2. literature survey - shodhgangashodhganga.inflibnet.ac.in/bitstream/10603/4525/13/13... ·...
TRANSCRIPT
30
2. LITERATURE SURVEY
2.1 GENERAL
A detailed report on the literature review covering relevant studies
available so far is presented in this chapter. The investigations were
made on flow through water distribution pipes regarding various
losses, hydraulic friction factor and unsteady turbulent flow.
Formation of biofilm and scale/deposits, their effects and prevention
in water pipe lines were also reported in literature. Research work has
been done on the various factors influencing the formation of scale in
water distributing pipes. Some modeling studies have been carried out
by eminent researchers in the field of hydraulic engineering. The
applications of an Artificial Neural Networks technique in water
resources engineering problems are also reviewed. Special emphasis is
given to the flow through PVC pipes.
Section 2.2 gives the loss coefficients for small diameter pipes.
Section 2.3 deals with the various factors influencing corrosion rate
and releasing of lead and iron from drinking water pipes. Section 2.4
enumerates the effects of biofilm/scale formation in pipes. Sections
2.5 and 2.6 mainly deal with the investigations on friction factor for
pipe flow and domestically used PVC pipes respectively. Section 2.7
describes some modeling studies on pipe flow and water resources
applications. Finally the chapter ended with an explanation on the
importance of present study.
31
2.2 LOSS COEFFICIENTS FOR SMALL DIAMETER PIPES
According to Gabriel Echavez (1997), there is an increase in losses
coefficient with age for small diameter pipes. The hydraulic behavior of
galvanized iron and copper pipes, of ages between 15 and 50 years
and of up to 50.8mm (2 in.) diameter was studied. It was found that
in the galvanized iron pipes, the diameter is reduced and both the
roughness and the losses coefficient increase with age. The increase in
roughness is substantial and follows a parabolic law instead of the
linear one usually found in the literature. For the copper pipes there is
no appreciable change in these variables with age. The apparition of
local narrowness or stenosis, due to age may cause considerable
increase in the losses coefficient and eventually even close the pipe
completely.
V.V.Nageswara Rao and G.K.Viswanadh (2004), reported the
variation of hydraulic loss coefficient in PVC pipes. The various
dissolved particles like salts, present in water, generally tend to
deposit on the walls of the pipes. Such salt deposits may strengthen
over a period of time, thus hindering the flow as well as the friction
factor. In the case of PVC pipes, it is observed that as the usage
increases, the diameter of the pipe is getting reduced because of the
deposits that are taking place along the inner walls of the pipes. The
deposition of salts again influenced by the factors such as quality of
water (presence of salts), velocity of flow, diameter of the pipe,
temperature changes etc., The Hazen-William coefficient may be
varying with the age of the pipe as the deposits are increasing within
32
the pipe. The H-W coefficient for new PVC pipe may be observed as
150 and for 10 years aged pipe, it was 133 and 6 years aged pipe it
was observed to be 142.
2.3 FACTORS INFLUENCING CORROSION SCALES IN WATER
DISTRIBUTION PIPES
In recent times, internal corrosion in water pipes is one of the
major contamination sources of tap water. Internal corrosion in
distribution networks has a close relation with the quality of water
flowing through the pipe. From the experimental investigation, it was
found that Dissolved Oxygen (DO) is the most critical parameter on
the corrosion of metal pipes. Internal corrosion of water pipes could be
effectively prevented by reducing DO concentration. The corrosion
products on the surface of copper pipe were appeared as light brown
and/or blue green color.
Corrosion rates for Galvanized Steel Pipe (GSP), Carbon Steel Pipe
(CSP) and Ductile Cast Iron Pipe (DCIP) were decreased to 72%, 75%
and 91% by reducing DO concentration from 9±0.5 mg/lit to 2±0.5
mg/lit respectively. Dissolved Oxygen significantly influences the
corrosion properties of water pipes. In DCIP, red rust (ferric oxide) was
visually observed on the surface of the pipe in the condition of 9±0.5
mg/lit of DO after 10 weeks of experiment.
Corrosion rates of water pipes decreased as the concentration of
hardness increased. Calcium carbonate deposited in the interface of
pipe surface and water. The deposition layer played a role to hinder
33
the electrochemical reaction between metallic pipes and water.
However, the continuous injection of calcium carbonate would result
in the formation of thick scale in the metallic pipes, which cause
serious increase of pressure.
Since pH is related with the formation of scale on the internal
surface of water pipe and metal solubility in water, pH adjustment has
been used as a corrosion control strategy of water pipes. (Haeryong
Jung, Unji Kim et al., 2009)
According to Lasheen MR et al., (2008), some factors influences
lead and iron release from drinking water pipes. It has been observed
by them that, the effect of stagnation time, pipe age, pipe material and
water quality parameters such as pH, alkalinity and chloride to sulfate
mass ratio on lead and iron release from different types of water pipes
used in Egypt. The different types of pipe materials used are Poly Vinyl
Chloride (PVC), Poly Propylene (PP) and Galvanized Iron (GI). Low pH
increases lead and iron from pipes. Lead and iron release decreases
with pH and alkalinity increases. It also increases with increasing
chloride to sulfate mass ratio in all the pipes. Lead and iron release
increases and then this release decreases with time. In general, GI
pipes showed to be the most effected by water quality parameters and
the highest iron release. PVC pipes are the most lead releasing pipes,
while PP pipes are the least releasing.
