]\gt ; On the Flow of Viscous Fluids through Smooth Pipes By CHARLES H. LEES , D.Sc .
, F.B.S. ( Received October 1 , 1914 .
) 1 .
The problem of the flow of a viscous fluid through a tube of circular section is of considerable interest both to physicists and to engineers .
Since Stokes*showed the connection between the viscosity of the fluid and the empirical formula by for the rate of slow or stream-line flow of the fluid through a capillary tube , the tube method has been one of the most useful and accurate for the determination of viscosity .
On the other hand , the extensive use of pipes for the transmission of gas , compressed air , water and oil , at speeds much above those which obtain in a Poiseuille experiment , rendered it necessary to investigate the laws of " " turbulent\ldquo ; flow of viscous fluids through tubes , and there are at the present time several formulae in use by engineers giving the mean rate of flow of such a fluid through a pipe in terms of the dimensions of the pipe and the difference of pressure between its two ends .
Of the purely empirical formulae at first used , that of Hagen , according to which the difference of pressure between two sections apart of a pipe of diameter was given by , where is a constant and the mean velocity , deserves mention owing to its approximation to the truth for incompressible liquids as shown by later work .
WeisbachS modified the older equation by introducing instead of , and this proposal , as will be seen in the sequel , is on the right lines , and gives a fair approximation to the facts so far as they relate to the effect of change of speed .
2 .
placed the power formulae on a rational basis by showing that if the fall of pressure per unit length of the pipe depended on powers of the density , viscosity , and mean speed of the fluid , and on the diameter of the pipe , it must be expressible as , where and are constants .
From the observations of Darcyl he deduced values of varying * Stokes , ' Camb .
Phil. Trans vol. 8 , p. 304 ( 1847 ) .
If are the pressures at the tube oameter dength tviscosity iows wocity vthen pomptes Bendus , ' through which a fluid Hagen , ' Math. Abh Berlin Akad .
( 1854 ) , p. 17 .
S Weisbach , " " Mechanics quoted Gibson , ' Hydraulics , ' p. 197 .
Trans , Scientific P. des Flow of Viscous through Smooth Circular Pipes .
47 from for glass and lead pipes to 188 for new and for old cast iron pipes .
This not proving sufficiently general to cover all the observations available , Unwin*proposed the modified expression , and found values of from for wrought iron and tin plate to for new and for old cast iron pipes , while varied from for tin plate to 116 for new and old cast iron and for wrought iron .
3 .
The important results which have been obtained by Dr. and by Dr. Stanton and Mr. at the National Physical Laboratory during the last three years from observations of the flow of air , oil , and water smooth brass pipes of diameters from to 12 cm .
at speeds from 5 to 5000 cm .
per second , seem by their accuracy and consistency to justify a re-examination of the subject with a view to the establishment if possible of a connection between the physical properties and mean speed of the fluid and the diameter , length , and pressure of the pipe , which shall hold with greater accuracy and over a wider range than the existing formulae have proved themselves capable of doing .
The data available for the examination of the flow in smooth pipes , e.g. drawn brass or lead or bitumen covered pipes , include , in addition to the results given by Stanton and Pannell , those of Saph and SchoderS on water in pipes from to 5 cm .
diameter , which agree very closely with those of Stanton and Pannell , except near the lower limit of speed when the flow is changing from the turbulent to the stream-line type ; those of on water in lead or bitumen covered pipes from 3 to 28 cm .
diameter , which give resistance to flow rather greater than those found by the former observers but varying with speed apparently according to the same law , and those of Beynoldsl on water in lead tubes and cm .
diameter , which give less resistance than any of the others but apparently show the same law of variation with speed .
Stanton and Pannell give a graphical comparison of all these results in fig. 5 , p. 212 , of their paper .
4 .
On the theoretical side of the investigation we have the " " Principle of Dynamical Similarity which , whether established by an examination of the * Unwin , ndustries , ' vol. 1 , p. 51 ( 1886 ) .
Resum\amp ; of the present position of the subject will be found in Unwin 's article " " Hydraulics in the ' Encyclopaedia Britannica , ' edition , and in Gibson 's ' Hydraulics .
' Stanton , ' Roy .
Soc. Proc , vol. 86 , p. 366 1911 ) .
Stanton and Pannell , ' Phil. Trans , vol. 214 , p. 199 1914 ) .
S Saph and Schoder , ' Amer .
Soc. Civil Engs .
Proc vol. 51 , p. 253 1903 ) .
Darcy , moires Acad. des Sciences , ' vol. 15 , p. 141 ( 1858 ) .
Reynolds , ' Phil. Trans , vol. 174 , p. 964 ( 1883 ) .
Dr. C. H. Lees .
On the Flow of viscosity , then fases Rhave tuesame where vpeed , iameter opipe athethe fiscous fluids through pipes , number oases vtheequations ootion oluid , bedoctrine oimensions oquantities , shows t where is the resistance per unit surface contact of fluid and pipe .
