Euopean Wate 57: 253-260, 2017. 2017 E.W. Publications Expeimental and numeical simulation of the flow ove a spillway A. Seafeim *, L. Avgeis, V. Hissanthou and K. Bellos Depatment of Civil Engineeing, Democitus Univesity of Thace, Kimmeia Campus, 67100 Xanthi, Geece * e-mail: athanseaf@hotmail.com Abstact: Key wods: This pape pesents the esults of modeling the flow ove a side-mounted Ogee type spillway using expeimental and CFD (Computational Fluid Dynamics) methods, specifically Ansys - CFX. This eseach aims to use CFD model to validate the simulation of the flow ove an Ogee type spillway which is installed in the Hydaulics Laboatoy of Democitus Univesity of Thace. To achieve this pessue-based, tansient simulation of the flow which involves an inteaction between liquid wate and ai, the Volume of Fluid (VoF) and the standad κ-ε tubulence numeical models, with standad wall functions, wee used, which simulate the fee suface of two-phase flow and the flow tubulence, espectively. Thee individual meshes with diffeent density in computational cells wee ceated (coase, medium and dense) using pismatic and tetahedal elements. Numeical esults of the volume factions, velocity and wate depths wee plotted fo vaious positions acoss the spillway and wee discussed though a validation pocess by compaing the CFD model esults with the esults obtained fom physical model. Even though some slight diffeences between numeical and expeimental esults wee evealed, the gaphical tends of the numeical esults emain easonably simila fo all expeimental tests, and had a close coelation to those achieved in the physical model. Ogee type spillway, expeimental simulation, numeical simulation, CFD, Ansys Fluent, VoF 1. INTRODUCTION Thee ae plenty methods to detemine the wate flow conditions at hydaulic stuctues. One way of achieving this detemination is to conduct field measuements. Howeve, this method is time consuming and sets as a pemise the existence of the stuctue. An altenative method is the use of expeimental simulations which equie lage buildings to be installed, and the installation and the maintenance of the model might be poved expensive. Finally, the most inexpensive and least time consuming method fo detemining the wate flow conditions is the numeical simulation by using Computational Fluid Dynamics (CFD) (Kanyabujinja, 2015 and Kumcu, 2017). This method, besides the need of a poweful compute, has not any specific constaints, and is one of the most pomising methods to validate the flow conditions obseved eithe duing field measuements o duing laboatoy expeiments. The geneal pupose of this pape is the investigation of wate flow conditions ove a sidemounted Ogee type spillway by using expeimental and CFD methods. The expeimental model, built in the Hydaulics Laboatoy, Depatment of Civil Engineeing, Democitus Univesity of Thace, and the CFD Ansys Fluent softwae wee used to detemine the flow conditions ove the spillway at the dam of Bekeche steam, in Ain Temouchent povince, Algeia. The poject was tested fo thee wate flow ates, coesponding to diffeent ecuence intevals. The physical model was constucted using the similaity law of Foude Manning, and the esults fom the numeical simulation aose by using the multiphase model Volume of Fluid (VoF) in combination with the κ-ε tubulence model. 2. EXPERIMENTAL METHOD 2.1 Expeimental set-up and similaity functions Expeiments wee caied out by Avgeis et al. (2015) in a built-in model, in Hydaulics
254 A. Seafeim et al. Laboatoy of Civil Engineeing Depatment, Democitus Univesity of Thace. This poject consists of the esevoi, the spillway, the side collection canal, the tansfe canal, the stilling basin and the tailace. The expeimental model was tested fo thee flow ates; 245 m 3 /s, 500 m 3 /s and 670 m 3 /s coesponding to ecuence intevals of 100, 1000 and 5000 yeas, espectively (Table 1). The constuction of the model and the laboatoy tests wee pefomed with a 1:50 scale between the model and the pototype and in accodance with the similaity law of Foude Manning. A full geometic and dynamic similaity between pototype and model can only be achieved when the Foude and Reynolds numbes ae equal. Howeve, fo fee suface flows, the inetial and gavitational similaities could be achieved if the equality of Foude numbe is maintained and Reynolds numbe is geate than 2000. Theefoe, by choosing the equality of Foude numbe as a citeion of dynamic similaity, a satisfactoy epesentation of the wate depths, the hydostatic pessues and the flow conditions can be achieved (Bellos, 2010). To achieve this similaity, the following Equations (1) (4) wee used: Q = L 2.5 (1) V 1 2 = L (2) t 1 2 = L (3) n 1 6 = L (4) whee Q m and Q p ae the wate flow ates in the model and the pototype, espectively, Q is the flow ate scale (Q m /Q p ), L is the length scale, V is the velocity scale, t is the time scale, and n is the Manning coefficient scale. By applying the scale 1:50, the atio of model lengths to pototype lengths is: L = 150 (5) Consequently, based on the above equations, the scales Q, V and t obtain the following values: Q 1 17700 V 17 t 17 (6) (7) (8) In Table 1, the model values Q m ae also given. 2.2 Expeimental measuements Measuements of wate flow ate, wate depth and pessue wee made at seveal sections along the expeimental set-up, including the esevoi, the spillway and the stilling basin. Howeve, this pape deals only with the spillway and wate flow conditions ove it. Specifically, instantaneous measuements of the wate flow depth in the downsteam side collection canal wee pefomed.