34
Cerrato JM, Reyes LP et al., (2006), studied the effect of PVC and
Iron materials on Mn (II) deposition in drinking water distribution
systems. The influence of Iron and PVC pipe materials on the
concentrations of soluble iron and manganese and the composition of
scales formed on PVC and Iron pipes were determined. Total Fe
concentrations were highest in water from iron pipes. Water samples
obtained from PVC pipes showed higher total Mn concentrations and
more black color than that obtained from iron pipes.
Manganese was incorporated into the iron tubercles and thus not
readily dislodged from the pipes by water flow. The PVC pipes
contained a thin surface scale consisting of white and brown layers of
different chemical composition. The brown layer was in contact with
water and contained 6% manganese by weight. Mn composed a
greater percentage by weight of the PVC scale than the iron pipe scale.
The PVC scale was easily dislodged by flowing water.
The various water quality parameters such as pH, temperature,
total dissolved solids and time of water circulation were all having an
effect on the migration of lead, tin and other metal stabilizers from
unplasticized polyvinyl chloride pipes. Water temperature did not
affect the migration of vinyl chloride monomer, unless it was raised to
high values (i.e.,450C).
35
2.4 EFFECTS OF BIOFILM/SCALE FORMATION IN PIPES
The formation of biofilms in drinking water distribution networks is
a significant technical, aesthetic and hygienic problem. Biofilms on
pipe walls in water distribution systems are composed of bacteria in a
polymeric matrix which can lead to chlorine demand, coliform growth,
pipe corrosion and water taste and odour problems.
According to Karloren et al., (2007), the formation of biological
films on the inner surface of the pipe not only affects the
characteristics of the surface, such as the roughness coefficient, but
may also promote the accumulation of organic and inorganic particles
because of its adhesive properties. These particles, in turn, can
contribute to sediment buildup, thus increasing the roughness
coefficient of the pipe surface. For clean 150 to 200mm PVC pipes, the
average value of Manning’s roughness coefficient is around 0.01, and
it was found to be independent of water depth, slope and pipe
diameter. The development of a biofilm on the internal surface of a
sewer pipe significantly affects the roughness coefficient. For small
diameter PVC pipes, covered with biofilm the minimum velocity and
shear stress to carry 90% of sand is around 0.55m/s and 1.4 N/m2
respectively. For the range of slopes studied, between 0.1% and 0.5%,
and for pipe diameters between 150 and 200mm, the shear stress
required to move particles of a given size is independent of slope and
pipe diameter.
36
Maggy NB Momba and N Makala (2004) compared the effect of
various pipe materials on biofilm formation. This investigation
indicated the colonization of all test pipe materials (PVC, uPVC,
MDPE, Cement and Asbestos cement) by coliforms and heterotrophic
plate count bacteria within 20min under chlorination treatment.
Statistical evidence showed that the generic type of pipe materials
greatly influenced the density of bacteria in the water system.
Cement-based materials (cement and asbestos) support less fixed
bacteria than plastic-based materials (PVC, uPVC and MDPE). No
significant difference in attached bacterial counts was found between
the generic types of pipe materials. The time factor also cannot be
ignored in determining the effect of pipe materials on biofilm formation
in potable water distribution systems. This study, therefore,
recommends the use of cement and asbestos cement pipe for the
distribution of chlorine-monochloramine treated water.
2.5 FRICTION FACTOR INVESTIGATIONS ON PIPE FLOW
The determination of frictional head losses is an important
engineering factor in the design of pipe networks that affects total cost
as well as the hydraulic balance of the network. Pipe sizing in a
network is dependent on the magnitude of allowable friction losses
calculated by the pipeline designer. Operating cost is inversely affected
by pipe diameter. As the pipe diameter increases for a given flow and
pressure head, frictional head losses per unit length decrease, thus
lowering the pump energy requirement. On the other hand, as the
37
diameter decreases the head losses per unit length increases and
hence the pump energy requirement is high.
