Calibration of the Merrill-G.A.L.C.I.T. Wind-Tunnel, and a Suggestion for a Variable Cross-Section on a Small High-Speed Wind-Tunnel Citation Schwarzenbach, Jean Christophe (1942) Calibration of the Merrill-G.A.L.C.I.T. Wind-Tunnel, and a Suggestion for a Variable Cross-Section on a Small High-Speed Wind-Tunnel. Engineer's thesis, California Institute of Technology. doi:10.7907/PGWH-E460. https://resolver.caltech.edu/CaltechETD:etd-11252008-083306 Abstract Speed, power and flow inclination calibration tests run in the Merrill-GALCIT wind-tunnel are described and results presented. A description of the new balance-system is included. Turbulence measurements with hot-wires were made in the tunnel at various speeds and with various screen configurations. Tests run in a small scale model of the projected cooperative wind-tunnel at subsonic and sonic velocities are briefly described. A suggestion for introducing variable working- section walls in the new scale model undergoing construction is discussed and presented. Item Type: Thesis (Engineer's thesis) Subject Keywords: (Aeronautical Engineering) Degree Grantor: California Institute of Technology Division: Engineering and Applied Science Major Option: Aeronautics Thesis Availability: Public (worldwide access) Research Advisor(s): Merrill, Albert Adams Group: GALCIT Thesis Committee: Merrill, Albert Adams (chair) Millikan, Clark Blanchard Liepmann, Hans Wolfgang Shevell, Richard Corrsin, Stanley Defense Date: 1 January 1942 Record Number: CaltechETD:etd-11252008-083306 Persistent URL: https://resolver.caltech.edu/CaltechETD:etd-11252008-083306 DOI: 10.7907/PGWH-E460 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 4663 Collection: CaltechTHESIS Deposited By: Imported from ETD-db Deposited On: 16 Dec 2008 Last Modified: 29 Jul 2025 22:52 Thesis Files Preview PDF (Schwarzenbach_jc_1942.pdf) - Final Version See Usage Policy. 2MB Repository Staff Only: item control page

CALIBRATION OF THE MERRILL-G.A.L.C.2I.T, WIND TUNNEL AND A SUGGESTION FOR A VARIABLE CROSS-SECTION ON A SMALL HIGH-SPEED WIND-TUNNEL Thesis by Jean Christophe Schwarzenbach In Partial Fulfillment of the Requirements for the Degree of Aeronautical Engineer California Institute of Technology Pasadena, California 1942 ACKNOWLEDGEMENTS The author wishes to extend his gratitude to Mr, Albert Merrill for his cooperation and assistance in the calibration work, and to express his admiration for Mr. Merrill's ceaseless work and greet resourcefullness, He also wants to thank Mr. Stanley Corrsin and Dr. Hans Liepmann, who spent considerable time and trouble in guiding the hot-wire measurements, as well as Dr, Fejer for his many helpful suggestions, The author further wishes to thank Mr, Richard Shevell for making possible the extremely interesting collaboration on the latter's project, Finally the author is deeply indebted to Dr. C.B, Milli- kan for hig: untiring assistance and advice, SUMMARY Speed, power and flow inclination calibration tests run in the Merrill-GALCIT wind-tunnel are described and results presented, A description of the new balance-system is included, Turbulence measurements with hot-wires were made in the tunnel at various speeds and with various screen configurations, Tests run in a small scale model of the projected cooperative wind-tunnel at subsonic and sonic velocities are briefly described, A suggestion for introducing variable working- section walls in the new scale model undergoing construction is discussed and presented, CONTENTS A. CALIBRATION OF THE MERRILL-GALCIT TUNNEL. I. II. III. IV. Ve vI. VII. General Description of the tunnel The balance system The velocity calibration 1. Full speed section survey 2. Axial calibration %, Speed variation 4. Insertion of screens The power calibration Determination of the flow inclination Turbulence measurements with hot-wires B. HIGH SPEED COOPERATIVE TUNNEL MODEL I. Description of tests II. Summary of tests III. The variable working-section wall IV. References. page 10 13 15 15 16 17 20 24 31 32 33 35 39 A, CALIBRATION OF THE MERRILL--G.A.L.C.I.T. WIND TUNNEL, Ie GENERAL. The purpose of the research forming the basis of this thesis: was tio investigate the new wind-tunnel built under the supervision of Mr. Albert Merrilland the G.A.L.C.I.T. at the Pasadena Junior College,and to present a complete calib- ration and description of the tunnel. At the time of the assignment the tunnel had just. been completed, but the power supply was not yet installed. Subsequently the tunnel cross-section was altered to accomodate models used in the investigation of the Tacoma bridge, thus: causing a considerable delay in the tunnel's final completion. Due to the expected increase in wind velocity it was felt that the balances used with the old,open jet,open return type tunnel would no longer be adequate and Mr. Merril under- took to design a new balance system ,in which work the author attempted to give some assistance. A description of the new balances is included. All the tests described in this paper were run with the glass: sidewalls of the working section in place, since higher wind velocities: as well as a more: undisturbed flow of air through the working section are obtainable by this configuration, ag compared to tests run with these sidewalls removed. During the month of May the author was requested to give some: assistance to Mr. Richard Shevell in his research re- garding the new cooperative wind-tunnel, and @ description and summary Of this work is included at the end of this manuscript, II, DESCRIPTION OF THE TUNNEL, The Merrill-GALCIT tunnel is located in the steam- house of the Pasadena Junior College, in Pasadena, California, It has a closed working section and a closed return, and is capable of developing wind~velocities up to 70 miles per hour, A section through the tunnel appears in figure l, The working section is rectangular, two feet by four feet, and the balance system is placed directly beneath it, Down- wind from the working section a diffusor leads to a screen, the purpose of which is to prevent loose objects from striking the airscrew, This latter is of four-bladed, wooden construction and is fitted with a small spinner, It is driven through a belt from an induction motor placed immediately below it, the belt system being encased in a manner to prevent pressure from escaping from the tunnel, The motor was built by General Electric, andhas the following specifications: 1150 RPM at full load 60 cycle, 3 phase 50 H.P, Following the propeller the crossection of the tunnel js further expanded as it follows around the four corners, These are fitted with bent sheet-metal vanes to guide the air- flow around them, and to prevent losses in the corners, The maximum cwossectional area of the tunnel is found immediately preceding the working section, and measures 48 square feet. The minimum area is 8 square feet and there fore: the contraction. ratio of the tunnel is 6:1. In the lower west corner,at the maximum cerosssaction~ al ares,two