]\gt ; Conditions the Boundary of Fluid in Turbulent Motion .
413 Again , the Fredbolm determinant is , and is consequently of class genre zero .
Lastly , it may be observed that The equation just written leads to interesting results , as to the relation between the singular values of and those of On the Conditions the of Fluid in Turbulent Motion .
By T. E. STANTON , F.R.S. , Miss DOROTHY MABSHALL , B.Sc. , and Mrs. C. N. BRYANT .
( Communicated from the National Physical Laboratory .
Received April 13 , 1920 .
) The experiments here described form part of a general research into the phenomena of skin friction of solid surfaces due to the flow over them of fluids whose motion , not in the immediate vicinity of the surface , is eddying or turbulent .
Considerable information has been obtained in recent years as to the magnitude of the frictional forces brought into existence in this condition of flow , and the manner of variation of these forces with the relative mean speed of surface and fluid , the roughness of the surface , and the physical characteristics of the fluid is fairly well known .
Practically nothing , however , is known about the mechanism by which the resistance to flow is transmitted to the bounding surfaces .
For speeds below the critical when the general motion of the fluid throughout is streamline in character , it is generally accepted that the layer of fluid in contact with the boundary is at rest lelative to it , as any slipping of finite amount would be detected in a variation from the Poiseuille law of the relationship between the diameter of a pipe and the time of efflux of a given volume of fluid .
At speeds above the critical , observations near the walls have shown that the mean velocity falls rapidly as the solid bounding surface is approached , and it has been suggested that at the walls there may exist a thin layer in VOL. XCVII.\mdash ; A. 2 Mr. T. E. Stanton and others .
On the which the flow is laminar in character , in which case , if there is no slipping , the frictional resistance would be determined from the slope of the velocity curve in the surface layer and the coefficient of viscosity of the fluid .
During the last few years several attempts have been made at the National Physical Laboratory to obtain evidence as to the truth of this assumption .
The method adopted has been to set up a condition of turbulent fluid motion over a surface of which the frictional resistance to the flow could be accurately determined , and to measure by means of a very fine Pitot tube the velocity of the fluid at a point as near the wall as possible .
The chief difficulties in the method are : ( 1 ) the interference in the flow in the neighbourhood of the boundary due to the presence of the tube itself ; and the fact that when such a tube was placed in a current of fluid in which the variation of velocity across the mouth of the tube was very great , it is by no means certain that the value of the speed deduced from the pressure in the Pitot tube would be the speed at the geometrical centre of the tube .
From the results of experiments made in on the distribution of the axial velocity of air flowing in pipes of cm .
and ( 4 cm .
diameter , it appeared that the thickness of the layer of fluid which was in a state of laminar motion at the walls , if it existed at all , was less than half a millimetre ; and when , by the appointment in 1916 of Miss C. N. Jones ( Mrs. Bryant ) as a research worker in fluid motion , an opportunity occurred of taking the matter up again , attempts were made to produce a Pitot tube sufficiently small to explore the region of these dimensions .
The smallest tube which could be used satisfacloriJy was one of which the external dimensions at the orifice were mm. and the internal diulensions mm. The first set of observations was directed to an estimation of the reliability of the instrument as an indicator of the square of the speed at the geometrical centre of the orifice of the Pitot tube , when this was placed close to the wall of the pipe .
For this purpose , the speed of the flow in the pipe ( cm .
diameter ) was kept below its critical value ( 580 cm .
per second ) , so that the distribution of velocity across any section could be calculated from the formula , where is the velocity at the axis of the pipe and is the radius of the pipe .
A satisfactory check between bserved speeds and theoretical speeds was obtained up to a distance of mm. from the walls .
It has been inferred by Morrow from observations on the distribution of velocity of water flowing through a pipe at speeds below the critical , that this theoretical 'Roy .
Soc. Proc , vol. 85 , p. 366 .
Conditions at the Boundary of Fluid in Turbulent Motion .
415 PITOT TUBE I SCALE TUSE SCALE Mr. T. E. Stanton and others .
On the distribution is not that which actually exists .
* This deduction , however appears to be due partly to a small inaccuracy in the calibration of the Pitot tube used and partly to the fact that in Morrow 's experiments the Pitot tube was situated at a section of the tube which was approximately 30 diameters from the inlet end of the tube .
It is almost certain that at this distance the radial distribution of velocity has not reached steady conditions .
No appreciable departure from the parabolic distribution of flow has been found in the National Physical Laboratory experiments at distances of 100 diameters from the inlet .
At distances closer to the walls than mm. the sensitiveness of the manometer employed was not sufficient to measure the differences between the Pitot and static pressures with sufficient accuracy for the purpose of making the desired comparison .
