91 Experiments on the Flow of Viscous Fluids through Orifices .
By G. F. Davidson , B.E. ( Sydney ) , B.A. ( Cantab .
) , 1851 Exhibition Research Scholar .
( Communicated by Prof. B. Hopkinson , F.R.S. June 26 , 1913 .
) Received May 23 , \#151 ; Read This paper deals with an experimental investigation of the flow of liquids through a round hole .
In most of the experiments a fairly thick oil was used , and by varying its temperature the kinematic viscosity could be varied about a thousand-fold .
It was possible by this means to change the character of the motion continuously from the type determined almost wholly by viscosity to the form in which the resistance was mainly due to ElecfrTcaf , time Sicjnals .
G ^.St AjJui Strip of blackened paper inertia , and to follow the corresponding change in the relation of resistance and flow .
When the experiments were first undertaken it was hoped by comparing experiments with orifices of different sizes , and in which the viscosity and rate of flow were suitably varied , to obtain confirmation of the law of similar motion in viscous fluids .
It was found , however , that this law did not hold , and the cause was discovered to be that the stress in this oil is not proportional to the rate of distortion .
Fig. 1 is a diagram of the apparatus used .
The tube A was 20 cm .
in diameter and 150 cm .
long , closed at the bottom by a horizontal brass plate with an orifice at its centre .
This tube was held in the centre of a tank 75 cm .
in diameter and 150 cm .
high .
The tank was a closed vessel with a small cock C for admitting air under pressure and an aperture B , 10 cm .
in diameter , by which the air pressure inside the tank could be suddenly released , he apparatus stood in a vertical position with the oiifice immersed , say , 10 or 20 cm .
below the surface of the fluid in the tank .
Aii pressure in the tank forced the fluid up in the tube A , so that the two sui faces of the fluid were about 100 cm .
apart .
On opening the aperture B the fluid in A flowed through the " drowned " orifice under a " falling head " until the two surfaces of the fluid came to the same level .
VOL. lxxxix.\#151 ; A. t Fig. 1 .
92 Mr. G. F. Davidson .
Experiments on the A float on the fluid inside the tube had a strip of paper attached to it which passed over a very freely running and evenly balanced aluminium wheel , and was held taut by a small weight attached to it .
This strip of paper was blackened and passed underneath an electrically operated stylus which marked on it equal intervals of time ( such as seconds for an orifice 2 cm .
in diameter ) as the level of the fluid fell in the tube A. The record was fixed with varnish , and from this the velocity of the surface of fluid in tube A can be found at any head , because zero head is iparked on the paper at the end of each experiment when the fluid is at the same level inside and outside the tube .
Corrections in head are made for the variation in level of the fluid in the outer tank as the column of fluid inside the tube falls .
Corrections in velocity are made for the velocity of approach .
The correction required owing to the retardation of the falling column giving a greater head than the record showed was quite negligible .
Knowing the dimensions of the tube and the velocity it was possible to calculate the actual discharge .
This divided by ^/ ( 2 \lt ; /H)* x area of the orifice gives the coefficient of discharge , which we call C^ .
Thus can be plotted against head .
Actually , the records were measured up and two curves drawn , showing velocity and head each to a time base .
Convenient readings of head were taken from these faired curves , and from the corresponding values the was calculated so that the C^ against head curves do not show experimental errors .
These errors were small , and the " fairing " of the original curves ( to a time base ) was \#166 ; almost unnecessary , the points being usually within 1 per cent , of the smooth curve .
Three orifices were used .
They were 4 , 2 , and 0'5 cm .
in diameter , and were made in plates 3'2 , 1*6 , and 04 mm. thick respectively .
They were made geometrically similar by having the edge of the orifice rounded to a semicircle whose diameter was the thickness of the plate .
The fluid used was a heavy engine oil of good quality , with a flash point of 260 ' C. About 60 gallons were used in these experiments .
It proved very suitable , and as it was kept in a closed vessel it remained perfectly clean and uniform during the 18 months it was under observation .
It showed no signs of streakiness , although , of course , considerable care was needed to get such a large quantity at a uniform temperature when it was very viscous .
In such cases it was " explored " with a resistance thermometer .
* Where H is the difference in level of the two surfaces of the fluid at any instant and a is the acceleration due to gravity .
Flow of Viscous Fluids through Orifices .
