Cold flow a practical solution

Transcription

1 Cold flow a practical solution Roar Larsen SINTEF Petroleum Research, Norway Are Lund SINTEF Applied Chemistry, Norway Carl B. Argo BP Exploration, United Kingdom Solutions for cold flow may allow the oil industry to use longer satellite tie-backs at lower costs, using uninsulated steel pipelines. One of the main challenges for implementing such solutions, are the handling of gas hydrates. We solve this by recirculating already cooled fluid with seed particles to quickly transform water in the wellstream to dry, inert hydrates. High-pressure tests have shown this to work for a wide range of conditions. The technology is simple, the pipeline round-trip piggable, and injection for other flow-assurance purposes is simplified. We describe the technology, with laboratory results, and with benefits for field developments. 1 BACKGROUND Multiphase wellstream transport exceeding the present-day subsea tie-back distances is of strategic importance for future deepwater field developments, and a possible enabler for economical exploitation of many marginal satellite fields and prospects at moderate water depths. The current direction of the industry is to develop new fields with multiple subsea tiebacks to host hubs, field centres or onshore facilities for final processing. This is often based on the requirement to make more effective use of existing infrastructure. One of the present key project stoppers for long distance tiebacks, is incomplete and very expensive technology for hydrate control. Removal of produced water upstream multiphase flowlines by future downhole and subsea separation systems will be a significant improvement, but will most likely not eliminate the problems of a free water phase completely. Hydrate inhibition with continuous injection of chemicals, combined with insulated or even heated pipelines, may remain a prerequisite and a cost-driving factor unless other, innovative solutions are developed. Cold flow, with slurry transport of hydrate particles and possibly other solids, is an attractive concept. However, it entails substantial challenges in the areas of hydrate formation, wax deposition, and other phenomena associated with low-temperature fluid flow. The goal of "pure" cold flow solutions is to find a technology solution which allows subsea field development based

2 on ultra-long cold multiphase wellstream transport, i.e. uninsulated pipelines, no heating requirements and no chemical additives. Since 1998, SINTEF has carried out a JIP project (CONWHYP Conversion of water to hydrate particles) to test and verify a radical new technology for solving hydrate problems in offshore oil pipelines. The underlying technology idea has been patented by SINTEF (1). The overall goal of the current project, called SATURN, is to bring cold flow technology from the laboratory to a state of offshore implementation. The project has so far focussed mainly on gas hydrate issues, but is currently extending the technology to incorporate other cold flow issues, e.g. wax and asphaltenes. 2 HYDRATE GROWTH PHENOMENA There is little information in the open literature about the details of hydrate growth in real systems. The processes described in the following, were first discussed by Lund and Larsen (2). 2.1 Normal hydrate formation in hydrocarbon systems Hydrates start to nucleate close to the hydrocarbon phase on a water droplet in gas, oil, or condensate phases (3). Hydrates grow along the surface of the droplet until it is completely covered with a thin hydrate layer. Water then penetrates from the interior of the water droplet to the hydrophilic hydrate surface next to the hydrocarbon phase through microperforations or small cracks in the hydrate film, as illustrated in Figure 1. The formation rate decreases as the hydrate layer increases in thickness, depending on the hydrate formation driving force and shear forces on the droplets. After a relatively short time no further conversion of water to hydrates is observed. Sugaya and Mori (4) have described the results of this effect for hydrates from hydro-fluorocarbons (HFCs), and Seleznev and Stupin (5) did the same for spraying of water droplets into a vessel with methane gas. Hydrocarbon phase Water droplet Crack or microperforation Hydrate layer Figure 1 Growth of hydrate layer on a water droplet surrounded by hydrocarbons. From (2).

