You are on page 1of 12

Paddy Water Environ

DOI 10.1007/s10333-014-0432-4

ARTICLE

Water management strategies for hydropower annexation


at existing irrigation dams in Japan
Tatsuki Ueda • Masahiro Goto • Atsushi Namihira •

Yuichi Hirose

Received: 18 June 2013 / Revised: 11 March 2014 / Accepted: 4 April 2014


 The International Society of Paddy and Water Environment Engineering and Springer Japan 2014

Abstract This study aims to propose a water manage- Introduction


ment scheme that will balance two objectives, irrigation
and hydropower, assuming a hydropower plant is annexed In the wake of the great East-Japan earthquake and tsu-
at an existing irrigation dam. A case study was conducted nami and the subsequent nuclear accident of Fukushima
at a dam in northeast Japan that has operated solely for in March 2011, wide regions across Japan suffered from
supplying irrigation water since construction. We simplify power shortages for several weeks. Since then, the Japa-
and focus the analyses by setting the current water man- nese society has realized the risk of relying too heavily on
agement scheme as a benchmark. We then propose two large thermal and nuclear power plants, which are often
alternative Scenarios designed to increase efficiency of located along the coast, for power. As an alternative
hydropower generation, and investigate their energy source of energy, attention is increasingly focusing on
potentials and economic performances. Scenario A adheres renewable energy generated at a small scale in scattered
to maintaining a dependable discharge throughout the year. locations. However, renewable energy (including large-
Consequently, it offers distinctively larger net benefits than scale hydropower) supplied only 8.9 % of electricity for
the current scheme, with a benefit–cost ratio (B/C) of 1.851 Japan in FY2008 (ANRE 2010). To facilitate the dis-
as compared to 1.497 for the current scheme. However, semination of renewable energy, the Japanese government
Scenario A has a side effect of insufficient recovery of introduced a feed-in-tariff scheme for renewables in July
reservoir storage at the start of the irrigation period in 2012.
drought years. Therefore, Scenario B tries to mitigate this Small- and medium-scale hydropower, defined here as
effect by occasionally cutting the dependable discharge to output of 30,000 kW or less, is one target in the scheme.
one half during the winter months. Scenario B, however, Hence sale prices were set at 34, 29, and 24 Yen/kWh (as
exhibits a smaller net benefit from hydropower generation of FY2012, tax excluded) for output scales of \200,
(with a B/C of 1.784) than Scenario A. Thus, there is a 200–1,000, and 1,000–30,000 kW, respectively, as com-
trade-off between the competing objectives of hydropower pared to previous prices of 8–10 Yen/kWh (Ueda et al.
and irrigation. A desirable water management scheme 2013). Although river water in Japan has been exploited
depends on which objective of water use farmers and other extensively to date, the Ministry of the Environment esti-
stakeholders regard as most important. mates that Japan still has reserves of 14,440 MW of small-
and medium-scale hydropower generation potentials (MOE
Keywords Dam  Hydropower  Irrigation  Renewable 2011) in rivers and canal networks conveying irrigation,
energy  Water management domestic, and industrial waters.
Irrigation dams constitute a part of such potentials with
an estimated output of 110 MW (NEF 2009). Hence the
T. Ueda (&)  M. Goto  A. Namihira  Y. Hirose Ministry of Agriculture, Forestry and Fishery (MAFF)
National Institute for Rural Engineering, National Agriculture
endeavors to exploit the potentials at existing irrigation
and Food Research Organization, 2-1-6 Kannondai, Tsukuba,
Ibaraki 305-8609, Japan dams by annexing small hydropower stations. Approxi-
e-mail: tued@affrc.go.jp mately, 1,500 irrigation dams had been constructed or

123
Paddy Water Environ

were under construction across Japan as of 2011, and Table 1 Specifications of the dam for the case study
80 % were designed exclusively for supplying irrigation Items Storage
water, i.e., without hydropower plants (Japan Dam capacity (m3)
Foundation 2012). On the average, an irrigation dam
Active storage 23,256,800
serves an irrigation area of 803 ha (71 % of which is
paddy field) with a mean effective storage of 1.76 million Reservoir water levels and storage capacities
m3 (MAFF 2005). Full water level (FWL) 299.5 (m) 23,355,800
Since most of these dams are designed solely for paddy Limited water level 294.4 (m) 16,906,480
in flood seasona
field irrigation, discharges tend to fluctuate by season.
Water level for the rule curve 280.0 (m) 4,295,700
Discharges increase during the paddy irrigation season on 10th September
from late April to early September to meet irrigation Minimum water level (MWL) 259.9 (m) 99,000
water demands, but decrease in other months to restore
Designed maximum tail water discharge Qbmax
reservoir storage for the next irrigation season. On the 3
Land preparation period 7.29 (m /s) –
other hand, hydropower plants are operated more effec- (26/April–10/May)c
tively under a steady discharge rate (US Army Corps of Paddy irrigation period 7.43 (m3/s) –
Engineers 1985). This implies a need to balance these two (11/May–10/September)c
contrastive objectives, i.e., irrigation and hydropower Non-irrigation period 7.43 (m3/s) –
generation, from the viewpoint of water management, if a (11/September–25/April)
hydropower plant is installed ex-post at existing irrigation Irrigation area 4,265 (ha) –
dams. a
Same as the spillway crest level. Applicable from 1/July to
Much research has been done on how to optimize the 30/September
design of newly-built multi-purpose dams for irrigation and b
During the land preparation and paddy irrigation periods, the
hydropower. For instance, Cai et al. (2003) and Rosegrant designed maximum tail water discharge Qmax shall coincide with the
et al. (2000) developed an integrated hydrologic–agro- maximum allowable withdrawal rate of a canal system at a down-
stream weir, which is authorized by the river regulating authority.
nomic–economic model for river basin management that (Note that water is first released from the dam into a river and
balanced multiple water uses, by building a model on a recollected at a weir downstream.) During the non-irrigation period,
large range of input data from root-zone crop water con- by contrast, the maximum allowable withdrawal rate downstream is
sumption to river hydrology. Similarly, Lall and Miller actually 0.832 (m3/s). Nevertheless, the dam operator sometimes
chooses to release water through a penstock in excess of this rate to
(1988) proposed a model to determine optimal storage regulate floods, etc. Therefore, in computing a tail water discharge
capacities for reservoirs, which maximized the expected with Eq. 1, we assume Qmax is 7.43 (m3/s) in this period
net benefits. Johnson et al. (1991) conducted similar opti- c
Periods broadly termed as the ‘‘irrigation period’’ in text and figures
mization studies with the objective of minimizing expected
shortages of water. Georgakakos et al. (1997) and Tilmant
and Kelman (2007) developed models to optimize water probable strong opposition by farmers. Therefore, using a
management for multi-reservoir systems. Such compre- current dam operation scheme as a benchmark may be
hensive approaches are certainly effective for planning helpful to determine any changes under an alternative
optimal water management schemes for a new dam or a water management scheme after installing a hydropower
river basin. plant, and to gain the understanding of farmers.
On the other hand, water management planning for ex- Following this line of argument, few studies have
post installation of a hydropower station at an existing investigated water management with hydropower annex-
irrigation dam may as well rely on a more focused ation at an existing dam. Cobaner et al. (2008) employed an
approach for some reasons. First, comprehensive models artificial neural networks method to predict hydropower
are too complex for a local administrative officer in charge potentials at several irrigation dams in Turkey. Their study
of a small irrigation dam. Therefore, a simple and acces- was based on monthly inflow data, but highly variable
sible model is needed. Second, unlike water management hydrologic conditions in Japan may require a model using
planning at a new dam, any alteration of a water man- daily data. Oven-Thompson et al. (1982) examined a
agement scheme as a result of hydropower annexation may modified water release plan at the High Aswan Dam to
be constrained by the current scheme whose sole purpose is increase the value of water, and recommended increasing
irrigation. For instance, Cobaner et al. (2008) noted that, in firm hydropower production by restructuring the current
such a situation, it was necessary to generate energy in cropping patterns in the Nile Delta. Although their
strict compliance with the irrigation schedule to avoid the approach provided valuable insights, the conclusions may

