]\gt ; INDEX SLIP .
COKER , E. CLE IRNT , S. B.\mdash ; An Experimental Determination of the Variation with Temperature of the Critical Velocity of FJOIY of Water in Pipes .
Phil. Trans. , , vol. 201 , 1908 , pp. EHlfBN S. B. , and COKEB , E. G.\mdash ; An Experimental Determination of the tion with Tenxperature of the Critical Velocity of Flow of Water in Pipes .
Phil. Trans. , , vol. 201 , 1903 , pp. 45-61 .
Flow of Water , Critical Velocity\mdash ; Variation with Temperaturs .
COKER , E. G. , and CLElrBNT , S. B. Phil. Trans. , , vol. 201 , 1903 , pp. 45-61 .
Viscosity , Relation to Critical Velocity for Regular Flow .
COKmR , E. G. , and CLHXKNr , S. B. Phil. , vol. 201 , 1903 , pp. 45-61 .
III .
Determination o.f the Variatiomoith emperature of the Critical Velocity of Ftow of Water in Pipes .
E. G. COKER , , D. Sc. , Assistant of Civil , and S. B. , B.Sc. , Dcnnonstrator in ) , both of by ofessor O REYNOLDS , F.R. eceived July November 20 , 1902 .
1 .
troduction .
THE motion of water in pipes and channels has been the subject of frequent investigation , both from ] theoretical and the experimental side , and it is well known that while in some cases theory and expel.iment are in exact accord , yet in many others the experimental results differ widely from the calculated .
In some cases , while the theory holds for one set of conditions , it is found not to hold for conditions which at first do not appear to be fundamentally different .
A striking instance is that of the flow of a viscous liquid through a pipe of circular section , a case for which a strict mathematical solution can be obtained under certain assumed conditions of flow .
Experiment shows that the theory is vel.ified if the pipe is of capillary bore and the motion small , while if the pipe is large and the motion appreciable , there is a discrepancy between experiment and calculation .
The discrepancy is due to the assumption that the motion is stream-line , a condition of things which is true for tubes of capillary bore , but in general is not true for tubes of appreciable diameter unless the motion is below a certain limit , fixed by the size of the pipe and the physical characteristics of the liquid .
Above this limit , the motion is eddying and the hydrodynamical equations no longer apply .
The change from stream-line to eddy or sinuous motion was first studied by SBORNE R who showed that the determining factors in the case of a circular pipe depended on the dimensions of the pipe and the viscosity of the water .
His results are based partly on deductions from the equations of motion for a viscous ' ' An Experimental Investigation of the Circumstances which determine whether the Iotion of shall be Direct or Sinuous , of the Law of esistance in Channels 'Phil .
Trans 1883 .
333 .
23.3.03 fluid ; thus , if we take the general equations of motion for an incompressible fluid subject to no external forces , as of type and iminate the sures from the equations , we obtain accelerations in terms of diflerent types .
Thus , if we take the middle term , , \mdash ; , and for wlite , we get \mdash ; Now , since and have the dimension of a velocity divided by a length and the other term has dimension of the square of a velocity , the relative values of these two terms are to one another as to , where is a length , say the radius of the tube .
The equations do not show in what way motion depends upon this relation , but it was inferred that the eddying motion must depend on some definite relation between and , expressible in the form , where is some constant .
The experimental observations were of two kinds , the earlier depending on the device of introducing a colour band into a glass pipe and observin the velocity at which break-down of the stream-line motion occurred , and the later method depending upon the fact that motion is associated with resistance proportional to the velocity , while for eddy motion the resistance is proportional to a higher power of the velocity .
Both methods showed that the critical velocity at which motion changed to eddy motion varied directly as the viscosity , and inversely as the radius of the pipe .
Object of the Experiments .
In the experimental verification of the temperature effect upon the critical velocity a satisfactory agreement was obtained with the formula , but as the range of temperature was extremely limited , it was pointed out that " " it would be desirable to make experiments at higher temperature ; but there were difficulties about this , which caused me , at all events for the time , to defer the attempt It does not appear that such experiments have since been made , and although the difficulties were not estimated lightly , it seemed worth while to attempt experiments through a much larger range of temperature .
