Solutions for Reliable and Accurate Measurement of Water Production in Subsea Wet Gas Fields

Transcription

1 Solutions for Reliable and Accurate Measurement of Water Production in Subsea Wet Gas Fields by Lex Scheers (Hint Services, The Netherlands) and Arnstein Wee (MPM, Norway) Abstract The use of high performance Multi-Phase Flow Meters (MPFM s) and Wet Gas Flow Meters (WGFM s) has an enormous economic impact, in terms of both CapEx and OpEx reduction, on subsea gas and oil developments. This was the main driver for a number of international oil companies to actively support the advanced MPFM and WGFM concept that was developed by MPM in Norway. The development and subsequent testing has resulted in an accurate metering system that can be deployed for both product/sales allocation as well as for flow assurance purposes (hydrate management, formation water breakthrough). A key feature of the MPM meter is that it has an extremely low sensitivity to changes in fluid properties; hence it does not require cumbersome and expensive subsea sampling. Despite this low fluid property sensitivity, the MPM metering concept also has a unique in-situ fluid property measurement on board to further reduce fluid property sensitivity to nil. 1 Introduction In the late 80 s the oil and gas companies started to develop, through extensive R&D activities, Multi-Phase Flow Meters (MPFM s) and Wet Gas Flow Meters (WGFM s), often in Joint Industry Projects with a particular vendor. The main drive here was to reduce costs by replacing complex and bulky test separators and to further simplify the upstream infrastructure. MPFM s and WGFM s are an order of magnitude lower in CapEx 1 and OpEx 2 than a fully equipped test separator and using these meters makes it possible to greatly simplify the production facilities with the removal of dedicated test lines in satellite developments. In addition, for offshore developments there is a significant benefit with the weight and space savings that can be achieved with the much smaller MPFM s and WGFM s. For subsea installations it is obvious that the use of test separators is not feasible at all and the use of MPFM s and WGFM s to measure the multi-phase and wet gas streams are the preferred option. Over the past decade some of the MPFM s and WGFM s have developed from prototypes into very mature, robust, advanced and field proven measurement devices. For subsea gas fields the measurement of water (but also condensate and gas) is relevant for two main reasons. The first is the production and sales allocation of the individual gas, condensate and water production to their respective sources (see 1.1) and the second is the flow assurance (see 1.2). 1.1 Production Allocation in a Complex Production Infrastructure The first applications for MPFM s were mainly for well testing and well/reservoir allocation. But with the steadily increasing experience and the more and more advanced technology, the application area for MPFM s has been extended from the well testing business to applications where the output of the meter actually determines money flow between oil companies or between companies and host government (e.g. sales allocation, transportation fees, custody transfer and royalty payments). The move of MPFM s into this application is a consequence of the fact that the upstream oil and gas business is becoming more complex in terms of 1 2 Capital Expenditure, costs for purchasing and installing a MPFM, this includes all hardware to operate the MPFM (sampling arrangements). Operating Expenditure, costs to operate a MPFM this includes all costs to maintain and operate the MPFM (including verification processes and sampling for fluid properties). Subsea Controls DownUnder 1 of 23

2 infrastructure, i.e. facilities are being shared between various producing companies and commingling is done much further in the upstream process. Today we see multi-phase production streams from different companies being commingled at 3 km of water depth and these applications call for a new measurement approach and different philosophies for allocating the condensate, gas and water to the respective sources (see Figure 1). In those complex subsea infrastructures the use of MPFM s is very beneficial from a project economics point of view but at the same time the use of these meters creates a few challenges from an operational point of view. As an example, in subsea applications it is expected that these MPFM s will continue to operate for a very long time without the need to retrieve the meter or to access the meter for maintenance, verification, sampling or calibrations. There exists a large diversity in applied physical concepts, which makes it difficult to select an adequate MPFM for a particular development. It is often difficult to see the wood from the trees and judge all the future operational effects on the MPFM in a remote and inaccessible location. One issue that is not well documented and still creates uncertainty in the long-term operation is the effect of changing fluid parameters on the oil/condensate, water and gas flow rate readings. It will be explained further below that mismanagement of these fluid properties can have a significant impact on the cash flows between operating companies or operating companies and host government and may jeopardise the flow assurance and eventually block subsea pipelines due to hydrate formation. Figure 1, MPFM s or WGFM s installed at the seabed at the entry of a commingled subsea pipeline will ultimately determine the money flow between the companies A, B and C. 1.2 Flow Assurance Aspects Flow assurance is a most critical task during deep water oil and gas production because of the high pressures and low temperature (approx. 4 C) involved. In order to ensure a continuous production of hydrocarbons from remotely located subsea wells, management of water production is essential. Water in the production lines can cause scale and hydrate formation, which can ultimately block subsea flow lines or subsea gathering lines. The subsequent financial loss from such a production interruption or facility damage can be astronomical. In order to optimise the scale and hydrate inhibition, it is important that water production rates are accurately measured. For many fields it is also important to know the salt content of the produced water in order to prevent corrosion in the pipelines and production facilities. Next to the above flow assurance aspect, knowledge about salt content is also important from a reservoir management perspective. Discrimination between salt and fresh water can be used to identify the origin of produced water, as either condensed water vapour or produced formation water. 1.3 OpEx and CapEx Considerations Once the decision has been made to use MPFM s or WGFM s in a project, either for well and reservoir management, for sales allocation, for legal requirements or for operational control (flow assurance), the MPFM selection process is started in order to determine which flow meter is best suited for the job. Given the large diversity in commercial MPFM concepts this is not an easy task and it is obvious that each metering concept Subsea Controls DownUnder 2 of 23

