See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/262116805 Turbulent Flow Simulations Through Tarbela Dam Tunnels Considering the Effect of Sediment Particles Conference Paper · July 2010 DOI: 10.1115/ESDA2010-24201 CITATIONS READS 2 48 2 authors: Muhammad Abid Adnan Noon COMSATS Institute of Informaton Technology, Wah Catt, Pakistan Kyungpook National University 251 PUBLICATIONS 1,253 CITATIONS 11 PUBLICATIONS 80 CITATIONS SEE PROFILE Some of the authors of this publication are also working on these related projects: Reservoir flow simulation View project Erosion Wear predictions for turbo-machinery in industry and power plants View project All content following this page was uploaded by Adnan Noon on 24 February 2017. The user has requested enhancement of the downloaded file. SEE PROFILE Proceedings of the ASME 2010 10th Biennial Conference on Engineering Systems Design and Analysis ESDA2010 July 12-14, 2010, Istanbul, Turkey ESDA2010-24201 TURBULENT FLOW SIMULATIONS THROUGH TARBELA DAM TUNNELS CONSIDERING THE EFFECT OF SEDIMENT PARTICLES Muhammad Abid Faculty of Mechanical Engineering GIK Institute of Engineering Sciences & Technology, Topi 23640, NWFP, Pakistan Tel:92-938-271858, Fax:92-938-271889, abid@giki.edu.pk ABSTRACT The sediments inflow in the Tarbela reservoir is resulting in reduction in water storage capacity and damage to the tunnels carrying water to the power generating units and ultimately to the plant equipment. The turbulent flow in tunnel-2 of the Tarbela Dam Project (TDP) is analyzed for the sediment particles tracking in ANSYS CFX. The particles flow in the tunnel is considered random and striking the tunnel walls at different impact angles. Langrangian particles tracking approach is used for the particles deposition, which is a oneway coupling phenomena as the carrier phase is unaffected by the sediment particles because of particle mass loadings less than 0.2 and unaltered flow field. Reynolds Stress Model (RSM) is used for turbulent modeling as it takes care of anisotropic effects near the tunnel walls and is used in flows with strong curvature, swirling flows, flows with strong acceleration/retardation. Keywords: Tunnels, Turbulent flow, Sediment particles, Turbulent modeling, Tarbela Dam NOMENCLATURE CD dp E FD g k m n p drag coefficient particle diameter erosion rate density drag force gravitational acceleration erosion model constant particle mass erosion model constant fluid pressure Adnan Aslam Noon Faculty of Mechanical Engineering GIK Institute of Engineering Sciences & Technology, Topi, NWFP, Pakistan Tel:92-938-271858, Fax:92-938-271889, maangiki@yahoo.com Rep p f Uc Up U β γ ρf ρp τp μ ν particle Reynolds number particulate phase volume fraction fluid phase volume fraction fluid velocity particle velocity particle relative velocity particle mass loadings particle impact angle fluid density particle density particle response time fluid dynamic viscosity fluid kinematic viscosity INTRODUCTION Tarbela Dam Project comprises of six tunnels, three of which are used for power generation and three for irrigation purposes. Tunnel number 2 is analyzed which is used for power generation. The maximum (in summer) and minimum (in winter) pool levels are 1550 ft and 1300 ft respectively. This tunnel is about half a mile long 858.7 m (2820 ft). The gross head is 444 ft. Intake portion is at an elevation of (1225 ft) above the sea level. The straight portion is placed at an elevation of (1112 ft). The diameter of water at inlet and outlet portion is 10.96 m (36 ft) and 4.87 m (16 ft) respectively. The average pressure difference between two levels i.e. ΔP is 950 kPa. The water is discharging at an average flow rate of 978.63 m3/s. The construction material is concrete with steel liner placed inside the concrete. The tunnel water is divided into six branch pipes which end at the turbine inlet section. This paper presents a computational fluid dynamics (CFD) based erosion prediction model and its application to water flow 1 Copyright © 2010 by ASME in the tunnel at different critical locations of the tunnel geometry specifically at the main bend, straight portion and branch pipes. This comprehensive procedure consists of three major components: flow simulation, particle tracking, and erosion calculation. The effect of the particle rebound model on the particle trajectories as well as erosion pattern in the main bend, straight portion and branch pipes is also investigated. PARTICLE SELECTION TRANSPORT MODEL During the analysis, considering wide particle size distribution, built-in erosion models in CFX were integrated with the Lagrangian particle tracking routines, and one-way coupled Eulerian-Lagrangian model was selected [1] for the current study. The choice of one-way or full coupling for sediment particles depends on the mass loading i.e. β and is defined as the ratio of the particulate mass per unit volume flow [2] to the fluid mass per unit volume flow and is expressed as; β= rp ρ p (1) rf ρ f If is less than 0.2, one-way coupling is used. Therefore oneway coupling of sand particles is used to visualize and analyze the particles once they enter the tunnel. One-way coupling simply predicts the particle paths during post-processing based on the flow field, but without affecting the flow field. In addition it is assumed that the particles do not interact with each other. As sediment particles concentration calculated is almost 1.0% [3], so β = 0.0255 is considered. Governing equations of fluid motion Navier–Stokes equations The governing equations of flow employed in CFX-11 are discussed in this section. The continuity and momentum equations are given in Eqs. (2) and (3), respectively ∂ρ + ∇. ( ρU ) = 0 ∂t ∂ ρU ( ) + ∇.