Multi-Objective Generation Scheduling of Hydro-Thermal System Incorporating Energy Storage with Demand Side Management Considering Renewable Energy Uncertainties

Atmospheric pollutants mainly produced by thermal power plants compel to utilize green energy sources such as renewable energy sources and hydroelectric plants in a power system. But due to blinking behavior of sources of renewable energy and due to very high rate of outages, it has a detrimental consequence on overall grid. Demand side management (DSM) programs decrease cost and improve power system security. This study proposes non-dominated sorting genetic algorithm-II (NSGA-II) to solve multiobjective scheduling of generation for fixed head hydro-thermal system integrating pumped hydro energy storage and sources of renewable energy taking into consideration the outage and uncertainty in presence of DSM. Numerical results of the test system attained using the proposed technique were compared with strength pareto evolutionary algorithm 2 (SPEA 2).


INDEX TERMS
Till today, the power plants based on fossil-fuel are the chief sources of generating power. But, these plants discharge sulfur oxides, nitrogen oxides and carbon dioxide to the atmosphere. These cause lethal damage to flora and fauna and global climate. These result into increased concern over ecological protection with several environmental amendments. For electric utilities, one major challenge is to decrease atmospheric pollution, for reducing acid rain and greenhouse gasses which is the aim of 1990 Clean Air Act. So modern's civilization wants quality electricity not only at low-cost, but pollution free. Many approaches are proposed to reduce pollution in the atmosphere [1]. The rapid increase of electric power demand, gradual reduction of fossil fuel and global warming have pushed energy based research in the direction of green energy. Because of this clean energy sources are achieving to meet the energy demand. Variability and irregularity turn out vital challenges to overcome the problem of scheduling. The grid may have detrimental effect due to this intermittent nature. It is overcome by using pumped hydro energy storage. There is always a possibility of high rate of outage in solar and wind power. Hence it is vital to study possibility of outage during generation scheduling. Optimal generation scheduling with renewable energy sources of a miniature autonomous system is discussed in [2]. Though, these sources are pollution free but their generation capability is low. Use of amalgam energy system i.e., thermal power integrating wind power [3], thermal power plant-solar PV plant [4], hydro-thermal integrating wind power [5] has swiftly enhanced. Pumped-storage-hydraulic (PSH) unit is attaining the mammoth attention all over the earth [6] primarily because of characteristic of energy storage. Main function of PSH units is to hoard low-cost excess energy during off-peak load levels as hydraulic potential energy pumping water from lower reservoir to upper reservoir. During peak load levels, stored hydraulic potential energy is utilized. PSH unit ordinarily works in daily or weekly. Operation over a period of a PSH unit reduces the fuel cost. In [8], Gradient search techniques and Lagrangian multiplier is used to get optimum hydro-thermal generation scheduling with PSH unit considering constraints. In [9] evolutionary programming technique has been employed for the same problem in hydrothermal system with PSH units. Mohan et al. [10] has shown that a pumped-hydro unit(PHU) can be used as peak-load management unit by shutting down electric power in turn to reduce the large deviation in frequency. Ma et al. [11] shows the pumped hydro storage system for solar energy infiltration and for mini sovereign systems. Multi-objective (MO) hydrothermal generation scheduling problem where cost and emission objectives are optimized simultaneously has been discussed by a number of researchers [12]- [19]. Simab et al. [12] have employed MO programming for pumped-hydro-thermal scheduling problem. Narang et al. [13] have discussed MO short term hydrothermal generation scheduling utilizing predator-prey optimization. Sun et al. [14] have applied an improved quantum-behaved PSO for economic emission hydrothermal scheduling problem. Zhang et al. [15] have presented gradient decent based MO cultural DE for shortterm hydrothermal optimal scheduling incorporating wind power and photovoltaic power. Dhillon et al. [16] applied a fuzzy decision method for deciding generation scheduling of a hydrothermal problem. Fuzzy satisfying method based on EP technique [17] is discussed for MO short-term hydrothermal scheduling problem. Crisscross PSO algorithm [18] is used for MO generation scheduling of pumped storage hydrothermal system incorporating solar units. Basu [33] has applied chaotic fast convergence evolutionary programming (CFCEP) for short-term hydrothermal scheduling. Kaur et al. [35] have applied chaotic-crisscross differential evolution (CCDE) algorithm for short-term hydrothermal scheduling. DSM programs have many advantages for example lessening the cost, improving the power system security [24], etc. A variety of classical methods like Newton's method [27], Lagrange multiplier method [28], dynamic programming [8] etc is employed to solve short-term fixed head hydrothermal scheduling. A variety of metaheuristic algorithms such as Hopfield neural network [29], artificial immune system [30], cuckoo search algorithm [31], fast convergence evolutionary programming with time varying mutation scale (FCEP_TVMS) [34] etc is employed to solve short-term fixed head hydrothermal scheduling problem. Modified cuckoo search algorithm [32] is discussed for MO short-term fixed head hydrothermal scheduling problem. The major aim of MO short-term generation scheduling of fixed head hydrothermal power system incorporating pumped hydro energy storage with and without demand side management (DSM) considering uncertainty and outage of renewable energy sources is to optimize total cost and emission echelon simultaneously over a scheduling period simultaneously satisfying various constraints. Here, nondominated sorting genetic algorithm-II (NSGA-II) is pertained to solve short-term MO generation scheduling of fixed head hydrothermal power system incorporating pumped hydro energy storage with and without DSM considering uncertainty and outage of sources of renewable energy Simulation outcomes of the test system are matched with that obtained by strength pareto evolutionary algorithm 2 (SPEA 2).
The major contributions of this manuscript can be stated as follows:  Multi-objective generation scheduling of fixed head hydrothermal system has been considered.  Ramp rate limit constraints of thermal generators have been taken into consideration.  Uncertainty and outage of renewable energy sources have been taken into account.
 The problem is solved with and without DSM.

