THE CSIRO S HYDRATES FLOW LOOP AS A TOOL TO INVESTIGATE HYDRATE BEHAVIOUR IN GAS DOMINANT FLOWS

Transcription

1 Proceedings of the 7th International Conference on Gas Hydrates (ICGH ), Edinburgh, Scotland, United Kingdom, July 7-,. THE CSIRO S HYDRATES FLOW LOOP AS A TOOL TO INVESTIGATE HYDRATE BEHAVIOUR IN GAS DOMINANT FLOWS Mauricio Di Lorenzo, Yutaek Seo and Gerardo Sanchez Soto CSIRO Earth Science and Resources Engineering Kensington, WA 65 AUSTRALIA ABSTRACT CSIRO has recently built a pilot size test facility to investigate natural gas hydrate behaviour in gas pipelines. It consists of a diameter, 4 meters long one pass flow loop in which compressed gas can be circulated in contact with water at pressure and temperature conditions in which most natural gas hydrates form. Hydrodynamic conditions can be set to achieve wavy and annular flow regimes typical of gas producing pipes. CSIRO s Hydrates Flow Loop has been designed to mimic as close as possible the conditions found at Australian offshore gas producing pipelines with less than % liquid load. In this work this test facility is described in detail. The temperature and pressure profiles along the test section of the flow loop are provided at different conditions and operational procedures to conduct steady state and transient tests are described. Some preliminary results obtained during the standard tests are presented and possible areas of application for the oil and gas industry are highlighted. Keywords: gas hydrates, flow loop, gas dominant flow NOMENCLATURE d pipe diameter [m] f friction factor P pressure [bar] s coat thickness T temperature [ºC] v gas velocity [m/s] P/l pressure loss [bar/m] gas viscosity [Pa s] gas density [kg/m 3 ] INTRODUCTION It has been recognized that a new flow assurance approach is required to reduce the cost of hydrate mitigation st rategies for the development of offshore deep water gas fields []. The concept of a risk based management of hydrates in pipelines is gaining acceptance in the O&G production community, based on the observation that the presence of hydrates does not always lead to pipeline blockages. Such approach must be necessarily built upon a sound knowledge of the mechanisms that govern hydrate formation kinetics and pipe clogging in gas production flow lines. Most of the research on hydrates behavior in flow lines has been limited to oil dominated systems in which the prevailing phase is black-oil or condensate. On the contrary, in gas flow lines, liquids (water and hydrocarbons) are usually present in small amounts which may increase with the production time. Even if gas dominated systems have some documented field data for hydrate blockage [], the authors are not aware of laboratory st udies performed in realistic conditions. These should be preferably conducted using dedicated flow loops designed to handle gas-water/oil systems at low temperatures and high pressures at hydrodynamic conditions representative of production scenarios. In Table flow loops used for hydrates studies are listed with their main specifications. It is indicated Corresponding author: Phone: +6 () Fax +6 () Mauricio.Dilorenzo@csiro.au