The difficulty of measuring the velocity at distances from the walls of less than mm. will be realised from the fact that , with a velocity at the axis of 1050 cm .
per second , the velocity at a distance of mm. from the wall is only 58 cm .
per second , corresponding with a pressure head on the manometer of mm. of water .
As the readings of the very sensitive manometer employed for the work could not be depended upon to an accuracy greater than mm. of water , it is obvious that no reliable comparison between the actual and theoretical speeds at this distance from the wall could be obtained .
In the second series of experiments , observations of the pressure differences between the Pitot and static pressure tubes , for a given distance of the geometrical centre of the former from the walls , were taken over as wide a range in the mean rate of flow through the pipe as possible .
At the same time , the corresponding opes of the static pressure gradient down the pipe were accurately determined .
From the latter measurements the values of the surface friction at the walls for different rates of flow were calculated .
The Pitot tube observations were then repeated for other distances of the tube from the walls ( Table I ) .
For each distance a curve of speed variation with surface friction was plotted , and from these curves it was possible to scale off , for any given value of the surface friction , the speeds corresponding to the different distances of the Pitot tube from the walls and so plot a curve showing the variation of speed with position of Pitot tube .
These curves are shown in fig. 3 .
It will be realised , of course , that these curves can only be taken as showing the radial distribution of velocity in the neighbourhood of the walls , under the assumption that the speed indicated by the Pitot tube is that which exists at the geometrical centre of the orifice , i.e. , that there is no interference with the flow near the walls due to the presence of the tube .
In ' Roy .
Soc. Proc , vol. 76 , p. 205 .
Conditions at the of a Fluid in Turbulent Motion .
417 Q ) Mr. T. E. Stanton and others .
On the Conditions the Boundary of a Fluid in Turbulent Motion .
419 the same figure are in dotted lines velocity slopes which would exist at the boundary if the flow were laminar .
These are caloulated from known frictional resistances and the value of the coefficient of viscosity for air .
Another interesting feature of the velocity distribution is the manner in which the speed at any given distance from the walls varies with the frictional resistance .
If the flow were laminar , the velocity at any point would , of course , be simply proportional to the surface friction .
For turbulent flow ib is known that the surface friction is proportional to the nth power of the mean rate of flow , where varies from to 2 , according to the diameter and roughness of the pipe .
This relation is usually obtained by plotting points of which the ordinates and abscissae are the logarithms of the surface friction and mean flow .
These points are found to be on a straight line of which the slope gives the value of .
Adopting the same method for velocities at a point , the curves shown on fig. 4 have been obtained .
These show that the relation between surface friction and speed at a point near the wall is of the same form as that .
T. E. Stanton and others .
On the between surface friction and mean flow , but that the value of iminishes .
the boundary is approached until , when the geometrical centre of the tube is mm. from the wall , the value of is .
It would appear , therefore , that at this distance the eddy motion has nearly disappeared .
As the experiments in this small pipe ( cm .
diameter ) were made at speeds not greatly exceeding the critical speed , i.e. , at values of V. between 2000 and 5000 ( where kinematic viscosity of air , mean rate of flow , and diameter of pipe ) , it was considered desirable to repeat the experiments at values of V. very much reater than this , for it is known that the distribution of the velocity across a pipe changes appreciably with the value of this function .
For this purpose a pipe of cm .
diameter was used .
Air was supplied to this by means of a high-speed Sturtevant fan which could maintain a constant rate of flow through the pipe of about 2900 cm .
per second , corre- sponding with a value of V. of 250,000 .
The same Pitot tube was used as in the small pipe , the method of attachment being shown in fig. 1 .
Experiments similar to those described above were made by Miss Marshall , who took up the work after the transfer of Miss C. N. Jones to another division of the laboratory .
The observations ( Table II ) were .
reduced in the Qame way as before , and the curves of variation of indicated speed with mean rate of flow and with position of Pitot tube are shown in fig. 5 .
On examining the curves in figs. 3 and 5 , and assuming the interference effects to be negligible , the following characteristics will be seen to be common to both : ( 1 ) The slopes of the tangents to the velocity curves at distances of mm. and abovs from the walls are very much smaller than would be the case if the flow were laminar at these distances from the wall .
For example , when the surface friction in the cm .
pipe is 10 dynes per square centimetre , the value of at mm. from the wall would be 57,200 if the flow were streamline in character .
The measured value of was 33,600 .
In the case of the ch pipe , the corresponding values are 211,200 for streamline motion and 90,000 actually observed .
( 2 ) The speeds calculated from the observations at distances of mm. and less from the walls are appreciably higher than would exist in laminar flow with the measured surface friction .
