Sunteți pe pagina 1din 10

J. Inst. Eng. India Ser.

A (May–July 2012) 93(2):111–120


DOI 10.1007/s40030-012-0013-8

ORIGINAL CONTRIBUTION

Optimal Reservoir Operation for Hydropower Generation using


Non-linear Programming Model
R. Arunkumar • V. Jothiprakash

Received: 2 January 2012 / Accepted: 23 July 2012 / Published online: 13 September 2012
Ó The Institution of Engineers (India) 2012

Abstract Hydropower generation is one of the vital Keywords Hydropower generation  Optimization 
components of reservoir operation, especially for a large Non-linear programming model 
multi-purpose reservoir. Deriving optimal operational rules Koyna hydroelectric project
for such a large multi-purpose reservoir serving various
purposes like irrigation, hydropower and flood control are
complex, because of the large dimension of the problem Introduction
and the complexity is more if the hydropower production is
not an incidental. Thus optimizing the operations of a Hydropower is one of the clean and green potential sources
reservoir serving various purposes requires a systematic for meeting the growing energy needs of a country. The main
study. In the present study such a large multi-purpose advantages of hydropower are the absence of pollution
reservoir, namely, Koyna reservoir operations are opti- during operation, its capability to respond quickly to
mized for maximizing the hydropower production subject changing demands, and its relatively low operating costs.
to the condition of satisfying the irrigation demands using a India is endowed with rich hydropower potential and it ranks
non-linear programming model. The hydropower produc- fifth in the world in terms of usable potential. In India, most
tion from the reservoir is analysed for three different of the reservoirs are having hydropower plants. However,
dependable inflow conditions, representing wet, normal only less than 25 % of the total possible potential had been
and dry years. For each dependable inflow conditions, developed or taken up for development [1] and the existing
various scenarios have been analyzed based on the con- hydropower plants are not operated to its full potential due to
straints on the releases and the results are compared. The several reasons. Some of them are (i) hydropower is pro-
annual power production, combined monthly power pro- duced through irrigation release in most of the reservoir;
duction from all the powerhouses, end of month storage (ii) hydropower and irrigation are having conflicting
levels, evaporation losses and surplus are discussed. From objectives and (iii) the disputes in sharing the available
different scenarios, it is observed that more hydropower water among different stakeholders. As per Asian Devel-
can be generated for various dependable inflow conditions, opment Bank (ADB) [1] report, nearly 78 % of the potential
if the restrictions on releases are slightly relaxed. The study remains without any plan for exploitation. Thus in the
shows that Koyna dam is having potential to generate more present scenario, there is a need to exploit the full potential
hydropower. of the existing hydropower plants and to optimize their
operations for maximizing the hydropower generation. In
India, most of the reservoirs are serving multiple purposes,
even if it was planned for a specific purpose at the time of
construction. Deriving the optimal operational rules for
large multi-purpose reservoirs serving various purposes
R. Arunkumar (&)  V. Jothiprakash, Fellow
such as irrigation, hydropower and flood control are com-
Department of Civil Engineering, Indian Institute of Technology
Bombay, Powai, Mumbai 400 076, India plex, because of the large dimension of the problem and
e-mail: arunkumar.r@iitb.ac.in conflicting objectives. Optimizing the operational plans of

