Вы находитесь на странице: 1из 5

Flow Analysis

Multi-phase Flow Analysis in Oil and Gas Engineering Systems and its Modelling
a report by

O r l a n d o F A y a l a , 1 L u i s F A y a l a 2 and O r l a n d o M A y a l a 1
1. Universidad de Oriente; 2. Pennsylvania State University

Two-phase flow is very common in industrial processes and its applications were already in use in ages as remote as the era of Archimedes. At the present time, many industrial processes rely on multi-phase phenomena for the transport of energy and mass or for material processing. During the last century, the nuclear, chemical and petroleum industries propelled intense research activity on the area. Their efforts have been aimed at the demystification of the mechanisms taking place during this complex flow situation. In the petroleum industry, two-phase flow can be found in a variety of situations. The three more common working fluids (oil, natural gas and water) can have four different two-phase flow permutations: gasliquid, liquidliquid, solidliquid and solidgas flows. A solid phase can be incorporated to the flow either from the reservoir itself (due to either drilling activities or sand formation during production) or from the formation of complex solid structures due to the prevailing production conditions (e.g. hydrates in natural gas flow or waxes and asphaltenes in oil flow). Oil and natural gas transportation typically deals with a gasliquid system of flow. Due to the deformable nature of fluids, the simultaneous flow of gas and liquid in a pipe represents a very complex process. As a

consequence of their deformable nature, gas and liquid may adopt a wide variety of spatial configurations, usually referred to as flow patterns. Multi-phase flow phenomena can be found in a wide range of length scales of interest. Therefore, the most suitable approach to study multi-phase flows will largely depend on the length scale of interest. Typically, in the petroleum industry, attention is given to large-scale phenomena in multi-phase flows, as no detailed flow behaviour is needed for routine design and operation. For instance, in pipeline networks we are interested only in the pressure drop and liquid hold-up. Other than the effect of the local flow pattern variables, detailed flow phenomena are not important. However, small-scale studies of multi-phase flows are very important because large-scale phenomena are controlled by small-scale physics. For instance, the transition from one flow pattern to another is driven by local small-scale phenomena. One of the most important problems to be addressed by the scientific community is the development of an improved understanding of transitions from one flow regime to another. This can be achieved only through small-scale studies of multi-phase flows. In addition, for the improved understanding of the operation of process equipment such as separators in the petroleum industry, it is necessary to understand the small-scale phenomena associated with separation.

Orlando F Ayala is a Consultant for the Venezuelan Petroleum Company. His research activities focus on the areas of natural gas, fluid mechanics, turbomachinery and heat transfer. He is a member of the National Association of Engineers of Venezuela. Professor Ayala holds an MSc degree in Mechanical Engineering from the University of Kansas. He was Professor of Mechanical Engineering at the Universidad de Oriente, Venezuela for 40 years before his retirement and chaired the Natural Gas Engineering graduate programme of Universidad de Oriente for 15 years.

The Growth of Multi-phase Flow Modelling The development of multi-phase flow large-scale analysis in the petroleum industry has been divided into three partially overlapping periods the empirical period, the awakening years and the modelling period1 which together encompass the second half of the past century. During the empirical period, all efforts were focused on correlating data from laboratory and field facilities in an attempt to encompass the widest range of operational conditions possible. The earliest attempt to empirically predict two-phase flow pressure drops for horizontal pipes is the well-known work of Lockhart and Martinelli. This correlation was followed by an innumerable number of new ones, which claimed to be progressively more applicable for a wider range of operational conditions. Being the first quantitative approach to two-phase flow modelling, Lockhart and Martinellis correlation became a classic against which subsequent correlations were compared. The fact is that most correlations are always best applicable for the conditions from which they were derived. It is worth mentioning the correlation developed by Beggs and Brill for predicting flow behaviour in inclined pipes. Along with a number of modifications applied to it, Beggs and Brills correlation became one of the most extensively used correlations. The correlation considers horizontal, vertical and inclined pipes, and the basic correlating parameter was the Froude number a dimensional number that is considered a measure of the influence of