small pressure-leads on each side of the tunnel record the momentary static pressure,and,as:is shown later thereby give a means of determining the dynamic pressure in the working section. These: leads are attached to an alcohol manometer,with a slanting glass tube,ca ,ibrated in centimeters. The fluid level in the tube is a measure of the pressure and has been denoted throughout. these tests ams the length 2. Thus. the pressure at the lead-orifices ,denoted by Py in inches of water is found as follows: let x = specific gravity of the alcohol @ = length in ¢entimeters of the alcohol colum Q= angle of the glass tube, then V4 at R= no - “ 2.54 & CX was measured with a vernier-protractor and was found to be: 8° 10' By & spring loaded rotary valve the static pressure manomder may be cut off from the pressure in the tunnel,so that it is not necessary to read this manometer at any particular mo- ment, since by a simple pull of a string the valve is suddenly closed,and the immobilzed length of fluid 4 indicates the pressure which existed at that’ moment. This device makes the operation of the tunnel by a single person possibles. III, THE BALANCE SYSTEM. Since the: author spent a considerable amount of time on preparing drawings and discussing details of the pro- jected. new balance system with Mr. Merril ,it seems logical to include: a description of the systeme. The old balances were built into a steel tubular fra,ework on wheels,and were: primarily capable of reading three. components, lift ,drag and pitching moment.e Yawrmoment. could be measured by modifying the pitching moment balance» All the balances are beam balances,rough adjustments made by mowing weights from one calibrated noth to another. Then by turning a calibrated screw: the final adjustment could be made and the force: read to within .001 lbs. This accuracy was not equaled in the application or in the transmitting of the forces, and the last figufe of these readings was: not signifigant. All the balances are damped in dashpots containing oil and equilibrium with no: forces acting was achieved by adding or removing lead washers from receptacles attached to the: balances. Disadvantages of this old system were: 1.) the framework was too small to house conveniently all the equipment. 2.) there was no possibility of measuring a side-force; 3) the model-support nec, was too small,particularly for the high speeds arrived at in the new: tunnel,allowing excessive model vibration, A host of smaller difficulties have also been noticed and the attempt has been made to eliminate them as far as possible: 10 from the new system. This new system was therefore built in the form of a steel -tube cube,of larger dimensions than the old frame. (Fig. 2.) Inside of this tube framework is hung a smaller cage, also cubic in shape. It is supended at the four corners by thin steel rods machbned down at both ends to allow unrestrained motion of the inner cage in all horizontal directions. This arrangement allows all horizontal force components and moments to be measured. The inside of the frame is very spacious and leaves ample room for: the installation of all the balZances,and the sensitivity of the inner cage to applied forces is very high,due to the length and fineness of the supporting rods. The balances have been designed with a view to pes- sible future changes. The first project is one in which side~ force, drag and yawemoment alone were to be recorded,and conse- quently the model support has been built into the inner cage such that the model can be rotated through 360°, and all three afore- “mentioned components measured. This is done with 2 side-force balances and a drag balance.The: side-force balances are placed on each side of the drag-balance and give between them both side-foree and yawing moment. It is to be remembered that since the drag~-balance is off center a correction to the yaw-moment must be included from this SOUrCee. | The new: balances differ from the old ones mainly in their type of suspension, since as is shown in figure 3, short ,steel flexing members are used,machined to a very small diameter at their focal point. These members are in compression,which is possible since they are flexing members at one point only,and all the motion is about this one point. This lends rigidity to the whole system, increases the accuracy and saves space. At present,for lack of time, the two side-force balances have been installed with roller-~-bearings instead of flexing members. As long as these bearings are new: the frictional losses will presumably be negligible,and it is pro- posed to replace them eventually by the fixtures described above. Another innovation is the installation of small chains, hanging from the ends of the beam balances. The end of such a chain passes over a pulley fastened to the balance-supporting structure, and. the length of such a chain end is a measure for the amount of chain weight applied to the balance. Thus the final adjustments are made without touching the balances,by merely pulling the chains. This is a decided advantage when unskilled operators are at the balances. | Lift and pitch-moment balances will be designed and installed as soon as needed, and if required a rolling-moment balance may also be added,thus enabling all six components to be measured simultaneously. IV. THE VELOCITY CALIBRATION. The primary objective of the calibration of this tunnel was to determine the velocity profile in the working section, so that the mean dynamic pressure over any model could be determined. Furthermore a means must be provided 60 establish the dynamic pressure acting on a model without having to insert a@ pitot-statice tube into the wind-stream, since this would frequently interfere with the tests, In the Merril- GALCIT tunnel the static pressure at. the maximum crosssection of the tunnel located immediately before 12 the working-section is taken as a measure of the vel- ocity at the model. Denoting the dynamic pressure in the working-section by q and the static pressure at the max- imum cross-section by h, then the desired redation is the ratio q/h as a function of position and engine spped. This ratio has been determined experimentally, by inserting a calibrated pitot-static tube of the Prandtl type furnished be the GALCIT into the working section, and measuring the static pressure simultaneously as described in section II. The pitot tube was movable in a fore and aft direction along the center-line of the tunnel, and could be placed anywhere in a section located such that the static orifice of the tube was 17.25 inches from the beginning of the working-~ section. As will be seen in the subsequent pages, this ratio q/h at full speed was found to be approximately 1.7 The assumption is usually made that the wind-stream in the working-section is at the normal atmospheric pressure. Cal- culating the value q/h that should result from this assumption it is seen: z Bernoulli : p+ Gv =pteu Continuity : AW= Arve 2. then: h=p-PR= fu], A. | where : pis