123
112 J. Inst. Eng. India Ser. A (May–July 2012) 93(2):111–120

such reservoirs serving both the irrigation and hydropower peak algorithm. The scientists have developed the optimal
are also complex, if the hydropower production is not an operation policies for a multi-reservoir hydropower system
incidental. Hydropower requires more head to be available in Maine, USA using the NLP technique and reported that
in reservoir for efficient operation, where as crop production the model allows the decision makers to adjust the opera-
requires more irrigation release. This calls for an efficient tions for different alternative operation scenarios [21].
and effective operation of reservoirs for the development of Some researchers have applied the mixed integer NLP for
the country in agricultural as well as power sector. short-term optimal operation of hydraulically linked
Optimizing the operations of a large multi-purpose hydropower reservoirs in Manitoba, Canada [22]. In their
reservoir requires a systematic study [2]. Several optimi- study, the hydraulic coupling between the cascades of
zation techniques have been applied to derive the optimal hydropower plants was established using the tail water
reservoir operational rules like linear programming (LP), elevation curve. Previous researchers have developed a
non-linear programming (NLP), goal programming (GP), monthly NLP optimization model, called SISOPT for the
chance constraint linear programming (CCLP), dynamic management and operations of the Brazilian hydro-power
programming (DP) and recently, the soft computing tech- system [14]. The formulated NLP model was first linear-
niques. Detailed reviews about these optimization tech- ized by two different linearization techniques and solved
niques are also available in the literature [3–5]. Even by LP method. It was reported that the LP model could be
though various techniques are available, the complexity of used in system capacity expansion studies, where storage
the model depends on the case study and its associated variation is not as important as the power production.
problems. Optimizing hydropower systems are more However, the NLP model performed better when checked
complex because of their non-linear objective functions with the historical operational records and concluded that
and constraints. However, the non-linear problem can be the NLP model is the most accurate and practically suitable
converted to linear through approximation and can be solved for real-time operation. A decomposition-coordination
using LP technique. Various studies have been reported for approach was applied by Ailing [23] to a multiple hydro-
optimizing hydropower systems using different techniques, electric reservoir systems of Yellow River, China.
for example, CCLP [6], mixed integer programming (MIP) After investigation, the scientists have developed a NLP
[7], stochastic dynamic programming (SDP) [8], bayesian model for a multi-reservoir system in upper Krishna basin
stochastic dynamic programming (BSDP) [9]. A review on for maximizing the irrigation, municipal and industrial
various techniques used for optimizing the hydropower releases and hydropower production [17]. On comparing the
production can be found from previous literature [10, 11]. results with the LP model, it was reported that the NLP
Among several techniques, NLP is widely applied for model resulted in less irrigation deficit with more hydro-
optimizing hydropower systems [12, 13], since it is most power production. The decomposition was done on the non-
accurate, involves no approximation and uses the physi- linear hydropower problem using linear approximation for
cally based non-linear energy production function [14]. optimal short-term hydropower scheduling of Multi-reser-
NLP technique is also used for many other applications voir system [24]. A non-linear multi-objective optimization
such as, optimal cropping pattern [15], multi-purpose res- model was developed by earlier researchers to optimize the
ervoir optimization [16], multi-reservoir optimization [17] annual scheduling of power generation in serial or parallel
and optimal design and operation of pumping stations [18] hydropower plants for dry, medium and wet scenarios [25].
apart from hydropower optimization. The multiple objectives were converted to a single objective
The researchers have developed a reliability model for using weighted sum method. It was reported that the wet
optimizing the operations of a multi-purpose reservoir for scenario resulted in increased amount of energy production
hydropower generation and flood control [19]. The model due to high inflow. A monthly mathematical model, called
was solved using a three level algorithm, which includes a HIDROTERM, was developed by the researchers to opti-
non-linear search, a separable programming and a LP. mize the management and operation of large hydrothermal
Earlier, the researchers have already compared the two systems [26, 27]. This study was further extended for more
non-linear methods namely, augmented price method R&D purposes [28, 29]. The researchers have compared the
(APM) and sequential quadratic programming (SQP) for two optimization models namely equivalent reservoir opti-
optimizing a large scale hydropower plant in Zambezi river mization model (EROM) and operational optimization
system, Africa [20]. Both the algorithms were applied for model (OPOM) for optimal hydropower production of São
different time horizons and inflow records. The researchers Francisco River Hydroelectric System, Brazil [30]. It was
have developed a non-linear optimization model for a reported that the OPOM procedure was more accurate than
multi-purpose reservoir operation [16]. It was reported that EROM, since power plants and reservoirs were represented
the developed model successfully integrated the behaviour as individual items of the system. However, when a quick
analysis algorithm, automatic differentiation and sequent estimate of the potential power production is needed, the

