RESILIENT INFRASTRUCTURE Tweens CERES, LONDON 2016 EFFECT OF WIND SPEED AND TERRAIN EXPOSURE ON THE WIND PRESSURES FOR ELEVATED STEEL CONICAL TANKS ‘Ahmed Musa Ph.D Candidate, Westen University, Canada Haitham Aboshosha Research Engineer, Boundary Layer Wind Tunnel Laboratory, Westem University, Canada ‘Ashraf El Damatty Professor and chair, Western University, Canada ABSTRACT Stee! liquid storage tanks in the form of truncated cones are commonly used as containment vessels for water supply ‘or storing chemicals. A number of failures have been recorded in the past few decades for steel liquid tanks and silos under wind loading. A stecl conical tank vessel will have a relatively small thickness making it susceptible to buckling under wind loads especially when they are not fully-filled. In this study, a wind tunnel pressure testis performed on ‘an clevated conical tank in order to estimate the extemal wind pressures when immersed into a boundary layer. The tested tank configuration represents combined conical tanks where the cone is capped witha eylinder. In addition, the cffect of terrain exposure and wind speed on the pressure values and wind forces is assessed, The mean and ms pressure coefficients are presented for different test cases in addition to the mean and rms total drag forces that are ‘obtained by integrating the pressure coefficient over the tank model’s surface. Itis found that the total mean and rms Buia TaD go |] & aevwrtistpusn) 0 of ag] & Buvtisbsnn ASCE SCE ( |_2= sce Py | WES asco) 5 50 # so Bao 320 Es E50 a a Vi, 20 20 V3 %, 0 wa 0 ‘SZ » @ , sp ~~~ 00 0s 10 0 0 20 30 40 50 Normalized wind speed Turulence Intensity (%) Figure 3: (a) Mean wind speed profiles; (b) Turbulence intensity for open and suburban terrain exposures ‘Typically, pressure taps on the model are connected through pneumatic tubing to electronic pressure seanners, each capable of handling 16 different taps. The data acquisition system records the digitized signals fom the scanning ‘modules ata scanning sampling rate of 400H7, The reference dynamic pressure in the free stream above the boundary layer is recorded similarly. All of this data is stored for the determination of maximum, minimum, mean and rms pressure values for each tap, More details on the experimental technique for pressure models can be found in BLWTL (2007) i @ 0) Figure 4: (a) Combined tank model, (b) Pressure taps’ layout (Dimensions are in em), (c) Cire distribution of pressure taps NDM-536-44. TEST RESULTS As mentioned, pressure time histories are recorded simultaneously at each pressure tap. These records are analyzed in ‘order to obtain mean and rms pressures in addition to total drag and lift forees acting on various components of the ‘conical tank model. The mean and rms pressures are reported in terms of mean and rms pressure coefficients Cand Goons defined as [1 Gon= EF C0/0 (2) Cm J4C C4 (0 — Cy) where (C,(0= pt) — p, /0.5 pV) with p(t) i the pressure measured at time instance , po is the mean static reference pressure, pis the air density, n is the number of recorded time increments, and ¥ isthe wind speed at 40 em height ‘which represents the roof height of the combined model 4.1 Mean pressures Distribution ofthe mean pressure coefficient Cym over the tank walls is investigated for different test eases. Effect of changing wind speeds and terrain exposure on Cm variation with the circumferential angle 0 is discussed. A circumferential angle 0-0" represents the windward side (ic. stagnation point), while 0-180" represents the leeward side angle. Figure 5 shows the effect of changing wind speed on Cy values where the horizontal axis represents the circumferential angle @ and the vertical axis represents the Cm values. Fach subplot shows Cux-0 relation for the three model components at different levels corresponding to certain wind speed. A typical Gym Variation with the angle 0 can be divided into four zones as shown in Figure Sd: zone 1 represents the positive pressure region starting from the stagnation point to the point of zero pressure, zone 2 extends from the end ‘of zone I tll the point of lowest pressure (maximum negative pressure), zone 3 extends from the end of zone 2 tll the point where the pressure becomes constant, and zone 4 represents the nearly constant negative pressure region acting ‘on the leeward cylinder region a 2 0 30 © 9% 120 150 180 0 30 90 120 150 180 0 (Degrees) Boegrees} as 15 (98) Versa 1 [Qe maent i = tc us) === = te (70%) — crn —— cam Ta aT cron 5 anes cr pen) © 30 60 90 120 150 180 0 30 6 0 170 150 :80 — 1) 0 (Degrees) 0 (Degrees) T= vc ae Figure 5: Distribution of Cj at different levels for open exposure test eases: a) Viar=4.6n0Vs ,b) Viar=9.Lmns, ¢) Vie 13.7m's, and d) Vis 18.35 NDM-536-5By investigating