193 ISSN 13921207. MECHANIKA. 2015 Volume 21(3): 193200 Detached Eddy Smulaton of the flow around a smplfed vehcle sheltered by wnd barrer n transent yaw crosswnd Maro Telenta*, Mataž Šubel**, Jože Tavčar***, Jožef Duhovnk**** LECAD, Faculty of Mechancal Engneerng, Unversty of Lublana, Slovena, E-mal: *maro.telenta@lecad.fs.un-l.s; **mataz.subel@lecad.fs.un-l.s; ***oze.tavcar@lecad.fs.un-l.s; ****ozef.duhovnk@lecad.fs.un-l.s http://dx.do.org/10.5755/01.mech.21.3.8942 1. Introducton Today s ground vehcles are senstve to crosswnd dsturbance. Nonetheless, the focus of vehcle development s on provdng lghter and streamlned vehcles. The vehcle s crosswnd senstvty deterorates due to vehcle s aerodynamc desgn mprovements. A lot of research s dedcated n analyzng the nfluence of the vehcle shape on the vehcle s crosswnd senstvty. There are two obectves for vehcle s aerodynamc desgn. The frst obectve s reducton of drag coeffcent n order to mprove fuel effcency. The second obectve s mprovement of drvng stablty where lft and ptch are mportant for the straght lne drvng, and yawng moment and sde force are mportant for the crosswnd senstvty [1]. However, these obectves represent conflctng goals. Therefore, n ths work other means for mprovng crosswnd senstvty are consdered. Instead of analyzng the nfluence of dfferent vehcle desgns on the vehcle s crosswnd stablty, wnd barrer s consdered for lowerng vehcle s crosswnd exposure. Prevous studes modeled flud flow through porous geometres whle not consderng the detals of the barrer s geometry. The man focus was to defne a sutable resstance model for a gven geometry of a barrer. Prevous studes, [2], [3], and [4] used the Reynolds averagng method wth turbulence closure for a two-dmensonal flud flow smulaton n whch the porous barrer was represented as a momentum snk. As stated n [5], numercal methods utlzng the momentum snk approach for wnd barrer modelng treat complex unresolved flow near and through the gaps at a superfcal level. A deeper understandng of the turbulent structure dynamcs s requred to evaluate the barrer shelterng effect. Author s prevous work [6], [7] addressed ths ssue. URANS numercal smulatons, verfed wth expermental data, were done modelng the flud flow through geometrcally accurate threedmensonal barrer model n order to resolve the flow near and through the porous barrer. The obectve was to nvestgate the nteracton between the bleed flow and the reverse flow for dfferent barrer confguratons. Present paper extents the research scope wth more advanced turbulence models, namely Detached Eddy Smulaton (DES). DES s computatonally more expensve than RANS. However, t provdes more accurate results and gves nformaton about the flow structures whch s out of reach for RANS methods. RANS provdes only the mean nformaton about the flow and the unsteady nformaton s lost. Also, flow calculaton accuracy s dependent on the turbulence model used. It s dffcult to defne a RANS model that accurately represents the Reynolds stresses n the regon of separated flow such as a wake behnd the barrer. In addton, complcated flow structures are developed n the wake regon behnd the barrer. These wake structures are domnated by large turbulent structures whch can be resolved by DES method. Increase n the computer capablty made DES smulaton possble nowadays. DES s used nstead of Large Eddy Smulatons (LES) snce LES s not feasble for hgher Reynolds number flow whch s the case n ths work. Therefore, DES was utlzed to smulate the tme-development of the flow around a generc vehcle behnd a wnd barrer subected to a sudden strong crosswnd. The nvestgaton scope n the current work ncludes vehcle shelter by geometrcally accurate wnd barrer. Present numercal smulaton mrrors the expermental work of [1] whch s further expanded wth barrer model