The problems of energy shortage and environmental pollution can no longer be ignored. How to make the best of energy and improve energy efficiency has always been a concern of researchers. The rapid development of electric vehicles (EVs) has made them an energy load that cannot be ignored. On this basis, an optimal configuration model of a multienergy virtual power plant (MEVPP) considering EV access is constructed to meet the multiple energy needs. To better consider EV users’ willingness to respond, this paper combines price demand response (PDR) with incentive demand response (IDR), establishes a fuzzy response model for EV charging and discharging under the joint response strategy (JRS), and analyzes the influence of user responsiveness and large-scale EV access on MEVPP planning and operation under different incentive levels. Meanwhile, to realize the low carbon, a stepped carbon trading mechanism (SCTM) is introduced. Based on the gazelle algorithm and mixed integer linear programming (MILP), the capacity and output of the system energy equipment are jointly optimized, and the running curve of MEVPP in a typical quarter is analyzed. The example analysis shows that the joint response strategy proposed reduces the operating cost by 7.1%, and the introduction of SCTM reduces the carbon emission by 13.7%, realizing the low-carbon and economic running of MEVPP.

1. Introduction

As the energy crisis intensifies and the greenhouse effect intensifies, it is particularly important to develop a sustainable energy system [1]. WT, PV, and other renewable energy (RE) have the characteristics of clean and renewable, to making the best of them is the key to realizing China’s “double carbon” goal. Under the new situation of energy development, VPP technology has developed rapidly because of its characteristics of integrating power generation resources and improving energy production and utilization [2]. Controllable loads such as EVs, RE, and energy storage (ES) have gradually become the main body of controllable aggregate resources inside VPP [3–5].

At present, domestic and overseas researchers have carried out some studies on VPP. Reference [6] makes a detailed analysis of the technical difficulties and solutions of trusted transactions in VPP and looks forward to its future development. Reference [7] presents an optimal allocation method for distributed generation and DR resources in a VPP. Reference [8] has built a VPP electric heating capacity optimal allocation model aiming at the economic optimization of WT transmission in regional grid protocol. Reference [9] explores the problem of optimal allocation of multiple power sources capacity in a VPP and thinks the uncertainty of RE output and electricity price. The energy storage characteristics of EVs make them an aggregable resource in VPP, which is not taken into account in the VPP studied in the above reference. Reference [10] proposes an optimal allocation method of ES in distribution networks considering the aggregation regulation of EVs, and results show that the carbon emissions are reduced and the photovoltaic utilization rate is improved. In Reference [11], the optimal allocation model of frequency modulation capacity of VPP with ES, EV, and thermal power units is established thinking the response uncertainty of EVs. Given the impact of renewable distributed power supply and large-scale EV access on the distribution network, Reference [12] proposes an optimal configuration method of distributed power supply and EV charging stations based on space-time characteristics and DR. Reference [13] sets up thinking EV-VPP photovoltaic high flexibility and access model of distribution network planning. The proposed model reduces the planning and running costs and increases the absorption rate of photovoltaics. However, the modeling and analysis process of the response behavior of EVs in the abovementioned reference is relatively simple. In real life, the response behavior of EV users will be affected by many factors, including charging cost and battery discharge loss. Therefore, the user’s willingness to respond should be fully considered, which will be conducive to formulating a reasonable response strategy to optimize the EV load.

The effective use of a variety of energy sources, including RE, is the focus of applied energy research, which can improve energy efficiency through the way of multienergy complementarity [14]. The operation process of traditional VPP lacks multienergy coupling supply and demand optimization means and flexible peak regulation strategy, while VPP uses information technology to provide a development direction for the transformation of the energy system structure [15–17]. Reference [15] has built an optimal scheduling model based on the cooperation of VPP and electricity-heat-gas IES, and the results show that the system has a better economy and environmental protection. In Reference [16], an multienergy virtual power plant (MEVPP) including a PV plant, WT plant, and ES battery was built, and a two-stage optimal scheduling model was established. In Reference [17], aiming at the problem of insufficient interaction between energy supply and demand in a comprehensive energy scenario, an optimal operation model of a VPP was established thinking the participation of multiple energy coupling devices and demand response. At present, most of the reference on MEVPP is studied from the perspective of optimizing operation, while research on optimizing configuration is rare. The optimal configuration is the premise and basis of optimal scheduling. Therefore, it is important to study the optimal allocation of MEVPP.

For the model solution method, most of the literature uses a two-layer optimization solution method. The upper layer mostly uses genetic algorithms, particle swarm optimize (PSO) algorithms, gray wolf algorithms, tree species algorithms, and so on [18]. The lower layer algorithms invoke the CPLEX solver for solving, and most of them have economy as the optimization objective, and some of the literature consider the indicators of power supply reliability, low carbon, safety, and so on [19–21]. However, all of the abovementioned algorithms were proposed very early, and there are inevitably problems of low solving efficiency and solving accuracy when solving the model.

On this basis, the MEVPP optimization model considering EV access is constructed to meet the various energy needs. To better optimize the EV load of the access system, this paper combines PDR and IDR to establish a fuzzy response model of EV load under JRS so as to better consider the user’s response willingness. The effects of user response and the number of large-scale EVs on MEVPP planning and operation under different incentive levels were analyzed. In this case, the JRS proposed can improve the low carbon economy and improve the rationality planning and operation. Meanwhile, the SCTM is introduced to further realize the low carbon. Based on the gazelle algorithm and MILP, the capacity and output of energy equipment of the system are jointly optimized, and the running curve of MEVPP in typical quarters is analyzed. Example analysis shows that the joint response strategy proposed can decrease the running expense of MEVPP by 7.1%, and the introduction of the SCTM reduces the carbon emission by 13.7%, which realizes the low-carbon economic running of MEVPP.