3 If a water droplet covered by a hydrate film hits e.g. a pipe or reactor wall in a turbulent system, the impact may create larger cracks in the film. Subcooled water inside the droplet will drain through these cracks and spread on the dry hydrophilic hydrate film. Hydrate-forming species close to the pipe wall may convert this water to hydrate quickly, often resulting in deposition of the hydrate/water droplet on the wall. Based on experimental studies (Aalvik (6), Fauskanger (7), Løken (8) Austvik et al. (9), Austvik (1), Lund et al. (11, 12), and Lund and Lysne (13)) there seems to be a large number of common basic elements in the conversion process. In turbulent liquid systems the water phase is often distributed in the hydrocarbon phase as rough, unstable water-in-oil emulsions. As surface tension of the droplets increases due to the hydrate layer, the water droplets agglomerate to larger droplets or water lumps in order to minimize surface area, as shown in Figure 2. In turbulent liquids these water lumps will change form, surface area, and volume continuously. The thin hydrate layer on the water lump will thereby often be broken, giving new water-hydrocarbon interfaces where more hydrates quickly form. The turbulent forces will also create small hydratecovered water droplets as illustrated in Figure 3. Due to the hydrophilic properties of the hydrate surface, these droplets will be absorbed in the water lumps giving a slush-like appearance (Aalvik (6)). Hydrocarbon phase Hydrate Waterphase Figure 2 Agglomeration of water droplets after hydrate initiation. From (2). Further growth and particle accumulation will make the outer area of the lumps stiffer. When these lumps collide with each other or with e.g. a pipe wall, free water from the lump interior will spread to the outer hydrate surface, acting as glue for agglomeration of the lumps to bigger lumps or plugs or to the pipe wall, by converting to hydrates. The hydrate layer covering lumps or plugs increases in thickness until internal pressure gradients due to capillary forces and volume changes breaks it down to smaller hydrate bits as illustrated in Figures 4 and 5. This process continues until the lumps have been broken down to a powder-like appearance assuming that flowing conditions can be maintained throughout the process. In a realistic industry case, the pipeline will more often than not be plugged before this stage is reached.

4 Hydrocarbon phase Hydrate Waterphase Figure 3 Hydrate-covered water droplets inside larger water lumps. The start of slush-like flow behaviour. From (2). Hydrocarbon phase Hydrate/water lump Micropores Hydrate Figure 4 Ongoing conversion of water to hydrate by water transport to surface of large lumps. From (2). 2.2 Implications for cold flow As is indicated by the previous discussion, if water is to be transported as a stable, more or less dilute hydrate slurry, the hydrates must contain no free water. This is in order to avoid any problems from deposits on the pipe wall or agglomeration at stop or start of the flow. In order to achieve this, free water will have to be converted to hydrates near the well in a fast and controlled manner, without any of the deposition-prone intermediate stages. This can be done without chemical additives, by using the fact that hydrate surfaces exhibit a high degree of hydrophilic behaviour (Hirata and Mori (14) and Lund et al. (15).

5 Figure 5 Break-up of large hydrate lumps. From (2). If a warm oil or condensate well stream containing free water droplets is mixed with a cooled well stream containing a large number of existing dry hydrate particles, the water will quickly coat the hydrate particles with a thin water film. If the temperature conditions are right, the water will be converted to hydrates by growing from the existing hydrate surfaces and outwards, as indicated in Figure 6. This should ensure that no free water is encapsulated within the hydrate particles, and therefore no later agglomeration or deposition can take place. Water Oil/Condensate Hydrate Figure 6 A water droplet wetting a dry hydrate particle. After wetting, the water layer is converted to hydrates from the existing hydrate surface and outwards. From (16).

