Numerical simulations of natural gas flow in pipe system with flowmeters

Numerical simulation of the flow behavior in part of large pipe system is presented in this article. Compressed natural gas is transported through the system in a dynamic unsteady way. Velocities at several points and velocity profiles at certain positions are monitored during the numerical simulation. The aim is to investigate the stability of velocity profiles at the positions of flowmeters in course of flow time. In addition, the possibility of flow conditioning in the system is presented and discussed.


Introduction
The natural gas transmission pipeline system is used for the international transportation of natural gas from gas fields over several transmission system operators (TSO) and distribution system operator (DSO) to the end user.Natural gas delivered at the entry/exit points from one to another system passes through a system of acceptance and transferal (custody transfer), i.e. it is quantitatively and qualitatively metered and measured at metering (delivery) stations.
Although today's gas market is based on trading in energy units (kWh) still volume flow metering is the basis for energy flow calculation.Thus precise gas flow metering at the station is the key to success.The most common are orifice meters, turbine meters and the latest trend the ultrasonic meters.Flow meters or metering system installed at entry/exit points are always of "class1" (metering uncertainty < 1 %, but operators shall keep it as low as possible) and operated in accordance with recommendations of international standards, e.g. in terms of using upstream and downstream straight lengths for gas flow profile development.
Fully developed flow profile, no swirls and pulsations are essential conditions for precise metering not only for orifice meters but mainly for ultrasonic meters.Although metering runs may be designed according to all standards and the meter manufacturer recommendation the gas flow behavior is always given by the configuration of the piping system up and down stream the metering run.
For this reason, the gas flow in the pipe system is investigated numerically in order to supply relevant information about any possible causes of measurement disturbances and it impacts.As a model exact configuration and dimensions of an existing metering station of an approximate capacity of 100 mio m 3 /day was used.Also tested scenarios were selected real operating conditions of the station.

Computational model
For such a piping system, 1D simulation approach is often used.However, in our case it was decided to model the piping system in fully 3D to be able to cover the gas behavior in detail in pipe branches, T-junctions, and mainly at the positions of gas flowmeters.
Computational model of part of complex metering station pipe system was created from scratch based on the design documentation and pipe system elements (Tjunctions, elbows) description from appropriate standards  Metering run itself consists of pipe DN 400 with several ball valves and pressure-flow regulating valve.Because ball valves are fully opened for several metering runs (see subsection 3 for operating scenarios) the smooth ideal pipe wall is used instead.The regulating valve is neglected for these simulations.The reason is the CAD data for detailed geometry of the valve are not available.Moreover, the valve is positioned far enough behind the flowmeters in the metering run so the influence to the velocity profiles at the flowmeters could be small, we suppose.
Individual metering runs are lead through T-junctions to main exiting pipe of DN 1200 and this main pipe is prolonged to its end after about 62 m (7).In the model, also several blind branches (8) are modeled in order to allow the 3D flow field to be simulated correctly.

Computational mesh
Computational mesh is created for each operating scenario.Operating scenarios differ in numbers and positions of used metering runs.Hence the computational model is modified for each scenario and mesh size varies about 10 million of 3D cells as it is shown in the table 1.At the picture 5 the surface mesh of the metering run and its cut is shown.You can see there is not so many cells in the metering run pipe, but much dense mesh could lead to enormous cell number of the whole geometry.The total cell mesh numbers for appropriate scenarios are shown in the table 1 below.

CFD numerical setup
Natural gas transported is treated as a pure methane in these simulations because this compound is present in natural gas by 96 %.Based on the operational pressure and temperature, methane density is determined as it is shown in the table 1.
The gas flow is simulated as a 2 nd order unsteady problem with a short steady simulation at the beginning.Turbulent flow is covered by the k-ω SST turbulence model.Spatial discretizations of 2 nd order is used.The flow is noncompressible because of relatively small transport velocities.
The computational time step for unsteady simulation is 0.002 s.The flow time of 4 s is computed to get the dynamical stable flow field.Than he unsteady data-sampling method is used during following 11 s of flow time to get the flow mean (average) values and its deviations.Also, the pressure and velocity values are monitored in many points in the system.
The huge number of small time steps to solve leads to numerically demanding simulations.For this reason, we run the CFD code in parallel using our computational Linux cluster, still the computational time counts to several days for every scenario.
During the unsteady run for last 11 s of the flow time, the velocity magnitude and velocity components were monitored in the points in certain places along the flow path in the pipe system.Here for clarity, only records of flow time from 4 s to 7 s are presented in the text below.

