Multiphase flow NTNU. Report year 2008

Transcription

1 NTNU Norwegian University of Science and Technology Faculty of Engineering Science and Technology Department of Energy and Process Engineering Multiphase flow PhD program: Multiphase Transport Dept. of Energy and Process Technology NTNU Report year 2008 January 2008 Ole Jørgen Nydal 1

2 Contents Status...3 Researchers...3 Project students...3 Meetings...4 Research activities...5 Laboratory...5 Flow modeling...5 Research Council projects...5 List of activities...5 Multiphase Flow Laboratory...18 The following is a summary on 2007 activities in the PhD program on Multiphase Transport, and includes reporting on the student projects at the Multiphase flow laboratory at NTNU, Department of Energy and Process Technology. The PhD activities are mainly sponsored by a group of companies. 2

3 STATUS Participants in the PhD program in 2007 have been BP Chevron ENI Hydro Total Scandpower SPT IFE RESEARCHERS The persons in Table 1 have been active in the multiphase flow group at NTNU in Table 1 Researchers 2007 Name Description Affiliation Finance Complete Ole Jørgen Nydal Prof. Supervisor NTNU Roar Larsen Proff. II. Supervisor NTNU/Sintef Total Tor Ytrehus Proff Co-supervisor NTNU Angela de Leebeeck PhD. Waves in a slug tracking scheme NTNU Total 2009 Xiaouju Du PhD. Numerical methods gas liquid flows NTNU ENI 2009 M. M. Shabani PhD Hydrate particles in multiphase flow NTNU BP 2009 George Johnson Post.Doc. Flow models NTNU Total/Hydro 2007 Kristian Holmås (Scp) PhD Multiphase CFD UiO IFE 2007 Håvard Holmås (IFE) PhD Numerics of stratified wavy flow UiO SPT 2008 Gandi Setyadi PhD Phase displacement effects in oilwater systems NTNU Chevron 2010 George Johnson started to work for Norsk Hydro in Porsgrunn on February 1 st, Gandi Setyadi arrived from Bandung Institue of Technology on September 3 rd starting on his PhD period at NTNU. Prof. Tor Ytrehus and prof. Stein Tore Johansen as co-supervisors for Gandi Setyadi. PROJECT STUDENTS Final year students participate in the research activities with a project in the fall and a MSc thesis in the spring. If the PhD s see a need for extended support, we also invite students form European universities. Table 2 shows the student activities in Table 2 Student projects Student projects Spring Marte Bø Andersen (Total)- Multiphase simulation software validation. Christine A Maren (SPT) Simulation of removal of water by oil stream Sebastien Chevanne (ENSAM) Experiments on pressure across roll wave front Andreas Hol Garder Experiments on pressure jump across wave front Marius Sæther (HIST) Solar water pumping: feasibility study 3

4 Trond Lauvaas (HIST) Fall Solar water pumping: feasibility study O.J. Nydal leaves for sabbatical period at UWA, Perth, Austalia MEETINGS Status meetings Progress reports are normally presented by the PhD s at two yearly meetings. The winter meeting was hosted by Chevron in Bakersfield, and included an introduction to heavy oil production at the site in Bakersfield. Thanks to Lee Ryhyne and his colleagues in Chevron! The fall meeting was cancelled, after O. J. Nydal left for a sabbatical period in Australia ( ). PhD defences Fabien Renault started to work with Banque de France, August 1 st The PhD defense for Fabien took place at NTNU on June 28 th.. Opponents were prof. R. Issa from Imperial College and Zhi L Yang from Sintef Materials and Chemistry. The defense was administrated by professor Muller, NTNU. Jørn Kjølaas went back to his position at Sintef Petroleum Research in The PhD defense for Jørn took place at NTNU on June 29 th. Opponents were prof. Y. Taitel from Tel Aviv University and R. Henkes from Shell Amsterdam. The defense was administrated by professor M. Golan, NTNU. We congratulate Fabien and Jørn with their PhD defenses! Visits We have had the following visits to the laboratory during 2007: Visits Date Petrobras June 11 th 2007 Chevron November 30 th, 2007 Gupkin University 14 February 14 th 2007 Hydro April 14 th 2007 University of KwaZulu Natal June 1 st 2007 Delft University April 26 th

