With the rapid growth of installed capacity for renewable energy sources such as wind power and photovoltaic power, the peak-to-valley load difference in the power grid has significantly increased (Lin et al., 2022), leading to a growing demand for the power grid’s ability to perform deep peak shaving with thermal power units (Wang J. et al., 2021). Renewable energy sources represented by wind power and photovoltaic power exhibit volatility and intermittency, resulting in frequent imbalances between supply and demand in the power grid and causing prominent issues of “curtailment of wind and solar power” (Haijiao et al., 2023). Coal-fired power plants must frequently participate in deep peak shaving to balance the intermittency and volatility of renewable energy generation (Zhang et al., 2022). In this context, Coal-fired power plants need to undergo flexibility upgrades to enable operation at lower loads (e.g., 30%–40% of rated load), thereby creating space for renewable energy grid integration (Zhiqiang et al., 2020).

Coal-fired power plants face two major issues during low-load operation: first (Haijiao et al., 2021), heating demand requires the unit to maintain a certain steam extraction volume, while deep peak shaving necessitates reducing power generation, leading to difficulties in thermal-electric decoupling for heating units. (Hou et al., 2024). Their peak shaving capacity is severely constrained by heating demand, resulting in limited load adjustment ranges under conventional steam extraction modes (Chen et al., 2020); second, under low-load conditions, unreasonable extraction pressure settings can reduce the unit’s thermal efficiency and economic performance (Gao et al., 2020). Steam extraction heating units urgently need to overcome this bottleneck through thermal-electric decoupling technology (Zheng et al., 2021; Song et al., 2021a).

Existing research indicates that technologies such as electric boilers (Wenping et al., 2018), thermal storage tanks (Sun, 2024), and molten salt thermal storage (Yang et al., 2024) can partially alleviate the thermal-electric decoupling contradiction,but their applicability to extraction heating units varies significantly: Electric boilers have fast response but high energy consumption; thermal storage tanks are suitable for short-term peak shaving with low initial investment; molten salt storage has large capacity but high operational costs (Wang C. et al., 2021). Additionally, the economic efficiency of steam extraction heating is influenced by main steam parameters, cylinder efficiency, and heat load distribution schemes (Elwardany, 2024), and when multiple units operate in coordination, mode combinations must be optimized to balance peak-shaving capacity and economic benefits (Song et al., 2021b).

In summary, under the requirements of low-carbon and flexibility (Jianglong et al., 2022), studying the regulation capacity, energy consumption characteristics, load distribution, and thermal economy of renovated heating units is of great significance (Weinberger et al., 2021; Zhang et al., 2024). So the objective of this study is to enhance the deep peak shaving performance and economic efficiency of thermal power heating units by developing an optimization model integrating decoupling technologies. The novelty lies in the construction of a dual-objective optimization framework, the integration of multiple thermal-electric decoupling schemes, and detailed scenario-based evaluation of dual-unit coordination strategies under different load conditions. This provides reference for the operation and profitability of Coal-fired power plants, promotes the consumption of new energy, reduces carbon emissions, and contributes to the realization of green and low-carbon energy development.

When performing thermodynamic modeling, the design operating conditions of the unit are typically used as the rated operating conditions (Huang et al., 2025). Under these conditions, various thermodynamic parameters—including main steam and reheat steam parameters, turbine extraction port parameters, condenser parameters, deaerator parameters, shaft seal parameters, and others—as well as data such as turbine pressure loss, heater parameters, and drain end difference, are all determined through the thermodynamic system’s heat balance diagram (Xilin et al., 2024). Based on the thermal balance diagram, a complete thermal power plant simulation model can be constructed in the EBSILON software. The simulation model of a 300 MW heating unit is shown in Figure 1. The 300 MW heating unit simulation model (Figure 1) includes core components: high-pressure cylinder (HP), intermediate-pressure cylinder (IP), low-pressure cylinder (LP), deaerator, heaters, and condenser. Key streams in Figure 1 are labeled as follows: (1) Main steam entering HP; (2) Extraction steam from HP; (3) Reheat steam entering IP; (4) Extraction steam from IP to deaerator; (5) Condensate water to the boiler, which together form the closed thermodynamic cycle of the unit.

The feasible range for mixed extraction heating of a 300 MW unit is shown in Figure 2.

As can be seen from the heating feasibility domain, by complementing Scheme 1 with Scheme 3, Scheme 1 is prioritized under THA conditions to ensure power generation efficiency. When pressure is insufficient, the system switches to Scheme 3, leveraging its lower power threshold to enhance peak-shaving flexibility, thereby balancing heating demand with the power generation adjustment range. This demonstrates that the coordinated operation of multiple schemes is an effective approach to achieving a flexible balance between heating and power generation.

