]\gt ; Surface Friction .
337 It will be understood from this problem that many similar problems in . .
unnery can be solved by the aid of this diagram with an accuracy probably pear enough for most practical purposes .
An accurately drawn ballistic diagram accompanies this paper and is nproduced in the folded plate , reduced to slightly less than half size .
It may be used in the way exemplified by for the solution of problems of gunnery relating to fire .
The scales to which the of $bis is drawn Velocity , 1 inch feet per second .
Distance , 1 inch feet .
Time , 1 inch second .
Energy , 1 inch foot-tons .
Auxiliary curves:\mdash ; , 1 inch unit .
, 1 inch unit .
Surface iction : Experiments with and Water in Pipes .
By CECIL H. , M.Sc .
, A.M.I.C.E. ( Communicated by Prof. J. E. Petavel , F.R.S. Received December 3 , 1915 .
) During the past hundred years much work , both theoretical and experimental , has been carried out with a view to determiuing the character of the laws the resistance to tangential motion between solid surfaces and liquids or gases .
A general relation between the dimensions of the surface , the velocity , the density of the flllid , and its viscosity had [ been surmised as a consequence of the laws of motion by Stokes , and Osborne Reynolds , but it was left to Lord Rayleigh* to show , from the principle of dynamical similarity , that the phenomena involved could be expressed definitely by a simple mathematical formula .
The laws governin the friction between solid surfaces and water have formed the subject of experimental ations by Froude , Osborne I Reynolds , Darcy , etc. , whilst the parallel case of the resistance to motion between solid surfaces and perfect gases has occupied the attention of Zahm , Brix , Stockalper , and others .
Practically all these investigators ( devoted their energies to experimental determinations of the friction in the mediunl which they employed , and it was not until the subject was taken * Phil. Mag p. 321 ( 1899 ) .
'Advisory Committee for Aeronautics Report , ' 1909-10 , p. 38 .
VOL. XCII .
338 Mr. C. H. Lander .
Surface : by Stanton and Pannell* that any attempt was made to investigate , under certain conditions , of the motion of fluids which widely mollgst the1nselves in their properties of densities and The same ators also took up the question of the limits of of the formulae currently accepted at the time and used in calculations of surface friction .
Experimental upon the filiction between steam and metal or other has been out arry and Carpenter , former on long pipes in the Lambrecht Mine at Anzin , the latter in the laboratories of the Sibley College .
Barry 's results have been reduced to an approximate formula by Ledoux .
These investigations extended only over a relatively small range of vclocity , and the accuracy of the results obtained was insufficient to establish any systematic departure from the velocity squared law ; the investigators therefore contented themselves with finding values of the constant suitable use at the particular mean velocity of their experiments , assulning that object of the present research was in the first place to obtain accurate for the resistance to flow in pipes conveying steam and other vapours with a to checking the accuracy with which Osborne Reynolds ' dimensional law of flow could be made to represent the friction of these fluids .
The work was afterwards extended to ranges over which this law did not accurately apply , and the results have been used to extend the work of Stanton and to the important case of friction between steam and metal surfaces .
Bxpressions for The expression cferred to above as given ) Lord Bayleigb and connecting the linear dimensions of the ) with the , the velocity and the 1-inematical viscosity of the fluid may ) written , where is a function of the one variable , resistance per unit area , densit of fluid , velocity , diameter of the pipe , kinematical viscosity of the fluid .
Trans , vol. 214 , p. 199 ( 1914 ) .
des Mines , ' Ser. 2 , 1892 .
Ledoux on " " Loss of Pressure in Pipes Proc. Inst. Mech. Eng. of ' 1893 .
Experiments with Steam and Water in Pipes .
Previous to the establishment of the above expression Reynolds had ihown from the theory of dimensions that , assuming the frictional resistance flow to vary as the power of the mean velocity , the relation between Ae resistance , the viscosity , the density , and the velocity would be erxpressed by assumption made by Reynolds is equivalent to taking the function of in Rayleigh 's formula to be the power or that In the reduction of his experimental work on the flow of water in pipes Reynolds* found that over the range of velocities used the expression held with considerable accuracy in any one pipe , but that varied in some manner with the roughness of the surface , and also increased with the pipe diameter .
