Experimental and Numerical Study on the E ﬀ ect of the Temperature-Control Curtain in Thermal Stratiﬁed Reservoirs

: The impoundment and power generation of dams cause the temperature of released water much lower than the original rivers in the thermal stratiﬁed reservoirs. In addition, the released low-temperature water would damage the downstream habitats of ﬁsh and other biological groups seriously. Available facilities, such as stop log gate intakes and multi-level intakes, are built to alleviate the problem. For overcoming the limitations of traditional facilities on construction conditions and the improved e ﬀ ect of water temperature, a new facility of the temperature control curtain (TCC), with the advantages of convenient regulation and no hydropower loss, has been proposed recently. However, to the author’s knowledge, the theory of TCC is not abundant, with incomplete experimental tests and few numerical simulations. In this paper, a rectangular tank is designed speciﬁcally to conduct experimental tests to verify the e ﬀ ects of TCC and explore its potential impacts on released water temperature (RWT) under four major inﬂuencing factors. The study results show that TCC has signiﬁcant e ﬀ ects on improving RWT, with a maximum increase of 8.3 ◦ C. In addition, a three-dimensional hydrodynamic model with the same size of experimental model is established for further research. The results show that RWT is mainly related to the temperature distribution of a reservoir and the water-retaining proportion of the curtain. Finally, a basic principle for TCC construction is proposed and all these laid an important theory foundation for its application in engineering practice.


Introduction
Hydraulic engineering projects aim to achieve efficient utilization of water resources. While generating economic benefits, they also change the natural properties of rivers and have significant impacts on the reservoir ecology [1][2][3][4]. Due to the construction of hydraulic structures, large amounts of water are stored behind dams, which results in deep water with low-velocity flow. These conditions mean that vertical temperature stratification frequently occurs in these deep reservoirs [5][6][7]. Water temperature impacts the transportation of oxygen and other nutrients, as well as biological and chemical reaction rates [8][9][10]. For example, temperature is closely related to the concentration of dissolved oxygen, nitrogen, and phosphorus in water [11,12]. Spawning and growing of fish depend on the water temperature also. Thus, cooler temperatures can change or potentially destroy the downstream habitats of fish and other biological groups, pushing them to the verge of extinction [13,14]. In addition, using cooler temperatures to irrigate can directly reduce the crop yields. Many hydropower engineering projects are affected by low-temperature released water like Glen Canyon Dam and Cherokee Dam in America and the Three Gorges Dam and San Ban Xi Dam in China [15][16][17][18]. Especially in China during A rectangular tank which is 7.2 m long, 0.6 m wide, and 1.3 m high was specially designed for the tests. It was composed of three parts: the steady-flow zone, test zone, and outflow zone ( Figure 1). The photos of experimental model are shown in Figure A1. There were two intakes, respectively, sitting on the top and the bottom in front of the steady-flow zone, which can pump water from different places to maintain a specific water stratification of the test zone. Several diversion tubes and a breakwater were used to represent the outlets with different heights, and each outlet was equipped with a sensor to measure the RWT. A nylon fabric curtain can be placed a certain distance from the outlet, with the bottom side fixed, the left and right top corners dragging fixed by ropes, and the top side opening. High-power electric heat tubes were used to form the thermal stratification before the test at the same time digital temperature sensors, electromagnetic flowmeters, and graduated scales were, respectively, used to measure the water temperature, flow rates, and water level. In order to observe the vertical distribution of water during the test, the digital temperature sensors were set as follow: there were 3 temperature sensors groups and 13 sensors in each group. The sensors were set vertically with intervals of 0.05 m between the depth of 0.05 and 0.65 m below the surface in each group.
Before the tests, each electric heat tube was dipped into water by ropes, one end of the ropes was tied to the tubes and the other tied to the top of tank. The tubes' immersion depth can be adjusted artificially by changing the length of ropes until the thermal structure of water was the same as target stratification. During the test, the high temperature water flowed into the steady-flow zone from the top intake, as the target thermal stratification, while the lowest from the bottom intake one. Thus, the steady thermal stratification can be maintained for more than 25 min according to repeated experiments, which was enough for a set of the scenario. For every test, flow and water temperature fields need approximately 5~10 min to stabilize, and the continuous monitoring was carried out for 20 min.