Figure S regarding the effect of wind speed on Cyn Variation, itis found that: the angle at which zone 2 starts shifts away from the stagnation point at higher levels and the maximum negative Cm value increases at higher Wind speeds; the base Cyn is reduced for higher wind speeds except for the case of the lower wind speed (Vier 4,6m’s), Under such a low speed the flow is in the sub-critical regime which may be the reason for that, On the other hhand, Figure 6 shows the effect of changing the terrain exposure on Cy values. Each subplot shows C0 relation for the three model components at different levels corresponding to the two considered terrain exposures, i., open and. suburban, a two wind speeds. By investigating Figure 6 regarding the effect of the terrain exposure on Crm variation for different model pars, itis found that: suburban exposure results in slightly lower positive and slightly higher negative Cyn Values due to the lower wind speeds compared to the open terrain exposure; the difference in Cym Values is reduced for higher levels as the wind speed profiles corresponding to the two exposures approach each other as we {0 far from the ground, ere 6 90 120 150 180 0 30 90 320 350180 ay] 8 (Degrees) 8 (Degrees) COREE an OD Saatar ape I — cn94) 0 3% 60 90 120 250 180 0 30 60 90 120 150 a9 © ———VC|B8») © (Degrees) 0 (Degrees) Figure 6: Distribution of Cy at different levels for testcase: (a) Open exposure (Visr=9.1nvs), b) Suburban, ‘exposure (Vier=9.1m/5), c) Open exposure (Vi47=18.3m/5), and d) Suburban exposure (Vi¢7=18.3mVs) 4.2 Fluctuating Pressure In this subsection, fluctuating pressures are examined and the variation of the rms pressure coefficient Cyne over the tank walls is investigated for different test cases. The effect of changing wind speed and terrain exposure on Cyn: variation with the circumferential angle @ is discussed. Figure 7 shows the effect of changing the wind speed on Cima: values where the horizontal axis represents the circumferential angle @ and the vertical axis represents the Crm values. Each subplot shows Cora-® relation for the three model components at different levels corresponding to certain wind speed. By investigating Figure 7 regarding the effect of the wind speed on Cm Variation, itis found that: Cy. distribution have a more clear peak plateau near ‘zones 2 and 3 for higher wind speeds for both lower eylinder and eonieal part; Comme values are higher in zones 2, 3 ‘and lower in zones 1,4 for higher wind speeds for both lower cylinder and conical part; Cyvas values over the upper cylinder are reduced in zones 2, 3 and 4, with the exception of the lower wind speed (i47-4.6m) in zone 4, for higher ‘wind speeds. NDM-536-6Figure 8 shows the effect of changing terrain exposure on Cys Values. Each subplot shows Cpm-0 relation for the three model components at different levels corresponding 10 the two considered terrain exposures, ic., open and suburban, at two wind speeds. By investigating Figure 8 regarding the effect of the terrain exposure on Cry, Variation, iis found that: Cyne lower cylinder values are smaller for the suburban exposure compared to the open terrain ‘exposure; Crm: Values for the conical part are higher for the suburban exposure in zones 1 and 2 and lower in zones 3 and 4 compared to the open terrain exposure except for the lower wind speed (vi4r=4.6m/5) where Gyms Values is lower for the suburban exposure in all zones; Com for the upper cylinder has lower values in zones 1, 2 and 3 and higher values in zone 4 for the suburban exposure compared to the open terrain exposure except for the lower wind speed (vie=4.6m'8) where Cm, values is lower forthe suburban exposure in all zones, or (@) Vier4.6e or 06 * og [OYerote pos oa net } fl © (Degrees) 6 (Degrees) as os gos gos tegen we ve ashe oe 0 2 == 0%) 8 (oeerees) Bocca’ ca Figure 7: Distribution of Cyn at different levels for open exposure test eases: a) Visr=4.6nvs ,b) Viar=9.1mis, (e) Ver" 13.7av, and d) Vigr=18.3mv3 NDM-536-7[op Openenpone Wie Tm WSs pee IS pos i 3 0 ] o2 : 0 ° i 2 90 20 150 ito 030 6 90 120 150 80 8 (oegrees) 8 (Degrees) bg POR ATTA og, [DabubereaR WaRToR Po os os Fos os oa 01 0 30 60 90 120 150 380 9 30 6 99 120 250 180 uC 3m @ (Degrees) 6 (Degrees) uc (ae Figure 8: Distribution of Cy, at different levels for test case: (a) Open exposure (Vi47-9. m/s), b) Suburban ‘exposure (Vie1=9.1m/5), c) Open exposure (Vig=18.3mv/5), and d) Suburban exposure (V.e7=18.3mVs) 4.3 Mean Drag Forces Drag force, Fy, acting on the tank is evaluated and reported in terms of the drag coefficient, Cy. The relationship between force Fp and coefficient Cy is given by Bl Ge where p is the air density, 7 is the reference wind speed and A is the projected area subjected to wind pressure, The