]\gt ; Motion of Solids in Fluids when the Flow is not By G. I. TAYLOB , M.A. ( Communicated by Prof. H. Lamb , F.R.S. Received April 13 , 1916 .
) The chief interest in the results obtained in the following pages lies in the fact that a mathematical result has been obtained concerning the motion of solids in fluids which is vel.ified accurately when recourse is had to experiment , with real solids moving in real fluids .
This is so exceptional a circumstance that it is hoped that the interest which it ives to the mathematical work will serve to extenuate , to a certain extent , the clumsiness of the methods employed .
The problem solved is two-dimensional .
An infinite cylindrical body of any cross-section moves in a uniformly flnid with its generators parallel to the axis of rotation .
The stream lines and the reaction between the solid and the fluid are found .
Suppose that a stream function has been found which represents the irrotational motion of an incompressible fluid when a cylindrical solid several cylindrical solids ) of the required cross-section is moved in an assigned manner starting from rest in a fluid which has a given boundary or has a given irrotational motion at infinity .
is a function of the -ordinates of a point in a plane perpendicular to the axis of rotation , and , the time .
Since the motion is irrotational satisfies the relation everywhere , and at the solid boundaries , where represents the velocity normal to the boundary of a point on the surface of cylindrical solid moving in the fluid , and repr sents the rate of change in measured in a direction along the solid boundary .
These , together with the conditions at infinity , if the fluid is unenclosed , are the necessary and sufficient conditions for determining .
The components of velocity of the fluid are then and Now consider the function , ( I ) where is a constant both in regard to space and to time .
It satisfies the dynamical equations of motion , for , which is constant ; and it is the stream function of the fluid motion obtained when the whole system represented by is rotated with uniform yular velocity VOL. XCIII.\mdash ; A. Mr. G. I. Taylor .
Motion of Solids in about the origin .
The boundary conditions of the rotating system evidently satisfied if the cylindrical solids move relative to the rotating stem in same way that they moved relative to fixed axes in the case of the motion represented by .
Hence it appears that the system consisting of the cylindrical solids and the fluid in which they move may be rotated without the motion of the fluid relative to the rotatin system , provided the cylinders are constrained to moye , relative to the rotating system , in the same way that they moved , relative to fixed axes , when the system was not If , however , the solids are free to move under the action of their own inertia and of the pressure of the fluid , the rotation will make a considerable difference to the relative motion of the solids and the fluid .
It therefore becomes important to find the pressure at any poin Let be the pressure at the point in the irrotatlonal case , and let be the pressure when the whole system is rotated .
The equations for and are , ( 2 ) ' ( 3 ) where and are the components of velocity in the rotational motion .
The symbol has been used to represent the rate of change , at a point fixed in space , in the component of velocity parallel to a fixed direction which momentarily coincides with the axis of This is not the same thing as .
Since may be regarded as being known in terms of the co-ordinates and , referred to axes , and the time , represents the rate of in the component of the velocity of the fluid which is parallel to the rotating axis of at a point which moves with the axes .
It is evident that and To find the value of and , consider the rate of , at a point fixed in space , in the component of velocity parallel to a fixed direction which momentarily makes an angle with the axis : * It will be shown later that this proposition cannot be extended to the case of the three-dimensional motion .
Fluids when the not Irrotational .
The component of the velocity of the fluid parallel to this direction is After a short interval of time , , the co-ordinates of the fixed point relative to the axes are and The components of velocity parallel to the rotating axes ( which now make an angle with their previous positions ) are and The component of velocity parallel to the fixed direction is therefore .
The rate of change in velocity parallel to the fixed direction is therefore Putting we find and putting Substituting these values in ( 3 ) , subtracting equations ( 2 ) and substituting for and , it will be found that - and These equations may be integrated in the form .
( 4 ) At this stage it is easy to prove that the proposition proved on p. 100 cannot be extended to the case of three-dimensional motion .
Let be the components of the velocity of a fluid in irrotational motion .
Suppose that the motion defined by is possible .
