Paper No.

6525

THEHYDRAULICRESISTANCE OF DRAINAGECONDUITS
by
Peter Ackers, MSc.( Eng.), A.M.I.C.E.
Principal Scientific Officer, Department of Scientific and Industrial Research,
Hydraulics Research Station
For written discussion

SYNOPSIS
A programme of tests on salt-glazed, spun precast concrete and pitch-fibre
pipes has now been completed at the Hydraulics Research Station: the results
are reported in the Paper, and show that the turbulent-transitional equation is
applicable, and that the eccentricity and spacing of the joints affect the overall
roughness. Data havebeenobtainedfrom the field indicating the jointing
standards ordinarily attained, and hence roughness values for new sewers are
suggested.Under part-full conditions, proportional velocities follow the
theoretical trend but aregenerally several units per centIOW, which is equivalent
to an apparentincrease in roughness for free-surface flow. Possible explanations
for this are considered.
Amplescope for further researchremains,particularly into the ageing
effects in sewers.

INTRODUCTION
THEtheoretical and experimental studies of the hydraulic resistance of sewers
and storm drains which are reported herein resulted largely from a request for
information fromthe Joint Committee on Rainfall and Run-off (of the Ministry
of Housing and LocalGovernment, and the Road ResearchBoard). An
intensive literature study was made of available equations; the rather complex
formula which was recommended as a result of these theoretical considerations
had to be expressed in simpler terms for it to meet the needs of the design
engineer, and so charts expressing the relationship of velocity and discharge to
conduit size and energy gradientfor a wide range of roughness were prepared on
the basis of the proposed equation.
2. The next stage of the problem was to supplement the available roughness
data forpipes and channels withnew experimental data onthe materials usedin
the construction ofsewers and storm drains; a programme of experimental
research was therefore put in hand at the Hydraulics Research Station to deter-
mine the discharge capacityof new salt-glazed pipes, precast-concrete pipesand
pitch-fibre pipes. The results of these laboratory tests and of some field tests
will be given, and it is hoped that additional data on old sewers will become
available from further field tests in a few years time.
307
Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 308 ACKERS ON THE HYDKAULlC RESISTANCE OF DRAINAGECONDUlTS

Hydraulic resistance
3. Manyempirical equations (references 1, 2, 3, 4, 5, 6, 7, 8) havebeen
proposed for the hydraulicdesignofpipes and channels, the basic Chew*
formula being one of the first.
v=C G f . . . . . . .
(1)
Exponential equations have generally been preferred, of the form
V = Crnzia . . . . . . . .
(2)
The most popular of these is the so called Manning equation

(3)
(Powellg has thrown some lighton the history101 11,1*.13* of this type of equation).
4. The most commonly used friction formula for sewers in this country is
Crimp and Brugess equation published in 1894 quite independently of the more
general two-thirds law equation. This was based on a re-analysis of earlier
data, including Bazins2 for part-fullsemi-circularchannels, and on original
experiments on a new brick channel: the equation was recommended for good
brickwork, butnodataon salt-glazedwarepipeswerepresented. In the
1930s the work of Karmanl4 and Prandtl15.16 led to an enlightened theory of
turbulent flow : when coupled with classic experiments by Stanton and Pannell17
on smoothpipes and by Nikuradse18on rough pipes, this momentum-exchange
theory of turbulence provided an almost complete accountof fluid resistance.
5. The turbulent flow pattern falls into three zones:smooth-turbulence,
when a laminar film at the boundary eliminates any roughness effect; rough-
turbulence, when resistance becomes independent of viscosity, the roughness
elements completelydisrupting the boundary film;and a transition between these
extremes. The plot of Nikuradses data for pipesroughened by aglued-on
layer of uniform sandis reproduced in most modern hydraulic reference works,
the variables being the following dimensionless parameters:
VD
Reynolds number,x = . . . . .
Friction factor, h = 2gDi/v2 . . . . . (4a)
Roughness ratio, Dlk
In fact, the transition zone betweensmooth and rough turbulence is complicated
becauseresistance is related not only to viscosity and roughness, but also
depends upon such factors as the shape, relative coverage and size distribution
of the roughening elements.
6. Colebrooklgconsidered the effectof roughness pattern on hydraulic
resistance, and, using data on commercial pipes20, he showed that most (but
not all)suchsurfacesfollowedsimilar transitions between the twobasic
turbulent laws (smooth and rough). He derived a transition equation
in conjunction with Professor White of Imperial College which has been in-
creasingly accepted for flow calculations in many countries21 over the last two
decades.
7. Consequently, the Author recommends this modern equation in preference
to purely empirical formulae: the reasons for so doing are set out in detail in a
Paper22 dealing more fully with the subject of fluid resistance than is possible
1 The references-are listed on pp. 334-336

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS 309
here. M o r r i ~ ~ 3 has
* ~ 4recentlysuggested further sub-divisionsofroughness
types, and if his theories are fully confirmed by experiment they will generalize
the estimation of friction losses even further.

The application of the Colebrook- Whiteequation to design

8. The transition equation for commercial surfaces in its general non-dimen-
sional form is:

If the dimensionless parameters are replaced by their engineering equivalents,

the following expression results,for pipeflow:

(The minus signs in equations (5) and (6)are cancelled by negative logarithms,
the functions in brackets being less than one.) It can be generalized25to apply
to non-circular sections by replacing D by 4m:

9. This transition equation has the advantage of representing the smooth

and rough zones as well as the intermediate conditions of turbulence, and as it
is applicable with any fluid it is practically universal. Unfortunately, its com-
plexity makes it unattractive for design purposes, although various attempts
havebeenmade at simplifying it*1*26327.2*. Thestudy at the Hydraulics
Research Station centred around the need to prepare charts on whichthe
scales were the engineering variables for discharge, velocity, conduit size and
gradient. The process by which these Wallingford charts29 were derived has
alreadybeendescribed in detail? it involvedre-expressing the Colebrook-
White equation interms of a new group of dimensionless flow parameters;
Velocity parameter, P = vk/v . . . .
. . . . (8)
Hydraulicmean depth parameter, M = m/k . . . (9)
Hydraulic gradient parameter, I = 2gik3/v2 . . . . (10)
Dischargeparameter, Q = 4 vqm/pkv . . . . . (1 1 )
These. have the advantage that, for any given class of construction (k-value),
and fluid (v-value), theyare in direct proportion to the four comparable engineer-
ing variables, V, m,i, and q. The following rational form of transition equation
results:
P = - 4 dmlog{+M+FE} 0.314

10. As it contains three variables only, this can be plotted as a family of

equal Z values on a chart with P and M as its axes. (Powell30 had previously
thought onthese lines, unknownto the writer). By tracing sectionsof the above
mentioned chart, and re-labelling the axes with values ofv, q, m , and i, a seriesof
simple design charts results.
11. It is as well to remember that some forms of roughness, e.g. corrugated
culverts, do not conform to the Colebrook-White transition equation, and the
approach suggested by Morris*3*24 may be more appropriate insome instances:

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 310 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS

experimentalevidence of this willbementionedlater.Nevertheless, the

Colebrook-White formula is applicable to almost all commercialpipelines
with adequate accuracy, and certainly with a much higher general confidence
than any other single equation.

