EMRP 2014 Metrological support for LNG custody transfer

EMRP LNG D1-4-2 Annual report May 2015 1/26

CESAME-EXADEBIT S.A.

43 route de l’Aérodrome

F. 86036 Poitiers Cedex

Tél. : 33(0)5 49 37 91 26

Fax : 33(0)5 49 52 85 76

E-mail : [email protected]

EMRP 2014 Metrological support for LNG custody transfer

and transport fuel applications ENG60 LNG

Studying of

Laser Doppler Velocimetry (LDV) for LNG flow measurement

(Task 1.4.2 WP1) R.MAURY

A.STRZELECKI Y.LEHOT O. VALET J.P. VALLET May 15th, 2015


The research leading to the results discussed in this report has received funding from the European Metrology Research Programme (EMRP). The EMRP is jointly funded by the EMRP participating countries within Euramet and the European Union


Summary

1. Summary of the Join Research Program 4 2. Main objectives of the LDV test campaign 5

3. Means of measurements 5 3.1. Description of the reference facility for flowrate measurements at Cesame Exadebit 5 3.2. Description of the simplified DN80 Cryogenic LDV Measurement Package 6 3.3. Description of the seeding system 8 3.4. Velocity measurements by means of the Laser Doppler Velocimeter 8

4. Velocity measurement in pipe by means of LDV 9

5. Test matrix during the experimental campaign: measurement / acquisition of test parameters (Pressure, Temperature, Flow velocity, reference flow rate) 10

6. Treatment of the raw LDV data 11 6.1. Volume measurement size 11 6.2. Power reduction of the laser through two portholes 11 6.3. LDV velocity profile decomposition into regions 11 6.4. LDV filtering method 12

6.5. LDV validation criterion 13

7. Similarity of the velocity profile between two different campaigns (2012 to 2015) 14 8. Analysis of the LDV measurements 15

8.1. Method 1: Radial integration of the velocity profile 15 8.1.1. Positive velocities 15

8.1.2. Negative velocities 16 8.1.3. Radial integration 16

8.2. Method 2: Single point measurement of the axial velocity 18 9. Conclusion 21 10. References 22

11. Appendix A 24


1. Summary of the Join Research Program Background Liquefied Natural Gas (LNG) is a strategic, and in the case of long distances, a more economical alternative for pipeline gas. It is also used to transport natural gas from and to locations where no pipeline infrastructure exists. After regasification of the liquid form, the natural gas is transported to the main users: power plants, industry and households. Recently the use of LNG as a cleaner transport fuel has been added to the list of important applications.

Need for the project In comparison with other commodities like natural gas or gasoline the total uncertainty of measured energy is high for LNG and has been estimated to be up to 1 %. The current lack of direct traceability to the SI also leads to the delayed introduction of new measurement methods in the LNG business. Therefore, a sound metrological framework is an indispensable element for the development of LNG as transport fuel, which is one of the pillars of the EU clean fuel strategy.

Scientific and technical objectives The aims of the JRP are to further develop the metrological framework for LNG, to contribute to a reduction of the measurement uncertainty of LNG custody transfer by a factor two (starting from 1 %) and to enable the development of LNG as a clean transport fuel.

Therefore, the JRP addresses the following objectives:

WP1 will develop and validate novel and traceable calibration standards of LNG mass and volume flow for vehicle fuel dispensing and ship bunkering. A mid-scale LNG mass and volume flow facility up to 200 m3/h (90 tons/h) will be built and validated. This standard will be traceable to the previously developed primary flow standard. It will furthermore be cross-validated using a new Laser Doppler Velocimetry (LDV) based standard. From the results a new ISO standard will be drafted (as part of WP5) to pave the way for industry wide adoption of the LNG flow measurement technology.

WP2 will develop and validate novel and improved methods for measuring LNG composition to address the online monitoring of the LNG quality and issues with sampling LNG. A LNG composition calibration system will be developed and integrated into the LNG flow facility. It will be cross-validated with a newly developed reference LNG liquefaction system. The latter will also be specially designed to validate Raman spectroscopy systems.

WP3 will develop a publicly available method for the determination of the methane number, including a correlation of the methane number to the LNG composition, in support of the use of LNG as transport fuel. From the results a draft ISO standard will be created (as part of WP5) to harmonise the definition and measurement methods used by gas engine manufacturers worldwide.