123
Paddy Water Environ

not be directly applicable to the situations of wet monsoon 1.10


Asian countries such as Japan. 1.05 (a)
Therefore, with a focus on the wet climatic conditions of 1.00
northeastern Japan, this study aims to propose a simplified 0.95
water management scheme that balances the two objectives 0.90
of irrigation and hydropower. It assumes ex-post annex- 0.85

ηq
ation of hydropower at an existing irrigation dam, and 0.80

employs current dam operation data collected on a daily 0.75

basis. This alternative scheme is then compared with the 0.70


0.65
current scheme with regards to hydropower generation
0.60
potentials and economic benefits and costs. 0 0.2 0.4 0.6 0.8 1 1.2
Relave discharge rq

Materials and methods 1.10


1.05 (b)
Data collection
1.00
0.95
A case study was conducted using water management data
0.90

ηh
collected at an anonymous irrigation dam in the Tohoku
region of northeast Japan. The dam is a concrete gravity 0.85
dam possessing a penstock (conduit for conveying tail 0.80
water discharge) with a Howell-Bunger valve and an
0.75
overflow spillway with radial gates. It was constructed in
0.70
1961 without a hydropower plant, and is still operated as an
0.4 0.6 0.8 1 1.2 1.4 1.6
irrigation-only dam. Selected specifications of the dam are Relave effecve head rh
shown in Table 1. The data collected consist of dam res-
ervoir water level, total discharge, and reservoir inflow rate Fig. 1 Turbine efficiency coefficients versus (a) relative discharge
recorded daily from April 26th, 1989 to April 25th, 2011. and (b) relative effective head for a Francis turbine (NEF 2002) gq:
Variable-discharge turbine efficiency coefficient; gh: Variable-head
The total discharge is defined here as a sum of tail water
turbine efficiency coefficient. Strictly speaking, turbines of different
and spillway discharges. The irrigation season starts on sizes have different optimum ‘‘turbine-specific speeds’’ (Mays 2001),
April 26th every year, so we treated a 1-year period starting which in turn have slightly different variable-discharge efficiency
on this date as a unit of analyses. curves (NEF 2002). However, this study omits such details, and
simply assumes a turbine-specific speed of 104 (m-kW), which is
considered representative within the range of the current investigation
Methods for estimating hydropower generation
potentials
It should be noted that undertaking hydropower gener-
Tail water discharge ation at an irrigation dam throughout 1 year would nor-
mally require specific permission from the river authority
Raw data of total discharge contain only a sum of tail water in Japan. This study simply presumes such permission is
discharge and spillway discharge, so we must distinguish granted, and focuses on the technical aspects of the
the former, which can later be used for hydropower gen- undertaking.
eration. (Spillway discharge can never be used for hydro-
power generation.) In doing this, we simply assume that, if Effective head
the total discharge, Qd, is larger than the designed maxi-
mum tail water discharge, Qmax (Table 1), then the dif- An effective (net) head is the useful energy for the turbine,
ference between Qd and Qmax constitutes a spillway and calculated by subtracting friction energy losses in a
discharge. Thus, tail water discharge is estimated as: penstock and other hydraulic components from the gross
Qt ðiÞ ¼ Qd ðiÞ if Qd ðiÞ\Qmax ðiÞ head, which is the difference between reservoir and tail
ð1Þ water levels (Mays 2001). (Tail water level is a water
Qt ðiÞ ¼ Qmax ðiÞ if Qd ðiÞ  Qmax ðiÞ;
surface level of a tailrace into which tail water is dis-
where Qt(i) is the tail water discharge on date i (m3/s), charged through a draft tube (Mays 2001).) Throughout
Qd(i) is the total discharge on date i (m3/s), Qmax(i) is the this study, we assume that the hydropower plant is installed
designed maximum tail water discharge on date i (m3/s). at the outlet of penstock, and friction energy losses are

