Channel confluences that are characterized by two channel flows converging into one flow are common in fluvial systems [1
]. The main channel and branch branches are connected by junctions to form water systems and even river networks. River networks are important parts of the surface water circulation system and they also represent important channels for watersheds, sediments, and other types of material transported in river basins [2
]. At river junctions, the turbulent mixing of flows is severe and there is a strong retention effect on sediment and pollutant transport [3
]. Thus, confluences have become a focus area for water conservation, shipping, and environmental protection.
Many types of junctions occur in natural rivers. Junctions can be categorized into either asymmetrical river confluences, where the post-confluence channel forms a linear extension of the upstream main channel, and Y-shaped confluences, where two tributaries are geometrically symmetrical. Both types of intersections are shown in Figure 1
In addition, there are differences in the hydrodynamic characteristics of the different types of confluences. For example, several studies have investigated asymmetrical junctions, and factors such as confluence ratio and confluence angle are proposed. In particular, Best [4
] proposed a model of the flow dynamics present at the river channel confluences based on six characteristic flow zones which include: regions with flow stagnation, flow deflection, flow separation, maximum velocity, downstream flow recovery, and several distinct shear layers. Moreover, Best and Reid [5
] conducted four flume experiments with different confluence angles to determine the features of the flow separation zone. The dimension was found to increase with both the junction angle and the contribution of the branch to the total discharge, but the shape remained more or less constant. In addition, a study by Qing et al. [6
] indicated that the discharge ratio affects the geometry and separation zone trend from the riverbed up to the surface. Furthermore, Best [7
] conducted an in-depth investigation of complex flow and sediment transport patterns, showing that the major controls for these processes are the junction angles and the ratio of the discharge in the two confluence channels.
In addition, the effects of differences in the depth between the main and branch channels have been extensively studied. These studies revealed that depth discordance had the effect of distorting the mixing layer as well as strengthening the turbulence intensity and shear stress between the confluent streams [8
]. Moreover, Hsu et al. [11
] proposed a mathematical formulation of the energy and momentum correction coefficients. The study was designed in order to determine the empirical relationship between the momentum transfer and the discharge ratio by studying a 90° equal-width open-channel junction. In a related study, Hsu et al. [12
] also showed that increases in the junction angle and downstream Froude number led to an increase in the depth ratio, and they proposed a suitable formula.
The confluence of current also forms unique phenomena such as helical cell and spiral flow. Stream flow analysis was furthered by Weber et al. [13
] who determined the three-dimensional flow field within a junction by using an acoustic Doppler velocimeter (ADV). In their report, the velocity vector field, turbulent kinetic energy, and water surface mapping were analyzed. Likewise, Liu et al. [14
] also conducted an experimental study of a 90° open channel confluence and found that the vertical distribution of the average velocity was influenced by both the discharge ratio and the flow pattern. Next, Biswal [15
] found that the secondary current and turbulent stresses were reproduced well by the hydraulic model, where they increased in the interface region as the relative flow ratio decreased. This was related to a study by Yuan et al. [16
] where several hydrodynamic and turbulence characteristics were analyzed. In this study, factors such as the turbulent kinetic energy, Reynolds shear stress, and turbulence spectrum were evaluated to obtain data from a T-shaped discharge-adjustable circulating flume. This experiment was conducted to determine the turbulent flow structure in the distorted shear layer. The results showed that a stronger helical cell was formed and it extended for a longer distance downstream when the branch channel had a higher flow rate than the main channel. In a related study it was found that the pressure gradient term was the primary factor that triggered the velocity redistribution, whereas the convective acceleration was the secondary term as the Froude number increased [17
The transport of pollutants at channel confluences is influenced by specific hydrodynamic characteristics [18
]. Biron et al. [19
] used a three-dimensional mathematical model of water flow to conduct numerical simulations in the laboratory, modeling the mixing of pollutants in an open channel. The results showed that the contaminant mixing process was faster when river bed depths were inconsistent between the main channel and tributaries. To address this issue, Isabel et al. [20
] established a three-dimensional hydrodynamic model to simulate the effects of contaminant diffusion in the Douro River estuary in Portugal. The results showed that a stable flow was most conducive to the diffusion of contaminants, thereby indicating that the water flow characteristics had important effects on the diffusion of pollutants. In a related study, a two-dimensional model was used to simulate the transport of pollutants under different flow rates at 90° intersections, where the polluted area decreased as the discharge ratio increased [21
]. This then affected the concentration of different cross-sectional contaminants according to the distance from the junction point. Based on the Reynolds averaged Navier–Stokes equations and Reynolds stress turbulence model, the distribution of contaminant concentrations is primarily controlled by the shear layer and two counter-rotating helical cells [22
]. This phenomenon is affected by the discharge ratio and bed morphology.