1) PROBABILITY DISTRIBUTION OF SOLAR PLANT AND WIND TURBINE
Due to their doubtfulness and irregularity of solar plant and wind turbine it is difficult to integrate them. Large reserve capacity margin is caused due to overestimation of renewable power which in turn instability in the steady state security if there is rise in demand, while underestimation outcomes loss of excess energy. During generation scheduling, both sum up to the total generation and operation costs . As a result, different uncertainty modeling, like Weibull, Beta, Lognormal and Gumbel probability distribution functions (PDFs), is implemented by many researchers to evaluate reserve cost and penalty cost for overestimation and underestimation respectively. Solar irradiation and wind speed are predicted to be well trailed by lognormal and Weibull PDFs respectively as in (1) and The o/p power [19] of k th wind turbine at time t for a given wind speed is affirmed as The o/p power [20] from m th solar plant at time t for a given irradiation G is affirmed by

4) POWER PROBABILITIES OF SOLAR PV PLANT
Probability of a PV power is equal to the value of corresponding solar power irradiation probability as in (5).

5) POWER PROBABILITIES OF WIND TURBINE
For the discrete zones wind power probabilities i.e., for first and third case of (3), is computed using (6) and (7) respectively [22].
The probability for WT power in the continuous region as second case in (3) is calculated as (8).

B. Outage modeling of solar PV plant and wind turbine
Unpleasant environmental state may force the renewable sources to face forced outage frequently. This forced outage modelling depends on three factors, viz., repairable failure, aging and weather dependency. Repairable forced outage rate is defined as (9) [25]. Hence, multi-factor independent outage is involved; the outage rate is estimated using the concept of union set. So, the forced outage rate of any renewable unit is cleared by (12).

C. Objective function and constraints
The multi-objective generation scheduling of fixed head hydrothermal power system with pumped hydro energy storage and renewable energy sources considering uncertainty and outage in presence of DSM is devised to optimize total cost and emission echelon simultaneously taking variety of constraints. objective functions and constraints taking into description with DSM and outage possibility.
(1) COST The total cost is affirmed as The cost function of fuel of i the thermal generator at time t , taking valve-point effect [36], is affirmed as (14) Reserve cost and penalty cost for overestimation and underestimation on dispatchable wind power [20] is given in (15)-(16) respectively.
Reserve cost and penalty cost for overestimation and underestimation on dispatchable solar power [20] is given in (17)-(18) respectively.
For assessment purposes, total emission of these pollutants For assessment purposes, total emission of these pollutants is affirmed as the summation of a quadratic and an exponential function [37]. Total emission of thermal generators is affirmed as subject to (i) Power balance constraints: Assuming , 0  t Ls when load curtailed due to DRP, . and when load is moved to base load demand, no load is curtailed .

Total transmission loss
Lt  can be calculated by utilizing and im  is the respective thermal, hydro, wind power, solar PV unit.