Previous evidence on sewer roughness

12. Theequation which is generallyapplied to sewerdesign in Britain
is the Crimp and Brugess equation,
V = 124 m0-57iO.5 . . . . . . . (13)
which is equivalent to equation (4) with n = 0.012.
The two-thirds exponent was derived by analysing earlier experimental data,
which these Authors supplemented by experiments on a new semi-circular brick
culvert. Their experimentsgaveacoefficient of 143, and similartests on a
brick sewer then28 years old gave a figureof 122. Contrary to popular opinion,
Crimp and Bruges did not experiment on salt-glazed pipes, and the arbitrary
application of their equation, which was conservative even for new brickwork,
to all forms of sewer construction has little justification.
13. Scobeys7 paper contained extensive data .on concrete pipelines, which
have since been re-analysed withinthe framework of the transition equation, so
that k-values can be quoted for the four classes of concrete suggested by him, as
in Table 1 below. It will be appreciated that in quoting k-values for a given
surface, the roughness is being expressed as the diameter of sand which, as a
uniform coating, would givethe same resistanceunder rough-turbulent flow. The
equivalent sand-roughness bears arelationship to the physical dimensionsof the
roughnessprojections, so thatavisualexaminationof asurfacewithtexturalrough-
ness might give a guide to its k-value. In Stricklers31 investigations ofnatural
channels, it was found thatk = dq0,i.e., the size exceeded by 10% of the exposed
bed material: in other engineering fields, one would therefore expect the larger
roughness elements to have a predominating influence. It can be shown22 that
4m
within the range 7 < < 130theMamingequationiswithin2% of thelogarithmic
form of rough-turbulent equation (rough turbulence is established fully when
vk exceeds 0.01 ft2/secwithwater at ordinary temperatures). Also n values
can be converted to equivalent sand roughness through:
1
n =- 32 k0.167 (ft
sec
units) . . , . . (14)
This permits experimental results in terms of the Manningcoefficient to be
converted to k-values (provided they relate to the rough-turbulent zone in the
appropriate relative roughness range),as has been done with Hortons32 recom-
mendations for brickwork in Table 1 . If the Crimp and Bruges data on brick-
built channels are compared;
New brickwork, n = 0-0096, k = 09008 ft
Old brickwork, n = 0.0122, k = 0.0036 ft
Hortons recommendations appear to be conservative.
14. The data available before the recent research on salt-glazed pipes were
no lessscanty than those on brickwork.Yarnell and Woodward33tested
4-in.to 12-in.-dia.butt-jointed drain pipes for the U.S. Department of Agriculture
In 1924, again in America,
in 1920, the larger sizes being in 2-ft long glazed units.

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS 3 11
Wilcox34 experimented with 8-in.-dia. salt-glazed and precast concrete pipes with
spigot and socket joints, under full and part-full conditions at gradients from 1
in 200 to 1 in 15. Although there is considerable scatter at proportional depths
under a half, average k-valueshave been estimated by re-analysing the American
data: Wilcox's socket-jointed glazed pipes had a roughness of 0.0010 ft when
full (n approx 0.010), whilst the butt-jointedpipesin the other seriesgave
k-0.0017 ft (n approx 0.01 1). In both cases the quality of pipe-layingis
uncertain; we have no record of the accuracy of alignment at thejoints, but the
photographs in the original report imply average to poor workmanship. More
recently Straub and Morris35336 havetestedconcretepipes and corrugated
metal culverts on a large scale. The latter type of roughness is a noteworthy
example of non-conformity with the Colebrook-White transition, and it would
be misleading to quote k-values in the present context. However, the Straub
and Morris data on 18-in., 24-in., and 36-in-dia. precast concrete pipelines have
been re-analysed to give the roughness included in Table 1.
l .-PREVIOUSEVIDENCE ON SEWER ROUGHNESS, CONVERTED TO
TABLE
EQWALENT SAND-ROUGHNESS

Authority Classification k-value : feet

Scobey Concrete Class 4. Monolithic con-
structionagainstoiled steel forms
with no surface irregularities, precast
pipe lines with no shoulders or de-
pressionsat the joints .
Scobey ConcreteClass 3. Monolithic 'coh
struction against steel forms, wet-mix
or spun precast pipes, or with cement
orasphaltcoating . 04013
Scobey Concrete
Class2.
Monoiithiccon:
struction against rough forms, rough-
textured precast pipes,or cement-gun
surface . . . . . . . . . om42
scobey ConcreteClass 1. Precastpipeswith
mortar squeeze at the joints . . . 0.017
Straub and Morris* Precast concrete pipes 18 in. dia. . . 0.0002
f ,, 24 dia.
in. . . O.OOO4
,, 36 in.dia. . . 04005
Horton Glikd briicwork . . . . . . 0.002 to 0.015
Horton Brickwork in cementmortar,brick
sewers . . . . . . . . 0.004 to 0.030
Crimp and Bruges New brickwork . . . . . 0.0008
28-year-old-smooth brickwork . 0.0036
Yarnell and Woodward Butt-jointedsalt-glazedpipes (l2 in:
dia.) . OW17
Wilcox Socket-jointed sait-giazeh pipes (8 in:
dia.) . . . . . . . . . 0~0010

* Note: the jointsof the %in. and 36-in-dia. pipes were pointed internally, those of
the 18-in. pipeline were not.
Laboratory installation
15. The most serious deficiency in this table is roughness data for typical
salt-glazedsewers, and consequently an experimentalinvestigation into such
pipes was included in the programme of the Hydraulics Research Station. It

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 312 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGECONDUITS

was considered that joint eccentricity would prove to be an important source

of resistance,so the test facilitieswere designed to allow the alignment of neigh-
bouring pipes to be adjusted. Although salt-glazed pipes were the major item
on the programme, the opportunity was taken of testing subsequently both
precast' concrete pipes with Cornelius-type rubber ring joints, and pitch-fibre
pipes. The laboratory installation is shown in Fig. 1.
16. For the first phase of the investigation, 180 ft of 12-in.-dia. salt-glazed
pipe to B.S.65: 1952 were obtained commercially (without special selection),
half in 3-ft long units and the remainder in 2-ft lengths. Each pipe was measured
before use. Several pipes were rejected at this stage as their dimensions were
appreciably outside the limits of the British Standard: the accepted pipes had
the average sizes, and ranges of variation given in Table 2. This shows that
commercial salt-glazedpipesare far from uniform. Moreover, thereis a tendency
for the spigot end of each pipe to be more oval than its socket end (presumably
a result of the manufacturing process).

TABLE
Z.-sALT-GLAZED PIPE DIMENSIONS

3-ft pipes 2-ft pipes

Dimension
inches inches
Overallaveragepipediameter . . . . . 12.08 12.12
Major-axisdiameter . . . . . . . . 11.90 to 12.23 11.94 to 12.31
Minor-axis diameter . . . . . . . . 11.83 to 12.12 11.91 to 12.19
Degree of ovality, i.e. major axis minus minor
axis . . . . . . . . . . Zero to 0.39 Zero to 038
Overall average pipe length :
before laying . . . . . . . . . 36.31 25.00
laid as . . . . . . . . . 36.43 25.15
Maximum length . . . . . . . . . 36.00 to 36.63 24.37 to 25.31
Minimum length . . . . . . 35.63 to 36.37 24.31 to 25.13
Degreeofskewness, i.e. max. length m'inus
min. length . . . . . . . . . . Zero to 0.37 Zero to 0.50

17. With these detailed pipe measurements, it was possible to calculate the
gap required between socketand spigot at four positions (top, bottom, andeach
side), in order to centre each pipe with its neighbour as accurately as possible.
Initially, the pipelinewaslaidwithnearly-true joints, although because of
differences in dimensions small lips remained at most joints. It was set on brick
piers and accuratelyaligned and levelledwithfoldingwedges to an invert
gradient of 1 in 100. Small timber wedges were used to adjust the joints, the
spigot/socket gap being measured by specially-made brasstaper gauges. Sealing
wasby means of bicycle inner tubes, inflated within the sockets; these were
extremely effective at heads up to several feet of water, and greatly facilitated
subsequent adjustments to the joints.