3 has its pro s and con s from a measurement point of view, from an engineering point of view and from a cost point of view. The cost of purchasing, installing and operating the MPFM s or WGFM s is one of the important aspects that needs extensive upfront consideration. There is a relatively large spread in purchasing costs (dependent on pressure rating and size) and subsea flow meters are often 3-4 times more expensive than the equivalent for a topside application. However, in the total life cycle cost of a MPFM or WGFM, the purchasing, engineering and installation costs (all considered CapEx) are a relatively small part. For most applications the costs to operate these MPFM s, i.e. maintenance, verification processes, sampling for fluid properties, etc. (all considered OpEx) for the years these MPFM s are in service, is often a magnitude higher. Hence, in the MPFM selection process it is strongly recommended to consider both the CapEx and OpEx very carefully. In the end it is the total life cycle cost that should be the driver from an economic point of view, this next to the performance driver (i.e. uncertainty, repeatability and availability). Two examples, although extreme, are given below for MPFM s or WGFM s that have similar performance in terms of uncertainty: 1) A low cost flow meter with high dependency on fluid properties. The additional hardware for extensive sampling and analyses for fluid properties will increase the CapEx slightly. The actual sampling and analyses during operation will make the OpEx relatively high, in particular in applications where accessibility to the MPFM is low and thus expensive (like subsea applications). 2) A higher cost flow meter, generally higher CapEx, but if the MPFM has no or minimum dependency on fluid properties or if the flow meter has the possibility of performing in-situ fluid property measurement, it will require less operational activity to determine these fluid properties and thus the OpEx will be much lower than in point 1 above. Hence, if fluid sampling and analyses is required to properly tune a MPFM or WGFM and the location is remote (subsea) this will increase the OpEx significantly. Executing subsea sampling at 2 to 3 km water depth for this purpose is not only very cumbersome and extremely expensive but also carries a significant environmental risk. In Figure 2Error! Reference source not found. the relation for CapEx and OpEx is plotted schematically as a function of meter performance in a subsea application. Figure 2, Schematic of the CapEx and OpEx for a MPFM or WGFM with high OpEx due to regular maintenance (e.g. calibration and sampling) and a meter that can cope with changing fluid properties and thus much lower OpEx and possibly slightly higher CapEx. 1.4 Why Wet Gas Flow Measurement is Important The production of natural gas from subsea developments is often done without any further subsea processing, i.e. all the produced fluids are transported from the wells by multi-phase flow lines to downstream processing facilities, either onshore processing plants or offshore facilities (gas treatment platforms or floating LNG). As indicated in 1.2 there is a high risk of hydrate formation which eventually might permanently block the multi-phase flow line and an adequate hydrate management strategy should be in place. Accurate metering of water production rates is a key element in securing and maintaining gas production and ensuring availability of the subsea system. Subsea Controls DownUnder 3 of 23

4 Next to the above flow assurance issue there obviously is the contractual and/or legal requirements to measure the individual well (reservoir) production to facilitate proper well, reservoir and concession allocation for reservoir management, operational control and production optimisation. What do we expect to be measured with a WGFM 3 system, all in either m 3 /d or kg/d: Gas flow rate, i.e. hydrocarbon gas including all other gaseous components like N 2, CO 2, H 2 S, etc. H 2 O vapour flow rate Condensate flow rate (hydrocarbon liquid) Condensed H 2 O flow rate (has no salinity) Formation H 2 O flow rate (often contains salt components) The latter two might be replaced by : Total liquid H 2 O flow rate (sum of condensed and formation water) Water salinity This because the condensed H 2 O does not contain any salt components and if the formation water has a known salinity the fraction of condensed and formation water can be calculated from the salinity of the total liquid H 2 O. 1.5 Requirements for WGFM s in Remote Locations For all WGFM applications, with limited or no accessibility, there are 3 important operating features that a WGFM needs to have from a measurement viewpoint. Mechanical engineering aspects related to the marinisation are not further discussed in this paper. Able to cope with changing fluid properties (changing PVT) In-situ verification/performance processes Redundancy Furthermore, only so-called in-line meters are considered, meters that require an upstream flow conditioning unit (e.g. a full or partial separation unit) with all the auxiliary equipment for control purposes are generally not considered attractive in subsea applications Cope with changes in fluid properties (PVT data or salinity) In single phase (or oil/water) flow rate measurement most of the used technologies show a dependency on the fluid properties, i.e. if the actual fluid property deviates from the fluid property that is used to configure the flow computer, the measurement will start to show systematic errors. Obvious examples are a simple orifice plate measurement with a wrong gas density or a Coriolis mass flow meter for the measurement of oil and water flow rates with wrong basic densities in the flow computer. In case of the Coriolis meter the flow computer needs the base density of oil and water, this together with the measured density and measured total mass flow rate of the oil/water mixture will result in the mass or volume flow rates of oil and water. In addition to the base densities (i.e. at 15 C and kpa) the Coriolis flow computer also needs to have correlations on board to convert the base oil and water densities to the oil and water densities at actual line conditions. Figure 3Error! Reference source not found. shows this phenomenon in a graphical way. If the Coriolis flow computer contains a value of 850 kg/m 3 and 1050 kg/m 3 for base oil and water density respectively, but the actual water density is only 1020 kg/m 3, the WLR will be measured with a systematic error of approximately -13%. As can be seen in Figure 3 the error or the misreading in WLR and thus in net-oil measurement is dependent on the actual WLR. A systematic WLR error of approximately 13%, at a high WLR (low oil-cut) of 94%, results in a systematic error in the net-oil measurement of over 200%. The above Coriolis example is very simple and straight forward; it can be simulated and understood by everyone. However, in the case of MPFM s or WGFM s this phenomenon is not very straight forward and is certainly not very transparent. Calculation algorithms, the two-phase flow models, the slip models and discharge coefficient calculations are not publically available because vendors consider this as their own intellectual property. But even if these calculation routines are available, it is often much more complex to run sensitivity analyses similar to the Coriolis meter example above. This because MPFM s and WGFM s often use a number of physical measurement principles (the building blocks) to determine the oil, water and gas flow rates. And these building blocks are certainly not the nice linear relations that we see in the density-wlr relation of the Coriolis meter example. 3 In the remainder of this paper the term WGFM will be used but most of the text is also applicable to MPFM s. Subsea Controls DownUnder 4 of 23