Above the critical velocity varied from for a smooth pipe inch diameter up to for Darcy 's rough pipes of 20 inches diameter .
Stanton and Pannell found that a considerable deviation from the simple exponential law became apparent when the velocity in the pipe varied to 3000 cm .
per second , but that for small ranges Reynolds ' formula fitted in with the experimental results to a fair of Limits of Present Experimental In the research the moving fluid was in most cases either dry saturated or slightly wet steam , and the experimental range extended habitually to 4000 cm .
per second , although in a few experiments a velocity of 6000 cm .
per second was obtained .
It was found undesirable to extend the work to speeds above this , since extremely speeds , such as can be obtained easily with steam , are associated with superheating down the pipe , making the reduction of the results obtained too complicated and unreliable .
The maximum value of , however , amounts to about C.G.S. units , since by at a pressure the value of the kinematical viscosity may be considerably rednced below that of air at atmospheric pressure and a increase in effected .
the low values of corresponding to the experimental pipes used , water was adopted as the moving fluid , and thus the curve extended backVards to values of this function corresponding to the point where the Oharacter of the curve changes , and below which the resistance varies directly as the velocity .
In all pipes values of the resistance corresponding to low velocities of 'Phil .
Traus 1883 , p. 935 .
.340 steam decreases , until eventually relial ) results become impossible to the serious eflects of condensation .
Condensation is also of importance when the experimental pipe is long since the effect of heat loss causes the steam to become wetter as it down the pipe , the specific volume and the density are altered , and the velocity is slightly from point to point .
The difficulty , however , measurmg the extremely small pressure differences which obtain in lengths of caused the author to adopt pipes varying in length from about 30 to 80 feet , and correcting for the effects of condensation and change in velocity in the manner hereafter described .
For the final plotting of the results , the experimental values were reduced to those obtaining over the first unit length of the pipe , thus making a comparison possible with other investigators ' results for air and water .
nental Apparatus .
apparatus consisted of three mild steel steam pipes of commercial quality , the joints made by screwed overlapping the ends of the two pipes , which were butted one against the other , leaving a uniform bore across the joint .
Each pipe was connected by means of a short leadingin with a large separator supplied with steam by a 300-H.P .
Babcock and Willcox boiler .
The large capacity of the boiler rendered it easy to maintain steady pressure conditions the trials , which frequently occupied upwards of one hour .
The measurements of pressure drop were made at points in the tube a definite distance apart .
At these points , A and ( fig. 1 ) , a number of small holes were drilled round the pipe , the inside being carefully smoothed down .
It follows from the work of Bramwell and on Pitot tubes , that by this method the static pressure at any point may be accurately determined .
The difference in pressure the two ends of the measured length was obtained for the bulk of the experiments by a -tube containing lmercury , ( fig. 1 ) , the two legs of the being connected by couplings Fi and and tubes and I to the of holes at the two " " take-off\ldquo ; points A and B. For the smaller pressure differences the -tube gauge was inverted ( fig. 2 ) , water being used to measure the difference in head , an 'Advisory Committee for Aeronautios , ' 1912-13 , p. 33 .
BramwelI and Fage on .
A DeteIsnination on the Whirling Table of the Pressure Velocity Constant for a Pitot ( Velocity Head and Static Press ) Tube ; and on the Absolute Measurements of Velocity in Aeronautical Work bke-off castings and and the pressure tubes and I. The level of the water in these chambers could be read on the gauge glasses and M. A scale for the -tube was marked off , being calculated in pounds per square inch , corresponding to mercury }ater , and scales for the glasses in pounds , to the water heads , the two latter bein reduced to the same zero by careful levelling .
will be seen , then , that the pressure difference at the two ends of the experimental pipe was equal to the reading in pounds of the mercury added to the difference in pounds between the two readings of the water glasses .