The heating feasibility domain for the 600 MW unit is shown in Figure 3.

When using mixed steam extraction for heating, the feasible heating range cannot necessarily be determined by the maximum power generation range, and some operating conditions may not be achievable.

To analyze the economic efficiency of heating units, it is first necessary to construct an economic evaluation model that takes into account factors such as electricity sales revenue, heat sales revenue, peak load compensation, fuel costs, and investment costs for heat-electricity decoupling equipment. The economic evaluation model is shown in Equation 1. E = C e P e + C h Q h + C p Δ P peak − C coal B coal − C OM V OM (1)

In the formula, E represents the net revenue of the thermal power plant, in yuan; C_e represents the revenue from electricity sales, in yuan/kWh; P_e represents the electricity generation of the thermal power plant, in MWh; C_h represents the revenue from heat sales, in yuan/t; Q_h represents the heat supply of the thermal power plant, in t; C_p represents the peak shaving compensation, in yuan/kWh; ∆P_peak is the deep peak shaving power generation capacity, measured in MWh; C_coal is the standard coal price, measured in yuan/t; B_coal is the coal consumption, measured in t; C_OM is the investment cost of the thermal-electric decoupling equipment, measured in yuan/kWh; V_OM is the capacity of the thermal-electric decoupling equipment, measured in MWh.

The low point of the case study unit’s electrical load occurs during the 8-h period from 23:00 to 7:00 at night, while the daytime peak load lasts for 16 h. During calculations, the electricity sales revenue is set at 0.36 yuan/kWh, the heat sales revenue at 200 yuan/ton, the peak shaving compensation at 0.396 yuan/kWh, and the standard coal price at 900 yuan/ton of standard coal.

When using a thermal storage system for thermal-electric decoupling, this scheme considers the daytime 16 h as the thermal storage system’s charging time and the nighttime 8 h as the thermal storage system’s discharging time, fully utilizing the thermal storage system’s peak-shaving and valley-filling regulatory role in the heating process.

Based on the unit’s operational conditions, it is assumed that the average power generation capacity of the unit is 200 MW, and the average heating capacity is 100 kg/s. The thermal storage tank has a capacity of 1,000 MWh. During the daytime 16 h, in addition to supplying 100 kg/s of heating steam externally, the unit provides 25 kg/s of heating steam to the thermal storage tank. During the night, the unit supplies 50 kg/s of heat externally for 8 h, with the remaining 50 kg/s supplemented by the thermal storage tank (Zhou et al., 2025). The specific calculation results for the unit’s performance indicators before and after the thermal storage tank are shown in Table 1.

After installing a thermal storage tank, compared to a conventional heating scheme: during the day, an additional 90 tons of standard coal will be consumed over 16 h, but at night, 126 tons of standard coal will be saved, resulting in a total reduction of 36 tons in coal consumption; fuel costs will decrease by 40,000 yuan, with significant savings at night, and net profits will increase by 40,000 yuan. However, considering the initial investment cost of the thermal storage tank is 500,000 yuan, plus subsequent operational and maintenance costs, the thermal storage tank must be used long-term for peak shaving to achieve benefits.

The molten salt thermal storage system has a capacity of 1,000 MWh. During the daytime 16 h, the unit supplies 100 kg/s of heating steam externally and provides 25 kg/s of heating steam to the molten salt thermal storage system. During the nighttime 8 h, the unit supplies 50 kg/s of heating steam externally, with the remaining 50 kg/s supplemented by the molten salt thermal storage system. The specific calculation results of the unit’s various indicators before and after the addition of the molten salt thermal storage system are shown in Table 2.

After adding molten salt thermal storage, compared to the conventional heating scheme: during the day, an additional 28 tons of standard coal will be consumed over 16 h, but at night, 126 tons of standard coal will be saved, resulting in a total reduction of 98 tons of coal consumption; fuel costs will decrease by 90,000 yuan, with significant savings at night, and net profits will increase by 90,000 yuan. However, considering the initial investment cost of 1.5 million yuan and the relatively high subsequent operational and maintenance costs, long-term use of auxiliary peak shaving is required to achieve benefits.

The electric boiler has a capacity of 1,000 MWh. During the daytime 16 h, the unit supplies heat at a rate of 100 kg/s, with the electric boiler remaining idle. During the nighttime 8 h, the unit supplies heat at a rate of 50 kg/s, with the remaining 50 kg/s provided by the electric boiler. At this point, the unit’s power output is 150 MW. The specific calculation results for various indicators of the unit before and after the addition of the electric boiler are shown in Table 3.