coefficient Cy is evaluated by integrating the pressure coefficient C, acting at various taps using the following tributary area around each tap. ‘expression, where A; i th 1608 9,4, /A Mean drag coefficient, Cu, can be evaluated from the Cy time series at each pressure taps’ level. Figure 9a shows the distribution of Car along the tank height for the eight tested cases corresponding to four wind speeds (i.e. Vir= 4.6, 9.1, 13.7, and 18.3 mvs) within the two considered terrain exposures (i.e. open and suburban). The following findings cean be obtained from this figure: (1) Cay distribution for the lower eylinder has a maximum value around the mid- hhcight except for the lowest wind speed (ic., vir4.6mvs) winere the max Cus is atthe top height, (2) For the conical part, Cay is increasing linearly with the eight excluding the lowest wind speed (Le, vier=4.6m/s), (3) For the top cylinder, Caw is reduced with the height, (4) Cay values for the suburban exposure are lower compared to the open terrain forall pars ‘As discussed earlier, the mean drag force for circular structures is dependent on Reynolds number Re which represents the ratio of inertial to viscous forces within the flow. As a result, the variation of the mean overall drag coefficient én with Reynolds number for different model parts is investigated. The mean overall drag coefficient Camm is ‘obtained for each part by integrating the Cin values shown in Figure 9a over the height of this part. Figure 9b shows the variation of Cipoe with Re for different parts. As the wind speed varies with height and the model diameter is not ‘constant forthe conical part, the considered Re is based on the wind speed and tank diameter atthe top of each part It is found that the mean overall coefficient is dependent on the Re and the following findings can be deduced from Figure 9b: (1) For the lower eylinder, Cam is reduced with increasing Re expect for the case of open terrain with NDM-536-8lowest wind speed (1e., vir=4.6mis), (2) For the conical part, Cjom is reduced with inereasing Re expect for the case of open terrain with lowest wind speed (i-e., viei=4.6m/5), (3)For the upper cylinder, Cin is slightly reduced with increasing Re, (4) Cayyy is found to be higher for all parts tested within the open terrain due to the higher wind speed profile compared to the suburban one. pense) =~ open 1m) sore Sita) Sten ns) TEST clita = Sabata(13 04) sa oy 00 oo 0s 10.18 a0 2s 0008 1.008 1.00806 Con R Figure 9: (a) Height-wise distribution of mean drag coefficient Cum, (b) Variation of mean overall drag coefficient Cao for each part with Re 444 Fluctuating Drag Forces In addition to the mean value, rms drag Cinq, coefficient is also evaluated from the drag time series. Figure 10a shows the distribution of Cans along the tank height for the eight considered test cases corresponding to four wind speeds Ge. Visr= 4.6, 9.1, 13.7, and 18.3 mis) within the two considered terrain exposures (ie. open and suburban). The following findings can be obtained with regards to the 1ms drag coefficient Cys: (1) Cam distribution for the lower cylinder has a maximum value around the mid-height, (2) For the conical part, Cim increases with the height, (3) For the top cylinder, Cigy reduces with the height, (4) Car Values for the suburban exposure are higher compared to the ‘open exposure duc to higher turbulence intensity. Similar to the case of mean overall drag, the variation of the rms overall drag Cane with Reynolds number for different model parts is investigated. The rms overall drag Cae Coefficient is obtained for each part by integrating the Cur Values shown in Figs. 10a over the height of this part. Figure 10b shows the vasiation of Casyers with Re for = tg Tir citar Sata) oo oo 3s * 06 = ae] os gn pos Bis os 2 10 02 5 ° ° 00 01 02 03 04 05 06 07 og L_______ oorso8—«.00EOS 0 Cems Re Figure 10: (a) Height-wise distribution of rms drag coefficient Crm, (b) Variation of rms overall drag coefficient ‘Caer: for each part with Re 5. CONCLUSIONS Inthis study, a wind tunnel testing on scaled elevated conical tank models is performed in order to estimate the external wind pressures in addition to the total wind forces, The model is tested within a boundary layer using two terrain, ‘exposures which are open and suburban in order to assess the effect of the terrain exposure on the wind pressure values. In addition, each wind tunnel test corresponding to a terrain exposure is repeated for different wind speeds in ‘order to assess the effect of changing Reynolds number on the measured pressures and the calculated forces. ‘The wind pressures’ time series are recorded and both mean pressure coefficient Ce and rms pressure coefficient hme are obtained over the model's surface for different