the pressure vis the velocity fis the density A is the cross-sectional area, and the subscript one denotes conditions at the maximum cross- section and two the conditions in the working-sectione 12 a, Since the ratio A,/ A, is the contraction, and-is equal to 1/6 , we find that a/fh =1.03 (theoretical.) This value shows a large disagreement with experiment. Mr. Merrill has calibrated the tunnel with the side-walls removed from the working-section and found values of q/h which agree closely with the theoretical value. Whereas he found approx~ imately the same dynamic pressure as found in these tests, the static pressure h was roughly one third higher than that arrived at by the authore With the side-walls removed the condition of atmospheric pressure in the working-section is certainly ful- filled. Consequently, since conditions through the eontraction must be similar in both cases, we must assume that the static pressure in the working-section with the side-walls in place must be below atmospheric. As discussed bn the chapter on tur- bulence , the working-section is vented to the atmosphere at its end. However this vent is not very large, and the only explanation that can be given for the low pressure is that the flow resistanc@:through this vent is too high to allow the in- flow to compensaté for the losses of air escaping through the leaks in the circuit, unless a large pressure difference exists between the: working-section and the atmosphere. A rough idea of the magnitude of this difference may be arrived at by as-~ suming: that thesame pressure-drop through the contraction occurs in the case of the tunnel with side-walls in or out. Then with q#12 lbs/ft® and g/h=1.7 we find h&7 lbs/ft . The pressure dif- ference between working-section and atmosphere is one third of this and therefore about 2.3 lbs/ft’. The values of afh were found as follows: 13 If: \- spec. gravity of alcohol in pitot-static manometer Pe static pressure in inches of water, Then: Gedy where he is the pressure difference between h Py maxe cross-section and atmosphere instead of working section,as used in the theory. The actual velocity calibration consisted of three parts: 1) qg/h calibration throughout the cross-section approximately in the model plane. 2) & survey lengthwise through the working-section at the centerline. 3) a determination of q/h at the various engine settings, 1.) Velocity distribution in model plane. Simultaneous measurements of dynamic pressure and static pressure were taken at 3 inch intervals in a horizontal plane and 4 4nch intervals in a vertical plane,excepting that near the walls in the: vertical direction another series of points were measured at two inches from the wall,to discover any possible drop in speed along the walls. The results of this survey have been plotted both in the: form of profiles showing the deviation from the average value of g/h along the horizontal lines in percent, (Fig. 4) and in the form of a contour diagram ,giving lines of constant q/h ,or for a given static pressure at any instant, giving lines of constant dynamic pressure. It is apparent that the largest deviation from the average value q/h®1.703 is approximately 6 % . However 1t will be noted that the profiles in the region of the model position srecmuch nearer the mean value. Furthermore the value 1.707 is the average of the whole cross-section. Thus if the average values of q/h tabulated below are used, this will cause the average de~ 14 viation to be within 2.9% of the value used on a span of 36 inches, for instance, and correspondingly less on smaller spans, Since upon repeat measurements at the same position during one test a maximum q/h deviation of 1,4% was found, it appears that the use of the tabulated values will be accurate within these limits, The contour-line diagram shows that the fluctuations in dynamic pressure are not local in character, but extend over large areas of the working section, It is probable, therefore, that by reforming the guiding vanes in the corners, which at present have been arbitrarily adjusted by hand, the magnitude of these areas could be reduced, However, as these results show, their effect is not excessive and other factors in the tunnel such as power fluctuations, balance readings etc. introduce errors of at least the same order. of magnitude, so that the correction was not deemed urgent. The values tabulated below are average values over an area extending from 4 inches above to 4 inches below, and a half-span north and south of the respective center-line, They are valid for flow cnnditions at approximately full speed, Model Span in inches | @) | 6 | 12 | 18 | 24 | 30 | 36 | 42 q/h li.6aa ju.712 |1.713 | 1.71711, 72211. 725!1. 7281 1.724 1S 2.) Axial Calibration The dynamic and static pressures were further determined, on the center line of the tunnel, extending from {!' te 28" from the beginning of the working section, The same equipment was used as for the section calibration, The results are plotted in figure 6. The accuracy of the test points is not high, although they are within the limits thet heve been experienced in other tests. This test was run on two consecutive days, since the author was unable to find sufficient time in which to run the tests in one stretch, due to the very marked power-drop effect if the tunnel was run over a prolonged period at full speed. (See flow-inclination) test.) Nevertheless the value of q/h = 1.675 at section 17.25 compares reasonably well with 1.688 found in the section calibration and somewhat less well with 1.718 for the speed calibration described below. However this only shows a variation from the mean at that particular point of 1.3%, Over the approximate model location q/h is within 1.5% of the axial gq/h, and is thus a@lso within tunnel accuracy, although an unsignificant increase from front to rear can be observed, probably due to the proximity of the contracting nozzle. Thus the q/h determined at section 17,25 can be considered valid at all stations in that region. 