6 3 THE SATURN COLD FLOW CONCEPT 3.1 Basic ideas Figure 7 gives a schematic overview of the processes described in the following. At the start of a hydrocarbon flow, either sub-sea, or onboard a minimum processing platform, or conceivably downhole in individual production wells, water separation is efficient enough that after cooling and condensation, no more than a certain amount of water (on the order of 1 to 2 vol %) is present in the fluid stream. Figure 7 Schematic description of the hydrate reactor concept. From (2). If water is to be transported as a hydrate slurry, the water to hydrate reaction has to be driven quickly to total completion such that no more solids will be able to deposit out. i.e. no water is available for further hydrate formation. The SATURN 1 concept of seeding and growing the hydrate from the inside and outwards eliminates the availability of free water outside a short reaction zone. In the rest of the pipeline, hydrates will only be seen as inert powder particles flowing along with the liquids. From a sub-sea well, or a minimum-processing platform, a water-containing hydrocarbon fluid stream is lead into the hydrate reaction part of the system, where hydrate particles in a cold fluid stream are mixed in (pumped from a downstream splitter). The water in the stream will be converted to dry hydrate particles well before it reaches the splitter, where some of the cold hydrocarbon fluids and dry hydrate particles are split off, and recirculated to the process starting point. In principle, the reactor is merely a continuous part of the un-insulated pipe. The particles escaping without being recirculated, will be solid, dry hydrate. The quick additional cooling in the bare steel pipes will have brought the system close to ambient temperature, and this will prevent further condensation of water from either liquid or gas hydrocarbon phases through the rest of the pipeline. The aim for the separation, or flow-splitting process is to have some of the fully converted particles (preferably the largest ones) proceed downstream, in an amount corresponding to the content of water in the inflow to the system. 1 The name for the concept was chosen due to the similarity with the rings of the planet Saturn (ice/hydrate particles in recirculation), and from the realization that Saturn was the old Roman god of "seeding and sowing".

7 The hydrate powder will not melt back to free water and natural gas until temperatures rise or pressures become too low - which will be at the end of the transport. The powder may e.g. be mechanically separated from the bulk liquid phase by a sieve (unlike dispersant-induced emulsions which are often difficult to break). Depending on the fluid system, the particle density may even be different enough from the bulk liquid that the particles may be drained from the bottom of the separator. 3.2 Field implementation The SATURN concept will not be limited to handling a single well or template, but can be used to produce a cold slurry from a chain of wells or templates possibly an entire field. The loop may then be enlarged so that only one split or recirculation point is needed. Implementation of the SATURN process must be evaluated in the initial stage of a field development design. Specifically, this means that one should be utilizing the distance between the templates for wellstream cooling and water-to-hydrate conversion. Thus, a first template gives the hydrates necessary for the second template and so on. The initiation mass in terms of hydrate slurry for the first template might as an extreme be returned all the way from the processing terminal - giving longer pipe lengths, but almost eliminating the need for subsea equipment (e.g. pumping). Also, such a solution eases chemical treatment of the fluids for other purposes, as well as most emergency operations. Figure 8 shows some possible implementations of SATURN. A large number of existing, new and planned oil or gas fields already use pipeline loop solutions of one type or another (e.g. Åsgard, Girassol, Ormen Lange). This ensures greater flexibility in production rates (one or two producers), the possibility of "round-trip" pigging from platforms, and eases injection of chemicals or even fluid replacement of the entire pipeline volume. A field solution with SATURN will demand the return flow of a certain percentage of the produced fluid, depending on the number of "inflow points" for warm fluids in the system. The more wells/templates to be incorporated, the less cold "hydrate fluid" is needed in the return flow. This will again of course be reflected in the economy of the development. Simulations on the behaviour of a real oil field (from the North Sea) using SATURN are shown in section 5.

8 SATURN visualization Shore/host platform or shallow water facility Subsea field layout ~1-2 km? "Large loop" Pipeline Detail C Detail A "small loop" Detail B Detail C Hydrate slurry for processing ONSHORE/SHALLOW HOST ONSHORE PROCESSING Alternative: Detail A Slurry in Slurry out Split Pump(s) Uninsulated well line(s) T =~ 4-6 C Cold hydrate particle stream Manifold Detail B Hydrate slurry to shore or shallow host Towards next well From last well Y split Alternative: Field hydrate reactor loop From last Slurry to shore/host PUMP (in shallow water or on platform) Slurry from shore/host To first well To 1st Figure 8 Possible field implementation of the SATURN process.

