Descrição: Sistema de padronização de transportadores de correia da Vale
Cementing HandbookDescripción completa
SPE-170853-MS Enhancements in Fraction Measurements and Flow Modeling for Multiphase Flowmeters D. Chazal, M. Fiore, G. Jolivet, A. Lupeau (OneSubsea), C. Toussaint, B. Fournier, and F. Hollaender, Schlumberger
Copyright 2014, Society of Petroleum Engineers This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Amsterdam, The Netherlands, 27–29 October 2014. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract Multiphase flowmeters (MPFMs) have been used since the early 1990’s in the oil and gas industry and have gained acceptance in many environments. They have been considered the primary metering option for a wide range of applications - from heavy oil to wet gas. The combination of operational benefits, measurement robustness, demonstrable accuracy, and auditability has improved their status from new, unproven technology to that of the premium mainstream metering option. With Wi th mo more re th than an a de decad cadee of exp exper erie ience nce acq acqui uire red d us usin ing g a com combi binat natio ion n ve vent ntur urii an and d gam gamma ma-r -ray ay multip mul tiphas hasee flo flowme wmeter ters, s, fur furthe therr gai gains ns in mea measur sureme ement nt qua qualit lity y and oper operati ationa onall rob robust ustnes nesss have been achieved. We will illustrate how enhancements in fraction measurements using multi-energy gamma ray attenuation and a more comprehensive analysis of the gamma-ray spectrum have been developed and implemented to provide better measurement accuracy and stability, leading to enhanced performances in multiphase flow measurements. Another area of improvement that has been pursued is in the field of modeling of multiphase flows through a venturi. Historically, single-phase flow equations have been used, being adapted to multiphase flows by semi-empirical means to account for their complexity. While such models have proven robust for most standard applications, they can reach some limitations in particular conditions. We will present how a more dynamic model that considers the nature of the flow has led to improved accuracy of multiphase flow measurements. We present the scientific basis for the new enhancements as well as illustrate the accuracy gains achieved based on hundreds of flow loop test points, ultimately leading to the quantification of the accuracy gains obtained through those technological improvements.
Introduction Since their development and introduction in the late 1980’s and early 1990’s, multiphase flow meters (MPFM) have generated significant interest in the oil and gas industry. The primary drivers were initially to obtain production estimates from individual wells in conditions where the use of test separators was impossible or prohibitively expensive (e.g. subsea installations and small offshore platforms with limited available space), but have since gained acceptance across the industry. It is estimated that more than 6,500
multiphase flow meters have been permanently installed with strong growth forecasted in years to come (Falcone et al., 2011; Yoder, 2013). This covers a wide range of applications from subsea wellhead or riser base installations, to surface wellheads, test headers, trunk line monitoring or custody transfer measurements (Al-Hassaini et al., 2012, Syre et al., 2013). In addition, MPFMs have been deployed as mobile testing units used in lieu of mobile test separators in applications from exploration and appraisal to production. The technologies are now being used in cases where deployment environments, facilities design, operational constraints and measurement quality are critical, making MPFMs the better option. For instance, a large number of subsea developments are now based on the use of commingled riser production where individual well testing is impossible using surface facilities and, therefore, subsea MPFM are proving extremely valuable (Jackson et al., 2012). The same logic of ease of use applies at surface in remote locations ( Navarette et al. 2010) but also in operations where MPFM have proven to be reliable and are even considered as references against tests separators (Al-Hammadi et al., 2012). This rise in interest by the industry has led to numerous developments in the search for the best technology, or technologies capable of providing multiphase flow measurements in a wide range of conditions with high levels of accuracy and moderate sensitivity to input parameters and their associated uncertainty. After early development efforts driven by academic institutions in the 1980’s and 1990’s, continuous work has been conducted both by large companies already present in the oil and gas industry (in services, instrumentation or facilities) as well as by smaller companies focusing solely on flow metering. The principles of multiphase flow meters relies on the combination of fraction and velocity measurements, and a wide selection of technologies have been used by manufacturers over the years, some being already at their third generation of MPFM. More details can be found in previous publications (Falcone et al., 2011; and Pinguet, 2010) but for the sake of illustration, velocity measurements used differential pressure measurements (venturi, V-cones), positive-displacement meters, cross-correlation techniques, ultrasonic Doppler measurements, Coriolis meters, passive and active ultrasonic measurements, or tracers. For fraction measurements, the range of technologies deployed is also wide from single to multi-energy gamma ray, electrical capacitance and inductance/conductance, nuclear magnetic resonance, microwave transmission techniques, and near infra-red transmission. Some specific principles were shown to have limitations when put to the rigor of field conditions, and several manufacturers have revised their MPFM architecture principles over the past few years. This included either replacing one or several core measurement blocks, or providing optional added measurements to improve accuracy, or completely changing the principles, either as a replacement of existing meters or to target specific markets (Agar, 2010; Emerson Process management, 2013). It is not surprising that such a large technological turnover has been observed in a still relatively young domain. Indeed, most of the MPFM are barely more than 10 years old, and still partly on the learning curve. What becomes interesting to analyze in such environments is explaining why some principles have failed as well as recognizing which principles have been kept. Two core technologies have proven reliable in the field of multiphase metering: venturi meters for flow velocities and gamma ray measurements to determine fractions of oil/water and gas. The MPFM providers that can be considered as market leaders use both of these principles. It may be argued that, in some instances, the use of radioactive-based measurements is considered optional; but in all cases this remains the best way of achieving optimal measurement accuracy (Pietro Fiorentini S.P.A., 2014; Emerson Process Management, 2012; Weatherford, 2010). A key benefit of fraction measurements based on gamma ray attenuation is that it offers a straight path of measurement across the fluids independently of their distribution, whether oil, water, or gas continuous and therefore offers the ability to identify dispersed fluids entrapped in other phases, which is particularly important when considering emulsions or foaming flows. Such measurements can also be performed with very good accuracy. Venturi meters benefit from a simplicity of structure and ability to operate without the need for intrusive or moving elements potentially prone to damage, as well as from not relying on strong assumptions about the