5 It was also demonstrated in an earlier publication [1] that if small volumes of one phase needs to be measured, e.g. the water flow rate in a wet gas stream for hydrate control, that it is critical to know the fluid property of the dominant gas phase extremely well. Small errors in that gas fluid property have a large impact on the water flow rate measurement (compare with the above 94% WLR net-oil example of the Coriolis meter). Figure 3, WLR determination using the density measurement of a Coriolis mass flow meter. The blue line is the flow computer and the red line is the actual density-wlr relation. Mismatch between flow computer and actual density results in systematic errors in WLR (and thus netoil) In-situ verification From above, it is clear that a change in fluid property, without updating the fluid property in the WGFM flow computer, can create unacceptable over- or under readings. It is not only important to know what the effect of a possible change in fluid properties is but it is also important to be able to correct for this change. Note that in conventional single phase fiscal flow metering or sales allocation metering any known systematic error in the system is expected to be corrected immediately. Today WGFM s are also used in that type of service and thus a similar approach should be followed. Furthermore, a feature that will allow the operator to verify the performance of the meter in terms of software, hardware and configuration data would be very beneficial. This will contribute to the confidence that the operated has in the flow rate data. Obviously, it would be more beneficial if all this verification could be performed without any other tools or support systems than the meter itself and particularly without complex and cumbersome sampling and analyses, i.e. true in-situ verification Redundancy The name of the game in subsea metering is redundancy. Subsea WGFM s are installed for a very long service life, up to years in some cases. The MTBF 4 specified by the vendors are also in that range but with the subsea WGFM technology having been around for less than that period of years it is difficult to prove that those theoretical MTBF numbers are absolutely right. In section 3.1 it is shown that a WGFM often contains a number of physical building blocks. If one building block fails it would be very beneficial if its functionality could be taken over by another building block or a combination of other building blocks. Having redundancy on board also allows the operator to execute verification checks on the meter (like mentioned in above) as a particular measurement can be done twice with the benefit of being able to compare the two readings. For those applications with no or limited access it is strongly recommended to apply concepts that do have intrinsic redundancy on board. In addition it is recommended to use virtual metering systems as back-up. Virtual metering systems are relatively low cost and easy to install, however their uncertainty is often difficult to assess. It is the author s opinion that a virtual metering system is not a replacement for conventional measurements or WGFM systems. However, by having a virtual metering system in place in addition to the primary metering system, there are possibilities to execute all the learning of the virtual metering system in the first years of a development; this 4 Mean Time Between Failure, the predicted elapsed time between inherent failures of a system during operation. Subsea Controls DownUnder 5 of 23

6 is the period it is expected that WGFM s still work properly. If over the years WGFM s will start to fail, then this initial learning may then be sufficient to fill in the gap that is created with a failed WGFM. From the above it is clear that changing fluid parameters, in-situ verification and redundancy are all closely linked. The MPM meter, that will be further discussed in paragraph 3 to 6, has all these features on board. 2 Wet-Gas Flow Measurement Technology In the early days of wet gas flow measurement usually single phase meters, like a Venturi, an orifice plate or a V-cone meter, were used to measure a wet gas stream [see 2.2]. However, the serious limitations with this rather simple and cheap technology have initiated further R&D work and have resulted in more advanced WGFM s that are able to measure individual gas, condensate and water flow rates in a wet gas stream [see 2.3] 2.1 What is the Definition of a Wet Gas If the Gas/Liquid stream at the point of measurement is mainly a gaseous phase but with small amounts of liquids present (either condensed hydrocarbons, condensed water or formation water), this is from a measurement point of view defined as a wet gas stream. Note that reservoir engineers quite often define wet gas as a particularly fluid type that delivers liquids after downstream processing; however at a far upstream point this stream could well be single phase (only gaseous phase), provided no formation water is produced. The latter application of single phase gas clearly does not call for advanced WGFM s but can be covered with conventional single phase flow meters. However, one must be absolutely sure that at a later stage no liquids will condense at the point of measurement. Wet gas is commonly defined by terms of the so-called Lockhart Martinelli (LM) parameter [2]. This is a dimensionless number that is used in liquid/gas calculations and basically expresses the liquid fraction in a wet gas stream. Its definition is where: m Liq liquid phase mass flow rate m Gas Q Liq Q Gas ρ Liq ρ Gas LVF GVF gas phase mass flow rate liquid phase volume flow rate (at actual conditions) gas phase volume flow rate (at actual conditions) liquid density (at actual conditions) gas density (at actual conditions) Liquid Volume Fraction Gas Volume Fraction Gas/Liquid flows with a LM between 0 and 0.3 are often classified as wet gas. LM values which are larger than 0.3 are usually defined as multi-phase. Often the Gas Volume Fraction (GVF) is used in wet gas flow meter discussions but this parameter cannot be directly converted to the LM, unless the pressure (and thus the gas and liquid density) is known. Practical values for GVF s in case of a LM equal to 0.3 are around the 92-94%. 2.2 Single Phase Gas Flow Meters used in Wet Gas Applications As most of these meters are momentum meters they all show an over-reading because of the presence of liquids. The over-reading (OR) is a function of the liquid loading or in other words is a function of the LM parameter [see 2.1]. where: Δp Two-Phase Pressure drop in Two-Phase Wet Gas Δp Gas Only Pressure drop with only Gas Subsea Controls DownUnder 6 of 23

7 Figure 4 shows the over-reading as function of LM parameter [3] and the philosophy is that, provided the liquid loading is known through special techniques like the tracer dilution method or conventional test separator measurement, the over-reading can be determined and subsequently the only gas flow rate can be calculated. Note that a key-step in this measurement method is that the wetness or LM parameter can be determined. For onshore and topside applications this is feasible (although cumbersome) but for subsea applications almost impossible. Moreover, with this approach there is no indication whether the liquid is condensed hydrocarbons, condensed water or formation water; basically it is a gas/liquid measurement. Hence, where individual condensate, water and gas flow rates are required, the use of single phase gas flow meters are not recommended and more advanced WGFM s need to be used. Figure 4, The over-reading of a Venturi as function of LM parameter (wetness). If LM can be measured the overreading can be corrected to get the dry gas flow rate. However, LM measurement is cumbersome or even impossible in subsea applications. 2.3 Dedicated Wet Gas Flow Meter used in Wet Gas Applications Looking to the WGFM 5 market there is a large diversity in the technology that is used. Flow rates (gas and liquid flow rates) and composition (oil/condensate, water and gas fractions) are required to calculate the individual oil/condensate, water and gas flow rates. For flow rate determination the majority of the concepts use either Venturi or V-cone mass or volumetric measurement methods. For the oil, water and gas composition measurements each vendor uses a different combination of physical measurement concepts; electric permittivity measurements (in many different forms), single energy gamma ray absorption and dual energy gamma ray absorption measurements are the most popular ones. A recent development, the MPM meter, uses advanced permittivity-based tomography methods, to determine not only the gas/liquid distribution in the pipe but also the WLR, salinity and composition. Next to these measurements, each vendor uses algorithms to describe the fluid dynamics of the gas/liquid flow (two-phase flow models or wet gas flow models). Hence, each WGFM is a combination of different building blocks, whereby each building block needs to have some information on the base fluid properties of the oil, water and gas. And it is not only the physical measurements for flow rates and composition that require these fluid properties, also the two-phase flow models or a simple factor like the Venturi discharge coefficient have a dependency on the fluid properties. It is these flow models that are far from transparent to the users as vendors consider these flow models as propriety information. For users a WGFM is basically a black box. In the WGFM flow computer the fluid properties are often inserted at 15 C and kpa condition, subsequently they are corrected to the actual flow line conditions before the oil, water and gas flow rate calculations are carried out at flow line conditions. Subsequently these actual flow rates are then corrected to standard conditions. From the above it is clear that those fluid properties and the PVT models are a key issue in the WGFM physical building blocks as well as in the calculation algorithms and will be further discussed in 4 and 5. 5 In the remainder of this paper the term WGFM will be used but most of the text is also applicable to MPFM s. Subsea Controls DownUnder 7 of 23