Luis F Ayala is Assistant Professor of Petroleum and Natural Gas Engineering at the Pennsylvania State University, US. He has also been an Instructor in the Chemical Engineering and Petroleum Engineering Departments at Universidad de Oriente. Professor Ayalas research activities focus on the areas of natural gas engineering, hydrocarbon phase behaviour, multi-phase flow, numerical modelling, and artificial intelligence. He is a Member of the Society of Petroleum Engineers), the American Chemical Society, the National Association of Engineers of Venezuela and the Canadian Petroleum Society. Professor Luis Ayala holds PhD and MSc degrees in petroleum and natural gas engineering from Pennsylvania State University and two engineering degrees with honours, one in chemical engineering (summa cum laude) and one in petroleum engineering (summa cum laude), from Universidad de Oriente.

Orlando M Ayala is Assistant Professor of Mechanical Engineering at the Universidad de Oriente. Professor Ayalas research activities focus on the areas of multi-phase flows, turbulent flows, computational fluid analysis, direct numerical simulation and heat transfer. He is a Member of the American Society of Mechanical Engineers and the National Association of Engineers of Venezuela. Professor Orlando M Ayala holds PhD and MSc degrees in Mechanical Engineering from the University of Delaware and an engineering degree in Mechanical Engineering with honours (cum laude ) from Universidad de Oriente. He has also been working on several engineering projects for the Venezuelan Petroleum Company.

TOUCH BRIEFINGS 2007

57

Multi-phase Flow Analysis in Oil and Gas Engineering Systems and its Modelling
gravity on fluid motion. In general, the reliance on the empirical approach was always limited by the uncertainty of their application to systems operating under different conditions than those from which the correlations were originally proposed. Nonetheless, calculating and designing flow lines in multi-phase production facilities on the basis of empirical correlations was the norm until well into the 1980s. The advent of the personal computer during the 1980s dramatically enhanced the capabilities of handling progressively more complex design situations, which is why this period has been called the awakening years. Much of the petroleum research on multi-phase
1

introduce a fully phenomenological description of how transitions occur among the different flow patterns was the work developed by Taitel and Dukler, which focused on horizontal and nearhorizontal pipes. The work of Taitel and Dukler is considered one of the classic papers in multi-phase predictions that began to incorporate more physical insight into analysis in the petroleum industry. This work led the way for subsequent research in the area, and most of their transition criteria are still in use in more recent two-phase flow models. Few years after that initial work, Taitel and co-workers extended the model for the vertical and nearvertical case and Barnea extended the phenomenological approach to the whole range of pipe inclinations in the 1980s. These three works are commonly referenced among researchers in the area, with a number of attempts at improvement. Additional steady-state comprehensive mechanistic models for twophase flow in vertical wells, horizontal pipes and deviated wells were presented by Ansari, Xiao, Kaya and co-workers in the 1990s. All these mechanistic models were developed at the Tulsa University Fluid Flow Projects and are usually referred to as TUFFP models.

flow during these years and the subsequent modelling period was enriched by the progress already made by the nuclear industry. Although the nuclear industry dealt with much simpler fluids (water and steam), it led the way towards more involved two-phase flow analysis in the petroleum industry. More fundamental multi-phase flow analysis approaches, such as two-fluid modelling, were already in use in the nuclear industry in the 1970s. These seed efforts are the genesis of the well-known fast transient two-phase-flow codes