2 that most of the available data from these facilities have been obtained at liquid dominant flow conditions. Flow loop Exxon s F.L. [3] Texaco s F. L. [4] Tulsa Univ. F. L. [5] Korea Gas Corp. [6] Intertek Westport Tech. Center [7] Lyre F. L. [8]-[9] Geometry /m (*) ID:.97 L: 83.8 ID:.49 L: 3.7 ID:.76 L: 48.8 ID:.5 L: 4. ID:.9.3 L: 85.3 ID:.49 L: 4. P/bar T/ C Liquid loading Velocity m/s P: %.5 T: P: 38 Liquid 6.7 T: dominant P: 38 5%.8 T: P: 5 5%.8 T: 7.78 C P: 68.9 %.3 T: P: - T: % - % Table. Flow loops applied to hydrate research (*) ID: internal diameter, L: length Gas: 6 Liquid:. 3 To the authors knowledge only the Lyre flow loop at the Institut Francaise du Petrole has produced preliminary unpublished results showing that natural gas hydrates can be formed and transported at high volume fractions of gas (up to 98%) and high velocity. The CSIRO s Hydrate Flow Loop has been recently commissioned as an experimental facility for investigation of natural gas hydrate formation and flow behavior in gas production pipelines. In this work this test rig is described and some preliminary results are presented to highlight its potential applications. FLOW LOOP DESCRIPTION CSIRO s hydrates flow loop allows a mixture of gas and liquids at high pressure to be circulated within a stainless steel inch flow line, 4 m long, connected to a gas circulation compressor and injection liquid pumps. A simplified layout of the rig is presented in Figure. The flow loop can handle non-corrosive gases, pure water and brines and model oils with water or oil-soluble additives. The loop can be pressurized using high pressure gas cylinders or the natural gas storage system. This consists of six gas cylinders in which the gas from the city network can be pressurized using two gas compressors. The flow rates of the gas and liquids in the loop can be adjusted to obtain liquid volume fractions from up to % in the flow line. Here the fluids mixture can be cooled by means of a chiller unit circulating liquid coolant through a 4 pipe-in-pipe system. This flow loop operates at temperature and pressure conditions where most natural gas hydrates form. In Table the main technical specifications of the flow loop are summarized. Figure. Simplified layout of the CSIRO s Hydrates Flow Loop The mixture of fluids and hydrates transported in the flow line is collected in a two phase separator tank at the flow line outlet where the mixture is separated into a gas and a liquid phase. The gas leaves the separator and flows through the gas compressor before being injected into the flow line, while the liquid collected in the separator is discharged into the liquid storage tank A U shaped by-pass deviation, meters long and.9 meter deep, is available to simulate a low point along the flow line. The pressure and temperature at different points in the flow line (PT- to PT-6 in Figure ) are measured using RTD sensors and pressure transmitters with accuracies of C and.3 bar respectively. The measurement thermowells in the test section are approximately 6 m apart. The gas and liquid flow rates are measured using a turbine gas flow meter and a positive displacement liquid flow meter with accuracies of.3% and % respectively. The pressure, temperature and flow rate readings are transmitted to the data acquisition system and stored in a PC. The fluids inside the loop can be inspected and video recorded from high pressure visualization windows at four different locations in the flow line (VW to VW4 in Figure ) using high speed video cameras with x magnification lenses. All the

3 videos are taken in a transmitted light configuration. Temperature range -8. to +3 ºC Pressure range to 7. bar (7 psi) Pressure drop < 3.8 bar ( psi ) Liquid flow rate.6 to m 3 /h (sup. velocity <.4 m/s) Gas flow rate 5 to Sm 3 /h Gas volume fraction > 9% Phases Inner diameter Test section length Material Total volume Test section volume (sup. velocity < 8.5 m/s) Water (brine)/model oils/ natural gas. m 4 m 36 Stainless Steel 6 L 5 L Table. CSIRO s Hydrates Flow Loop technical specifications Further details can be found in Ref. [9]. EXPERIMENTAL PROCEDURES Two types of tests have been performed so far in the flow loop. One test is referred to as continuous flow test for studies under steady state flow conditions, similar to those found in a producing gas pipe. The second type is referred to as restart test to investigate transient flow conditions, simulating pipeline restart operations after a shut-down period. The experimental procedures are as follows. Continuous flow tests Initially all liquids are cleared from the tests section. The flow loop is pressurized to the test pressure. The test section is cooled down to the test temperature. In these tests the gas is circulated at a constant flow rate of 6. acfm (.7 m 3 /min) until a steady-state gas temperature profile is established along the test section. Water is injected at a constant flow rate of l/min in these tests. Temperature and pressure along the test section and the flow rates of gas and water are measured and logged into the PC during the experimental time (around minutes) High speed videos at 5 frames/second are recorded at the viewing windows. Restart tests The restart tests are performed using the U-bend section of the flow loop, where liquids can be accumulated and cooled down at static conditions and maintained at low temperature for a coolingdown time period, before restarting the gas flow. The test section is first cleared from all liquids and the low point is filled with water (7 ml in these tests). This corresponds to an in-situ water volume fraction of 7% (.44% volume fraction over the whole loop). The cooling-down period is initiated by cooling the flow loop from ambient temperature to the target value for the test. T he cooling rate is around.8 C/min. During this period the fluids in the low point are visually inspected through the viewing window and snap-shots are taken using the video camera located at the low point window. Once the required cooling-down period has elapsed, six hours for these tests, the compressor is turned on and the gas is forced to flow through the low point of the flow line at a constant flow rate. High speed videos at 5 frames/second are recorded for further analysis of the fast process occurring at restart. During the test the gas flow rate, pressure and temperature along the flow loop are measured and the readings stored in the PC. MATERIALS Deionised water and gas from the city network saturated with water was used in all these tests. The dry gas composition is shown in Table 3. Components Composition (mol%) CH C H C 3 H 8 3. i-c 4 H.4 n-c 4 H 9 i-c 5 H.4 n-c 5 H. C 6. CO.9 N.59 Table 3. City gas composition used in these tests

