the effect of a gauze of arbitrary shape (and therefore arbitrary ... The trouble with the origin&i gauzes of Baines was that the velocity profile decayed with ...
AmPosphcric Environment, Pergamon Press 1948. Vol. 2, pp. 73-76. Printed in Great Rritain.
RESEARCH NOTE METHODS OF PRO~~~~~~
PROFILES IN
(Received 15 August 1967)
O~ER the last 60 yeara those concerned in the science of ‘*aircraft aerodynamics” have gradually been improving their ability to specify the correct conditions for wind tunnel experiments and have beets able to interpret the results more surely. It is hardly surprising that those concemed in the infant science of “industrial aerodyuamics” should be unsure at this stage. They can build upon the aircraft technology, but must advance step by step, checking that the assumptions made in the aircraft field apply to their own and that the various parameters have the same relative importance. Where this does not happen, they must start at&h. For most of its flying time, an aircraft moves at high speed relative to the air at a height where turbulence, except for clear air turbulence, is very low. Accordingly, for his laboratory experiments, the aircraft aerodynamic& requires wind tunnels which have constant velocity across the working section and a low level of turbulence (altbough at times turbulence is deliberately added so that tests carried out at a low Reyuolds number, usually subcritical, will correspond more nearly to flight conditions at a much higher Reynolds number). One branch of “industrial aerodynamics” has to do with the performance of structures, cables or plumes in the atmosphere and it is important for the industrial aerodyuamicist to he able to model the atmosphere iu the wind tunnel. The chief characteristics of the lower atmosphere are a variation of mean velocity with altitude and a high level of turbulence, also vary@ with altitude. There are several methods of producing such a variation and this paper attempts to list the major methods and to give references to papers describing a few detailed attempts of each type. The first problan is to determine what the atmosphere is like that we are trying to simulate, The atmosphere itself varies from second to second and place to place, so the best that can be done is to define a “standard atmosphere” with constants and indices such that a reasonable likeness to the atmosphere at a given place and time can be produced. The major parameters which are used to detine the atmosphere are either velocity terms or density terms and their variations with altitude are important. The velocity terms include mean velocity, intensity of turbulence, scale of turbulence and spectnun of turbulemx, and the density terms either density or temperature. At present, nearly all the effort is concernad with velocity terms and this paper attempts to compare and contrast the different approaches. THE VELOCITY
PARAMETERS
The first attempt of which I am aware of thickening artificially a turbulent boundary layer was done by RLWSANO~P and DIEHL(1952). They were trying to produce a fully developed turbulent boundary layer of 3 in. thickness in which to calibrate hot wire anemometers. From their conclusions I quote “Almost any kind of obstruction on a surface would produce a thickened turbulent flow, but only for a certain class did downstream effects disappear in a reasonably short distance. This class appeared to be characterized by the ability to absorb euergy from the mean flow without introducing large scale disturbance, Two examples of this class are as follows: 1. A fine-mesh srzreen placed so that the plate and screen form a narrow wedge opening into the wind. 2. Sand roughness cemented to the surface. From this point the techniques have split into four main types: 1. Gauzes or honeycombs; 2. Rods; 3. Flat plates; 4. Obstructions (with or witbout mixing); and we will follow the attempts under these headings. 73
74
Research Note 1. GAUZES
OR HONE’
COMBS
Gauzes had long been used to remove velocity changes f om an airstream, for example in the work of COLLAR(1939). Their use to produce velocity gradient was iried by WXNTERMTZ and RAMSAY (1957) who found neither of the obstruction meth& above suitable either due to inconvenience or probable inability to repeat conditions reliably or on grounds of space. They decided ro retard the flow in the boundary region by means of projecting screen rings. All screen rings were made from X-mesh woven wire cloth, and they conclude “A novel method of generating specific boundary layer conditions by means of project@ annuiar screens has been applied to diSii research.” ELD~ fl9S9) studied ~~rcti~y the effect of a gauze of arbitrary shape (and therefore arbitrary ~~bution of blockage) and related his theory to a series of known experiments, and BAINES(1963) showed the results of experiments conducted using curved gauzes. The trouble with the origin&i gauzes of Baines was that the velocity profile decayed with distance do~tr~ and he had to work fairly close behind the screens prod the profile. Recently, however (see WHITB~ZXD,1%7), Bainea has written computer programmer which allow him to c&ulate the shape of a screen for any required velocity profile. Honeycombs have also been used: Baines (see Wm-ra~cu>, 1967) used a honeycomb graded so that the cell kngth decreased with height and showed it was a practical proposition. An intere&ng modification of honeycombs was used by STARR(1966) who cast a series of rubber pfugs and laboriously filled up cells of his honeycomb by trial and error to produce the required veIocity profile. 