Both collectors were carefully lagged to reduce the condensation to a minimum .
The amount of water densed was collected from the gauges by the cocks , and measured by weighing in a known quantity of water .
In the upper of the two collectors the condensed water was due entirely to the heat loss from the tor itself , and may be taken as a measure of the steam flow the pressure holes , C. The amomlt was small , and such that the dynamic head corresponding to its velocity did not in any case exceed per cent. of the least pressure differenc observed .
The pressure , as measured , could therefore be considered static .
In the lower of the two collectors the water collected was due both to a proportion of the water condensed in the experimental main and to that condensed in the collector .
It is clear that the former will make no difference to the measurements of the static pressure , and the latter will only affect the readings to the same extent as that in the high-pressure collector .
The total amount of water collected in the low-pressure collector was added to that discharged in the hot well , since all results have been reduced to the initial velocity in the pipe , and this water in the of steam has already passed the high-pressure take-off point .
It is obvious that proper precautions must be taken to ensure that water and mercury arc continuous in the pressure-measnring devices , and not intercepted by spaces .
After passing the second of the two take-off points , the steam was led into a separator , and thence passed through a plug valve to the condenser .
The plug valve was arranged to act as a throttling calorimeter , a pressure and a thermometer pocket being inserted in the pipe on the low-pressure side .
Condensed water was pumped out of the condellser by an air pump , and collected in a calibrated tank .
The separator was fitted with a gauge glass and a scale calibrated to read in pounds collected in the separator .
The whole of the piping was well clothed with slag wool wrapped with canvas , the thickness of the covering averaging inches throughoub . .
number of experiments were also carried out on the bare pipe .
Mr. C. H. Lander .
Surface Friction : The condition of thoe inlet steam was adjusted by means of a throttle valve on the inlet pipe , situated at a distance of about 6 feet from the high-pressure measuring point , the method being to obtain a few rees of superheat at a thermometer pocket and Bourdon feet from the high-pressure measuring point .
Dry steam at the entry of the experimental length was thus ensured .
Any instability of flow due to the thermometer would die down in the 20 or more diameters before it reached the take-off holes , and in any case its effects would be negligible in view of the long experimental pipes used .
Finally Bourdon pressure were connected directly to the takeoff muffs in order to obtain readings at high velocities .
Results .
The observations taken con.sisted of the initial temperature and pressure of the steam , the pressure difference between the take-off points , the dryness on the second separator and the weight of steam that condensed in the lower of the two take-off muff chambers ( see Columns 1 , 3 , 7 , and 8 of Table ) .
Tests were also made on the heat loss from the main by allowing an extremely small flow through the pipe , collecting at the same time the condensed water .
The method of inferring the velocity from the volume passing and the diameter of conduit necessitates a very accurate determination of the latter since the value of depends on the fifth power of the diameter .
For this purpose a measured length of the pipe was filled with water , emptied and the quantity accurately weighed , the diameter being inferred from the volume .
The mean result obtained from six determinations was taken as the correct value .
This method appeared to the author to be the most suitable one , since by leaving a wet surface ou the interior of the pipe it approximated more accurately to the actual diameter and state of the surface wheu steam was flowing .
Condensation continually takes place and a onary film of water is of necessity adhering to the fhness of the interior surface .
of Experimental Values .
In order to detelmine the form of the law governing the frictional losses in any one pipe at ven initial steam pressure Reynolds ' form of 's law was adopted , since the range of velocity was small and well within the limits of accuracy of the exponential law .
then that .
let weight of steam condensed per second per linear foot of pipe , this being the intrinsic amount of water present due to the loss through the pipe covering partly neutralised by drying effects of friction , weight of dry steam entering pipe per second , .
weight of dry steam flowing past a given point distant along pipe ; then The relation between the volume of unit weight of dry steam and its absolute pressure has for this purpose been taken as that iven by constant , since the change in specific volume down the pipe is small .
