas pressure in the riser and the accumulated hydrostatic head in the riser. This study also introduced a quantitative index to denote the degree of severe slugging to quantify the severity of a given slug. The work described in Taitel et al. (1990) discussed the definition of slug, as Bøe criterion assumes that if the liquid column is constant with no change in the running conditions, a constant steady state is assumed to exist. However, this assumption was disproved by the work in Taitel et al. (1990) where it was observed that a cyclic process still can exist even when the liquid column is constant. Another study (Jansen et al., 1996) concludes that Bøe Criterion is only valid when no elimination methods are applied, which gives limitations for the practical use of it. Another criterion was proposed by Taitel (1986); Taitel et al. (1990). This detection method is based on the gas holdup at the riser base; the void fraction of a Taylor bubble that penetrates into the riser during slugs. A Taylor bubble (also called gas slug) is the large asymmetric bullet-shaped bubble under gas-liquid multi-phase flow, which occupies almost the entire cross-section of the pipeline and has a length several times larger than the pipeline diameter (Liao and

Zhao, 2003). This criterion only determines the severe slugging from the superficial liquid velocity and is hence useless when the gas velocity has to be considered. In Fuchs (1987) a criterion was developed considering the severe slug’s release time. The release time is the time when the slug tail moves from the horizontal pipeline to the riser leading to the gas surge phase of the slug cycle. This physical behavior however is not unique for the severe slugging and can not always be distinguished from hydrodynamic slug or bubble flow. The work in Jansen et al. (1996) introduced a modified version of the criterion from Taitel et al. (1990), where valve choking and gas lifting are included; however only with constant values. In Asheim (1988) two factors was introduced for detection of gas-lift instabilities. Stable flow is ensured if at least one of the factors are above a value of 1. In recent years other methods have also been used for specific subsections of the well-pipelineriser constructions, such as: Vertical well casing-heading instability (Fairuzov et al., 2004; Mahdiani and Khamehchi, 2015), and slugs induced by S-shaped risers (Montgomery, 2002; Tchambak, 2004). A newer topic is to detect the slug regime online for controller decision making. In Pedersen et al. (2014a) two simple separate criteria were provided and designed for supervisory online detection as part of a supervisory anti-slug control scheme. The study proposed two new slug criteria for real-time detection of the slug regime: (i) The first criterion is based on the bottom pressure’s changing rate over time and is in other words increasing proportional to the frequency of the pressure oscillations. A threshold was introduced for determining for the minimum frequency for the severe slugs to occur. (ii) The second criterion was based on the pressure drop over the riser, as the hydrostatic pressure in the riser changes rapidly when slug exist. Both principles require a riser low-point pressure measurement, although the first criterion also could work for the topside pressure measurement. As the slug criteria in general only are applied in the early design stage the methods can only predict the steady-state flow regimes under many assumptions, such as constant inlet flow rates and no applied closed-loop scenarios with sensor information and feedback controllers. Online slug detection methods do exist but slug prediction for controller design is still in a premature stage. 4. Dynamic models The dynamics of flow in multi-phase pipeline systems have been investigated for many years and is still an actively discussed topic. The early studies focused mainly on the simulations of the steady-state performance of the process, see Taitel et al. (1980); Viggiani et al. (1988). Since the early1990’s the focus changed to the simple transient models. The work in Sarica and Shoham (1991) presented a transient model for pipeline-riser systems based on one-dimensional gravity-dominant flow in both the pipeline and riser. The model was tested against experimental data from a testing facility constructed in Vierkandt (1988). The results showed that the model could satisfyingly predict pipeline pressure transients, liquid accumulation, slug length, and cycle time for all flow conditions tested. Knowledge of the detailed physical size and dimensions are required for usage of the model, as the pipeline inclination angle was proven to be very sensitive to uncertainties. Furthermore the model was compared to a model developed in Jansen (1990). It was concluded that the model examined in Sarica and Shoham (1991) could successfully predict severe slug better than previous studies, such as Bøe (1981). However, the results also showed that during some slug scenarios the model suffers from non-convergence, possibly because the flow is not as gravity-dominant anymore in the applied momentum equations. Thus the model’s accuracy is questionable inside the slug flow region. All these older models’ limitation is that they were not developed for anti-slug control but

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

rather to simulate the flow in the pipelines, and thus not validated with manipulated variables in closed-loop. In recent years commercial computational multi-phase flow simulation programs such as SPT group’s OLGA (Bendiksen et al., 2013) and Kongsberg’s LedaFlow (Danielson et al., 2011; Kongsberg, 2014; McArdle, 2014) have been used to compare the different low-dimensional control-oriented models. Both software simulators are based on continuity equations and have been verified based on data from a large-scale testing facility, a SINTEF Multiphase Flow Laboratory at Tiller (SINTEF, 2014), located in Trondheim, Norway. In Belt et al. (2011) a detailed comparison was made between OLGA 5.3 and LedaFlow based on tests on several platforms, both from laboratory and field data. Both OLGA and LedaFlow performed equally good in relative simple cases, but also equally poor in complicated cases where the gas phase is dominant and as a consequence the multi-phase flow is a varying mixture of churn and slug flow. The study concludes that OLGA 5.3 and LedaFlow predictions are of the same level in general (Belt et al., 2011). Many of the control-oriented models presented have been compared to either OLGA or LedaFlow simulations. 4.1. Control-oriented models Control-oriented models describe a wide branch of relatively simple models which can be applied in numerous control design schemes where the model is included in the control development (and sometimes in the implementation as well) to archive acceptable reference tracking performance while guaranteeing stability. In Storkaas et al. (2003) a low-dimensional differential-algebraic model was presented for severe slugging in a pipeline-riser based on mass balance equations. The mass conservation equations are shown in equation (1) where the total liquid change is equal to the difference between the injected and outlet flow rates, equation (2) for the gas mass change in the horizontal pipeline, and equation (3) for the gas mass change in the riser. The model also uses a valve equation similar to equation (4). The model was compared to OLGA simulations and to the scaled medium-sized testing facility constructed by SINTEF, where both comparisons showed good consistency with the model and the data. As the main goal of this model was to apply it for anti-slug control designs, it was required that it was able to not only describe the (undesired) slug behavior, but also describe the desired flow regimes. This was achieved with acceptable success, and several anti-slug controllers were designed based on this model (Sivertsen and Skogestad, 2005; Storkaas, 2005).

d mL = wL,in − wL,out dt

(1)

d mG1 = wG,in − wG1 dt

(2)

d mG2 = wG1 − wG,out dt

(3)

A similar model, which also was based on mass balance equations and the detailed physical structure of a pipeline-riser facility was studied in Jahanshahi and Skogestad (2011), where a 4 state model was developed. Fig. 5 illustrates this modeling principle, where the system is divided into two subsystems; one describing the horizontal and inclination pipeline, and one describing the vertical riser. Each subsystem is modeled with two mass equations, one for the mass of gas, and one for the mass of the liquid; thus the 4 states are carried out based on the mass balance equations. The figure also shows the switching mechanism leading to riser-induced slugging, which indicates if liquid is blocking the gas. Besides proposing this new model, the study compared the

275

new model and other low-dimensional control-oriented models with the OLGA. All the models compared in Jahanshahi and Skogestad (2011) are using the mass balance equations. The comparison include the following models: The model in Storkaas et al. (2003) which already has been examined, a 4state dynamic model (Eikrem, 2008) extended from a model developed in Eikrem (2006), a simple 3-states model (Kaasa and V. Alstad, 2008) heavily depending on 7 tuning parameters, another 4-state model (Silva and Nydal, 2010), and a 3-state model (Meglio et al., 2009) based on the principle of a virtual valve. In Jahanshahi and Skogestad (2011) it is concluded that models with many tuning parameters such as the models in Kaasa and V. Alstad (2008) with 7, and Meglio et al. (2009) with 5 tuning parameters, can obtain similar results as with the OLGA simulations. However the many tuning parameters indicate an unclear physical understanding of the slugging principle in addition to the problems linked to a high amount of tuning parameters. The models from Eikrem (2008) and Silva and Nydal (2010), both with 3 tuning parameters, do not match the OLGA simulations as good as the other models. This was also concluded in Jepsen et al. (2013), where a controller was developed based on the model examined in Eikrem (2008); it was hard to make the model fit the real data. In Jahanshahi and Skogestad (2011) the work concluded that their proposed model and the more simple model developed in Meglio et al. (2009) seemed to be the best, considering the trade-offs between complexity and the number of tuning parameters. As the model developed in Jahanshahi and Skogestad (2011) was limited to the pipeline-riser construction the model was extended to a 6-state model in Jahanshahi (2013) for modeling the entire well-pipeline-riser system. However, in this extended model the average mass ratio of gas and liquid produced from the reservoir is assumed to be a known parameter, which often is unknown in reality. The extended model was compared to OLGA simulations and it was concluded that it could predict the steady-state and the bifurcation point for the choke valve with a good accuracy. In Jahanshahi and Skogestad (2014) more simulation results with the model was shown and a tuning guide for the four adjustable parameters was developed. The tuning was based on the tuning parameters’ impact on the model’s physical equations; a correction factor for the liquid level in the pipeline, a production valve constant for flow estimation, and the coefficients for gas and liquid flow through the riser low-point, respectively. In Table 1 an overview comparison of the examined models is shown to summarize the model evaluation. The work in Meglio et al. (2009) had their model modified and used for control development successfully in Meglio et al. (2012a) and Meglio et al. (2012b), respectively. The key characteristics of this modified model is its simplicity, because a virtual valve is introduced in the bottom of the riser which emulates the gas blocking during slugging until the accumulated gas pressure overcomes the hydrostatic pressure of the riser. Even more important is that this model does not depend as much on the physical structure as the model proposed in Jahanshahi and Skogestad (2011), thus it can handle both pipeline-riser and well facilities. Hence this model is more flexible and faster to tune when big physical uncertainties of a considered system exist. In Biltoft et al. (2013) describes the model tuning of the modified (Meglio et al., 2009) model and the work in Pedersen et al. (2014a) constructed a controller based on the tuned model. Fig. 6 illustrates the principle of the virtual valve introduced in Meglio et al. (2009). An alternative concern is to employ Partial Differential Equations (PDE) models as control-oriented models. The study in Sinègre et al. (2006b) developed a PDE model to predict slugs in gas-lifting wells where the stability analysis was performed through small gain theorem. The work examined in