Scope of the Although it would be eminently satisfactory to make experiments throughout the whole range of temperature of water , yet the experimental difficulties of maintaining 'Phil .
Trans , 1895 , p. 131 .
'Phil .
Trans 1883 , p. 977 OF THE CRITICAL LOCITY OF WATEIi WITH a uniform temperature in the pipe increase in a much greater ratio than the crease of temperature beyond , say , C. , and there are other difficulties , due to convection and evaporation , which made it desirable to limit the investigation , at any rate for the time , to a range of to C. With these limits it was found that the decrease or increase of temperature along the pipe , when thickly lagged , was inconsiderable , and the correction to be applied was therefore small and not likely to cause an appreciable error .
In order to carry on experiments at a higher temperature , it would apparently be necessary to surround the experimental tube with a waterjacket maintained at the same temperature as the water in the tank , drop of temperature along the pipe would be so considerable as to seriously increase the chances of error .
Mefhod of Eaperiment .
The principle of the method is the same as iginally d by OSBORNE REYNOLDS , but the manner of carryin out the work differed somewhat in detail .
, The method of colour bands is unsuitable for water at a high temperature , as it is impossible to eliminate the effect of conduction and convection , and the water consequently never comes to rest ; moreover , experiments by this method lead to a different form of the criterion , , the maximum limit at which stream-line motion is possible , while experiments on the variation in resistance of pipes lead to the minimum criterion , , that at which eddies change to steady motion .
This latter method is more likely to be accurate , the maximum velocity of stream-line motion depends upon external causes , which influence it to a remarkable extent .
Experiments were made with the tank in the laboratory to discover how far streamline motion could be carried under favourable conditions ; the tank rests directly upon the ground , and after water at temperature of the room .
had been allowed to stand therein for two or three days , stream-line motion in pipes could be maintained at higher velocities than that given by the upper limit formula for the critical velocity , viz. : the units being metres and degrees centigrade , a result no doubt due to the complete vibration in the tank , which was founded on rock , and also the freedom of the water from sediment .
Moreover , it is easy to lower the critical velocity by subjecting the water to a disturbing cause ; thus fine matter in suspension in the water will lower the critical velocity .
Tapping the pipe or interposing therein a piece of wire gauze will also act likewise ; in fact , the point of break-down can be varied within wide limits according to the circumstances .
Whatever be the disturbing causes , however , if stream-line motion exists , the 'Phil .
Trans 1884 , p. 957 . .
CLEMENT ON TIIE yARIATION relation of slope to velocity is a perfectly definite one at a definite temperature for the flux , being expressed by the equation If write and , we obtain , and this relation between and , plotted logarithmically , is , at a definite temperature , a line inclined at to the axes .
Slightly ) critical velocity , it can be shown experimentally that no definite relation exists , but well above this point , where the motion is perfectly eddying , it can be shown imentally that the relation between and is a perfectly definite one at a temperature , and is expressed by some straight line inclined at an angle , where is a constant for any particular pipe .
It therefore appears that the minimum critical velocity is the intersection of the two branches of the logarithmic homologue ; and throughout this paper this point has been taken as the critical velocity for the temperature considered .
As the experiments below the critical velocity require apparatus for measuring pressures of extreme accuracy and limited range , while above the critical velocity the limit of accuracy is relatively less important and the range is large , it simplifies matters to take a series of runs at different temperatures below the critical velocity without any change , and afterwards to take runs aboye the critical velocity .
With this method the variation of temperature during a single series is small , and the correction to a standard temperature is generally negligible .
used in the ) The experimental tank , fig. 1 , is of cast-iron , 5 feet square in section and about 30 feet in height , its base resting upon the earth , so that the water in it is not easily disturbed by external causes .