123
J. Inst. Eng. India Ser. A (May–July 2012) 93(2):111–120 113

EROM approach is a simpler alternative to model the sys- the first base load hydel station with 4 9 70 MW capacity.
tem. All these studies show that NLP technique provides The second stage (stage II) was designed to harness the
better results for hydropower optimization. It is becoming capacity of 4 9 80 MW. For these two stages the dam,
increasingly important to optimize the operations of reser- headrace tunnel, surge well pressure shafts and tailrace are
voirs for effective, efficient and economical management common and hence referred as single powerhouse PH I in
and operations for various purposes. Among the various this study. Thus, the combined total capacity of both the
purposes of a reservoir, hydropower plants play a significant stages is 600 MW. The full installed capacity of PH I
role in supplying the national power grid, especially the high (stage I and II) supplying base power to the grid was found
valued peak demand loads. Therefore, planning and opera- insufficient to cope up with the peak demand of the state.
tion of hydropower reservoirs is more focused on peak load Hence, it was decided to meet the demand during peak
power generation. Hence, in the present study it is aimed to hours by converting the hydropower stations into peaking
optimize the hydropower generation of a large scale stations. This has given rise to stage IV (hereafter referred
hydropower plant using NLP model. The present study also as PH II) with a capacity of 4 9 250 MW in KHEP. Apart
focuses on the effect of different inflow conditions for var- from power generation, there is also a need to release for
ious operating scenarios in peak hydropower production. irrigation and downstream riparian rights on eastern side,
Thus, three different dependable inflow levels such as 50 % which has fertile land as compared to barren exposed rock
(wet), 75 % (normal) and 90 % (dry) are considered to study covers, and undulating terrain on western side. Hence, the
the performance of the developed NLP model. The results of Koyna Dam Power House (KDPH) (hereafter referred as
this study can be used as initial trajectory and the results can PH III) was constructed with a capacity of 2 9 20 MW to
be further improved, and also provides a basis for develop- generate hydropower through the irrigation releases. The
ing operational plans using soft computing techniques. stage III is at Kolkewadi dam with a capacity of
4 9 80 MW. All the releases required for this stage III is
made from Kolkewadi dam and receives most of its inflow
Study Area from the tailrace water from PH I and PH II of the Koyna
dam. In the present study, only the Koyna dam power-
The Koyna Hydro Electric Project (KHEP) in Maharashtra, houses are considered and the total installed capacity of
India is considered as the case study. The KHEP is the Koyna dam considered is 1,640 MW (PH I, PH II and PH
lifeline of Maharashtra, which has four stages to a total III). The power production at Koyna dam is not continu-
capacity of 1,960 MW [31]. The Koyna reservoir situated on ous, since they are all peaking stations. They are operated
the west coast of Maharashtra, India, alone has three pow- only during the peak hours i.e. during the morning and
erhouses, two on the western side and one at the dam foot on evening where there is a peak demand. It is also to be
the eastern side of the reservoir. The location of Koyna dam noted that the power generation at PH III in the eastern
and its powerhouses is shown in Fig. 1. The Koyna stage I is side of the dam is incidental, since it is through irrigation

Fig. 1 Location of powerhouses of Koyna dam

123
114 J. Inst. Eng. India Ser. A (May–July 2012) 93(2):111–120

release, and the major powerhouses release are to be made The release for irrigation should be greater than or equal
in the western side of the dam. to the demand for irrigation during the time period ‘t’.
R3;t  IDt t ¼ 1; 2; . . .12 ð5Þ
Model Formulation where R3;t is the eastward irrigation release from Koyna
Dam during the time period t and IDt is the demand for
The capacity of a hydropower plant is primarily the func- irrigation during the time period t.
tion of the hydraulic head and the flow rate through the The diversion of huge quantity of water to the western
turbines. According to Loucks et al. [2], the hydropower side for power production has resulted in series disputes.
production during any time period t for any reservoir is To ensure adequate water for irrigation on eastern side and
dependent on the installed plant capacity, flow through the other downstream rights, the westward diversion was
turbines, average effective storage head, number of hours restricted to certain limit [31]. Therefore, diversion of large
operation, plant factor (the fraction of time that energy is quantity of water to westward for power production was
produced) and a constant for converting the product of restricted to 1,912 9 106 m3 and the irrigation release
flow, head and plant efficiency to electrical energy. Thus, should be 850 9 106 m3. This Krishna water tribunal
the hydropower produced [32] (PHt) in terms of kilowatt- constraint is given as:
hours (kWh) during any time period t is expressed as
X
12
PHt ¼ 2725  Rt  Ht  g ð1Þ ðR1;t þ R2;t Þ  Rw;max ð6Þ
t¼1
where, Rt is the release to powerhouse during the time period
t, Ht is the net head available during the time period t and g is X
12
R3;t  IDmax ð7Þ
the plant efficiency. In the present study, the average head t¼1
available during a particular time period t is expressed as
function of storage. Then the net head is estimated by where, Rw;max is the maximum annual volume of water that
deducting the tail water level and the frictional losses. can be diverted to the western side for power production
The objective of the present study is to maximize the and IDmax is the maximum annual volume of water to be
hydropower production from all the three powerhouses and released for irrigation in the eastern side.
it is expressed as: The reservoir storage St during any time period t should
not be less than the minimum storage (Smin ) or dead storage
X
12
and should not be more than maximum storage (Smax ) or
Max Z ¼ PH It þ PH IIt þ PH IIIt ð2Þ
t¼1
capacity of the reservoir. This physical constraint is given by:

where, PH It ; PH IIt and PH IIIt is the power produced from Smin  St  Smax t ¼ 1; 2; . . .12 ð8Þ
PH I, PH II and PH III, respectively during the time period The evaporation loss (Et) during the period t is
t in terms of kWh. The above objective function is sub- expressed as a function of initial and final storage [33]
jected to various constraints. during that particular time period. A regression equation
The head available in the reservoir should be greater has been developed between the monthly evaporation loss
than the minimum drawdown level of the powerhouse for and the average storage in that month. The regression
any time period t. This is expressed as equation is expressed as:
Hn;t  MDDLn;t t ¼ 1; 2; . . .12; n ¼ 1; 2; 3 ð3Þ  
St þ Sðtþ1Þ
Et ¼ a t þ b t t ¼ 1; 2; . . .12 ð9Þ
where Hn;t is the head (m) in the reservoir for the 2
powerhouse n during the time period t and MDDLn;t is the
where, at and bt are the regression constants during the time
allowable minimum drawdown level for the powerhouse n
period t estimated by regression analysis.
during the time period t. The power production during any
For any given time period t, the final storage in the res-
time period t should be less than or equal to the maximum
ervoir should be equal to the inflow, releases and other losses
power generating capacity of the power plant
of the system. This mass balance constraint is given as:
PHn;t  P maxn;t P maxn;t ¼ 1; 2; . . .12; n ¼ 1; 2; 3
X
3
ð4Þ Sðtþ1Þ ¼ St þ It  Rn;t  Ot  Et
n¼1
ð10Þ
where PHn;t is the power produced (kWh) in the power-
t ¼ 1; 2; . . .12; n ¼ 1; 2; 3
house n during the time period t; Pmaxn;t is the maximum
capacity of power generation (kWh) for the powerhouse n where, Sðtþ1Þ is the final storage in the reservoir during the
during the time period t. time period t (106 m3); St is the initial storage in the

123
J. Inst. Eng. India Ser. A (May–July 2012) 93(2):111–120 115

reservoir during the time period t (106 m3); It is the inflow minimum are almost same which shows that the data is near
into the reservoir during the time period t (106 m3); Rn;t is normally distributed. From the analysis, it is observed that
the release to the power house n during the time period t the reservoir receives 95 % of the inflow during the mon-
(106 m3); Ot is the overflow from the reservoir during the soon season (5 months) and the remaining 5 % during the
time period t (106 m3) and Et is the evaporation losses from non-monsoon period (7 months). Thus, the reservoir is
the reservoir during the time period t, (106 m3). completely dependent on monsoon inflow. Therefore, it is
The overflow occurs when the final storage exceeds the very much important to store the water to meet the unex-
reservoir capacity. This overflow constraint is given by: pected demands during the non-monsoon periods. Hence, in
real-life the reservoir is always maintained with good stor-
Ot ¼ Sðtþ1Þ  Smax t ¼ 1; 2; . . .12 ð11Þ age. The initial storage for optimizing the models is taken as
and Ot  0 t ¼ 1; 2; . . .12 ð12Þ corresponding dependable inflow year starting month stor-
age observed in the reservoir. The Koyna reservoir evapo-
where, Smax is the capacity of the reservoir during time ration is expressed as the function of average storage [33]
period t (106 m3) and Sðtþ1Þ is the final storage in the res- (initial and final storage) of the reservoir during the partic-
ervoir during time period t (106 m3). This final storage is ular time period t, as showed in Eq. 9. The constants in the
initial storage for the next time period t ? 1, when there is equation at and bt are estimated through regression analysis
no overflow. If overflow occurs then Smax will be the initial of the evaporation during a particular time period against the
storage for the next time period t ? 1 for the reservoir. average storage during that time period using 49 years of
observed data. The values of the regression constants for
various time periods are given in Table 1.
Based on the constraints of the above formulated NLP
Results and Discussion
model, several scenarios are analyzed for each dependable
inflow considering with and without restriction on releases
In the present study, the operational plans of Koyna reservoir
to the westward and eastward sides. These different sce-
are optimized for maximizing the hydropower production
narios will be helpful in assessing the full potential of the
using a monthly time step NLP model. The above formu-
system and are given as:
lated NLP model is solved using LINGO/Global solver [34].
The developed NLP model is optimized for 50 % (wet), Scenario 1: No eastward and westward binding
75 % (normal) and 90 % (dry) dependable inflows, esti- constraints on releases are considered
mated by Weibull’s method [35] from 49 years of inflow Scenario 2: Only annual binding constraint on irrigation
data. The box and whisker plot of Koyna reservoir inflow is (eastward) releases is considered
shown in Fig. 2. From the figure, it can be observed that the Scenario 3: Both monthly and annual binding constraint
average inflow is higher during monsoon periods. The on irrigation (eastward) releases is considered
maximum inflow is observed in the month of August. It is Scenario 4: Both westward and eastward binding
also observed that some outliers are found in few months. constraints on releases are considered as per
During the non-monsoon periods, the mean, maximum and the Krishna tribunal [31]