After installing an electric boiler, compared to the conventional heating scheme: the standard coal consumption during the day remains the same at 16 h, but an additional 57 tons of standard coal will be consumed at night; fuel costs increase by 50,000 yuan, net profits decrease by 50,000 yuan, and the initial investment cost for the electric boiler is 800,000 yuan, with ongoing operational and maintenance costs. However, the electric boiler has superior deep peak shaving performance, making it suitable for use when a significant increase in peak shaving depth is required.

To further leverage the deep peak shaving performance advantages of the electric boiler, the nighttime unit peak shaving depth was reduced from 40% to 30%. The specific calculation results for all indicators are shown in Table 4.

Under 30% load conditions, compared to 40% load conditions: nighttime consumption will be reduced by 32 tons of standard coal; fuel costs will decrease by 30,000 yuan, deep peak shaving compensation will increase by 90,000 yuan, and net benefits will increase by 120,000 yuan.

As shown in Figure 4, based on the comparison and analysis of the above four sets of data, it can be concluded that when the depth of peak shaving for heating units is small, using thermal storage tanks and molten salt thermal storage yields better returns. Among these, thermal storage tanks require lower initial investment and have lower operational costs, but molten salt thermal storage offers better long-term returns. Electric boilers do not yield good returns when the depth of peak shaving is small, but as the depth of peak shaving increases, the returns gradually improve.

The core objective of constructing a dual-machine online operation system is to synergistically optimize thermodynamic efficiency and economic indicators. The fixed maximum load unit serves as the base load unit, continuously outputting stable thermal energy while maintaining optimal efficiency. The parallel variable load unit assumes the peak shaving function, dynamically adjusting to respond to load changes. The key advantages of this configuration are: 1) Base-load units avoid efficiency penalties caused by frequent load adjustments; 2) Peak-shaving units limit their operating range to control the decline in combustion efficiency; 3) The dual-heat-source parallel structure enhances system redundancy, enabling the system to maintain 50% heating capacity even in the event of a single-unit failure, significantly improving supply reliability (Yi and Wen, 2024). This coordinated operation strategy fundamentally achieves a Pareto improvement between static efficiency optimization and dynamic response capability, providing a feasible engineering solution for flexible regulation of heating units under high renewable energy penetration scenarios.

Optimization objective 1 function is to optimize deep peak shaving performance, i.e., minimize the power generation of the interconnected system: min ⁡ P L = ∑ i N P i (2)

In the Formula 2: N represents the number of heating units in the online operation system; P_L represents the generating capacity of the heating units in the online operation system (unit: MW); P_i represents the generating capacity of the ith unit (unit: MW);.

Optimization objective 2 is to minimize the power generation coal consumption rate. The power generation coal consumption rate is a parameter that characterizes the energy consumption characteristics of power generation units. In heating units, selecting the power generation coal consumption rate as the optimization objective not only reflects the characteristics of energy utilization in a cascade system but also allows the power generation coal consumption rate of units under the same load demand to serve as an evaluation criterion for assessing the thermal economic efficiency of different units. When calculating the coal consumption rate for power generation, the thermal consumption of the unit must first be calculated, and the heating quantity subtracted to obtain the thermal consumption for power generation. The formula for calculating the thermal consumption rate for power generation of the unit is: q t , i = Q T , i − Q g , i η B , i η P , i P i (3)

In the Formula 3: q_t,i is the thermal consumption rate of the ith unit (unit: kJ/kWh); Q_T,i is the thermal consumption of the ith unit (unit: kJ); Q_g,i is the heat supply of the ith unit (unit: kJ); η_B,i is the boiler efficiency of the ith unit; η_P,i is the pipeline thermal efficiency of the ith unit.

The formula for calculating the coal consumption rate of the unit is: b t , i = q t , i 1000 Q 0 (4)

In the Formula 4: b_t,i is the coal consumption rate per unit of electricity generated by the ith unit (unit: g/kWh); Q₀ is the lower heating value of standard coal (unit: kJ/kg), which is taken as 29271.2 kJ/kg according to data from the National Bureau of Statistics.

The objective function for Optimization Goal 2 is to minimize the total coal consumption rate per unit of electricity generated by the online operation system is: min b t = ∑ i N b t , i · P i P L (5)

In the Formula 5: b_t is the total coal consumption rate of the online operating system (unit: g/kWh).

