You are on page 1of 9

STATE OF THE ART OF COMPARATIVE ANALYSIS BETWEEN HYDRAULIC

MODELS SUBJECT TO THE EFFECT OF PRESSURE ON FLOW LEAKS


1
ZAMORA ATAO Frank Renzhon, 2POMACANCHARI VALDEON Pamela,
3
VILLAR ROJAS Wendy

SUMMARY
The effect of pressure on flow leaks is due to pressure fluctuations, which causes the increase in
velocity, as well as the flow rate. To be clear, this effect in a simple pipe is much more feasible to
calculate having the mathematical formulation, however, in a pipe network, it becomes very
cumbersome, due to the complication of the network where nodes are already present, therefore the
help of models is necessary, where its application will allow to calculate and detect these flow leaks.
The model in pipe networks, which has to do with the application of pressure and flow, is the
hydraulic model.
This article presents an approach to the development and comparative analysis between hydraulic
models subject to the effect of pressure on flow leaks in pipe networks.

Keywords: flow leakage, pressure, hydraulic models.


INTRODUCTION

One of the essential fluids for our subsistence is water. Due to the increase in population, engineering,
in its ability to seek solutions, so that the population can be supplied with this essential liquid, has
designed pipe networks, which are immersed in various inconveniences, such as breakdowns, failures
and leaks. flow. Water leaks can cause discomfort in the economy of the population, since the
information on the real consumption of water and therecorded by measurement systems are not the
same. These originate from various causes, but the one that stands out the most is that caused by
pressure fluctuation.
There are different models for the detection of flow leaks in pipe networks, however, the model
that is related to flow leaks caused by pressure is hydraulic. The application of this model allows
certain programs to locate and detect flow leaks in pipe networks. To develop the hydraulic
modeling of the pipe networks, the static, quasi-static and dynamic models are taken into account,
these are the ones that will actually determine the effect of pressure on flow leaks, due to their
characteristics that present with respect to the flow velocity in time.
In the article carried out, a state of the art will be developed with respect to the comparative
analysis of the aforementioned models, since these will determine the effect of pressure on flow
leaks. During the research we used a series of works carried out by experts on the subject that
addressed theoretical aspects.

THE EFFECT OF PRESSURE ON FLOW LEAKS

A. Mathematical model of the effect of pressure in simple pipes


There are several factors that influence a pipe break, below are mathematical models, of two very
important factors, in pipe break.
a. Influence of pressure on rupture: In this case, the overpressure due to water hammer is taken
into account.
The water hammer is the sudden change in pressure, due to a change in speed that occurs quickly.
This is one of the main causes of pipe failures.
J. Twyman (2018) in his study of "Water hammer in a pipe network due to the rapid closure of a
valve" presents the following model:

 Equation for a simple node


The term node (or junction) indicates the location where the edge sections meet. For frictionless multi-
pipe joining, the equation(one) (Karney and McInnis 1992, quoted by J. Twyman, 2018):

𝐻𝑃𝑡+∆𝑡 = 𝐶𝑐 − 𝐵𝑐 ∗ 𝑄𝑒𝑥𝑡 (1)


Where t = current simulation time, = junction height, and are known constants, y = external flow
(positive when assumed to be from junction). The equation𝐻𝑃𝑡+∆𝑡 𝐶𝑐 𝐵𝑐 𝑄𝑒𝑥𝑡 (one) lets calculate 𝐻𝑃𝑡+∆𝑡
at the junction for the number of pipes found at the node.
 Equation for the pond
If the hydraulic losses between the pond and the pipe joint are negligible, the nodal height equal to
the level of the pond surface, then the following expression is valid (Karney and McInnis, 1992,
quoted by J. Twyman, 2018):

𝑄𝑒𝑥𝑡 = (𝐶𝐶 − 𝐻0 − 𝐵0 ∗ 𝑄𝑒 ) ∗ (𝐵𝑐 + 𝐵𝑐 )−1 (2)