2. Structural Diagram of MEVPP Considering EV Access

Figure 1 displays the structure of the MEVPP build. The energy facilities included in the MEVPP model are as follows: WT, PV, combined heating and power (CHP), gas boiler (GB), absorption cooler (AC), electric cooler (EC), electric boiler (EB), electric ES (EES), cold ES (CES), and heat ES (HES). Thus, a multienergy coupled VPP polymerization unit is constructed.

Details are in the caption following the image — **Figure 1**
Open in figure viewer PowerPoint

Consider the MEVPP structure diagram of EV access.

By using advanced information technologies such as the internet of things and big data, VPP establishes an interactive platform for information flow and energy flow to realize the operation and management of VPP. MEVPP manages a variety of aggregated controllable resources within it through unified scheduling, which can achieve efficient complementary utilization of heterogeneous energy and meet a variety of energy consumption requirements.

2.1. EV-JRS Model

Monte Carlo simulation is a simulation statistical method whose practical application scenario is to simulate unavailable data [22]. Using this method to simulate the EV group load, the specific modeling process is referred to in Reference [23]. See Table 1 for EV’s detailed arguments.

Table 1. EV-related parameters table.

Parameter name	Parameter value
l_s	0.25 Kw · h
C_ev	38 Kw · h
	0.9
	0.9
	10 Kw
	10 Kw

When EVs participate in DR, they are not only constrained by the objective response capability but also affected by the subjective willingness of users. In this paper, we consider the EV type as a private car, which has a long residence time to participate in the system scheduling after finishing a day trip. Some studies only guide EV charging and discharging through PDR, and while PDR has a limited role in guiding EV charge and discharge activities, it is necessary to optimize EV charging and discharging behaviors by combining with IDR [24].

Setting the incentive level is a binary function about charging and discharging subsidies, and the incentive level is set as follows [25]:

()

where: x_jl is the incentive level; α and β are the charging and discharging subsidy attraction factor; s_c is the charging subsidy; s_d is the discharging subsidy; φ_c is the charging loss cost; and φ_d is the discharging loss cost.

The charging subsidy guides the users’ charging response, and its subsidized price is set as follows:

()

where

is the saturation charging allowance, above which the economy of the system will be reduced.

The discharge subsidy guides the users’ discharging response and takes into account the compensatory charging behavior caused by EV discharges. The subsidy price is set as follows:

()

where

is the saturation discharge allowance, above which the economy decreases;

is the compensatory charging expense; and ω is the charging billing tariff.

In this paper, the degree of response describes the consumer’s desire to respond, and the vague mapping relationship between the consumer’s response state and responsivity is displayed in the following equation:

()

where ω(i) is the response state of the user, w(i) = 0 represents reject response; ω(i) = 1 represents response; σ(i) is the consumer responsivity; and σ_T is the user response degree threshold.

The user responsiveness at different incentive levels can be denoted as

()

where k_r is the slope of responsivity for the excitation level; x_jl, max is the level of stimulation in response to saturation; and Δσ(i) is the maximum deviation of responsivity.

The pattern of Δσ(i) change is shown in the following equation:

()

where k₁ and k₂ are the proportionality coefficients of the changes in incentive prices before and after the solid bootstrap effect, respectively. The relevant parameters involved in this section are referred to in Reference [25].

In this way, the user’s response will usually be σ_down(i) ≤ σ(i) ≤ σ_up(i). In practice, it can be modeled and analyzed through the distribution law of historical data. Uniform distribution is used to describe it simply as follows:

()

2.2. SCTM

At this stage, China, mainly through the government regulatory authorities, has emission reduction obligations to market players to allocate free initial carbon quota and market players under their practical carbon emissions to attend the carbon trade market for carbon quota-free trade [26]. Adoption of the baseline methodology for determining systematic carbon allowances and the carbon emissions mainly come from purchased electricity, the CHP equipment, and GB.

The allocation of nonremunerated carbon credits under carbon trading is as follows:

()

where E_e, E_CHP, and E_GB are the unremunerated carbon emission allowance of purchased electricity, CHP, and GB, respectively; E is the total unremunerated carbon emission; q_l is the unremunerated carbon emission per unit of electricity;

is the amount of electricity purchased; θ is the amount of unremunerated carbon emission unit of heat; ζ is the conversion coefficient of the electricity generation of the CHP equipment to the amount of heat generated;

are the heat and electricity output of the CHP; and P_GB,t is the heat output of the GB.

Carbon trading costs are modeled as follows [27]:

()

where E^′ is the actual overall carbon emissions, same as that used to calculate E, the difference is that the carbon emission coefficients taken in Equations (9)–(11) are different values; C_C is the carbon trading expense; ρ^c is the ladder carbon price;

is the base price of carbon trade; l is a carbon emission range; m is the increase factor of carbon price; and u is the number of the intervals. The related parameters are referred to [26].

3. MEVPP Two-Tier Planning Model

3.1. Upper-Tier Capacity Allocation Model

3.1.1. Objective Function

The upper-tier capacity allocation phase aims to minimize the total system expense, which mainly includes the system equipment investment expense and the operating expense per unit period.

