This study is motivated by the need to understand the temperature dynamics and warm-water temperature withdrawal. This study also recognizes the need for an environmental assessment of the proposed temperature control schemes at New Fengman Dam. An unsteady three-dimensional (3D) non-hydrostatic model is used in the present study to predict the hydrodynamics and thermal dynamics in the forebay and intakes of the New Fengman Dam. The numerical model is validated using hydrodynamic data collected from a 1:120 entire physical model and 1:30 local model in the present paper. The numerical and experimental results indicate that the Stop Log Gate has no effect on warm-water withdrawal. After dam reconstruction, the reservoir release temperatures can be increased from 4.0 °C to 6.0 °C to improve the habitat of native fishes. The cancellation of the Stop Log Gate program is recommended; the old dam meets the performance objectives for temperature increase.

INTRODUCTION

The presence and operation of hydropower plants in natural environments cause several ecological problems. In reservoirs, the outlets might release water that is too warm or too cold for downstream ecosystems, depending on reservoir temperature at the intake. It is extremely detrimental to the health of fish in rivers, especially during their migration seasons (Van Der Kraak et al. 1997; Deng et al. 2004; Thorpe 2004; Vermeyen 2006; Sherman et al. 2007; Olden & Naiman 2010). Managing reservoirs explicitly for downstream temperatures is an increasingly important objective of reservoir operators for riverine ecosystem management.

With important implications of water temperature for biotic responses, it became clear that the thermal characteristics of reservoirs play a crucial role in managing downstream temperatures (Caissie 2006). The thermal stratification of a reservoir is the result of various physical processes that distribute heat from the reservoir surface to its deeper layers (Casamitjana et al. 2003; Olden & Naiman 2010; Rheinheimer et al. 2015). These processes depend not only on meteorological variables such as wind velocity, and short and long wave radiation, but also on the biochemical characteristics of the water body.

Many numerical studies related to water temperature have been conducted over the years to study the thermal characteristics in reservoirs. Dake & Harleman (1969) conducted experiments and analyzed thermal stratification in lakes. Orlob & Selna (1970) constructed a mathematical model to study the thermal behavior of freshwater impoundments. Their models were based on one-dimensional (1D) vertical transport equations, and they considered net heat inputs from the atmosphere to compute the heat flux from solar radiation, convection, and evaporation. Imberger et al. (1978) proposed a Lagrangian 1D model of the dynamics for medium-size reservoirs and proposed the mixed layer concept for the first time. Chen et al. (1998) employed a modified prediction model for the Miyun Reservoir. Li & Shen (1990) utilized a 1D vertical model to predict the temperature in the Heihe Reservoir. Vermeyen (2006) used a similar approach in evaluating the selective withdrawal system of the Hungry Horse Project. Casamitjana et al. (2003) employed a 1D model to predict the thermal structures of reservoirs during maximum water demand. Based on these studies, the 1D vertical model considers only vertical stratification and cannot be used to indicate horizontal and longitudinal variation.

A 2D model should be used to study stratification dynamics subject to horizontal advection. Laterally averaged 2D models have been extensively applied to predict water dynamics in reservoirs that have slight changes in their lateral indices. Farrel & Stefan (1988) employed a 2D standard K-epsilon (κ-ɛ) turbulence model to predict density current and temperature stratification. Deboltsky & Neymark (1994) modeled reservoir temperatures using a 2D thermal model. Bormans & Webster (1998) predicted the dynamics of temperature stratification in lowland rivers using a 2D model that considers radiative fluxes at free surfaces. Chen et al. (1997) utilized a 2D turbulent model to predict temperature in the Hongfeng Lake. Deng et al. (2003a, 2004, 2006) applied the Navier–Stokes and energy equations with the Boussinesq assumption to predict the water temperature in the Zipingpu, Ertan, and Xiluodu reservoirs. Such a 2D model considers the buoyancy effect, short and long wave radiation, and heat flux by evaporation at the free surface. Ma et al. (2008) used a laterally averaged 2D CE-QUAL-W2 reservoir model to predict the thermal stratification and water temperature profiles in the Kouris Dam in Cyprus.

2D models obtain superior results in predicting water temperature in reservoirs. However, water flow in complex geometries (e.g., the intake structure region) is highly three-dimensional. Moreover, the additional complexity caused by thermal stratification cannot be accurately captured by traditional depth or width-average models. Therefore, 3D models are used to predict flow fields and temperature dynamics, particularly where they accurately predict complex domains with several hydraulic structures. Deng et al. (2003b) compared the 3D Spalart Model, Standard κ-ɛ Model and Reynolds Stress equation Model (RSM) in simulating the thermal density current in a reservoir; the Standard κ-ɛ Model was quick to converge and suitable for engineering applications. Haque et al. (2007) utilized a 3D Reynolds-averaged Navier–Stokes (RANS) turbulence model to simulate the flow and temperature distributions within the powerhouse units and forebay of the McNary Dam on the Columbia River. Politano et al. (2006) developed a 3D hydrodynamic and heat transport model to predict water temperature on the McNary Dam. An incompressible RANS solver was utilized for buoyant flows, and the Boussinesq approach and a Standard κ-ɛ Model with wall functions were employed for solar radiation and convective heat transfer at the free surface. Politano et al. (2008) used a similar approach to predict temperature profiles in the forebay and within the gatewells of the McNary Dam. Ren et al. (2007) utilized a 3D thermal model based on the arithmetic operator splitting technique to predict the temperature distribution in the Ertan Reservoir. Jin et al. (2000) employed the environmental fluid dynamics code (EFDC) to predict temperature profiles along water columns in Lake Okeechobee. Between others, Çalışkan & Elçi (2009) and Gao et al. (2012) used EFDC to simulate the hydrodynamics of a stratified reservoir.

Except for many numerical studies, there were also a few experimental researches on the hydrodynamics of a stratified reservoir. Johnson (1980) studied the density current using an experimental flume. Ettema et al. (2004) used a unique hydraulic model to simulate the stratified water-temperature condition. Gao et al. (2010) developed a reservoir model to predict the water temperature released from the multi-level intake. Physical model tests of water temperature were also conducted by Liu et al. (2012) to analyze the multi-level power intake of the hydropower station.

Water temperature has been modeled at various levels of complexity. Note that numerous studies have been conducted to characterize thermal stratification in lakes and oceans, but only a few numerical studies focus on temperature dynamics in hydropower reservoirs, particularly on water temperatures released from the forebay. Moreover, in -previous studies, the grid is either based on the σ-coordinate system or too large in scale. As such, some hydraulic structures are simplified for computation, which in turn increases errors for complex flows. Since the flow in the forebay near the powerhouse and into the intake structures is an extremely complex and highly 3D flow, both vertical motions and accelerations cannot be ignored, and the use of the hydrostatic assumption is not valid here. Given the lack of appropriate models, the present study develops a full 3D thermal model for predicting complex hydrodynamics and temperature distribution using a state-of-the-art computational fluid dynamics (CFD) tool. In addition to that, the present study also develops two hydraulic models to study the hydraulic and water temperature dynamics. No prior hydraulic model of comparable size and complexity has been used for studying the management of warm-water release from a thermally stratified reservoir.

STUDY AREA

Fengman Dam (43°43′N, 126°41′E), built in 1937, is one of the largest hydroelectric power facilities in China located on the Songhua River. Figure 1 shows the location of Fengman Dam. The State Electricity Regulatory Commission defined Fengman Dam in 2007 as a ‘sick dam’ because of some structural problems that compromised safety. Given the importance of Fengman Dam and the seriousness of possible accidents it may cause, the National Development and Reform Commission approved a reconstruction program for the dam in 2010. The new dam rebuild project started, in 2012, and is located 120 m downstream of the old dam. The old dam is a concrete gravity dam, and the maximum value of the dam height is 91.7 m with a multiyear-average annual inflow of 442 m3/s. For the old Fengman power station, the multiyear-average annual natural runoff volume is 13.9 billion m3, having an installed capacity of 1,002.5 MW. The maximum value of the new dam height is 94.9 m, and the multiyear-average annual inflow and natural runoff volume are also 442 m3/s and 13.9 billion m3, respectively. However, the installed capacity has increased from 1,002.5 MW to 1,480 MW for the new dam. The new dam is mainly used for power generation, but is also used for flood control, irrigation, urban water supply, and environmental protection. In order to reduce the impact on the ecological environment downstream as far as possible, only sections of the old dam with an elevation of 237.5 m and a length of 594.0 m were demolished. This reconstruction program is rare both in China and the world. Figure 2 presents an aerial view of Fengman Dam and its forebay.
Figure 1

Map of the Songhua River Basin showing the location of Fengman Dam.

Figure 1

Map of the Songhua River Basin showing the location of Fengman Dam.

Figure 2

Aerial view of the Fengman Dam and its forebay: (a) 3D view of the old dam, (b) location and orientation of the old and new dam (simulated).

Figure 2

Aerial view of the Fengman Dam and its forebay: (a) 3D view of the old dam, (b) location and orientation of the old and new dam (simulated).

From the power station operation before, several ecological problems have been observed to occur. The lowest migration temperature needed for Xenocypris argentea Gunther is 16 °C, and 18 °C and 20 °C for Cyprinus carpio and Aristichthys nobilis, respectively. However, the tailwater temperature withdrawn from Fengman Dam is only 10–15 °C, which is too cold to delay their migration season. Scientists believed that increasing the water temperature in the river would improve the habitat for the native fishes (Vermeyen 1999). Because of that, the Ministry of Environmental Protection recommended that the Fengman Dam Reconstruction Engineering Construction Bureau investigate ways to control the temperature of water released from New Fengman Dam in future to solve the ecological problems. Selective water withdrawal from different depths of a dam is widely used in managing reservoirs to meet the demands of different water usages. When water is withdrawn from different depths in the dam, the flow patterns and water temperature in a reservoir are usually altered. At present, a selective withdrawal system is controlled by the Stop Log Gate, which is shown in Figure 3. The workers can adjust the layer number (Ln) of the Stop Log Gate to control the temperature of water released from the forebay.
Figure 3

Representation of selective withdrawal system for Fengman Reservoir.

Figure 3

Representation of selective withdrawal system for Fengman Reservoir.

The scope of this work is the experimental and numerical study of hydrodynamics and temperature distribution in the New Fengman Dam forebay. The main goal is to predict the water temperature discharged from the powerhouse and to ultimately propose solutions to increase the water temperature. Temperature dynamics with and without the Stop Log Gate are presented and discussed.

MATHEMATICAL MODELING

The 3D model is based on the commercial code FLUENT 6.3 (Fluent Inc., USA). The implicit RANS solver employs a cell-centered finite-volume scheme and a hybrid unstructured mesh. The continuity equation is satisfied using the SIMPLE pressure–velocity algorithm. An additional scalar transport equation is solved to account for temperature effects through the Boussinesq assumption (Haque et al. 2007). The boundary conditions are programmed through user-defined functions.

This numerical model includes the main features of the New Fengman Dam, including the forebay, old dam, six powerhouse units, and new dam (which will be built in the future). The length and width of the forebay are 1.5 km and 1.8 km in the present model, respectively. Mean depth is approximately 80 m, and free surface area is 2.58 × 106 m2. The demolished elevation and length of the old dam is 237.5 m and 594.0 m, respectively. Figure 4 shows the simulation of the main power structures and forebay of New Fengman Dam.
Figure 4

Illustration of the main power structures and forebay of the New Fengman project.

Figure 4

Illustration of the main power structures and forebay of the New Fengman project.

The computational grid scale should reflect the effect of the hydraulic structure and topography on the flow and temperature fields in the simulated region. Economic feasibility should also be considered. Thus, a hybrid grid with approximately 1 million cells was employed to mesh the forebay and intake regions. The longitudinal and lateral grid sizes of the forebay region were 5 m to 30 m and 5 m to 10 m, respectively. The minimum and mean sizes at depth direction were 0.25 m and 0.3 m, respectively. The complex geometry regions near the powerhouse were the key positions for the present calculation, and required refined grids with minimum and maximum sizes of 0.25 m * 0.5 m * 0.5 m and 1 m * 1 m * 1 m, respectively. The cell sizes along the vertical direction, except near the bottom and free surfaces, were decreased. Mesh refinement along the vertical direction near the free surface was conducted to resolve the strong thermal gradients, which was essential to ensure consistency with the field data. Figure 5 shows the mesh for the New Fengman Hydropower Station.
Figure 5

(a) 2D mesh in all numerical regions, (b) 3D mesh in the intake area, (c) longitudinal section X–X showing mesh in one of the intake bays.

Figure 5

(a) 2D mesh in all numerical regions, (b) 3D mesh in the intake area, (c) longitudinal section X–X showing mesh in one of the intake bays.

Hydrodynamic model

The flow field is solved with the incompressible RANS equations using the Boussinesq approach.

Continuity equation: 
formula
1
Momentum equation: 
formula
2
where represents velocity, acceleration due to gravity, p fluid pressure, T temperature, β the thermal expansion coefficient, and effective viscosity, with and denoting molecular and eddy viscosity, respectively. represents water density at reference temperature T0.
The area between the old and new dam causes the flow to be highly three-dimensional, along with vortexes and backflow, and exhibits strong anisotropy. The field data on the Fengman Reservoir shows a low velocity, particularly in the bottom area, which results in a low Reynolds number. The standard κ-ɛ model can simulate a fully developed turbulent flow well, which fits a high Reynolds number with local isotropy (Rodi 1984). However, the standard model fails when used for flows with a Reynolds number. The RNG (renormalization group)-based κ-ɛ turbulence model is derived from the Navier–Stokes equations using a rigorous statistical technique (i.e., renormalization group theory). The analytical derivation results in a model with constants that differ from those in the standard κ-ɛ model; the model also has additional terms and functions in the transport equations for κ and ɛ (Fluent Inc. 2006). RNG provides an analytically derived differential formula for effective viscosity to account for the effects of the low Reynolds number, thereby allowing the model to handle low Reynolds numbers and near-wall flows (Yakhot & Orzag 1986). Given the advantage of RNG, the RNG κ-ɛ turbulence model is applied with standard wall functions in the present study. Kinetic energy κ and turbulent dissipation rate ɛ are obtained from: 
formula
3
 
formula
4
In Equation (3), Gb denotes the generation of turbulent kinetic energy caused by buoyancy, which is given as , where represents the Prandtl number. The buoyance term Gb is the key factor when the stratification is stable because it can restrain the generation of turbulence and decrease the heat that is passed down. In the model, the default constants , , , , and are used.

Temperature model

Temperature is calculated from the energy conservation equation for incompressible flows: 
formula
5
where represents effective conductivity, with and denoting molecular and turbulent thermal conductivity, respectively.

Boundary conditions

Free surface

Considering the lower water surface fluctuation in the Fengman forebay, in the present study the free surface was considered as a flat surface, which was modeled using a rigid-lid approximation. The heat transfer at the free surface includes short wave radiation, long wave radiation, evaporation, and the convection process. The transfer process is shown in Figure 3. In the present study, the evaporation and heat transfer coefficients are used to calculate the heat flux at the free surface (Chen & Mao 1990).

The evaporation coefficient is calculated from 
formula
6
and the heat transfer coefficient at the free surface is calculated from 
formula
7
with in and in .

In these equations, b is calculated from , where P and v are the atmospheric pressure (hPa) and wind speed (m/s) at 1.5 m above the water surface, respectively. is the radiation coefficient at the water surface, with the constant value of 0.97.

is the Stefan–Boltzmann constant: .

is given by 
formula
8
where Ta is the atmospheric temperature (°C) at 1.5 m above the water surface, and Ts represents the temperature at the water surface (°C).
The heat loss by evaporation is calculated from 
formula
9
where es is the saturated water vapor pressure (hPa) at the water surface temperature Ts, which is calculated from 
formula
10
ea is the water vapor pressure (hPa) at 1.5 m above the water surface, which is calculated from 
formula
11
In this equation, Td is the dew point temperature (°C), which is computed from relative humidity RH and atmospheric temperature Ta: 
formula
12
The coefficient k is provided by the saturated water vapor pressure es, which is calculated from (10) and the temperature at the water surface Ts: 
formula
13
According to meteorological data, such as atmospheric pressure, wind speed, vapor pressure and so on, it is calculated that the heat transfer coefficients in May, June, July, and August are 25.78 W·m−2·°C−1, 30.05 W·m−2·°C−1, 33.82 W·m−2·°C−1 and 32.21 W·m−2·°C−1, respectively.

Inlet

The total power discharge and measured temperature profiles are specified at the upstream inflow section. During May, June, July, and August, the average power discharges during daily hours were 2,012 m3/s, 1,846 m3/s and 1,721 m3/s in a high flow year (HFY), median flow year (MFY) and low flow year (LFY), respectively. The corresponding Reynolds numbers based on the inflow hydraulic diameter are Re = 1.1 × 106, 1.0 × 106 and 0.9 × 106, respectively. The turbulent variables are assumed as zero at the upstream end.

Warm-water withdrawals would likely occur during the months of May, June, July, and August. Based on the field data measured in front of the dam from 1951 to 2004, the multi-year mean variation of water temperature against depth during different months is shown in Figure 6, which specifies the temperature profiles at the inlet. In spring and summer, the reservoir surface heats up from incoming solar radiation and conduction from warm air, forming a warm upper layer (epilimnion) and a cold lower layer (hypolimnion), which are separated by a transitional zone (thermocline). In the epilimnion, such as in the Fengman Reservoir, it is formed at approximately the first 30 m beneath the free surface, gradually. There is often a clear temperature gradient in the epilimnion, which depends on mixing by surface winds or other mechanical forces and reservoir depth. As shown in Figure 6, the temperature gradients in May, June and July are 0.2, 0.59 and 0.72 °C/m, respectively. Because of the characteristics of atmospheric conditions, the temperature gradient becomes greater, and the epilimnion and hypolimnion are not distinct any longer in August.
Figure 6

Water temperature against depth.

Figure 6

Water temperature against depth.

Outlet

The outlets are defined as outflows with a specified discharge. The river discharge was distributed evenly to the six powerhouse units, such as each unit operated at 335 m3/s in HFY, and 308 m3/s and 285 m3/s in MFY and LFY, respectively.

Walls and river bed

A no-slip condition and zero heat flux are imposed on all the walls and on the forebay bed. The standard wall function is used for the viscosity layer near the wall. The value of the roughness coefficient for the river is 0.033, and is 0.014 for the concrete structures.

Numerical method

According to the experimental results, the influenced range of temperature distributions is less than 600 m from the old dam, which implies zero or slight changes in temperature distributions before reconstruction. Variations are observed only near the intake structure. Therefore, the temperature distribution and water level are initialized in full field to accelerate the convergence. After a retreatment of the Fengman Dam, the characteristics of the water level do not change, which means that an operation from 242.0 m to 263.5 m is maintained. Figure 7 shows the variation of water level with three kinds of flow years, which are HFY, MFY and LFY, respectively.
Figure 7

Water level variation against flow years (high, median, and low).

Figure 7

Water level variation against flow years (high, median, and low).

Numerical calculations were performed at the China Institute of Water Resources and Hydropower Research (IWHR). The solutions were obtained using a fixed time step of 25 s. Most of the simulations employing meshes with 1 million cells were run in real time using the four processors of a Linux PC cluster. The average time to get a solution starting from scratch was about three days. This is a reasonable time, which allows the model to be used as a design tool by the hydropower industry.

EXPERIMENTAL METHODS AND FACILITIES

The hydraulic model encompassed an 8.0 m by 6.0 m area, as shown in Figure 8. The area of the model was needed to ensure that the model had a sufficient volume of warm water and cold water to maintain the temperature profile in the model, and thereby enable water to outflow from the model intake and attain a steady temperature. The model also had to be adequately deep so that the flow field could be simulated accurately. The hydraulic and temperature dynamics had to be considered in the model. After considering some constraints including the Reynolds number of the flow approaching the intake, a scale of Xr = 120 (=prototype 1ength/model length) for the entire reservoir experiment (Figure 8(a), six powerhouse units) and Xr = 30 for the local experiment (Figure 8(b), three powerhouse units) were selected. Considering roughness similarity, the sidewalls and bed were made of concrete (with roughness coefficient n = 0.014), and the intake structures and Stop Log Gate were made of plexiglass (with roughness coefficient n = 0.008).
Figure 8

Experimental set-up model: (a) photograph of entire model experiment; (b) photograph of local model experiment.

Figure 8

Experimental set-up model: (a) photograph of entire model experiment; (b) photograph of local model experiment.

The physical model included three parts, which were the supplying zone, heating zone and testing zone as shown in Figure 8(a). In the supplying zone, water was supplied by a 12 kW capacity and 0.10 m3/s rated discharge pump, and the discharge was measured with an in-line magnetic flow meter. In order to provide a relatively uniform stream at the upstream, a special entrance device (pipe and plate with an orifice) was used in the supplying zone. In the present experiments, a special thermal system was designed in the heating zone. The thermal system was composed of ten layers of U type heating rods and thermal baffles between the layers. The U type heating rods had three levels of power, 0.5 kW, 1.0 kW and 2.0 kW. The working power levels were adjusted according to the experiments needed. Five monitoring sections were set in the testing zone, which were respectively 5.8 m, 4.2 m, 2.5 m, 1.2 m and 0.5 m in front of the New Fengman Dam. Water temperatures were collected in real time and dealt with using a DJ800 data collecting system, which was designed by the hydraulics department of IWHR (IWHR 2010). The temperature measurement was accurate within ±0.1 °C, with a maximum range of 50.0 °C, which could meet the demand. The water temperature withdrawal was measured at the downstream of the tail pipe. To simulate the density stratified flow in this intake, a conversion formula of temperature stratification between the model and prototype was derived by buoyancy similarity (Liu et al. 2012). Only when the real-time collected water temperature distribution in the model was consistent with the distribution of that in the prototype, were water temperatures withdrawal collected in the next 120 seconds. In order to ensure the accuracy and representativeness of the results, the collected time interval was set as 0.1 s, in other words, there were 1,200 experimental points every time. Therefore, the water temperature withdrawal in a run was equal to the average value of the 1,200 experimental points.

In the local model, the reservoir was made of steel plate, and the intake structures and Stop Log Gate were also made of plexiglass. The velocity profiles were measured using a P-EMS electromagnetic flow velocimeter with an accuracy of 0.002 m/s and 2.5 m/s of maximum velocity. In any experiment, the discharge and the water level in the forebay were maintained constant so as to keep inflow conditions constant. The experimental conditions are shown in Table 1.

Table 1

Experimental conditions in prototype

  Month Water level (m) Discharge (m3·s−1Stop Log Gate (m) 
LFY May 248.40 6 × 335 12/0 
Jun 248.31 6 × 335 12/0 
Jul 249.34 6 × 335 12/0 
Aug 251.85 6 × 335 15/0 
MFY May 260.05 6 × 308 24/0 
Jun 259.22 6 × 308 24/0 
Jul 258.90 6 × 308 21/0 
Aug 260.99 6 × 308 24/0 
HFY May 260.76 6 × 285 24/0 
Jun 260.50 6 × 285 24/0 
Jul 260.50 6 × 285 24/0 
Aug 261.17 6 × 285 24/0 
  Month Water level (m) Discharge (m3·s−1Stop Log Gate (m) 
LFY May 248.40 6 × 335 12/0 
Jun 248.31 6 × 335 12/0 
Jul 249.34 6 × 335 12/0 
Aug 251.85 6 × 335 15/0 
MFY May 260.05 6 × 308 24/0 
Jun 259.22 6 × 308 24/0 
Jul 258.90 6 × 308 21/0 
Aug 260.99 6 × 308 24/0 
HFY May 260.76 6 × 285 24/0 
Jun 260.50 6 × 285 24/0 
Jul 260.50 6 × 285 24/0 
Aug 261.17 6 × 285 24/0 

NUMERICAL AND EXPERIMENTAL RESULTS

The numerical and experimental results were compared with each other. In the following, case I is used here to refer to the scheme with the Stop Log Gate, and case II the scheme without the Stop Log Gate. Because the maximum gradient in the epilimnion occurred during July, the flow and temperature characteristics in July were selected in the present study as representative to analyze.

Velocity profiles in Fengman forebay

The water velocity profiles were conducted at different typical years. The profiles of relevant sections were compared with each other. Figure 9 shows a comparison of the predicted and measured velocity magnitudes at sections A and B, and Figure 10 presents the section positions. The agreement among the velocity magnitudes of all the sections was extremely good.
Figure 9

Comparison of the predicted and measured velocity magnitude at sections A and B: (a) section A and (b) section B ((-1) represents case I, while (-2) represents case II).

Figure 9

Comparison of the predicted and measured velocity magnitude at sections A and B: (a) section A and (b) section B ((-1) represents case I, while (-2) represents case II).

Figure 10

Section positions in the Fengman Reservoir.

Figure 10

Section positions in the Fengman Reservoir.

Figures 11 and 12 show the predicted velocity vector field for July in typical flow years. Two reference vectors at 0.1 m/s and 0.5 m/s were used because of the magnitude difference between the old (less than 0 coordinate axis in longitudinal) and new (more than 0 coordinate axis in longitudinal) reservoir. The old and new reservoirs are defined in Figure 10.
Figure 11

Velocity vector fields from the numerical calculation for July in typical flow years with the Stop Log Gate: (a) HFY, (b) MFY, and (c) LFY.

Figure 11

Velocity vector fields from the numerical calculation for July in typical flow years with the Stop Log Gate: (a) HFY, (b) MFY, and (c) LFY.

Figure 12

Velocity vector fields from the numerical calculation for July in typical flow years without Stop Log Gate: (a) HFY, (b) MFY, and (c) LFY.

Figure 12

Velocity vector fields from the numerical calculation for July in typical flow years without Stop Log Gate: (a) HFY, (b) MFY, and (c) LFY.

Figures 11(a) and 12(a) present the velocity vector field for two cases, respectively, in HFY. The water withdrawal elevation was 244.0 m for case I and 220.0 m for case II. The high velocity region in the old reservoir ranged from the free surface to 40 m underwater with a big backflow beneath. The largest velocity value was found at the free surface, and the minimum and maximum values were 0.05 m/s and 0.15 m/s, respectively. The velocity distributions for two cases in the old reservoir were similar, whereas those in the new reservoir exhibited several differences. Figure 11(a) shows that the velocity distribution for case I was an inverted triangle. The region had a higher velocity than the top of the Stop Log Gate. The velocity value rapidly decreased beneath the top of the Stop Log Gate and reached approximately zero near the bottom. The velocity distribution for case II was different. The withdrawal elevation decreased from 244.0 m to 220.0 m, which implies a uniform velocity distribution in the new reservoir. Figure 12(a) shows that the velocity values above 220.0 m had slight differences. A comparison of velocity values at the withdrawal section of the two schemes shows that maximum velocity occurred at the median for case I and at the bottom for case II. Consequently, the withdrawal schemes only slightly affected velocity distribution in the upstream region but significantly affected the region, between the old and the new dam.

Figures 11(b), 11(c) and 12(b), 12(c) present the velocity vector field in MFY and LFY, respectively. The velocity distributions were similar to those in HFY. The water level was lower in LFY than that in other years; the water head at the old dam was small, and the withdrawal elevation in case I decreased from 244.0 m to 232.0 m. The high velocity region in the old reservoir ranged from the free surface to 30 m underwater in LFY with a decrease of 10 m. Figure 11 also shows that the Stop Log Gate only slightly affected velocity distributions in the old reservoir. A comparison of the schemes shows that the velocity value decreased to approximately 1.0% error at sections 0–600 to 0–700. Thus, the range influenced by the withdrawal scheme was approximately 650 m in front of the old dam of the Fengman Hydropower Station.

Temperature profiles in Fengman forebay

The numerical and experimental water temperature profiles at relevant sections were compared with each other. The simulated temperature magnitude profiles (continuous line) at the two sections (the positions of the sections are indicated in Figure 10) situated upstream and downstream of the old dam were compared with the measurements (symbols), as shown in Figure 13. It was pointed out that the predicted results were consistent with the scale model data.
Figure 13

Comparison of the predicted and measured water temperature magnitudes at sections A and C: (a) section A, (b) section C ((-1) represents case I, while (-2) represents case II).

Figure 13

Comparison of the predicted and measured water temperature magnitudes at sections A and C: (a) section A, (b) section C ((-1) represents case I, while (-2) represents case II).

Figures 14 and 15 show the predicted temperature contours in July in typical flow years. The water temperature profiles slightly changed in the old reservoir during the withdrawal period with two schemes; this result is in accordance with the experimental results of the 1:120 physical model. Similar results were also obtained from the numerical and experimental reports for the Jinping Project, which is another large hydropower station in China (Deng 2003; Liu et al. 2012).
Figure 14

Temperature contours from the numerical calculation for July in typical flow years with Stop Log Gate: (a) HFY, (b) MFY, and (c) LFY.

Figure 14

Temperature contours from the numerical calculation for July in typical flow years with Stop Log Gate: (a) HFY, (b) MFY, and (c) LFY.

Figure 15

Temperature contours from the numerical calculation for July in typical flow years without the Stop Log Gate: (a) HFY, (b) MFY, and (c) LFY.

Figure 15

Temperature contours from the numerical calculation for July in typical flow years without the Stop Log Gate: (a) HFY, (b) MFY, and (c) LFY.

Figures 14(a) and 15(a) show the temperature contours in HFY with the two schemes, respectively. Note that the temperature contours in the old reservoir were horizontal. The water temperature distribution exhibited only slight changes at a certain distance from the old dam, which is a decrease in the value of the water temperature at the free surface relative to the far field. The value decreased from 26.6 °C to 24.6 °C near the old dam with a variation of approximately 2.0 °C. The variation value at the free surface decreased as the distance increased. No effect was observed on the temperature distribution when the distance was approximately 600 m in front of the old dam. The distribution in the new reservoir significantly differed from that in the old reservoir. The thermal stratification in the new reservoir was weaker than that in the old reservoir, and the difference between the surface water and bottom temperature significantly decreased. The difference in case I decreased from 19.3 °C to 5.9 °C, and that in case II from 19.3 °C to 5.3 °C. The water in the new reservoir was extremely turbulent during the withdrawal period. Different temperature distributions formed in this region because of the turbulence and warm water in the old dam. Figures 14(b), 14(c) and 15(b), 15(c) present the temperature contours in MFY and LFY, respectively.

A comparison of cases I and II shows that the temperature distributions were nearly the same in the old reservoir. Thus, the Stop Log Gate only slightly affected the temperature distribution in the old reservoir. However, the temperature contours in the new reservoir were different for the two cases. The average temperature in case I was lower than that in case II, with values of 18.5 °C and 19.4 °C in HFY, respectively. The water temperature at the bottom was 16.8 °C in case I and 17.6 °C in case II. Therefore, the Stop Log Gate had no role in warm-water withdrawal in the New Fengman Project. This case differs from cases where power stations use the Stop Log Gate to obtain warm water (e.g., Jinping hydropower station) (Liu et al. 2012).

Water temperature withdrawn by different schemes

Warm-water withdrawals would likely occur during the months of May, June, July, and August. Figure 16 shows the average water temperature released from the new dam obtained from numeration and experimentation. The maximum and minimum errors between numerical and experimental are 4.4% and 0.0%, respectively. However, the difference between case I and case II obtained by numerical and experimental methods is only 0.1 °C for the maximum. It is pointed out that the numerical and experimental data were in good agreement with each other, which indicated the results were precise. The mean temperatures released from the old dam before reconstruction were 6.9, 10.7, 14.5, and 17.8 °C in May, June, July, and August, respectively. According to the aforementioned temperature needed for native fishes, the water temperature should be higher than 16.0 °C, and some should be higher than 20.0 °C (Refer to part 2). It was indicated that the temperature condition could not meet the requirements of lots of native fishes. By contrast, the temperatures from the new dam were all higher than the previous temperatures, particularly in June, July, and August. It is researched that the temperatures should increase from 4.0 °C to 6.0 °C after Fengman Dam is rebuilt. The water temperatures withdrawn with and without the Stop Log Gate during July and August all met the temperature requirements for native fishes in typical years.
Figure 16

Release temperature in different typical years (°C).

Figure 16

Release temperature in different typical years (°C).

A comparison of cases I and II shows that the temperatures withdrawn were almost the same; sometimes, the temperature value in case II was even higher than that in case I. The maximum and minimum absolute differences were 0.3 °C and 0.0 °C, respectively. The numerical results were in good agreement with those obtained from the experiment, which indicates that the effect of warm water withdrawn using the Stop Log Gate was insignificant. These results suggest that the Stop Log Gate program be cancelled.

DISCUSSION AND ANALYSIS

The demolished gap and new reservoir were the main factors that affected velocity profiles at the upstream of the intake, which in turn affect warm-water withdrawal. The elevation of the demolished gap in the old dam was 237.5 m, and water depth above the gap was from 10 m to 20 m. The velocity within this range was relatively large in the old reservoir, with a maximum magnitude of 0.12 m/s to 0.15 m/s. The value gradually decreased with depth. Because of radiation and evaporation and so on, the water temperature at the surface was higher than that at the bottom during the months of May, June, July, and August. The high temperature layer was at approximately the first 30 m beneath the free surface, which agreed with the high velocity layer. As such, an increased amount of warm water was injected into the new reservoir because of the presence of high surface velocity and the old dam. The mean temperature in the new reservoir was consequently higher than that in the old reservoir, and warm water was withdrawn downstream. Therefore, the water temperature withdrawn would be higher than that through the old Fengman Dam. The reservoir release temperatures before the reconstruction were approximately 7.0, 10.7, 14.5, and 17.8 °C in May, June, July, and August, respectively. These temperatures could be increased by as much as 4.0 °C to 6.0 °C to improve the habitat of native fishes.

For a traditional multi-level intake project, the water from the top layer would be withdrawn using a Stop Log Gate, which contained high temperature water. By contrast, the main flow occurred in the middle layer for the withdrawn scheme without a Stop Log Gate. As such, the water temperature through the Stop Log Gate would be higher than that without the Stop Log Gate. However, the water temperatures withdrawn for cases I and II only had slight differences in the New Fengman Hydropower Station, unlike in the case of a traditional multi-level intake project. As shown in Figure 16, the maximum absolute difference was only 0.3 °C, and the water temperature for case II was higher than that in case I under some conditions, due to the topographical particularities. It is known that the water temperature discharged depends on the thermal flux from the upstream. Based on the thermal equilibrium principle, the heat withdrawn downstream was equal to that through the gap in the old dam plus the heat that remained in the region between the old dam and the new dam minus the heat exchanged between the water body and the environment. Because of the short distance between the old dam and the new dam, the latter had a slight influence on the total heat flux, which could be neglected. During the New Fengman Hydropower station operation, the temperature distribution in the new reservoir became stable, i.e. the heat in this region was a constant. As such, the heat through the gap directly determined the value withdrawn downstream, i.e. the average temperature through the gap was approximately equal to that withdrawn downstream. Figure 17 shows the temperatures through the gap, and it is indicated that the values that occurred in case I and case II are nearly same. The withdrawal scheme had a slight influence on the water temperature withdrawn downstream for the New Fengman Hydropower Station.
Figure 17

Average temperature through the gap in HFY (°C).

Figure 17

Average temperature through the gap in HFY (°C).

Figure 16 also shows that the water temperatures withdrawn in various typical years were different, which may be caused by the initial water level. Figure 7 shows that the difference in the water level in HFY and MFY was relatively small, while the decrease in the water level in LFY was relatively large, particularly above the gap. The water level was approximately 20 m above the gap in HFY and MFY but only around 10 m in LFY. Thus, the water temperature above the gap in LFY was higher than that in other typical years. Through the analysis in the above paragraph, it can be deduced that the water temperatures withdrawn would be higher in LFY than those in other typical years.

CONCLUSIONS

The Fengman power project differs from other traditional projects and will result in two dams after the reconstruction. However, the key problems in this reconstruction project are ecological. The flow and temperature dynamics in the old and new reservoirs of the Fengman Hydropower Station were numerically studied using an unsteady, non-hydrostatic 3D model. This study predicted the temperature distribution in the reservoir and determined the water temperature withdrawn downstream. This study also compared two cases and planned and drafted an environmental assessment of the proposed schemes for temperature increase. The modeling approach and main characteristics of the CFD model were presented.

The numerical model was validated through the hydrodynamic data collected using 1:120 and 1:30 physical models. The velocity and temperature profiles of the reservoir were satisfactory, as predicted by the model. The distributions of the velocity vector field of the two cases in three typical flow years in different months were simulated. The distributions were not affected by the Stop Log Gate scheme. Relative to the experimental data, the range influenced by the withdrawal scheme was approximately 650 m in front of the old dam for the New Fengman Hydropower Station. Similar results were also obtained from the numeration of temperature distributions.

The water temperatures withdrawn in the two cases were evaluated for the New Fengman Hydropower Station. The numerical and experimental results show that the water temperatures withdrawn were nearly the same. The Stop Log Gate had no effect on warm water withdrawal. From the reconstruction of the dam, the reservoir release temperatures could be increased by as much as 4.0 °C to 6.0 °C to improve the habitat for native fishes. This study indicates that the Stop Log Gate program should be cancelled. The thermal flux through the gap directly determined the value withdrawn downstream. The old dam at the front met the performance objectives for temperature increase.

The simulation results of this study can be used in combination with field-collected water quality and biological information to support decisions and alleviate the occurrence of adverse thermal conditions at the New Fengman Dam. This type of CFD model is applied to address a wide range of ecological problems caused by large hydropower dams.

ACKNOWLEDGEMENTS

This work was supported by the National Natural Science Foundation of China (Grant No. 51309256, 51269028, 51569029). Major State Basic Research Development Program of China (Grant No. 2016YFC0401708). The field data in the work are offered by Fengman Dam Reconstruction Engineering Construction Bureau.

REFERENCES

REFERENCES
Bormans
M.
Webster
I. T.
1998
Dynamics of temperature stratification in lowland rivers
.
Journal of Hydraulic Engineering
124
(
10
),
1059
1063
.
Caissie
D.
2006
The thermal regime of rivers: a review
.
Freshwater Biology
51
,
1389
1406
.
Çalışkan
A.
Elçi
S.
2009
Effects of selective withdrawal on hydrodynamics of a stratified reservoir
.
Water Resources Management
23
,
1257
1273
.
Casamitjana
X.
Serra
T.
Colomer
J.
Baserba
C.
Pérez-Losada
J.
2003
Effects of the water withdrawal in the stratification patterns of a reservoir
.
Hydrobiologia
504
,
21
28
.
Chen
H. Q.
Mao
S. M.
1990
Studies on the Evaporation and Heat Loss Coefficient from Water Surface
(Transactions of the Special Report)
.
Chen
X. H.
Liu
M. N.
Lin
Y. S.
1997
Study on two-dimensional water quality distribution in reservoirs
.
Journal of Hydraulic Engineering (Transactions of the JHE)
4
,
9
16
.
Chen
Y. C.
Zhang
B. X.
Li
Y. L.
1998
Study on model for vertical distribution of water temperature in Miyun Reservoir
.
Journal of Hydraulic Engineering (Transactions of the JHE)
9
,
14
20
.
China Institute of Water Resources and Hydropower Research (IWHR)
.
2010
DJ800 Data Collecting System Users Guide
.
Beijing
.
Dake
J. M. K.
Harleman
D. R. F.
1969
Thermal stratifications in lakes: analytical and laboratory studies
.
Water Resources Research
5
(
2
),
484
495
.
Deboltsky
V. K.
Neymark
R. V.
1994
Thermally non-uniform flow in water reservoir: numerical investigation
.
Journal of Hydraulic Research
32
(
1
),
25
40
.
Deng
Y.
2003
Study on the water temperature prediction model for the huge and deep reservoir
.
Sichuang university
,
Chengdu
(Transactions of the Doctoral Thesis)
.
Deng
Y.
Li
J.
Li
K. F.
Zhao
W. Q.
2003a
Simulation on thermal stratification of Zipingpu reservoir
.
Water Resources and Hydropower Engineering (Transactions of the WRHE)
34
(
9
),
50
52
.
Deng
Y.
Li
J.
Luo
L.
Zhao
W. Q.
2003b
Temperature prediction model for reservoirs
.
Journal of Hydraulic Engineering (Transactions of the JHE)
7
,
7
11
.
Deng
Y.
Li
J.
Luo
L.
2004
Simulation on thermal stratification of the river-like deep reservoir
.
Journal of Hydrodynamics (Transactions of the JH)
19
(
5
),
604
609
.
Deng
Y.
Li
J.
Li
R.
Li
K. F.
2006
Impact of reservoir operation on the water temperature downstream of Xiluodu power station
.
Journal of Sichuan University (Transactions of the JSU)
38
(
5
),
65
69
.
Ettema
R.
Muste
M.
Odgaard
A. J.
Lai
Y.
2004
Lake Almanor cold-water feasibility study: hydraulic model
.
IIHR Technical Report No. 438. IIHR – Hydroscience & Engineering
.
The University of Iowa
.
Farrel
G. J.
Stefan
H. G.
1988
Mathematical modeling of plunging reservoir flows
.
Journal of Hydraulic Research
26
(
5
),
525
537
.
FLUENT Inc
.
2006
Fluent 6.3 Users Guide
.
FLUENT Inc., Lebanon, New Hampshire
.
Gao
X. P.
Chen
H.
Wang
A. R.
Dong
S. R.
Zhang
Z. L.
Zhao
Y. N.
2010
Experimental study on water temperature released from multi-level intake structure of Nuozhadu hydropower Station
.
Journal of Hydroelectric Engineering (Transactions of the JHE)
29
(
3
),
126
131
.
Gao
X.
Zhang
S.
Zhang
C.
2012
3-D Numerical simulation of water temperature released from the multi-level intake of Nuozhadu hydropower station
.
Journal of Hydroelectric Engineering (Transactions of the JHE)
31
(
1
),
195
201
.
Imberger
J.
Panerson
J.
Hebbert
B.
1978
Dynamics of reservoir of medium size
.
Journal of the Hydraulics Division, ASCE
104
(
5
),
725
743
.
Jin
K.
Hamrick
J.
Tisdale
T.
2000
Application of three-dimensional hydrodynamic model for Lake Okeechobee
.
Journal of Hydraulic Engineering, ASCE
126
(
10
),
758
771
.
Johnson
B. H.
1980
A review of multidimensional reservoir hydrodynamic modeling
. In:
Proceedings of the symposium on surface water impoundments
. pp.
497
507
.
Li
H. E.
Shen
J.
1990
Research on 1D water temperature model of reservoir and temperature prediction for Hei River Reservoir
.
Journal of Shanxi Institute of Mechanical Engineering (Transactions of the JSIME)
6
(
4
),
236
243
.
Liu
H. T.
Sun
S.
Wang
X. S.
Xia
Q. F.
Jiang
H.
2012
Water temperature model test for a multi-level power intake of large deep reservoir
.
Journal of Hydroelectric Engineering (Transactions of the JHE)
31
(
1
),
129
134
.
Ma
S.
Kassinos
S. C.
Kassinos
D. F.
Akylas
E.
2008
Effects of selective water withdrawal schemes on thermal stratification in Kouris Dam in Cyprus
.
Lakes & Reservoirs: Research and Management
13
,
51
61
.
Orlob
G. T.
Selna
L. G.
1970
Temperature variation in deep reservoirs
.
Journal of the Hydraulics Division, ASCE
96
(
2
),
391
410
.
Politano
M.
Haque
M.
Constantinescu
G.
Weber
L.
2006
A Three-Dimensional Thermal Model for McNary Dam
.
World Environmental and Water Resources Congress. Omaha
,
Nebraska
,
USA
.
Politano
M.
Haque
M.
Weber
L.
2008
A numerical study of the temperature dynamics at McNary Dam
.
Ecological Modelling
212
,
408
421
.
Ren
H.
Chen
Y.
Liu
Z.
2007
Temperature simulation in large reservoir
.
Journal of Hydrodynamics (Transactions of the JH)
22
(
6
),
667
675
.
Rheinheimer
D. E.
Null
S. E.
Lund
J. R.
2015
Optimizing selective withdrawal from reservoirs to manage downstream temperatures with climate warming
.
Journal of Water Resources Planning and Management, ASCE
141
(
4
),
631
639
.
Rodi
W.
1984
Turbulence Models and their Application in Hydraulics – A State of the Art Review
.
University of Karlsruhe
,
Karlsruhe
.
Thorpe
J. E.
2004
Life history responses of fishes to culture
.
Journal of Fish Biology
65
(
s1
),
263
285
.
Van Der Kraak
G.
Pankhurst
N. W.
Wood
C. M.
McDonald
D. G.
1997
Temperature effects on the reproductive performance of fish
. In:
Global Warming: Implications for Freshwater and Marine Fish
(C. M. Wood and D. G. McDonald, eds)
.
Cambridge University Press, Cambridge, UK
.
Vermeyen
T. B.
1999
An Overview of the Design Concept and Hydraulic Modeling of the Glen Canyon Dam Multi-Level Intake Structure
. In:
ASCE's Waterpower ‘99 Conference Proceedings
.
Las Vegas, NV
.
Vermeyen
T. B.
2006
Hungry Horse selective withdrawal system evaluation 2000–2003
.
US Department of the Interior Bureau of Reclamation Technical service center Hydraulic Investigation and laboratory Group
,
Denver, Colorado
.
Yakhot
V.
Orzag
S. A.
1986
Renormalization group analysis of turbulence: basic theory
.
Journal of Scientific Computing
1
(
1
),
3
51
.