276

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

Fig. 5. Schematic representation of model parameters from Jahanshahi and Skogestad (2011), where (a) is the non-slugging flow regime and (b) is the slugging flow regime.

Table 1 Overview of examined low-dimensional control-oriented models. Models

Storkaas et al. (2003)

Jahanshahi and Skogestad (2011)

Eikrem (2008)

Kaasa and V. Alstad (2008)

Silva and Nydal (2010)

Year Extended from State equations Tuning parameters

2003 3 + 1 alg. 5

2011 4 4

2008 Eikrem (2006) 3 3

2008 3 7

2010 4 3

Meglio et al. (2012b) 2009 Meglio et al. (2009) 3 5

Meglio et al. (2011) developed a low-dimensional PDE model which comprises the gas mass fraction, the pressure, and gas velocity as states. Compared with numerical simulations the model proves to be accurate according to oscillation frequencies and shapes, see Meglio et al. (2011). Another PDE model was developed in Nemoto and Balio (2012), which is based on two switchable states: One where the gas is able to penetrate into the riser (steady flow), and another in which there is a liquid accumulation preventing the gas from penetrating into the riser (severe slugging). The model considers the liquid penetration length and the liquid height in the riser, thus the model can distinguish different kind of slugs.

Most of the examined models are based on mass balance equations, thus the liquids and gasses inflow rates have to be measurable, which is a rare case on offshore platforms. The pressure measurements solely, can only estimate the total combined mass flow rate, and not the ratio of gas and liquid. An example of estimating the total mass flow rate from the pressure measurements can be seen by investigating the model constructed in Meglio et al. (2009) with the adjustments made in Meglio et al. (2012b), where a nonlinear valve equation is stated as in Eq. (4). This mass flow estimation can be very useful, however it should be notices that this relationship is designed for one-phase liquid flow estimation.

4.2. Challenges for the control-oriented models

ωout = CA(ρ (P1 − P2 ))1/n z

Control-oriented slug modeling is a difficult task and none of existing models are adequate for control when several uncertainties have to be taken into account. This lack of robustness can cause erroneous simulations and ultimately poor control performances. Some of the main issues will be addressed here.

where ωout is the combined gas and liquid mass flow through a choke valve, ρ is the density, P2 is the pressure after a valve, P1 is the pressure before the choke valve, z is the choke valve opening percentage, A is the cross-section area and n has a value of 1 for laminar flow and 2 for turbulent flow.

(4)

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

277

Fig. 6. Schematic diagram of the riser system from Meglio et al. (2009) including the virtual valve at the bottom of the riser.

4.2.1. Observers Not all pressure transmitters are available either, especially for the part of the pipeline located subsea. For this reason several studies have focused on developing observers to estimate key point pressures and flow rates. It has to be noted that the observers’ performance heavily depends on the precision of the models used, which consequently is the observers biggest limitation. The work in Meglio et al. (2010) designed a high-gain observer for estimating the bottomhole pressure in wells by using only topside measurements. However, no proof of convergence was carried out when only using the topside pressure. In Grimstad and Foss (2014) an adaptive extension to the observer in Aamo et al. (2004b) was carried out. The adaptive observer estimates the well flow rate and downhole pressure from topside measurements on gas-lift wells. The study concludes that the observer successfully can predict the model states with measurement noise tested through simulations. However the observer is limited to the specific structure on which the well model is validated, it only works with a constant reservoir pressure, and the adaptive extension reduces the robustness of the observer. A similar observer was developed in Riccio et al. (2015) for offline identified (Nonlinear) Autoregressive Exogenous (ARX/NARX) models. The work in Mansoori et al. (2015) studied different transients of the bottomhole pressure in wells using system identification techniques to estimate a reservoir model. The model used for the system identification technique is a PDE model and is hard to implement for online estimation due to the heavy computation load required. Hence the study is more relevant for offline pressure transient analysis. The studies in Barbosa et al. (2015); Rezende et al. (2015) used Neural Networks to predict the bottomhole pressure in a oil well using three topside measurements: A topside flow transmitter, a pressure and a temperature transmitter. Thus the method require a lot of topside measurements as well as a large data set for sufficient model training. In both (Jahanshahi, 2013; Terese Vardenær Syre, 2012) riser observers for the low-point pressure were designed and applied as controlled variable in several control schemes. Not all wells are limited by few measurements; Some new wells have more transmitters integrated in the facility. They are

commonly refereed to as “Smart wells” or “Intelligent wells” (Jansen, 2001), and have downhole transmitters to monitor well and reservoir conditions, in addition to manipulatable devices to regulate the inflow of fluids from the reservoir to the well. These Smart wells are also being used to improve the effect of artificial reservoir flooding, generally water flooding. A water flooding well’s main objective is to use water injection in the reservoir to increase the oil production by increasing the reservoir pressure (Doren et al., 2011; van Essen et al., 20 06; 20 09; Grema and Cao, 2013; Zandvliet et al., 2006). A side benefit is that the increased reservoir pressure can be helpful to eliminate the slug (van Essen et al., 2009; Grema and Cao, 2014; Jansen et al., 2008). Equivalent to the “Smart wells” there exist “Smart risers”; The work described in Johal and Cousins (2001) patented an intelligent production riser for deep-water oil & gas fields where gas-lifting, slug catching, and measurements are combined into one big riser system; Three sets of pressure and temperature transmitters are installed at the top, center, and base of the riser through the outer pipe wall.

4.2.2. Model identification Another practical issue arises as the well-pipeline-riser models heavily depend on the initial conditions of the masses of all phases in the pipelines, which also can be hard to estimate in an online manner. In practice, the initial states are calculated by solving the model for steady-state and the initial state will be approximated by this solution. All the models examined in Section 4 are based on the physical structure design of the system which is not always known in details, thus the uncertainties can lead to inaccurate predictions. Due to the trade-off between complexity and precision, the dynamic models are limited by the various tuning parameters, forcing a lot of effort into tuning the models for various system operating conditions. Without a proper data set to validate the model’s tuning parameters the model can be very imprecise. An alternative approach to obtain the model in an online manner was examined in Ogazi et al. (2009b), where identification of a linear unstable model using a closed-loop step test was carried out. In a similar way in Jahanshahi and Skogestad (2015) the simplified model was identified by the relay-feedback method.

278

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

Table 2 Comparison between the passive slug elimination methods based on advantages and disadvantages. Method

Pipe shape change

Venturi / Permanent choke

Slug catcher

Homogeniser

Examined in

Adedigba (2007); Makogan (2007); Xing et al. (2013a, 2013b, 2013c) 2007–2013 Changes flow regime upstream riser Some

Almeida and Gonalves (1999a, 1999b); Jansen et al. (1996)

McGuiness and Cooke (1993)

Hassanein and Fairhurst (1998)

Sarica and Tengesdal (20 0 0); Tengesdal et al. (2002)

1996–1999 Increased back pressure

1998 Changes multiphase fluids to a homogeneous fluid All

20 0 0–20 02 Reduces hydrostatic pressure

No multiphase

Works for most vertical sections, flexible to disturbances Costly installation and maintenance, complicates the flow assurance

Year Technique

Advantages

Simple, no problems for pigging operations

Cheap, Simple

1993 Physically filters the flow oscillations Depends on slug catcher’s size Eliminates slug at early stage

Disadvantages

Costly installation, does no eliminate all slugs

Bad for pigging operations, reduces the production rate

Costly installation and maintenance, requires extra pipelines

Stable regions∗

∗

All

Bad for separation

Self-gas lifting

Some

Indicates the realistic potential for elimination of slugs in the flow map region.