Generalization of Thermal Stratification
Based upon previous studies, thermal stratification influences the effect of TCC mostly. However, it is complicated owing to a variation of the topography and climates of a reservoir. To generalize the most representative thermal stratification, five hydraulic engineering projects which have the low-temperature released water problems were selected. They are all river-type reservoirs, with the dam heights more than 150 m, including Sanbanxi, Nuozhadu, Ertan, and Xiluodu projects in China and the Glen Canyon Dam in America. As shown in Figure 2a, May and August were chosen to represent the spring and summer, respectively, when the lower RWT was obvious.
The thermal stratification characteristics of these reservoirs can be generalized into two types: single thermocline and double thermocline. Single thermocline (S1) distribution refered to a clear vertical thermal stratification with a three-layer structure, comprising an upper mixed layer, a middle thermocline layer, and the hypolimnion layer at the bottom of the reservoir. This stratification was more stable and the thermocline had uniform temperature gradients. Double thermoclines (S2) included two thermoclines, upper and lower. There was a mixed layer between the two areas with more sharp gradients.
Considering the size of tank, two curves with the characteristics of S1 and S2 stratification were designed, respectively, and the curves were shown in Figure 2b. Through adjusting the depth and location of high-power electric heat tubes, the vertical temperature measured in the tank at the beginning of tests were almost the same as the designed curves ( Figure 2b). High-power electric heat tubes were used to form the thermal stratification before the test at the same time digital temperature sensors, electromagnetic flowmeters, and graduated scales were, respectively, used to measure the water temperature, flow rates, and water level. In order to observe the vertical distribution of water during the test, the digital temperature sensors were set as follow: there were 3 temperature sensors groups and 13 sensors in each group. The sensors were set vertically with intervals of 0.05 m between the depth of 0.05 and 0.65 m below the surface in each group.
Before the tests, each electric heat tube was dipped into water by ropes, one end of the ropes was tied to the tubes and the other tied to the top of tank. The tubes' immersion depth can be adjusted artificially by changing the length of ropes until the thermal structure of water was the same as target stratification. During the test, the high temperature water flowed into the steady-flow zone from the top intake, as the target thermal stratification, while the lowest from the bottom intake one. Thus, the steady thermal stratification can be maintained for more than 25 min according to repeated experiments, which was enough for a set of the scenario. For every test, flow and water temperature fields need approximately 5~10 min to stabilize, and the continuous monitoring was carried out for 20 min.

Generalization of Thermal Stratification
Based upon previous studies, thermal stratification influences the effect of TCC mostly. However, it is complicated owing to a variation of the topography and climates of a reservoir. To generalize the most representative thermal stratification, five hydraulic engineering projects which have the low-temperature released water problems were selected. They are all river-type reservoirs, with the dam heights more than 150 m, including Sanbanxi, Nuozhadu, Ertan, and Xiluodu projects in China and the Glen Canyon Dam in America. As shown in Figure 2a, May and August were chosen to represent the spring and summer, respectively, when the lower RWT was obvious.
The thermal stratification characteristics of these reservoirs can be generalized into two types: single thermocline and double thermocline. Single thermocline (S1) distribution refered to a clear vertical thermal stratification with a three-layer structure, comprising an upper mixed layer, a middle thermocline layer, and the hypolimnion layer at the bottom of the reservoir. This stratification was more stable and the thermocline had uniform temperature gradients. Double thermoclines (S2) included two thermoclines, upper and lower. There was a mixed layer between the two areas with more sharp gradients.
Considering the size of tank, two curves with the characteristics of S1 and S2 stratification were designed, respectively, and the curves were shown in Figure 2b. Through adjusting the depth and location of high-power electric heat tubes, the vertical temperature measured in the tank at the beginning of tests were almost the same as the designed curves ( Figure 2b