Density of dry vapour Volume of weight of dry vapour But the actual substance past any section of the pipe is a mixture of steam and a small proportion of fine drops of water .
The density of the mixture , and therefore .
The velocity of the fluid .
Inserting these values of and in the dimensional formula , we ' which may be written Assuming the viscosity to be sensibly constant down the pipe*and integrating between lengths and , pressures and , we get In order to plot the results the formula was put in the form , ( 1 ) for any given value of the diameter and the pressure .
values of * The justification for his assumption may be inferred from an inspection of the final plot , fig. 3 .
At low pressures in small pipes , i.e. where this error would have its greatest effect , the value of works out at about .
At pressures of kgrm .
per sq .
cm .
reatest observed dence oressure wkgrm.sq.corresponding theoretical change oiscosity .
Since tev iscosityappearsinthe formula with the index or , or say 0.5 per cent. , the maximum relative of error will be If now represents the pressure at a point distant unit from we get , ( 2 ) but oyel unit length at the entry to the experimental length approx. Dividing ( 1 ) by ( 2 ) we are thus led to the equation since small ; from which the drop in pressure over unit length at the entrance to the pipe can calculated .
The principal values of the resistance per unit surface and the velocities , with values of and , are given in Tables I , II , III , and .
Details of the rednction of typical experimental values will be found in Table In reductions , values for the viscosities* at the various steam temperatures have been obtained by linear extrapolation of the results given by Puluj , Kundt and , and Meyer and Schumann for temperatures and 10 respectively , this method being justified since i.nspection of the esultant curve ( fig. 3 ) shows it to be comparatively insensitive to slight inaccuracies in the values of the viscosity , thus an error in the estimation of the viscosity as great as 10 per cent. would only cause a horizontal shifC of the point involved by , an amount which in view of the comparatively slow change of ordinate with abscissa is negligible and commensurate with the accuracy of other parts of the ation .
In the bulk of the expsrinlents bhe pressure head corresponding to the increase in velocity did not exceed per cent. In the cases , however , where pressure drops were experienced a correction was applied to allow for the proportion of the pressure drop necessary to accelerate the steam from the initial to the final velocity .
values of the viscosities for saturated steam at temperatures above C. are not accuratoly known .
From experiments carl.ied out below that temperattlre it appears , however , to follow a law and being approximately and respectively .
Mr. C. H. Lander .
Surface : Comparison of tlRsults with those of other nenters on Water and In fig. the values of all the experimental results have been plotted with as ordinates and as abscissae .
The mean results as calculated from the formulae Ledoux for their steam velocity are also plotted on the diagram .
The boundary curves for Stanton and Pannell 's results upon a drawn brass pipe are shown , together with their interpretation of ' experiments upon lead pipes .
A curve has also been drawn illustrating Stanton's* experiments on artificially roughened pipes , in which by cutting screw threads in alternate directions on the inside of the pipe he succeeded in obtaining two pipes in the resistance varied directly as the velocity squared .
The resistance for these pipes as given by Stanton is expressed by the formula where is the velocity at the axis .
From the curve in Stanton 's paper giving the radial distribution of velocity in these pipes the value of the mean velocity has been determined , the horizontal line in fig. 3 the resistance in the screwed pipes reduced to terms of the mean velocity .
It will be noted that the author 's points for steam lie with fair accuracy upon a smooth curve slightly higher than Stanton and Pannell 's boundary curves for low values of becoming proportionately higher for large values .
This characteristic is quite in conformity with the curves for rough and smooth brass pipes , the rough pipe offering proportionately much more istance than the smooth pipe at the high values of It has been shown by Lees that the results of Stanton and Pannell 's experiments can be represented with considerable accuracy by an expression of form the constants for air and water in smooth brass pipes being , and .
With a modification of the constants the same expression represents the mean of the present experiments , the line in fig. 3 having been plotted using the values : From the general trend of their curves for high values of Stanton and Pannell predicted a limit beyond which the resistance would be constant and hence independent of .
