This paper presents recent attainments of one of the main components of the POSEIDON project launched in 1984 by TOTAL, IFP and STATOIL : the multiphase pump. After successful bench tests, the pump had been installed on a real field in Tunisia. Four thousand running hours without mechanical problems have provided a better understanding of how a multiphase pump behaves when placed in a production network. INTRODUCTION With the necessary development of small oil accumulation, either onshore or offshore and the increasing depth of offshore exploration especially in Brazil and in the Gulf of Mexico, it is widely accepted that there will be a need to pump the effluent from the wellhead to the nearest processing facility without treatment on site. This results from both a technical limitation of the traditional platforms when faced with deeper waters and from economic considerations of lower cost offered by this new technique. For shallower waters, even when a platform could be envisaged, the size of the accumulations often rules out the development because of economic constraints. As the maximum possible distance that can be covered by a multiphase effluent with the natural pressure appears to be limited to about 15/20 km of pipe line (15 km for DON, BP), a satellite developments could use a new type of technology to add energy to the flow at the beginning of the line. The available economic studies [I] show that the majority of reserves in the North Sea are at a distance of 50 km or less from existing facilities, and that 80% of the discoveries with a total of 10 million of barrels of reserves are located withing 35 km of them, thus giving access to additional oil at acceptable cost, should an industrial boosting technique be made available to the operators. The POSEIDON project was created in 1984 by TOTAL, IFP and STATOIL to review all the problems to be solved to build a multiphase subsea station. One of the major element of such a station was the multiphase pump. At that time no pump capable of handling a multiphase fluid was available and hence a major effort was agreed upon by the associates to develop an emerging technology in order to reach a very ambitious goal : a pump able to create a differential pressure on any liquid, with a gas fraction from O to 95% in steady state flow and the capability of handling void fraction of 100% in case of slugs of pure gas. The available technology of the end at the 70?s was an ESP (electric submersible pump) able to pump a fluid with GLR of 3(gas to liquid ratio) under down hole conditions. Today the POSEIDON multiphase pump is industrially proven, has

Full scale gas kick experiments have been performed. The objectives of the experiments were to gain more knowledge during the circulation out of gas kicks in a horizontal well through the study of the behaviour of the two-phase flow at high pressures. The horizontal wellbore was simulated on the surface by a 200 m long flow loop (casing) with an inside diameter of 9 1/4". The inside of the well had a drillstring with a 50 m long drill collar section. The drillstring assembly could be rotated. Also, the last 50 m of the well near the bit was inclined upward with 4 from horizontal for part of the experiments. The wellbore was designed for pressures up to 170 bars. The gas used during tests was air. Two different liquids; water and CMC polymer solution, were tested. During the experiments the main parameters varied were mud circulation rates, gas injection rates, system pressure and drillstring rotation. Three test categories were performed. During DISPERSION tests, single bubbles of gas were injected and circulated out in order to study the gas dispersion along the horizontal well. During FLOW TESTS, a continuos injection of gas was performed to study gas velocity, flow regime and pressure loss. During CLEAN-OUT TESTS, methods for circulating out stationary gas positioned in traps in the well were studied. New correlations for gas transport velocities and frictional pressure losses have been developed based on these data as well as data from two low pressure flow loop tests. Results from the full scale tests as well as the new correlations are presented in this paper. Introduction Gas kicks may develop into major well control incidents. A gas kick situation in a horizontal well may for the future occur more frequently than today. The industry is moving towards fewer appraisal wells and early production. This may lead to pressure surprises. Also horizontal well drilling in reservoirs with water or gas injection or in depleted reservoirs may lead to surprises related to pressure. Studies related to kick development and control have been performed. These include full scale experiments, development of kick simulators and the related gas rise velocity models. Studies on kick control in horizontal wells have been few. These include computer simulations, well control evaluations and experimental studies. The results from the experimental studies have been implemented into a horizontal kick simulator. These experiments were performed at low pressures. and in a 12 m long model loop. It is important to study gas kick development and control in horizontal wells, for two reasons mainly. First, development of general procedures and guidelines for well control of horizontal wells are needed, procedures like minimum mud velocities for gas clean out in a given mud/well; or procedures to clean out gas when minimum clean out velocity is unobtainable etc. Secondly, further development of models related to gas kick development and control in horizontal wells are needed. Such models will be important parts of and inputs to the development of advanced kick simulators for horizontal wells. Horizontal wells vary tremendously related to angle, build-up, length, detailed profile and existence of gas traps in the well. In order to develop well specific well control plans and procedures. simulations with an advanced kick simulator for horizontal wells is necessary. The most important models are those for gas transport velocities, frictional pressure losses and removal rate by gas dissolution. Gas transport velocities are significantly higher for annuli than for pipe flow. P. 765

Since May 1991 a POSEIDON pump of the rotodynamic type has been running on the SIDI EL ITAYEM field in TUNISIA, logging more than 3500 operating hours (end of November 91) without a major problem. This pump is one of the practical results of the POSEIDON project, launched in 1984 by TOTAL, IFP and STATOIL, to develop the concept of multiphase pumping: the final target being a subsea multiphase boosting station. Prior to the Tunisian test, this pump had been tested satisfactorily on the two phase loop of IFP at SOLAIZE (FRANCE), and it was this which decided the partners to send the prototype, wihout any modification, to the real active field test rig specially prepared to accommodate multiphase equipment.