AERODYNAMICS AND AEROACOUSTICS OF AIRFLOW

  • 格式:pdf
  • 大小:2.94 MB
  • 文档页数:15

AERODYNAMICS AND AEROACOUSTICS OF AIRFLOWOVER A HUMAN HEADFUYIN MA *,JIU HUI WU *,‡,FU GANG *,CHANG AN BAI †and LI QI †*School of Mechanical Engineering State Key Laboratory for Strength and Vibration of Mechanical Structure Xi’an Jiaotong University,Xi’an 71009†Shanghai HikeyTech,Inc.,Shanghai 200235‡ejhwu@ Received 29December 2013Revised 14March 2014Accepted 19March 2014Published 11April 2014Airflow should be affected by ear,nose,etc,and aerodynamic sound could be brought when the human head and the airflow occurs at a relative speed movement.Because the drag force of the head is large,in cycling competitions and other high-speed sport the diversion hats should be worn.After the speed of airflow reaches to a certain speed,due to the strong airflow inter-ference,the aerodynamic noise could be brought and it could greatly impact on the athletes to make action decisions.In this paper,computational fluid dynamics (CFD)method is used for solving the aerodynamic behavior of a human head under different air velocities,and the pressure on head surface and the airflow around the head are calculated.Then,the above-mentioned conditions of different aerodynamic sound are solved by the finite element/infinite element method (FEM/IFEM),the points in the canal entrance of the two ears are picked up for collecting the SPL spectral curves,and the sound distribution of the horizontal plane and the median plane are drawn.Previous studies showed that the aerodynamic noise brought by head spoiler is obvious at low frequencies,because the influence of the head,if the airflow speed is greater than a certain value,the aerodynamic noise could not only increase,but also should be substantially reduced.The results are useful for athlete diversion cap design and spatial hearing research in which the influence of the airflow should be considered.Keywords :Aerodynamic;aeroacoustics;human head;drag coefficient;CFD.1.IntroductionThe previous biomechanical study is mainly focused on the dynamics behaviors of bone joint motion,1,2the viscoelastic mechanical behavior of the muscle tissue,2mobility and rheological behavior of the blood of the cardiovascular system.How-ever,for the hydrodynamic behavior of the organism in the fluid (air and water)and the interaction with the fluid,when the relative movement occurs,the relative research is poor.Many sports projects involved faster relative motion between the human and air,such as sprinting,cycling race,skating competitions.A similarJournal of Mechanics in Medicine and BiologyVol.14,No.5(2014)1450068(15pages)°c World Scientific Publishing Company DOI:10.1142/S0219519414500687J . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .situation also existed in water,especially when swimming,the body could be obstructed by water,and the score of the athletes are also affected.In fact,the \skin-like "swimsuit of the famous Olympic champion Michael Phelps aroused widespread interest,which is a typical example of combination of the fluid dynamics and human behaviors.In 1990,Ennos 3studied the unconventional air dynamic behavior of insects in flight,found the different lift mechanisms in Drosophila and other insects flying comparing with the lift generated by aircraft.There are also lots of studies regarding the aerodynamic and aeroacoustics of airflow over a subject,such as the aircraft disturbed by the airflow.4–8In 2005,Usherwood et al.9studied the mechanical behavior of greyhound racing at high speed,but did not consider the aspect of the aerodynamic.At present,some researches focused on the aerodynamic counter-attack damage by the flow behavior of the human body when the pilots worked in aircraft,through the drag coefficient of the head and limbs to evaluate the aero-dynamic behavior of the human body.10Such researches are generally done through wind tunnel testing on dummies,the purpose is to develop the deflector shields,play a protective role on the pilots.Fluid dynamics behavior in water is studied by more researchers.11–13The first aid is to optimize the performance of submarines and other underwater weapons through the bionics research,the other is to develop the low resistance swimsuit for athletes.Head especially ear plays a very important role to locate the sound source po-sition.Through the reflection of sound waves and other factors,the desired effect should be achieved to determine the position and distance of the sound source.14,15In the past,people studied the sound source localization effected by the head without considering the influence of the airflow.16However,in some special sports environment,the influence of the airflow is obvious.In addition,when the head is disturbed by the airflow,a great aerodynamic sound will generate at the higher airflow speed,which depends on the aerodynamic behavior of the head.In this paper,computational fluid dynamics (CFD)method is used for solving the aero-dynamic behavior of a human head under different air velocity,which is useful for athlete diversion cap design and spatial hearing research in which the influence of the airflow should be considered.2.Materials and Methods2.1.Model descriptionThe HRTF research is usually processed in an auditory spatial coordinates system,of which the origin is located at the midpoint of two ears,X -axis is pointed to the right side,Y -axis to the front side and Z -axis to the up side.There are three characteristic planes,which are the horizontal plane,median plane and lateral plane,as shown in Fig.1.We built a finite element method (FEM)solution model through a three-dimensional reverse head,in which the ears,nose and mouth areF.Ma et al.J . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .enclosed,and the impedance of the hair and skin is ignored i.e.,the overall head is rigid,the influence of the material is ignored and only the shape of the head is considered.The method of the aerodynamic sound solving requires an unsteady CFD tur-bulent calculation,and the element size of the CFD model needs to satisfy the following relationship between the steps of sampling time:U Át Áx %2:5;ð1Þwhere U is the fluid velocity,Át is the step of sampling time and Áx is the element size.For an acoustical FEM model,atleast six elements per wavelength are re-quired.So the upper frequency which the model can precisely consider is determined by the element size.The head surface is discreted by the triangular elements,and in order to get a more detailed expression for the geometric features of the auricles,the maximum element size of the head is 2mm,as shown in Fig.2(a).Create a cube space,in which the length is 1.3m,the width and height are 0.5m,each,and the origin of auditory coordinates is located in the midpoint of the two ears.Cube surface is discreted by triangular elements too,and the maximum element size is 20mm.Then the mesh of sound propagation space should be generated by the elements oftheFig.2.The mesh of the head and the fluidfield.Fig.1.Space coordinates of head.Aerodynamics and Aeroacoustics of Airflow Over a Human HeadJ . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .cube and head surface which is limited by the maximum size of 20mm,as shown in Fig.2(b).Thus,the CFD head model for the aerodynamic behavior solving is obtained,in which the solid elements constitute the fluid field region,the outermost cube boundary wall surface and the head rigid wall surface constitute the grid wall boundary.The model of Fig.2(b)is also used in acoustical calculation.The difference is that in the acoustic model,the outer wall of the cube is the infinite element surface,where the sound wave could propagate along the original path spread and without any reflection.Since the maximum element size of the entire model is 20mm,so according to the wavelength of six per unit to count,the model can be considered to 2820Hz.Calculation model based on the requirement of six elements per wave-length,divided into 1,791,864linear tetrahedral elements,solved the acoustic field distribution up to 2820Hz.2.2.The solving model and the boundary conditions In fluid field solving,the front side surface is set as the velocity inlet,and a velocity value is given;the behind side surface is set as the pressure outlet,a standard atmospheric pressure is generally given as the original pressure value,the around sides and the head surface are set as the rigid wall and the internal flow area is defined as the air material under normal temperature and pressure.The air is considered as incompressible fluid.Because the surface mesh of the head is dense enough,representing the influence of the boundary layer has been considered,and no boundary mesh added separately.The boundary conditions of the CFD model are shown in Fig.3.Three groups of velocity are selected as 5,12and 21m/s.One hundred steps of the steady iteration are calculated at first until the flow field to stabilize,then do not need to re-initialize the model and start the transient unsteady calculation.According to Eq.(1),at different wind conditions the time-step is different.Unsteady fluid field is solved by the k À"standard models,in-cluding the turbulent kinetic energy equation and the diffusion equation,the for-mula is as the following 17:@@t ð k Þþ@@x i ð k ui Þ¼@@x j þ t p r ;k @k @x jþG k þG b À "ÀY m þS k ;ð2ÞFig.3.The boundary conditions of the CFD model.F.Ma et al.J . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .@@t ð "Þþ@@x i ð "ui Þ¼@@x j þ t p r ;" @"@x jþC 1""k ðG k þC 3"G b ÞÀC 2" "2k þS ";ð3Þwhere G k is the turbulent kinetic energy generated by laminar flow velocity gradi-ent,G b is the turbulent kinetic energy generated by buoyancy,Y m is the fluctuation generated in compressible turbulence,C 1e ;C 2e and C 3e are some constant,p r ;k and p r ;"are the Prandtl number of the kinetic energy equation and diffusion equation,i and j are subscript,u i is the absolute velocity component in x i direction,c ui is the turbulent dissipation rate under u i ,k ui is the turbulent kinetic energy under u i ,s e is the component of the turbulence source term, is the dynamic viscosity.The model is solved by Fluent software in a normal workstation.Data of each solution step are required when solving the aeroacoustics,so we need to save each iteration results of unsteady iterative solution.Due to the number of elements in model is large,all the results not only should consume a lot of disk space,but also should lead to slow solution in the aeroacoustics solving,so just save the pressure data of each step.The points in the canal entrance of the two ears are picked up for collecting the SPL spectral curves,and the sound distribution of overall field could be obtained.Acoustic finite element analysis software Actran can complete the FFT trans-formation directly for unsteady field results obtained in the CFD solution,and transform the data into the frequency domain pressure fluctuations.Then through the Lighthill acoustic analogy method (corresponding to the low Mach number conditions),analogy aeroacoustics monopole,dipole sound source term can be obtained by the pressure fluctuations,and the sound distribution could be solved.The flow field is set to a Lighthill volume,the outer surface of the cube is set to an infinite element surface.The aeroacoustics solving model is as shown in Fig.4.Fig.4.The aeroacoustics solution model for Actran software.Aerodynamics and Aeroacoustics of Airflow Over a Human HeadJ . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .After an aeroacoustics solving,the sound distribution of various frequencies and the spectral curve of the sound pressure at any point of space can be obtained.The aeroacoustics calculation could cost lots of time,the more the computation itera-tions,the greater the time costs,while too little iterations can cause the instability results.In this study,three groups of velocity corresponding to the CFD calcula-tions are selected as5,12and21m/s.SPL spectral curves in the points in the canal entrance of the two ears are picked up,and the sound distribution sectional contours of some special frequency points are drawn.3.Results and Discussion3.1.The grid-independency testingIn order to validate the solution model,grid-independency is tested in this section. Because the solution accuracy of the aeroacoustics depend on the accuracy of the CFD solution,only the CFD model in different mesh densities is tested.There are three groups mesh density chosen,171,827,1,438,825and1,791,864elements,re-spectively.Figure5shows the mesh of these groups.The last group is used in later sections.For the condition with the velocity of21m/s,after1000times unsteady iteration with the time-step of0.0003s,Fig.6shows cross-sectionalflow velocity contours in the median pared the maximum velocity values of each group with Fig.7(c),it suggests that the results of thefirst group is very different with the results of the other two groups.It means the mesh density of thefirst group is not enough to complete the solution.And both the contours shape and the velocity values of the second and third groups are close to each other,so the mesh density of the last group is enough to support the later calculations.3.2.Fluidfield distributionIn order to obtain the change in theflowfield around the head more clearly,the cross-sectionalflow velocity contours of each given velocity of inlet conditionsin Fig.5.Three groups mesh density for these testing,171,827,1,438,825and1,791,864elements,respectively.F.Ma et al.J.Mech.Med.Biol.214.14.Downloadedfromwww.worldscientific.combyHUAZHONGUNIVERSITYOFSCIENCEANDTECHNOLOGYon6/22/15.Forpersonaluseonly.median plane and the pressure contours of head surface are drawn and as shown in Fig.7.From the velocity contours of Figs.7(a)–7(c),it suggests that with the increase of flow speed,the interference of the head and airflow become stronger.At 5m/s the entire flow field is disturbed by the head;when the velocity increases to 12m/s and 21m/s,only parts of flow field is more significantly disturbed,and at the top,local high velocity region appeared.Overall,mutual interference behaviors of the head and flow are obvious at the higher velocity,the closer the front face,the slower of the flow;at the top area,the closer to the head,the higher the flow;at the rear area,the closer to the head,the slower the flow.In addition,regardless of the speed is high or slow,a low-wake zone at the rear side could be generated.From the pressure contours of Figs.7(d)–7(i),it suggests that in the front face region of the head,the pressure is greater than in the top and rear regions,i.e.,the Fig.6.The cross-sectional flow velocity contours in the median plane of the first twogroups.(a)(b)(c)(d)Fig.7.The cross-sectional flow velocity contours of each given velocity of inlet conditions in the median plane and the pressure contours of the head surface.