1.Peak load demand constraint: The total power generation P_L of the heating units in the online operating system meets the external peak load demand.

In the Formula 6: P_T is the rated total power generation capacity of the online operation system (unit: MW); k is the peak shaving depth.

2.Heating balance constraint: The total heating capacity M_L of the heating units in the online operation system is equal to the heat load demand of the surrounding industrial parks.

In the Formula 7: M_i is the heat supply of the ith unit (unit: kg/s).

3.Single unit heat supply constraint: There is a maximum value for the heat supply of each heating unit.

In the Formula 8: M_i,max is the maximum heat supply capacity of the ith unit (unit: kg/s).

4.Minimum steam flow constraint for the low-pressure cylinder: The steam flow into the low-pressure cylinder of the steam turbine must be maintained above the minimum steam flow required for safe operation of the low-pressure cylinder.

In the Formula 9: M_i,IL,min represents the minimum steam flow rate required for the safe operation of the low-pressure cylinder of the ith unit (unit: kg/s); M_i,IL represents the steam inlet flow rate of the low-pressure cylinder of the ith unit (unit: kg/s).

First, the minimum power generation of the online operation system is set as the optimization objective, with heating balance constraints, single-unit heating constraints, and minimum low-pressure cylinder steam inlet flow rate constraints as the constraint conditions. The genetic algorithm is used as the solution algorithm to optimize the minimum power generation of the online operation system. It is then determined whether the minimum power generation meets the peak shaving demand constraints. If it does not meet the constraints, the process is terminated directly. If it meets the constraints, the minimum coal consumption rate of the online operation system is set as the optimization objective. with peak shaving demand constraints, heating balance constraints, single-unit heating constraints, and minimum steam inlet flow rate constraints for the low-pressure cylinder as constraint conditions, and using a genetic algorithm as the solution algorithm, the optimal allocation of heating steam extraction volume for the heating units and the minimum power generation coal consumption rate of the online operation system when meeting peak shaving demand are optimized. The optimization process is shown in Figure 5.

The following describes the online operation of heating units of different models, using a 300 MW unit and a 600 MW unit. Based on whether the units are operating under maximum boiler load conditions, two operational modes can be distinguished:

Operational Mode 1: The 300 MW unit serves as the base-load unit, operating under maximum boiler load conditions, while the 600 MW unit acts as the peak-shaving unit, operating under variable conditions to assist the 300 MW unit in deep peak shaving.

Operation Mode 2: The 600 MW unit operates as the base-load unit at the boiler’s maximum load condition, while the 300 MW unit operates as the peak-shaving unit, adjusting its operating conditions to assist the 600 MW unit in deep peak shaving.

The constraint conditions are set as follows:

1.Peak-shaving demand constraint: Peak-shaving depth is 50%; the rated generating capacity of the combined operation system is 900 MW.

Based on the configuration of the electric boiler, four configuration schemes can be identified. Using Operation Mode 1, each of the four configuration schemes is optimized, with the optimization results shown in Table 5. This table is intended to compare the impact of four different electric boiler configurations on system output and efficiency under Operation Mode 1. The data helps determine which schemes satisfy deep peak shaving demands while balancing coal consumption.

The optimization results indicate that when the online operation system is not equipped with an electric boiler, the power generation capacity is 515.282 MW, which does not meet the 50% deep peak shaving requirement. This conclusion is consistent with that of a case study involving the online operation of two units of the same model. To meet the deep peak shaving requirement, the online operation system needs to be equipped with thermal-electric decoupling technology. Among the configuration schemes that meet the 50% deep peak shaving requirement, Configuration Scheme 2 lacks further peak shaving capability, while Configuration Schemes 3 and 4 have good deep peak shaving capability. Configuration Scheme 3 can achieve a 30% peak shaving depth, and Configuration Scheme 4 can achieve a 20% peak shaving depth. Analyzing the optimization results of the four configuration schemes, as shown in Figure 6, it can be observed that when different models of dual units operate in Mode 1, the system prioritizes allocating thermal load to the 300 MW heating unit to achieve the minimum coal consumption rate, thereby keeping the 300 MW heating unit in maximum heating mode.

An analysis of the economic efficiency of the above four configuration schemes per unit of time is shown in Table 6. Configuration scheme 1 does not meet the deep peak shaving requirements and does not receive deep peak shaving compensation. As shown in Figure 7, configuration schemes 3 and 4 rely on deep peak shaving compensation and have good net returns.

Using operating mode 2, the four configuration schemes were optimized separately, and the optimization results are shown in Table 7. This table evaluates the same configuration schemes under Operation Mode 2. It illustrates the variation in heating output and system coal consumption when roles of the 300 MW and 600 MW units are reversed.