5. Slug elimination methods The elimination methods can be divided into two major categories: Passive approaches, and active approaches. Passive approaches: The slug is avoided by process changes. No actuators are involved. Active approaches: The slug is eliminated by manipulated actuators. In some cases actuators are being used, while process changes also are being implemented. To avoid confusion this will be categorized as active control. 5.1. Passive elimination Elimination of severe riser slug, by creating physical adaptions in the process, has been investigated for a long time. Early studies (Yocum, 1973) identified several different process changes which still are being used in plants today to eliminate the slug. The passive methods can be divided into three categories: 1. Reduce incoming line diameter near the riser to establish stable flow regime. 2. Create dual multiple risers, instead of a single riser. 3. Liquid remix device, which mixes the fluids in the riser base to avoid accumulation, hence preventing the stratified flow to cause severe slugging. These three ideas are the base concepts of all the methods explained in this section. Research on passive elimination methods have been made in Xing et al. (2013b), where different flow conditioners have been investigated. A flow conditioner is defined as a passive method in which a device is installed in the pipeline to change the existing flow regime upstream the riser. In the following a list of passive elimination methods is examined. Table 2 shows a brief comparison between the passive slug elimination methods based on the advantages and disadvantages. Helix-shaped pipe. The gas/liquid stratified flow can be modified effectively by non-straight pipe sections. In Adedigba (2007) it was proved that the slug region in the flow map can be reduced by the use of a helical-shaped pipeline, because the added helical shape in some cases changes the flow regime downstream the helical pipeline section. Placing the helical pipe upstream of the riser was concluded to be the most efficient for reducing the slug region. Wavy pipe. Placing a wavy pipe at the pipeline close to the riser base (Xing et al., 2013b). This way small artificial slugs are created to avoid big severe slugs created with the accumulation at the riser base. Experiments proved a reduced operating region of severe slugging (Xing et al., 2013a). A numerical study using a CFD

model for the wavy pipe was examined in Xing et al. (2013c). The numerical study’s primary goal was to find the optimal position and dimensions of the wavy pipe. Venturi-shaped device. The work in Almeida and Gonalves (1999a, 1999b) proposed and patented a venturi-shaped device including a convergent nozzle section followed by a divergent diffuser section. This device is supposed to be located near to the riser base. Venturi-shaped devices can give a pressure drop causing a mixing effect and converting the stratified flow to a turbulent flow. This pressure drop, however, also minimizes the production and cause problems for pigging operations. Pipeline as flow conditioner. Another flow conditioner was patented in Makogan (2007). The patented pipeline construction upstream the riser bottom was developed to act as a flow conditioner by adding a small trapezium bend to the pipeline itself. Based on their results it was stated that this device could eliminate severe slugging by establishing short mini-plugs. The volume of each liquid slug could be sufficiently small to be transported by the gas pressure building up behind it. Consequently severe slugging in the riser could be changed into plug flow or intermittent flow. Thus the idea was to minimize the slug to plugs rather than completely eliminating it. Permanent choking. The work described in Jansen et al. (1996) proposed permanent choking as an effective way to avoid the slug flow. The method is closely related to a venturi-shaped pipe. The main idea is to change the flow condition by creating back pressure. This however is not the optimal solution in terms of production rate, and even if the valve is choked to the optimal point (open-loop bifurcation point) the approach is not robust as any variations in fluid velocities can cause the slug to reoccur. Slug catcher. Slug catchers are the most commonly used passive slug elimination methods. They work as buffer tanks and sometimes even pre-separates the liquid and gas. The slug catchers operate as physical low-pass filters, such that the high-frequency oscillating inflow will be filtered out to a smooth outflow. An early stage separation method was carried out and implemented in McGuiness and Cooke (1993) to prevent multiphase flow completely from the transportation pipeline. The idea was to separate the oil, water and gas in a separator at the top of the well. This method is effective but very expensive as the cost of multiple single-phase pipelines are much more expensive than a single multi-phase pipeline. It also increases the frequency of pigging operations. Besides, it is more expensive to have slug catchers subsea, and for early-stage separators there are in many cases no well platform above sea level. Thus the biggest limitation of the subsea slug catchers is the economic cost.

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

Homogenising the multiphase flow. The work examined in Hassanein and Fairhurst (1998) presented a method to avoid slug formation by attenuating the non-homogeneous liquid and gas into one homogeneous fluid; thus eliminating the multiphase flow. The idea was to reduce the surface tension of the fluid by injecting a surfactant which could change the fluid into foam, hence making the fluid homogeneous. One main limitation using this method is the separation of the material at the topside separation process, ultimately reducing the product quality. Self-gas lifting. Self-gas lifting is an artificial way of making the slug cycle smaller, and thus the amplitude of the pressure oscillations. It reduces the static head (weight of liquid) by creating a smaller pipeline feeding from the main pipeline to the riser, where a one-way rectifier is linked to the riser to ensure one-directional flow. This way the gas will accumulate for a shorter period of time due to the feeding to the riser. Furthermore no external gas lift supply is required. Such a method was proposed in Sarica and Tengesdal (20 0 0) which also proved that it could give a smoother start-up transient where flow blow-outs often exist. The negative features of the self-gas lifting is the difficulties of pigging and the cost of the extra pipeline. The work in Tengesdal et al. (2002) also investigated the possibility of avoiding the gas compressing in the pipeline by separating the gas upstream of the riser base and re-injected into the riser. The re-injected gas reduces the hydrostatic pressure created by the liquid in the riser. Hence the slug formation is inhibited and if slug occurs the pressure amplitude will be reduced. 5.2. Active elimination Active elimination is the elimination technique where actuators are controlled in a feedback loop mainly with pressure, temperature and flow transmitters. The placement of the sensors varies depending on which specific platform is studied but is often located topside on the platforms. The impact between the inputs and the outputs has been studied in the recent years. The input-output controllability of a system can be evaluated quantitatively by calculating minimum achievable peaks of different closed-loop transfer functions (Skogestad and Postlethwaite, 2005). These values show the physical limitations of a system’s controllability which depends on the location of poles and zeros in the open-loop system. 5.2.1. Active valve choking Choke valve methods are the most investigated actuator for slug elimination. This is mainly due to the fact that valve choking is a cheap, easy and flexible implementation solution. It is however hard to use the choke valves to eliminate the slug without also reducing the production rate. The work in Storkaas (2005) using theoretical examination, Storkaas and Skogestad (2008) using PDE model analysis, and Jahanshahi et al. (2012) using be Ordinary Differential Equations (ODE) model analysis, investigated the Input-Output controllability of a topside control choke valve with different measurements in order to find the most suitable control feedback loop for stabilization. All studies conclude that both topside pressure and flow measurements have limitations and have suggested a combination of these measurements to develop cascaded Single-Input-MultipleOutput (SIMO) control system for the well-pipeline-riser system. It was also suggested that a subsea control choke valve could be used as a possible manipulated variable, both in a configuration as only actuator, and in a configuration combining the subsea and the topside control valve as actuators in a Multiple-InputSingle-Output (MISO) control scheme. The work described in Jahanshahi (2013); Jahanshahi et al. (2012) applied the controllability analysis on two case studies: A well-pipeline-riser system

279

and a gas-lifting oil well system. It was determined that in both cases the bottomhole/riser-base pressure measurements are the best for Single-Input-Single-Output (SISO) topside control valve control. If it is available the downstream flow rate of the choke valve is preferable, and it was concluded in Jahanshahi (2013); Jahanshahi et al. (2012) that a combination of the bottomhole pressure and the outlet flow rate gives the best result for the well-pipeline-riser system. If the subsea pressure measurements are not available, combining the topside pressure and the flow rate gives a satisfactory result too. The problems using only the topside pressure measurement is linked to the inverse response of the non-minimum phase system due to right-half plane zeros and low Signal-to-Noise ratio (SNR) (Jahanshahi, 2013; Jahanshahi et al., 2012). However, in Meglio et al. (2012b) the performance of the observer controller using the topside pressure measurement does not seem to be influenced by the unstable zero, caused by the controlled variable, which have been proposed in Storkaas (2005). The work in Helgesen (2010); Sivertsen and Skogestad (2005) also proposed various combination for cascade control of topside measurements. Thus there are still ongoing discussions regarding which controlled variable to apply for optimal closed-loop performance. Other measurement techniques have been considered for the antislug control as well: In Hedne and Linga (1990) and Meglio et al. (2012b) the pressure drop over the riser was proposed as controlled variable, and the experimental study in Pedersen et al. (2015b) proposed an Electrical Resistance Tomography (ERT) transmitter for online slug detection. The study investigated the ERT’s ability to handle the 3-phase flow on a lab-scaled testing facility. It was concluded that the ERT technique successfully can detect the slug if the liquid phases are well-mixed and the oil-to-water ratio is low. Both assumptions are realistic for mature wells.