WP4 will validate and improve models for LNG density prediction and associated uncertainty evaluation. A new set of measurement data will be created with an uncertainty 5 times lower than existing data and an improved correlation between density and composition will be established. A cryogenic Speed of Sound (SoS) measurement system will also be developed to establish an improved correlation between SoS and density.

WP5 is designed to maximize the impact for the JRP, by disseminating the results and by strengthening links to industrial stakeholders.

WP6 describes the management of the JRP. The selected methods will ensure an effective and efficient project management.


Cesame Exadebit in the JRP: A very promising alternative to the state-of-the-art static volume measurements is the dynamic principle of flow metering. WP1 addresses the great technological challenge of creating traceability for LNG flow meters that currently does not exist anywhere in the world. Providing a direct link to SI with a very small uncertainty and disseminating that link to a range of flows has never been done before and will be a unique achievement. The project will develop the know-how to ultimately provide traceability to the full range of LNG flows. The goal of this work package is the development of metrologically-sound traceability schemes for LNG flow metering. A novel cryogenic flow metering technology, Laser Doppler Velocimetry (LDV), will be explored as promising an alternative to ultrasonic and Coriolis flow metering. LDV as a flow measurement technology has already been demonstrated in high pressure natural gas with an uncertainty of 0.1 – 0.2 % (Dopheide et al., Metrologia 1993/1994, 30, pp. 453-469) but its extension to cryogenic temperatures is challenging and will be checked for its feasibility. LADG has performed a feasibility study of LDV technology applied to LNG flow metering during the first LNG program (2011-2014) [XXXX]. The study has been focused on the technological challenges and solutions for extending the LDV method to cryogenic temperatures, and on the estimation of the uncertainty that can be realistically achieved with such a system. This report is dedicated to the experimental campaign conducted in the Cesame Exadebit in 2015.

2. Main objectives of the LDV test campaign An experimental campaign has been carried out during the first semester of 2015 in Cesame Exadebit facility in Poitiers. The tests have been done using pressurized air from 1 to 10bar. During these experiments, the main objectives were:

• Build a validated database, treated with stochastic tools allowing results interpretation with a high level of confident.

• Validation of impermeability and vacuum level in the mock-up. • Retro diffusion validation with the full optics set of the laser (power reduction). • Results confrontation between the set of experimental data from 2012-2015 campaign. • Programming a stochastic tool in order to treat the data acquired with the Dantec system. • Fit of the axial velocity with an analytical function to get a better integration of the velocity

profile in order to get the mass flow (Qv). • Attempt to increase the Reynolds Number. • Determine the plus/minus and improvements of the proposed method of calculation.

3. Means of measurements 3.1. Description of the reference facility for flowrate measurements at Cesame Exadebit The pressurized calibration facility for medium and high flowrates at CESAME EXADEBIT can generate flowrates from 8 m3/h to 80000 m3/h (normal conditions). A set of twelve Venturi nozzles (nominal flowrate: 1.5 to 1000 m3.h-1.bar-1) operating in sonic conditions is used for the determination of the standard mass flowrate. The test pressure range is from 1 bar up to 45 bar (absolute). Compressed dry air stored in a 110 m3 vessel under 200 bar (absolute) is used as the test fluid. The air coming from the storage vessel goes through the valves and the heating control system. This one


adjusts the suitable temperature and pressure upstream the nozzles automatically. The pipe lines bear the reference nozzles chosen according to the flow patterns to be generated for the tests. The longest testing pipeline is 50 m long with nominal diameters from DN25 up to DN300. The meter under test is placed on a pipeline downstream the set of nozzles. This configuration allows a comparison between the reference and tested device mass flows. The pressure and the temperature can be measured at the level of the meter in test in order to determine the volume flowrate going through. The real gas effects are taken into account by applying compressibility factor corrections to the thermodynamic conditions where the measurement is taken. A set of control valves placed downstream the tested instrument allows adjustment of the suitable back pressure for calibration [between 1.5 to 10 bar (absolute pressure) for these tests]. This operation is automated and controlled from a board located in a room near the test rig. Data acquisition and calculations are performed by an automatic computing system located in the same control room. These nozzles are traceable to National Standards by mean of a (P, V, T, time) method.

