395 © IWA Publishing 2013 Water Science & Technology | 67.

2 | 2013

Verification of equations for incipient motion studies

for a rigid rectangular channel
Charles Hin Joo Bong, Tze Liang Lau and Aminuddin Ab. Ghani

ABSTRACT
Charles Hin Joo Bong (corresponding author)
The current study aims to verify the existing equations for incipient motion for a rigid rectangular
Tze Liang Lau
channel. Data from experimental work on incipient motion from a rectangular flume with two Aminuddin Ab. Ghani
River Engineering and Urban Drainage Research
different widths, namely 0.3 and 0.6 m, were compared with the critical velocity value predicted by Centre (REDAC),
Universiti Sains Malaysia,
the equations of Novak & Nalluri and El-Zaemey. The equation by El-Zaemey performed better with an Engineering Campus,
Seri Ampangan,
average discrepancy ratio value of 1.06 compared with the equation by Novak & Nalluri with an average
14300 Nibong Tebal,
discrepancy ratio value of 0.87. However, as the sediment deposit thickness increased, the equation Penang,
Malaysia
by El-Zaemey became less accurate. A plot on the Shields Diagram using the experimental data had E-mail: bhjcharles@feng.unimas.my

shown the significant effect of the sediment deposit thickness where, as the deposit becomes
thicker, the dimensionless shear stress θ value also increased. A new equation had been proposed
by incorporating the sediment deposit thickness. The new equation gave improved prediction with
an average discrepancy ratio value of 1.02.
Key words | critical velocity, incipient motion, rigid rectangular flume, sediment deposit thickness

INTRODUCTION

Open drain system has been conventionally used to
convey storm water runoff in urban areas in developing
countries like Malaysia. Open drain system could be
quite efficient in rapid removal of surface runoff; however,
sediment deposition tends to build up after a period of
time. Sediment deposition in sewers and open drains
has been identified as one of the factors of flash flooding
in urban areas (Campisano et al. ; Liew et al. ;
Rodríguez et al. ). To reduce sedimentation, a constant
minimum velocity of 0.9 m/s has been recommended by
the Department of Irrigation and Drainage (DID) Malay-
sia for the purpose of self-cleansing design (Ab. Ghani
et al. ). The adoption of a constant minimum value,
however, does not take into account the characteristics
of the sediment and the hydraulic aspect of the channel
(Butler et al. ; Tranckner et al. ; Goormans
et al. ). The design criteria through incipient motion
equations might be a more viable approach as these
equations take into account some aspect of the sediment
and channel characteristics. Vongvisessomjai et al. ()
provides a review of existing self-cleansing design criteria
(Nalluri et al. ; Ota & Nalluri ) for sewers based
on both the critical velocity and critical shear stress
approach. Current developments in turbulent flow model-
ling induced by 2D and 3D roughness (Bhaganagar & Hsu
; De Marchis et al. ; De Marchis & Napoli )
have demonstrated that random irregularities of the wall,
such as in sewers and open concrete drains, due to sedi-
ment deposition caused reduction in the streamwise
velocity with respect to smooth wall condition. This
resulted in modification of incipient motion which devel-
oped in the near wall region. The implication of this
modification could be pursued as future work.
Vast literature exists on the subject of incipient motion
of sediment; however only a few studies which are used
for comparison with the results of this study are mentioned,
paying attention to the criteria for beginning of motion using
relationships for critical velocity Vc. Determining when
incipient motion starts is quite subjective. Visual obser-
vation normally categorised the intensities of movement
near critical condition into: (i) no movement, (ii) weak
movement, (iii) medium movement, and (iv) general move-
ment (Kramer ). Two distinct bed load movement
modes have been observed during critical condition,
namely discrete movement of individual grains and collec-
tive movement as bed forms (James & Minter ;

doi: 10.2166/wst.2012.580
 396 C. H. J. Bong et al. | Incipient motion studies for a rigid rectangular channel
James et al. ). Though various efforts have been made to
standardise incipient motion condition, Shields Diagram
(Shields ) was widely accepted for design purposes.
With the introduction of subsequent data, additional scatter
had been introduced into the Shields Diagram. To take into
account the scatter, Paphitis () developed an envelope
for the Shields Diagram with an upper and lower limit and
also a single curve representation by using data sourced
from 29 researchers.
Using sediment with sizes ranging from 3.6 to 37.2 mm,
and for the condition of both smooth and rough fixed bed
with touching sediment particles for both circular and rec-
tangular flume, Novak & Nalluri () developed the
following equation for predicting critical velocity:
 