8 The remainder of this paper will discuss a highly advanced WGFM, manufactured by MPM in Stavanger Norway, that not only has unparalleled accuracy and repeatability but also has unique features on board to determine in-situ the fluid properties and thus eliminate the need for fluid property sampling. 3 The MPM solution for wet gas flow measurement In 2003 MultiPhase Meters (MPM) was founded in Stavanger, Norway. MPM s business concept was to develop a high-end combined MPFM and WGFM, which at that time was a clear expressed need by the oil companies. The MPM technology is based on a completely new concept for multi-phase and wet gas metering (7 patents + licensed technology from oil companies) and was developed and qualified in several operator driven Joint Industry Projects. For each JIP project, six to nine international oil companies participated with expert advice, field experience and financial support. MPM is currently focusing on the manufacturing and delivering of the MPM meter (both for topside and subsea applications). The success of the company resulted in FMC Technologies acquiring the company in 2009 and since then MPM is a wholly owned subsidiary of FMC Technologies. The MPM meter is a combined MPFM and WGFM, which means that the meter can be software configured to operate using either its multi-phase or wet gas models or to switch automatically between the two. In addition, the Droplet Count concept [4] further enhances the range of the MPM meter models to the ultra-high GVF range (see Figure 5). The hardware part of the MPFM and the WGFM are identical and the difference between the two meters is simply the software. Figure 5, The MPM meter is a combined multi-phase and wet gas flow meter. Hardware is the same but software is different. 3.1 The building blocks of the MPM concept The MPM meter uses a combination of a Venturi flow meter, a gamma-ray detector, a multi-dimensional and multi-frequency dielectric measurement system [5], and advanced wet gas flow models [6] [7], which are combined to a multi-modal parametric tomographic measurement system. The Venturi is used to create a radial symmetrical flow condition in the 3D Broadband section downstream the Venturi, which is the natural flow condition if the pipe were infinitely long. These flow conditions are ideal when using tomographic inversion techniques. The 3D Broadband system is a high-speed Electro-Magnetic (EM) wave based technique for measuring the Water/Liquid Ratio 6 (WLR), the water salinity and the liquid/gas distribution within the pipe cross sectional area (see Figure 6). By combining this information with the measurements from the Venturi, accurate flow rates of oil/condensate, water and gas can be calculated. The measurement is based on permittivity measurements performed simultaneously at many measurement frequencies and over many measurement planes (3D) within the meter. The measurement frequencies cover a range of 20-3,700 MHz. 6 WLR is the ratio of Water over total Liquid at actual conditions, watercut is the same ratio but at standard conditions. Generally WLR is not equal to watercut. WVF is water over total volume at actual condition. Subsea Controls DownUnder 8 of 23

9 Figure 6, The 3D Broadband tomography TM measurements using a range of high frequency electromagnetic waves on multiple planes is the core of the MPM multiphase and wet gas flow meter. The measurements are used to determine WLR, composition, salinity and gas/liquid distribution in the pipe. The MPM meter has a dual mode functionality, which means that the meter it is a combined multi-phase and a wet gas meter. In wet gas mode, the MPM meter can either operate in three-phase mode or in two-phase mode. In three-phase mode, the meter measures all the fractions of the flow (oil/condensate, water and gas). Two-phase mode is used for in-situ verification purposes, and requires the Gas Oil Ratio (GOR) as an additional configuration parameter. The GOR is typically calculated based on the well composition. At ultra-high GVF, when the liquid volume is extremely small compared to the gas volume, the Droplet Count functionality significantly improves the measurement resolution of the liquid fraction. By using Droplet Count, the MPM meter can make precise measurements of minuscule liquid volumes in a GVF range where no other technology is capable of making true three-phase measurements. The method is even highly tolerant towards changes in fluid PVT properties, such as the oil/condensate and gas density and water properties. This is achieved through a patented (pending) methodology with a significantly higher resolution on mixture density, as compared to gamma based density measurements, and for which the liquid metering accuracy actually increases with increasing GVF. More information about the MPM meter can be found in reference [8], [9], [10], [11], [12], [13], [14]. 3.2 Configuration requirements All multiphase meters require a set of configuration data or PVT data describing the properties of the different flow constituents (oil/condensate, water and gas). The PVT data (fluid properties) are used for two purposes, namely: 1) Calculation of fluid properties at actual temperature and pressure conditions (temperature and pressure at the location of the multi-phase meter) which is used as input for the complex physical flow rate calculations. 2) Conversion of the measured flow rates at actual temperature and pressure conditions to the standard conditions, such as 15 ºC and kpa. The second point is common to all meters (also conventional flow metering) and is a general requirement when converting flow rates of hydrocarbons and water from one set of temperature and pressure conditions to another one. This will not be further discussed. However, the first point is different for the various metering technologies that are available as it is linked to the basic physical measurement system of a WGFM. Some technologies may require more detailed and more accurate PVT configuration data, whereas other technologies are more tolerant to variations in the PVT properties [1], [8], [9], [15]. This will be further discussed in 4 below. As a consequence, the field experience gained with one type of metering technology may not be relevant for meters using other measurement principles. Until recently there has been little guidance in the industry related to the required precision in PVT data for the various metering technologies and there has been a Subsea Controls DownUnder 9 of 23

