Академический Документы
Профессиональный Документы
Культура Документы
Abstract — In recent years, a substantial amount of originally for hot reserve of the power system and load
photovoltaic (PV) generations have been installed in power leveling. The energy is stored in the form of water
systems. However, the power output from the PV is random
and intermittent in nature. Therefore, the PV generation pumped up from a lower reservoir to an upper reservoir.
poses many challenges to the power system operation. To The PSHPP is usually operated as a generator in the
evaluate the impact of the behavior of PV on power supply daytime and a pump in the nighttime. A new operation
reliability, we have developed a power supply reliability scheduling of PSHPP is proposed to solve the problem in
evaluation model considering a large integration of PV this paper. It will be operated with the proposed pattern
generation into power system. As a result, the power supply
reliability is getting worse as the PV penetration level solving the surplus power problem caused by the PV
increases. To solve this issue, we have proposed that pumped generation. The authors have already proposed how the
storage hydro power plant (PSHPP) is used effectively to PSHPP operation pattern is changed effectively to
improve the reliability. In this paper a cooperative improve the power system reliability in case of a large
scheduling method for the PSHPP and thermal power plants integration of PV into the power system [1]. However,
that makes it possible to improve both reliability and
economy is presented. the scheduling method for obtaining the optimal PSHPP
operation pattern that makes it possible to improve both
Keywords — Monte Carlo Simulation, Optimal Scheduling, reliability and economy has not been developed yet. In
Pareto Optimal Solutions, Photovoltaic Generation, Pumped
Storage Hydro Power Plant, Surplus Power Problem this paper, we propose a new method for planning PSHPP
operation pattern considering the surplus power problem,
power supply reliability and fuel cost reduction, which is
I. INTRODUCTION represented by pareto optimal solutions. The power
In recent years, a large amount of photovoltaic (PV) supply reliability and the fuel cost are estimated for each
and wind power generations have been installed in power PSHPP operation plan by using Monte Carlo simulation.
systems around the world. However, Japan is not an
appropriate site to install a large amount of wind power II. SIMULATION MODEL
generations. Therefore, the Japanese government aims at
installing a large amount of PV in power system. A. Pumped Storage Hydro Power Plant
Since the power output from the PV is random and In PSHPP, the energy is stored in upper reservoir in
intermittent in nature, the PV generation poses many the form of water. Therefore, it should be operated within
challenges to the power system operation. When the load the capacity of the upper reservoir. In this paper, the
demand is very small and at the same time the PV amount of the stored water in the upper reservoir is
generates a large amount of power, the balance between considered as energy. It can be operated 6 to 8 hours
power supply and demand cannot be maintained. This continuously when the operation starts from the full
problem is called “surplus power problem". In addition to capacity of the upper reservoir. Therefore, the capacity of
the surplus power problem, it is also necessary to the upper reservoir is assumed to be 7-hour operation of
consider the intermittent and uncertain output the rated power at generator mode.
characteristic of PV. In general, there are two types of PSHPP, fixed speed
Therefore, various kinds of countermeasures have been one and adjustable speed one. The fixed speed PSHPP
considered when a substantial amount of PV generation is cannot adjust the input power when it operates as a pump.
installed into a power system. Though Battery Energy The adjustable speed one can adjust the input power. It is
Storage System (BESS) could be considered as a solution coming into the limelight because of its adjustable
characteristics in the pump mode [2]. Therefore, the
of these issues, their high cost is preventing them from
PSHPP is assumed to be an adjustable speed PSHPP in
being considered as a solution in most situations. This
this paper.
paper proposes an effective use of Pumped Storage
Hydro Power Plant (PSHPP). PSHPP is installed
100
90
80
PV Output [%]
70
60
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time
70 Rainy
60 the PV output and the load demand could fluctuate
50
40 randomly in Monte Carlo simulation. Therefore, the load
30
20
dispatch of generators is changed and its operational cost
10 rises or falls at each trial. In this paper, the weekly fuel
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 cost is defined as the mean value of fuel costs of all trials
Time
Fig. 3 PV Output for Three Kinds of Weather excluding the trial where the generator failure occurs.
This is because the fuel cost increase is the unexpected
The generator failure occurs during Monte Carlo factor due to the generator failure. In other words, the
simulation at the failure rate λ described by (3). In (3), outage cost is not determined uniquely.