()

where F is the total expense; F_T is the investment expense of device; F_Y is the running expense per unit cycle; N is the number of kinds of the device; C_i,c is the capacity of device; C_i,p is the unit price of the capacity of the device; r is the discount factor, which is set to 5%; Y is the life cycle of the device; s denotes the number of typical scenarios; num_s indicates the number of days in a representative scenario; and F_Y,s is the daily operating cost under the scenarios.

3.2. Lower-Tier Optimization Operational Model

3.2.1. Objective Function

Based on the configured device’s original capacity, the MEVPP lower-tier optimized operation model targets the lowest daily running costs.

()

where C_P is the expense of purchased energy; C_M is the device running and maintenance expense; C_DR is the expense of EV demand response; p_t is the price of electricity; q_t is the price of gas;

is the volume of purchased gas; m_i is the unit running expense of device; P_i,t is the power of the device; and C_ev,c and C_ev,d are the charging and discharging response capacity.

3.2.2. Restrictive Condition

3.2.2.1. Energy Balance Constraints

()

where:

are the charge and discharge power of EES and single EV;

are the output power of PV and WT;

is the electrical load; and

are the consuming power of EB and EC.

3.2.2.2. Heat Balance Constraints

()

where:

are the charge and discharge heat power of HES; P_EB,t is the heat production power of EB;

is the heat load; and

is the thermal consumption power of AC.

3.2.2.3. Cold Energy Balance Constraints

()

where P_AC,t and P_EC,t are the cooling power of AC and EC;

are the cooling power of CES charge and discharge; and

is the cooling load.

3.2.2.4. Gas Balance Constraints

()

where

is the gas load and

are the gas consumption power of CHP and GB.

3.2.2.5. Unit Climbing Constraints

()

where

is the output power of the device and

are the uphill and downhill power limit values of the equipment.

3.2.2.6. Energy Storage Constraints

()

where

is the ES power of the ES;

are the charge and discharge power and efficiency of the ES facility; and

are the upper and lower capacity bound of the ES facility.

3.2.2.7. EV Demand Response Constraints

Vehicles participating in the system’s optimized scheduling must have an EV battery level that meets the users’ expectations before the next day’s trip. The EV must also meet the battery capacity limit constraints and maximum charge and discharge power constraints.

()

where C_ev,t is the battery capacity of EV;

are the charge and discharge efficiency and time of EV; C_ev,max and C_ev, min are the upper and lower bounds of EV battery capacity; and

are the maximum charge and discharge power of EV.

4. Model Solution

For the optimal planning model proposed, the upper model is addressed by employing the gazelle optimization algorithm; the lower model is addressed by invoking the CPLEX solver through MATLAB.

4.1. Gazelle Optimization Algorithm (GOA)

For the study of MEVPP capacity allocation, a large amount of references has used traditional algorithms such as PSO and genetics for optimization and solution. The GOA proposed in 2023 is used to solve the established two-tier allocation model. The GOA is a novel, simple, and search-capable global stochastic optimization algorithm [28]. GOA is explored and developed with the core idea of gazelle’s behavior of escaping from predators as follows.

4.1.1. Initialization

GOA utilizes randomly initialized gazelles as hunt individuals, and hunt individuals are n × d matrices of candidate solutions with the following expression:

()

where X is the position vector matrix of candidate populations; x_i,j is the randomly generated vector location of the ith population in jth dimension; i = 1, 2, ⋯, n; j = 1, 2, ⋯, d, where n is the number of populations; and d is the number of dimensions. In each iteration, each individual generates a candidate solution. The optimal solution is named the top gazelle, and an elite matrix is constructed for the next step of the search with the following expression:

()

where

is the location of the apical gazelle. GOA deems both predators and gazelles to be searching individuals that move in the same direction when a predator is detected stalking a gazelle.

4.1.2. Development Phase

This phase is the foraging phase, and the gazelle performs Brownian motion. The expression can be denoted as follows [29]:

()

where g_i+1 and g_i are the solution of the next iteration and the present; s is the predation speed; R_B is the vector of random figures of; R is the vector of uniform random figures within [0, 1]; and E_i is the elite matrix.

4.1.3. Discovery Phase

Assuming that a sudden change in direction occurs at each iteration as the number of iterations is odd, the GOA shifts in one direction. As the number of iterations is even, the GOA moves in the other direction. The gazelle reacts first, while the predator reacts slower, so it is considered to run first in Brownian motion and then in Levy flight [30]. When the gazelle finds the predator, its behavior can be denoted as follows:

()

where μ is a constant; S is the largest speed gazelle can attain; and R_L is a vector of stochastic digits on account of the Levy distribution. Predator chasing gazelle can be denoted as follows [31]:

()

where iter, max _iter are the present and largest iteration figure.

Let PSRs represent the predator success rate, which can be utilized to restrain the GOA from falling into a local optimum. In the gazelle escape phase, it was found that the PSRs in pursuing gazelles was 0.34, which affected the gazelle’s escape ability, and the position update formula can be denoted as follows [32]:

()

where UB and LB are the upper and lower limits of the population; U is a binary vector; r is the stochastic figure in [0, 1]; and g_r1 and g_r2 are the stochastic index of the gazelle matrix.

4.2. Solution Process

For the two-layer model, an optimization algorithm with the GOA and CPLEX solver in the upper and lower tiers is used to solve the model. The solution procedure of the optimization model is given in Figure 2.