Fig. 1 – Diagram of the calibration facility for medium and high flowrates at CESAME EXADEBIT

3.2. Description of the simplified DN80 Cryogenic LDV Measurement Package The model is composed of three parts:

- The cryogenic seeding part - The conditioning part (containing the convergent) with the measuring cross-section - The divergent part.


The seeding part is equipped with an access for the seeding probes in cryogenic conditions, and with two windows for particles visualization. The conditioning part is provided with windows which allow passage of laser beams for measuring the velocity profile at the exit of the convergent. The downstream part of the Cryogenic LDV Measurement Package contains the divergent. These three parts are located inside a vacuum chamber to ensure thermal insulation. The entire model is equipped with pressure and temperature taps:

- Upstream the convergent (P, T) - Throat of the convergent (P) - Downstream the divergent (P, T).

Fig. 2 – (a) Description of the simplified Cryogenic LDV Measurement Package and (b) 3D model – horizontal cut

Full sketch of the experimental setup in Cesame Exadebit facility:

Fig. 3 – Sketch of the experimental setup in Poitiers (2015)

- P1 : Upstream pressure of venturi (RPM4) - P2 : Downstream pressure of venturi Turbine G250 - T1 – T3 : Upstream and downstream temperature ( cryogenic PT100 probes) - ∆P1 : Upstream pressure – pressure at the nozzle exit - Laser Dantec - focal 160 mm - NI card to acquire the PT100 probes + P2


3.3. Description of the seeding system For these tests with air under pressure up to 10 bar seeding is done by generating micronic particles of DHES Di (2-ethylhexyl) sebacate, sebacic acid (photo below) 8D upstream the Cryogenic LDV Measurement Package.

Fig. 4 – Seeding System operating up to 10 bar maximum working backpressure

3.4. Velocity measurements by means of the Laser Doppler Velocimeter The velocity profiles are measured by means of a Laser Doppler Velocimeter DANTEC (Figure 6) in the backscattering mode (Figure 5) with the following specifications:

- Wavelength of the laser line = 532 nm green line of a frequency doubled Nd:YAG laser - Focal length = 160 mm - System configuration = backscattering mode - Data acquisition and signal processing = DANTEC BSA Flow Software - Traverse system controlled from the PC running BSA Flow Software for laser

displacements - Size of the measurement volume: l = 0.0496 mm and L = 0.4105 mm - Interfringe spacing = 2.217 µm.

Fig. 5 – Backscatter configuration


Fig. 6 – LDV System in operation

4. Velocity measurement in pipe by means of LDV Two methods are examined to determine the volume flowrate from velocity measurements performed by the LDV:

1. Integration of the velocity profile measured downstream of the throat, 2. Calculation of the volume flowrate from a local velocity measured downstream of the throat.

For the first method, the volume flowrate is obtained by integrating the flow velocity across section S where R is the limit of integration:

R

0V v(r)rdr2πQ

For the second method, the basic idea of the previous designed convergent is to get very symmetrical velocity profiles to allow a very repeatable and fast profile measurement. Once the boundary layer has been measured, the volume flowrate measurement can be reduced to a single point measurement (center line velocity measurement):

v R πQ 2

V

For a given Reynolds number, the output velocity is given by the relation

2

V

dπ4Qv


and the ratio between the velocity on the axis vaxis (measured by the LDV system) and the output velocity v

is a constant function of the Reynolds number Red

)A(Rev

vd

axis

with

dμπQ4

μρdvRe m

d

These relations allow the calculation of the volume flowrate from the velocity measured at one point downstream the throat on the axis of the pipe.

)A(RevRπvRπQ

d

axis22

V

To implement this second method, it is necessary to establish the correlation function between the volume flowrate and the local velocity measured and the influence of the Reynolds number.

5. Test matrix during the experimental campaign: measurement / acquisition of test parameters (Pressure, Temperature, Flow velocity, reference flow rate) The mean and the RMS velocity profiles are measured in a specific cross-section downstream the throat of the convergent of the Cryogenic LDV Measurement Package for each condition below:

Upstream convergent Pressure

Nominal Velocity at the throat of the convergent without

seeding

Reynolds number in the pipe

(D = 80 mm)