0:4
Vc d
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:5 50
gd50 ðSs
1Þ R
where Vc is the critical velocity [m/s], g is the acceleration of
gravity [m/s2], d50 is the representative diameter of the sedi-
ment [m], Ss is the specific gravity of the sediment [–] and R
is the hydraulic radius of the flow section [m].
El-Zaemey () conducted an experiment for incipient
motion of touching grouped particles for sediment sizes ran-
ging from 2.9 to 8.4 mm on both smooth and rough rigid bed
in a circular flume with flat bed and obtained the following
relationship:
 
0:34
Vc d
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:75 50
gd50 ðSs
1Þ R
Both Equations (1) and (2) to predict the critical velocity
are of the type:
 b
Vc d
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ a 50
gd50 ðSs
1Þ R Vc , ρ, ρs , d50 , R, γ s , ts
relating the particle critical Froude number (left hand side)
with the relative roughness (right hand side).
Using the sediment sizes commonly found in Malaysian
drains for the experimental work in a rigid rectangular
flume, Ab. Ghani et al. () maintained that Equation
(2) gave better prediction for the critical velocity than
Equation (1). However, as the sediment deposit thickness
increased, the prediction by Equation (2) became less accu-
rate (Ab. Ghani et al. ). This shows that a new equation
is needed to take into account the presence of different sedi-
ment deposit thickness. Current available literature is
Water Science & Technology | 67.2 | 2013
lacking on the study of the influence of sediment deposit
thickness on the critical velocity.
The influence of sediment deposit thickness on the criti-
cal velocity was observed and a new equation was proposed
by incorporating the effect of sediment deposit in this paper.
The experimental works were carried out in two rectangular
channels with different width using sediment size com-
monly found in Malaysian urban concrete drains. The
observed critical velocity data were then compared with
the predicted value from existing incipient motion equations
in the literature (Equations (1) and (2)) to verify the applica-
bility of the existing equations.
DIMENSIONAL ANALYSIS
(1)
According to Yalin (), the ‘components’ to describe a
two-phase phenomenon involving fluid and sediment are:
(i) fluid, (ii) noncohesive granular medium and (iii) flow.
To derive the expression in the form of Equation (3), fluid
is defined by its density ρ; the noncohesive granular
medium is defined by its density ρs and representative
size/diameter d50; and the flow is defined by the hydraulic
radius of flow area R and gravity acceleration g. The critical
velocity Vc is expressed as Vc ¼ f ðρ, ρs , d50 , R, gÞ. By choos-
ing to replace g with γ s ¼ gðρs
ρÞ, where the specific
weight for the sediment γ s could be excluded as it would
be a constant and can be eliminated from the analysis
(2) (Azmathullah et al. ), the dimensionless equation is
obtained in the form of Equation (3).
In order to incorporate the effect of sediment deposit
thickness, another characteristic parameter – the sediment
deposit thickness, ts – is added, obtaining the following par-
ameters:
(3)
(4)
By choosing Vc, ρ and d50 as the repeating variables
and applying the Buckingham π Theorem while excluding
the specific weight of the sediment γ s , as it would be
a constant, the following dimensionless variables are
obtained:
Vc R ts
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , , (5)
gd50 ðSs
1Þ d50 d50
with an additional term which is the relative thickness of the
sediment deposit.
 397 C. H. J. Bong et al. | Incipient motion studies for a rigid rectangular channel
METHODOLOGY
The current experimental work involved a rectangular open
channel flume with the dimensions 6 m (L) × 0.6 m (W) ×
0.4 m (D). The velocity and discharge values in the channel
were obtained from an electronic current meter which
could be further verified with the discharge calculated from
the V-notch installed downstream of the channel. The non-
cohesive sediment used had d50 sizes of 0.81, 1.53 and
4.78 mm with a specific gravity of 2.54, 2.55 and 2.57,
respectively. The sediment used had a geometric standard
deviation σg value of 1.31 mm for d50 ¼ 0.81 mm (uniform
sand); σg value of 2.63 mm for d50 ¼ 1.53 mm (slightly nonu-
niform for mixture of sand and gravel); and σg value of
1.27 mm for d50 ¼ 4.78 mm (uniform gravel). The exper-
iment was run at four different slopes (1/200, 1/350, 1/500
and 1/1,000). Initially the sediment was positioned in one
layer (i.e. sediment thickness, ts ¼ d50 for each of the respect-
ive sediment mixtures used). Later, the sediment was placed
in different thicknesses, namely 5, 10 and 24 mm. The total
number of runs for the experiment was 48. The selection of
sediment size, slope and sediment thickness was based on
site observations by Ab. Ghani et al. ().
In running the experiment, the sediment was placed and
levelled to the required thickness starting at about 2.5 m
from the upstream inlet of the flume as shown in Figure 1.
The total length of the sediment bed was about 2.1 m. The