10 perception that all MPFM s and WGFM s are equally influenced by variations in the PVT. Vendors have been challenged to be more open on the influence of changing fluid properties on their concept; and standardization bodies should provide more steering and guidelines on how to test and report this issue [16]. 4 Sensitivity of WGFM s to gas fluid properties (PVT data) In single phase flow measurement the sensitivity of a measurement for variations in other parameters like temperature, pressure, fluid properties, etc. needs to be considered in order to judge the robustness of the measurement. Often temperature and pressure are compensated for through additional measurements. The influence of changing fluid properties, like density and viscosity are often not compensated for but obviously needs to be known in order to judge the robustness of the measurement. In WGFM s this is not very straight forward and is certainly not very transparent. Calculation algorithms, the two-phase flow models, wet gas flow models and discharge coefficient calculations are not publically available because vendors consider this as their own intellectual property. But even if these calculation routines were available, it is often much more complex to run sensitivity analyses as WGFM s often use a number of physical measurement principles (the building blocks) to determine the condensate, water and gas flow rates. It was also demonstrated in earlier publications [1] that if small volumes of one phase needs to be measured, e.g. the water flow rate in a wet gas stream for hydrate control, that it is extremely important to know the fluid properties of the dominant gas phase extremely well. In other words, only small errors in the dominant gas fluid property have a large impact on the determination of the much smaller water flow rate. 4.1 Uncertainty (random errors) and systematic errors Each measurement has intrinsic measurement errors, and so have WGFM s. This measurement uncertainty is something that we, in principle, cannot influence and is given by Mother Nature. Examples are the noise in electronic signals, noise generated by electronic components and radioactive decay in case a WGFM concept is based on gamma ray absorption. Only measurements over a longer period and under stable conditions can reduce these uncertainties, but in a fast changing multi-phase flow environment these long measurements are generally not desired. The uncertainty specifications as quoted by the vendors often refer to the above mentioned uncertainty (also known as random errors 7 ). Next to these random errors, there are systematic errors. These are errors that do have a well-defined cause, i.e. something is systematically wrong. This can be a bias in the calibration, a wrong Venturi discharge coefficient, a wrong parameter in the flow computer, etc. It is common practice in fiscal or sales allocation applications, that systematic errors are removed immediately once they are detected. A wrong fluid parameter in the WGFM flow computer should also be considered as a systematic error, i.e. the meter systematically shows an over- or under-reading. Any deviation between the fluid property that sits in the WGFM flow computer and that same fluid property in the flow line will create a systematic error in the oil, water or gas flow rate measurement. Figure 7 illustrates this graphically, the uncertainty of the fluid property (x 1 to x 2 ) and the uncertainty of the WGFM will result in the uncertainty of the output of the WGFM (a 1 to a 2 ). However, what is the guarantee that the basic fluid property (x 0 ) used to configure the flow computer will stay in-line with the actual flow line fluid property over the years that the WGFM is operated. Moreover, sometimes the WGFM measures a commingled flow and then the fluid property depends on the relative contribution of each of the streams. In other situations the salinity of the production water changes because of water injection (different salinity). In particular at locations where there is no access to the meter, or access is difficult and expensive, it is extremely important to know what the impact is of a change in fluid properties. Examples are deep subsea installations where it is very cumbersome and extremely expensive to run a sampling exercise in order to determine a new value for a fluid property. Also for remote and unmanned operations or applications where sampling cannot be tolerated from a safety point of view (e.g. operations with a high H 2 S composition in the gas) it is important to make an upfront consideration regarding the actual potential change in fluid properties, the type of WGFM concept and the WGFM sensitivity to changing fluid properties. 7 Random indicates that they are inherently unpredictable, and have null expected value, namely, they are scattered about the true value, and tend to have null arithmetic mean when a measurement is repeated several times with the same instrument. Subsea Controls DownUnder 10 of 23

11 Figure 7, A change in the actual fluid property, without updating the WGFM flow computer parameters, has an effect on the output of a WGFM. This is a systematic effect, and results in systematic over- or under-reading of the individual flow rates 4.2 Dual Energy Gamma Ray Absorption technology For WGFM s using the dual energy gamma ray absorption concept, it is common to illustrate the effect of changes in fluid properties using the so-called calibration triangle. A typical calibration triangle for a dual gamma system with a 30 kev low energy and 100 kev high energy is shown in Figure 8. The 100% water, condensate/oil and gas points (in terms of count rates at the two energy levels) define the composition triangle and any point is the triangle represents a water, oil/condensate and gas composition. For this figure, the mass absorption coefficients for oil, water and gas have been calculated using the National Institute of Standards and Technology (NIST) database [17]. Decane (C 10 H 22 ) is used for condensate and Methane (CH 4 ) is used for gas and the water is saline water containing 10% NaCl by weight. Clearly, changes in the base fluid properties for condensate, water and gas will have an effect on the 100% single phase calibration points of the composition triangle. Hence, if a change in fluid property is not accounted for by updating the position of the calibration point, it will introduce a systematic error in the calculation of the phase fractions of condensate, water and gas. This will be demonstrated with the following examples: 1) Condensate and Gas with 0-30% (by weight) variations in dissolved H 2 S 2) Condensate and Gas with 0-30% (by weight) variations in dissolved CO 2 3) Water salinity variations from 0 25% NaCl (by weight) All the cases have been calculated at an operating temperature of 50 ºC and operating pressure of 80 bar and Calsep PVTSim was used to calculate the condensate and gas density based on the composition of Methane and Decane with the various concentrations of H 2 S and CO 2. Figure 8 shows the effect on the calibration triangle for a 30/100 kev Dual Energy Gamma Ray Absorption system. It can be seen that H 2 S dissolved in the oil has a large impact on the low energy calibration point for condensate, resulting in a large systematic error in Water/Liquid Ration (WLR). It can also be seen that the triangle becomes a straight line when the H 2 S content is approximately 12%. At this point the measurement concept will be unable to perform any calculation of the WLR. Similarly, dissolved H 2 S in the gas also causes the calibration point to shift. Dissolved CO 2 in the gas, on the other hand, have only a small impact. 4.3 Single Energy Gamma Ray Absorption and Permittivity technology A similar consideration, like in 4.2 above, can also be constructed for the concept applied in the MPM meter, see Figure 9. The Y-axis is the product of the density and mass absorption for the 662 kev single energy gamma ray measurement and the X-axis is the high frequency permittivity. The water permittivity will change as a function of the measurement frequency as described by the Debye Relaxation Law for calculation of the effective permittivity of water [13]. Since the MPM meter uses many simultaneous measurement frequencies, there will be many simultaneous calibration triangles. In this example, a measurement frequency of 3,000 MHz has been used. It is also worth mentioning that the fractions are not a linear function of the value of the X and Y axis and it will also be different in oil and water continuous flow. However, the calibration triangle is a useful tool to illustrate how the measurements are influenced by changes in the condensate, water and gas fluid properties. Subsea Controls DownUnder 11 of 23