2. RODS When using gauzes, the spacing of the wires was constant over the gauze (except in the few examples where gauzes were hammered to increase this bIockage locahy or even produce a graded bIockage). The vertical wires were mainly useful to position the horixontal rods evenly. If the diameter of the rods was increased, this support became unnv and the result is a bank of rods. These were exam&d theoretizally by OWEN and Zmrnrr~wrcz (1957) and a means of producing a linear velocity gradient detailed. Their theoretical results were corroborated by experimentsdescribed in their orighml paper. Thereafter many people, Lrvxsk~ and T(19&u) at Manchester and BROWN at Bristol, used their method, though in the limit final corrections to spacings were usually made by trial and error. Meanwhile, WIEGHARDT (1953) had shown that the performance of screens could be related to the and Lxx (1966) performance of the wires of which they had been made. This probably led Cm to adapt Elder’s method to produce a theory applicable to rods which produced good results except in the regions of high shear such as close to the surface. As this is a region of pan interest, Cowna.x~ (1%7) made a completely new approach which has resulted in a method of producing the desired velocity profile accurately, with fairly simple calculations. Most of the above exp&menters used circular rods, although the use of slats is quite possible and was used, for exampIe, by %X’ION. 3. FLAT
PLATES
The use of gauzes and rods, although the simplest, easiest aad most compact method, traJ the drawback that it is impossible to vary the turbufence in scale, ma&u& and spectrum once the size oftherodshiub&nChoSgl.~c~SQLki9thed~~offhet~~dthena~lfIcqumcy is the shedding frequency of the vortices. It would be extremely de&able to be able to vary velocity profile and intensity of turbulence profile independently. To this end, a grid of Sat plates, the diiierential spacing of which produced the velocity gradient has been built (STROMand ILem, 1967; and STROM,1%2). Here the ve&ity gradient is produced by boundary layer action rather associated with boundary iayers and not wakes. Gonsequently~ thanwakeactionandhaatheturbuknce it is possibk to produce velocity probes with a much smaller intensity of turbulence than with rods. Ho~v~~itisusuallytheaucthatmon:~b~b~~,andInthiscueitcank,~~ by pla&ql either turbularce or vortex generators (depend& upon se&e) upon the Sat piatee. Usk~ this method LUND (1967) was abk to produce the velocity and turbuknce proiue that ht ~8~t.04 but
Resarch Note
75
the~ofturbuleeawaosmaa~scalewiationswaeobtaincdbyusingaretucn~owtype windcunml9ndobtainineeitherreaectioasfromsomepartofthetunncloronthevariationscoming round again in mod&d form. An interesting feature of Lloyd’s work was the relationship between the velocity profile once produced and the subsequent floor roughness. He showed that it was possible to Lpnaate a profile at the start of a working section whose floor had a given surface roughness and, if the two were tobethesamc. compatibk, for the velocity protie at the end of the working section40 ft doe He ~ucce+&dtoalesscrextentinrepeatingthiswiththe turbulenccprofile.ThisabilitytoProduce a non-developing boundary layer is essential in some work on pollutants where conditions must be constantoveralongdistance. Lloyd also demonstrated the effect of change of surface roughness; a new boundary layer growing up under the original one with a d&rent velocity profile. Tbis is the case for example after the wind has,blown over the sea, it comes to land and blows over a city, and can readily be reproduced in the wind tunnel. 4. OBSTRUCTION Following the original work of Rlebanoff and Diehl, Jm and FRANCK(1963) suggested there was no quick method and the only way to obtain a correct boundary layer was to grow it with the correct surface roughness. Most experimenters cannot spare sufticient distance for this in their tunnels, but others have used this approach including Bains at the University of Toronto, and Davenport at the University of Western Ontario (see WRITBREAD, 1967), but in the latter’s case the 80 ft fetch was not sulilcient and had to be augmented by a step at entry to the tunnel. Lrvmev and TuRmR (19646) used a step to produce a velocity protie at the entry to a diffuser, but were happy for the profile to decay. In this issue, Anr+m-r(1967) described work at C.E.RL. in which he used a solid step coupled with a few large vortex generators to stir up the flow. This had the advantage that it produced large scale turbulence, but produced a marked variation of mean conditions across the tunnel. He also had recourse to one circular rod parallel to and above the step to assist the mean velocity distribution. Armitt also d&usscs the parameters he considers important for the various problems to be studied in the simulated atmospheric boundary layer. OTHER