For example , in the cm .
pipe , with a frictional resistance of 10 dynes per square centimetre , the calculated speed at mm. from the walls was 795 cm .
per second .
In laminar flow with this value of the surface friction the speed would be 443 cm .
per second .
In the case of the 5-inch pipe at a value of the surface friction of dynes per square centimetre , and at a distance of mm. from the Conditions at the Boundary of a Fluid in Turbulent Motion .
421 Mr. T. E. Stanton and others .
On the wall , the corresponding values are 1230 for the calculated speed and 1050 for the speed in streamline flow .
To sum up the evidence afforded by the curves of figs. 3 and 5 , it would appear that if it may be assumed that the speed indicated by the Pitot tube is the speed of the fluid at its geometrical centre during the experiment , streamline motion if it exists must be confined to a region of less than mm. from the boundary , and further , the observations are not inconsistent with a finite amount of slip at the boundary .
It was vident , therefore , that the proof of the existence or otherwise of streamline motion at the mdary would involve a closer exploration of the region near the boundary than had hitherto been possible .
This , however , could not be done with the available appliances to the fact that any further reduction in the cross dimensions of the Pitot tube would have rendered it unworkable owing to the time fluctuations of pressure in the room .
After some consideration it was thought that if the wall of the Pitot tube adjacent to the wall of the pipe could be removed and its place taken by the wall of the pipe itself , it would be possible to obtain readings at distances of and mm. from the wall .
The arrangement of tube previously used was accordingly modified and the final form adopted is shown in fig. 2 .
This was set up in the 5-inch pipe and a series of observations ( Table III ) similar to those previously described were made and reduced in the same way as before .
The final results are shown in the curves of fig. 6 which give the variation of the velocity calculated from the new Pitot tube pressures with the surface friction and with the distance of the geometrical centre of the tube opening from the wall of the pipe .
From an examination of these curves it will be seen that by means of the new device the nature of the flow in the region of the boundary is revealed to a ve17 much greater extent than was possible with the original form of Pitot tube , and that the new evidence afforded by the velocity estimations up to within mm. from the wall strengthens considerably the probability of the existence at the boundary of alayer of fluid of finite thickness which is in streamline motion .
The existence of a slip at the boundary is not , however , decidedly negatived , as it will be seen from fig. 6 that the values of the speeds at min. from the wall are much higher than those corresponding with the velocity slope at the boundary calculated from the known viscosity of air and the measured friction .
In order to obtain more definite evidence on this point , the possibility of a a calibration of the Pitot tube in a current of air in which the distribution of velocity up to the boundary was known , was re-examined , and in view of the success which had attended the use of the new form of Pitot tube , it was decided to attempt the fitting of a tube of this Conditions at the Boundary of a Fluid in Turbulent Motion .
423 Mr. T. E. Stanton and others .
On the Conditions at the of a Fluid in Turbutent Motion .
425 form to a pipe of about cm .
diameter .
The ruction of this attachment , it is needless to say , required considerable skill and accuracy , and much credit is due to Mr. H. G. Pincott of the Mechanical Staff of the Engineering Department for the success with which it was carried out .
The diameter of the pipe used was cm .
, the Pitot tube ( here referred to as No. was mm. wide and the upper wall had a thickness of mm. In the experiments with this pipe the outlet end was connected by a rubber tube to a gas holder of which the rate of displacement could be varied by means of the tension in the supporting cord .
It was possible , therefore , in these tests to make a series of observations at a constant value of the surface friction and there was no necessity to reduce the actual observations by preliminary plotting to obtain these results .
The first series of tests were made with the Pitot tube at a given distance from the wall and over as large a range in the rate of flow as possible ( Table ) .
Ihese observations were then repeated for another distance of the tube from the wall .
The results are shown plotted in fig. 7 , in which the ordinates of the plotted points are the speeds indicated by the Pitot tube and the abscissae the rates of mean flow .
In the same figure are also plotted the values of the surface friction obtained from the fall of static pressure in the pipe ( Table V ) .
It will be seen that at rates of flow below the critical , when the flow over the whole pipe was streamline in character , the plotted poirts in the velocity ctlrves Table .
III Pitot in cm .
Pipe .
Mr. T. E. Stanton and .
On the onditio , the Boundary of Fluid in rbulent M. 427 Table .
III Pitot in cm .
Pipe .
Table .
III Pitob in cm .
Pipe .
lie on straight lines passing approximately through the origin .
It may be concluded , therefore , that these calculated speeds are the values of the speeds existing at a certain point near the boundary , this point may not be the geometrical centre of the Pitot tube orifice .
The desired calibration of the Pitot tube appeared , therefore , to be a simple matter , and was made as follows .