123
Paddy Water Environ

15 % of the gross head, which is a rule-of-thumb value for A

rough estimation (J-Water 2006). Therefore, the effective

Tail water discharge


F Rated discharge
head is calculated as: D D’

for hydropower
Usable range
he ðiÞ ¼ ½hd ðiÞ  h0   0:85; ð2Þ
where he(i) is the effective (net) head on date i (m), hd(i) is Lower limit of
the dam reservoir water level on date i (m), h0 is the tail E
G operable discharge
E’
water level (m).
C
O H B
Turbine selection Days (365 d)

Based on the dataset and Eqs. 1 and 2, possible ranges of Fig. 2 Schematic diagram for the usable range of tail water discharge
for hydropower generation. Tail water discharges are arranged in
tail water discharge and effective head are expected to be
descending order
0.6–7.4 m3/s and 19–48 m, respectively. These are broadly
within the application ranges of a Francis turbine (NEF
2002), one of the most popular turbines for hydropower
plants in Japan. Therefore, a Francis turbine is assumed of tail water discharge is hypothetically set between D–D0
throughout the following analyses. and E–E0 . Accordingly, usable water for power generation
is the area DFGHO, given that a portion of discharge in
Turbine efficiencies under variable discharge and head excess of D–D0 can be bypassed through another pipeline.
Thus, the results in Fig. 2 imply that an identical discharge
The power-generating efficiency of a turbine varies with pattern such as the A–C curve can have different hydro-
discharge and effective head, and a turbine is only capable power generation potentials depending on the scales of a
of operating satisfactorily over limited ranges of discharge turbine. (Note that the scale of a turbine is roughly pro-
and head (US Army Corps of Engineers 1985). Figure 1 portional to its rated discharge; a larger turbine holds
shows the efficiency curves of a Francis turbine (NEF higher lines for D–D0 and E–E0 .)
2002), which can be approximated as: As for the rated effective head her, Fig. 1b shows that
the operable range of effective head extends from 0.5her to
gq ¼ 3:0449 rq4 þ 7:657 rq3  7:4317 rq2 þ 3:628 rq 1.7her with the optimal efficiency attainable in between
þ 0:1711 ð3Þ (i.e., exactly at her). This implies that it is usually ineffi-
2
R ¼ 0:9997; cient to set her equal to the maximum effective head
obtainable at the full reservoir water level, and therefore her
gh ¼ 0:9313 rh4 þ 4:5696 rh3  8:5793 rh2 þ 7:2 rh
 1:2577 has to be determined, together with the rated discharge,
ð4Þ
through trial and error to optimize target indicators as
R2 ¼ 0:9998;
discussed later.
Qp he
rq ¼ ; rh ¼ ;
Qr her
Power and energy generated
where gq is the variable-discharge turbine efficiency coef-
ficient, gh is the variable-head turbine efficiency coeffi- Whereas theoretical output (power) generated with a tur-
cient, Qp is the discharge used for power generation (m3/s), bine is the product of discharge by effective head and the
Qr is the rated (maximum) discharge for power generation gravitational acceleration constant, actual output is con-
(m3/s), he is the effective (net) head (m), her is the rated strained by efficiency factors. When a turbine is operated at
effective head (m). the rated discharge Qr and the rated effective head her, a
Figure 1a indicates that the ratio of the lower limit of typical combined efficiency of turbine and dynamo is
operable discharge to the rated discharge is 0.2, and the approximately 0.7 (J-Water 2006). Therefore, an actual
coefficient gq largely becomes smaller as the discharge output generated at a certain discharge Qp and an effective
falls. This has a significant implication for a hydropower head he can be estimated by further incorporating Eqs. 3
plant at an irrigation dam with variable discharges, because and 4 as follows:
only a portion of the tail water discharge can be utilized for P ¼ g  Qp  he  0:7  gq  gh ; ð5Þ
power generation. This is schematically depicted in a dis-
tribution diagram for tail water discharges that are sorted in where P is the electrical output (power) (kW), g is the
a descending order (Fig. 2). In this figure, the usable range gravitational acceleration constant.

123
Paddy Water Environ

Summing up the output over a period of 1 year gives the FS ¼ 11:5  E50  0:95; ð10Þ
annual energy production:
where FS is the fossil fuel saving (Yen/y), E50 is the annual
X
365
energy production in a median year (kWh/y), 0.95 is the
E¼ ½PðiÞ  24; ð6Þ
i¼1
plant utilization factor for allowing maintenance breaks.

