1. Introduction
Transient events caused by air pockets inside pipelines have gained attention in the last decades since pressure surges can originate catastrophic scenarios that can produce the failure of hydraulic installations [
1]. Most analysed scenarios assume the worst possible conditions, so the resulting air pocket pressure peaks represent the maximum values in water pipelines [
2,
3], which must be considered when selecting the pipe resistance class [
4]. This makes studying this problem of practical interest for water utilities [
5]. To avoid issues caused by entrapped air pockets, air valves are commonly installed at various points in the system [
6,
7]. These valves allow air to be expelled, reducing pressure surges in installations [
8]. It is crucial to note that air valves do not always provide the expected reliability and can sometimes cause severe problems when they are undersized or when filling operations are not performed according to recommendations proposed by the American Water Works Association (AWWA) [
9].
Despite the advantages of using air valves [
10], they also present several challenges [
11]. For instance, sizing and selecting air valves must be done carefully. Estimating the amount of expelled air is also challenging [
12,
13]. Oversizing, especially during air expulsion, can cause high overpressures, making it dangerous to select a valve that is either too small or too large [
4].
Another significant issue is the proper static and dynamic modelling of these valves. While some manufacturers provide characteristic curves in their catalogues, these data are often unreliable. Various laboratories and universities, including the Polytechnic University of Valencia, have conducted static tests on different valves, and the results generally show significant discrepancies from the manufacturers’ data [
14]. This can lead to critical problems due to incorrect selection based on unsuitable characteristic curves. The situation regarding dynamic characterisation is even worse since there is a lack of available information. Manufacturers overlook this aspect, and there is still much to be done from a research perspective.
Besides the difficulties in sizing air valves and the unreliability of manufacturers’ characteristic curves, other issues include the “dynamic closure” of these devices (an air valve closing prematurely due to a lifting effect on the float, leaving a dangerous air pocket inside water installations), the poor correlation between the nominal diameter of an air valve (usually matching the connection diameter) and its expulsion capacity, and inadequate or non-existent maintenance, which can lead to malfunction when an air valve is needed [
4]. Therefore, while air valves in hydraulic installations are highly recommended, they do not guarantee total safety. For the reasons stated above, these devices can cause more issues related to pressure surges than those they are intended to solve.
During the last decades, different mathematical models have been proposed. Martin (1976) [
15] proposed the first system for analysing two-phase flows in confined pipeline systems, establishing a differential equations system composed of the continuity and momentum formulations and the thermodynamic behaviour of the air phase. Liou & Hunt (1996) [
16] developed a rigid column model, applicable when high flow velocities and small pipe diameters are present. Izquierdo et al. (1999) [
17] investigated scenarios involving irregular pipe profiles with entrapped air using the rigid column model. This model considers a perpendicular air-water interface (piston flow) representing rapid transients where some pipe sections are filled with air. In contrast, others are entirely occupied by water.
Wylie & Streeter (1993) [
18] were the first to develop an elastic model that incorporates the elasticity of the fluid, focusing solely on single-phase flows. Zhou, Liu, and Karney (2013) [
19] extended this elastic model to analyse filling processes by combining the continuity and momentum equations for the water phase with the polytropic equation governing the behaviour of the air phase. This model was developed for both simple and irregular pipe cases, accounting for entrapped air pockets within the system. More recently, Zhou et al. (2020) [
12] studied, using a simple pipe, the effects occurring during rapid filling phenomena in an experimental setup measuring 8.862 m in length and 400 mm in internal diameter, with a 10-mm orifice added at the end to expel airflow.
The effects of blocking columns in confined hydraulic installations represent another critical issue that requires careful consideration. This is because the elasticity of air is significantly higher than that of water [
20], leading to scenarios where water can easily compress air columns, potentially causing pressure surges in hydraulic systems. Although the effects of blocking columns have been explored regarding various entrapped air pockets using both elastic and rigid models [
17,
19], the literature reveals a gap in addressing these issues in systems with undulating pipe profiles, particularly when air valves or orifices sizes are installed during filling operations.
This research presents a mathematical model for studying transient events with entrapped air pockets, including air valves. This study provides a valid model for analysing transients generated by
entrapped air pockets in pipelines with irregular profiles and
installed air valves [
21,
22]. The proposed model’s generalised equations, including the air valve’s behaviour, are presented. The main advantage of the proposed model over previous ones is that it is suitable for considering the effects of blocking columns and air valves. Furthermore, this system of equations is particularised for a specific case: an installation with two entrapped air pockets and an air valve installed at the highest point of a water installation with an irregular profile. The problem involves the presence of a second air pocket [
23]. A polytropic evolution was considered to characterise the behaviour of air pockets during transient events. A specific case is analysed using the obtained system of equations and its corresponding numerical resolution. The selected case study corresponds to a water installation located at the hydraulic laboratory at the Technical University of Valencia, Spain. Transient events are analysed considering two scenarios: first without air valves and then considering an air valve installed at the highest point in the pipeline.