Where Ho = pond height at the beginning of the time step, Bo = ∆t / 2Ar, with Ar = pond cross-
sectional area, y = initial external discharge. It should be noted that, if the constant Bo is zero, the
height of the tank is independent of the external discharge. The term Ho can be a constant or a known
function of time (for example, waves in a pond). Equation (2) can be used to represent constant head
reservoirs, storage tanks, and simple equalization tanks.𝑄𝑒 .
b. Mathematical model of the relationship between pressure and flow in a leak.
LJ Montoya and RD Montoya (2012)mention that there is a series of equations for the leakage flow
rate, which presents a potential relationship as shown in equation (3) for continuous pressures in the
system and different types of failure, where both the coefficient and the exponent of the relationship
presented in equation (3) are dependent on pressure, since pressure can cause elastic changes and in
the shape of the orifice.

𝑄𝑓 = 𝑐𝑃 𝑏 (3)

Where is the leakage flow rate, c is the emitter coefficient of the leak, P is the pressure, b is the
leakage exponent.𝑄𝑓

B. Mathematical model of the effect of pressure in pipe networks


According to Nava and Delgado (2018), the leakage of the flow in a pipe network is a phenomenon
that causes the waste of water, that is, it is the amount of water that escapes from a pipe supply system,
these leaks depend on many factors such as bad connections between pipes, corrosion whether
external or internal, excessive load, high pressures, external temperature, the presence of breaks or
some structural failure in the pipe, where the most important factor is pressure, among more high is
the pressure the flow in the leak will be greater and vice versa. According to studies carried out,
approximately 50% of water is wasted due to these causes.
Thus, the factor causing these leaks to occur is the pressure and at the same time, this is related to the
flow, because the relative drop along the pipe is twice the relationship between the flow that is leakage
and the total of the transfer in the conduction, it should be emphasized that if there is an increase in
flow it does not necessarily mean that there is a leak; Each signal has to be carefully analyzed.
In a pipe network system, the leak appears at any point in the network structure, for this reason
mathematical networks are formulated aided by simulation models that define the variation of flow
and pressure, these models allow the realization of a case-by-case analysis If this is used to establish
a degree of reliability or assumption of risks, it should be noted that a model is not an exact reflection
of reality since some aspects such as fluctuations due to errors in the equipment and instantaneous
consumption than the model are not considered. not plasma. An increase in flow rates together with
a decrease in pressure implies a detection.
Martínez, (2000) (as cited in Rondán, 2016) mentions that in order to take into account the sensitivity
of leaks to pressure variation, leaks should be treated as a pressure-dependent demand, such as It has
been described in the previous section. The simplest approach to modeling leaks in a hydraulic model
is to use equation (4) for flow emitters. This equation, included in the EPANET library and any
commercial program that uses its calculation algorithm, makes it possible to simulate the outflow
through a nozzle or orifice discharging to the atmosphere; so they can also be used to simulate a leak
in a pipe connected to a node.
Then equation (4) is determined in the following expression:

𝑄𝑓,𝑗 = 𝑐𝑗 𝑃𝑗 𝑏 (4)

Where is the leakage flow at node j, is the emitter coefficient at node j, and which depends on the size
and shape of the leak hole, is the pressure at node j, b is the leakage exponent.𝑄𝑓,𝑗 𝑐𝑗 𝑃𝑗

From the given equation, new variations and improvements are proposed, according to the variables
that are given.
Almandoz et al, (2003) (cited by Rondán, 2016) propose to assign the leaks in the model using a
global emitter coefficient, adjusted to each node by means of a weighting factor of the weighted length
of the pipes connected to the node with respect to the length total distribution network.
𝑏 𝑏
𝑗
𝑄𝑓 = 𝑐 ∑𝑖=1 𝐿̅𝑖 𝑃𝑖 (5)

𝐿
𝐿̅𝑖 = 𝐿 𝑖 (6)
𝑇

Where in equation (5) is the leakage flow, is the global leakage coefficient, is the length weighting
factor for node i, is the pressure at node i, b is the leakage exponent, is the number total knots in the
model. In equation (6) it is 50% of the length of the pipes connected to node i, it is the total length of
pipes in the model.𝑄𝑓 𝑐𝐿̅𝑖 𝑃𝑖 𝑏𝑗 𝐿𝑖 𝐿 𝑇

Giustolisi et al (2008) (cited by Rondán, 2016) propose equation (7) to separately consider bottom
leaks and breaks along the pipeline. And it uses the equation for flux emitters with the theoretical
leakage exponent (0.5) for breaks, and the expression proposed by Germanopoulos (1985) for bottom
leaks.