2.5.1 Hydraulic Friction Factors for Pipe Flow
An over or under estimation of frictional head loss caused by water
flow in pipes will result in the selection of larger or smaller pipe and
pump sizes than actually needed, thus affecting total cost and
pressure distribution in the network. Different friction head-loss
equations used to estimate the energy loss in a certain pipe give
significantly different estimates depending on pipe size and water flow
rate. In this study friction factors were developed to be used with the
Hazen-Williams (H-W) and the Manning (Mn) equations for Cast- iron
and PVC pipes that give similar calculated head losses to that of
Darcy-Weisbach (D-W) equation. The H-W friction factor varied from
134 for 25 mm ID PVC pipes to 150 for 1050 mm ID PVC pipes and
from 111 for 25 mm ID Cast-iron pipes to 135 for 1050 mm ID cast-
iron pipes. Similarly, the Mn friction factor increased with diameter
and varied from 0.0083 for 25 mm ID PVC pipes to 0.0102 for 1050
mm ID PVC pipes and from 0.0099 for 25 mm ID Cast-iron pipes to
0.0113 for1050 mm ID cast-iron pipes. These friction coefficients
should provide satisfactory results (Kamand 1988).
2.5.2 Simple and Accurate Friction Loss Equation for Plastic Pipe
Engineers are continually searching for simpler calculation methods
that do not compromise accuracy. Some simple methods of calculating
pipe friction loss has been found, but the more complex Darcy-
38
Weisbach equation is the most universally accepted. The insertion of
the Blasius friction into the Darcy-Weisbach (D-W) equation results in
a combined equation with the following advantages.
1. It is theoretically sound and dimensionally homogeneous. Both
the Blasius and D-W equations have theoretical bases.
2. It is very accurate for plastic pipe when Reynolds numbers are
less than 1, 00,000. The Reynolds number limit is nonrestrictive
for irrigation-system design using pipes smaller than 80mm.
3. It is conveniently written in readily available terms: flow rate,
length and diameter.
4. It can be easily corrected for viscosity changes directly or by
referring to the included table of correction factors.
(Von Bernuth R.D.,1990)
2.5.3 Effect of Bed-Load Movement on Flow Friction Factor
The presence of bed-load movement increases the flow friction
factor. For pipe flows, the friction factor is larger for flows with bed-
load movement than those without bed-load movement at the same
Reynolds number. For clear-water flow in pipes without bed-load
movement, the measured relationship between the flow friction factor
and the Reynolds number fit the theoretical relationship well. All flows
studied were in the hydraulically smooth region. The increase in
friction factor in flows with bed-load movement is attributed to the
collision of sediment particles in motion, which consumes additional
39
energy in the flow. The increment in the flow friction factor can be
represented with the relative flow friction factor, f/fc. Its value depends
on the volumetric bed-load concentration and the size of the particle.
For a certain sediment size, the larger the concentration, the bigger
the increase. For a given concentration, the bigger the size of the
sediment, the bigger the increase. An empirical equation was fitted for
calculating the friction factor for flows with bed-load movement. If the
bed-load concentration, C, the diameter of the bed-load size, d, and
the flow friction factor of the equivalent flow without bed-load
movement is known , the flow friction factor can be estimated using
the proposed equation.( T.Song, Y.M. Chiew and C.O.Chin,1998)
2.6 FRICTION FACTOR STUDIES ON DOMESTICALLY USED PVC
PIPES
The friction factor in the case of PVC pipes was studied initially by
Von Bernuth and Tanya Wilson in 1989. The study of uniform pipe
flow condition needs two different investigations. (1) The flow velocity
and shear stress distribution in the cross section and (2) the flow-
resistance law i.e., the estimation of the f, friction factor of Darcy-
Weisbach uniform flow equation. In the past these two investigations
were separately developed although they are conceptually linked. If
the flow velocity distribution is known, the mean flow velocity can be
evaluated by integration and as a consequence the flow resistance law
can be deduced (Marchi 1961).
40
The flow-resistance law has been theoretically deduced for some
cross sectional shapes (circular and very wide rectangular) and under
specific boundary conditions because in these cases the velocity
profile is known. For example, the Prandtl-Von Karman theory of
turbulence can be applied to deduce the velocity profile, which has a
logarithmic shape; by integrating this logarithmic distribution for a
circular cross section of a smooth pipe, the Prandtl flow-resistance law
is obtained (Nikuradse 1933).
In the past the incomplete knowledge of the flow velocity profile
encouraged the empirical analysis of the friction factor measurements.
Empirical relationships like Blasius, Hazen-Williams and Manning’s
equations (Kamand 1988, Von Bernuth 1989) were developed. Some of
the relationships, characterized by a simple mathematical form,
neglect the variability of the friction factor with the flow conditions
and as a result, they only allow accurate head loss estimates for given
ranges of velocity and pipe diameter.
Blasius (1911) made a critical survey of the then existing
experimental results and arranged them in dimensionless form in
accordance with Reynolds law of similarity. The following empirical
equation called Blasius formula, was established f = c/Rem
where ‘c’ is the Blasius coefficient and ‘m’ is the exponent, which is
valid for the frictional resistance of smooth pipes of circular cross
section. According to his result, the dimensionless coefficient of
41
resistance in a pipe is a function of the Renolds number only. It is
found that the Blasius formula is valid in the range of Reynolds
number ≤ 105
Explicit relationships based on the Colebrook-White formulation
are available (Swamee and Jain 1976) and they are not more
complicated than the Blasius relationship (1913). However for
Reynolds numbers Re less than 105, the simple Blasius equation,
which allows accurate friction factor estimates for small diameter
plastic pipes (Watters and Keller 1978; Von Bernuth and Wilson 1989)
can be used (Schlichting 1979).
