A Dispatching Optimization Model for Park Power Supply Systems Considering Power-to-Gas and Peak Regulation Compensation

: To ensure the stability of park power supply systems and to promote the consumption of wind / photovoltaic generation, this paper proposes a dispatching optimization model for the park power supply system with power-to-gas (P2G) and peak regulation via gas-ﬁred generators. Firstly, the structure of a park power system with P2G was built. Secondly, a dispatching optimization model for the park power supply system was constructed with a peak regulation compensation mechanism. Finally, the e ﬀ ectiveness of the model was veriﬁed by a case study. The case results show that with the integration of P2G and the marketized peak regulation compensation mechanism, preferential power energy storage followed by gas storage had the best e ﬀ ect on the park power supply system, which minimized the clean energy curtailment to 11.18% and the total cost by approximately $120.190 and maximized the net proﬁt by approximately $152.005.


Introduction
Due to the fluctuation and randomness of wind/photovoltaic power generation, the issue of energy curtailment in China is still serious [1]. According to data from the National Energy Administration of China, the amount of wind power curtailment reached 10.5 billion kW·h in the first half of the year 2019 [2]. Wind/photovoltaic power curtailment is an obstacle to the development of clean energy in China and may cause greater economic losses. Therefore, research on the consumption of wind/photovoltaic power generation is still of great practical significance.
Power-to-gas (P2G) technology can, to a great extent, reduce wind/photovoltaic curtailment. Unconsumed electricity can be converted into natural gas via P2G and then into electricity in reverse through gas-fired generators (GFG), which plays a positive role in the consumption of clean energy generation [3]. Therefore, P2G can change the coupling modes of different energies [4]. In the previous studies on P2G, Gholizadeh et al. [5] evaluated and enhanced the security of P2G in networked energy hubs. Yang et al. [6] and Ye et al. [7] proposed optimal dispatching models for the multi-energy systems, which provided a new optimization method for the flexible operation of integrated energy systems. P2G technology can promote the level of renewable energy consumption in a hybrid system, as demonstrated by Marco et al. [8] and Hassan et al. [9]. Fischer al. [10] looked into the use of on-site

Structure of the Park Power Supply System with the P2G
The power supply system includes a wind turbine (WT), photovoltaic unit (PVU), energy storage device (ESD), P2G, gas storage tank, and GFG (see Figure 1). Priority is given to wind/photovoltaic power supply in the park, and the surplus power is input into the P2G or stored in the ESD. P2G is set to convert the surplus power into CH 4 , and CH 4 is only used for power generation. The surplus CH 4 will be stored in the gas storage tank. Also, the WT, PVU, P2G, and gas storage tank (P2G and gas storage tank are defined as PGST in this paper) only participate in the internal transactions, and the GFG participates in the ancillary service market. When the wind/photovoltaic power generation cannot meet the power load of users in the park, the GFG provides the peak regulation service.

Structure of the Park Power Supply System with the P2G
The power supply system includes a wind turbine (WT), photovoltaic unit (PVU), energy storage device (ESD), P2G, gas storage tank, and GFG (see Figure 1). Priority is given to wind/photovoltaic power supply in the park, and the surplus power is input into the P2G or stored in the ESD. P2G is set to convert the surplus power into CH4, and CH4 is only used for power generation. The surplus CH4 will be stored in the gas storage tank. Also, the WT, PVU, P2G, and gas storage tank (P2G and gas storage tank are defined as PGST in this paper) only participate in the internal transactions, and the GFG participates in the ancillary service market. When the wind/photovoltaic power generation cannot meet the power load of users in the park, the GFG provides the peak regulation service.

Output Models of Equipment
The available equipment in the constructed park power supply system includes clean energy units (WT and PVU), GFG, PGST, and ESD. Before building the dispatching optimization model, the output models of different kinds of equipment are given below.
The WT output is affected by the wind speed. When wind speed is within the acceptable range for the unit, the power output of the unit increases as the wind speed increases. However, if it exceeds the acceptable range, that is, the wind speed is too slow or too fast, the WT will not start in order to avoid damage to the body. The specific output model is detailed in [27]. The PVU output model generally coincides with the β distribution, which is detailed in [28].
The generation efficiency of GFG is greatly affected by its output, which decreases as the output decreases. The third-order efficiency model is adopted in this paper, which can better analyze the impact of output fluctuations on the system. The specific model [29] is where ( ) GT F t is the natural gas consumption of the GFG at time t (kW·h); , ( ) P2G is one of the key conditions for realizing bidirectional energy coupling in the power-gas system. The reaction processes are detailed in [30]. The comprehensive energy conversion efficiency of power-to-CH4 is about 45-60%. The power-to-CH4 model and the state of gas storage [25] are illustrated as