5.2.1.1. Topside choke valve. Choking the pipeline upstream the riser has proven effective in eliminating the slug flow (Almeida and Gonalves, 1999a, 1999b). It is however an expensive solution to have a low-point choke valve placed subsea, hence the placement of the choke valve at the riser base is a rare case. The topside placement of the choke valve is more common in offshore constructions. The anti-slug control using a topside choke valve has been studied for many years; it was first suggested in 1979 in Schmidt et al. (1979). By choking the valve the pressure drop across the choke increases. This will reduce the gas velocity in the riser and complicates the gas tail from penetrating the riser base (Fargharly, 1997). The work examined in Havre and Dalsmo (2001) proved from OLGA simulations that manipulating a topside choke valve with feedback control from pressure measurements could eliminate the slugging flow. The study concluded that feedback control can improve the production rate compared with permanent choking. The selection of transmitters for the feedback control has been a big discussion for several years, as a seabed (low-point) pressure transmitter rarely is available. In a similar way the downhole pressure of a well is often not available. For this reason many studies are working on estimating the seabed/downhole pressure from a topside pressure transmitter for the purpose of regulating the topside choke valve; The work in Jahanshahi et al. (2013a) developed and compared different observer schemes for a pipeline-riser system which includes a standard Luenberger filter and a Unscented Kalman Filter (UKF), the work examined in Eikrem et al. (2004b) proposed a PI control solution using a state observer to estimate the downhole pressure of the well, and the work described in Scibilia et al. (2008) developed a high gain observer to estimate the downhole pressure in a well. Hence the issue of not having the bottomhole or low-point pressure measurements are often handled by the extension of an observer to the controller.

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

Riser base pressure

280

Pstable

: Open-loop manual choking : Unstable equalibrium

Pslug,max Pslug,avg

Pslug,min Bifurcation point

Valve opening

Fig. 7. A general schematic showing a Hopf Bifurcation map with corresponding bifurcation point.

Hopf Bifurcation maps are often applied to determine which choke valve openings results in stable flow regions with associated pressure. A Hopf Bifurcation is a dynamic system which loses it’s stability as a pair of complex conjugate eigenvalues of the linearized system cross the imaginary axis of the complex plane in a pole-zero plot. Fig. 7 shows a hypothetical Hopf bifurcation map for a riser. In the unstable region the maximum pressure (Pslug, max ) of a slug cycle is indicated by the top line and the corresponding minimum pressure (Pslug, min ) is indicated by the bottom line. Besides the Hopf Bifurcation map an unstable equilibrium is represented by the dashed line. The unstable equilibrium is indicating the continuation tendency from the stable region in the unstable region. The bifurcation point indicate the switching point between the stable and unstable flow and the corresponding production loss of severe slug due to the higher average pressure during a slug cycle (Pslug,avg ). It has to be noticed that the bifurcation point can be moved to higher choke valve openings by applying feedback control. For this reason many studies such as Jahanshahi (2013); Ogazi (2011) have focused on designing controllers to move the open-loop bifurcation point to a more desirable location in the closed-loop system. However, the higher opening at which the valve operates the less robust the system is Jahanshahi et al. (2012). The work examined in Ogazi (2011) considered the topside separator gas outlet control valve, as an alternative solution as the anti-slug control actuator. It was proved from simulations that the gas outlet valve successfully could stabilize the flow at a slightly higher valve opening than the topside valve upstream of the separator. It was however concluded that even though gas outlet valve could guarantee stable flow at a higher opening the riser base pressure was actually lower using the valve upstream the separator. This was due to the difference in the individual valve characteristics and the fact that the gas valve also gives back pressure to the separator. An Input-Output controllability analysis was also made to evaluate which of the two topside control valve would be the most optimal, and it was concluded that both valves possesses the ability to stabilize at high openings. Similarly, in Molyneux and Kinvig (20 0 0) a control method to handle the slug flow by manipulating the gas outlet valve of the separator was patented. The gas outlet valve, however, is often dedicated to pressure, level or smooth flow control out of the separator, see AL-Hatmi and Tham (2006). In Jahanshahi et al. (2013b) feedback linearization was used to design a control law; thus developing a nonlinear model-based control. The controller design is using two pressure measurements, the riser base pressure and the topside pressure. The controller was able to stabilize the slug flow with high choke valve openings and thus avoided the low choke valve openings’ production reduction. The work examined in Ogazi et al. (2009a)

also investigated the possibility of using large valve openings to maximize the oil production rate while also eliminating the slug; thus working in the open-loop unstable (slugging) region by moving the bifurcation point with a feedback loop. The study concluded that the percentage improvement in oil production compared with manual choking will increase as the well pressure declines. This means that adopting the proposed control scheme is more beneficial for mature oil fields than for relatively new fields. The work developed in Enricone Stasiak et al. (2012) also developed a topside control design for minimizing either the flow or pressure oscillations, while keeping the choke valve opening higher than the opening at the beginning of the limit cycle. Hence the control aims to suppress the oscillations while keeping the choke opening operating around a high desired opening value. The work described in Havre (2007) patented a dynamic feedback controller using a topside control choke valve as an actuator and either the topside pressure, flow, or the differential pressure over the valve as controlled variables. If the measurements observe a sudden drop, a decision is made whether a liquid blockage in the flow line is present (if the drop is big enough). If a liquid blockage is observed or predicted the valve opening will increase. The changing magnitude of the valve opening is determined by the actual measurement drop. Further manipulation of the valve is inhibited until a time penalty has expired. A similar method was considered in Pedersen et al. (2014a) where a controller was combined with a supervisor which determines if the pressure varies too much in a pipeline-riser system, thus using the pressure variations as the controlled variables for the supervisor. The supervisor was combined with a switching PID controller for two different objectives: One for the slug elimination and one for production rate optimization. By manipulating the choke valve the severe slug was eliminated and at steady-state the production rate was increased by 7.8% compared with a constant fully open choke valve. The study concluded that this controller finds the optimal open-loop bifurcation point with a fixed choke valve opening but a further increase in production rate can be possible if a varying valve opening is considered to move the bifurcation point further into the open-loop unstable region. The use of supervisory control has also been applied in several other control schemes: In de Oliveira et al. (2015) the supervisory controller was used to increase a robust adaptive controller’s robustness moving the system to a safer operational point, and in Campos et al. (2015) a supervisor was part of a safety control scheme and was combined with an optimal control scheme to handle uncertainties in the offshore plants. As PID is the most commonly used controller in the industry, several studies considered a PID control scheme but have included more advanced tuning techniques. In Godhavn et al. (2005) three PI tuning methods were proposed for eliminating slug. The tuning methods applied different controlled variable used for the feedback signals; a PI for volumetric flow to stabilize the flow, a pressure PI controller to stabilize the pressure, and a pressure and volumetric flow cascade PI controller where the outer loop maintains a stable pressure and the inner loop handles the stable flow. However, actually this cascaded configuration is redundant as a stable pressure also naturally will cause a stable flow (implying there is a constant pressure source for the inlet flow), and thus there is no need for an extra feedback loop. For the volumetric flow a topside flow transmitter or a densitometer are required, and for the pressure a seabed pressure measurement is required. Thus the tuning method is limited to a system with many measurements available. In Jahanshahi et al. (2014) a new IMC-PIDF controller (Internal Model Control tuned PID with a low-pass filter) was developed as an extension to the tuning methods proposed by Godhavn et al. (2005). Their tuning method was compared with other optimal tuning methods and it was concluded that the robustness of the controller

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

: Region of optimimum lift gas utillization

Oil production rate

Unstable

Stable

Maximum oil production rate

Maximum PVPAT

Gas injection rate Fig. 8. A performance graph of a hypothetical gas-lift well including the PVPAT and maximum oil production. The dashed line indicates the actual production during slugs.

could be further improved by the extension of a H∞ loop shaping to improve the robust stability and performance. The study in Abardeh (2013) also examined robust IMC-PIDF controller solutions for S-shaped risers. The same approach was carried out in Jahanshahi and Skogestad (2015) where a comparison between IMC-PIDF and two H∞ controllers were tested. The control experiments were carried out on the same medium-scaled S-shaped riser as in Abardeh (2013), but also on a small-scaled L-shaped riser. It was concluded that the performance of all three controllers were good, but that the H∞ loop-shaping controller gave the best robust performance. The work in Eikrem et al. (2004b) developed a PI controller for wells using topside pressure measurement to estimate the downhole pressure of a well. Since then several studies have worked with PID controllers in combination with observers to stabilize well pressure and flow, see Eikrem et al. (2004a) and Aamo et al. (2004a) summarized in Eikrem (2006) where Extended Kalman Filters (EKF) are used to estimate either the downhole pressure measurement or used as an estimate of the masses in the system (the states). A reduced order observer is then introduced as an alternative to the EKF. Both observer methods are used in combination with several controllers to stabilize the mass in the well; thus controlling the states of the system by feedback control of a topside valve. It has to be noted that gas lifting in the well also is considered with the topside control valve as a combined MultipleInput-Multiple-Output (MIMO) solution. 5.2.2. External gas lifting The external gas lifting serves two purposes: (i) Enabling the mature and depleted reservoirs with low pressure to produce by injecting gas at the bottom of the well, and (ii) preventing riser-induced slugging by injecting gas at riser base. In Jansen et al. (1996) it was concluded that gas lifting eliminates severe slugging by increasing the velocity and reducing the liquid holdup in the riser, but also that large amounts of injected gas is needed to stabilize the flow. Hence gas lifting will reduce a vertical pipeline’s hydrostatic pressure and stabilize the flow in the direction of the superficial velocity of the gas. In Hu (2004) it was proved that the two main methods to obtain stable flow in the well are by using waterflooding to increase reservoir pressure or by increasing gas injection using gas-lifting. It was also concluded that a smaller annulus volume could improve the stability, as it is another approach to increase the superficial velocity. Fig. 8 illustrates a typical gas-lift performance relationship (LPR) of a hypothetical gas-lift well. The solid line shows the operators’ estimated production rate from