The optimization results indicate that when the online operation system adopts Operation Mode 2, Configuration Scheme 1, Configuration Scheme 2, and Configuration Scheme 4 cannot meet the 50% deep peak shaving requirement. Although Configuration Scheme 3 can meet the 50% deep peak shaving requirement, it lacks the ability for further peak shaving. Analyzing the optimization results of the four configuration schemes, as shown in Figure 8, it can be observed that when different models of dual-unit systems operate in online mode using Operation Mode 2, the system adjusts the 300 MW heating unit to operate at 70% heating capacity to achieve the minimum coal consumption rate.

An analysis of the economic efficiency of the four configuration schemes mentioned above within a unit of time is shown in Table 8. None of the four configuration schemes in Operation Mode 2 met the peak shaving requirements. As shown in Figure 9, the net profit was at a relatively low level in the absence of peak shaving compensation.

A comprehensive analysis of the above eight sets of optimized data, as shown in Figure 10, reveals that operating mode 1 has better peak shaving performance than operating mode 2. Under this operating mode, both configuration scheme 3 and configuration scheme 4 have good deep peak shaving performance, with the latter having superior deep peak shaving capabilities.

In order to conduct a comprehensive technical and economic comparison of Configuration Scheme 3 and Configuration Scheme 4 under Operating Mode 1, the peak shaving demand was segmented and restricted, and Configuration Scheme 4 was optimized. The optimization results are shown in Table 9.

As shown in Figure 11, the optimization results indicate that as the power generation capacity of the online operating system decreases, the coal consumption rate of the system gradually increases. When the peak shaving depth of Configuration Scheme 4 is 35%, the coal consumption rate of the system is 249.306 g/kWh. When the coal consumption rates are comparable, Configuration Scheme 3 demonstrates superior peak shaving performance.

An analysis of the economic efficiency of the above three configuration schemes per unit time is shown in Table 10. As the peak shaving depth increases, the net revenue of the online operation system relying on peak shaving compensation continues to increase, as shown in Figure 12.

Based on the above analysis, it can be concluded that when operating dual-unit systems of different models, to optimize the technical and economic benefits of the system, the smaller-scale heating unit should be designated as the base-load unit, operating at the maximum boiler load condition, while the larger-scale heating unit should serve as the peak-shaving unit, operating under variable conditions to assist the base-load unit in deep peak shaving. When meeting deep peak shaving requirements, equipping the peak shaving unit with an electric boiler can optimize the system’s thermal economic efficiency. Equipping both units with electric boilers can further enhance the system’s deep peak shaving capability, but this will significantly reduce the system’s thermal economic efficiency.

Based on multi-objective collaborative optimization theory, this study systematically investigates the key technical issues related to enhancing the deep peak-shaving performance and optimizing the thermal economy of online-operating heating units. By constructing a dual-objective optimization model that minimizes power generation and coal consumption rates, combined with genetic algorithms and a thermal system simulation platform, global optimization of unit operating parameters under multiple steam extraction schemes was achieved. The study focused on dual-unit online systems of the same and different models, conducting multi-scenario simulation experiments to reveal the mechanisms by which thermal-electric decoupling technology influences system peak shaving capacity and energy consumption characteristics. The simulations were carried out using the EBSILON software.

A comprehensive technical and economic optimization method for deep peak regulation performance of heating units based on genetic algorithms and multiple steam extraction schemes was proposed. It was demonstrated that fixing the base-load unit at the maximum boiler load condition can maintain the optimal thermal efficiency range, which improves simulation accuracy by 8.7% compared to traditional method. While adopting a sliding pressure regulation strategy for the peak regulation unit can reduce the system’s power generation capacity by 23.6%–41.2%, verifying the technical feasibility of differentiated operation strategies.

A coupling pattern between unit capacity ratios and operating modes was identified: when 300 MW and 600 MW units adopt the “small units base load + large units peak shaving” mode, the system’s net revenue increases by 67.3% compared to the reverse configuration, providing important reference value for optimizing systems with mismatched units.

The novelty of this study lies in that the proposed comprehensive technical and economic optimization method provides quantitative basis for selecting interconnection system configuration schemes and lays the theoretical foundation for establishing coordinated operation modes between power and heat systems under the new power system framework. Furthermore, this method has potential practical significance in reducing operational costs, enhancing the adaptability of power plants under fluctuating renewable generation scenarios, and supporting policy-driven targets for carbon emission reduction.