Output Models of Equipment
The available equipment in the constructed park power supply system includes clean energy units (WT and PVU), GFG, PGST, and ESD. Before building the dispatching optimization model, the output models of different kinds of equipment are given below.
The WT output is affected by the wind speed. When wind speed is within the acceptable range for the unit, the power output of the unit increases as the wind speed increases. However, if it exceeds the acceptable range, that is, the wind speed is too slow or too fast, the WT will not start in order to avoid damage to the body. The specific output model is detailed in [27]. The PVU output model generally coincides with the β distribution, which is detailed in [28].
The generation efficiency of GFG is greatly affected by its output, which decreases as the output decreases. The third-order efficiency model is adopted in this paper, which can better analyze the impact of output fluctuations on the system. The specific model [29] is where F GT (t) is the natural gas consumption of the GFG at time t (kW·h); g i,g (t) is the output of the GFG at time t (kW·h); and η G (t) is the generation efficiency of the GFG at time t (%). P2G is one of the key conditions for realizing bidirectional energy coupling in the power-gas system. The reaction processes are detailed in [30]. The comprehensive energy conversion efficiency of power-to-CH 4 is about 45-60%. The power-to-CH 4 model and the state of gas storage [25] are illustrated as Processes 2019, 7, 813 where Q P2G (t) is the CH 4 production of the P2G at time t (m 3 ); E P2G (t) is the power consumption of P2G at time t (kW·h); ϕ P2G is the P2G conversion efficiency (%); HCV is the high calorific value of natural gas (39 MJ/m 3 ) [31]; S g (t) is the gas volume in the gas storage tank at time t (m 3 ); S g (T 0 ) is the initial gas volume (m 3 ); and Q o (t) is the gas volume from the gas storage tank to the GFG at time t (m 3 ).

Peak Regulation Compensation Mechanism
Due to China's complicated electricity market, which is still developing, the peak regulation service (belonging to the ancillary service market) is an effective way to promote the consumption of clean energy generation. According to the "Market Regulation Measures for Power Peak Regulation Service" in a certain area of China, the peak regulation service is divided into basic (unpaid) and paid peak regulation based on a load rate of 52% (i.e., a peak regulation rate of 48%). The basic peak regulation service is provided when the peak regulation rate of a unit is less than 48%; the paid peak regulation service is provided when the peak regulation rate is more than 48% or a unit is started up or shut down for peak regulation.
The compensation mechanism involves units submitting the day-ahead quotations for peak regulation, and then the peak regulation service is provided according to the order from low to high. The current clearing price is taken as a settlement price, that is, the maximum quotation for the peak regulation compensation occurring in the unit in the statistical period of the same day. The peak regulation service follows the principle of "basic peak regulation taking precedence over paid peak regulation, and low-priced peak regulation taking precedence over high-priced peak regulation, on the premise of the safe operation of the power grid". The peak regulation compensation model is described as where P a (t) is the compensation price at time t ($/kW·h); P 1 , P 2 , and P 3 are the current clearing prices under different peak regulation rates ($/kW·h); R load is the load rate of unit w at time t (%); C(t) is the capacity of unit w (kW); and g(t) is the available output of unit w at time t (kW·h).

Objective Functions
As illustrated above, to reduce clean energy wastage by GFG actively participating in peak regulation and gain more profit at the same time, the dispatching optimization model (including the objective functions and constraints) based on the system net profit, the wind/photovoltaic curtailment rate, and the GFG cost is constructed.
The economic benefit is the most important focus of the park power supply system; therefore, the net profits are considered. Firstly, the net profit model of WT and PVU is given by where Z wpv is the net profit of the WT and PVU ($); g j,w (t) and g m,pv (t) are the outputs of the WT and the PVU at time t, respectively (kW·h); p park (t) is the internal power price in the park ($/kW·h); and c w and c pv are the unit costs of the WT and the PVU ($). The unit cost covers unit power generation, operation, and maintenance costs. The GFG costs mainly cover power generation, operation, and maintenance costs, and the income is from selling electricity and obtaining peak regulation compensation, so the net profit model is given by where Z g is the net profit of the GFG ($); P a (t) is the peak regulation compensation price of the GFG at time t ($/kW·h); G i,g is the compensable electricity at time t (kW·h); c g is the power generation cost of the GFG ($); ρ is the natural gas price ($/m 3 ); and c * g is the unit operation and maintenance cost ($). Therefore, the net profit of the park is given by Therein, where R is the system's net profit ($); c PGST and c ESD stand for the costs of the PGST and the ESD respectively, both of which only refer to the operation and maintenance costs ($); c * PGST and c * ESD are the unit operation and maintenance costs of the PGST and the ESD ($); and g ESD (t) and g PGST (t) are the storage capacity (kW).
Following the concern on net profits in the park, the clean energy consumption is also what this paper mainly focuses on; therefore, the objective of clean energy curtailment rate is given by where I is the clean energy curtailment rate (%); U c is the actual clean energy curtailment (kW·h); and U t is the total clean energy power generation (kW·h).
To ensure more electricity from clean energy units can be consumed, the GFG acts as a bridge between clean energy units and P2G; therefore, its cost is one of this paper's focuses. The objective of the minimum GFG costs is analyzed to motivate more willing cooperation with clean energy units. The corresponding objective is given by