where P(i) is the electrical output on date i (kW), E is the Costs


annual electrical energy production (kWh/y).
In evaluating the costs, we confine our analyses to costs for
Indicator for power generation efficiency constructing and running a hydropower plant. Additional
costs for maintaining grid system stability (e.g., preparing
A hydropower plant can be operated at full capacity with reserve capacity to backup intermittent hydropower gen-
the rated discharge and the maximum effective head (at full eration) that may be incurred on a grid operator (UKERC
reservoir water level). An output generated under such 2006) are not considered. This is because penetration of
conditions can be estimated by: intermittent renewable energy is still low in Japan, and
Pmax ¼ g  Qr  hemax  0:7; ð7Þ addition of a small-scale hydropower to the grid is unlikely
to affect the grid stability at present.
where Pmax is the maximum power (kW), hemax is the
As for the capital costs, we include only costs related to
maximum effective head (m).
hydropower plant construction and exclude those for dam
Power generation during a period of 1 year is the most
construction, because we assume ex-post annexation of a
efficient if the plant is always operated at full capacity.
power plant at an existing dam.1 Such costs have been
Hence such efficiency is indicated by the capacity factor
estimated by NEF (2002) with an assumption that the plant
CF, which is a ratio of the actual annual energy production
is constructed on the ground surface with a single Francis
E to the theoretical maximum energy production attainable
turbine as follows:
at full capacity:
Yh ¼ 0:084  ðPmax Þ0:830 ; ð11Þ
E
CF ¼  100 ð%Þ ð8Þ 1:49
Pmax  24  365 Yb ¼ 0:0595  ðQr  h2=3emax Þ ; ð12Þ
Methods for evaluating economic performances  0:648
Pmax
Ye ¼ 12:8  pffiffiffiffiffiffiffiffiffiffi ; ð13Þ
hemax
Benefits
where Yh is the cost for constructing housing of the power
Benefits of hydropower generation consist of capacity credit plant (million Yen), Yb is the cost for constructing the
and fossil fuel saving (UKERC 2006). Capacity credit is basement for machinery (million Yen), Ye is the cost for
avoided capital and other fixed costs for constructing thermal electrical equipment (million Yen).
(fossil fuel) power plants, because a portion of thermal Using a method of MAFF (1995), the above costs are
capacities can be displaced by a hydropower plant in the long amortized over 30 years at a discount rate of 3 % and
run, if hydropower provides stable output comparable to combined with annual maintenance costs to form an annual
thermal plants. Therefore, capacity credit is usually given to total cost TC.
the stable portion of hydropower outputs. This study follows Finally, the net benefit NB and the benefit cost ratio (B/
Nonaka and Asano (2011) and estimates capacity credit as: C) are evaluated as:
CC ¼ 6; 200  P90 ; ð9Þ NB ¼ ðCC þ FSÞ  TC, ð14Þ

where CC is the capacity credit (Yen/y), 6,200 is the capital B=C ¼ ðCC þ FSÞ=TC ð15Þ
cost for a displaced fossil fuel power plant (Yen/(kW y)),
P90 is the output that can be maintained over 90 % of the
days during a drought (90 percentile) year (kW).
1
Fossil fuel saving is a direct benefit of generating Exactly speaking, a cost is also necessary to install a new pipeline
that branches off the existing penstock and diverts water to the
electricity with hydropower, because every kWh (energy) hydropower plant, since the plant is assumed to use only a portion of
generated can in principle avoid equivalent fuel uses as the tail water discharge at times. However, we design to construct the
well as associated greenhouse gas emissions at a thermal plant as close as possible to the outlet of penstock, so that the length
power plant. Nonaka and Asano (2011) estimated this of the branch pipe would be minimal. Our rough estimation indicates
that such piping would cost less than a few percent of the power plant
benefit in a Japanese context at around 11.5 (Yen/kWh). construction in total. Therefore we omit it here for simplicity,
We accordingly estimate an annual fossil fuel saving as: although it may be a subject for future studies.

123
Paddy Water Environ

(a) (a)

(b)

(b)

Fig. 3 a Daily reservoir inflow and b total discharge (actual records


for each year). Note the total discharge indicates a sum of tail water
and spillway discharges

Results and discussions (c)

Descriptions of water management schemes

Overview of the current water management scheme

Figure 3a shows reservoir inflow rate during a 22-year


period. Watershed of the dam has a wet cold climate with a
mean annual precipitation of approximately 1,900 mm,
which turns to snowfall in winter months. Thus, from late
December to early March, reservoir inflows are generally
low as snow accumulates in the mountain slopes. The snow
melts from March to early May, thereby augmenting the
reservoir inflow. From mid-June through October, occa-
Fig. 4 Daily dam reservoir water level. a Actual records under the
sional high inflows are observed due to typhoons as well as current water management scheme; b simulated values for alternative
other heavy rain events. water management scheme (Scenario A); c ibid (Scenario B). Note
Figure 3b shows total discharge and Fig. 4a variations Thin solid curves values for each year; bold solid curve daily median
values; dashed line the rule curve
in the dam reservoir level. Currently, the rule curve for dam
operation, illustrated in Fig. 4a, is applied rather heuristi-
cally as explained below. The prime target of water man- of the stipulated flood season (see Table 1), to make
agement is to restore the reservoir storage on April 26th, capacity for flood control. Toward the end of the irrigation
the first day of the irrigation period, to the full water level season on September 10th, the dam operator continues to
(FWL). Enough water can then be released to meet irri- release water satisfying the irrigation demands but not in
gation demands thereafter. In the meantime, the reservoir excess to avoid possible water shortages at the end of the
level must be lowered to 294.4 m by July 1st, the first day irrigation season. (Occasional spikes in total discharges