ntroducton n computatonal doman. The goal was to analyze the barrer nfluence on aerodynamc loads development that vehcle s subected to. Wnd tunnel testng of a vehcle n crosswnd was done n [1] where transent yaw crosswnd scenaro on a smplfed vehcle shape correspondng to sport utlty vehcle was performed. The flexblty of the CFD makes transent crosswnd studes easer to realze than expermental studes. However, no DES studes wth tme-dependent boundary condtons nvestgatng the wnd barrer applcatons were found pror to ths work. To date, transent crosswnd wth DES method has been nvestgated on a bus geometry [8], smple vehcle shapes [9], and hgh speed trans [10], [11]. In ths work, unsteady crosswnd smulaton for sheltered movng vehcle behnd the wnd barrer subected to a determnstc gust wnd represented by a contnuous and smooth step functon s reported. Advanced boundary condtons are mplemented to smulate a gust wnd propagatng through the computatonal doman. Two types of vehcle aerodynamc research are found: tme averaged aerodynamcs testng and transent aerodynamc testng [12]. Tme averaged technque subects the vehcle to a constant yaw angle, whereas wth the transent technques the vehcle s subected to a rapd change n yaw angle. Although there are many wnd nduced traffc accdents every year, the wnd barrer shelter effect on vehcle aerodynamc s not yet properly nvestgated. Avalable studes are stll nadequate to gve the complete pcture of the flow structures around the movng vehcle behnd the wnd barrer n gust wnd condtons.
194 2. Numercal methods Delayed Detached Eddy Smulaton (DDES) s performed to study the flow behnd the wnd barrer. DDES s a hybrd technque for predcton of separated turbulent flows at hgh Reynolds numbers. Development of ths technque was motvated by the prohbtve computatonal costs of applyng Large-Eddy Smulaton (LES). Thus, hgh-reynolds number separated flows have been predcted usng steady or unsteady Reynolds-averaged Naver-Stokes equatons (RANS and URANS). However, the dsadvantage of the RANS methods appled to massve separatons s that the statstcal models are desgned and calbrated on the bass of the mean parameters of thn turbulent shear flows contanng numerous relatvely standard eddes. Such eddes are not representatve of the comparatvely fewer and geometry-dependent structures that typcally characterze massvely separated flows. Reynolds-averaged Naver-Stokes (RANS) equatons, Eq. (1) and Eq. (2): Δ s the maxmum edge length of the local computatonal cell, d dstance from the wall, d ~ length scale, r d parameter, f d functon, and C DES =0.65 s model constant. 3. Barrer and vehcle model One barrer confguraton model was used n the numercal smulaton, as seen n Fg. 1, a. It conssts of fve horzontal bars, all wth 90 nclnaton angle. The barrer length s L = 9.5L V, where L V = 0.48 m s the vehcle length. Porosty of the barrer s 25%. The computatonal doman s also shown n Fg. 1, b, where the barrer model length L and heght H are 4.56 m and 0.4 m, respectvely. Vehcle model has a box-lke geometry where the wdth W V and the heght H V are the same wth value of 0.2 m. Ground clearance of the vehcle s 0.03 m. In order to avod movng mesh, vehcle poston s fxed and the velocty s mposed at the doman nlet. u x 0, (1) 2 u u p u u t x x x x ' ' uu, x (2) where p s averaged pressure, u s averaged velocty, s the ar densty, dynamc vscosty, and Reynolds ' ' stresses are u u. The fltered Naver-Stokes equatons for LES, Eq. (3) and Eq. (4), are as follows [13]: u 0, (3) x u 1 p uu t x x u u, x x x where the velocty u s separated nto the fltered, resolved part (4) u ~ and sub-fltered, unresolved part u, and s knematc vscosty. The swtch between RANS and LES n DDES, Eq. (5) and Eq. (6), s expressed as follows [14]: where: d d f max 0,d C, (5) d 3 DES f 1 tanh 8r, (6) r d