Mass flow in metering runs
Mass flow in kg/s for all scenarios is shown in the table 2 together with the percentage of the total flow.We can see the flow is relatively well balanced for every scenario.With higher number of metering run, the mass flow gets a bit lower, it means the system pressure loss differs for individual metering run flow streams.

Flow behavior in the pipe system
Although the flow situation differs a bit for each scenario, we can now focus only to scenario 2 results to show the main flow behavior in the complex pipe system.In this scenario, the metering runs #1, #3, and #6 are active.Behind the metering runs in the main exit DN 1200 pipe the partial streams from metering run interact again causing huge flow fluctuations, see picture 11.Another view to the exit pipe is at the picture 12. Instantaneous velocity magnitude is presented here.At the picture 13 is the average velocity magnitude field, you can see much smoother flow compared to instantaneous field.
At the picture 14, velocity magnitude deviation field is presented showing the places where the flow fluctuates a lot.The graph 15 of the velocity magnitudes course be-

Velocity profiles at flowmeters
The main interest is given to the flow situation in the metering runs at the places where flowmeters are.Also here the situation is presented only for scenario 2 because for 02054-p.4This is caused by relatively high transport velocities and relatively high gas density.The gas inertial forces are quite high and flow disturbances are traveling in the individual metering runs as it was described above.

EPJ Web of Conferences
When we look to the velocity profile in the metering run cut at the position of the flowmeter, we notice the ve- Namely from deviations pictures we can see the flow changes are located near the metering run walls.The same situation is illustrated by the picture 16, where the isosurface of instantaneous velocity with the magnitude of 22 m/s is colored in the metering run #1.The spiral character of this isosurface points to near-wall flow disturbances traveling in the metering run.

Conclusions
In this work the build-up of the pipe system model of the border station metering part is presented.The natural gas flow is simulated for given operational scenarios.
The results show the gas flow is unsteady and highly chaotic in the pipe system mostly because of the geometry changes and pipe branching.In almost all monitored points there are velocity fluctuations of low frequencies and relatively high amplitudes.Due to relatively high gas transport velocities and high operational gas density the disturbances are not calmed down and they are traveling in the pipe system [2].
That's why at the positions of the flowmeters in the individual metering runs the instantaneous velocity profiles are changing in time and are distant from the ideal velocity profiles even the average velocity profile is close to awaited ideal velocity profile.This could result into some inaccuracy in case the flowmeter awaits the ideal stable velocity profile for its measurement technique.
Next work will be focused on the method of possible velocity profile fluctuation reduction by using some appropriate flow conditioner [3], [4], [5].The real pipe system does not allow to place the intended flow conditioner to arbitrary place in the metering run, so we are a bit limited in our further investigations.
We must admit the numerical simulations are influenced by relatively sparse computational mesh to cover the flow dynamics properly.Also, the numerical models 02054-p.5 EFM 2013

Acknowledgement
The result was developed within the CENTEM project, reg.no.CZ.1.05/2.1.00/03.0088,co-funded by the ERDF as part of the Ministry of Education, Youth and Sports OP RDI programme.
[1].Only relevant parts are chosen to cover the investigated pipe system.Nevertheless the model covers the pipe system of area about 31 m × 101 m.Horizontal high differences are about 4.5 m.

Figure 1 :
Figure 1: Geometry of piping system

Figure 2 :Figure 3 :
Figure 2: Combined mesh elements type in the pipe system

Figure 4 :
Figure 4: Surface mesh near the metering run starts

Figure 5 :
Figure 5: Surface mesh of the metering run

Figure 6 :Figure 7 :
Figure 6: Velocity field at the inlet part

Figure 8 :Figure 9 :
Figure 8: Velocity field in front of metering runs

Figure 12 :
Figure 12: Instantaneous velocity in exit pipe

Figure 13 :
Figure 13: Average velocity in exit pipe

Figure 15 :
Figure 15: Velocity behavior at the outlet DN 1200 pipe

Figure 19 :
Figure 19: Deviation of velocity in metering run #1

Figure 22 :
Figure 22: Deviation of velocity in metering run #3

Table 1 :
Three operational scenarios description

Table 2 :
Mass flow [kg/s] in metering runs