EXPERIMENTAL PROGRAMME
18. The first series of tests was made on the salt-glazed pipes with joints set
as accurately as possible, the pipe-full runs covering hydraulic gradients from

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS 3 13

3
f

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 314 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS

0.001 to 0.020. Similarly, data for full and part-full flowwere obtained after
imposing eccentricityat the joints, so covering the following conditions inturn :
Near-perfect joints
0.4 in. step-up at each joint
0.3 in. step-up at each joint
0.2 in. step-up at each joint
0.35 in. step-down at each joint
025 in. alignment error at each joint
Random steps, up and down, at each joint, with average step of
0.193 in.
Joint eccentricities of random amount and in random direction,
with average of 0.194 in.
19. For the next phaseof the investigation, 120 ft of 1Zin.-dia. spun-concrete
pipes in 6-ft long units were installed at a gradient of 1 in 100 in place of the salt-
glazedpipestestedpreviously. The pipesweregenerally in accordancewith
B.S.55:1955, the internal finishbeing a good-qualityfine-texturedcement
surface. Joints were of the Cornelius type, i.e. spigot and socket joints sealed
with a solid rubber ring. These automatically ensure accurate centring of each
pipe with its neighbours, so the pipeline tested had joints which were virtually
free from lips.Moreover, it wasaligned and levelled to within0.0625 in.,
representing the best attainable quality of workmanship.
20. After completing tests on these concrete pipes, 96 ft of pitch-fibre pipes
were laid. These were of 6 in. dia., the largest size available commercially in
Britain, and they were in units 8 ft long. The quality of pitch-fibre pipes and
their joints is controlled under B.S.2760: 1956; the pipes tested had a glassy
internal surface of pitch, and the machined taper joints provided near-perfect
mating between the pipe units.

TABLE
3.-cONCRETE AND PITCH-FIBRE PIPE DIMENSIONS

Concrete pipes: Pitch-fibre pipes:

Dimension
1 inches
inches
Overall average internal diameter . . . . 11.96 6.15
Major-axis diameter . . . . . . . . 11.75 to 12.11 6.1 1 to 6.22
Minor-axisdiameter . . . . . . . . 11.74 to 12.11 6.08 to 6.17
Degree of ovality, i.e. major axis minus minor
axis . . . . . . . . . . . . Zero to 0.11 Zero to 0.13
Length of units . . . . . . . . . 72.0 96.6

21. The dimensions of the pipes were checked before laying, and were as
shown in Table 3. The objects of these laboratory tests were as follows:
To determine the effective roughnessof salt-glazed, concrete, and pitch-
fibre pipelines when flowing full.
To investigate the effect of the spacing and accuracy of alignment of
their joints.
(3) To determine their effective roughness under part-full conditions, and
the relation between proportional discharge and proportional depth.

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS 315
Analysis of the experimental data
22. The primary data for pipe-full conditions were analysed withinthe frame
work of the Colebrook-White equation, by reducing the measured discharge,
hydraulic gradient and water temperature for each run to friction factor and
Reynolds number (q/A= V , h/L= i, 2gDi/vZ= h, vD/v=K where the kinematic
viscosity is read from a standard graphagainsttemperature). X wasthen
plotted against R for each seriesin the range l05 <K< 8 X 105. By calculation
from equation 6, a k-value was deduced from each run and a weighted average
taken for the series, treating the 3-ft long and 2-ft long pipes separately where
appropriate. These mean k-values were used to calculate the transition curve
which best fitted theplotted data for each series of tests.
23. Under part-fullconditions, the depthsmeasured at the piezometer
tappings were averaged (again keeping 3-ft and 2-ft pipes separate, and omitting
the upstream tappings at which uniform conditions had not been established),
and expressed as a proportion of the full diameter. Similarly, discharge was
expressed as a fraction of the pipe-full discharge obtained with the hydraulic
gradientparallel to the invert: this,whendivided by the proportional area
corresponding to the average depth, yielded the proportional velocity. Hence
q, and v, could be plotted against D, for each series.
24. In reducing these part-full data to equivalent roughnesses, equation (7)
was used in place of equation (6): knowing discharge, depth of flow and water
temperature for each run, the only unknown is the roughness k , which couldthus
be evaluated.
25. The aboveprocedurewasused for eachseries of tests, although the
Reynoldsnumberscoveredwith the smallerdiameterpitch-fibrepipes were
naturally less (4 X l04 < < 4 X 105). It had been expected that the salt-glazed
pipes wouldoperate in the fully-developed rough-turbulentzone, but they actually
proved smoother than anticipated so that the range which could be covered by
the available pumping capacity lay entirelyin the transition zone.

Experimental results for salt-glazed ware

26. The distinction between the Manning type of exponential equation and
the transition law is obvious when plotted in the form h against X. The former
appears as a series of linesof constant h-value, their ordinates depending on the
value of ,o.,
n
,whilst the transition law producesa family of curvesfor different
relative roughness values.
27.On plotting the experimental data for fullpipes it wasimmediately
apparent that they did not conform with the Manning equation, but were in
closeragreementwiththeform of the Colebrook-White transition. Fig. 2
gives the results for the pipeline of 3-ft long units,for series A, B, and H, and Fig.
3 gives the corresponding results for the 2-ft long pipes. In both diagrams, the
best-fit transition curves have been added, and the lines of constant h-value for
Mannings n of 0.008, 0.009,0.011, and 0.012 (the latter being equivalentto the
Crimp and Bruges equation) are shown for comparison, for this 1 ft-dia. pipe.
The data for the other series of tests have been omitted for the sake of clarity,
but show a similar trend.
28. The meanroughnessvaluesobtained for pipe-fullconditionswith
velocities in the range 1 to 10 ftlsec are given in Table 4.
29. Thus, withbadly-made joints, the effectiveroughness of the pipein

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 316 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS

2-ft units is distinctlygreater than that in 3-ftunits,confirming that joint

spacing has an effect on the hydraulic performance of a pipeline. However,
with near-perfectjoints, the 3-ft-long pipesare very similarto the 2-ft-long ones.
Their /c-valuesof 090013 ft andOW014 ft respectively are comparable with the
usually recommended values for uncoated steel and spun-bitumen-lined pipes,
so that salt-glazed pipes, when carefully laid
and new, are considerably smoother
than the previous design methods assumed: this is clear from the Mannings n
values on the right of Figs 2 and 3. It will also be noted that the American
data on8-in. pipes mentioned earlier (Wilcox34, Table 1) gave a higher roughness
figure.
I

v1
W
0
p.
0.030
3
0.012
.- 0.025 U
21: 0.01 I &
-
N
,, EI.
5 0.020 0
0.010 Y
CL

e
U
r:
F
5
z 0015 0.009 2z
Q+
<
U X
E 0.008 +
t
-I
0.010
10 z X 10 3 X 10 4 x 10 5.x l0 6 X 10 E x 10 10 23
REYNOLDS NUMBER: 8 - $ 2
FIG.2.-FRICTION DATA FOR SALT-GLAZED PIPES IN 3-FT-LONG UNITS

TABLE
4.-MEAN ROUGHNESS VALUES, k, FOR PIPE-FULL
CONDITIONS (SALT-GLAZED)
~~

Series Average lip atjoints: Roughness k in ten-thousandths of a foot

feet
3-ft-long pipes 2-ft-long pipes
-
A 0.007* 1-3 1.4
B 0.033 4.1 5.4
C 0.025 3.1 3.3
D 0.017 1.9 2.3
E 0.029 3.2 4.0
F 0.021 2.3
G 0016
H 0016
~~ ~

* The residual lip at invert and soffit with joints centredas accurately as possible.
30. The experiments under part-full conditions have been expressed in terms
of proportional velocity and discharge: Fig.4 for series A is typical of the results
for salt-glazed pipes, and the experiments confirm the theoretical curves for V,
and q p against D , based on the representation of the size of cross-section by
hydraulic mean depth, as in equation (7).

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGECONDUITS 317

0,030
-
0
4
0 012 $
q;"""
N

-< 0-020
0-01I - 2Y
OL

0
0010 8c
54 c"
;0015 0.009 2
Z
0 z
5h 0-008 :
5
0-0I 0 2
10' 2 x 10' 3 X 10' 4 x 10' 5 X 10' 6 X 10' 8 x 10' 10' g
2
REYNOLDS N U M E E R , R = g B
FIG. Q.-FRICTION DATA FORSALT-GLAZEDPIPES IN 2-m-LONG UNITS

PROPORTIONAL DISCHARGE q p

PROPORTIONALVELOCITY Vp

FIG. 4.-PROPORTIONAL VELOCITY AND DISCHARGE FOR PART-FULL FLOW IN SALT-GLAZED

PIPES

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 3 18 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGECONDUITS

31. Because the discharge of a circular section conduit at 0.8 proportional

depth equals the pipe-full discharge,it generally proved impracticableto conduct
tests at depth exceeding 0.80. Moreover, at depthsbelow 0.20 the analysis
becomes very sensitive to errors in depth measurements, and the k-values at these
low discharges are unreliable. The average roughnesses obtained under part-
full conditions are listed in Table 5, for D, > 0.2.