The results for both situations are compared, observing the effect of air valves on the transient events. When the air pocket’s location coincides with the air valve’s position, the device allows free air expulsion, significantly smoothing the air compression process (reducing pressure peaks). However, when the water arrives and the air valve closes, the air pocket is again confined between two water columns with no escape to the atmosphere, leading to a new transient evolution similar to those studied without valves, generating the corresponding pressure peaks based on the conditions at the time of valve closure. Finally, this research analyses the same installation with different valve sizes. This allows for studying the effect of air valves’ sizes and characteristics (mainly the expulsion coefficient and outlet size) on various transient events.
2. Mathematical Model
This section shows the mathematical model’s development considering the air valves’ behaviour. In this sense, a pipeline with an irregular profile and an air valve at the highest point in the installation is considered. At this point, an air pocket is initially trapped between two water columns (
Figure 1).
The governing equations are presented below, considering the following assumptions: (i) water movement is simulated using the rigid water column model; (ii) a polytropic law is suitable for following the behaviour of entrapped air pockets; (iii) the air-water interface is considered perpendicular with the main direction of the pipe, which is applicable for pipelines with small internal diameters; (iv) a friction factor using steady flow conditions is considered; and (v) an isentropic flow is considered for the expelled air.
For a filling column, there are two equations:
1. Equation of the rigid model characterising the motion of the filling column driven by an energy source (
= upstream pressure in the pipeline;
= air pocket pressure)
where
= water velocity,
= friction factor,
= internal pipe diameter,
= difference elevation between two points of a water column,
= gravity term,
= water density,
= length a water column, and
= gravitational acceleration.
2. Position of the filling column
where the subscript 0 refers to an initial condition.
For an air pocket, there are three equations:
3. Evolution of the air pocket (assuming a polytropic evolution)
where
= air mass,
= polytropic coefficient, and
= air pocket volume. The subscript 1 refers to the condition of the air pocket. The polytropic coefficient varies from 1.0 (isothermal evolution) to 1.4 (adiabatic condition).
4. Continuity equation for an air pocket (considering that the air pocket density (
) and air density at the outlet section (
) are equal)
where
= air velocity and
= cross-sectional area of an air valve.
Considering that:
where
= cross-sectional area of the pipe,
= position of the water column, and
= water velocity of a blocking column.
Solving for the air density:
5. Equation modelling the behaviour of the air release valve
During the pipeline filling process, it is typical that air valves operate in the subsonic flow range (
< 19.55 m = 1.918 bar). The behaviour of expelled air in the subsonic region follows a simple expression that relates the expelled airflow
under normal conditions (
atm and 15 °C, that is, a density
kg/m
3) with the pressure inside a pipeline
. The corresponding equation is given by:
where
= expelled coefficient of an air valve and
= atmospheric pressure.
From this, the air expulsion velocity
:
where
= air density at atmospheric conditions (1.205 kg/m
3).
Equation (11) can be expressed as:
If a filling operation were a perfectly controlled process, it would proceed slowly, and air valves would operate in the subsonic range. However, when a pipeline filling occurs more abruptly, the pressure in hydraulic installations may exceed the subsonic flow limit (
); in this case, the expelled air is found in the sonic flow regime. The behaviour of an air valve in the sonic region fits well with a straight line, meaning the relationship between the expelled airflow under normal conditions
and the pressure inside in a pipeline
is linear. To extend the previous equation and thus have a valid expression for the sonic flow region, this line has the same slope as the one at the limit point (
) when the flow is subsonic. Therefore, the line should pass through the point
and
m. The slope of this line is:
Thus, the expression that simulates the behaviour of the expelled air when an air valve is operating in the sonic region (when
) is:
Expressing this line in terms of the air velocity
is as follows:
For a blocking column, there are two equations:
6. Equation of the rigid model characterizing the movement of a blocking column (
* is the pressure of the air pocket and
is atmospheric pressure)
where
= length of the blocking column and
= gravity term of the blocking column.