The steps can be denoted as follows:

(1)
Enter the WT, PV, and load forecasts for each typical day, electricity and gas prices, and utility parameters for energy coupling equipment;
(2)
Initialize the GOA parameters, set the digit of algorithm populations, the largest number of iterations, and the upper and lower bounds of the search space for decision variables;
(3)
Population initialization, random initialization of gazelle populations;
(4)
Determine whether the termination condition is met; if so, output the optimal solution and running scheme; if not, continue with (Step 5);
(5)
Substituting the upper-level decision variables into the lower-tier constrains the tier optimization model;
(6)
Invoke the CPLEX solver to address the lower model and get the optimal operating scheme and daily operating cost;
(7)
Calculate the objective function value and construct the elite matrix for the next search step;
(8)
Determine whether r is less than 5; if less, then carry out the exploitation operation; if not less, then determine whether the number of iterations is odd; if odd, then use Formula (37) to explore the operation; and if even, then use Formula (38) to investigate the operation;
(9)
Update the gazelle position, top gazelle, and elite matrix. Then, renew the gazelle position again using the PSRs and Equation (40) to prevent it from falling into a locally optimal solution;
(10)
Skip to Step 4.

5. Calculus Analysis

A residential park in northern China is used as an algorithm for simulation analysis. A year is divided into summer, winter, and transition season, and a typical day in summer is selected for analysis. In this paper, the wind and light output scenes are clustered by the Latin hypercube method, and the scene is reduced by the probabilistic distance fast reduction method.

The load curves of cold, heat, electricity, and gas on this typical day are shown in Figure 3. The price of gas is constant at 3.5 yuan/m³, and the time of use price curve is shown in Figure 4. For details about device parameters in the system, see Table 2. Assume that the EV holding capacity of the park is 30.

Table 2. Energy coupling equipment parameter table.

Energy equipment	Longevity (year)	Unit investment cost (yuan · Kw⁻¹)	Unit operation and maintenance costs (yuan · Kw⁻¹)	Efficiency (%)
WT	25	6000	0.01	20
PV	25	12,700	0.01	30
CHP	15	4600	0.015	Electronic: 55, hot: 45
GB	20	310	0.01	80
EB	15	2100	0.01	88
EC	15	900	0.02	80
AC	15	1050	0.01	90
EES	20	1440	0.01	90
HES	20	90	0.013	95
CES	20	136	0.01	95

5.1. Typical Daily Operating Curve Analysis

To prove the effectiveness of the proposed strategies, different DR strategies were developed to guide EV charging and discharging behavior. The policies are as follows: Strategy 1: no DR; Strategy 2: EV-PDR; and Strategy 3: EV-JRS. As shown in Figure 5, by subsidizing the load transferred by the EV, the joint response strategy proposed can better optimize the load of EVs, thus balancing the peak-valley difference of the electrical load of the system.

MEVPP optimizes system operation by using JRS to optimize EV’s load; the scheduling results of the system’s load are shown in Figure 6.

In Figure 6(a), the WT prioritizes the energy supply to the system for 1–6 h, and the shortfall of electricity is mainly satisfied by the system purchasing energy. At the same time, the system schedules EVs to charge at this time, and the EES stores the surplus power. The 11–13 and 18–22 time periods are the peak periods of power consumption, and the system schedules PV, WT, CHP, and EES to coordinate the power to supply the energy. In the 18–22 time periods, some private car EVs have finished their trips, and the system schedules the EV discharging to participate in its own optimized operation.

As shown in Figure 6(b), the heat demand in periods 1–4 is low and is mainly supplied by EB and HES. In the rest of the day, influenced by the CHP unit cogeneration, thermal energy is jointly supplied by the CHP unit and GB, while the system is equipped with a specific capacity of HES to flexibly charge and discharge the heat so that enhance the economy of the system operation.

From Figure 6(c), in the 1–6 time periods, the tariff valley hours, the EC mainly supplies the cold energy, the spare cold energy is stored by the CES. In the 12–14 h, the demand for cold energy is higher, and the CES releases the previously stored cold energy and cooperates with the cold output of the AC for a joint supply of cold energy. In the rest of the time, the cold energy demand is mainly met by the AC output.

From Figure 6(d), in 1–3 time periods, the purchased gas of the system is primarily used to satisfy the gas load demand of the customer side and the gas consumption of the GB. In the remaining period, CHP unit and GB operations consume a certain amount of natural gas, so the number of purchased gas units increases. Because the high power demand and low heat demand, the output of the CHP unit has thermoelectric coupling, so the CHP became the main gas-consuming equipment of the system in this period.

To confirm the superiority of the GOA, the PSO and wind-driven optimization (WDO) are, respectively, used to solve the model. Set the population and iterations number of the algorithms to 50 and 200. As shown in Figure 7, the WDO is easy to fall into local optimal values, and its convergence time is short. While the PSO does not easily fall into local optimal values, its convergence performance is poor. The convergence time of GOA is between the two, but its convergence ability and convergence accuracy are better than the two.

5.2. Analysis of Optimized Configuration Results

We set up different cases for comparative analysis as follows: Case 1: EV-no DR, traditional carbon trading; Case 2: EV-PDR, traditional carbon trading; Case 3: EV-JRS, traditional carbon trading; and Case 4: EV-JRS, cascade carbon trading (the case in this paper). The obtained system configuration results with annualized cost and carbon emissions are given in Tables 3 and 4.