123
Paddy Water Environ

during this period (Fig. 3b) are spillway discharges for


lowering the reservoir level to the spillway crest of hd
Qin(i-1)
294.4 m.) On September 10th, the rule curve cuts a level of Yes
280 m (Fig. 4a), which is the midpoint between the FWL hd<hmin Qt=0
and the minimum water level (Table 1). However, in terms No
of reservoir storage capacity, this level (280 m) corre- No Yes
sponds to just 18 % of the active storage (Table 1). Thus, Qin(i-1)≥Qmax
Yes
the remaining storage at this level is reserved as a safety hd≥hR hd≥hs
margin for drought years. Yes
No No
After the irrigation period is over, the operator continues
1/Jan- 1/Apr- 26/Apr- Yes
to release water to keep the reservoir level as low as pos-
31/Mar 31/Dec 10/Sep
sible until December 1st (Fig. 4a). Such a release appears No Yes
Yes No
to be rather wasteful but in fact rational, because the sole No
objective of dam operation during this period is to run the hd≥hR-5
dam safely (i.e., prevent a dam overflow) in heavy floods. Yes
No
Nevertheless, it is possible to operate the dam at higher
Qt=Qmin Qt=Qd Qt=Qmax
reservoir levels during this period without compromising
safety, as we will see later. Hence, we later try to amend
Fig. 5 Algorithm for determining the tail water discharge. Dashed
dam operation during this period to add the other objective line indicates a flow applicable to Scenario A only, whereas gray lines
of hydropower generation. and boxes imply flows applicable to Scenario B only. Qin(i-1)
From December 1st to early April, the main objective of reservoir inflow on the previous day, Qt tail water discharge, Qmax
operation is the restoration of reservoir storage for the next designed maximum tail water discharge, Qd dependable discharge,
Qmin minimum discharge (=Qd/2), hd dam reservoir water level, hmin
irrigation season, so water release is cut to a minimum minimum water level, hR reservoir level on the rule curve, hs spillway
(Fig. 3b). As a result, water level is eventually restored to crest level
294.4 m, the crest level of the spillway, by early April
(Fig. 4a). Reservoir storage is then allowed to increase other hand, to protect the farmers’ interests, we choose to
with the snowmelt waters to the FWL before the irrigation maintain the basic principles of the current scheme during
season starts (Fig. 4a). the irrigation period.
Accordingly, a new rule curve is set up during the
Simulation on alternative water management schemes critical period, and the old curve is otherwise maintained.
The new curve is an upward sloping line connecting the
This section presents a method for modifying the current two reservoir levels of the old curve on September 11th and
water management scheme discussed above, assuming a March 31st (Fig. 4b, c). In line with this curve, a
hydropower plant is newly annexed at the dam. A hydro- ‘‘dependable’’ discharge rate that could be released con-
power plant can be operated most efficiently when output is tinuously throughout the critical period is set as follows:
maintained as constant as possible. In practice, this goal DD ¼ ðCICSÞ=ðCP  24  3600Þ; ð16Þ
translates into an objective for maximizing the ‘‘firm out-
put,’’ an output that is expected throughout the year (US where DD is the dependable discharge (m3/s), CI is the
Army Corps of Engineers 1985). In terms of dam opera- cumulative reservoir inflow during the critical period (90
tion, this requires identifying the maximum possible flow percentile value, m3), CS is the target cumulative incre-
rate that can be discharged throughout the ‘‘critical per- ment in reservoir storage during the critical period (m3), CP
iod,’’ the most adverse stream flow period during the year is the number of days in the critical period (September
(US Army Corps of Engineers 1985). 11th–March 31st).
In the context of our case study, the total discharge is the Since the target increase in the reservoir level is from 280
lowest during the winter months, because the reservoir to 294.4 m during the critical period (see the rule curve in
inflow is the lowest due to snow accumulating in the Fig. 4b), CS is set at a reservoir storage capacity between
mountains, and because the reservoir storage needs to be these two levels (see Table 1). Accordingly, the dependable
restored for the next irrigation season. Therefore, we fix the discharge is calculated as 2.08 m3/s. (However, it is
critical period between September 11th and March 31st, exceptionally set at 5.06 m3/s from April 26th to June 30th
non-irrigation period including the winter months. Since (calculated similarly with Eq. 13), when large amounts of
the reservoir storage is currently left idle during this period, water must be released to meet irrigation demands).
we modify the water management plan by adding the Figure 5 shows the algorithm for determining tail water
additional objective of hydropower generation. On the discharge. The basic principle is to maintain a dependable

123
Paddy Water Environ

discharge throughout the year. In addition, the algorithm


tries to simulate, albeit crudely, the current scheme during
the irrigation period by augmenting discharge to the max-
imum Qmax when water level is above the rule curve or the
spillway crest level. On the other hand, for the winter
period between January 1st and March 31st, two scenarios
are set. Scenario A adheres to the principle of maintaining
a dependable discharge throughout the year regardless of
the reservoir level relative to the rule curve. By contrast,
Scenario B compromises this principle by reducing the
discharge to half, if the reservoir level drops 5 m or more
below the rule curve. In this way, this scenario prioritizes
Fig. 6 Variations in daily tail water discharge. The graphs show the
restoring the reservoir level along the rule curve more median of daily data for 22 years
swiftly and reliably for the start of the irrigation period in
April.
To determine the spillway discharge, we simply adopt
the observed data rather than conduct detailed flood simu-
lations, because the necessary (every hour or minute) inflow
data are not available. Thus, we assume that the spillway
discharge is equal to the difference between the observed
total discharge and the tail water discharge calculated
above, provided that the reservoir level is above both the
rule curve level and the spillway crest level. The spillway
discharges calculated by this method comply with the dam
operation regulations imposed by the river authority.
Calculations are started on April 26th every year with
actual data of the reservoir water level hd, to avoid influence Fig. 7 Distribution of daily tail water discharge arranged in
from the preceding year’s results. After the tail water dis- descending order. The graphs show the median of daily data for
22 years
charge is determined with the algorithm of Fig. 5, the cumu-
lative total discharge of the day is summed, and then hd discharges and rated effective heads. Therefore, the rated
calculated on the next day by employing the continuity discharge Qr varies from Qmax to 20, 30, …, 70 percentile
equation (Mays 2001) and the H–V curve of the reservoir. The discharges Qt20, Qt30, …, Qt70 evaluated with Fig. 7. (10
calculation is repeated until April 25th of the following year. percentile values are omitted as they are exactly the same
as the 20 percentile values.) Meanwhile the rated effective
Turbine design and hydropower generation head her varies from the maximum effective head hemax to
performances 0.9, 0.8, …, 0.5 times that value. However, our subsequent
analyses discussed below indicate that relevant indicators
Under conditions of variable discharges, selection of the rated are broadly optimized at her of 0.7hemax. Therefore, to
discharge and the rated effective head has important ramifi- simplify the argument, we confine discussions of analytical
cations in determining the turbine efficiency (see Eqs. 3 and results to this condition for her.2 The above procedures are
4). Thus, this section provides a case study on these relation- repeated for Scenarios A and B for later discussions.
ships under the current water management scheme. Variations
in the current discharge (Fig. 6) indicate an increase starting
from early March first for release of snowmelt water, then 2
Among indicators discussed later, both the net benefit and the
supply of irrigation water throughout the irrigation period, and benefit-cost ratio (with regard to the current scheme and Scenarios A
finally maintenance of a low reservoir level until mid- and B) are all maximized when her is set at 0.7hemax, with the only
exception for the net benefit of Scenario B, which is maximized, with
December. The discharge is subsequently kept very low until
her of 0.8hemax, at 18,886 thousand Yen/y. However, the difference in
early March. When these discharges are arranged in a that variable between the 0.7hemax and 0.8hemax cases are minimal
descending order as shown in Fig. 2, the distribution curve (0.57 %) and hence adopting the 0.7 hemax case instead (as shown in
follows a smooth curve until an approximate 80 percentile Table 2) does not affect the conclusion in any way. We similarly
confirmed that the 0.7hemax case scored the optimal value or very
value (Qt80), or the 292nd day (Fig. 7).
close to the optimum for other indicators as well. Therefore, we omit
With the above data, we then estimate the power gen- the estimated values with her save 0.7hemax as it greatly simplifies
eration performances under various conditions of rated Table 2 and discussions in text.