4 SII hydrates are formed with this gas. The hydrate dissociation curve, as obtained using Infochem Multiflash Vr.3.6.6, is shown in Figure. The pressure gradient due to the gas only is.75 bar/m, according to the Darcy-Weisbach equation. The distance between P and P6 is 33.4 m. Loop pressure drop P-P6 (bar) Expimental data Model calculations Time (min) Figure. Hydrates dissociation curve for the city gas RESULTS Continuous flow experiments The results obtained from two continuous flow tests performed at 3 bar (5 psi) at different temperatures are reported to show the hydrate flow characteristics as indicated by the pressure drop behavior. First the results obtained in a test performed at a pressure of 3 bar and a temperature of 8 ºC in a liquid/gas point of the phase diagram are shown as a reference. In Figure 3 the pressure drop across the whole test section (P-P6) is presented. In this test gas is continuously circulated. Water injection started at t= min and stopped at t= min. In the same graph the values for the pressure drop calculated for a natural gas system and a gas-water system at the experimental conditions of this test are shown according to the Beggs and Brill s model []. The input values for the model parameters are listed in Table 4. Ga s flow rate (m 3 /min).7 Water flow rate (l/min). Ga s density (kg/m 3 ) Water density (kg/m 3 ) Ga s viscosity (Pa s) -5 Water viscosity (Pa s). -3 Table 4. Parameters used for pressure drop calculations in the liquid/gas system At these conditions the model predicts a liquid holdup of 6% and a pressure loss of.37 bar/m. Figure 3. Pressure drop across the loop in the gas/liquid test. In Figures 4 to 7 the evolution of pressure drop at different sections of the flow line is presented. The data correspond to an experimental time of about minutes from the time when water started to be injected in the loop. The temperature profiles at the same sections are presented in Figure 8. The temperature values reported in the graphs in Figure 8 for each section refer to a mean value estimated from the temperature data at steady-state. In Figures 9 to the pressure drop data for a test conducted at a lower temperature profile and the same pressure are presented. The temperature data for this test are shown in Figure 3. Pressure drop P-P (bar) T=6. C Figure 4. Pressure drop P-P as a function of time at P=3 bar T=6 ºC Pressure drop P-P3 (bar) T=3. C Figure 5. Pressure drop P-P3 as a function of time at P=3 bar T=3 ºC

5 Pressure drop P3-P4 (bar) T=. C Figure 6. Pressure drop P3-P4 as a function of time at P=3 bar T= ºC Pressure drop P -P3 (bar) T=8. C Figure. Pressure drop P-P3 as a function of time at P=3 bar T=8 ºC Pressure drop P4-P6 (bar) T=8. C Figure 7. Pressure drop P4-P6 as a function of time at P=3 bar T=8 ºC Temperature ( C) T T T3 T4 T Figure 8. Temperature profiles in each section during the first test at P=3 bar Pressure drop P -P (bar) T=. Figure 9. Pressure drop P-P as a function of time at P=3 bar T= ºC Pressure drop P3-P4 (bar) T=7 C Figure. Pressure drop P3-P4 as a function of time at P=3 bar T=7 ºC Pressure drop P4-P6 (bar) T=4. C Figure. Pressure drop P4-P6 as a function of time at P=3 bar T=4 ºC Temperature ( C) T T T3 T4 T6 Figure 3. Temperature profiles in each section during the second test at P=3 bar Different flow regimes have been observed in the continuous flow experiments, which are represented in Figures 4, 5 and 6. In the

6 dispersed flow regime most of the water is immediately converted into hydrates and hydrate particles are pneumatically conveyed by the gas at high velocity. In Figure 4 at time T= the image from the window VW, located 5 m downstream the injection point, shows some hydrates stuck at the window. In the following snapshots hydrate particles, with a size of several millimeters, can be seen passing through the window from the right to the left. The time at which each photo is taken from the first one is also indicated below each snapshot. The velocity of these particles can be estimated to be very close to the superficial velocity of the gas (8.5 m/s). The dispersed flow regime is observed at lower temperatures (7 ºC and below) compared to the slurry flow regime. T= mm T=.67ms T=.ms T=.68ms T=3.35ms T=4.ms Figure 5. Snapshot showing a hydrate slurry at viewing window VW (P=3 bar, T= ºC) A we bcam attached on the viewing window VW4, located at the test section outlet, was used to monitor the fluids and hydrates coming out loop. In these experiments no liquid water has been observed at this point of the flow line. In Figure 6 a sequence of snapshots is presented where a hydrate slug can be seen passing through the window out of the test section. Figure 6-A exhibits the image of the window VW4 just before its arrival. Some hydrates particles can be seen at the test section outlet on the right. In Figure 6-B the hydrate slug is passing at the window and the photo in Figure 6-C shows the window just after it went through. -A- -B- Figure 4. Sequence of snapshots showing hydrate particles flowing through window VW (P=3 bar T=7 ºC - dispersed flow regime) A picture of the hydrate slurry observed in the viewing window VW is shown in Figure 5, which wa s obtained at a temperature of ºC. Small particles can be seen through the water film wetting the window. -CFigure 6. Sequence of photos showing a hydrate slug flowing out of the test section