Mass flowrate through the sonic

nozzles

P

bar(a)

v

m.s-1

ReD

Qm

kg.s-1

2.0

11.30 5.95E+04 0.0338

21.38 1.13E+05 0.0642

59.75 3.12E+05 0.1764

5.0

10.80 1.44E+05 0.0813

21.13 2.81E+05 0.1590

58.82 7.61E+05 0.4323

10.0

10.54 2.78E+05 0.1582

20.51 5.40E+05 0.3062

58.53 1.52E+06 0.8620 Table 1: Measurement conditions


During this campaign, several parameters have been tested. Indeed, influence of two portholes in the optical path has been studied (light power reduction, path modification…) and two kinds of measurements have been realized to ensure the feasibility of both methods to calculate the mass flow:

- Radial profile of the axial velocity : for profile integration – Method 1 - Axial velocity on the centerline : for velocity at the nozzle exit – Method 2

6. Treatment of the raw LDV data 6.1. Volume measurement size The size of the volume measurement can be a source of uncertainties. Indeed, it has to be correlated with the measured velocity gradient. The dimensions are calculated with:

sin (𝛳

2) = √(𝐹² +

𝐷2

4) 𝑠𝑢𝑐ℎ 𝑑𝑦 =

4𝐹𝜆

𝜋𝐷𝑓 cos (𝜃2)

If we consider the laser and focal dimensions, the volume measurement gets these dimensions:

- 𝑑𝑥 = 0.049 mm (tiniest side of the volume) - 𝑑𝑦 = 0.41 mm (larger side of the volume)

During the campaign, the shear layer is about 7.5mm for the thinnest configuration. It means that the optic is correctly defined for our experiments since there is a magnitude order between 𝑑𝑦 𝑎𝑛𝑑 𝛿𝜔. We can get about 15 points to determine the velocity gradient. It is possible to increase the number of points in the sheared region by modifying the focal length (80 mm for example, 160 mm in our case) but it reduces the velocity number and the Reynolds number. We have the opportunity to use the 360 mm focal length to increase the velocity number but the size of the volume measurement increases as well. The separation scale is not so clear by using this second optic and it may provide several measurements mistake. Indeed, 𝑑𝑦 = 1.5 mm (larger side of the volume) in this case when 𝛿𝜔 = 7.5mm.

6.2. Power reduction of the laser through two portholes During this test campaign, the experimental setup is close to the real condition of utilization because two portholes are used to mimic the laser path (vacuum chamber between the portholes). The power of our laser is enough to perform accurate LDV measurements trough this configuration of windows. 6.3. LDV velocity profile decomposition into regions The LDV data need to be treated in order to avoid bias. Indeed, stochastic tools can be helpful to eliminate unphysical value in a set of data. The LDV software gives the opportunity to delimitate areas in which parameters can be tuned as requested: velocity center, velocity span, min and max record length, the sensitivity and the gain of the LDV signal. The figure 7 above gives an example of the benefits to do such decomposition of the flow in the sheared region.


Fig. 7 (a) – LDV measurements without any region

(b) – LDV measurements with region It is clear that the velocity profile decomposition improves consequently the quality of the results. The RMS is particularly smoothed with this technique. Adjusting parameters of the “burst waveform” give a higher level of thrust in the acquired results.

6.4. LDV filtering method The LDV velocity histograms are Gaussian. It means that the probability density looks like the figure above. It is possible to filter the raw data by removing values considered out of the relevant boundaries. Indeed, if a 𝛿𝜎 is applied with (𝛿=3), 99% of the data are conserved and 1% are considered as false.

Fig. 8 (a) – 𝛿𝜎 LDV filtering and (b) data removed by the filter


As an example, the figure 9 (a) above presents the temporal signal of the filtered and unfiltered LDV measurement on the centerline axis whereas the figures (b) and (c) show the Mean and RMS velocity profiles of both filtered / unfiltered configurations.

Fig. 9 – (a) Temporal signal, (b) Mean velocity profile and (c) RMS velocity profile for both filtered

and unfiltered configurations 6.5. LDV validation criterion

The LDV system gives the opportunity to stop the data acquisition by two manners. Indeed, you can give a number of samples that have to be acquired or you can limit the time of acquisition if the data rate is too low. In both cases, we need a validation criterion to go to the next point of the experimental test matrix. There is a validation criterion in the software regarding the number of “burst” which are kept or removed. Cesame decided to add another criterion based on the convergence of the mean and RMS value of each measured points. The driven equations for this convergence are reminded below:

𝑒𝑟(𝑥) = [∑ (𝑈(𝑥)

𝑁𝑒) − 𝑈𝑚𝑜𝑦]

𝑁𝑒

𝑖=1

𝑒𝑟(𝑥) = [√∑ (𝑢(𝑥) − 𝑈𝑚𝑜𝑦)²

𝑁𝑒

𝑖=1

𝑁𝑒− 𝑈𝑟𝑚𝑠]


If the convergence is monotonic and lower than 1% during the last 10% of samples, we consider the acquisition point has validated. The figures below show the convergence of mean and RMS profile at a singular point in the shear layer for 10000 samples.

