Documente Academic
Documente Profesional
Documente Cultură
Diploma Thesis Norwegian University of Science and Technology Department of Petroleum Engineering and Applied Geophysics July 2003
Acknowledgement i __________________________________________________________________________ I wish to thank Professor Jon Steinar Gudmundsson for being my supervisor. I am grateful for his enthusiasm, an ocean of suggestions, and excellent supervision throughout this work. Thanks for my supervisor in Poland Dr. Ing. Czeslaw Rybicki to recommend me for Erasmus Link to Norway scholarship and his efforts to enable my take on the study at NTNU. I am thankful to ING AG Leipzig and Norwegian University of Science and Technology for financing my scholarship. I wish to thank Mr Wolfgang Laschet from Office of International Relations for his help to organize my stay in Norway. I also want thank to Professor Danuta Bielewicz and Professor Jan Falkus for their efforts and engagements into international cooperation between universities, and for their appreciate help to surmount the official adversity. In addition I want to thank Jon Rnnevig, Kjell Korsan and Harald Celius from Markland AS, for their guidance into the computer simulations and suggestions towards the obtained results.
ii List of Contents __________________________________________________________________________ List of contents Acknowledgement .i List of contents .ii Nomenclature.v Abstract..1 Introduction ...2 1. Metering of Multiphase Wells 3 1.1 Introduction.....5 1.2 Overview of multiphase metering.......................5 1.3 Challenges and accuracy.............7 1.4 Metering techniques............9 1.5 MFMs Projects..........12 2. Pressure Pulse Technology 13
2.1 Pressure Pulse method...............14 2.2 Water-hammer effect.............14 2.3 Theory and equation......................15 2.4 Pressure surge in wellbores...............16 2.5 Mass and volume flowrates...............................................................17 2.6 Flow condition analysis.....................19 2.7 Concluding remarks..................................20 3. Geothermal Applications 27
3.1. Geothermal energy...........................28 3.2 Geothermal well flow........................................................29 3.3 Well performance .....................................................................32 4. Multi-phase Flow in Wells 37
4.1. Introduction..38 4.2. Main difficulties.. 38 4.3. Phase behaviour... 39 4.4. Definition and variables.. 42 4.5 Fluid properties.... 44 4.6. Flow patterns... 44 4.7. Pressure gradient..47 4.8. Multiphase flow models...48 4.9. Duns and Ros correlation for multiphase flow in oil wells..51 4.10. Duns and Ros modifications..51 4.11. Orkiszewski correlation for multiphase flow in geothermal wells....52
5.1 Introduction 55 5.2 Compressibility of two-phase mixtures. 55 5.3 Compressibility of steam-water system 57 5.4 Acoustic velocity models .. 62 5.5 Attenuation mechanisms of sound wave 66 5.6. Concluding remarks. 68 6. Case studies 74
6.1 Calculation purpose75 6.2 Water-hammer and line packing in oil wells..75 6.3 Water-hammer and line packing in geothermal well106 7. Discussion 156
7.1 Multiphase flow correlations... 157 7.2 Acoustic velocity profile...158 7.3 Line packing 159 7.4 Size of the pressure pulse..161 8. Conclusions 9. References Appendix A Multiphase Metering Projects Appendix B Duns and Ros, Orkiszewski - Multiphase Flow Correlations 162 164 174 183
B.1 Duns and Ross Correlation...184 B.2 Orkiszewski Correlation...192 Appendix C Sound Wave Propagation Process in Steam Water Mixture Appendix D PipeSim 2000-Multiphase Flow Simulator 195 198
D.1 PipeSim Well Performance Analyses.199 D.1.1. Fluid Properties Correlations...199 D.1.2 Advanced calibration data....203 D.2 Profile model..205 D.2.1. Detailed model205 D.2.2. Simplified model.206 D.3 IPR Data.207 D.4 Matching option.208 D.5 VLP correlations and applications.210
E.1 Introduction....215 E.2 Governing equations .215 E.3 The computational models of HOLA 3.1..218 E.4 Heat loss parameters..218 E.5 Wellbore geometry219 E.6 Feedzone properties...220 E.7 Velocities of individual phases..221 E.8 Productivity Index estimation222 Appendix F Simulation results in oil wells 226
F.1 Well A1..227 F.2 Well A2......234 F.3 Well B240 F.4 Well C....246 Appendix G Simulation results in geothermal wells 252
G.1 Well D1.....253 G.2 Well D2.....256 G.3 Well E1......259 G.4 Well E2......262 G.5 Well F1......265 G.6 Well F2......268
Nomenclature v __________________________________________________________________________
a acoustic velocity
d diameter
f friction factor
g - absolute gravity h enthalpy H liquid holdup ID inner diameter k permeability
K slip ratio
KS isentropic compressibility KT isothermal compressibility
L length
t time
T temperature
u velocity
WC - water cut V volume x mass fraction z - direction opposite to gravity
Nomenclature vi __________________________________________________________________________ Greek letters: void fraction water-oil volumetric factor specific heats ratio dynamic viscosity
v specific volume
density
Abstract 1 __________________________________________________________________________ Multiphase flow measurement is of vital importance in petroleum and geothermal industry. Overview of currently available metering techniques has been made in present work. Pressure Pulse method is a new developed method which propose a different approach to measure two-phase flow in wells. The pressure effects after rapid valve closure that built up the method were illustrated. The inspection of the types of geothermal reservoirs allowed characterizing typical parameters of high enthalpy geothermal well. The difficulties to predict the multiphase flow in wells are presented together with description of the definitions and variables that need to be calculated. Multiphase flow models were examined and two most appropriate correlations have been selected for oil and geothermal wells. The speed of sound in two-phase mixtures was calculated. The available models to estimate acoustic velocity were studied and verified with respect to their limitations. The compressibility of steam-water system under the well flow conditions, required for calculations was derived from thermodynamics definitions. The simulations were performed in PipeSim 2000 and HOLA 3.1 programs for oil and geothermal wells respectively, in order to demonstrate the Pressure Pulse method. The case studies include three different North Sea oil wells and likewise three typical high enthalpy geothermal wells. Inflow performance and tubing performance calculations allowed extending the calculation for different diameters and flowrates. The results are presented in form of the tables and plots. Obtained results for oil and geothermal cases were compared to each other. All parameters that affect the acceleration pressure (pressure increase after rapid valve closure) and pressure built up in wells are discussed. The work ends with conclusions towards the performed calculations and gives the assessment for possible application of the Pressure Pulse method to meter the flow in two-phase geothermal wells.
Introduction 2 __________________________________________________________________________ Pipe-flow mixtures of crude oil, gas and water are common in petroleum industry, and yet their measurements nearly always present difficulties. The traditional solution is first to separate the components of the flow, and then measure the flow rate of each component using conventional single-phase flow meters. This method is both inconvenient and expensive to use for well monitoring. In addition the separation is not accurate, about 10% (Millington, 1999). Current multiphase meters have similar accuracy, they employ the complex techniques, and some of them contain the dangerous radioactive materials as discussed in Chapter 1. In geothermal wells producing steam and water mixture under various operating conditions the capability accurately measure the flow is also of value importance for several reasons similar to petroleum industry. These are general evaluation of the geothermal reservoir under proper reservoir management, optimalisation of the wellbore design from well deliverability considerations and minimization of scale deposits in the wellbore (Ragnarsson, 2000). The background for this thesis work is a new method to measure multiphase follow (Gudmundsson and Falk, 1999; Gudmundsson and Celius, 1999), developed at Norwegian University of Science and Technology (NTNU). The metering method is simple, requires little space, and is cost effective with at least the same accuracy as the competitors (Gudmundsson and Celius, 1999). The multiphase flow meter is based on measurements of pressure magnitude and pressure build-up. The output is velocity and density of the gasliquid flow. This thesis work concerns multiphase metering, specifically pressure transients caused by a rapid valve closure in oil and geothermal wells. Pressure propagation in fluids is closely related to sound velocity. The acoustic velocity in two phase mixtures varies significantly from this in single liquid or gaseous phase, and depends on physical properties of every mixture constituents. Available models for acoustic velocity in two-phase mixtures need to be verified according to their limitations in order to find the most appropriate for particular calculations.
Introduction 3 __________________________________________________________________________ Multiphase flow is a complex, turbulent and highly nonlinear process, which can not be fully described mathematically due to increased numbers of flow parameters. Computer simulations base on semi-empirical correlations were developed in order to predict the pressure and fluid parameters changes across the wellbore. Calculating flowing pressure profiles in oil wells, phase transfer between oil and gas requires a rather simple treatment, and is accomplished trough the use of solution gas-oil ratio Rs relationship. In geothermal wells, however phase transfer between water and steam attains critical importance and calculations must incorporate the steam tables accurately. Pressure profile calculations for geothermal wells vary from those for oil well in another important aspect in that the temperature of the fluid must be computed precisely. This thesis describes how the Pressure Pulse method can be used to meter the flow in high enthalpy two-phase steam-water geothermal wells similarly to oil wells. The calculations performed aim to estimate the size of pressure pulse after the valve closure and determine the parameters affecting the early pressure build-up.
Chapter 1
Metering of Multiphase Wells
1. Metering of Multiphase Wells 5 __________________________________________________________________________ 1.1 Introduction Multiphase means a single component existing in a variety of phases such as steam, water and ice. In the oil industry multiphase refers to a stream of fluid containing a liquid hydrocarbon phase (crude or condensate), a gaseous phase (natural gas, and non hydrocarbon gases), a produced water phase, and solids phase (sand, wax, or hydrates). In general the quantities of solids produced are minimal and thus have less impact than the liquid and gas phases. In present thesis work some simplification will be made and two phase, liquid phase and gas phase will be considered. The mixture of two immiscible fluids will be termed as liquid phase, regardless to components number. The mixture of gases flowing together will usually, unless there is a large density difference and little turbulence, diffuse together and can be treated as single homogenous phase (McNeil, 1990). Multiphase measurement is the measurement of the liquid and gas phases in a production stream without the benefit of prior separation of the phases before entering the meter. 1.2 Overview of multiphase metering While two-phase and multiphase flows have been common throughout petroleum industry for many years, there has until very recently been little or no demand for real-time metering of such flows. Traditionally the problem was circumvented by separating the flow into its constituent components, which allowed straightforward single phase metering techniques to be used (Theuveny et al., 2001). This approach was very practical and effective, but did give rise to processing systems which were quite inflexible in terms of their capability to handle fluctuating flowrates, varying water contents, and changes in the physical properties of production fluids. However in the early years of offshore North Sea production this was not a major problem, and at that time - pre 1980 - there was little or no impetus to develop more sophisticated metering technology that could perhaps dispense with separation equipment and expensive metering facilities (Falcone et al., 2002).