Table 3. Collocation results in different scenarios.

Configuration capacity (Kw · h)	Case 1	Case 2	Case 3	Case 4
CHP	721	681	644	668
GB	388	401	451	437
EB	100	83	42	69
EC	543	543	548	557
AC	441	457	450	532
BT	245	234	222	247
HSE	300	232	238	253
ISE	300	292	284	277

Table 4. Annualized costs and carbon emissions for different scenarios.

Parameter name	Case 1	Case 2	Case 3	Case 4
Investors (ten thousand yuan)	102.4	100.27	97.74	100.59
Energy purchase (ten thousand yuan)	479.34	454.78	395.89	385.29
Operation and maintenance (ten thousand yuan)	14.11	13.95	13.73	13.93
Demand response (ten thousand yuan)	0	5.84	21.17	21.17
Carbon trading (ten thousand yuan)	102.17	100.06	94.75	132.93
Operation (ten thousand yuan)	595.62	574.63	525.54	553.32
Carbon emissions (t)	3929.78	3848.49	3644.24	3392.67

In Table 3, the different response strategies for EVs have different impacts on the configuration results of the system. Case 3 shows a decrease in the configured capacity of CHP and EB units compared with Cases 1 and 2. The configured capacity of the GB increases due to its lower unit configuration cost. Case 4 introduces a stepped carbon trading mechanism based on Case 3. GB produces carbon emissions during operation, while EB is relatively cleaner. The configured capacity of CHP and EB under Case 4 is increased, given the CHP units’ coupled heat and power output.

As shown in Table 4, different EV response strategies have different impacts on system annualized expenses and carbon emissions. Case 3 decreases the investment expenses of the system by 4.6% and 2.5%, and the operation cost by 11.8% and 8.5%, respectively, compared with Cases 1 and 2. This shows that JRS is effective in guiding EV charging and discharging behavior, which has a positive influence on the planning and running of the system. Meanwhile, carbon emissions in Case 3 are decreased by 7.3% and 5.2% compared with Cases 1 and 2, which also illustrates the effectiveness of the system in realizing carbon emission reduction by scheduling orderly charge and discharge of EVs. Compared with Case 3, the Case 4 further reduces the carbon emissions.

5.3. Sensitivity Analysis

5.3.1. Affect of Different Charging and Discharging Incentive Subsidies

In Section 5.2, the system configuration results and operating costs under different EV response strategies are obtained. In this section, the EV-JRS model established in Section 2.1 will be used to analyze user responsiveness under different incentive levels. Figure 8 shows a boundary graph of user responsiveness at different incentive levels. With the improvement of the incentive level, the user’s willingness to respond is increasing. When the incentive level is increased to a certain value, the user’s response reaches a saturated state.

The user responsiveness and system expense under different incentive levels as shown in Table 5 (discharge subsidy constant 0.27 yuan/KWh) and Table 6 (charging subsidy constant 0.25 yuan/KWh), respectively. The charge and discharge response capabilities under different charge and discharge subsidy prices are shown in Figure 9. The charge–discharge response capacity under different charging subsidy prices is shown in Figure 9(a). The response power is shown in Figure 9(a)(A) and 9(a)(B). The charge–discharge response capacity under different discharge subsidy prices is shown in Figure 9(b). The response power is shown in Figure 9(b)(A) and Figure 9(b)(B).

Table 5. Different charging subsidy prices.

Charging subsidies (yuan/KW · h)	0.25	0.30	0.35	0.40	0.45
User responsiveness (%)	33	42	53	70	96
Investor (ten thousand yuan)	107.34	105.25	103.32	102.73	101.24
Demand response (ten thousand yuan)	11.14	15.54	20.67	27.14	37.89
Operation (ten thousand yuan)	546.41	543.46	538.38	543.36	549.29

Table 6. Different discharge subsidy prices.

Discharge subsidies (yuan/KW · h)	0.27	0.32	0.37	0.42	0.47
User responsiveness (%)	33	41	48	59	74
Investor (ten thousand yuan)	107.34	105.92	104.53	103.01	102.52
Demand response (ten thousand yuan)	11.14	14.97	17.84	22.02	26.79
Operation (ten thousand yuan)	546.35	543.99	540.17	542.31	544.01

Table 5 shows that as the charging subsidy price increases, users’ willingness to respond continues to increase. The system can decrease the investment expense by scheduling the load of EVs. On the contrary, the running expense of the system is first reduced and then increased. This is because when the charging subsidy is set to 0.35 yuan/Kwh, the charge and discharge response of the EV access system reaches saturation. When the charging subsidy price continues to increase, the subsidy cost of MEVPP operators to charge and discharge EVs continues to augment, and the operating expenses begin to increase.

Table 6 shows the system costs at different discharge subsidy prices. With the increase of the discharge subsidy price, the charge and discharge response ability of EVs increases, and the changing trend of system investment and operating cost is the same as when the charging subsidy price changes. As shown in Figure 9(a)(B) and 9(b)(B), unlike the charging subsidy incentive, the guiding effect of the discharge subsidy incentive on EV discharge is somewhat weakened during load peaks because EV users are affected by the cost of battery loss. When the discharge subsidy is set to 0.37 yuan/Kwh, the charge–discharge response capacity of the EV access system reaches saturation, and the system has a good economic benefit.