It is provided with a steam heater for the inflowing LANB 's namics , ' p. 521 .
OF THE VELOCITy OF WATER WITH MPERATURE .
water , and there is a direct steam connection to the boiler room , so that steam can be lown directly into the tank .
About 8 feet above the base there is an opening , the middle of one side , through which the tube was inserted , its bell-mouth being placed at the centre ; and at suitable distances apart pressure chambers , , were formed and connected up to the -gauge E. The flow of water was controlled by a valve , , and on the prolongation of the pipe a three-way plug yalye , , was inserted , so that the water could run to waste through the pipe , or could be discharged , by the pipe , into the glass flask K. The handle of this tap was provided with a flexible brass plate , , in circuit with a , so that at the middle of its swing a circuit was completed by the contact of the brass strip with the pipe , and a record was obtained on the drum of the chronogra , ph .
This latter instrument was furnished with two pens , marking in opposite directions , one ticking seconds and the other operating at the beginning and end of each run .
This arrangement tends to prevent errors in reading .
The pressure chalnbers were of a special design and consisted of three separate pieces , the outer one ( A ) of which couples the parts and together , leaving a continuous opening , , which may be of any required width .
In the present case , the two sides the slit were separated by an interval not than inch , so as to prevent , as as possible , any interference with stream-like flow .
The part is recessed to form a pressure cha1llber , , connected to the gauge by an opening , .
The parts and are faced so that when drawn together by the coupling , they form a water-tight joint at and the ends of the pipe are screwed into corresponding recesses in and C. This form of chamber has several advantages .
The continuous opelling gives an accurate mean value of the pressure , and it can be faced without any burr ; moreover , it may be readily disconnected for inspection .
The pipe was of brass , without seam , and 6 feet in length between the pressure chambers ; its mean diameter was determined by first weighing empty and then full of mercury .
The mean diameter thus determined was inch .
The , surement of Pressure .
The accurate determination of the pressures at the given sections of the pipe is matter of considerable difficulty , especially at the very low differences of head required the accurate determination of the slope of pressure at velocities below the critical velocity .
At the higher pressures , -tube containing mercury was VOL. CCI.\mdash ; A. found to answer requirements .
Errors due to the inequalities of the ube were got rid of by measurements taken on both tubes , while a suitable correction was made temperature .
At the low velocities , considerable difficulties were experienced .
The difference of heads between the two sections at the lowest velocities was about inch of mercury , and as this must be read very accurately , it became a matter of such difficulty that a new gauge was made and filled with carbon bisulphide .
A number of trials were made , but it was found that the carbon bisulphide was very sluggish in action , and , unless a very considerable time was allowed between every two successive runs , its could not be relied upon .
Another and more serious defect was the shape of the falling meniscus , which rarely assumed its proper form , so as to afford a definite measurement .
This was due to adherence of the carbon bisulphide to the glass ; and in spite of repeated cleanings with different re-agents , no decided improvement was made and the gauge was abandoned .
A return was made to the mercury gauge , and the cathetometer was replaced by micrometer , which had been carefully calibrated beforehand .
These afforded much better results , but the observations were still irregular .
Finally , the solution of the difficulty was found by turning the -gauge upside down and imprisoning in its upper part a fixed column of air above the water in both limbs of the gauge .
At first sight this might not seem to be a good arrangement , slnce any small variation of temperature will affect the imprisoned volume of air considerably , but this affects both legs equally , and there is no error from this cause .
A possible source of error is the creeping of air from the pipe to the gauges .
This is extremely unlikely , as the air in this ca , se must first descend .
If any liberation of air occurred from the water , its effect in altering the gauge would only be momentary .
In practice , this gauge proved extremely sensitive and bhe readings could be repeated very accurately .
The cathetometer used in reading the heights of the liquid in the -tubes was of a standard pattern made by the Cambridge Scientific Instrument Company and capable of reading to of a millimetre .
Stream-line Flow .
The determination of the relation of slope to pressure , .
water in motion flowing through tubes of more than capillary size , is rendered somewhat difficult because of the smallness of the difference of pressure required to produce the flow .
