Wave Speed Calc For Water Hammer Anal
Wave Speed Calc For Water Hammer Anal
Wave Speed Calc For Water Hammer Anal
In order to accurately solve the water hammer problem Para resolver en forma precisa el problema de golpe de ariete
using the Method of the Characteristics MOC is usando el Método de las Características MC es necesario que
necessary to fulfil with the so−called Courant condition el número de Courant Cn = 1 en cada tubería del sistema. El
which establishes mandatorily that Cn = f(a) = 1 in each valor de Cn depende de la velocidad de la onda a, cuyo valor
pipeline of the system, where a is the wave speed. The value depende a su vez de las propiedades del fluido (densidad,
of Cn is dependant of a whose value depends in turn on módulo de compresión) y de las características físicas de cada
the fluid properties (density, bulk modulus) and physical tubería (módulo de elasticidad, diámetro, espesor, condición
characteristics of each pipeline (elasticity modulus, de apoyo). Debido a que las redes de distribución de agua
diameter, wall thickness, supporting condition). Because generalmente tienen muchas tuberías diferentes, y por tanto,
water distribution systems usually has many different muchas velocidades de onda distintas, cumplir con Cn = 1 en
pipes, and therefore, many different wave speeds, it can cada tubería se torna una tarea muy difícil, más aún cuando
be said that fulfil with Cn = 1 in each pipeline is a very la solución mediante el MC necesita un paso de tiempo Δt
difficult task, more when the solution by MOC needs a común para todas las tuberías del sistema. Una forma de
common time step Δt for all pipe sections of the system. A solución a este problema es aplicar el método de ajuste de
way of solution to this problem is applying the method of la velocidad de la onda que consiste en modificar el valor de
the wave−speed adjustment that involves modifying the a en cada tubería en un cierto porcentaje hasta obtener Cn=
value of a in each pipe section in a certain percentage up 1. Con esto se garantizan resultados óptimos en términos
to obtain Cn = 1. With this procedure optimum results are numéricos, pero ¿es posible decir lo mismo en términos
guaranteed in numerical terms, but it is possible to say físicos?. La pregunta que se plantea es: ¿qué parámetros
the same in physical terms? The question which arises is: dentro de la fórmula de a deben (o pueden) ser cambiados sin
what parameters within the formula of a must (or can) be exceder los valores característicos del material componente
changed without exceeding the characteristic values of de la tubería?. En este trabajo se muestran algunos casos
the component material of the pipes?. This work shows donde la modificación de a puede alterar significativamente
that in some cases the wave speed modification can la magnitud de los parámetros que definen su valor, dando
significantly alter the value of the parameters that define lugar a valores que pueden ser físicamente incompatibles,
a, leading to values that can be physically inconsistent, ficticios o sin aplicación práctica.
fictitious or without practical application.
Palabras clave: velocidad de la onda, golpe de ariete,
Keywords: wave speed, water hammer, Courant number número de Courant
86
Twyman, J. (2016). Obras y Proyectos 20, 86-92
pipes with a range of wave speeds and lengths (Karney related with the boundary conditions of specific devices,
and Ghidaoui, 1997). There are two strategies to deal with describe the phenomenon of wave propagation for a water
this problem. The first strategy is apply the method of the hammer event.