5 RESEARCH ACTIVITIES Laboratory A short description of the multiphase flow laboratory is given in Appendix1. Among the laboratory staff, the following persons have participated in the activities in the multiphase flow laboratory: Name Description Erling Mikkelsen Verksmester Laboratory responsible Knut Glasø Verksmester Mechanical engineering Per Bjørnaas Engineer Labview, Netlab Helge Laukholm Engineer Electronics The laboratory activities in 2007 were mainly: Small scale experiments on selection of hydrate model particles Preparations for transient particle measurements in the laboratory Measurements of pressure across roll waves Purging of air by a water flow at constant inlet pressure: W-pipe geometry Decay of slugs in a downwards inclined section Demonstrations in NTNU courses Flow modeling Flow modeling/simulation activities are made in the frameworks of: Industrial simulators Some student projects make use of the OLGA and LEDA programs Special purpose flow modeling in Matlab Matlab provides an efficient framework for prototyping of model concepts, for investigation of special problems and for experimental data analysis. Examples in 2007 are: o Simplified models for pipeline-riser slugging o Two-fluid model for pipeline-riser slugging o Testing of some particular numerical issues: iterative, non-staggered. o Incompressible model for the water-particle transient experiments C++ Plug tracking methods The activity on modeling hydrate plug release during depressurization is made within a slug tracking model, and an object oriented implementation in C++. Research Council projects Petromaks Petromaks is funding a project on gas hydrates (model particles) in multiphase flow (Scandpower, NTNU, BP). The project includes the PhD (M.M. Shabani) sponsored by BP. The NRC project provides funding for experimental costs. Centre for Research based Innovation FACE NTNU (Multiphase laboratory and Ugelstad laboratory) forms, together with SINTEF and IFE, a Centre for Research Based Innovation on Multiphase Flow Assurance (FACE), focusing on complex mixtures. The duration for the centre is 5-8 years. LIST OF ACTIVITIES 5

6 PhD Title: Slug Initiation Methods for Slug Tracking Schemes Name: Fabien RENAULT Supervisor: Ole Jørgen Nydal Objective Investigate slug initiation models, in order to improve the precisions of the slug tracking methods. Background As general purpose flow models need criteria for flow regime transitions, so do slug tracking codes need models and procedures for slug initiation. How, when and where to initiate the liquid slugs in the stratified flow pipe are the main questions. Theoretical Investigations Stability analysis The work has covered stability analysis of stratified flows, as a method for generating flow regime transition boundaries. This method has weaknesses for high gas density flows, where the transition to slug flow often evolves from wave coalescence, and not from unstable smooth stratified flow. Two fluid model Solution of a two fluid model has been investigated, as a general method for slug initiation prediction. The evolution of waves into slugs can be simulated using a fine grid. Hybrid two fluid and slug tracking model A combined Lagrangian slug tracking model (slugs are followed) and slug capturing model (slugs are automatically initiated) has been formulated, implemented and tested. The screen-shot on the right shows slug initiation in a 2 cm diameter air/water horizontal 30 m long pipe (pipe holdup in black, pressure profile at the bottom and inlet pressure time series on top). Experiments Initiation in bends: Influence of the inclination angle and upstream volume on the length of the generated slugs. The experiments were carried out with the help of two students from ENSAM (France). This project was awarded the third best ENSAM project prize. The results compare well with predictions Initiation in an horizontal pipe: Observation of slug initiation mechanisms in a 3 cm pipe: videos, photos, and holdup traces were collected for comparisons with numerical simulations. Experiments Model x/d=20 x/d=80 x/d=120 Completion PhD defense took place in June 28 th,

