In heavy oil production, steam-based recovery methods are increasingly used, due to the smaller environmental impact and ability to reach deeper accumulations compared to open -pit mining. In steam assisted bitumen production a large number of injection and producer wells are required. In order to ensure optimum heating, the steam flow rate and quality (the fraction of steam in vapour phase) injected at the various wells needs to be accurately controlled. At the moment there are no adequate and cost-effective metering solutions available for measuring both steam flow and steam quality.

Vortex flow meters are well known for measuring the flow of a fluid, in particular a single phase flow. Vortex flow meters comprise a sensor body that is placed in the flowing fluid and that generates vortices at a vortex shedder that is part of the sensor body. The vortices generate a vibration of a flexible vane that is attached to the vortex shedder. The vibration at the eigenfrequency is detected by a sensor placed on or in the vane. A computer is used to process the detected vibration. International patent application WO2011/043667 for example discloses a flow meter and method in which a fibre Bragg grating is used as sensor for detecting the eigenfrequency from which the density of the fluid cab be estimated in the case of a single phase flow. Summary of the invention

According to one aspect, the invention relates to a method for determining at least one physical property of a gas-liquid two-phase flow using a vortex flow meter comprising a flexible vane, the method comprising measuring a frequency spectrum of the vane movement and is characterised by determining a gas quality of the gas-liquid two-phase flow by

- determining a vortex peak quality, and

deriving the gas quality from the vortex peak quality.

The gas may be the vapour of the liquid. An example of such two phases is a fluid comprising condensed steam (water) and saturated steam (vapour) or any other mixture of a gas and a liquid. The two phases may be mixed, for example small water droplets in air, known as mist. The two phases may be separated from each other, for example a layer of water flowing at the bottom of a pipe, often referred to as stratified flow, and the pipe further comprising gas.

The gas quality is the ratio between mass fraction of the gas versus the total gas and liquid flow. For example the amount of water in air. The same holds for steam. Steam quality gives the ratio in mass flows between saturated steam and the total mass flow of water (w) and saturated steam (ss). The steam quality changes when water condenses out of steam as P, T conditions change.

In a vortex flow meter, the vortices generate a characteristic vortex frequency, also known as a characteristic vortex shedding frequency. This frequency depends on design criteria of the flow meter, in particular the sensor body including the vane, more in particular geometry and dimensions of the sensor body, and on physical properties of the fluid flowing around the sensor body. In addition to its maximum, the vortex peak can be characterised by its peak quality, viz. by a parameter that relates to the shape of the peak, in particular to the sharpness of the peak. The peak quality may for example be defined as the ratio between the frequency at which the peak has its maximum value and the width of the peak at 3 cLB below the maximum.

An advantage of determining a vortex peak quality is that the inventors have found that the vortex peak quality is indicative for the gas quality of a flow comprising two phases. In particular they have found that the vortex peak quality is indicative for the amount of steam in a two phase flow comprising water and air at elevated temperatures.

An embodiment of the method comprises determining the vapour flow velocity by

detecting the vortex shedding peak frequency, determining the liquid area of the gas flow,

calculating the gas flow velocity using the vortex shedding peak frequency and the liquid area of the gas flow

An advantage of using the vortex shedding peak frequency also for determining the gas flow velocity is that with one single measurement, viz. detecting the vortex shedding peak frequency, both the gas flow and the gas quality can be determined.

An embodiment of the invention comprises determining the gas flow velocity by

detecting the eigenfrequency of the vane,

determining the liquid area of the gas flow,

calculating the gas flow velocity using the

eigenfrequency and the liquid area of the gas flow. This embodiment provides an alternative for using the vortex shedding peak frequency or it may be used in combination with using the vortex shedding peak frequency in order to improve the accuracy and/or reliability of the flow measurement.

Further, the method may comprise a step of detecting the eigenfrequency of the vane for calculating a gas/droplet mixture density, without using a determined liquid area of the gas flow. In a preferred embodiment of the method the vortex flow meter comprises an optical sensor for measuring the vane movement.