93 A wide range of viscosities was obtainable by varying the temperature .
The value of p/ p was about 200 at 11 ' C. , and 0T4 ( about the same as for air ) at 110'.* The viscosity was measured by allowing 126 c.c. of the oil to flow under a small head through a tube 6'2 cm .
long and 6T mm. in diameter .
The viscometer was simply a spherical bulb with two tubes attached to it opposite one another , and held in a vertical position .
Errors were looked for both with water and oil , on account of wall effect , vortex motion , and baffling action by the bottom of the tank .
The wall effect was examined by putting sleeves of various diameters inside the large tube A , and observing when a variation in the coefficient of discharge due to this cause became perceptible .
It was found to be negligible .
The absence of vortex motion , that is the fluid swirling round so that a particle follows a spiral path towards the orifice , was proved by observing small floating bodies .
A baffle plate , distant only one diameter of the orifice away from it on the downstream side , had no effect on the C\lt ; * .
Accordingly , the results are considered to be accurate within 1 per cent , for one infinite fluid flowing into another through this sort of round-edged orifice .
The working of the apparatus was tested with water flowing through sharp-edged orifices of 2 cm .
and 4 cm .
diameter .
The results agreed well with those obtained by other experimenters .
It is interesting to note that with this arrangement it is possible to measure the Gd at much lower heads than in the case of an orifice discharging into air , because the surface of the fluid can be remote from the orifice , and is not affected by the motion there .
For water the for a " drowned " orifice as compared with a " free " discharge is only reduced about 1 or 2 per cent. A non-viscous fluid would have a constant Cd at all heads ( somewhere about 07 for these orifices ) until the pressures are such as to produce appreciable changes of density .
The value which the Gd should have when flowing through an orifice in a thin plate into space has not been calculated for a circular orifice , although it has been for a long rectangular one .
Many experiments have been made to determine this C which is in the neighbourhood of 06 for circular orifices with sharp edges .
With viscous fluids at low heads viscosity becomes relatively important , and the effect of inertia negligible .
The velocity then is proportional to the Throughout the paper the word " viscosity " is used to signify the kinematic viscosity pjp .
denoted by v. Mr. G. F. Davidson .
Experiments on the head , and the curve Cd against head ( H ) should be a parabola , since Cd is really a measure of Results for Oil.\#151 ; Fig. 2 shows a set of curves of Crf against head for the orifice , 4 cm .
in diameter .
The dotted lines for the curves at viscosities of 24'5 , 497 , and 206 The numbers in brackets are the values of the kinematic viscosity of the fluid in C.G.S. units .
Orifice drowned Head cms .
3 2 mm.thick .
Fig. 2 .
C.G.S. units are the parabolas which the curve approximately follows under the low heads .
That it does not do so completely is due to the fact that the viscosity of the oil is not independent of the rate of distortion , a point which is referred to later .
It is interesting to note the relative effects of the viscosity and inertia forces .
Up to a Gd of , say , 0'3 the curve rises sharply , fulfilling more or less the parabolic law , but above this the inertia forces tend to flatten it until at a Cd of 065 the inertia effect is predominant , and the curve continues as an approximately straight line parallel to the H axis .
The general effect of viscosity is to turn the curves downwards .
Fig. 3 , however , shows a curve for water , whose viscosity is about O'Ol , which actually is curved in the opposite direction , and the flat part of its curve has a Cd below 066 , while the viscous oils rise to 07 .
Flow of Viscous Fluids through Orifices .
On this figure also are shown three curves for fluids discharging through this round-edged orifice into air .
It is interesting to note the small difference between the two curves for water .
The two curves for oil discharging freely WATER C-OI ) FREE .
WATER DROWNED ( -01 ) 4CM .
ORIFICE Discharging to air , and drownedr HEAD CMS .
Fig. 3 .
can also be compared with the corresponding ones for the drowned orifice , and it is to be observed that the difference becomes more marked as the viscous forces become more important .
One would expect the flat part of the C^-H curve to attain the same value for all fluids .
The curious thing is that , for a mobile fluid like water , flowing through a 4 cm .
orifice , the flat part of the curve is considerably lower than that for the viscous oils .
The water curve ( even for the drowned orifice ) also curves upwards , and at low velocities rises above the upper limit prescribed by inertia .