1. Metering of Multiphase Wells 6 __________________________________________________________________________ During the 1980s the process of gradually declining oil production from the major North Sea fields started, so in the interests of operational cost effectiveness, there was a move to use existing platform based process plant for other production roles (Steward, 2003). To maintain production levels, smaller satellite fields which were previously uneconomic to produce on a stand alone basis, were tied back to existing platform based infrastructure. From a technological point of view this introduced a step change in the complexity of production. There were now numerous fields, typically with quite different oil properties, water contents and gas fractions, all being produced through process plant designed for the early years of single-well production (Theuveny et al., 2001). Furthermore, the water contents and gas fractions started to increase, and this exacerbated the production problems even further. It began to emerge quite quickly that more operationally flexible multiphase technologies were going to be needed, if not immediately, certainly within five to ten years (Steward, 2003). For existing platforms the prime purpose of this new technology would be to improve processing flexibility, and for new field developments the aim would be to completely eliminate the need for costly and bulky platform based process plant (Falcone et al., 2002). The ultimate aim was of course to move towards remote subsea instrumentation. The key driver at all times being lower production costs through reduced initial capital expenditure, and reduced operating manpower. To take up these challenges, the growth in multiphase research and development since the early 1980s has been exponential, especially with regard to metering, and today there are a variety of multiphase flowmeter (MFM) installed onshore and offshore. It appears to be no reduction in new metering developments (Steward, 2003). However the actual growth rate of installations has been lower than initial industry forecast suggested (Falcone et al., 2002). Oil companies have been hesitant to invest in expense meters with limited tracks record. Figure 1.1 shows actual trend up, tied with very low level of utilisation MFM technology before the 2000. The reasons may be the fact that when operators decide between a traditional approach to the production facilities and one including MFM, must compare the capital and operating expenses of each solution. Very little operational history of MFM cause difficult to predict the operating costs (Jamieson, 1999). This difficulty results from relatively low number of MFM applications worldwide, allow claiming that widespread implementation of MFM
1. Metering of Multiphase Wells 7 __________________________________________________________________________ cannot take place until expertise is spread more widely trough oil industry (Falcone et al., 2002).
900
Number of MFM istalations
800 700 600 500 400 300 200 100 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 0 1 5 0 2 5 0 10 3 1 17 6 5 29 24 22 52 65 20 66 78 43 103 129 55 210 542
Figure 1.1 Grow rates of MFM installations (Falcone et al., 2002) 1.3 Challenges and accuracy The level of difficulties in accurately measuring the multiphase stream is increased dramatically over single phase measurement. Single phase fluids can be quantified by knowing about the pressure, fluid density, viscosity, compressibility and geometry of the measurement device (Williams, 1994). Unfortunately, multiphase fluids do not act in the same manner as single phase fluids and above variables of each phase would not quantify multiphase flow. Multiphase flow is a complex, turbulent, highly non linear process. Williams (1994), and also King (1999) give the brief description of the processes that may take place as the different phases flows simultaneously. The phases interact with each other gas may evolve out of the
1. Metering of Multiphase Wells 8 __________________________________________________________________________ solution, or is absorbed into the liquid, waxes and hydrates may precipitate etc. If the single component exists in the two phases there is significant mass transfer and thus mixture quality may be considered variable. The components do not mix homogenously, and tend to remain separate, the water does not mix well with the oil, and gas remains separate from the liquid phase. Both phases flow at the different velocities. It is common for gas and liquid to flow at the different rates. Very complex flow regimes can exist and are dependent on the relative velocity of the phases, fluid properties, pipe configuration and flow orientation. The mentioned above and other relevant to the multiphase flow parameters are described in Chapter 4, which deals with multiphase flow in wells. The parameters definitions and relationship between them are given together with the correlations developed in order to predict the flow behaviour. Expectations of MFM performance in the early days were concluded and sometimes in the fiscal range of accuracy. Such levels of accuracy were, and never will be achievable by present technology (Steward, 2003). Over the last ten years, a gradually more realistic assessment of uncertainty capabilities has evolved. To date, no international regulations for MFM accuracy has been delivered. Varying level of accuracy requirements exists in multiphase measurement depend on how the information will be utilized. Essentially, three main accuracy requirements exist for metering multiphase fluids (Falcone et al., 2002): approximately 5 -10% for reservoir management, approximately 2-5% for production allocation, and approximately 0.25-1% for fiscal metering, are anticipated to be required.
However because of high complexity of multiphase mixtures it may be optimistic to claim that the above ranges of accuracy apply to any regime and for any chemistry of the fluids.
1. Metering of Multiphase Wells 9 __________________________________________________________________________ 1.4 Metering Techniques Under the multiphase flow circumstances the following parameters are required to compute flowrates of each phase: the cross-sectional area of the pipe occupied by each phase the axial velocity of each phase density of each phase.
The cross-section area and phase velocities give the volumetric phase flowrates. The product of phase densities and phase volumetric flowrates gives the phase mass flow rate. Unfortunately, at the present time there is no method of measuring phase fraction directly, they are derived from two independent measurements, coupled with the continuity equation which requires the sum of oil water and gas phase fraction to equal unity. Typically, two variables independent are the density of the entire flow, and the water content in the liquid phase. Once these are measured, some simple mathematical analysis allows the individual phase fractions to be calculated. With these technology limitations, projects aimed at developing multiphase meters have tended to adopt one of two metering strategies (Millington 1999): A set of sensors that take volume measurements, which when combined are capable
of isolating the individual phase fractions. A combination of flow models and velocity measurements are used to derive the phase velocities as functions of time. To determine densities, temperature and pressure are measured and assumed equal in all phases. A set of sensors which again take volume measurements, but which also require flow
to be conditioned such that only one mixture velocity measured is assumed to be required . The overall mixture density is considered representative of the three individual phases. Phase fraction data is required as above.
1. Metering of Multiphase Wells 10 __________________________________________________________________________ Following sensors and techniques are commonly used: Gamma Densitometers consist of radioactive source and detector, placed so that the beam passes trough the flow and is monitored on the opposite side of the multiphase mixture. The amount of radiation that is absorbed or scattered by the fluid is a function of both fluid density and energy level of the source. Typical radioactive sources used include isotopes of caesium, barium or americium. Single energy gamma sensors are those that incorporate only one source or monitor only one energy level from source. These devices are often used to measure the density of the multiphase mixture. Dual energy gamma sensors measure the absorption of two separate energy levels. The two energy levels are provided either by two isotopes or by a single isotope that has two discernible levels. If two energy levels are far enough apart, these two independent absorption measurements can be used to determine the oil, gas, and water phase volume fractions. The densitometers are frequently calibrated by filling the device with known fluids, typically gas (or empty pipe) and water. Capacitance Sensors measure the dielectric properties of fluid. Each sensor consists of a pair of metal plates or electrodes. These are mounted on the pipe wall or are otherwise located so that the fluid occupies the space between them. The capacitance of the fluid is measured by varying the voltage difference between the plates and measuring the resulting electric current between them. From the capacitance, the dielectric constant of the mixture can be calculated. Since the dielectric constant of the mixture is a known function of the composition, this information can be used to calculate the volume fractions of oil, gas, and water phases. This technique will work for mixtures in which the liquid (oil/water mix) is oil continuous. Since the water phase is a much better conductor of electricity, water continuous mixtures will effectively "short" the capacitance plates rendering the measurement ineffective. For water continuous liquids, an approach based on conductance is used (see below). Conductance / Inductance Sensors use an electrical coil around the pipe to induce a current in the flowing multiphase mixture. The magnitude of this induced current is related to the dielectric constant of the mixture, which can be used to determine the (mixture composition as with the capacitance and microwave sensors.
1. Metering of Multiphase Wells 11 __________________________________________________________________________ Microwave Sensors measure the dielectric properties to help determine the phase fractions of the multiphase mixture. The sensor consists of emitters and receivers (antennae) of electromagnetic waves in the MHz or GHz range (microwaves). The dielectric constant of the mixture is a function of both the frequency of the waves and the mixture conductivity. The measured dielectric constant is a volume weighted average of the individual phase dielectric constants. The conductivity and dielectric constant of the water phase is a function of salinity. As such, meters that use this technique either need brine salinity as a calibration variable or have some other way of estimating it on-line. Cross Correlation Techniques use two similar measurements, each in a different axial location in the pipe. By comparing the two measurements, the velocity of the flow feature is determined, for example, the time required for a bubble to travel between the two sensors. Implicit in this technique is a measurable amount of non-homogeneity in the multi phase flow. For this reason, many available meters require the Gas Volume Flow (GVF) to be within certain limits, far enough from the pure liquid (GVF = 0) and pure gas limits (GVF = l) that the flow does not appear homogeneous to the sensors. Gamma densitometers, microwave sensors, and capacitance sensors are used in MPM systems for cross correlation. Venturi Meters consist of a gradual restriction in the flow path, followed by a gradual enlargement. For single phase flows, the pressure drop across the restriction is a straightforward function of the velocity and density of the fluid. For multiphase flows, the analysis is more complicated. The gradual restriction in the flow path makes the Venturi meter slightly intrusive to the flow. Positive Displacement Meters (PD) rely on the metered fluid to rotate mechanical gears or rotors in the flow path. Each rotation of the rotor corresponds to a known amount of volume passing through the meter. PD meters are commonly used in single phase service. For full well stream production, risks due to erosion and blockage should be considered.
1. Metering of Multiphase Wells 12 __________________________________________________________________________ 1.5 MFMs Projects A very limited amount of information is available on MFMs performance. In the oil and gas sector where competition is always intense a black box MFMs packages are usually offered, where very little is unveiled. A brief description of some MFMs projects that are now commercially available is given in Appendix A, together with the tables containing comparison of the methods with regard to the techniques that are used for measurement purposes.
Chapter 2
Pressure Pulse Technology
2. Pressure Pulse Technology 14 __________________________________________________________________________ 2.1 Pressure Pulse method Multiphase metering in oilfield operation is of considerable interest in petroleum industry as described in Chapter 1. As the response for these needs new method called Pressure-Pulse has been developed at NTNU by Professor Gudmundsson. The method is based on the propagation properties of pressure waves in gas-liquid media. Waves generated in gas-liquid mixture flowing in a pipe at a speed of sound will propagate as pressure pulses (Gudmundsson and Celius, 1999). These effects called water-hammer and line packing are described precisely below. The method has been tested in several offshore platforms including Gullfalks A, Gullfalks B, and Oseberg B, with positive repeatable results similar to the theoretical models (Gudmundsson, Falk, 1999). Total of 800 tests were run on 12 different gravel packed wells. No negative effects were observed on the production system or the reservoir during the 11-month test period. The method has the advantage of being simple, low-cost, and gives the same accuracy as the competitors (Gudmundsson and Celius, 1999). Pressure is the easiest parameter to measure in the production of oil and gas. It can be measured in pipelines, flowlines and wellbores; at wellhead, chokes, manifolds, and separators. The widespread use of the quick acting valves in the oil industry to open, close, and control pipeline and wellbore flow, has made it possible to harness the information contained in the rapid pressure transients when a valve is activated (Gudmundsson et al., 2002). 2.2 Water-hammer effect The water-hammer effect can be caused by a rapid closure a valve in pipe line with flowing liquid. The immediate pressure increase created by the valve is referred as the acceleration pressure-pulse pa. Wylie and Streeter (1993) described how this increase in pressure travels in the pipe with the velocity of sound, and stop the flow as it passes. The instant the valve is closed, the fluid immediately adjacent to it is brought to rest by the impulse of the higher pressure developed at the face of the valve. As soon as the first layer is stopped, the same action is applied to the next layer of fluid bringing it to rest. In this manner a pulse wave of
2. Pressure Pulse Technology 15 __________________________________________________________________________ high pressure is visualised as travelling upstream at same sonic velocity. However in long pipe flows with high frictional pressure loss the accelerational pressure-transient is attenuated and does not stop the flow completely. Yet, since the fluid must stop by the valve, there is a continuous pressure increase near the valve also after is wholly closed. The name of these phenomena is line packing. Figure 2.1 illustrates the water-hammer effect. 2.3 Theory and equation Water-hammer phenomena, line packing and pressure pulse velocities are essential for the new multiphase method. Water-hammer pressure transient can be found using homogenous continuity equation at high pressure well conditions fluids are well mixed and thus homogenous continuity equation can be applied (Falk, 1999). Continuity equation
p p u +u =0 + a2 x x t
(2.1)
(2.2)
The characteristic pressure pulse velocity running upstream the valve is Thus,
p a 2 u u p =0 + + t a u t a u t p u = a t t
x = a u. t
(2.3)
(2.4)
2. Pressure Pulse Technology 16 __________________________________________________________________________ During quick valve closure the velocity jump is u = -u in a short period of time t. The water-hammer due this retardation is
p a = a u
This equation is generally known in literature as Joukowski equation. Momentum conservation principle is given as
(2.5)
u u u u 1 p dz +u + = f g t x x 2d dx
(2.6)
In steady-state turbulent pipe flow frictional pressure gradient is represented by DarcyWeisbach equation
p f
f u2 2d
(2.7)
where, f is the dimensionless friction factor. The frictional pressure gradient is made available to measure when the flow is brought to the rest after valve closure. The line-packing pressure increase, in liquid-only flow represents the pressure drop with distance in the pipeline. In two-phase flow line-packing is more complicated and in addition to frictional pressure gradient it contains also increase in waterhammer with upstream distance. In vertical gas-liquid wells pressure increase with depth and hence the water-hammer changes with depth (Gudmundsson and Celius, 1999).