distribution and stability of various phases in flow such as would be the case of ultrasonic, Doppler or cross-correlation-based measurements (Huang et al., 2013). While those two technological blocks have been at the core of most of the multiphase meters, the implementations have been various. First, in a majority of cases, a single-energy gamma ray measurement is used (i.e., based on the determination of attenuation at a single photon energy level). In those cases, only a single piece of information is obtained from that measurement, typically related to the split between gas and liquid phases linked to mixture density, and additional information is required to determine the fraction of three phases. This usually comes in the form of a measurement focusing on water content based on the contrast in electromagnetic properties between water and hydrocarbons, or can also consist of a dedicated water-cut measurement to split liquids between water and oil. This generally requires performing measurements at multiple locations since the mechanical design required to collocate pressure tapping, gamma ray measurements, and electromagnetic probes or electrodes becomes very difficult to realize when considering the various metallic penetrations required. In such situations, the multiplicity of measurements raises issues regarding the representativeness of correlating observations performed at multiple locations at the same time in inherently chaotic and unstable flows. Those observations, combined with significant in-house development work on alternative technologies and subsequent analysis, led to focusing on a simple design that allowed for all measurements to be performed at a high frequency at a single measurement cross-section. This removes the need for stringent flow conditioning or complex correlative interpretation of measured flow velocity and fractions. The multienergy gamma ray measurement principle was employed at the throat of the venturi tube, providing both total flow rate and all three fractions at a single cross-section. This technology was commercially introduced in 2000 (Theuveny et al., 2001) and has since been used extensively in wide-ranging applications after an early focus on standard black oil wells. The technology principles have been proven robust enough to cover applications from heavy oil (Pinguet 2012) to gas condensates (Lomukhin 2011) while providing excellent metrological performance. However, while the fundamental principles are sound and validated by the multiphase flow metering industry and by returns of experiences, there are still gains achievable through improvements of the acquisition and interpretation of the acquired signals. The summary of several years of work in that field that have led to the launch of an evolution (JPT 2014). The initial interest in MPFM focused on high-value wells producing at high rates and with low to moderate water cuts, with expectations of the meters to provide reasonable accuracy for monitoring purposes. Today, more and more applications call for deployment in either low net oil producers or, at the other end of the spectrum, for better accuracy to address fiscal or custody transfer applications, which have been tackled as part of this work. In the following, and after providing some background on the main principles used, the work performed on three main axes will be presented: enhancements in fraction measurements via a more comprehensive interpretation of the measured gamma ray response; refinement of the flow modeling under multiphase conditions, and improvements in the configurability and ease of maintenance of the meters.
Principles of multienergy gamma ray/venturi MPFM The fundamental interpretation workflow of multienergy gamma ray/venturi MPFMs can be decomposed in three main blocks: Determination of flowing phase fractions from the gamma ray measurements using at least two distinctive energy levels ● Determination of total mass rate flowing through the meter, leading to the determination of the flow rates of each phase at metering conditions after the application of slippage effect ●
Conversion of flow rates from metering pressure and temperature to the appropriate pressuretemperature standard conditions.
Fraction determination using gamma ray measurements
Determining flowing fractions based on multienergy gamma ray measurements relies on the response of fluids to two different phenomena occurring within the range of photon energy in use: Compton (incoherent) scattering and photoelectric absorption. Both phenomena are related to the interaction between gamma ray photons and electrons present in the atoms of the flowing mixture. For further information, the reader can refer to Knoll (2000). The primary quantity of interest is the linear attenuation of each phase, which is the product of the density of the fluids by its mass attenuation (attenuating power for a unit of density). The attenuation of a fluid to gamma rays is obtained from the Beer-Lambert law (Lambert 1760): (1) where I o is the intensity of the incident photon beam; I the intensity of the transmitted beam; and A the absorbance of the matter. The absorbance can, in turn be expressed as the product of the linear attenuation characterizing the fluids and thickness of the fluid penetrated. The linear attenuation coefficient can be further expressed as the product of fluid density by its mass attenuation coefficient, leading to an expression of measured counts over a given interval of time: (2) where N o is the number of photons entering the matter; N the number of photons transmitted; d the thickness of material penetrated; the density of the matter; and its mass attenuation coefficient of the matter. When considering gamma rays passing through a mixture of fluids, the resulting attenuation can be expressed as the sum of the contribution of each atom/molecule or phase present in flow as: (3) where i is the index of the component and i the volumetric fraction of that component. It is interesting to note that this equation can be re-written in terms of mass fraction i by factoring in the mixture density mix: (4) The mass attenuation coefficient of a given fluid is a strong function of the energy of incident photons and by extension of the dominating physical phenomena taking place. Photoelectric absorption is an effect where an incident photon is absorbed by an electron (bound or orbital), leading to the electron being ejected from the atom. The probability of having photoelectric absorption is higher when the energy of the photon is close from the binding energy of the electron. This phenomenon is more likely with atoms of high atomic number because the binding energies of bound electrons are within the range of energies of gamma ray considered at the lower energy level. The mass attenuation due to photoelectric absorption observed at a given energy level can be considered as proportional to the cube of the atomic number, thus showing significant contrast between various atoms depending on their atomic number. In particular, this provides a good contrast between hydrocarbons (dominated by carbon, atomic number 6) and water (dominated by oxygen, atomic number 8). Compton scattering, or incoherent scattering, is a different effect in which the photon is not absorbed but collides with a free or weakly-bound electron and transfers part of his energy to it, thus coming out at a different angle and with lower energy/frequency. The probability for this to occur is higher when the