Model Scenarios
The factors influencing RWT mostly in a reservoir are the property of curtains and the discharge rules besides thermal structure [31,32]. Thus, four characteristic variables were proposed to represent the main influencing factors.
• Curtain position (CP), means the distance from the curtain to the outlet, with the unit of m.
• The water-retaining proportion (Pr), means the proportion of retaining water depth of TCC to the overall depth of the water, with the unit of %. • Outlet submerged depth (OSD), means the depth from the water surface to the centerline of the outlet, with the unit of cm. • Released flow quantity (RQ), means the flow quantity released from the outlet, with the unit of m 3 /h. By comprehensive consideration of the possible flow fields of the real flow fields, model similarity scales, and conditions of construction sites for the typical reservoirs with low released temperature water problems, a series of scenarios were designed in Table 1. Three typical distances of CP were selected, which were 1.0, 2.5, and 3.5 m. From an engineering view, for most of reservoirs the closer to upstream the larger river cross-section is, the more difficult the construction is, thus the place of TCC will not be very far from the outlet. Our discussion about the influence of curtain position on TCC was carried out on this basis. Pr can be set from 55% to 90%. During summer and autumn, when the lower RWT is obvious, the spillway outlet is the usual way of discharge, thus OSDs were set from 24 to 40 cm, while the water level maintains 90 cm during all the tests. General outflow discharged from reservoirs has a large range, the range of released water quantity we discussed should be in the suitable range of a thermal stratification reservoir troubled by low released water temperature. Thus, take the example of the Sanbanxi project, outflow changes between 125 to 870 m 3 /s [33], and so the RQ was set from 1.6 to 11 m 3 /h. The tests were carried out inside a laboratory, so only air temperature was considered in meteorological conditions with a daily measurement.

Model Scenarios
The factors influencing RWT mostly in a reservoir are the property of curtains and the discharge rules besides thermal structure [31,32]. Thus, four characteristic variables were proposed to represent the main influencing factors.

•
Curtain position (CP), means the distance from the curtain to the outlet, with the unit of m.

•
The water-retaining proportion (Pr), means the proportion of retaining water depth of TCC to the overall depth of the water, with the unit of %.

•
Outlet submerged depth (OSD), means the depth from the water surface to the centerline of the outlet, with the unit of cm.

•
Released flow quantity (RQ), means the flow quantity released from the outlet, with the unit of m 3 /h. By comprehensive consideration of the possible flow fields of the real flow fields, model similarity scales, and conditions of construction sites for the typical reservoirs with low released temperature water problems, a series of scenarios were designed in Table 1. Three typical distances of CP were selected, which were 1.0, 2.5, and 3.5 m. From an engineering view, for most of reservoirs the closer to upstream the larger river cross-section is, the more difficult the construction is, thus the place of TCC will not be very far from the outlet. Our discussion about the influence of curtain position on TCC was carried out on this basis. Pr can be set from 55% to 90%. During summer and autumn, when the lower RWT is obvious, the spillway outlet is the usual way of discharge, thus OSDs were set from 24 to 40 cm, while the water level maintains 90 cm during all the tests. General outflow discharged from reservoirs has a large range, the range of released water quantity we discussed should be in the suitable range of a thermal stratification reservoir troubled by low released water temperature. Thus, take the example of the Sanbanxi project, outflow changes between 125 to 870 m 3 /s [33], and so the RQ was set from 1.6 to 11 m 3 /h. The tests were carried out inside a laboratory, so only air temperature was considered in meteorological conditions with a daily measurement. Notes: Curtain position (CP); water-retaining proportion (Pr); water-retaining proportion (Pr); released flow quantity (RQ); released water temperature (RWT).