wave-speed adjustment MWSA where one of the pipeline
properties is altered (usually wave speed) to satisfy Wave speed
exactly the Courant condition. The second strategy is For water, without presence of free air or gas, the more
interpolating between known grid points allowing Courant general equation to calculate the water hammer wave speed
numbers less than one. At first glance the MWSA appears magnitude in one-dimensional flows is (Watters, 1984):
simpler because is non-dissipative and non-dispersive and
in theory only consists in modify the value of the wave
speed in a certain percentage to meet Cn = 1. Nevertheless,
this procedure distorts the physical characteristics of the
problem (Ghidaoui and Karney, 1994). In other words, with a the wave speed, K the volumetric compressibility
changing a involves altering, in physical terms, the modulus of the liquid, ρ the liquid density, e the pipe wall
value of one or more of the parameters that are part of its thickness, E the pipe elasticity modulus (Young); ψ a
formulation such as fluid density or the elastic modulus factor related with the pipe supporting condition which can
of the constituent element of the pipe. More clearly, the be calculated from general expressions (see Table 1) being
modification of a in numerical terms involves altering the case 2 more conservative from an engineering point of
the initial physical conditions of the system, leading to a view. Equation (3) supposes that:
solution that may be correct in numerical terms (to meet • Pipe has a thin internal wall, condition which is met
Cn), but incorrect in physical terms because the problem is when D/e > 40 (Watters, 1984) or when D/e > 25
solved using parameters with unreal magnitudes. (Wylie and Streeter, 1978).
Governing equations of transient flow • Pipe remains full of water during the transient event;
that is, no separation of the water column is generated,
When analyzing a volume control it is possible to obtain a
which means that at all times the pressure is greater
set of non-linear partial differential equations of hyperbolic
than the vapour pressure.
type valid for describing the one-dimensional 1D transient
flow in pipes with circular cross-section (Chaudhry and • Water has small air content, so that the magnitude of
Hussaini, 1985): the wave speed may be assumed constant.
• The pressure is uniform across any section of the pipe.
It means that inertial forces associated with radial
motion of the fluid are negligible (Skalak, 1955).
Equation (3) includes Poisson’s ratio effect but neglects
the motion and inertia of the pipe. This is acceptable for
where the partial differential equations (1) and (2) rigidly anchored pipe systems such as buried pipes or
correspond to the continuity and momentum (dynamics), pipes with high density and stiffness, to name only a few.
respectively. Besides, H is the piezometric head, a is the Examples include major transmission pipelines like water
wave speed, c = (gA/a), where g is the acceleration of distribution systems, natural gas lines and pressurized and
gravity, A is the pipe cross-section, Q is the fluid flow and surcharged sewerage force mains. However, the motion
R = f/2DA, f is the friction factor (Darcy-Weisbach) and D and inertia of pipes can become important when pipes
is the inner pipe diameter. The subscripts x and t denote are inadequately restrained (unsupported, free-hanging
space and time dimensions, respectively. Partial differential pipes) or when the density and stiffness of the pipe is small
equations (1) and (2), in conjunction with the equations (Ghidaoui et al., 2005).
87
Twyman, J. (2016). Wave Speed Calculation For Water Hammer Analysis.
Obras y Proyectos 20, 86-92
Table 1: Expressions for ψ according to the pipe supporting dispersion) in the solution (Goldberg and Wylie, 1983).
condition (Watters, 1984; Pierre, 2009) The MOC stability criterion states that (Watters, 1984):
Case Pipe supporting condition
Pipe anchored at the upstream end only
88
Twyman, J. (2016). Obras y Proyectos 20, 86-92
practitioners to think of interpolation as a numerical device = 1000 kg/m3. The Young’s elasticity modulus E is the
with only numerical side effects. In general, all common relationship between the force increment and the unitary
interpolation procedures result in numerical dissipation strain (Martínez and Azuaga, 1997). E has the same value
and dispersion, and they considerably distort the original for a tension or compression, being a constant as long as
governing equations. The interpolation procedures the force does not exceed a maximum value called elastic
effectively change the wave speed (Ghidaoui and Karney, limit (Hooke’s law). The formula for calculating the
1994). In summary, interpolation fundamentally changes elasticity modulus is:
the physical problem and must be viewed as a nontrivial
transformation of the governing equations. Because this
topic is beyond the scope of this paper, more information
will not be included here. In the following paragraphs, the where E is the modulus of elasticity, F is the force, A0 is the
main parameters of the wave speed in (3) will be briefly surface (area) where the force is applied, ΔL is the length
analyzed, showing their characteristic values. variation and L0 is the initial length. Typical values of E for
some materials are shown in Table 2.