t d. 05 2 2 U, U, k d d. (7) ' Fg. 1 (a) Barrer confguraton and (b) computatonal doman 4. Solver A commercal CFD code, Ansys Fluent 14.5, was used to solve ncompressble Naver-Stokes equatons of flud moton. It uses cell-centered numercs, va a segregated approach, on a collocated, unstructured grd. The Delayed DES (DDES) method wth the standard Spalart Allmaras (SA) model was used. The dffusve fluxes of the momentum and turbulent equatons are dscretzed usng the central dfference (CD) scheme. The bounded central dfference scheme was used for convectve fluxes n LES and RANS regons. Second order upwnd was used for the spatal dscretzaton of the convecton terms of the turbulence model. Second order upwnd scheme s less dffusve and offers more accurate soluton over the frst order upwnd scheme. The least square method was used for the gradent method. Standard pressure nterpolaton was used. Tme ntegraton was performed wth the second-order backward Euler scheme and the bounded second-order mplct Euler scheme for turbulence varables. The DDES smulaton s ntalzed wth steady state RANS smulatons. The SST k- turbulence model s used n RANS and smulatons are run untl convergence s reached. In the DDES method, pressure mplct wth splttng of operator (PISO) algorthm was set for the pressure-velocty couplng. The tme step s chosen to comply wth the solver requrements for stablty. Addtonally, recommendatons for tme step sze are followed from [15]. The tme
step was set to 0.0001 s, and the CFL number for ths tme step was approxmately 0.5. 5. Grd 195 0,35 0,3 0,25 y (m) The ICEM-CFD commercal grd generator software was utlzed to create the numercal grd, as seen n Fg. 2. A hgh qualty unstructured hexahedral grd was created followng the recommendaton for grd generaton from [15]. Flow regons wth dfferent grddng requrement exst n the numercal smulaton. The grd should follow the recommendatons as far as possble to be effcent for the DES method. The mesh conssts of O-grd and H-grd topologes. Ths allowed for a fner grd close to the wnd tunnel walls and model surface because a no-slp boundary condton s used on those surfaces. Values of y + are set low (below 1) near the ground, the barrer and the vehcle surfaces. Refnement zones are created n crtcal areas, such as separatons and wakes. Only one grd was created for the numercal smulaton. However, the grd senstvty was conducted n a pror numercal smulaton where only the vehcle was analyzed under steady crosswnd. Hence, a grd wth 33 mllon elements was created, where the smallest element sze was 4 mm. 0,2 1,6 2,1 2,6 3,1 3,6 /η Fg. 3 Profle of the rato between the vehcle and the barrer modeled contrbuton s larger. Hgher turbulence vscosty s located near the vehcle and represents the RANS area, whereas lower turbulence vscosty represents the LES area. Fg. 4 shows the nstantaneous turbulence vscosty between the vehcle and the barrer. As one can see from Fg. 4, the lower regon represents the RANS regon near the barrer, and the upper part represents the RANS regon of the vehcle. In-between s located the LES regon. y (m) 0,4 0,3 0,2 0,1 0 5 10 15 20 25 30 35 40 45 ν t ν Fg. 4 Non-dmensonal nstantaneous turbulence vscosty Fg. 2 Hexahedral grd wth barrer and vehcle detals 5.1. Assessment of the grd resoluton The resoluton characterstcs of the grd near the walls were dscussed above by reference to the grd spacng n wall unts. In the nteror of the flow, the grd resoluton can be assessed by comparng the grd spacng to an estmate of the Kolmogorov length. The Kolmogorov length scale, Eq. (8), characterzes the length scale of the dsspatve moton. 