2.4 Pressure surge in wellbores
Using a high sampling rate and high resolution pressure gauge, pressure buildup is possible to record. A typical pressure-pulse technology set up is shown in Figure 2.2. It contains a quick-
acting valve and two pressure transducers A and B upstream of the valve taking samples in micro to mili seconds time period. Today technology can definitely provide such high sampling gauges. A valve is termed quick-acting if it closes completely before waves are reflected from up-stream or downstream. If there are reflections before valve is closed, the pressure on closing will be affected (Gudmundsson and Falk, 1999). The example of measured pressure from two transducers is shown on Figure 2.3. Pressure Pulse is measured at two locations spaced 83.35m up-stream a quick acting valve. The speed of sound may be estimated from cross correlation between the signals. In this example figure this is 170 m/s. By knowing the mixture density, acoustic velocity for the mixture and pressure increase due to acoustic term during a quick shut-in, mixture velocity may be calculated at the wellhead (Gudmundsson, 1999). The studies of Khokhar (1994) suggest that the phenomena like wellbore storage, skin effect, and phase redistribution that occur after the well shut have no effect on the pressure technique. Whereas, the pressure-pulse method dependents more on mixture composition and gas-liquid ratio of the well fluids which influence the acoustic velocity. The speed of sound in two-phase mixtures, and its dependency on fluid properties and PVT conditions will be investigated in Chapter 5 of this work. In the Pressure-Pulse method the sound speed can be determined from cross-correlation of two pressure signals from locations A and B, as indicated in Fig. 2.4. The testing of the Pressure-Pulse method on several North Sea fields has resulted in measurements that make this possible (Gudmundsson, Falk, 1999).
2.5 Mass and volume flowrates
The mass flowrate in a pipe of constant cross-sectional area can be obtained directly from the Joukowski water-hammer equation, when the sound speed is also determined from crosscorrelation of the measured delay time between two signals from transducers A and B. A kg a s
m = p a
(2.8)
The continuity principle dictates that the mass flow rate at the valve is the same as the mass flow rate at other locations. Mixture density and the mixture velocity can be also obtained from the measurements f L p a
2 2
mix =
2 d a p f
kg m3
(2.9)
v mix =
2 d a p f kg 3 f L p a m
(2.10)
Knowing the density of individual phases of the fluid mixture, void fraction can be calculated
L mix L g
(2.11)
Flow rates in petroleum industry are traditionally expressed in volumetric quantities. The mass flowrate and the volumetric flowrate of the liquid are related trough relationship
m Sm 3 s
q=
(2.12)
Treating about volumetric flow rates requires volumetric factor to be taken into consideration. Volumetric factor B(p,T), indicates the effects of pressure and temperature changes, from reservoir to stock-tank conditions. Thus, volumetric flowrates for oil can be calculated as
Sm 3 q o ( p, T ) = o Bo ( p , T ) s
(2.13)
o ( p, T ) =
o + g R s ( p, T ) kg
Bo ( p, T ) m3
(2.14)
(2.15)
Sm 3 q g ( p, T ) = q o GOR R s ( p, T ) B g ( p, T ) s
(2.16)
where:
Sm 3 gas R( p, T ) - amount of dissolved gas in oil 3 Sm oil Sm 3 GOR gas oil ratio at standard conditions 3 m
If the oil produced contains water, also watercut WC [%] need to be known.
2.6 Flow condition analysis
The pressure profile in a pipeline can be used to detect and monitor solid deposits as shown on Figure 2.5. Deposits will change the frictional pressure drop in the affected interval both by change pipe roughness and by reducing the tubing diameter. This will show up as increase in the line packing gradient in the affected region. When the valve is activated the pressure is measured resulting in a pressure time log. The pressure - time log is then converted into pressure - distance log. Those give the location and extend of the deposits in a pipeline. Pressure Pulse testing can be also used in gas lift wells for flow rate metering and flow conditions analysis. An examination of the line packing pressures makes it possible to identify the location of gas injection points, and asses the status gas lift valves (Gudmundsson
et al., 2002). Figure 2.6 shows an example of the simulations for three different valve locations. The bubble point depth may be identified from line packing as it appears with the peak on the time derivative plot. Figure 2.7 shows typical bubble point pressure response experienced during the fields tests.
The water-hammer theory treats pressure-pulse propagation in single phase flow, and has also been directly extended to multiphase flow. This theory is important for the new multiphase meter. Multiphase flow models like the drift flux model, the homogenous model and certain forms of the two-fluid models could predict pressure pulse. However, the assumptions of the one pressure in the one-dimensional two-fluid model are not appropriate. This thesis uses the homogenous model, where due to large pressure surge fluid homogeneity and continuity may be assumed. The multiphase models are described in Chapter 4 of this work that treats about multiphase flow in wells.
Figure 2.6 Simulation results for three different valve locations (Gudmundsson et al, 2001)
Figure 2.7 Typical bubble point pressure response during the field tests.
Chapter 3
Geothermal Applications
Geothermal energy is one of the cleaner forms of energy now available in commercial quantities. The use of this alternative energy source, with low atmospheric emissions, has a beneficial effect on our environment by displacing more polluting fossil and nuclear fuels. Thermal energy carried in the produced fluid can be used for direct heating in residential, agricultural, and industrial applications; or the thermal energy of higher temperature systems can be used to produce electricity. Rapidly growing energy needs around the world will make geothermal energy exceedingly important in several countries. For example in Iceland provides 50% of the total power supply, and 86% energy used for space heating (Ragnarsson, 2000). The production of electricity requires a greater concentration of energy than other applications. If hot fluid is available in great enough quantities, a geothermal power plant can be installed that uses the produced steam directly to drive a turbine generator system. Geopressured geothermal reservoirs are closely analogous to the geopressure oil and gas reservoirs. Fluid caught in stratigraphic trap may be raised to litostratic pressure due to overburden pressure. Such reservoirs are given fairly deep (over 2,000 m), so that the geothermal gradient can give temperature over 100oC (Grand, 1982). A number of such reservoirs have been found in drilling for oil and gas. These reservoirs derive their heat from the terrestrial heat flux, and are widespread throughout the world. It occurred not economic to exploit even the most favourable reservoirs for a long time, but over the last decade many projects arise to utilise their energy (Dickson and Fanelli, 2001). In some places over the world high temperature over 250 oC geothermal reservoirs occur. That heat source may be either an abnormal high geothermal gradient or volcanism nature. Those fields usually display surface activity when high temperature fluid systems transfer heat to the surface from crustal rocks heated by magmas and are mainly located in six countries: United States, Mexico, New Zealand, Philippines, Iceland and Italy (Gudmundsson and Ambasth, 1986).
High temperature geothermal reservoirs can be liquid and vapour dominated this mean that can have liquid only or steam-water feedzone (Gudmundsson, 1989). Steam dominated reservoirs are relatively rare, and most geothermal fields are water-dominated, where liquid water at high temperature, but also under high (hydrostatic) pressure, is the pressurecontrolling medium filling the fractured and porous rocks. When liquid water flows into a geothermal well, the water will remain liquid up the wellbore until reaching a depth where the pressure is equal to the saturation pressure. The pressure decreases as the water moves toward the surface and at this depth the liquid water will start to flash to form a steam. It will continue to flash until reaching the wellhead, surface pipeline, and eventually the steam separator. Beginning with liquid water, the first flashing results in comparatively small amounts of stream that flows as a bubbles trough a continuous column of water. With pressure drop towards the surface more steam evaporates and thus flow changes the regime into slug and steam continuous annular flow (Gudmundsson, 1989). The two-phase output from geothermal wells is piped to a separator to produce steam for electric power generation. The liquid water separated from the steam is disposed of at the surface or injected back into the reservoir. Reinjection of the geothermal liquid back into reservoir after use has a number of purposes. The most important ones being (Eliasson, 2001): disposal being use liquid without polluting the environment, sustenance of the reservoir pressure to counteract with drawn down and surface subsidence, mining of heat stored in hot formations simultaneously extend the useful life of the reservoir. The most common approach for measuring flow rate similarly to the petroleum industry where gas is separated from oil, in geothermal applications separator is also used, where steam-water mixture is separated into a flow of water and steam at the pressure of separator. The flow of each phase can be then measured individually using pressure differential devices.
High temperature wells are typically drilled in four stages: (Figure 3.1) a wide hole to a depth of 50-100 [m] into which is cemented the surface casing. a narrower hole to a depth of 200-600 [m] into which the anchor casing is cemented. a narrower hole still to a depth of 600 to 1,200 [m] which carries a cemented casing called the production casing finally the production part of the wells drilled into and/or trough the active aquifer. This part carries a perforated liner that is hung from the production casing reaching almost to the well bottom. On top (wellhead) the well is fitted with expansion provision and a sturdy sliding plate valve (master valve). It is also commingled to a muffler, usually of a steel cylinder fitted with an expanding steam inlet pipe to low down the fluid on entry. Steam capacity of these wells commonly range between 3 30 [kg/s] (1.5 MWe 15 MWe) (Eliasson, 2001). The drilling programs are of two types: Standard: Wide: Surface casing 22 nominal diameter in a 24 hole. Anchor casing 18 nominal diameter in a 20 hole. Production casing 13 3/8 nominal diameter in a 17 1/2 hole Liner 9 5/8 nominal diameter in a 12 1/2 hole. Surface casing 18 nominal diameter in a 20 hole. Anchor casing 13 3/8 nominal diameter in a 17 1/2 hole. Production casing 9 5/8 nominal diameter in a 12 1/2 hole Liner 7 nominal diameter in an 8 1/2 hole.