Euopean Wate 57 (2017) 255 Table 1. Recuence inteval and wate flow ates of the pototype (Q p ) and the model (Q m ) Recuence inteval [yeas] Q p [m 3 /s] Q m [m 3 /s] Q m [l/s] 100 245 0.014 13.86 1000 500 0.028 28.28 5000 670 0.038 37.90 3. NUMERICAL METHOD 3.1 Numeical model simulation Computational Fluid Dynamics (CFD) is a compute-based tool that is used to epesent and analyze systems that involve fluid flow, heat tansfe and chemical eactions, by using numeical methods that ae based on patial diffeential equations descibing these systems (Vesteeg and Malalasekea, 2007). CFD techniques ae govened by a numbe of diffeential equations which must be solved in each contol volume, and thei fundamental pinciples emain simila to those applicable to the expeimental model. These equations include mass consevation o continuity, momentum and enegy equations, also known as the Navie - Stokes equations. In ode to tansfom a eal hydaulic stuctue into a computable model, Ansys Fluent softwae was used. Ansys Fluent is a CFD package ceated by Ansys Inc., which is based on finite volume method and uses the Volume of Fluid (VoF) methods and the standad "κ-ε" tubulence numeical models to simulate the fee suface flow and the flow tubulence, espectively (Fluent, 2009). 3.2 Geomety and mesh CFD model geomety was built with the same dimensions of the physical model (Figue 1-[1]), by using the Ansys design modele. In the geomety builde, both suface and body wee fozen to allow a fluid flow to pass though the modeling domain. The accuacy of a CFD solution depends upon the numbe of cells in the mesh. In ode to analyze the fluid flow, the domain had to be split into smalle cells within which the govening equations would be solved. The aeas of high solution inteest have been meshed up with the fine mesh, while the coase mesh was implemented in aeas with less solution inteest. The main featues of inteest in the numeical domain ae the spillway cest, the downsteam side collection canal whee the tubulent phenomenon occus, and the egion of inteface between the two phases (wate and ai). Fo the 3D model, thee individual meshes with diffeent density in computational cells wee ceated (coase, medium and dense) using pismatic and tetahedal elements (Figue 1-[2]). Accoding to Ansys Inc. (2008), the list of quality citeia fo mesh metic includes: element quality, aspect atio, Jacobean atio, waping facto, paallel deviation, maximum cone angle, skewness, and othogonal quality. In this wok, computational cell skewness has been adopted. Cell skewness is defined as the measued diffeence between the shape of the cell and the shape of an equilateal cell of equivalent volume. Highly skewed computational cells can educe the accuacy and stability of the solution. The ange of equisize skewness lies between 0 and 1 whee cell skewness should not exceed 0.98. Fo all cases, computational cells (mesh) had equisize skewness less than 0.24 which indicates a good quality of mesh. 3.3 Multiphase modeling Multiphase modeling is a technique which simulates flow in the simultaneous pesence of diffeent phases. All thee phases (gas, liquid and solid) ae identifiable with a distinct paticula
256 A. Seafeim et al. inetial esponse to an inteaction with the flow and the potential field. Thee ae fou main categoies of multiphase flows; gas-liquid, gas-solid, liquid-solid and thee-phase flows (Muone and Villedieu, 2011). Such complex systems have been made possible with the availability of two boad appoaches fo the numeical calculation, namely Eule-Eule (Volume of Fluid model o VoF model) and Eule-Lagange (Discete Phase Model) appoaches. The Volume of Fluid model (VoF) was chosen to simulate the multiphase flow. Two Euleian phases, ai and wate, wee defined as pimay and seconday phase, espectively. Figue 1. [1]: Thee - dimensional CFD model geomety, SW and SE view. [2]: Thee - dimensional CFD model mesh, SW and SE view. 4. RESULTS The investigations conducted in the physical and numeical modeling wee focused on the validation of the flow conditions ove an Ogee type spillway fom the physical modeling though a compaison of esults obtained fo equivalent flow ates of CFD modeling. In the spillway model, wate suface pofiles of each flow ate wee investigated along the spillway chute fom upsteam to downsteam of the spillway. It should be noted that the wate suface pofiles obtained along the spillway efe to the flow depth measued in the downsteam side collection canal. The wate pofiles ae pesented in Figues 2-[1], 2-[2] and 2-[3] and the constant line epesents the eadings obtained though physical modeling, while the solid fills of the chats demonstate the esults obtained though CFD modeling. Thee individual suface planes whee ceated fo the detailed display of the numeical esults. The exact locations ae 0.1, 0.6 and 1.1 metes fom the uppe side of the spillway. In the spillway numeical analysis, volume factions cases fo both pimay and seconday Euleian phases wee examined. The seconday phase (fluid tap wate) volume faction esults ae pesented in Figues 3-[1], 3-[2] and 3-[3].