123
Paddy Water Environ

Table 2 Hydropower generation potentials and economic performances of the schemes


Qmax Qt20 Qt30 Qt40 Qt50 Qt60 Qt70

Rated discharge Qr (m3/s)


Current 7.43 7.29 6.21 5.50 4.76 3.95 3.18
Scenario A 7.43 7.43 7.29 5.08 5.08 2.08 2.08
Scenario B 7.43 7.43 7.29 7.29 5.08 2.08 2.08
Maximum power Pmax (kW)
Current 1,950 1,913 1,629 1,443 1,249 1,036 834
Scenario A 2,440 1,950 1,913 1,333 1,333 546 546
Scenario B 2,440 1,950 1,913 1,913 1,333 546 546
Annual energy production E (MWh/y)
Current 5,971 5,982 5,898 5,721 5,348 4,682 3,873
Scenario A 6,181 6,181 6,152 5,446 5,446 3,478 3,478
Scenario B 6,133 6,133 6,103 6,103 5,416 3,411 3,411
Capacity factor CF (%)
Current 35.0 35.7 41.3 45.3 48.9 51.6 53.0
Scenario A 36.2 36.2 36.7 46.6 46.6 72.8 72.8
Scenario B 35.9 35.9 36.4 36.4 46.4 71.3 71.3
Benefit, gross total TB (=CC ? FS) (1,000 Yen/y)
Current 65,230 65,349 64,431 62,505 58,424 51,149 42,310
Scenario A 68,850 68,850 68,549 60,996 60,996 39,722 39,722
Scenario B 67,000 67,000 66,673 66,673 59,930 38,296 38,296
Benefit, capacity credit CC (1,000 Yen/y)
Current 0 0 0 0 0 0 0
Scenario A 1,324 1,324 1,333 1,494 1,494 1,721 1,721
Scenario B 0 0 0 0 766 1,033 1,033
Benefit, fossil fuel saving FS (1,000 Yen/y)
Current 65,230 65,349 64,431 62,505 58,424 51,149 42,310
Scenario A 67,526 67,526 67,216 59,501 59,501 38,001 38,001
Scenario B 67,000 67,000 66,673 66,673 59,165 37,263 37,263
Cost, total TC (1,000 Yen/y)
Current 54,805 54,016 47,836 43,666 39,207 34,160 29,150
Scenario A 54,805 54,805 54,016 41,151 41,151 21,463 21,463
Scenario B 54,805 54,805 54,016 54,016 41,151 21,463 21,463
Net benefit NB (=TB - TC) (1,000 Yen/y)
Current 10,426 11,334 16,595 18,839 19,217 16,989 13,160
Scenario A 14,045 14,045 14,533 19,844 19,844 18,258 18,258
Scenario B 12,196 12,196 12,658 12,658 18,779 16,832 16,832
Benefit–cost ratio B/C (=TB/TC)
Current 1.190 1.210 1.347 1.431 1.490 1.497 1.451
Scenario A 1.256 1.256 1.269 1.482 1.482 1.851 1.851
Scenario B 1.223 1.223 1.234 1.234 1.456 1.784 1.784
Median values of output (i.e., solid curves in Fig. 9) are used for calculations, except for the capacity credit that is estimated on the basis of 90
percentile values of output (i.e., lower limit of gray areas in Fig. 9) (see text). All values are evaluated at a rated effective head of 0.7hemax (see
text)

The results are shown in Table 2. For the current and costs between the optimum Qr of Qt50 and larger Qr
scheme, the annual energy production is maximized at (say, Qt20), a much smaller cost TC of Qt50 (39,207 thou-
5,982 (MWh/y) with a rated discharge (Qr) of Qt20. In sand Yen/y) more than offsets the corresponding reduction
contrast, the net benefit is maximized at 19,217 (thousand in the gross benefit TB to augment the net benefit NB of
Yen/y) with a smaller Qr of Qt50. Comparing the benefits Qt50. Figure 8 further depicts variations in the daily outputs