MTBFi is the mean time between failure of generator i
and MTTRi is the mean time to repair of generator i
III. OPTIMAL OPERATION SCHEDULING
1 1 (3)
i , i
MTBFi MTTRi A. Pareto Optimal Solutions
During Monte Carlo simulation, if the generator failure In general, improvement of power system reliability
occurs at time t, the other generators are re-dispatched and reduction of operational cost are competing.
immediately. Figure 4 shows how the generators are re- Optimization of the PSHPP scheduling has no unique
dispatched when generator G3 is tripped at time t, for solution. This type of problem is a multi-objective
example. First, Generator G1, G2 and G4 are re- optimization problem. In this paper, pareto optimal
dispatched immediately. If imbalance of power supply solutions are obtained.
and demand cannot be avoided by the re-dispatching,
PSHPP is operated immediately to avoid it as an B. Quantization of PSHPP operation pattern
emergency control. In the next time t+1, the remaining
generators are committed to compensate for the The operation scheduling of PSHPP and thermal
unavailable generator. The tripped generator is restored at power plants are quantized as shown in Fig. 5. PSHPP
the restoration rate μ after time t+1.
operation pattern is quantized by 6 steps at each time t.
Generator Failure Generators Re-dispatch
Thermal power plants operation is represented as a
First Step Second Step number of surplus generators from the cost minimize case
Supply Demand Supply Demand
determined by dynamic programming [3].
Emergency
Shortage
Supply
Control
PSHPP
G4
G5
G4
・・・
G3
Load Demand
Load Demand
Load Demand
Load Demand
G4
Maximum Output
Maximum Output
G2
168hours
G2
1 2 3 4 5 6 7 8 9 10
G2
G2
・・・
Pump
Time
-100%
G1
G1
G1
G1
PSHPP 5 4 4 3 1 0 0 1 1 5 ・・・ 4
Thermal 0 2 2 0 0 1 0 1 1 0 ・・・ 0
Time t Time t+1
Surplus Generators
The power supply reliability is evaluated through Fig. 5 Quantization of Operation Scheduling
LOLP (Loss of Load Probability) index which can be
computed from (4). LOSS is the total supply interruption C. Optimization Problem
period. N is the number of trial. T is the simulation period.
The trial is repeated 10,000 times until LOLP is The optimization problem for operation scheduling of
converged to the steady-state value. PSHPP and thermal power plants is then formulated as
follows:
LOSS (4)
LOLP
N T Control variable:
Ps Ps2 Ps3 ... Pst
The weekly fuel cost is calculated at each trial during v 1
Ts3 ... Tst
(5)
Monte Carlo simulation. If the power output of PV and Ts1 Ts2
Objective functions is not good at local search. Therefore, this paper proposes
Minimize: that GA and TS are combined to obtain pareto optimal
N T I PSHPP operation patterns. Tabu search is a local search
f1 technique which enhances the performance using
n 1 t 1 i 1 memory structures named “Tabu List”. Once a potential
{u F i ( Pt i,n ) uti,n (1 u(it 1),n ) S i }
i
t ,n (6) solution has been determined, it is marked as "taboo" so
that the algorithm does not visit that possibility
repeatedly.
N An algorithm overview of the proposed method is
Lo n shown in Fig. 6.
f2 n 1
(7)
N T
Subject to:
Reservoir [MWh]
2000
Cap. of Upper
2000
1500
1500 1000
500
1000
0
500
Sa t Sun Mon Tue Wed Thu Fr i
Summer (August) Day
Spring (May)
0
Sa t Sun Mon Tu e We d Thu Fr i
(a) PV 0MW in Summer
Day
2000
Cap. of Upper
1500
Figure 7 depicts two different load conditions for two 1000
seasons, i.e. summer (August) and spring (May). The 500
0
system peak load is 3200 MW. In the simulation, the Sa t Sun Mon Tue Wed Thu Fr i
Day
amount of PV penetration is set to be 0MW and 1000MW.