Model Description
At the meantime, Flow-3D was used to build a three-dimensional hydrodynamic and temperature numerical model with the same size of the experimental test model (Figure 3a), for the further research of TCC. The model was divided into 145 longitudinal segments, 45 vertical layers, and 20 lateral segments. Based upon the phenomenon of the experimental test, TCC appeared arc-shaped because of the thrust force by flow and its flexibility, and the fields around the curtain were the most concerned. Thus, for the test zone, 45 segments were set in the upstream forebay zone and 25 segments were set in the curtain zone to give a better show of the arc-shaped curtain for obtaining more details (Figure 3b).
The grid-independence tests were taken before the numerical simulation. For working condition A1, the RWT with different grid sizes were compared, as shown in Table 2. It can be concluded that variation in RWT basically did not with the grid size increase and decrease, both vertical and longitudinal; however, the water temperature of the outflow changes by up to 2%. Therefore, the mesh generation method described in this paper was determined by considering the calculation precision and calculation amount comprehensively.
Since the mechanical characteristics were not concerned, the curtain was simplified as a rigid arc-shaped structure without regard to flexibility.
The boundary conditions and scenarios were set the same as the experimental tests. After repeated adjustment, the bottom Manning coefficient was set as 0.01, the coefficient of bottom heat exchange was set as 0.3 W/(m 2 °C ), and dynamic viscosity and thermal conductivity of fluid were set according to water properties at different temperatures.  In order to cover the shortage of experimental tests, the numerical model's scenarios were as a supplement and improvement to experimental tests. Thus, each variable set multiple scenarios; taking Pr as an example, the numerical model scenarios were set the same as model tests and with 95%, 92%, 80%, 62%, 50%, and 48% added at the same time. Through a lot of numerical simulation, the accuracy of the physical model was verified and the principle of TCC was summarized as well.

Model Verification
To verify the accuracy of the numerical model, simulated vertical temperature distributions before and after TCC were extracted and compared with the values measured in experimental tests under the same scenario. The scenarios selected were A1 and B1, and results were shown in Figures 4 and 5. The root mean square error (RMSE) and coefficient of determination (R 2 ) were used to evaluate the matching degree [34,35]. The goodness-of-fit tests and the statistic error labeled in figures showed that the 3-D numerical model matched the experimental model well and can accurately simulate the temperature fields.

Parameters
The Size of Grid Since the mechanical characteristics were not concerned, the curtain was simplified as a rigid arc-shaped structure without regard to flexibility.
The boundary conditions and scenarios were set the same as the experimental tests. After repeated adjustment, the bottom Manning coefficient was set as 0.01, the coefficient of bottom heat exchange was set as 0.3 W/(m 2 • C), and dynamic viscosity and thermal conductivity of fluid were set according to water properties at different temperatures.
In order to cover the shortage of experimental tests, the numerical model's scenarios were as a supplement and improvement to experimental tests. Thus, each variable set multiple scenarios; taking Pr as an example, the numerical model scenarios were set the same as model tests and with 95%, 92%, 80%, 62%, 50%, and 48% added at the same time. Through a lot of numerical simulation, the accuracy of the physical model was verified and the principle of TCC was summarized as well.

Model Verification
To verify the accuracy of the numerical model, simulated vertical temperature distributions before and after TCC were extracted and compared with the values measured in experimental tests under the same scenario. The scenarios selected were A1 and B1, and results were shown in Figures 4 and 5. The root mean square error (RMSE) and coefficient of determination (R 2 ) were used to evaluate the matching degree [34,35]. The goodness-of-fit tests and the statistic error labeled in figures showed that the 3-D numerical model matched the experimental model well and can accurately simulate the temperature fields.

The effect of TCC and Cause Analysis
The comparison of the RWT results under different scenarios from physical models were shown in Table 1 and Figure 6. For different Pr values, RWTs were increased by 6.8~8.3 • C under S1 thermal distribution and 2.3~3.5 • C for S2 distribution. TCC had a significant effect on RWT, but the degree of improvement was related closely to the thermal stratification of the reservoir.