However, these previous studies have all considered asymmetrical river confluences, whereas relatively few studies have investigated symmetrical river confluences known as Y-shaped confluences. Guo et al. [23
] employed a model to study the three-dimensional hydraulic characteristics of the flows at a Y-shaped junction using data obtained with an acoustic Doppler velocimeter (ADV). According to their results, the stagnation zone, flow deflection zone, flow separation zone, acceleration zone, and other zones could be detected at the “tributary inclined mainstream”-shaped junction. The overall flow was characterized as a spiral flow, which is one of the main properties downstream of a Y-shaped junction. As the discharge ratio became higher and there was bed discordance, the trend of the spiral flow was weakened. Rhoads [24
] suggested that helical motion can enhance the mixing patterns at confluences. This was based on observations of the downstream persistence of a well-defined mixing interface at two symmetrical confluences, and the disruption of this interface at asymmetrical confluences. Likewise, Geberemariam [25
] suggested that for a 90° junction, the separation zone area and discharge ratio are indirectly proportional due to the recirculating flow, low pressure, and minimum velocities near the T- and Y-junction areas.
Current methods based on the hydrodynamic characteristics of water flows in river intersections can be divided into prototype observational data analyses [26
], physical model tests [28
], and numerical simulations [30
]. Previous studies on Y-shaped river channels mainly focus on hydrodynamic characteristics, but there are few studies on the law of pollutant mixing. In order to address the lack of previous analyses of Y-shaped junctions, flume experiments were performed to investigate the hydrodynamics and contaminant transportation at a 60° channel confluence. We distinguished the differences in the two types of confluences in order to facilitate pollutant management at river confluences.
2. Materials and Methods
Eight representative channel confluences in Xitiaoxi watershed served as the study sites for the research. The Xitiaoxi river basin, one of the most important tributaries in the Lake Taihu Basin, is a typical dendritic river network (see Figure 2
). It is located in Huzhou city, Zhejiang province, and supplies water for residents’ daily living. The classification and morphological analysis of the river network indicate that many confluences exist in Y-shaped form of the Xitiaoxi river basin. Natural confluence river intersection angles ranged between 30° and 90°, with a statistical average of 60°.
In order to ensure that this study had practical significance, the flume model was designed based on the morphological characteristics of the water system in the Xitiaoxi river basin. Table 1
shows the specific data, which indicate that the natural confluence intersection angles ranged between 30° and 90°, with a statistical average of 60°. Thus, the confluence angle employed in the model was 60°.
A type of flume with a symmetrical confluence was used to perform hydrodynamic disturbance and contaminant transport experiments. The experimental device was designed according to the results of the field investigation. This experiment was a basic investigation where the model and statistics were designed according to a scale of 1:250. According to the gravity similarity criterion, the generalized flat bottom flume fixed bed model was used for the test. The current study is concerned with results of a 60° confluence angel, although the natural confluence river intersection angles ranged between 30° and 90°. The two tributaries were 22 cm and 26 cm wide, respectively, and 3 m long, while the post-confluence channel was 40 cm wide and 6 m long (see Figure 3
). It is common for the post-confluence channel to be slightly wider than the branch channel at natural river confluences, a phenomenon referred to as “downstream hydraulic geometry” [32
]. This experiment was carried out using a flume that is awaiting national invention patent of China approval (No. 201810287680.8).
Head tanks on the both branch channels supply the discharge. To ensure a fully developed flow entered into the junction branches, energy dissipator and a sufficiently long channel were placed at the branch channel inlets. Water was pumped from the tank to the two branches through polyvinyl chloride pipes of 110 mm in diameter, and the flow discharges were monitored by two ultrasonic flowmeters and pump-value systems. The water level in the downstream main channel was controlled by an adjustable tailgate (h = 15 cm).