9 4 LABORATORY TESTS Some of our experimental results, and details of the laboratory set-up, have been reported earlier, in (16). The facility is essentially a 2.5 cm ID loop, with a pressure limit of 1 MPa. The loop is liquid-filled (about 2 litres), generally with Exxsol D8 and methane and propane. The majority of the experiments have been conducted in confidential projects, but in the following we give an overview of the experiments performed, and the most important results obtained. 4.1 Experiments and results A number of experiments lasting from 1 hours to several days were performed. The subcooling below equilibrium at operating conditions of 7 MPa and 4 C was usually greater than 1 C. Given some equilibration time, water was injected at carefully controlled rates. Conversion of the water droplets to hydrate particles, and their deposition and agglomeration characteristics were then studied by visual observations, both real-time and by video recordings. Pressure drops, as well as the temperature at several points were monitored Dry hydrate behaviour Normally, the hydrates produced were dry, loose agglomerates which dispersed throughout the bulk volume, and easily flowed with the liquid phase. Provided that a (low concentration) starting mass of hydrates was present, the production of this type of hydrates was easily achieved and significant amounts of water could be handled. However, when water was injected without a starting mass present, deposition of hydrates was frequently observed. Withdrawal of hydrate mass from a given experiment for use in later ones was very successful. The pressurized withdrawal cylinder was stored cold for up to several weeks, filled with concentrated hydrate slurry, without creating problems with the re-injection. The hydrates retained their dry, non-sticking nature during storage, and flowed just as easily as before when injected back into the loop. Also, the re-start of a hydrate-filled system after stagnant conditions over night was usually successful. Problems in the loop only occurred when free water was present Experiments on water content The water content in the incoming stream that can be handled successfully, depended on a number of factors. Most important was the amount of hydrate already present in the reactor (i.e. the amount coming from the recirculation). Other factors were mixing efficiency between oil and water, and the level of turbulence in the reactor itself. The whole range from large droplets to finely dispersed mist was tried in the water injection system, and the best strategy seemed to lie somewhere in-between these extremes. It was established over the course of the experiments that a hydrate volume fraction of about.3 % in the flowloop was necessary for the system to handle any water injection at all. Any buildup of hydrate mass before that (without chemicals) was very time-consuming, and had to be done with extreme care.

10 4.1.3 Hydrate particle morphology The morphology results from the experiments indicated a connection between the particle shape and the injection temperature (an effect of the subcooling, or the driving force for crystallization). At low injection temperature (high driving force) the particles were branched, and "snowflakelike" in appearance. Although still dry and transportable, these shapes resulted in more flocculation of particles. Even though these aggregates were readily broken apart, it is a less desirable shape than more rounded particles. Higher temperatures for the injection (lower driving force), and possibly also for the bulk fluid, seemed to give rounder, more solid particles Salt and black oil effects As hydrates form, salt is excluded from the hydrate phase, and an increasing concentration of salt in the remaining water may eventually block further hydrate formation. This will result in free water in the pipeline. Theoretically, this should not be a problem for the hydrate handling capabilities of SATURN, as the problem with free water only arises when it is capable of forming hydrate (thereby gluing hydrate particles together or to the wall). The periodical injection of saline water (NaCl solution) at rates comparable to those used for fresh water did not reveal any change in hydrate behaviour at low free water contents. The initial salt concentration in the water was 3.5 weight %, to mimic seawater. During the experiments, this concentration would increase towards the NaCl solubility limit. As the amount of concentrated free salt water increased there was a tendency for the hydrate particles to stick together in loosely adhered agglomerates. There were no observations that indicated that the hydrates were adhering more to the walls of the loop than before. In a system where the greater part of the moist hydrate will be transported out of the loop system, the free saltwater content will reach a stationary level, and the salt concentration will be such that no hydrates can form from this water. The final experiment was performed on a real hydrocarbon field fluid supplied by BP. Both fresh and salt water were used in the experiment. Fresh water was used to build up the hydrate mass. It was clear that the tendency of the hydrate to adhere to the PMMA window sections was greatly reduced compared to the model system experiments with Exxsol D8. Injection of salt water gave the same behavior as for the Exxsol D8 system. The loop was able to handle quite high water injection rates without any plugging tendencies being observed. 4.2 Remaining challenges and further work The SATURN concept remains an active on-going project, where the goal is to perform pilot testing, then field trials, and reach commercial maturity within the next few years. There is a realization that for cold flow to become practical, it is necessary to have an integrated solution for hydrates, wax, and any other solids that may precipitate out in the fluid system (e.g. asphaltenes and scale (inorganic salt deposition)). There are already indications that the SATURN technology will be able to incorporate these effects, and this is currently the key project priority.