incident photon energy is significantly larger than the binding energy of the electron. The attenuation of photons due to Compton scattering is essentially related to the electron density of the fluids and thus to a large extent to the density of the fluids themselves. This is not a perfect correspondence though, as the electron density varies between atoms (Figure 1). In particular, hydrogen (which does not have any neutrons) has a higher electron density compared to other atoms. This means that, as shown in Figure 1, there exists a contrast between light hydrocarbons (with a high ratio of hydrogen to carbon atoms), Figure 1—Ratio of electron over molar weight for different atoms and long hydrocarbon chains (with nearly two hydrogen molecules. atoms per carbon atom), and other molecules such as CO2 not bearing any hydrogen. Other effects can take place such as pair production or Rayleigh (coherent) scattering but those are either not active in the range of energies considered here or have a negligible contribution and will not be discussed further. To illustrate how different components respond at different photon energy levels, Figure 2 shows the mass attenuation coefficient of various fluids (CH2 representing long-chain hydrocarbons that would be found in oil, CH4 as an analog to gas, H2O for water, and CO2 as a reference of hydrogen-free Figure 2—Mass attenuation of various fluids (dashed photoelectric; dotted Compton; solid line total attenuation). molecule) at different energy levels, splitting the contribution of each effect. This clearly shows how the combination of high-energy gamma rays measurements (above 70 keV) with low energy measurements can provide solid indications about oil, water and gas content in flow. The high-energy measurements show similar mass attenuations for the various fluid types and the measured linear attenuation will essentially depend on the mixture density (even though it cannot provide a direct measurement of density) and will therefore strongly relate to the gas fraction. The low-energy measurement will additionally show a strong contrast between hydrocarbons and water and can therefore be used to determine the water fraction. Using those two independent measurements as well as knowing that the sum of oil, water, and gas fractions is equal to one, the flowing fractions are then determined by solving the following system of equations:
This can then be used to determine mixture density and perform flow-rate calculations based on the measurement of differential pressure across the venturi. Flow rate calculations at metering conditions Determining flow rate using a venturi tube is customarily used in the oil and gas industry as well as many others. It relies on Bernoulli’s principle, which states that a change of velocity in a fluid flowing through a pipe will induce a change in pressure. When combined with the mass conservation equation, this lead to the typical venturi equation:
(6) where Q is the mass flow rate; Ath the throat cross-sectional area; Dth and Din the throat and inlet diameters of the venturi tube respectively; P the measured differential pressure; g the acceleration of gravity; is the mixture density, and h the height difference between the inlet and throat pressure tapping points used to measure differential pressure. The frictional losses in the venturi and the effect of compressibility are taken into account by means of two corrective factors respectively called discharge coefficient and expansion factor. Discharge coefficient and expansion factor are normalized in ISO standard 5167 (ISO 2003). The performance of venturi flow meters relies on the uncertainty of: machined venturi geometry ● performance of the transmitters used ● knowledge of the fluid properties inputs (density and viscosity) ● discharge coefficient and expansion factors estimation ●
For multiphase flows, it is necessary to adapt the standard venturi equation to account for additional losses due to the turbulent nature of multiphase flow in addition to discharge and expansion terms (Atkinson et al., 2000). This leads to determining the total flow rate through the meter that is then split into the flow rate of gas, oil and water according to measured phase holdups and the application of a slippage relationship. It is assumed that there is no slippage between oil and water thanks to flow conditioning via a blind-T spool located at the inlet of the meter that provides efficient mixing. It is important to note that the determination of a slip law is not straightforward. The strength of the current model relies on a very large database of tests point (over 12,000 test points) considering the largest possible range of flow conditions. Trying to map velocities based on multiple cross-pipe measurements has not yet been achieved at the required resolution level to allow for such claims. Instead, the solution consisting of performing all measurements at a single pipe location and at a high frequency allows for the local application of slippage without relying on a specific flow distribution assumption. Conversion of flow rates to standard conditions Performing flow-rate conversions from measurement conditions to an agreed set of standard conditions is a requirement for any metering device to provide normalized data for production and reservoir engineering use, or for production reporting purposes. This is well-known when using test separators where the flow rate of oil measured at separator pressure and temperature is converted to standard conditions either explicitly through the application of a shrinkage factor or implicitly by calibration of the sensor readings against a tank measurement, yielding a combined meter factor (CMF) accounting for both sensor calibration against test fluids and oil shrinkage. Fluid properties are also required on reported gas measurements to account for gas deviation but also to consider the amount of gas released from the oil when flashing live oil to standard conditions (the dissolved gas-in-oil ratio being known as Rst or GOR2 in well testing parlance). This conversion process relies on a mass-conserving fluid properties model, ensuring that a consistent set of fluid properties is used for the various properties. A comprehensive conversion process is illustrated in Figure 3, and can be summarized in two main behaviors:
volumetric change of fluids between metering conditions and standard conditions: oil and water shrinkage (bo and bw respectively) and gas expansion (bg) factors ● phase changes characterizing the amount of gas in solution in the liquids (Rst and Rwst for gas dissolved in oil and water respectively) as well as liquid condensing from the gas phase (rgmp for liquid condensate, rgwmp for water steam condensing out from the gas). ●
Figure 3—Conversion process from multiphase conditions to standard conditions.