5.3.2. Influence of Different Carbon Trading Base Prices

Based on Case 4, this paper simulates the planning results and system operation under different carbon prices.

In Figure 10, with the increase of trading price, the installed capacity of CHP units and EB appears an overall rising trend, while the installed capacity of GB appears a declining trend. As the continuous improvement of the carbon trading floor price, the system decreases the purchase of coal power with higher carbon emissions per unit by increasing the configuration capacity of CHP units fueled by natural gas and achieves low carbon emissions. Meanwhile, due to the influence of thermal and electrical cogeneration of CHP units, the GB configuration capacity of the system is reduced during operation, and there is a certain carbon emission, while the relatively environmentally friendly EB increases the configuration capacity to make up for the insufficient heat output that may be caused by the reduction of GB. As the augment of carbon trading price, the continuous increase of cogeneration units and EB equipment with higher unit configuration cost is the main reason for the continuous increase of system investment cost.

As shown in Figure 11, as the augment of carbon price, the system presents a trend of increasing annual operating expense and decreasing carbon emissions, and carbon cost first increases and then decreases. When the trading price is 400 yuan/t, the carbon emission fee reaches its peak. Since then, as the price continues to augment, carbon emissions continue to reduce under the influence of the economic operation so that the carbon trading cost begins to decline, while the annual operating cost of the system has a steady growth trend.

5.3.3. Impact of EV Holdings on System Operation

To analyze the effect of EV access numbers on the economic running, assuming that the numbers of EVs in the park is 30, 40, 50, 60, and 70, respectively, and the system is connected for simulation and comparison. As shown in Figure 12, the annual demand response cost augments as the number of EV system connections. When the access volume of EVs is 50, the annual running expenses is the lowest, and the carbon trading expenses is also low at this time, indicating that the access volume of EVs and the system achieve better supply–demand interaction and coordination, and the EV access volume reaches saturation. After that, as the number of EV access increased, the annual running expenses of the system began to increase, and the annual carbon trading costs also began to rise.

6. Conclusion

This paper is verified by simulation and concluded as follows:

(1)
For EV load, the joint response strategy developed in this paper can well optimize EV charging and discharging behavior, and the proposed strategy can enhance the economy and low carbon to a certain extent. Meanwhile, considering EV users’ willingness to respond, the formulation of reasonable incentive subsidy prices is conducive to improving the rationality of MEVPP planning and running.
(2)
Compared with traditional carbon trading, SCTM has a better low-carbon emission effect. It is worth noting that different carbon trading prices affect the planning and operating costs of the system. MEVPP operators should set reasonable carbon trading prices to balance the low-carbon and economic running.
(3)
A better-performance GOA is selected to handle the model. Compared with the PSO and WDO, the GOA has better convergence ability and convergence accuracy and enhances the accuracy of model solving.

At present, this paper assumes that the number of EV access systems is small and does not consider the influence of the uncertainty of user response. Future studies will consider the influence of large-scale EV access and the uncertainty of EV user response on the system. Further in-depth research on the low-carbon performance can be carried out, and the subsequent research can conduct in-depth studies and discussion on the carbon emissions of the whole life cycle of WT, PV, and EV.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding

This research was funded by the National Natural Science Foundation of China, grant number 52167014; the Key Research and Development Program of Gansu Province, grant number 23YFGA0084; the Major Science and Technology Special Project of Gansu Province in China, grant number 23ZDGA007; the Industrial Support Plan Project of Gansu Province in China, grant number 2024CYZC-18; and the Joint Research Fund Project of Gansu Province in China, grant number 24JRRA827.

Acknowledgments

This research was funded by the National Natural Science Foundation of China, grant number 52167014; the Key Research and Development Program of Gansu Province, grant number 23YFGA0084; the Major Science and Technology Special Project of Gansu Province in China, grant number 23ZDGA007; the Industrial support plan project of Gansu Province in China, grant number 2024CYZC-18; and the Joint Research Fund Project of Gansu Province in China, grant number 24JRRA827.

Open Research

Data Availability Statement

The data used to support the findings of this study are included within the article.

