WCCM V Fifh World Congress on Compuaional Mechanics July 7-12, 2002, Vienna, Ausria Dam Flooding Simulaion Using Advanced CFD Mehods Mohamed Gouda*, Dr. Konrad Karner VRVis Zenrum für Virual Realiy und Visualisierung Forschungs GmbH A-8010 Graz, Inffeldgasse 16/II e-mail: gouda@vrvis.a Dr. Reinhard Taschl AVL Lis GmbH Hans-Lis-Plaz 1, A-8020 Graz Key words: CFD, flooding simulaion, muli-phase, finie volume, conservaion equaions Absrac Flooding, be hey naural or caused by dam breaing, are a hrea o manind civilizaion all over he world. Today, he need for more and more waer sorage and elecriciy as a modern sources of energy in he densely populaed regions are compelled o build dam along large rivers around living spaces which in urn, increases he ris of more injury o he residens. According o U, beween 1963 and 1992 floods illed more people han any oher naural disaser [1]. The complexiy of he environmenal flows and he insufficien nowledge of he underlying physical mechanisms induce sill large uncerainies in boh he assessmen and policies for miigaion of flood hazard. This paper presens he VRvis flooding simulaion projec o develop a mehodology and hydrodynamics soluion based on an he AVL CFD pacage ( SWIFT) o develop flood hazard maps.
Mohamed Gouda, Dr. Konrad Karner, Dr. Reinhard Taschl 1 Inroducion This VRvis simulaion sudy aims on he developmen of a flood simulaion mehodology o evaluae hazard zone maps for he ciy areas where no hisory of flooding are documened. Ris analysis based on hese maps deermine condiions which cause flooding and define he degree of he flood ris in differen regions of he ciy. A simulaion mehodology based on Compuaional Fluid Dynamics (CFD) is presened here. The echnique was evaluaed based on hydraulics laboraory measuremen and used for a real ciy area. 2 Flood Map Evaluaion In areas wih large river basins or low land areas by he coas, floods are hazards of grea significance in he case of dam failure caused by poor design or a major even such as an earhquae desroys he flood pah. The aim of he VRVis (research cener of Virual Realiy and Visualizaion) flood simulaion projec is o develop a numerical soluion o evaluae flood zone maps as an imporan componen of a flood warning sysem. Regulaory auhoriies responsible for flood ris assessmen and flood warning, and he insurance agens are he arge requiring flood ris assessmen echniques. The resuls from his projec enable engineers o generae flood zone maps using he CFD simulaion. The simulaion resuls lead o following benefis: More accurae flood ris assessmen for civil engineering developmens; A more accurae base for calculaion of financial ris. Improved evaluaion of flood ris in a given area. More accurae flood warning sysems, greaer overall appreciaion for he impac of floods. 3 umerical Model The CFD simulaion of flow is a non-rivial echnique when dealing wih non-linear ransien problems; lie in he case, he fluid inerfaces are unnown or moves hroughou he flow, as in he case of muliphase flow processes. The Swif simulaion pacage is based on a Finie Volume approach and is based on general conservaion equaions of mass, momenum and energy, for muli-phase flows. In he Finie Volume mehod, The soluion domain is subdivided ino a finie number of conrol volumes. The conservaion equaions are applied o each conrol volume unis and he compuaional node is a he cener of conrol volume. A direc applicaion of Reynolds Theorem leads o he general form of he differenial conservaion law for an inensive propery φ [2], [3]. We consider air and waer as wo separae phases during he flooding process. I is possible o calculae he volume fracion disribuion for each phase by using he SWIFT muli-phase module. Conservaion equaions for momenum, mass and urbulence are solved for each phase. 2
