Journal of the Persan Gulf (Marne Scence)/Vol. 3/No. 10/December 2012/6/1-6 Numercal Study on Resstance of a Bulk Carrer Vessel Usng CFD Method Ebrahm, Aboozar 1- Faculty of Marne Engneerng, Chabahar Martme Unversty, Chabahar, IR Iran Receved: July 2012 Accepted: November 2012 2012 Journal of the Persan Gulf. All rghts reserved. Abstract Bulk carrers have an mportant role n nternatonal martme transport. In ths paper, we carred out a numercal study on a model of bulk carrer vessel and calculated total resstance of the model. A model of a bulk carrer vessel wth length of 2.76 m, breadth of 0.403 m and draught of 0.173 m was selected for numercal modelng. Numercal work was done by commercal CFD software ANSYS FLUENT 13 based on fnte volume method (FVM). Applyng VOF model, FLUENT solved RANS equatons and accordngly, free surface waves around vessel and total resstance were calculated. A fully structured mesh was used whch generated n GAMBIT pre-processng software. Numercal soluton was done for 5 dfferent speeds from 0.2 to 0.95 m/s. Also, mesh dependency was studed by usng four mesh szes and fnally, a mesh wth 755,000 hexahedral cells was selected for all speeds. For verfcaton of numercal results, expermental data from test of bulk carrer model were used. Model tests were carred out n towng tank of Galat Unversty n Romana. Moreover, free surface waves from numercal study were compared wth waves captured from model tests. The results showed that CFD results had an error up to 12 % relatve to model test results and wave patterns around vessel were very smlar to that of model test. Keywords: Bulk Carrer Vessel, Resstance, CFD, FLUENT, Model Test 1. Introducton Bulk carrers contrbute to a substantal porton of world martme transport. The world's bulk transport reached mmense proportons n 2005 n whch, 1.7 bllon metrc tons of coal, ron ore, gran, bauxte, and phosphate were transported by bulk carrer shps (UNCTAD, 2006). Today, the world's bulker fleet ncludes 6,225 shps of over 10,000 tons deadweght and represents 40% of all shps n terms of tonnage and 39.4% n terms of vessels. * E-mal: ab_ebrahm@cmu.ac.r In recent years, the numercal soluton method and specally Computatonal Flud Dynamc method (CFD) have developed for studyng hydrodynamcs of shps. The CFD method n comparson wth model test has great advantages, such as lower costs, accessblty and more vsble detals n results e.g., pressure and velocty contours, vectors and gradents. One of the mportant weaknesses of ths method s that t s less accurate. Therefore, expermental data should be used for verfcaton of CFD results. In ths study, we use CFD method for calculatng resstance of a bulk carrer 1
Ebrahm / Numercal Study on Resstance of a Bulk Carrer Vessel Usng CFD Method model and then, verfy results by expermental data from model test. Numercal study on resstance of a shp has been performed n several cases. Thornhll et al. (2003) used a fnte volume code to smulate the flow around a plannng vessel at steady speed s through calm water, usng 3D unstructured hybrd mesh. Force and moment data from the smulatons were used n an teratve schemee to determne the dynamc equlbrum poston of the model at selected speeds. Van et al. (1998) measured flow around a VLCC tanker model. Ogwara et al. (1994) carred out a study on seres of 60 and calculated pressure dstrbuton around ts hull. Jones and Clarke (2010) used FLUENT code for numercal smulaton of flow around a modern naval shp, DTMB 5415 and calculated waves and free surface around t. Banks et al. (2010) calculated components of contaner shp resstance numercally. They used ANSYS CFX for numercal smulaton and compared CFD results wth expermental results. Hakan et al. (2007) modeled a shp hull by usng FLUENT commercal code and calculated resstance of model. Fnally, they compared numercal results wth expermental data. Obrea et al. (2005) carred out a seres of model tests n a bulk carrer vessel. In ths study, we used results of these tests for verfcaton of numercal results. Accordngly, we used the geometry of expermental model for numercal modelng. 2. Model Geometry A model of a Panamax bulk carrer shp was used n ths study, that was a 1: 80 scale of full f scale shp. The bulk carrer model s shown n Fgure 1. The characterstcs of the shp and of the model are presented n Table 1. Fg. 1: Model of bulk carrer shp (Obrea et al., 2005) 2 Table 1: Full scale shp s and modell man characterstcs (Obrea et al., 2005) Man characterstcs Length of waterlne ( L WL ) Full scale shp 220.915 [m] Model 2.761[m] Length between perpendculars ( L B BP) 217.30 [m] 2.716[m] Breadth of waterlne (B) Draught mean (D) 32.20 13.83 [m] [m] 0.403[m] 0.173[m] Trm (t) 0 [m] 0 [m] Volumetrc dsplacement ( ) 82626 [m ] 0.161 [m 3 ] Block coeffcent (CB) Mdshp secton coeffcent (C M ) Waterlne coeffcent (C WP ) 0.837 0.995 0.923 0.837 0.995 0.923 3. Numercal Soluton In recent years, the CFD methods and applcaton of these methods for hydrodynamc study of shpss have developed. Although, ths method n comparson wth model test s less accurate, t can be used for studyng detals of flow characterstcs, e.g. drag and lft forces, pressure and velocty contourss and waves. In ths study, free surface flow around a model was smulated usngg FLUENT CFD software. Ths commercal software solves Reynolds-Averagedd Naver-Stokes (RANS) equatons n computatonal doman and computes pressure and velocty n alll nodes and accordngly calculates total resstance r of model. For capturng surface waves around model, Volume of Flud (VOF) method was used. 2