The effect of TCC and Cause Analysis
The comparison of the RWT results under different scenarios from physical models were shown in Table 1 and Figure 6. For different Pr values, RWTs were increased by 6.8~8.3 °C under S1 thermal distribution and 2.3~3.5 °C for S2 distribution. TCC had a significant effect on RWT, but the degree of improvement was related closely to the thermal stratification of the reservoir. In order to reveal the reason why TCC has a good effect on RWT, baseline scenarios A0 and A1 were analyzed primarily to make a contrast with others. The water upstream of the dam was subjected to a pulling force as a function of the discharge through the power station and to buoyancy differences caused by density stratification [36]. When the thermal stratification of water was strong, vertical mixing was inhibited, and the flow field in front of the dam was characterized by stratified water intake. As shown in Figure 7a, the water flow was stratified, and the pulling due to discharge of water only drove the water within a certain vertical range. However, when the TCC was set as shown in Figure 7b, the flow field changed dramatically. The flow of the water in front of and above the TCC was strengthened and the longitudinal velocity of all sections were enhanced. Then the major plume of the faster water from upstream flowed across the top of the TCC and plunges (mainly) into the layers level with the outlet. Between the outlet and the TCC, the water pulled by the discharge was still on the same horizontal line as the outlet, with an enlarged range of influence. However, it could also be concluded from the results of the physical model tests and numerical simulation that the lateral gradient of the water temperature and lateral flow velocity in the water were small and negligible. Therefore, the analysis below was focused on the action in the longitudinal and vertical directions. In order to reveal the reason why TCC has a good effect on RWT, baseline scenarios A0 and A1 were analyzed primarily to make a contrast with others. The water upstream of the dam was subjected to a pulling force as a function of the discharge through the power station and to buoyancy differences caused by density stratification [36]. When the thermal stratification of water was strong, vertical mixing was inhibited, and the flow field in front of the dam was characterized by stratified water intake. As shown in Figure 7a, the water flow was stratified, and the pulling due to discharge of water only drove the water within a certain vertical range. However, when the TCC was set as shown in Figure 7b, the flow field changed dramatically. The flow of the water in front of and above the TCC was strengthened and the longitudinal velocity of all sections were enhanced. Then the major plume of the faster water from upstream flowed across the top of the TCC and plunges (mainly) into the layers level with the outlet. Between the outlet and the TCC, the water pulled by the discharge was still on the same horizontal line as the outlet, with an enlarged range of influence. However, it could also be concluded from the results of the physical model tests and numerical simulation that the lateral gradient of the water temperature and lateral flow velocity in the water were small and negligible. Therefore, the analysis below was focused on the action in the longitudinal and vertical directions.
The comparison of the RWT results under different scenarios from physical models were shown in Table 1 and Figure 6. For different Pr values, RWTs were increased by 6.8~8.3 °C under S1 thermal distribution and 2.3~3.5 °C for S2 distribution. TCC had a significant effect on RWT, but the degree of improvement was related closely to the thermal stratification of the reservoir. In order to reveal the reason why TCC has a good effect on RWT, baseline scenarios A0 and A1 were analyzed primarily to make a contrast with others. The water upstream of the dam was subjected to a pulling force as a function of the discharge through the power station and to buoyancy differences caused by density stratification [36]. When the thermal stratification of water was strong, vertical mixing was inhibited, and the flow field in front of the dam was characterized by stratified water intake. As shown in Figure 7a, the water flow was stratified, and the pulling due to discharge of water only drove the water within a certain vertical range. However, when the TCC was set as shown in Figure 7b, the flow field changed dramatically. The flow of the water in front of and above the TCC was strengthened and the longitudinal velocity of all sections were enhanced. Then the major plume of the faster water from upstream flowed across the top of the TCC and plunges (mainly) into the layers level with the outlet. Between the outlet and the TCC, the water pulled by the discharge was still on the same horizontal line as the outlet, with an enlarged range of influence. However, it could also be concluded from the results of the physical model tests and numerical simulation that the lateral gradient of the water temperature and lateral flow velocity in the water were small and negligible. Therefore, the analysis below was focused on the action in the longitudinal and vertical directions.