Leveraging information from the full gamma ray spectrum The metering principles described above have been those used since the first commercial deployment of our multiphase metering technology and though the range of application has been extended over the years, the fundamental principles have not changed (Atkinson 1999). One characteristic of a multiphase meter is the use of a 133Ba isotope as a source of gamma rays. This source has several advantages: a half-life of over ten years, allowing it to be used for at least twenty to thirty years without replacement, while not creating major disposal issues (a source can be used for several half-lives as long as enough counts are measured without significant effect on the signal to noise ratio) ● the possibility to use a low-activity source (10mCi) thanks to a design using pressure-retaining ceramic windows that allow for the passage of gamma rays, removing the need for a high-activity source required when having to penetrate metal pipes ● a highest gamma ray energy level around 380 keV, reducing the need for heavy shielding 241 ● no associated emission of alpha particles such as emitted by Am that are extremely harmful or 133 even significant emission of beta particles ( Ba is generally considered as a gamma ray only source) ● a spectrum that bears several energy lines within the range of interest, at around 32 keV (in fact several lines at very similar energy levels) and 81 keV respectively, in addition to several lines at higher energy levels above 270 keV, with the dominant one being at 356keV. ●
With this combination of HSE and information content benefits, 133Ba proved to be a very suitable isotope to the purpose of fraction determination through dual-energy gamma ray measurements, providing more measurements than required for cross-validation purposes or to potentially obtain additional information about the flow content (Pinguet et al., 2010). However, there are specific challenges associated with the complexity of the spectrum of this barium isotope. The primary one is related to the imperfections of gamma ray detectors. A scintillation detector operates as follows (Knoll 2000): Each gamma ray entering the detector if first transformed into light with an intensity proportional to the energy level of the incoming photon. ● That light then enters a photomultiplier tube (PMT) that converts that light into electrons and amplifies the signal, thus generating an electrical pulse ● That pulse is then converted into a shaped pulse that is analyzed through an electronics board to interpret the signal and provides counting statistics as a function of incoming gamma ray energy level ●
Figure 4 —Gamma-ray measurement chain when using a scintillation detector.
This process is illustrated on Figure 4. As is the case with any measurement, there are imperfections related to such a nuclear system. Two of which are of particular interest here and are related to the crystal and to the photomultiplier tube. Gamma rays emitted by radioactive isotopes are generated at very specific energy levels driven by the isotope decay mechanism and associated quantum effects. When considering a narrow beam, those enter and exit the flowing fluids without mod- Figure 5—Energy spectrum deposited on a crystal from a single-energy incident photopeak. ification of their energies; however their measurement itself does affect the captured spectrum. Considering the crystal first, a large majority of incident gamma rays may be properly captured and recorded but the process of Compton scattering may take place in the crystal itself, leading to the generation and deposition of gamma rays of lower energy levels inside the crystal. In that situation, the detector would record the energy of the main incident photon but also that of the scattered photon. Those “secondary” recordings can only occur at energy levels equal to or lower than a specific limit, called the Compton edge and corresponding to the energy of the photon generated from a head-on collision with an electron. In addition to that effect, multiple collisions may also occur, creating measurements spanning the full range of energies up to the total incident energy. This means that when measuring the gamma-ray spectrum of a single-energy photon source, a continuous spectrum of light will be emitted by the scintillator crystal and will be fed to the photomultiplier tube (Figure 5). Then, in the process of converting the incident light into an electrical pulse, spectral smearing occurs due to the limited resolution of the measurement system. The output of the PMT is, in essence, a number of electrical charges proportional to the number of gamma rays deposited in the crystal. However, due to the statistical nature of the electron multiplication process, the output charge can vary from one event to the other. This fluctuation follows a Poisson process which results into a broadening of the whole spectrum. The recorded spectrum then loses in sharpness and covers a wide range of energies besides the primary incident one of interest (Figure 6). The phenomena described above influence any radiation measurement, where the non-idealities of the system have to be taken into account when interpreting the data. It is not possible to focus the measurement on a narrow range of energies since this would mean ignoring a potentially large number of photons whose signal has been smeared out of the range of recording, while considering a wide-enough window requires applying a correction for signal stemming out of single or multiple Compton scattering. When considering measurements targeting multiple levels of energy, this implies also that measurements performed at lower energy levels have to be corrected for contamination by Compton scatter from higher energy gamma rays. Since those effects are related to the detector system, they can be studied and properly accounted for through calibration to determine correction coefficients. Such corrections do require proper detector control to account for aging of the scintillation crystal and associated electronics,
Figure 6—Impact of measurement chain on recorded gamma ray spectrum.