281

steady-state simulations. Here the oil production rate increases rapidly with the gas injection rate at first and then tends to level off before reaching a peak. If the gas injection is further increased the production will gradually decrease. This production decrease is caused by the increased gas-induced friction which cannot be compensated by the reduced hydrostatic pressure in the well tube. The operation points for maximum Present Value Profit After Tax (PVPAT) and maximum oil production rate are marked on the graph. In the daily production, PVPAT is often the value aimed for, even though this value can be located in the slugging region. The figure also shows the region of the optimum gas lift utilization, where the physical and economic uncertainties are being taken into account. As there often exist problems determining the exact PVPAT due to the uncertainties, the operators consequently define this acceptable region for the gas-lift well to operate in based on the estimated process uncertanties. The dashed line indicates the actual production rate during gas-lifting slug flow, where the slugs decreases the production rate dramatically. Clearly, the predicted LPR can mislead the operators to aim for an unstable operating point causing slug flow and production decrease. Besides, the gas-lifting is often limited by the gas compressors’ capability, and thus the optimal gas-lift inflow is not always feasible. In Johal et al. (1997) a new method to lift gas in a riser was presented. The method specifically aims at riser gas-lifting for deep-water oil fields. The investigation introduced “Multiple Riser Base Lift (MRBL)” as an alternative to the traditional “Riser Base Gas Lift (RBGL)” where the Joule-Thompson cooling (a change of temperature for a fluid when it flows through a valve) can cause problems for the gas at the control valves due to the possibility of hydration. The proposed method diverts a stable multi-phase flow stream to the nearest pipeline-riser system where severe slug is experienced. This obviously demands multiple installed transportation pipelines with the costly possibility of diverting each single multi-phase flow steam, respectively. In Cousins and Johal (20 0 0) a device called “A slug catcher, a multiphase flow meter and a riser base gas lift” was patented. The device is a riser which provides the benefits from all three components in one device. This device consists of two concentric pipelines where the inner one is a riser production tube and the outer one is dedicated for the gas-lifting and flanged off at the riser bottom to create an annulus. They are linked through a number of holes both at the bottom and the top of the riser. The topside of the outer pipe is connected to an inlet source with a valve to compress the gas and thus control the gas injected. The device is intended to reduce the fluid density in the riser tube (the inner pipe) by the use of the compressed penetration gas from the outer pipe. The corresponding gravitational pressure drop will ease the lifting of the liquid. This injection will continue until a steady-state is reached. In Johal et al. (1997) a new method to lift gas in a riser was presented. The method specifically aims at riser gas-lifting for deep-water oil fields. The investigation introduced “Multiple Riser Base Lift (MRBL)” as an alternative to the traditional “Riser Base Gas Lift (RBGL)” where the Joule-Thompson cooling (a change of temperature for a fluid when it flows through a valve) can cause problems for the gas at the control valves due to the possibility of hydration. The proposed method diverts a stable multi-phase flow stream to the nearest pipeline-riser system where severe slug is experienced. This obviously demands multiple installed transportation pipelines with the costly possibility of diverting each single multi-phase flow steam, respectively. In Cousins and Johal (20 0 0) a device called “A slug catcher, a multiphase flow meter and a riser base gas lift” was patented. The device is a riser which provides the benefits from all three components in one device. This device consists of two concentric pipelines where the inner one is a riser production tube and the outer one is dedicated for the gas-lifting and flanged off at the riser bottom to create an

282

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

annulus around the pipes. They are linked through a number of holes both at the bottom and the top of the riser. Here the topside of the outer pipe is connected to an inlet source with a valve to compress the gas. The device is intended to reduce the fluid density in the riser tube (the inner pipe) by the use of the compressed penetration gas from the outer pipe. The corresponding gravitational pressure drop will ease the lifting of the liquid. This injection will continue until a steady-state is reached. The work described in Krima et al. (2012) developed several PI controller for the gas-lifting focusing on the mitigating hydrodynamic slug in OLGA simulations. The controllers were based on different riser measurements as the controlled variables. The work concluded that using a topside hold-up transmitter as the controlled variable resulted is the best control solution. The differential pressure across the riser was mentioned as an acceptable alternative. It was however concluded that a good control design for the topside control choke valve reduced the requirement of injection gas, and thus a combined control design for both actuators was the optimal solution. 5.2.3. Combination of gas-lifting and topside choking For the well, the work examined in Pagano et al. (2008) developed a model-free MIMO PI-controller for both gas-lifting and topside choking. The valve manipulating the gas-lifting inlet was controlled to stabilize the gas injected to the production tube and the topside choke valve was used to stabilize the topside pressure. The main idea was to use a Variable Structure Control (VSC) law that introduced sliding bifurcation on the system to control the limit cycles. The idea was based on an approach developed in Angulo et al. (2005). The work proved that Hopf Bifurcation is mainly caused by low gas inflow rates. The downhole measurements are not necessary for the control scheme. Furthermore, the controller proved to also work for re-starting the well, where it reduces the start-up time without the occurrence of start-up slug. The study examined in Pedersen et al. (2014b) developed a new way finding the optimal boundaries between the slugging and non-slugging regions. The idea was to combine the flow map and bifurcation map into a single map to obtain the surface at which the slug occurs. 5.2.4. Shell’s slug suppression system (S3) Shell has designed a slug suppresson system, see Kovalev (2003). The system is a combination of physical process changes and feedback control. A 2-phase gas-liquid mini-separator is installed to separate the liquid from the gas upstream the firststage 3-phase oil-water-gas separator. Between the two separators there are two choke valves: One for the gas pipeline and one for the liquid pipeline. This way the liquid injection into the first stage separator is controlled, to stabilize the height of the liquid, while the gas injection is used to compensate for the possible slugs. This is an advantage as the gas flow is much easier to control when no liquid is in the same pipeline. This slug suppression system was successfully implemented and the study concluded that the system eliminated all types of slug and improved the production rate of both oil and gas. The economic loss from the extra equipment has to be mentioned as a huge disadvantage, because this also introduce extra maintenance of equipment. 6. Conclusion and future work The work described in this paper examined the key challenges related with severe slug flow in offshore multiphase oil & gas pipeline transportation systems. Slug modeling is a challenging task and is the groundwork for anti-slug control. The main issue in severe slug modeling is addressed to the robustness; the