(b) PV 1000MW in Summer
2000
Cap. of Upper
3.8 2000
Cap. of Upper
Weekly Fuel Cost [Million $]
PV1000MW
3.7 1500
3.6 1000
3.5 500
3.4 0
Sa t Sun Mon Tue Wed Thu Fr i
3.3 Day
3.2
(d) PV 1000MW in Spring
3.1
3 Fig. 9 PSHPP operation patterns
2.9
0 0.05 0.1 0.15 0.2 0.25
LOLP
(a) Summer In summer, the PSHPP is operated as a pump in the
2
PV0MW
nighttime and as a generator in the daytime on weekday
1.9
to supply the heavy summer load when PV penetration is
Weekly Fuel Cost [Million $]
PV1000MW
1.8
1.7 0MW. The role of the PSHPP in this case is load leveling
1.6 within a week. It is the conventional role of PSHPP in the
1.5
power system. Compared with the operation patterns
1.4
1.3 between LOLP best case and cost best case, they are not
1.2 so much difference. The near optimal operation pattern is
1.1 determined uniquely in summer season when PV
1
0 0.05 0.1 0.15 0.2 0.25 penetration is 0MW. In contrast, when the PV penetration
LOLP
increases to 1000 MW in summer, the PSHPP is operated
(b) Spring as a pump in the daytime and as a generator in the
Fig. 8 Pareto Optimal Solutions nighttime in the LOLP best case in spite of no surplus
power. However, PSHPP is hardly operated in the cost VI. CONCLUSION
best case.
In spring, the PSHPP is hardly operated when the PV This paper presents a cooperative scheduling method
penetration is 0MW because the operation of the PSHPP that optimizes both PSHPP operation and hot reserve
does not contribute much to the fuel cost reduction. When capacity of thermal power plants. The method makes it
the PV penetration increases to 1000 MW, PSHPP is possible to improve both reliability and economy in the
hardly operated in the cost best case, either. The PSHPP power systems with a large integration of PV. The
has to be operated as a pump in the daytime both on the simulation results show that the proposed method is very
weekday and weekend in the LOLP best case. This effective in terms of the reliability of power system
pattern is similar to that of summer pattern. operation. The total fuel cost of thermal power plants
increases in order to secure the power supply reliability
C. Generation Scheduling
using operation of the PSHPP in power system with a
The weekly thermal generation schedules for the large penetration of PV.
obtained PSHPP operation patterns are shown in Fig. 10.
The LOLP best case in both summer and spring season
when PV penetration is 1000MW are discussed here. In
REFERENCES
Fig. 10, the PSHPP pushes up the load demand in the
daytime and pushes it down in the nighttime. The [1] R. Aihara, A. Yokoyama, F. Nomiyama and N. Kosugi,
deducted load demand shown by solid black line falls “Impact of Operational Scheduling of Pumped Storage
Power Plant Considering Excess Energy and Reduction of
below the minimum output of thermal plants in the Fuel Cost on Power Supply Reliability in a Power System
daytime, which shows that the surplus power occurs. with a Large Penetration of Photovoltaic Generations”, in
However, in this case, the PSHPP can be operated for Proc. 2010 International Conference on Power System
both PV output decrease and increase by emergency Technology, Hangzhou, China, Oct. 2010.
operation. It is concluded that modification of the [2] T. Kuwabara, A. Shibuya, H. Furuta, E. Kita and K.
operation pattern of the PSHPP is effective for the Mitsuhashi, ”Design and dynamic response characteristics
of 400 MW adjustable speed pumped storage unit for
surplus power problem.
Ohkawachi Power Station, ” IEEE Transactions on
Maximum Output of Thermal Minimum Output of Thermal
Energy Conversion, vol. 11, No. 2, 1996.
Hydro
Load Demand (w/o PV, w/o Pump)
Nuclear
Load Demand (with PV, with Pump)
[3] Allen J. Wood and Bruce F. Wollenberg, "Power
Load Demand (with PV, w/o Pump) Generation, Operation, and Control", John Wiley& Son,
Power Output / Load Demand [MW]
4000
1983
3500
3000
[4] Reliability Test System Task Force of the Application of
2500
Probability Methods subcommittee, ”IEEE
2000
RELIABILITY TEST SYSTEM,” IEEE Transactions on
1500 Power Apparatus and Systems, vol. PAS-98, No. 6, 1979.
1000
500
0
Sa t Sun Mo n Tu e We d Th u Fr i
Day
(a) PV 1000MW in Summer, for Best LOLP
4000
3500
3000
2500
2000
1500
1000
500
0
Sa t Sun Mo n Tu e We d Th u Fr i
Day
(b) PV 1000MW in Spring, for Best LOLP
Fig. 10 Weekly Generation Schedule