Experimental evaluation of two-layer air bubble curtains to prevent seawater intrusion into rivers

BC

Bubble column

DO

Dissolved oxygen

SCW

Sharp-crested weir

SWI

Seawater intrusion

ABC

Air bubble curtain

Q_a

Air bubble discharge

Q_s

Seawater discharge

Q_f

Freshwater discharge

LOP

Length of a river which polluted by seawater

IPM

Image processing method

ρ_s

Density of seawater

ρ_f

Density of freshwater

V_frontal

Frontal velocity of saline wedge zone

V_s

Velocity of the saline wedge at the beginning of intrusion

KH

Kelvin–Helmholtz instability

H

Hight of river's water

I₀

Intensity of struck light

I

Intensity of light transmitted from material

ɛ

Ability to absorb of the absorbing material

b

Length of the path in which light is absorbed

C

Concentration of the material

Reo

Reynolds number of freshwater

Fr

Froude number of freshwater

Fr_d

Densimetric Froude number

Due to the population growth, the water resources are in danger of overuse and contamination (Karamouz et al. 2016). In coastal areas, a significant portion of the water consumption is supplied by groundwater resources (Erostate et al. 2020). Burgan et al. (2017) conducted a statistical study about extreme events in some rivers that indicated the frequency of extreme events like drought increased from 1981 to 2000 which can cause destructive results. Chebana & Ouarda (2021) studied hydrological events like floods and droughts in Iran. They developed a model that considers the non-stationary new risk assessment framework. The final model constitutes a lognormal distribution for the margins with a linear trend in the peak series, stationary for the volume series, and a quadratic trend in the logistic Gumbel copula parameter for the dependence structure (Chebana & Ouarda 2021). Overexploitation of groundwater resources, sea-level rise (SLR) due to climate changes, extreme conditions such as drought and floods lead to seawater intrusion (SWI) in coastal zones. SLR could directly affect the serviceability of infrastructures of coastal regions; effects may include intrusion of seawater into drinking water resources (Nazarnia et al. 2020). Soil may solute by infiltrating water, and its chemical contents affect the environment (Ekeleme et al. 2021). Sharp differences between fresh and seawater densities result in the formation of density currents and lead to intrusion of the tip of the seawater into the freshwater in rivers and estuaries (Zhang & Savenije 2019). SWI could damage water bodies and the neighbouring regions along rivers. Currently, more than 100 counties and locals worldwide are endangered by SWI (Barlow & Reichard 2010; Abd-Elhamid et al. 2016; Han & Currell 2018; Palacios et al. 2020; Bhagat et al. 2021). This issue has received more attention than ever, fostering several developments. The increasing SLR and salination of groundwater resources in coastal areas and rivers has made researchers study this phenomenon (van der Ven et al. 2018; Zeynolabedin et al. 2021a, 2021b).