An advantage of using an optical sensor in comparison to other strain measuring sensors, for example electrical sensors, is that optical sensors have a low loss in signal over long distances allowing powerful processing methods to be used and making it possible to do signal processing remotely. Furthermore optical fibres are hardly prone to electromagnetic interference that may obscure the electrical measurement or the transfer of measurement data by electronic means. Further, the use of optical signals is preferred in environments where there is a high risk for ignition of fire or an explosion, for example due to the presence of highly explosive gasses.

In an embodiment of the method the flow meter comprises a strain sensitive fibre optical sensor, the sensor preferably comprising a grating such as a Bragg grating.

In a fibre optic sensor, the optical fibre is not only used for transmitting a signal but also for transducing a response into an optical signal. Advantageously, a gas/droplet mixture density may be calculated using the detected eigenfrequency of the vane if the vortex peak quality is above a set quality threshold, applying an accurate measurement technique when a liquid area in the gas flow has no significant

contribution to the measurement.

Further, the gas quality may be derived from the vortex peak quality if the vortex peak quality is below a set quality threshold, then switching to a measurement technique that may be more accurate when the liquid area in the gas flow has a significant contribution to the measurement. If the vortex peak quality subsequently increases to above the set quality threshold, the method may include then switching again to calculating the gas/droplet mixture density using the detected

eigenfrequency of the vane.

The vortex peak quality threshold may be set depending on experimental data interrelating the contribution significance of liquid area in the gas flow to vortex peak quality values. A vortex flow meter comprising an optical fibre, more in particular an optical fibre comprising a Bragg grating, has several advantages in comparison to other strain measuring sensors. The signal processing can be done remotely based on the unreduced signal, allowing powerful processing methods to be used. No local electronics which makes the meter in particular suitable for high -temperature environments and hazardous zones. A further advantage of using optical fibre technology is that multiple sensors, for example multiple Bragg gratings, can be read out with one single fibre. Using fibre optics to determine the movement of the vane is also advantageous because optical fibres have a low weight and a limited effect on the movement of the vane. Brief description of the figures

Figure 1 shows an embodiment of a vortex flow meter, Figure 2 shows the sensor body of the flow meter,

Figure 3 shows an example of a frequency spectrum of a vane, Figure 4 shows schematically water hold up in a pipe comprising the flow meter,

Figure 5 shows experimental data of Vortex Peak Quality plotted against liquid hold-up and gas quality, and

Figure 6 shows experimental data of a vane eigenfrequency plotted against mixture (droplets and gas) and gas quality.

Detailed description of the invention

In figure 1 an embodiment of a vortex flowmeter is shown.

Preferably the orientation of the vane (l l) is horizontal but the vane may also be oriented otherwise provided that its movement is not hindered by liquid or other materials for example deposited on the walls of the pipe (12) in which the sensor body is mounted, more in particular of bottom side of the wall of the pipe. The preferred orientation with respect to the gravitational field is indicated by arrow g.

Figure 2 shows the sensor body of the flowmeter. The sensor body, also called shedder bar, comprises a bluff body (21) for generating the vortices and a vane (l l). The vane comprising a strain sensor (22). Preferably, the strain sensor is a fibre optical sensor comprising a Bragg grating. Such a flowmeter allows for example to provide an independent measurement of steam quality being the mass fraction of water in the total flow (gas mass fraction). A computer (not shown) may be used programmed with a program to make the computer process strain sensor results to measure the steam quality and perform all computations described herein. The flow meter, the program and the computer that processes the strain sensor results form a flow meter system. Secondly, the steam quality measurement can be used to correct the over reading, commonly observed in conventional vortex meters when used in wet gas flows. Applying the correction results in much more accurate wet gas flow measurements.

The fibre-optics based flowmeter is specifically suitable to give an indication of the amount of liquid present in a two phase flow besides the gas flow measurement which is also provided by conventional vortex meters. This property of the fibre optic vortex flowmeter is advantageous for use in steam-based oil recovery methods.