TABLEMEAN ROUGHNESS VALUES, k, FOR PART-FULL

COND~~IONS(SALT-GLAZED)

Series 1 Average
lip
joints:
at Roughness, k in ten-thousandths of a foot
feet
l 3-ft-long pipes ~ z-ft-long pipes
A 0407*
B 0.033
C 0.025
D 0.017
E 0.029
F 0.021 1.9
G 0.016 1.o
H 0.016 2.0

* The residuallip at invert and soffitwith joints centred as accurately as possible.

32. Althoughthere is no great change of the meanroughnessoverthis
range of proportional depths, there is a slight tendency for part-full k-values to
be higher than their pipe-full values, except when the joint eccentricity consists
of steps-down at each joint (Series E), as might be expected. In practice, the
pipe-fullroughnessvaluesmaybeused for hydraulic calculations at partial
depths with reasonable accuracy.

Experimental results for concrete pipes

33. Theadvantage of the Colebrook-White equation over the Manning
formula was againdemonstratedbytheseexperiments,asshown in Fig. 5,
v)

L
h
0.030
5
0.0120
S

5Ny 0 2 , 0.01 I U2-

5 0.020 d
0.010 E
4
9
Y
0.015 0.009
0 z
45 0.008
2
4
E
z
,!w Io*
YI
-J
OOlOl
10' 3x'105 rlx'los sX/loT .5!105 ,!,os 2
3
REYNOLDSN U M B E R : R = ~
2
R G , 5.--FRIWON DATA FOR SPUN PRECAST CONCRETE
PIPES

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRALNAOE CONDUITS 319
The mean roughness value proved to be 0.00013 ft, i.e. slightly smoother than
the American data given in Table 1 (Straub and Morris) and equivalent to the
bestof the salt-glazedseries. Our pipes,with their rubber-ring joints auto-
matically ensuring accurate alignment, probably had less joint resistance than
their American counterparts; these self-centring joints made it unnecessary to
investigate a range of pipe-laying standards as was done for salt-glazed ware.
34. Part-bore conditions are plotted in Fig. 6. Velocities and discharges

PROPORTIONAL DISCHARGE pp

PROPORTIONAL VELOCIT? Vp

FIG. 6.-hOPORTIONAL VELQUTY AND DISCHARGE FOR PART-FULL FLOW IN CONCRETE

PIPES

in the concrete pipeline tended to be several per cent lower than the values
predicted from the pipe-full roughness, i.e. there is an apparent increase in
roughness under part-full conditions from k=0.00013 ft to k=OW045 ft. The
theoretical curves in Fig. 6 based on the latter value are close to the experimental
POiQb

Experimental results for pitch-fibre pipes

35. It was anticipated that the glassy interior surface of the pitch-fibre pipes
would resultin a very low roughness;in fact, when the datawere compared with
the Colebrook-White equation, negative k-values resulted, implying the impos-
sible condition of smoother-than-smooth pipes. The capacity of these
pipes is about 2% greater than the predicted value with k=O. It is conceivable

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 320 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGECONDUITS

that a systematic experimental error occurred whichwouldexplain the dis-

crepancy, but it is more probable that theusually accepted valueof the constant
in the basic equation is not exact. Putting k=O in equation (5), we obtain for
the smooth-turbulent equation
1
- = 210gK dX-0.8 . . . . . . (15)
dX
The data forpitch-fibre pipes is in better agreement with
1
z/x = 2-04 l o g x d1-0.8 . . . . . (16)
This is still remarkably similar to thebasic solution for smooth pipes derivable
from the theory of turbulence (reference 16)
1
- = 2.03 IOgRdx-1.0 . . .
dX
. (17) .
These experimental results have been plotted in Fig. 7, to which equations (15)
-VI

4 X 10 5 X 10 6 x 10 8 X1010 2 X 10 3 X 10 4 X 10 3
REYNOLDS NUMBER:R= % 0
FIG. 7.-FRICTION DATA FOR PITCH-FIBRE PIPES

and (16) have been added for comparison. Fig. 8 gives the part-full results:
again, there is an apparent increase in roughness at partial depths, as with the
previous materials, yielding a mean k-value of O.OOO11 ft. The experimental
data follow the theoretical proportionality functions closely when compared with
a full pipe with k = O.OOO11 ft, but proportional velocities and discharges are
about 5% lower than the smoother-than-smooth equation would predict.

The efect of joint eccentricity and spacing on roughness

36. The linear roughness measure, k, originated from Nikuradses18 experi-
ments on pipes roughened withsand grains of uniform size, the size varying from
test to test. These grains were juxtaposed when embedded in the fixing glue, so
that the ratio of roughness height to pitch was not one of the variations investi-
gated by Nikuradse. The Colebrook-White transition equation, though using a
similar linear roughness measure,is particularly applicable to pipes withisolated
protuberances. An alteration in the pitch of these separate roughness elements,
i.e. an increase in the number per unit pipe length, would be expected to give a

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 Rti. 13-1NTERNAL VIEW OF SALT-GLAZEDPIPELINE SHOWINti UNUSUALOVALITY

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 FIG.IGINTERNAL VIEW OF SALT-GLAZEDPIPELINESHOWINGYARNHANGING
FROM ONEJOINT

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS 321
changed equivalent k-value: the relative roughness of a pipe would be afunction
of e[D, e/py and e/pTyas well as of the shape of the elements, where:
e = roughness height
e= peripheral roughness dimension (summation)
p =wetted perimeter
p= longitudinal pitch of elements
Provided the roughness elements are sufficiently isolated for no sheltering
by those upstream to occur, one mightexpect the followingsimplescalar
relationship to hold:
k e e e
Relative roughness - - X - X 7 X shape factor
D - D P P
projected area of protuberances height
i.e. k2: X shapefactor . (18)
pitch perimeter

PROPORTIONAL DISCHARGE pp

PROPORTIONAL VELOCITY Yp
FIG. PROPORTIONAL VELOCITYAND DISCHARGE FOR PART-FLTLLFLOW IN PITCH-FIBRE
PIPES

Note: qp shown thus 0, and V, shown thus A, are expressed as proportions of pipe-
full flow estimated from the Colebrook-White equation withk=00001 ft.
qp shown thus X and V, shown thus - .- are expressed as proportionsof pipe-full
flow estimated from equation(16). - = 2.04 IogKdX -0.8, on which the theoretical
z/A
proportionality curves are based.
22
Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 322 ACKERS ONTHE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS

37. In the particular case of eccentric joint roughness under consideration,

this anticipated result can be expressed as

where e is the joint eccentricity and p is the joint spacing.