Mr. G. I. Taylor .
tion of Solids in as before , it will be found that the pressure equations can be reduced to These are not consistent unless and are independent of ; that is , unless the motion is two-dimensional .
Let us now apply ( 4 ) to find the resultant force and couple which the fluid pressure exerts on a solid moving in a rotating fluid .
Let , and be the resultant forces and couple due to fluid on the solid in the case when the system is not and supposed to act at the centre of gravity of the area of the crosssection of the solid .
Let , and be the corresponding quantities in the case when the system rotating .
If represents the angle between the normal to the surface of the solid and the axis , then and are the co-ordinates of a point the surface referred to axes parallel to the axes of and , and passing through ; and the integrals are taken round the surface of the solid .
the value of given by ( 4 ) may be orated .
Thus ( 5 ) Now since .
Also where A is the area of cross-section of the solid , and are the coordinates of centroid .
Fluids when the Flow is ?
2 may be rated by parts .
It then ecomes ( 6 ) since .
evidently vanishes .
represents the velocity of the fl normal to the surface of the solid .
The boundary condition which must be satisfied by is where is the angular velocity of ) body .
Substituting iu ( 6 ) and that \amp ; d and , it will be found that The these rals vanishes and the second be written Now -A and , since , is centroid of the area of cross-section .
Hence from ( 5 ) ) Similarly it will be found that It will be noticed that and are the components of a ARE radially , being the distance of from the centre of rotation .
Also 2 and are the components of a force 2 AQ acting at right angles to the direction of motion of relative to the axes , the relative velocity of C. Now consider the couple due to rotation .
Substituting from ( 4 ) , .
Mr. G. I. Taylor .
Motion of Solids in Neglecting all terms which contain only powers of or of and integrating the second integral by parts , this becomes Now and ; also since leduces to The forces due to fluid pressure , which act on a body moving in an assigned manner in a rotating fluid , may therefore be regarded as being made up as follows:\mdash ; ( 1 ) The forces , and the couple which would act on the body if it moved in the same way relatively to the fluid at rest .
( 2 ) A force equivalent to acting towards the centre of rotation through C. ( 3 ) A force acting at in a direction perpendicular to the relative motion of and the rotating axes , and directed to the left if the rotation of the fluid is anti-clockwise .
We can solve any problem on the motion of cylindrical solids in a rotating fluid if we can obtain a solution of a similar problem respecting the motion of the solids in a fluid at rest .
Now , consider the forces and the couple which is necessary to apply to a solid body of mass , in order that it may move in an assigned manner relatively to axes .
Suppose that a force and a couple must be applied at its centre of ravity , in order that it may move in the assigned manner relatively to fixed axes .
The additional force which it is necessary to apply when the system is rotating uniformly with angular velocity may be shown to consist of a force perpendicular to the direction of the velocity of the centre of gravity relative to the rotating system , together with a force acting through the centre of ravity towards the centre of rotation .
* For they vanish when integrated round a closed contour .
Ftuids whoen the is not Irrotational .
It will be noticed that , if the position of the centre of gravity of the solid coincides with the centroid of its cross-section , and if the mass per unit length of the solid is equal to , that is to say , if the mass and the centre of gravity of the solid are the same as those of the fluid displaced , then these forces are the same as those which on the solid .
to the additional pressures in the fluid due to its rotation .
These considerations lead to the conclusion that , if a solid of the same density as the fluid be moved , a certain path by certain assigned external forces , then a uniform rotation of the whole system , the external force , makes no difference to the path which the solid pursues relative to the system .
This theorem applies only to the case of two-dimensional motion .
In the case of a finite cylinder , for instance , it seems almost obvious that pressures due to the rotation must fall off towards its ends .
It is natural to suppose , therefore , that the reaction of the fluid would not sufficient to hold a finite cylinder in its path when the whole system is rotated .
The case of a sphere in a rotating fluid presents considerable mathematical difficulties , but the initial motion has been ated by .
J. Proudmall , who has kindly conscnted to allow the author to make use of his results , , they are not yet published .
* He finds that , if sphere of volume starts from rest iu the fluid and moves with uniform velocity a straight line relative to the system , it is acted on initially by a force directed towards the centre of the rotation which is at a distance from the centre of the sphere ) by a force acting in a direction perpendicular to its path .
