With longer exploration of o®shore gas wells, the amount of liquid, especially water, in the product stream increases. This is mainly due to the fact that water is usually used to keep the well under pressure. The increase in liquid contaminants requires improvement of current separation methods. In commonly used separators gravity is used to separate the dispersed phase. To allow for su±cient settling time, gas velocities in these separators must be low. As a result these devices are voluminous, heavy and expensive. With increasing liquid amount, the capacity of these devices is no longer su±cient. As a result more and/or heavier separation devices are needed and sometimes it even requires heavier and more expensive supporting structures. In some cases, although considerable gas reserves are still present, the exploitation of a well has to be stopped, as current gas treatment techniques are not economically viable. In order to make the exploitation of older wells pro¯table, the o®shore industry is searching for e±cient, compact phase separation devices. The goal of this study is to design and test a compact and e±cient device for separating condensed liquid phases from natural gas. For this purpose the Rotational Particle Separator (RPS) principle is used. Due to the high centrifugal force created in the separator and the small radial distance the particles have to travel to reach a collecting wall the separator can be designed compact, while at the same time a high e±ciency can be reached. Before the separator was designed, two issues were resolved. At ¯rst, the behavior of the liquid ¯lm inside the channels of the ¯lter element was investigated by using an analytical model based on the Nusselt analysis. The model is valid for small liquid loads and with the model the amount of liquid leaving the ¯lter at the in- and outlet can be determined. Secondly, the order of magnitude of the smallest droplets present in the gas phase was determined by considering nucleation, condensation and coagulation of water vapor in methane. For the conditions considered the droplets reach a minimum size of 1 micron, provided that the coagulation time is su±ciently large. For the design of all main components of the separator general design relations were derived. With these relations the pressure drop, angular speed and separation e±ciency of the separator can be predicted as a function of the °ow rate. Based on the design relations, the design criteria and the o®shore process conditions, a full- scale prototype was built, which is capable to handle the volume °ow of one wellhead under high pressure (80 bar) and which separates particles down to 2 micron. The design involves a so-called naturally driven RPS, which means that the separator is driven by a swirl generated in the °ow upstream of the separator. As a consequence no external motor and therefore no shaft, which needs to be sealed, is required. The performance of the prototype was tested in two test loops. In a low pressure loop, which was built at the Technical University Eindhoven, both the hydrodynamic and separation performance of the prototype were measured. In a high pressure loop, which was available at CDS Engineering, the hydrodynamic performance of the prototype was measured. In this test loop the o®shore operating conditions can be simulated. Comparing the hydrodynamic performance of the prototype with the the- oretical model shows good agreement. The separation performance of the prototype was determined by injecting dust particles in a size range between 1-10 micron in the low pressure test loop. Subsequently the particle size distribution both up- and down- stream of the separator was measured with an impactor. From those measurements the e±ciency of the prototype as a function of the separated particle size was calcu- lated. The measurement results were compared with both an analytical model and with previous measurement, which were performed with an externally driven RPS. The current measurement results show good agreement with both the previous results and the analytical predictions.
|Qualification||Doctor of Philosophy|
|Award date||12 Dec 2005|
|Place of Publication||Eindhoven|
|Publication status||Published - 2005|