3 14, (8) where s the dsspaton rate and 1.51110 m 2 /s s the knematc vscosty. Fg. 3 shows the vertcal profle of the rato between the vehcle and the barrer, along cuts through the shear layer and n the regon beyond the wall. A substantal part of the dsspaton s resolved where the grd spacng s 12 η. One can see from Fg. 3 that ths level of dscretzaton was acheved wth levels of < 12 over the entre doman. Further support for the grd resoluton s provded by the rato t, as seen n Fg. 4, whch gves an ndcaton of the rato of resolved and modeled contrbutons to the dsspaton. When eddy vscosty s larger, the RANS 5 Grd dependence study was not performed. Instead, the largest feasble grd sze was used, n ths case hexahedral grd wth 33 mllon elements. Snce the sze of the grd nfluences the numercal results, only one numercal smulaton wth largest possble grd sze was performed. Instantaneous and turbulent flow s analyzed n the present work. The boundary condtons are tme-dependent and therefore the data s nstantaneous. 6. Boundary condtons The dmensons of the computatonal doman were set to smulate open doman, as seen n Fg. 1, b. The upstream nlet and outlet were postoned at a dstance of 0.5L upstream and L downstream from the barrer, respectvely. The doman heght and wdth were 10H and 20H, respectvely. No-slp boundary condtons were appled on the vehcle and the ground surfaces. Approprate boundary layer and blockage rato were smulated. Velocty nlet boundary condton wth a unform velocty profle was specfed at the doman nlet; n ths case the upstream and the lateral sdes of the doman. A small turbulence ntensty of 0.3% was mposed at the nlet, whch corresponds to the expermental case. A pressure outlet was appled at the doman outflow. The nflow velocty was u 13 m/s and maxmum crosswnd velocty was ucrosswnd 4. 73 m/s. The correspondng Reynolds number based on the barrer 5 heght s Re H 1. 7 10. To smulate transent gust wnd, transent boundary condtons on the upstream and the lateral sdes of the doman are used. In addton to the front velocty nlet, dentcal velocty nlet boundary condtons are specfed on the lateral sdes of the doman. A slp wall v
196 boundary condton was set at the top surface of the doman. A movng wall boundary condton wth 13 m/s streamwse velocty was set for the ground and the barrer surfaces, whereas a statonary wall was set for the vehcle surfaces. A movng grd was avoded wth ths boundary confguraton. No wall functons were used for boundary layer modelng. 6.1. Unsteady crosswnd A sngle strong gust wnd was smulated. The crosswnd scenaro smulates the expermental study done by [1]. A vehcle model s propelled at a constant velocty through the wnd tunnel exhaust. It depcts the scenaro represented n Fg. 5, a. The gust wnd conssts of a et flow and two mxng layers. The et flow was modeled as a step functon where the smooth transtons represent the mxng layers, Fg. 5, b. The smooth transtons were modeled as cosne functons [9]. The maxmum crosswnd velocty length was set to 5L V, and the cosne perod to 1.5L V. The maxmum crosswnd velocty was set to correspond to the 20 yaw angle of the ncomng wnd n respect to the vehcle. The yaw angle value of 20 s consdered the most crtcal for vehcle safety. At the front nlet, the crosswnd s only a functon of tme, whereas at the lateral sde nlets the crosswnd s a functon of the tme and space. The transent boundary condton was ntroduced n the solver va user defned functons (UDFs)., (10) 2, (11) 2 1 u u, (12) 2 x x 1 u u. (13) 2 x x 7.1. Transent crosswnd The present work focuses on understandng of the flow mechansms, whch occur when the gust wnd acts on the barrer and the sheltered vehcle. A tme-dependent gust wnd was ntroduced as boundary data after the headwnd was frst run through the computatonal doman. Fg. 6 shows the propagaton of the gust wnd through the computatonal doman n four dfferent tme moments: 1. no crosswnd, 2. crosswnd approachng the barrer, 3. crosswnd approachng the vehcle, 4. crosswnd behnd the barrer and the vehcle. 