Euopean Wate 57 (2017) 257 Figue 2. A compaison of wate pofiles downsteam of the spillway which wee fomed though the expeimental and numeical simulation fo flow ate: [1] 245 m3/s, [2]: 500 m3/s, [3]: 670 m3/s.
258 A. Seafeim et al. Figue 3. Numeical wate volume faction distibution (density contous) on the spillway fo flow ate: [1] 245 m 3 /s, [2]: 500 m 3 /s, [3]: 670 m 3 /s. The numeical velocity esults ae pesented in Figues 4-[1], 4-[2] and 4-[3]. The highest velocity value was 1.986 m/s.
Euopean Wate 57 (2017) 259 Figue 4. Numeical velocity distibution (density contous) on the spillway fo flow ate: [1] 245 m3/s, [2]: 500 m3/s, [3]: 670 m3/s. 5. CONCLUSIONS A study of flow inteaction with scaled model of a side-mounted Ogee shape spillway was pesented. A fully thee-dimensional CFD model was set up and the numeical model was validated against expeimental infomation. Most of the simulation esults ae found to be in easonable ageement with the expeimental ones. The outcomes of the simulations pemitted the evaluation of the designed stuctues. The study of the flow suchage at diffeent flow ates indicated that the wate suface pofiles obtained fom CFD modeling ae simila to the physical model. Numeical esults of wate volume faction and velocity ae easonable and inteelated both with wate pofiles downsteam of the spillway.
260 A. Seafeim et al. Based on the numeical esults, the design of the Ogee shape spillway was found to be appopiate fo all simulation conditions. Although physical model studies may be moe expensive and time consuming than computational modeling, they ae still cucial fo poviding data fo numeical model calibation and validation studies. The unique combination of computational expetise with physical flow modeling can be applied in ode to povide cost effective, pactical solutions to spillway flow poblems. Thus, CFD modeling can be used as a complementay tool along with physical modeling to solve complex flow poblems of spillways. REFERENCES Avgeis, L., Bellos, V., Xiloudis, N., Papalexopoulos, V., Bensasson, A. and Bellos, K., 2015. Expeimental simulation of spillway at Bekeche dam. Poceedings of the 3d Joint Confeence, 13th Confeence of the Geek Hydotechnical Association, 9 th Confeence of the Geek Committee fo Wate Resouces Management and 1st Confeence of the Hellenic Wate Association, Athens, Geece, 10-12/12/2015, pp. 403-410 (in Geek). Bellos, K., 2010. Open-Channel Hydaulics. Depatment of Civil Engineeing, Democitus Univesity of Thace, Xanthi, Geece (in Geek). Fluent, 2008. Ansys Fluent 6.3 Use's guide. Ansys Inc., USA. Fluent, 2009. Ansys Fluent 12.0 Theoy Guide, S.1. Ansys Inc., USA. Kanyabujinja, P. N., 2015. CFD Modelling of Ogee Spillway Hydaulics and Compaison with Physical Model Tests. Maste of Engineeing Thesis, Stellenbosch Univesity, Stellenbosch, Westen Cape, South Afica. Kumcu, S. Y., 2017. Investigation of Flow Ove Spillway Modeling and Compaison between Expeimental Data and CFD Analysis. KSCE Jounal of Civil Engineeing, 21(3), 994 1003. Muone, A. and Villedieu, P., 2011. Numeical Modeling of Dispesed Two-Phase Flows. Aeospace Lab. Vesteeg, H. and Malalasekea, W., 2007. An intoduction to Computational Fluid Dynamics: The finite volume method. Pentice Hall, Halow, England.