The difference may be increased by using a long length of pipe , or by using apparatus of extreme accuracy .
The disposition of the permanent apparatus in the laboratory prevented the use of a pipe more than 6 feet in length between the pressure chambers ; and at first considerable difficulty was experienced in obtaining OF THE CRITICAL yELOCITy OF WATER WITH TEMPERATURE .
consistent results , but after many trials this was accomplished .
A disturbing cause , which could not be altogether avoided , was the rise or fall of the temperature of the water as it flowed along the pipe ; a fall if the temperature of the room was above the temperature of the water , and vice versed .
This was partly removed by covering the pipe with thick cotton-wool lagging overlaid with flannel , and in order to obtain a mean value of the temperature of the water in the pipe a long-stem thermometer was fixed in the tank and another was immersed in the outflowing water , and a mean value of these readings was taken as the true temperature in the pipe .
3 .
A plot of the variations obtained is shown in fig. 3 , in which the line AB gives the temperature of the outflow water , CD the temperature of the tank , and OF the corresponding temperature of the room .
The relation between the tank temperature and outflow temperature is shown to be practically a linear one , thereby warranting the correction .
In all , ten series of runs were made at temperatures covering the range , and the results obtained are recorded in the following table , and are shown on fig. 4 .
TABLE I. OF { CRITICAL VELOCITY OF WATER WITH RATURE .
TABLE I.\mdash ; continued .
In this table the obseryations are recorded in Columns 2 , 3 , 4 and 5 , and from Column .
and 4 the mean values of the velocity of the water in feet per second have been calculated , and the arithms of these quantities are given in Column 6 .
The observed differences of head given in Column 5have been reduced to feet of water , , to correspond , and the values of are given in Column 7 .
In most cases , owing to the large volume of water in the tank ( usually not less than 300 cubic feet ) , the temperature remained remarkably steady during the runs forming a series , and no correction for temperature was necessary , and none was made unless the temperature differed more than C. In some cases , howeyer , much greater variation was met with , especially at the higher temperatures , and correction was necessary , not only in this series , but in the second series when the motion was eddy or sinuous .
The correction factor to } ) applied may be obtained as follows:\mdash ; If we assume that in stream-line motion or sinuous motion the total resistance depends on powers of the pipe radius , the kinematic viscosity , the density and the velocity , we may write , with the usual notation , where radius of the pipe , coefficient of kinematic viscosity , density , mean velocity of water along the pipe , constant .
Dimensionally this equation becomes [ giving the relations and therefore For the case of stream-line motion , giving For the case of sinuous motion is greater than unity , and we may write the equation , where Taking logarithms , we get Differentiating with ) constant , we obtain Now , therefore , and approximately , therefore OF THE CRITICAL VELOCITY OF WATER WITH TEMPERATURE .
Hence For stream-line motion , and the values of and taken from formula are The value of for water is approximately 00048 , and hence the second term is never important , and the first term is only important where is large ( and the term is not small ) .
For stream-line motion the correction for a difference of 1o at C. is approximately028 , while at C. it is 017 .
The observations recorded in the table , corrected to a mean temperature , are in fig. 4 .
diagram shows the logarithmic homologues for stream-line motion in the pipe at ten different temperatures between C. and C. , and these are represented by a series of straight lines equally inclined to the axes of co-ordinates .
The mean temperatures to which the observations have been reduced are as follows:\mdash ; These lines were ] found to agree closely with the formula where is the radius of the pipe , and the pressures at the ends , the length or pipe , and is the coefficient of viscosity .
* Corrected Nov. 14 , 1902 , as pointed out by tho Referee .
Added Nov. 14 , 1902 . .
cit. ante ( p. 48 ) .
PROFESSOR .
G. COKER AND MR. R. B. CLEMENT LE ENT yARIATION OF This reement is shown clearly by fig. , where the of the homologues with the zero ordinate are plotted with reference to the } ) erature abscissa , and are compared with the intersections determined from the equation above , taking the value of according to POISEUILLE'S formula .