12 Figure 8, Effect on the calibration triangle for a Dual Energy Gamma Ray Absorption system (30 kev/100 kev) with changes in the H 2 S, CO 2 and NaCl concentrations Figure 9, Effect on the calibration triangle for a high frequency permittivity and 662 kev single energy gamma ray absorption system with changes in the H 2 S, CO 2 and NaCl concentrations For the high frequency permittivity/660 kev gamma ray system, a 0-30% change in the H 2 S content has a marginal impact on the gas calibration and condensate calibration point. The latter will cause a small shift in both the oil/gas ratio and the WLR measurement. For the Dual Energy Gamma Ray Absorption system, dissolved H 2 S in the gas has quite a large impact on the low energy calibration point for the gas, whereas dissolved H 2 S has a very small impact on the gas point for the high frequency permittivity/662 kev gamma system. In both cases, dissolved CO 2 in the gas have a small impact [18]. 4.4 Comparison The example above illustrates the need for frequent (and expensive) fluid sampling when using a 30/100 kev Dual Gamma system, particularly for applications with changing fluid properties. The example also demonstrates that the field experience gained with one type of multi-phase meters may not be directly transferred to meters based on other metering principles. It is also worth noting that a dual gamma system which uses higher energy levels will be less sensitive to fluid properties, whereas a dual gamma system that uses lower energy levels will be highly sensitive. The fluid property (PVT) robustness of the MPM meter, applying high frequency permittivity (multi-frequency and multi-dimensional) and 662 kev gamma ray absorption measurements, is a huge advantage in those Subsea Controls DownUnder 12 of 23

13 applications with no physical access to the meter and with no possibilities to sample fluids to determine these fluid properties. Despite the fact that the sensitivity of the MPM meter is extremely low, there are additional features that are implemented in the meter and that will further enhance the performance and further reduce the sensitivity of the meter to changing fluid properties. This the in-situ measurement of fluid properties, this makes it possible to use a more generalised fluid compositions for calculation of oil and gas properties, particularly when this is combined with automatic configuration of water and gas PVT properties using the salinity and in-situ gas measurement functionality of the meter. 5 In-situ verification The MPM meter has three methods for in-situ measurement of fluid properties which greatly increases the robustness of the measurement: 1) Measurement of salinity of the water phase. This is an in-line and continuous measurement, whereby separate methods are used for water continuous conditions (high WLR), for multi-phase flow conditions, and for wet gas flow conditions. 2) Measurement of gas density and permittivity by utilizing the DropletCount (see 5.2) method to detect periods with 100% single phase gas in the meter. During these periods, the actual gas permittivity and actual gas density measurements are used to measure, verify and correct the PVT calculated values for permittivity and density (which are part of the meter configuration). A similar procedure can also be used for oil, then the oil permittivity and oil density are measured and the DropletCount function is used to detect periods of single phase liquid (gas free liquid). However, if the oil contains some water (water in-oil emulsion) this will introduce a bias in the measurement of the oil PVT data. Therefore it is always recommended to perform a manual inspection of the oil measurement to validate the measurement. 3) In wet gas, the MPM meter incorporates three different methods for measurement of the fractions and the individual flow rates of gas, condensate and water. These methods are in-line and continuous, and are performed under normal flowing conditions; a) two-phase mode with Gas-Oil Ratio (GOR) input b) three-phase mode c) three-phase mode with Droplet Count These three methods behave differently when changes are introduced in the PVT configuration data and it is this different behaviour that subsequently can be used to determine the correct PVT configuration data. More detailed information regarding the in-situ verification can be found in [19] below. In-situ verification for water and gas fluid properties is critical for accurate wet gas flow metering and is further discussed in more detail in paragraph 5.1 and Measurement of water properties The MPM meter can measure the conductivity of the produced water. The measured conductivity is subsequently converted to salinity and the water density is calculated assuming a certain type of salt (e.g. NaCl, which is dominating most production water). Based on the measured temperature and salinity of the water, the viscosity of the water can also be calculated by the meter. The measurement method is based on measurements with multiple radio frequencies and is part of MPM s patented 3D Broadband technology. The MPM meter can automatically measure the water conductivity and density in the high WLR flows, i.e. watercontinuous emulsions. For the lower WLR s, i.e. oil- or condensate-continuous emulsions, the water conductivity has little effect on the measurement uncertainty. If, however, the WLR is expected to increase during the life of the field and the flow changes from oil continuous to water continuous, it is very important to configuring a MPFM or WGFM with the correct water conductivity. In the MPM meter the water conductivity and water density are automatically and continuously measured and subsequently updated in the flow computer configuration data. This eliminates the risk of getting incorrect flow rate measurements as a consequence of incorrect configuration data. Equally important, it also eliminates the need to take and analyse samples of the produced water, in order to update the configuration data when the WLR is increasing or when the salinity of the produced water is changing. This is very valuable for unmanned and remote operations (i.e. subsea installations). The WLR at which the flow changes from oil-continuous to water-continuous depends on the application and the actual fluid properties, but normally it occurs when the WLR is in the 30-60% range. If slugging is expected, measurement of the water conductivity could be important even for lower WLR s. The reason is that Subsea Controls DownUnder 13 of 23

