Numerical Simulation of Flow in Flowrate Measurement Section of Natural Gas Pipelines

2006Petroleum ScienceVol 3 No. 3Numerical Simulation of flow in Flowrate measurement Section ofNatural Gas PipelinesLi zhenlin and zhang yongxueaculty of mechanical and Electronic Engineering, China University of Petroleum, Beijing 102249, China)Abstract: The orifice-plate flowmeter and ultrasonic flowmeter are used widely for natural gas flowrate measurement,and the measurement accuracy is affected greatly by flow state. Numerical simulation was used to study the flow ofnatural gas in the diffusion pipe, and the length of the irregular flow induced by the diffuser or rectifier was computedSimulation results indicated that the flow in the diffusion pipe was three-dimensional turbulent flow and the steady stateflow was restored at 17 pipe-diameters downstream of the diffuser. The rectifiers equipped in the diffusion pipe showedgood rectification effect. notwithstanding the induced irregular flow, Downstream the rectifier. the flow becamesymmetrical and uniform in a shorter length than the case without a rectifier. For the diffusion pipe equipped with platerectifier, tube rectifier and tube -plate rectifier, the lengths at which uniformly distributed flow was restored were 12, 6and 5 pipe-diameters downstream the rectifier respectively. On the basis of simulation results, the minimum installationlength for flowmeters equipped in the diffusion pipe was determined. This provides a new method for improving naturalgas measurement accuracyKey words: Natural gas measurement, computational fluid dynamics(CFD), rectifier, numerical simulation, diffusionkinematic viscosity was 1.603x10 m/s. The acoustic1 Introductionvelocity in natural gas was 450 m/s. Because theIt is important to measure the flowrate of natural gascorresponding Mach number was much smaller than 0.3,during its production, transportation, and consumptithe flow in the diffusion pipe was assumed to be steadyAccurate and reliable measurement of natural gasand incompressible. Since the measurement accuracyrectly influences the economic benefit of both was the main purpose of the study, the flow in thesuppliers and consumers. Many experiments on natural diffusion pipe could be simplified as steadygas measurement have been performed based on variousflowmeters in recent years. The experimental resultsThe governing equations for steady incompressibleshow that the measurement accuracy of flowmeters is turbulent flow can be written in a unified form, i.e. theaffected by many factors. However, the state of flow in standard k-g turbulent modelthe pipeline, which is caused by the baffle, has greatera(ou)ainfluence than other factors. Except for the volumetricSflowmeter, the flowrate displayed by almost all otherflowmeters is usually calculated with empirical where o is the universal variable; u, is the velocityequations or mathematical models, which are builtcomponent, I is the generalized diffusion coefficientbased on the ideal flow. The measurement accuracy ofWhen o is 1,(u, v, w), k and E, Eq (1)becomesflowmeters is determined mainly on the agreementthe continuity equation, three momentum componentbetween the real flow and the ideal model. In this paper, equations, turbulent kinetic energy equation andthe flow in the measurement section of a diffusion pipe turbulent dissipation rate equation respectivelwith various rectifiers was analyzed with computationalIn this study, CFD Software FLUENT was used tofluid dynamics(CFD)solvegoverning equations. To improve the2. Mathematical modelcomputation stability and precision, the convectionterms of momentum equations were discretized byThe ambient temperature was 22C, and the pressure using the QUICK scheme (Leonard, 1979), and theand average flow velocity of natural gas at the inletconvection terms of k and e equations werethe diffusion pipe0.8 MPa and 8 m/s respectively. discretized by using the second-order upwind schemeThe density of natural gas was 0.68 kg/m, and the The convergence criteria were that the maximum中国煤化工CNMHGPetroleum Science2006relative errors of the continuity equation u, u, w. k determining the irregular flow region in the diffusionand e equations were less than 10pipe with various rectifiers is a key procedure forThe boundary conditions and governing equations numerical simulationtogether formed an integrated mathematical model foFor several classical baffles. their standardthe above-mentioned. Uniform velocity profile was installation lengths are determined by experiment andassumed at the inlet for simulation. The second experience. For example, for a novel orifice-plateboundary conditions were applied to the outlet velocity. flowmeter, its standard installation length of a confluentThe non-slip boundary conditions were introduced at pipe is 145 pipe-diameters of the straight pipe. Limitedhe turbulent kinetic energy per unit local space is likely to preclude installing the flowmetermass, k and turbulent kinetic energy dissipation rate this far away from the rectifier. Analyzing the flow inper unit mass, e at the inlet boundary were given the pipe with CFD would contribute to determining adirectly by the velocity profile On the solid boundary, shorter installation length and improve the measurementk and e were determined by the standard wall accuracy of flowmeterfunction( Chieng and Launder, 1980). Other boundaryconditions were Neumann boundary conditions3.1 Computational gridFig. 