Analysis of Influencing Factors
The comparisons between numerical simulation results and physical model tests in different scenarios were shown in Figures 8a, 9a, and 10a. The flow and temperature fields of the different scenarios were shown in Figures 8b, 9b, and 10b.

Analysis of Influencing Factors
The comparisons between numerical simulation results and physical model tests in different scenarios were shown in Figures 8a, 9a and 10a. The flow and temperature fields of the different scenarios were shown in Figures 8b, 9b and 10b.

Analysis of Influencing Factors
The comparisons between numerical simulation results and physical model tests in different scenarios were shown in Figures 8a, 9a, and 10a. The flow and temperature fields of the different scenarios were shown in Figures 8b, 9b, and 10b.

Effect of CP
From the results of A1~A3 and B1~B3 (Table 1), the change of the curtain's position had little effect on RWT, and it only had influence on the stable-time of water after the set of TCC. Thus, in practical engineering, the position of TCC should be determined after considering the complex terrains, economic condition, waterway transport, and other specific cases in actual engineering.

Effect of Pr
Under the difference of thermal stratifications, both experimental tests and numerical simulation results had the same trend-that increasing Pr will increase the RWT dramatically (Figure 8a).
The flow velocity distributions of the water modelled with different Pr values were partly shown in Figure 8b. It can be seen that the flow velocity of the water above the TCC was significantly higher than that of the water in the lower layer. The curtain prevented the lower section of the water column from flowing downstream, thus separating it from the warmer upper water stratum. In addition, the water flow over the TCC was drawn by the tractive action of outlet. The increase of Pr caused the flow area above the TCC to decrease, which increased the flow velocity. As seen from the temperature field (Figure 8b), with the increase of Pr, higher temperature surface water flowed into the plume downstream of the curtain; therefore, the average temperature of this water was higher.
Meanwhile, there were critical values of Pr; we can see from the line graph in Figure 9a that the trend lines levelled off at both ends. There was a sensitive zone where the changes of Pr and RWT were positively correlated, and out of the zone the change of Pr had no effect on RWT; the critical value of Pr was 90% and 65%. This indicated that only the top of TCC was in the thermocline; its ability to regulate RWT was explicit.
As a result, the Pr obviously influenced the surface velocity and had a significant effect on the flow field. Most significantly, the increase of Pr blocked the flow of lower layers, allowing only hightemperature layers to flow around the curtain, thus increasing the RWT.

Effect of OSD
As the OSD increased, the RWT decreased, but the range of variation was less than 1 °C , as shown in Figure 9a.
The OSD values of 44 and 28 cm in the numerical simulation were compared and the flow field distribution was shown in Figure 9b. It was concluded that the change of OSD had little effect on the flow field upstream of the curtain, and the change mainly affected the outflow position of curtaindownstream water. Thus, all else equal, as the OSD increased, the RWT decreased. But the change

Effect of CP
From the results of A1~A3 and B1~B3 (Table 1), the change of the curtain's position had little effect on RWT, and it only had influence on the stable-time of water after the set of TCC. Thus, in practical engineering, the position of TCC should be determined after considering the complex terrains, economic condition, waterway transport, and other specific cases in actual engineering.