observation section for the sediment bed was 0.6 m (L) ×
0.6 m (W) and located at 3.5 m from the upstream end of
the flume. An extension of about 1 m of the sediment bed
upstream of the observation section was to reduce the
effect of sudden change of channel bed material. It was
observed from preliminary experiments that the flow is
fully developed at 3.5 m from the upstream inlet of the
flume. Before the pumps were turned on, the sediment
Figure 1 | Schematic diagram of the flume.
Water Science & Technology | 67.2 | 2013
sample was moistened by filling the flume with water from
a water hose so as to minimise the initial filling surge.
Water level and discharge were slightly increased while uni-
form flow was maintained. This was repeated until some
particles started to move. The definition used in this exper-
iment was of general movement as defined by Kramer
() via visual observation.
Another set of experimental data was obtained from
Salem () where a rectangular flume with dimensions
10 m (L) × 0.3 m (W) × 0.45 m (D) was used. The non-cohe-
sive uniform sediment used had d50 sizes of 0.55, 0.97,
1.80, 3.09 and 4.78 mm and specific gravity ranging from
2.301 to 2.569. The experiment was run at six different
slopes (1/500, 1/600, 1/750, 1/1,000, 1/1,200 and 1/1,700)
with four different thicknesses, namely one layer (ts ¼ d50),
5, 10 and 24 mm. The total number of runs for the exper-
iment was 119. The definition used by Salem () for
incipient motion was also of general movement as defined
by Kramer ().
The total number of run from the two experiments was
167. The two sets of data were plotted on the Shields Dia-
gram to allow comparison with the existing data from the
literature. Verification with the existing equations by
Novak & Nalluri () and El-Zaemey () for predicting
critical velocity was done to determine which existing
equations give better approximation by comparing in
terms of the discrepancy ratio value. The discrepancy ratio
value was calculated using Equation (6). Improvement of
the existing equation was done by incorporating the sedi-
ment deposit thickness, ts, as one of the factors for
incipient motion. Multiple linear regressions were carried
out using MINITAB to derive the improved equation:
Vc predicted (m=s)
Discrepancy ratio ¼ (6)
Vc observed (m=s)
 398 C. H. J. Bong et al. | Incipient motion studies for a rigid rectangular channel
RESULTS AND DISCUSSION
Comparison with Shields Diagram
Figure 2 shows the data from the current study when plotted
on the Shields Diagram. The upper limit, lower limit and
single curve on the Shields Diagram were obtained using
the relationship developed by Paphitis (). From Figure 2,
about 55% (92 out of 167) of the data for the current study
falls within the upper and lower limits. It could also be
observed that, generally, as the thickness of the sediment
deposit increased, the critical shear stress required for inci-
pient motion also increased as evidenced in the increase
of the dimensionless shear stress θ value (the triangle and
round shape markers representing sediment deposition
thickness of 10 and 24 mm respectively, generally situated
higher in the vertical axis of Figure 2 than the diamond
and square shape markers for sediment deposition thickness
of d50 and 5 mm respectively). This could be due to the
greater friction that existed between the sediment particles
as the sediment deposit thickness increased. The effect of
‘support’ from neighbouring particles also increased as the
sediment deposition thickness became thicker and thus
the particles became more stable.
Figure 2 | Data for current study as plotted on the Shields Diagram.
Water Science & Technology | 67.2 | 2013
Comparison with existing critical velocity equations
Figure 3 shows the comparison between the observed criti-
cal velocity and the critical velocity predicted by using
equations by Novak & Nalluri () and El-Zaemey
(). It could be observed that the equation by Novak &
Nalluri tended to underpredict the critical velocity values
compared with the observed values where most of the
points plotted were below the line where the discrepancy
ratio equals one. The average discrepancy ratio value given
by Novak & Nalluri () was 0.87. The equation by El-
Zaemey performed better giving an average value for the dis-
crepancy ratio of 1.06. Comparison in terms of slope for the
best fit line also shows that velocity predicted by El-Zaemey
() had a slope which was closer to the value 1 (the slope
for the line of discrepancy value equals 1) than the velocity
predicted by Novak & Nalluri (). Though the equation
of El-Zaemey () was developed using a circular channel,
the presence of a flat rigid bed in a circular channel tends to
simulate the incipient motion condition in a rigid rectangu-
lar channel.
The results also conformed to the previous finding that as
the thickness of the sediment deposit increased, the predic-
tion using the equation by El-Zaemey () became less
 399 C. H. J. Bong et al. | Incipient motion studies for a rigid rectangular channel Water Science & Technology | 67.2 | 2013