It is necessary to characterize the coastal aquifer and the SWI to evaluate and mitigate the environmental risks of this phenomenon. Different approaches are applied by researchers in this regard. Zheng et al. (2021) studied the optimal location of cut-off walls for SWI using the Kriging surrogate model simulation and optimization model. The results show that the outputs of the Kriging surrogate model and the variable density groundwater simulation model for the same cut-off wall design fit well, and the average relative error of the two outputs is only 2.2% which proves that it is feasible to apply the Kriging surrogate model to this problem. Yusuf et al. (2022) provide a physical simulation to demonstrate the sand column performance of a recharge reservoir to control seawater. The results revealed a reasonably close match between physical and computational modelling. It was also found that the more sand columns and higher the reservoir water level resulted in decreased seawater penetration length. In a numerical simulation, Panpan et al. (2021) studied the controlling of SWI by using hydraulic and physical barriers. The results show that when the recharge well is located near the toe of the seawater wedge of 40 cm from the coastline and 5 cm from the surface, the optimal performance of the recharge scheme is achieved with a repulsion rate of up to 21.5%. When the physical barrier is located 10 cm from the coastline, and the penetration depth is 35 cm, the toe of the seawater wedge is effectively driven to the coastline with a repulsion rate of up to 81.8%. Llopis-Albert & Pulido-Velazquez (2015) used the numerical models to investigate SWI. They showed that the sharp interface approach can provide accurate results in modelling SWI when applied to transient problems. Chun et al. (2018) assessed the impacts of climate change and SLR on SWI in a coastal aquifer by using the Saturated-Unsaturated Transport (SUTRA) model. The results showed that the largest increase in salinity occurred (40.3%) from the scenario of 57 cm in SLR. Abd-Elhamid et al. (2019) evaluated the impact of reservoir dam on SWI in the Nile Delta Aquifer, Egypt using a numerical method. They concluded that the maximum groundwater drawdown will reach 2.65 m and seawater will advance significantly by constructing the new dam. Torres-Martinez et al. (2019) developed a geophysical model to assess SWI in a coastal aquifer in Mexico. This model was based on a density-dependent flow model to simulate groundwater flow and SWI. The results indicated that due to the high rate of pumping for agricultural use in the past decade, a large cone of depression had formed, leading to the SWI advancement on the order of 6–8 km inland in the northern-central part of the aquifer. Zeynolabedin et al. (2021a, 2021b) assessed the uncertainty of geoelectrical models in modelling SWI in the coastal aquifer of Chaouia, Morocco. The general results indicated that the seawater has intruded about 2 km into the shore. The uncertainty analysis showed that there was a sensible difference between the maximum and minimum extent of SWI (maximum 9% variation in the area of SWI) which should be noted when doing water research management studies.

Besides modelling SWI and assessing the current status of coastal aquifers, researchers proposed different appropriate mitigation strategies in previous studies. One of the common alternatives to solving or reducing this negative effect is to decrease the intrusion length of the saline wedge (Nakai & Arita 2002; Zeynolabedin & Ghiassi 2019). The initial solution to cutting down the SWI length is creating a mound or a weir on the bottom of the channels, which also has side effects such as reducing the flood run-off capability of the channels. A superior alternative for narrow rivers is using air bubble curtains (ABCs). Not only is this solution easy to build but also it does not decrease the flood run-off capability (Nakai & Arita 2002). Moreover, it increases the dissolved oxygen (DO) by injecting bubbles into the water, which is notable in enhancing water quality (Larsen et al. 2019). Most of the studies carried out on the application of ABCs have focused on the density current in lock exchanges (Sasaki & Asaeda 1993; Asaeda et al. 1997; Oldeman et al. 2020). Nakai & Arita (2002) investigated steady saline wedge behaviour with the addition of ABC to find out more about the flow mechanism around an air curtain. Based on the findings, ABCs can significantly prevent SWI in different scenarios. Recently, Talebi & Salehi Neyshabouri (2021) studied the effect of ABCs on SWI in a numerical assessment with various seawater density rates and with river water levels as the main variable. The results showed that the ABC method is a good solution for SWI challenges. Bacot et al. (2022) established a formal analogy between bubble curtains and air curtains and unified the two frameworks. By means of small-scale laboratory experiments conducted in a channel with freshwater and brine solutions, they studied how effectively a bubble curtain acts as a separation barrier for a wide range of density differences as well as different air fluxes and water depths. The results showed high applicability of this approach. Talebi et al. (2022) investigated the effect of seawater density and air bubbles discharge on the performance of the air bubbles curtain using a numerical model. The results showed that the ABCs can prevent salinity intrusion by forming a vertical flow. Also, they concluded that there is an optimal discharge that discharges greater than it has insignificant effects on ABCs performance.