7 PhD Title: Incompressible multiphase flow models Name: Trygve Wangensteen Supervisor: Ole-Jørgen Nydal Objective Explore methods for transient multiphase flow simulations. Background Many multiphase flow phenomena can be formulated with an incompressible flow model. These models can be used as stand alone models to simulate problems like wavy flow and oil-water flow, or the models can be used as sub models in gas liquid-liquid flow to describe the liquid-liquid flow. To solve these models, numerical methods are needed. These methods differ in accuracy, and robustness and some methods are more applicable for some problems than other. Activities Unit cell modelling Methods for implementing a unit cell slug flow model (or a general algebraic slip relation) on a transient two-fluid model have been investigated. The method in OLGA is based on deriving an interface friction factor from a slip relation. This method has some weaknesses, in particular for cases where the slip relation gives zero relative velocity, where correction schemes must be made. This problem has been solved by using the unit cell holdup instead of the unit cell slip as basis for deriving the interface closure relations. Numerical methods A variety of numerical methods based on the characteristics of the model s Jacobian has been investigated: Godunov, ROE/FDS and different variations of second order accuracy. These methods work very well when the models have distinct eigenvalues, but they fail for singular problems, like transition to single phase flow. A new numerical method (Harmonic scheme) based on the slip velocity is developed. This method, compared with the Jacobian based schemes, is simpler and robust for flow phenomena where the Jacobian based schemes fail. For special flow problems, like wavy flow, the Harmonic scheme is less accurate than the Jacobian based schemes. Numerical Computations Liquid fraction ( ) nd order 1st order Length (m) Wave propagation in stratified flow, FDS scheme Mixture velocity (m/s) Experiment Simulation Plot Time (s) Unstable oil-water flow with pressure dependent inlet condition Harmonic scheme U sg =0.25,U sl = U sg =0.25,U sl = U sg =1,U sl = U =2,U =1 sg sl U sg =2,U sl = U sg =3,U sl = Length (m) Slug flow, horizontal pipe, holdup profiles, Harmonic scheme. Completion Trygve completed his work in 2005, and the PhD defense will take place in

8 PhD Title: Gas hydrate plugs in multiphase flow lines Name: Jørn Kjølaas Supervisor: Ole-Jørgen Nydal Background The PhD work concerns prediction methods for the flow dynamics of a plug propagating in a two phase flow line. Single sided depressurization across a hydrate plug represents a potential hazardous operation, as the may release and propagate at high speeds through he pipeline. Objective The objective of the PhD work was then to establish a numerical model for the plug release problem, and compare predictions with available experimental data. Activities and results The original slug tracking scheme, which was based on uniform pressure in the bubble, was compared with the Statoil experiments, where a model plug (400 g) was inserted into a vertical pipe (7 m high) and subject to a high pressure difference (up to 30 bar) by the sudden opening of a valve. The Figure shows that including a steady state pressure drop term in the bubble lead to improved predictions. Co mputed and measured plug velocity in Statoil hydrate gun experiment However, when comparing the modified model with the deep star data, the compression wave had to be included in order to reach the measured high plug velocities. The scheme was therefore modified to include a full two fluid model in the bubble region, and the energy equation was implemented as well. Experiments Some small scale experiments were made, where foam plugs were subject to a differential pressure, resulting in a very rapid expulsion of the plug. This was made with and without initial liquid in a pipe bend. Comparisons between experiments and simulations show that the initial transient is well reproduced, but the further damping is too low in the computations. Small scale set-up for plug release experiments. Computations and experiments. Completion PhD defense took place on June 29 th,

9 PhD Title: Numerical simulation of two phase pipe flows Name: Angela De Leebeeck Supervisor: Ole Jørgen Nydal Background The focus of this work is on numerical issues of transient two phase flow, in a continuation of previous work at EPT. A particular problem in slug tracking models is the phenomena of large waves. At the point of transition to slug flow, where slugs are formed from stratified flow or decay from slug flow, the waves may be on the same scale as slugs and transport liquid. Large waves can also be important for slug initiation. One approach to modeling the large waves is to think of wave tracking as analogous to slug tracking. An existing slug tracking scheme developed within the multiphase flow group can be used to test the wave tracking model. Objective The objective of this PhD work is to develop a dynamic wave tracking model for two phase pipe flow where large waves are on the same scale as slugs. Activities Study of background material Literature related to experiments and modeling of roll waves in two phase flow was reviewed. The performance of an existing steady-state roll wave model from G. Johnson (2005) was compared to commercial multiphase flow modeling software OLGA, and gas-condensate experimental data. This comparison will be presented at the 6 th North American conference on Multiphase Technology in Banff, Canada in June Experiments Experimental data on roll waves in two phase flow at atmospheric pressure were conducted, results were analyzed, and it was found that large roll waves on the same scale as slugs could be tracked experimentally and they had an associated pressure jump similar to slugs. The experiments were carried out by Andreas Hoel Gaarder as part of his Masters student project in the Multiphase Flow lab at NTNU. The results will contribute to the wave tracking model developed in this PhD work. These experiments were the subject of a conference paper presented at the 16 th Australasian Fluid Mechanics conference in Australia in December 2007 Modeling The existing slug tracking code is being investigated with thought to how a wave tracking model can be implemented. One possibility for modeling the observed pressure jump in large waves is to think of gas flow over a wave as similar to an orifice. Courses Four courses, Multiphase Transport, Thermodynamics of Hydrocarbon Mixtures, Modelling Multiphase Flow, and Kinematics and Dynamics of Ocean Surface Waves, were completed between January and December