1 shows a sketch map of the diffusion pipe3. Numerical resultsequipped with a plate rectifier. The origin of coordinatesIn order to measure the flowrate accurately in theis at the center of the inlet. Thirty-five orifices, with adiffusion pipe with various rectifiers, the flowmeter diameter of 12 mm, are distributed symmetrically in themust be located at a specific length downstream of the plate rectitier. Computational grids were generated byrectifier, to avoid the irregular flow region where theusing algebralidcross-sections of the diffusion pipe, as shown in Fig. 2plate rectifier3000ig. 1 Sketch map of the diffusion pipe with a plate rectifier (Unit: mm)少像数慢像像(a)z=200mm(b)z=300mm(c)z=832mm(d)z=2000mmFig. 2 Computational grids for cross-sections of the diffusion pipe with a plate rectifierthe circular pipe. Fig. 3 indicates that the distribution of3.2 Diffusion pipe without a rectifieraxial velocities changed obviously in the diffuser and itsFig. 3 shows the axial flow velocity profile in the downstream region that extended approximately 5-6diffusion pipe without a rectifier. The velocity profile is pipe diameters. Flow recirculation occurred in thethe distribution of velocities in the axial direction diffuser, and with increasing flow cross-sectional area(velocity component w) over a cross section (x=0) of the velocity in the diffuser decreased. It could beH中国煤化工CNMHGVol 3 No. 3Numerical Simulation of Flow in Flowrate Measurement Section of Natural Gas Pipelinesderived from the Bernoulli equation that the pressure at the diffuser, uniformly distributed velocity was restored,the outlet increased, thus making a negative pressure so the flowmeter would be located approximately 17gradient. This is the leading cause of flow separation pipe-diameters downstream of the diffuserand recirculation. About 17 pipe-diameters downstream10.08.06.04.01.00.00.03-0.020.0100.010.020.030.060.04-0.0200.020.040.060.06-0.04-0.0200.020.040.06a)z=200mm(b)z=300mm(c)z=600mm2.01.01.01.00.0-0.06-0.040.0200.020.040.06-0.06-0.040.0200.020.040.06-0.060.04-0.0200.020.040.06(d)z=900mm(e)z=1500mm()z=2000mmFig. 3 Axial velocity profile of natural gas along the diffusion pipe without a rectifierFig. 4 shows the distribution of velocities of natural gas Downstream the diffuser, u and v decreased gradually untiin the horizontal and vertical directions (velocity they were approximately zero, where uniformly distributedcomponents u and v) over cross-sections of the diffusion flow was restored and the velocity components u and v nearpipe without a rectifier. The velocity components u and v the pipe boundary were greater than those in the centerwere relatively greater at 2=330 mm because of the diffuser. because of the viscous boundary layer1.0z=330n△△A△△△MMMA△△。z1000mmz=2000mmz=1000mm△2=2000mm-0.06-0.04-0.0200.020.040.06-0.06-0.04-0.0200.020.040.06Fig. 4 Distribution of velocity components u and v over the cross-sections of the diffusion pipe without a rectifier中国煤化工CNMHG82Petroleum Science2006through the rectifier. Uniformly distributed velocity was3.3 Diffusion pipe with a plate rectifierrestored about 12 pipe-diameters downstream theFig. 5 shows the axial velocity profile of natural gas rectifier. In other words, the flowmeter could be locatedflowing in the diffusion pipe with a plate rectifier(as a shorter length downstream the plate rectifier than theshown in Fig. 2). The axial flow velocity of natural gas diffusion pipe without a rectifier.was symmetrically distributed when natural gas passed3.01.00-0.03-0.02-0.010.010.020.0-0.06-0.04-0.0200.020.040.06-0.06-0.04-0.0200.020.040.06(a)z=200mmb)2=300mm(c)z=832mm2.53.000.06-0.04-0.0200.020.040.06-0.06-0.04-0.0200.020.040.0606-0.04-0.0200.020.040.06(d)x=900mm(e)z=1500mm(z=2000mmFig. 5 Axial velocity profile of natural gas flowing in the diffusion pipe with a plate rectifierCompared with the flow field in the diffusion pipeFig. 