Journal of Hydrology 601 (2021) 126599

Contents lists available at ScienceDirect

Journal of Hydrology
journal homepage: www.elsevier.com/locate/jhydrol

Research papers

Multi-objective optimal operation of cascade hydropower plants

considering ecological flow under different ecological conditions
Lei Yu a, b, Xiufeng Wu a, *, Shiqiang Wu a, Benyou Jia a, *, Guoyi Han c, Peng Xu a, Jiangyu Dai a,
Yu Zhang a, Fangfang Wang a, Qianqian Yang a, Zehui Zhou d
a
State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering, Nanjing Hydraulic Research Institute, Nanjing 210029, China
b
College of Water Conservancy and Hydropower Engineering, Hohai University, Nanjing 210098, China
c
Stockholm Environment Institute, Stockholm 11523, Sweden
d
School of Earth Sciences and Engineering, Hohai University, Nanjing 211100, China

A R T I C L E I N F O A B S T R A C T

This manuscript was handled by Huaming Guo, Maintaining ecological flow and mitigating impacts on river health are among the core requirements for sus­
Editor-in-Chief, with the assistance of Xiaohui tainable hydropower development. Designing specific operational modes to fit specific ecological conditions of
Lei, Associate Editor rivers remains a key practical challenge for hydropower operation, especially in terms of balancing the various
tradeoffs among multiple objectives in the context of coordinated operation of cascade hydropower plants.
Keywords:
Taking the Yalong River, China as an example, this study develops a decision-making support model that can
Ecological operation
quantitatively analyse the tradeoffs between the hydro-energy generation and river ecosystem projection under
Multi-objective operation
Cascade hydropower plants different hydrological conditions. The model contains three main components – calculation of ecological flow,
Ecological flow multi-objective optimal operation, and determination of optimal schemes. For the ecological flow determination,
Tradeoff relationship we propose a revised variable monthly flow method (RVMF), which incorporates variations of hydrological
Yalong River conditions, also combines the classification concept of the Tennant method. For these objectives, we consider
cascade power generation, the assurance rate of cascade power generation, and the assurance rate of ecological
flow. The preliminary application in the Yalong River case has shown that (1) with the increase in environmental
flow requirements of the river, the cascade power generation has dropped significantly, and the most affected is
the Jinping-II hydropower plant; (2) there are clear tradeoffs between the cascade power generation and the
assurance rate of cascade power generation, and the cascade power generation and the assurance rate of cascade
power generation under different riparian ecological conditions; and (3) being able to quantify those tradeoffs
provides crucial scientific evidence for stakeholder dialogue and informed decision making for hydropower
operation modes that can achieve the best desirable balances between power generation and river health
conservation.

1. Introduction hydropower projects but also analyse the benefit tradeoff between
different operation targets. Besides, the operation can also be optimised
Concepts such as river health and ecological flow are current aca­ by considering the environmental flow requirements (EFRs) of the river
demic research hotspots (Arthington et al., 2006; Huang et al., 2014; reach, which provides vital technical support for decision-making
Poff and Matthews, 2013; Rees et al., 2020; Tan et al., 2018) and are also among operators of cascade hydropower plants (Jiang et al., 2019;
the focus of attention for the comprehensive benefits of water conser­ Ding et al., 2020; Chen et al., 2014).
vancy and hydropower projects (Ren et al., 2019; Renofalt et al., 2010; Currently, most hydropower plants have multiple functions such as
Parish et al., 2019). The multi-objective optimal operation of cascade power generation, flood control, water supply, and pursue the high-
hydropower plants considering ecological flow can not only explore the efficiency utilisation of water resources and maximising economic
possible comprehensive benefits of ecology, power generation, flood benefits (Zhang et al., 2019), however, the modes of operation of these
control, water supply, shipping, and tourism for water conservancy and hydropower plants do not adequately consider the EFRs of the

* Corresponding authors.
E-mail addresses: [email protected] (X. Wu), [email protected] (B. Jia).

https://doi.org/10.1016/j.jhydrol.2021.126599

Available online 26 June 2021

0022-1694/© 2021 Elsevier B.V. All rights reserved.
L. Yu et al. Journal of Hydrology 601 (2021) 126599

dewatered river reach downstream of the dam, causing varying degrees existing studies have assessed the tradeoff between human needs and the
of damage to the river ecosystem (Zhou and Guo, 2013; Magilligan & riverine ecosystem with optimizing the operation strategies. Xu (2020)
Nislow, 2005; Grill et al., 2019), for instance, dam construction blocks taken a nondominated sorting genetic algorithm (NSGA-II) to solve the
river flow, cause changes in hydrological conditions and the biological multi-objective optimal operation to tradeoff between EFRs and agri­
habitat, affect the spawning and reproduction of migratory fish, leading cultural irrigation. Xu et al. (2020a) developed a multi-objective opti­
to changes in the population and structural characteristics of the river mization model to quantitatively analyse the tradeoffs between
biome (Gibeau et al., 2017). Therefore, researchers take ecological hydropower generation and biodiversity conservation. The research
factors into account during hydropower plant operation (Xu, 2017). results showed that the optimal operation of reservoirs is effective for
Since Schluter (1971) first proposed the theory that the operation of maintaining ecological flow in the downstream river reach of hydro­
reservoirs should maintain the diversity of the riverine ecosystem, Petts power plants.
and Geoffrey (1996) proposed an ecologically acceptable ecological flow The main aim of this study is to quantify the tradeoffs between the
method and shown that 60% of the available resources are needed to hydro-energy generation and river ecosystem projection under various
maintain the river as a trout stream. Hughes and Ziervogel (1998) first riparian ecological conditions in different hydrological conditions
established an ecological operation model of the reservoir (i.e., the following a decision-making support model. In this study, a revised
DAMIFR model), which can simulate the effects of a variety of different ecological flow method is firstly proposed to determine the EFRs under
release priorities. Recently, ecological and environmental protection has different ecological conditions in downstream river reach of hydro­
received more attention and attention, and related research results have power plants. Then, a decision-making support model is developed,
become more abundant. Steinschneider et al. (2014) developed a large- which contains three main components – calculation of ecological flow,
scale optimal operation model to explore the contribution of optimal multi-objective optimal operation, and determination of optimal
operation management of large-scale hydropower plants to ecological schemes. Finally, the cascade hydropower plants of the lower reaches of
benefit. Tsai et al. (2015) employed a hybrid artificial neural network to the Yalong River Basin in China are taken a research case-study, and the
quantify the relationship between flow regimes and fish communities multi-objective optimal operation model is solved by NSGA-II algorithm,
and guided the ecological operation of the Shihmen Reservoir in and the tradeoffs between three optimal objectives are evaluated. The
northern Taiwan. Huang et al. (2019) quantitatively analysed the present research results can provide scientific evidence for the ecolog­
tradeoffs among social, economic, and ecological benefits during the ical restoration and protection of the lower reaches of the Yalong River,
flood season for the Three Gorges Project, China, and the results indi­ and at the same time, provide a reference for the ecological operation of
cated that a smaller flood peak leads to less sediment and P transports to cascade hydropower plants under various ecological conditions.
the downstream. Similarly, Zhang et al. (2019) established ecological
operation models of cascade hydropower plants and quantified the 2. Methodology
interaction between power generation and degree of ecological flow
satisfaction under different modes of operation. Recently, Feng et al. An ecological flow method (revised variable monthly flow, RVMF)
(2020) developed a novel multi-strategy gravitational search algorithm was proposed for a variety of ecological conditions based on the variable
and applied it to the ecological operation problem of the Wu hydro­ monthly flow (VMF) method and was combined with the classification
power system, which can make obvious reductions in the inappropriate concept in the Tennant method. This paper established six ecological
ecological water volume. scenarios based on the ecological flow calculation results of the RVMF
The determination of ecological flow is the prerequisite and basis for method. A multi-objective optimal operation model was constructed
the study of the ecological operation of hydropower plants (Zhang et al., under three objectives (i.e., cascade power generation, assurance rate of
2019). The ecological flow of the river is defined as the proper flow cascade power generation, and assurance rate of cascade power gener­
needed for rivers to maintain the health of the river ecosystem and ation), which was solved by the NSGA-II algorithm to obtain the Pareto
ensure the survival and development of human beings (Poff and Mat­ optimal sets of solutions. The optimal or recommended schemes were
thews, 2013). The main ecological flow methods are hydrological determined by a multi-objective decision-making method (the entropy-
methods, hydraulic methods, habitat simulation methods, and holistic weighted TOPSIS method). These methods or models were established
methods (Pastor et al., 2014), respectively. Among these methods, the to analyse the relationship between power generation and ecosystem
hydrology method is the earliest method to calculate ecological flow, protection quantitatively. The flow chart through the methods used in
and it is also the most widely used method because it has the advantages this study is shown in Fig. 1.
of a lower data demand, simple operation, and lower cost (Pastor et al.,
2014). Zhang et al. (2019) comprehensively considered the 7Q10 2.1. Calculation of ecological flow
method (hydrological method), RVA (hydrological method), and Riv­
er2D model (habitat simulation method) to compute the comprehensive At present, there are more than 200 ecological flow methods
ecological water demand of the study area, and established three (Tharme, 2003), which are mainly classified into four types: hydrolog­
ecological operation models to quantify the relationship between power ical methods (Tennant method (Tennant, 1976), IHA, and RVA); hy­
generation and ecological system protection. The research results draulic rating methods (R2CROSS); habitat simulation methods (WUA
showed that power generation and ecological demand are mutually and River2D); and holistic methods (ELOHA) (Pastor et al., 2014). These
restrictive and conflicting. Shadkam et al. (2016) calculated the envi­ methods have their advantages and disadvantages, and the advantage of
ronmental flow requirements of high-salt lakes based on the variable hydrological methods is that they are fast, straightforward ecological
monthly flow (VMF) method (hydrological method) and estimated that flow methods for use in preliminary assessments or when ecological data
3.7 × 109 m3 a-1 water was needed to preserve Urmia Lake. sets are unavailable.
Until recently, most research on the ecological operation of hydro­ The Tennant method (Tennant, 1976) is an empirical relationship
power plants employs a single target ecological flow to characterise the between flow and habitat quality established based on observational
EFRs of the dewatered river reach downstream of the dam, such as data and is the most widely used hydrological method in the world. The
ecological base flow, which is used to maintain the vital ecosystem of the method is simple and convenient to use because it only needs historical
river, ignoring other ecological conditions (e.g., optimum, excellent, and flow data to determine the environmental flow requirements (EFRs) of
good ecological conditions) (Wu et al., 2011). Also, relatively few the river, which have macroscopic guiding significance. The Tennant
studies have focused on the tradeoffs between hydropower production method qualitatively describes the habitat as eight ecological conditions
and river ecosystem projection considering different EFRs of the based on a specific percentage of the annual mean flow (MAF), which
downstream river reach for the cascade hydropower plants. Some are severe degradation, poor or minimum, fair or degrading, good,

2
L. Yu et al. Journal of Hydrology 601 (2021) 126599

Fig. 1. The flow chart through the methods used in this study.

excellent, outstanding, optimum range, and flushing or maximum, low-flow months (Pastor et al., 2014). This paper referred to the
respectively. However, the method does not reflect the high-flow, in­ ecological flow classification standard of the Tennant method and used
termediate-flow, and low-flow months of the ecological flow during the 10% of MMF as the minimum ecological flow, which was a poor
year. Thus, the Tennant method is suitable for large and perennial rivers, ecological condition. At the same time, five levels were set from the
but not for seasonal rivers. ecological condition “fair” to the ecological condition “optimum”, by
The variable monthly flow (VMF) method is an ecological flow gradually increasing the MMF by 10%. When the ecological flow is
method developed by Pastor et al. (2014), which was a hydrological 100% MMF, the optimum ecological condition is reached (Table 1).
method. This method uses an algorithm to classify the flow regime into The specific calculation formula of ecological flow is as follows:
high, intermediate, and low-flow months and takes intra-annual vari­
EF = αMMF (1)
ability into account by allocating EFRs with a percentage of mean
monthly flow (MMF), which increases the protection of freshwater where EF is the ecological flow of the river (m3/s); α is the flow coeffi­
ecosystems during the low-flow season. The VMF method offers better cient (%), see Table 1 for specific values. For example, when the hy­
performance compared with the other three widely used hydrological drological season is high-flow months and the ecological condition of
methods (Tennant method, Smakhtin method (Smakhtin et al., 2004), the river is poor, α is equal to 10; MMF is the mean monthly flow (m3/s).
and Tessmann method), however, it is designed to achieve a “fair”
ecological condition for river flow, which does not take into account
other ecological conditions, such as poor, good, and excellent 2.2. Multi-objective optimal operation model for cascade hydropower
conditions. plants
Although the grading standard of the Tennant method is not uni­
versal, the grading idea of an arithmetic progression is based on many 2.2.1. Objective functions
field observations, hydraulic analysis, and statistical analysis, which The main objective of cascade hydropower plants operation is to
have been universally recognised (Tharme, 2003). In the present work, maximize the power generation benefits, so this study considers cascade
we revised the VMF method based on the ecological flow standard of the power generation as an objective in multi-objective optimal operation.
Tennant method and proposed an ecological flow method (Revised VMF, The main goal of the ecological operation of hydropower plants is to
RVMF) for multiple ecological conditions, that is, based on a specific ensure that the discharge flow meets the water demands of the down­
percentage (10%) of the MMF, the ecological conditions of the river stream river ecosystems. Thus, the ecological benefits of the operation of
were qualitatively described by six levels: poor, fair, good, excellent, cascade hydropower plants are expressed by the assurance rate of
outstanding, and optimum, respectively. The VMF method allocated ecological flow of the downstream river in this study. Also, this study
water for freshwater ecosystems in fair ecological conditions, and the considers various ecological conditions in the downstream river.
corresponding ecological flow in different months is 30% MMF in high- Moreover, In addition, for in-conduit hydropower plants, assurance
flow months, 45% MMF in intermediate-flow months, and 60% MMF in output may not be met when large amounts of water are used to meet the
water needs of downstream river ecosystems rather than for power

3
L. Yu et al. Journal of Hydrology 601 (2021) 126599

Table 1
Ecological conditions and corresponding flow of RVMF method with MAF (the mean annual flow) and MMF (the mean monthly flow) (% MMF).
Hydrological season Calculation algorithm Ecological conditions in the river

Poor Fair Good Excellent Outstanding Optimum

High-flow months MMF greater than 0.8MAF 10 30 40 50 60 70

Intermediate-flow months 0.4MAF < MMF ≤ 0.8MAF 15 45 55 65 75 85
Low-flow months MMF ≤ 0.4MAF 20 60 70 80 90 100

generation. Therefore, we took the assurance rate of cascade power time t, respectively (m).
generation as one operation objective. The three objectives are in a (3) Discharge flow constraints
mutually competitive relationship, and the corresponding functions are
0 ≤ Qi,t ≤ Qmax (8)
described as follows: i,t

(1) Maximum cascade power generation (CPG) 3

where Qmax
i,t is the maximum discharge from reservoir i at time t (m /s).
n ∑
∑ T
(4) Output constraints
E1 = max ηi Hi,t Qi,t Δt (2)
i=1 t=1 0 ≤ Ni,t ≤ Ni,max (9)

where E1 is the maximum annual power generation of cascade hydro­ where Ni,max is the maximum power output limit of hydropower plant i
power plants (108 kWh); n is the number of reservoirs; T is the numbers at time t (MW).
of months in a year (T = 12); ηi is comprehensive power coefficient of the
ith hydropower plant; Hi,t is the average head at time t (m); Qi,t is the 2.2.3. Constraint handling strategy
turbine release while producing power (m3/s); Δt is the time-step (s). Here, we developed a constraint handling method through constraint
(2) The maximum assurance rate of cascade power generation transformation based on the characteristics of the actual operation of
(ARCPG) cascade hydropower plants. The optimal operation of cascade power
plants mainly includes three constraints of water level, discharge flow,
1 ∑ n ∑ T
E2 = max pi,t × 100% (3) and output. For the constraints of water level, discharge flow: firstly, we
n × T i=1 t=1
employed the water level as a decision variable; then, we converted the
{ maximum flow constraint into a water level interval through the water
0, Ni,t < Ni,min balance equation; finally, the intersection of the water level constraint
pi,t = (4)
1, Ni,t ≥ Ni,min
interval and the water level interval obtained by the discharge flow
constraint conversion is taken to obtain the feasible region of the deci­
where E2 is the assurance rate of cascade power generation to the total
sion variable. For the output constraint, the upper limit is taken when
number of periods; Ni,min is guaranteed output of the ith hydropower
the output at time i is greater than the upper limit of the output of the
plant (MW)
hydropower plant. In this way, the numerous constraints of the original
(3) The maximum assurance rate of ecological flow (AREF)
problem are converted into the water level restrictions of each reservoir
1 ∑ n ∑ T in each operation period, which is relatively simple to deal with.
E3 = max mi,t × 100% (5)
n × T i=1 t=1
2.2.4. Solution method
{
0, EF i,t < EFT i,t At present, the research into the algorithm for solving the optimal
mi,t = (6) operation model of hydropower plants has achieved fruitful results.
1, EF i,t ≥ EFT i,T
According to the number of objectives, it is mainly divided into single-
where E3 is the assurance rate of ecological flow in the river section to objective optimisation algorithms (linear programming, dynamic pro­
the total number of periods; EFi,t and EFTi,t are the ecological flow and gramming, non-linear programming, optimal recursive algorithm, etc.)
target ecological flow of the downstream river of plant i at time t, and multi-objective optimisation algorithms (multiple target genetic
respectively (m3/s), and the EFTi,t is determined by the RVMF method algorithms, multi-objective particle swarm optimisation, multi-
proposed in this study. objective differential evolution algorithms, etc.). The above optimal
operation model belongs to a multi-objective, multi-constrained, multi-
2.2.2. Constraint conditions dimensional, and non-linear complex optimisation problem. Thus, we
The multi-objective optimal operation model is subject to the selected the fast elitist multi-objective genetic algorithm (NSGA-II) (Deb
following constraints: et al., 2002), which has the advantages of rapidity and good conver­
(1) Water balance constraint gence. Also, the algorithm is widely used in solving the multi-objective
( ) optimal operation of hydropower plants (Boyaghchi and Sohbatloo,
Vi,t+1 = Vi,t + Ii,t − Qi,t Δt − Ei,t − Li,t (6) 2018; Huang et al., 2019). The general procedure of NSGA-II is shown in
Fig. S1 (modified from Li et al. (2018)).
where Vi,t+1 and Vi,t refer to the average storage of reservoir i in the
(t + 1)th and tth time steps, respectively (m3); Ii,t and Qi,t are the average 2.3. Multi-objective decision-making model
inflow and outflow of reservoir i at time t, respectively (m3/s); Ei,t is the
evaporation of reservoir i at time t, (m3); Li,t is the leakage of water from Pareto optimal solutions could be obtained after solving the multi-
reservoir i (m3). objective optimal operation problem of cascade hydropower plants.
(2) Water level constraints Decision-makers need to finally obtain the optimal or recommended
Zi,tmin ≤ Zi,t ≤ Zi,tmax (7) schemes through the multi-objective decision-making model. The tech­
nique for order performance by similarity to ideal solution (TOPSIS)
where Zi,t is the average water level of the reservoir i at time t; Zmin method had no strict restrictions and requirements in terms of the
i,t and
number of indicators, also referred to as criteria, samples, and data. The
Zmax
i,t are the minimum and maximum water level limits of reservoir i at
method has been widely used in multi-index and multi-scheme decision

4
L. Yu et al. Journal of Hydrology 601 (2021) 126599

analysis problems and is also an effective and standard decision-making where diag(W) is the diagonal matrix corresponding to the indicator
optimisation method (Liu and Yin, 2019; Xu et al., 2020b). The present weight (W).
research employed the entropy-weighted TOPSIS method, which could Step 3. Calculate the ideal and anti-ideal solutions.
avoid the irrationality of the subjective weighting of the traditional Calculate the ideal and anti-ideal solutions (S+ and S− ) based on the
TOPSIS method (Hussain et al., 2020). weighting matrix, S+ = [s+ 1 , s2 , ⋯, sn ], S = [s1 , s2 , ⋯, sn ]. The ideal
+ + − − − −

values (s+
j and s −
j ) are estimated by using Eq. (15).
2.3.1. Entropy weight method ⎧ +
The entropy method is an objective weighting method, which only ⎨ sj = max (rij )
(17)
1≤i≤m
relies on the discreteness of the data itself, and has the characteristics of ⎩ s−j = min (rij )
strong operability and objectivity. In the evaluation process, the higher 1≤i≤m

the degree of dispersion of an indicator is, the greater the weight of the Step 4. Compute the Euclidean distance. The separation of each
indicator (Hussain et al., 2020). The specific calculation steps are as alternative from the positive and negative ideal solution is given below,
follows: respectively.
Step 1. Indicator normalisation. If the aij is the benefit indicator, the √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
indicator normalisation is followed as Eq. (10). If the aij is the cost in­ √∑
√ n
(18)
√ 2
dicator, the indicator normalisation is followed as Eq. (11). D+i = (S+ j − rij ) , i = 1, 2, ⋯m
j=1
aij − min(aij )
aij = ( ) (10) √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
max aij − min(aij ) √∑
√ n
D−i =√ (S−j − rij )2 , i = 1, 2, ⋯m (19)
( )
max aij − aij j=1
aij = ( ) (11)
max aij − min(aij ) Step 5. Calculate the performance score value.
Step 2. Calculate the proportion of the ith sample value with nor­ D−i
malisation under the jth indicator (pij ). Ci = , i = 1, 2, ⋯m (20)
D+
i + D−i
aij
pij = ∑n (12) where Ci is the performance score value, Ci ∈ [0,1].
i=1 aij
Step 6. Finally, the ranking of alternatives is set based on the per­
where aij is the ith sample value with normalisation under the jth formance score value.
indicator.
Step 3. Calculate the entropy value of the jth indicator. 3. Case study

1 ∑
n
ej = − pij lnpij (i, j = 1, 2, ⋯n) (13) 3.1. Study area
lnn i=1

We selected the cascade hydropower plants in the lower reach of the

where ej is the entropy value of the jth indicator; n represents many Yalong River as a case study. The Yalong River is the largest tributary of
alternatives. the Jinsha River in China, with a total length of 1571 km, a natural drop
Step 4. Calculate the weight of the jth indicator (uj ) of 3830 m, an annual runoff of 60.9 billion m3, a drainage basin area of
1 − ej approximately 136,000 km2, and average annual precipitation of 500 to
uj = ∑n (14) 2470 mm. The basin has abundant water resources and is one of the most
n− j=1 ej
critical water resources for strategic reserve areas in China. Five cascade
hydropower plants have been built in the lower reach of the Yalong
2.3.2. TOPSIS method
River, namely Jinping-I, Jinping-II, Guandi, Ertan, and Tongzilin,
Hwang and Yoon (Hwang and Yoon, 1981) proposed the TOPSIS
respectively. Jinping-I is a yearly regulating reservoir, Ertan is a sea­
method, which is a powerful and commonly used analysis method for
sonal regulating reservoir, Guandi and Tongzilin have daily regulating
multi-objective decision-making problems based on the idea that the
capacity, and Jinping-II is an in-conduit hydropower plant with a
best alternative should have the smallest distance (i.e., Euclidean dis­
dewatered river reach of 119 km between the water intake and the plant
tance) from the ideal solution. The TOPSIS method is used to estimate
(Yu et al., 2019).
distances to both ideal and anti-ideal solutions and is a useful and
The case study is generalised into the form of “two reservoirs and five
straightforward method for ranking many desirable alternatives to
levels”, that is, the reservoir capacities of Jinping-II, Guandi, and
identify the closest-to-ideal solutions (Hussain et al., 2020). The main
Tongzilin reservoirs are generalised to equivalent storage capacity, and
steps of the TOPSIS method could be elaborated as follows:
the corresponding reservoir water level is a fixed value (taking the
Step 1. Construct a standardised initial matrix (Z)
average of the normal water level and the dead water level) throughout
⎡ ⎤
z11 z12 ⋯ z1n the entire period. The distribution of hydropower plants on the Yalong
⎢ z z ⋯ z2n ⎥
Z = ⎣ 21 22 (15) River and their characteristic parameters are shown in Fig. 2 and
⋮ ⋮ ⋱ ⋮ ⎦ Table 2.
zm1 zm2 ⋯ zmn
Step 2. Construct a normalised weight matrix 3.2. Basic data
Here, the entropy method was used to calculate the indicator weight
(W = [w1 ,w2 ,⋯wn ]), then the weighted normalised decision matrix was The flood season in the Yalong River Basin lasts from June to
estimated by using Eq. (14) with the performance value of the normal­ October, and the reservoir reaches its normal water level at the end of
ised matrix. October. Therefore, the hydrological year and the scheduling period of
{
R = Z × diag(W) the basin were set to run from November to October. Five typical years
i = 1, 2, ⋯m; j = 1, 2, ⋯n (16) were selected based on the streamflow data from the cascade hydro­
rij = wj × zij
power plants from 1958 to 2018, i.e., wet year (2012, P = 10%), rela­
tively wet year (2008, P = 30%), normal year (2015, P = 50%),

5
L. Yu et al. Journal of Hydrology 601 (2021) 126599

Fig. 2. Study area.

relatively dry year (2002, P = 70%), and dry year (2006, P = 90%), used
Table 2
for input of the multi-objective optimal operation model. The typical
Characteristics of hydropower plants on the lower reaches of the Yalong River.
yearly inflow to the Jinping-I hydropower plant is shown in Fig. S2.
Index Jinping- Jinping- Guandi Ertan Tongzilin
I II
3.3. Scenarios
Normal water level 1880 1646 1330 1200 1015
(m)
Flood limited water 1859 – – 1190 –
Three periods were classified based on the RVMF method in the
level (m) Yalong River Basin, i.e., high-flow months (June to October),
Dead water level (m) 1800 1640 1328 1155 1012 intermediate-flow months (May, November, and December), and low-
Total storage (108 77.60 0.14 7.29 57.90 0.72 flow months (January to April). The EFRs of the Jinping-I are listed in
m3)
Table 3. When the ecological condition of the river is poor, the average
Regulation ability Yearly Daily Daily Seasonally Daily
Installed capacity 3600 4800 2400 3300 600 ecological flow is 14.58% MMF; when the ecological conditions are
(MW) optimum, the average ecological flow is 83.75% MMF. The Jinping-II is
Annual power 166.20 237.60 110.16 168.84 29.75 an in-conduit hydropower plant, and the diversion gate is located 7.5 km
generation (108 downstream of the dam site at Jinping-I. When the Jinping-II Power
kWh)
Comprehensive 8.50 8.65 8.50 8.60 8.50
plant diverts the water to generate electricity, it will form a dewatered
power coefficient river reach of 119 km (Dahewan Reach) between the water intake and
the plant. The relationship between the ecological flow of the dewatered
river reach and Jinping-I and II power plants can be expressed as:

Table 3
Ecological flow at the downstream reach of Jinping-I (m3/s).
Ecological Scenario Average ecological Month
conditions in the flow (%MMF)
Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct
river

Poor S1 14.58 125.7 77.6 75.2 67.4 68.8 90.6 110.9 177.0 285.0 268.0 268.0 165.0
Fair S2 43.75 377.1 232.7 225.6 202.2 206.4 216.0 332.6 531.0 855.0 804.0 804.0 495.0
Good S3 53.75 460.9 284.4 263.2 235.9 240.8 317.1 406.5 708.0 1140.0 1072.0 1072.0 660.0
Excellent S4 63.75 544.7 336.4 300.8 269.6 275.2 362.4 480.4 885.0 1425.0 1340.0 1340.0 825.0
Outstanding S5 73.75 628.5 387.8 338.4 303.3 309.6 407.7 554.3 1062.0 1710.0 1608.0 1608.0 990.0
Optimum S6 83.75 712.3 439.5 376.0 337.0 344.0 453.0 628.0 1239.0 1995.0 1876.0 1876.0 1155.0

6
L. Yu et al. Journal of Hydrology 601 (2021) 126599

Table 4 paper, the RVMF method and the Tennant method were used to calcu­
The rate of power generation water consumption of cascade hydropower plants late the ecological flow under the six scenarios of the Dahewan Reach,
under scenarios S1-S6 in a normal year (m3/kWh). and the calculation results were compared and analysed. The calculation
Efficiency index Scenario results of the RVMF method were consistent with the changing trend of
S1 S2 S3 S4 S5 S6
natural flow in Dahewan Reach, which could reflect the seasonal change
of ecological flow to a certain extent. In contrast, the change of the
Rate of water consumption 2.78 2.92 2.99 3.07 3.15 3.25
Tennant method is singular during a given year, only reflecting the
difference between flood season and dry season (Fig. 3).
Table 4. The natural flow characteristics of rivers constitute the integrity of
the ecosystem. Hydrological rhythm refers to the periodic and rhythmic
QJP1,t ≥ θ × Qe,t , t = 1, 2, ⋯T (21) changes in the hydrologic regime of rivers and lakes and is closely
connected with the riverine ecosystem. When the hydrological rhythm
QJP1,t − QJP2,t = θ × Qe,t , t = 1, 2, ⋯T (22) changes, it will affect the reproduction, growth, and development of
aquatic organisms to a certain extent (Wheeler et al., 2017). The dis­
where QJP1,t is the outflow of Jinping-I at time t (m3/s); Qe,t is the
tribution of hotspots on monthly scales of standardised natural flow and
ecological flow of Dahewan Reach at time t (m3/s); QJP2,t is diversion
ecological flow (divide by the maximum value of the corresponding
flow of Jinping-II at time t (m3/s); θ is the slack variable, θ ∈ [0,1], and
flow) is shown in Fig. 4. The comparative analysis shows that the hy­
its value is equal to the ecological flow satisfaction of the cascade hy­
drological rhythm of natural MMF increases month-by-month from
dropower plants.
March to July, and decreases month-by-month from August to February.
According to the ecological conditions in the river, six scenarios of
The RVMF method simulates the hydrological rhythm of ecological flow
ecological flow were established, namely S1-S6, as listed in Table 3.
in a manner consistent with the natural flow. In contrast, the calculation
Scenario S1 means that only the minimum ecological flow is considered
results of the Tennant method do not change significantly during the
for the operation of hydropower plants (reference scenario), which is
year. This means that the Tennant method cannot reflect the natural
also the current operating mode of most hydropower plants considering
change in hydrological rhythm, which may cause the abundance of some
ecological flow. At the same time, S6 is the scenario where the ecological
aquatic organisms that are highly dependent on hydrological rhythms to
flow reaches the level of the optimum ecological conditions in the river.
decrease or disappear. Together, the RVMF method is deemed to have
been more reasonable than the Tennant method.
4. Results At present, there have been some relevant research results on
ecological base flow in the Dahewan Reach of the Yalong River. Wang
4.1. Rationality of ecological flow method et al. (2007) calculated the minimum ecological flow in Dahewan Reach
using hydraulic methods, and their results showed that 45 m3/s of
The determination of the ecological flow is the prerequisite and basis discharge from the sluice gate in the dry season, can meet the living
for the study of the ecological operation of hydropower plants. An conditions of fish of about 95% of the river section under hydraulic
ecological flow method (RVMF) was proposed under a variety of factors prevailing in the dewatered river reach of Jinping-II hydropower
ecological conditions, and its rationality was the critical point of the plant. Chen et al. (2014) combined the low power generation loss and
paper. Also, Jinping-II is an in-conduit power plant, which has formed a optimal ecological flow plan and recommended a discharge flow of
119 km dewatered river reach (Dahewan Reach) between the water 36–50 m3/s in the Dahewan Reach in the dry season, which can ensure
intake and the power plant. The mode of operation of the power plant that at least 50% of the target fish habitat in the river is protected. Be­
had the most significant influence on the ecology of the Dahewan Reach. sides, the actual operation of the ecological sluice gate in the Jinping-II
Therefore, Dahewan Reach was taken as an example to discuss the ra­ hydropower plant must pass 45–74 m3/s to ensure the minimum
tionality of the RVMF method. ecological flow in the Dahewan Reach. Here, we employed the RVMF
The Tennant method is an empirical relationship between ecological method to calculate the minimum ecological flow (the ecological con­
flow and habitat quality based on observational data and is also the most dition is poor); the dry season flow is 67.4 m3/s, which agrees with
widely used hydrological method in the world (Tennant, 1976). In this

Fig. 3. The monthly ecological flow of Dahewan Reach among various scenarios.

7
L. Yu et al. Journal of Hydrology 601 (2021) 126599

Fig. 4. Hydrological rhythm simulation effect based on the (a) RVMF and (b) Tennant methods.

previous studies (Chen et al., 2014; Wang et al, 2007) and is also closer
to the actual operating results.
To further illustrate the rationality of the RVMF method, we selected
a case study to compare and analyse the calculation results of ecological
flow. Wan and Xia (2007) calculated the minimum ecological flow at the
Reba dam site in the upper reaches of the Yalong River through five
ecological flow methods, which are Tennant method, Texas method
minimum flow method, hydraulic radius method, and wetted perimeter
method, respectively. Fig. 5 shows the comparison of the annual average
minimum ecological flow at the Reba dam site calculated by different
methods. It can be seen that the minimum ecological flow at the Reba
dam site is between 36.3–46.8 m3/s, with an average value of 41.2 m3/s,
which is consistent with the results calculated by the RVMF method in
this study (43.7 m3/s). In summary, the RVMF method proposed is
reasonable and reliable to a certain extent.

4.2. Cascade hydropower plants operation results analysis

In a normal year, the Pareto non-inferior set of solutions and the best
compromise solutions of the six ecological scenarios (S1-S6) are shown
in Fig. 6(a)-(f): the CPG and the ARCPG restrict and conflict with each
other, showing a significant negative correlation, that is, as the CPG
increases, the ARCPG decreases. In contrast, the correlation between the
Fig. 5. Comparison of annual average minimum ecological flow by six methods
CPG and the AREF is insignificant, and the correlation between the in the Reba dam site of Yalong River.
ARCPG and the AREF is also insignificant. Besides, this study constructs
the three objectives of CPG, ARCPG, and AREF as the evaluation in­
kWh, 93.33% and 100%; 584.17 × 108 kWh, 78.33%, and 100%,
dicators system of operation schemes, and uses the Pareto non-inferior
respectively. The AREFs for scenarios S1-S6 are all 100%, which shows
set of solutions as the input of the multi-objective decision-making
that, while the cascade hydropower plants in the Yalong River are
model. The optimal schemes and their values of scenarios S1-S6 are
pursuing power generation benefits, they can meet the optimum
determined by the entropy-based TOPSIS model, as shown in Table S1.
ecological condition of the river; however, the power loss of the cascade
And their weights as shown in Table S2. Among them, the CPG, the
hydropower plants is relatively significant compared with S1 at this time
ARCPG, and the AREF for scenario S1 and scenario S6 are 777.25 × 108

8
L. Yu et al. Journal of Hydrology 601 (2021) 126599

Fig. 6. Pareto front of the NSGA-II and best compromise solutions with the entropy-weighted TOPSIS method concerning the cascade power generation (CPG),
assurance rate of cascade power generation (ARCPG), and assurance rate of ecological flow (AREF) under scenarios S1-S6 in a normal year.

(193.08 × 108 kWh). Also, as the demand for the ecological flow of electricity generation) (Chang et al., 2017), as summarised in Table 4. It
rivers increases, the power generation of the Yalong River cascade hy­ can be seen from the table that the ecological flow demand of rivers
dropower plants has dropped significantly. increases and the power generation efficiency of cascade hydropower
Besides, we analysed the changes of power generation efficiency plants decreases. Among them, scenario S1 has the highest power gen­
indicators of cascade hydropower plants for optimal schemes under six eration efficiency, and its rate of water consumption is 2.78 m3/kWh;
ecological scenarios, namely the rate of power generation water con­ and scenario S6 is the lowest, with a rate of water consumption of
sumption (the water through the turbine per unit of hydropower 3.25 m3/kWh.

9
L. Yu et al. Journal of Hydrology 601 (2021) 126599

Furthermore, Fig. 7 shows the power generation of five power plants in the wet year has the least influence on cascade power generation.
under scenarios S1-S6 in a normal year. Only the power generation of
Jinping-II hydropower plant has changed significantly, from 253.41 (S1) 4.3. Relationship between abandoned water and ecological flow
to 63.92 × 108 kWh (S2), and the other four plants have little change.
This shows that different EFRs in the downstream river reach mainly The issue of abandoned water is unavoidable in the sustainable
affect the power generation of Jinping-II hydropower plant, and then development of hydropower and the operation and management of
affect the CPG. The main reason is that the Jinping-II hydropower plant power plants (Ye et al., 2018; Jiang et al., 2018). For example, the total
is an in-conduit power plant. When the optimal ecological condition (S6) energy loss of abandoned water in Sichuan Province (China) in 2016 was
of the river is reached, most of the outflow from Jinping-I hydropower 287.3 × 108 kWh (Ye et al., 2018), which is close to 0.4 times the
plant is used for the ecological needs of the Dahewan Reach, rather than average power generation of the lower cascade hydropower plants in the
the power generation of the Jinping-II hydropower plant. This means Yalong River, resulting in much waste of clean energy, which is not
that for the dammed hydropower plant (e.g., Jinping-I), the tradeoff conducive to the sustainability of hydropower development. Here, the
relationship between ecological flow and power generation is weak status of abandoned water from cascade hydropower plants in the lower
because the power flow can meet the EFRs of downstream rivers. reach of the Yalong River under five hydrological years and six
However, for the in-conduit power plant (Jinping-II), there is a signifi­ ecological scenarios is analysed, and the relationship between aban­
cant tradeoff relationship between ecological flow and power genera­ doned water and ecological flow is explored.
tion, so the ecological problems of its dewatered river reach need to be Table 6 lists the abandoned water from the cascade hydropower
paid attention to. plants in the lower reach of the Yalong River during five hydrological
Jinping-I and Ertan hydropower plants are controlled reservoirs with years and under six ecological scenarios. The issue of abandoned water
strong regulation ability in the lower reach of the Yalong River Basin: in the lower reach of the Yalong River mainly occurs in wet and rela­
their operating water level and outflow are closely related to the power tively wet years, while less water is discarded in normal, relatively dry,
generation of downstream power plants and the ecological flow of the and dry years. Among them, there is almost no abandonment in the dry
river. Fig. S3 shows the monthly operating water levels of Jinping-I and year. Also, we found that the degree of water abandonment in different
Ertan hydropower plants under scenarios S1-S6 in a normal year: the ecological scenarios and hydrological years is inconsistent. In a wet year,
water level of the reservoir drops to the lower limit under the six sce­ the annual abandoned water shows a significant downward trend from
narios, which vacates enough storage capacity to ensure the safety of scenario S1 to S6, from 318.57 to 246.72 × 108 m3, respectively.
flood control during the flood season. At the ending of the flood season, However, scenario S3 is the smallest amount of abandoned water among
the reservoir will gradually store water, and it will be stored to the the six scenarios for other typical years.
normal water level to ensure water use during the dry season by the end The better to explore the problem of abandoned water for hydro­
of October. Moreover, Fig. S4 shows the outflow from Jinping-I and power, the hydrological year in which the most severe abandonment
Ertan hydropower plants under scenarios S1-S6 in a normal year: the occurs was taken as an example, i.e., the wet year, and the distribution
outflow from the reservoir during the dry season is less than the annual characteristics of the abandoned water of cascade hydropower plants
mean flow at the dam site, while the opposite is true in the flood season, during the year were analysed. The water is mainly abandoned in the
that is, the outflow from the reservoir is greater than the annual mean lower reach of the Yalong River from July to October during the flood
flow at the dam site under all six scenarios. season, and no water is abandoned in the dry season. The amount of
Table 5 lists the cascade power loss in five typical years under sce­ abandoned water in July is the largest, and the amount of abandoned
narios S2-S6 compared to S1 (reference scenario). The annual average water from July to October shows a decreasing trend. Also, we found
cascade power losses under scenarios S2-S6 are 60.20, 89.41, 115.19, that the cascade abandoned water decreases with the decrease of the
136.69, and 161.91 × 108 kWh, respectively, which indicates that the average ecological flow under flood season, showing a positive corre­
ecological flow of the river has a significant effect on power generation, lation (Fig. S5).
and the operation of the cascade hydropower plants is considered such
that the better the EFRs of the river, the higher the cascade power loss. 5. Discussion
Besides, the effect of ecological flow on the cascade power generation is
inconsistent in different typical years, and the release of ecological flow The traditional operation of cascade hydropower plants reduces the
peak discharge in flood seasons and increases the water volume in the
dry season through the storage function of reservoirs. At the same time,
these plants bring huge power generation benefits but ignore or un­
derestimate the ecological environment impact. Also, the modes of
operation of cascade hydropower plants are closely related to the
ecological health of the river. In the present study, the main goal of
optimal operation of cascade hydropower plants is to pursue the maxi­
mization of comprehensive benefits of hydro-energy generation and
ecological flows and explore the tradeoff relationship of three optimal
objectives under six scenarios (S1-S6) in different hydrological condi­
tions. Here, we took a normal year as an example, to illustrate the
tradeoffs between two optimal objectives, i.e., the CPG and the ARCPG,
the CPG and the AREF, and the ARCPG and the AREF
It can be seen from Fig. 6 that these three objectives have a certain
degree of tradeoff relationship. Therefore, when analysing the tradeoff
relationship between the two objectives, it is necessary to exclude the
influence of the third objective on the two objectives. For the ARCPG
and the AREF (discrete value), we employed the controlled variable
method, such as analysing the tradeoff relationship between the CPG
and the ARCPG under the same AREF. At the same time, the appropriate
Fig. 7. Power generation of five power plants of optimal schemes under sce­ number of representative values is selected so that they can represent
narios S1-S6 in a normal year (108 kWh). the entire range. For the CPG (continuous value), we selected scenarios

10
L. Yu et al. Journal of Hydrology 601 (2021) 126599

Table 5
Cascade power loss of scenarios S1-S6 compared to S1 (reference scenario) in five typical years (108 kWh).
Scenario Typical years Annual average

Wet year Relatively wet year Normal year Relatively dry year Dry year

S1 0 0 0 0 0 0
S2 37.58 51.78 68.57 70.2 72.85 60.20
S3 61.95 81.61 98.09 100.05 105.33 89.41
S4 86.43 112.37 129.38 130.86 116.89 115.19
S5 111.79 141.86 159.95 160.72 107.61 136.39
S6 137.63 174.01 193.08 185.04 119.81 161.91

5.2. Tradeoff between cascade power generation and the assurance rate
Table 6
of ecological flow
Annual abandoned water of cascade hydropower plants under scenarios S1-S6 in
typical years on the lower reach of the Yalong River (108 m3).
To illustrate the tradeoff relationship between CPG and the AREF,
Scenario Typical years the Pareto non-inferior set of solutions are presented in Fig. 9. As seen in
Wet Relatively wet Normal Relatively dry Dry this figure, there is a significant tradeoff between the CPG and the AREF
year year year year year under scenarios S2-S6, that is, when the CPG increases, the AREF de­
S1 318.57 93.15 8.12 1.42 0.00 creases sharply. It is added that the number of non-inferior solutions in
S2 276.40 70.02 7.12 1.52 0.08 scenario S1 is small, so we did not analyse it here. This means that CPG
S3 268.63 69.32 6.95 0.41 0.00 has a remarkably adverse impact on the AREF. Thus, it is necessary to
S4 261.02 71.14 7.37 1.16 0.00
S5 254.56 70.65 16.47 2.01 0.00
consider the tradeoff between the two objectives when decision-makers
S6 246.72 70.98 22.26 8.60 6.79 choose the appropriate operation mode of cascade hydropower plants.

5.3. Tradeoff between the assurance rate cascade power generation and
S2, S4, and S6 as representatives to illustrate the tradeoff relationship
the assurance rate of ecological flow
between the ARCPG and AREF.

To illustrate the tradeoff relationship between the ARCPG and the

5.1. Tradeoff between cascade power generation and the assurance rate
AREF, we had standardised the CPG, which can more clearly show the
of cascade power generation
tradeoff between these two objectives under the same CPG. The Pareto
non-inferior set of solutions are presented in Fig. 10. It is difficult to find
To illustrate the tradeoff relationship between CPG and the ARCPG,
a significant tradeoff relationship between the two objectives from the
the Pareto non-inferior set of solutions are presented in Fig. 8. As seen in
figure. Thus, decision-makers do not consider the tradeoff between the
this figure, there is a significant tradeoff between the two objectives
two objectives when they choose the appropriate operation mode of
under scenarios S1-S6, that is, when the CPG increases, the ARCPG de­
cascade hydropower plants.
creases. Also, it is found that these Pareto fronts have some turning
Overall, there is a significant tradeoff relationship between the CPG
points (dotted circle) under scenarios S1-S6. For instance, for scenario
and the ARCPG, and the CPG and the AREF. That is, as the CPG in­
S1, the ARCPG slowly decreases as CPG increases from 773.80 to
creases, the ARCPG and the AREF show a downward trend. It is worth
777.25 × 108 kWh. However, the ARCPG decreases sharply after CPG
noting that there is a clear turning point in these Pareto frontiers (Fig. 8).
increases beyond 777.25 × 108 kWh (Fig. 8). This means that CPG
However, there is no obvious tradeoff between the ARCPG and the
should be controlled at this turning point to avoid severe damage to the
AREF. Therefore, this reminds decision-makers of the operation for
power generation reliability of cascade hydropower plants.
cascade hydropower plants to focus on the tradeoffs between the CPG
and the ARCPG, and the CPG and the AREF.

Fig. 8. Pareto front of the NSGA-II concerning cascade power generation (CPG) and the assurance rate of cascade power generation (ARCPG) under scenarios S1-S6
in a normal year.

11
L. Yu et al. Journal of Hydrology 601 (2021) 126599

Fig. 9. Pareto front of the NSGA-II concerning cascade power generation (CPG) and the assurance rate of ecological flow (AREF) under scenarios S2-S6 in a
normal year.

obtained by the entropy method tended to be ecological indicators

(Table S2). Therefore, the combined weighting method (coupled sub­
jective and objective weighting method) needs to be considered in the
next research to make the calculation of weights more reasonable.
Also, we only analysed the tradeoffs between the two objectives, not
the overall tradeoff relationship between the three objectives. In future
studies, in combination with external factors such as climate change,
population growth, and the low-carbon economy, the uncertain analysis
of the cascade optimal operation model considering multiple ecological
flow warrants further research. Further, how to quantify the economic
benefits of ecological flow may be very important to the research into
the benefit tradeoff of hydro-energy generation and river ecosystem
projection for cascade hydropower plants.

6. Conclusion

An ecological flow method (RVMF) was proposed for various

ecological conditions in the river based on the VMF method and com­
bined with the classification idea of the Tennant method, and six
Fig. 10. Pareto front of the NSGA-II concerning the assurance rate of cascade ecological scenarios (S1-S6) were established with concomitant calcu­
power generation (ARCPG) and the assurance rate of ecological flow (AREF) lation results. A decision-making support model was developed to
under scenarios (S2, S4, and S6) in a normal year. quantify the benefit tradeoffs between the optimal objectives. The main
results were as follows:
5.4. Limitations and future research directions
(1) The RVMF method qualitatively describes river habitats based on
This study designs a set of methods for in-depth study of the multi- a specific percentage of monthly mean flow (MMF) into six levels:
objective ecological operation of cascade hydropower plants, poor, fair, good, excellent, outstanding, and optimum. The cor­
including ecological flow method, multi-objective optimal operation responding average values of ecological flow are 14.58, 43.75,
model, multi-objective solution algorithm, and multi-objective decision- 53.75, 63.75, 73.75, and 83.75 (% MMF), respectively. The
making method. The advantages and disadvantages of these methods analysis of the seasonal variation of runoff and the hydrological
will be briefly discussed below. The revised VMF method (RVMF) rhythm of natural runoff shows that the RVMF method is
originally proposed in this study aims to calculate the ecological flow reasonable and has been applied in the Yalong River Basin.
under different river ecological conditions for the study area with only (2) A multi-objective optimal operation model was established to
hydrological data. However, the rationality of this method needs to be minimise the contradiction between the hydro-energy generation
further demonstrated in combination with more measured data, and river ecosystem projection. The Pareto non-inferior set of
including hydrology, hydraulics, and ecology, to realize its application solutions for five typical years was obtained, and the entropy-
and promotion. The NSGA-II algorithm has the advantages of rapidity weighted TOPSIS model was used to determine the optimal
and good convergence, but its global search ability is poor, and it is easy schemes. The results of the optimal operation showed that as the
to appear “early converging” phenomenon. The entropy-weighted EFRs of the river increase, the CPG decreases significantly, and
TOPSIS method has no strict restrictions and requirements in terms of the most affected hydropower plant is the Jinping-II.
the number of indicators, also referred to as criteria, samples, and data. (3) The tradeoffs between hydro-energy generation and river
Also, this method can avoid the irrationality of the subjective weighting ecosystem projection with three optimal objectives were quan­
of the traditional TOPSIS method. However, we found that the weights tified under different scenarios (S1-S6) in five typical years. There

12
L. Yu et al. Journal of Hydrology 601 (2021) 126599

are clear tradeoffs between the CPG and the ARCPG, and the CPG Hussain, J., Zhou, K., Guo, S.L., Khan, A., 2020. Investment risk and natural resource
potential in “Belt & Road Initiative” countries: a multi-criteria decision-making
and the ARCPG under different riparian ecological conditions.
approach. Sci. Total Environ. 723, 13. https://doi.org/10.1016/j.
However, the tradeoff between the ARCPG and AREF is not scitotenv.2020.137981.
significant. Hwang, C., Yoon, K., 1981. Multiple Attribute Decision Making: Methods and
Applications. Springer, Berlin, Germany.
Jiang, Z.Q., Wu, W.J., Qin, H., Zhou, J.Z., 2018. Credibility theory based panoramic
Declaration of Competing Interest fuzzy risk analysis of hydropower station operation near the boundary. J. Hydrol.
565, 474–488. https://doi.org/10.1016/j.jhydrol.2018.08.048.
Jiang, Z., Liu, P., Ji, C., Zhang, H., Chen, Y., 2019. Ecological flow considered multi-
The authors declare that they have no known competing financial objective storage energy operation chart optimization of large-scale mixed
interests or personal relationships that could have appeared to influence reservoirs. J. Hydrol. 577, 123949. https://doi.org/10.1016/j.jhydrol.2019.123949.
the work reported in this paper. Li, M.D., Fan, J.T., Zhang, Y., Guo, F., Liu, L.S., Xia, R., Xu, Z.X., Wu, F.C., 2018.
A systematic approach for watershed ecological restoration strategy making: an
application in the Taizi River Basin in northern China. Sci. Total Environ. 637–638,
Acknowledgement 1321–1332. https://doi.org/10.1016/j.scitotenv.2018.04.411.
Liu, J., Yin, Y.u., 2019. An integrated method for sustainable energy storing node
optimization selection in China. Energy Convers. Manage. 199, 112049. https://doi.
We much appreciate the contributions of the experts who partici­
org/10.1016/j.enconman.2019.112049.
pated in this review. This study was supported by the National Key R & D Magilligan, F.J., Nislow, K.H., 2005. Changes in hydrologic regime by dams.
Programme of China (2017YFC0404605), the National Natural Science Geomorphology 71 (1-2), 61–78.
Foundation of China (51709178), the China Postdoctoral Science Parish, E.S., Pracheil, B.M., McManamay, R.A., Curd, S.L., DeRolph, C.R., Smith, B.T.,
2019. Review of environmental metrics used across multiple sectors and geographies
Foundation (2019 M661884), the Science and Technology Support to evaluate the effects of hydropower development. Appl. Energy 238, 101–118.
Program of Jiangsu Province (SBK2020041282), the financial support https://doi.org/10.1016/j.apenergy.2019.01.038.
from Nanjing Hydraulic Research Institute (Y120006) and the Post­ Pastor, A.V., Ludwig, F., Biemans, H., Hoff, H., Kabat, P., 2014. Accounting for
environmental flow requirements in global water assessments. Hydrol. Earth Syst.
graduate Research & Practice Innovation Programme of Jiangsu Prov­ Sci. 18, 5041–5059. https://doi.org/10.5194/hess-18-5041-2014.
ince (KYCX20_0488). Petts, G.E., 1996. Water allocation to protect river ecosystems. Regulated Rivers Res.
Manag. 12 (4-5), 353–365.
Poff, N.L., Matthews, J.H., 2013. Environmental flows in the Anthropocence: past
Appendix A. Supplementary data progress and future prospects. Curr. Opin. Environ. Sustain. 5 (6), 667–675. https://
doi.org/10.1016/j.cosust.2013.11.006.
Supplementary data to this article can be found online at https://doi. Rees, G.N., Cook, R.A., Ning, N.S.P., McInerney, P.J., Petrie, R.T., Nielsen, D.L., 2020.
Managed floodplain inundation maintains ecological function in lowland rivers. Sci.
org/10.1016/j.jhydrol.2021.126599. Total Environ. 727, 138469. https://doi.org/10.1016/j.scitotenv.2020.138469.
Ren, K., Huang, S.Z., Huang, Q., Wang, H., Leng, G.Y., Cheng, L.Y., Fang, W., Li, P., 2019.
References A nature-based reservoir optimization model for resolving the conflict in human
water demand and riverine ecosystem protection. J. Cleaner Prod. 231, 406–418.
https://doi.org/10.1016/j.jclepro.2019.05.221.
Arthington, A.H., Bunn, S.E., Poff, N.L., Naiman, R.J., 2006. The challenge of providing
Renofalt, B.M., Jansson, R., Nilsson, C., 2010. Effects of hydropower generation and
environmental flow rules to sustain river ecosystems. Ecol. Appl. 16 (4), 1311–1318.
opportunities for environmental flow management in Swedish riverine ecosystems.
https://doi.org/10.1890/1051-0761(2006)016[1311:TCOPEF]2.0.CO;2.
Freshw. Biol. 55, 49–67. https://doi.org/10.1111/j.1365-2427.2009.02241.x.
Boyaghchi, F.A., Sohbatloo, A., 2018. Assessment and optimization of a novel solar
Schluter, U., 1971. Uberlegungen zum naturnahen Ausbau von Wasserlaufen. Landschaft
driven natural gas liquefaction based on cascade ORC integrated with linear Fresnel
and Stadt 9, 72–83.
collectors. Energy Convers. Manage. 162, 77–89. https://doi.org/10.1016/j.
Shadkam, S., Ludwig, F., van Vliet, M.T.H., Pastor, A., Kabat, P., 2016. Preserving the
enconman.2018.02.013.
world second largest hypersaline lake under future irrigation and climate change.
Chang, J.X., Li, Y.Y., Yuan, M., Wang, Y.M., 2017. Efficiency evaluation of hydropower
Sci. Total Environ. 559, 317–325. https://doi.org/10.1016/j.scitotenv.2016.03.190.
station operation: A case study of Longyangxia station in the Yellow River, China.
Smakhtin, V., Revenga, C., Döll, P., 2004. A pilot global assessment of environmental
Energy 135, 23–31. https://doi.org/10.1016/j.energy.2017.06.049.
water requirements and scarcity. Water Int. 29 (3), 307–317. https://doi.org/
Chen, D., Chen, Q., Li, R., Blanckaert, K., Cai, D., 2014. Ecologically-friendly operation
10.1080/02508060408691785.
scheme for the Jinping cascaded reservoirs in the Yalongjiang River. China. Front.
Steinschneider, S., Bernstein, A., Palmer, R., Polebitski, A., 2014. Reservoir management
Earth Sci. 8 (2), 282–290. https://doi.org/10.1007/s11707-013-0396-5.
optimization for basin-wide ecological restoration in the Connecticut River. J. Water
Deb, K., Pratap, A., Agarwal, S., Meyarivan, T., 2002. A fast and elitist multiobjective
Res. Plann. Manag. 140 (9), 04014023. https://doi.org/10.1061/(ASCE)WR.1943-
genetic algorithm: NSGA-II. IEEE Trans. Evol. Comput. 6 (2), 182–197.
5452.0000399.
Ding, Z., Fang, G., Wen, X., Tan, Q., Lei, X., Liu, Z., Huang, X., 2020. Cascaded
Tan, Q.F., Lei, X.H., Wang, X., Wang, X., Wang, H., Wen, X., Yi, J., Kang, A.Q., 2018. An
hydropower operation chart optimization balancing overall ecological benefits and
adaptive middle and long-term runoff forecast model using EEMD-ANN hybrid
ecological conservation in hydrological extremes under climate change. Water
approach. J. Hydrol. 567, 767–780. https://doi.org/10.1016/j.jhydrol.2018.01.015.
Resour. Manage. 34 (3), 1231–1246. https://doi.org/10.1007/s11269-020-02496-6.
Tennant, D.L., 1976. Instream flow regimens for fish, wildlife, recreation and related
Feng, Z.K., Liu, S., Niu, W.J., Li, S.S., Wu, H.J., Wang, J.Y., 2020. Ecological operation of
environmental resources. Fisheries 1 (4), 6–10.
cascade hydropower reservoirs by elite-guide gravitational search algorithm with
Tharme, R.E., 2003. A global perspective on environmental flow assessment: emerging
Levy flight local search and mutation. J. Hydrol. 581 https://doi.org/10.1016/j.
trends in the development and application of environmental flow methodologies for
jhydrol.2019.124425.
rivers. River Res Appl 19 (5–6), 397–441. https://doi.org/10.1002/(ISSN)1535-
Gibeau, P., Connors, B.M., Palen, W.J., 2017. Run-of-River hydropower and salmonids:
146710.1002/rra.v19:5/610.1002/rra.736.
potential effects and perspective on future research. Can. J. Fish Aquat. Sci. 74 (7),
Tsai, W.P., Chang, F.J., Chang, L.C., Herricks, E.E., 2015. AI techniques for optimizing
1135–1149. https://doi.org/10.1139/cjfas-2016-0253.
multi-objective reservoir operation upon human and riverine ecosystem demands.
Grill, G., Lehner, B., Thieme, M., Geenen, B., Tickner, D., Antonelli, F., Babu, S.,
J. Hydrol. 530, 634–644. https://doi.org/10.1016/j.jhydrol.2015.10.024.
Borrelli, P., Cheng, L., Crochetiere, H., Ehalt Macedo, H., Filgueiras, R., Goichot, M.,
Wan, D.H., Xia, J., 2007. Analysis of ecological water requirements in Yalong River of
Higgins, J., Hogan, Z., Lip, B., McClain, M.E., Meng, J., Mulligan, M., Nilsson, C.,
West Route of South-to-North Water Transfer Project. Engineering Journal of Wuhan
Olden, J.D., Opperman, J.J., Petry, P., Reidy Liermann, C., Sáenz, L., Salinas-
University 6, 1–5 (in Chinese).
Rodríguez, S., Schelle, P., Schmitt, R.J.P., Snider, J., Tan, F., Tockner, K., Valdujo, P.
Wang, Y.R., Li, J., Li, K.F., Rui, J.L., 2007. Ecological water demand of reducing reach of
H., van Soesbergen, A., Zarfl, C., 2019. Mapping the world’s free-flowing rivers.
Yalong river downstream of jinping waterpower station stage II. Resources and
Nature 569 (7755), 215–221. https://doi.org/10.1038/s41586-019-1111-9.
Environment in the Yangtze Basin 1, 81–85 (in Chinese).
Huang, L., Li, X., Fang, H.W., Yin, D.Q., Si, Y., Wei, J.H., Liu, J.H., Hu, X.Y., Zhang, L.,
Wheeler, K., Wenger, S.J., Freeman, M.C., 2017. States and rates: complementary
2019a. Balancing social, economic and ecological benefits of reservoir operation
approaches to developing flow-ecology relationships. Freshw. Biol. 8, 1–11.
during the flood season: A case study of the Three Gorges Project. China. J. Hydrol.
Wu, X., Huang, X., Fang, G., Kong, F., 2011. Optimal operation of multi-objective
572, 422–434. https://doi.org/10.1016/j.jhydrol.2019.03.009.
hydropower reservoir with ecology consideration. J. Water Resour. Prot. 03 (12),
Huang, S.Z., Chang, J.X., Huang, Q., Wang, Y.M., Chen, Y.T., 2014. Calculation of the
904–911.
instream ecological flow of the wei river based on hydrological variation. J. Appl.
Xu, W., 2020. Study on multi-objective operation strategy for multi-reservoirs in small-
Math. 2014, 1–9. https://doi.org/10.1155/2014/127067.
scale watershed considering ecological flows. Water Resour. Manage. 34 (15),
Huang, Z., Xie, Z., Zhang, C., Chan, S.H., Milewski, J., Xie, Y., Yang, Y., Hu, X., 2019b.
4725–4738. https://doi.org/10.1007/s11269-020-02685-3.
Modeling and multi-objective optimization of a stand-alone PV-hydrogen-retired EV
Xu, C., Xu, Z., Yang, Z., 2020a. Reservoir operation optimization for balancing
battery hybrid energy system. Energy Convers. Manage. 181, 80–92. https://doi.
hydropower generation and biodiversity conservation in a downstream wetland.
org/10.1016/j.enconman.2018.11.079.
J. Cleaner Prod. 245, 118885. https://doi.org/10.1016/j.jclepro.2019.118885.
Hughes, D.A., Ziervogel, G., 1998. The inclusion of operating rules in a daily reservoir
Xu, C., Ke, Y., Li, Y., Chu, H., Wu, Y., 2020b. Data-driven configuration optimization of
simulation model to determine ecological reserve releases for river maintenance.
an off-grid wind/PV/hydrogen system based on modified NSGA-II and CRITIC-
Water SA 24, 293–302.

13
L. Yu et al. Journal of Hydrology 601 (2021) 126599

TOPSIS. Energy Convers. Manage. 215, 112892. https://doi.org/10.1016/j. Zhang, H.X., Chang, J.X., Gao, C., Wu, H.S., Wang, Y.M., Lei, K.X., Long, R.H., Zhang, L.
enconman.2020.112892. P., 2019. Cascade hydropower plants operation considering comprehensive
Ye, Y., Huang, W., Ma, G., Wang, J., Liu, Y., Hu, Y., 2018. Cause analysis and policy ecological water demands. Energy Convers. Manage. 180, 119–133. https://doi.org/
options for the surplus hydropower in southwest China based on quantification. 10.1016/j.enconman.2018.10.072.
J. Renew. Sustain. Energy 10 (1), 015908. https://doi.org/10.1063/1.5024256. Zhou, Y.L., Guo, S.L., 2013. Incorporating ecological requirement into multipurpose
Yu, L., Jia, B.Y., Wu, S.Q., Wu, X.F., Xu, P., Dai, J.Y., Wang, F.F., Ma, L.M., 2019. reservoir operating rule curves for adaptation to climate change. J. Hydrol. 498,
Cumulative environmental effects of hydropower stations based on the water 153–164. https://doi.org/10.1016/j.jhydrol.2013.06.028.
footprint method-Yalong River Basin China. Sustainability 11, 12. https://doi.org/
10.3390/su11215958.

14