But in order a sphere of the same density as the fluid , that is , one whose mass is may nlove along a path relative to the system , it mnst be acted on by a force directed towards the centre of rotation and by a force perpendicular to its path .
The forces due to fluid pressure are not sufficient to supply the second of these .
If , therefore , the sphere were drawn the fluid by means of a string , it would not move in the direction } string was pulling it , but would be deflected to the left if the fluid were rotating clockwise , and to the if were anticlockwise .
On the other hand , if cylinder of the same density were drawn the ting fluid , the force necessary to hold it in its straight would be supplied by the fluid pressure .
The cylinder would therefore move the fluid in the direction the string pulling it .
* Since the above was written Mr. oudman 1 results .
They appeared in ' Roy .
Soc. Proc , vol. 92 , pp. 408-424 .
Mr. G. I. Taylor .
Motion of Solids in These conclusions have been tested and completely verified by means of experiments made by the author in the Cavendish Laboratory .
with water in a rotating tank .
a Tank of A glass tank full of water was mounted so that it could be rotated about a vertical axis at various speeds by means of an electric motor .
The speeds varied from 2 to 6 seconds per revolution .
Two bodies were prepared , one and the other sphelical .
The former consisted of a piece of thinwalled brass tube about 6 in .
iu .
stopped at the end with waxed cork , while the other was a spherical glass bulb .
They were weighted until they would fall very slowly water , and the positions of the were adjusted till they would stay almost at rest in any position in .
the water .
The centres ravity of the bodies were then coincident with the centres of ravity of the water displaced by them .
A sinlple mechanism was next devised to tow them through the tank from one end to the other .
It consisted of a wood pulley about 4 inches in diameter , mounted on a vertical spindle lich was driven into a wood bridge , fixed to the tank over the middle of it .
This spindle coincided with the axis of ation of the tank .
Cotton was then wound round the pulley , passed through some small rings screwed into a board fixed to one end of the tank , and led horizontally along the tank to the cylinder or sphere , which was fixed at the other end .
The body was held in a holder while the tank and water were being brought to a state of uniform rotation .
A device was arranged so that the holder could release the body and at the same moment the wood pulley on which the cotton was woumd could be fixed in space .
As the tank was then rotating round the pulley the cotton wound up round it , and pulled the bodies along the middle of the tank from one end to the other .
Result.\mdash ; It was found that the cylinder through the middle of the tank .
Even when the tank was very rapidly the cylinder always passed over the central line .
The sphere , however , was violently deviated to the left ( the tank was rotating clockwise ) .
When the tank was rotated quite slowly , about once in 6 seconds , the sphere would not quite touch the side , though it ' came up to the stop at the other end from a direction less than away from the central line .
When the tank rotated more rapidly the complete path could never be seen , because the sphere always hit the side of the tank before it had gone more than a few inches in the direction along which the cotton was trying to pull it .
After striking the side of the tank the sphere would follow the side along , touching all the time , Fluids when the Flow is not till it to a position close to the other end where the string was pulling in a direction making an angle of about with the side of the tank .
It would leave the side and approach the point towards which the cotton was pulling it along a curved path .
The accuracy with which the experiments just described verify the hydrodynamical theory of rotating fluids is at first sight ulost .
Besides the fact that there is apparently no case in which experiments made with real solids moving in real fluids agree with the predictions of hydronamics , it is known that the stream lines of a real fluid round a circular cylinder in particular bear no resem'ulance to the stream lines used in the ordinary hydrodynamical theory .
It will be noticed , however , that in order that there may be agreement between theory and experiment in the articulal .
respect to whicn attention has been drawn .
it is that the actual flow pattern shall be the same as the flow pattern contemplated in the ordinary hydrodynamical theory .
All that is necessary is that the flow pattern in the case of the cylinder shall be two-dimensional , while that in the the sphere shall be three-dimensional .