3.) Calibration For Different Speeds Having determined the full speed characteristics, it was now important to find values of q/h at lower speeds, and check the theoritical hypothesis that it is constant with the speed, The motor as installed at present has only ten different 1o speeds, spotted somewhat at random within the speed range. Two runs from full speed through all settings were made, and the average values of both runs used as the basis of the resulting curves and values, The pitot-static tube was again placed on the tunnel center-line, and the result can be compared to those found under 1.) and 2.) above. The agreement is reasonable. Thus the primary objective of this test was to furnish the data for figure 7 in which d/h is plotted as a function of the static pressure, It is immediately apparent that whereas this represents & linear function, it is not a constent. This is dissappointing, and must be accounted ror®%S1ower speeds, The procedure to find the average span q/h at a slower speed would be then as follows: Find (q/h)q, for the h in question. Multiply the (4/h)s5en average found on the tabulation pagei¢by the ratio: a gince 1,688 is the average q/h at the center line at full speed, found from the section calibration. It is in this manner that the values of dynamic pressure and velocity were found for the power calibrations, 4,) Since the wind-stream of the tunnel appears to be quite rough as indicated by model vibrations and the fluctuation of the fluid level in all manometers, a fine screen was inserted in the wind-stream, to reduce the unevenness, The screen was placed right at the beginning of the working section, It nonsisted of a fine precision wire mesh, 30 wires per inch, The pitot-static tube was placed as usual at 17 the center-line at station 17.25. Values of g/h versus h have been plotted as they were for the case with no screen, (Figure 7, ) Also values of dynamic pressure versus h, together, with the values found in the absence of the screen have been plotted, (Figure 8.) The loss in dynamic pressure due to the screen is high, amounting to about 2/3 of the screenless case, However the readings were extremely consistent, indicating a very smooth flow, No fluctuations were observed in either of the manometers, and wind velocities of 45.5 feet per second were obtained, despite only proceeding as far as engine step 5, since the screen was not very rugged, Therefore it would seem desirable to run tests with the screen in,when this speed is adequate for the purpose, since the ease with which scale and manometer readings are made will greatly increase the accuracy of all results, V. THE POWER CALIBRATION It was desired to find the power actually delivered to the tunnel, to determine the dependence of dynamic pressure on the revolutions per minute and to find the RPM for the various engine settings. It was not known whether the engine was actually de- livering the rated 30 horsepower or not, and the possibility of a speed increase from this source was considered, With the help of Mr. Holtzklaw of the Electrical Engineering Department, runs were made determining voltage, amperes and kilowatt input, into the primary and secondary windings of the induction motor, by using an analyser, Simultaneously RPM and To) static pressure readings were teken for each engine setting. An efficiency was assumed and the engine output determined, Ap- parently there is no procedure other than this to determine the output of the induction motor, unless a dynamometer is used, a procudure too complicated to warrant consideration, The power was computed as ¢ HP =17 (Kp-K,)E 75 where Kpy= primary kilowatts Kg= secondary kilowatts KE = voltage Lf i efficiency, The power of step 10, or maximum output, was found to be 31,2 HP, This was for I= 75 Amps, E = 331 Volts RPM = 1156 Since the engine is rated at SOHP for B = 220, I= 72, RPM = 1150, this result is reasonable, To find the dynamic pressure at the center line and the average dynamic pressure for the whole tunnel the method outlined in section IV, 3, was used, The average value of q/h for the whole tunnel was found to be 1.7035. Accordingly for full speed condition the dynamic pressure in this test was found to be: q=12.7 lbs/ft” The kinetic energy thus passing through the working wection per second is found: K.E. | _m_ lve sec, ~ sec,2 where m = 9 volume 2@Av sec, sec, 19 accordingly a =qAv ftelbs-. GAY — 19,1 HP Thus the energy ratio of the tunnel at full speed is: 19.1 _ This extremely low value of the energy ratio is difficult to ex- plain. By including the belt-efficiency it might be raised some 20 % at best,and the ratio is still very poor. The many air~leaks through aceess dowrs are certainly sources of energy losses,as is the heavy screen inserted in the diffusor. It is made of heavy flattened wire and certainly has a high resistance. Finally a large part of the loss must be attributed to the propeller, which due to lack of time was not specifically designed for the tunnel, but merely taken over from the old installation.This energy loss is probably also directly connectéd with the separation phenomena described in the section on turbulence. The results obtained at different engine speeds have been plotted off figure 9, where the velocity and q are measured at the tunnel center-line. By comparing the test-points, corresponding to different motor: settings giving q,of this figure to those of figure 7, it will be noticed that. a certain engine setting does not correspond to a definite dynamic pressure.e Thus it is impos- sible to run the tunnel at an exact, predetermined speed. Theo- _retically this would not be too serious, since force coefficients and other dimensionless values are supposedly independent of the velocity. Needless to say this is however not strictly true, as may be seen from the fact that the value of q/h decreases with decreasing wind-speed. Therefore accurate test results can only be arrived at if constant dynamic pressure can be maintained. 20 To correct this limitation, the installation of a sensitive speed control is necessary. This is being considered at the present time, Such a device would also eliminate to some extent the serious difficulty encountered during many of the tests described herein, namely the marked fluctuation of power, From these tests and also others it was apparent that after the motor had been run for an extended period at full speed, the RPM dropped markedly, and if it was allowed to rest and cool it would recover its former velocity. This is probably due to overheating in the engine windings, since furhhermore the motor is installed on e badly ventilated besement floor, and closely hemmed in by the tunnel walls, Therefore if prolonged tests at full speed are desired, an improved method of ventilation should be devised, possibly in the form of a fan, The method of finding the dynamic pressures and velocities corresponding to these tests ns been discussed before, VI. DETERMINATION OF THE FLOW INCLINATION: It 1s important to determine the flow-inclination with respect to the horizontal through the working section, The method chosen consisted of determining the polar of a model wing, first in a normal position and subsequently inverted, Thus, if a flow inclination existed, the components Cy and CL at any particular given true angle of attack would have a different ratio if the wing is normal or inverted, The angle between the resultant forces is twice the flow inclination angle, The wing used had a 24" span and a wing area of .613 ft.” It was mounted in a heavy block, and could be rotated 2i through 360° and inverted. The mounting block was adjustable and calibrated, so that the angle of attack could be read from any arbitrary horizontal setting, The whole model was mounted on an arm of approximately $" diameter. The test was run twice, the results of the first test appearing to be unsatisfactory, since power fluctuations amounting to 17% velocity difference were observed, These