( ρU ×U )= p σ =− ρ I+ T µ [∇U + ∇U ] ρ ( ) (4) Generally, a small, rigid spherical particle entrained in turbulent pipe flow encounters many forces, some of which can be justifiably neglected in the particle equations of motion in this study [4]. These neglected terms include pressure (buoyancy) force, virtual mass force, the Basset force, and Brownian diffusion. Gravitational settling and the Saffman lift force are also neglected. Therefore, the governing particle equation of motion is given as follows: dU p = FD (U c − U p ) dt (5) The drag force per unit mass and is defined as; FD = and τp = 1 τp CD Re p (6) 24 is the particle response time and is defined as; ρ p d p2 18µ (7) Rep is the particle Reynolds number based on the relative velocity between the particle and the carrier phase and is defined as; Re p = d p (U − U p ) ν (8) Discharge coefficient is implemented in CFX by the Schiller Naumann correlation. As the Schiller Naumann correlation is derived for flow past a single spherical particle, it is only valid in the dilute limit of very small solid phase volume fractions. EROSION MODELING (2) ( ) (3) B + ∇. − ρ u × u + σ t ∂ U is the instantaneous velocity vector, U is the mean velocity component, u is the fluctuating velocity component due to turbulence (i.e., U= U + u ). ρ u × u is the Reynolds stress; and the stress tensor, is given by In the present study, for the water flow conditions considered as the sand concentration is fairly small so that the effect of sediment particles on the carrier fluid is assumed negligible. Thus, one-way coupling method is employed to calculate sediment particles trajectories in this study [5]. Finnie erosion model is used which require the less number of model constants and is easy to implement. One-way coupling assumes that the presence of solid particles has little effect on the flow field. Usually, for a given mass of sediment, the trajectories of tens of thousands of particles that are randomly distributed at the inlet is determined to obtain statistically representative sediment 2 Copyright © 2010 by ASME impingements on the wall in order to acquire representative erosion profiles for the geometry. Impingement information, such as particle impact speed, impact angle, and impact locations, are obtained from the particle trajectory calculations. The impingement information is applied to erosion models to finally predict the erosion caused by sediment particles within the entire simulated geometry. The properties of pipe wall material as well as particle shape can be accounted for to quantify the erosion. Governing Equation to predict erosion rate is given below as; E =k ⋅ V p2 ⋅ f ( γ ) particle tracks in conjunction with a transient flow. Therefore to develop an instantaneous erosion map at several points in time the flow field was saved after every 4 sec and particle tracks were then run on each of these result files as post-processing. Length 858.70 m (9) where, 1 1 f ( γ ) = cos 2 ( γ ) , if tan ( γ ) > 3 3 = f ( γ ) sin ( 2γ ) − 3sin 2 ( γ ) , if tan ( γ ) < 1 3 In the present study the velocity power n was set to 2.0 and the constant k was set to 1.0. In the CFX implementation an overall erosion rate at each point on the surface is then found by multiplying E by the mass flow carried by the Lagrangian particle impacting the surface, and then summing over all particles. This ultimately leads to an erosion rate density variable with units of kgs-1m-2 which can be displayed in the post-processor. (a) MODELING AND ANALYSIS Flow simulation of the continuous fluid (carrier fluid) is the first step of the CFD-based erosion prediction procedure. The conservation equations (Navier–Stokes equations) for mass, momentum and fluid turbulence were solved within the commercial code CFX-11 using a finite volume technique. Convection terms in the momentum equations were discretised using a second-order accurate scheme. Detailed modeling [6] of tunnels is done in Pro-E software [Fig. 1] then model is imported into ICEM CFX for meshing [7], then mesh is imported into ANSYS CFX for detailed analysis. Computational grid of the geometry containing approximately 2855120 tetrahedral elements is used [Fig.2]. Boundary conditions at the tunnel inlet, outlet and at the wall are specified. A velocity of 7.57 m/s and a pressure of 578 kPa are specified at the tunnel inlet. A zero pressure is specified at the tunnel outlet because it is exposed to the atmosphere. The particles were assumed to be randomly distributed at the inlet and the particle velocity distribution was assumed to be identical to that for the fluid phase. Standard no-slip wall functions were applied at all solid surfaces for the fluid phase and the coefficient of restitution for the particles was left at the default value of 1.0. (b) (c) Fig. 1: Tunnel Model; (a) main view, (b) intake, (c) outlet branches The transient analysis was run for 20 sec of real time using time steps of 0.1 sec with convergence