Compressibility is the property of a fluid to change its
volume due to the pressure (Del Valle, 2010). For problems Table 2: Typical values for E (Larock et al., 2000)
involving the effect of water hammer is necessary to Material E, Pa
take into account the compressibility of water, which is Steel 2.077 ∙ 1011
inversely proportional to its bulk modulus of elasticity and Copper 1.1 ∙ 1011
is defined mathematically as: Bronze 1.0 ∙ 1011
Asbestos cement 2.3 ∙ 1010
Fiberglass reinforced 9.0 ∙ 109
PVC 2.8 ∙ 109
Polyethylene 8.0 ∙ 108
where v is the specific volume and P is pressure. The bulk
modulus of elasticity K is:
When a sample of material is stretched in one direction it
tends to get thinner in the other two directions (Figure 1).
The Poisson’s ratio is the ratio of the relative contraction
strain (or transverse strain) normal to the applied load. It
The equation (7) represents the relative change in a fluid can be expressed as:
volume per unit of applied pressure. The negative sign is
because as the pressure increases, the volume decreases
and vice versa. The ε units are the same for pressure. At a
temperature of 20°C and atmospheric pressure (1 bar) the where u is the Poisson’s ratio, εt is the transverse strain and
bulk modulus of water is K = 2.07 ∙ 109 Pa. The density of εL is the longitudinal or axial strain. Strain can be expressed
water is the weight of the water per its unit volume: as:
with ρ the density, m the fluid mass and V the fluid volume. where dL is the change in length and L is the initial length.
The fluid density is function of pressure and temperature For isotropic materials the Poisson’s ratio is in the range of
(especially in gases), it increases with increasing pressure 0 to 0.5 (Greaves et al., 2011). Table 3 shows some typical
and it decreases with major temperature. At atmospheric values of u.
pressure and temperature of 4°C the water density is ρ
89
Twyman, J. (2016). Wave Speed Calculation For Water Hammer Analysis.
Obras y Proyectos 20, 86-92
Assumptions
In order to obtain Cn = 1, the following assumptions will be
taken account: i) it is not possible to modify Δt, L and N;
ii) water parameters such as density ρ or bulk modulus K
are known and unalterable; iii) there are not availability of
Figure 3: Variation of wave speed versus Poisson’s ratio
90
Twyman, J. (2016). Obras y Proyectos 20, 86-92
Conclusion
The MWSA distorts the physical characteristics of the water
hammer problem. Due to this, it is recommendable that in
Figure 4: Variation of wave speed versus modulus of elasticity the process of discretization (Δx, Δt) of the pipe network,
(k/ms2 = Pa)
necessary to solve the water hammer in pipe networks by
MOC, before deciding to apply the MWSA to obtain Cn = 1,
Even though, in the analyzed example u should vary 120%
the analyst must see if the final values adopted to calculate
to get a value of 0.66 (out of range) for a = 1289 m/s and
a are consistent and appropriate, both in numerical and
therefore Cn = 1. Figure 4 shows that +5% variation of
physical terms. Otherwise it would solve a very different
the wave speed to obtain Cn = 1 implies to increase the
problem originally raised with implications for all stages
value of elasticity modulus up to E = 3120 GPa, which
of design or verification of the system. Before changing
corresponds to a variation of 50% of its initial value, being
the value of a, it is important to check the implications
out of range because the existence of a material more rigid
of changing its magnitude. At this point, it is important to
than steel and equally efficient and useful as a constituent
know what parameters of its formulation are known and
element of a pipe is unknown. Another interesting point
can be considered as unalterable (pipe length, diameter or
from Figure 4 is that the allowed variation of a falls
wall thickness) and check what of the other parameters can
between 0 and −11%, which only allow incorporate pipes
be modified by analyzing its variation range and level of
of copper (E = 1100 GPa) or bronze (E = 10000 GPa) into
reality.