control-oriented models might be precise for specific running conditions but imprecise for others. The running conditions, such as production mass flow from the well into the pipeline system, are often estimated from observers if no direct measuring equipment is installed. These observers are also limited by the precision of the corresponding model. For the slug elimination many different methods have been examined. The two main active approaches for anti-slug control are control of a topside control choke valves and external gas lifting in a well or riser. The control of these and their corresponding controlled variables have been discussed in details in this paper. As the gas lifting often is limited by the compressors’ capacity the slug elimination often rely solely on the choke valves where the reduced production rate can be a problem. It is observed that most investigated controllers aim for two objectives: (1) Eliminate the slug flow, while (2) optimize the oil & gas production rate. It is concluded that the main limitation of the examined controllers are their lack of robustness; the more aggressive the controllers focus on optimizing the production the closer the system gets to their limits and hence shifts to slug flow for any model deviations or process changes. Many possible solutions in modeling and eliminating the slug have been examined. Even though anti-slug control has been heavily investigated over the last decade, obtaining the most robust and optimal solutions are still unsolved problems. As the global oil & gas resources reduce with time the motivation for deeper oil & gas drillings increase, thus the slugs in the future will get even more severe with longer vertical wells and risers. For this reason it can be predicted that slug elimination and especially anti-slug control will be an even more crucial and necessary topic in the future. Acknowledgment The authors would like to thank the support from the Danish National Advanced Technology Foundation (via PDPWAC Project (J.nr. 95-2012-3)). Thanks also go to our colleagues J.P. Stigkær, A. Aillos, C. Yigen, K. G. Nielsen and P. Molinari from Maersk Oil A/S, our colleagues P. Sørensen, A. Andreasen, J. Biltoft and S.A. Meybodi from Ramboll Oil & Gas A/S, and C. Mai, L. Hansen, K.L. Jepsen and H. Enevoldsen from Aalborg University, for many valuable discussions and technical supports. References Aamo, O., Eikrem, G., Siahaan, H., Foss, B., 2004a. Observer design for gas lifted oil wells. In: The 2004 American Control Conference, Boston, USA. Aamo, O., Eikrem, G., Siahaan, H., Foss, B., 2004b. Observer design for multiphase flow in vertical pipes with gas-lift - theory and experiments. Journal of Process Control. Abardeh, M.E., 2013. Robust control solutions for stabilizing flow from the reservoir: S-Riser experiments. Norwegian University of Science and Technology, Department of Chemical Engineering Master’s thesis. Adedigba, A.G., 2007. Two-phase flow of gas-liquid mixtures in horizontal helical pipes. Cranfield University Ph.D. thesis. AL-Hatmi, N., Tham, M., 2006. Controllability and resiliency aspects of gravity three-phase horizontal separators. USTARTH. Al-safran, E.M., Taitel, Y., Brill, J.P., 2004. Prediction of slug length distribution along a hilly terrain pipeline using slug tracking model. J. Energy Resour. Technol. Trans. ASME 126, 54–62. doi:10.1115/1.1649971. Almeida, A., Gonalves, M., 1999a. Device and method for eliminating severe slugging in multiphase-stream flow lines.Patent: US6041803A. Almeida, A., Gonalves, M., 1999b. Venturi for severe slugging elimination. In: The 9th International Conference on Multiphase Production, pp. 149–158. Angulo, F., di Bernardo, M., Fossas, E., Olivar, G., 2005. Feedback control of limit cycle: a switching control strategy based on nonsmooth bifurcation theory. IEEE Trans. Circuit Syst. 52 (2), 366–378. Arnold, R., Sandmeyer, D., Eickmeire, J., 1972. Production problems of a high pressure, high temperature reservoir. Asheim, H., 1988. Criteria for gas-lift stability. J. Petrol. Technol. 40, 1452–1456. Baardsen, I., 2003. Slug regulering i to phase stroemning - eksperimentell verifikasjon. Norwegian University of Science and Technology Master’s thesis. Bai, Q., Bai, Y., 2014. Subsea Pipeline Design, Analysis, and Installation. Gulf Prof. Publishing.

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284 Bailey, R., Lu, Z., Shirzadi, S., Ziegel, E., 2014. From market-basket analysis to wellhead monotoring: Use of events to increase oil recovery. In: SPE Intelligent Energy Conference and Exhibition, Utrecht, The Netherlands. Barbosa, B.H.G., Gomes, L.P., Teixeira, A., Aguirre, L.A., 2015. Downhole pressure estimation using committee machines and neural networks. In: 2nd IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, Florianopolis, Brazil, pp. 292–297. Belt, R., Djoric, B., Kalali, S., Duret, E., Larrey, D., 2011. Comparison of commercial multiphase flow simulators with experimental and field databases. In: BHR Groups Multiphase Production Technology Conference in Cannes 15. Bendiksen, K., Malnes, D., Moe, R., Nuland, S., 2013. Information for Beginning with OLGA - “The Dynamic Two-Fluid Model OLGA: Theory and Application”. SPE Production Engineering. Biltoft, J., Hansen, L., Pedersen, S., Yang, Z., 2013. Recreating riser slugging flow based on an economic lab-sized setup. In: IFAC International Workshop on Periodic Control 5th, pp. 47–52. Bøe, A., 1981. Severe slugging characteristics; part i: flow regime for severe slugging; part ii: point model simulation study. Selected Topics in Two Phase Flow, NTH, Trondheim, Norway. Campos, M., Takahashi, T.R., Ashikawa, F., Simoes Neto, S., Meien, O.F.V., Sant Anna Stender, A., 2015. Advanced anti-slug control for offshore production plants. In: 2nd IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, Florianopolis, Brazil, pp. 83–88. Cousins, A., Johal, K., 20 0 0. Multi purpose riser. Pub. No. EP1022429 A1, mentor subsea technology services Inc., Houston, Texas. EP Patent App. EP20 0 0 030 0126. Danielson, T.J., Bansal, K.M., Djoric, B., Duret, E., Johansen, S.T., 2011. Testing and qualification of a new multiphase flow simulator. In: Offshore Technology Conference, Houston, Texas. Dansk Undergrunds Consortium, DUC, 2012. DUC in the North Sea. Doren, J.F.V., den Hof, P.M.V., Jansen, J.D., Bosgra, O.H., 2011. Parameter identification in large-scale models for oil and gas production. In: 18th IFAC World Congress, Milano, Italy, pp. 10857–10862. Eikrem, G., Aamo, O., Siahaan, H., Foss, B., 2004a. Anti-slug control of gas-lift wells - experimental results. 6th IFAC Symposium on Nonlinear Control Systems, Stuttgart, Germany. Eikrem, G.O., 2006. Stabilization of gas-lift wells by feedback control. Department of Engineering Cybernetics, Norwegian University of Science and Technology Ph.D. thesis. Eikrem, G. O., 2008. Eikrem riser model. http://www.nt.ntnu.no/users/skoge/diplom/ prosjekt08/tuvnes/Eikrem_risermodel_matlab/. Eikrem, G.O., Imsland, L., Foss, B., 2004b. Stabilization of gas lifted wells based on state estimation. IFAC. Enricone Stasiak, M., Pagano, D.J., Plucenio, A., 2012. A new discrete slug-flow controller for production pipeline risers. In: IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, pp. 122–127. van Essen, G., Zandvliet, M., Jansen, J., 2006. Robust optimization of oil reservoir flooding. In: Computer Aided Control System Design, IEEE International Conference on Control Applications, pp. 699–704. van Essen, G.M., den Hof, P.M.V., Jansen, J.D., 2009. Hierarchical economic optimization of oil production from petroleum reservoirs. International Symposium on Advanced Control of Chemical Processes (ADCHEM 2009), Istanbul, Turkey. Fairuzov, Y., Guerrero-Sarabia, I., Calva-Morales, C., Carmona-Diaz, R., Cervantes-Baza, T., Miguel-Hernandez, N., Rojas-Figueroa, A., 2004. Stability maps for continuous gas-lift wells: A new approach to solving an old problem. In: Society of Petroleum Engineers (SPE) Annual Technical Conference and Exhibition, Houston, Texas. Fargharly, M.A., 1997. Study of severe slugging in real offshore pipeline riser-pipe system. PE Middle East Oil Show, Paper SPE 15726. FMC Technologies, 2010. Separation Systems - Solids Handling. Commercial brochure. Foss, B., 2012. Process control in conventional oil and gas field - challenges and opportunities. In: Control Engineering Practice 20, 4th Symposium on Advanced Control of Industrial Processes (ADCONIP), pp. 1058–1064. Fuchs, P., 1987. The pressure limit for terrain slugging. In: BHRA International Conference on Multiphase Flow 3rd, pp. 65–71. Godhavn, J.-M., Fard, M.P., Fuchs, P.H., 2005. New slug control strategies, tuning rules and experimental results. J. Process Control 15, 547–557. Grema, A., Cao, Y., 2014. Optimal feedback control for reservoir waterflooding. In: The 20th International Conference on Automation and Computing (ICAC’14). Grema, A.S., Cao, Y., 2013. Optimization of petroleum reservoir waterflooding using receding horizon approach. In: 8th IEEE Conference on Industrial Electronics and Applications (ICIEA), pp. 397–402. Grimstad, B., Foss, B., 2014. A nonlinear, adaptive observer for gas-lift wells operating under slowly varying reservoir pressure. In: 19th World Congress, The International Federation of Automatic Control (IFAC’14), Cape Town, South Africa, pp. 2824–2829. Hassanein, T., Fairhurst, P., 1998. Challenges in the mechanical and hydraulic aspects of riser design for deep water developments. In: IBC UK Conf. Ltd. Offshore Pipeline Technology Conference. Havre, K., 2007. Method, computer program product and use of a computer program for stabilizing a multiphase flow. US Patent 10433653. Havre, K., Dalsmo, M., 2001. Active feedback control as the solution to severe slugging. In: SPE Annual Technical Conference & Exhibition, New Orleans, Louisiana. Havre, K., Sotrnes, K. O., Stray, H., 20 0 0. Taming slug flow in pipelines. Hedne, P., Linga, H., 1990. Suppression of terrain slugging with automatic and manual riser choking. In: Advances in Gas-Liquid Flows, pp. 453–469.