Wide tubing configurations has been initially implemented in order to cut down the frequency of wellbore cleaning due to calcium carbonate scale depositions (Gudmundsson, 1986). These wide 13 3/8 production casing has been particularly developed in Iceland and
the narrower 9 5/8 production casing is reported by many authors as the typical (Uphady et al., 1977), (Gudmundsson and Thrainsson, 1988) . In present work, data about the wells was taken from Icelandic sources for 13 3/8 production casing, from two fields Reykjanes and Svartsengi. Nevertheless the deliverability considerations, presented in next section allowed finding the operating parameters assuming the 9 5/8 production casing and thus simulations covered the both typical tubing sizes. Geothermal wells have total mass flowrates are greater than oil and gas wells, primarily due to width casing configuration presented above that allows yield such high mass flowrates. The calculations made in this work confirmed that diameter change from 13 3/8 to 9 5/8 allows yield almost double output. Typical exploitation parameters gained from literature are placed in the Table 3.1 Table 3.1 Typical exploitation parameters
Variables Range
total mass flowrate wellhead pressure wellhead temperature wellhead enthalpy well depth
12.9 - 68.6 [kg/s] 2.3 - 56.5 [bar] 150 - 250 [oC] 965 - 1966 [kJ/kg] 913 - 2600 [m]
Geothermal reservoir data inspection presented here is required to characterise typical geothermal well that can be used in pressure pulse simulations. Simulations will be done for various mass flowrates, and wellhead pressures. In order to predict the above parameters, deliverability method developed in petroleum industry and widely applied also for geothermal reservoir engineering is necessary.
The production of liquid water from a geothermal reservoir depends on the reservoir pressure, the flow of fluid trough the feedzone into the well, and then up the wellbore to the surface. These three elements of deliverability are called reservoir, inflow and vertical lift performance respectively. The production output test gives the deliverability at the time of testing. As the production proceeds with the time the deliverability is like to change because of drawdown in reservoir pressure. The prediction of the reservoir pressure with time is the subject of reservoir modelling, and is not necessary to be discussed here. The fluid entering flowing well in liquid dominated reservoir contains pressurized water. Nevertheless when well flowing pressure pwf decrease below saturation pressure psat a twophase mixture of steam vapor and liquid water flows into the wellbore as a result of flashing outside the wellbore. The flashing occurs over a relatively short distance near the wellbore. This indicates that rapid pressure drop and radial-flow effects in the wellbore region may control the output characteristics of geothermal well (Gudmundsson, 1986). The inflow performance curve for geothermal well is composed of two forms of flow behavior, depending upon whether the flowing pressure is above or below the saturation pressure of the geothermal fluid. Above the saturation pressure a linear relationship as assumed between the mass flowrate m and the well flow pressure pwf. The inflow performance curve for geothermal well is composed of two forms of flow behavior, depending upon whether the flowing pressure is above or below the saturation pressure of the geothermal fluid. Above the saturation pressure a linear relationship is assumed between the mass flowrate m and the well flow pressure pwf. In general the mass flowrate increase when pressure difference enlarges as can be expressed:
m = PI ( p r p wf
where:
kg ) s
(3.1)
kg m - mass flowrate s
p r - average reservoir pressure [bar ]
This equation applies for single-phase Darcy flow into the wellbore. If the pressure in a near wellbore distance decreases below bubble point pressure the slope of inflow performance curve is assumed to become more negative. This indicates that when steam-water mixture enters the wellbore, the resistance to flow is grater than for liquid only flow for the same flowrate. It is like a solution-gas drive reservoir in petroleum industry, and thus equation from petroleum industry can be adopted, after some modifications. The orginal form of the equation for oil and gas is given below
1 2 k h B o o pr re 3 ln r +s w 4 kr o 2 k h B o o pb re 3 ln r +s w 4
qo =
( p r pb ) +
(p
2 b
p wf
) sm
2 pb
s
3
(3.2)
where:
k permeability [m2] h thickness [m] dynamic viscosity [Pas] re effective radious [m] rw well radious [m]
Using HOLA 3.3 simulator described in Chapter 6, it is possible to estimate Productivity Index PI for given wellhead flow conditions.
In Darcy law for two phase flow the fluids are assumed to flow practically independently of each other. The fundamental law is then applied to the two phase flow individually. In geothermal simulation studies of two phase reservoir flow, the relative permeability for steam and water need to be defined. The following expression gives the total mass flowrates:
k k dp kg m = A k rw + rs w s dL s w s psat
(3.3)
Thus, to calculate curve for liquid only feedzone, when well flow pressure pwf above saturation pressure, the following equation can be used:
m = PI ( p r p wf
kg ) s
(3.4)
And as the well flow pressure pwf decrease below saturation pressure the equation (3.3) can be used in form
sm 3 s
(3.5)
where: psat saturation pressure [bar] Gudmundsson et al. (1986) in their study of relative permeabilities give the necessary relations. The relative permeability ratio of vapor and water can be calculated from equation
Sw 1 w = K s 1 Sw
kr w kr s
(3.6)
Assuming that there is no interaction between the flowing phases, that is steam and water are assumed to flow independently, the retaliations can be made
kr w + kr s = 1
(3.7)
Relative permeability for steam and water can be found from following functions:
S w < 0.4,
kr w = Sw kr w = Sw
0, 6
; ;
0,7
S w < 0.2,
kr w = Sw
0 , 77
The other way to calculate the two-phase performance curve part, below the saturation pressure is to use the Vogel empirical relationship obtained for the situation when gas is coming out of the solution (Gudmundsson, 1986).The equation has form
2
p wf m = 1 0 .2 p m max sat
p 0.8 wf p sat
(3.8)
Where mmax is the ideally maximum flowrate obtained assuming pwf = 1[bar] Vertical lift performance curves were calculated for particular wellhead pressures pwh, using HOLA 3.1 wellbore simulator. Then MATHLAB 6.5 program has been used to calculate the deliverability curves. The mass flowrate of steam and water from geothermal reservoirwellbore system is given by well operating point, determined by the intersection of the IPR and VLP curves.
Chapter 4
Multiphase Flow in Wells
Two-phase flow occurs commonly in the petroleum, geothermal, chemical, civil, and nuclear power industries. In the petroleum and geothermal industry, two-phase flow is encountered in well production, transportation, processing systems. The complex nature of two-phase flow challenges production engineers with problems of understanding, analyzing, and modelling two-phase-flow systems. The calculation and prediction methods that are discussed in this chapter were developed for petroleum industry. Geothermal applications also base on this method, however due to different water and crude nature a different approach is required in some cases.
4.2. Main difficulties
When two or more phases flow simultaneously in pipes, the flow behaviour is much more complex than for single-phase flow. Phases tend to separate because of differences in density. Shear stresses at the pipe wall are different for each phase as a result of their different densities and viscosities. Expansion of the highly compressible gas phase with decreasing pressure increases the in-situ volumetric flow rate of the gas. As a result, the gas and the liquid phases normally do not travel at the same velocity in the pipe, upward flow the less dense, more compressible, less viscous phase tends to flow at a higher velocity than the liquid phase, causing a phenomenon known as slippage. However, for down flow, the liquid often flows faster than the gas. Perhaps the most distinguishing aspect of multiphase flow is variation in the physical distribution of the phases in the flow conduit characteristic known as flow pattern or flow regime (Brill, 1999). During multi-phase flow through pipes, the flow pattern that exists depends on the relative magnitudes of the forces that act on the fluids. Buoyancy turbulence, inertia, and surface-tension forces vary significantly with flow rates, pipe diameter, inclination angle, and fluid properties of the phases (Brill, 2001). Several different flow patterns can exist in a given well result of the large pressure and temperature changes the fluids encounter (Manabe et al., 2001). Especially important is the significant variation in
pressure gradient with flow pattern. Thus, the ability to predict flow pattern as a function of the flow parameters is of primary concern. Analytical solutions are available for many single-phase flow problems. Even when empirical correlations were necessary (i.e., for turbulent-flow friction factors), the accuracy of prediction was excellent. The increased complexity of multiphase flow logically resulted in a higher degree of empiricism for predicting flow behaviour. Many empirical correlations have been developed to predict flow pattern, slippage between phases, friction factors, and other such parameters for multi-phase flow in pipes. Virtually all the existing standard design method relies on these empirical correlations. However, since the mid-1970s, a dramatic advance have taken places that improve understand the fundamental mechanisms that govern multiphase flow. These have resulted in new predictive methods that rely much less on empirical correlations. This chapter introduces and discusses basic definitions for parameters unique to multiphase flow in pipes. Flow patterns are described in detail, including methods available to predict their occurrence. The use of empirical correlations based on dimensional analysis and dynamic similarity performed by software used in this work are presented.
4.3. Phase behaviour
Two-phase can be interpreted as a single component like a water and its vapour steam, and a complex mixture of various components like a hydrocarbons composition. Geothermal fluid or complex mixture of hydrocarbon compounds or components can exists as a single-phase liquid, a single-phase gas, or as a two-phase mixture, depending on the pressure, temperature, and the composition of the mixture (Campbell, 1994). Unlike to a single component or compound, such as water-steam system, when two phases exist simultaneously a multicomponent mixture will exhibit an envelope rather than single line on a pressure/temperature diagram. Figure (3.1) gives a typical phase diagram for a multicomponent hydrocarbon system. Shapes and ranges of pressure and temperature for
actual envelopes vary widely with composition. Figure (3.1) permits a qualitative classification of the types of reservoirs encountered in oil and gas systems. Typical oil reservoir has temperatures below the critical temperature of the hydrocarbon mixture. Volatile oil and condensate reservoirs normally have temperatures between the critical temperature and the cricondentherm for the hydrocarbon mixture. Dry gas reservoirs have temperature above the cricondentherm (Campbell, 1994). Many condensate fluids exhibit retrograde condensation, a phenomena in which condensation occurs during pressure reduction rather than with pressure increase, as for most gases (Firoozabadi, 1999). This abnormal or retrograde behaviour occurs in a region between the critical and the cricondentherm, bounded by the dewpond curve above and, a curve below formed by connecting the maximum temperate for each liquid volume percent. As pressures and temperatures change, mass transfer occurs continuously between the gas and the liquid phases within the phase envelope of Fig. 3.1. All attempts to describe mass transfer assume that equilibrium exists between the phases. Two approaches have been used to simulate mass transfer for hydrocarbons the "black-oil" or constant-composition model and the (variable) compositional model (Brill, 1999). Each is described in the following sections.