Von Bernuth and Wilson (1989), by using the experimental data of
Urbina (1976), Paraquiema (1977) and Norum (1984) assumed a
constant 'm' exponent of the Blasius (1913) equation equal to 0. 25;
the ‘c’ coefficient is variable and according to these experimental
results for PVC pipes, it is in the range of 0.281 to 0.345. The
curvature of the theoretical Prandtl equation (1935) on a graph log f
versus log Re can be filled by using an exponent 'm' of Blasius
equation variable with Re. The variability of 'm' can be explained by
the relationship between the velocity profile and the flow-resistance
law. Schlichting (1979), by using Nikuradse's (1933) data, showed that
by applying a power law velocity distribution with an exponent n=l/7,
and 'm' value equal to 0. 25 are obtained. According to Monin and
Yaglom (1971), because a simple relationship exists between 'm' and
'n', the Blasius equation (1913) has a theoretical basis.
42
A question to be resolved is whether the friction factor estimate for
irrigation design for PVC pipes should account for the deposits
accumulated in pipes which are used domestically for the flow of
ground water, since the variation of friction factor if any, can
significantly affect the head loss estimate. In particular, an
experimental study is needed to verify whether for a given pipe
diameter and discharge, the variation of friction factor with time of
usage can be explained by the formation of deposits and its
roughness.
In the case of domestically used PVC pipes of small diameters,
the Blasius equation is a good predictor for the estimation of
friction factor. In this particular study, the m exponent of Blasius
equation is taken as 0. 25 and the variation of the multiplicative
constant is noted.
Basically in the present study, the friction factor investigations as
well as head loss calculation equations were formulated. The main
aspect of deposition has not been taken into the preview of the study.
As mentioned earlier in this work, the deposition thickness, type of
slats and their characteristics, and rate of development of deposit
thickness mainly influenced by the water quality parameters, time of
running, pressure in the pipes, temperature etc. If a relationship is
established between the age and other parameters the equations
resulted above would be more useful. The same data may be collected
for artificially deposited pipes. Equivalent roughness coating materials
43
can be chosen and applied on the inner walls of the PVC pipes. The
friction data can be obtained for uniformly increasing deposit
thickness and at the same time higher Reynolds numbers can be
attained by using longer diameter pipes.( V.V. Nageswara Rao,2003)
2.6.1 Variation of Blasius Coefficient (c) in PVC pipes
According to V.V.Nageswara Rao (2003), In the case of
domestically used PVC pipes of small diameters, the Blasius
equation is a good predictor for the estimation of friction factor.
In this particular study, the ‘m’ exponent of Blasius equation is
taken as 0. 25 and the variation of the multiplicative constant is
noted.
(A) Variation of 'c' with the diameter
It can be clearly observed that when the area of cross section
of the flow reduces the head loss vis-a-vis the friction factor
increases. The variation of ‘c’ value which is used to calculate the
friction factor has been follows a 4th degree polynomial equation
given as follows:
c = 3. 4896D4 - 30.404D3 + 98.21 D2 - 139.31 D + 73.528
in which D is the diameter of the pipe in centimeters.
(B) Variation of 'c' with deposit thickness
It is very important to find the friction factor when the deposit
thickness is increased, because as discussed in the previous
section, the friction factor as well as the 'c' value increases with
44
the thickness of the deposit in the PVC pipe. The deposit thickness
has been observed to be increased to 1 mm in the case of 26 mm
pipe when they are used for 11 years. This deposition is gradual
and the relation has been given as follows:
c = 24.547 t2 - 2.031 t + 0.3195
in which’t’ is the thickness of the deposit formation in cm.
The deposit thickness is increased to 0. 49 mm after 3 years of
usage in the case of 20 mm pipes. 16mm pipe for a total period of
8 years was considered with two pipes in between this period with
3 years and 5 years also has been taken. In all these three cases
the thickness of the deposition has been measured. The total
accumulation is observed to be 0. 4 mm during this 8 year period
and followed a trend line relating ‘c’ and ‘t’ in the following form:
c =164.3 t2- 3.28 t+ 0.4
in which ’t’ is the thickness of the deposit formation in cm.
(C) Variation of 'c' with age of the pipe
When the pipe is domestically used depending on the quality
of ground water the deposits are formed inside the pipes. When
the head loss calculations are required it is necessary to have an
idea of the value of c for different conditions because this value is
varying with respect to diameter, thickness of the deposits etc.,
Instead, if a direct relation exist between the age of the pipe and 'c'
it can directly be calculated by knowing simply the age of the pipe.
45
Even the same thing can be substituted in the head loss equation
and the head loss can directly be calculated with the known age of
PVC pipe. The Blasius constant c is given by following equation:
c = 0. 0075 A + 0. 3067
in which A is the age of the pipe in years.