WCCM V, July 7-12, 2002, Vienna, Ausria The conservaion law of mass expresses he fac ha, in a fluid sysem, mass can no disappear from he sysem nor be creaed. The conservaion of mass, or coninuiy is defined by: αρ + α ρ v = Γl 1,l (where =1,,) (1) α is he volume fracion of phase, v is he velociy phase, and Γ l represens he inerfacial mass exchange beween phases and l. The compaibiliy condiion mus be saisfied: α = 1 (2) = 1 The coninuiy equaion conains he unseady erm and he convecion erm, while he diffusion erm is no presen, because mass is ranspored only hrough convecion [4]. The differenial equaion governing he conservaion of momenum in a given direcion for a ewonian fluid wih consideraion of shear and normal sresses can be wrien as α ρ v + αρ v v = (3) α p + α ( τ + T ) + αρg + M l + v Γl 1,l 1,l where g is he graviy vecor, M l represens he momenum inerfacial ineracion beween phases and l, and p is pressure [5]. Pressure is assumed uniform for all phases. τ is he shear sress in phase ; µ is molecular viscosiy; T is Reynolds sress and µ urbulen viscosiy. The urbulence ineic energy equaion yields: αρ + α ρ v = µ α µ + + α α ρ + + Γ σ P K l l 1,l 1,l The producion erm due o shear, P, for phase is equal o: (4) P The urbulence dissipaion equaion is: T : v = (5) 3
Mohamed Gouda, Dr. Konrad Karner, Dr. Reinhard Taschl α 1 α µ + α ρ C P + α µ + σ α C v 2 + ρ Dl 1,l 2 α + C Γl 1,l C 1...4 are he closure coefficiens in he - urbulence model. ρ = 4 ρ v (6) 4 Model Validaion Experimenal daa [6] obained from a laboraory es faciliy for dam brea case sudies are compared wih he SWIFT simulaion resuls. The invesigaed es case comparises of combines a square shaped upsream reservoir and a 45 bend channel (see Figure 1). Oule P5 Tan P1 P2 P3 P4 4.0 m 2.4 m 4.2 m Figure 1: Plane view of he channel The iniial condiions are waer a res wih he free surface 0.25 m above he bed level in he upsream reservoir and 0.01m waer deph in he channel. All boundaries are solid non-slip walls expec he oule, which is considered as saic pressure. Figure 2 shows he assumed waer surface wih volume fracion equal 0.5. The color scaling describes he velociy of he fron. Red color shows he maximum value of he velociy and blue he minimum value. 4
WCCM V, July 7-12, 2002, Vienna, Ausria =0.6 s =1.7 s =2.0 s =2.4 s Figure 2: CFD simulaion resuls The waer displacemen fron afer dam breaing is of ineres. Here, he displacemen ime in he es faciliy wih he simulaion resuls a defined monior poins is compared and shows good agreemen beween measuremen and simulaion (see Figure 3). Time [s] 3.0 2.5 2.0 Measuremen Simulaion 1.5 1.0 0.5 0.0 1 2 3 4 5 Observaion Poins Figure 3: Comparison of displacemen ime beween es and simulaion resuls 5
Mohamed Gouda, Dr. Konrad Karner, Dr. Reinhard Taschl 5 Case Sudy A real ciy area is he subjec of his simulaion sudy. A river ha flows hrough cener of he ciy, is used by several local power saions. The case of dam breaing in he upper par of he river and flooding of he cener is simulaed. Figure 4: Simulaion domain is ciy cener area The grid for he analyzed area was generaed by using AVL FAME (Flexible Auomaic Meshing Environmen). I conains 225K hex elemens wih 2.2m in each direcion. The analyzed domain is 800m x 800m and he heigh of mesh is exended 11m higher han he riverbed. Figure 5: Grids in physical plane The forcing condiion is a discharge of 1100 m3/s, which applied suddenly by a flood wave ha moves down he cener of he ciy. The ime sep is 2 seconds and he oal duraion of simulaion is 4 minues. The oal CPU ime for he simulaion is 82 hours on an IBM power3 wih a 200MHz cloc. Figure 6 shows he degree of flooding in he differen regions by volume fracion scaling. Red color shows 100% waer and blue 100% air regions. 6
WCCM V, July 7-12, 2002, Vienna, Ausria Figure 6: Compued volume fracion of he waer during he flood phase 6 Conclusions A Finie Volume algorihm based on conservaion equaions for muli-phase flow has been derived, discussed and applied. The high compuaional efficiency of his mehod has made i possible o provide fine deails of he waer circulaion, velociy and pressure around he buildings in he ciy area during he flooding process. This simulaion mehod enable he regulaory auhoriies o assess a flood zone warning in he areas where no flooding hisories are documened. References [1] Blyh; A Telenewor for Acquisiion, Processing and Disseminaion of Earh Observaion Daa for Monioring and Emergency Managemen of Floods Hydrological Processes 11, 1359-1375, (1997). [2] Paanar; umerical Hea Transfer and Fluid Flow; ISB 0-07-048740-5. [3] B. Basara; AVL SWIFT Theory Boo; AVL-Lis GmbH, (2001). [4] Schiefermueller; High Order Differencing Schemes and heir Applicaion o Muli-Dimensional Flow Problems; AVL LIST GmbH, (1993). [5] A. Alajbegovic ; AVL Muli-phase Theory Boo; AVL-Lis GmbH (2002). [6] P. Berufa and P. Garcia-avarro; Two Dimensional Dam Brea Flow Simulaion; In. J. umer. Meh. Fluids; 33:35-57 (2000). 7