Table 1 Estimated values of evaporation constants for different


months
Month at bt

June 3.00 0.0074


July 3.06 0.0068
August 3.62 0.0068
September 3.43 0.0069
October 3.38 0.0076
November 3.01 0.0076
December 3.01 0.0078
January 3.37 0.0075
February 3.48 0.0066
March 4.35 0.0068
April 4.42 0.0064
Fig. 2 Box and whisker plot of Koyna reservoir inflow for different May 4.07 0.0073
time period

123
116 J. Inst. Eng. India Ser. A (May–July 2012) 93(2):111–120

substantially reduced the releases on western side and the


The results of the developed NLP model for these four annual power production is also decreased from scenario 1
scenarios under different dependable inflow conditions are for all the inflow conditions. The reduction in power pro-
discussed in the following section. duction is also due to that the powerhouses on the western
side are having high capacity and net head than PH III.
Annual Power Production Since, the irrigation releases are not made in all the months
in the above scenario 2, both the irrigation release con-
The resulted annual hydropower production from various straints (Eq. 6 is not considered) are taken into account in
scenarios for different dependable inflow conditions are scenario 3. The additional irrigation constraint has forced
given in Table 2. From the table it can be seen that the the model to release for irrigation every month as per the
power production decreases with increase in the inflow monthly demand. The maximum annual power production
dependability. This is due the fact that the quantity of in this scenario is 3956.13 9 106 kWh for 50 % dependable
inflow decreases with the increase in the dependability inflow. However, there is a slight reduction in power pro-
level. Also, the various scenario shows that the power duction compared to scenario 2, due to the variation in
production decreases with the increase in the irrigation monthly releases and storage levels. In all the above sce-
release. Since, major powerhouses are in western side and narios, there is no restriction on westward release for power
the irrigation releases are on the eastern side of the reser- production and hence the total release on westward is more
voir. This also shows that the power production and the the tribunal limit. In order to restrict the release on the
irrigation are having conflicting objectives in this case. In westward as per the tribunal, in scenario 4 all the binding
scenario 1, no restrictions (Eqs. 5, 6 and 7 are not con- constraints (Eqs. 5, 6 and 7) on releases are considered. The
sidered) are considered on releases either on westward or maximum annual power production in this scenario is
eastward. In scenario 1, the 50 % dependable inflow has 2685.58 9 106 kWh for 50 % dependable inflow. Since the
produced a maximum of 4824.48 9 106 kWh hydropower, releases are restricted to less than the capacity of major
better than the other inflow series. However, the irrigation powerhouses, the power production has been considerably
release on the eastern side is very less and also for only decreased. Also, in scenario 4 the variation in power pro-
three months. For this scenario, the model has not resulted duction among the different inflow conditions is less com-
in any irrigation releases to the eastern side for 75 and pared to other scenarios. For 90 % dependable inflow there
90 % dependable inflows. This may be due to that no is no variation in power production among the scenario 3 and
binding constraints are considered on the irrigation relea- 4. This shows that the when the inflow is very low, the
ses. This shows that the model tend to release more for constraints on releases does not affect the power production.
power production than irrigation. Since there is no release On the other hand, this has resulted in increased storage in
for irrigation on the eastern side, there is no power pro- the reservoir and also overflows in few months.
duction from PH III for all the three inflow conditions.
In order to achieve irrigation release, in scenario 2, the Combined Monthly Power Production
annual irrigation release constraint (Eqs. 5 and 6 are not
considered) is considered. The maximum annual hydro- The total monthly power production from all the three
power production in scenario 2 is 4332.28 9 106 kWh is powerhouses for 50 % dependable inflow condition is
also for 50 % dependable inflow. In this scenario, the model given in Fig. 3. From the figure, it is observed that all the
has resulted in the annual irrigation release as per the tri- scenarios, except scenario 4, have resulted in maximum
bunal for all the inflow conditions, but not for all the time production during monsoon period due to high inflow and
periods. Only few months have irrigation release. This has storage. For most of the months, the scenarios 1 and 2 have
produced almost equal power production. The scenario 1
has resulted in good power production for all months
Table 2 Annual hydropower production from various scenarios for expect May. This may be due to the less storage at the end
different inflow conditions
of the season. The scenario 4 has constant power produc-
Scenario Annual hydropower production for various dependable tion for most of the monsoon months and much lesser than
inflow levels (106 kWh)
other scenario due to the restriction in releases on either
50 % 75 % 90 % side. This shows that the scenario 4 keeps storing the water
during monsoon and hence during non-monsoon seasons
1 4824.48 4300.68 3535.64
the power production is higher.
2 4332.28 3294.05 2520.24
The combined monthly power production from all the
3 3956.13 3268.42 2493.76
three powerhouses for various scenarios of 75 % depend-
4 2685.58 2673.25 2493.76
able inflow is given in Fig. 4. From the figure, it is

123
J. Inst. Eng. India Ser. A (May–July 2012) 93(2):111–120 117

Fig. 3 Monthly power production from various scenarios for 50 % Fig. 5 Monthly power production from various scenarios for 90 %
dependable inflow dependable inflow

shows that even under unrestricted flow on westward, the


system is not able to produce hydropower due to the less
inflow.