283

Helgesen, A.H., 2010. Anti-slug control of two-phase flow in risers with: Controllability analysis using alternative measurements. NTNU, Norway Master’s thesis. Hewitt, G.F., Roberts, D.N., 1969. Studies of Two-Phase Flow Patterns by Simultaneous X-Ray and Flash Photography. Defense Technical Information Center. Hill, T.J., Wood, D.G., 1994. Slug Flow: Occurrence, Consequences, and Prediction.. University of Tulsa Centennial Petroleum Engineering Symposium. Hu, B., 2004. Characterizing gas-lift instabilities. Norwegian University of Science and Technology (NTNU), Department of Petroleum Engineering and Applied Geophysics Ph.D. thesis. Hu, B., Golan, M., 2003. Characterizing gas-lift instabilities by dynamic simulation. 2-Fall Gas-Lift Workshop, Kuala Lumpur, Malaysia. Husveg, T., Rambeau, O., Drengstig, T., Bilstad, T., 2007. Performance of a deoiling hydrocyclone during variable flow rates. Miner. Eng. 20, 368–379. Isaac, O.A., Cao, Y., Lao, L., Yeung, H., 2011. Production potential of severe slugging control systems. In: 18th IFAC World Congress, pp. 10869–10874. Jahanshahi, E., 2013. Control Solutions for Multiphase Flow - Linear and nonlinear approaches to anti-slug control. Norwegian University of Science and Technology, The Faculty of Natural Sciences and Technology, Department of Chemical Engineering Ph.D. thesis. Jahanshahi, E., Oliveira, V.D., Grimholt, C., Skogestad, S., 2014. A comparison between internal model control, optimal pidf and robust controllers for unstable flow in risers. In: 19th World Congress, The International Federation of Automatic Control (IFAC’14), pp. 5752–5759. Jahanshahi, E., Skogestad, S., 2011. Simplified dynamical models for control of severe slugging in multiphase risers. In: 18th IFAC World Congress, pp. 1634–1639. Jahanshahi, E., Skogestad, S., 2014. Simplified dynamic models for control of riser slugging in offshore oil production. Soc. Petrol. Eng. (SPE 172998) 40–55. Jahanshahi, E., Skogestad, S., 2015. Anti-slug control solutions based on identified model. J. Process Control 30, 58–68. Jahanshahi, E., Skogestad, S., Grtli, E.I., 2013a. Anti-slug control experiments using nonlinear observers. In: American Control Conference (ACC), Washington DC, USA, pp. 1058–1064. Jahanshahi, E., Skogestad, S., Grtli, E.I., 2013b. Nonlinear model-based control of two-phase flow in risers by feedback linearization. In: IFAC Symposium on Nonlinear Control Systems 9th, pp. 301–306. Jahanshahi, E., Skogestad, S., Helgesen, A.H., 2012. Controllability analysis of severe slugging in well-pipeline-riser systems. In: IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, NTNU, Trondheim, Norway, pp. 101–108. Jansen, F.E., 1990. Elimination of severe slugging in a pipeline-riser system. University of Tulsa M.S. Thesis. Jansen, F.E., Shoham, O., Taitel, Y., 1996. The elimination of severe slugging - experiments and modeling. Int. J. Multiphase Flow 22, 1055–1072. Jansen, J.D., 2001. Smart wells. Jaarboek van de Mijnbouwkundige Vereeniging, Delft 32–43. Jansen, J.-D., Bosgra, O.H., den Hof, P.M.V., 2008. Model-based control of multiphase flow in subsurface oil reservoirs. J. Process Control 18, 846–855. Jepsen, K., Hansen, L., Mai, C., 2012. Modeling and Control of an Unstable Gas-lifted Oil-producing Well. Technical Report. Aalborg University Esbjerg, Department of Energy Technology. Jepsen, K., Hansen, L., Mai, C., Yang, Z., 2013. Emulation and control of slugging flows in a gas-lifted offshore oil production well through a lab-sized facility. In: IEEE International Conference on Control Applications (CCA), pp. 906–911. Johal, K., Cousins, A., 2001. Intelligent production riser. US Patent 6253855. Johal, K., Teh, C., Cousins, A., 1997. An alternative economic method to riserbase gas lift for deep water subsea oil/gas field developments. Society of Petroleum Engineers (SPE), Offshore Europe, Aberdeen, United Kingdom. Kaasa, G.-O., V. Alstad, e. a., 2008. Attenuation of slugging in unstable oil wells. Nonlinear Control. 17th IFAC World Congress, Seoul, Korea. Kang, C., Wilkens, R., Jepson, W.P., 1996. The effect of slug frequency on corrosion in high pressure, inclined pipelines. NACE International. Paper no. 20 Kongsberg, 2014. LedaFlow - The new multiphase simulator. User guide (2014). Kovalev, K., 2003. The slug suppression system in operation. Society of Petroleum Engineering (SPE), Offshore Europe 2003, Aberdeen, UK. Krima, H., Cao, Y., Lao, L., 2012. Gas injection for hydrodynamic slug control. In: IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, 1, pp. 116–121. Li, N., Guo, L., Li, W., 2013. Gasliquid two-phase flow patterns in a pipelineriser system with an s-shaped riser. Int. J. Multiphase Flow 55, 10. Liao, Q., Zhao, T., 2003. Modeling of taylor bubble rising in a vertical mini noncircular channel filled with a stagnant liquid. Int. J. Multiphase Flow 29, 411–434. Mahdiani, M., Khamehchi, E., 2015. Preventing instability phenomenon in gas-lift optimization. Iranian Journal of Oil & Gas Science and Technology 4, 49–65. Makogan, B., 2007. Patent on device for controlling slugging. http://patentscope. wipo.int/search/en/WO2007034142. Malekzadeh, R., Henkes, R., Mudde, R., 2012. Severe slugging in a long pipelineriser system: experiments and predictions. Int. J. Multiphase Flow 46, 9–21. Mansoori, M., den Hof, P.M.J.V., Jansen, J.D., Rashtchian, D., 2015. Pressure transient analysis of bottomhole pressure and rate measurements using system identification techniques. SPE Journal, SPE-176031-PA. McArdle, S., 2014. Multiphase flow gets closer to reality. EPmag.com Operator Solutions 74–75. McGuiness, M., Cooke, D., 1993. Partial stabilization at st. joseph. In: Proceeding of the 3rd International Offshore and Polar Engineering Conference, pp. 235–241.

284

S. Pedersen et al. / International Journal of Multiphase Flow 88 (2017) 270–284

Meglio, F.D., Kaasa, G., Petit, N., Alstad, V., 2010. Reproducing slugging oscillations of a real oil well. In: 49th IEEE Conference on Decision and Control, pp. 4473–4479. Meglio, F.D., Kaasa, G.-O., Petit, N., 2009. A first principle model for multiphase slugging flow in vertical risers. In: Joint 48th IEEE Conference on Decision and Control and 28th Chinese Control Conference, pp. 8244–8251. Meglio, F.D., Kaasa, G.-O., Petit, N., Alstad, V., 2011. Slugging in multiphase flow as a mixed initial-boundary value problem for a quasilinear hyperbolic system. In: American Control Conference on O’Farrell Street, San Francisco, CA, USA. Meglio, F.D., Kaasa, G.-O., Petit, N., Alstad, V., 2012a. Model-based control of slugging: advances and challenges. In: 2012 IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, pp. 109–115. Meglio, F.D., Petit, N., Alstadb, V., Kaasab, G.-O., 2012b. Stabilization of slugging in oil production facilities with or without upstream pressure sensors. J. Process Control 22, 809–822. Mokhatab, S., 2010. Severe Slugging in Offshore Production Systems. Nova Science Publishers, Inc. Mokhatab, S., Poe, W.A., 2012. Handbook of Natural Gas Transmission and Processing. Gulf professional Publishing. Mokhatab, S., Towler, B.F., 2007. Severe slugging in flexible risers: Review of experimental investigations and olga predictions. Petrol. Sci. Technol. 25, 867–880. Molyneux, P. D., Kinvig, J. P., 20 0 0. Method and apparatus for eliminating severe slugging in a riser of a pipeline includes measuring pipeline pressure and operating a valve. Patent: GB2358205, App. no. GB20 0 0 0 013331 20 0 0 0602. Montgomery, J., 2002. Severe Slugging and unstable flows in an S-shaped riser. Cranfield University, England Ph.D. thesis. Nemoto, R.H., Balio, J.L., 2012. Modeling and simulation of severe slugging with mass transfer effects. Int. J. Multiphase Flow 40, 144–157. Ogazi, A., Cao, Y., Yeung, H., Lao, L., 2009a. Slug control with large valve opening to maximise oil production. In: Society of Petroleum Engineers (SPE), Offshore Europe Oil & Gas Conference & Exhibition held in Aberdeen, UK. Ogazi, A., Ogunkolade, S., Cao, Y., Lao, L., Yeung, H., 2009b. Severe Slugging Control through Open Loop Unstable Pid Tuning to Increase Oil Production. BHR Group, pp. 17–32. Ogazi, A.I., 2011. Multiphase Severe Slug Flow Control. Cranfield University, School of Engineering, Department of Offshore, Process and Energy Engineering Ph.D. thesis. de Oliveira, V., Jäschka, J., Skogestad, S., 2015. An autonomous approach for driving systems towards their limit: an intelligent adaptive anti-slug control system for production maximization. In: 2nd IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, Florianopolis, Brazil, pp. 104–111. Pagano, D.J., Plucenio, A., Traple, A., Gonzaga, C.A., 2008. Controlling oscillations and re-starting ooperation in gas-lift wells. XVII Congresso Brasileiro de Automtica, Juiz de Fora, Brasil. Pedersen, S., Durdevic, P., Yang, Z., 2014a. Learning control for riser-slug elimination and production-rate optimization for an offshore oil and gas production process. In: The 19th World Congress of the International Federation of Automatic Control, pp. 8522–8527. Pedersen, S., Durdevic, P., Yang, Z., 2015a. Review of slug detection, modeling and control techniques for offshore oil & gas production processes. In: 2nd IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, 48, pp. 89–96. Pedersen, S., Mai, C., Hansen, L., Durdevic, P., Yang, Z., 2015b. Online slug detection in multi-phase transportation pipelines using electrical tomography. In: 2nd IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, Florianopolis, Brazil, pp. 165–170. Pedersen, S., Stampe, K., Pedersen, S.L., Durdevic, P., Yang, Z., 2014b. Experimental study of stable surfaces for anti-slug control in multi-phase flow. In: International Conference on Automation & Computing, 20. Cranfield University. Plucenio, A., Ganzaroli, C.A., Pagano, D.J., 2012. Stabilizing gas-lift well dynamics with free operating point. In: Proceedings of the 2012 IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, pp. 95–100. Pots, B., Bromilow, I., Konijn, M., 1987. Severe slug flow in offshore flowline/riser systems. SPE (Soc. Petrol. Eng.) 2:4, 347–356. Rassenfoss, S., 2014. E&P emphasis is now on the P. J. Petrol. Technol. 66 (10), 80–81. Rezende, E.F.A., Teixeira, A., Mendes, E., 2015. Br-sensor: An on-line data-driven soft sensor of downhole pressure. 2nd IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, Florianopolis, Brazil. Riccio, B.F., Teixeira, A., Teixeira, B.O.S., 2015. Downhole pressure soft-sensing using interacting multiple modeling. 2nd IFAC Workshop on Automatic Control in Offshore Oil and Gas Production, Florianopolis, Brazil. Sagatun, S., 2005. Riser slugging - a mathematical model and the practical consequences. SSPE J. Prod. Facil. 26 (2), 95–109. Sarica, C., Shoham, O., 1991. A simplified transient model for pipeline-riser systems. Chem. Eng. Sci. 46, 2167–2179. Sarica, C., Tengesdal, J.Ø, 20 0 0. A new technique to eliminate severe slugging in pipeline/riser systems. In: SPE Annual Technical Conference & Exhibition, SPE 63185, pp. 633–641. Schmidt, Z., Brill, J.P., Beggs, H.D., 1979. Choking can eliminate severe pipeline slugging. Oil Gas J. 12, 230–238. Schmidt, Z., Brill, J.P., Beggs, H.D., 1980. Experimental study of severe slugging in a two-phase flow pipeline riser-pipe system. SPE J. 20, 407–414.