7. Results The frst 9000 tme-steps are run wth only headwnd and were not consdered because these tme-steps correspond to a transent perod when the flow s unsteady. Ths corresponds to one flow-through tme,.e., the tme needed for one partcle to go through the entre computatonal doman. Afterwards, gust wnd was ntroduced n the computatonal doman. The components of the aerodynamc forces proected on the vehcle axs are the drag n the streamwse drecton, the sde force n the lateral drecton, and the lft n the upward vertcal drecton. Coherent structures of the flow are nvestgated by usng the second nvarant of the velocty gradent, the Q-crteron, Eq. (10). The vsual nspecton of the turbulence structures was done usng the so-surfaces of the Q-crteron. 1) poston 2) poston 3) poston 4) poston Fg. 6 Gust wnd propagaton colored by the velocty magntude: 1 - no crosswnd, 2 - crosswnd approachng the barrer, 3 - crosswnd approachng the vehcle, 4 - crosswnd behnd the barrer and the vehcle 7.2. Numercal accuracy Fg. 5 (a) Crosswnd scenaro and (b) representatve velocty profle The defnton of the Q-crteron [16]: Q C 2 2 Q, (9) where C 0. 5, ε s the absolute value of the stran rate, Q and s the absolute value of the vortcty. Fg. 7 shows the agreement among the characterstc ponts of the expermental measurements and the numercal data for no-barrer scenaro. The gust length n the numercal smulaton s longer for two vehcle lengths than the one from the expermental measurements, and adustments on the tme axs were performed for the expermental data to evaluate the agreement between the expermental data and the numercal results. As one can see, present results are n good agreement wth the expermental measurements. In addton, Fg. 7 shows the sde force coeffcent trace for the fxed yaw tests. The sde force coeffcent value for the fxed yaw test was tme averaged. As one can see from Fg. 7, the sde force coeffcent for transent yaw test dsplays an overshoot compared
197 to the fx yaw test, whch ndcates the mportance of performng the transent yaw tests for vehcle crosswnd senstvty. 1,55 cs p p3 p 1,35 1,15 numercal 0,95 smulaton 0,75 experment 0,55 p1 0,35 fx yaw test 0,15-0,05 0,000,030,050,080,100,130,150,180,200,230,250,280,300,33 tme (s) Fg. 7 The vehcle s sde force coeffcent wthout the wnd barrer (p1, p2, p3, p4 are vehcle postons relatve to the gust wnd) 7.3. Aerodynamc forces and moments 0,20 cyaw 0,15 0,10 0,05 tme (s) 0,00 0,00 0,03 0,05 0,08 0,10 0,13 0,15 0,18 0,20 0,23 0,25 0,28 0,30 0,33-0,05 no barrer -0,10 wth barrer -0,15 a 0,07 croll 0,06 no barrer 0,05 wth barrer 0,04 0,03 0,02 The axs system for the force and moment coeffcents s the center axs of the vehcle model, where a postve sde force occurs wth postve wnd loadng and a postve yawng moment occurs when the nose of the model turns leeward. The non-dmensonal coeffcents, c F and c M, for the forces and the moments, are defned n 0,01 0,00-0,010,00 0,03 0,05 0,08 0,10 0,13 0,15 0,18 0,20 0,23 0,25 0,28 0,30 0,33 tme (s) b 0,18 0,15 1,25 1,15 cd 0,13 0,08 0,95 0,05 0,85 0,03 0,65 no barrer wth barrer 0,10 1,05 0,75 cptch no barrer wth barrer tme (s) 0,55 0,00 0,03 0,05 0,08 0,10 0,13 0,15 0,18 0,20 0,23 0,25 0,28 0,30 0,33 0,00-0,030,00 0,03 0,05 0,08 0,10 0,13 0,15 0,18 0,20 0,23 0,25 0,28 0,30 0,33 tme (s) -0,05-0,08 c a 1,55 cs 1,35 1,15 0,95 no barrer 0,75 wth barrer 0,55 0,35 0,15-0,05 0,00 0,03 0,05 0,08 0,10 0,13 0,15 0,18 0,20 0,23 0,25 0,28 0,30tme 0,33(s) b 0,60 cl 0,50 0,40 0,30 0,20 0,10 tme (s) 0,00-0,100,00 0,03 0,05 0,08 0,10 0,13 0,15 0,18 0,20 0,23 0,25 0,28 0,30 0,33-0,20 no barrer -0,30 wth barrer -0,40-0,50 c Fg. 8 Aerodynamc forces on the vehcle wth and wthout the barrer (p1, p2, p3, p4 are vehcle postons relatve to the gust wnd): a - drag force coeffcent, b -sde force coeffcent, c - lft force coeffcent Fg. 9 Aerodynamc moments on the vehcle wth and wthout the barrer (p1, p2, p3, p4 are vehcle postons relatve to the gust wnd): a - yaw moment coeffcent; b - roll moment coeffcent; c - ptch moment coeffcent Eq. (14) and Eq. (15): cf F, qa (14) cm M, qa (15) where q s the dynamc pressure of the ncomng wnd, Eq. (16): q 1 2 ur, 2 (16) where F s the force (drag, lft, and sde force), M s the moment (roll, ptch, and yaw), s the densty of ar, A s the frontal area of the vehcle model, and u R s the resultant velocty. The non-dmensonal pressure coeffcent:
198 c p p p, (17) q where p s the pressure n the free-stream. Fgs. 8-9 show the aerodynamc force and moment coeffcents on the vehcle wth and wthout the barrer, respectvely. As one can see, all the aerodynamc loads are lower when the barrer s ntroduced except for the ptch moment coeffcent. Vehcle crosswnd stablty s nfluenced manly by the yaw moment coeffcent and the sde force coeffcent. Hence, vehcle crosswnd stablty s mproved sgnfcantly wth the barrer shelter. 7.4. Vsualzaton of the turbulent structures and pressure mappng Fg. 10 shows the vehcle poston relatve to the gust wnd. Poston 1 represents the vehcle at the begnnng of the gust wnd. Poston 2 represents the start of the vehcle experencng the maxmum velocty of the gust wnd. Poston 3 represents the mddle of the gust wnd, poston 4 represents the end of the gust wnd maxmum velocty, at the poston 5 the vehcle starts to ext the gust wnd, and at the poston 6 the vehcle completely exts the gust wnd. Fgs. 11-14 show the vsual representaton of the large-scale flow structures usng the so-surfaces of the Q-crteron colored by CFL value. The Q-crteron value n ths work s 100 000 s -2. One can see from Fgs. 11-14 that the flow s complex and unstable. Also, the flow s remarkably changed wth the wnd barrer ntroducton. There are domnant and well defned coherent vortces orgnatng from the front surface of the vehcle at postons 3, 4 and 5 when there s no wnd barrer shelter. These structures are nclned n the drecton of the resultant drecton of the gust wnd. Ths s not the case for the wnd barrer scenaro where prevalent vortcal structures are small and parallel to the vehcle travelng drecton for the entre duraton of the gust wnd. In partcular, the flow structures of the sheltered vehcle durng the gust wnd are smlar to that before the sheltered vehcle entrance nto the gust. Also, one can notce the addtonal vortces formed between the vehcle and the barrer. a) poston 1 b) poston 2 c) poston 3 d) poston 4 e) poston 5 f) poston 6 Fg. 12 Top vew of the turbulence structures around the vehcle wth the barrer Fg. 10 Vehcle poston relatve to the gust wnd a) poston 1 b) poston 2 a) poston 1 b) poston 2 c) poston 3 d) poston 4 c) poston 3 d) poston 4 e) poston 5 f) poston 6 Fg. 11 Top vew of the turbulence structures around the vehcle wthout the barrer e) poston 5 f) poston 6 Fg. 13 Isometrc vew of the nstantaneous pressure coeffcent c p on the vehcle wthout the barrer
199 a) poston 1 b) poston 2 c) poston 3 d) poston 4 e) poston 5 f) poston 6 Fg. 14 Isometrc vew of the nstantaneous pressure coeffcent c p on the vehcle wth the barrer Fgs. 13-14 show the pressure dstrbuton on the vehcle surfaces. The pressure dstrbuton ndcates the nfluence of the vortces on the vehcle s aerodynamc loads. As one can see, pressure coeffcents are much lower for the case wthout the barrer shelter and consequently vehcle experences larger aerodynamc loads. Also, one can notce n Fg. 14 that at the postons 5 and 6 hgher values of the negatve pressure coeffcent are dsplayed; however, these values occur for a very short tme and on a small area of the vehcle surface. In addton, one can see from Fg.14 that at the poston 4, the large negatve pressure coeffcent values on the vehcle s leeward surface correspond to the sde force coeffcent and yaw moment coeffcent peak values. However, these peak values are stll lower than those for the case wthout the wnd barrer. 