7 Restart experiments Restart tests have been performed at a pressure of 97 bar and a temperature of 3.5 ºC at two different in-situ water volume fractions of 7% and % (volume of water/volume of. the void horizontal section of the U-bend). Figure 7 shows the temperature profiles at two points: T5 (in the horizontal section of the U-bend) and T6 (downstream the U-bend). The temperature quickly drops from 5 ºC (in the hydrate free region of the phase diagram) to a steady-state value between 4 and ºC in 5 minutes before rising sharply at the restart time, (after about six hours of cooling-down period. The oscillations in the steady-state temperature profile are due to the chiller cycling and the increase at restart is attributed mainly at the injection of hotter gas. Both restarts have been performed at the maximum gas flow rate of 6. acfm during a time period varying from to 5 minutes. increases with time more bubbles are formed enhancing the gas-water interfacial area and the hydrate formation rate. Pressure drop P4-P6 (bar) 3.5 water vol. fract.=7% water vol. fract.=% Figure 8. Pressure drop across the U-bend with the horizontal section 7% and % filled with water (P=97 bar, T=3.5 ºC) Temperature ( o C) T5 T6 T= ms T=9 ms Restart time Figure 7. Temperature as a function of time at points T5 and T6 during the shut-down and restart test at P=97 bar, T=3.5 ºC. The pressure drop values at restart across the U- bend (P4-P6) are presented in Figure 8. As soon as the gas is restarted (at Time: minute) the pressure drop sharply rises to stabilize later on with a small increasing trend over the experimental time. The pressure drop at restart can be estimated to be around bar for the 7% water cut test and bar for the % water cut. A sequence of snapshots taken at restart during the test at % water cut is presented in Figure 9. The time elapsed from the restart is indicated below each photo. At time T= the viewing window at the U-bend (VW3) can be seen fully flooded with water, as the pipeline is. The following photos show how the gas bubbles break through the water upon restart as soon as the gas phase reaches the U-bend. As the turbulence T=3 ms T=6 ms T=9 ms T=9 ms Figure 9. Se quence of snapshots showing the water-gas mixing and hydrate formation at restart in the test at % water volume fraction, P=97 bar, T=3.5 ºC.

8 The results from two shut-down and restart tests conducted at similar conditions are shown to explore the effect of different restart gas rates. In the lower velocity test, the gas flow rate was set at 4 acfm (gas velocity: 5.4 m/s) and the pressure and temperature at.3 bar and ºC respectively. In the higher velocity restart the maximum gas flow rate of 6. acfm (gas velocity: 8. m/s) wa s used and the conditions are: pressure P=.3 bar and temperature T= ºC. In both tests the in-situ water volume fraction is 7%. Figure shows the pressure drop (P4-P6) after restart obtained as a function of time in these tests. The pressure drop is consistently higher during the first 3 minutes in the test performed at a lower restart rate. Finally a sequence of snapshots obtained after restart in the higher velocity test is presented in Fig.. The time elapsed from the restart is indicated below each photo. Pressure drop (P4-P6) (bar) gas velocity=8. m/s gas velocity=5.4 m/s Figure. Pressure drop across the U-bend for two tests at different restart rates. The image taken at the T= ms exhibits the three phases present after the system has been maintained at P=.3 bar and T= ºC for six hours. The gas phase is on top, the liquid water phase at the bottom and a hydrate layer, about mm thick, is extended over the gas-water interface. Hydrate dendrites also appear below the hydrate layer protruding into the water phase. The following images show how the hydrate layer is disrupted by the incoming gas stream and the following strong mixing process taking place due to the turbulence in the flowing gas-liquid mixture. Fresh gas-liquid interface is generated during the process at a high rate allowing for a continuous and fast hydrates formation. After only.3 seconds the window is completely covered with hydrates. T= ms T= ms T=4 ms T=5 ms T=55 ms T=6 ms Figure. Se quence of snapshots taken upon a high velocity restart at P=.3 bar and T= ºC. DISCUSSION The preliminary results obtained in these tests show that hydrates in pipelines can display a variety of behaviors even in a relatively short experimental time under dynamic conditions as those simulated in this flow loop. At this moment a model to describe this complex behavior is not available. The main challenge here is to couple the kinetics of hydrate formation with mass and heat transfer processes determined by the thermodynamic and hydrodynamic conditions in gas dominant flows at high velocity. A qualitative analysis of the data obtained can provide useful insights at the phenomenological level. The first striking evidence is that hydrates form and evolve under the experimental conditions simulated in this flow loop at much shorter time scales (fraction of a second to several minutes) compared to oil dominated flow loops (hours to days).