Hence , if be taken as the abscissae , and as the ordinates , every case of motion of a viscous fluid through a pipe will be represented by a point of the diagram , and the whole of the points will lie on a curve if the principle is strictly applicable to such cases , but will cover a strip of width which will increase as the applicability of the principle decreases .
The observations made by Stanton and by Stanton and Pannell on the flow of air and water through pipes show that the locus of the points in the , if , the velocity , be taken as the mean velocity over the whole section of the pipe , is a narrow strip the width of which is only of the order of the errors of experiment , and that for cases in which is the same , the law of distribution of velocity over the cross section is the same .
Understanding by in the above statement of the principle , the mean velocity of flow , one of the most impoltant results of the National Physical Laboratory experiments is the proof of the applicability of the Principle of Similarity to the turbulent flow of fluid in pipes , although the eddies set up in the pipes may not have dimensions in the same ratios as the diameters of the pipes , which the principle strictly demands .
Taking then to bs a function of , that is , it remains to determine the simplest form of the function which will embody the results of Stanton and Pannell and of Saph and Schoder , which are in close agreement and appear to be much more accurate than the results previously available .
The curve given by the former observers connecting with ests a relation of the hyperbolic type , and on plotting ordinates against as abscissae the curve was found to be slightly concave upwards , showing that the resistance per unit surrace cannot be expressed as a power of the mean velocity throughout the whole of the experiments .
As the curve given by Stanton and Pannell gests that approaches a finite limit other than zero as ncreases indefinitely , curves of as ordinates against * Stokes , 'Camb .
Phil. Soc. Trans vol. 9 , Pt. 2 , ; 'Math .
and Phys. Papers , ' vol. 3 , .
I7 .
Helmholtz , ' Berl .
Ber 1873 , p. 60 ] ; 'Wiss .
Abh vol. 1 , p. 168 .
Bayleigh ; ' Aeronautics Report , ' 1911-1912 , p. 97 .
Viscous Fluids through Smooth Circutar Pipes .
as abscissae were then plotted with various values of , and it was found that with the curve became a straight line , the slope of which indicated that plus times was a constant , or that Hence or .
( 1 ) Since this expression satisfies the principle of similarity which Stanton and Pannell have shown holds for the mean speed of flow of air and water pipes of diameters between and 12 cm .
, the constants of the above equation hold over the whole of this range at least , and probably this type of relation with a small change of the constants will apply over a much wider range .
The viscosity of the fluid only enters into the expression through the term , and has only a small influence on the result .
the temperature as C. the values of for water and air are and respectively , and the values of , and respectively .
Hence for water at C. and for air at C. and 76 cm .
pressure In many technical cases it is not , the resistance per unit area of surface exposed to the fluid , which is required , but the fall of pressure along a given length of pipe due to it .
Taking the length of pipe to be short enough to allow changes of density of th .
fluid to be neglected we have or and for any fluid .
( 2 ) In the case of water at C. this reduces to ; and in the case of air at C. and 76 cm .
pressure , .
VOL. XCI.\mdash ; A. Dr. C. H. Lees .
On the Flow of In order to show how nearly the formula reproduces the observations of Stanton and Pallnell their original is reproduced in the accompanyi 'ure , with a line drawn on it to show the values deduced from the It will be seen that the agreement is as close as could be expected .
expresslon .
It will be noticed from the form of the final expression for the fall of pressure along the tube , that this fall will , for elocities , be approximately proportional to a power of the velocity between the and the 2nd , and that as the speed or diameter increases or the kinematical viscosity decreases the law of variation will approximate more and more closely to the second power .
This result was obtained by Stanton and Pannell* in the case of the flow of water through a pipe of cm .
diameter .
For mean speeds of 58 , , 900 and 2250 cm .
per second the powers of the velocity were respectively .
Gibson found the same result for the flow of brine of various strengths through smooth brass pipes of diameters from to inch .
Thus for the smallest pipe for speeds between 2 and 10 feet per second the resistance was found to vary as the power of the mean velocity and for the largest pipe and the same speeds as the power .
He was unable to detect any influence of the kinematical viscosity on the value of the index from water to the yest brine he used , for which the density was and the kinematical viscosity times that of water .
In order to arrive at some quantitative estimate of the agreement of these results and the formula given , it is necessary , since most observers have embodied their results in single power of the velocity formulae , to find in each case the best single power formula which will represent the complete formula over the limited range of any particular set of experiments .
The accuracy of the representation will be sufficient for the present purpose if the single power expression has its constants and so determined that its value and the value of its differential coefficient.with respect to are identical , a particular value of given by the experi1nents , with the value given by the complete expression for the same value of That is ) and or Hence .
* Stanton and Pannell , ' Phil. Trans , vol. 214 , \fnof ; g Gibson , ' Inst. Engs .
' Feb. , 1914 .
Viscous Fluids through Smooth Circular Pipes .
Using this expression , we have for the best values of at the points given by Stamton and Pannell the following Table:\mdash ; ' Calculated .