Fig. 10 – (a) Mean velocity convergence (b) RMS velocity convergence over 10000 samples

7. Similarity of the velocity profile between two different campaigns (2012 to 2015) First of all, we need to find the centerline of the jet extent. To do this, we did a radial velocity profile and we determine aerodynamically the jet center. The figure above presents the center research and the correction that we made to determine the jet axis.

Fig. 11 – (a) Determination of the jet axis (b) Corrected velocity profile

Now, we want to compare the velocity profile of the 2012 campaign with these results. We fix parameters such the velocity and the upstream pressure and realize the measurement. The experimental process is identical to the last campaign. The figure below presents the comparison of the axial velocity for 57 m/s and 5 bar. The curves have been dimensionless for the velocity and the (X,Y) directions.

0

5

10

15

20

25

0 10 20 30 40

Velo

ctiy

[m/s

]

Y [mm]

Sér…

0

5

10

15

20

25

0 10 20 30 40

Velo

city

[m

/s]

Y [mm]

Série1

∆y=1mm


Fig. 12 – Velocity profile comparison between two different campaigns (2012-2015) at 5bar, 57m/s

and x/D=0.3 We can notice that the number of points in the 2015 campaign has been considerably increased in order to get a better definition of the shear layer. The radial extend has also been increased to get physical event that occurs in the optical path.

8. Analysis of the LDV measurements 8.1. Method 1: Radial integration of the velocity profile

8.1.1. Positive velocities Cesame wants to find an analytical function that can fit the experimental velocity profile in order to get a higher resolution for the integration method. Based on Maury analysis, Cesame proposes to fit the experimental profile with two functions – one for the positive velocities and one for the negatives velocities. The figure below presents the experimental velocity profile and the fit of the positive velocities.

Fig. 12 – Experimental velocity profile and analytical fit of the positive velocities

-0,4

-0,2

0

0,2

0,4

0,6

0,8

1

1,2

-3 -2 -1 0 1 2 3 4 5

Ve

loci

ty [

U/U

j]

r/R

Essais Octobre 2014

Essais Octobre 2012

October 2015

October 2012


The analytical function for the positive velocities can be calculated with the equation below:

𝑓(𝑥) = [𝑑 (1 + (1 + 𝑐(tanh(𝑥)2)) tanh (𝑏 ((0,5 + 𝑎

𝑥) − (

𝑥

0,5 + 𝑎))))]

From our perspectives, this analytical function represents correctly the main aerodynamic features of the flow.

8.1.2. Negative velocities The experimental profile of the axial velocity shows a negative area around r/D>1.4. This location where the analytical fit does not match with the experimental data is named as ∆. The analytical function that fits this velocity profile is defined as:

𝑓(𝑥) = 𝑎𝑥3 + 𝑏𝑥2 + 𝑐𝑥 + 𝑑 The figure below presents the fit of the experimental results in the negative velocities for 20m/s and 2 bar of upstream pressure.

Fig. 13 – Experimental velocity profile and analytical fit of the negative velocities

The R² coefficient gives in indication about the accuracy of the fit. If it is close to 1, we consider that is fit is good enough for the negative values of the velocity. Finally, a β coefficient has to be determining in order to match both analytical expressions in terms of magnitude. Indeed, the location ∆ where the fit of the positive velocity divergences from the experimental data is not the same for every experiments.