Black-Oil Model
The term black oil is a misnomer and refers to any liquid phase that contains dissolved gas, such as hydrocarbons produced from oil reservoirs. These oils are typically dark in colour, have gravities less than 40 API (824.97 kg/m3), and undergo relatively small changes in composition within the two-phase envelope (William and McCain, 2002). A better description of the fluid system is a constant-compositional mode. For black oils with associated gas, a simplified parameter Rs has been defined to account for gas that dissolves (condenses) or evolves (boils) from solution in the oil. This parameter, Rs can be measured in the laboratory or determined from empirical correlations. Because the black-oil model cannot predict retrograde condensation phenomena, it should not be used for temperatures approaching the critical-point temperature. A second parameter, called the oil formation volume factor Bo also has been defined to describe the shrinkage or expansion of the oil phase. Oil volume changes occur as a result of changes in dissolved gas and because of the compressibility and thermal expansion of the oil. Dissolved gas is by far the most important factor that causes volume change. Oil formation volume factor can be measured in the laboratory or predicted with empirical correlations (Brill and Mukherjee, 1999). Once the black-oil-model parameters are known, oil density and other physical properties of the two phases can be calculated. When water also is present, solution gas/water ratio, Rsw, and water formation volume factor, Bw, can be defined. Brill and Mukherjee (1999) also give correlations for these parameters and physical properties of the water. The amount of gas that can be dissolved in water and the corresponding possible changes in water volume are much smaller than for gas/oil systems (William and McCain, 2002).
Compositional Model
For volatile oils and condensate fluids, vapour-liquid equilibrium (VLE) or "flash" calculations are more accurate to describe mass transfer than black-oil-model parameters. Brill and Mukherjee (1999) provide a description of VLE calculations. Given the composition of a fluid mixture or "feed," a VLE calculation will determine the amount of the feed that exists in the vapour and liquid phases and the composition of each phase. From
these results, it is possible to determine the quality or mass fraction of gas in the mixture. Once the composition of each phase is known, it also is possible to calculate the interfacial tension and densities, enthalpies, and viscosities of each phase. Brill and Mukherjee (1999) also give methods to predict these properties. VLE calculations are considered more rigorous than black-oil model parameters to describe mass transfer. However, they also are much more difficult to perform. If a detailed composition is available for a gas/oil system, it is possible to generate black-oil parameters from VLE calculations. However, the nearly constant compositions that result for the liquid phase and the increased computation requirements make the black-oil model more attractive for non-volatile oils (Brill, 1999).
4.4. Definition and variables
When performing multiphase calculations, single-phase flow equations often are modified to account for the presence of a second phase. This involves defining mixture expressions for velocities and fluid properties that use weighting factors based on either volume or mass fraction (King, 1990). When gas and liquid flow simultaneously up a well, the higher mobility of the gas phase tends to make the gas travel faster than the liquid. This is a result of the lower density and viscosity of the gas. The slippage between both phases in defined as the ratio of the gas velocity to the liquid velocity
uG uL
K=
(4.1)
where, uG gas velocity [m/s], uL liquid velocity [m/s]. The mass fraction of flowing phases is defined as the ratio of gas mass flowrate to the total mixture flowrate:
(4.2)
where, mG gas mass flowrate [kg/s], mG liquid flowrate [kg/s]. The gas mass flowrate is related to the volume flowrates with expression
mG = u G G AG
(4.3)
(4.4)
where AG and AL are the cross sectional area occupied by gas and liquid phase respectively. Under steady state condition the slippage between both phases result in a disproportionate amount of the slower phase being present at any given location in the well. Gas void fraction can be defined as the fraction of pipe cross sectional area occupied by gas. Substitution of the equations (4.3) and (4.4) to the equation (4.2) results in the void fraction given by
x
x + K G (1 x ) L
(4.5)
The opposite value to the gas void fraction is the liquid holdup defined similar way as the cross section area occupied by liquid or volume increment that is occupied by the liquid phase
H L = (1 )
(4.6)
The gas void fraction and liquid holdup can be distinguished in horizontally oriented pipes where stratification occurs due to gravity. In vertical wellbore two-phase turbulent flow under
high velocities, both phases may be considered as a homogenous mixture (King, 1990). Two phases may be assumed to flow at the same mixture velocity with no slippage between.
4.5. Fluid properties
A numerous equations have been proposed to describe the physical properties of gas/liquid mixtures. The following expression has been used to calculate in multi-phase flow mixture density
M = G + (1 ) L
(4.7)
The two phase viscosity is the property expressed per mass unit and thus was calculated from equation
1 x
(4.8)
When performing the temperature change calculations for multi-phase flow in geothermal wells, it is necessary to predict the enthalpy of the multiphase mixture. Also most VLE calculation method for oil wells includes a provision to predict the enthalpies of the gas and liquid phases. Enthalpy of the mixture was calculated from equation
ht = x hG + (1 x ) hL
4.6. Flow patterns
(4.9)
Prediction the flow pattern that occurs at a given location a well is extremely important. The empirical correlations or mechanic model used to predict flow behaviour varies with flow pattern (Gomez, 2001). Essentially all flow pattern predictions are based on data from lowpressure systems, with negligible mass transfer between the phases and with a single liquid phase (Brill, 1999). Consequently, these predictions may be inadequate for high-pressure,
high production-rates, evidently high-temperature geothermal wells, or for wells producing oil and water or crude oils with foaming tendencies, respectively (Manabe et al., 2001), (Gudmundsson and Ambastha, 1984), (Aggour, 1996). A consensus exists on how to classify flow patterns (Brill, 1999). For upward multi-phase flow of gas and liquid, most investigators now recognize the existence of four flow patterns: bubble flow, slug flow, churn flow, and annular flow. These flow patterns, shown schematically in Fig. (4.2) and Figure (4.3) are described next. Slug and churn flow are sometimes combined into a flow pattern called intermittent flow. It is common to introduce a transition between slug flow and annular flow that incorporates churn flow. Some investigators have named annular flow as mist or annular-mist flow. Flow in vertical and horizontal or inclined pipes exhibits different behaviour. The distribution of the multiphase contents across the pipe in vertical flow regimes is randomly chaotic, and the phases show no preferences for the one side of the pipe or another. The exception to the random distribution is annular flow where at very high flow rates gas occupies the centre of the pipe. There may be large discontinuities that pass along the vertical pipe or wellbore, as when gas flows much faster than liquid in slug and churn flow regime. In non vertical flow random distribution of the phases across the pipe is replaced by gravity segregation by the phases.
Bubble Flow
bubbles in a continuous liquid phase. Based on the presence or absence of slippage between the two phases, bubble flow is further classified into bubbly and dispersed - bubble flows. In bubbly flow, relatively fewer and larger bubbles move faster than the liquid phase because of slippage. In dispersed bubble flow, numerous tiny bubbles are transported by the liquid phase, causing no relative motion between the two phases.
Slug Flow
gas pocket a plug of liquid called a slug, and a film of liquid around the bubble flowing downward relative to the Taylor bubble. The Taylor bubble is an axially symmetrical, bullet-
shaped gas pocket that occupies almost the entire cross-sectional area of the pipe. The liquid slug, carrying distributed gas bubbles, bridges the pipe and separates two consecutive Taylor bubbles.
Churn Flow
Churn flow is a chaotic flow of gas and liquid in e which the shape of both the
Taylor bubbles and the liquid slugs are distorted. Neither phase appears to be continuous. The continuity of the liquid in the slug is repeatedly destroyed by a high local gas concentration. An oscillatory or alternating direction of motion in the liquid phase is typical of churn flow.
Annular Flow Annular flow is characterized by the axial continuity of the gas phase in a
central core with the liquid flowing upward, both as a thin film along the pipe wall and as dispersed droplets in the core. At high gas flow rates more liquid becomes dispersed in the core, leaving a very thin liquid film flowing along the wall. The interfacial shear stress acting at the core/film interface and the amount of entrained liquid in the core are important parameters in annular flow.
The pressure gradient equation for multi-phase flow can be modified from single-phase flow. Considering the fluids to be a homogenous mixture the equation may be written
du dp f M u M = + M g sin + M u M M dL dL 2d
2
(4.10)
For vertical flow = 90o, dL =dz and the equation for pressure gradient can be written as
dp dp dp dp = + + dz dz f dz el dz acc
(4.11)
The pressure-gradient equation for single-phase flow in pipes was developed by use of the principles of conservation of mass and linear momentum. The same principles are used to calculate pressure gradient for multiphase flow in pipes. However, the presence of an additional phase makes the development much more complicated. The pressure-drop component caused by friction loses requires evaluation of two phase friction factor. The pressure drop caused by elevation change depends on the density of the two phase mixture which may be calculated from equation (4.7). The pressure drop caused by acceleration component in normally negligible as is considered only for cases of very high flow velocities (King 1990).
4.8. Multiphase flow models
Early investigators treated multiphase flow as a homogeneous mixture of gas and liquid. This approach did not recognize that gas normally flows faster than liquid. The no slip approach tended to under predict pressure drop because the volume of liquid predicted to exist in the well was too small (Brill, 1999) Improvements to the no-slip methods used empirical liquid holdup correlations to account for slippage between the phases. Although liquid holdup and friction effects were often dependent on the flow pattern predicted by empirical flow-pattern maps, in general these methods still treated the fluids as homogeneous mixture (Falk 1999). Treating the fluids as a homogeneous mixture is often unrealistic, resulting in poor predictions of flow behaviour (Brill, 1999). A trend to improve flow-behaviour predictions has emerged that is a compromise between the empirical correlations and the two-fluid approach. The methods used to predict pressure gradient can be classified as empirical correlations and mechanistic models (Gomez, 1999). The empirical ones are based on experimental data, and are suitable for preceding steady-state flow. The mechanistic multiphase flow models include the two-fluid model, the drift-flux model and the homogenous model (Manabe et al., 2001). These can be developed from physical relationship with mass, momentum and energy conservation of each phase resulting in local, instantaneous equations. The mechanistic modelling approach still requires use of some
empiricism, but only to predict specific flow mechanisms or closure relationships (Gomez, 1999). The conservation laws are connected by interaction laws between the phases and between the fluid and wall. A popular approach is to average the conservation equations over the cross sectional area to get a one-dimensional model. The mechanistic models differ from each in how they implement the conservation laws (Falk, 1999). The two-fluid method uses one conservation equation for each phase, the drift-flux method uses the sum of the momentum equations in addition to energy and mass conservation for each phase, while the homogenous flow model uses only the sum of all phases for each conservation law. The homogenous flow model is a simplification, assuming the same flow velocity for all phases (Gould, 1970), thus needs neither interfacial friction nor drift flux terms. The empirical correlations can be placed in one of three categories (Brill,1999): I category - no slip, no flow pattern consideration. The mixture density is calculated based on the input gas/liquid ratio. That is, the gas and liquid are assumed to travel at the same velocity. The only a correlation required is for the two-phase friction factor. No distinction is made for different flow patterns. II category - slip considered, no flow pattern considered. A correlation is required for both liquid holdup and friction factor. Because the liquid and gas can travel at different velocities, a method must be improved to predict the portion of the pipe occupied by liquid at any location. The same correlations used for liquid holdup and friction factor are used for all flow patterns. III category - slip considered, flow pattern considered. Not only are correlations required to predict liquid holdup and friction factor, but methods to predict which flow pattern exists are necessary. Once the flow pattern is established, the appropriate holdup and friction factor correlations are determined. The method used to calculate the acceleration pressure gradient also depends on flow pattern.
The following list presented on Figure (4.3) gives the published empirical correlations for vertical upward flow and the categories in which they belong.