Figure 3 | Comparison between observed and predicted critical velocity for (a) using the Novak & Nalluri (1984) equation and (b) using the El-Zaemey (1991) equation.

accurate. This could be seen in Figure 4(a) where for the thick- sediment deposit thickness increased, the slope for the best
ness of d50 (one layer), the best fit line for the data had a slope fit line for the points as predicted by the equation of El-
value of 0.8401 which was close to 1. However, as the Zaemey () started to move away from the slope value
 400 C. H. J. Bong et al. | Incipient motion studies for a rigid rectangular channel
Figure 4 | Comparison between observed and predicted critical velocity by using the El-Zaemey (1991) equation for different sediment deposit thickness: (a) d50, (b) 5 mm, (c) 10 mm and
(d) 24 mm.
of 1 and became milder with the worst case in Figure 4(d) for
the thickness of 24 mm with a slope value of 0.5019.
Improvement to critical velocity equation
To improve the existing critical velocity equation, the effect
of sediment deposit thickness was incorporated into the
equation. Hence, using the dimensionless variables in (5)
and MINITAB to perform the regression analysis, the fol-
lowing form of critical velocity equation was obtained:
 
0:167  0:0378
Vc d ts
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 1:17 50 (7)
gd50 ðSs
1Þ R d50
where Vc is the critical velocity [m/s], g is the acceleration of
gravity [m/s2], d50 is the representative diameter of the sedi-
ment [m], Ss is the specific gravity of the sediment [–], R is
the hydraulic radius of the flow section [m] and ts is the sedi-
ment deposit thickness [m]. The term on the left hand side
of the equation is the particle critical Froude number
while the first term on the right hand side is the relative
roughness and the second term is the relative thickness.
Water Science & Technology | 67.2 | 2013
The correlation test showed that the particle critical
Froude number was correlated to the relative roughness
and relative thickness with Pearson correlation values of

0.582 (p-value ¼ 0.000) and 0.496 (p-value ¼ 0.000),
respectively. In comparison, another equation in the form
of Equation (3), which was the modification of the equation
available in the literature (Novak & Nalluri ; El-Zaemey
), was developed using the data from the current study
and the following was obtained:
 