Another curious point is that at high heads with drowned orifices ( as with air* ) the curve is not truly flat , but varies slightly as the head increases .
These variations are probably caused by the viscous forces increasing the coefficient of contraction Cc , and decreasing the coefficient of velocity C\#187 ; .
For the coefficient of discharge , = Cv xCc .
A viscous medium round the issuing jet , and the viscous resistance near the surface of the plate in the neighbourhood of the orifice , will both tend to ^ Watson and Schofield , 'Proc .
Inst. Mech. Eng. , ' May , 1912 .
Mr. G. F. Davidson .
Experiments on the increase the contracted area of the jet by giving the fluid a rotation outwards .
In this way the Gc may be increased , while the inertia forces are so large as to keep C\#187 ; nearly unity .
Hence , at high velocities , the more viscous the fluid , the higher the flat part of the curve will be , as is shown by water and oil in these curves .
It is curious to find a friction effect like viscosity actually increasing the quantity discharged per second .
At low heads with water , this same effect causes the Cd to rise above the value for the flat part of its curve , but , with a more viscous fluid , the Gvis so reduced that it overcomes the increase in Gc\gt ; and so the Cd is reduced and the curve is concave downwards .
It is of interest to apply to these results the principle of dynamical similarity.* Consider the motion in an experiment in which the fluid has viscosity v and the orifice is of diameter d. If the length scale of the motion be altered in the ratio L , and the velocity at every point multiplied by a constant Y , a geometrically similar motion will be derived going through a similar orifice of diameter L d.This second motion will in general be dynamically possible if , and only if , the viscosity of the fluid in which it occurs is Ei\gt ; , where YL/ E is unity .
Further , if this condition be fulfilled , the pressure at corresponding points , and , therefore , the heads under which these two motions go on , will be in the ratio V2 .
Conversely , if H be the head in the first case , then the similar motion so derived from it will be that which actually occurs when fluid of viscosity Ei^ flows through a tube of diameter Ld under a head V2H .
And if in any two experiments the head , the viscosity , and the diameter are so related that d*yiL/ v is the same for both , then the motions must be similar in these two cases , and the velocities at corresponding points , and , therefore , the average velocities which then hold , will be in the ratio of the square roots of the heads .
Since Cd is equal to average velocity ( 2yH ) , it follows that it must be the same in both cases .
That is , Cd should be the same in all cases for which YL/ v is the same , where V might be taken indifferently as representing the average velocity of the flow , or the square root of the head , L as the diameter , and v as the kinematic viscosity .
In figs. 4 and 5 the values of Cd found in these experiments are plotted against VL/ *\ From these figures it is seen that Cd varies from 02 to 0-4 when VL/ = 2 , from 0'4 to 0'6 when YL / v = 10 , and so on .
The experimental errors are considered to be well within 5 per cent. * Dr. Stanton , 'Trans .
Inst. Nav .
Archs .
, ' March 22 , 1912 .
, Flow of Viscous Fluids through Orifices .
97 The variation of C a in different experiments for which YLJv is the same can only he explained by the failure of one or more of the assumptions underlying the application of the principles of dynamical similarity to fluid motion .
These assumptions are : ( 1 ) That the stress varies as the rate of distortion .
.5 .4 .2 I 0 2 4 6 8 v|_ 10 12 14 16 18 20 C.C.S. units Fig. 4 .
For values of Cd above -5 see Fiq 5 .
J______I ; i l i_____L Jj___A 1 i Vaiuesfortha '/ 2cm orifice J___________________________________________________i______________________________________________ !
_ __________________________________________________1___________________________________________________I \#166 ; L. T5 o Til T i ir l#T Val ues for the '6 .
cm .
orifice 60 vl so ~"=C-C-S units Fig. 5 .
( 2 ) That gravity does not determine the motion except in so far as it produces differences of pressure at points where the fluid is at rest .
That the second assumption is fulfilled in these experiments was proved by reversing the flow .
The same results were obtained whether the flow was upwards or downwards through the orifice .
That the first assumption is not fulfilled for this oil is suggested by the divergence of the C^-H curve ( fig. 2 ) from the parabola , when the inertia forces are unimportant .
In order to determine the amount of variation of viscosity and its possible 98 Experiments on the Flow of Viscous Fluids through Orifices .
effects an absolute viscometer was made .