fluctuations were of a slow nature, but prohibited accurate determination of both lift and drag since, due to the excessive model vibration explained below, equilibrium could only be established by observing the light-flashes of the balance over a period of many seconds, attempting to achieve an equal flash distribution, Since this could not be done on both lift and drag balance simultaneously. it was essential to have a constant dynamic pressure over the period of balance-adjusting. On the date of this first test, April 6, 1942, power fluctuations were dbbserved throughout all Pasadena, and therefore the author decided to repeat the experiment. Unfortunetely on the second test completely constant dynamic pressure was still not achieved. However points were rechecked in instances where an excessive difference appeared, and the results are believed to be as accurate as the equipment permits, A glence at these results will show that hhis accuracy is very low, since despite all precautions, it would be senseless to attempt to establish any value for the flow inclination other ‘than that it is within 20' of arc, The reasons for the in- sensitivity of this test sre briefly. 22 1.) Assymetry and high tare drag of the model support system. 2.) Weakness of the model support. 3.) Speed fluctuations of the induction motor giving varying dynamic pressure. The assymetry of the support system can be the only explanation for the fact that the polars of the two wing systems differ as they do. A free wing whose force components are mea- sured by two different coordinate systems must determine two polars. Each point of the one polar must correspond to a point on the other,both corresponding to &@ certain absolute angle of attack.Then the angle between the two corresponding force com- ponents is the angle between the coordinate systems,and must be constant for all points of the polar. It is immediately apparent from Figure 10 that,whereas the experimental points of the differ- ent wing orientations fit quite readily into a smooth polar, these polars cross each other at two places other than at zero lift. This is a eontradiction of the theory indicated above,and points to a difference in aerodynamic conditions between normal and inverted attitudes. The effect of the high tare drag, amounting to 96.5% of the total drag at minimum drag position is obviously to destroy the significance of the megsurements. Thus the low drag position of the polar is very approximate in character. The socket for supporting the model was mounted on 4 rod of approximately #" diameter,and this rod was too light for the forces experienced in the present tunnel. Thus excessive model fluctuation was possible,amounting to amplitudes at the wing 23 tips of 2 inches or more, This often took the form of a constant lutter, and made a reading of the balances an exceedingly difficult and tedious task. In the new balance system this support will be more rigid and this difficulty will be largely eliminated. Finally during the course of this test the power-input was found to drop off as the motor became overheated, This drop was peculiar in character and amounted to a definite drop in dynamic pressure of approximately 8.5%. If the motor was allowed to cool the former higher pressure niveau could be attained, The test was run in this manner, but this requirement introduced the necessity of making the balance adjustments rapidly to prevent overheating, giving rise to the errors discussed in the description of test number one, | In consideration of the foregoing it would seem unwise to draw any further conclusion from these tests than that there is no flow inclination which would affect the results of measurements. obtained with the balances used herein and the power supply in the described condition, The flow inclination is less than 20! of arc, appearing to be as an average 15' above the horizontal and downstream directions, Computations: Measured L, D, £ L = CG, 5 q Dd =p 5S gq ~ 1204 where: L=Lift measured in lbs. on scales 24 D = Drag ad = tare-drag = .03054 q, from previous model calibrations 2 tn i wing Area = .6135 ft. gq = dynamic pressure in los. /ft.* 1 = static pressure in centimeters 2 1 Ly. 14205 = hi .14205 = sin, 8° 1o! Qs qi 62.4 12 f= specific gravity of fla@id = .814, (measured) gq" for 24" span = 1,722 (see page I+) nh" VII, TURBULENCE MEASUREMENTS The fact that there are no honey combs or screens inserted into the wind before the contraétion, and the fact that the tunnel walls ere in many instances quite rough, led to the questions whether the turbulence level of the tunnel was not quite high, Furthermore models and manometer readings oscillated ex- cessively, and it was thought that considerable light could be shed on the cause of these phenomena by turbulence measurements, The method that was adopted is the usual hot-wire re- sistance measurements, Only a brief description is included since the method has been extensively described in other publications, (ref,No.3), It consists of placing a very fine wire, (,00025 inches in diameter in this case) and very short, (2mm.) at right anlges to the velocity which is to be determined, The resistance of the wire is a function of its temperature, and this, in turn, @s a function of the wind-velocity component at right-angles to the 25 wire-axis, Therefore by measuring the change in resistance in the wire, we can calculate the velocity fluctuation, A very slight current is maintained in the wire, and the resulting voltage fluctuations are amplified, compensated for time-lag and recorded, The turbulence has been defined as a percentage of the mean stream velocity U, and has been presented by measuring and tabulating the quantities: | © | 2 u 100 and Vv 100 U i where u' is the difference between the mean-stream velocity U and the instantaneous velocity at any point, and v is the velocity at right angles to the tunnel axis. Mr, Stanley Corrsin and Dr, Hans Liepmann, both of the aeronautics department of the GALCIT, made the measurements described in this chapter possible, They hed been making extensive hot-wire measurements at the GALCIT and furnished the equipment and supervised the tests run in the Merrill-GALCIT tunnel, In addition to the usual hot-wire equipment an oscilloscope was used to show the nature of certain flow irregulsrities, which had been noticed by violent fluctuations of the galvanometer in- dicator, The hot-wires were mounted on the center-line of the tunnel, approximasely in the position at which the velocity cali- brations were made, namely 17.25 inches aft of the beginning of the working-section, Runs were made at different speeds, with the glass ‘sidewalls in place and the tunnel clear, Subsequently measurements were taken with a 30-mesh precision copper- wire screen, the same as was used in the velocity calibration, 26 as well as with a cheese-cloth screen, The screens were inserted into the wind-stream at the entrance of the working~section, Their wooden one inch by one inch frames