State-of-the-art large-scale multi-phase flow modelling in the oil and gas industry is largely based on mechanistic models.
RELAP4, RELAP5, RETRAN, MEKIN, COBRA, CATHARE and TRAC in use today in the nuclear industry. Nowadays, the petroleum industry might be ready to explore new research avenues in multi-phase flow analysis, with the incorporation of the increasingly sophisticated modelling tools that have become available in the last few years. Modern multi-phase flow analysis models the flow of oil and gas The modelling period, which extends up to the present day, refers to the growing tendency of introducing more physically based (mechanistic or phenomenological) approaches into multi-phase flow calculations. The main goal remains an attempt to reduce the impact of empirical correlations on multi-phase predictions. Stateof-the-art multi-phase modelling efforts can be studied in two different but interrelated fields of interest: small-scale and largescale, depending on the length scale of interest to the modeller. During recent years, the oil and gas industry has paid particular attention to large-scale modelling of multi-phase flows. However, small-scale modelling promises to bring important physical insights into the quest for more accurate and reliable modelling of multiphase flow in the oil and gas industry in the foreseeable future. Large-scale Interest State-of-the-art large-scale multi-phase flow modelling in the oil and gas industry is largely based on mechanistic models. One of the distinguished features of a mechanistic model is the need for a reliable tool for the prediction of flow pattern transitions for a given set of operational conditions. Perhaps the earliest attempt to In the last decade, a great deal of attention has also been devoted to mechanistic or phenomenological models i.e. models trying to capture specific features of individual flow patterns in which simplified conservation equations are invoked while the main focus is the prediction of pressure drop and hold-up. However, in previous decades, the challenge of modelling two-phase flows by invoking such fundamental laws had been circumvented by reliance on empirical and semi-empirical correlations, especially in the oil industry. through pipelines by invoking the basic principles of continuum mechanics and thermodynamics. Depending on how these equations are applied and how the interactions between phases are described, the most widely used two-phase models are the homogeneous model (flow treated as a single phase with averaged fluid properties), drift-flux model (flow described in terms of an averaged local velocity difference between the phases), separated model (phases considered to be flowing in separated zones of the channel) and two-fluid model (a multi-fluid model that considers two flowing phases and their interactions). Nowadays, there are also a number of commercially available twophase flow packages, which include various features intended to accomplish specific tasks. Examples include OLGA, TACITE, PEPITE and PIPESIM, among others.

58

HYDROCARBON WORLD 2007

Multi-phase Flow Analysis in Oil and Gas Engineering Systems and its Modelling
Perhaps one of the most fundamental and rigorous approaches to the study of large-scale multi-phase flow currently in use in the petroleum industry is the two-fluid model. In the two-fluid model, separate conservation equations (mass, momentum and energy) are written for each of the two phases for a total of six equations. These equations are coupled with terms describing the interaction between phases. In this two-phase flow method of analysis, as well as in all the others, empiricism cannot be completely avoided, since additional closure relationships are needed. Empiricism Small-scale Interest and Computational Physics The study of small-scale multi-phase flow has proved to be extremely difficult for researchers due to the elusive nature of the phenomena are a continuum and invokes the basic laws of continuum mechanics in one dimension coupled with a thermodynamic phase behaviour model. In their work, the required semi-empirical relationships needed to give mathematical closure to the model are discussed in detail.

Perhaps one of the most fundamental and rigorous approaches to the study of large-scale multi-phase flow currently in use in the petroleum industry is the two-fluid model.
comes into the picture during attempts to model the variety of constitutive relationships that show up in conservation equations. For instance, Ayala et al.2 have presented a unified two-fluid model for the analysis of natural gas flow in pipeline in multi-phase flow regimes. Their formulation assumes that both gas and its condensate and the inherent limitations of experimental set-ups. A great deal of progress has been made on the development of useful small-scale experimental studies, but numerical experiments or models still remain the most effective way of studying such detailed flow behaviour. The challenge of modelling small-scale multi-phase flow

Engineering for mobility

www.mubea.com

Mubea is specialised in the production of high quality disc springs (Belleville Washers) for all different kinds of applications in the Oil and Gas Industry. Special materials and geometries up to 800 mm outside diameter can be offered to suit your requirements. Mubeas intensive knowledge about shot-peening of highly stressed springs guarantees a long fatigue life as well as reliability in performance. Our complete in-house process and our well equipped test laboratories help to almost fulfill all of our customers demands. Mubeas highly qualified engineering team can offer you the best design and solution wherever you need high forces in a small environment.