8. Concluson In ths work, a geometrcally accurate threedmensonal wnd barrer model was used n the numercal study. An advanced aerodynamc CFD smulaton, DES, was utlzed n analyzng the transent crosswnd scenaro. Tme-dependent boundary condtons were used to smulate the gust wnd propagaton. One barrer confguraton was analyzed. Hexahedra grd wth 33 mllon elements was used n the present numercal study. Aerodynamc coeffcents, pressure mappng, and flow vsualzaton are used to analyze the effects of the transent crosswnd on the vehcle aerodynamcs. The goal of ths research was to quantfy and vsualze the barrer s shelter nfluence on the vehcle aerodynamcs n transent crosswnd scenaro. LES s not feasble for hgher Reynolds number flow whch s the case n ths work and DES provdes an accurate predcton of the dynamc change n the aerodynamc coeffcent. Therefore, DES was used to nvestgate the nfluence of the barrer shelter on the vehcle aerodynamcs n gust wnd. The am was to study the flow around the vehcle n unsteady wnd scenaro and sheltered vehcle s aerodynamc load development. Vortcal structures are sgnfcantly changed wth the wnd barrer applcaton compared to the no-barrer case. Large and well defned vortcal structures develop when the vehcle s not sheltered by the wnd barrer. These vortcal structures ncrease the negatve pressure on the vehcle s surface whch promotes ncrease of the vehcle aerodynamc loads. The present work shows that the vehcle aerodynamc force and moment coeffcents are much lower wth the barrer shelter except for the ptch moment coeffcent. Moreover, vehcle s crosswnd stablty s mproved wth ntroducton of the wnd barrer snce vehcle s yaw moment coeffcent and sde force coeffcent are sgnfcantly lower. The present work provdes nsght nto the vortcal structure development whch s result of the nteracton among the vehcle, the barrer and the gust wnd. It offers deeper understandng of the flow mechansm around the sheltered vehcle n the transent crosswnd. In partcular, the dynamcs of the coherent structures, as well as the flow structure evoluton and nteracton wth the wnd barrer ntroducton for the vehcle protecton s presented. No pror work to date has analyzed the tme development of the vehcle aerodynamc loads behnd the geometrcally accurate wnd barrer n gust wnd condtons. References 1. Chadwck, A. 1999. Crosswnd Aerodynamcs of Sport Utlty Vehcle, PhD Thess, Cranfeld Unversty. 2. Packwood, A. 2000. Flow through porous fences n thck boundary layers: comparsons between laboratory and numercal experments, J. Wnd Eng. Ind. Aerodyn. 88: 75-90. http://dx.do.org/10.1016/s0167-6105(00)00025-8. 3. Fang, F.-M.; Wang, D.Y. 1997. On the flow around a vertcal porous fence, J. Wnd Eng. Ind. Aerodyn. 67-68: 415-424. http://dx.do.org/10.1016/s0167-6105(97)00090-1. 4. Huang, L.-M.; Chan, H.C.; Lee, J.-T. 2012. A numercal study on flow around nonunform porous fences, J. Appl. Math. p.12. http://dx.do.org/10.1155/2012/268371 5. Bourdn, P.; Wlson, J.D. 2008. Wndbreak aerodynamcs: Is Computatonal Flud Dynamcs Relable?, Boundary Layer Meteorol. 126: 181-208 http://dx.do.org/10.1007/s10546-007-9229-y. 6. Telenta, M.; Duhovnk, J.; Kosel, F.; Šan, V. 2014. Numercal and expermental study of the flow through a geometrcally accurate porous wnd barrer model, J. Wnd Eng. Ind. Aerodyn. 124: 99-108. http://dx.do.org/10.1016/.wea.2013.11.010. 7. Telenta, M. 2014 Wnd barrer motorway protecton. LAP LAMBERT Academc Publshng. 8. Hassan, H.; Kranovć, S. 2007. DES of the flow around a realstc bus model subected to a sde wnd wth 30 yaw angle, The ffth IASME/WSEAS Internatonal Conference on Flud Mechancs and Aerodynamcs, Athens.