𝑄𝑓,𝑖𝑗 = 𝛽𝐿 𝐿𝑖𝑗 (𝑃̅𝑖𝑗 )𝑏 + 𝑐𝐿 (𝑃̅𝑖𝑗 )0.5 (7)

Where is the emitter coefficient corresponding to the sum of the coefficients of the existing breaks
along the tube.𝑐𝐿
Tabesh et al (2009) (cited by Rondán, 2016) propose a global leakage coefficient calculated from
equation (9), and that uses the value of the leakage flow estimated at the instant of minimum night
flow from the equation (8). The leakage assignment at each node is carried out using equation (10),
based on the emitters equation defined in EPANET (Rossman, 2000).

𝑄𝑓,𝐶𝑀𝑁 = 𝑄𝐶𝑀𝑁 − 𝑄𝑢𝑠𝑜 𝑛𝑜𝑐𝑡𝑢𝑟𝑛𝑜 (8)


𝑄𝑓,𝐶𝑀𝑁
𝐶𝑓 = 𝑏𝑗 𝑏𝑃 𝐿𝑖𝑗
(9)
∑𝑖=1(∑𝑗=1 𝑃𝑖,𝐶𝑀𝑁 𝑁 )
2

𝐿𝑖𝑗 𝑏
𝑏
𝑄𝑓,𝑖 = ∑𝑗=1
𝑃
𝑐𝐿 𝑃𝑖 (10)
2
Where is the leakage flow at the instant of minimum night flow (CMN), it is the minimum night flow
injected into the distribution network, used to estimate the real losses in distribution systems of
continuous operation, it is the flow associated with legitimate uses during the CMN instant, it is the
global leakage coefficient, it is the total number of nodes in the model, it is the number of pipes
connected to node i, it is the length of the pipe j connected to node i, it is the pressure in the node i at
CMN instant, b is the leakage exponent, is the leakage flow rate at node i for all instant
𝑄𝑓,𝐶𝑀𝑁 𝑄𝐶𝑀𝑁 𝑄𝑢𝑠𝑜 𝑛𝑜𝑐𝑡𝑢𝑟𝑛𝑜 𝐶𝑓 𝑏𝑗 𝑏𝑃 𝐿𝑖𝑗 𝑃𝑖,𝐶𝑀𝑁 𝑄𝑓,𝑖 𝑡, is the pressure in the node i for all times 𝑃𝑖 𝑡.

Rondán (2016) deduces that the simplest approach consists of assuming the same emitter coefficient
for all nodes of the model in equation (11). Wu et al (2011) (cited by Rondán, 2016) mention that this
approach does not consider the physical characteristics of the distribution network, such as the length
or the structural condition of the pipes, the number of service connections.
𝑏
𝑄𝑓 = 𝑐 ∑𝑖=1
𝐷
𝑃𝑖 𝑏 (11)

Where is the leakage flow in the distribution network, c is the global emitter coefficient, is the number
of demand nodes of the model, is the pressure at node i, b is the leakage exponent.𝑄𝑓 𝑏𝐷 𝑃𝑖

HYDRAULIC MODELS OF PIPELINE NETWORKS


In order to control these leaks, it is necessary to implement leak control programs, since they
contribute to the prevention of problems that arise in the pipes and at the same time allow the external
damage that may arise to be anticipated.

For the detection of leaks, a correct handling must be carried out by means of continuous pressure
and flow readings with precise meters, theoretically it must be taken into account that the area of
some failures varies with pressure.