Effect of Pr
Under the difference of thermal stratifications, both experimental tests and numerical simulation results had the same trend-that increasing Pr will increase the RWT dramatically (Figure 8a).
The flow velocity distributions of the water modelled with different Pr values were partly shown in Figure 8b. It can be seen that the flow velocity of the water above the TCC was significantly higher than that of the water in the lower layer. The curtain prevented the lower section of the water column from flowing downstream, thus separating it from the warmer upper water stratum. In addition, the water flow over the TCC was drawn by the tractive action of outlet. The increase of Pr caused the flow area above the TCC to decrease, which increased the flow velocity. As seen from the temperature field (Figure 8b), with the increase of Pr, higher temperature surface water flowed into the plume downstream of the curtain; therefore, the average temperature of this water was higher.
Meanwhile, there were critical values of Pr; we can see from the line graph in Figure 9a that the trend lines levelled off at both ends. There was a sensitive zone where the changes of Pr and RWT were positively correlated, and out of the zone the change of Pr had no effect on RWT; the critical value of Pr was 90% and 65%. This indicated that only the top of TCC was in the thermocline; its ability to regulate RWT was explicit.
As a result, the Pr obviously influenced the surface velocity and had a significant effect on the flow field. Most significantly, the increase of Pr blocked the flow of lower layers, allowing only high-temperature layers to flow around the curtain, thus increasing the RWT.

Effect of OSD
As the OSD increased, the RWT decreased, but the range of variation was less than 1 • C, as shown in Figure 9a.
The OSD values of 44 and 28 cm in the numerical simulation were compared and the flow field distribution was shown in Figure 9b. It was concluded that the change of OSD had little effect on the flow field upstream of the curtain, and the change mainly affected the outflow position of curtain-downstream water. Thus, all else equal, as the OSD increased, the RWT decreased. But the change was not that obvious. It can be explained at this point that the released water mainly gathered by the nearby layer of the outlet, so RWT mainly depended on the thermal stratification of the forebay. As previously seen in Figure 5, the vertical temperature was very close when the depth from the water surface was between 2 and 40 cm. Thus, during the OSD value range set in this study, the change of OSD had little effect on RWT.

Effect of RQ
Under both the S1 and S2 distribution, RWT increased as the RQ increased (Figure 10a). The maximum and minimum values of RQ (11 and 1.6 m 3 /h) were compared and the resulting flow field distributions and temperature field distribution were shown in Figure 10b. The factor of RQ had a great influence on the flow field. For the upstream waters, increasing RQ enlarged the range of current water and caused an increase in mean velocity. For the downstream waters, with the increase of RQ, the pulling force resulting from the discharge of water strengthened and a light eddy appeared between the main downstream flow and the curtain. For the temperature field shown in Figure 10b, the increase of RQ accelerated the flow of higher-temperature water at the top of the curtain, thus reducing the time of water after TCC reached stability. But the influence of RQ on RWT was small. The difference of RWT in experimental tests and numerical simulation was less than 1 • C.
In practical engineering, the RQs were decided with the consideration of downstream flood control capacity, power generation capacity, the benefit of the eco-environment, and so on. RQ may have a large variation range, but the final RWT would not change a lot.

Construction Principle of TCC
To sum up, construction principles for the application of TCC in practice are proposed as follows. First of all, the position of TCC had little influence on the improvement effect of RWT. So it should be determined after considering the complex terrains, economic condition, waterway transport, and other specific cases. Secondly, Pr should be set between 65% and 90%, whereby the larger the Pr chosen, the higher RWT was. Thirdly, the OSD and RQ had little effect on RWT, so in practical engineering its variation will not influence the improving effect of TCC on RWT.

Conclusions
The use of TCC is an effective and practical method to raise the RWT in thermal stratified reservoirs. To provide the theoretical basis for the engineering application of TCC, experimental tests are carried out, whose results fully demonstrated the effectiveness of the TCC in increasing the RWT. For its further research, a 3-D numerical hydrodynamic and water temperature model is built with the same size of the physical model, whose accuracy is verified through comparing with experimental test results. In contrast with previous simulations of TCC, this study describes an arc-shaped curtain, which is closer to reality. Then the influences of different factors on increasing the RWT are studied.
According to the analysis of experimental tests and numerical simulation, the conclusions are summarized as follows: 1.
The regulating effects of TCC on the RWT is closely related to the temperature structure of waters. In S1 distribution, the maximum temperature increased by up to 8.3 • C, while this is only 3.5 • C under the S2 distribution.