References

1 Zhang J. and Liu Z., Low Carbon Economic Scheduling Model for a Park Integrated Energy System Considering Integrated Demand Response, Ladder-type Carbon Trading and Fine Utilization of Hydrogen, Energy. (2024) 290, https://doi.org/10.1016/j.energy.2024.130311.
10.1016/j.energy.2024.130311
Web of Science® Google Scholar
2 Song T., Liu Z., and Song Z., Research and Thinking on the Aggregation and Dispatching Control Framework of Virtual Pow Er Plant’s Large Scale Flexible Resources, Electric power. (2024) 57, no. 01, 2–8.
Google Scholar
3 Wei X., Pan Z., Wang B. et al., Review on Virtual Power Plant Resource Aggregation and Collaborative Regulation Using Cloud-Tube-Edge-End Architecture, Journal of Global Energy Interconnection. (2020) 3, no. 06, 539–551.
Google Scholar
4 Zhao J., Xiang J., Tang Z. et al., Practice Exploration and Prospect Analysis of Virtual Power Plant in Shanghai, Electric power. (2023) 56, no. 2, 1–13.
Google Scholar
5 Ma L., Liang A., Wang L. et al., Review of Research on Regulation Technology of Adjustable Load Resources in Public Buildings, Electrical Measurement & Instrumentation. (2023) 60, no. 5, 1–10.
Google Scholar
6 Wang D., Ai Q., Tao Y. et al., The Technical Challenges and Solutions for Trusted Transactions in Virtual Power Plants, Automation of Electric Power Systems. (2024) .
Google Scholar
7 Nosratabadi S. M., Hooshmand R. A., Gholipour E., and Parastegari M., A New Simultaneous Placement of Distributed Generation and Demand Response Resources to Determine Virtual Power Plant, International Transactions on Electrical Energy Systems. (2016) 26, no. 5, 1103–1120, https://doi.org/10.1002/etep.2128, 2-s2.0-84940825093.
10.1002/etep.2128
Web of Science® Google Scholar
8 Hong W., Xiao W., Bing L. et al., Optimal Configuration Method of Electric Heating in a Virtual Power Plant with Peak Load Regulation Considering a Wind Power Delivery Protocol, Power System Protection and Control. (2021) 49, no. 19, 135–144.
Google Scholar
9 Wei Z., Chen Y., Huang W. et al., Optimal Allocation Model for Multi-Energy Capacity of Virtual Power Plant Considering Conditional Value-At-Risk, Automation of Electric Power Systems. (2018) 42, no. 4, 39–46.
CAS Google Scholar
10 Luo S., Teng J., and Tan Z., Optimal Allocation of Energy Storage in Distribution Network Considering Aggregate Regulation of Electric Vehicles, Energy Storage Science and Technology. (2023) 12, no. 11.
Google Scholar
11 Kang D., Wei W., and Fan Y., Optimal Configuration of Frequency Modulation Capacity of Virtual Power Plant Participating System Based on TVF-EMD, Modern Electronics Technique. (2023) 46, no. 5, 180–186.
Google Scholar
12 Li L., Tong W., and Xian C., Multi-objective Optimal Allocation of DG and EV Charging Station Based on Space-Time Characteristics and Demand Response, Electric power automation equipment. (2021) 41, no. 11, 48–56.
CAS Google Scholar
13 Yan H., Hu J., Huang D. et al., Distribution Network Planning Considering the Flexibility of EV Virtual Power Plants and High Penetration of PV, Electric Power Construction. (2022) 43, no. 11, 14–23.
Google Scholar
14 Lei Y., Wang D., Jia H. et al., Multi-objective Stochastic Expansion Planning Based on Multi-Dimensional Correlation Scenario Generation Method for Regional Integrated Energy System Integrated Renewable Energy, Applied Energy. (2020) 276, https://doi.org/10.1016/j.apenergy.2020.115395.
10.1016/j.apenergy.2020.115395
PubMed Web of Science® Google Scholar
15 Guo J., Wang J., and Zhang H., Scheduling Optimization Strategy Based on Virtual Power Plant for Multi-Energy Collaborative System, Electric Power Construction. (2022) 43, no. 12, 141–151.
Google Scholar
16 Ling Z., Xing W., Ying D. et al., Optimal Dispatch of Multi-Energy Virtual Power Plant Considering Time-Of-Use Electricity Price and CSP Plant, Electric Power Construction. (2022) 43, no. 04, 119–129.
Google Scholar
17 Lin C., Bing W., and Nan W., Optimal Operation Strategy of Virtual Power Plant Considering Integrated Demand Response in Demand Side, Science Technology and Engineering. (2021) 21, no. 17, 7133–7142.
Google Scholar
18 Yang Y., Li Q., Pu Y. et al., Optimal Configuration of CHHP Island IES Considering Different Charging Modes of Electric Vehicles, Power System Technology. (2022) 46, 3869–3880.
Google Scholar
19 Ji L., Liang X., Xie Y., Huang G., and Wang B., Optimal Design and Sensitivity Analysis of the Stand-Alone Hybrid Energy System with PV and Biomass-CHP for Remote Villages, Energy. (2021) 225, https://doi.org/10.1016/j.energy.2021.120323.
10.1016/j.energy.2021.120323
Web of Science® Google Scholar
20 Graça Gomes J., Xu H., Yang Q., and Zhao C., An Optimization Study on a Typical Renewable Microgrid Energy System with Energy Storage, Energy. (2021) 234, https://doi.org/10.1016/j.energy.2021.121210.
10.1016/j.energy.2021.121210
Google Scholar
21 Kong J., Kim S. T., Kang B. O., and Jung J., Determining the Size of Energy Storage System to Maximize the Economic Profit for Photovoltaic and Wind Turbine Generators in South Korea, Renewable and Sustainable Energy Reviews. (2019) 116, https://doi.org/10.1016/j.rser.2019.109467, 2-s2.0-85072983436.
10.1016/j.rser.2019.109467
Web of Science® Google Scholar
22 Wang D., Ge Y., Cao J., Lin Q., and Chen R., Charging Load Forecasting of Electric Vehicles Based on Sparrow Search Algorithm-improved Random Forest Regression Model, Journal of Engineering. (2023) 2023, no. 6, https://doi.org/10.1049/tje2.12280.
10.1049/tje2.12280
Google Scholar
23 Li Z., Li C., Zhang B. et al., Research on New Energy Vehicle Charging Prediction Based on Monte Carlo Algorithm and its Impact on Distribution Network, Frontiers in Energy Research. (2023) 11, https://doi.org/10.3389/fenrg.2023.1269041.
10.3389/fenrg.2023.1269041
Web of Science® Google Scholar
24 Zhou X., Huang X., Zhang Y. et al., Real-time Scheduling and Optimization Model of Electric Vehicles Based on Fuzzy Evaluation of Demand Response Potential, Electric Power Automation Equipment. (2022) 42, no. 10, 30–37.
Google Scholar
25 Wang J., Jia Y., Mi Z. et al., Reserve Service Strategy of Electric Vehicles Based on Double-Incentive Mechanism, Automation of Electric Power Systems. (2020) 44, no. 10, 68–76.
CAS Google Scholar
26 Lin S., Wen S., Zhu M. et al., Optimal Allocation of Electric-Thermal Hybrid Energy Storage for Seaport Integrated Energy System Considering Carbon Trading Mechanism, Journal of Shanghai Jiao Tong University. (2024) 1–25.
Google Scholar
27 Chen J., Hu Z., Chen J. et al., Optimal Dispatch of Integrated Energy System Considering Ladder-type Carbon Trading and Flexible Double Response of Supply and Demand, High Voltage Engineering. (2021) 47, no. 9, 3094–3106.
Google Scholar
28 Agushaka J. O., Ezugwu A. E., and Abualigah L., Gazelle Optimization Algorithm: a Novel Nature-Inspired Metaheuristic Optimizer, Neural Computing & Applications. (2023) 35, no. 5, 4099–4131, https://doi.org/10.1007/s00521-022-07854-6.
10.1007/s00521-022-07854-6
Web of Science® Google Scholar
29 Qin S., Zeng H., Sun W., Wu J., and Yang J., Multi-Strategy Improved Particle Swarm Optimization Algorithm and Gazelle Optimization Algorithm and Application, Electronics. (2024) 13, no. 8, https://doi.org/10.3390/electronics13081580.
10.3390/electronics13081580
Web of Science® Google Scholar
30 Syama S., Ramprabhakar J., Anand R., and Guerrero J. M., A Hybrid Extreme Learning Machine Model with Levy Flight Chaotic Whale Optimization Algorithm for Wind Speed Forecasting, Results in Engineering. (2023) 19, https://doi.org/10.1016/j.rineng.2023.101274.
10.1016/j.rineng.2023.101274
Web of Science® Google Scholar
31 Wu D., Wu L., Wen T., and Li L., Microgrid Operation Optimization Method Considering Power-To-Gas Equipment: An Improved Gazelle Optimization Algorithm, Symmetry. (2024) 16, no. 1, https://doi.org/10.3390/sym16010083.
10.3390/sym16010083
Web of Science® Google Scholar
32 Hasanien H. M., Alsaleh I., Tostado-Véliz M., Alassaf A., Alateeq A., and Jurado F., Optimal Parameters Estimation of Lithium-Ion Battery in Smart Grid Applications Based on Gazelle Optimization Algorithm, Energy. (2023) 285, https://doi.org/10.1016/j.energy.2023.129509.
10.1016/j.energy.2023.129509
Web of Science® Google Scholar