It consisted of a glass tube 10| cm .
long , and 2 mm. diameter , through which 220 c.c. of oil were caused to flow by air pressure .
The air pressure came from a reservoir of 40,000 c.c. capacity , and was measured by a manometer .
The volume of oil was measured in a spherical bulb of 220 c.c. capacity .
There was one of these bulbs connected to each end of the capillary tube by about 4 cm .
length of piping 1-|- cm .
internal diameter .
The viscometer was contained in a large water-bath to keep its temperature constant .
The temperature was kept constant and the time of flow was measured .
The head varied from under 4 cm .
to over 52 cm .
of mercury .
The viscosity was calculated from the ordinary expression for flow in tubes and reduces in this case to v \#151 ; constant x time of efflux x head .
The critical velocity ( mean ) would be about x 105 cm./ sec. in this apparatus .
Since the actual mean velocity never exceeded 15 cm./ sec. , and v was over 9 , there was no possibility of eddying motion .
The apparatus was designed to give roughly the same average rates of distortion as occurred in the principal experiments with the orifices .
Owing to lack of time the variation of v was only thoroughly examined at one temperature , namely 25 ' C. A few readings were taken at 35 ' C. ( where v is about 4 ) which showed the same sort of variation as those at 25 ' C. , but not to such a marked extent .
The values of v found varied from 13-7 to 9'4 , and are shown on fig. 6 plotted against the rate of distortion , both being in C.G.S. units .
The value of v found in the first viscometer fits nicely on this curve .
-----value obtained in original experiment .
Oil at 25'C .
Rate of Distortion ( in secs .
) = 8V/ 3 ( Rad. of Tube ) .
Fig. 6 .
The variation of viscosity with rate of distortion precludes any application of the laws of similar flow to these experiments , because the rate of distortion varies from point to point , and the corresponding changes in v will alter the form of the flow .
But generally speaking the effect of the changing viscosity is that the values of v assumed in calculating the results exhibited graphically in figs. 4 and 5 , are too high , the error being greater at the higher viscosities If allowance be made for this by taking in each case the value of v correEffect of Rotating Magnetic Insulator Magnetic Field .
99 sponding to the actual average rate of distortion in that experiment , the result is to reduce somewhat the range of values of VL/ i ; corresponding to any \#171 ; iven Crf .
Thus the variation of viscosity may be said to furnish a qualitative explanation of the large divergences in this oil from the accepted laws governing similar motions .
It is perhaps worthy of notice that there were indications of a want of proportionality between stress and rate of strain in this oil even when it was as mobile as air .
This work was done at the Engineering Laboratory , Cambridge , and I wish to express my thanks to Prof. Hopkinson for his kind help and inspiring interest .
On the Electric Effect of Rotating a Magnetic Insulator a Magnetic Field .
By Marjorie Wilson , B.A. , M.Sc .
, and H. A. Wilson , D.Sc .
, E.R.S. , Professor of Physics , Rice Institute , Houston , Texas , U.S.A. ( Received May 26 , \#151 ; Read June 19 , 1913 .
) In a previous paper* by one of us it was shown that when an insulator of specific inductive capacity K rotates in a magnetic held there is an electromotive force induced in it equal to that in a conductor multiplied by 1 \#151 ; K-1 .
The object of the experiments described below was to measure the induced electromotive force in a magnetic insulator rotating in a magnetic held parallel to the axis of rotation .
According to the theory based on the " principle of relativity " this induced electromotive force should be equal to that in a conductor multiplied by 1 \#151 ; ( yaK)-1 , where / / .
is the magnetic permeability of the insulator , whereas on the theory of H. A. Lorentz and Larmor the appropriate multiplier appears to be 1 \#151 ; K-1 , as for a non-magnetic insulator.f Ao insulator is known for which yu , differs appreciably from unity , so that it was necessary to construct a sort of model of a magnetic insulator .
The insulator adopted consisted of wax , in which a large number of small steel spheres was embedded .
The spheres were \#163 ; inch in diameter , and each one * " On the Electric Effect of Rotating a Dielectric in a Magnetic Field , " by H. A. Wilson , ' Phil. Trans. , ' 1904 , A , vol. 204 .
t M. Abraham , ' Theory der Elektrizitat , ' vol. 11 , p. 322 .