2.
The CP has a little difference on RWT, the position of TCC should be determined after considering the complex terrains, economic condition, waterway transport, and other specific cases in actual engineering.

3.
The Pr directly affects the vertical range of water entering the discharge water. As the Pr increases, more upper, high-temperature water flows to the downstream of TCC drawn by the force of outlet. But there are critical values of Pr: When the value is out of the range 90%~65%, the impacts of Pr on RWT is negligible.

4.
When the water is released by spillway, variation of OSD has little effect on the thermal stratification of the forebay, which causes little impact on improving RWT. 5.
The change of RQ has a great influence on the flow field, and reduces the time that the forebay water takes to reach stability. But the effect of RQ on RWT is small; thus, in practical engineering, the variation of RQ can be decided by other complex factors and will not influence the improvement effect of RWT.
Based upon the above conclusions, the following principles are proposed for the application of TCC in thermal stratified reservoirs. The position of the curtain should be decided with comprehensive consideration of complex terrains, economic condition, waterway transport, and other specific cases. In spring and summer, maximize the Pr of the TCC to achieve the best effect on RWT without influencing the flow, and increase the RQ of the water to achieve the final (raised) RWT as soon as possible. The change of OSD and RQ can be decided by actual demand, which will not influence the effect of TCC on improving RWT. For those thermal stratified reservoirs troubled with the problem of RWT, the application of TCC is an effective solution.
Meanwhile, there are many points worth further study. For example, the paper is carried out based on a model with the generalized topography and temperature structure; the influence of real terrains on RWT should be studied in future. In addition, in winter, some reservoirs have the problem of high-temperature of released water, which is the opposite of our focus; the feasibility of TCC's in this condition is worth researching. Furthermore, the impact of secondary flow on wall-bounded turbulence have obtained some achievements [37], and the role of data uncertainties in identifying the logarithmic region of turbulent boundary layers have been researched [38]; the further research on the secondary flow's impact on the flow field of reservoir within TCC can be expected.
The research results presented herein, as well as those of future research for more complex terrains and construction methods, will offer powerful support for the application of TCC to regulate the RWT.

Conflicts of Interest:
The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Appendix A
The photos of experimental model tests are shown in Figure A1.
Appl. Sci. 2019, 9, x FOR PEER REVIEW 13 of 15 5. The change of RQ has a great influence on the flow field, and reduces the time that the forebay water takes to reach stability. But the effect of RQ on RWT is small; thus, in practical engineering, the variation of RQ can be decided by other complex factors and will not influence the improvement effect of RWT.
Based upon the above conclusions, the following principles are proposed for the application of TCC in thermal stratified reservoirs. The position of the curtain should be decided with comprehensive consideration of complex terrains, economic condition, waterway transport, and other specific cases. In spring and summer, maximize the Pr of the TCC to achieve the best effect on RWT without influencing the flow, and increase the RQ of the water to achieve the final (raised) RWT as soon as possible. The change of OSD and RQ can be decided by actual demand, which will not influence the effect of TCC on improving RWT. For those thermal stratified reservoirs troubled with the problem of RWT, the application of TCC is an effective solution.
Meanwhile, there are many points worth further study. For example, the paper is carried out based on a model with the generalized topography and temperature structure; the influence of real terrains on RWT should be studied in future. In addition, in winter, some reservoirs have the problem of high-temperature of released water, which is the opposite of our focus; the feasibility of TCC's in this condition is worth researching. Furthermore, the impact of secondary flow on wall-bounded turbulence have obtained some achievements [37], and the role of data uncertainties in identifying the logarithmic region of turbulent boundary layers have been researched [38]; the further research on the secondary flow's impact on the flow field of reservoir within TCC can be expected.
The research results presented herein, as well as those of future research for more complex terrains and construction methods, will offer powerful support for the application of TCC to regulate the RWT.

Conflicts of Interest:
The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Appendix A
The photos of experimental model tests are shown in Figure a1.