For a certain value of the mean rate of flow below the critical value , a series of readings of the Pitot tube pressures were made at different distances of the geometrical centre of the Pitot tube from the wall ( Table ) .
From the equation of the distribution of velocity in streamline flow , , the values of were calculated for each Mr. T. E. Stanton and others .
On the value of obtained from the observations .
Points were then plotted whose ordinates were the values of , and whose abscissa were the corresponding distances of the geometrical centre of the Pitot tube from the wall .
This process was repeated for different values of the mean flow , all below the critical value , and the mean curve through all the plotted points was taken to be the calibration curve for the Pitot tube .
This curve is shown in fig. 8 .
It will be seen that for of the Pitot tube of the order of mm. the assumption that the calculated speed is that which exists at the geometrical centre of the orifice is not greatly in error , but that when the opening is of the order of mm. , the interference with the flow is so openinQ considerable that the calculated speed is that which exists at the edge of the Pitot tube furthest from the wall .
A series of observations were then made at known values of the surface friction , and with the centre of the Pitot tube at different distances from the wall ( Table .
The speeds calculated from these observations are shown plotted in two methods in fig. 9 .
On the left-hand side of the figure , the abscissae of the points are the distances of the centre of the Pitot tube from the wall , and , on the right-hand side , the abscissae are the " " effective \ldquo ; of the Pitot tube from the wall , as scaled off from the calibration curve of fig. 8 .
Conditions at the Boumdary of a Fluid in Turbulent fotion .
429 VOL. XCVIL\mdash ; A. Mr. T. E. Stanton and .
On the Conditions at the of Fluid in Turbulent Motion .
4:31 It will be obvious , of course , that the application of this calibration curve to velocities indicated by the Pitot tube when in corresponding positions , but with the flow turbulent , is open to criticism , since this would involve the assumption that the nitude of the interference for any given distance from the wall is the same for turbulent motion as for laminar motion .
This criticism would certainly need very careful consideration if the degree of turbulence at these positions were of the same order as that at the centre of the pipe .
It is clear , however , from the curves of fig. 4 , that the turbulence is greatly diminished in intensity at distances from the wall as small as those under consideration , from the fact that in this region the surface friction is proportional to , where is the observed speed , and is little in excess of unity .
For this reason , it is here assumed that the calibration curve of fig. 8 can be applied in experiments with values of flow above the critical with reasonable accuracy .
On examination of the two sets of curves , it will be seen that the effect of the correction for interference in the slope of the curves at distances from Mr. T. E. Stanton and .
On the o r Conditions at the of a Fluid in Turbulent Motion .
433 434 Conditions at the Boundary of a in Turbulent the wall of the order of mm. is practically negligible , as would be expected , but that , at distances of the order of mm. , the slopes , when corrected for interference , are considerably increased , and approximate fairly closely to the values which would exist in laminar flow with the same surface friction .
Further , there is no indication of the existence of any slip at the boundary .
It was concluded , therefore , that in the turbulent flow of air through this small pipe speeds of about double the critical speed , the exploration of the flow near the boundary by means of a fine Pitot tube showed unmistakably the existence at the boundary of a finite layer of fluid in laminar motion and having zero velocity at the boundary .
As the rates of flow in the experiments in the cm .
pipe were not greatly in excess of the critical value , it was decided to repeat the experiments in the cm .
pipe , and for this purpose the No. 3 Pitot tube was transferred to this pipe and a series of observations made with it at fairly rates of flow ( Table VIII ) .
The results are shown plotted in fig. 10 and the values of the velocities reduced from these curves to show the variation of the speed at constant values of the surface friction are shown in fig. 11 , in the same way results iliustrated in fig. 9 , i.e. , on the left are shown the curves of velocity variation with ( distance of centre of Pitot from wall , and on the right the same results corrected for the interference , on the assumption that this is unaffected by the radius of curvature of the walls .
This assumption may not be true to a degree of accuracy , but it is obviously the only one which could be made under the circumstances .
It will be seen that when the correction for interference is made , the curves of velocity distribution tend to a definite slope at the boundary which is identical with that which would exist in a layer of fluid in laminar motion and having the same surface friction , a result which is in agreement with that obtained from the observations in the cm .
pipe .
The general conclusions which are arrived at from the whole series of experiments here described are that in turbulent motion there exists at the boundary a layer of fluid of finite thickness , which is in laminar motion , and which has zero velocity at the boundary .
Expressing this in symbols , the boundary condition is where the origin is taken in the boundary and is measured along the normal , is the velocity parallel to the boundary , is the coefficient of viscosity of the fluid , and is the intensity of the surface friction .