11 Some of the most important remaining challenges being dealt with in the current and near-future stages of the project are: - the influence of other solids in the system, most importantly wax and asphaltenes - evaluation of the range of oil and system characteristics which are suitable for the concept - determination of maximum gas-oil ratios (GORs) which will still allow effective transport of the liquid/solid slurry - identification of practical implementation aspects, and interfacing with engineering companies for the building of needed facilities - evaluation of any new challenges faces by endpoint processing facilities when they receive a cold slurry instead of the usual warm fluid mixture The final two points from the list above are challenges which will have to met in co-operation with engineering companies. Particularly in the area of subsea processing of wellstream fluids (e.g. subsea separation), SATURN and other emerging subsea technologies may have significant synergy effects. First-generation subsea separator concepts are already demonstrated to be effective enough for SATURN purposes, while SATURN itself promises to solve the problems of residual water. This removes the need for further costly and complicated subsea separation improvement. 5 FIELD IMPLEMENTATION MODEL 5.1 Model description As an aid for developing subsea oil and condensate fields by the SATURN loop concept, a flow sheet simulator has been made. The simulator is a sequential modular type, implemented in Matlab. Any number of hot fluid (oil/condensate/water) streams from subsea wells, templates or fields to a SATURN loop may be simulated by the program. A basic simulation may e.g. contain the following types of simulator units: Sources: Mass flow, composition, temperature and thermodynamic properties of the streams are specified (for both the warm stream from the well/template, and the cold recirculation stream). Mixers: Used mainly for calculations of mixing temperatures. Plug flow reactors: Conversion to hydrate and heat exchange with the environment is calculated. Pipes: Heat transfer to the environment is calculated, but not conversion to hydrate. To simulate a chain of templates or wells, the required units of suitable lengths can be coupled in any desired number or combination, One additional module is the hydrate formation rate model. This uses fundamental driving forces (chemical potentials), specific fluid composition, and correlations from experimental data to give a state-of-the-art model for hydrate kinetics. This is an in-house proprietary SINTEF model which is currently unpublished. 5.2 Model results The North Sea case which is simulated here is briefly described by the "Standard value" column in Table 1. The variation range for parameter sensitivity evaluations is also included. We have