Stanton aannell .
Observed The agreel1lent is as good as could be expected , when we rennember that the values are each calculated for a point of the resistance velocity curve , while the observed values lepresent means over considerable crths of the curve .
Taking Gibson 's observations on smooth pipes of diameters and cm .
, through which water or brine passed , with a mean speed of 180 cm .
per second , we have the following Table:\mdash ; Mean Speed 180 The agreement is satisfactory , and the values of the indices for water and lor the strongest brine are sufficiently alike to account for the difficulty in distinguishing them experimentally .
Caruthers*found for the flow of Texas fuel oil through rough pipes of rs from 2 to 10 feet that the resistance varied as .
This expression lies outside the range of the present formula , but Caruthers does not attach much weight to the indices he adopts .
5 .
So { ar as the effect of temperature on the difference of pressure is concerned , it is almost entirely confined in the case of liquids to the term Roy .
Soc. Proc ( 1912 ) .
Flow of Fluids throu.yh Smooth Circular Pipes . .
For water has the values , 0179 , 0148 at , and C. respectively , and the values of the first term of the expression for the fall of pressure will be in the same ratios .
Since the ratio .
of the value of this ternl to that of the complete expression decreases as the diameter of the tube increases and as the speed increases , the effect of temperature on resistance of the pipe will be smaller as the diameter increases and as the speed incl.eases .
In the case of a smooth brass pipe cm .
in diameter , which vater flowed , with speeds from 120 to 1000 cm . .
second , found that ) relative resistances at and .
were as 1 : 0.87 : .
According to the formula given above , they should be at the .
and as 1 : at the limit of speed .
The agreement is as close as could be expected without a more iled examination of the distribution of 's results about the mean he gives .
It is evident from formula that as the resistance approximates more closely to the law as diameter of pipe or velocity of the fluid it increases , the of chan of temperature on the first term ecolnes relatively less important .
Thus bIair states that when the resistance val.ies ils the power of the speed it decreases per cent. ' degree Centigrade and the resistance valies as the power of the velocity it decreases per cent. per degree .
It is evident that the mnla reproduces the experimental facts at known much more closely than any previous expression has succeeded in doin .
As it been deduced from experiments on smooth it is not strictly applicable to rough pipes , although measurements on pipes give esults which } ) } ) to follow the same general laws as those for smooth pipes .
greater recision in the laws of flow through such can ) attained it will be necessary to come to some as specification and measurement of the degree of roughness of a surface .
6 .
The conclusions arrived at may be summarised as follows : ( 1 ) The difference of pressure in dylles per square between sections distant cm .
from each other along a pipe of diameter cm .
which a fluid density may be considered constant over the length and whose kinematical viscosity is is flowing with mean velocity cannot be represented by a single power of the velocity but requires for its expression a formula of the type where , and are constants .
* Mair , ' Inst. Civil Engs .
Proc vol. 84 , p. 424 ( 1886 ) .
The Silver Voltameter .
5 3 ( 2 ) To the extent to the of Dyllanlical Similarity is applicable to the flow of fluid in tubes , and ?
be absolute constants applicable to all fluids and all tubes .
Stanton and Pannell 's results that over a wide ( 3 ) As the velocity and diameter increase and as the viscosity decreases the pressure diffel.ence varies more nearly as ( 4 ) The effect of temperature on the pressure difference decreases as velocity and diameter increase as the kinelnatical viscosity decreases .
Sifver .
Part III.\mdash ; The Solrent Silver te By T. Received Read June In an earliel ' papel by .
E. Smith* a second by the in conjunction with Mr. nith it was shown that the of silvel deposited by a given current was influenced in a very important rree 1 impurities in the silver nitrate solutions .
The most important impurities are those which are capable of exerting a reducing action upon the silver nitrate .
But there is also a group of substances to be considered which are soluble in silver nitrate solutions , almost insoluble in water : these are precipitated when the silver nitrate solutions are impoverished at the cathode by the passage of the current and cause an appreciable increase in the weight of the deposit .
They may be removed by diluting the silver nitrate solutions , filtering off precipitated impurities , evaporating the filtered solution and draining the crystals of silver nitrate which separate ; any impurities that may have escaped precipitation will be freely soluble in the concentrated mother-liquor from which the crystals have } ) arated , and may be got rid of in the usual way by draining on the pump and cautiously with water .
In view of the importance of these impurities in the experimental deterwtion of the electrochemical equivalent of silver , and the interest of the * Part I , " " A Comparison of Various Forms of Voltameters by F. E. Smith , A.R.C.Sc .
, and ' ' A Determination of the Electlo-chemical quivalent of Silver by F. E. Smith , A.R.C.Sc .
, and T. Mather , F.R.S. Phil. Trans ( 1908 ) ; Part II , " " The Chemistry of the Silver Voltameter by F. E. Smith , A.R.C.Sc .
, T. M. Lowry , D.Sc .
Phil. Trans ( 1908 ) ) .