6 indicates that the fluctuation of velocitwithout a rectifier, the flow recirculation in the diffuser components u and v were intense at z=330 mm becausewas weaker because of the blockage of the plate of the diffuser, and slight at z=860 mm because of theectifier, which made a lower negative pressure gradient plate rectifier. Downstream the plate rectifier, thein the diffuserelocity components u and v decreased to o2330mm-A-22000mm首-0.06-0.04-0.0200.020.040.06-0.06-0.04-0.0200.020.040.06Fig. 6 Distribution of velocity components u and v over the cross-sections of the diffusion pipe with a plate rectifierH中国煤化工CNMHGVol 3 No. 3Numerical Simulation of Flow in Flowrate Measurement Section of Natural Gas Pipelines83Compared with the flow in the diffusion pipe 19 orifices(20 mm), with a total length of 220 mmwithout a rectifier, the rectification induced by the plate Fig. 8 shows the axial velocity profile of natural gasrectifier was obviousflowing in the pipe. Recirculation flow was hardlyobserved at the diffuser. About 6 pipe-diameter3.4 Diffusion pipe with a tube rectifierdownstream the tube rectifier, uniformly distributedFig. 7 shows a sketch map of the diffusion pipe velocity was restored. The distribution characteristics ofequipped with a tube rectifier. The origin of coordinates velocity components u and v in the pipe were similar tois at the center of the inlet The tube rectifier contained Fig. 6500Fig. 7 Sketch map of the diffusion pipe with a tube rectifier (Unit: mm)048-0.03-0.020.0100010.020.030.06-0.040.0200.020040.06-0.06-0.040.0200020.040.06v, m(a)z=200mm(b):=300mm(c)z=1052mmE1.000ax.020o2o04060-0.06-0.04-0.0200.020.040.06-0.06-0.040.0200.020040.06(d)z=1100mm(e)z=1600mm(f)z=2000mmFig. 8 Axial velocity distribution of natural gas flowing in the diffusion pipe with a tube rectifierrectifier was 40 mm long and the plate rectifier was 133.5 Diffusion pipe with a tube-plate rectifiermm long. The total length of the tube-plate rectifierFig. 9 shows a sketch map of the diffuswas 355Fig. 10 shows the axial flow velocityequipped with a tube-plate rectifier. The origin of profile of natural gas flowing in the pipe. About 5coordinates is at the center of the inlet. The tube pipe-diameters downstream the tube-plate rectifier, theH中国煤化工CNMHGPetroleum Science2006velocity component w over the cross section was diffusion pipe with various rectifiers were alsoalmost constant. This velocity distribution wasimulated. The three-dimensional turbulent flowfavorable for natural gas measurement. The fields in different cross-sections were not identical,distribution characteristics of velocity components u although the diffusion pipe and the rectifier wereand v were similar to Fig. 6both axially symmetricalThe flow fields over other cross-sections of thetube-plate rectifieroutlet3552000Fig9 Sketch map of the diffusion pipe with a tube-plate rectifier (Unit: mm)6.03.02.02.02.002.00.06-0.04-0.0200.020.040.060.06-0.04-0.0200.020.040.060.06-0.04-0.0200.020.040.06(a)z=300mm(b)z=832mm(c)z=900mm2.02.0s1.0-1.0-0.06·0.04-0.0200.020.040.06-0.06-0.04-0.0200.020.040.060.06-0.04-0.0200.020.040.06(d)z=1187mm(e)z=1200mm()z=1700mmFig. 10 Axial velocity distribution of natural gas flowing in the diffusion pipe with a tube-plate rectifier2) Various types of rectifiers showed obvious4. Conclusionsrectification effect, notwithstanding the induced irregular1) The flow in the diffusion pipe wflow Downstream the rectifier the flow becamethree-dimensional turbulent flow. Steady state flow wassymmetrical and uniform in a shorter distance than therestored at a specific length downstream the diffuser. To case without a rectifier. For the diffusion pipes with platemeasure the nature gas flowrate accurately, the flowmeter rectifier, tube rectifier and tube-plate rectifier, the lengthsmust be located long enough downstream the irregularat which uniformly distributed flow was restored were 12flow region induced by the diffuser. A length of seventeen 6 and 5 pipe-diameters downstream the rectifierspipe-diameters downstream of the diffusion pipe may be respectively, much smaller than the case without a rectifierrequired to obtain an acceptable velocity profileThe tube-plate rectifier showed the best rectificationH中国煤化工CNMHGVol 3 No. 3Numerical Simulation of Flow in Flowrate Measurement Section of Natural Gas Pipelineseffect, and the tube rectifier was the secondabout the first authorReferencesLi Zhenlin. born in 1967Chieng C. C. and Launder B. E.(1980) On the calculation ofreceived his m. s degree from theturbulent heat transport downstream from an abrupt pipeChina University of Petroleum inexpansion. Numerical Heat Transfer, 3, 189-2071993. He is an associate professor inLeonard B. P(1979)A stable and accurate convective modelingthe China University of Petroleumeth. Appl. Mech Eng, 29, 59-98Van Doormaal J.R. and Raithby G. D.(1984) Enhancement offluid machinery and oil gasflowrate measurement. E-mailflows. Numerical Heat Transfer, 7, 147-163zhenlinli@263.net( Received January 20, 2006)中国煤化工CNMHG