Storage Curves

The resulted end of month storage levels from various


scenarios for 50 % dependable inflow condition is given in
Fig. 6. From the figure, it can be seen that all the scenarios
have resulted in maximum storage at least in one month.
Even though all the scenarios have resulted in maximum
storage, the overflow occurred only in scenario 4 during the
month of August for 50 % dependable inflow. This is due
to that, in scenario 4, the releases are restricted strictly to
the tribunal limits and this has increased the storage in the
Fig. 4 Monthly power production from various scenarios for 75 %
dependable inflow

observed that initially the power production is same for all


the scenarios and the power production increases with
increase in the inflow for scenario 1, 2 and 3. Like 50 %
dependable inflow condition, the power production from
scenario 4 is constant during the monsoon season and there
is high variation during non-monsoon season. Also, there is
a substantial power production during the non-monsoon
periods where there will be peak demand. Figure 5 shows
the combined monthly power production from various
scenarios for 90 % dependable inflow. Unlike other two
inflow conditions, the power production from 90 %
dependable inflow is constant at the start and end of the
season for all the scenarios. It is observed that the overall
monthly maximum power production is lesser than 50 and
75 % dependable inflow. Also, the scenarios 3 and 4 have Fig. 6 Resulted end of month storage from various scenarios for
almost produced same hydropower in all the months. This 50 % dependable inflow

123
118 J. Inst. Eng. India Ser. A (May–July 2012) 93(2):111–120

reservoir leading to overflow. However, there is no over-


flow in scenario 4 for 75 and 90 % dependable inflows,
since the inflow is decreased. For all the scenarios, the
storage during the non-monsoon period is decreasing since
there is very less inflow and only the carry over storage is
available for release. The resulted final storage for different
time periods from various scenarios for 75 % dependable
inflow is given in Fig. 7. For both the inflow condition,
except scenario 4, all other scenarios have resulted in a
similar trend in final storage. From the figure, it can be seen
that the resulted final storage for scenario 1, 2 and 3 is
much lesser than the scenario 4. This is due to the increased
release in all the scenarios except scenario 4. It is to be
noted that in scenario 4, due to restriction, the release is
very less and subsequently the storage is more. Also, the
difference in maximum storage between scenario 4 and Fig. 8 Resulted end of month storage from various scenarios for
other scenarios is very high for 75 % dependable inflow. 90 % dependable inflow
Thus in case of 50 % dependable inflow, the surplus is over
and above the complete utilization of the powerhouses, Fig. 8. From the figure, it is observed that all the scenarios
whereas in case of 75 % dependable inflow, the flow is have resulted almost same maximum storage during the
underutilized and resulted in more storage. The available month of August. It is also observed that the storage levels
storage in the reservoir shows that still more hydropower for scenario 3 and 4 are same for all the months. The
can be generated in all the scenarios during the monsoon resulted storage is much less than 50 and 75 % dependable
season. During monsoon period, the variation in storage inflow condition. Unlike other inflow condition, the final
level is very less for all the scenarios. However, during the storage at the end of the season is same as initial storage for
non-monsoon period, the variation is very high. Also for all all the scenarios. It is also seen that there are no surplus for
the scenarios, except scenario 4, the final storage at the end 90 % dependable inflow level, since the inflow is less and
of the season is same as the initial storage thus representing fully utilized for power production. This shows that the
the steady state policy. However, in scenario 4, due to release constraint restricted the power production half of
restriction on releases, the model keeps on storing the the possible power production. Instead of producing power,
water resulting in excess storage at the end of the season. huge storage is maintained in the reservoir. Thus, it may
The resulted end of month storage levels from various concluded that there is still more power production
scenarios for 90 % dependable inflow condition is given in potential from Koyna dam.