Reviewing the literature revealed that most previous studies focused on a single linear ABC, while ignoring the optimal location of ABC installation along a polluted river. This is important in terms of economic approaches because the length of the intrusion in actual conditions sometimes exceeds tens of kilometres. The cost of providing airflow in ABC in different locations vary since intrusion fluxes might decrease by moving in the upstream direction. Also, the minimum value of airflow to prevent SWI in different scenarios is another question that directly affects the feasibility of such ideas in field projects. In order to fill these gaps, this experimental study is focused on investigation of the optimal location of ABC regarding the changes in the density of seawater, velocity, and depth of the river by using two different linear ABCs. Also, this study attempts to investigate the optimal air bubble flow for controlling SWI.

Two linear air curtains were separately installed perpendicularly to the stream-wise direction at the bottom of the flume to create air. The distance from the first ABC (ABC-I) and the second ABC (ABC-II) to the downstream edge was 50 and 100 cm, respectively. The curtains had five circular holes. The distance between the holes was 1 cm with each hole 1 mm in diameter. An air compressor was used to create bubbles in the flume. A flowmeter measured the airflow rate (see Figure 1).

The experimental conditions are shown in Table 1. The variables of the study included freshwater discharge per unit width (q_f), river depth (H), and seawater density (ρ_s). Some experimental assumptions made are as follows:

• The slope of the flume is zero.

• The cross-section is rectangular.

• There is no friction between the flume and water.

• The injected air pressure is 1 atm and the water temperature is around 25 °C.

The present study explored optimum airflow and suitable location of ABC as a sustainable solution in narrow rivers. Different parameters like seawater density, power of freshwater, power of seawater, and airflow rate directly affect SWI issues. So, analysis of the presented results in the previous section is present in this section.

The optimum ABC airflow for preventing SWI in different ABC locations depends on the seawater density. As shown in Figures 5(b) and 6(b), the seawater density directly affects the value of airflow for preventing SWI at both ABC-I and ABC-II. As indicated, when the ρ_s/ρ_f ratio increased from 1.006 to 1.010, the V_frontal/V_s ratio, Q_a/Q_s for the ABC-I, and Q_a/Q_s for the ABC-II are raised 60, 58, and 70%, respectively. An increase in saline water density leads to more SWI, which was also reported by Talebi & Salehi Neyshabouri (2021). This may be related to an increase in the momentum and convection of the saline water, as a result of which a higher volume of saline water is shaped around the ABC location, requiring more airflow to prevent SWI. Also, the inertia of the saline wedge rises with an increase in seawater density. All the above-mentioned causes some environmental phenomena like KH and Entrainment could not mix the fresh and seawater in their interface (in the higher range of seawater density), so more airflow is required to prevent SWI.

Besides the seawater density, the power of the seawater affects the SWI at the tip of the saline wedge. As reported in Figure 5(b) in the ABC-I, the effect of seawater density in low Fr ranges plays a significant role. This is related to the lack of freshwater power and the high power of seawater. So, both the density and power of the seawater have a direct interaction: seawater power exponentially increases with a rise in seawater density. Contrary to ABC-I, the ABC-II scenario is reverse behaviour. As shown in Figure 6(b), in the greatest Fr cases (Fr > 0.031), seawater density changes do not significantly affect the process of SWI. It means that a rise in the seawater density does not increase the power of the seawater. The reason for this behaviour is KH in the tip of the saline wedge, which caused seawater to be mixed with freshwater. In other words, the rate of the entrainment and KH are higher than in other regions of the saline wedge. So, an increase in seawater density in the ABC-II does not considerably affect seawater power.

Another factor affecting the optimum ABC airflow value in preventing SWI is the power of freshwater. Mainly, the freshwater power depends on freshwater discharge, velocity, river level, and Fr number. As shown in Figures 5(b) and 6(b), increasing the Fr number in both gates caused the rate of the airflow (Q_a/Q_s) to decrease. As mentioned, when the Fr number changes from 0.035 to 0.027, the V_frontal/V_s ratio increases 92%, 67% in cases where ρ_s/ρ_f is 1.010 and 1.006, respectively. These values for the ABC-I raise about 40 and 68%, respectively. Raising the Fr number of the freshwater boosts the freshwater advection; as a result, a higher shear stress rate is exchanged at the interface between the freshwater and seawater. This helps the seawater mass at the boundary layer return downstream by the KH phenomenon and entrainment, which is in line with the finding of Nakai and Arita (2002) and Talebi et al. (2022). Also, based on Figure 4(b), increasing in Fr number leads to a decrease in the saline wedge's frontal velocity. Increasing in Fr number accelerates the rate of the KH and entrainment, so the velocity of the saline wedge tip reduces. All the mentioned scenarios indicate that an increase in the freshwater power leads to a decrease in the intrusion rate. So, these scenarios subsequently led to less SWI intrusion and less demand for ABC airflow. This is in line with the study by Talebi & Salehi Neyshabouri (2021).