12 simulated 3 wells/templates in the same SATURN loop, with a 22.5 cm ID pipeline. Several of the parameters used are based on our experimental results. In these calculations, the flow from each template is equal, unless otherwise stated. Table 1 SATURN simulation parameter range Parameter Standard value Variation Sea temperature 4 ºC -2, 2, 6, and 8 ºC Wellstream temperature 4 ºC 2, 5, and 6 ºC Recirculation flow temperature 4 ºC 6, 12, and 14 ºC Reaction rate for hydrate formation Calculated from our model, and normalized to 1.1,.25,.5, and.75 Well water cut 13 wt%.5, 3, 5, 1, and 2 wt% Recirc. hydrate fraction 23 wt % 5, 1, 15, and 2 wt% Overall heat transfer coeff. Calculated for realistic bare steel, and normalized to 1.15,.2,.3,.5, and.75 Production rate Actual field estimate,.5,.8, Recirculation (cold stream) rate normalized to times the template production, normalized to 1 2, and 3.1,.2,.4,.5,.7,.8,.9, 1.1, 1.2, 1.3, 1.4, 1.5, 1.7, 2, and The first template Only the relevant results relating to cooling efficiency and pumping requirements are discussed here. The other parameters showed little or trivial influence on the reactor length. For the standard case, a reactor length for elimination of all water, of 274 m was calculated. The first set of results concerns the effect of different temperatures for the sea, the wellstream, the recirculating flow, see Figure 9. Here, the reaction zone length (for complete water conversion) is plotted as a function of other parameters. For the sea temperature, the variation is not very large, from 28 m for 2 ºC, to 365 m for 8 ºC. As most realistic cases will lie within these bounds, this parameter is judged not to be limiting. Varying the temperature of the incoming wellstream over a large interval also is encouraging; going to the high (basically "uncooled") temperature of 6 ºC increases the reactor length by less than 1 m, while colder wellstream of course helps the process by cutting down on the required length. Variations of this magnitude should be easily designed into the base case for a given application, with very reasonable safety factors. Equally encouraging is the observed effect of the temperature of the hydrate-carrying recirculation cold-stream: increasing the temperature to 12 ºC only adds 16 % to the length. Figure 1 shows the results for the reactor length when two of the most important parameters are studied: the production rate and the recirculation rate. These show a linear influence, and can be balanced against each other.

13 [m] T sea [C] [m] T wellstream [C] [m] T recirculation coldstream [C] Figure 9 The required reaction zone lengths are plotted on the y axes, as functions of the sea temperature (top left), the wellstream temperature (top right), and the temperature of the recirculation stream (bottom). Figure 11 shows how some other parameters influence the length of the reaction zone. The reaction rate has to drop well below half that of our best estimate in order to have a notable effect on the reactor. The water cut in the production stream shows intuitively an almost linear relation with the length of the reaction zone. The hydrate fraction in the cold recirculation flow shows a very flat response, indicating a low sensitivity The second template Continuing downstream, the next template has to be phased in. Continuing on from the 1st mixing and hydrate formation, the cold stream will have grown in volume (and velocity), and will generally be even better able to handle the cooling and conversion at the next template provided that the distance between templates is large enough so that the fluids have cooled down enough. As seen in Figure 9, the cooling does not have to reach seabed temperature, as significantly warmer temperatures are tolerated. A large number of simulations have been run to study the temperature effects. One example is seen in Figure 12.

14 [m] Production rate [/orig] [m] Cold stream rate [/orig] Figure 1 The required reactor lengths are plotted on the y axes, as functions of the production rate from the well/template (left), and the recirculation stream rate(right). The temperatures of the streams are as in the standard case (4 ºC well, 4 ºC cold). When all water has been converted to hydrates (after about 3 m), the temperature drops rapidly towards the sea-bottom conditions (here 4 C), going towards 1 m downstream. The previous section showed that even a temperature of 14 C was acceptable coming into the reaction zone, and this is reached already 5 m downstream, and this will correspond to an acceptable distance between templates, if needed. This relation has been studied for a range of initial coldstream rates to the first template, sea temperatures, heat transfer coefficients, and lengths of the first template reaction zone. Figure 13 shows a typical example, for a case with mostly standard values of the parameters. The results in Figure 13 show that when the distance between the templates approaches 9-1m, the second template is no longer directly influenced by the conditions at template 1. The fact that the cold-flow is almost 2 times larger than into template 1, gives a shorter reaction zone for this template. This will be a common feature templates further downstream in the chain have a larger window of operability of its parameters, due to the beneficial impact of the larger volume of cold-flow. This means that e.g. templates with higher production and/or larger water cuts, etc., can be handled as long as they are not the first template in the loop.

