Object Oriented Simulation of Multiphase Flow P.Klebert and O.J.Nydal Norwegian University of Science and Technology (NTNU) 7491 Trondheim Norway Abstract Pipelines in offshore oil and gas production systems normally carry a multiphase mixture of oil water and gas. To remove liquid accumulation in the lower portions of hilly terrain, pigging operations are performed. This operation keep the pipeline free of liquid, reducing the overall pressure drop and by the same increase the pipeline flow efficiency. It can be used for wax control or oil production as an artificial lift system. A pigging model has been developed for predicting the dynamics of a pigging operation. For that purpose, our slug tracking model based on an object-oriented approach is used.
1. Introduction Pigging operation in lab activities consist in the launch of water inflatable rubber, neoprene or polyethylene spheres. Those are pushed by the incoming two-phase mixture. Several flowing zones are represented (fig.1)during this operation [1], [2]: - An undisturbed flow zone is located far from the pig. The effects of this pig are not yet felt. - In front of the moving pig a liquid slug zone is created and grows more and more. - The last zone is a redeveloping two-phase flow zone: a very low liquid holdup zone forms since the pig removes most of the liquid phase. By this way, pipelines subject to pigging are, always, operated under transient conditions. The aim of this project is to apply our novel and alternative method for slug flow modeling (slug tracking model) in pipelines and improve current understanding of the pigging hydraulics within the pipeline-riser system to ensure an efficient and safe operation. Undisturbed upstream flow Transient Two phase Flow Pig Section Liquid Slug section Undisturbed downstream flow Transient Two phase Flow Fig 1. Schematic view of the physical model for the pigging Phenomena
2. Slug tracking Model 2.1 Background Slug flow is pattern commonly encountered in offshore multiphase flow lines. It is characterized by an alternate flow of liquid slugs and large gas bubbles, resulting in an unsteady hydrodynamic behavior. The complexity of a slug flow is due to its dynamic structure showing large time fluctuations in characteristic parameters at given positions. Slug Flow may result from hydrodynamic slugging or from terrain slugging. Hydrodynamic slugging is encountered in straight pipes when separated flow becomes unstable and the flow regime changes to slug flow. Terrain slugging depends of the topography of a pipeline introduces other length and times scales to the flow in addition to the characteristic scales for hydrodynamic slugging. Liquid may accumulate at low points along the pipeline, build up to large slugs propagate through the line. Both phenomenon are well known and have received much attention from engineers. A few transient multiphase flow simulators for dynamic analysis of petroleum production systems are commercially available (e.g. OLGA, Tacite, PLAC). Common to the available transient models is that they are general purpose models based on numerical integration of general conservation equations for multiphase flows (mass, momentum and energy). In this class of models normal slug flow is regarded as an averaged flow, where information on individual slugs are lost and only average pressure drop and liquid fractions (holdup) are computed in time along the pipeline. Long liquid slugs are captured in the numerical solution, provided they cover many grid cells. Length and time scales in the numerical discretization will be larger than the scales of normal slugging. Because of the complexity of those codes (as they have to cover a wide range of flow simulations) and computationally demanding and they also require a quite large amount of detailed input information (discretization, initial and boundary conditions, fluid properties and simulation control parameters), we propose to explore a novel and alternative methods for slug flow modeling in pipelines. 2.2 description of the slug tracking model The starting point for the modeling will be slug flow, rather than general two phase flow, and the special structure in the phenomenon (slugs and large bubbles) will be fully exploited. Slugs and bubbles are modeled as separate objects which are coupled by
exchange of mass and momentum (through their borders). The slug and bubble structure of the data suggests an object oriented implementation tracking general numerical objects in a pipeline. The objects will be liquid slugs, bubbles as well as solid objects (pigs for inspection or removal of liquid in pipelines) and have characteristics data (holdup, gas and liquid velocities, densities, pressure and front velocities). Those objects are dynamically created and deleted according to the flow pattern at the location of the object. The view of the flow is now a sequence of slugs and bubbles where slugs and bubbles are the computational objects, which are linked with mass and momentum balances. This methodology is very different from the methods applied in current simulators solving discretized general balance equations on a fixed numerical grid. In our model, the pressure of the compressible gas in bubbles is solved from the gas mass balances and the liquid velocity in slug, from the momentum balance. 2.3 Pigging Model Implementation The Object Oriented structure of our code is very helpful for the implementation of our pigging model. The basic computations objects are liquid slugs and gas bubbles. The pig is defined as a particular kind of slug, with different physical characteristics : volumetric mass, friction factor, a holdup fix to one and no gas entertainment. Two different ways for the build of the slug, in front of the pig, are implemented: for that purpose two others borders are created : a bubble/pig border and a pig bubble border. The first way for the implementation of the building of a slug is to prevent the mass transfer : by this way the holdup in the bubble in front of the pig is going to grow and make this bubble transform into a slug. The disadvantage of this buildup is that if the size of the bubble in the neighbor of the pig is too long, it can take times to reach the critical value of the holdup. For this reason another way has been implemented : as for the severe slug model in bends, the concept of an accumulation border is used [3], [4], [5] : depending of the direction of the flow and inclination of the pipeline, the border may be filled (liquid will be taken out of the pipeline) or become empty (liquid substituted back into the pipeline ) ; but if sufficient liquid has been accumulated than all the liquid is substituted back in front of the pig as a slug.
Because of the lack of information available in the literature concerning experimental data we have decided to perform our own experiments in our lab and be able to compare accurately experimental and numerical data. Pictures below present the build up of a slug in front of the pig. We used our S-Riser configuration to perform those numerical and experimental works. Pig Bubble Pig Created Slug a) Slug build up in front of the pig a) Created Slug is pushed by the pig Fig 2 Graphical view of the pig motion [1] - Pigging Dynamics in two-phase flow pipelines : Experiments and modeling Kazuioshi Minami and Ovadia Shoham in 68 th Annual Technical Conference and exhibition of the SPE, Houston Texas, 3-6 October 1993. [2] Transient Operations and Modelling of Multiphase Pigging in Pipes PCR Lima and H.Yeug. BHR-Group 1998 Multiphase Technology [3]- Object Oriented Dynamic Simulation of Slug Flow O.J.Nydal and S.Banerjee Proc. 2nd Int. Conf. on Multiphase Flow 95, April 3-7 1995, Kyoto, Japan 4]- Dynamic slug tracking simulations for gas-liquid flow in pipelines O.J.Nydal and S.Banerjee. Chem.Eng.Communication 1996 Vol 141-142 [ 5] - "Experiments and Modelling of Gas-Liquid Flow in a S-shape Riser". O.J. Nydal, M. Audibert and M. Johansen. In 10th Int. Conf on Multiphase Technology, Cannes (FRANCE), June 2001.