9 The data obtained in the steady state tests indicate that the presence of hydrates can be detected from the pressure drop measurements. The values obtained in the short 6 meters sections (P-P, P- P, P3-P4) and in the long one (P4-P6) are well in excess compared to the values expected in absence of hydrates (. bar for the short and bar for the long section). Furthermore the time evolution of the pressure drop in the presence of hydrates deviates from a constant trend, and exhibits a much complex behavior, including the presence of transient peaks or steady increase over time. These measurements, complemented by visual observations, suggest that, at the experimental conditions of these tests, the system evolves from a slurry flow regime where hydrate formation is limited, at temperatures between 6 and ºC (subcooling < 8.7 ºC), to a dispersed flow regime where most of the water is converted to hydrates, at temperatures below 8 ºC (subcooling >.8 ºC). It is considered that the increase in the pressure drop in gas dominant conditions is mainly due to a restriction of the flow area of the pipe. The presence of particle dispersed in the gas flow or in the water phase will contribute to the pressure loss, but to a lesser extent. A reduction of the flow area in gas pipelines at hydrate forming conditions has been linked to the build up of a hydrate coat on the pipe wall (stenosis buildup) []. Depending on the mechanical properties of the coat this may or may not withstand the shear imposed by the gas stream and parts of it can be peeled off and transported downstream (sloughing). The presence of the peaks in the pressure drop readings, that has been associated with hydrate formation in production pipelines [3, 4], could be explained by such a mechanism. Using the Darcy-Weisbach equation: P v f l d () and the correlation for the friction factor in the turbulent regime: d v f.79, () the pressure loss can be related to the hydrate coat thickness, s according to [4] p l p l h d s d 4.75 (3) Here subscript refers to the hydrate free pressure loss, and the subscript h to the pressure loss in the presence of the hydrate coat. Using: (p) =. bar and (p) = bar for the short and long sections and (p) h = 4 bar as a typical height value of the peaks in Figures 7 and, the thickness of the transient deposits in the pipe walls can be estimated to be 4.8 mm for the short section P-P3 (Figure ) and 3.7 mm for the long section P4-P6 (Figure 7). This highly idealized model assumes that the coat is equally distributed along the pipe section, which may not be case, but these results indicate that the restriction to the flow should be quite substantial compared with the flow loop diameter. On the other hand, the slug flow behavior observed in Figure 6 could indicate that the hydrates accumulate preferentially in localized sections of the flow line, building up a loose permeable plug until the rising pressure is enough to dislodge it and transport it downstream. In this case slug flow pneumatic conveying theories should be used to predict the pressure drop. The shut-down and restart tests are more difficult to analyze. The visual observations suggest that hydrates are formed at a very high rate during the high velocity restart. Small fragments of hydrate particles and gas bubbles can be observed in the sub-cooled water phase within an extremely short time frame (Figure ). A slug develops and a hydrate-water-gas slurry starts to form under the intense mixing and agitation occurring in the low spot. This slurry is flushed downstream and appears to thicken with time sticking at the window and covering the glass completely. Hydrates may accumulate at hot spots such as the pipe line bends downstream the U-bend section from P5 to P6 and form restrictions to the gas flow. In a previous work evidence has been presented that indicates that the pressure drop through the U- bend can be related to the formation rate driven by high subcooling conditions [5]. In this work the transient tests have been performed at the close values of the subcooling (between 5 and 7 ºC). The results obtained with different amount of water in the low spot indicate that the pressure drop increases with the water content in the pipe, which can be expected. Even if initially the