In summary, the optimum location of the ABC and optimum value of the airflow to control the SWI is affected by some parameters like seawater density, power of the freshwater, and power of the seawater, which are studied in this section.

Many countries worldwide have been endangered because of SLR and climate change in recent years. SWI, as a result of SLR and climate change, can damage freshwater resources along rivers. Developing a sustainable solution to alleviate this concern has attracted a lot of attention. In the same line, the present study explored optimum airflow and suitable location of ABC as a sustainable solution in narrow rivers.

In conclusion, the present study investigated the effect of ABC location and airflow demand on SWI. To this end, several experimental tests were performed by assuming different values for the river and airflow discharges as well as for the salinity. In addition, two ABCs were located near and far from the downstream. The results showed that the optimal value of the air bubbles required to control SWI depends mostly on the density of seawater, with more airflow required to control denser flows. In other words, when the ρ_s/ρ_f ratio increased from 1.006 to 1.010, the V_frontal/V_s ratio, Q_a/Q_s for the ABC-I, and Q_a/Q_s for the ABC-II are raised by 60, 58, and 70%, respectively. Also, when the Fr > 0.031, seawater density changes do not significantly affect the process of SWI.

Another important parameter is the freshwater discharge and its corresponding Froude number. In the experiments, Fr numbers were chosen based on a low velocity of freshwater under estuaries and laboratory limitations. Results indicate that when the Fr number changes from 0.035 to 0.027, the V_frontal/V_s ratio increases by 92 and 67% in cases where ρ_s/ρ_f is 1.010 and 1.006, respectively. These values for the ABC-I raise about 40 and 68%, respectively. The results show that as the Froude number of freshwater increases, the SWI value and the airflow required to control it decrease. In addition, the ABC closer to the river downstream requires higher airflow rates to completely control the SWI and decrease the pollution extension.

Eventually, the regression equations for predicting the SWI parameters were derived based on the flow properties. Clearly, the equations have been derived based on some general assumptions and laboratory limitations, for example, flume geometry, Froude number, etc., yet they can be utilized efficiently for the initial design and for predicting the required range of the airflow and approximate location of the ABCs to control unwanted SWI properly. It goes without saying that the exact design of the optimal point in a real case and its corresponding airflow parameters depend on local conditions such as river geometry, water quality, financial costs for ABC construction, environmental issues, etc. This system may only be applicable to narrow rivers in some specific ranges.

Finally, this study has some limitations. All the tests studied the SWI in a limited density range of the sweater. At the same time, the seawater concentration in an actual condition is sometimes four times greater than the studied density. For this study, NaCl is considered for seawater synthesis, while natural seawater may have different properties in the field. Also, experiments were conducted in a specific range of the Fr number of freshwater. In addition, some physical assumption considered in this investigation includes: (1) the slope of the flume is zero, (2) the cross-section is rectangular, (3) there is no friction between the flume and water, (4) the injected air pressure is 1 atm, and (5) the water temperature is around 25 °C. Future studies are recommended to investigate other unknown parameters, including the effects of vegetation as a natural barrier against SWI prevention. Besides, investigation on the cross variation of airflow and its effect on SWI patterns can be interesting due to the unbalanced seawater and river velocity profile distribution.

This study was supported by the Tarbiat Modares University, Tehran, Iran.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.