In an area where the similar conditions of pressure diameters
and the quality of water prevails this equation gives very accurate
results since the correlation coefficient is very high (0. 991). In the
case of 20 mm diameter PVC pipes the relationship between the
age(A) and coefficient 'c' is given by the equation:
c = 0. 0047 A +0. 326
In 16 mm diameter pipes the relationship between the age (A)
and the value of 'c' coefficient is obtained to be slightly complex
and follows a 3 degree polynomial equation which can be written
as:
c =0.0004 A3- 0.0026A2+0. 166 A +0.342
In all the cases, the friction factor calculated by the above Blasius
coefficients and the friction factor measured in the experiments are to
be nearly close to the line of agreement.
2.6.2 Variation of Friction Factor in PVC pipes
The data obtained for various diameter pipes and time of usages
has been analyzed for the calculation of friction factor by using
Darcy-Weisbach equation. These friction factor values are drawn to
the respective values of Reynolds numbers and relationship is
46
obtained in the form of a power law by taking the exponent 'm' as 0.
25 in the Blasius equation.
In the case of new pipes, it is clearly evident from the literature that
the data follows the Blasius equation in the calculation of friction factor.
By fixing the value of 'm' as 0. 25 the value of constant 'c' has been
analyzed for various diameter pipes in new condition. In the present
study it has been observed that the constant 'c' is slightly different from
that of the values suggested by Von Bernuth R.D, and Tanya
Wilson(1989) and going well with the value suggested by
Bagarellow et al.,(1995).
It has been observed from the study that the friction factor is
increasing with the time of usage for each diameter of the pipe. It is
obvious that when the deposits are formed the diameter decreases and
hence the head loss increases. More over the deposition roughness
also influences the friction factor. In the case of 26 mm pipes because
of the deposits the diameter of the pipe is gradually decreasing from
26mm to 24mm of internal diameter in a total period of usage of 11
years. The variation of' ‘c' for calculating the friction factor in
deposited pipes varying from 0. 304 for new pipe to 0. 39 for 11 years
used pipes with values in between.
For 20 mm pipes the value of 'c' varies from 0. 326 for new pipes to
0.345 for 3 years used pipe. Similarly the 'c' values in the case of 16
mm pipes changes from 0.342 to a value of 0. 524 for 8 years used
pipe. It is clearly evident that the friction factor increases steeply in
47
the case of smaller diameter pipes as the time of usage is increasing.
This is mainly because of the reduction in the diameter of the pipe.
Even there is a slight change in the reduction of diameter, the friction
factor changes rapidly (V.V. Nageswara Rao, 2003).
2.6.3 Hazen-Williams coefficient for PVC pipes
Trevor C. Hughes and Roland W. Jeppson (1978), reported the
field measured friction losses in three one-mile sections of small
diameter PVC pipe which had been in service for 10 years. Hazen—
Williams and Darcy-Weisbach equations were examined to
provide a framework for comparing Hazen-Williams Coefficients
recommended by pipe manufacturers to those obtained by the field
measurements. The conclusion is that the Hazen-Williams
Coefficient of 150 recommended by most PVC pipe manufactures is
too high for the diameter-velocity combinations encountered in
rural dead-end small diameter lines. The measured coefficients
averaged 133 which is close to that predicted by superimposing
H-W coefficients on the Moody Diagram from which the friction
factor for the Darcy-Weisbach equation is obtained. In this
study head loss measurements were taken along three
approximately one mile sections of pipe which consisted of 12m
(40 feet) lengths of PVC with glued joints. The three sections
included 1617m (5385 feet) of 60mm (2.5 inch) nominal diameter,
1333m (4440 feet) of 50mm (2 inch) and 1465m (4880 feet) of
40mm (1.5 inch) pipe all had been in service for 10 years. The
48
actual inside diameters were 67.4mm (2. 655 inches), 54. 6mm (2.
149 inches) and 43. 7mm (1.720 inches) respectively. The tests
were conducted in the ranges of Reynolds numbers between
14,350 to 41,100 and the estimated Darcy f values were between
0.0213 to 0.0280. Pipe sections approximately 0.6m (2 feet) long
were cut from the three PVC pipe diameters during installation of
master meters. The visual inspection of pipe sections cut from
these three lines appeared to be identical to new pipe in terms of
smoothness and total absence of deposition or corrosion. This had
been one of the very earlier studies on friction of PVC pipes after several
years of usage. But in this study it had been observed that the
deposition was absent inside the pipe.
2.7 MODELING STUDIES ON PIPE FLOW
There has been an extensive effort in the last decade to develop
theoretical models for fluid flow problems in pipes. In recent years
there has been success in the prediction of output results in the pipe
flow. In this section some modeling studies on flow through pipes have
been reviewed.