presented definite obstacles to the wind-stream Since they were not recessed into the tunnel wall. However due to the large distance from these disturbances to the hot-wires, the errors introduced from this source are con- sidered negligible. During the course of measuring the u! component it be- came immediately apparent that there was not only the normal turbulence of more or less constant amplitude with respect to time, which is present in most tunnel wind-streams, but that si&perimposed on this oscillation was another of much larger amplitude, and extremely aperiodic in character, These bursts of turbulence were observed in the oscilloscope and appeared to be similar in nature to the smaller scale turbulence, although having a many times larger amplitude, These bursts would last approximetely from + to + seconds and appear at intervals of less than five seconds, The cause of these bursts of severe turbulence is not as yet apparent, The investigators placed light threads on the walls of the contraction and of the working section and also inserted them on the ends of rods into the free stream, These threads behaved in marked accord with the oscilloscope observations, lying quietly in the wind stream for the majority of the time, and occasionally flapping quite violently and changing their direction, so that in such instances of turbulence they actually no longer were parallel to the walls or tunnel axis, as the flow characteristics would lead to expect, but were at an angle of perkaps some 10 or 15 degrees to their normal positions, Furthermore all threads under observation seemed to change direction simultaneously and in the same direction, and these shifts were accompanied by a burst of 27 turbulence observed in the oscilloscope. The v' measurements also showed a large-scale turbulence superimposed on the basic, but instead of appearing in isolated bursts, the intense turbulence lasted for more prolonged periods of time. Since the flow was seen to change direction,by observing the threads, a v-component of velocity would thus be introduced, and this would naturally be recorded by the hot-wire, This the excessively high v! components of velocity found in these measure- ments is a result of the flow-direction changes, which in turn appears to coincide with the u' bursts of turbulence, No more than some speculative suggestions can be offered as explanation, As has been described before, the static pressure at the maximum cross section is measured at all times, and certain oscillations spaced at similar time intervals as the turbulence bursts described above, may be observed at higher speeds, How- ever no correlation can be noted, since of course the manometer system has a considerable time-lag. It is improbable that the motor could give cause to pulsations of such short duration. No reason can be given to expect the propeller to cause such irregular propulsive forces, although it is doubtless also a source of turbulence, It might appear however that certain unstable separa- tion phenomena could give rise to these occurrences, The thread investigation showed a definite separation at the end of the working section, where an air-slot of one inch width is located, The diffusor level is approximately 1/8" below the working section floor, thus presenting in addition to the air slot a marked dis- 28 continuity and the air entering the tunnel at this point causes a large vortex which actually occasionally blew the threads upstream. That this separation could caryy all the way around the tunnel and cause the flow-irregularities des- cribed above is probably not likely. In any ¢éase however such a disturhance is detrimental to the diffusor flow and thus to the energy ratio, which, as we have seen, is extremely low. Due to the insufficiency of the present air- slot it would seem extremely desirable to design a new one, of larger dimensions. This slot should be built to allow the air to enter the tunnel tangentially to the diffusor wall, thus reducing the probability of causing separation at this point. Furthermore by carefully sealing all air leaks the necessity of a large vent could be reduced. Possibly by cutting some small holes in the end of the diffusor and allowing the air to be drawn into the tunnel at this point by the propeller, the static pressure in the working-section could be raised to atmospheric without necessitating the presence of a large slot. When this large separation vortex was noticed the slot was taped over,.in the hope of eliminating the dis- turbance and reducing the turbulence. However, although the separation point seeped to vanish, as indicated by the threads held into the wind-stream, no change was noteice- able in the turbulence characteristics. The most violent thread oscillations were observed in the contraction and since it is a short but high-ratio contraction, 29 one might suppose that this leads to separation at opposite walis at different moments, In this manner changes in flow direction would be caused, as well as high turbulence, This would be sub- stantisted by the fact that there is no honeycomb to smooth the flow in the tunnel, end the vanes inserted in the corners are rough and irregular, Thus improvement would presumably be obtained by making a gentler contraction or introducing a honey-comb after the last corner before the working section, There is no room for both contraction-change and honey-comb, since an effective comb should be followed by a straight portion of tunnel, to allow decay of the eddies introduced by it. The Merrill-GALCIT tunnel cannot be lengthened, since,for one reason,the building accomodating it will not permit it. it might therefore be best to first introduce a honey-comb in the tunnel as it now stands, and if this improve- ment is not satisfactory, to subsequently lengthen the contraction section, It is interesting to note thet the best flow conditions were found with the 30-mesh precision screen, Both turbulence coefficients were reduced approximately to half their previous values, and the violent bursts were largely eliminated, Thus for slower speed tests the use of the screen is strongly advised, By inserting a similar screen in the tunnel immediately beforé the contraction the turbulence might be reduced with much less resulting loss in dynamic pressure, The insertion of the cheese-cloth screens proved to in- crease the turbulence as compared to the wire-mesh, although an 390 improvement over the free tunnel condition Was still apperent, This increase of turbulence over the wire mesh may be explained by the fact that the cloth-mesh is probably extremely irregular in character, allowing large-scale velocity differences over the whole screen, and introducing irregularity from this source, Furthermore due to the highly flexible character of the cloth, small oscillating motions such as flapping and stretching will presumably occur constantly, giving rise to small pressure changes. These disturbing effects would not be present in the wire-mesh screen, The