38. The experiments on salt-glazedpipes, running full,haveprovidedthe
data in Fig. 9, where k is compared with e2/p. The method of least squares
gives the following equation:
k = 0.00012f0.75 e2/p (ft units) . . (19) . .
oooo6

v0005

0 0004

E
*
g 00003
P
c)

S 00002

oooo1

-0

JOINTECCENTRICIN ANDSPACING PARAMETER 5:FEET

FIG.CORRELATION BETWEEN JOINT ECCENTRICITY, SPACING, AND ROUGHNESS

The fmt component, OW012ft, may be regarded as the basic resistance of a

e2
salt-glazed surface, with the second component,0.75 7 ft, being the contribution
P
of joint eccentricity to roughness. This empirical relationship should be applied
only to eccentrically jointed pipelines of a similar scale to the experimental
one:it is not ageneralequivalent-roughness equation. The only other
investigation known to the Author which could be examined in support of the
hypothesis embodiedin equation (18) was Schichtlings37 work on several forms
of roughness, the pitch of the elements being one of the variables. He dealt
with spheres, hemispheres, cones, short walls and long walls, applying these
roughnesses to one face of a rectangularconduit anddeducing their roughnesses
from velocity traverses taken under rough-turbulent
- conditions. The k-values
projected area -height
.pitch
quoted by Schichtling have
been plotted in Fig.
10
against
The mean lines of unit slope corresponding to equation (18) fit the experiments
reasonablywell: the graph for spheresshows the shelteringeffectwhen e/p

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS 323
exceeds 0-4. Although this information providessomeconfirmation of the
form of equation (1 8 ) for isolated roughness elements,
the proportionality factors
for Schichtlings formsof roughness seem highin comparison with the eccentric
joint data: however, the latter form of roughness may well be that much less
effective as a turbulence stimulation than the former.
Further data on salt-glazedpipes
39. With the help of the Ministry of Works, some field experiments38on the
head losses at manholes were carried out recently on a new 6-in.-dia. drainage
system and, although not specifically designedfor the purpose, these experiments

I Short angles V7
X Long angles I10

G 001 0.0 I 0.1

PROJECTED AREA
PERIMETER PITCH

FIG. ~O.-VARIATIONOF EQUIVALENT ROUGHNESS WITH HEIGHT AND PITCH OF ELEMENTS

(AFTER SCHLICHTING)

provided a rare opportunity for the deduction of roughness coefficients. The

site was at Bramshill Park, Hampshire, where a pond at the head of the system
provided an ample water supplyfor several days work, and the duplicate tanks
provided at the new treatment -workscould be used for volumetric discharge
measurements. These
experiments
havebeen
described
elsewhere38: the
equivalent k-valuesfor this 6-in. salt-glazed sewerin 3-ft units laid at a gradient
of about 1 in 60 under ordinary contract conditions were as follows:
Pipe running full, mean k = 0.00013 ft (n N _ 0.009)
Pipe part full, mean k = 0.00076ft (n N 0.010)

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 324 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS

These field tests were not carried out under such regulated conditions as the
laboratory experiments previously described, and the sensitivity of the equiva-
lent roughness values to minor discrepancies in head loss, discharge, and depth
measurements might explain the difference between the Bramshill data and our
laboratory tests: the roughness values for the 6-in. pipe in the field are lower
when full, and higher when part-full, than for the 12-in. laboratory line.
40. To provide further correlation between the laboratory work and ordinary
construction standards, aspecialcamerawasconstructed for taking photo-
graphs inside 12-in.-dia. pipelines. Illumination was by a ring electronic flash
tube encircling the camera lens; the camera was an automatic Shackman
taking 35 mm film; the flash, shutter, and film transport were operated from a
remote control lead; the whole arrangement was mounted in a wheeled trolley
just under I2 in. in diameter and some 30 in. long. With the co-operation of a
City Authority, photographsof the joints of four lengths of surface water drain
were taken andthe misalignment at the joints was measured,the data so extracted
being listed in Table 6.

TABLE
6.-JOINT CONDITIONS OBTAINED IN THE FIELD

Sewer Sewer Sewer Sewer

A C D
Diameter,
inches
Lengthofpipeunits,
. . . .
feet . .
12
3
1 1; 12
3
12
3
Gradient . . . . .
Datelaying
of . . . .
..
. . 1 in200
Jan. 1960
1 in 200
July 1959
Pipe quaIity . . . seconds seconds seconds seconds
Number of joints photographed 37 23 8 39
Maximummisalignment,feet . O@%O 0.058 0.051 0048
Averagemisalignment,feet . 0,027 0037 0030 0.024
Root-mean-squaremisalign-
ment,feet . 0.027 0.039 0.031 0027
Predicted k-value inew) feet 0.0003 0.0005 O.OOO4 0.0003
EquivalentMannings II when
just full (new) . . . . . 0.0095 0.0101 0.0097 0.0095

41. Included in Table 6 are the k-values for sewerspredictedfrom

equation (19), but itmust be rememberedthat these would represent conditions
when new. The carrying capacity may reduce with age, especially if the uneven
joints attract local accumulationsof debris or detritus, but the increase of k with
age cannot bepredicted from presentknowledge.Twotypical photographs
(between pp. 320 and 321) are given in Figs 13 and 14, illustrating faults such
as an exceptional case of ovality and a festoon of tarred yam hanging from
one joint. These were second quality pipes, but even the nominally British
Standard pipes obtainedforourlaboratory investigationincludedmany
examples of oval spigot ends, the worst of which were rejected.

Suggested values for the roughness of new sewers

42. If the average joint eccentricity and the unit pipe length are known for a
section of new salt-glazed sewer, the anticipated roughness is given by equation
(19). Even with the careful control of joint alignment in our laboratory tests,

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS 325
the average lip could not be reduced below 0 . 0 0 7 ft: it is most unlikely that the
highest standard of workmanship in the field would better e= 0.010 ft, from which
equation (19) gives the most optimistic design value of k = 0.00015 ft for new
12-in. pipes in 3-ft lengths. On the other hand, the worst conceivable standard
of pipe-laying might produce joint eccentricities equallingthe socket clearance,
i.e. with the outer surface of the spigot touching the inner surface of the socket.
Using the joint dimensions specified in Table 1 of B.S.65 :1952, the k-values for
new pipes badly laid cannot then exceed the figures in Table 7.

TABLE FOR THE POOREST

~ . - ~ E D I C T E D ROUGHNESS VALUES
STANDARD OF PIPE-LAYING
(New salt-glazed pipes to B.S.65: 1952)

Pipe Assumed lip at joints Predicted k-value:feet

diameter feet
inches Pipes in 3-ft units Pipes in 2-ft units
3 0026 0.0003 00004
6 0.036 O.OOO4 04006
9 0.042 OM06 O~OoO8
12 0052 0.0008 0.001 1
18 0.052 0.0008 0.0011
24 0.062 0~0011 OW15
30 0.062 0.0011 0001 5
36 0-083 0.0018 0.0026

43. Now the field measurements of joint eccentricity led to estimated rough-
ness of between 0.0003 ft and 0.0005 ft (12-in. dia.), and the Bramshill tests
gave k=0.00013 ft (new 6-in. pipes). Thus, the available evidence leads to the
following recommendations on the values of k for clean new salt-glazed pipes
flowing full:
Allsizesof salt-glazed pipes withverygoodworkmanship . . 0.0002 ft
Normal workmanship:
Less than 24-in. dia. in 3-ftunits} . . . . . . . OQ005 ft
Less than 12-in. dia. in 2-ft units
24-in. dia. and over in 3-ft units
}
12-in. dia. and over in 2-ft units . . . . . . . . 0~0010ft
Poor workmanship :
Less than 24-in. dia. in 3-ft units
Less than 12-in. dia. in 2-ft units
. . . . . . . 0.001Oft
24-in. dia. and over in 3-ft units} . . . . . . . . OG020 ft
12-in. dia. and over in 2-ft units
44. Similarly, for new concretepipesflowingfull, the following
k-values are recommended :
All sizes of precasr concrete pipe, with very smooth internal
surface and good workmanship, such as is attainable with
accurate self-centring joints . . . . . . . . . . OtK@2ft