Scibilia, F., Hovd, M., Bitmead, R., 2008. Stabilization of gas-lift oil wells using topside measurements. IFAC World Congress 17, 13907–13912. Silva, C.M.D., Nydal, O., 2010. Dynamic multiphase flow models for control. In: BHRG North American Conference on Multiphase Technology. Banff, Canada 7th. Sinègre, L., Petit, N., Menegatti, P., 2005. Distributed delay model for density wave dynamics in gas lifted wells. In: IEEE Conference on Decision and Control, and the European Control Conference, 44. Sinègre, L., Petit, N., Menegatti, P., 2006a. Predicting instabilities in gas-lifted wells simulation. In: Proceedings of the 2006 American Control Conference, pp. 5530–5537. Sinègre, L., Petit, N., Menegatti, P., 2006b. Predicting instabilities in gas-lifted wells simulation. In: American Control Conference, Minnesota, USA, pp. 5530–5537. SINTEF, 2014. The large scale loop. http://sintefloops.com. Sivertsen, H., Skogestad, S., 2005. Cascade control experiments of riser slug flow using topside measurements. In: Triennial World Congress, Prague, Czech Republic 16th, p. 6. Sivertsen, H., Storkaas, E., Skogestad, S., 2010. Small-scale Experiments on Stabilizing Riser Slug Flow. Chemical Engineering Research and Design, pp. 213–228. Skogestad, S., Postlethwaite, I., 2005. Multivariable Feedback Control: Analysis and Design. Wiley & Sons, Chichester, West Sussex, UK. Storkaas, E., 2005. Anti-slug control in pipeline-riser systems. Norwegian University of Science and Technology Ph.D. thesis. Storkaas, E., Skogestad, S., 2008. Controllability analysis of two-phase pipeline-riser systems at riser slugging conditions.. Control Eng. Pract. 15, 567–581. Storkaas, E., Skogestad, S., Godhavn, J.-M., 2003. A low-dimensional dynamic model of severe slugging for control design and analysis. In: 11th International Conference on Multiphase flow (Multiphase ’03), pp. 117–133. Sun, J.-Y., Jepson, W.P., 1992. Slug flow characteristics and their effect on corrosion rates in horizontal oil and gas pipelines. In: SPE 24787, pp. 215–228. Taitel, Y., 1986. Stability of severe slugging. Int. J. Multiphase Flow 12, 203–217. Taitel, Y., Barnea, D., Dukler, A.E., 1980. Modeling flow pattern transitions for steady upward gasliquid flow in vertical tubes. AIChE J. 345–354. Taitel, Y., Dukler, A.E., 1976. A model for predicting flow regime transitions in horizontal and near horizontal gas-liquid flow. AIChE J. 22, 47–55. Taitel, Y., Vierkandt, S., Shoham, O., Brill, J.P., 1990. Severe slugging in a riser system: Experiments and modeling. Int. J. Multiphase Flow 16, 57–68. Tchambak, E., 2004. Mitigation of severe slugging using gas injection. Interim PhD review, Report No. 2. Cranfield University, England. Tengesdal, J.O., Sarica, C., Thompson, L., 2002. Severe slugging attenuation for deepwater multiphase pipeline and riser systems. In: PE Annual Technical Conference and Exhibition. Paper SPE 87089 Terese Vardenær Syre, 2012. Anti-Slug Control with Non-Linear State Estimation. Norwegian University of Science and Technology (NTNU) Master’s thesis. TTA-Group, 2008. TTA-Group Consortium: Technology strategy for deepwater and subsea production systems 2008 update. TTA7. Vierkandt, S., 1988. Severe slugging in a pipeline-riser system, experiments and modelling. The University of Tulsa Master’s thesis. Viggiani, M., Mariani, O., Battarra, V., Annunziato, A., Bollettini, U., 1988. Pipeline simulation interest group (psig) - a model to verify the onset of severe slugging. PSIG Annual Meeting. Whitney, S., Larnholm, P., 2015. Split flow pipe separator with sand trap. WO Patent App. PCT/US2015/013,883. Wilhelmsen, M., 2013. Control Structure and Tuning Method Design for suppressing Disturbances in a multi-phase Separator. Norwegian University of Science and Technology, Department of Engineering Cybernetics Master’s thesis. Xiaoming, L., Limin, H., Huawei, M., 2011. Flow pattern and pressure fluctuation of severe slugging in pipeline-riser system. Fluid Flow Transport Phenomena, Chin. J. Chem. Eng. 19, 26–32. Xing, L., Yeung, H., Shen, J., Cao, Y., 2013a. Experimental study on severe slugging mitigation by applying wavy pipes. Chem. Eng. Res. Des. 91, 18–28. Xing, L., Yeung, H., Shen, J., Cao, Y., 2013b. A new flow conditioner for mitigating severe slugging in pipeline/riser system. Int. J. Multiphase Flow 51, 65–72. Xing, L., Yeung, H., Shen, J., Cao, Y., 2013c. Numerical study on mitigating severe slugging in pipeline/riser system with wavy pipe. Int. J. Multiphase Flow 53, 1–10. Xu, Z., Golan, M., 1989. Criteria for operation stability of gas lift. SPE 19362. Yang, Z., Juhl, M., Løhndorf, B., 2010. On the innovation of level control of an offshore three-phase separator. In: 2010 IEEE International Conference on Mechatronics and Automation, pp. 1348–1353. Yang, Z., Stigkær, J.P., Løhndorf, B., 2013. Plant-wide control for better de-oiling of produced water in offshore oil & gas production. In: 3rd IFAC International Conference on Intelligent Control and Automation Science, 3, pp. 45–50. Yocum, B., 1973. Offshore riser slug flow avoidance: Mathematical model for design and optimization. SPE 4312, London, UK. Zandvliet, M., Bosgra, O., van den Hof, P., Jansen, J., Kraaijevanger, J., 2006. Bang-bang control in reservoir flooding. In: 10th European Conference on the Mathematics of Oil Recovery (ECMOR X), Amsterdam, The Netherlands. Zhou, X., Jepson, W.P., 1994. Corrosion in Three-phase Oil/water/gas Slug Flow in Horizontal Pipes. NACE International. Paper no. 26