123
Paddy Water Environ

B differs from Scenario A only by a temporal cut in the


dependable discharge for parts of February and March to
facilitate swift recovery of the reservoir storage.
Likewise, differences in the reservoir level are most
visible in the non-irrigation period from September to April
(Fig. 4). Timing of reservoir level restoration before the
irrigation period tends to be deferred in Scenarios A and B
(Fig. 4b, c) relative to the current scheme (Fig. 4a). This
tendency is stronger for Scenario A (Fig. 4b) due to
adherence to releasing a dependable discharge throughout
the winter (see Fig. 5). As a consequence, in a drought (90
percentile) year, the reservoir storage on April 25th (day
before the irrigation period starts) is recovered to only 96 %
of the current level. Scenario B, by contrast, is designed to
Fig. 8 Variations in outputs under various conditions of rated mitigate this effect by reducing the dependable discharge to
discharges (current water management scheme). Rated effective head one half under drought conditions (see Fig. 5). Accordingly,
is fixed at 0.7hemax
the recovery of reservoir storage (Fig. 4(c)) is swifter than
Scenario A, so that exactly 100 % of the current storage
under different conditions of Qr. Notable output peaks in level is recovered on April 25th in a 90 percentile year.
the energy maximization case (Qt20) imply that power During the irrigation period, in the meantime, both
generation potentials across the irrigation period are fully Scenarios A and B release water in exactly the same pattern
utilized by choosing a larger turbine (i.e., larger Qr). By (Fig. 6) due to their identical algorithm for determining tail
contrast, the benefit maximization case (Qt50) cuts these water discharge (Fig. 5). Both Scenarios appear to supply
peaks off by choosing a smaller turbine (smaller Qr). Thus, at least the current level of water during most of the period
the lower cost of this case can be attributed in part to the (Fig. 6), even though the discharge curves are not smooth
higher capacity factor CF (Table 2), because if the output due to employment of a simplistic algorithm (Fig. 5). In
curve is flatter, the factor tends to be higher. quantitative terms, the average discharges across the period
These results demonstrate that the optimal scale of the for both Scenarios (5.99 m3/s) are larger than the current
turbine depends on whether energy maximization or benefit level (5.26 m3/s) by 14 %, and therefore we may well
maximization is emphasized. The latter objective for benefit judge the Scenarios are able to supply enough irrigation
maximization appears to be more important under the cur- water on average. Nevertheless, the same period has
rent circumstances in Japan. This is because revenue from 18 days (i.e., 13 % of the period) in which the Scenarios
hydropower generation by irrigation dams is often expected fall short of the current level of water supply by more than
to help finance routine operations of local Land Improve- 20 %. In this regard, future studies should improve the
ment Districts (farmers’ water management associations), simplistic algorithms of this study in order to fine-tune the
and farmers may opt for an economically efficient way of water supply schedule during the irrigation period by
power generation. Thus, we focus on an objective for ben- incorporating crop water demands etc.
efit maximization in the subsequent discussions. In summary, the main effect of adopting the new Sce-
narios in lieu of the current scheme appears in the non-
Comparisons of water management schemes irrigation season, although there are minor ramifications in
the irrigation period as well. Comparing the performances
Effects on reservoir water level and tail water discharge of Scenarios A and B in the non-irrigation period, Scenario
A may cause negative impacts in a drought year, because it
Variations in reservoir water level and tail water discharge fails to recover the full reservoir storage at the start of the
under different water management schemes are depicted in irrigation period, as a result of adherence to the principle to
Figs. 4 and 6, respectively. Differences in the tail water maintain a dependable discharge throughout the year.
discharge are most distinctive during the non-irrigation Scenario B, by contrast, is successful in mitigating this
period (Fig. 6). Both Scenarios A and B aim to store spare adversity by somewhat compromising the principle.
inflow in autumn and release it later to maintain a dependable
discharge throughout the winter, especially from December Effects on hydropower generation
through March. Thus, a steady release of a dependable dis-
charge (at 2.08 m3/s) can have a profound effect on hydro- Hydropower generation performances of the current and
power generation, as discussed in the next section. Scenario alternative schemes are presented in Table 2. In terms of

123
Paddy Water Environ

(a) throughout the year, thereby keeping the turbine operating


1000
at an almost full capacity.
As for economic performances, the net benefit of each
800

scheme is maximized with the same Qr of Qt50, while the


Output (kW)

B/C with Qt60. In terms of the net benefit, Scenario A


600

scores highest (19,844 thousand Yen/y), followed by the


current scheme (19,217) and then Scenario B (18,779). As
400

for the B/C, Scenario A similarly scores highest (1.851),


followed by Scenario B (1.784) and then the current
200

scheme (1.497). The best performance by Scenario A can


be attributed in part to the largest capacity credit awarded
0

J F M A M J J A S O N D among the cases. This must be derived from the relatively


Months
stable output of Scenario A during drought months and
1000

(b) years, because the credit is only given to stable outputs (see
Eq. 9).
800

Therefore, to observe stability in outputs, daily


hydropower generation is simulated for a 22-year period
Output (kW)
600

following the flowchart of Fig. 5, with maximizing con-


ditions for the B/C (i.e., Qr of Qt60). Then, the median as
400

well as 10 and 90 percentile values of outputs are


extracted for each day and graphed in Fig. 9. The results
200

reveal a sharp contrast between the current scheme and


the two Scenarios. The current scheme (Fig. 9a) shows
0

wide margins between the 10 and 90 percentile values,


J F M A M J J A S O N D
Months particularly during the winter months, indicating output
instability. Consequently, the scheme suffers from zero
1000

(c)
capacity credit (Table 2). By contrast, both Scenarios A
and B (Fig. 9b, c) demonstrate relatively narrower mar-
800

gins implying stable output across months and years.


Output (kW)

Hence such stability renders capacity credits as well as


600

lower costs (due to higher capacity factors) to the two


Scenarios (Table 2).
400

Therefore, under the B/C maximizing conditions of Qt60,


a stable output gives favorable scores to Scenarios A and B
200

over the current scheme in terms of capacity credit and


costs. This more than compensates for their smaller fossil
0

J F M A M J J A S O N D fuel saving (due to the smaller annual energy production)


Months
than the current scheme. Thus, the two Scenarios enjoy
10−90 percentile Median
larger B/C.
Fig. 9 Variations in output under different water management
Meanwhile, the differences between Scenarios A and B
schemes (B/C maximization case). a Current water management (Fig. 9b, c) are noticeable from January to March, when
scheme; b Alternative scheme (Scenario A); c ibid (Scenario B) temporal cuts of discharge in Scenario B (Fig. 6) destabi-
lize the output. This leads to a lower capacity credit as well
as a lower capacity factor (hence a higher cost) for Sce-
annual energy production, the difference between the nario B relative to Scenario A under the B/C maximizing
schemes is not distinctive. Thus, the alternative Scenarios conditions (Table 2).
do not clearly affect augmentation of energy production. In summary, Scenario A shows the best performance in
On the other hand, the difference in the capacity factor is terms of both the net benefit and the B/C. This must be
noticeable especially at a rated discharge of Qt60 or smaller, chiefly derived from the stable output throughout the year.
implying efficient hydropower generation for Scenarios A Scenario B suffers from a smaller net benefit compared to
and B. This can be explained by referring to Fig. 7. If the Scenario A due to the compromise in the discharge plan in
rated discharge is set at this value (2.08 m3/s) in the two winter (Fig. 5). Nevertheless, Scenario B is still superior to
Scenarios, a uniform discharge can be maintained almost the current scheme at least in terms of the B/C (although