Figure 7—Incident spectrum of 133Ba.
affecting the efficiency of the system as well as the conversion from charged pulse energy to the appropriate gamma ray energy level. When considering the incident spectrum of a 133Ba isotope, a significant number of different gamma rays are generated by the decay process of that radioactive source as shown on Figure 7. It is particularly clear that the higher-energy gamma rays will create a measured spectrum overlapping with the lower levels of energy used for interpretation purposes (the group in the range from 30 to 34 keV and the two peaks at 81 keV). Historically, this effect has been accounted for through the determination of appropriate correction factors obtained from extensive laboratory measurements on the detector system used, and by means of detector control in terms of temperature regulation and signal gain control. The interpretation is then based on the measurement of gamma ray counts over specific windows of energy around the peaks of interest, with the higher energy emissions (above 240keV) being taken as one. While this approach has proven over the 14 years of deployment in the field, it was clear that this remained an imperfect way of properly accounting for incident gamma rays and that many incident counts of moderate intensity were ignored in the process. To close that gap, it was decided to update the acquisition and interpretation principles to perform a more comprehensive analysis of the measured signal. Fundamentally, the effect of detector imperfections in measurements can be understood since this depends on well-known Compton scattering effects in the energy deposition process in the crystal, and can be modelled. Similarly, spectral smearing is an intrinsic property of the detector system that can also be accounted for. The implication of this ability to properly model the measured spectrum obtained from a mono-energetic beam is that when considering a source emitting gamma rays a multiple energies, the measured spectrum can be modeled by superposition of signals. Considering that the counts deposited on the crystal at a level of energy e due to an incident gamma ray at a level of energy e’ can be expressed by the crystal response function h(e,e’) the deposited signal d(e) can be expressed as: the sum of the contributions from all gamma rays :
Figure 8 —Decomposition of the full measured spectrum into the contribution of individual incident gamma ray energies.
(7) where i(e’) denotes the intensity of incident gamma rays at an energy e’. At a second stage, the measured response s(e) can then be considered as the convolution product of the deposited signal with the spectral smearing kernel (PMT response) g(e,e’) as: (8) As the incident signal is not continuous but consists of a discrete set of energies, and considering that the measurement will bin data into sets corresponding to specific energy ranges, we can express the above equations in discretized matrix form: (9) where S is the vector of measured gamma ray counts at various energy levels of dimension nm; G is the PMT response kernel; H is the deposition kernel; and I is the incident gamma ray intensity vector of dimension nin. To accurately determine the true incident gamma rays present in the vector I , two modifications were required in the acquisition and interpretation chain. The first is to acquire not just gamma ray counts over specific ranges of energies but to perform a full spectrum measurement across the range from very low energies to significantly above the highest peak (at 384keV for 133Ba). In the new system, the measured spectrum is captured over 512 bins of energy ranges using a dedicated multichannel analyzer (MCA), thus providing a detailed signal for analysis. Having the measurement available, it was also required to build an adequate interpretation scheme to obtain the incident energy vector I from the measured smeared signal S and the characteristic response matrices of the detector system G and H through a deconvolution process minimizing the appropriate error function to match noisy measurements to the modeled response. We can note that since nm is significantly larger than nin, we have a strongly over-determined system to solve. This process is illustrated on Figure 8, showing a measured spectrum (gray bars) decomposed into the contribution of the various incident energy levels presented in Figure 7. The modeled spectrum is the sum of the gamma rays emitted by each contributing energy line, whose intensity is determined as part of the solving process.
The acquisition system and interpretation model were extensively tested in static laboratory conditions, using materials of known attenuation for reference to ensure the correctness of the response against expected attenuation, as well as performing extended stability tests of the measurement system under elevated temperatures to simulate accelerated aging of the components. This comprehensive formulation enable significantly improving the quality of measurements. First, moving away from windowed counts around the energies of interest to the comprehensive interpretation of the full spectrum means that every measured photon is now used in the analysis. This leads not only to better measurement statistics- only functions of the intrinsic behavior of naturally-decaying isotopes and not anymore of the counting system - but also provides a better analysis of the signal between its individual components, splitting the various energy lines at very high levels of energy as well as adding low-intensity intermediate lines at 53keV and 161 keV that were previously not taken into account. Second, as the detailed spectrum is measured, the health status of the detector can easily be assessed and its controls automated to ensure that measurement efficiency is maintained via feedback control loops. Besides ensuring the long-term stability of the measurement, this also significantly reduces the need for thermal regulation of the detector system. This makes it possible to use a simpler system, operated at ambient conditions while providing excellent measurement stability and consistency. Third, as the signal quality is enhanced, the technology can be used in an extended range of application where a low mixture attenuation is measured, typically small pipe diameter allowing targeting low producers, as well as providing better measurement quality with low-attenuation fluids, typically in high gas content applications. When either d or the linear attenuation product . in equation (2) is small, the difference between emitted and measured counts N-N o is small. Since this is the signal of interest, a measurement error of only a few counts can have significant impact on the accuracy of the measurement, which has historically been a limiting factor to the development of small-size meters. With an enhanced measurement based on the full-spectrum deconvolution, it was possible to expand the range of operations down to at least 2-in nominal pipe size (19 mm diameter at the throat of the venturi) while keeping high metrological performances.