10 PhD Title: Numerical issues in transient two phase flow model Name: Xiaoju Du Supervisor: Ole Jorgen Nydal Background Pipeline simulators are commercially available, and in extensive use by petroleum engineers worldwide. Transient and dynamic flows are particular classes of flow problems where the numerical methods become important. The PhD work will target some specific numerical issues, some of which are also weak points in the commercially available programs. Objective The objectives are to test numerical methods for one-dimensional simulation of multiphase pipe flows. The work will be related to standard two fluid schemes and to front tracking schemes. Activities Some numerical issues of practical importance in transient flow are, or will be considered: Iterative versus Non-Iterative A non-iterative scheme (e.g. OLGA) will be compared with an iterative scheme regarding computational efficiency and robustness. Direct schemes require time step control to limit the evolution of numerical errors, whereas iterative schemes remove the numerical error in one time step by iteration. Test will be made for single phase flows (with varying compressibility) and for two phase flows. Staggered versus Non-staggered Staggered representation of velocities and pressure removes checker board problems, but introduces other problems in large pipe bends. Suggestions for non-staggered grids in the literature will be compared with staggered schemes for a selection of cases. Two fluid model versus Drift flux model A two fluid model (separate momentum equations for gas and liquid) will be compared with a mixture momentum equation and slip relation. Of special interest is the case when the same basic scheme is applied for both separated flows (with friction closure laws) and for mixed flows (with slip relations), and how such schemes compare with mixture schemes (both formulations active, depending on the regimes). Other issues Other issues are, adaptive grid for gravity dominated flows, transient flows with steady gas approximation, multilevel modeling and level gradient driven flows. 10

11 PhD Title: Hydrate particles in multiphase flows Name: M. M. Shabani Responsible: Roar Larsen / Ole Jorgen Nydal Background Gas hydrate management is a very critical issue in the design and operation of multiphase pipelines. Classical hydrate management strategies aims at avoiding the formation of hydrates in the pipes. The cold flow methods take the opposite approach, by allowing controlled formation of hydrates which are transportable, and which do not agglomerate into plugs. If future pipelines may carry multiphase mixtures with gas hydrate, then the prediction tools must be extended accordingly. As part of such an activity, a PhD study will focus on liquid-particle flows. Objective The objective of the study is to perform flow experiments with hydrate-like particles, with emphasis on transient flow conditions (e.g. shut-down, start-up, settling/dispersion, slugging). Activities Particles Different types of model hydrate particles provided by SINTEF Material and Chemistry Group were experimentally studied. Most of the model particles had hydrophobic surface and could not be used in water without any modifications. Model particle with hydrophilic surface was considerably heavier than water. Surface treatment was suggested by SINTEF Material and Chemistry Group to be performed on one of the provided model particles. High concentration chemicals are used in this treatment process and therefore the multiphase flow laboratory is under some modifications to be able to handle such process. The treatment process will be done in January Laboratory As shown in the following figure, the experimental set was designed and different parts of the system including peristaltic pump, motor, propeller, pressure gauge, safety valve, transparent pipe, hose were purchased. The experimental set is under construction and primary experiments will be made in February Instrumentation Video-based observations will be made, as the test pipes are transparent. Pressure recording is part of the standard instrumentation. Inlet slurry flow rate is measured by the peristaltic pump. Conductance rings will be used to measure the particle concentration along the pipe. Flow experiments Experiments will primarily be made to study the development of the hydrate plug at the bends in different slopes and velocities. In continue, transient liquid-particle flows experiments (shut-in / restart / rate changes) will be done in varying pipe geometries (e.g. bends). Flow modeling The experiments will be compared with available flow models: OLGA and research codes at NTNU.