15 [m] WC [wt%] [m] Hydrate fraction [-] [m] Reaction rate [/orig] Figure 11 The required reactor lengths are plotted on the y axes, as functions of the normalized hydrate reaction rate (bottom), the hydrate fraction in the cold stream (top right ), and the water cut in the wellstream (top left).

16 Figure 12 Temperature evolution downstream from the 1st template. The dashed line shows the equilibrium temperature for hydrate formation The third template The inclusion of the third template is not very different from the second one. The situation is actually even more favourable, as the cold flow is now even more substantial, and will cool even more efficiently. In Figure 14, we show a case where templates 1 and 2 have been placed only 227 m apart (minimum distance for all water from 1 to be converted before reaching 2), with standard values for most parameters. The results indicate that a distance between templates 2 and 3 as short as 2 m is enough to be in a reasonable parameter range, and that again the variation levels off with maximum robustness for parameter variations after about 1 m. The considerations in these sections show that when the process has handled the first template, there are few problems in phasing in more warm wellstreams further downstream provided that the flow rates and pressure drops in the single pipe remain manageable. Any adverse effects from these factors can be countered by increasing the pipe size, or by branching.

17 Reaction zone length, template 2 [m] Temp. cold stream no. 2 [C] [m] 15 6 [C] Distance between template 1 and 2 [m] Figure 13 Data for template 2, downstream from template 1. The x-axis shows the distance between the templates. The temperature of the cold stream into template 2 (right axis) and the length of the reaction zone downstream from template 2 (left axis) are plotted [C] Temp. of cold stream no. 3 [C] Reaction zone length, template 3 [m] [m] Distance between template 2 and 3 [m] Figure 14 Coldstream temperature into template 3 (left axis), and reaction zone length at template 3 (right axis) as a function of the distance between templates 2 and 3 in the extreme case of templates 1 and 2 being placed as close together as possible.

18 6 COLD FLOW BENEFITS BP has carried out internal reviews on cooling rates and the economic prize of cold flow technology, using the assumption that a viable subsea technology could be developed. In terms of benefits for deepwater oil developments, BP believes that cold flow could offer in particular two main benefits: 1. Being an enabling technology for longer distance tie-backs. 2. Lower cost tie-backs. Cost savings of several tens of millions of US dollars are thought to be possible over conventional dual flowline insulated designs versus single bare steel pipe. Tie-backs are at present restricted to rather short distances, and cold flow will help extend the reach from existing installations towards perhaps 7-1 km for oil systems. It is immediately obvious that leaving out heating or insulation systems is a cost saver some industry rules-ofthumb indicate USD.5 million/km savings in going from insulated pipe to bare steel pipe, and USD 1 million/km if the alternative is a bundled pipe. In a more general discussion, the potential benefits of cold flow through SATURN or other means can be grouped into three broad categories: improvements in HSE and environmental aspects, operational aspects, and system-wide cost savings. The most significant aspects from an HSE or environmental viewpoint, are that new surfacepiercing structures can be eliminated through enabling direct subsea production to shore, shallow water or a host platform with available capacity. This type of safer operation removes people from deepwater offshore operations. In addition, our cold flow concept is a greener solution, due to the elimination or reduction of many chemical additions. Also, produced water is probably easier to handle and process as a solid. The operational advantages are first of all due to the elimination of hydrate and wax blockage risks. This means that a host of injection and control systems will become unnecessary. Successful cold flow means a simpler, steady operation of a low-maintenance system in thermal equilibrium with its surroundings. All of the above factors also contribute towards making cold flow an even more economically attractive solution. In addition, it seems probable that efficient cold flow may be the deciding factor a project enabler in terms of getting distant and/or marginal satellite fields to become economically viable. 7 CONCLUSIONS We have presented a laboratory-proven concept for solving some of the main challenges associated with cold flow slurry transport of liquid hydrocarbons and solids over long distances without heating, insulation, or chemical additions. The SATURN cold flow technology uses seeding and cooling by recirculation of precipitated solids (most importantly gas hydrates) to effectively form transportable, non-depositing particles over a very short transport distance: a