, nents 'n ; ith Vorte , 'Rings The theory explained on p. 1051eads to the conclusion that if a homogeneous solid , which is not cylindrical , be projected in a fluid of the same derlsity as itself it will be deviated , to the left if the rotation is clockwise , and to the right if the rotation is anti-clockwise , of the path it would pursue through the fluid if the whole system not rotating .
vortex ring affects the fluid round it in much the same way as a solid of the same dimensions as the cyclic portion of the flow system .
If it is projected through a fluid at rest it travels along a straight line .
We should expect , therefore , that if a vortex ring were projected through a rotating fluid it would follow a curved path relative to ) fluid , to the left the fluid were rotating clockwise .
This conclusion was tested experimentally and found to be correct .
small vortex box with a rubber top and a circular hole in the side was made .
This was filled with a solution of fluorescein and placed in one end of the tank , which was filled with water and held fixed .
On striking the rubbel lightly a vortex ring was produced which travelled straight down tlJe tank and struck the middle of the opposite end .
The same experiment was repeated when the tank and vortex box were rotating .
On tapping the box , rings started out in the same direction as before , but were deflected in a curved path , so that they hit the side of the end of the tank .
By the box quite htly and rotating the Mr. G. I. Taylor .
Motion of Solids tank fail .
lie rapidly the rings could be made to turn in such small circles that they came round and struck the vortex box again without the side of the tank on the way .
They would , in fact , turn in a circle whose diameter was only about four times the diameter of the rings .
It was ) ointed out by Dr. F. W. Aston , to whom the writer was showing this experiment , that the rings appeared to remain parallel to a plane fixed in space , while the rest of the fluid rotated .
He suggested that the yyroscopic action prevented the ring from being deviated this plane , and that in order that the ring might move relative to the fluid in a direction perpendicular to its plane it would have to move through the fluied along a curved path .
Jlolion of a which has a Jfotion at Infinity but does not Vecessarily Ras Whole .
The results in the rest of this paper have no immediate practical interest .
The author entered on the ation with a view to an idea of how the instability which is known to exist in a uniformly laminar flow would be likely to manifest itself , and to find out the characteristics of the motion of solids in fluids , which have been discussed in the first part of this paper , have any counterpart in the case of solids moving a fluid whose undisturbed motion is a uniform laminar flow .
The problem of the motion of a circular cylinder in a rotationally fluid divides itself naturally into two parts , that of finding the stream function for a boive motion of the cyiinder , and that of .
the force which the pressure associated with that stream function exerts on the cylinder .
The stream function for a certain type of rotational flow in which the vorticity is uniform will now be found .
Let be the polar co-ordinates of a poin referred to axes through the centre of the cylinder , and let be the co-ordinates of the centre of the cylinder referred to fixed axes , so that the equation represents a line parallel to the axis of , at a distance / from it .
Consider the stream function .
( 7 ) It satisfies the equation everywhere .
If , therefore , constants , etc. , be so chosen that the boundary condition ( S ) satisfied where being the radius of the cylinder , then is the Fluids ?
) the Flow is not stream function which represents the lnotion of a fluid which , if the cylinder were , would be moving in accordance with the velocities iven by the stream function . .
( 9 ) Now ( 8 ) must be satisQed for all values of ; hence we may equate coefficients of , and , separately to zero .
In this way the following relations between the constants are determined:\mdash ; ( 10 ) It will be noticed that , the stream metion of the fluid before the introduction the cylinder , is expressed in terms of co-ordinates referred to moving axes .
In order to find the motion of a cylinder in a fluid whose undisturbed motion before the introduction of the cylinder is known with reference to fixed axes , we must transform ( 9 ) so as to give in terms of co-ordinates and referred to the fixed axes used to fix the position of the cylinder .
The transformation is performed by then becomes .
( 11 ) If the motion of the fluid before the introduction of the cylinder be iven by the function where are given constants , we find , by coefficients of in ( 11 ) and ( 12 ) , the relations d in terms of and ?
, Solving ( 10 ) and ( 13 ) we obtain the following values of , 1 ) , ( 14 ) Mr. G. I. Taylor .
of Solids in Hence the stream function is obtained for the motion of a cylinder in a fluid whose undisturbed motion may be expressed by a stream function of the .