test cases. By integrating the pressure coefficient, the mean drag coefficient Cay is obtained at different heights and the following is concluded: (1) Cig forthe lower cylinder has ‘2 maximum value around the mid-height, while inereasing linearly with the height for the conical part except for the lowest wind speed, and reduced with the height for the upper cylinder; (2) Cay values for the suburban exposure are Tower compared to the openterrain. Integrating the mean drag Cay over the height of each part wll result in the overall drag coefficient Cin corresponding to each model part and it is concluded that Camm is higher for the case of open ‘exposure due to the higher wind speed profile compared to the suburban ot By plotting Cin for each part against Re, the following is concluded: (1) For the lower cylinder, Camis reduced with increasing Re for different testing levels expect for the case of open terrain with lowest wind speed; (2) For the conical part, Canm is reduced by increasing Re for tae cascs where the lower cylinder is included expect for the case open terrain with lowest wind speed; (3) For the upper cylinder, Cio is slightly reduced for higher Re values Following the same integration procedure for rms pressure coefficient, rms drag coefficient Cum at different levels is ‘obtained in addition to rms overall drag Capone and the following is concluded: (1) Ciny distribution for the lower cylinder has a maximum value around the mid-height; (2) Cj, increases with the height forthe conical part; (3) Cams is reduced with the height for the top cylinder; (4) Car values for the suburban exposure are higher compared to the ‘open terrain due to higher turbulence intensity. Regarding the overall drag coefficient, the following is concluded: (1) For the upper cylinder, Cass is reduced with increasing Re expect for the case of within open terrain with lowest wind speed, (4) Casemris higher for the case of suburban exposure due to the higher turbulence intensity. By plotting Cin: for each part against Re, the following is concluded: (1) For the conical part, Carymsis reduced with increasing Re; (2) For the upper cylinder, Cine is teduced with increasing Re expect for the test case within open terrain corresponding to lowest wind speed NDM-536-10REFERENCES, Asi 1012. Wind Tunnel Testing for Buildings and Other Structures, Reston, VA: ASCE, Baban, F. and So, R, 1991. Aspect ratio effect on flow-induced forces on circular cylinders in a cross-flow. Experiments in Fluids, 10:313-21 BLWTL, 2007. Wind Tunnel Testing: A general outline. Western University, Boundary Layer Wind Tunnel ESDU, 1980. Mean forces, pressures and flow field velocities for circular eylindrical structures: single cylinder with ‘two-dimensional flow. Engineering Sciences Data Unit , ESDU data item 80025. Faleinelli, OLA. et al, 2011. Influence of Topography on Wind Pressures in Tanks Using CFD. Latin American Applied Research, 41:379-88 Flores, F.G. and Godoy, L.A., 1998, Buckling of short tanks due to hurricanes. Engineering Structures, 20(8):752:60. Fox, T.and West, G., 1993a, Fluid-induced loading of cantilevered circular eylinder in a low-turbulence uniform flow. Part 1: mean loading with aspect ratios in the range 4 to 30, Journal of Fluids and Structures, 7:1-14, Godoy, L.A., 2007, Performance of Storage Tanks in Oil Facilities Damaged by Hurricanes Katrina and Rita, J Perform, Constr. Facil.,21(6):441-49. ‘Macdonald, P.A. eta., 1988, Wind Loads on Circular Storage Bins, Silos And Tanks: I. Point Pressure Measurements ‘on Isolated Structures. Journal of Wind Engineering and Industrial Aerodynamics, 31:165-88. ‘Okamoto, T. and Sunabahiti, ¥., 1992. Vortex shedding ftom a circular eylinder of finite length placed on a ground plane, Journal of Fluids Engineering, 114:512-21 Parammasivam , KM. and Tamura, Y., 2007. Aerodynamic Characteristics of a Conical Water Tank. In Proce ofthe Annual Meeting, Japan Society of Fluid Mechanies., 2007. dings Rossana , J.C. and Godoy, L., 2010. Wind buckling ofimetal tanks during their construction. Thin-walled structures, 48(6):453-459. Sabransky, IJ. and Melbourne, W.H., 1987, Design Pressure Distribution on Citeular Silos With Conical Roofs. Journal of Wind Engineering and Industrial Aerodynamics, 26:65-84. Sundaravadivel, T.A. etal, 2009, CED Prediction of Wind Pressures on Conical Tank. Journal of Wind Engineering, 6545-55, ‘Uematsu, Y. ot al., 2014. Design wind foree coefficients for open-topped oil storage tanks focusing on the wind- induced buckling, Jounal of Wind Engineering and Industrial Aerodynamics, 130:16-29. Uematsu, Y. et al, 2015. Design wind loads for open-topped storage tanks in various arrangements. Journal of Wind Engineering and Industrial Aerodynamics, 138:77-86, NDM-536-11