19 reaction zone on the order of a couple of hundred metres. Simulations of field applications of the concept show great promise, and substantial benefits both economically and otherwise have been discussed. The economic impact is particularly associated with being able to use bare steel pipelines for the transport. 8 ACKNOWLEDGMENTS The authors are grateful for project financing over the years by Saga Petroleum, Norsk Hydro, BP, and the Research Council of Norway (RCN). RCN was also instrumental in providing funding for project start-up. FMC Kongsberg Subsea was a valuable technology discussion partner in an early project phase. Special thanks are due to Kai W. Hjarbo and Vibeke Andersson for carrying out many of the experiments, John Morud and Paal Skjetne for developing the simulation model, and Marita Wolden for performing a large number of the simulations. 9 REFERENCES (1) A. Lund, D. Lysne, R. Larsen, and K.W. Hjarbo, Method and system for transporting a flow of fluid hydrocarbons containing water, Norwegian patent no. NO , British patent no. GB 2,358,64, Eurasian Patent no (2) Lund, A. and Larsen, R.: Conversion of Water to Hydrate Particles Theory and Application, paper presented at 14 th (2) Symposium on Thermophysical Properties, Boulder, June (3) B. Kvamme, Proc. 2 nd Int. Conf. Nat. Gas Hydrates, Toulouse, France, June 2-6, 1996, p (4) M. Sugaya and Y.H. Mori, Chem. Eng. Sci., 51 (1996) (5) A.P. Seleznev and D.Yu. Stupin, Kolloid Zh., 39 (1977) (6) J. Aalvik, Diploma Thesis, Dept. of Refrigeration Eng., The Norwegian Inst. of Technology, Trondheim, Norway, (7) R.F. Fauskanger, Diploma Thesis, Dep. of Chem. Eng., The Norwegian Inst. of Technology, Trondheim, Norway, (8) K.-P. Løken, Diploma Thesis, Dep. of Chem. Eng., The Norwegian Inst. of Technology, Trondheim, Norway, (9) T. Austvik, A. Lund, D. Lysne, E. Lindeberg, and K.-P. Løken, SINTEF Report STF21 A96 (199), Trondheim, Norway. (1) T. Austvik, Dr.ing. Thesis, Division of Thermodynamics, The Norwegian Univ. of Sci. and Technology, Trondheim, Norway, (11) A. Lund, D. Lysne, Ø. Grande, and T. Austvik, SINTEF Report STF21 A9234 (1992), Trondheim, Norway. (12) A. Lund, D. Lysne, T. Austvik, and B. A. Ardø, SINTEF Report STF21 A914 (1991), Trondheim, Norway. (13) A. Lund, D. Lysne, SINTEF Report STF21 A9273 (1992), Trondheim, Norway. (14) Hirata, A. and Mori, Y.H.: How liquids wet clathrate hydrates: some macroscopic observations, Chem. Eng. Sci. (1998) 53, No. 14, 2641.

20 (15) A. Lund, O. Urdahl, O. Lier, L.H. Gjertsen, T. Jakobsen, and J.A. Støvneng, Proc. 7nd Int. Offshore and Polar Eng. Conf., Honolulu, USA, May 25-3, 1997, p. 11. (16) Larsen, R., Lund, A., Andersson, V., and Hjarbo, K.W.: "Conversion of water to hydrate particles", SPE 7155, 21 SPE Ann. Techn. Conf. and Exhib., New Orleans, Louisiana, 3 Sep-3 Oct.