(a)Cross-sectional velocity contours at speed of 5m/s;(b)cross-sectional velocity contours at speed of 12m/s;(c)cross-sectional velocity contours at speed of 21m/s;(d)cross-sectional pressure contours at speed of 5m/s;(e)cross-sectional pressure contours at speed of 12m/s;(f)cross-sectional pressure contours at speed of 21m/s;(g)the pressure contours of head surface at speed of 5m/s;(h)the pressure contours of head surface at speed of 12m/s;(i)the pressure contours of head surface at speed of 21m/s.Aerodynamics and Aeroacoustics of Airflow Over a Human HeadJ . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .pressure of these regions is small.With the increase of the velocity,the pressure of the flow field becomes uniform,only the regions close to the head retain some uneven distribution.Further,from the figure,it also suggests that the pressure around the nose region is greater than other regions,i.e.,the influence of the airflow is more obvious;and the pressures of the auricle regions are relatively small,the influence of the airflow is weak.It indicates that the auricle structure of the aero-dynamic characteristics is relatively optimized compared with the overall head,so in the high-speed movement process it could not generate a greater additional re-sistance caused by the ear.3.3.Sound field distribution According to Eq.(1)the minimum allowable iteration time-step of CFD solution can be determined,and this step directly determines the cutofffrequency of theacoustical model.So at the speed of 5m/s,the minimum iteration step is 0.01s and the highest cutofffrequency of the acoustical model is 100Hz;at the speed of 12m/s,the minimum iteration step is 0.004s and the highest cutofffrequency is 250Hz;at the speed of 21m/s,the minimum iteration step is 0.002s and the highest cutofffrequency is 500Hz.For the sound field solution,the speed of 5m/s corre-sponds to the calculation frequency from 10–100Hz,steps by 2Hz;12m/s corre-sponds to the calculation frequency from 10–250Hz,steps by 5Hz;21m/s corresponds to the calculation frequency from 10–500Hz,steps by 5Hz.The bin-aural frequency response functions of the monitoring points are shown in Fig.8(a).(e)(f)(g)(h)(i)Fig.7.(Continued )F.Ma et al.J . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .The sensitivity frequency band of the auditory is mainly located in the range of 1000–5000Hz,especially from 1000Hz to 3000Hz.In order to obtain the sound field distribution of the higher frequency band,only the higher speeds of 12m/s and 21m/s conditions are solved up to cutofffrequency of 3000Hz.The binaural fre-quency response functions of the monitoring points are shown in Fig.8(b).(a)(b)Fig.8.The binaural frequency response functions of the monitoring points at different velocity condi-tions.(a)The results in different cut-offfrequency;(b)the results in upper cut-offfrequency of 3000Hz.Aerodynamics and Aeroacoustics of Airflow Over a Human HeadJ . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .Figure 8(a)suggests that the total SPL gradually increase with the decrease of frequency,and followed a stable trend.In the low frequency the SPL is very large,at 12m/s conditions,the maximum SPL reaches to 115dB.When the velocity increases from 5m/s to 12m/s,the SPL gradually increases with the speed below 100Hz;when the velocity increases to 21m/s further,the SPL decreases unsteadily.Furthermore,the figure also suggests that the SPL is greater in the low frequency ranges,and the SPL is lower in high frequency ranges,of which the human auditory is more sensitive to sound than other ranges.Just based on this characteristic,the SPL of the A-weighting is attenuated a lot in the low frequency ranges.In addition,except at 12m/s conditions,there exists a band where the SPL of left ear is smaller than the right ear,under the remaining frequencies,is greater than the SPL of the right ear,which is probably caused by vortex turbulence of the asymmetric flow field.Figure 8(b)suggests that the spectrum changes very largely above 400–500Hz,and appears as a significant valley.At 12m/s,the right ear appears as a deep valley at about 450Hz,whereas the left ear appears as a deep valley at about 1800Hz.At 21m/s,the left ear appears as deep valley at about 1300Hz.From the attenuation trends of the SPL,the value of SPL is no more than 80dB at 12m/s above 500Hz,and the minimum value is only 50dB;while at 21m/s,the maximum value islessFig.9.The sound distribution contours in the median and horizontal plane at 42,82and 100Hz with the velocity of 5m/s.Sound distribution contours in the median plane at (a)42Hz,(b)82Hz,(c)100Hz;sound distribution contours in the horizontal plane at (d)42Hz,(e)82Hz,(f)100Hz.F.Ma et al.J . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .than 40dB,the minimum reaches nearly to 0dB.This indicates that,on the one hand,the aerodynamic noise is mainly obvious in the low frequency ranges;on the other hand,since the influence of the head,when the airflow speed is greater than a certain value,the aerodynamic noise should not only increase,but also should be substantially reduced.In order to obtain the sound field distribution under different flow conditions at different frequencies,Fig.9shows the sound distribution contours in the median and horizontal planes at 42,82and 100Hz frequencies with the velocity of 5m/s;Fig.10shows the sound distribution contours in the median and horizontal planes at 55,155and 250Hz frequencies with the velocity of 12m/s and Fig.11shows the sound distribution contours in the median and horizontal planes at 55,255and 500Hz frequencies with the velocity of 21m/s.The figure indicates the spoiler role of the head dues to a large vortex appearing in front of the head of the sound field and a series of vortices in behind,similar to the horseshoe vortex of a flow field.In comparison,when the flow velocity is slow (5m/s),vortex distribution is more obvious in front and behind sides and exhibit both an obvious high sound pressure region and a low sound pressure region,the maximum SPL of the high sound pressure region exceeds 100dB,but the SPL of the low sound pressure region is less than 10dB.When the velocity becomehigherFig.10.The sound distribution contours in the median and horizontal plane at 55,155and 250Hz with the velocity of 12m/s.Sound distribution contours in the median plane at (a)55Hz,(b)155Hz,(c)250Hz;sound distribution contours in the horizontal plane at (d)55Hz,(e)155Hz,(f)250Hz.J . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .(21m/s),sound pressure distribution becomes more uniform in front of the head region,and behind of the head a higher sound pressure shaded region appears,and a low pressure zone appears behind of this shaded region.From the sound field dis-tribution in horizontal plane,it suggests that in the low speed condition,the sound field of the left and right sides are asymmetric;but in the high speed conditions,the sound field tends to symmetric.These results are also indicated from the frequency response curves of Fig.8,when the velocity reaches to 21m/s,the difference of the SPL between the two ears is small,barely 1dB.The numerical model of this study has been validated by the results reported in the literature,in which the airflow has not been considered.By comparing thecurves of Fig.8with the curves of Fig.12,it suggests that there are some similar trends of the binaural frequency response functions,such as the valley feature.However,the valley will move along the low frequency direction when the head is influenced by the airflow.In Fig.8,the first valley of the right ear at 12m/s is generated at 460Hz,and the first valley of the left ear at 12m/s is generated at 1800Hz,and the first valley of Fig.12is generated at 1900Hz.The first valley of the left ear is most close to the one of Fig.12.The main reason is that the sound source of this study is the aeroacoustic which is brought by airflow,but the sound source of Ref.14is a fixed plane-type soundwave.Fig.11.The sound distribution contours in the median and horizontal plane at 55,255and 500Hz with the velocity of 21m/s.Sound distribution contours in the median plane at (a)55Hz,(b)255Hz,(c)500Hz;sound distribution contours in the horizontal plane at (d)55Hz,(e)255Hz,(f)500Hz.J . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .4.Conclusions Through the aerodynamic sound field and CFD flow field calculation and analysis,the distribution of flow field and sound field can be summed up,including the following points:(1)The distribution of flow field indicates that with the increase of flow speed,the interference of the head and airflow become stronger.At high flow speed,the interference is obvious,showing offin front of the face,the closer to the face,the slower the airflow;in the top region,the closer to the head,the faster the flow;behind the head,the closer both head,the slower the speed.(2)With the increase of the velocity,the whole flow field pressure becomes uniform except the region near the head.This shows that the auricle structure of the aerodynamic characteristics is relatively optimized,and in the high-speed movement process they could not generate a greater additional resistancecaused by the ear.(3)From the binaural frequency response curves,it indicates that,on the one hand,the aerodynamic noise mainly distributes in the low frequency ranges;on the other hand,with the influence of the head structural characteristics,when the airflow is higher than a certain speed,aerodynamic noise should not continue to increase,but should be greatly reduced.In the low frequency ranges,with the less sensitive human auditory,noise level is relatively higher,in the medium and high frequency ranges,with the more sensitive human auditory,SPL is rela-tively lower.This rule is consistent with a weight of human auditoryexperience.Fig.12.A group HRTF data at 240 azimuth,in which the fixed sound source has been considered.The solid line and the dotted line show the results obtained by the source independent method and the newly proposed method in Ref.14,respectively.J . M e c h . M e d . B i o l . 2014.14. D o w n l o a d e d f r o m w w w .w o r l d s c i e n t i f i c .c o mb y H U A Z H O N G U N I V E R S I T Y O F S C I E N C E A N D T E C H N O L O G Y o n 06/22/15. F o r p e r s o n a l u s e o n l y .。