B.Bros (1999) presented a numerical model for the description of
fluid flow, and suspended and bed-load sediment transport. Density
effects are included in the momentum (Reynolds) equations and in the
turbulence (k and €) equations. Changes in bed levels are calculated
from sediment continuity, and the finite-element grid is adapted to the
geometry. The flow at a surface mounted cylinder in a steady flow is
49
predicted in good agreement with experiments. Scour calculations are
performed for a cylinder in a steady flow with its underside placed at
the level of the original flat bed. Predicted scour at a pipeline in steady
flow is in good agreement with laboratory measurements reported in
the literature. The routine use of numerical models in solving scour
problems for simple as well as complicated flow and bed conditions
will require physical insight and more modeling effort as well as
computer power.
M.Rashidul Islam and M.Hanif Chaudhry (1998), developed a
dynamic computer model to predict the constituent concentration at
various points and nodes in a pipe network under slowly varying flow
conditions. Unlike presently available steady-state or extended-period
water quality models, this model uses the slowly varying flow
conditions and is more appropriate for real-life applications to typical
distribution systems. In this model, a steady-state analysis of the pipe
network is performed first to determine the initial conditions. Then,
the dynamic governing equations, including the inertial effects, are
solved subject to the proper boundary conditions. The model is
applied to two typical pipe networks to simulate the transport and
decay of chlorine, and the results are compared with another model
which uses the standard extended-period simulation technique. The
results are found to be in good agreement at the beginning of the
simulation. The model may be used to analyze the propagation and
decay of any substance with a first-order reaction rate.
50
C.J.Kerr et al., (1999), predicted the relationship between the pipe
material and biofilm formation in a laboratory model system. The aim
of this study was to compare biofilm accumulation and heterotrophic
bacterial diversity on three pipe materials- cast iron, Medium Density
Poly Ethylene (MDPE) and unplasticised Poly Vinyl Chloride (uPVC) -
using a laboratory model system run over a short period (21days) and
a longer period (7months). Number of biofilm heterotrops accumulated
more slowly on uPVC and MDPE than on cast iron at the low flow rate
used in this study. The suggestion that the biofilm on uPVC is less
stable than on MDPE or cast iron requires further investigation.
R. Banki et al., (2007), presented a rigorous mathematical model
for the prediction of wax deposition in pipelines for laminar flow. In
the last 15 years, there have been a number of studies in modeling of
the deposition processes in flow lines. There was an extensive effort in
the last decade to develop theoretical models for wax deposition
calculations. There are two main processes that affect the deposition
of wax in flow lines: 1) heat transfer and 2) species flow. The transient
deposition of each component is calculated from the solution of the
coupled momentum, energy and species balance equations, and a
thermodynamic wax precipitation model at the local level. A detailed
numerical algorithm to solve the mathematical model is also provided.
51
2.8 APPLICATIONS OF ANN TECHNIQUE IN HYDRAULIC AND
HYDROLOGIC ENGINEERING MODELS
C.S.V.Subrahmanya Kumar (2009), applied an Artificial Neural
Networks technique for the performance evaluation and comparison of
different types of hydrological models. The study focuses on the
application of Artificial Neural Networks (ANNs) to formulate various
hydrologic models. The results of the study reveal that ANN technique
has the capability to capture the non-linear dynamics of a hydrologic
system. It can be used as a versatile tool to solve diverse problems.
The predictions of hydrological variables in the study are to the
reasonable accuracy. The study revealed that the error convergence
was quicker in the models.
Artificial Intelligence (AI) techniques are widely used in solving
various problems of Civil Engineering such as water management and
modeling. One of such AI techniques is an Artificial Neural Networks.
The Artificial Neural Network (ANN) is an information processing
system that roughly emulates the behavior of a human brain by
replicating the operations and connectivity of biological neurons.
ANNs have an ability to capture a relationship from given patterns and
hence this makes them suitable for employment in the solution of
large-scale complex problems such as pattern recognition, nonlinear
modeling, classification, association and control.
52
The ANN training is fundamentally a problem of nonlinear
optimization, which minimizes the error between the network output
and the target output by repeatedly changing the values of ANN’s
connection weights according to a predetermined algorithm. Error
back propagation is by far the most widely used algorithm for
optimizing feed forward ANNs.
For the application of ANN one may refer to Murry (1995). Most of
the ANNs applications related to the fields of water engineering were
presented in Negm et al. (2001). Therefore, only few related studies
were briefly reviewed here for the benefit of the reader.
The ASCE Task Committee (2000) investigated the role of artificial
neural networks (ANNs) in hydrology. According to the basic review
paper prepared by the committee, ANNs are gaining popularity, as is
evidenced by the increasing number of papers on this topic appearing
in hydrology journals, especially over the last decade. In terms of
hydrologic applications, this modeling tool is still in its nascent stages.
The practicing hydrologic community is just becoming aware of the
potential of ANNs as an alternative modeling tool. The paper is
intended to serve as an introduction to ANNs for hydrologists. Apart
from descriptions of various aspects of ANNs and some guidelines on
their usage, this paper offers a brief comparison of the nature of ANNs
and other modeling philosophies in hydrology. A discussion on the
strengths and limitations of ANNs brings out the similarities they have
with other modeling approaches, such as the physical model.