Enhanced flow modeling under multiphase conditions Multiphase flow modeling derived from Bernoulli equation can have some limitations when the whole flow passing through the venturi becomes less homogeneous. In most cases the nature of the multiphase flow produces heterogeneous distribution of velocities. This can affect the robustness of the flow model if based on a genuine homogenous distribution assumption. This loss of homogeneity is extremely complex to predict and model. There have been multiple attempts to map flow regimes as a function of superficial gas and liquid velocities, but such approaches remain qualitative and are only applicable for the conditions where they have been established. Furthermore, the flow patterns depend on a large number of parameters such as fluid density contrast and viscosities, relative velocities of each phase, interaction between the phases, and flow conditioning prior to the metering section. Some example behavior are illustrated on Figure 9 and Figure 10 and often several different flow patterns can be observed during a single flow sequence as large-scale production instabilities occur. Besides flow pattern maps, the use of dimensionless parameters such as the Lockhart-Martinelli parameter, the Froude number or Ohnesorge number are often referred to as ways of assessing the interaction forces between fluids and consequentially to evaluate the nature of the flow. Current flow modeling derived from Bernoulli’s equation was based on the use of semi-empirical models not assuming any particular fluid distribution and tuned to flow-loop measurements. To properly build and validate those models it is crucial to test in a wide range of flow conditions that can be observed in various flow loops. The operating pressure, fluid types, or piping layout of various test loops allow assessing meter performances under a wide range of conditions, starting from standard fluids at moderate pressures available in multiple flow labs but with different installation effects such as long slug lines,
viscous flows (Atkinson et al., 2000), or wet gas and high pressure conditions (Brister 2013). In addition, validation under real conditions can also be performed from field testing (Theuveny et al., 2001), but in those situations, the reference measurements would have uncertainty levels significantly higher than in flow loop conditions, and results have to be taken with proper care and cannot be used for the purpose of model improvements (Hollaender 2013). Such extensive testing in multiple facilities is required to avoid building models that are biased to the behavior of one or two facilities. To properly account for the inherently unsteady nature of multiphase flows, the approach consisting in using collocated fractions and velocity measurements and performing the interpretation at a high data acquisition frequency to properly account for rapid changes has proven very robust to a large number of flow conditions. The collocation of measurements is critical to ensure consistency between the observations with the throat of the venturi being the ideal measurement point. This is where the flow Figure 9—Types of flow-regimes that exist in vertical flow conditions. is best conditioned thanks to the acceleration experienced by the fluids and where measurements are little affected by upstream flow patterns. As seen in Figure 10 - c, the fluid distribution changes significantly between the inlet, throat, and outlet section of the venturi tube. At the inlet, a vertical line of liquid can be clearly seen. This line is in fact static water and does not move along with the other fluids at the core of the flow. At the venturi throat, the liquid film wetting the wall of the pipe thins down and a higher dispersed fraction is observed due to the acceleration of the fluids. On the divergent section, a near-cylindrical core of gas-dominated fluids flows at a high speed at the center of the pipe while liquid droplets are shed to the near-wall region and fall back toward the venturi throat against the main flow direction. This example illustrates how the location of fraction measurement can strongly influence the process of flow dynamics estimation when correlating different measurements exposed to different behavior. Using more specific flow models in conditions where the nature of flow can be well estimated has significant value, for instance in the presence of wet gas flow (Cadalen, 2009), where fluids can be assumed to be distributed between an annular liquid layer with a few entrained gas bubbles and a gas-dominated core with entrained mist. Having that knowledge enables the application of flow models with a well-refined velocity profile across the pipe and a proper conversion of measured holdups to flowing fractions under dynamic conditions. An opportunity of improving the computation was identified in the recognition of the acting flow-regime and proper consideration regarding the intermittency of flow. This can be evaluated directly from the high frequency nuclear measurement. Recognizing the occurrence of permanent flow as shown in Figure 11-a is quite straightforward, and such information can be used to refine the flow model. Similarly, in the presence of intermittent flows, the frequency of variations can be used to refine the flow calculations and properly account for the nature of flow. Such observations linking the behavior of flow through a high-frequency interpretation of flowing fractions proved instrumental in ensuring that both the fractions used as input to the flow model and the flow calculation process provides optimal results.
Figure 10—Example of flow patterns from high-speed capture - (a) bubbly flow, (b) slug flow and (c) annular dispersed.
Figure 11—Nuclear response with stable flow (a) and very unstable flow (b) over a 10-mn period.
Simplification of the MPFM system Beside improving the acquisition and interpretation methods to offer more robust and accurate measurements, the review of the metering system also considered improvements to the metering hardware around three main axes: simplification of the meter arrangements ● ease of operation and maintenance ● modularity to adapt to operators’ specific requirements ●
The physical assembly is presented in Figure 12, with the main components highlighted:
Figure 12—Physical architecture of the meter.
the venturi body of the meter with the associated measurements of pressure, differential pressure and temperature acquired via a multivariable transmitter (within the blue border). Two pressure tapings at the inlet and throat of the venturi are used to measure flowing pressure and differential pressure. An optimum place for the thermowell temperature sensor was made at the outlet of the venturi, near the throat. ● the gamma detector with multichannel analyzer (with red border) is located at the level of the throat of the venturi, opposite to the gamma-source holder under the cover on the right ● the flow computer, along with power and data connectors is located in the small box with the green borders ●
To simplify the system, a single multi-variable transmitter is used to provide the standard measurements (pressure, differential pressure, and temperature), making the system more compact and reducing the number of spares if replacement is required. The nuclear detector itself is also fairly standard and does not use thermal regulation to maintain its response since the full-spectrum analysis allows for the proper tracking and scaling of its response. This has two benefits: limiting the temperature exposure of associated electronics and minimizing power supply requirements. The flow computer has been upgraded to follow industry advances, minimizing its size while increasing its data storage capacity to provide full redundancy in data capture. It is attached to the meter through a mounting base that makes it hot-swappable, meaning that it can be replaced on site without requiring bleed-off and depressurization of the meter, without creating any risk of sparks in a Zone 1 area. A significant change in the design of the meter was made to provide significant configurability following each operators’ own standards and requirements. Historically, there has been some level of customization required on each group of manufactured meters to apply specific standards for individual projects. This has an impact on meter delivery time and cost due to re-engineering requirements, which should not be the case for widely deployed technologies. This was achieved by a modular design, making it possible to build meters combining a large number of options: meter size ● piping connection type ● material of construction ● certification standards ●
isolation system of sensors from flowing fluids ● various options on data connectivity, local display, and power supply ●
Those changes were implemented with considerations that, with a multiphase metering market requiring several hundreds of meters to be manufactured each year, efficiency gains were required in the design, manufacturing, and maintenance process. Therefore, a rationalization of the MPFM system was called for.