Journal of the Persan Gulf (Marne Scence)/Vol S.3/No. 10/December 2012/6/ /1-6 The VOF method smulates the moton of two fluds by solvng a sngle set of momentum equatonss and trackng the volume fractons of each of the fluds throughout the doman. A volume fracton functonn was defned for each of the fluds n the smulaton andd was set to unty f the gven flud occuped a cell volume or was set to zero otherwse. When the nterface between two fluds cut through a computatonal cell the value of the functonn for a partcular flud representedd the fracton of the cell volume occuped by that flud Jones and Clarke (2010). Equatons (1) and (2) were the ncompressble RANS equatons n tensor form and Equatonn (3) was the equaton of volume fracton transport Perc and Ferzger (2002). ( U ) ( UU U ) P U U t x x x x x x U 0 x c ( cu ) 0 t x The volume fracton c was defned as (V ar /VV total ) and the flud densty, ρ, and vscosty, µ, were calculated as Hakan et al. (2007): c ar (1 c) water c ar water (1 c) (5) External forces appled to the flud were represented as f, whch nclude buoyancy forces due to dfferences n densty and momentum sources. The effect of turbulencee on the flow s represented n Equaton (2) by the Reynolds stress tensor u u. 4. Computatonal Doman The computatonal doman and ts dmensons are dsplayed n (Fgure 2). These dmensons were selected accordng to Van et al. (1998). The doman was defned by a cube volume whch the model volume subtracted from t. Accordng to VOF model, from the bottom of the doman to the ntal water surface was occuped by the water and the rest by the ar. The hull and flow feld were symmetrc about the u' u' ' f (1) (2) ' (3) (4) 3 center plane of model.. Therefore, ust half of computatonal doman was consderedd for numercal treatment and the symmetry boundary condton appled at center plane. Ths could ncrease rate of computatons and reduce tme of soluton. Fg. 2: Dmensons of computatonal doman The GAMBIT Pre-processng software was used for mesh generaton. A structured mesh was used n doman whchh all of the cells were hexahedral type. At stern and bow b sde of model and near free surface, meshes have refned n all states. For checkng grd ndependency, we examned 4 grd szes. s By vsual nspecton of o the wave pattern and detaled comparson of o other results, such as pressure and velocty contours and resstance, we selected the optmum mesh. For example, total resstance of model due to dfferent grd szes s plotted n Fgure 3 for V=0.9. It seemed by ncreasng grd sze over 755000 cells, varatonn of resstance would be nconsderable. Although, byy usng hgher grd szes, free surface waves could be b captured more accurately. Fnally, ths sze of mesh was selected for numercal modelng. Detals of the grd are dsplayed n Fgure 4. A tme-dependent soluton wth tme step of 0.005 second was used u for numercal soluton. Montorng of the resstance and free surface waves around model was used to know k when the problem had reached a statonary soluton. We observed that after 20 seconds, resstance of model m and surface wavess dd not change consderablly.. Therefore, ths tme was selected for
Ebrahm / Numercal Study on Resstance of a Bulk Carrer Vessel Usng CFD Method modelng at all speeds. Numercal modelng was performed for 6 speeds between Fr = 0.09 (V=0.59 m/ /s) and Fr = 0.18 (V=0.95). k model was selected for numercall smulaton. For near wall treatment, t Standard Wall Functons weree used. In numercal soluton, trm and snkage of model were kept fxed. Although, t could produce errors n results, but becausee of our lmtatons and complexty of 6-DOF modelng of hull, we appled ths strategy. Also, because the vessel was dsplacement and Froude number was under 0.3, t seemed that ths assumpton was ustfable. 5. Boundary Condtons Fg. 3: Varaton of resstance force due to mesh sze Fg. 4: Detals of mesh For turbulence modelng, standard, RNG and Realzable k models were used n numercal soluton. Results showed that there was not much dfference between results of these models, but the appled computng tme for standard k model was shorter than that of others. Accordngly, standard 4 The non-slp boundary condton wass appled on the hull surface of o the model, that s, the flud partcles on the body move wth body velocty. Inlet and outlett of doman were dvded ntoo two boundares by the ntal locaton off free surface. At nlet, the upper part was entrance of ar and the lower was for water. The velocty nlet condton was mposed to both parts. The statc pressure at the outlet boundary was defnedd as a functon of water volume fracton. The movng wall boundary condton was appled to sde walls, top and bottom of doman and fnally symmetry for center plane of doman. A varety of Pressure-Velocty Couplng schemes were avalable n FLUENT. The PISO scheme was selected att ths study. For pressuree and momentum spatal dscretzatons, body force weghted and second order upwnd methods weree selected, respectvely. The moton off the free surface flow was governed by gravtatonal force. f Hence,, the boundary condtons to be mposed must m take nto account the gravty effects. For ths purpose, thee computatonal doman was modeled as an a open channel flow, whch was also consstent wth expermental setup. 