The sensor body of the fibre optical flow meter may comprise a triangular bluff body and a vane that is mounted downstream to the bluff body. State of the art vortex meters typically use piezo-based electronics to measure forces on a bluff body positioned in a pipe. The flowmeter preferably comprises Fibre Bragg Gratings (FBGs) instead of conventional force sensors. These consist of an optical fibre imprinted with a grating pattern in the core of the fibre. This fibre may be located inside the vane When light of a broad-wavelength spectrum is fed into the fibre, the grating reflects only light of a wavelength which corresponds to the spacing of the grating. Deformation, more in particular elongation or compression of the grating will change the spacing and hence will lead to a change in reflected wavelength. This allows the fibre to act as a strain gauge, measuring the dynamic loads imposed by the vortex shedding.

Vortex flow meters are in general used for single phase flow conditions due to the high overreacting which occurs as liquid fraction increases. The fibre optic flow meter has shown the capability of

measuring the actual flow velocity in single phase conditions irrespective of the orientation of the vortex shedder bar. In two phase flow the vortex shedder bar of the fibre optic flow meter preferably is mounted horizontally to avoid interference between the liquid film that may be present at the bottom of the pipe in which the meter is mounted. To avoid high liquid holdups or slug flow at lower flow rates it the meter may be placed in a slight downslope for example between 0.5 and 5 degrees, preferably at about 1 degree.

The vane of the shedder bar is used to pick up the vortex shedding frequency caused by the bluff body. The vortex shedding frequency f is proportional to the local velocity:

f =— D u

body wherein Sr is the Strouhal number (based on actual fluid velocity at shedder bar), Dbody the diameter of the bluff body of the sensor body, and U the superficial fluid velocity.

In figure 3 an example is shown of the spectrum obtained when taking a sample for 60 seconds. In this example, two characteristic peaks can be observed. First,

a vortex -shedding peak at f _v = 196 Hz and secondly a peak at fm = 984 Hz, which is the peak corresponding to the mechanical eigenfrequency of the vane. In state of the art flow meters, it is this eigenfrequency that is used to a determine physical property, for example the density of a flowing fluid.

The exact peak position may be obtained by a polynomial fit of the peak in order to determine the position of the peak.

The Vortex Peak Quality factor Q _vp is defined as: -3dB wherein f _v is the vortex-shedding peak frequency and AfedB is the peak width at 3dB below the peak maximum.

In figure 3, the crosses show the -3dB limits that are used to compute Afsd of the vortex-shedding peak. In this case -3dB width of the vortex-shedding peak is 12 Hz. The vortex-shedding peak maximum is at at f _v = 196 Hz, resulting in a Vortex Peak Quality factor (Q _vp ) of 16.3.

The inventors have found that there is a relationship between the vortex peak quality and the gas quality. Once this relationship has been determined for a specific embodiment of the flow meter, this relationship can be used to determine the gas quality of a two phase flow by measuring the frequency spectrum of the vane movement and determining the vortex peak quahty from the spectrum. The relationship between gas quality and vortex peak quality may be stored in a memory of the flow meter or otherwise. The stored information may comprise more than one such a relationship, for example for different two phase systems. A computer (not shown) may be used with a computer program to make the computer use the information from the memory to estimate the gas quahty from the vortex peak quality. The relationship may be a linear relationship but the relationship may also be a higher order polynomic. The inventors have found that for an air and water, more in particular air and steam, a linear relationship provides good results.