Constraints
The constraints used in this paper include the supply and load balance, the running constraints of different types of equipment, and the reserve capacity of the system.
The constraint of the system power balance is stated as where g i,g , g j,w , and g m,pv are the available outputs of GFG, WT, and PVU respectively (kW·h); and D is the load demand (kW·h). The running constraints of GFG, WT, and PVU are detailed in [32]. The constraint of the system's reserve capacity is where D(t) is the load demand at time t (kW·h); R(t) is the system reserve demand at time t (kW); l is the line loss rate (%); θ is the unit self-use rate (%); and g max (t) is the maximum output at time t (kW·h).
Since the objective functions with the above constraints are mixed integer programming problems, the General Algebraic Modeling System (GAMS) software is a good solution tool. The mixed-integer programming models and the GAMS are detailed in [33].

Scenario Settings
To verify the effectiveness of the proposed model, four scenarios based on different types of storage equipment and the different use priorities were set up (see Table 1).
The different types of storage equipment were as follows: • In Scenario 1 (S1), the park power supply system employed an energy storage device (ESD) to store the unconsumed wind/photovoltaic power generated to be sold in the next period; • In Scenario 2 (S2), the park power supply system employed a P2G and a gas storage tank (PGST) to convert the unconsumed wind/photovoltaic power generated into natural gas to be stored in the gas storage tank as the fuel for power generation.
The different storage equipment use priorities were as follows: • In Scenario 3 (S3), the park power supply system employed a PGST and an ESD. The unconsumed wind/photovoltaic power generated was preferentially converted into natural gas via the P2G and stored in the gas storage tank as the fuel for power generation, and then the surplus was stored in the ESD.

•
In Scenario 4 (S4), the park power supply system employed a PGST and an ESD. The unconsumed wind/photovoltaic power generated was preferentially stored in the ESD, and then the surplus was converted into natural gas via the P2G and stored in the gas storage tank as the fuel for power generation.

Basic Data
A park power supply system used in an area of China was selected for the case study. In this park, clean energy units, gas-fired generators (GFGs), an ESD, and a PGST were employed. The clean energy units contained a wind turbine (WT) and a photovoltaic unit (PVU), and the PGST included a P2G and a gas storage tank. Besides, there were four GFGs in the park (the specific parameters are shown in Tables 2 and 3). The capacity of the ESD was 1000 kW, the maximum charging/discharging power was 200 kW, the charging/discharging loss was 0.4, and the initial energy storage was set to be 0; the unit cost of operation and maintenance was 0.071 $/kW·h.  The power load demand and the output of clean energy units in this park are illustrated in Figure 2. Given the uncertainty of wind and photovoltaic power generation, the output data of the clean energy units on a typical day were selected for this research. GFGs satisfied the load demand that WT and PVU could not cope with. be 0; the unit cost of operation and maintenance was 0.071 $/kW·h.  The power load demand and the output of clean energy units in this park are illustrated in Figure  2. Given the uncertainty of wind and photovoltaic power generation, the output data of the clean energy units on a typical day were selected for this research. GFGs satisfied the load demand that WT and PVU could not cope with. Based on the load demand in the park, a day was divided into three periods: 8:00-15:00 and 19:00-22:00 were the peak periods, 23:00-4:00 was the valley period, and 5:00-7:00 and 16:00-18:00 were the flat periods. For the given periods, the power price data are presented in Table 4, according to the actual prices of a province in China. In addition to power prices, the natural gas price was 0.488 $/m 3 (for convenient calculation, it was converted into the unit calorific value price, 0.049 $/kW·h). Besides selling electricity, GFGs participate in peak regulation to gain profit, which increases clean energy consumption. In the light of peak regulation compensation mechanism, GFGs propose their quotations the day before operation. The quotation standards under different peak regulation rates are shown in Table 5. According to the load demand and peak regulation rate of each unit, the day-ahead peak regulation quotations of all GFGs during each period were generated randomly, and these are displayed in Table 6.