10 interfacial area for gas-water contact is smaller and hydrates do not appear to have formed during the cooling-down period from visual observation (see Figure 9), as soon as the gas flow is restarted, conditions for a high rate of hydrate formation should be achieved due to the intense mixing and bubbling apparent from the snapshots. The results in Figure, showing a larger pressure drop obtained at a slower restart rate compared to the faster one, could be surprising. Even if the hydrate formation rate is expected to be higher at a faster restart due to a higher intensity of the turbulence, the residence time of the subcooled water should be shorter in this case and a lesser amount of hydrates available to deposit at the pipe wa lls could be produced. More data and theoretical modeling is needed to elucidate these important aspects of transient operations in the presence of hydrates. It has been suggested that high velocity startups could be less prone to hydrate blockages than the low velocity ones [6], but the evidence is still scarce. CONCLUSIONS Preliminary results obtained in the CSIRO s Hydrate Flow Loop has been shown in this work to highlight the potential application of this facility for investigations on hydrates behavior in gas dominated pipelines. These results indicate a complex, fast evolution of the hydrates over time occurring at a relatively short time scale at the conditions of the tests (high velocity, moderate to high subcooling). It has been demonstrated that useful information can be extracted from the pressure drop profiles together with visual observations and a preliminary description of the phenomena taking place at steady-state and transient conditions has been provided as a first step to the implementation of a mathematical model. More data and further theoretical development are needed for such purpose. For practical flow assurance applications it is envisaged that flow loop can be used to fine-tune hydrate inhibition strategies using chemicals at conditions representative of gas production pipelines, after pre-screening with conventional laboratory st udies using autoclaves or rocking cells. The CSIRO s Hydrates Loop has been recently upgraded to allow for longer experimental times and a research program on natural gas hydrates transportability is underway. REFERENCES [] Sloan D. A changing hydrate paradigm-from apprehension to avoidance to risk management. Fluid Phase Equilibria 4; 8: [] Sloan D. et al. Natural Gas Hydrates in Flow Assurance. New York: Elsevier Inc.,. [3] R.L. Reed et al. Some Preliminary Results from a Pilot-Size Hydrate Flow Loop. Annals of the New York Academy of Sciences 994; 75: [4] P.N. Matthews et al. Flow Loop Experiments Determine Hydrate Plugging Tendencies in the Field. Annals of the New York Academy of Sc iences ; 9: [5] M. Volk et al. Risk Based Restarts of Untreated Subsea Oil and Gas Flowline in GoM Final Report, Hydrate Flow Performance JIP, University of Tulsa, March 7 [6] J-H. Lee et al. Effect of Flow Velocity and Inhibitor on Formation of Methane Hydrates in High Pressure Pipeline. J. Ind. Chem. Eng. ; 8 (5): [7] J.W. Nicholas et al. Measuring Hydrate/Ice Deposition in a Flow Loop from Dissolved water in Live Liquid Condensate. AIChE J. 9; 55 (7): [8] T. Palermo et al. Flow Loop tests on a Novel Hydrate Inhibitor to be deployed in the North Sea ETAP Field. Annals of the New York Academy of Sc iences ; 9: [9] M. Di Lorenzo et al. Dynamic Behaviour of Natural Gas Hydrates Turing Simulated High Velocity Restart Operations in a Flow Loop. Proceedings of the 7th International Conference on Ga s Hydrates, Edinburgh, Scotland, UK, July 7-, [] Beggs H. D. & Brill J. P. A study of twophase flow in inclined pipes. Journal of Petroleum Technology, Transactions 973; 5: [] Sloan E.D. and Koh C.A. Clathrate Hydrates of Natural Gases, 3 rd Ed. CRC Press Taylor & Francis Group 8, 654:656 [] Ballard A.L. Flow-Assurance Lessons: The Mica Tieback. OTC 8384 (6) [3] Ref.. Appendix A. [4] Dorrstewitz F. And Mewes D. The Influence of Heat Transfer on the Formation of Hydrate Layers in Pipes. Int. J. Heat Mass Transfer 994; 37(4): 3-37 [5] Di Lorenzo M. et al. Hydrate Formation Characteristics of Natural Gas during Transient Operation of a Flow Line. SPE 3968 ()

11 [6] E. Leporcher et al. Multiphase Flow: Can we take advante of hydrodynamic conditions to avoid hydrate plugging during deepwater restart operations? SPE () ACKNOWLEDGEMENTS The authors thank CSIRO for permission to publish this paper