López et al(nineteen ninety five)mention that the mathematical model of a network will allow the
case-by-case simulation referred to above to be carried out on it and that it will serve to establish the
degree of reliability or assumption of risks that a given system has when less from a theoretical point
of view. The modeling of a network can be performed using a static, quasi-static or rigid model
(Cabrera et al. 1994, cited by López et al, 2016), however, in recent decades the dynamic model has
also been implemented.
A network is analyzed when, based on certain physical characteristics, it is well (the data of the
elements that make up the system m such as pipes, tanks and valves are known in detail) and the
demand that it must satisfy (including in the same their spatial and temporal distribution) it is intended
to know their answer. A response that is generally specified in pressure levels, also distributed in
space and time. The analysis, of the weather conditions hardly if they change, should be carried out
with the static model and, otherwise, with dynamic models. And even though, in some way, any static
analysis is ideal (everything changes over time), the study being much simpler, when the network
operating functions vary slowly, it is always the most convenient to perform.
A. Static model
Also called steady-state models. They are models that collect the state of a system at a certain moment
in which the parameters and variables are fixed (consumption, levels, etc.), without taking into
account the possible variations that they may suffer (Vegas, 2012).
Almandoz and Cabrera et al (2006) mention that in this section a quick review of the basic foundations
of the mathematical formulation of the problem of analysis in permanent regime of complex hydraulic
systems such as water distribution networks to large urban centers is carried out. . It is studied through
the so-called static models or models of hydraulic balance. This type of model allows predicting the
response of the system (internal flows that circulate through its elements, as well as the pressures in
the nodes), knowing their configuration and characteristics (behavior and interconnection of the
different elements), under an operating situation of determined operation (consumption status and
conditions defined at the power points).
The main elements that are part of a hydraulic network and that have relevance in the problem of
static analysis are: pipes, tanks, valves and pumps. All these elements are considered static in this
type of analysis and, therefore, their behavior is described through algebraic relationships.
Thus having the hydraulic equilibrium balance, which according to Almandoz and Cabrera et al
(2006), the mathematical formulation of the static analysis problem consists of establishing a system
of equations so that the number of unknowns equals the number of independent equations. Different
formulations have been proposed to solve the problem, which fundamentally differ from each other
in the treatment of the system of equations. Typical difficulties arise from non-linearity and the large
number of equations that generally have to be solved.
Next, there is a system of equations:
∑𝑛𝑗=1 𝑄𝑖𝑗 = 𝑆𝑖 ; 𝑁 = 𝑁𝐽 + 𝑁𝐹 𝑒𝑐𝑢𝑎𝑐𝑖𝑜𝑛𝑒𝑠
{ (12)
𝐻𝑖 − 𝐻𝑗 = 𝑓(𝑄𝑖𝑗 ) ; 𝐿 𝑒𝑐𝑢𝑎𝑐𝑖𝑜𝑛𝑒𝑠

Where the index j refers to all nodes directly connected to node i. We will consider an internal flow,,
positive when the fluid circulates from node i to node j (divergent flow from the node) and negative
in the opposite case (flow converging at the node). For external flows,; the opposite criterion is
adopted so that consumption is always negative. While N corresponds to the total number of nodes
in the system. Identified the L lines and the N nodes of the system (N = NJ + NF, where NJ is the
number of connection nodes and NF is the number of nodes of fixed head). On the other
hand,𝑄𝑖𝑗 𝑆𝑖 𝐻𝑖 − 𝐻𝑗 indicates the energy difference (piezometric heads) and represents the flow
function.𝑓(𝑄𝑖𝑗 )