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 326 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS
Normal workmanship, with spigot and socket joints (eGO.05
. .
ft, pipe length 2 6 ft), or ogee joints pointed oninside. 0~0005ft
Poor workmanship (pipe lengtha6 ft) or somewhat rough-
texturedpipes . . . . . . . . . . . . . 0.0010 ft
New pitch-fibre pipesare virtually smooth, i.e. k =0, but the chart given in refer-
ence 29 for k = O~oooO1ft will give a closeapproximation to the original capacity
of this class of pipe.
45. It will be appreciated that the above recommendations make no allowance
for the effects of branch connexions on the overall roughness,nor is any account
taken of the uncertain effects of ageing. Self-cleansing sewers will obviously
show less deterioration than pipes at slack gradients, and initial differences in
the surface textures of the available materials might be reduced as sliming takes
place. The effects of branch junctions are equally conjectural, but the writers
guess is that well-spaced connexionsinto a main sewer might increase the overall
k-value by between O.OOO1 ft and 04005ft.
46. Fig.11 has been prepared to showhow the capacity ofnewsewers
calculated from the transition equation using these new roughness data exceeds
estimates using the Crimp and Bruges equation. Velocities for full pipes are
plotted against diameter for three gradients (1 in 10, 1 in 100, and 1 in 1,OOO);
the Colebrook-White equation has been used with k-values of 0.0002 ft and
0.0010 ft, and Mannings n-values of 0.010 and 0012 are shown, the latter being
equivalent to the Crimp and Bruges recommendation.
Flow under part-full conditions
47. Having designed a gravity pipeline to carry a specified discharge when
just full, the depth offlow for some lesser discharge is often required: such
problems are readily solved if charts are available giving proportional velocity
and discharge in terms of proportional depth. The Manning equation is adapted
thus to part-full pipes by putting
VP = VO m P0.67 . . . . . . . (20)
where the suffix p denotes values as proportions of pipe-full, and the suffix o
denotes full-bore conditions. Unfortunately the transition equation does not
lend itself to such simplification, although the proportional relationships for
the rough-turbulentandsmooth-turbulent zones arenot undulycomplex
(reference 29, p. 5), and by analogy give an approximation for the transition
zone :

{-+
0-
k 1,100Dio433
}-(ft units, water at 15C) .
NOWm p is a function of flow depth, but instead of single curves for vP and qp
versus dp,under turbulent conditions we really have a family of curves,
the third
parameter being 0 which itself dependson pipe size, roughnessand gradient (see
reference 29, Fig. 5). In fact, allmembers of the family are closelyspaced,
and very similar on average to equation (20), so the distinction is somewhat
academic.
48. Theabove theoreticalconsiderationswere not entirelyconfirmed by
experiment,however, as the apparent roughness under part-fullconditions

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS 327
proved to be slightlygreater than when running full-bore. There are four
possible explanations for the discrepancy of several per cent.
(U) Thetransformation of the circular-section equation to non-circular
conditions, equation (6) becoming equation (7) may be an over-
simplification.
(6) There may be (and clearly is,for some ofthe salt-glazed pipe tests)a real
change in effective roughnessas the wetted perimeter changes.

PIPE DIAMETER:INCHES

FIG.11 .-ck"ARISON OF FLOW FORMULAE FOR NEW SEWERS

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 328 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS

(c) Additionalenergylossesmayoccur under free-surfaceconditions

because of surface waves,an effect which would depend uponFroudc
number.
(d) Secondary currents in free-surface flow depressthe filament of maximum
velocity and induce transverse surface curvature.
(e) The measured central depth may not accurately represent the mean
water level, either because of the stationary wave pattern under (c),
or the surface curvature under (d).
49. Consideringthesepossibilities further, although the process of con-
verting a pipe formula to an open-channel one may appear axiomatic, there is
the implied assumption that integrating the equation for the turbulent velocity
distribution over a non-circular section would yield the same numericalconstants
as for a circular section, an assumption which KeuleganS9 has shown to be
inaccurate in the cases he considered, namely trapezoidal channels and very
wide ones. The Keulegan hypothesis may go part-way towards explaining the
difference, but is definitely not the whole story.
50. Turning to the second possible explanation, although it is inconceivable
that there should be a real difference in the surface roughness of the spun-
concrete or pitch-fibre pipelines between full-bore and part-bore conditions, it
is obvious that the very reverse holds for the salt-glazed pipes with eccentric
joints, especially whenthe imposed mis-alignments were all in the same direction.
With steps-up at each joint, the roughness truly increases as the depth reduces
whilst with steps down the reverse trend would be expected. The summarized
data inTables 4 and 5 for series B, C, D, and E reflect this real changeof rough-
ness, which is shown by the plot of k against proportional depth for series C
(2-ft pipes) given in Fig. 12. Modifying equation (19) on the lines of equation
(18) for part-full conditions, the theoretical variation of roughness with depth
can be calculated,as given in Fig. 12. (It has been assumedthat the contribution
of negative steps to roughness is half that of positive steps.) Despite the scatter
due largely to the inherent sensitivity of the linear measure of roughness, the
theoretical trend finds general confirmation.
51. Considering next the effectswhichmay arise from surface waves and
secondary flow, resulting in a depression of the maximum velocity filament,
the Froude numbers for half-full conditions in eachseries of tests were as
follows :
Salt-glazed ware (seriesG : 2 ft) 1.9 > F,, > 1a 7 (series B: 2 ft)
Spun precast pipes: F,, = 1a 7
Pitch-fibre pipes:F,,= 1.7
The Froude numbers are almost independent of flow depth for dp< 0-5: for
dp> 0.5, F, gradually reduces to become unity at dp=0.9 approx. As these
donot coverasignificant range of variation, an analysisof the variation
of resistance with Froude number is not feasible. It was shown by Powell's40
tests in a triangular-section channel that the apparent roughness was dependent
on Froude number, the maximum effect being at F,,= 1.0, with critical flow
yielding a higher resistance than sub-critical in general.
52. The dependenceof the headloss in free-surfaceflow on the Froude
number has been investigated by Chen Che-Pen41 who, from experiments on a

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE H r n R A u L r c RESISTANCE OF DRAINAGE CONDUITS 329
small
rectangular-section
glazed
channel,
derived
the
following
empirical
relationship:

where h0 is the friction factor in a circular-section pipe of the same hydraulic

mean depth under similar conditions.
53. Applying this modification to the basic turbulent equation, he obtained
from field data on a concrete-lined channel k=0.001 ft (compare with Table 1,

1.5 2
ROUGHNESS k : FEET
FIO. 12.-VARIATION OF R O U G W S WITH PROPORTIONALDEPTH; ECCENTRIC-JOINT TYPE
ROUGHNESS IN SALT-GLAZED PIPES

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 330 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS

class 3 and 4), whilst the unmodified equation gave an apparent roughness of
0.016 ft.
54. The new experimental data on the roughness of pipes flowing full can be
similarly modifiedto cover part-full conditions at a Froudenumber of 1.7 using
Chen Che-Pens empirical relationship, and the anticipated apparent part-full
roughness can then beestimated from the basic equation. The comparison
between the estimated apparent roughness (with Froude effects) and the experi-
mental values are as follows:

k in feet Experimental, Estimated by modifying pipe-full

part-full values for Froude effects
Concrete pipes 000045 040035 (0.00013 when full)
Pitch-fibre pipes 00001 1 O.ooOo6 (smooth when full)

55. It thusappearsthatFroude effectsmaybe the mainsource of the

differences in the apparent roughness, between full-bore and free-surface con-
ditions, although there are objections to the general application of Chen Che-
Pens empirical equation.
56. The fourthhypothesis about the cause of the apparent increase in resist-
ance was the transverse curvature of the water surface. NOsuch measurements
were takenin the testsdescribed, butonanother occasion the transverse
surface profile in a verysteep15-in.-dia.concretepipeline (i= 1 in 22) was
measured at several discharges, and showed appreciable curvature.
57. This effect would be far less pronounced at the lower velocities which
obtained in the friction investigation reported here, but nevertheless might have
been sufficient to explain the small residual differences.
58. If the surface curvature is convex upwards, this would lead to a positive
error in the estimation of the flow cross-section from a measurement of depth
at the centre. Thus the mean velocity would be underestimated and the rough-
nessoverestimated. A stationary pattern of surface waves-the familiar
diamond pattern seen in steep channels with joint roughness-can lead to this
kind of error, the sign dependingon whether the depth measurement happensto
coincide with a crest or trough in the pattern.
Comparison of experimental data with Morriss new concept
59. Recently, Morris24 has severely criticizedthe Colebrook-White equation,
despite the wide range of experimental data on commeIcial pipes which are in
general agreement, on the grounds that some types of roughness give different
resistance laws which this generaltransition formula does
not adequatelyrepresent.
It cannot be denied, of course, that the transition law is not truly universal:
Colebrook and Whiterecognized that where the roughnesselements were
spacedcloselyenough to givemutualinterferencebetween their wakes, a
different function applied, as Nikuradse18 had already demonstrated, and the
equation recommended was intendedto apply to roughnesses formedof isolated
elements. Colebrooks analysis of commercial pipe data led him to conclude:
It is seen that although some of the pipes do not agree very closely with the
mean curves, some having too rapid a transition and others too slow, there
appears to besufficientevidence to justify the adoption of thegivenmean
transition law together with the mean k-values. Now Morris classifies three