Method Poettmann and Carpenter Baxendell and Thomas Fancher and Brown Hagedorn and Brown Gray Asheim Duns and Ros Orkiszewski Aziz Chierici Beggs and Brill Mukherjee and Brill
Figure 4.3 Published Vertical Flow Correlations Categories The following sections of this chapter present method to predict pressure gradients and presents the methods applied for calculations that were performed. In present work Duns and Ros, 1963 for oil wells and also Orkiszewski, 1967 for geothermal well methods was applied. Those authors summarized numerous investigations that have described flow patterns in wells and made attempts to predict when occur. Both correlations were verified many times
since the time that was developed by other authors and by the industry elaborating the software commercially available. Those methods are recommended for vertical wells calculations and contain modifications that improve accuracy. (PipeSim Manual), (Gudmundsson and Oritz, 1984) (Uphady, 1977)
4.9. Duns and Ros correlation for multiphase flow in oil wells
Duns and Ros method was chosen for oil wells calculations performed in present work. This
method is ranged to III group from Figure (4.3), which is assumed to give the most appropriate issues. The method is a result of an extensive laboratory study in which liquid holdup and pressure gradients were measured. About 4,000 two-phase-flow tests were conducted in a 185-ft (56,39m)-high vertical-flow loop. Pipe diameters ranged from 1.26 to 5.60 in. and included two annulus configurations. Most of the tests were at near-atmospheric conditions with air for the gas phase and liquid hydrocarbons or water as the liquid phase. Liquid holdup was measured by use of a radioactive-tracer technique. A transparent section permitted the observation of flow pattern. For each of three flow patterns observed, correlations were developed for friction factor and slip velocity, from which liquid holdup can be calculated. Duns and Ros performed the first dimensional analysis of two-phase flow in pipes. They identified 12 variables that were potentially important in the prediction of pressure gradient. Performing a dimensional analysis of these variables resulted in nine dimensionless groups. Through a process of elimination, four of the groups were identified as being important and were used to select the range of variables in the experimental program. Equations presented in Appendix B for this method gives those four groups.
4.10. Duns and Ros modifications
Two proprietary modifications of the Duns and Ros method have been developed but are not available in the literature. The first, known as the Ros field method, involved modifications based on carefully obtained data from 17 high-GOR vertical oil wells. In a joint Mobil-Shell study undertaken between 1974 and 1976, a modification resulted in the Moreland-MobilShell method (MMSM) (PipeSim Manual). In this study, 40 vertical oil wells, including the
17 used in the Ros field method, and 21 directional wells were selected as the basis for the modifications. The MMSM method includes liquid- holdup correlations derived from the data for bubble and slug flow that are simpler in form than those used in the original Duns and Ros method (Brill, 1999). Possible discontinuities at flow-pattern-transition boundaries also were removed.
4.11. Orkiszewski correlation for multiphase flow in geothermal wells
For Geothermal Two-phase flow calculations Orkiszewski correlation was used. This method was recommended by Uphadhay and Hartz (1977). Their work contains comparison of calculated and observed flowing pressure profiles for geothermal wells located in the United States and Philippines. Comparisons were included for tubular flow as well as flow trough the casing-tubing annulus. Their work revealed that for tubular flow, the Orkiszewski correlation makes the best prediction, whereas for annular flow, no clear choice of correlation can be made. Similar work has been done by Ambastha and Gudmundsson (1986), they measured the flowing pressure profile data from many geothermal wells around the world, covering a vide range of flowrate, fluid enthalpy and wellhead pressures. The authors reported a good accuracy of Orkiszewski correlation in estimating the downhole conditions. The capability to accurately predict flowing pressures in geothermal wells producing steam and water a mixture under various operating conditions is of value importance for several reasons similar to petroleum industry. These are general evaluation of the geothermal reservoir under proper reservoir management, optimalisation of the wellbore design from well deliverability considerations and minimization of scale deposits in the wellbore (Ragnarsson, 2000). The predictive capability is especially important because of the difficulty of running flowing pressure surveys in geothermal wells. These wells are characterized by very high fluid velocities, which sometimes makes impractical for pressure recorders to traverse downward in the well. There have been the cases of pressure recorders thrown out of the wellbore due to fluid velocities (Uphadhay, 1977). In calculating flowing pressure profiles in oil wells, phase transfer between oil and gas
requires a rather simple treatment, and is accomplished trough the use of solution gas-oil ratio Rs relationship. In geothermal wells, however phase transfer between water and steam attains critical importance and calculations must incorporate the steam tables accurately. Pressure profile calculations for geothermal wells vary from those for oil well in another important aspect in that the temperature of the fluid must be computed precisely. Orkiszewski tested several published correlations with field data and concluded that none was sufficiently accurate for all flow patterns (Orkiszewski, 1967) He then selected what he considered to be most accurate correlations for bubble and mist flow and proposed a new correlation for slug flow. Orkiszewski used the Duns and Ros flow-pattern transition for the boundaries between slug flow and mist flow, including the transition region between them. Equation defined these are given in Appendix B together with Duns and Ros correlation description. For the boundary between bubbly flows, he chose these criteria established by Griffith and Wallis.
Chapter 5
Speed of Sound in Two-Phase Mixtures
A book on physical acoustic (Trusler, 1991), defined sound as infinitesimal pressure waves propagating trough a medium with a characteristic speed; the velocity of sound depending on media. In present work a pressure wave caused by a rapid valve closure travels at the velocity of sound trough the two phase mixture. This velocity is dependant on the compressibility and densities of both phases. These flowing phases may have different structures; usually it is of the gas bubbles or slug surrounded by the liquid phase. The gas present in liquid phase cause a marked increase of damping. The examination of this phenomena showed that this damping is due to increase of distortion of the liquid separating the bubbles. The pressure variations act almost entirely on the volume of gas and scarcely at relative incompressible liquid (Firoozabadi, 1999)
5.2. Compressibility of two-phase mixtures
The velocity of sound is defined as the square root of the derivative of pressure with respect to the density at constant entropy (Henry et al. 1977)
p a= S
(5.1)
The second law of the thermodynamics tells that process must be isentropic due to there is no temperature gradient except inside the wave itself. Therefore instead of differentiate the density, the sonic velocity can be related to the properties of the fluid. Using isentropic compressibility Ks the speed of sound formula can be written
1 KS
a2 =
(5.2)
(5.3)
where v is the specific volume [m3/kg] and n is the composition vector, which is defined by n = (n1, n2, n3, , nc), where c is the total number of components and ni is the number of moles of each component i of the mixture. The expression for the two-phase gas liquid mixture compressibility will be 1 v M K S M = v M p S ,n
(5.4)
vM is the total specific volume of the gas liquid phases in the mixture. Similarly, the isothermal compressibility of a two-phase multicomponent system is defined by 1 v M K T M = v M p T ,n
(5.5)
The isothermal compressibility represents the volume change caused by small change in the pressure of the closed system at constant temperature. The isothermal and isentropic compressibility in the single phase state are related by a simple expression (Firoozabadi, 1999) KT = Cp CV KS (5.6)
Where Cp and CV are the heat capacities [kJ/kgK] at constant pressure and constant volume, respectively. The derivation of (5.6) can be found in (Firoozabadi, 1999).
Since Cp Cv, then KT KS. In reservoir engineering applications isothermal compressibility is often used to describe the fluid compressibility away from the wellbore. Inside the well, due to expansion, the fluid may undergo heating or cooling and process may become nonisothermal. If the heat loss can be neglected, the isentropic compressibility may better represent the pressure and volume changes. Practicing engineers in order to obtain the two-phase compressibility often use the following relationship (Firoozabadi, 1999)
KT M = KT G SG + KT L SL
(5.7)
where SG and SL are the volumetric fractions of the gas and liquid components, respectively. The equation is invalid where there is mass transfer between the phases (Firoozabadi, 1999) what is of vital importance for water and its vapor system discussed in further section.
5.3. Compressibility of steam-water system
Calculations of the sound speed in two-components two-phase systems is an easy procedure if adiabatic equation of state data are available to calculate compressibility, because pressure and temperature may be considered to be independent variables in such systems. Calculation of the sound speed in one-component two-phase system is more difficult matter because the pressure and temperature are not independent variables and are related by the Gibbs equation for equilibrium between both phases (Kieffer, 1977). The complex physical process which occurs during propagation of sound wave in watervapor two-phase system is given in Appendix C. Propagation of the sound wave in the fluids is accomplished by compression or rarefaction. If steam-water system remains in thermal equilibrium where variables characterizing the system may be considered to follow the saturation line, there must be mass transfer between the phases, since the fraction of steam in mixture changes due to evaporation or condensation process. When the pressure wave passes the adiabatic compression causes pressure increase in both phases; as a result the water
becomes subcooled, and the steam becomes superheated. The induced temperature difference between the steam ad water phases leads to the heat transfer from superheated steam to subcooled water (Kieffer, 1977). Depending on the original phase composition, whether steam or water was the dominant phase heat transfer will cause water evaporation or steam condensation. These heat and mass transfer cause that both water and steam are restored to the saturation line. Similar process takes place in cause of pressure reduction. In general consideration the evaporation and condensation can not take place instantaneously, since transportation of heat and mass can only occur at a finite speed. The time period in flashing water to steam or condensing steam to water is important in determination the degree of equilibrium obtained in the sound wave. Since condensation and evaporation generally proceed at different rates, it should be expected that compression and rarefaction waves behave differently. Experiments have confirmed this theory that finite amplitude rarefaction waves in steam water mixtures have lower velocities than compression waves because rarefaction waves tend to maintain continuous equilibrium (McWilliam and Duggins, 1969). A mixture of liquid and its vapor may respond to pressure disturbances by equilibrium and nonequilibrium state. Nearly high frequencies waves the process of pressure wave propagation follows fast and may be considered adiabatic where there is no equilibrium between the phases. In this case the mass transfer can be neglected and thus calculations of speed of sound are greatly simplified. The pressure pulse waves are low frequency (Gudmundsson and Celius, 1999; Falk 1999) thus equilibrium response needs to be considered and the mass transfer between the liquid and its vapor occurs in a time short with comparison to the acoustic wave period. Due to complex physical process related to the acoustic wave propagation in one-component system different approach than for two-component need to be considered in order to find steam-water compressibility. Experimental data on sonic velocity in steam-water system at geothermal wellbore conditions are not readily available. To solve this problem theoretical calculation need to be carried out to find the compressibility of this system for ideal thermodynamics conditions. The method to derive the compressibility of a steam-water mixture in contact with reservoir rock was elaborated by Grant and Sorey (1979). Similar
approach was adapted to geothermal wells by Gudmundsson et al., (1998). The studies below are based on the work of these authors. Two-phase fluid in geothermal well can be considered to be in thermodynamics equilibrium where the flow is homogenous, steady-state and one dimensional. Assuming adiabatic conditions no heat loss and gain a balance equation can be written for steam-water mixture flowing from one infinitesimal cross section to another (Gudmundsson et al., 1998) x1 hG 1 + (1 x1 ) hL 1 = x 2 hG 2 + (1 x 2 ) hL 2 (5.8)
where x represents the mass fraction of the steam, and hG and hL [kJ/kg] are the enthalpy of steam and vapor and liquid water, respectively. The equation can also be written as hL 1 + x1 hLG 1 = hL 2 + x 2 hLG 2 (5.9)
the hLG [kJ/kg] is the latent heat of vaporization which may be assumed to change negligible between adjacent cross-sections; from one infinitesimal cross-section to another. Thus the equation (5.17) can be rewrite in form hL 1 + hL 2 = (x 2 x1 ) hLG (5.10)
Since for liquid water a change in enthalpy is equal to the heat addition at constant pressure h Cp = T p
(5.11)
(5.12)
When pressure is lowering the dominant volume change is that caused by phase change; flashing liquid into steam vapor. As water before changed phase has occupied a volume mvW, after that occupies larger volume being in gas phase mvG. This increase of volume can be written as V = m (vG v L ) The mass of water that changed phase is simply
m = ( x1 x 2 ) m
(5.13)
(5.14)
As shown in previous section the compressibility may be defined in several ways depending on what physical property is assumed to be constant, temperature, enthalpy or entropy. The question arises what conditions may be found as the fluid flows in the wellbore. Usually as defining compressibility the temperature is assumed constant, giving an isothermal compressibility. For two-phase wellbore flow without heat loss or gain adiabatic process seems to be most appropriate. From thermodynamics we know that an isentropic process is both adiabatic and reversible. The flow in pipes, pipelines and wells, frictional pressure loss makes the process non-reversible. This aspect of fluid flow is particularly important in situation where rapid pressure drop occur for example in nozzles (Watters, 1978). It may be considered less important in situations where the pressure changes gradually with distance for example if wellbores and long pipelines. Since the sonic velocity of steam-vapor flowing in the well may be approximated to isentropic conditions for which the acoustic velocity is defined in equation (5.1).