14 if the water comes in slugs, then the WLR during the slug can be well above the water-continuous threshold. If so, and if the water conductivity is incorrectly specified, the oil and water flow rates will show systematic errors. Another benefit of the method is that the water conductivity is measured at actual temperature conditions, thus avoiding discrepancies in the correlations/models which convert the conductivity from one temperature to another. As an example, it is common to use the conductivity at 25 ºC as a configuration parameter in the flow computer, since the water conductivity in most cases is measured in a laboratory at room temperature and converted to 25 ºC. The MPM meter requires the water conductivity at actual conditions; hence the water conductivity needs to be converted from 25 ºC to the actual line temperature. This conversion correlation/model may not be 100% correct and might introduce a secondary source of error for meters which rely on the water conductivity as a configuration parameter. This is avoided when the water conductivity is continuously measured at line conditions. The water conductivity/salinity measurement is based on a patented method using dielectric measurements carried out locally at the pipe wall, using a differential principle with one transmitting and two receiving antennas. Electromagnetic phase measurements are performed over a broad frequency range and each measurement frequency provides a separate independent equation. All the measurements are combined in such a way that the measured water conductivity represents a best fit of the measured water fraction for all the measurement frequencies, assuming that the ratio between the real and imaginary part of the dielectric constant of the multi-phase mixture is related to the ratio between the real and imaginary part of the dielectric constant for pure water. The curves in Figure 10 illustrate the basic measurement principles. Using the 3D Broadband (see Figure 6), many cross-sectional planes are measured and analysed simultaneously to determine the liquid and particularly the water content. Some of the measurements are based on frequency sweeps, which are performed in each direction with a step in the phase of the electromagnetic waves. The frequency location of a differential phase shift between two receiving antennas is related to the water fraction of the wet gas and the slope of the phase shift vs. frequency is related to the conductivity of the water fraction. An increase in water conductivity causes a decrease in the slope of the curve. The measurement is based on a differential measurement within the pipe. Hence, any discrepancies in the cables, antennas and electronics are cancelled out. The water salinity is then obtained from the measured conductivity and measured temperature. Figure 10, Water conductivity measurement principle for wet gas applications. The frequency is related to the water fraction while the slope of the phase shift vs. frequency is related to the conductivity of the water This measurement can be performed in all the 27 measurement directions used by the 3D Broadband TM system (see Figure 6). In order to maintain a high speed measurement principle, of the measurements are performed and 5-10 of the measurements are used. This is considered an appropriate trade-off between speed and number of measurements. For each measurement direction a so-called normalised S-factor is calculated. The S-factor is a number which is related to the slope of the frequency sweeps. It is defined to be one for Subsea Controls DownUnder 14 of 23

15 fresh water and deviates from one for increasing salinities. Typical, the normalised factor increases when the salinity increases. The salinity measurement is further described in reference [9], [11], [12] and [20]. 5.2 Measurement of gas fluid properties A patented (pending) method for the in-situ measurement of gas properties has been developed and implemented in the MPM meter [21]. The method uses the Droplet Count function [4] to detect short periods of time where pure gas flows through the measurement section of the meter. Alternatively the meter can be bypassed and gas filled during a scheduled shut-in of the well or during the passage of long gas slugs. Figure 11 shows the measured GVF and the liquid detection signal, called the Liquid Index, for a gas filled period in the measurement section. The yellow line is the threshold value for gas detection. When pure gas is present, the permittivity and the density of the gas are measured using the 3D Broadband section and the gamma densitometer. The Droplet Count is so sensitive to droplets that it immediately detects when condensation of liquid (due to falling temperature) starts to occur, such that the in-situ gas measurement can be halted. Figure 11, Example of measured GVF and Liquid Index during an in-situ gas fluid property measurement at the end of a test period with only gas in the WFGM. Since the 3D Broadband section performs measurements of permittivity at multiple frequencies and on multiple measurement planes, many different measurements of the gas permittivity and gas density can be made. These all should give the same result, and thus the in-situ gas measurement has a built-in quality verification function. Such measurements are also used to verify the integrity of the 3D Broadband sensors. The in-situ measurement can either be used to calculate correction factors to the input configuration gas density and gas permittivity or to adjust the composition of the well fluid and generate new look-up tables using the Calsep PVTSim routines. Two methods of the in-situ gas measurement have so far been implemented; a manual procedure and a method based on automatic update. The automatic method is well suited for applications where frequent variations in the gas properties are expected. In the manual version, an in-situ report is generated, where the in-situ measurements are documented together with a calculation of the effect any changes in the gas configuration data may have on historical measurements. A recommendation for potential corrective action is added to the report. If the operator approves the corrective actions, the in-situ measured corrections to the gas density and gas permittivity are implemented in the MPM meter, stamped with the date and time. This manual procedure ensures full traceability of any changes performed on the gas configuration data, and the procedure is particularly suited for those applications where the MPM meter is used for fiscal applications. This procedure is typically used as a part of the commissioning of the meter. Most MPM meters are preconfigured with the field PVT data prior to delivery. The MPM meter is then fit for service immediately at start-up of the wells. Following successful commissioning of the field and the individual wells, in-situ measurements can then be inspected to validate the pre-configured PVT data in the meter. Evaluation of the in-situ gas measurements may also be performed on a regular basis as part of the metering quality assurance plan. Using pre-agreed acceptance limits for the in-situ gas measurements, allows the operator to efficiently process the in-situ reports. This procedure also ensures that the operator has full documentation of the validity of the configuration data for the meter. Documentation of the integrity of the Subsea Controls DownUnder 15 of 23

16 measurements from the meter is also obtained by inspecting the historical trend of the multi-frequency and multi-directional measurements from the 3D Broadband sensor in gas. Figure 12 shows the liquid detection signal and GVF measurement for a short shutdown period. In the figure on the left side, the green and blue line is the liquid detection signal and the straight red line is the pure gas threshold value. Immediately after the shutdown, there is a short period with pure gas in the meter where an in-situ measurement is performed. After approximately 20 minutes condensation of liquid start to occur due to the temperature drop in the meter and the liquid raw signal moves above the red liquid detection threshold and thus halting the in-situ measurement. In the figure on the right it can be seen that the GVF from the MPM meter (blue line) and the reference of K-Lab (red line) after the in-situ calibration matches much better. Another example of the in-situ verification/calibration is shown in Figure 13Error! Reference source not found.. In the figure on the right side the reference WVF (red line) is approximately % abs both before and after the shutdown. As can be seen from the graph of the WVF measurement, there is a negative bias on the measured WVF from the MPM meter (blue line) prior to the in-situ measurement. This is due to the error in the input configuration data for gas causing a negative bias on the zero point for the WVF measurement. After the gas permittivity and density have been automatically corrected with the in-situ measurement, the negative bias is removed and the measurement from the MPM meter follows the reference measurement of 0.002% abs water (red line). Figure 12, In-situ measurement after a shut-in, a short period of time there is no liquid which allows gas fluid properties to be measured in-line (K-lab Nov 2008). Figure 13, WVF shows initially a negative bias and after the in-situ gas fluid property measurement this bias has been removed (K-lab, Nov 2008). Subsea Controls DownUnder 16 of 23

17 5.3 Using the three different measurement modes As mentioned above, the MPM meter incorporates three different methods for measurement of the fractions and flow rates (2-phase mode, 3-phase mode, 3-phase mode with DropletCount ) which all are influenced differently by errors in the PVT configuration data. In [8] the influence of PVT configuration parameters in wet gas mode was analysed and reported in 2-phase mode and 3-phase mode without DropletCount. In 2- phase mode operation, the gas density and GOR input mode have the largest impact on the measurement result and the 3-phase mode and 2-phase mode are influenced differently by errors in the PVT configuration data. The water fraction shows little influence by errors in the gas density in 3-phase mode operation, whereas it has a significant impact on the water fraction in 2-phase mode. Similarly, the gas density have a large relative impact on the liquid flow rate in 3-phase mode without droplet count, but shows little influence on the measured liquid flow rate when the DropletCount function is used in 3-phase mode, or when the meter is operating in 2-phase mode. All these differences in behaviour can be used to investigate the most likely source for the discrepancy and provide an estimate of corrections to the PVT configuration data used by the meter. This is done by using logged raw data within the meter, which is then used to reprocess the measurements in 2-phase mode, 3-phase mode and 3-phase mode with DropletCount for a range of gas densities and GOR inputs. The density and GOR range is typical selected such that it covers values around the PVT calculated values. The measured water fraction and liquid flow rate is plotted vs. the density and GOR range used in the reprocessing for all measurement modes, and by inspecting the interception points between the modes, it is possible to obtain an estimate of the gas density and GOR input which minimizes the discrepancy between the different measurement modes. These data can be used together with in-situ gas measurement to validate the PVT data and even trend changes in the PVT data. 6 Experiences with the MPM Wet Gas Flow Meter Over the last years the MPM was testing extensively at various test loops and various field locations around the world. In 6.1 and 6.2 below some experiences gained with the MPM meter in wet gas applications are summarized. Further information on test programmes that have been executed with the MPM meter are reported in [5], [8], [10], [11], [13] and [22]. 6.1 High pressure Wet Gas (South West Research, Texas) In Q the MPM meter was extensively tested in South West Research Institute (SWRI), Texas. The main objective was to verify the performance of the MPM Meter for multi-phase and wet gas conditions at high pressure. The test conditions were quite representative for typical wet gas fields. Another important aspect of this test was to verify that the MPM Meter, when delivered from MPM s premises and installed and configured as per the standard procedures, will perform as per specifications Figure 14, The deviation between MPM liquid flow rate and reference liquid flow rate as function of GVF Subsea Controls DownUnder 17 of 23

18 Figure 15, The deviation between MPM gas flow rate and reference gas flow rate as function of GVF Figure 16, The deviation between MPM WLR and reference WLR as function of GVF Figure 17, The deviation between MPM WVF and reference WVF as function of GVF Subsea Controls DownUnder 18 of 23

19 The entire test was performed as a blind test, i.e. the reference and MPM data were handed over to Statoil, and were only disclosed to MPM upon final completion of the entire test. A total of 50 test points were performed at wet gas flow conditions with fresh water and three different water salinities. The main conclusion drawn from the test at SWRI is that the MPM Meter performed in accordance with expectations and specifications. Graphs for the liquid, gas, WLR and WVF measurements as function of the GVF are plotted in Figure 14 to Long term unattended performance (K-Lab, Norway) Figure 18 shows the measured water fraction from a 10 MPM meter installed in the flowloop at K-Lab, Norway. The graph is for a period of 14 days in November 2008 and has been made with in-situ gas property measurements active. Initially, the water fraction varies in the range from % by volume and for the remaining 14 day period the water fraction is mainly well below 0.01 %. The pressure during this period varied from 25 to 55 bar. From the graph it is seen that the water fraction measurement tracks the small variations in the water fraction well and a stable zero point is maintained for the entire period, despite significant changes in the operating conditions (pressure and thus gas fluid properties) of the flowloop. Figure 19 below shows the same 10 MPM meter, but now in September 2009, after 10 months continuous operation with in-situ gas measurements. The setup configuration data of the meter (containing the fluid properties) has been untouched during this entire period despite significant changes in the pressure and the continuous changes in the gas composition due to frequent loading and discharge of the gas in the flowloop. Figure 18, The WVF of a 10 MPM meter and the reference measurements as measured during a 14 day period in Nov 2008 at K-lab, Norway. Insitu fluid property measurements and Droplet count are used. Figure 19, Same 10 MPM meter as in Figure 18 but now 10 months later. Significant changes in gas properties and pressures occurred but with the in-situ verification and the DropletCount the MPM meter is able to cope with that. Subsea Controls DownUnder 19 of 23

20 The pressure varied from 28 to 74 bar during the test in September 2009 without any noticeable effect on the zero point on the water fraction measurement. The water fraction measurement of the meter is generally within ±0.02% abs of the K-Lab reference measurement for the entire GVF and pressure range with a zero point stability well within 0.002% abs. Another example with the in-situ fluid property measurement is demonstrated in Figure 20, here the sensitivity of GVF is plotted and it can be seen that this is better than 0.002% absolute. Similarly the zero measurement of the WVF measurement is approximately 0.001% absolute, thus an extreme small bias. However, a change in WVF of only 0.002% can be easily detected. The last example shows the GVF measurement during May 2009 as reported by Statoil [23]. It can be clearly seen that after 6 months of operation, without changes in the meter configuration, the GVF measurement is remarkably well in line with the K-lab reference measurement (Figure 21). Figure 20 An example of the sensitivity in GVF and WVF of the MPM meter compared with the reference measurements in K-Lab, Norway Figure 21 Example of the GVF measurements with the MPM meter after 6 months of operation with without any manual configuration changes but with in-situ fluid property measurements. Subsea Controls DownUnder 20 of 23