4) CONSTRAINTS OF PUMPED-STORAGE
PSH unit depends entirely on water which is pumped to an PSH unit depends entirely on water which is pumped to an upper reservoir from lower one. When the unit changes from generating mode to pumping mode or vice-versa, the unit is made off for an hour called as change-over time.  (29) In this problem. the initial and final volume of water of upper reservoir of the PSH unit is considered as same

5) GENERATION LIMITS max min
hj hjt hj

8) DEMAND SIDE MANAGEMENT
Demand side management [23] plays an important role in power system. Demand side management alters customers' electricity consumption patterns to produce the desired changes in the load shapes of power distribution systems. The changes in the final consumption profile will depend on the planning objectives and operation of the utility companies. Demand side management focuses on utilizing power saving technologies, electricity tariffs, monetary incentives, and government policies to mitigate the peak load demand instead of enlarging the generation capacity or reinforcing the transmission and distribution network. To mitigate system instabilities brought about by increasing electricity demand, a suitable objective of demand side management activities could be to change the shape of the load demand curve by reducing the total load demand of the distribution system during peak periods, and shift these loads to be served during more appropriate times in order to reduce the overall planning and operational cost of the network. Time of use (TOU) DR program [24] is the most common price based programs that aims to improve and control subscribers' consumption by changing the electricity price in different time periods. This is actually achieved by motivating the consumers that their electricity price will be reduced. Therefore, this program implements DR programs by informing the consumers about electricity prices. In this type of DR programs, the electricity price depends on when electricity is used. Consumers are heavily charged for power consumption during peak period. Therefore, they are encouraged to reduce their consumption during peak hours and shift their suspended loads to off peak hours. In the TOU program, the electricity tariff varies in different time periods. These tariffs are usually obtained through power generation and transmission cost in these periods. In TOU programs, electricity tariffs are usually pre-determined for several months, years, and different seasons. Here, DR program is used to smooth the load curve by shifting loads from peak hours to off peak hours and, thus, reduce operating costs. As a result the power demand curve is flattened. The TOU program is designated by the equation (35) and constrained by equations (36)- (39).

III. SOLUTION METHODOLOGY A. Multi-objective Optimization Principle
The majority of the actual-world problems engross optimization of a number of noncommensurable and conflicting objective functions simultaneously where a set of optimal solutions is produced in place of one optimal solution because no solution can be looked upon as superior than any other with respect to all objective functions. The problem of Multi-objective optimization comprises a no. of objectives and a number of equality and inequality constraints and is affirmed as:

B. Non-dominated Sorting Genetic Algorithm-II
Srinivas and Deb [38] established nondomoinated sorting genetic algorithm (NSGA) Nondomination is exploited based on grade decisive factor of solutions, and fitness sharing is exploited for diversification control in the explore space. As NSGA depends heavily upon fitness sharing parameters, Deb et al. [39] established NSGA-II, which producing more reliable solution than its precursor. NSGA-II flow chart is depicted in Fig.1.
Here, the problem is solved with and without DSM. The minimum and maximum forecast limits of solar irradiation and wind velocity [33] are illustrated in Fig. 2 and Fig. 3 respectively. A sudden change in wind speed can be noticed at 16 th hour in Fig. 3. Such high wind speed generally results into turbulent weather condition and causes renewable unit failure. The failure probabilities for PV and WT units [33], which can be fetched from weather dependent historical data, are portrayed in Fig. 4. The forced outage rates of PV and WT units [33] are presented in Fig. 5

MW
Operating limits: PHP is permitted to work at −100 MW while pumping. Reservoir starts at 3000 acre-ft and at the end of 24 hour it must be at 3000 acre-ft. Spillage is not considered and water inflow rate is neglected.
Solar-wind-hydro-thermal-pumped storage generations with DSM acquired from cost minimization and emission minimization by using RCGA are summarized in Table 1 and Table 2 respectively. Solar-wind-hydro-thermalpumped storage generations with DSM acquired from cost and emission objectives optimized simultaneously by using NSGA-II and SPEA 2 are summed up in Table 3 and Table  4 respectively. Solar-wind-hydro-thermal-pumped storage generations without DSM acquired from cost minimization and emission minimization by using RCGA are summarized in Table 5 and Table 6 respectively. Solar-wind-hydrothermal-pumped storage generations without DSM acquired from cost and emission objectives optimized simultaneously by using NSGA-II and SPEA 2 are summed up in Table 7 and Table 8 respectively. Cost, emission and CPU time acquired from cost minimization i.e. economic dispatch, emission minimization i.e. emission dispatch and economic emission dispatch where cost and emission objectives are optimized simultaneously with and without DSM are summed up in Table 9. The cost convergence and emission convergence characteristics with and without DSM acquired by utilizing RCGA have been revealed in Fig. 6 and Fig. 8 respectively. Figure 7 and Figure 9 reveal the distribution of 20 nondominated solutions attained from the economic emission dispatch with and without DSM in the final iteration of proposed NSGA-II and SPEA2. VOLUME XX, 2017 1       Table 7: Thermal-hydro-wind-solar-pumped storage generation (MW) acquired from economic emission dispatch without DSM using NSGA-II