12 PhD Title: Study of Phase Displacement Effects in Oil-Water System Name: Gandi. R. Setyadi Main Supervisor Co-Supervisor : Ole Jorgen Nydal : Tor Ytrehus Stein Tore Johansen Background Accumulation of water in low spots (e.g. in jumpers connecting well heads and pipes), represent a risk for gas hydrate formation risk As the part of flow assurance in gas production activities, flushing out water accumulated in manifold jumper is carried out to prevent line plugging because of the formation of hydrate during shutdown. In this study, oil will be utilized to remove the accumulated water in jumper geometries during the experiments with low flow rate as the practical limitation. The specific character of each of oil-water flow pattern and regime in jumper geometries will be investigated in particular with respect to phase displacement, under transient and steady state conditions. Flow model will be implemented to reproduce the experiments and for practical usefulness. Objective The objectives of the study are to perform limited number of transient flow experiments of water flushing using oil, design and implement flow model to reproduce the experiments and make a comparison with existing models in 1D and 3D. Activities Laboratory The experiments will be carried out in NTNU multiphase laboratory for oil-water flow system, with some modifications which include pipe outlet modification to facilitate water accumulation in the geometries, and low rate oil injection pump. Six (6) cm diameter transparent pipe will be utilized, equipped with impedance probes and video image analysis instrumentation. Flow experiments The flow experiments include the oil-water flow pattern identification in straight pipes near horizontal, some transient experiments of flushing water pipe using oil in straight pipes and jumper geometries Flow modeling Flow model to reproduce the experiments result will be developed using MATLAB for 1D incompressible two fluid model with simple friction closure relations, with friction relations from numerical integration of 2D velocity profiles and transient effects (oil-water front propagation). OLGA and FLUENT will be used as references. 12

13 PhD Title: Modeling of two-phase flow using CFD Name: Kristian Holmås, IFE Supervisor: Jan Nossen IFE, Hans Petter Langtangen, UiO, Ruben Schulkes, UiO Background This PhD project is part of the Horizon project at the Institute for Energy Technology (IFE) and Scandpower Petroleum Technology (SPT). The Horizon project represents a completely new approach for multiphase pipe flow models in that it allows for multi-d simulations of large pipeline systems with the computer speed of 1-D models, realized through pre-integration of velocity profiles prior to run time. In the Horizon project CFD models are used as numerical experiments. CFD allows for the generation of detailed results for virtually all locations within a flow system. At present multi-d CFD models are not functional in studies of large pipeline systems due to long simulation times. However, CFD models can give a better qualitative understanding of multiphase flow phenomena, which is needed to improve the accuracy of the 1-D multiphase flow models. Objective The main objective of this PhD project has been to extract 3-D effects (velocity profiles, turbulence profiles etc.) from the results of the CFD simulations and use this information as an aid to establish accurate cross-sectional models for subsequent pre-integration. Activities The PhD project was started in February 2004 and the thesis is now ready to be delivered. The candidate has tested and further developed various CFD techniques for simulation of two-phase flow. The main focus is on two-phase stratified flow in channels and pipes. The thesis is divided into two parts: Numerical methods for interface problems: In the first part, numerical methods for interface problems are studied and developed. This work has led to three papers. Paper 1 showed the ability of the level set method to predict evolving interfaces. The main result in the first part of the thesis is the sharp interface finite element method that has been developed. It was first presented in Paper 2 for 1-D elliptic problems involving interfaces with discontinuous coefficients and source terms. Later it was extended to 2-D problems and tested on moving interfaces in Paper 3. The sharp interface method was shown to be a significant improvement compared to the standard smooth interface methods in the simulation of two-phase stratified flow. Modelling of two-phase stratified flow: In the second part of the thesis, the physics of turbulent two-phase stratified flow is studied. This work has also led to three papers. Paper 4 (and some additional work) shows the importance of interface turbulence conditions in modelling of stratified two-phase flow. The main result in the second part is the modelling approach using separation of variables and power law expressions for the velocity and eddy viscosity in one space direction, which was presented in Paper 6. Inspired by the work on stratified channel and pipe flow of Biberg (2007), which was studied in Paper 5, a semi-analytic model for fully developed stratified pipe flow was developed. The method provides the potential for efficient simulation of two-phase stratified flow in large pipeline systems with the consistency and accuracy of a CFD model. 13