0:228
Vc d
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 1:015 50 (8)
gd50 ðSs
1Þ R
Figure 5 shows the plot of the observed critical velocity
with the critical velocity predicted by Equation (7) and the
best fit lines for the velocity predicted by both Equations (7)
and (8). Comparing the slope for the best fit line for Equations
(7) and (8), Equation (7) performed better than Equation (8)
as the slope value was closer to the value of 1.
Equation (7) also performed better than the equation by
El-Zaemey () in terms of average discrepancy ratio value
 401 C. H. J. Bong et al. | Incipient motion studies for a rigid rectangular channel
Figure 5 | Comparison between observed and predicted critical velocity for Equations (7) and (8).
with a value of 1.02. In terms of the increased sediment
deposit thickness, Equation (7) performed better than the
equation by El-Zaemey () as shown in Figure 6. Improve-
ment can be seen where the data point moving further away
from the discrepancy ratio line with the slope value of 1 and
forming a best fit line with increasingly milder slope as the
sediment deposit thickness increased was not as significant
as the one seen in Figure 4. The slope for the best fit line for
the points calculated using Equation (7) varied from 1.016
(for sediment deposition thickness of d50) to 0.6612 (for
sediment deposition thickness of 24 mm) which was an
improvement compared with using the equation by El-
Zaemey (). Table 1 shows the comparison of the discre-
pancy ratio value obtained by using the equations by Novak
& Nalluri () and El-Zaemey (), and Equation (7).
Generally, Equation (7) performed better than the equations
by Novak & Nalluri () and El-Zaemey () after taking
into account the thickness of sediment deposit.
The reason why the equations by the previous research-
ers did not include the effect of sediment deposit thickness
is due to the assumption that sediment deposit will not
build up under the condition of steady flow with critical vel-
ocity (such as the condition of the experiment in the
laboratory). However, the flow in open drains on site tends
to vary and any changes in cross-section (such as bend and
Water Science & Technology | 67.2 | 2013
entering a culvert) will induce sediment deposition. Thus,
for the incipient motion equations to be applicable for the
design of open drains, the effect of sediment deposit thick-
ness needs to be incorporated into the equations. Another
interesting observation was that the predicted critical vel-
ocity by Novak & Nalluri () and El-Zaemey () does
not vary much and seems to assume a flat trend as shown
in Figure 3 at higher critical velocity value (observed critical
velocity roughly more than 0.5 m/s). This is less evident for
Equation (7), as seen in Figure 5, suggesting the new equation
is also representative for higher flow regime. Further study on
the common sediment deposit thickness that is found in open
drains could be carried out to determine the suitable sedi-
ment deposit thickness to be used in the new equation for
the purpose of self-cleansing design.
CONCLUSIONS
The current study aims to verify the applicability of the exist-
ing incipient motion equation to predict critical velocity for
sediment in a rigid rectangular channel. By comparison with
the observed critical velocity data from the experimental
work in this paper, the equation by El-Zaemey () per-
formed better with an average discrepancy ratio value of
 402 C. H. J. Bong et al. | Incipient motion studies for a rigid rectangular channel Water Science & Technology | 67.2 | 2013

Figure 6 | Comparison between observed and predicted critical velocity by using Equation (7) for different sediment deposit thickness: (a) d50, (b) 5 mm, (c) 10 mm and (d) 24 mm.

Table 1 | Comparison of discrepancy ratio value at different sediment deposit thickness effect of the slope for best fit line for the points predicted
using the new equation deviating from the value of 1 as
Discrepancy ratio
Sediment deposit thickness
the sediment thickness increased is not as significant as
Equation d50 5 mm 10 mm 24 mm Average the one predicted by El-Zaemey ().
Novak & Nalluri () 0.90 0.82 0.85 0.93 0.87
El-Zaemey () 1.11 1.00 1.03 1.11 1.06
Equation (7) 1.06 0.96 0.99 1.07 1.02 ACKNOWLEDGEMENTS