This is equivalent to the resistance varying as the square of the velocity .
In the steam curves , owing probably to the ' Roy .
Soc. Proc , vol. 85 , p. 'Roy .
Soc. , vol. 91 , p. 46 .
352 Surface Friction .
pipes being htly rougher , this limit would appear to be reached earlier and the ultimate value of to be considerably above that corresponding to the smooth pipes used by Stanton and Pannell .
The characteristics of the various curves appear to denote that the limiting value of beyond which the resistance varies as the velocity squared becomes less with the roughness of the pipe surface , until , for a very rough surface , the resistance val.ies as the velocity squared for values only exceeding by small amounts that corresponding to the critical velocity .
The hness of the steam pipe surface appears to be about the same as that of Darcy 's pipes , the bulk of which were bitumen covered .
points out that if the law of similarity is to apply it must extend even to the proportionality of the projections and hollows which constitute the rou,.lmess of two given surfaces .
Since in the present experiments three different diameters of pipe were used , the surfaces of which were probably of the same of roughness , it appears that , as was observed by Stanton and Pannell , in this the effect of any want of proportionality of roughness is small .
The close reement between the results obtained by steam and water in the same mains will also be noted .
Conclusions .
Further direct evidence has been produced to demonstrate the truth of the dimensional law , which is now shown experimentally to extend to the case of saturated vapoul .
S. The fact that for the first time a range of pressures extending from about 11- ) inches of vacuum to .
per square inch above atmosphere has been used for the gas experiments is also not without interest .
It is hoped that the work will prove of value to those who require data as to the resistance in steam and other mains , and who up to the had to content themselves with a velocity squared law constants for few values of the velocity .
Again , .the further demonstration of law of dynamical similarity appears to open up a wide field for testing purposes .
The metLod of inferring th resistance of bodies of val.ious shapes to the flow of air from experiments made on models in water has been in use for some time .
The present work , by demonstrating the truth of the law as applied to the flow of steam and water in pipes , opens up the possibility of inferring the resistance of complicated steam passages , such as are met with in tnrbines , by tests on models the law of similarity .
It is hoped in the future to extend the work on lines .
Converging Sequences in of In conclusion the author desires to express his sincere thanks to Prof. E. Petavel and the authorities of the Uniyersity of Manchester for the valuable granted for the carrying out of the work , and also to express his gratitude to the Council of the Royal Society for grant to defray the cost of the special apparatus needed .
on xistence of Sequences Succession.s By Prof. W. H. , Sc.l ) .
, Received Angust 1 .
The object of the note is I ; prove , in large important , a succession of functions can be shown to contain sub-sets of functions which converge .
The interest attaching to the question is well known , and is sufficientl illustrated by the use I have nladc of these considerations , for example , in my paper on the conditions that a metrical series should have a certain form , as well as elsewhel.e .
theorem on the subject is , as is there pointed out , due to 2 .
Theorem th tinuovs on the , there is countable set of everywhere , such that the of th , function at any ) not to set is the limit of the of th ' function at set in neighbourhood of the point that up to th In fact , since the limits of approach are the same on the left and on the , except at a countable set of points , it follows that , except a countable set of points , the fnnction is both lowel semi-conl , inuous and upper semi-continuous ; in other words , that it is , ) at the points of such a set .
Add to this set , should it not be everywhere , any countable everywhere dense set , and we get such a set as contemplated in the enunciation .
For a point not belonging to it the value is the unique limit of values in the eighbourhood equally , whether we confine our attention to points on the countable set or not .
A revision of one part of the proofs , due to .
Grace Chish eiveot on October 25 , 1915 , and has been porated in the paper .
Published in these oceedings , vol. 88 .
The theorunl in .
to Theorem 4 of para .
of the present communication is there utilised , but its proof is in part based ou an roneous theorem , qnoted from the ' Proc. Loudou Math. Soc. ' VOL. XCII.\mdash ; A. 2