14 PhD Title: Simulation of waves in pipe flow with an Incompressible Two-Fluid Model Name: Håvard Holmås, Scandpower Petroleum Technology (SPT) Supervisors: Hans Petter Langtangen, UiO, Ruben Schulkes, UiO, Magnus Nordsveen, SPT Objective Investigate new physical formulations and implement higher order numerical methods for resolving waves in multiphase flow pipelines. Performed work The PhD project was started in August The first ten months were spent on continuation of work performed by the PhD candidate in his Master project. Effects of adding a diffusion term to an incompressible two-fluid model were investigated. It was found that second order diffusive terms can remedy the ill-posedness of the model by stabilizing short wavelengths. Simulations confirmed the mathematical analysis by demonstrating convergent numerical solutions for flow conditions beyond the inviscid Kelvin-Helmholtz line. The next step in the PhD project was to study and implement a pseudospectral numerical method for the same two-fluid model. The particular numerical method is well-known in other areas in physics (e.g. turbulence modeling, meteorology, oceanography). To the knowledge of the candidate, however, it has not previously been applied to the modeling of multiphase pipe flow. The motivation for using this method was a need for a numerical scheme that could combine a high degree of accuracy with flexibility in terms of handling different types of equations. Simulations have provided valuable insight with respect to the modeling of transient waves with two-fluid models. Finally, the candidate has in Q3 and Q4 of 2007 been doing some work on the Biberg pre-integrated stratified two-phase flow model and the effects of including profile factors in the momentum flux terms. The Biberg model has been implemented in the pseudospectral numerical framework, and a stability analysis of the model has been carried out. It has been found that the viscous Kelvin- Helmholtz neutral stability line of the model agrees very well with the onset of roll-waves in stratified flow. Courses The compulsory PhD courses have been completed. They include UNIK4600 Mathematical Modeling of Physical Systems, INF-MAT4350 Numerical Linear Algebra and MEK4320 Hydrodynamic Wave Theory. All courses were taken at the University of Oslo. Publications Analysis of a 1D incompressible two-fluid model including artificial diffusion. H. Holmås, T. Sira, M. Nordsveen, H. P. Langtangen, R. Schulkes. IMA Journal of Applied Mathematics (accepted). A high-accuracy and efficient numerical method for 1D multiphase flow. H. Holmås, D. Clamond, H. P. Langtangen. 6th International Conference on Multiphase Flow 2007, Leipzig (published). A pseudospectral Fourier method for a 1D incompressible two-fluid model. H. Holmås, D. Clamond, H. P. Langtangen. International Journal for Numerical Methods in Fluids (accepted). Stability analysis of the Biberg pre-integrated stratified two-phase flow model including profile factors. H. Holmås, D. Biberg, R. Schulkes, G. W. Johnson, T. Sira. BHRG-conference 2008, Banff (accepted). 14

15 STUDENT PROJECTS VALIDATION OF FLOW PERFORMANCE ENGINEERING SOFTWARES Marte Bø Andersen Supervisors: OJ Nydal, Alexandre Goldszal (Total) Jon-Ingar Monsen (Total) A selection of cases were simulated with OLGA and LEDA, both steady state and dynamic. The cases included a field case showing severe slugging, experiemental cases on pressure drop and holdup in inclined flows, and an expeirmetnal case on air-water flow in an S-shaped riser. MArte did her work in the offices of Total in Stavanger DYNAMIC SIMULATION OF INCOMPRESSIBLE TWO PHASE PIPE FLOW Maren Christine Arnulf (SPT) Supervisors: OJ Nydal, Trygve Wangensteen (SPT) Christine acquired some experimental data on the flushing of a water filled pipe with an oil stream during her project work at NTNU. Christine spent her MSc period with SPT at Kjeller, comparing her data with OLGA and OPUS (next OLGA version) simulations. Comparison, Usl 0.5 m/s Hold-up H - 1 Measured H - 1 OPUS H - 1 OLGA H - 4 OPUS H - 4 OLGA H - 4 Measured Time (s) Hold-up (water fracion) plot at the positions of the first and last conductivity probe, for superficial oil velocity 0.5 m/s, measured, and predicted by OLGA and OPUS The computed front propagation is slower than in the experiments for both programs. OPUS (two liquid momentum equations) gives a holdup profile which is closer to the experiments than OLGA does (mixture liquid momentum equation and slip relation). Tests with a second order, explicit mass integration scheme gives a sharper initial drop in water fraction with time (for low flow rates). EXPERIMENTS ON ROLL WAVES IN AIR-WATER FLOWS Andreas Hol Gaarder Supervisors: OJ Nydal, Angela De Leebeeck Angela is considering numerical wave tracking schemes, and looking for simplified integral wave models for that purpose. In support of Angelas work, Andreas made some air-water woll wave experiments in the laboratory, where the main purpose was to measure the pressure across the wave 15