Test results After significant work on the fundamentals of the data acquisition and interpretation was performed, a first set of experimental prototypes was built and the meters were used for a first round of flow-loop testing (over 300 test points in three different flow loops) as well as for stability testing. From the lessons learned with those units, some revisions were applied to both the meter design and to the data acquisition and interpretation methodologies and final versions of the meter in various sizes were built and used for testing Laboratory results in flow loops
The first method of validating performances of multiphase meters consists of using them in flow loops with high accuracy references to determine metrological specifications. This was initially done in four different flow loop facilities, with more planned to cover the largest possible range of fluids and flowing conditions. To date, over 600 test points have been captured in different loops with different sizes of meters, increasing on a monthly basis. The output parameters evaluated in multiphase flow loops can be several but, for the purpose of illustration of the improvements achieved only two that are straightforward to evaluate will be shown, first the water/liquid ratio (WLR) to evaluate the gains achieved from the enhanced fraction measurement, second the liquid volumetric rate to illustrate the quality of the flow model over a wide range of conditions. It is important to note that all the points shown below correspond to relatively short acquisition periods, in the range of 7 to 20 minutes. In those situations measurement noise plays a non-negligible part, in particular under elevated GVF conditions. Another point to note is that the lack of data at low gas fractions is essentially due to limitations of the test facilities. In order to perform a representative measurement, each individual single-phase meter has to be within its range of calibration and operation. When considering in particular a small meter size, this means that the loop may reach its lower limits in terms of range of achievable gas rates and thus cannot be operated on the low end of the GVF range. Figure 13 shows the WLR (water/liquid ratio) comparison between meter readings and flow-loop data. The data here comes from four different flow-loops, at low (8-10 bara) and moderate (18-20 bara) operating pressures, with light as well as viscous fluids and with various installation setups leading to both stable and unstable flows. This shows that the deviations are small and for GVF values below 95% provide an uncertainty of measurement of less than 2%. This is a reduction of the uncertainty of one third compared with previous specifications and clearly shows the gains achieved. Figure 14 shows results obtained on total mass flow rates with the 2-in (19 mm throat diameter) meter in three different flow loops. The data there shows no particular bias and excellent performances validating the quality of the flow model in that situation. An improvement in measurement performances compared with a previous meter version is observed, even though the small size creates a more challenging environement for the measurement. Field deployment In addition to testing in a controlled environment, multiple units have been deployed in several locations across the world to be exposed harsh field conditions. Those units were skidded and are used as mobile test units to evaluate the response of the hardware system as well as to assess the representativeness and
Figure 13—Comparison of water liquid ratio (WLR) differences between MPFM and flow-loop measurements for a 4-in meter.
Figure 14—Comparison of total mass rate differences between MPFM and flow-loop measurements for a 2-in meter
Figure 15—Picture of a unit carrying two meters (a 19 mm and a 40 mm venturi) used for mobile field deployment.
reliability of measurements under challenging environments in terms of ambient temperatures, humidity, shock, and vibrations. Figure 15 shows a picture of a skid bearing two different meter sizes, providing a large operational turndown of 50:1 with a compact setup that has been used for field introduction and validation. To date, no equipment failure has been observed from these units, which are used daily. Since this is not in itself a new technology but an enhancement on a proven one, the units have been used only to demonstrate the operability of the meter.
Conclusions and way forward Multiphase metering using a combination of multienergy gamma ray measurements and a measurement of differential pressure across a venturi tube, collocated at the throat section, has proven to be a robust way of obtaining high-quality multiphase flow rate measurements. Even with fifteen years of experience and continuous improvement, some enhancements in the data acquisition and interpretation processes yielded a significant improvement in the metrological performances. The simple architecture of the meter as well as the clear benefits of using collocated measurements allowed for a rationalization of the hardware systems are of particular interest for a mature market. While the gains achieved are significant, work is still on-going to optimize measurement quality and its use. The enhancements in the fraction measurement process in particular offer significant opportunities to improve performances, but also offering additional deliverables as well as usability enhancements.
Acknowledgments Special thanks to Michel Bérard, Sebastien Cadalen, Bruno Pinguet, Marcus Rossi, and the initial Schlumberger engineering team in Bergen (Norway) involved in the project development process for mechanical design, instrumentation, and testing.