engineering for mobility

Multi-phase Flow Analysis in Oil and Gas Engineering Systems and its Modelling
resides in the finite nature of the computer power typically available to the modeller and the difficulty of tracking separated phases (and interfaces between them) with sharply different properties. The interplay of these two factors has historically limited the complexity of the systems that can be studied using small-scale simulation. However, during the last decade, major progress has been achieved by implementing a variety of numerical techniques, which typically depend upon the flow pattern type that prevails under the conditions of the study. The study of small-scale phenomena started when a group of scientists at the Los Alamos National Laboratory began to develop the basis of Computational Fluid Dynamics (CFD) in the early and mid-1960s. In multi-phase flow modelling within small-scale interest, the Navier-Stokes equations with the appropriate boundary conditions are solved through a suitable numerical method e.g. finite volumes, finite differences, finite elements or spectral methods. The main problem arises when considering that some boundary conditions are time-dependent, since they are located at phase boundaries, which are free to move, deform, break up or coalesce. Different methods have been proposed; here we mention a few of them. Small-scale modelling typically takes advantage of certain multiphase flow conditions that can greatly simplify the modelling process. For example, it is possible to simplify the Navier-Stokes equations by ignoring inertia completely (Stokes flow) or by ignoring viscous effect (inviscid flows) in the limit of low and high Reynolds numbers, respectively. These two limiting cases are typically studied with boundary integral methods. The study of dispersed flows, for example, can be made especially amenable to small-scale simulation since the study of one of the phases (i.e. the dispersed phase) can be greatly simplified. Two main methods are used to order to accurately capture the new fluid positions at each timestep. At every time-step, the grid is refitted and adjusted to match the location of the new, displaced boundaries. In the 1980s, Ryskin and Leal used this method to study the steady rise of buoyant, deformable, axisymmetric bubbles, while Oran and Boris studied the break-up of a two-dimensional drop. A similar approach, called front tracking, is also used, where a separate front marks the interface but a fixed grid is used for the fluid within each phase; however, the fixed grid is modified near the front so a single grid line follows the interface.

Large-scale modelling includes the use of transient and steady-state two-fluid models, as well as a variety of steady-state mechanistic models.
The most common small-scale modelling approach discretises the flow domain using a regular and stationary grid i.e. the wellknown Eulerian frame of reference for fluid motion. The first smallscale Eulerian method proposed was the marker-and-cell (MAC) method, where marker particles distributed uniformly in each fluid were used to identify each fluid. Using this method, in the late 1960s Harlow and Shanon studied the splash when a drop hits a liquid surface. The MAC method has become obsolete since then and has largely been replaced by others that use marker functions instead e.g. the so-called volume-of-fluid (VOF) method. In the VOF method, the transition between two fluids takes place within the context of one grid cell. The main problem associated with this is the difficulty of maintaining a sharply defined boundary between two flowing fluids. In order to address this difficulty, level-set (LT) methods use continuous rather than discontinuous marker functions in order to identify the fluids. The use of continuous marker functions creates smooth transition zones between the two fluids of interest and avoids the difficulty of maintaining a sharply defined boundary. Some other small-scale modelling approaches use the Lagrangian frame of reference for fluid motion. In Lagrangian methods, the numerical grid follows the fluid and deforms with it. In this approach, the motion of the fluid interface needs to be modelled in simulate dispersed flows: the Eulerian-Eulerian or the EulerianLagrangian approach. In the Eulerian-Eulerian approach, separated equations are solved for the dispersed and the continuous phase. No attempt is made to resolve the detailed motion of the particles, and thus closure relations are necessary for the unresolved motion and the forces between the particle and the continuous phase. The closure relations are determined through experimental correlations (similar to the computation of turbulent flows using Reynolds average Navier-Stokes equations). In the Eulerian-Lagrangian approach, the dispersed phase is represented by points moving inwards and otherwise constantdensity flow i.e. the so-called point particle approximation. The particles are followed using a Lagrangian approach and the forces (such as drag forces) on the particle are specified by analytical and experimental models. In some cases, the particles are assumed to have no effect on the fluid flow, but in other cases the forces from the particles are added to the right-hand side of the Navier-Stokes equation of the continuous phase. However, none of these approaches can model the detailed flow around the particle that affects the interactions of nearby particles. These interactions are important to understand particle coalescence, which is the first step towards a possible flow pattern change. Ayala et al. 3 have recently developed a first attempt to incorporate such interactions through a new hybrid direct numerical