7. Position of a blocking column
Therefore, a system composed of 7 equations needs to be solved. The 7 unknown variables for the analysed problem are:
Two unknowns for the filling column:
Three unknowns for the air pocket:
Two unknowns for the blocking column:
This system of equations remains valid until the filling column reaches the position of an air valve. At that moment, the water pressure will cause the float to rise, closing the valve and nullifying the air expulsion velocity ().
In this new situation, the equation modelling the air valve’s behaviour (Equations (12) and (15)), depending on whether it operates in the subsonic or sonic zone, respectively, is removed from the system. Then, a system of 6 equations with 6 unknowns (
,
,
,
,
) needs to be solved. The continuity equation for the air pocket (Equation (9)) can be neglected because the computation of the air density
is no longer necessary; thus, Equation (3) becomes:
Equation (18) maintains the assumption of a polytropic evolution when an air valve closes and an air pocket is trapped. A water utility’s standard procedure for cleaning and maintaining water pipelines involves starting them up. This process begins when the pipeline is at rest because the regulating valves are closed.
4. Analysis of Results and Discussion
4.1. Practical Application
The proposed model is applied to study the filling of a system with four pipe branches of constant slope (), two air pockets (), and one air valve () installed at the highest point in the pipeline of irregular profile.
An experimental facility was configured at the Technical University of Valencia, Spain (see
Figure 8). This installation is fed using a pump, which was tested in the experimental facility. The pump curve is
. It discharges into the atmosphere in a reservoir. The pipeline has a total length (
of
, a diameter (
of
, and a friction factor (
of
. Once a steady-state regime is established, the flow rate is
(with a velocity of
). The steady state was reached at the end of the transient event, which was considered by numerical and experimental results.
For the practical application, the initial positions of the blocking columns and, consequently, the initial sizes of the air pockets are:
(,, )
(, , )
Once the pump is turned on, the discharge electro-valve is opened, thus beginning the pipeline’s filling operation.
4.2. Installation without Air Valves
Solving the corresponding system of equations yields the results depicted graphically in
Figure 9. This figure presents the pressure evolution of the two air pockets throughout the transient process. To measure the pressure pulses of the air pockets, transducers Tr3, Tr4, and Tr5 were used [
20]. At the onset of the transient event, air pocket 1 is positioned at the location of Tr5, and the peak value predicted by the proposed model is 18.577 m at 0.19 s, which closely matches the experimental measurement of 17.484 m. A similar correspondence is observed with air pocket 2 and transducer Tr3. It is crucial to emphasise that monitoring air pockets should consider the different transducers used. In this sense, the proposed model was used.
Figure 9 shows, in purple colour, the difference between the computed (continuous line) and measured (discontinuous line) position of transducer Tr5, where the proposed model can follow the air pocket pulses during the transient event. The proposed model is suitable for predicting extreme pressure pulses for both air pockets. Similarly,
Figure 10 shows the water velocity evolution of the three water columns.
Due to the small size of the studied installation, the transient process lasts only a few seconds. An adiabatic evolution was considered during the experimental stage since entrapped air pockets are rapidly compressed during the transient event, considering that there is no air expulsion.
Table 1 presents the times at which the three water columns reach the downstream end of the pipeline.
During the initial instant of the transient event, the air pockets compress notably (especially the first one) and reach their maximum pressures.
Table 2 presents these maximum values and their corresponding time occurrences.
4.3. Installation Using an Air Valve
This section presents the results considering an air valve located at the highest point of the installation (
), which is suitable for expelling air.
Table 3 presents the main characteristics of the air valve. An isothermal process was assumed for this analysis, as the air release valves expel air, reducing the likelihood of abrupt changes in the entrapped air pockets.
The air valve has a coefficient , which is suitable of expelling an airflow rate of approximately () with a differential pressure of ().
The numerical resolution yields the transient evolution shown in
Figure 11 (pressure of the two air pockets) and
Figure 12 (water velocities of the filling column and the two blocking columns).
It is observed that with the presence of the air vent, the second blocking column comes out of the installation at time s (30% later than filling without the air valve).
The maximum values of air pocket pressure heads considering the installation of an air valve are presented in
Table 4.
Initially, the position of the air valve coincides with the second air pocket, so the device is opened and begins expelling air until the corresponding blocking column arrives (). Throughout this time, the presence of the air valve produces a lower peak of pressure heads compared to when it is neglected. The explanation is straightforward: as the installation fills, the air pocket compresses, causing high-pressure peaks when no air valves are present. The air valve allows the expelling of airflow, preventing the air pocket from compressing as much.