8.1.3. Radial integration The volume flowrate is obtained by integrating the flow velocity across section S where R is the limit of integration:

Vel

ocity

[m/s]

∆+r/D


R

0V v(r)rdr2πQ

R is the last point in the optical path measured by the LDV system. It is around 1.9 (r/D). The integration has been realized by two manners: a classic method and Simpson integration. The results are presented in the figure below for 𝑅𝑒 = 2,6. 105. Fig. 14 – (a) Experimental velocity profile and analytical fit, (b) Comparison of the mass flow rate

with two integration method The figure shows that the fits are correctly calculated and they represent the main aerodynamic features of the jet development. The integral calculation shows that the Simpson approach is more representative to our flow than the rectangle integration. The results are better with the Simpson integration and the comparison with the mass flow from the standard facility is close to the one calculated with the LDV system. The difference is around 2% for this Reynolds number. NOTE: There is an important assumption made with this calculation. Indeed, the optical path is not 2pi periodic. If we want to correctly calculate the mass flow with accuracy we need to make correction due 3D model shape. This procedure has been realized for the test matrix defined in table 1 and the graphics are visible in appendix A.

𝑸𝒗𝒎𝒖𝒄𝒌−𝒖𝒑 93,55 95,14

𝑸𝒗𝒔𝒕𝒂𝒏𝒅 97,05

1,037 1,02

𝑅𝑒 = 2,6. 105

Vel

ocity

[m/s]

Analytical function Exp

Simpson integration

Fit of experimental data


The graphic 15 below is a synthesis of all the results presented above and in appendix A. It is a representation of the Qv ratio versus Reynolds Number.

Fig. 15 – Qv ratio as a function of Reynolds number for all Air test in CESAME

8.2. Method 2: Single point measurement of the axial velocity The goal of this method is to only do one point measurement to determine the correlation function A for every Reynolds Number. The single point measurement is taken in the potential core which is “quasi irrotational”. As a reminder, the mass flow can be calculated using:

)A(RevRπvRπQ

d

axis22

V

and

)A(Rev

vd

axis .

The Reynolds number is increased from 1.4E+5 to 1.5E+6. The correlation function is defined in this range of Re due to facility limitation. In order to correctly determine the velocity at the throat, an axial velocity profile has been made to insure that the potential core was irrotational. The figure 16 presents the evolution of the axial velocity at 5 bar for 57 m/s. We can see that there is a slight increase of the velocity when the probe is getting close to the throat (2m/s - +3%). This phenomenon can be probably explained by the low contraction ratio of the convergent.

0,85

0,875

0,9

0,925

0,95

0,975

1

1,025

1,05

1,075

1,1

1,125

1,15

0,0E+00 5,0E+05 1,0E+06 1,5E+06 2,0E+06

Qv

rati

o

Re

Méthode 1

Int simpson

10 bar where fit was not optimal


To consider the correct velocity at the throat (where LDV cannot measure due to beams convergence), a fit has been defined. It is a polynomial expression of the second order.

Fig. 16 – Axial velocity profile on centerline at 57 m/s – 5 bar

Thanks to this measurement, the comparison of the axial velocity on the centerline can be related to the mean velocity given by the standard facility. All the results are summarized in the table 17 below:

Essai PMaq Re-col Papp-ref Tapp-ref Qv-App-

ref Qv-Etal tuy*

Qv-ref-condition

maquette(col) vitesse

débitante Vaxe Vaxe/Vdéb

Nb bar (capt-10 bar) °C m3/h m3/h m3/h m/s m/s

1 4,9969 1,44E+05 4,9950 18,15 48,95 45,016 48,968 10,80 11,160 1,034

2 4,9964 2,81E+05 4,9904 17,80 95,68 91,746 95,808 21,13 21,670 1,026

3 4,9949 4,14E+05 4,9825 17,20 140,51 136,106 140,908 31,07 31,730 1,021

4 4,9936 5,41E+05 4,9719 19,28 186,30 182,003 187,236 41,28 42,040 1,018

5 4,9903 7,61E+05 4,9473 19,56 264,02 258,897 266,756 58,82 59,120 1,005

6 9,9946 2,78E+05 9,9937 19,82 47,78 45,295 47,810 10,54 10,840 1,028

7 9,9929 5,40E+05 9,9842 20,17 92,88 90,221 93,022 20,51 21,035 1,026

8 9,9910 7,94E+05 9,9694 22,01 138,37 135,499 138,785 30,60 31,120 1,017

9 9,9877 1,06E+06 9,9485 21,43 184,08 181,327 185,040 40,80 41,250 1,011

10 9,9801 1,52E+06 9,9021 19,56 262,56 258,899 265,448 58,53 59,210 1,012

Fig. 17 – Synthesis of the centerline measurement with the LDV package

The ratio is then characterized for a Reynolds number range from 1.44E+05 to 1.52E+06. The ratio (Vaxis/V) is around 1.02 in average. By using this correction, we can tune the mass flow measurement to accuracy fit the mass flow of the standard facility using equations:

)A(RevRπvRπQ

d

axis22

V and )A(Re

vv

d

axis

y = -0,0052x2 - 0,1382x + 58,214R² = 0,9991

56,5

57

57,5

58

58,5

59

59,5

-15 -10 -5 0 5 10

Ve

loci

ty [

m/s

]

Position X [mm]

Polynomial fit at 57 m/s - 5bar

Exp

Poly. (Exp)


The figure 18 is a synthesis of the evolution of the A(Re) coefficient as a function of Reynolds number.

Fig. 18 – Synthesis of the evaluation of the correlation function which allows to access mass flow In this graph, it is clear that there is a slight divergence between two set of experiments. Indeed, in order to increase the Reynolds number, the upstream pressure has been increase from 5 to 10 bar. The correlation function A(Re) is not exactly of the same order. For example, at Reynolds number 5E+05 there is a difference of 0.015. We need to add the error bars for each point in order to determine if there is a common range of correlation function to determine a general law for the flow meter. It is important to add that the uncertainties are going to significantly drop with the increase of the Reynolds number because the sheared velocity region is going to be reduced. Furthermore, those experiments have been carried out with air. In a close future, we are going to use cryogenic fluid which provides a significant reduction of the sheared region (superfluid). This phenomenon is going in a good direction for the accuracy of our flow meter.

0,950

0,970

0,990

1,010

1,030

1,050

0,00E+00 2,00E+05 4,00E+05 6,00E+05 8,00E+05 1,00E+06 1,20E+06 1,40E+06 1,60E+06

Qm

rat

io

Re

Qv ratio as a function of Reynolds number

10 bar 5 bar

The trend is mostly the same between – decreases of ratio with increase of Re. However, there is a difference with upstream pressure

Need to add error bar on this plot


9. Conclusion A novel cryogenic flow metering technology with Laser Doppler Velocimetry (LDV) has been explored as an alternative to ultrasonic and Coriolis flow metering. CESAME EXADEBIT (LNE-LADG) performed a feasibility study of LDV technology applied to LNG flow metering which is directly traceable to SI units. The study focused on the technological challenges and solutions for extending the LDV method to cryogenic temperatures, and on the estimation of the uncertainty that can be realistically achieved with such a system. The cryogenic LDV measurement system was manufactured and allows the performance of measurements in the LNG unloading conditions on a test bench in the laboratory CESAME with a substitution fluid (dry air). Similarity of the Reynolds number was used to simulate the LNG flow conditions with pressurized air at CESAME. This experiment has been realized with the objectives to:

1. Build of validated database, treated with stochastic tools which allow results with a high degree of confident.

2. Validation of the permeability of the mockup for the vacuum needs. 3. Validation of the backward scattering with the two set of portholes. 4. Results consistency between two different experimental campaigns. 5. Programming statistic tools to post process the raw data acquired with the Dantec system. 6. Try to determinate an analytical function to fit the axial velocity profile in order to access the

mass flow. 7. Increasing the Reynolds number.

This study has developed an expertise on the velocity measurement by LDV for cryogenic applications which can be directly used to characterize the flow calibration facilities for LNG which are currently under development within the framework of this project. Furthermore, this technique can also be used to study the influence of installation conditions (coplanar and no coplanar elbows, valves, pressure reducers...) on flowmeters used with LNG. Finally, the cryogenic LDV measurement system is traceable to SI units (time and length) and use a different technology to the flowrate references (mass and time) used in future calibration facilities for LNG.