Nomenclature A D d(e) e e’ g(e,e’) G
I o nin nm
N N o Q q s(e) S
area / absorbance diameter deposition spectrum measured photon energy level source photon energy level spectral smearing impulse response spectral smearing kernel matrix crystal deposition impulse response crystal deposition kernel matrix intensity of photon beam incoming gamma ray vector intensity of source photon beam number of incident energy lines number of measured energy bins number of photons (counts per second, cps) number of photons emitted by the radiation source mass flow rate volumetric flow rate smeared spectrum measurement vector volumetric fraction, or holdup (-)
P P mix Subscripts i in g o th w
mass fraction (-) differential pressure density (kg/m3) mixture density (kg/m3) mass attenuation coefficient (cm2/g or m2/kg) index number inlet gas oil throat water
References Agar, J. 2010. New Coriolis Based Multiphase Flow Meter for Heavy Oil Mature Fields. Paper SPE 135992, presented at the SPE Russian Oil & Gas Technical Conference and Exhibition, Moscow, 26-28 October. Al-Hammadi I.T., Handak Aly S., Khan M.N., Gurses H., Baslaib A. and Younes M.K. 2012. Validation of Coriolis and V-Cone Meters Using Multi Phase Flow Meters (MPFM) in ADCO’s North East BAB (NEB) Field – A Case History. Paper SPE 162200, presented at the Abu Dhabi International Petroleum Exhibition & Conference, Abu Dhabi, 11-14 November. Al-Hosani M.A. and Khan M.N. 2012. Evolution of the Smart Production Testing Strategy in ADMA-OPCO, paper SPE 161237, presented at the Abu Dhabi International Petroleum Exhibition and Conference, Abu Dhabi, 11-14 November. Atkinson, I. 1999. New Generation Multiphase Flowmeters from Schlumberger and Framo Eng. AS. Paper presented at the 17th International North Sea Flow Measurement Workshop, Oslo, Norway. 25-28 October. Atkinson, D.I., Bérard, M. and Ségéral, G. 2000. Qualification of a Nonintrusive Multiphase Flow Meter in Viscous Flows. Paper SPE 63118, presented at the SPE Annual Technical Conference and Exhibition, Dallas, 1-4 October. Brister, C. 2013. Wet Gas Performance of a Low Energy Gamma Ray Multiphase Meter. Paper OTC 24352, presented at the Offshore Technology Conference Brasil, Rio de Janeiro, 29-31 October. Cadalen, S. 2009. Method for Determining Flow Rates for a First Gas Phase and at Least One Second Liquid Phase Present in a Multiphase Flow. International (PCT) Patent No. WO 2010/034939. Emerson Process Management, 2012. Roxar Multiphase Meter 2600 datasheet . Emerson Process Management, 2013. The Roxar Multiphase meter 2600, commercial brochure. Falcone, G. and Harrison, B. 2011. Forecast Expects Continued Multiphase Flowmeter Growth. Oil & Gas Journal, Vol. 109, Issue 10. Hollaender, F. 2013. Field Testing Multiphase Flow Meters: Lessons Learned and Best Practices. Paper ADIPEC-691, presented at the Abu Dhabi International Petroleum Exhibition & Conference, Abu Dhabi, 10-13 November. Huang, S., Xie, C., Lenn, C., Yang, W. and Wu, Z. 2013. Issue of a Combination of Ultrasonic Doppler Velocity Measurement with a Venturi for Multiphase Flow Metering. Paper SPE 16442, presented at the Middles East Oil and Gas Show and Conference, 10-13 March. International Organization for Standardization (ISO). Measurement of Fluid Flow by Means of Pressure Differential Devices Inserted in Circular Cross-Section Conduits Running Full. ISO standard 5167, 2003, Geneva
Jackson M., Napalowski R., Paris N. and Moksnes P.O. 2012. Operational Experience with Subsea Multiphase Flow Meters in the Pyrenees Development, Offshore Western Australia. Paper SPE 158518, presented at the SPE Asia Pacific Oil and Gas Conference and Exhibition, Perth. Knoll, G.F. 2000. Radiation Detection and Measurement. 3rd edition. Wiley and Sons, New York. Lambert, J.H. 1760. Photometria sive de Mensura et Gradibus Luminis, Colorum et Umbrae. Sumptibus viduae Eberhardt Klett. Lomukhin, A., Romashkin, S., Rymarenko, K. and Afanasiyev, V. 2011. Experience of Multiphase Flow Measurement Systems Application in Arctic Conditions. Paper SPE 149922 presented at the SPE Arctic and Extreme Environments Conference & Exhibition, Moscow, 18-20 October. Navarette, C. and Gamboa, E. 2010. Application of Multiphase Meters in Gas Condensate Wells with High Flow. Paper SPE 139391, presented at the SPE Latin American and Caribbean Petroleum Engineering Conference, Lima, 31 November-3 December. Pietro Fiorentini SpA. 2014. Multiphase Flow Measurements. http://www.fiorentini.com/media/files/ 995_flowatch_2014_lr.pdf , commercial brochure, Pinguet, B. 2010. Fundamentals of Multiphase Metering . Sugar Land, Texas, Schlumberger. Pinguet, B. 2012. More than 10 Year’s Experience with Multiphase Flowmeter in Heavy Oil with Either Cold or Thermal Production. Paper WHOC12-131, presented at the World Heavy Oil Congress, Aberdeen, 10-13 September. Pinguet, B., Bornia, O. and Caldera, J. 2010. The Measurement of 4-Phases in Diluent Environment with Only the Use of a Venturi and Multi-Energy Gamma-Ray Combination. Paper presented at the 9th South East Asia Hydrocarbon Flow Measurement Workshop, Kuala Lumpur, 2-4 March. Surface Multiphase Flowmeter. 2014. Journal of Petroleum Technology, Vol. 66, No. 2 (February): 28 Syre, B., Andersen, O. and Fosse, S. 2013. Using Multiphase Meters for Fiscal Purposes – A Field Experience. Paper presented at the 31st International North Sea Flow Measurement Workshop, 22-25 October. Theuveny, B., Ségéral, G. and Pinguet, B. 2001. Multiphase Flowmeters in Well Testing Applications. Paper SPE 71475, presented at the SPE Annual Technical Conference and Exhibition, New Orleans, Louisiana, 30 September-3 October. Weatherford, 2010. Alpha VSRD Multiphase Flowmeter and Alpha VSR Wet Gas Flowmeter, commercial brochures. http://www.weatherford.com/Products/Production/FlowMeasurement/AlphaVSRDMultiphaseFlowmeter//index.htm. Yoder, J. 2013. The Many Phases of Multiphase Flow Meters. Pipeline & Gas Journal, Vol. 240, No. 7.