Fig. 7 Resulted end of month storage from various scenarios for Fig. 9 Resulted evaporation from various scenarios for 50 %
75 % dependable inflow dependable inflow

123
J. Inst. Eng. India Ser. A (May–July 2012) 93(2):111–120 119

Evaporation Loss The resulted evaporation from various scenarios for


90 % dependable inflow condition is given in Fig. 11.
The evaporation loss during a particular time period is Contrasting to the above inflow conditions, the variation in
expressed as a function of initial and final storage in the evaporation loss is very less among different scenarios for
reservoir during that particular time period. The resulted 90 % dependable inflow condition.
evaporation loss from various scenarios of power produc-
tion for 50 and 75 % dependable inflow condition is shown
in Figs. 9 and 10, respectively. From the figures, it can be Conclusion
seen that the evaporation has followed a similar trend of
storage. This shows that the storage in the reservoir plays a In the present study, the Koyna reservoir operations are
significant role in reservoir evaporation, since the water optimized for maximizing hydropower production satisfying
spread area of the Koyna reservoir is very high. The the irrigation demands using a NLP model. Among different
evaporation for scenario 4 is much higher than all other dependable inflow conditions, 50 % dependable level has
scenarios for the both inflows, since the storage is also resulted in more power production due to the high inflow.
higher for scenario 4. The model resulted in a maximum power production of
4824.48 9 106 kWh when there is no restriction in releases.
From the study, it is observed that the increase in release for
irrigation has decreased the power production considerably.
The comparison of monthly power production shows that
during the monsoon seasons the power production is higher
than the other months. This is due the storage and the high net
head available during that period. This study shows that the
power that can be produced with all the release constraints is
only 55 % of the power production potential of Koyna res-
ervoir. The developed model has also resulted in good stor-
age at the end of the season and available storage can be
further utilized for more hydropower generation. The study
shows that Koyna dam is having more potential to generate
more hydropower. Further, the results of this study can be
used as initial trajectory and provides the basis for the other
methods like recent soft computing techniques through
which the results can be further improved.
Fig. 10 Resulted evaporation from various scenarios for 75 %
dependable inflow
Acknowledgments The authors gratefully acknowledge the Minis-
try of Water Resources, Government of India, New Delhi, for spon-
soring this research project through the Indian National Committee on
Hydrology. The authors also thank the people who concern related to
Koyna Hydroelectric Project and Koyna Dam for providing necessary
data.

References

1. ADB. Hydropower development in India: a sector assessment.


Asian Development Bank, Publication Stock No. 031607, Phil-
ippines, 2007
2. D.P. Loucks, J.R. Stedinger, D.A. Haith. Water Resources Sys-
tems Planning and Analysis. (Prentice Hall Inc., Englewood
Cliffs, 1981)
3. W.W.G. Yeh. Reservoir management and operation models: a
state-of-the-art review. Water Resour. Res. 21(12), 1797–1818
(1985)
4. R. Wurbs. Reservoir-system simulation and optimization models.
J. Water Resour. Plan. Manag. ASCE 119(4), 455–472 (1993)
5. D. Rani, M.M. Moreira. Simulation–optimization modelling: a
Fig. 11 Resulted evaporation from various scenarios for 90 % survey and potential application in reservoir systems operation.
dependable inflow Water Resour. Manag. 24(6), 1107–1138 (2009)