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAlNAGECONDUITS 331
types of roughness for rigid boundaries by their mechanism of turbulence.
Closelypackedelementssuch as Nikuradsesuniform sand causemutual
interference; this he callswake-interference.Secondly, if the roughness
leaves an essentiallyevensurfacewithdepressions or grooves, the flowwill
skim the surface leaving vortices in dead-water pockets in the depressions; this
Morris calls quasi-smooth. His third category is isolated roughness, in
which the well-separated projections are interspaced with a smooth wall, each
element making an independent contribution to resistance: With regard to
typical commercialconduit surfaces, wrote Morris, it is apparent thatisolated
roughnessflowoccursovermost of them,includingconcrete,rivetedsteel,
wood-stave, cast-iron, brick, etc., includingalsoconduitswhich are nearly
smooth but have protrusions at the joints. For sewers, then, our main con-
cern at this stage is whether the Colebrook-White isolated-roughnesstransition
equation is the best available, or does the equation Morris suggested have any
advantage over it?
60. The latter equation treats the effect of the protuberances as an addition
to the smooth-pipe resistance of the foundation pipe, thus:
h = h, (1+67.21E) . . . . . . * (24)
where h, is the smooth-turbulent friction factor at the appropriate obtained
from :
- z/x
1 = -2 lOgRTx 2.51

and E is defined as

where e/p is the ratio of the total peripheral length of the roughness elements
to total wetted perimeter, and e/p is the ratio of roughness height to spacing.
Two noteworthy features are (i) that according to equation (24) a plot of log h
against logR for an isolated-roughness surface should give a curve parallelto the
smooth law of equation (25) never reaching the square-law region of rough-
turbulence and (ii) that the relativeroughnessdefined as e / D has no direct
bearing on the problem.
61. The former feature is given some support by the tests on eccentrically
jointed salt-glazedpipes, the experimental data for whichfollowedsteeper
transitions than the Colebrook function (Figs 2 and 3) and did not reach the
square law region withthe available discharge,though they might havedone had
higher discharges been available. On the other hand, the concrete-pipe data
in Fig. 5 do follow the transition function veryclosely, although again the
rough-turbulent region was not attained. A comparison of the accuracy of fit
of the alternative equations has been made, and the results are summarized
in Table 8 in terms of the standard deviation of the experimental data from the
theoretical curves with the best fitroughness coefficients,k and Erespectiveiy.
62. As a result of this comparison, it maybeconcluded that there is no
significant difference between the accuracies of the Colebrook-White equation
and the more-recent Morris equationfor fine-textured concrete pipes (each being
much preferable to the Manning formula), but that the eccentric-joint type of
roughness yields a somewhat steeper resistance function than the standard
transition law would imply, which is more accurately represented by Morriss

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 332 ACKERS ON THE HYDRAULICRESISTANCE OF DRAINAGE CONDUITS

TABLE
S.--COMPARISON OF EXPERIMENTAL DATA WITH COLEBROOK-WIFITE
AND MORRIS EQUATIONS
-
Class of pipe Colebrook-White
k x 104 (I
I
_-
l E
Moms

feet per cent

12-in. concrete . . . . . . . 1.26 4.0 17.2
18-in. concrete
36-in. concrete l
24-in.concreteSeereference35
Mean .
1 -99
3-62
5.32
1.1
2.8
12.4
35.0
35.1
5.1
* 3.3%
12-in. salt-glazed, 3-ftunits
Series A . . . .
. . . 1.30 15.1 2.1
B. . . . . . . . 4.12 36.8 3.0
c. . . . . . . . 3.12 32.7 3.6
D . . . . . . . 1e94 22.4 2.7
E. . . . . . . . 3.25 1.9 31.9
. . . . . . .

1
F. 2.34 24.3 1.8
G . . . . . . . 2.54 26.9
H . . . . 2.16 24.3 1a9
12-in. salt-glazed, i-ft & i s 1.6
Series A . . . . . . . . 1.37
B. . . . . . . . 5.44
,--
L . . . . . . . . 3.32
D . . . . . . . 2.3 1
E. . . . . . . . 404
F. . . . . . . . 2.43
G . . . . . . . 216
H . . . . 2.28
Mek

equation (2.2%mean deviation compared with3.4%). However, in view of the

general acceptanceof the Colebrook functionfor so many commercial surfaces,
the writer suggests that the 1% or so loss of accuracy for this class of material
is acceptable in return for the convenience of auniversal equation.

SUMMARY AND CONCLUSIONS

63. The following conclusions may be drawn from this research:
(a) Flow conditions in newsewers are turbulent-transitional and neither
the Crimp and Bruges nor Manning formulae fit the experimental
data.
(b) The preferred equation is that proposed by Colebrook and White, for
which design charts are available, but Morriss more recent equation
is a slightly better fitto the eccentric-joint type of roughness.
(c) Previous evidence on roughness is acceptable for many types of con-
struction (Table l), and this has been supplemented bynew research
on three classes of pipeline.
( d ) Salt-glazed ware produces roughness values dependent on joint eccentri-
e2
city and spacing: there is good correlation between k-value and 7.
P

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGECONDUITS 333
(e) Measurements of the jointing standards under ordinary contracting
standards have provided design recommendations, depending on the
class of workmanship (see 0 43) of 0.0002 ft < k<0.0020 ft for new
pipes running full.
(f)Smooth textured precast-concrete pipes with selfcentring jointshave a
roughness of O W O 2 ft when full, which may increase to 0*0010ft
with less accurate joints.
(g)Pitch-fibrepipes are virtually smooth: a design chart based on
k=0*00001ft is suitable for full-bore computations.
(h) The above recommendations make no allowance for ageeffects or
branch connexions, which the designer must assess.
( i ) The laboratory investigation of part-bore flow at Froude numbers of
about 1.7 gave an apparent increase in roughness, i.e. a reduction in
velocity of several per cent as compared with the theoretical propor-
tionality functions.
( j ) There is a definite need for field data on old sewers: the writer would
be grateful if engineers would offer for publication any such infor-
mation in their possession.

ACKNOWLEDGEMENT
64. The writer is grateful to those members of the staff of the Hydraulics
Research Station whohaveassisted inthe experimentalwork, particularly
to Mr B. A. Say, Assistant Experimental Officer.
65. This work was carried out as part of the research programme of the
Hydraulics Research Board of the Department ofScientific and Industrial
Research, and is published with the permission of the Director of Hydraulics
Research.