6. Results and Dscusson D 6.1. Numercal Results R One way for verfcaton of numercall results was comparng waves generatedd around model. The predcted waves around a model of bulk carrer at speed
Journal of the Persan Gulf (Marne Scence)/Vol S.3/No. 10/December 2012/6/ /1-6 of V=0.95 m/s and wth avalable expermentall data showed generally good agreement (Fgures 5a and b). In these fgures, wave systems of bow and stern were obvous. resstance of bulk b carrer model from CFD method s up to 5 percent dfference wth expermental resultss n lower speeds, but after that, t reached to 13%. The reason of these errors may be weakness of ths CFD code for calculatng wave makng resstance. However, n low speeds whch wave makng resstance had smaller magntude; CFD results showed less dfference wth expermental results. We could seemngly reduce errors by usng the modern computers for numercal modelng and usng fner grds. 6.2. ITTC 57 Method Fgs. 5a (top) and b (bottom): Comparson of waves from CFD method and model test The model test was an accurate and relable method for measurng and nvestgatng of thee shp resstance. For verfyng of numercal results nn ths study, we used results of model testss accomplshed by (Obrea et al., 2005) n towng tank of Galat Unversty. The total resstance calculated by CFD method s shown n Fgure 6 and compared wth the expermental results. Another way w for estmatng resstance of a shp was applyngg common emprcal formulas, such as ITTC 57 method. In ths study, frctonal resstance of model for comparson by CFD results were nvestgated. Frctonal resstance of shp (RF) was calculated from Equaton (6): R C (0.5 SV 2 m ) F f Coeffcentt of frctonal resstance (Cf) was computed by emprcal formula presented by ITTC- 57 as shown n Equaton (7): C f 0.075 (log Rn 2) 2 (6) (7) In Fgure 7, 7 frctonal resstance from ITTC 1957 method was compared c wth frctonal resstance from CFD method and there was good correlaton betwen them. In all speeds, theree was aboutt 8% dfference between results. Fg. 6: Comparson of total resstance from CFD and model test As shown n Fgure 6, there was lttle dfference between expermental and numercal fndngs. Total 5 Fg. 7: Comparson of frctonal resstance from CFD and ITTC 57 method
Ebrahm / Numercal Study on Resstance of a Bulk Carrer Vessel Usng CFD Method 7. Concluson In ths paper, an nvestgaton was carred out on resstance of a bulk carrer vessel usng CFD method by FLUENT software. Total resstance of model was calculated for dfferent speeds and surface waves around model were obtaned numercally. Fnally, CFD results were compared wth expermental data. Analyss showed that n low Froude numbers, CFD results had a dfference up to 5% relatve to the model test results. In hgher Froude numbers, error ncreased up to 13%. The frctonal resstance of model calculated by ITTC 57 method was compared wth CFD results. There was good agreement between results of CFD and ITTC method. References Banks, J., Phllps, A. and Turnock, S., 2010. Free surface CFD predcton of components of shp resstance for KCS. 13 th Numercal Towng Tank Symposum, Germany, pp.1-7. Hakan, O., Bayraktar, S, and Ylmaz, T., 2007. Computatonal nvestgaton of a hull, 2 nd Internatonal Conference on Marne Research and Transportaton, Italy, pp.145-150. Jones, D.A., Clarke, D.B., 2010. Fluent code smulaton of flow around a naval hull, the DTMB 5415. Techncal report, Defense Scence and Technology Organzaton. DSTO-TR-2465, 30P. Obrea, D., Popescue, G, Ionas, O. and Pacuraru, F., 2005. Expermental technques for vortces nvestgaton around the shp model, Workshop on Vortex Domnated Flows, Romana. Ogwara, S. and Katan, H., 1994. Pressure dstrbuton on the hull surface of seres 60 model. Proceedngs of CFD Workshop Tokyo, pp.350-358. Perc, M., Ferzger, J.H., 2002. Computatonal Methods for Flud Dynamcs, Sprnger Pulcaton, 3rd edton, 426P. Thornhll, E., Bose, N., Vetch, B. and Lu, P., 2003. Planng hull performance evaluaton usng a general purpose CFD code, 24 th Symposum on Naval Hydrodynamcs, the Natonal Academy of Scences, Japan. Unted Natons Conference on Trade and Development (UNCTAD), 2006, 11 P. Van, S.H., Km W.J., Km, D.H., Ym, G.T., Lee, C.J., and Eom, J.Y., 1998. Flow measurement around a 300K VLCC model. Proceedngs of the annual sprng meetng, Ulsan, pp.185-188. Ebrahm / Numercal Study on Resstance of a Bulk Carrer Vessel Usng CFD Method Journal of the Persan Gulf (Marne Scence)/Vol.3/No. 10/December 2012/6/1-6 Journal of the Persan Gulf (Marne Scence)/Vol. 3/No. 10/December 2012/6/1-6 6