123
Paddy Water Environ

superiority is reversed when the performance is measured J Water Resour Plan Manage 129(1):4–17. doi:10.1061/
in terms of net benefits). (ASCE)0733-9496(2003)129:1(4)
Cobaner M, Haktanir T, Kisi O (2008) Prediction of hydropower
energy using ANN for the feasibility of hydropower plant
installation to an existing irrigation dam. Water Resour Manage
Conclusions 22:757–774. doi:10.1007/s11269-007-9190-z
Georgakakos AP, Yao H, Yu Y (1997) A control model for
dependable hydropower capacity optimization. Water Resour
Focusing on hydropower annexation to an existing irriga- Res 33(10):2349–2365. doi:10.1029/97WR01713
tion dam, the present study proposed alternative water Japan Dam Foundation (2012) Dam Yearbook 2012 [In Japanese]
management schemes that can maintain a dependable dis- Johnson SA, Stedinger JR, Staschus K (1991) Heuristic operating
charge throughout the year for efficient hydropower gen- policies for reservoir system simulation. Water Resour Res
27(5):673–685. doi:10.1029/91WR00320
eration, while preserving the principal function of J-Water (2006) Small hydropower handbook. In: The Japan confer-
supplying irrigation water. The results suggest that, if we ence for small hydropower (J-Water), Ohm-sha, Tokyo, Japan
adhere to the principle of maintaining a dependable dis- [In Japanese]
charge throughout the year (Scenario A), we should be able Lall U, Miller CW (1988) An optimization model for screening
multipurpose reservoir systems. Water Resour Res
to obtain a distinctively larger net benefit from hydropower 24(7):953–968. doi:10.1029/WR024i007p00953
than the current scheme. However, the Scenario also has a MAFF (1995) Manual for planning and design technologies for small-
side effect of insufficient recovery of reservoir storage at scale power plant in irrigation facilities. The Ministry of
the start of the irrigation period in drought years. When we Agriculture, Forestry and Fishery, Japan [In Japanese]
MAFF (2005) Register of agricultural dams, March 2005. The
try to mitigate this effect (Scenario B), the net benefit from Ministry of Agriculture, Forestry and Fishery, Japan [In
hydropower is reduced relative to Scenario A. Japanese]
Therefore, we face a trade-off between the competing Mays LW (2001) Water resources engineering, 1st edn. Wiley, New
objectives of hydropower and irrigation. Hence a desirable York
MOE (2011) Study of potential for the introduction of renewable
water management scheme depends on which objective of energy FY 2010. The Ministry of the Environment [In Japanese
water use farmers and other stakeholders regard as most with English summary]
important. If they can accept a possible reduction in cul- NEF (2002) Small- and medium-scale hydropower guidebook. The
tivation area in drought years, then one may well choose New Energy Foundation [In Japanese]
NEF (2009) Survey on hydropower potentials with unused water
Scenario A to obtain the maximum benefit from hydro- heads FY 2008. A report commissioned by the Agency for
power. If, on the other hand, they place priority on securing Natural Resources and Energy, The New Energy Foundation [In
water for irrigation every year even after hydropower Japanese]. http://www.enecho.meti.go.jp/hydraulic/data/dl/hou
annexation, then it is better to choose Scenario B or even kokusho.pdf Accessed 1 June 2013
Nonaka J, Asano K (2011) Evaluation of competitiveness of solar
maintain the current scheme. power against grid power. SERC Discussion Paper 11028, the
In this way, the methodology and outcome of this study Central Research Institute of the Electric Power Industry, Tokyo
is useful in aiding decision-making of stakeholders facing [In Japanese]
such a trade-off. We suggest that the present study be Oven-Thompson K, Alercon L, Marks DH (1982) Agricultural vs.
hydropower tradeoffs in the operation of the High Aswan Dam.
further developed by (1) incorporating detailed simulation Water Resour Res 18(6):1605–1613. doi:10.1029/
of crop water demands during the irrigation period; (2) WR018i006p01605
evaluating costs for reducing irrigation water supply in a Rosegrant MW, Ringler C, McKinney DC, Cai X, Keller A, Donoso
drought year, and compare this with the benefits of main- G (2000) Integrated economic-hydrologic water modeling at the
basin scale: the Maipo river basin. Agric Econ 24:33–46. doi:10.
taining a dependable discharge for hydropower; and (3) 1111/j.1574-0862.2000.tb00091.x
applying the methods of this study to other regions with Tilmant A, Kelman R (2007) A stochastic approach to analyze trade-
different hydrological conditions to determine whether the offs and risks associated with large-scale water systems. Water
conclusions remain viable. Resour Res 43:W06425. doi:10.1029/2006WR005094
Ueda T, Goto M, Namihira A, Hirose Y (2013) Perspectives of small-
scale hydropower generation using irrigation water in Japan. Jpn
Agric Res Q 47(2):135–140
UKERC (2006) The costs and impacts of intermittency: an assess-
References ment of the evidence on the costs and impacts of intermittent
generation on the British electricity network. The UK Energy
ANRE (2010) Energy in Japan. The Agency of Natural Resources and Research Centre, Imperial College, London. http://www.ukerc.
Energy, Japan. http://www.enecho.meti.go.jp/topics/energy-in- ac.uk/Downloads/PDF/06/0604Intermittency/0604Intermitten
japan/english2010.pdf Accessed 1 June 2013 cyReport.pdf Accessed 1 June 2013
Cai X, McKinney DC, Lasdon LS (2003) Integrated hydrologic– US Army Corps of Engineers (1985) Hydropower, Engineer Manual
agronomic–economic model for river basin management. No. 1110-2-1701

123

You might also like