When the water column arrives, the air valve closes and remains closed until another air pocket arrives (). The air pockets compress, usually with corresponding pressure peaks, since there is no air outlet through the air valve during this period. From the moment the air valve opens again, coinciding with the position of the first air pocket, the pressure in this pocket quickly decreases. The effect of this new opening on the other air pockets is relatively small. The pressure evolution in the first air pocket is smooth until the filling column reaches the pair valve position and closes again ().
Figure 13 depicts the flow rate and velocity of air expelled through the air valve during the filling procedure. It is observed that during the first opening of the air valve, the maximum expelled flow rate is
. When the maximum pressure reaches the second air pocket, the maximum flow rate is simultaneously expelled. On the other hand, when the air valve opens for the second time, the pressure in the first air pocket is much higher, consequently increasing the expelled airflow as well (
with
= 146.7 m/s).
4.4. Comparison between the Transient Event Considering and Neglecting the Air Valve
After separately analysing the pipeline’s filling without air valves and considering them, a comparison was performed, as shown in
Figure 14 and
Figure 15.
Table 5 shows the maximum value of air pocket pressure reached for the two analysed scenarios.
As expected, the first pressure peak is slightly reduced with the installation of the air valve (16.093 m versus 16.286 m). It should be noted that when the hydraulic installation starts, the filling column abruptly compresses the first air pocket. Still, the first water column reacts much more slowly due to its greater inertia. Therefore, the air valve has almost no effect on the pressure of the first air pocket during these times. Significant differences appear when the first blocking column moves significantly, compressing the second air pocket and causing air to be expelled through the air valve. The second pressure peak is much smaller, with 15.997 and 18.230 m values occurring at 0.695 and 0.500 s, respectively. When the air pocket reaches the air valve position and opens, the pressure drops to nearly atmospheric values. The air pocket begins to compress once more when the air valve closes again as the filling column reaches its position.
Table 6 presents the maximum pressures reached for the second air pocket in the two analysed situations (
Figure 15).
The effect of the air valve’s installation is evident from the beginning. From the start of the transient event until t = 0.460 s (the period during which the air valve is expelling air), the evolution of the air pocket pressure is much smoother, reaching a maximum value of 11.538 m, which is significantly lower than when the installation fills without the air valve ( 14.540 m). Once the air valve closes, the air pocket compresses again, generating the corresponding pressure peaks.
The installation of the air valve affects the velocity of the filling column. The first blocking column’s velocity slows down slightly after the beginning of the hydraulic event. However, the movement of the second blocking column is significantly slower when the air valve is installed, as illustrated in
Figure 18.
4.5. Comparison of Various Air Valve Sizes
The hydraulic event was analysed, considering the characteristics of the air valve presented in
Table 3. The influence of various characteristic curves of air valves is performed.
Table 7 shows three different air valves: the one already studied, a smaller one, and a larger one.
With these expulsion coefficients, for a gauge pressure in the pipe of 0.5 bar, air valve A can expel an airflow rate of , air valve B can expel an airflow rate of , and air valve C can expel an airflow rate of .
Due to their significantly different expulsion capacities, the development of the transient generated will vary notably from one scenario to another.
Table 8 shows when each air valve opens and closes during the conduit’s filling process.
Figure 19 illustrates the pressure evolution of the first air pocket, depicted for the four analysed scenarios: installation without an air valve and considering the three air valves. As observed, during the initial instants of the hydraulic event, the influence of the air valve is practically negligible. When the air valve closes, a compression process of the air pocket begins, which varies for each scenario depending on the specific conditions of the closure’s occurrence. For instance, with the most oversized air valve, a higher second peak pressure is reached (
= 18.280 m), slightly exceeding the pressure obtained without air valves.
Figure 20 presents the effect of the air valve size on the transient occurring during the filling manoeuvre. It shows the pressure evolution of the second air pocket.
Table 9 shows the value of the first pressure peak, the time when the maximum pressure is reached, and the time for the first closure of the air valve.
Table 9 shows the influence of the air valve size on the hydraulic event. The greater the air valve’s expulsion capacity, the lower the pressure peak. This peak is reached earlier and the air valve closes sooner due to the arrival of the water column.
When the air valve closure is produced, the air pockets compress, reaching their corresponding pressure peaks. Considering the air valve closure, the highest-pressure peak in the hydraulic installation occurs using the most oversized air valve (see
Figure 20); since the air valve is not active, the air cannot be expelled, reducing the pressure peaks.
Figure 21 illustrates the expelled flow rate of the three air valves during the transient event, while
Figure 22 shows the air velocity through the outlet orifice.