B. Quasi static model


In water distribution systems, it is found that the principle of ensuring the supply at all times leads to
a very fine mesh and an oversized network characterized by stagnant zones and long retention periods
(Cohen, 1993 cited by Almandoz and Cabrera et al, 2006). It happens that the consumptions (and,
therefore, the flows that circulate through the pipes of the system and the pressures) do not remain
constant, but instead present important daily fluctuations. On the other hand, the levels of the water
levels in the regulation tanks suffer variations due to the instantaneous balance of the mass of water
in the system. The pumping stations start or stop (or there is variation in the rotation speed and / or
number of pumps in operation), and the valves are maneuvered to properly operate the system.
According to Vegas (2012), the quasi-static models or models in extended period (example,
EPANET), where the inertia of the fluid in the system is neglected and whose analysis model for each
instant of time is a static model, where the variation between one instant and the next, it comes from
updating parameters such as levels in tanks, demands at nodes, etc., between one calculation instant
and the next.
Almandoz and Cabrera et al (2006) mention that the quasi-static model superposes different static
snapshots, it is very useful in the analysis of the evolution of the behavior of pressure hydraulic
systems on the long-time scale, with a view to obtaining the management and the most appropriate
design (especially the regulation volumes of the tanks). To do this, the evolution of the variables
associated with the network (pressures, levels and flows) must be simulated for a long period of time
(24 hours, for example), complying with the mass balance in the system in accordance with the
consumption forecasts.
Taking into account the analysis of the mathematical formulation of the problem, the equation that
governs the filling / emptying of a tank whose level is admitted varies slowly over time (for which
inertia can be neglected) takes the form:
𝑑𝑧𝑖
𝐴𝑑𝑡 𝑑𝑡
− ∑𝑛𝑗=1 𝑄𝑖𝑗 +𝑆𝑖 = 0 (13)

Where Ja is the cross-sectional area of the reservoir i, z is the water level in it, n is the number of
pipes that converge (diverge) in the reservoir, to which an internal flow rate Q, is associated, is the
contribution (or demand) to the deposit, which in this type of model can be a known function of time
- autonomous non-dynamic boundary condition.𝐴𝑑𝑡 𝑆𝑖
C. Dynamic model
Also called transient regime models, water hammer or distributed parameter model. In these models
the temporal variation of the system parameters intervenes, which logically also induces the temporal
variation of the network variables. The analysis of this type of model is often a discrete analysis, since
it is necessary to solve the equations of the system for each instant of time. They are applied in sudden
maneuvers of the analyzed system. These models have more terms than static analysis models when
proposing the characteristic equations system of each one of them; Likewise, it takes a greater
mathematical difficulty and time spent in its resolution (Vegas, 2012).
Almandoz and Cabrera et al (2006) mention that, when modeling the movement of fluids inside
pressure pipes, the "quasi-static" dynamic model does not include two especially significant aspects
that must be taken into consideration. when the changes to be modeled are faster. In particular the
inertia term and the elastic term. And so, when it is desired to study the pressure variations along the
network due to changes in demand or to the variation of water levels in the head reservoirs, it seems
clear that these variations are slow enough for the changes of speed and pressure that they induce in
the circulating water and in the nodes of the network are very discrete. And in consequence, the effects
due to the inertia of the water (it is a high-density fluid whose changes in speed imply a consumption
or contribution, depending on the sign of the acceleration, of energy to the system as a whole) and
the elasticity of the system ( when both the fluid and the pipe are compressed, they have the capacity
to store energy that they return as soon as the pressure drops) it is not necessary to include them in
the model. But of course, there are situations in which the flow changes are very important and the
model must include the two preceding effects (inertia and elasticity). This is the case of a blowout in
a pipeline, a sudden maneuver of a valve, the start or stop of a pump or, finally, the well-known
problem of pathogenic intrusion, with water entering the pipeline, due to a depression , who has
previously escaped. Depending on whether only one or both effects are included, we arrive at the two
traditional dynamic models that are presented below. The first one takes into account only one of the
two effects, the inertia of the water in the pipe, and is known as the rigid model (in the past it was
called the mass oscillation model) while the one that considers the two effects is the model elastic,
better known for water hammer.
To the mathematical modeling of the lines that each of these models makes, it is necessary to add the
boundary conditions of each case, some boundary conditions that include the disturbing source that
must be modeled in such a way that it rigorously reflects the generation of the disturbance and its
subsequent modification. The enormous variety of elements that we can find in practice (valves,
suction cups, pumps, boilers, etc.) makes the mathematical representation of the boundary conditions
one of the most complex aspects.
COMPARATIVE ANALYSIS BETWEEN HYDRAULIC MODELS SUBJECT TO THE
EFFECT OF PRESSURE ON FLOW LEAKS
This section shows a comparative analysis of the profiles obtained, as a function of the effect of
pressure on flow leaks, describing the alterations in the pressure wave when water leaks occur, as
well as the advantages and disadvantages between the exposed models. Next, the aforementioned
analysis will be detailed, as well as its application to the detection of flow leaks and why its relevance
in this study.