60

HYDROCARBON WORLD 2007

Multi-phase Flow Analysis in Oil and Gas Engineering Systems and its Modelling
simulation (HDNS). This approach consists of direct numerical simulation of the undisturbed continuous phase flow and an analytical representation of local disturbance flows induced by the particles. In addition, a relatively new method in small-scale modelling is the Lattice Boltzmann method. Lattice Boltzmann methods are based on kinetic theory and thus no Navier-Stokes equations are solved. Instead, the method considers a typical volume element of fluid to be composed of a collection of particles that are represented by a particle velocity distribution function for each fluid component at each grid point. In this novel approach, the rules governing the motion and collisions of these particles are designed in such a way that the time-average motion of the particles is consistent with the Navier-Stokes equation. Concluding Remarks The most popular modelling approach nowadays in the oil and gas industry above and beyond the use of long-established, fully empirical equations is large-scale mechanistic modelling. Large-scale modelling includes the use of transient and steady-state two-fluid models, as well as a variety of steady-state mechanistic models. However, large-scale multi-phase flow modelling relies much more heavily on empirical or semi-empirical correlations to model the phenomena than small-scale multi-phase analysis does. Small-scale multi-phase flow analysis relies on the direct solution of the most fundamental fluid dynamic equations, thus greatly reducing the need for empiricism. The limitation of small-scale multi-phase analysis resides in its scope, which is not currently amenable to the study of large industrial systems. The oil and gas industry relies on large-scale analysis and does not currently use small-scale methods for the simulation and modelling of oil and gas systems, but the physical insights that can be obtained by small-scale simulation are invaluable. It is widely expected that the demystification of small-scale intricacies of multi-phase flow phenomena can greatly help large-scale modelling in the foreseeable future. The simultaneous implementation of large-scale and small-scale simulation represents a powerful combination that can significantly improve our understanding of multi-phase flow phenomena. Small-scale simulation, for example, could play an important role in significantly improving the nature and reliability of the semi-empirical relationships needed by large-scale simulation models. Small-scale simulation can also define more reliable flow pattern transition models, which are the backbone of the large-scale multi-phase flow simulators in use today.

1. 2.

Brill J, Arirachakaran S, State of the Art in Multiphase Flow, J Pet Tech , 1992;44(5):53841. Ayala LF, et al., Low-liquid Loading Multiphase Flow in

3.

natural Gas Pipelines, J Energy Resources and Tech , 2003;125:28493. Ayala OM, et al., A Hybrid Approach for Simulating Turbulent

Collisions of Hydrodynamically-interacting Particles, J Comput Phys , 2007;doi:10.1016/j.jcp.2006.11.016.

NEPTUNE OCEANOGRAPHICS LTD


PROVIDING SERVICES TO THE OFFSHORE INDUSTRY

Leaks in your subsea pipeline installations?


Neptune Oceanographics can find them
Techniques available include: Fluorescence Acoustics Direct hydrocarbon detection (oil and gas) Differential temperature NOL maintain a continuous programme of development of new techniques for subsea leak detection
Neptune Oceanographics Ltd, Sapharey House, Charlbury, Oxon, OX7 3SX, UK Tel +44 8453 707177. Fax +44 8704 581979. Email info@neptune.gb.com www.neptune.gb.com Neptune Oceanographics ltd is an ISO 9001 : 2000 certified company
First Point Registration 20436

Вам также может понравиться