The two particular cases which are of the greatest interest are those of uniform rotation , for which , and uniformly laminar flow , for which the rate of shear .
Before , these cases , however , it is necessaiy to find an expression in terms of the force on the cylinder .
In tTeneral there does not appear to be a pressure like 's for the case of irrotational motion , or the expression given in equation ( 4 ) for pressure in a rotating fluid .
It is necessary to back to equations of motion of the fluid .
If the rate of in pressure along a direction which makes an with the axis of represented the symbol representing an element of in the direction , then the equation of motion is , where represents the component of velocity of the in the direction Its value may be found in terms of by the equation .
( 15 ) may be written where esents , as before , the rate of in at a point fixed in space .
If are the changes in the co-ordinates of a fixed point in time , ( 17 ) represents the rate of in at a point fixed relative to the axes .
The value of may ) obtained by the expression ( 15 ) with respect to time , which occurs in all terms which contain / 20 , or The values of and may be found by resolving the velocity of , the centre of the cylinder , along and perpendicular to Tlaus ; substituting in ( 17 ) , Fluids when the Flow is not substituting this in ( 16 ) , Now represents the component , normal to the surface , of the relative velocity of the fluid and the cylinder .
It must therefore vanish .
Hence and substituting for irom ( 10 ) , If be put equal to , we obtain the variation in pressure round the cylinder in the form .
( 18 ) If and represent the components of the resultant force acting on the cylinder due to pressure , These may be integrated by parts .
then becomes and ( 19 ) By substituting the value obtained for in ( 18 ) we can find the force exerted by fluid pressure when the , cylinder has any motion for which a stream function can be found .
This method will be applied to two particular cases : In Case ( 1 ) the general motion of the fluid is one of uniform rotation .
This problem been solved already in the first part of this paper , but it seems worth while to verify the calculation .
In Case ( 2 ) the general motion of the fluid is one of uniform shearing .
Case l.\mdash ; The stream function of the general motion of the fluid is Mr. G. I. Taylor .
Motion of Solids in .
Ill this case , then , and stream function of the lnotion round the cylinder is Substituting in ( 18 ) the value be found , and substituting this value in ( 19 ) it be found that , If these expressions be transformed by the transformation , so that , are the polar co-ordinates of a point referred to axes which rotate with the fluid , it will be found that the forces , may be resolved into components along , and perpendicular to it where This agrees with the results obtained on p. 104 , for the force whose components are and may be regarded as being made up in the following way : ( 1 ) a force acceleration of the cylinder relative to the rotating axes ) ; ( 2 ) a force acting towards the centre of rotation ; and ( 3 ) a force velocity of the cylinder relative to the rotating axes ) acting at right angles to the direction of relative motion .
These are evidently the same as the three forces discussed on p. 104 .
2.\mdash ; The general motion of the fluid is one of uniform .
The fluid moves parallel to the axis of with velocity , which increases at a uniform rate as increases .
In this case , which be written this with ( 12 ) it appears that and , while .
Hence from ( 7 ) and ( 14 ) Hence differentiating and putting Fluids when the Flow is not Hence from ( 18 ) Hence from ( 19 ) .
( 20 ) This result will now be applied to find the motion of a cylinder of the same density as the fluid when it is projected from the origin with velocity whose components are and The equations of motion of the cylinder are 01 The first of these may be integrated in the form constant .
That is to say , the component parallel to the axis of of the relative velocity of the cylinder and the fluid is constant and equal to U. The acceleration of the cylinder in the direction of the axis of ' is constant and equal to .
If , that is to say , if the cylinder is shot off in a direction perpendicular to the direction of shear , then the component of velocity parallel to the axis of is constant , and the fluid pressure is just sufficient to give the cylinder the acceleration , which is necessary in order that the velocity of the cylinder relative to the fluid round it may remain constant .
This property of uniformly fluids appears to be analogous to a certain extent to the property of rotating fluids discussed on p. 105 .