In an embodiment of the method, the gas flow is determined in addition to the gas quality. One of the steps in this method is detecting the vortex shedding peak frequency. This peak frequency may already be available in case that the gas quality has been determined first. Another step in this method comprises determining the liquid area of the fluid flow. The liquid area is the cross section area of the liquid in the pipe (12) at the position of the flow sensor. In the two phase system, a fraction of the liquid (41) may be at the bottom of the pipe as shown in figure or the liquid may be attached to the inner wall (43) of the pipe. Also shown is the shedder bar (21). The liquid area Au _q can be determined based on the properties of the fluid, for example by using specific models implemented in software such as OLGA. Such a software may provide the liquid hold up, viz. the fraction of the pipe cross section that consists of liquid. As an input for the determination of the liquid area an estimated velocity U _S g,dr _y may be used, viz.

The liquid area may also be determined otherwise, for example by a sensor detecting the presence of the liquid layer.

The gas velocity U _S g, _W et is calculated using the detected vortex shedding frequency f and the liquid area Au _q according to

The method can advantageously be applied in air-steam two phase systems. More in particular the inventors have successfully applied the method at temperatures up to 355 °C and pressures up to 140 bara. Typically the steam velocities where between 1 and 14 m/s. Steam qualities between 40 and 100 % were measured.

The inventors have found that the peak quality can be used to correct measured gas velocity for liquid hold-up because of an observed correlation between the hold-up and the steam quality on a steam-air system.

The fibre optic vortex meter can be applied in two phase flows at both low pressure, for example air-water, and high pressure, high temperature conditions, for example wet steam. The liquid holdup can be estimated from the measured steam quality and flow rate using simplified hold-up models based on multiphase flow simulations.

The method is in particular suited for determining the quality of a wet gas or steam flow. Secondly, the method can be used to correct the over reading commonly observed in many flowmeters when used in wet gas flows.

Experimental data

To evaluate the actual upstream superficial velocity and the gas quality using the spectrogram obtained with the vane attached to the shedder bar, the following procedure can be applied:

- First, determine the characteristics of the vortex shedding:

o Vortex peak frequency

o Vortex Peak Quality factor (Q _vp )

- Then estimate the velocity assuming single phase flow from Usg, dry= (fDIST)*(Atube Abody)IA _t ube, with A _tu be the tube area [m ² ] and Abody the blockage area of the meter [m ² ], the vortex shedding peak frequency [Hz], D the diameter of the shedder bar [m] and Sr the Strouhal number of the actual velocity at the vortex meter [-]. - Then determine the gas quality of the two phase flow e.g. by using the fit function of the Vortex Peak Quality factor (Qvp).

- Then evaluate the liquid correction factor based on a simplified model between the gas quahty (lambda) and the liquid hold-up (alpha-1).

- Followed by a second estimate of velocity assuming two phase flow conditions Usg, wet= (£D/Sr)*(Atube ^' Abody A]i _gu id)/Atube with Aiiquid the liquid area (alpha-1* A _tu be) [m ² ].

Results obtained in dry-air flow experiments show that the deviations between vortex flow measurements and reference flow values as a percentage of the reference flow value remain well within ±3% against the reference meter. This performance is expected of a vortex shedding flow meter in single phase flow.

Results obtained from the low pressure flow experiments in two- phase flow (air and water mixtures) show that without correction over- readings in the measured flow velocity may occur of up to 40% at increasing liquid holdup. But as long as the holdup remains below 5% the measurement error remains within ±5%.

Figure 5 shows the experimental data of Vortex Peak Quality factor (Qvp) plotted against liquid hold-up, left-hand side, and gas quahty, right-hand side. A clear relationship occurs between the liquid content in the two-phase flow mixture and the response of the vane.

The change in the peak-quality is presumably caused by interference between the liquid distribution around the shedder bar (i.e. stratified film and waves at the wall/bottom of the flow line) affecting the uniformity of the velocity profile around the shedder bar, leading to a broader frequency peak.

Figure 6 shows the experimental data of the eigenfrequency of the vane plotted against mixture density (droplets and gas), left-hand side, and gas quality, right-hand side. On the other hand, the eigenfrequency of the vane is known to be affected by the added mass of the medium surrounding the vane. For example, as more droplets are entrained in the gas flow, the eigenfrequency is reduced.