10. References [1] Symposium on long range and short range optical velocity measurements. Saint-Louis, 15-18 sept. 1980. ISL Report 117/80 [2] International symposium on applications of laser anemometry/techniques to fluid mechanics. Lisbon (Portugal), juillet 1982, 1984, 1986, 1988, 1990, 1992, 1994, 1996 [3] International laser anemometry symposium. Miami, 17-22 nov. 1985. ASME-FED, vol. 33 [4] International conference on laser anemometry: advances and applications. Manchester 1985, Glasgow 1987, Swansea-Wales 1989, Cleveland-USA 1991 [5] International specialists meeting on the use of computers in laser velocimetry. Saint-Louis, 18-20 mai 1987. ISL Report 105/87 [6] Congrès francophone de vélocimétrie laser. Marseille 1988, Meudon 1990, Toulouse 1992, Poitiers 1994, Rouen 1996 [7] Laser velocimetry. Livre VKI Lecture series 1991-05. Institut Von Karman, Belgique [8]Possibilités d’étalonnages du CETIAT – Accréditation COFRAC n°2-58 - http://www.cofrac.fr/annexes/sect2/2-58.pdf [9] A. Pedišius, V. Janušas, and A. Bertašienė - Low Air Velocity Measurement Characteristics’ Variation Due to Flow Regime - World Academy of Science, Engineering and Technology 40, 2008 [10] Possibilités d’étalonnages du CNAM/INM-LNE – http://inm.cnam.fr/ [11] Possibilités d’étalonnages du LNE – Accréditation COFRAC n°2-35 - http://www.cofrac.fr/annexes/sect2/2-35.pdf [12] Germany’s new Optical Primary National Standard for Natural Gas of high pressure at pigsar™, Harald Müller, Volker Strunck, Rainer Kramer, Bodo Mickan, Dietrich Dopheide, Hans-Jürgen Hotze, Physikalisch-Technische Bundesanstalt (PTB), Bundesallee 100, 38110 Braunschweig *E.ON Ruhrgas AG, Haltener Str. 125, 46284 Dorsten [13] Uncertainty of NIST Airspeed Calibrations- 2008 – www.nist.gov [14] LNG Custody Transfer Handbook Third Edition (2010) [15] http://www.elengy.com [16] G. Bewley - Using frozen hydrogen particles to observe rotating and quantized flows in liquid helium – Thèse - (Dec 2006) [17] M. Paoletti - Visualisation of superfluid helium flow, august 2008 [18] M. Paoletti - Experimental Characterization of trubulent superfluid helium, 2010


[19] N. Fdida, B. Patte, J.B. Blaisot, E. Rouland and V. Bégin - Velocity measurements of bubbles in liquid nitrogen - ILASS – Europe 2010, 23rd Annual Conference on Liquid Atomization and Spray Systems, Brno, Czech Republic, sept 2010 [20] B. Rousset, D. Chatain, L. Puech, P. Thibault, F. Viargues, and P.E. Wolf, Visualization in Cryogenic Environments: Application to Two-Phase Studies, Cryogenics, Vol.49, p.554-564, 2009 [21] A. Melling - Tracer particles and seeding for Particle Image Velocimetry, oct 1997 [22] http://www.oxinst.com - Windows for Cryogenic environnments [23] http://encyclopedia.airliquide.com/encyclopedia.asp [24] http://www.dantecdynamics.com [25] A. Boutier, H. Royer - Visualisations et mesures optiques en aérodynamique - Techniques de l’Ingénieur, traité Mesures et Contrôle - R 2 160


11. Appendix A In this section, the experimental velocity profiles and their analytical fit are presented. The results are presented for an increase of inlet pressure and velocity.

-4

-2

0

2

4

6

8

10

12

14

0,0 0,5 1,0 1,5 2,0

Ve

loci

ty [

m/s

]

r/D

2 bar - 10m/s

Exp

fct analytique

-15

-5

5

15

25

35

45

55

65

0,0 0,5 1,0 1,5 2,0

Vit

ess

e [

m/s

]

r/D

2 bar - 57 m/s

fct analytique

Exp

𝑅𝑒 = 5,3. 104

Exp Analytical function

𝑅𝑒 = 3,1. 105



-4

-2

0

2

4

6

8

10

12

0,0 0,5 1,0 1,5 2,0

Ve

loci

ty [

m/s

]

r/D

5bar - 10 m/s

Exp

fct analytique

-20

-10

0

10

20

30

40

50

60

70

0,0 0,5 1,0 1,5 2,0

Ve

loci

ty [

m/s

]

r/D

5 bar - 57 m/s

Exp

fct analytique

𝑅𝑒 = 1,3. 105


𝑅𝑒 = 7,7. 105



-15

-5

5

15

25

35

45

55

65

0,0 0,5 1,0 1,5 2,0

Ve

loci

ty [

m/s

]

r/D

10 bar - 57 m/s

fct analytique

Exp


𝑅𝑒 = 1,6. 106

Documents

EMRP 2014 Metrological support for LNG custody transfer