Wind/Photovoltaic Curtailment
WT, PVU, GFG, and ESD were the units providing power. The power output of PGST was in the form of gas-fired power generation. Figure 3 shows the output of different types of equipment in each scenario, and Table 7 shows the clean energy curtailment rates of different scenarios. The results indicate that the available outputs of the WT and PVU were basically the same in different scenarios, but the wind/photovoltaic curtailment differed accordingly. The curtailment rate in the S1 was the highest, reaching 16.70%, whereas when the PGST was employed, the curtailment rate decreased (12.37% in S2), as did the GFG output. Moreover, in the scenario where both the PGST and ESD were employed, because of the different priorities of using the two devices, different results occurred for the wind/photovoltaic curtailment and the GFG output. The curtailment rate in S3 decreased to 11.59% compared with that in S2, but that in S4 decreased more-11.18%. Thus, the method of 'energy storage preferentially, then gas storage' has the best effect on reducing wind/photovoltaic curtailment.

Wind/Photovoltaic Curtailment
WT, PVU, GFG, and ESD were the units providing power. The power output of PGST was in the form of gas-fired power generation. Figure 3 shows the output of different types of equipment in each scenario, and Table 7 shows the clean energy curtailment rates of different scenarios. The results indicate that the available outputs of the WT and PVU were basically the same in different scenarios, but the wind/photovoltaic curtailment differed accordingly. The curtailment rate in the S1 was the highest, reaching 16.70%, whereas when the PGST was employed, the curtailment rate decreased (12.37% in S2), as did the GFG output. Moreover, in the scenario where both the PGST and ESD were employed, because of the different priorities of using the two devices, different results occurred for the wind/photovoltaic curtailment and the GFG output. The curtailment rate in S3 decreased to 11.59% compared with that in S2, but that in S4 decreased more-11.18%. Thus, the method of 'energy storage preferentially, then gas storage' has the best effect on reducing wind/photovoltaic curtailment.

GFG Costs
For different scenarios, the GFG costs during different periods are shown in Table 8.  Table 8 indicates that, with the participation of the PGST, the GFG costs decreased significantly, especially in flat and peak periods. Compared with S1, the GFG costs in S2 decreased by $220.160, because the GFG used gas from the P2G instead of purchasing it from the outside; thus, its generation cost decreased. The GFG costs in S3 and S4 were both lower than that of S2 by $41.430 and $72.680, respectively, which indicates that with the combination of the ESD and the GPST, the costs were further reduced. Additionally, compared with S3, the costs in S4 decreased by $31.249, so the different priorities of using the PGST and the ESD affect the GFG costs, and the method of 'energy storage preferentially, then gas storage' has a better effort.  4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

GFG Costs
For different scenarios, the GFG costs during different periods are shown in Table 8.  Table 8 indicates that, with the participation of the PGST, the GFG costs decreased significantly, especially in flat and peak periods. Compared with S1, the GFG costs in S2 decreased by $220.160, because the GFG used gas from the P2G instead of purchasing it from the outside; thus, its generation cost decreased. The GFG costs in S3 and S4 were both lower than that of S2 by $41.430 and $72.680, respectively, which indicates that with the combination of the ESD and the GPST, the costs were further reduced. Additionally, compared with S3, the costs in S4 decreased by $31.249, so the different priorities of using the PGST and the ESD affect the GFG costs, and the method of 'energy storage preferentially, then gas storage' has a better effort.

System Economic Benefits
In view of the above four scenarios, taking the transactions in the electric energy market and ancillary service market into account, the results of system economic benefits in different scenarios are shown in Table 9. As presented in Table 9, the scenario with PGST (namely, S2) earned $116.090 more and cost $66.034 less than S1 where the PGST was not considered. This demonstrates that the P2G can effectively reduce the total cost and increase the net profit. Under the joint operation of the PGST and the ESD, the method of 'energy storage preferentially, then gas storage' (namely, S4) minimized the total cost of the system and maximized its net profit.

Conclusions
To promote clean energy consumption to mitigate resource wastage and encourage gas-fired generators to collaborate with clean energy units, the construction of a dispatching optimization model with P2G and the peak regulation compensation mechanism for GFG was carried out in this paper. The model tends to pursue the maximum net profit of the park power supply system and the minimum clean energy curtailment rate and GFG costs. After conducting a case study, it was found out that systems with P2G can effectively promote clean energy consumption in comparison with ones involving no P2G. Additionally, with the joint operation of the PGST and the ESD, the method of 'energy storage preferentially, then gas storage' can maximize benefits.
Due to the high installation capacity of clean energy units but insufficient peak regulation sources and immature market mechanism development in China, marketized compensation is still an important means to encourage conventional generators to participate in peak regulation for clean energy consumption, and P2G not only increases the consumption but also reduces GFG costs, being a novel means in the power industry. As P2G technology develops (lower investment and higher conversion efficiency), it will be more practical and popular in the future.
In future research, the effect of P2G on CO 2 emission reduction (environmentally) will be taken into account and how much it can gain for the park power supply system in terms of green certificate trading and carbon trading (economically) will be investigated. Also, more marketized peak regulation compensation means and more types of peak regulation units based on local resource advantages will be discussed, respectively.