PARAMETERS
So far I have described the attempts to alter the velocity profiles in a wind tunnel without mentioning the variation of wind direction with height. The other major pammemr is the temperature or, more precisely, the variation of temperature with height upon which depends the stability of the atmosphere. The only worker to have attempted this, to my knowledge, is Strom. The real problem in this simulation is the secondary flows induced in the wind tunnel in the form of large longitudinal VoltiCCs. The general opinion
of workers in this field, without experimental evidence to support it, is that: (i) mostcriticalcascsoccuri.nhighwinds;and (ii) in most high winds the stability of the atmosphere is neutral. It could be that workers take this view as they are at present unable to simulate a lapse rate in the wind tunnel; fimdamental work must be carried out to test this hypothesis and establish its validity. CONCLUSIONS I have outlined several of the many attempts to produce velocity gradients in a wind tunnel. Can reach any conclusion as to the relative merits of each? At present the answer is “no”, because we do not know the relative importana of each parameter. It is outside the scope of this note to list the major problems and try to assess the important sin& k&Y pammeters in each. In time this will be done and then it will be posstble to define a “best buy” for a particular problem or for general use. Until then, the workers in this field must use a variety of methods. de6ning the conditions of the test exactly, study the problems and compare their model results with full scale measurements. In conclusion let me add a personal note The production of the profile in the wind tunnel is only the start; I am convinced that uptodate methods of data acquisition and processing are vital if we we
76
Research Note
are to uaderstand fully tbc probIems we study. This goes for fu&scale experiments as we& A great deal is to be kamed by measand studying the vazying quantities, particularly at &e very short time scale-s; our mcawremmt techniques must be better than our representations in t&e wind tuneI. ~~t~~t
of Aermmticnt
E@neering,
T. v.
hW$ON
University of Bistol, J3riszol8, England. REFERENCES m
J. (1967) Tbc simulation of the atmospheric boundary layer in a wind tu.mxeL~t~s~~ric Envirmme~r, 5 49-71. Be W. D. (1963) Effect of velocity distribution on wind loads and flow patterns on buildings. Paper 6 of Int. Conferenceon WindEfkcts @IIus ~Struct~es t&da the NPL. H.M.S.O., London. BROW~JD. I17returbuient jet in a linearly sheared cross $5~. Ph.D. Theis, University of Bristol. In preparation. m D. J. and Luz B. E. (1%6) Production of dsar profiles in a wind tunnel by cylindrical rods placed normal to the stream. $1. R. Aeronaut, Sot. 65,724-72X Cow A. R. (1939) The effect of a gauze on the velocity distribution in a uniform duct. R. & M. 1867. C~WDREYC. F. (1967) A simple method for the design of wind tunnel velocity profile grids. NPL Acre Note 1095. ELXXRJ. W. (X959) Steady flow tbrougb non-~0~ gauzes of arbitrary shape. J. &id Meek. 5,355”368. JENSENM. and F~CIC N. (1963) Model Iaw and experimental technique for detetmination of wind loads on buildings. Paper IS, Symposium on Whd Eg~ts OR 3~s and Structures heid at NPL. H.M.S.O., London. KLEB~OFF P. S. and D=a 2. W. {1952) Some features of arti&cially tbiekened fully developed turbulent boundary layers with zero pressure gradient. N.A.C.A. Rept. 1110. Lnz#x J. L. and -R J. T. (1-j The genaratiou of symmetrica duct velocity @‘o&b of b&b uniform shear. J. FIuid me&. Zo, 201-298. Lxvsss~ J. L. and TURNERJ. T. (1964b) The e&et of velocity profiie decay on shear flow in dawrs. Int. J. mech. Sci. 6, 371-379. Leon, A. R. J. M, (1%7) The generation of shear flow in a wind tunnel. f2. Jl. R. met. Sot. 93, 79-96. OWEN P. R. and Znxrnvscz H. R. (1957) Tbe production of u&form shear flow iu a wind tunnet. J. Fluid Me&. 2,521-531. SEXTON D. E. Design and performance of a wind tunnel for building research. &!g. Res. CWT. Papers, Research Series 18. STARRM. R. (1966) The G~ruGter~ftc~ of diem #ws past circukw c_&%derrs. Ph.D. IbeSis, Department of Civil Engineer&, Univezsity of B&tol. STROMG. H. and KAPLIN E. J. (1960) Convective turbtience wind tunnel project. Arg. NGL Lab. ANL 6199. STROMG. H. (1962) scale model wind ttmne1 studies in atmospberie di@usion phenomena. Praceedings of 7th ~~ra~li~ Cunfwence. W~IBREZADR. E. (1967) Report of a visit to Canada and U.&A. NPL Aero. Note 1051. WEGHARDT K. E. G. (1953) On the resistance of screens. Aeronaut. @wt. 4, 186-192q Wmarn~z F, A, L. and RAMMY W. J. (1957) E$fccts of inlet on co&al diffusers. Jl. R. ueronuut. Sue. 61, 116-124.