53
The role of ANNs in various branches of hydrology has been
examined here. It is found that ANNs are robust tools for modeling
many of the non-linear hydrologic processes such as rainfall-runoff,
stream flow, ground-water management, water quality simulation, and
precipitation. After appropriate training, they are able to generate
satisfactory results for many prediction problems in hydrology. A good
physical understanding of the hydrologic process being modeled can
help in selecting the input vector and designing a more efficient
network. However, artificial neural networks tend to be very data
intensive, and there appears to be no established methodology for
design and successful implementation. For this emerging technique to
find application in engineering practice, there are still some questions
about this technique that must be further studied, and important
aspects such as physical interpretation of ANN architecture, optimal
training data set, adaptive learning, and extrapolation must be
explored further. The merits and limitations of ANN applications have
been discussed, and potential research avenues have been explored
briefly.
H. M. Nagy et al., (2002), used an Artificial Neural Model to
estimate the natural sediment discharge in rivers in terms of sediment
concentration. This is achieved by training the network to extrapolate
several natural streams data collected from reliable sources. The
selection of water and sediment variables used in the model is based
on the prior knowledge of the conventional analyses, based on the
54
dynamic laws of flow and sediment. Choosing an appropriate neural
network structure and providing field data to that network for training
purpose are addressed by using a constructive back-propagation
algorithm. The model parameters, as well as fluvial variables, are
extensively investigated in order to get the most accurate results. In
verification, the estimated sediment concentration values agree well
with the measured ones. The model is evaluated by applying it to
other groups of data from different rivers. In general, the new
approach gives better results compared to several commonly used
formulas of sediment discharge.
Murat Alp and H.Kerem Cigizoglu (2007), provided two simulations
for suspended sediment load by applying ANN methods using hydro
meteorological data. Estimates of sediment load are required in a wide
spectrum of water resources engineering problems. The nonlinear
nature of suspended sediment load series necessitates the utilization
of non-linear methods for simulating the suspended sediment load. In
this study ANNs are employed to estimate the daily total suspended
sediment load on rivers. Two different ANN algorithms, the Feed-
Forward Back-Propagation (FFBP) method and the Radial Basis
Functions (RBF), were used for this purpose. The neural networks are
trained using rainfall flow and suspended sediment load data from the
Juniata Catchment, USA. The simulations provided satisfactory
simulations in terms of the selected performance criteria comparing
well with conventional multi-linear regression. Similarly, the
55
simulated sediment load hydrographs obtained by two ANN methods
are found closer to the observed ones again compared with multi-
linear regression.
Gokmen Tayfur and Vijay P. Singh (2006), developed an ANN and
Fuzzy Logic (FL) models for predicting event based rainfall runoff and
tested these models against the Kinematic Wave Approximation
(KWA). A three-layer feed-forward ANN was developed using the
sigmoid function and the back propagation algorithm. The FL model
was developed employing the triangular fuzzy membership functions
for the input and output variables. The fuzzy rules were inferred from
the measured data. The measured event based rainfall-runoff peak
discharge data from laboratory flume and experimental plots were
satisfactorily predicted by the ANN, FL, and KWA models. Similarly, all
the three models satisfactorily simulated event-based rainfall-runoff
hydrographs from experimental plots with comparable error
measures. ANN and FL models also satisfactorily simulated a
measured hydrograph from a small watershed 8.44 km2 in area. The
results provide insights into the adequacy of ANN and FL methods as
well as their competitiveness against the KWA for simulating event-
based rainfall-runoff processes.
Song Songbai and Cai Huanjie (2006), used an ANN model to
assess sustainable utilization of regional water resources. First,
stochastic method was used to form enough assessment indexes
sequence. Second, by using back-propagation networks, assessment
56
indexes sequences and their assessment grade values were considered
as input layer units and output layer units to develop ANN
assessment model. Finally, the model was applied to assess
sustainable utilization of water resources in Hanzhong basin and
Huaihe watershed in China. The assessment results showed that the
model was practical and convenient to use. It is feasible to use the
model to assess the sustainable utilization of regional water resources.
M. Nasseri et al., (2008), developed neural networks to simulate
the rainfall field and Back Propagation (BP) algorithm coupled with
Genetic Algorithm (GA) used to train and optimize the networks. The
technique will be implemented to forecast rainfall for a number of
times using rainfall hyetograph of recording rain gauges in the Upper
Parramatta catchment in the western suburbs of Sydney, Australia.
Results of the study showed the structuring of ANN network with the
input parameter selection, when coupled with GA, performed better
compared to similar work of using ANN alone.
The friction factor of an open channel flow is generally affected by
the Reynolds number and the roughness conditions, and can be
decided by laboratory or field measurements. During practical
applications, researchers often find that a correct choice of the friction
factor can be crucial to make a sound prediction of hydraulic
problems. In this paper, a three-layer ANN was set up to predict the
friction factors of an open channel flow, with the Reynolds number
and the relative roughness as two input parameters.