percentage turbulence found for the different con- ditions is listed below, That the resulting values are reasonably precise is indicated by the fact that the free-tunnel results at different speed check very well with one-another, { t2 2 CONDITION Uf.p.8. % 5 4%, ‘5 TT No screen 32.2 .o7 No screen 51.6 205 2.8 No screen 77.7 60 50-mesh precision 53.8 eo? 1.4 50-mesh precision 45,5 ool 02 1,9 1.7 Or cheese-cloth B, TESTS RUN IN A SMALL MODEL OF THE COOPERATIVE WIND-~TUNNEL AND A SUGGESTION FOR A VARIABLE WORKING-SECTION WALL, 31 32 I, DESCRIPTION OF TESTS During the month of May the author was called upon to give some assistance to Mr, Richard Shevell in his research problem concerning the new cooperative wind-tunnel, the design of which is being studied by the GALCIT at present, The tunnel is to be a high speed tunnel, with a working section of approximetely 8 feet by 12 feet, The effects of compressibility are of extreme importance since the velocities considered approach that of sound. Because little data is available on tests in this region it was decided to build a small model of the contraction, working-section and diffusor, and to endeavor to gather information on certain pro- blems, (Figure 11) These were briefly to find the effect of the high speeds on the energy ratio, on the diffusor efficiency, and to find a diffusor form which eliminated or reduced seperation at the high speeds, Furthermore there should of course be no pressure gredient in the working-section, Mr, Shevell's report on the subject is complete and accurate, and an attempt at duplication by the author of this paper would be useless, Therefore merely a brief outline is in- cluded, Eighteen small orifices spaced approximstely 2 inches apart along the walls of the plaster-cast tunnel model were the means of finding the local static pressures. Plugs in the entrance and working sections were provided to permit the intro- duction of pitot-static tubes and total-head manometers, and a glass window was provided at the working section to permit obser- vation of the proceedings, The power wes supplied from a large 33 pressure tank, and the model made small enough, so that sonic velocities could be attained by the volume of air pumped into the tank by the compressor, Thus the model working section was built 5,096 by 2.195 inches, II, SUMMARY OF TESTS A brief summary of the results obtained follows: 1.) A Mach number of 1.05 was obtained at the end of the working section, The Mach number M is defined as: Me ¥ v=wind velocity a a=velocity of sound For the airstream to return to the high pressure and corresponding low velocities of the exit within the expanding diffusor, it must create shock-waves with their attendant pressure jump. TheSe, the observers believe, were made visible by the oil droplets present in the wind, which flowed in small, slow streams along the tunnel walls, At high speeds these streams stopped abruptly in a transverse line behind a pitot-static tube, supposedly due to the pressure-gradient of the shock wave caused by the disturbance, Furthermore an arch believed to be condensed water- vapor was seen to encircle the pitot-static tube at Mach numbers in the vicinity of one, possibly resulting from pressure and density gradients in shock-waves, 2.) Separation was found to exist near the end of the diffusor. 3.) The curves of pressure against tunnel position found in these tests show a cnnstant pressure existing up to Mach numbers around ,82 and thence an increasingly steep and ot and negative pressure gradient through the working section, This is attributable to the increasing effect of the boundary layer, Since with increasing M larger changes in pressure are caused by small changes in crossectional area, as is seen from the following equations: ue, , xdf p = pressure 1-M” be f=crossectional area k=heat constant ,1.4 for air The curves found are extremely similar to those found by Dr, Ackeret of the E,T.H. in Zurich, Switzerland, (Ref. 1.) and thus give a good check to the measurements made with his : tunnel, Due to this increasing sensitivity to changes in crossectional area it follows that it is not possible to build a working section with fixed walls, such that there will be no pressure gradient through it, for all Mach numbers up to one, This difficulty is further aggravated if disturbances such as models or pitot-static tubes are inserted into the tunnel. There ~ fore if it is nevertheless wished to run significant tests at these velocities, there is no alternative but to adjust the tunnel crossection to the particular velocity, and, which is still more difficult, to the particular model, The tests run with the small model described before having shed such a wealth of information im such a simple arrangement, it has been decided to construct another, wooden: model, constructed more carefully and including various improvements over the old arrangement. Among these the problem of affording a 35 variable wall for the test section is receiving considerable attention, and the author has been asked to include a workable suggestion on this subject, TII. THE VARIABLE CROSSECTION, It has been explained above why good results could not be obtained without a variation in crossection, What then are the problems involved in affording this variable area? In the ease of the full scale tunnel these problems are obviously primarily the enormous size of this varying wall, and the fact that the tunnel must be accessible through the walls that are to be varied, Thus the wall would have to be split in parts, and furthermore be flexible to avoid discontinuities, It would have to be rigid enough to withstand the relatively large negative pressures inside the tunnel, amounting to about $ an atmosphere, unless the space outside of the wall were also held at this low pressure, a scheme involving a multitude of new design problems, The crossection being somewhat octagonal in shape, a discontinuity must arise at the joining of wall and corner fillet, unless the fillet is attached to the wall, in which case it would have to be flexible too. These and other obstacles could probably only be overcome at the price of tremendous cost increase and delay. An easier scheme might prove to be the introduction of preformed walls, fitting within the normal, fixed ones, in which case a different wall would possibly be needed for each speed considered as well as for each model, The author kiows too little about the problem to go into further detail, However he will attempt to outline a possibility for the model tunnel. 