NOTATION
A =cross sectional area
b =water-surface width withfreesurface flow
C = Chezy's coefficient in v =
c' = coefficient in friction equation
D = pipe diameter
e' e
E = a roughness parameter, after Morris, defined as Cd- X 7 with C, a
P P
shape coefficient
e = actual height of roughness
e' = sum of roughness dimensions over pipe perimeter
V
= Froude number, -
F,

g
E
= gravitational acceleration
h = head loss in length L of pipe
2gik3
Z = hydraulic gradient parameter, 7
i = energy gradient, surface slope in uniform flow in open channels

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 334 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS

k = roughness (Nikuradses linear measure),the equivalent sand-roughness

m = hydraulic mean depth, Alp
M = hydraulic mean depth parameter, m/k
n = roughness parameter in Manning equation
0 = sufix denoting circular-section pipe running full-bore
P = wetted perimeter
P = suffix denoting proportional values in part-full circular sections
P = longitudinal pitch of roughness elements
4nqm
Q = discharge parameter, -
Pkv
4 = discharge
R. = Reynolds number vD/v
P = velocity parameter, vk/u
V = mean velocity
E, B = exponents in exponential equation
V = kinematic viscosity of fluid
A = friction factor, 2gDi/v2or 8gmilv2
6 = a parameter in the proportionality function
As = friction factor in a smooth pipe (smooth turbulent flow)
& = friction factor in an open channel, with allowance for Froude effects
U = mean deviation, root-mean square value

REFERENCES
1. A. ChCzy. Report, unpublished. h o l e Nationale des Ponts et ChaussCes,1775
2. H. Bazin,Rechercheshydrauliques. MCm.AcadSci.,Paris, 1865.
3. W. R. KutterandE.Ganguillet,Versuch zur AufstellungeinerForme1der
gleichformigerBeneigungdesWassers.Bern, 1877. English translation: R.
Hering and J. C. Trautwine,A general formula for the uniform flow of water in
rivers and channels. Wiley, New York, 1916.
4. R. Manning, On the flow of water in open channels and pipes. Proc. Instn civ.
Engrs, Ire., 1891, vol. 20,pp. 161-207; 1895,vol. 24, pp. 179-205.
5. W. S. Crimp and C. E. Bruges, A new formula for the flow in sewers and water
mains. Proc. Instn civ. Engrs, vol. 122, 1894-95, Part W , pp. 198-201.
6. A. Hazen and G. S. Williams, Hydraulic Tables. Wiley, New York, 1905,1908,
1920.
7. F. C. Scobey, The flow of water in concrete pipes. U.S. Dept of Agriculture,
Bull. 852, 1920, 100 pp. incl. appendixes and discussion.
8. A. A.Barnes, Hydraulic flow reviewed. Spon, London, 1916.
9. R.W. Powell, History of Mannings formula. J. Geophysical Research, vol. 65
(Apr. 1960), No. 4, pp. 1310-1311.
10. A.Flamant,MCcanique appliquk, hydraulique. Baudry, Paris, 1891.
1 1 . W.Wilcocksand P.Holt,ElementaryHydraulics.Nationalprinting office,
Cairo, 1899.
12. R. B. Buckley, Design of channels for irrigation or drainage. Spon, London,
1911.
13. P. A. M. Parker, Control of water. Routledge, London, 1913.
14. Von Karman,MechanischeAhnlichkeitundTurbulenz.Nachr.Ges.Wiss.,
Gottingen, 1930, EnglishTrans.Tech.Memo. Nat. Adv. C o r n . Aero.,Wash
1931, No. 611.
15. L. Prandtl,NeuereErgebnissederTurbulenzforschung. Z . Ver.dtsch.Ing.,
vol. 77, 4 Feb., 1933, No. 5, pp. 105-114.

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 ACKERS ON THE HYDRAULIC RESISTANCE
OF DRAINAGECONDUITS 335
16. 3. A.Bakhmeteff,Themechanicsofturbulentflow.PrincetonUniversity
Press, U.S.A., 1936.
17. T. E. Stanton and J. R. Pannell. Similaritv of motion in relation to the surface
friction of fluids. PM. Trans. (A), 1914, vol. 214, pp. 119-224.
18. J. Nikuradse, Stromungsgesetze in
rauhen
Rohren.
Forschungsheft Ver.
dtsch. Ing., No. 361, 1933. (Abstracted in Zeitschrift V.D.I., vol. 77, No. 39,
Sept. 1933, pp. 1075-1076).
19. C. F. Colebrook and C. M. White, Experiments with fluid friction in roughened
pipes. Proc. Roy. Soc. (A), 1937, vol. 161, pp. 367-381.
20. C. F. Colebrook, Turbulent flowin pipes with particular referenceto the transition
regionbetweenthesmoothandroughpipelaws. J. Instnciv.Engrs., 1939,
vol. 11, pp. 133-156.
21. L. F. Moody, Friction factors for pipe flow. Trans. Amer. Soc., Mech. Engrs.,
1944, vol. 66, pp. 671-684.
22. P.Ackers,Resistanceoffluidsflowinginpipesandchannels.Hydraulics
Research Paper No. 1, London, H.M.S.O., 1958.
23. H. M. Morris, A newconcept of flowinroughconduits.Proc.Amer.Soc.
civ. Engrs., 1954, vol. 80, Separate No. 390, 31 pp.
24. H. M. Morris, Design methods for flow in rough conduits. Proc. Amer. Soc.,
civ. Engrs, 1959, vol. 85, HY7, pp. 43-62.
25. E. S . Crump, Flow of fluids in conduits and open channels. Proc. Instn civ.
Engrs, Part 3, 1956, vol. 5, pp. 522-526.
26. H. Rouse,Evaluationofboundaryroughness.Proc.2nd.Hyd.Conf.,Iowa,
1942, pub. 1943, Bulletin 27, pp. 105-116.
27. S . T. Bonnington, Guide to the calculation of head losses in piping systems.
British Hydromechanics Res. Assn. publication TN445.
28. P. A. Lamont, A review of pipe-friction data and formulae, with a proposed set
of exponential formulae based on the theory of roughness. Proc. Instn civ.
Engrs, Part 3, 1954, vol. 3, pp. 248-274.
29. P. Ackers, Charts for the hydraulic design of channels and pipes. Hydraulics
Research Paper No. 2, London, H.M.S.O., 1958.
30. R.W. Powell, Diagram determines pipe sizes directly. Civil Engineering (New
York), vol. 20, September 1950, pp. 45-46.
31. A.Strickler,BeitragezurFragederGeschwindigkeitsformelundderRauhig-
keitszahlen fur Strome, Kanaleund
geschlossene
Leitungen.Schweitz.
Bauztg, 1924, vol. 83, NO.23, pp. 265-268.
32. R. E. Horton, Some better Kutters formula coefficients. Engg News, February
24, 1916, pp. 313-374.
33. I). L. Yarnell and S. M. Woodward, The flow of water in drain tile. Bulletin
Dept. Agriculture, U.S.A., 1920, NO. 854.
34. E. G. Wilcox, A comparative test of the flow of water in 8-in. concrete and
vitrifiedclaysewerpipe.University of WashingtonEnggExp.Sta.,Bull.
21,1924.
35. L. G. StraubandH. M. Morris,Hydraulictestsonconcreteculvertpipes.
St Anthony Falls Hyd. Lab, Minneapolis, 1950, Technical paper 4, Series B.
36. L. G . Straub and H. M. Morris,Hydraulictestsoncorrugatedmetalculvert
pipes. St AnthonyFallsHyd.Lab,Minneapolis, 1950, Technicalpaper 5,
Series B.
37. H.Schlichting,Experimentalinvestigationoftheroughnessproblem.Trans.
and Abst. by V. L. Streeter, Proc. Amer. Soc. civ. Engrs, vol. 63, Nov. 1937,
pp. 16-31 (supplement).
38. p. Ackers, An investigation of head lossesat sewer manholes. Civil Engineering
and Public Works Review, vol. 54, 1959, pp. 882-884 (July), and pp. 1033-1036
(Sept.).
39. G . H.Keulegan,Lawsofturbulentflowinopenchannels. J. Res. Nat. Bur.
Stand. 1938, vol. 21, pp. 707-741.

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.
 336 ACKERS ON THE HYDRAULIC RESISTANCE OF DRAINAGE CONDUITS
40. R. W. Powell and C. E. Posey, Resistance experiments in a triangular channel.
Proc. Amer. Soc. civ. Engrs, Paper 2018, May 1959, with discussion No. 1959.
41. Chen Che-Pen, The difference between the head losses in a pipe with a circular
cross-section and those in a shallow open channel. La HouiUe Blanche, 14th
year, No. 6, November 1959, pp. 820-826.

The Paper,which was receivedon 18 October, 1960, is accompanied by twelve

sheets of drawings and diagrams, from which the half-tone page plates and the
Figures in the text have been prepared.
Written discussion on this Paper should be forwarded to reach the Institution
before 15 September, 1961, and willbe published in or after January, 1962.
Contributions shouldnot exceed 1,200 words.-Sac.

Downloaded by [ University College London] on [21/09/16]. Copyright ICE Publishing, all rights reserved.