All articles

Configuration capacity (Kw · h)	Case 1	Case 2	Case 3	Case 4
CHP	721	681	644	668
GB	388	401	451	437
EB	100	83	42	69
EC	543	543	548	557
AC	441	457	450	532
BT	245	234	222	247
HSE	300	232	238	253
ISE	300	292	284	277

Configuration capacity (Kw · h)	Case 1	Case 2	Case 3	Case 4
CHP	721	681	644	668
GB	388	401	451	437
EB	100	83	42	69
EC	543	543	548	557
AC	441	457	450	532
BT	245	234	222	247
HSE	300	232	238	253
ISE	300	292	284	277

Joint Optimization of Multienergy Virtual Power Plant Configuration and Operation Considering Electric Vehicle Access

Abstract

1. Introduction

2. Structural Diagram of MEVPP Considering EV Access

2.1. EV-JRS Model

2.2. SCTM

3. MEVPP Two-Tier Planning Model

3.1. Upper-Tier Capacity Allocation Model

3.1.1. Objective Function

3.2. Lower-Tier Optimization Operational Model

3.2.1. Objective Function

3.2.2. Restrictive Condition

3.2.2.1. Energy Balance Constraints

3.2.2.2. Heat Balance Constraints

3.2.2.3. Cold Energy Balance Constraints

3.2.2.4. Gas Balance Constraints

3.2.2.5. Unit Climbing Constraints

3.2.2.6. Energy Storage Constraints

3.2.2.7. EV Demand Response Constraints

4. Model Solution

4.1. Gazelle Optimization Algorithm (GOA)

4.1.1. Initialization

4.1.2. Development Phase

4.1.3. Discovery Phase

4.2. Solution Process

5. Calculus Analysis

5.1. Typical Daily Operating Curve Analysis

5.2. Analysis of Optimized Configuration Results

5.3. Sensitivity Analysis

5.3.1. Affect of Different Charging and Discharging Incentive Subsidies

5.3.2. Influence of Different Carbon Trading Base Prices

5.3.3. Impact of EV Holdings on System Operation

6. Conclusion

Conflicts of Interest

Funding

Acknowledgments

Open Research

Data Availability Statement

References

Figures

References

Related

Information

Configuration capacity (Kw · h)	Case 1	Case 2	Case 3	Case 4
CHP	721	681	644	668
GB	388	401	451	437
EB	100	83	42	69
EC	543	543	548	557
AC	441	457	450	532
BT	245	234	222	247
HSE	300	232	238	253
ISE	300	292	284	277