Effects of TDC on thermal regimes and stratification stability
The box diagram in Figure 4A shows the distribution of water temperature at different elevations during the same period as in Figure 2. The elevation of 250 m was close to the surface layer of the reservoir, and the elevation of 220 m was close to the bottom layer. The average and median water temperatures at an elevation of 250 m in Period C were 22.3 °C and 22.8 °C, respectively, which were significantly higher than those in Period A, which were 19.7 °C and 20.4 °C (`p=0.008), respectively. Moreover, the water temperature in Period C had a larger fluctuation range from 14.1 °C to 29.8 °C, which was related to the use of upper water temperature and meteorological factors. As the elevation decreased, the mean and median water temperatures in Period C decreased and were 1.2 °C and 1.4 °C, 2.9 °C and 2.7 °C, and 7.0 °C and 7.1 °C lower than those in Period A, respectively, which was due to the strong buoyancy inhibition of the up moving thermocline in Period C and the more uniformly mixed water of the deep layer. Among them, the 240 m elevation difference between Periods A and C was insignificant according to the paired T test (p>0.05) because the driving force of the water temperature change was gradually transformed from the various dynamic effects of the water surface to thermal diffusion caused by the temperature difference and vertical mixing due to mid-level flow.
The TDC caused changes in water temperature at various elevations (Fig. 4B) and changed the reservoir stability characteristics accordingly. The SI represents the work performed to achieve further mixing over the depth range of the entire water column without adding or reducing heat. Generally, the larger the SI is, the greater the amount of work the reservoir can do to resist buoyancy, and the more stable the water column. With the TDC, the increase in the intake layer reduced the thermal mixing of the bottom water temperature, and the water column was more stable.
Fig. 4. Variation in WT at different elevations (A). (a) 250 m, (b) 240 m, (c) 230 m, and (d) 220 m. The box symbol denotes the mean value of the dataset, with the horizontal lines in the box denoting the 25th, 50th, and 75th percentile values. SI changes (B) in front of the new dam (scatter points) and old dam (columns, red for Period A and blue for Period C). Temperature profile (C) in front of the new dam and old dam in Period C for eight different years. The red dotted line represents the dismantling elevation of the old dam. Accumulated thickness (D) of all layers including the epilimnion, metalimnion and hypolimnion in front of the two dams and the TSI comparisons during period C.
The TDC formed a pool with a storage capacity of approximately 1.5 million m3 between the two dams that was fully connected with the reservoir area in front of the old dam above the demolition elevation of 240.2 m and was blocked with the water bodies beneath 240.2 m. The difference in front of the two dams confirmed the water temperature effect of new dam released water (Fig. 3). From the eight prototype observation data, it can be seen that with the demolition elevation of 240.2 m as the boundary, the water temperature difference between the two vertical lines was small above 240.2 m but significant below it. The largest difference occurred at the bottom of the reservoir, with an average maximum difference of 8.4 °C. The obstruction of the old dam directly caused the water temperature to rise below the elevation before the removal of the new dam. It is worth noting that the difference in water temperature in Sep. was greater than that from Jun. to Aug., only because Fengman Reservoir was a dimictic reservoir where water was turned over twice a year. There was a cumulative effect on temperature stagnation in front of the old dam. The temperature difference continued from the stratification to the back and had a cumulative impact on the temperature hysteresis in front of the old dam.
Based on the analysis of temperature profiles in front of the two dams, the vertical water column was divided into the epilimnion, metalimnion and hypolimnion (FIG. 4D). It can be seen that the average thickness of thermocline layer in front of the old dam was 25.8m, larger than that before the new dam (8.9m) due to insufficient mixing between the bottom and the surface. The thickness of the epilimnion and hypolimnion was opposite. However, the TSI in front of the new dam (0.71 °C/m on average) was larger than that before the old dam (0.67 °C/m), which wasn’t consistent with the rule of SI. The reason was that SI in Figure 4B was the stability of water column 50m underwater, while TSI was the average strength of metalimnion on water column monitored, and the water depth and metalimnion thickness in front of the new dam and the old dam were different. The increase of metalimnion thickness may lead to weak average TSI. Therefore, combined with TSI and SI analysis, it can be seen that the water temperature stability in front of the new dam was significantly weaker than that in front of the old dam after the coexistence of two dams.
Effects of TDC on vertical mixing
During the high-temperature period, heat transfer and vertical mixing affected the downward transfer of heat to the deeper water layer (Prats & Danis, 2019). TDC mainly regulated the thermal state by releasing the epilimnion and metalimnion water bodies, which had a significant impact on the vertical water temperature structure in front of the dam. Figure 5A shows the distribution comparison of thermocline parameters between the TDC and ODC. Selecting a buoyancy frequency equal to 1.5×10-2 s-2 as the threshold, the upper and lower bounds of the thermocline were greater than 1.5×10-2 s-2 and less than 1.5×10-2 s-2 in the beginning, and the difference was the thickness of the thermocline. From Jul. to Aug., when stratification was strongest, the heat exchange between the epilimnion and hypolimnion was restrained, and the vertical mixing was weakened after the epilimnion and metalimnion water bodies were taken. The thermocline was concentrated between 4 m and 30 m underwater, with an average thickness of 26 m. With the ODC, the use of cold water led to a deeper thermocline lower boundary (37 m) and a thicker thermocline (35 m). From 1952 to 1954, the depth and thickness of the thermocline were 37 m and 35 m, 39 m and 39 m, and 47 m and 47 m, respectively.
Fig. 5. Comparison of thermocline parameters between the TDC and ODC, blue represents the thermocline with no TDC, red represents the thermocline with TDC, and purple represents the overlap, which is realized by the transparency of the two colors.
In other reservoirs, selective withdrawal measures also reduced the thickness of the thermocline. After the front retaining wall was adopted in the Dongqing Reservoir, the thickness of the thermocline increased by 11.8 m (Yang, 2021). The change in thermocline thickness was caused by the difference in the thermal state of the reservoir. Due to the temperature dependence of the coefficient of thermal expansion, the temperature gradient was stronger, and the average temperature was higher, resulting in increased stratification (Schwefel et al., 2016).
Using EEMOT, the vertical equivalent elevation points in front of the dam corresponding to the WWT were counted based on 6 groups of data with the ODC (including Feb. and Jun. to Oct.) and 8 groups of data with the TDC (Fig. 6). With the ODC, the equivalent elevation ranged from 221 m to 226.5 m with an average elevation of 223.4 m, which was located 8.4 m above the bottom elevation and 0.4 m above the top elevation of the old water intake. With the TDC, the water was mainly taken from above the demolition gap of the old dam, with an average equivalent elevation of 243.0 m, 2.8 m above the demolition elevation, and the equivalent elevation varied from 241.4 m to 246.7 m. Comparatively, the average equivalent elevation of Period C rose by 19.6 m. This long-term action of taking surface water would further increase the thickness of the hypolimnion of the reservoir after the operation of the new power station to further reduce the heat storage capacity of the reservoir.
Fig. 6. Comparison of the equivalent elevation and operating water level during period A and period C.
Engineering applications and further considerations
The withdrawal mode of TDC was especially proposed for the treatment of sick large dams in large reservoirs, in which there were few cases in the world. However, similar to many selective withdrawal facilities in principle, such as stop-log gates, temperature-control curtains and retaining walls, it raised the position of the mean plug flow and thermocline layer and inhibited vertical thermal mixing, simultaneously improving the withdrawal water temperature and mitigating water environmental hazards caused by deep withdrawal. The old dam acted as a front retaining wall in front of the new dam. Therefore, the water taken from the new dam was taken from the epilimnion, while the water taken from the old dam was taken from the hypolimnion (Fig. 7.), which made up for the defect that the front retaining wall can only use surface water (Yang, 2021). Under the engineering background of the Fengman Reservoir and the expiration of the service life of an increasing number of large dams, there are no available treatment methods. This study can provide a reference for the treatment of sick large dams, which can solve the problem of reservoir diseases and improve the adverse withdrawal water temperature caused by dam construction.
Fig. 7. Schematic diagram of the selective withdrawal effect. Red arrows represent the approximate position of the withdrawal layer.
From the perspective of sensitive objects, there was a rime landscape in winter and fish spawn in summer in the Fengman Reservoir. The maintenance of the rime landscape required long-distance water vapor exchange to ensure that the river channel was not frozen. The WWT in the freeze-up period was 0.2 °C lower than that in the ODC period on average (Fig. 3). In the process of cooling along the way, the time and freezing distance to 0 °C would lead to insufficient water vapor conditions in the river section and threaten the formation of rime. Therefore, how to improve the WWT and eliminate its negative impact on the rime landscape in winter should be considered. Corresponding to the inversion stratification in front of the dam in winter, the water intake of the old dam was low, and the higher temperature water was discharged, while the new dam released the lower temperature water of the epilimnion and metalimnion. Therefore, it was recommended to select the old power station and old dam for operation in winter to increase the WWT. In spring and summer, fish spawn was affected by the low-temperature water taken from the bottom of the old dam. The new dam should be used as much as possible in summer, and high-temperature water can be available.
Additionally, Fengman Reservoir started the construction of the main works in Oct. 2014, and the new power station was put into use in Jul. 2019. During a period of nearly five years, the reservoir ecosystem was in a transition stage from one steady state to another, including the special operation mode of the reservoir and the changes in the thermal structure and thermal state of the downstream river, which directly led to changes in the downstream spawning ground, rime landscape and other sensitive factors, such as water quality factors. The interpretation of this transitional stage also had a significant impact on the research on the treatment of the coexistence of the two dams. Therefore, it is necessary to integrate more prototype observation data to analyze the water temperature and water quality process in the transition stage.