As will be described, two parameters could be used for

determining the steam quality; the Vortex Peak Quality factor and the mechanical eigenfrequency of the sensing assembly (shedder bar and vane).

It has been shown that the mechanical eigenfrequency clearly relates to the fluid density in pressurised gas flows, which make this parameter a suitable candidate to be considered for mixture -density measurements and hence steam quality measurements in highly

entrained steam flows, like mist flows.

All flow conditions were simulated with OLGA. In all steam flow tests the entrainment (the fraction liquid droplet in the steam) were predicted to stay below 50%. This means that in all cases the liquid was predominantly flowing at the wall/bottom of the flow meter in a rivulet as is illustrated in 4 (stratified flow).

A clear correlation was shown between the reference steam quality and Qvp*Usg derived from the measured data (called SQmeas). Note that correlation holds in wet-steam but gets undefined at dry-steam conditions (dots in the red circle). Higher flow velocities work in favour of SQmeas, reducing the deviations to within ±20% (absolute). Note that in the field the flow velocities are expected to be around 10 m/s (and 15 m/s as the upper range). The observation that the correlation works better at higher velocities is most likely caused by the fact that the holdup varies much more at low velocities and that at these conditions the hold-up is generally much higher.

In the case of wet-steam the flow velocity was calculated directly from the observed vortex frequency, using the expression U = f*D/Sr. In this case the over-reading of the flow velocity can be as high as 60% compared to the reference data at higher velocities (U > 7 m/s) the error reduces down to 40% due to a lower holdup.

Based on OLGA simulations it was found that at higher velocities, the liquid holdup was primarily dependent on the steam quality. Based on this knowledge correlations were developed to correct for holdup resulting in a much reduced over-reading and errors typically less 10%.

This approach and holdup correction may be applied for (conventional) vortex meters when the actual steam quality is known. However, when the actual steam quality is not known an estimated steam quality may be used based on the method outlined above.

Resulting deviations are less than +/- 10% for all flow velocities higher than circa 7 m/s whereas if no hold-up correction is applied the errors are in the order of +40%.

The flow meter yields two benefits over conventional vortex meters. Firstly, the ability to provide an independent measurement of the quality of a wet gas or steam flow. Secondly, this flow quality measurement can be used to correct the over-reading commonly observed in many flow meters when used in wet gas flows, resulting in much more accurate flow measurements. The flow meter provides measurements of the actual gas velocity and the gas quality. The latter measurement is a unique feature of this flow meter and is not possible using a conventional vortex meter.

The mechanical eigenfrequency shifts with the effective density of the surrounding fluid around the sensing element of the flow meter and can therefore be used to determine the droplet concentration which is related to flow quality. Therefore, this parameter is most useful at conditions where the entrainment rate is high (> 50%), for which we have found evidence at high velocities, in the air-water experiments.

The second parameter that can be derived from the sensing element is the quality factor of the vortex shedding peak. The quality factor is the width of the peak compared to the central frequency. It was shown that an estimate of the gas quality can be made at low

entrainment rates (< 50%). In wet steam flow, this was successful at velocities exceeding 7 m/s.

Tests demonstrated that at high steam qualities (above 98% which is 'dry' steam) errors in steam flow measurements are less than 3% for velocities above 7 /s, ^' between 2 and 7 m/s errors are less than 5%.

At lower steam qualities the errors in steam mass flow

measurement increased to 60%, similar to conventional vortex meters. This effect is due to the liquid holdup and (shedder bar) meter blockage which increase the local vapour velocity. The steam tests confirmed the ability of the flow meter to give an estimate of steam quality with errors within 20% absolute. Steam quality was estimated by making use of the Vortex Peak Quality factor only since high entrainment (> 50%) could not be reached in the test loop. At higher entrainment rates (and thus velocities than could be tested in this loop), the mechanical eigenfrequency may provide an independent measure of steam quality as demonstrated in the low pressure air-water tests.