36 The problems here are virtually the same but in a much smaller scale, First the question arises as to what is to be achieved by the variable wall, The primary objective is to main- tain an even pressure throughout the section, This is relatively easily obtained for any Mach number less than one, for the empty tunnel, However when a model is introduced, there will again appear a gradient, and this could only be removed by practically continous position control along the wall, This is of course not practical, There would not be a great value in achieving a constant pressure for an empty tunnel only, but it is the authors opinion that any correction to the tunnel wall will certainly cause the critical Mech number, at: which a gradient appears, to be raised, particularly if the sdjustment can be made in the presence of the disturbing element, Whereas a cure for the problems is not contemplated, the variable wall will certainly provide good test conditions at higher Mach numbers, This improvement will be the better, and the value of the small scale results the greeter, if the sensitivity of the wall-control is high, Therefore fixed wall corrections, 48 heve also been suggested, would merely prove the varieble correction and not the tunnel wall. The results would furthermore not be conclusive, since differences in fit and in surface roughness might well cause variations com- parable to the corrections contemplated. On the other hand if a flexible wall is inserted, the smallest effects of position change can be. observed, Finally, by such a device the airflow may first be chosen, and subsequently the walls adjusted as desired, whereas if an insert must be made, the wind-speed must be adjusted to the 37 wall configuration, a cumbersome procedure in the case of the model in question, For these reasons the author suggests a tunnel with a thin flexible metal wall, adjustable by thumb-screws, To afford the possibility of not only adjusting the minimum crossection, but of also making some allowance for the presence of a distur- bance in the tunnel, two thumb-screws are suggested. This will also permit test-material to be gathered on the desirable shape of a corrected wall. Figure 12 shows the projected wall. Since many other tests are to be made with the tunnel, it is extremely desirable to be avle to remove this whole apparatus and replace it by either glass windows for Schlieren photographs, by plugs for pitot-static tubes and so forth, It is suggested that a standard fitting be constructed in whtch all these devices will be interchangeable, possibly among all four walls, The problem of the corner fillets, the author believes ean not be completely solved, There will of necessity be a dis- continuity when the wall is projected beyond the normal wall. In this suggestion small sheet~metal covers are extended to the point of maximum wall-deflection, to prevent an air gap at a sharp metal edge, connecting to a cavity, in the working section, Some such seal is almost certainly essential, The secondary . fillets on this small seal, (réfer to figure 12), can be removed, and tests can show whether their presence or the presence of a ‘free sheet edge is more detrimental to the flow, Due to the low pressure in the tunnel, it is desirable that the chamber behind the flexible wall be vented to the tunnel, 38 to prevent pres®ure loads from affecting the wall. However the author believes that there will be sufficient leakage around the top of the well to fulfill this condition, By calibrating the thumb-screws the exact wall position ean be known at any instant, In an article by Dr. Gasperi, (ref. 2) the possibility is indicated of making a workable supersonic tunnel, with only one of the four walls adjustable. Although it is realized that an inclination will be given to the airstream, nevertheless the desirability of leaving as many walls clear for other equipment as possible calls for the first experiments to be made with only one adjustable wall, Should this prove unsatisfactory it would be an easy matter to duplicate the arrangement on the opposite wall, if the fittings heve been made identically. To find the necessary wall travel the following rough calculations were made: Assume M=.9, which represents the region under considera- tion, Assume the changes in pressure and area are small compared to the original values: 2 = ane vp KM”, | OF F=crossectional area=6,45 in“ Dp Tom’ * OF k =1,4 for air up, the pressure drop through the working section was found to be 4,3 cm of mercury on the three high speed runs, p=o51 cm of mercury = average pressure in working section at the high-speed runs, VF =,053 in’ = VF. .0407" necessary wall travel, 1.501 Thus .1" wall travel will be ample. The author suggests using 39. .O1" sheet steel, for the flexible wall. Since in the model in which the experiments were made Mach numbers larger than one were achieved downstream from the working section, the device suggested in these pages will presumably afford the possibility of attaining supersonic velocities in the working section itself, Undoubtedly the corner-fillet distur- bances will be more noticeable if operation of the model tunnel is contemplated in this regime, but at the same time very inter- esting data could be collected on the effect of sonic velocities and shock-waves on the velocity-distribution, the boundary layer and the diffusor flow, In short a host of problems await the experimenter, and whether or not this device proves successful, the variable wall for a high speed tunnel will ultimately become essential, and every venture in this field will shed added light on the nature of the problem, REFERENCES J. Ackeret : Windkanaele fuer hohe Gesechwindigkeiten. (Convegno di scienze fisiche, Roma, 1936.) Ref, 2. M. Gasperi : (Convegno di scienze fisiche, Roma, 1936.) Ref. 3. : NACA Technical Report 320. yoo — 6 YdOLOW Madd NOIL93¢ V ONIMYOM v4 L. I SJINVIVS { ASWwa y | L Se Zo a tr \ | QNIM ‘~T .a4unois v=, J1v9¢ TINNAL ONIM LiIDtw5 — Wawa icam WIND, IND Jeo SUPPORT h = \ } |i wOwXKwvxXxK=——— << —— I~ pRAG BALANCE y ~~ INNER CAGE ——~—__] tae a | SUPPORTING © ROD 5 n 4" ~~ YAW MOMENT & SIDE FORCE BALANCES EAST rh OUTER CAGE -— — O, Uo _U_ JU fioor NEW BALANCE SYSTEM ( SCALE 5) FIG. 2 (scHEMATIC ) NOTE: LIFT € ROLLING MOMENT BALANCES NOT DRAWN , CORRESPONDING TO O€VELOPEMENT To THis DATE —"s § BALANCE ~ARM ROD TRANSMITTING pe- FORCE FLEXING MEMBER an TO BALANCE-ARM BASE FIXED ne 2 TO OUTER CAGE DRAG- BALANCE DETAIL (rut. scare) FIG. 3 rv h dae fbr pe see pee faye La te 4 NORTH SOUTH DYNAMIC PRESSURE DISTRIBUTION SECTION 17.25 NUMBERS REFER TO LINES OF CONSTANT VY, FIG. 5 2 +4 S & 1o a 14 | starr OF WORKING SECTION | } ‘ J. n ] 4 | nl a | a 1G 18 20 22 at 26 28 STATION IN INCHES AXIAL _ CALIBRATION FIG. G - eee i hg WJ abe Lee fopep dens ret 1 eit gees b t pee enema i H a 1h +. To Je bf = aed T Hoye iti ia im are 4 2 pl ated ean 4 bef + : : Q : i = £ ,1 > 7! auno - avs FANNNL-ONIM dAlLVaaddOO> 320 W|GOW NOLLDIS «=SNIMYOM (MOGNIM SsvID HAIM) SRION ¢SS299v WIOW iSWI~ UALSVId ANVL YOSSIVdWOD eh awn TJGOW ISNNNL-aNIM aAliveadood sIwes TIN TIVM SJIDGVISWA LN@WaAOW SMauDS «OONISArOW TIWM MOTIV OL Maus MalA NwId ONititd SI@WAOWIY se¢_[SA 4 + pt I | \ it I fl ip ' a2 J i | { | ouone DNIMYOM Vv 7 cesuosune OL 03.1~4aMsS) —i——— NoMloayid ss GNIA VIWM aTIVWIBVA ~ —_ ~ - -_ - “> F3BNNOAL W-W MBIA ViwM DIBVISWA STIVA SIGIXs1IS 4UOd SIw3aS Yiv cc 1 | 4 [ siztid waNYvo? G | NOI123S ONINNOM | n cy Co (J