At temperatures between and C. the agreement is close , the values at and do not correspond very well , and there is a very fair agreement at the temperatures .
The dotted lines on fig. 4 are the logarithmic at the temperatures of and C. respectively , and these have been by aid of figs. 4 and 5 , in order to determine the intersections with the for eddy motion also plotted on fig. 4 , and which are referred to in the next of Slopc to ) locity for A second series of experiments was commenced with ] water eddy Jnotion to determine the relation between the loss of head and the yelocity at a number of temlJeratures within the range .
It was extremely difficult to control the tempel.ature , and so attempt }lade to obtain a series of runs with temperatures corresponding exactly to those obtained Fig. 5 .
for stream-line motion , nor was this necessary , as the logarithmic homologue for motion , corresponding to a similar one for eddy motion , was obtained by interpolation from figs. 4 and 5 .
The observations were made under precisely the same conditions as before , except that in the pressure mercury in contact with PROFLSSOli .
G. AND MR. .
B. MEANT ON THE VARIATION water used instead of water in contact with air and the gauge was in its normal position with the connecting below .
After some preliminary ials , seven complete series of runs were made , covering of temperature , and the observations made are recorded in Table Il .
, and the logarithmic plots are shown on fig. 4 .
As slight differences of temperature were now of much less importance , the plotted results lie much better upon the mean line , and enable the value of to be determined with considerable accuracy .
TABLE II .
seconds .
Time , dischargedof water , OF THE CRITICAL VELOCITY OF WATER WITH TEMPERATURE .
TABLE II.\mdash ; continued .
The slopes of the lines for the different temperatures are shown in the accompanying table , and their mean value is The different values obtained are no doubt due to temperature errors , and this view is confirmed when it is seen that the variations from the mean value are greater the further the temperature of the water is from the temperature of the laboratory .
As there seems no reason to pose that the value of varies with the the logarithmic humologues have been drawn at a mean inclination of in determining the critical velocity .
The corrections to be applied for differences of temperature are now much smaller , and for the mean value of is found to be 0076 for a difference of 1o from a temperature of C. , and the same difference at C. The observations for each series of runs in Table II .
have been plotted to a mean temperature like those for stream-line motion , the temperatures corresponding to the interpolated homologues for stream-line lnotion described above , Critical Vctocity .
It has been ) ointed out in an earlier section that no attempt was made to determine the velocity at which -line motion broke down , but that the intersections of the sets of lines above and below Cl.itical velocity were used to determine the lninimum critical velocity .
method of procedure amounts to the determination of the curve of intersection of two families of straight lines , whose positions are experiment lly determined , and it is clear that if the points of intersection lie upon some straight line in the logarithmic plot , the variation of the critical velocity must follow the viscosity of the water linearly , while , if they do not , the law cannot be a linear one .
Fioa ' .
4 shows the observations for stream-line flow , and the lines representing eddy motion are drawn thereon , and produced to meet the interpolated lines for Fig. 6 .
-line fiow ( shown dotted ) ; the points of nleeting are found to lie very approximately upon the straight line AB .
It is therefore apparent that these intersections vary as the viscosity , and they afford a verification of the formula .
This is brought out clearly by fig. 6 , in which the velocities so found are plotted directly with respect to temperature .
As will be seen , less weight is given to the OF THE CRITICAL VELOCITY OF WATER WITH TEMPERATURE .
observations in the neighbourhood of , for the results on streamline motion are probably less accurate , as was pointed out in a previous section .
The law of variation is found to be approximately which agrees closely with the variation in the viscosity of water , .
It may therefore be concluded that for small pipes , ovel the range of temperature examined , the critical velocity of water varies directly as the viscosity .
In conclusion the authors desire to thank Professor BovEY for placing the hydraulic laboratory of McGill University at their disposal , and also Dr. BARNES , gave much assistance during the gress of the work .