Taking the compressibility from equation (5.3), and eliminating (x2-x1) from equation (5.23) by substituting equation (5.20), the compressibility may be written in form (Gudmundsson et al., 1998) 1 m C p L T (vG v L ) VM p hLG
Ks =
(5.16)
At all condition fluid maintain equilibrium between the phases and follows the saturation line thus
p p = T T sat
(5.17)
Ks =
(5.18)
p may be calculated using steam tables what bring considerable inconvenience in T sat
(5.19)
Compressibility expressed in this form is convenient for numerical calculations. The thermodynamics properties values can be obtained from wellbore simulator output file and then sonic velocity can be calculated.
5.4 Acoustic velocity models
The formulas derived for acoustic velocity base on experiments performed and varies with respect to the components that was applied and flow patterns encountered during the experiments. The flow regimes are described in chapter 4 of this work which deals with multiphase flow in wells. Also the fact is important, whether homogeneity can be assumed or not. For homogenous flow the slippage between the phases may be neglected. The study of different models that was made here has the purpose to find the most appropriate model for sonic velocity in one component steam-water mixture and two component gas-oil mixtures. In present thesis acoustic velocity in steam-water mixture was calculated from Wood (1941) equation. He derived the equation for the velocity of sound in a homogenous two component media, based on air-water experiments (Wood, 1944). The author proposes instead of differentiating the density of liquid mixture as shown in (5.1) relate the sonic velocity directly to the properties of the gases and liquids. This model assumes that overall compressibility of the mixture is related to the compressibility of the constituents by the relation K S M = K S G + K S L (1 ) (5.20)
aWood =
G + (1 ) L G aG 2
+ 1
(5.21)
L aL 2
where is the void fraction and is the density, with the subscripts G and L indicating the gas and liquid, respectively. Nakoryakov et al. (1993) provided that Woods equation can be derived from equation (5.1) assuming the mass fraction of each phase reminded constant what is not such a good assumption for steam-water system. Another theoretical limiting case for vapor-liquid had the changes in density exclusively included by variation in the gas-mass fraction due to flashing. In experiments however practically nobody has manage to observe disturbances propagating with this flashing velocity (Falk, 1999). Semenow and Kostern (1964) found experimentally sonic velocity in steam water flow agreeing well with Woods estimate, and Noryakov et al. (1993) claimed that acoustic of Wood was the low frequency limit for bubbly flow when the wave process were isothermal. For high frequencies the wave process was adiabatic without energy transfer leading to the frozen velocity not treated here. (Nakoryakov et al., 1993). The dependence of propagation velocity on flow regimes demonstrates analytical expression derived by Henry for bubbly flow using a slip flow model (Falk, 1999). 1
a Henry =
(1 (1 K ))
K G + (1 ) L
1 K + 2 G aG L aL 2
(5.22)
The author claimed that interfacial momentum transfer exhibited a strong influence on the propagation velocity, and that different flow structures would have different interfacial drag during the passage of the wave. Assuming homogenous flow with sleep K=1, equation (5.29) reduces to Woods equation (5.28). However formula for acoustic velocity in slug flow overestimates the results giving velocities close to the velocity of sound in gas phase (Falk, 1999). As the slug flow is mostly encountered flow pattern in flowing geothermal wells, this model may bring the wrong results. There are a few examples of analytical models for sonic velocity in gas liquid mixtures in literature (McWilliam and Duggins, 1969; Kiefer, 1977; Firoozabadi 2000). These models
are based on equation (5.1) and relate the density of the mixture to the densities gases and liquids. To make the differentiation in equation (5.1) possible, the ideal gas law was employed for the gas and adiabatic state equation for the liquid. The derived expression for the density of mixture was then differentiated with respect of the pressure. Henry et al. (1974) in the book of thermodynamics properties of hydrothermal systems showed that steam express the significant deviation from the perfect gas under high pressure conditions. Such high pressures may be expected in geothermal wells treated in this work, thus these models also do not occur to be the most applicable. It is well known that properties of the natural gas also deviate from perfect gas for high pressures, thus from the same purpose acoustic velocity in oil - gas mixture was calculated using the formula developed by Gudmundsson and Dong (1993). The authors similarly to Wood relate the sonic velocity directly to the properties of the gases and liquids (Gudmundsson and Dong, 1993). Their formula presented below, was developed for gas-oil mixture and assumes that liquid phase may contain water. The thermodynamics compressibility and densities of mixture components are related assuming the phases are homogenously distributed in the liquid. From equation (4.7) the density of gas/liquid mixture is given by
M = G + (1 ) L
(5.23)
where is the gas/liquid void fraction, and the subscripts G and L stand for gas and liquid, respectively. If the liquid phase contains both water and oil, the density of the liquid phase can be similarly obtained from
L = W + (1 ) O
(5.24)
where is the water-oil volumetric fraction and the subscripts W and O stands for water and oil, respectively.
M = G + (1 ) [ W + (1 ) O ]
The relationship (5.7) can be written for: gas-oil-water mixture as
(5.25)
(5.26)
where the S volumetric fraction of the component is in fact the void fraction and SO and SW is the liquid holdup (1-). Thus (5.26) can be alternatively written as
KT M = KT G + (1 )KT L
(5.27)
If the liquid contains water and oil, its isothermal compressibility can be expressed as K T L = K T W + (1 ) K T O (5.28)
For gas-oil-water mixture, the heat capacities at constant temperature and constant volume may be expressed as
(5.29)
(5.30)
Where x is the gas-liquid mass fraction and y is the water-oil mass fraction. It should be noted that mass fraction instead void fraction is employed in these equations because the ratio of specific heats is a property based on mass unit (Gudmundsson, 1993). From equations (5.2) and (5.6) speed of sound for mixture is given as
a2 =
M KT M
(5.31)
(5.32)
Substituting (5.25), (5.26) and (5.32) to equation (5.31), speed of sound for gas-oil-water mixture can be written as
a=
[x C
[ G + (1 ) ( W
pG
+ (1 x ) y C p W + (1 y ) C p O /[x CV G + (1 x ) ( y CV W + (1 y ) CV O )] + (1 ) O )] [ K T G + (1 )( K T W + (1 )K T O )]
)]
(5.33)
5.5 Attenuation mechanisms of sound wave
When sound spreads trough the media there are a diminution mechanisms of intensity or attenuation as the distance from source increase due to loses of mostly frictional character. All media capable of transmitting sound, are limited in extend and sooner or later the wave must stop or change from a medium to another. Thermal conductivity and shear viscosity attenuate the wave motion. Both the speed and attenuation of the sound waves depend on the frequency. General considerations say that pressure waves will pass through structures if they are of the frequency for which the
wavelength is larger than the structure. If the structure is a bubble or a slug in gas-liquid flow, a pressure pulse needs to have a wavelength greater than the size of the bubble and slug to propagate through the gas-liquid flow (Gudmundsson and Celius 1999). In multiphase flow there are additional attenuation mechanisms (Falk, 1999) which include: viscous drag steady interfacial drag added mass (interior effect) boundary layer around a particle/drop (Basset force) interfacial heat exchange compressibility of each phase concentration gradient effects phase transitions deformation and fragmentation of bubbles and drops reflections at the interface
Knowledge of which of these attenuation mechanisms are the most important can greatly simplify the models for sound wave propagation. The relative importance, however, depends on the media, flow pattern and the frequency. In approach towards zero frequency where all processes take place slowly, the compression and expansion of the fluid occur reversibly and adiabatically. Trusler (1991) claimed that boundary-layer absorption was the most important attenuation mechanism for low-frequency waves in tubes with single phase flow. However, at high frequencies the fluid cannot maintain local equilibrium. Consequently, some of the energy is dissipated and high frequency waves will be highly attenuated by reflections of the interfaces and can not propagate very far in bubbly and slug flow (Falk, 1999). The dominant frequency in lowpressure air-water flow in pipelines has been shown to be in the range 1-10 Hz (Dong and Gudmundsson 1993). Similar results were reported by Falk (1998). Therefore, pressure
waves in gas-liquid flow are infrasonic (lower frequency than audible sound which is about 20,000Hz). Numerical calculations in bubbly flow showed that the slippage between the phases has little influence on the wave structure (Falk, 1999), though it can only be neglected when the liquid viscosity is considerable higher than of water, or if the bubbles are small. In bubbly flow the main attenuation mechanism was heat exchange between the gas bubbles and surrounding liquid. Kiefer (1977) showed addition theory that added liquid mass is important for the pressure wave propagation. Calculations of speed of sound in bubbly flow (McWilliam and Duggins, 1970) and (Kiefer, 1977) showed that surface tension is important for small bubbles, while at large bubbles only liquid compressibility is important. Firoozabadi (2000) also revealed that that capillary pressure may affect the two-phase compressibility only in porous media at reservoir conditions outside the wellbore where phases interface is curved.