123
120 J. Inst. Eng. India Ser. A (May–July 2012) 93(2):111–120

6. K. Sreenivasan, S. Vedula. Reservoir operation for hydropower 22. R.S.V. Teegavarapu, S.P. Simonovic. Short-term operation model
optimization: a chance-constrained approach. Sadhana 21, for coupled hydropower reservoirs. J. Water Resour. Plan.
503–510 (1996) Manag. ASCE 126(2), 98–106 (2000)
7. J. Yi. Mixed integer programming approach to optimal short-term 23. L. Ailing. A study on the large-scale system decomposition—
unit commitment for hydropower systems. KSCE J. Civ. Eng. coordination method used in optimal operation of a hydroelectric
2(3), 335–346 (1998) system. Water Int. 29(2), 228–231 (2004)
8. B. Zahraie, M. Karamouz. Hydropower reservoirs operation: a 24. S. Liu, J. Wang, Z. Liu. Short-term hydropower optimal sched-
time decomposition approach. Sci. Iran. 11, 92–103 (2004) uling of multireservoir system using a decomposition approach.
9. P.P. Mujumdar, B. Nirmala. A bayesian stochastic optimization Fourth International Conference on Natural Computation, IEEE,
model for a multi-reservoir hydropower system. Water Resour. 2008, pp. 565–570
Manag. 21(9), 1465–1485 (2007) 25. S.A.A. Moosavian, A. Ghafari, A. Salimi, N. Abdi. Non-linear
10. J.A. Momoh, M.E. El-Hawary, R. Adapa. A review of selected multiobjective optimization for control of hydropower plants
optimal power flow literature to 1993 part I: nonlinear and qua- network. ASME International Conference on Advanced Intelli-
dratic programming approaches. IEEE Transactions on Power gent Mechatronics, 2008, pp. 1278–1283
Systems 14(1), 96–104 (1999a) 26. M.T.L. Barros, R.C. Zambon, J.E.G. Lopes, P.S.F. Barbosa,
11. J.A. Momoh, M.E. El-Hawary, R. Adapa. A review of selected A.L.F. Francato, W. Yeh. Model to optimize large hydrothermal
optimal power flow literature to 1993 part II: Newton, linear system operation considering water and environment sustain-
programming and interior point methods. IEEE Trans. Power ability. Proceedings of World Environmental & Water Resources
Syst. 14(1), 105–111 (1999b) Congress, Honolulu, 2008
12. C.R. Gagnon, R.H. Hicks, S.L.S. Jacoby, J.S. Kowalik. A non- 27. M.T.L. Barros, R.C. Zambon, J.E.G. Lopes, P.S.F. Barbosa,
linear programming approach to a very large hydroelectric sys- A.L.F. Francato, W. Yeh. Impacts of the upstream storage res-
tem optimization. Math. Program. 6, 28–41 (1974) ervoirs on itaipu hydropower plant operation. Proceedings of the
13. J.A. Tejada-Guibert, J.R. Stedinger, K. Staschus. Optimization of World Environmental & Water Resources Congress 2009, Kansas
value of CVP’s hydropower production. J. Water Resour. Plann. City, 2009
Manag. ASCE 116(1), 52–70 (1990) 28. R.C. Zambon, M.T.L. Barros, P.S.F. Barbosa, A.L. Francato,
14. M.T.L. Barros, F.T.C. Tsai, S.L. Yang, J.E.G. Lopes, W.W.G. Yeh. J.E.G. Lopes, W.W.G. Yeh. Planning Operation of Large-Scale
Optimization of large-scale hydropower system operations. J. Water Hydrothermal System. World Environmental and Water
Resour. Plan. Manag. ASCE 129(3), 178–188 (2003) Resources Congress 2011: Bearing Knowledge for Sustainability,
15. G. Paudyal, A.D. Gupta. A nonlinear chance constrained model 2011, pp. 3026–3035
for irrigation planning. Agric. Water Manag. 18(2), 87–100 29. R.C. Zambon, M.T.L. Barros, J.E.G. Lopes, P.S.F. Barbosa,
(1990) A.L. Francato, W.W.G. Yeh. Optimization of large-scale hydro-
16. A. Sinha, B. Rao, C. Bischof. Nonlinear optimization model for thermal system operation. J Water Resour. Plan. Manag. 138(2),
screening multipurpose reservoir systems. J. Water Resour. Plan. 135–143 (2012)
Manag. ASCE 125(4), 229–233 (1999) 30. J.L.B. Brandão. Performance of the equivalent reservoir model-
17. M.G. Devamane, V. Jothiprakash, S. Mohan. Non-linear pro- ling technique for multi-reservoir hydropower systems. Water
gramming model for multipurpose multi-reservoir operation. Resour. Manag. 24(12), 3101–3114 (2010)
Hydrol. J. 29(3–4), 33–46 (2006) 31. Koyna Hydro Electric Project Stage—IV, Irrigation Department,
18. M. Moradi-Jalal, M.A. Mariño, A. Afshar. Optimal design and Government of Maharashtra
operation of irrigation pumping stations. J. Irrig. Drain. Eng. 32. S. Vedula, P.P. Mujumdar. Water Resources systems: Modelling
ASCE 129(3), 149–154 (2003) Techniques and Analysis. (Tata McGraw-Hill Publishing Com-
19. S.P. Simonovic, R. Srinivasan. Explicit stochastic approach for pany Ltd, New Delhi, 2005)
planning the operations of reservoirs for hydropower production. 33. V. Jothiprakash, R. Arunkumar, A.A. Rajan. Optimal crop
Extreme Hydrological Events: Precipitation, Floods and Droughts, planning using a chance constrained linear programming model.
Proceedings of the Yokohama Symposium, 1993, pp. 349–349 Water Policy 13(5), 734–749 (2011)
20. E. Arnold, P. Tatjewski, P. Wołochowicz. Two methods for 34. Lingo User Guide. (Lindo systems, Inc., Illinois, 2011), p. 834
large-scale nonlinear optimization and their comparison on a case 35. K. Subramanya. Engineering Hydrology. 3rd edn. (Tata
study of hydropower optimization. J. Optim. Theory Appl. 81(2), McGraw-Hill Education Pvt. Ltd., 2008), p. 452
221–248 (1994)
21. C.S. Peng, N. Buras. Dynamic operation of a surface water
resources system. Water Resour. Res. 36(9), 41–60 (2000)

123

S-ar putea să vă placă și