The availability of the steam quality measurement allows further corrections to the velocity and mass flow measurement. The hquid holdup can be estimated from the measured steam quality and flow rate using simplified holdup models based on multiphase flow simulations using OLGA. Note that these simulations must be done in advance and must cover the range of conditions encountered by the flow meter. Making these corrections reduced the errors in steam (vapour) mass flow measurement to +/- 10% for higher velocities (Usg > 7 m/s).

Besides being able to measure volume flow rate in a (near)-single phase flow, the meter was specifically designed to give an indication of the amount of liquid present in a two phase flow. This was verified in both low-pressure air-water flow loop tests and in high pressure, high temperature, wet steam tests.

The experiments show that the flow meter yields two benefits over conventional vortex meters. Firstly, the ability to provide an independent measurement of steam quality - the mass fraction of steam in the vapour phase. Secondly, this steam quality measurement can be used to correct the over reading, commonly observed in conventional vortex meters when used in wet steam flows. Applying the correction results in much more accurate wet steam flow measurements. The sensor developed for the steam flow meter has a triangular vortex bluff body with a downstream vane. The latter attribute differs from a conventional vortex meter, which typically uses piezo-based electronics to measure forces on a bluff body positioned in a pipe.

In the proposed steam flow meter, the conventional force sensors are replaced by a fibre optic pick-up (Fibre Bragg Grating, FBG) embedded inside the downstream vane. Elongation or compression of the grating, caused by the induced micro-motion of the vane due to vortex shedding, will change the spacing of the grating and hence the wavelength of the back reflected light at FBG. This allows the fibre to act as a local strain gauge, measuring the dynamic loads imposed on the vane by the vortex shedding

Vortex flow meters are normally used for single phase flow.

Generally, vortex flow meters are not used in two-phase flow metering mainly because of the high over-reading inherent to these flow conditions. However, the proposed fibre optic flow meter has overcome these limitation as will be demonstrated below.

Assuming that the flow meter is used in a horizontal steam flowline, the vortex shedder bar of the fibre optic flow meter is mounted horizontally to avoid interference between the liquid film and the shedder bar. To mitigate high liquid holdups or slug flow at lower flow rates it is advised to place the meter in a slight downslope.

The vane of the shedder bar is used to pick up the vortex shedding frequency caused by the bluff body. The vortex shedding frequency is proportional to the local velocity f = Sr*U/Dbody wherein Dbody is the diameter of the bluff body. In addition, the vane is also used to determine the gas quality of the two phase flow. The gas quality lambda, or steam quality in case of a (two phase) steam flow saturated steam vapour (m _ss ) and water (m _w ) can be expressed using:

lambda=mss/(m _w +mss)

The shape of the vortex peak can be used to estimate the amount of liquid, and is described by a dimensionless value named 'Vortex Peak Quality factor' (Q _vp ) . Thus the interaction with the vane of the shedder bar and the liquid layer flowing at the wall/bottom of the flow meter can be used.

Alternatively, the interaction with the vane of the shedder bar and the mist flow can be used. The amount of this interaction can be determined by the relative change in mechanical eigenfrequency of the vane.

Atube crossectional tube area [m ² ]

Abody crossectional area shedder bar (D*ID) [m ² ]

Aiiquid cross sectional area liquid

(al* Atube) [m ² ]

D shedder bar diameter [m]

ID tube diameter [m]

mss Mass saturated steam

rriw Mass water

(condensed steam)

Sr Strouhal number (based on actual fluid velocity at shedder bar [-]

Usg superficial gas velocity upstream SmartFlow [m/s]

Us _g> dry superfical gas velocity upstream of SmartFlow meter (assuming dry gas) [m/s]

U _S g_wet superfical gas velocity upstream of SmartFlow meter (assuming wet gas) [m/s]

Qvp Vortex Peak Quality factor (peak frequency/width peak at -3dB) [-]

SQ steam quality [-]

f vortex shedding frequency [Hz]

alpha-1 liquid hold-up [-]

lambda Quality [-]