the model because their rigidity. That is, by including less
rigid pipes into the model (PVC for example, with E = 2.8
GPa), the variation of a would stay outside the allowable
References
Chaudhry, M.H. (1979). Applied hydraulic transients. Van
range ±15% recommended by authors such as Wylie and
Nostrand Reinhold, New York
Streeter (1978). The modification of u or E values leads to
Cn = 1, assuring the optimum results in numerical terms. Chaudhry M.H. and Hussaini M.Y. (1985). Second–order
Nevertheless, in this case the application of the method accurate explicit finite–difference schemes for waterhammer
of wave−speed adjustment had a cost in physical terms analysis. Journal of Fluid Engineering 107(4), 523 - 529
because Cn = 1 was obtained from parameters (E or u) Del Valle, V. (2010). Fluidos. Apuntes editorial Universidad
out of range. In general, analysts tend to forget such cost Tecnológica Nacional, Tucumán, Argentina
because the MWSA has been recommended in the pipeline
Ghidaoui M.S. and Karney B.W. (1994). Equivalent differential
literature (Karney and Ghidaoui, 1997), without giving
equations in fixed−grid characteristics method. Journal of
further details about its physical limitations. Finally,
Hydraulic Engineering 120 (10), 1159 - 1175
another point is that physical limitations of the (4) and
(5) show up to where it is possible modify a in order to Ghidaoui, M.S., Zhao, M., McInnis, D.A. and Axworthy, D.H.
avoid an out of range value. For example, in the case (2005). A review of water hammer theory and practice. Applied
of u (Figure 3), the range of permitted variation of a is Mechanics Review 58(1), 49 - 76
very restricted, between −1 and +1%. For the case of E
Goldberg, D.E. and Wylie, E.B. (1983). Characteristics method
(Figure 4) such range varies between 0 and −10%. This
using time–line interpolations. Journal of Hydraulic Engineering
means that the range of variation of a between −15 and
109(5), 670 - 683
91
Twyman, J. (2016). Wave Speed Calculation For Water Hammer Analysis.
Obras y Proyectos 20, 86-92
Greaves, G.N., Greer, A.L., Lakes, R.S. and Rouxel, T. (2011). Skalak, R. (1955). An extension of the theory of water hammer.
Poisson’s ratio and modern materials. Nature Materials 10(11), Tech. Report No. 15, Columbia University
823-837
Twyman, J., Twyman, C. y Salgado, R. (1997). Optimización
Karney B.W. and Ghidaoui M.S. (1997). Flexible discretization del método de las características para el análisis del golpe de
algorithm for fixed–grid MOC in pipelines. Journal of Hydraulic ariete en redes de tuberías. XIII Congreso Chileno de Ingeniería
Engineering 123 (11), 1004 - 1011 Hidráulica, Universidad de Santiago de Chile, 53 - 62
Larock, B.E., Jeppson, R.W. and Watters, G.Z. (2000). Hydraulics Watters, G.Z. (1984). Analysis and control of unsteady flow in
of pipeline systems. CRC Press, Boca Raton, Florida, USA pipelines. 2nd edition, Butterworth−Heinemann, USA
Martínez, P. y Azuaga, M. (1997). Medición del módulo de Wood, D.J., Lingireddy, S., Boulos, P.F., Karney, B.W. and
elasticidad de Young. Apuntes laboratorio IV, Departamento de McPherson, D.L. (2005). Numerical methods for modeling
Física, UBA transient flow in distribution systems. Journal of the American
Water Works Association 97 (7), 104 - 115
Pierre, B. (2009). Pressure waves in pipelines and impulse
pumping: physical principles, model development and numerical Wylie, E.B. and Streeter, V.L. (1978). Fluid transients. McGraw–
simulation. Doctoral thesis, Norwegian University of Science Hill, USA
and Technology, Trondheim
92