The authors would like to thank Universiti Sains Malaysia

1.06 compared with the equation by Novak & Nalluri ()
with an average discrepancy ratio value of 0.87. As the sedi-
ment deposit thickness increased, the equation by El-
Zaemey () became less accurate. This could be seen by
comparing the slopes of the best fit line for the points pre-
dicted by the equation of El-Zaemey (), where the
slope values deviated from the value of 1 and became
milder as the sediment deposition thickness increased. A
new equation was proposed by incorporating the sediment
deposit thickness. The new equation gave improved predic-
tion with an average discrepancy ratio value of 1.02. The
for the financial support under the RU-PRGS grant No.
1001/PREDAC/8044050. Special appreciation goes to Dr
Darrien Mah Yau Seng of Universiti Malaysia Sarawak for
proofreading the manuscript.
REFERENCES
Ab. Ghani, A., Salem, A. M., Abdullah, R., Yahaya, A. S. &
Zakaria, N. A.  Incipient Motion of Sediment Particles
Over Loose Deposited Beds in a Rigid Rectangular Channel.
8th International Conference on Urban Storm Drainage.
Sydney, Australia.
 403 C. H. J. Bong et al. | Incipient motion studies for a rigid rectangular channel
Ab. Ghani, A., Zakaria, N. A., Kassim, M. & Nasir, B. A. 
Sediment size characteristics of urban drains in Malaysian
cities. Urban Water 2, 335–341.
Ab. Ghani, A., Zakaria, N. A. & Kassim, M.  Sediment
deposition in a rigid monsoon drain. International Journal of
River Basin Management 6, 23–30.
Azmathullah, H. H. Md., Deo, M. C. & Deolalikar, P. B. 
Neural networks for estimation of scour downstream of a ski-
jump bucket. Journal of Hydraulic Engineering 131, 898–908.
Bhaganagar, K. & Hsu, T. J.  Direct numerical simulations of
flow over two-dimensional and three-dimensional ripples and
implication to sediment transport: Steady flow. Coastal
Engineering 56, 320–331.
Butler, D., May, R. & Ackers, J.  Self-cleansing sewer design
based on sediment transport principles. Journal of Hydraulic
Engineering 129, 276–282.
Campisano, A., Creaco, E. & Modica, C.  Laboratory
investigation on the effects of flushes on cohesive sediment
beds. Urban Water Journal 5, 3–14.
De Marchis, M. & Napoli, E.  Effects of irregular two-
dimensional and three-dimensional surface roughness in
turbulent channel flows. International Journal of Heat and
Fluid Flow 36, 7–17.
De Marchis, M., Napoli, E. & Armenio, V.  Turbulence
structures over irregular rough surfaces. Journal of
Turbulence 11, 1–32.
El-Zaemey, A. K. S.  Sediment Transport over Deposited Bed
Sewers. Ph.D., University of Newcastle upon Tyne.
Goormans, T., Engelen, D., Bouteligier, R., Willems, P. &
Beramont, J.  Design of self-cleansing sanitary sewer
systems with the use of flushing devices. Water Science and
Technology 60, 901–908.
James, C. S. & Minter, W. E. L.  Experimental flume study of the
deposition of heavy minerals in a simulated Witwatersrand
sandstone unconformity. Economic Geology 94, 671–688.
James, C. S., Bulovic, N. & Naidoo, E.  Modes of bed-load
movement on a smooth rigid surface. Journal of Hydraulic
Engineering 137, 1283–1286.
First received 11 June 2012; accepted in revised form 29 August 2012
Water Science & Technology | 67.2 | 2013
Kramer, H.  Sand mixtures and sand movement in fluvial
models. Transaction, ASCE 100, 798–878.
Liew, Y. S., Selamat, Z., Ab. Ghani, A. & Zakaria, N. A. 
Performance of dry detention pond: Case study of Kota
Damansara, Selangor, Malaysia. Urban Water Journal 9,
129–136.
Nalluri, C., Ghani, A. A. & El-Zaemey, A. K. S.  Sediment
transport over deposited beds in sewers. Water Science and
Technology 29, 125–133.
Novak, P. & Nalluri, C.  Incipient motion of sediment particles
over fixed beds. Journal of Hydraulic Research 22, 181–197.
Ota, J. J. & Nalluri, C.  Urban storm sewer design: Approach
in consideration of sediments. Journal of Hydraulic
Engineering 129, 291–297.
Paphitis, D.  Sediment movement under unidirectional flows:
An assessment of empirical threshold curves. Coastal
Engineering 43, 227–245.
Rodríguez, J. P., McIntyre, N., Diaz-Granados, M. & Maksimović, Č .
 A database and model to support proactive management
of sediment-related sewer blockages. Water Research 46, 4571–
4586.
Salem, A. M.  Incipient Motion over Loose Deposited Beds in
a Rigid Rectangular Channel. MSc Thesis, Universiti Sains
Malaysia.
Shields, A.  Anwendung der Ahnlichkeitsmechanik und
der Turbulenzforschung auf die Geschiebebewegung.
Mitteilungen der Preussischen Versuchsanstalt fur Wasserbau
und Schiffbau, No. 26, Berlin, Germany (English Translation
by W. P. Ott & J. C. Uchelon, California Institute of
Technology, Pasadena, Calif.).
Tranckner, J., Bonisch, G., Gebhard, V. R., Dirckx, G. & Krebs, P.
 Model-based assessment of sediment sources in sewers.
Urban Water Journal 5, 277–286.
Vongvisessomjai, N., Tingsanchali, T. & Babel, M. S.  Non-
deposition design criteria for sewers with part-full flow.
Urban Water Journal 7, 61–77.
Yalin, M. S.  Mechanics of Sediment Transport. Pergamon
Press, Great Britain.