A. Effect of pressure on flow leaks


In the pipes there is constant flow movement, which is why it is subject to events such as pressure
fluctuations, especially when there are no valves to regulate the pressure increase, which can lead to
leaks in the pipes. This happens because as the pressure increases, the exit velocity through the orifice
through which the leak is caused increases, as well as the increase in the flow rate.
From the study of hydraulic models of pipe networks and mathematical models, it is observed that
the flow regime of the static model is a stationary or permanent regime, in which the velocity acting
on the flow is constant at a certain node time. A node in pipeline networks, therefore, there would not
be the case of a leak, although this fact does not occur in our reality because it is something ideal. As
for the quasi-static model, there is a stationary regime, however, moderate pressure fluctuations are
observed, therefore, the flows that circulate through the system pipes and the flow velocities do not
remain constant, this fact It is the most common according to our reality. On the other hand, the
dynamic model has a transitory or non-stationary regime,
B. Application to flow leak detection
From the study of hydraulic models of pipe networks and mathematical models, it is observed that
the quasi static model is the one that stands out the most because it is in line with our reality, that is
why software programs such as EPANET, WaterCAD, etc., make use of this model in the design of
pipe networks, in the analysis, modeling and management of pressure networks. For the flow leak
detection application, the program that has stood out the most for years is EPANET.
However, these models are not an exact reflection of reality either, since, although it can show an
analysis of the evolution of pressure and flow at each point, the measurements that are made in reality
would present a fluctuation due to errors in the devices and consumption. snapshots that the model
cannot reproduce, but it can make a diagram regarding the analysis of flow leakage in a pipe network.
C. Advantages and disadvantages between the exposed models
From the study of hydraulic models of pipe networks, it is observed that the static model can facilitate
its use in didactic problems, in some subject; However, if you want to apply according to reality, the
use of the quasi-static model is more acceptable, on the other hand, the dynamic model, just as the
static model does not correspond to reality.
CONCLUSION
From the authors reviewed, it is concluded that it is important first, to take into account the
mathematical models that are formulated, since these will allow the development of the applicative
models, in our case, a study of the hydraulic models in the pipe networks was made, focusing on
static, quasi-static and dynamic models; The theoretical investigation of these models, allowed us to
have a clearer approach in the comparative analysis of these, since the relationship of the flow velocity
in the pipe with respect to time in each model was different, this was the important factor to specify
the effect of pressure on flow leaks in pipe networks, as well as flow behavior.
BIBLIOGRAPHY
J. Twyman., "Water hammer in a pipe network due to the rapid closing of a valve" Construction
Engineering Magazine, vol. 33, no. 2,2018.
LJ Montoya and RD Montoya., "Effect of pressure on water leaks in a simple pipe system" Revista
Ingenierías Universidad de Medellín, vol. 11, no. 20, pp. 77-86, 2012.
Almandoz Berrondo, FJ, EÇA GUIMARAES DE ABREU, JD, Cabrera Marcet, E., Cobacho Jordán,
R., & Cabrera Rochera, E. (2005). "Analysis, design, operation and management of water networks
with EPANET" (VOLUME I, VOLUME II). Reference Manual Collection.
Nava, J., Delgado, X. (2018) Location of leaks in drinking water networks. Journal of the scientific
popularizer, 4 (1), 2869-2873.
Rondán Galán, E. (2016). State of the art in the calibration of hydraulic models. Leak modeling with
Epanet. Research Deposit of the University of Seville.
López, PA, Vela, AF, & Iglesias, PL (1995). Analysis of security in supplies and its application to
the detection of leaks. Water Engineering, 2 (3), 7-22.
Vegas Child, Ó. T. (2012). Tools to help the sectorization of water supply networks based on graph
theory applying different criteria. Institutional Repository of the Polytechnic University of Valencia.

You might also like