achieved in 3-5 iterations per time step. In CFX-11 it was not possible to use Lagrangian 3 Copyright © 2010 by ASME periphery where the velocity has it highest value. The pressure reduces again at the main branch and at the outlet branches. Erosion rate density profiles at the critical locations are shown in the Fig. 5 below. Highest value of the erosion rate density is 6.23 e-5 kgs-1m-2. The value of E changes abruptly at the main branch and at the outlet branches due to the high impactvelocity and impact angle at these locations. Results for velocity, pressure and erosion rate density are also tabulated in Table 1. (a) (a) (b) (c) Fig. 2: Tetrahedral Mesh; (a) inlet, (b) branch pipe, (c) outlet branches (b) RESULTS AND DISCUSSION Velocity profiles for the the critical locations (main bend, branch pipe and outlet branches) are shown in the Fig. 3 with sediment particles flowing through the tunnel. Maximum value of the velocity is 49.76 ms-1 at the inner periphery of the of the main bend. The velocity decreases after the main bend to 31.70 ms-1 and finally reduces to 18.88 ms-1 when the water flow is fully developed at 150 m from the vertical section. The velocity increases again at the branch pipe and at the outlet branches. Pressure profiles for the the critical locations are shown in the Fig. 4 with sediment particles flowing through the tunnel. Maximum value of the pressure is 803 kPa at the fully developed flow location i.e 150 m from the vertical section. The minimum pressure is found to be 33 kPa at the main bend inner (c) Fig. 3: Results at main bend; (a) velocity, (b) pressure, (c) erosion rate density 4 Copyright © 2010 by ASME (a) (a) (b) (b) (c) (c) Fig. 4: Results at branch pipe (a) velocity, (b) pressure, (c) erosion rate density Fig. 5: Results at outlet branches; (a) velocity, (b) pressure, (c) erosion rate density 5 Copyright © 2010 by ASME Table 1: Velocity, Pressure and Erosion rate density results Location Velocity (V) ms-1 Pressure (P) kPa Erosion rate density (E) 10-5 kgs1 -2 m 1 2 3 4 5 6 7 8 48 36 36 36 36 38 35 35 29 791 164 164 164 29 164 29 6.2 4.5 3.1 3.1 4.7 4.9 2.5 4.1 • • • EXPERIMENTAL VALIDATION The simulation results are validated through the experimental work [8]. The straight portion of the pipe used in the experiment is modeled, meshed and analyzed in the way the tunnel is analyzed. The comparison between the experimental and simulation results is shown in the Table 2. REFERENCES [1] Table 2: Comparison between experimental and simulation results Experimental Simulation % Difference Erosion rate 0.16 0.1473 7.94 density (E) -1 -2 g hr m [2] [3] [4] ACKNOWLEDGEMENT Authors acknowledge Pak-US project for providing financial support and TarbelaDam personnel for all technical support. CONCLUSIONS • • The commercial CFD code CFX-11 is used to investigate the reasons of highly localized erosion along critical locations including main bend, branch pipe and outlet branches. The motion of sediments particles is predicted using an Eulerian-Lagrangian approach in conjunction with a Reynolds stress turbulence model (RSM), and an erosion map is developed using the Finnie erosion model. The results are shown for eight different locations of critical importance with sediment particles flowing through [5] [6] [7] [8] 6 View publication stats the tunnel where the velocities and pressures are changing, causing the erosive damage at these locations. The erosion rate density is the maximum at the main bend and branch pipes due to several reasons like the higher impact velocity and impact angle, the lower pressure and the production of turbulent eddies. Numerical simulations as well as experimental erosion tests are performed. Comparisons show that the CFD-based erosion prediction procedure is able to reasonably predict the erosion profile and satisfactorily capture the trend of erosion with respect to the carrier velocity with an error of about 8%. The modeling was able to successfully predict the cause of the erosion and was subsequently used in the development of a flow simulation. This resulted in increased confidence in CFD as a tool in the engineering design process, rather than only to investigate problems after these are occurred. Dosanjh, S., and Humphrey, J.A.C. The influence of turbulence on erosion by a particle laden fluid jet, Wear. Vol.102, 1985, pp. 309-330. Gary BrownUse of CFD to predict and reduce erosion in an industrial slurry piping system. Fifth International Conference on CFD in the Process Industries CSIRO, Melbourne, Australia 13-15 December, (2006) Tarbela Dam Sediment management Study, TAMS Consultant Inc. Volume 2, March 1998. Xianghui Chen, Brenton S. McLaury, Siamack A. Shirazi, Application and experimental validation of a computational fluid dynamics (CFD)-based erosion prediction model in elbows and plugged tees, Computers & Fluids 33 (2004). pp. 1251–1272. Latif Bouhadji, Three Dimensional Numerical Simulation of Turbulent Flow over Spillways, ASLAQFlow Inc. Sidney, British Columbia, Canada, 1999. Pro/Engineer. Wildfire Release 4.0 © 2009. ANSYS CFX Solver Modeling Guide. ANSYS CFX Release 11.0. © 1996-2006 ANSYS Europe Ltd. R.J.K Wood and T.F. Jones Investigations of sandwater induced erosive wear of AISI 304L stainless steel pipes by pilot-scale and laboratory-scale testing. (2004), pp. 1-35. Copyright © 2010 by ASME