57
The Levenberg–Marquardt (LM) learning algorithm was employed to
train the model by using laboratory experimental data, and the
trained network was tested by a single set separated from the rest of
the data and a good correlation between the experimental and
predicted results has been obtained. Finally, the ANN simulated
results were compared with the calculated results obtained by the
empirical formula and both comparisons showed that the ANN model
can be used to predict the non-linear relationship between the friction
factor and its influencing factors correctly, once enough samples are
provided. The successful application proved that ANN model can be
used in engineering practice as a convenient and effective method,
and those traditional hydraulic problems which are mostly based on
laboratory tests can be analyzed by ANN modelling.
According to A.H. Lobbrecht and D.P. Solomatine (1999), ANN and
Fuzzy Adaptive Systems (FAS) appeared to be efficient alternatives for
using optimal control algorithms in real-time tasks. The results
obtained for the control of water levels show that ANN and FAS are
able to replicate the behaviour of the Aquarius control component at
one-two time steps (1 hour) ahead with the accuracy in the range 90-
97%. This gives the possibility to replace the slow computational
components by the fast-running trained intelligent controllers and
thus to simplify the use of Aquarius in the real time control tasks.
Abdeen, M. A. M.(2001), introduced the use of ANN technique to
model and predict the hydraulic characteristics of the water surface
58
profile in natural open channels. Synthetically generated data was
used in the study to show the applicability of using ANN technique for
modeling natural open channel behavior. The study implemented ANN
technique to predict flow depths and average flow velocities along the
channel reach when the geometrical properties of the channel cross
sections were measured or vice versa. The results of this study show
that ANN technique is capable, with small computational effort and
high accuracy, of predicting the different hydraulic characteristics of
irregular open channels.
Abdel-Azim M. Negm and Mohamed A. Shouman used ANNs to
model the characteristics of the submerged hydraulic jump formed
over roughened bed with regular staggered roughness elements.
Multilayer feed forward neural network with back propagation
learning algorithm is used to model the characteristics of such
submerged hydraulic jump. A network of size of 5-5-4 is found
suitable for this purpose with 3300 iterations and sigmoidal (tansh)
activation function. The results of the trained, verified and tested ANN
model are compared to the experimental measurements. Also, results
from previously developed models based on statistical methods are
compared to results of ANN model. The training data set is used to
develop multiple linear regression models for the jump characteristics
in terms of the input variable to the network. The MLR models are
tested using the validation and test data sets are then used to
compare the results of the ANN model. The results indicated that the
59
ANNs are powerful tools for modelling of submerged hydraulic jumps.
Through the sensitivity analysis that conducted using the ANN model,
the most contributing variables to characteristics of the submerged
jump over rough bed are specified.
The work reported by C.S.V.Subrahmanya Kumar and
G.K.Viswanadh (2008), is aimed to predict the reservoir stage using an
Artificial Neural Network Approach. The potential of ANN model which
belongs to the class of data-driven approaches for predicting daily
reservoir stage in Osmansagar reservoir, Hyderabad has been
presented in this study. An ANN using Levenberg-Marquardt
algorithm with back propagation is adopted for the study. The ANN
stage prediction model was developed using stage data for a period of
76 years. The model was trained using the data for a period of 50
years. The trained ANN was then tested for 26 years and the results
were compared with the observed values of the corresponding period.
The results suggests that a three layer feed forward ANN having single
hidden layer with two neurons can effectively be used to predict the
reservoir stage. The Nash coefficients of efficiency of the ANN model
were found to be 0.971 and 0.987 during training and testing. The
correlation coefficient between the observed and computed stage
series is 0.986 during training and 0.994 during validation. The study
revealed that the ANN model developed gives the best prediction of
reservoir stage.
60
2.9 NEED FOR PRESENT STUDY
In the present scenario of water distribution systems, the PVC
pipes are in abundant usage for water supply, in wide range of variable
conditions. Because of the presence of salts in the water that they carry,
the inner walls of the PVC pipe is getting deposited with the salts that
are present in the water, in due course of time. The extent of deposition
has been observed to be high when the water is having more amount of
dissolved solids (salts). The thickness of the deposits inside the walls
of the pipe may vary depending on various factors such as water
quality, time of service of the pipes, flow pressure in the pipe,
temperature of water etc. Because of this deposition the friction inside
the pipe will also be changing since initially the PVC pipe is observed
to be smooth inside when compared to the surface of the salt coat
which is being formed with usage.
Though, considerable amount of research work is reported
regarding experimental studies on flow-resistance law and friction
factor studies in PVC pipes, the causes for the deposit formation and
the amount of thickness of deposits have not been reported in
literature. Also the reasons for deposit formation have not been
mentioned. If the probable reasons are investigated and modeled, the
criterion to minimize the formation of these deposits can be suggested.
In this context it becomes necessary to investigate the reasons for this
deposit formation.