5.6 Concluding remarks
1. The acoustic velocity in steam-water two phase systems is more complicated than for twocomponent systems. In two-component system gas being dissolved under high pressure comes out of solution as the pressure decrease and is characterized by solution gas-oil ratio Rs. In one component system gas emerges due to evaporation process and if the pressure change is not of the high frequency the equilibrium is maintained between the phases. As the system responses with equilibrium for pressure change the temperature effects and mass transfer between both phases is essential for calculations. 2. The study of models made in this chapter showed that there is not one best model for acoustic velocity in one component steam-water mixture. The Wood equation was chosen for calculations. The choice of this model is mainly due to Semenow and Kostern (1964) experiments that showed a sound velocity in steam water flow agreeing well with those estimated from Woods model. However the other authors show the limitation of this model in many areas. The speed of sound in oil gas mixtures was calculated from formula
Developed by Gudmundsson and Dong (1993). This formula is reported by authors to be in good agreement with measured data. Both chosen models require the properties of each constituent phase to be known. The procedure necessary in order to calculate these properties was fully described in this chapter for both gas-oil and steam-water cases. For oil and gas two-phase flow thermodynamics compressibility can be readily obtained from simulations performed. Unfortunately software available for geothermal simulations is not such sophisticated as those available for petroleum applications and thus steam compressibility need to be calculated from equation derived in this chapter. 3. The acoustic velocity in liquid single-phase is higher from acoustic velocity in pure gas phase due to significantly lower compressibility of liquid phase. At the presence of only one percent by volume gas in the form of gas bubbles the acoustic velocity decreases dramatically and two-phase system presents different character from each of the constituent phases. The explanation is the fact that such two-phase system has density of liquid and compressibility of gas. The calculated acoustic velocities for gas-oil and steam-water mixtures for wide range of void fraction values are presented on Figure 5.1 and Figure 5.2. The plots also shows that depression in the sonic velocity is less if the pressure is increased while retaining the same void fraction, although the sensitivity to change in pressure becomes progressively diminished. 4. The plots of acoustic velocity vs. void fraction for oil-gas and steam-water systems revealed different behavior for high void fraction values. For hydrocarbons two-phase fluids mixture the plot has less slope which becomes more flat as the pressure is higher. For steamwater mixture this shape is more inclined towards vertical. The calculations made showed out that this effect may be important for lower pressure values where water dryness has higher values and steam occupies main volume of the pipe. This high value void fraction changes has more effect on acoustic velocity than pressure changes and thus acoustic velocity decrease with pressure increase across the wellbore. This problem will be presented precisely together with calculations made in Chapter 8.
5. Gudmundsson and Dong model relate the sonic velocity directly to the properties of the gases and liquids in the same manner as Woods model. In fact assuming that mixture do no contain water and = 0, the equation may be expressed in the same form as Woods. The difference is in the approach towards the thermodynamic process that occurs during the sound propagation. Compressibility is often defined as the small volume change than occur in closed system at constant temperature. The second law of the thermodynamics tells that sound propagation process must be isentropic due to there is no temperature gradient except the wave itself. Woods proposed to calculate speed of sound separately for both phases from isentropic compressibility. Gudmundsson and Dong equation use the isothermal compressibility and then transform the equation to the isotropic condition using the specific heats ratio as shown in equation (5.31). The possible error due to assuming isothermal process instead isentropic is up to 7% for gas-oil mixtures and even up to 14% for steam-water depending on the pressure. Table 5.1 contains calculated results for pressure equal 52.4 bar, for gas-oil mixture and Table 5.2 shows similar calculations for steam-water mixture at the 45.1 bar. The differences between calculated sound speed values assuming whether isothermal or isentropic process are also shown in form of plots by Figure 5.3 and 5.4 for gas-oil and steam-water respectively.
1200 1100 1000 900 acoustic velocity [m/s] 800 700 600 500 400 300 200 100 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0.9
void fraction
Figure 5.1 Calculated acoustic velocities Vs void fraction for oil-gas mixture
1200 1100 1000 900 acoustic velocity [m/s] 800 700 600 500 400 300 200 100 0 0 0.1 0.2 0.3 0.4 0.5 void fraction 0.6 0.7 0.8
0.9
Figure 5.2 Calculated acoustic velocities Vs void fraction for steam-water mixture
800
700
500
400
300
200
100
0 0 0.1 0.2 0.3 0.4 0.5 void fraction 0.6 0.7 0.8 0.9 1
Figure 5.3 Calculated acoustic velocity for gas-oil mixture assuming isothermal and isentropic process of sound propagation
800
700
500
400
300
200
100
0 0 0.1 0.2 0.3 0.4 0.5 void fraction 0.6 0.7 0.8 0.9 1
Figure 5.4 Calculated acoustic velocity for steam-water mixture assuming isothermal and isentropic process of sound propagation
Chapter 6
Case Studies
The purpose of these calculations is to estimate water-hammer and line packing effects when valve has been fully closed. The results from these calculations should give a picture of these phenomena in two-phase wells. Then similar calculations for oil and geothermal wells may be compared witch each other. A description of the effect should give a result which can be linked to field measurements to see if any other factors might affect the pressure gradient after the valve closure. Offshore Pressure Pulse tests have been made in several North Sea wells to validate the theoretical simulation. The tests issues were described in two confidential reports and then published (Gudmundsson and Falk, 1999), (Gudmundsson and Celius, 1999). Geothermal wells are planed to be tested for Pressure Pulse method in summer 2003. The water-hammer effect is caused by rapid closure a valve in flowing pipe. The pressure increase is dependant on fluid density, velocity and sonic velocity in the flowing well as shown in Chapter 3. These will be calculated here. In liquid flow, line packing is the increase in pressure caused by a wall friction when a valve is closed and the fluid is stopped by a pressure wave which is emitted from the valve in the instant of closure. In liquid/gas twophase flow, the increase in pressure is the sum of pressure drop due to wall friction and the pressure drop due to interfacial friction. The interfacial friction is an unknown quantity, while the frictional pressure drop can be considered a known quantity.
6. Case Studies oil wells 76 __________________________________________________________________________ 6.2 Water-hammer and line packing in oil wells
The three programs used were PipeSim 2000, GOW 3.0 and Excel. PipeSim is a multiphase flow simulator. The futures and modes of the program used in present work are described in Appendix D. PipeSim simulates the flow of an oil/gas/water mixture in a well or pipeline. The program takes the well fluid data and uses them to calculate various properties of the fluid and system. PipeSim was also used for inflow performance (IP) and tubing performance (VLP) in order to estimate the flowrates for different wellhead pressures. The GOW program is published by Gulf Publishing Co. Houston, Texas. It allows calculating the various parameters of many substances including oil, gas and water. The PipeSim output file does not give the values of oil and gas compressibility and these were found from this program for given PVT conditions taken from well profile. The Excel spreadsheet was used in present work for additional calculations of acoustic velocity. The acoustic velocity was calculated using formula reported by Dong and Gudmundsson (1993) and given in equation (5.40) in this work. Excel was also used to plot the results. The above programs may be substituted by similar programs. Prosper Multiphase Flow Simulator, designed by Petroleum Experts Ltd. Edinburgh gives similar output files suitable to present calculations. Only the PipeSim was available at universitys computers thus calculations placed in this thesis work hail from PipeSim. Excel may be replaced by almost any spreadsheet that can plot and calculate. Also GOW has equivalent programs. To run PipeSim the fluid, completion and production data are necessary. The data used in these simulations for one well was obtained from work of Jonsson (1995). This contains the data about production, completion and fluid properties from the Draugen oil field. Two other wells were simulated based on data from work of Falk (1999). These data were gained during the Pressure Pulse Method tests on Gullfaks and Oseberg platforms. In addition some
reference values for these oil fields were taken from the Skjveland and Kleppe Monograph (1992) that contains the characteristics of the most North Sea fields. PipeSim was used to determine the properties of the oil and gas at different depths in the producing well. Simulations gave the values at 50 meter intervals. Oil pressure, temperature, density, void fraction, gas and oil heat capacities and compressibility are the properties required to calculate the acoustic velocity. The last two were computed in GOW and then entered into the Excel spreadsheet. The void fraction is also not given directly in PipeSim output file but as water cut is 0 (assumed in all simulations done), the void fraction and the liquid holdup add to one. Excel was used to plot the results. The depth, acoustic velocity and the pressure drop due to friction were entered into spreadsheet. The frictional pressure drop is available on the PipeSim output file. These values are then used to calculate travelling time for a pressure pulse down to a certain point and up again. Calculated results were plotted against each other. Depth and time are plotted on the x axis and total pressure drop and acoustic velocity on the y axis. The total pressure drop versus time gives the effect of line packing. PipeSim gives the output file compatible with excel format thus the results are presented in two forms; Excel spreadsheet and Excel plots. Excel spreadsheet contains the calculated values of various properties affecting the acoustic velocity and line packing listed in columns and is available from all simulated oil wells in Appendix F of this work. These values were directly used to plot the results. Table 6.1 contains the data about the wells that was chosen to illustrate the pressure pulse method in oil wells. These data include well depth, tubing inner diameters, SCSSV depth and inner diameters, reservoirs pressures and temperatures necessary to run simulations in PipeSim. Table 6.2 contains the molecular composition of the well fluids used in simulations. PipeSim base on the reservoir data to predicts the well flowing pressure at the bottom of the well and then simulate the flow in vertical well. Table 6.3 presents the calculated parameters. The wellhead pressures, temperatures and flowrates are presented in the table together with parameters that make up water-hammer: density, acoustic velocity and flow velocity.
Table 6.1 Well geometry and reservoir data for computer simulations
WELL SYMBOL
Depth H (m) Tubing ID (inch) SCSSV depth (m) SCSSV ID (inch) Reservoir pressure pres (bar) Reservoir temperature Tres (oC)
Table 6.2 Molecular composition of the well fluids (measured as a mole fraction of the gas phase at stock-tank conditions)
WELL SYMBOL
A1
A2
B 0.307 0.996 4.045 0.881 0.556 0.511 0.662 0.293 1.014 2.869 4.064 3.257
C 0.320 0.620 3.83 0.94 0.57 0.56 0.6 0.28 0.92 2.28 4.05 3.36
0.090 0.090 0.280 0.280 6.14 4.60 0.92 2.31 0.99 1.35 2.04 3.15 3.35 2.18 6.14 4.60 0.92 2.31 0.99 1.35 2.04 3.15 3.35 2.18
WELL SYMBOL
A2 158.5 71.0 5125.0 52.755 20.35 0.601 322.4 8.782 102.7 2.91
B 258.7 72.0 2165 22.708 123.5 0.210 616.2 2.906 288.4 5.16
C 228.97 72.85 1024.9 10.794 65.12 0.252 623.5 1.607 192.9 1.93
Inlet pressure - pwf (bar) Inlet temperature Twf (oC) Stock tank oil flow rate - QL (Sm /day) Total mass flowrate - m (kg/s) Wellhead pressure pwh (bar) Void fraction Mixture density M (kg/m3) Fluid mean velocity u (m/s) Acoustic velocity a (m/s) Water-hammer pa (bar)
Stock tank gas flow rate - QG (million m /day) 0.3796 0.26651 0.20352 0.0641
Figure 6.1 contains two plots that present estimated water-hammer and line packing for oil wells taken into considerations in present work. The first pressure increase on the plots is the water hammer effect after valve closure, and then the long line packing shows the pressure build up due to friction. The rapid pressure increase in line packing starts at the time where bubble point is reached.
180 160 140 120 pressure [bar] 100 80 60 40 20 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 tim e elapsed [s] Well A2 Well A1 Well C Well B
35.0 32.5 30.0 27.5 25.0 22.5 pressure [bar] 20.0 17.5 15.0 12.5 10.0 7.5 5.0 2.5 0.0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 tim e elapsed [s] Well C Well A1 Well A2 Well B
Calculated results are presented for each well in the form of plots. The list of plots is given below: