Download as pdf or txt
Download as pdf or txt
You are on page 1of 41

Accepted Manuscript

China’ energy-water nexus: Hydropower generation potential of joint operation of the


Three Gorges and Qingjiang cascade reservoirs

Yizi Shang, Shibao Lu, Yuntao Ye, Ronghua Liu, Ling Shang, Chunna Liu, Xianyong
Meng, Xiaofei Li, Qixiang Fan

PII: S0360-5442(17)31652-3
DOI: 10.1016/j.energy.2017.09.131
Reference: EGY 11625

To appear in: Energy

Received Date: 25 May 2017


Revised Date: 2 September 2017
Accepted Date: 26 September 2017

Please cite this article as: Shang Y, Lu S, Ye Y, Liu R, Shang L, Liu C, Meng X, Li X, Fan Q, China’
energy-water nexus: Hydropower generation potential of joint operation of the Three Gorges and
Qingjiang cascade reservoirs, Energy (2017), doi: 10.1016/j.energy.2017.09.131.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Hydropower generation potential of joint operation of Three Gorges and
Qingjiang cascade reservoirs

Graphical Abstract:

PT
RI
U SC
AN
M
D
TE
C EP
AC
ACCEPTED MANUSCRIPT

1 China’ energy-water nexus: Hydropower generation potential of joint operation of the Three
2 Gorges and Qingjiang cascade reservoirs
3 Yizi Shanga,b, Shibao Luc*, Yuntao Yea*, Ronghua Liua, Ling Shangd, Chunna Liua, Xianyong Menga, Xiaofei Lie, Qixiang
4 Fanb
a
5 State Key Laboratory of Simulation and Regulation of Water Cycles in River Basins, China Institute of Water Resources and Hydropower
6 Research, Beijing 100038, China (Email: yzshang@foxmail.com(Dr. Yizi Shang); yeyuntao@iwhr.com(Dr. Yuntao Ye))
b
7 China Three Gorges Corporation, Beijing 100038, China
c
8 School of Public Administration, Zhejiang University of Finance and Economics, Hangzhou 310018, China (Email:
9 lu5111284@aliyun.com(Dr. Shibao Lu))
d
10 College of Computer and Information, Hohai University, Nanjing 210000, China
e
11 CABR Technology Co., Ltd, China Academy of Building Research, Beijing 100013, China

PT
12 Abstract: The completion of Shuibuya, the last hydropower station to be constructed in the Qingjiang
13 cascade, signifies China’s successful development of the world’s largest mixed cascade hydropower
14 generation system. Joint operation of such cascaded hydropower stations is considered necessary to

RI
15 improve hydropower output in China. In this study, two modeling methods – routine and optimal
16 operations – were adopted, based on existing rules of reservoir operation, to determine the effects of
17 joint operation. To investigate the realistic and potential effects, the two models were computed using

SC
18 observed runoff data (1951–2009). The potential effects were identified by comparing the total
19 hydropower generation two single cascades and their cogeneration. Under the routine operational
20 mode, the incremental power generation of joint operation would be 5.73 × 108 kWh, an increase of
21 5‰ compared to isolated operation. However, even in isolation, if reservoir operation is optimized

U
22 through the dynamic programming algorithm, the incremental power generation of the cascade would
23 be 42.24 × 108 kWh, which is seven times that of joint operation with routine reservoir operation
AN
24 (5.73 × 108 kWh). The results showed that, under current reservoir operating rules, there is little room
25 for improvement in hydropower generation, although joint operation could increase hydropower
26 generation to a certain extent, especially in the reservoirs’ refill and release stages. China must amend
27 its existing operational mode for reservoirs to enhance the economic benefits of cascade hydropower
M

28 stations. Furthermore, to meet increasing demands for both water and energy, carefully considered
29 planning of constructing clusters of these stations is required.
30 Keywords: China; hydropower; Three Gorges cascade; Qingjiang cascade; reservoir joint operation.
D

31
32 1 Introduction
TE

33 Hydropower is a clean, renewable energy source that is attracting much attention worldwide. As
34 traditional energy resources dwindle, many countries have afforded priority status to hydropower
35 generation, intensifying their use of hydropower resources [1]. Cascade development is at the
EP

36 forefront of developing and utilizing river hydropower resources [2], and cascade hydropower station
37 clusters within a river basin highlight the comprehensive benefits of such systems. Cascaded
38 hydropower stations must utilize compensation coordination among the reservoirs with regard to the
39 overall benefit to operational management, to optimize the hydropower resources’ use and maximize
C

40 each unit’s efficiency [3]. Therefore, assessing the joint generation potential of cascaded hydropower
41 stations is key to such entities’ efficient operation [4]. Each hydropower station in a cascade is a link
AC

42 between a reservoir and the power grid. As there is variation in both the water supply from an
43 upstream reservoir and the power load of the grid, the generating status of each hydropower station
44 changes dynamically [5]. Thus, the maximum power generation of an entire cascade, which is the
45 objective of hydropower station regulation, could be temporary and dynamic [6]. Dynamic adjustment
46 of the individual generating units makes it difficult to assess the benefits to power generation of joint
47 operation. In addition, the upstream and downstream water levels of the reservoirs within a cascade
48 are connected, and cascaded hydropower stations jointly bear the power grid load. Therefore, the
49 relationship between the hydropower resources and the generation potential of hydropower stations
50 within a cascade is complex [7], which increases the difficulty of research [8]. Accordingly, to
51 determine the benefit to power generation of joint operation of cascaded reservoirs, two basic
52 problems must be addressed: 1) choosing the assessment method, and 2) establishing how best to
53 realize the maximum power generation of the entire cascade system (power generation benefits).

1
ACCEPTED MANUSCRIPT
54 It is generally accepted that joint operation has the potential to significantly improve the power
55 generation of cascaded hydropower stations [9,10]. Based on many years’ experience of development
56 and operational management of cascaded hydropower stations, numerous European and American
57 countries actively promote joint operation and integrated management [11]. Relevant systems in these
58 countries are operational on the Columbia River in the USA (42 cascades), Parana River basin in
59 Brazil (30 cascades), and Yenisei River in Russia (9 cascades). Statistical data indicate that, compared
60 with the isolated operation of individual hydropower stations, joint operation of cascaded stations
61 could increase their power generation benefits by 4%–9% [12]. In 2009, the Shuibuya hydropower
62 station in the Qingjiang (QJ) cascade was formally linked to the energy generation grid, marking
63 China’s achievement in constructing the world’s largest mixed cascade hydropower station cluster

PT
64 system [13]. Currently, the Three Gorges (TG) and QJ cascades operate in isolation; however, China
65 is expected to improve the power generation benefits of such cascades by adopting joint operation
66 procedures [14].

RI
67 Various factors could be exploited in the joint operation of the power stations of the TG and QJ
68 cascades. For example, differences in the geographical positions of the reservoirs, their regulation
69 performances, and the power stations’ power-bearing loads could be used for mutual coordination and

SC
70 mutual compensation related to flood control [15], power generation [16], water supply [17], and
71 ecological protection [18], thus realizing the comprehensive benefits of a cascaded hydropower
72 station cluster [19]. Chen et al. [20] indicate that, in the flood season, joint operation of the two
73 cascades could improve flood-regulating capacity in the upstream reaches of the Yangtze River,

U
74 potentially improving, in turn, the flood control standard of the downstream watercourse from a
75 10-year flood to a 113-year flood. Regarding the non-flood season, Guo et al. [21] indicate that joint
AN
76 operation of the two cascades would improve the various water demands in the downstream reaches
77 of the Yangtze River. The above-mentioned research focused mainly on flood defense and ecological
78 protection; little attention was given to systematic assessment of the overall improvement of power
79 generation benefits. Here, power generation benefits are considered to stem from the water
M

80 compensation of two cascades’ joint operation, while ensuring flood control, shipping, and water
81 supply safety [22]. Currently, the reservoirs of the TG and QJ cascades are operated according to the
82 designed reservoir operating rule curves. In practice, the operational processes of the reservoirs,
D

83 power stations, and power generation enterprises are optimized manually and regulated according to
84 the reservoir operating rule curves. In other words, the generation potential of cascaded power stations
85 is based on the possible increase in their power generation under the restrictions of the reservoir
TE

86 operating rule curves and power generation plans [23]. Calculation of the incremental power
87 generation resulting from joint operation of cascades under current reservoir operational modes, and
88 assessments of the power generation potential and increased benefits, have not been subject to prior
89 research; thus, they constitute the fundamentals of this study.
EP

90 Through a holistic assessment, the current study’s principal contribution is to fill a knowledge gap
91 concerning the TG–QJ cascade’s power generation benefits, based on the current operational situation
92 of the cascaded reservoirs. The assessment provides a technical reference that could be used in
C

93 restructuring China’s hydropower system. Our research method could also be adapted to improve
94 research on mixed cascade power stations’ power generation potential in other countries. The
AC

95 remainder of this article is structured as follows. Section 2 presents related work. In section 3, we
96 provide an overview of five hydropower stations of the two cascades, and introduce the data series
97 used in this research. In section 4, we present our assessment methodology, separately introducing the
98 assessment methods for routine operation (subsection 4.1) and optimal operation (subsection 4.2). The
99 results analysis and discussion are presented in section 5. Finally, section 6 concludes this paper.

100 2 Related work


101 Various models and methods exist for assessing the optimal operation of hydropower stations, and
102 common algorithms include linear programming (LP) [24], nonlinear programming (NLP) [25],
103 dynamic programming (DP) [26], and heuristic algorithms [27,28]. The LP algorithm is the most
104 widely used optimization solution [29,30]: it is superior for solving multidimensional spatiotemporal
105 problems, and it can ensure that the obtained result is a global optimal solution. Barros et al. [31]

2
ACCEPTED MANUSCRIPT
106 established a joint operation model for 75 hydropower stations in Brazil, which they used to optimize
107 the operation of the entire power generation system. They found that the optimized system
108 significantly improved power quantity. At present, the power supply from such systems accounts for
109 92% of Brazil’s entire power supply. For a mathematical mode, the LP algorithm requires a linear
110 relation between the objective function and the constraint condition. However, reservoir operation for
111 maximum power generation quantity (power generation benefits) is a nonlinear objective [32].
112 Therefore, when the LP algorithm is adopted, the nonlinear relation must be linearized using the
113 piecewise linear and first-order Taylor expansion method [33]. The use of NLP can overcome optimal
114 operation problems, which include the nonlinear relation [34]. For example, Zambon et al. [35] used
115 NLP to develop the HIDROTERM model for optimizing the joint operation of Brazil’s hydrothermal

PT
116 power system. This system includes 127 hydropower stations and various thermal power stations. The
117 power generated by these stations is transmitted via a unified power grid to the national power
118 consumption centers. The optimal model’s objective is to minimize the difference between power

RI
119 generation and power grid demand. However, the NLP algorithm has strict requirements regarding a
120 power system’s power generation characteristics [36]. For example, the mathematical function
121 describing power generation must be differentiable and continuous; consequently, NLP is not used
122 widely [37]. Moreover, as the operational model of cascaded power stations has strong non-convexity

SC
123 [38],the NLP algorithm might be unable to obtain a global optimal solution [39].
124 The DP algorithm was proposed by Bellman in 1957 [40]. Since then, it has been used widely for
125 optimizing the operation of hydropower station systems [41,42]. However, it suffers from the

U
126 so-called “curse of dimensionality” [43], which restricts the DP algorithm’s application to cooperative
127 generation by multiple power stations. This is because when the number of power stations increases,
AN
128 the state space increases exponentially and, consequently, the computer memory demand exceeds the
129 hardware capacity of modern computers [44]. To overcome this dimensionality problem, various
130 changes to the DP algorithm have been suggested, including incremental dynamic programming [45],
131 dynamic programming with successive approximation [ 46 ], discrete differential dynamic
M

132 programming [ 47 ], and the progressive optimality algorithm [ 48 ]. Most of these algorithms
133 approximate the optimal solution by reducing the state space size, premised on presenting an initial
134 solution and then using an iteration method [49,50]. However, when the number of reservoirs
D

135 increases, larger computational memory is required for the calculation (the curse of dimensionality),
136 meaning that DP algorithms are potentially unsuitable for implementation in stand-alone computers
137 [51]. Mousavi and Karamouz [52] adopted the DP algorithm for long-term research on the operational
TE

138 mode of five hydropower stations in Iran. They reduced the algorithm’s calculation overhead by
139 identifying and avoiding unfeasible capacity state combinations. Globally, the scale of cascaded
140 hydropower station systems (already built or under construction) is large [53]. To assess such systems’
141 power generation potential, the mathematical function describing the operation of the reservoirs and
EP

142 the generation of the hydropower stations typically has multidimensional spatiotemporal
143 characteristics [54,55]. In adopting the DP algorithm, the problem of dimensionality would appear in
144 the computational process [56]. However, using a simplified simulation could lead to incomplete
C

145 assessment of the real power generation benefits [57,58]. Therefore, to guarantee the calculation
146 results’ accuracy and reliability, this research adopted the DP algorithm and used a supercomputing
AC

147 cluster to avoid the problem of dimensionality [59,60].

148 3 Project and data

149 3.1 Overview of the TG-QJ cascade


150 The TG–QJ cascade is the world’s largest mixed hydropower station cluster. The total installed
151 capacity of the cascaded power stations is about 2800 × 104 kW. The TG and QJ cascades, located in
152 China’s Yangtze River, are geographically proximate. However, because the five cascaded
153 hydropower stations were constructed to serve different purposes, their modes of operation differ
154 considerably. The positions of the cascaded hydropower stations are shown in Fig. 1. The TG cascade
155 comprises the Three Gorges and Gezhouba hydropower stations (hereafter, TGP and GZB,
156 respectively). The TGP dam is the world’s largest comprehensively utilized hydro-junction, and its
157 main benefits are in flood control, power generation, and shipping [61]. In the flood season, the TGP

3
ACCEPTED MANUSCRIPT
158 reservoir’s primary function is flood control, while power generation is secondary. In the event of a
159 flood, the reservoir impounds the flood water, causing its water level to rise. After a flood, the
160 reservoir is then emptied in preparation for the next flood event. During the dry season, power
161 generation and shipping become the primary focus. The TGP reservoir’s water level is maintained as
162 high as possible. During power generation, the TGP hydropower station generates large and violent
163 discharge flows. To avoid adverse effects on shipping activity, the reverse regulating GZB
164 hydropower station, located downstream of the TGP dam, functions to regulate the discharge flows
165 from the TGP hydropower station. Thus, rather than generating power, the objective of the GZB
166 hydropower station is to create downstream water flow conditions suitable for shipping.

PT
167 <Place Fig. 1 around here>
168 The Qingjiang River is a large tributary of the Yangtze River, along which are located (from upstream
169 to downstream) the cascaded Shuibuya, Geheyan, and Gaobazhou hydropower stations (hereafter,
SBY, GHY, and GBZ, respectively). SBY is the leading hydropower station of the QJ cascade, and it

RI
170
171 has capacity to perform multiyear watercourse runoff regulation. The second hydropower station in
172 the cascade (GHY) has capacity to regulate annual watercourse runoff. Both SBY and GHY are large
173 hydropower hubs with power generation as their primary function. However, they also provide

SC
174 benefits regarding flood control and shipping. GBZ is the reverse regulating hydropower station of
175 GHY, and it has capacity to regulate daily watercourse runoff. The reservoir regulation parameters of
176 the TG and QJ cascaded hydropower stations are listed in Table 1:

U
177 <Place Table 1 around here>
AN
178 3.2 Data acquisition and management
179 The data used in this study were obtained from the TG cascade dispatching center of the China Three
180 Gorges Corporation and the QJ cascade dispatching center of the Hubei Qingjiang Hydropower
181 Development Co., Ltd. The data and key parameters used in this study are as follows:
M

182 (1) Reservoir storage capacity-water level in front of the dam curve;
183 (2) Discharge flow-water level behind the dam curve;
D

184 (3) Output limit curve: the installed units of the TGP and GZB hydropower stations are different
185 models. The products of the weighted means of the different units’ output limit curves and the total
TE

186 unit quantity of the hydropower stations were adopted to prepare the output limit curves;
187 (4) Water head loss: reservoir inflow-water head loss curve of the TGP and GZB hydropower stations:
188 1.3 m for SBY, 0.6 m for GHY, and 0.4 m for GBZ;
EP

189 (5) Installed capacity and guaranteed output of the hydropower stations;
190 (6) Runoff data: hydrographic data in 10-day cycles for 48 hydrological years from June 1951 to
191 2009.
C

192 TG cascade: from 2000 to 2002, daily flow rate data from the hydrological station located at Yichang,
193 Hubei province, were adopted and the 10-day flow rate obtained after treatment. From 2003 to 2010,
AC

194 the daily inflow data of TGP were adopted and the 10-day inflow obtained after treatment.
195 Geographically, TGP and GZB are very close; therefore, afflux in the zone cannot be considered. The
196 discharge flow rate of TGP was regarded as the inflow of GZB.
197 QJ cascade: for the period from 2000 to 2007, the GHY inflow, the concurrent flow of the zone
198 between SBY and GHY and the zone between GHY and GBZ were calculated using the following
199 formulae: SBY inflow = 0.7526 × GHY inflow; flow in the zone between SBY and GHY = GHY
200 inflow – SBY inflow; flow in the zone between GHY and GBZ = 0.0845 × GHY inflow. From 2008
201 to 2010, the 10-day inflow of SBY was adopted; flow in the zone between SBY and GHY = GHY
202 inflow – SBY discharge; flow in the zone between GHY and GBZ = GBZ inflow – GHY discharge;
203 (7) Initial and stop water levels: the flood limit water level was adopted: 145.00 m for TG, 64.20 m
204 for GZB, 391.80 m for SBY, 192.20 m for GHY, and 78.55 m for GBZ;

4
ACCEPTED MANUSCRIPT
205 (8) Water level control conditions: reservoir operation curves were adopted to formulate the initial
206 water levels. In addition, special requirements for water level limits were set, e.g., during the flood
207 season, the upper limit of GZB’s water level was 64.80 m, whereas, during the non-flood season, this
208 limit was around 64.20 m for GZB and 78.55 m for GBZ; and
209 (9) Minimum discharge control condition.
210 It must be noted that because runoff was changed after construction of the cascaded hydropower
211 stations, most reservoir hydrological stations are now submerged. Therefore, to ensure consistency of
212 the hydrological data, this study mainly used runoff data acquired before 2009. The complete data
213 sequence covered the period 1951–2009.

PT
214 4 Methodology

215 4.1 Implementation of the routine operation method for assessment

RI
216 4.1.1 Operating rule curves

SC
217 Currently, the TG and QJ cascade reservoirs are operated based on operating rule curves. Generally, in
218 these curves, time (month and 10-day periods) is plotted as the abscissa, and reservoir water level or
219 storage capacity is plotted as the ordinate. Reservoir storage curves are drawn for different periods to
220 guide reservoir operations [62]. The operating rule curves show the relationship between the decision

U
221 variables (output, water supply, and discharge) and state variables (reservoir water level, inflow, and
222 time) in the operation. For example, in the designed operating rule curves for the TGP reservoir (Fig.
AN
223 2), months are plotted as the abscissa and reservoir storage level is plotted as the ordinate. The TGP
224 reservoir water level control curve is determined based on flood control, shipping, power generation,
225 water supply, and other regulatory objectives.
226 <Place Fig. 2 around here>
M

227 As shown in Fig. 2(a), operating rules require the reservoir water level to be lowered to 145 m (the
228 flood-limited water level) from the end of May to the beginning of June. In October, the reservoir
229 water level should be raised gradually to the normal water level of 175 m. From November to the end
D

230 of April in the following year, the reservoir water level should be maintained as high as possible to
231 generate maximum electrical power. Even though the reservoir water level will be lowered further, it
TE

232 should not fall below 155 m before the end of April. Fig. 2(b) shows how to use the operating rule
233 curves to guide the reservoir refill operation. According to the scheme, when the reservoir water level
234 is above the upper boundary curve (zone I), water should be spilled to lower the water level below its
235 normal value (175 m). The power station generates optimum output when the reservoir water level is
EP

236 below the lower boundary curve (zone III). If the water level lies in zone II, the generators are
237 activated to maximize electrical output. The designed operating rules can be regarded as a standard
238 operating policy.
C

239 4.1.2 Evaluation method and calculation procedure


240 The designed reservoir operating rule curves shown in Fig. 2 are actually a curve cluster presenting
AC

241 the safe operating conditions for the reservoir. In practice, the China Three Gorges Corporation
242 formulates an optimal operating curve for the reservoir via an integrated decision-making process,
243 premised on ensuring safe reservoir operation and maximal economic benefits. The technical working
244 chart of this process is shown in Fig. 3, in which the left-hand side depicts the process of formation of
245 actual reservoir operating rule curves. To properly evaluate the cascades’ current power generation,
246 the influence of manual intervention on the operational power generation process must be considered
247 in the evaluation and calculations. Therefore, this study adopted the actual operating rule curves for
248 evaluating and calculating the cascaded hydropower stations’ actual power generation.
249 <Place Fig. 3 around here>
250 This study used historical actual daily flow data of the TG and QJ cascades from 1951 to 2009, and
251 calculated the output regulation of manual intervention per day. Based on the regulation performance
252 and the mode of combination of the hydropower stations in the cluster, the stations were divided into

5
ACCEPTED MANUSCRIPT
253 compensating and compensated stations, and the compensation sequence was determined. Then, based
254 on a designated dry period (or cluster of continuous dry years), the operating rule curves of the
255 hydropower stations with regulated compensation were drawn. Finally, the output processes of the
256 compensated hydropower stations during the same period were added. Based on this sum, the
257 compensation regulation of all compensating hydropower stations, according to the compensation
258 sequence, was calculated to obtain the total output process of the cluster. First, equal discharge runoff
259 regulation was used to calculate the total output process of the TG cascade from upstream to
260 downstream. Then, based on the reservoir operating rule curves of the regulated hydropower stations
261 in the QJ cascade (SBY and GHY) during the given compensation sequence, the TG cascade’s output
262 was determined through a balance computation with regard to the overall guaranteed output of the two

PT
263 cascades. Finally, the total guaranteed output and the compensation benefits of the hydropower station
264 cluster were calculated.
265 The TGP hydropower station is located on the mainstream of the Yangtze River. Therefore, the

RI
266 reservoir water level and flow should meet the river’s mandatory requirements for shipping, water
267 supply, flood control, and ecology. The power generation regulation of the TGP hydropower station is
268 not large. Furthermore, GZB is a runoff hydropower station with almost no regulatory capacity.

SC
269 Therefore, the TG cascade is regarded as an integral compensated hydropower station cluster. In the
270 QJ cascade, both SBY (multiyear regulation) and GHY (annual regulation) have good regulatory
271 performance; thus, this cascade is regarded as a compensation cascade. The principle of the
272 compensation regulation calculation is based on minimizing electricity loss. If the Qingjiang River

U
273 has abandoned water, the output should be increased to reduce the electricity loss caused by water
274 abandonment. If there is excess capacity or if the Qingjiang River has no abandoned water while the
AN
275 TG cascade has some, the peak should be regulated to reduce electricity loss from water abandonment
276 by the TG cascade. To obtain the actual value of the compensation electricity, the surplus peak
277 regulation capacity of the QJ cascade and the compensation demand of the TG cascade were
278 calculated.
M

279 4.2 Implementation of optimal operational method for assessment


280 Power generation and economic benefit are two important indicators with which to assess a power
D

281 station’s economic operation. In this study, the maximum power generation and maximum benefit of
282 the reservoir cluster were taken as objective functions of the joint operation. Flood control, water
283 supply, shipping, and other objectives of the reservoir cluster were regarded as constraint conditions.
TE

284 The dynamic programming method was adopted to solve the optimization problem. To improve model
285 calculation efficiency, a parallel computing platform proposed by the author was adopted [63].

286 4.2.1 Problem formulation


EP

287 4.2.1.1 Objective function


288 Maximum power generation model:
C

289 The objective function is: max ∑ ∑ , = max ∑ ∑ , × ∆ (1)
AC

290 where Ei,t is the power generation of the hydropower station of reservoir i at time t, Ni,t is the output of
291 the hydropower station of reservoir i at time t, ∆t is the time interval, I is the number of hydropower
292 stations, and T is the number of time intervals.
293 Maximum power generation benefit model:
294 The objective function is: max ∑ ∑ , × , (2)
295 where Pi,t is the electricity price of reservoir i at time t.

296 4.2.1.2 Constraints


297 Water quantity equilibrium equation: , = , + , − ,  × ∆ (3)
 
298 Storage capacity constraint: , ≤ , ≤ , (4)

6
ACCEPTED MANUSCRIPT
 
299 Flow constraint: , ≤ , ≤ , (5)
 
300 Output constraint: , ≤ , ≤ , (6)
301 Boundary condition: , = ,! (a); , = ,"! (b) (7)
302 where Vi,t-1 and Vi,t are the initial and final storage of reservoir i at time t; Qi,t is the inflow of reservoir

303 i at time t; Ri,t is the outflow of reservoir i at time t; , is the storage corresponding to the
304 allowable minimum water level of reservoir i at time t (generally, it is the storage capacity

305 corresponding to the dead water level); , is the storage capacity corresponding to the allowable
maximum water level of reservoir i at time t (generally, it is the storage corresponding to the normal

PT
306
307 storage water level during the non-flood season and to the flood control limit water level during the

308 flood season); , is the lower limit of the outflow of reservoir i at time t, which is generally

309 required according to the requirements of downstream water supply, irrigation, and shipping; ,

RI
310 is the upper limit of the outflow of reservoir i at time t, which is generally required according to the

311 requirements of downstream maximum discharge and flood control; , is the lower limit of the

312 output of reservoir i at time t; , is the upper limit of the outflow of reservoir i at time t; Vi,initial is

SC
313 the initial storage of reservoir i; and Vi,final is the final storage of reservoir i.

314 4.2.2 Optimization method and calculation procedure

U
315 The DP algorithm is a conventional optimization method used in the field of water resources. To
316 optimize reservoir operation, DP can divide an original problem into subproblems in several stages,
AN
317 and sequentially obtain solutions according to the typical characteristics of a multistage
318 decision-making process. A discrete form of DP can effectively handle nonlinearity, non-convexity,
319 discontinuity, and other functional relations between an objective function and constraint conditions.
320 Enumeration of all state combinations can ensure global optimality of the problem. Therefore, it is
M

321 used widely in optimizing the operation of reservoirs (a cluster). For example, the variable definition
322 and recursive solving method of the DP algorithm in the maximum power generation model are as
323 follows:
D

324 Step 1: Select t according to operation requirements and runoff data;


325 Step 2: Select initial water level Zt-1 or reservoir storage capacity Vt-1 as state variables. The value can
TE

326 be varied within the allowable scope of reservoir water level or storage capacity. The scope is
327 generally discretized into equal Mt-1 state point sets or fixed steps;
328 Step 3: Select power generation flow qt as a decision variable at stage t;
EP

329 Step 4: Set the equation of water balance (3) as the state transition equation;
330 Step 5: For a staged decision problem with given initial and final states, the solution is obtained
331 through a back-stepping algorithm, the recursive equation of which is as follows:
C

∗ & ',
332 $∗ & ' = max()  & , * ' + $+  (8)
AC

333 where Nt(Vt-1, qt) is the output in initial state Vt-1, with decision variable qt at stage t; the second term is
334 the sum of the optimal output for the remaining period (i.e., from stage t+1 to the final stage) and its
335 initial state is Vt; similarly, ft*(Vt-1) is the sum of the optimal output of reservoir storage capacity Vt-1
336 from the beginning of stage t to the final stage; and Ω is the allowable decision set of decision
337 variable qt in a given Vt-1 situation, which meets the above constraint conditions.
338 The calculation of the DP algorithm includes two steps. The first is to obtain the optimal output
339 Nt*(Vti-1) between state Vti-1 at stage t and the final state at the final stage, in inverse time sequence
340 from the final state at the final stage, using Eq. (8). The second step is to obtain the optimum strategy
341 {qt} and change process of the optimal state {Vt} within the time sequence.
342 4.3 Method to assess hydropower generation potential
343 The TG cascade and the QJ cascade have yet been operated jointly. Given the current isolated mode of

7
ACCEPTED MANUSCRIPT
344 their operation, the research team formulated the reservoir operating curves for the joint operation of
345 the two cascades under the negotiation framework (Fig. 3), and co-established the finalized operation
346 scheme with the cascade operation center of the China Yangtze Power Co., Ltd. For the isolated
347 operation of the two cascades, the actual reservoir operating curves were adopted to assess power
348 generation. For both isolated and joint operation, the power generation results calculated with the DP
349 algorithm are regarded as the maximum.
350 The method to assess hydropower generation potential is as follows. First, the power generation
351 benefits rendered by the isolated operation and joint operation of the two cascades were respectively
352 calculated using reservoir operating curves. Second, the power generation benefits of the cascaded

PT
353 system were investigated using optimal operational theories and methods. Finally, hydropower
354 generation potential was assessed through comparison with the current operational mode’s power
355 generation benefits, to evaluate the generation potential of the entire cascaded hydropower system.

RI
356 5 Results and discussion
357 5.1 Assessment results of the routine operation method
358 5.1.1 Calculation results of historical power generation of all cascade hydropower stations

SC
359 Under their current operational mode, the TG cascade and the QJ cascade generate power in isolation.
360 The multiyear average power generation of the TG and QJ cascades is 977.3 × 108 and 82.0 × 108
361 kWh, respectively (total power generation: 1059.3 × 108 kWh). If the TG and QJ cascades were to

U
362 operate jointly, the total multiyear average power generation would be 1065.0 × 108 kWh (TG cascade:
363 983.4 × 108 kWh; QJ cascade: 81.6 × 108 kWh). Comparisons of the multiyear average power
AN
364 generation without and with joint operation of the TG and QJ cascades are shown in Fig. 4.
365 <Place Fig. 4 around here>
366 The blue bars in Fig. 4 show the actual power generation of the cascaded hydropower stations
M

367 evaluated according to the current individual operational mode. The yellow bars show the change in
368 power generation of the cascaded hydropower stations after joint operation, accounting for the
369 compensation role between the cascades. Red figures in brackets represent negative values. As
370 depicted, after joint operation, power generation of the TG cascade would increase by an annual
D

371 average increment of 6.2 × 108 kWh. Conversely, power generation of the QJ cascade would reduce
372 by 0.5 × 108 kWh. Therefore, the total incremental power generation of the joint TG–QJ cascade
TE

373 would be 5.7 × 108 kWh. The historical actual power generation of all the cascaded hydropower
374 stations and the simulated power generation of the cascades operating jointly are shown in Fig. 5.
375 <Place Fig. 5 around here>
EP

376 As shown in the power generation comparisons of all the hydropower stations, after joint operation of
377 the TG and QJ cascades, the annual average power generation of TGP and SBY would increase by
378 6.86 × 108 and 0.68 × 108 kWh, respectively. Further, the annual average power generation of GZB,
379 GHY, and GBZ would decrease by 0.68 × 108, 0.71 × 108, and 0.42 × 108 kWh, respectively. In
C

380 general, if the TG and QJ cascades operated jointly, the annual average power generation would
381 increase by 5.73 × 108 kWh. This means that the annual power generation after joint operation would
AC

382 be 5.4‰ MORE THAN the current annual power generation.

383 5.1.2 Analysis of joint operation’s impact on power generation of all cascade hydropower stations
384 The upstream and downstream hydropower stations in a cascade obviously have mutually influential
385 and restrictive effects. This section analyzes the effects of the increase in power generation of all the
386 cascaded hydropower stations to show the coupled relationship between the upstream and
387 downstream hydropower stations, and to improve awareness of the effects of coordinated operation
388 between cascades. The relationships of power generation change for all the hydropower stations in the
389 TG and QJ cascades are shown in Figs. 6 and 7.
390 <Place Fig. 6 around here>
391 <Place Fig. 7 around here>

8
ACCEPTED MANUSCRIPT
392 This study used the Pearson correlation coefficient to evaluate the power generation compensation
393 effect among the cascaded hydropower stations (Table 2). In general, the degree of correlation of
394 variables can be evaluated by categorizing the values as follows: 1) 0.8–1.0 denotes extremely strong
395 correlation, 2) 0.6–0.8 denotes strong correlation, 3) 0.4–0.6 denotes intermediate correlation, 4) 0.2–
396 0.4 denotes weak correlation, and 5) 0.0–0.2 denotes extremely weak or no correlation.
397 <Place Table 2 around here>
398 The Pearson correlation coefficient of the power generation increase between TGP and GZB is 0.45,
399 which indicates intermediate correlation. During power generation, the annual average power
400 generation of TGP would increase by 6.86 × 108kW·h, giving it the highest rank among the five

PT
401 hydropower stations in the QJ–TG cascade. TGP is a main contributing unit to the increase in power
402 generation. The power generation increase of TGP and GZB shows a negative correlation. This means
403 that GZB, as a reverse regulating reservoir of TGP, would contribute a certain water head for the
increase in power generation of TGP after joint operation of the five reservoirs. In certain situations,

RI
404
405 to meet the full load demand for power generation of TGP, GZB might have to abandon certain water
406 used for power generation to meet the requirements of flood control and ecological water demand in
407 the downstream reaches of the Yangtze River.

SC
408 Annual average power generation of the TG cascade would increase by 6.2 × 108 kWh, and that of the
409 QJ cascade would decrease by 0.5 × 108 kWh. This indicates that after joint operation of the TG and
410 QJ cascades, to guarantee the TG cascade’s output, the QJ cascade would lose certain power

U
411 generation benefits. However, the Pearson correlation coefficient of incremental power generation
412 between the TG and QJ cascades is only –0.18, which denotes extremely weak correlation. This
AN
413 indicates that under the current operational mode, these two cascades have no obvious regulating
414 relations, although the relation between the two cascades does show some reverse regulating
415 characteristics in certain years.
416 The Pearson correlation coefficient of the power generation increase between SBY and GHY is 0.63,
M

417 which indicates strong correlation. However, GBZ has no obvious correlation with SBY or GHY,
418 which reflects the randomness of the power generation operation of GBZ. The QJ cascade urgently
419 needs to strengthen the coordinated operation of the hydropower stations to improve the entire
D

420 cascade’s power generation benefits. In joint operation of the five reservoirs, the incremental power
421 generation of the cascade would be seen mainly the hydropower stations of the leading reservoirs, i.e.,
TE

422 the TGP reservoir in the TG cascade and the SBY reservoir in the QJ cascade. To a certain degree, the
423 other reservoirs would act as reverse regulating reservoirs of these leading reservoirs.

424 5.1.3 Analysis of water head change of all cascade hydropower stations in non-flood season
EP

425 As seen from the above section’s analysis, after joint operation of the five hydropower stations, the
426 power generation of TGP and SBY would increase and that of GHY, GBZ, and GZB would decrease.
427 Power generation is mainly increased through the mode of reservoir operation, i.e., the reduction of
C

428 possible abandoned water and an increase in the water head.


429 In any given year, there is variation in the guaranteed objectives and operational modes of the
AC

430 reservoirs in different periods. For example, during the flood season, the main objective of reservoir
431 operation is flood control; thus, operating the TG and QJ cascades to withstand large floods with
432 lower frequency would be the primary aim. In the non-flood season, the main objectives of reservoir
433 operation are meeting the social and economic demands for water, ecological water supplementation,
434 improving power generation benefits, and other basic demands, which require the coordination and
435 optimization of multiple objectives. Thus, reservoir operation in the non-flood season must be flexible,
436 and flexible operational measures could obviously improve the hydropower stations’ power
437 generation. This section discusses the reasons for the increase in power generation after joint
438 operation of the cascaded hydropower stations: its analysis focuses on the water head change of the
439 five reservoirs during the non-flood season and the reservoir operation processes in typical
440 hydrological years.
441 <Place Fig. 8 around here>

9
ACCEPTED MANUSCRIPT
442 The water heads of all the cascade hydropower stations in the non-flood season were analyzed
443 statistically, and the results of this analysis are shown in Fig. 8. In Fig. 8, “Actual” refers to the
444 average water head during the isolated operation of the cascades, “Simulated” refers to the water head
445 of the simulated joint operation of the hydropower stations in the non-flood season, and “Multiyear
446 average” refers to the multiyear average of water heads in the non-flood season. As Fig. 8 shows, after
447 joint operation of the TG and QJ cascades, the multiyear average water head of TGP and SBY in the
448 non-flood season would be, respectively, 2.32 and 9.02 m higher than before joint operation, whereas
449 that of GZB, GHY, and GBZ would be 0.07, 0.39, and 0.22 m lower, respectively. In general, the
450 changes in water head agree with the calculated results of the rise in power generation presented in
451 section 3.2.2. The water head of the leading reservoir would obviously increase, and the other

PT
452 reservoirs in the cascades would reverse regulate the hydropower stations of the leading reservoirs.
453 The water heads of the reverse regulating hydropower stations would reduce by certain degrees. To
454 further analyze the influence of joint operation of the cascades on water heads in the flood season, the

RI
455 standard deviations of water heads before and after joint operation of the cascades were calculated
456 (Table 3).
457 <Place Table 3 around here>

SC
458 Standard deviation is a parameter for measuring the degree of dispersion from the mean. A large
459 standard deviation is indicative of considerable difference between most values and the mean; a small
460 standard deviation implies that most values approximate the mean. In this analysis, the standard
461 deviation can be regarded as a stable indicator of water head change. The larger the standard deviation,

U
462 the greater the water head change in the non-flood season; thus, reducing the safety guarantee rate of
463 power generation. Conversely, a smaller standard deviation indicates that the interannual change in
AN
464 water head is more stable and, therefore, the risk of negative effects on safe power generation is lower.
465 As listed in Table 3, after joint operation of the TG and QJ cascades, the standard deviations of the
466 five reservoirs in the non-flood season reduce. This indicates that after joint operation of these two
467 cascades, the water head change in the non-flood season could be reduced by hydrological
M

468 compensation between the cascades to improve power generation stability. Furthermore, it is also
469 indicated that joint operation of the cascades would greatly influence the water heads of TGP, SBY,
470 and GBZ: their standard deviations would decrease from 1.92 to 1.80 m, 6.89 to 3.88 m, and 1.99 to
D

471 1.06 m, respectively.


472 Next, two aspects are analyzed in detail: changes in the water head of these three hydropower stations
TE

473 at different stages in the non-flood season, and the influence of the joint operational mode of the
474 cascade reservoir clusters on the increase in power generation. The non-flood season of the Yangtze
475 River basin can be divided into three stages: October to December, January to mid-April, and
476 mid-April to the end of May. In the first stage, when the water level is relatively stable, the TGP
EP

477 reservoir could operate at the specified high-water level. In the second stage (the dry season), the
478 water level of the TGP reservoir should be lowered to compensate the demands of socioeconomic
479 development and the environment downstream. In the third stage, the water level falls; therefore, the
C

480 reservoir’s water level will decrease rapidly. During this period, the reservoir should be emptied
481 rapidly to prepare for flood control in the flood season. The water head changes of TGP, SBY, and
AC

482 GBZ at different stages in the non-flood season are shown in Figs. 9–11.
483 <Place Fig. 9 around here>
484 <Place Fig. 10 around here>
485 <Place Fig. 11 around here>
486 Based on Figs. 9–11, the value of the water head change of the hydropower stations before and after
487 joint operation of the TG and QJ cascades can be calculated using the following formula:
488 Increased water head = Multiyear average of actual value – Multiyear average of simulated value
489 The calculated results show that TGP’s water head would increase in the first, second, and third stages
490 of the non-flood season by 0.05, 3.57, and 3.96 m, respectively. SBY’s water head would also increase
491 in the first, second, and third stages of the non-flood season by 6.99, 12.21, and 4.79 m, respectively.
492 The change in GHY’s water head would be inconsistent. In the first stage of the non-flood season, the

10
ACCEPTED MANUSCRIPT
493 water head would increase by 0.46 m; however, in the second and third stages, it would decrease by
494 0.57 and 1.79 m, respectively.
495 According to the above results, it is suggested that in the first stage of the non-flood season, the
496 compensation benefits from the QJ cascade to the TG cascade would not be obvious. After joint
497 operation, TGP’s water head would only increase by 0.05 m. From the analysis based on Table 2,
498 SBY’s water head in the first stage would increase by 6.99 m, which is approximately the standard
499 deviation of the annual average water head. This indicates that SBY would not compensate TGP.
500 Furthermore, in this stage, the water heads of both SBY and GHY would increase, indicating that
501 GHY would not compensate SBY. In general, in the first stage of the non-flood season, the QJ cascade

PT
502 would not compensate the TG cascade. This means the power generation of all the hydropower
503 stations in the QJ cascade would be uncoordinated. Thus, the two cascades and all hydropower
504 stations within these two cascades would be essentially isolated.
In the second stage of the non-flood season, the water head of TGP would increase by 3.57 m, which

RI
505
506 is obviously higher than the standard deviation of the annual average water head of 1.92 m (1.80 m
507 after joint operation). This indicates that, in this stage, active compensation would be developed
508 between the TG and QJ cascades. The water head of SBY would also increase considerably by 12.21

SC
509 m. At this time, the water head of GHY, as the reverse regulating hydropower station of SBY, would
510 reduce by 0.57 m, indicating that GHY would regulate SBY’s water head to increase SBY’s power
511 generation.

U
512 In the third stage of the non-flood season, TGP’s water head would increase by 3.96 m, and GHY’s
513 would increase by 4.79 m. This indicates that the TG and QJ cascades would develop active and
AN
514 effective compensation, and that the cascaded hydropower stations would operate in a coordinated
515 mode.

516 5.1.4 Analysis of TGP’s power generation operation process in typical hydrological years
M

517 After joint operation of the TG and QJ cascades, TGP’s annual average incremental power generation
518 would be 6.86 × 108 kWh, placing it first among the five hydropower stations of the QJ–TG cascade.
519 This section analyzes the power generation characteristics of TGP in different typical hydrological
D

520 years. Here, the years 1998, 1981, and 1994 were adopted as years typical of high, normal, and low
521 flow, respectively, in the Yangtze River basin. Changes in the power generation parameters before and
TE

522 after joint operation were analyzed to improve understanding of the stage characteristics of TGP’s
523 incremental power generation. Tables 4–6 list TGP’s power generation parameters before and after
524 joint operation.
525 <Place Table 4 around here>
EP

526 <Place Table 5 around here>


527 <Place Table 6 around here>
C

528 Analysis of the data in Tables 4–6 reveals the following. First, the changes in TGP’s water head and
529 discharge would commence in January. Second, from January to mid-April, after joint operation of the
AC

530 five reservoirs, TGP’s discharge would be less than under individual operation, corresponding to
531 decreased power generation. During this period, water level upstream of the reservoir would be high.
532 Third, considering the changes in discharge and output during the 10 days in mid-February and in
533 March, the output would be higher because a higher water head is accumulated during January to
534 mid-April[1]. In general, the benefits of joint operation would manifest mainly in the falling stage.
535 Because water quantity and water head in the previous period would be high, the TGP’s water head
536 would decrease rapidly in the falling stage. Both discharge and water head would be significantly
537 higher than under individual operation.

[1] In China, the reservoir operations shall be planed periodically, and are required to submitted the
plans to River basin Administrative Agency ten days ahead of the actual operation.

11
ACCEPTED MANUSCRIPT
538 5.1.5 Analysis of influence on guarantee rate of power generation
539 According to the requirements of the State Power Grid, the authorized output of TGP and GZB is
540 4990 and 768 MW, respectively, and the total guaranteed output of the TG and QJ cascades should be
541 no less than 6386.8 and 628.8 MW, respectively. After joint operation of the TG and QJ cascades, the
542 total guaranteed output of the five hydropower stations should be no less than 6386.8 MW. This
543 section presents statistical data of power generation output from the five hydropower stations before
544 and after joint operation. Safe power generation probability is calculated based on the State Power
545 Grid’s requirements for the cascaded power stations’ output. Results of the calculations and
546 evaluations of the safe power generation guarantee rates for joint and individual cascade operation are

PT
547 shown in Fig. 12.
548 <Place Fig. 12 around here>
549 In Fig. 12, “Annual” refers to the calculated result of the annual guarantee rate for a given operational

RI
550 mode, and “Overall” refers to the multiyear average guarantee rate for a given operational mode. As
551 shown in the figure, after joint operation of the TG and QJ cascades, the total guarantee rate of power
552 generation of the five hydropower stations would reach up to 99.9%, which is higher than the total

SC
553 guarantee rate of the TG cascade (98.2%) and far higher than that of the QJ cascade (71.9%). The
554 annual guarantee rate would never fall below 90% after joint operation. Under individual operation of
555 the QJ cascade, the annual guarantee rate does fall below 50% at times. Therefore, it can be concluded
556 that the safety and reliability of the power system would be improved considerably by joint operation

U
557 of the two cascades. AN
558 5.2 Assessment results of the optimal operational method

559 5.2.1 Results calculated with maximum power generation model


560 On the premise of guaranteeing the firm output, the maximum power generation mentioned in this
M

561 paper refers to the maximum multiyear average power generation of a single cascade or the jointly
562 operated cascades. To ensure comparability of the results in this section with those of the routine
563 operation discussed in section 4.1, the research benchmark mentioned in section 5.1 was adopted here
D

564 (i.e., meeting the State Power Grid’s minimum requirements for guaranteed output). The minimum
565 output of the two cascades (6658.8 MW) was adopted in calculating the power generation of joint
TE

566 operation. For single cascade operation, the minimum guaranteed outputs of 6386.8 MW and 628.8
567 MW were adopted for calculating the power generation of the TG and QJ cascades, respectively. The
568 calculation results are shown in Fig. 13.
569 <Place Fig. 13 around here>
EP

570 In Fig. 13, “Isolated” represents the calculation result of power generation in isolated operation, and
571 “Incremental” refers to the incremental power generation after joint operation compared with isolated
572 operation. As shown in Fig. 13(a), when the two cascades are in joint operation at minimum output
C

573 (6658.8 MW), the total power generation of the two cascades would be 1104.16 × 108 kWh, which is
574 2.63 × 108 kWh higher than under isolated operation. The total power generation in the flood season
AC

575 would decrease by 1.53 × 108 kWh, while that in the non-flood season would increase by 4.16 × 108
576 kWh (Fig. 13(b)). Fig. 13(d) shows that after joint operation, the total power generation of the QJ
577 cascade in the flood season would decrease by 1.56 × 108 kWh, while that of the TG cascade would
578 remain unchanged (Fig. 13(c)). During the non-flood season, the total power generation of the QJ and
579 TG cascades would increase by 2.20 × 108 and 1.96 × 108 kWh, respectively (Fig. 13(c) and (d)).
580 After analyzing the results in terms of reservoir operation processes, the reason for the TG cascade’s
581 unchanged power generation during the flood season was found to be flood control operation
582 management of the Yangtze River. According to the current operation regulation of the TG reservoir,
583 its water level from June to September in the flood season should be strictly controlled at 145 m or
584 lower. During this period, TGP cannot increase the water head for power generation. Therefore, even
585 if the optimal operation algorithm were adopted, there would be little room for improvement in TGP’s
586 power generation.

12
ACCEPTED MANUSCRIPT
587 The Three Gorges hydropower station undertakes primary responsibility for flood control of the
588 Yangtze River. By the end of July, after the main flood season, the hydropower stations of the QJ
589 cascade are not restricted by the water level limit for flood control. Their reservoirs can operate with
590 high water levels, providing the possibility of increasing power generation output during the
591 non-flood season. In other words, the increased power generation of the QJ cascade during the dry
592 season is caused by its sharing of flood control responsibility with the TG cascade after joint
593 operation. Incremental power generation before and after joint operation of the TG and QJ cascades is
594 shown in Fig. 14.
595 <Place Fig. 14 around here>

PT
596 Fig. 14 shows the assessment results of the potential for increased power generation after joint
597 operation from 1951–2009. In the figure, “TG cascade” refers to the total incremental power
598 generation of TGP and GZB after joint operation, and “QJ cascade” refers to the total incremental
power generation of SBY, GHY, and GBZ after joint operation. “In total” refers to the total

RI
599
600 incremental power generation of the five hydropower stations in these two cascades after joint
601 operation. The multiyear average incremental power generation of the TG–QJ cascade would be 2.65
602 × 108 kWh (TG cascade: 2.24 × 108 kWh; QJ cascade: 0.41 × 108 kWh). The incremental power

SC
603 generation of the TG cascade accounts for 84.5% of the total of the combined TG–QJ cascade.
604 5.2.2 Results calculated with maximum power generation benefit model
605 An assessment model formulated to calculate the power generation benefits of the TG cascade, QJ

U
606 cascade, and the TG–QJ cascade under joint operation. The same assumption (meeting the guaranteed
607 output of hydropower stations) of the maximum power generation model in section 4.2.1 was used for
AN
608 this section’s maximum power generation benefit model. The maximum power generation is the
609 maximum multiyear average power generation benefit of the cascaded hydropower station system.
610 The calculation results of the power generation benefits are shown in Fig. 15.
M

611 <Place Fig. 15 around here>


612 Under joint operation of the two cascades at minimum output (6658.8 MW), the total power
613 generation benefits of the two cascades would be CNY 28.759 billion (USD 4.312 billion), which is
D

614 approximately CNY 72 million (USD 10.8 million USD) higher than under isolated operation[2]. In
615 the flood season, the value would be reduced by approximately CNY 63 million (USD 9.4 million),
TE

616 whereas it would be increased by approximately CNY 134 million (USD 20.1 million) in the
617 non-flood season (Fig. 15(b)). As shown in Fig. 15(c), after joint operation, the flood season power
618 generation benefits of the TG cascade would remain unchanged, but those of the QJ cascade would
619 decrease by CNY 62 million (USD 9.3 million) (Fig. 15(d)). During the non-flood season, the
EP

620 incremental power generation benefits of the TG and QJ cascades would reach CNY 54 million (USD
621 8.1 million) and CNY 81 million (USD 12.1 million), respectively. Combined with the analysis of
622 power generation in section 5.2.1, it is suggested that under joint operation, partial power generation
623 benefits move from the QJ cascade in the flood season to the TG and QJ cascades in the non-flood
C

624 season. On the whole, the potential power generation benefits of the two cascades would be realized
625 mainly during the non-flood season.
AC

626 5.3 Generation potential calculation and results analysis


627 To assess the power generation benefits of the TG and QJ cascades’ joint operation, this study
628 considered the hydropower stations’ power generation before and after joint operation. This was
629 accomplished using historical runoff data, together with routine and optimal operational methods. The
630 routine operational method is that currently adopted by the TG and QJ cascades for power generation.
631 The optimal operational method was based on the DP algorithm. The calculation results of power
632 generation are summarized in Table 7:
633 Routine reservoir operation method: When the TG and QJ cascades operate in isolation, their

[2]The USD to CNY exchange rate ranges from 6 to 7, and the current exchange rate of 6.67 is used
in this paper.

13
ACCEPTED MANUSCRIPT
634 multiyear average power generation is 977.3 × 108 and 82.0 × 108 kWh, respectively (total: 1059.3 ×
635 108 kWh). If the TG and QJ cascades were operated jointly, their total multiyear average power
636 generation would be 983.4 × 108 and 81.6 × 108 kWh, respectively (total: 1065.0 × 108 kWh). Thus,
637 the power generation of the jointly operated TG and QJ cascades would increase by 5.73 × 108 kWh
638 (5‰) compared with their isolated operation.
639 Optimal reservoir operation method: When the TG and QJ cascades operate in isolation, their
640 multiyear average power generation is 1015.36 × 108 and 86.17 × 108 kWh, respectively (total:
641 1101.53 × 108 kWh). If operated jointly, the total multiyear average power generation of the TG and
642 QJ cascades would be 1017.59 × 108 and 86.57 × 108 kWh, respectively (total: 1104.16 × 108 kWh).
Thus, the power generation of the jointly operated TG and QJ cascades would increase by 2.63 × 108

PT
643
644 kWh (2.4‰) compared with their isolated operation.
645 Comparison of the two reservoir operation methods: If reservoir operation were based on the
reservoir operating curves, the incremental power generation of the TG-QJ cascade would be 42.24 ×

RI
646
647 108 kWh, which is seven times that of the joint operation without changing the reservoir operation
648 mode (5.73 × 108 kWh). After joint operation, if reservoir operation were based on the reservoir
649 operating curves obtained from calculations using the DP optimization algorithm, the TG-QJ

SC
650 cascade’s incremental power generation would be 39.14 × 108 kWh, which is 15 times that of the joint
651 operation without changing the reservoir operation mode (2.63 × 108 kWh).
652 Comparison of the calculation results indicated that the operational mode of the reservoirs has

U
653 considerable impact on the hydropower stations’ power generation. Although a coordinated mode of
654 power generation between the cascades could increase power generation to a certain degree, the ratio
AN
655 of increase would not be large. As listed in Table 7, under the routine operational mode, the
656 incremental power generation of joint operation would be 5.73 × 108 kWh, an increase of 5‰
657 compared to isolated operation. Under an optimal operational mode, the incremental power generation
658 of joint operation would be 2.63 × 108 kWh, an increase of 2.4‰ compared to isolated operation.
M

659 Therefore, although joint operation of the two cascades could increase the total power generation,
660 changing the existing mode of reservoir operation could result in a greater increase in power
661 generation. Even if the cascades operate in isolation, if reservoir operation is optimized through the
DP algorithm, the incremental power generation of the cascades would be 42.24 × 108 kWh, which is
D

662
663 seven times that of the joint operation without changing the reservoir operation mode (5.73 × 108
664 kWh). After joint operation, if reservoir operation is optimized using the DP algorithm, the
TE

665 incremental power generation of the TG–QJ cascade would be 39.14 × 108 kWh, which is 15 times
666 that of the joint operation without changing the reservoir operation mode (2.63 × 108 kWh).
667 <Place Table 7 around here>
EP

668 6 Conclusions
669 In general, active change to the current reservoir operation mode would increase the power generation
670 of the cascaded hydropower station clusters more than the development of a joint operational mode.
C

671 Furthermore, the potential space in the storage and falling periods of the five reservoirs’ joint
672 operation is much larger in comparison with isolated operation of the TG and QJ cascades. However,
AC

673 the actual annual power generation of the TG and QJ cascades has been at a high level. Under the
674 restrictions of the existing operational regulations, the power generation benefits of jointly operating
675 the TG and QJ cascades offer certain, but not large, potential for improvement. Therefore, this study
676 recommends that the current reservoir operation mode should be changed.
677 In fact, the reservoir storage capacity and runoff regulation capability of the five TG–QJ cascade
678 reservoirs are quite different. Joint operation of the two cascades could provide a regulatory role in
679 “storage capacity compensation” and “hydrological compensation.” Therefore, power generation
680 could increase to a certain degree. However, the TG and QJ cascades are geographically close, being
681 both located within the Yangtze River basin, which means they have similar hydrographic
682 characteristics and that their degree of complementation is not great. In addition to power generation,
683 these two cascades also have roles in flood control, water supply, ecological protection, shipping, and
684 other tasks. The current reservoir operational mode safely determines the water level control line of

14
ACCEPTED MANUSCRIPT
685 the TG reservoirs according to the pertaining regulation objectives, which restricts flexibility. Thus, it
686 is suggested even if these two cascades were operated jointly, without changing the existing reservoir
687 regulation mode, the percentage increase in power generation would be small. Nevertheless, since the
688 TG–QJ cascade is very large, even if the operational modes of cascade reservoirs were not changed,
689 the absolute value of the incremental power generation (5.73 × 108 kWh annually) would be
690 considerable. Therefore, this research supports the conclusion that “joint operation of the two
691 cascades is necessary.”
692 The Chinese government has decided to intensify the development of hydropower as an important
693 source of renewable energy. It has proposed to build 13 national cascade hydropower bases before

PT
694 2050, including 366 hydropower stations with installed capacity of up to 94 million kW. However, to
695 ensure successful realization of future cascade development, China must urgently consider the water–
696 energy nexus of hydropower generation from various perspectives, e.g., flood control, shipping,
697 ecology, and water supply. The operation of cascaded hydropower stations is quite different from that

RI
698 of isolated hydropower stations in non-cascaded developments. The operation of cascaded
699 hydropower stations should meet the safety requirements of the reservoir area, dams, flood discharge,
700 power units, and shipping. It should also provide comprehensive benefits to flood control, power

SC
701 generation, water supply, shipping, and ecological protection. This means that reservoir hydropower
702 stations in cascade are restricted in their operation by flood control, shipping, power generation, and
703 other conditions. The situation is not one in which the greater the number of installed cascaded
704 hydropower stations and the larger the scale of the cluster (i.e., the higher the installed capacity), the

U
705 larger the actual power generation of each cascaded station. Although the above factors are directly
706 related to the increase in actual power generation, there is no necessary connection. Thus, it is
AN
707 suggested that in planning hydropower development, considered research and analysis must be
708 conducted.
709 Acknowledgements
M

710 This work was supported by the National Natural Science Foundation of China (Grant Nos.: 51579248 and
711 51379219).
712 References
D

[1]Baños R, Manzano-Agugliaro F, Montoya FG, Gil C, Alcayde A, Gómez J. Optimization methods


TE

applied to renewable and sustainable energy: A review. Renewable and Sustainable Energy Reviews,
2011, 15 (4): 1753-1766.
[2]Fang Y P, Deng W. The critical scale and section management of cascade hydropower exploitation in
Southwestern China. Energy, 2011, 36(10):5944-5953.
EP

[3]Tauxe G W, Inman R R, Mades D M. Multiple objectives in reservoir operation. Journal of the Water
Resources Planning and Management Division, 1980, 106 (1): 225-238.
[4]Loucks D P, Van Beek E, Stedinger J R, Dijkman J P M, Villars M T. Water resources systems planning
and management: an introduction to methods, models and applications. Paris: UNESCO, 2005.
C

[5]Tejada-Guibert J A, Stedinger J R, Staschus K. Optimization of value of CVP's hydropower production.


Journal of Water Resources Planning and Management, 1990, 116 (1): 52-70.
AC

[6]Yeh W W G. Reservoir management and operations models: A state-of-the-art review. Water Resources
Research, 1985, 21 (12): 1797-1818.
[ 7 ]Choong S M, Shafie A E. State-of-the-Art for Modelling Reservoir Inflows and Management
Optimization. Water Resources Management, 2015, 29(4):1267-1282.
[8]Cheng C T, Shen J J, Wu X Y, Chau K W. Operation challenges for fast-growing China's hydropower
systems and respondence to energy saving and emission reduction. Renewable and Sustainable Energy
Reviews, 2012, 16 (5): 2386-2393.
[9]Li F F, Shoemaker C A, Wei J H, Fu X D. Estimating maximal annual energy given heterogeneous
hydropower generating units with application to the Three Gorges system. Journal of Water Resources
Planning and Management, 2013, 139 (3): 265-276.
[10]Trott W J and Yeh W WG. Optimization of multiple reservoir system. Journal of the Hydraulics
Division, 1973, 99 (10): 1865-1884.
[11]Giuliani, M., Castelletti, A., Pianosi, F., Mason, E., and Reed, P. Curses, tradeoffs, and scalable
Management: Advancing evolutionary multiobjective direct policy search to improve water reservoir

15
ACCEPTED MANUSCRIPT

operations. Journal of Water Resources Planning and Management, 2016, 14(2): 04015050.
[12]Chang X L, Liu X H and Zhou W. Hydropower in China at present and its further development.
Energy, 2010, 35 (11): 4400-4406.
[13]Li F F, Wei J H, Fu X D, Wan X Y. An effective approach to long-term optimal operation of large-scale
reservoir systems: Case study of the Three Gorges system. Water Resources Management, 2012, 26
(14): 4073-4090.
[14]Jiang Z Q, Ji C M, Sun P, Wang L P, Zhang W K. Total output operation chart optimization of cascade
reservoirs and its application. Energy Conversion and Management, 2014, 88:296-306.
15] Guo S L, Chen J H, Liu P, Li T Y. Joint Operation of the Multi-Reservoir System of the Three Gorges

PT
and the Qingjiang Cascade Reservoirs. Energies, 2011,4(7), 1036-1050.
[16]Ma C, Lian J J, Wang J N. Short-term optimal operation of Three-gorge and Gezhouba cascade
hydropower stations in non-flood season with operation rules from data mining. Energy Conversion &
Management, 2013, 65:616-627.

RI
[17]Zheng Y, Fu X D, Wei J H. Evaluation of power generation efficiency of cascade hydropower plants: A
case study. Energies, 2013, 6(2):1165-1177.
[18]Mateo C, Hanasaki N, Komori D, Tanaka K, Kiguchi M, Champathong A, Sukhapunnaphan T,
Yamazaki D, and Oki T. Assessing the impacts of reservoir operation to floodplain inundation by

SC
combining hydrological, reservoir management, and hydrodynamic models. Water Resources Research,
2014, 50(9):7245-7266.
[19]Yeh W W G, Becker L. Multiobjective analysis of multireservoir operations. Water Resources
Research, 1982, 18 (5): 1326-1336.

U
[20]Chen J H, G S L, Li Y, Liu P, Zhou Y L. Joint operation and dynamic control of flood limiting water
levels for cascade reservoirs. Water Resources Management, 2013, 27(3):749-763.
AN
[21]Zhou Y L, Guo S L, Liu P, Xu C Y. Joint operation and dynamic control of flood limiting water levels
for mixed cascade reservoir systems. Journal of Hydrology, 2014, 519(3):248-257.
[22]Tsai W P, Chang F J, Chang L C, Herricks E E. AI techniques for optimizing multi-objective reservoir
operation upon human and riverine ecosystem demands. Journal of Hydrology, 2015, 530:634-644.
M

[23]Mateo C, Hanasaki N, Komori D, Tanaka K, Kiguchi M, Champathong A, Sukhapunnaphan T,


Yamazaki D, and Oki T. Assessing the impacts of reservoir operation to floodplain inundation by
combining hydrological, reservoir management, and hydrodynamic models. Water Resources Research,
D

2014, 50(9):7245-7266.
[24]Needham J T, Watkins Jr D W, Lund J R, Nanda S K. Linear programming for flood control in the
Iowa and Des Moines rivers. Journal of Water Resources Planning and Management, 2000, 126 (3):
TE

118-127.
[25]Chu W S, Yeh W W G and Rossman L A. A nonlinear programming algorithm for real-time hourly
reservoir operations. JAWRA Journal of the American Water Resources Association, 1978, 14 (5):
1048-1063.
EP

[26] Hall W A, Butcher W S, Esogbue A. Optimization of the operation of a multiple‐purpose reservoir


by dynamic programming. Water Resources Research, 1968, 4 (3): 471-477.
[27]Nicklow J, Reed P, Savic D, Dessalegne T, Harrell L, Chan-Hilton A, Karamouz M, Minsker B,
Ostfeld A, Singh A, Zechman E. State of the art for genetic algorithms and beyond in water resources
C

planning and management. Journal of Water Resources Planning and Management, 2009, 136 (4):
412-432.
AC

[28]Jain S K, Das A and Srivastava D K. Application of ANN for reservoir inflow prediction and operation.
Journal of Water Resources Planning and Management, 1999, 125 (5): 263-271.
[29]Tu M Y, Hsu N S and Yeh WWG. Optimization of reservoir management and operation with hedging
rules. Journal of Water Resources Planning and Management, 2003, 129 (2): 86-97.
[30]Li X, Li T J, Wei J H, Wang G, Yeh W G. Hydro unit commitment via mixed integer linear
programming: A case study of the Three Gorges project, China. IEEE Transactions on Power Systems,
2014, 29(3): 1232-1241.
[31]Barros M T L, Tsai F T, Yang S, Lopes JEG, Yeh W W. Optimization of large-scale hydropower system
operations. Journal of Water Resources Planning and Management, 2003, 129 (3): 178-188.
[ 32 ]Arunkumar R, Jothiprakash V. Optimal reservoir operation for hydropower generation using
Non-linear programming model. Journal of The Institution of Engineers (India): Series A, 2012,
93(2):111-120.
[33]Yoo J H. Maximization of hydropower generation through the application of a linear programming

16
ACCEPTED MANUSCRIPT

model. Journal of Hydrology. 2009, 376, (1-2):182-187.


[34]Luenberger D V, Ye Y Y. Linear and Nonlinear Programming (Third Edition). International Series in
Operations Research & Management Science, Springer, 2008.
[35]Zambon R C, Barros M T L, Lopes J E G, Barbosa P S F, Francato A L, Yeh W W G. Optimization of
large-scale hydrothermal system operation. Journal of Water Resources Planning and Management,
2011, 138 (2): 135-143.
[36]Cai X M, McKinney D C, Lasdon L S. Solving nonlinear water management models using a combined
genetic algorithm and linear programming approach. Advances in Water Resources, 2001,
24(6):667-676.

PT
[37]Catalão J P S, Mariano S J P S, Mendes V M F, Ferreira L A F M. Parameterization effect on the
behavior of a head-dependent hydro chain using a nonlinear model. Electric Power Systems Research,
2006, 76 (6): 404-412.
[38]Grygier J C, Stedinger J R. Algorithms for Optimizing Hydropower System Operation. Water

RI
Resources Research. 1985, 21(1):1-10.
[39]Han S P. A globally convergent method for nonlinear programming. Journal of Optimization Theory
and Applications, 1977, 22(3):297-309.
[40]Nemhauser G L. Introduction to dynamic programming. New York: John Wiley, 1966.

SC
[41]Zhao T T G, Cai X M, Lei X H, et al. Improved dynamic programming for reservoir operation
optimization with a concave objective function. Journal of Water Resources Planning and Management,
2012, 138 (6): 590-596.
[42]Chandramouli V and Raman H. Multireservoir modeling with dynamic programming and neural

U
networks. Journal of Water Resources Planning and Management, 2001, 127 (2): 89-98
[43]Bellman R. Adaptive control processes: a guided tour. Princeton, NJ: Princeton University Press, 1961.
AN
[44]Labadie J W. Optimal operation of multireservoir systems: State-of-the-art review. Journal of Water
Resources Planning and Management, 2004, 130 (2): 93-111.
[45]Larson RE. State increment dynamic programming. New York: Elsevier Science, 1968.
[46]Larson RE and Korsak AJ. A dynamic programming successive approximations technique with
M

convergence proofs. Automatica, 1970, 6 (2): 245-252.


[47]Heidari M, Chow V T, Kokotović P V, Meredith D D. Discrete differential dynamic programing
approach to water resources systems optimization. Water Resources Research, 1971, 7 (2): 273-282.
D

[48]Howson H R and Sancho N G F. A new algorithm for the solution of multi-state dynamic programming
problems. Mathematical Programming, 1975, 8(1):104-116.
[49]Delipetrev B, Jonoski A, Solomatine D P. A novel nested dynamic programming (nDP) algorithm for
TE

multipurpose reservoir optimization. Journal of Hydroinformatics, 2015, 17(4):570-583.


[50]Haddad O B, Hosseini-Moghari S M, Loáiciga H A. Biogeography-Based Optimization Algorithm for
Optimal Operation of Reservoir Systems. Journal of Water Resources Planning and Management, 2016,
142(1): 04015034.
EP

[51]Li X, Wei J H, Li T J, Wang G Q. A parallel dynamic programming algorithm for multi-reservoir


system optimization. Advances in Water Resources, 2014, 67(4):1-15.
[52]Mousavi S J and Karamouz M. Computational improvement for dynamic programming models by
diagnosing infeasible storage combinations. Advances in Water Resources, 2003, 26 (8): 851-859.
C

[53]Maxwell R M. A terrain-following grid transform and preconditioner for parallel, large-scale,


integrated hydrologic modeling. Advances in Water Resources, 2013, 53:109-117.
AC

[54]Simonovic S P. Reservoir systems analysis: closing gap between theory and practice. Journal of Water
Resources Planning and Management, 1992, 118 (3): 262-280.
[55]Kollet S J, Maxwell R M, Woodward C S, Smith S, Vanderborght J, Vereecken H, Simmer C. Proof of
concept of regional scale hydrologic simulations at hydrologic resolution utilizing massively parallel
computer resources. Water Resources Research, 2010, 46 (4):475-478.
[56]De Farias D P, Van Roy B. The linear programming approach to approximate dynamic programming.
Operations research. 2013, 51(6): 850-865.
[57]Maxwell R M. A terrain-following grid transform and preconditioner for parallel, large-scale,
integrated hydrologic modeling. Advances in Water Resources, 2013, 53:109-117.
[58]Li X, Wei J H, Fu X D, Li T J, Wang G Q. Knowledge-Based Approach for Reservoir System
Optimization. Journal of Water Resources Planning and Management. 2014, 140(6):04014001
[59]Hinneburg A, Keim D A. Optimal Grid-Clustering: Towards Breaking the Curse of Dimensionality in
High-Dimensional Clustering. Proceedings of the 25th International conference on Very Large

17
ACCEPTED MANUSCRIPT

Databases, 1999, 506-517.


[ 60 ]Zhang Y K, Jiang Z Q, Ji C M, Sun P. Contrastive analysis of three parallel modes in
multi-dimensional dynamic programming and its application in cascade reservoirs operation. Journal of
Hydrology, 2015, 529 (part A) 22-34.
[61]Huang H, Yan Z. Present situation and future prospect of hydropower in China. Renewable and
Sustainable Energy Reviews, 2009, 13 (6): 1652-1656.
[62]Oliveira R and Loucks D P. Operating rules for multi-reservoir systems. Water Resources Research,
1997, 33 (4): 839-852.
[63]Shang Y Z, Guo Y X, Shang L, Ye Y T, Liu R H, Wang G Q. Processing conversion and parallel

PT
control platform: a parallel approach to serial hydrodynamic simulators for complex hydrodynamic
simulations. Journal of Hydrodynamics, 2016, 18(5):851-866.

RI
U SC
AN
M
D
TE
C EP
AC

18
ACCEPTED MANUSCRIPT

Figure Captions:

Fig. 1 Location of TG-QJ cascaded power stations.


Fig. 2. Designed operating rule curves of the Three Gorges reservoir.
Fig. 3. Technical line and work flow: (a) Formation process of actual reservoir operating rule curves
and (b) Calculation flow chart.
Fig. 4. Comparison of multiyear average power generation before and after joint operation of the TG
and QJ cascades (1951–2009).
Fig. 5. Comparison of actual power generation of all cascade hydropower stations and simulated

PT
power generation after joint cascade operation (1951–2009).
Fig. 6. Incremental power generation of TG cascade. (a) Column chart of incremental power
generation of all hydropower stations in TG cascade; (b) Area graph of incremental power generation
of TG cascade.

RI
Fig. 7. Incremental power generation of QJ cascade: (a) Column chart of incremental power
generation of all hydropower stations in QJ cascade; (b) Area graph of incremental power generation
of QJ cascade

SC
Fig. 8. Comparison of average water heads of all cascade hydropower stations in non-flood season
and average water heads of simulated cascade joint operation in non-flood season.
Fig. 9. Water head change comparison for main cascade hydropower stations in the first stage of
non-flood seasons, before and after joint cascade operation.

U
Fig. 10. Water head change comparison for main cascade hydropower stations in the second stage of
non-flood seasons, before and after joint cascade operation.
AN
Fig. 11 Water head change comparison for main cascade hydropower stations in the third stage of
non-flood seasons, before and after joint cascade operation.
Fig. 12. Guarantee rate of safe power generation of TG cascade, QJ cascade, and jointly operated
TG-QJ cascade.
M

Fig. 13 Comparison of multiyear average power generation before and after joint operation of TG and
QJ cascades (1951–2009).
Fig. 14 Incremental power generation before and after joint operation of TG and QJ cascades.
D

Fig. 15. Comparison of multiyear average power generation benefits before and after joint operation
of TG and QJ cascades (1951–2009).
TE

Table Captions:

Table 1 Reservoir regulation characteristics of cascade hydropower stations


Table 2 Pearson correlation coefficients of power generation increase of all cascade hydropower
EP

stations.
Table 3 Standard deviation of annual average water head of all cascade hydropower stations before
and after joint operation in flood season.
C

Table 4 Power generation parameters of TGP before and after joint operation of the TG and QJ
cascades in a typical high-flow year (1998).
Table 5 Power generation parameters of TGP before and after joint operation of the TG and QJ
AC

cascades in a typical normal-flow year (1981).


Table 6 Power generation parameters of TGP before and after joint operation of the TG and QJ
cascades in a typical low-flow year (1994).
Table 7 Summary of calculation results of power generation (×108kW·h)

19
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC

Fig. 1 Location of TG-QJ cascaded power stations.

20
ACCEPTED MANUSCRIPT

175
Upper boundary curve
Reservoir storage level (m)

165
Zone(II)
Zone(III)
155

PT
Lower boundary curve

RI
145
Month Oct
UBC 175

SC
LBC 145
(b)
175

U
II II
Reservoir storage level (m)

III
AN
165
M

155
IV
D
TE

145 II
Month Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
UBC 145 145 145 145 175 175 175 175 175 175 175 155
EP

LBC 145 145 145 145 145 156.3 169.6 166.6 160.9 155 155 145
(a)

Fig. 2. Designed operating rule curves of the Three Gorges reservoir.


C
AC

21
ACCEPTED MANUSCRIPT

Start Start

cascade
cascade
TG-QJ

TG
QJ
Cascade

PT
Weather forecast Hydrological monitoring
Reservoir discharge=Const Reservoir discharge=Const

Flood defense

Water supply

Eco-system
Navigation
Reservoirs compensation
priority determination

RI
……

Basin rainfall Reservoirs inflow


Adjusting given reservoirs Adjusting given power Adjusting given reservoirs
discharge? generation? discharge?

SC
Integrated decision - Cascaded reservoirs state
making (now and future) Equal discharge runoff Equal discharge runoff
Power generation regulation
regulation regulation

U
Reservoirs discharge (Q) &Water level (Z)
Actual operating
rule curves

AN
Power grid SBY-GHY-GBZ reservoirs &
requirement SBY-GHY-GBZ reservoirs TG-GZB reservoirs
TG-GZB reservoirs
Power generation plan Interactive Facilities operation plan

M
No Yes No
No Yes Satisfied the requirement of
Satisfied the requirement of Satisfied the requirement of
Generator units Sluice & gate reservoirs rule curves? power grid? reservoirs rule curves?

D
Yes Yes Yes

TE
Power generation Power generation Power generation for
for QJ cascade for TG and QJ cascades TG cascade
Reservoirs storage
EP
End
End
C

(a) Formation process of actual reservoir operating rule curves. (b) Calculation flow chart.
AC

Fig. 3. Technical line and work flow: (a) Formation process of actual reservoir operating rule curves and (b) Calculation flow chart.

22
ACCEPTED MANUSCRIPT

PT
(a) TG-QJ cascade

RI
U SC
(b) TG Cascade
AN
M
D
TE

(c) QJ Cascade
Fig. 4. Comparison of multiyear average power generation before and after joint operation of the TG
and QJ cascades (1951–2009).
C EP
AC

23
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
(a) TGP (b) GZB

M
D
TE
C EP
AC

(c) SBY (d) GHY (e) GBZ

Fig. 5. Comparison of actual power generation of all cascade hydropower stations and simulated power generation after joint cascade operation (1951–2009).

24
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
(a) (b)

Fig. 6. Incremental power generation of TG cascade. (a) Column chart of incremental power generation of all hydropower stations in TG cascade; (b) Area

M
graph of incremental power generation of TG cascade.

D
TE
C EP
AC

(a) (b)

Fig. 7. Incremental power generation of QJ cascade: (a) Column chart of incremental power generation of all hydropower stations in QJ cascade; (b) Area
graph of incremental power generation of QJ cascade

25
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
(a) TGP (b) GZB

M
D
TE
C EP
AC

(c) SBY (d) GHY (e) GBZ

Fig. 8. Comparison of average water heads of all cascade hydropower stations in non-flood season and average water heads of simulated cascade joint
operation in non-flood season.

26
ACCEPTED MANUSCRIPT

PT
RI
U SC
(a) TGP (b) SBY (c) GHY

AN
Fig. 9. Water head change comparison for main cascade hydropower stations in the first stage of non-flood seasons, before and after joint cascade operation.

M
D
TE
C EP
AC

(a) TGP (b) SBY (c) GHY

Fig. 10. Water head change comparison for main cascade hydropower stations in the second stage of non-flood seasons, before and after joint cascade
operation.

27
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
(a) TGP (b) SBY (c) GHY

Fig. 11 Water head change comparison for main cascade hydropower stations in the third stage of non-flood seasons, before and after joint cascade operation.

M
D
TE
C EP
AC

(a) TG cascade (b) QJ cascade (c) TG cascade and QJ cascade

Fig. 12. Guarantee rate of safe power generation of TG cascade, QJ cascade, and jointly operated TG-QJ cascade.

28
ACCEPTED MANUSCRIPT

(a) TG-QJ cascade

PT
RI
(b) TG-QJ cascade

U SC
AN
(c) TG cascade
M
D
TE

(d) QJ cascade
EP

Fig. 13 Comparison of multiyear average power generation before and after joint operation of TG and QJ cascades
(1951–2009).
C
AC

Fig. 14 Incremental power generation before and after joint operation of TG and QJ cascades.
29
ACCEPTED MANUSCRIPT

PT
(a) TG-QJ cascade

RI
U SC
AN
(b) TG-QJ cascade
M
D
TE

(c) TG cascade
C EP
AC

(d) QJ cascade
Fig. 15. Comparison of multiyear average power generation benefits before and after joint operation
of TG and QJ cascades (1951–2009).

30
ACCEPTED MANUSCRIPT

Table 1
Reservoir regulation characteristics of cascade hydropower stations

Cascade Name of hydropower Total reservoir storage Utilizable capacity Coefficient of reservoir Hydropower station type c

PT
station capacity (×108m3) (×108m3) a storage capacity (β)b
393 165 3.66%
TG Three Gorges (TG) Seasonally regulating

RI
cascade hydropower station
7.11 0.848 0.02%

SC
Gezhouba (GZB) Runoff hydropower station
45.89 23.83 25.2%
QJ Shuibuya (SBY) Multiyear regulating

U
cascade hydropower station

AN
33.3 19.41 18%
Geheyan (GHY) Yearly regulating
hydropower station

M
4.85 0.54 0.39%
Gaobazhou (GBZ) Daily regulating hydropower
station

D
a. Utilizable capacity: volume of reservoir from normal storage level to dead water level.

TE
b. Coefficient of reservoir storage capacity β: proportion of utilizable capacity to multiyear average inflow quantity.
c. Based on reservoir regulation performance, the hydropower stations can be divided into daily regulating, seasonally regulating, yearly regulating, and
EP
multiyear regulating hydropower stations. Daily regulating: reservoir has certain regulating reservoir storage capacity; runoff in one day and night is
reallocated, i.e., the regulation cycle is 24 hours. Yearly regulating: regulating reservoir storage capacity can reallocate the reservoir inflow in one year
according to social water demand. Seasonally regulating: hydropower station reservoir has certain yearly regulation capacity and can impound excess water in
C

the flood season. When flooding occurs, the reservoir should abandon water and discharge the flood. Multiyear regulating: the reservoir storage capacity is
large enough to impound excess water accumulated in the reservoir over many years. Water is supplemented in the dry season. Allocation of water in many
AC

dry years is called multiyear regulation. The hydropower stations are divided based on the coefficient of reservoir storage capacity, i.e., daily regulating or
runoff hydropower station (β≤2%), seasonally regulating hydropower station (2% < β ≤ 8%), yearly regulating hydropower station (2% < β ≤ 20%), and
multiyear regulating hydropower station (β > 20%).

31
ACCEPTED MANUSCRIPT

Table 2 Table 3
Pearson correlation coefficients of power generation increase of all cascade Standard deviation of annual average water head of all cascade hydropower
hydropower stations. stations before and after joint operation in flood season.

PT
Items Pearson Correlation Coefficient r Items Standard deviation (m)
TGP-GZB -0.45 Before joint operation After joint operation

RI
TGP-SBY -0.22 TGP 1.92 1.80
SBY-GHY 0.63 GZB 0.41 0.40
SBY-GBZ -0.19 SBY 6.89 3.88

SC
GHY-GBZ 0.08 GHY 1.99 1.06
TG Cascade-QJ Cascade -0.18 GBZ 0.24 0.21

U
Table 4

AN
Power generation parameters of TGP before and after joint operation of the TG and QJ cascades in a typical high-flow year (1998).

M
Single cascade Joint cascades
Outflow Output Water head
Month Period Inflow
difference difference difference
Outflow Output Water head Outflow Output Water head

D
TE
First ten
14580 8392 6951 96 8392 6951 96 0 0 0
days
EP
Middle ten
October 16000 9812 8671 102 9812 8671 102 0 0 0
days
C

Late ten
10897 6879 6386 107 6879 6386 107 0 0 0
days
AC

First ten
9778 9778 9225 109 9778 9225 109 0 0 0
days
November
Middle ten
7212 7212 6833 110 7212 6833 110 0 0 0
days

32
ACCEPTED MANUSCRIPT

Late ten
6335 6335 6009 110 6335 6009 110 0 0 0
days

First ten

PT
6330 6330 6005 110 6330 6005 110 0 0 0
days

RI
Middle ten
December 5822 5822 5526 110 5822 5526 110 0 0 0
days

SC
Late ten
4941 5745 5438 109 5745 5438 109 0 0 0
days

U
First ten
4936 5820 5470 109 5820 5470 109 0 0 0
days

AN
Middle ten
January 4300 6068 5641 107 5184 4838 108 -884 -803 0
days

M
Late ten
5127 5931 5453 106 5127 4768 107 -804 -686 1

D
days

TE
First ten
4306 6074 5524 105 5190 4809 107 -884 -715 2
days
EP
Middle ten
February 3981 5749 5144 103 5749 5268 106 0 124 2
days
C

Late ten
3963 6173 5429 102 5068 4593 105 -1105 -836 3
days
AC

First ten
3538 6190 5325 99 5306 4749 103 -884 -577 4
March days

Middle ten 3223 5875 4903 96 5875 5146 101 0 243 5

33
ACCEPTED MANUSCRIPT

days

Late ten
3236 6451 5174 93 5647 4811 98 -804 -363 6
days

PT
First ten
3600 6252 4797 89 5368 4457 96 -884 -340 7
days

RI
Middle ten
April 5154 6922 5125 86 6038 4931 94 -884 -194 9

SC
days

Late ten
9845 7193 5333 86 6309 5210 95 -884 -123 10

U
days

AN
First ten
10418 6882 5354 90 10418 8754 97 3536 3400 7
days

M
Middle ten
May 11194 14730 11367 89 18266 14546 92 3536 3178 3
days

D
Late ten
14609 19431 13834 82 20235 14507 83 804 673 1
days

Table 5 TE
EP
Power generation parameters of TGP before and after joint operation of the TG and QJ cascades in a typical normal-flow year (1981).
C

Single cascade Joint cascades


AC

Outflow Output Water head


Month Period Inflow
difference difference difference
Outflow Output Water head Outflow Output Water head

First ten
October 19010 12822 10571 95 12822 10571 95 0 0 0
days

34
ACCEPTED MANUSCRIPT

Middle ten
14330 8142 7206 102 8142 7206 102 0 0 0
days

Late ten days 11066 7048 6542 107 7048 6542 107 0 0 0

PT
First ten
8505 8505 8041 109 8505 8041 109 0 0 0

RI
days
Novemb
Middle ten

SC
er 8559 8559 8091 109 8559 8091 109 0 0 0
days

Late ten days 6422 6422 6091 110 6422 6091 110 0 0 0

U
First ten

AN
5859 5859 5561 110 5859 5561 110 0 0 0
days
Decemb
Middle ten

M
er 5003 5887 5572 109 5003 4754 110 -884 -818 0
days

D
Late ten days 4833 5636 5298 109 5636 5336 109 0 38 1

TE
First ten
4280 6048 5623 107 5164 4855 109 -884 -768 1
days
EP
January Middle ten
3635 5403 4951 106 5403 5024 108 0 73 2
days
C

Late ten days 3512 5923 5322 104 5119 4692 106 -804 -630 2
AC

First ten
3533 6185 5417 101 5301 4784 104 -884 -633 3
days
Februar
y
Middle ten
4056 5824 4986 99 5824 5167 103 0 181 4
days

35
ACCEPTED MANUSCRIPT

Late ten days 4230 6440 5401 97 5335 4674 101 -1105 -727 4

First ten
4111 5879 4830 95 4995 4339 100 -884 -491 5
days

PT
March Middle ten
4347 6115 4905 93 5231 4502 99 -884 -403 7

RI
days

Late ten days 4155 6566 5106 90 5762 4883 98 -804 -223 8

SC
First ten
5334 6218 4703 87 5334 4474 97 -884 -229 10
days

U
April Middle ten

AN
5141 6909 5115 86 6025 5027 96 -884 -88 11
days

Late ten days 6035 6919 4993 83 6035 5004 96 -884 11 12

M
First ten
8550 6782 4924 84 6782 5665 97 0 742 13

D
days

TE
May Middle ten
9605 6953 5231 87 14909 12087 94 7957 6856 7
days
EP
Late ten days 8667 14293 10378 84 15901 11685 85 1607 1307 1
C

Table 6
AC

Power generation parameters of TGP before and after joint operation of the TG and QJ cascades in a typical low-flow year (1994).

Single cascade Joint cascades Water


Outflow Output
Month Period Inflow head
difference difference
Outflow Output Water head Outflow Output Water head difference

36
ACCEPTED MANUSCRIPT

First ten days 17710 11522 9515 95 11522 9515 95 0 0 0

October Middle ten days 21270 15082 13251 102 15082 13251 102 0 0 0

PT
Late ten days 13884 9865 9135 107 9865 9135 107 0 0 0

RI
First ten days 8468 8468 8006 109 8468 8006 109 0 0 0

November Middle ten days 9717 9717 9168 109 9717 9168 109 0 0 0

SC
Late ten days 9165 9165 8656 109 9165 8656 109 0 0 0

U
First ten days 9025 9025 8525 109 9025 8525 109 0 0 0

AN
December Middle ten days 7335 7335 6948 110 7335 6948 110 0 0 0

Late ten days 5937 5937 5635 110 5937 5635 110 0 0 0

M
First ten days 5647 5647 5362 110 5647 5362 110 0 0 0

D
January Middle ten days 5203 6087 5760 109 5203 4943 110 -884 -817 0

TE
Late ten days 4448 6056 5670 108 5252 4972 109 -804 -698 1
EP
First ten days 3834 5602 5171 107 5602 5247 108 0 76 2

February Middle ten days 3803 5571 5067 105 5571 5143 107 0 75 2
C

Late ten days 5455 6560 5890 104 5455 4997 106 -1105 -893 2
AC

First ten days 5202 6086 5421 103 5202 4767 106 -884 -654 3

March Middle ten days 4182 5950 5234 102 5066 4626 106 -884 -608 4

Late ten days 4437 6045 5226 100 5241 4750 105 -804 -476 5

37
ACCEPTED MANUSCRIPT

First ten days 5316 6200 5279 98 5316 4796 104 -884 -482 6

April Middle ten days 5999 5999 5078 98 5999 5407 104 0 329 6

PT
Late ten days 6232 6232 5274 98 7116 6383 104 884 1109 6

RI
First ten days 8163 8163 6885 98 13467 11634 100 5304 4749 2

May Middle ten days 8254 13558 11024 94 13558 11024 94 0 0 0

SC
Late ten days 11086 18319 13381 84 18319 13381 84 0 0 0

U
Table 7

AN
Summary of calculation results of power generation (×108kW·h).

Routine operation Optimal operation

M
Isolated Joint
operation operation Incremental power generation Isolated operation Joint operation Incremental power generation

D
TE
TG cascade
977.25 983.44 6.19 1015.36 1017.59 2.23
QJ cascade
82.03 81.58 -0.46 86.17 86.57 0.4
EP
In total
1059.29 1065.02 5.73 1101.53 1104.16 2.63
C
AC

38
ACCEPTED MANUSCRIPT
Hydropower generation potential of joint operation of Three Gorges and
Qingjiang cascade reservoirs

Highlights

• China focusing increasingly on joint operation of cascaded hydropower stations.


• Investigating the increased power generation potential of joint cascade operation.

PT
Dynamic programming algorithm and super-computing avoid dimensionality
problem.
• Optimizing reservoir operation yields greater gains than joint operation.

RI
• Growing water and energy demands require careful construction of station clusters.

U SC
AN
M
D
TE
C EP
AC

You might also like