The coordinate system defined for this testing had the positive x-axis oriented in the downstream direction of the main channel. The positive y-direction points to the left branch wall opposite of the channel junction. Thus, the positive z-axis is upward in the vertical direction. The origin from which all points are measured was the bed at the right branch corner of the channel junction (see Figure 4
Two discharge ratios are considered in this study. In Case 1 the left branch channel, Ql
, has a discharge of 10 m3
/h, and that of the downstream main channel, Qt
, is 30 m3
/h, yielding a discharge ratio (q = Ql
) of 0.33. In Case 2, the left branch channel, Ql
, the discharge is 20 m3
/h and for Qt
it is 40 m3
/h, yielding a discharge ratio of 0.5 A combined flow usually comprises a subcritical flow and turbulent flow. Therefore, when designing the flume model, the Froude number was maintained at less than 1 to ensure that the flow was subcritical. The Reynolds number was greater than 1000 to ensure that the flow was turbulent. Within the two discharge ratios mentioned above, the contaminant test used a mixture of sodium chloride, ethanol, and water as the contaminant tracer, which together made the density approximately equal to that of the water body cycled in flume. The tracer was stored in the water tank and discharged into the flume by the tracer discharge device which is plastic tube with discharge hole. The tracer emission manner for the tracer considered both the point source and the line source. According to the different discharge mode, the discharge position was set on different sides of the branch channel. The specific contaminant tracer discharge conditions are shown in Table 2
According to preliminary studies, the discharge outlets are usually located at half of the water depth in the same direction as the flow of water, which makes the distribution of the emissions more uniform. Therefore, under the different test conditions, the discharge ports were fixed at half of the water depth and the pollutant discharge qc was constant at 0.1 L/s.
A Vectrino acoustic Doppler velocimeter (ADV) was used to measure three-dimensional flow velocities at a series of grid-defined points, which were taken in lines at 13 cross sections (M1–M13), with the near bed locations being more closely spaced as shown in Figure 4
a. Each channel cross section consisted of seven evenly spaced vertical profiles, as shown in Figure 4
b. In each line, the lowest measurement point is located at 1.5 cm above the bed due to the measuring requirements of the ADV, and other points are located in the line with a vertical interval of 1.5 cm. This testing grid produced approximately 888 velocity measurement locations for each flow condition studied. The velocity measurements were taken at each sampling location for 30 s at a sampling rate of 50 Hz.
In the contaminant transport experiment, a multi-point electrolytic conductivity meter was used to measure the conductivity of the water. This device measured the water conductivity and temperature, and the data were converted into the corresponding pollutant concentrations for further analysis. The sampling data are collected over duration of 180 s, in which approximately 200 instantaneous datums were acquired from each measurement point (average value of the final instantaneous data), thereby ensuring that accurate mean data were obtained.
According to the morphological characteristics and similarity theory for the Xitiaoxi water system as a typical dendritic river network area, as well as the actual size of the test site, a physical model test system was developed for a 60° Y-type intersection in order to analyze the hydrodynamic characteristics. Furthermore, pollutant discharge devices and multi-point conductivity meters were employed in pollutant blending tests. Several conclusions can be drawn based as follows.
(1) The water flow in the 60° Y-shaped confluence was vulnerable to flow velocity separation at the junction, thereby forming a small-range low-flow region located in the two branches downstream of the water flow intersection. After the intersection, the flow direction moved downward near the left side, whereas the flow direction was upward near the right bank.
(2) A double-spiral flow with equivalent strength appeared at the intersection and a spiral flow with a counter-clockwise vortex flow was present downstream. The vortex intensity decreased with the downward flow of the water and the rotation trend of the spiral flow decreased as the discharge ratio increased.
(3) The contaminant concentration band that appeared after the confluence tended to be compressed and it then diffused downstream, where it was generally twisted. A high concentration pollutant zone is likely to occur at the junction and downstream near the junction.
(4) The discharge of pollutants from the point source on the inner bank on the left branch was more conducive to the transport and mixing of the contaminants than that from the outer bank on the left branch, and the concentration was lower after mixing evenly, whereas the opposite was found on the right branch.
(5) The discharge was higher from the line source on the left branch than that from the right branch. A higher concentration contaminated area was readily produced near the junction, but the overall concentration was lower after mixing, which is more beneficial for the transport and blending of contaminants. When the pollutants were released from the left branch, the discharge ratio was smaller, the pollution belt width was smaller, and the mixing was more rapid and effective. This was more favorable for the identification and treatment of pollutants, but the opposite was found for the right branch.
The spiral flow and its effect on the transport and blending of contaminants were the main differences between Y-shaped and asymmetrical river confluences. The results obtained for 60° and 90° Y-shaped intersections were similar, but the size of the separation zone at 60° and the vortex tendency of the overall spiral flow were reduced as compared to at 90°.