200 9. Favre, T. 2011. Aerodynamcs smulatons of ground vehcles n unsteady crosswnd, Stockholm. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:dva- 50242. 10. Kranovć, S. 2008. Computer smulaton of a tran extng a tunnel through a varyng crosswnd, Internatonal ournal of alway 1(3): 99-105. 11. Kranovć, S.; Rngqvst, P.; Nakade, K.; Basara, B. 2012. Large eddy smulaton of the flow around a smplfed tran movng through a crosswnd flow, J. Wnd Eng. Ind. Aerodyn. 110: 86-99 http://dx.do.org/10.1016/.wea.2012.07.001. 12. Šušteršč, G.; Prebl, I.; Ambrož, M. 2014. The snakng stablty of passenger cars wth lght cargo tralers, Stronšk vestnk-journal of Mechancal Engneerng 60: 539-548. http://dx.do.org/10.5545/sv-me.2014.1690. 13. Yogn P. 2010. Numercal nvestgaton of flow past a crcular cylnder and n a staggered tube bundle usng varous turbulence models, Lappeenranta 14. Spalart, P.; Deck, S.M.S.; Squres, K.D. 2006. New verson of detached-eddy smulaton, Resstant to Ambguous Grd Denstes Theortcal and Computatonal Flud Dynamcs 20: 181-195. http://dx.do.org/doi 10.1007/s00162-006-0015-0. 15. Spalart, P. 2000. Strateges for turbulence modellng and smulatons, Int. J. Heat Flud Flow. 21: 252-263. http://dx.do.org/10.1016/s0142-727x(00)00007-2. 16. Cuctore, R.; Quadro, M.; Baron, A. 1999. On the effectveness and lmtatons of local crtera for the dentfcaton of a vortex, Eur. J. Mech. B/Fluds. 18(2): 261-282. http://dx.do.org/10.1016/s0997-7546(99)80026-0. Maro Telenta, Mataž Šubel, Jože Tavčar, Jožef Duhovnk DETACHED EDDY SIMULATION OF THE FLOW AROUND A SIMPLIFIED VEHICLE SHELTERED BY WIND BARRIER IN TRANSIENT YAW CROSSWIND S u m m a r y Ths paper numercally nvestgates the flow around a movng vehcle sheltered by a wnd barrer. Flow control strategy amed at reducng the aerodynamc loads on the road vehcles requres a detaled knowledge of the reference flow. Exploratory study s conducted n understandng the flow physcs nvolved n the wnd barrer vehcle protecton. Transent nature of the crosswnd gust s represented by the tme-dependent boundary condtons n Detached Eddy Smulaton. Three methods to quantfy and vsualze the effects of the transent crosswnd on the movng vehcle are consdered: aerodynamc coeffcents, pressure mappng, and flow vsualzaton. Aerodynamc forces and moments actng on the vehcle are consderably lower compared to the case wthout the barrer shelter. Vehcle s crosswnd senstvty s mproved by wnd barrer applcaton n crosswnd protecton. The am of ths paper s to quantfy and vsualze the barrer nfluence on the movng vehcle aerodynamcs n transent crosswnd scenaro. Keywords: wnd barrer; vehcle aerodynamcs; DES, gust wnd; coherent structures. Receved December 17, 2014 Accepted March 20, 2015