16 front. Roll wave experimets have previously been made at high gas densities at IFE (George Johnson), but not with rapid pressure measurements. The results of Andreas shows a pressure jump in the wave front, with similarities to pressure jumps at slug fronts, see the Figure for an example. Figure Overlapped signals for liquid fraction time trace 3 and 4 and the pressure time trace and the matching video snapshot for the 0 degrees, Usg = 5.89 m/s and Usl = 0.17 m/s. The pressure signal is multiplied by a factor of 10 so that it can be seen on the same plot as the liquid fraction. ECOULEMENTS MULTIPHASIQUES PETROLIERS Sebastien Chevanne Jean Marie was a visiting student from ENSAM, Paris. He did some experiments in a small scale setup, continuing a study of downwards flow regime as function of inlet flow regime. An additional setup was also constructed, to study purging of air by the inflow of water at constant pressure. This is to simulate filling of a sub sea pipeline with water. Some trial experiments were made, but the work should be continued with another student. 16

17 17

18 Attachment to report 2007: Laboratory MULTIPHASE FLOW LABORATORY The main purpose of the loop is to provide a flow laboratory for students (projects, MSc, PhD.) and a demonstration facility which can be used in courses. The main specifications are given in the table below. Table 1: Specifications of multiphase flow loop Test sections Straight Acrylic Straight Steel Flexible PVC S-riser L-riser Fluids Flow rates air water oil Instrumentation dp Absolute p Impedance ring probes Quick closing valves Optical Pressure Process Liquid circulation Air Separation Buffer tank Data acquisition and control System Control NetLab 6 cm ID and 3 cm ID, 16 m long, inclination 6 cm ID, 16 m long, inclination, coated and non-coated 5 cm ID, 30 m long configurable, transparent Acrylic, 5 cm ID, 16 m long, 7 m high Acrylic, 5 cm, 16m long, 7 m high, +- 1 degree first part Air from central system, tap water, oil m/s (vortex meter and coriolis meter) m/s (2 electromagnetic meters) m/s (2 coriolis meters) slow response fast response Electronics for capacitance and conductance Manually operated Video, Cameras and optical box for refraction index matching. Arrangement for moving camera Atmospheric 4 centrifugal pumps, frequency controls 2 dosage pumps Overflow arrangement for pressure controlled liquid flow Central supply, 7 bar 3 m 3 tank for gravitational oil/water separation Air ventilation system Air buffer tank on low pressure side, for variation of the frequencies of terrain slugging Lab-view based interface, Separate HP logging, Fast DAQ cards in PC s Field point modules. Control valves, pump frequency regulation Remote lab control through web site The loop can be operated both manually and automated, also from a web site. Oil and water are circulated with centrifugal pumps and metered before mixing with air at the inlet of the test sections. The air is taken from the central supply at the university and the pressure is reduced to about 3-4 bar in a buffer tank. The air and the liquid streams can be routed to the straight pipes (6 cm ID) mounted on an aluminum beam, or to a pipe configuration mounted on the wall (5 cm ID). The wall arrangements are flexible. With the available bends which have made in the workshop, we can have L-shaped and S-shaped geometries. Impedance probes are used for holdup measurements, flush ring probes and external clamp-on probes. 18

19 Moving and stationary cameras, with optical boxes and light boxes. MINI-LOOP A small-scale, table-top, two phase flow loop has also been instrumented. The loop can be dismantled and transported in a tailor made case. This also makes it easy to bring the loop into lecture rooms. The main use of the loop is for educational purposes. Typical two phase flow situations can be demonstrated and students can perform exercises using the loop. Typical cases are: Flow regimes (limited due to small diameter) Flow development in undulating geometries Terrain slugging Gas lift Gravity dominated flows in undulating pipelines Table 2: Specifications of mini-loop Test sections Flexible PVC Configurable geometry. 16 mm ID, 1-3 m long Straight Acrylic Several sections which can be joined in any desired geometry. 12 mm ID Fluids Air and colored water Instrumentation Absolute pressure at inlet Flow meter air Flow meter water Clamp-on light diodes for slug detection Process Air Compressor Water Pump Separators Acrylic tanks Tanks Buffer tank for adjusting compressibility for terrain slugging Support Aluminum mountings 19

20 20