Energy Optimization Model of Multi-Energy System Considering Distributed Photovoltaic Coordinated Carbon Reduction

Abstract

This report addresses the challenges of energy optimization in thermal power plant stations operating under carbon reduction constraints, focusing on complex nonlinear, multivariable, and strongly coupled characteristics within multi-energy systems (MES). A novel energy optimization method incorporating distributed photovoltaic (PV) coordination for carbon reduction is proposed. The method establishes an energy characteristic model of MES, analyzes interactions among system components, and develops a carbon emission intensity model. An energy optimization model is formulated with objectives to minimize operating costs and maximize carbon emission reductions, solved using a deep reinforcement learning algorithm. Validation is performed via simulation based on operational data from a power plant in Northeast China, demonstrating the method’s effectiveness in supporting Sustainable Development Goals (SDGs), particularly SDG 7 (Affordable and Clean Energy) and SDG 13 (Climate Action).

Introduction

Research Background and Problem

The global transition towards clean and low-carbon energy systems necessitates shifting from controlling energy consumption to controlling carbon emissions, aligning with SDG 13. Thermal power plants remain critical energy sources in the medium term, requiring enhanced carbon reduction strategies while maintaining energy balance. The integration of renewable energy sources such as wind and PV demands flexible upgrades in thermal power plants, including heat energy storage, electric boilers, waste heat recovery, and carbon capture technologies, contributing to SDG 7 and SDG 13.

However, multi-energy balance optimization in systems coupling wind, PV, electricity, heat, and storage presents significant challenges, increasing the complexity of carbon reduction efforts.

Literature Review on Carbon Reduction Technologies for Thermal Power Plants

• Studies have explored coupling regulation characteristics of carbon capture power plants and scheduling models to achieve low-carbon economic operations.
• Pollution reduction and carbon efficiency evaluation methods for coal-fired power plants have been developed.
• Load distribution models minimizing carbon emissions and generation costs have been proposed.
• Lifecycle assessments of coal-fired units with carbon capture and photovoltaic integration have been conducted.
• Carbon reduction strategies under carbon trading mechanisms and CCUS technologies have been investigated.
• Coordinated optimization scheduling models coupling renewable energy and carbon capture have been studied to reduce carbon intensity.
• Low-carbon retrofitting schemes such as sludge co-firing, biomass coupling, and rooftop PV installation have been evaluated.

These efforts contribute to SDG 7 and SDG 13 by promoting cleaner energy production and reducing greenhouse gas emissions.

Research Gaps and Contributions

1. Integration of diverse resources (distributed PV, battery storage, heat storage) to support low-carbon MES operation remains underexplored.
2. Coordination among multiple ‘source-grid-load’ units to enhance carbon reduction requires further study.
3. Addressing nonlinear and multivariable factors in low-carbon energy optimization with extensive equipment and operational data is challenging.

This report proposes an energy optimization model of MES considering distributed PV coordination and carbon reduction, linking operational output, energy balance, and carbon emissions. The model employs deep reinforcement learning (specifically DDPG algorithm) for solution and is validated with real operational data from a Northeast China power plant. The approach advances SDG 7 and SDG 13 by optimizing energy use and reducing carbon emissions.

Multi-Energy System Model

System Structure

The MES comprises combined heat and power (CHP) units, renewable energy equipment (distributed PV), heat storage, electric boilers, heat recovery boilers, battery energy storage, carbon capture equipment, and multi-energy loads. It serves internal energy needs and exports electricity and heat to residents, supporting SDG 7.

Energy Supply Equipment Models

1. CHP Unit Model: Describes the relationship between electrical and heat energy outputs.
2. Electric and Heat Recovery Boilers: Models heat output and waste heat reuse efficiency.
3. Heat and Battery Energy Storage: Models charging/discharging dynamics, efficiencies, and storage limits.
4. Carbon Capture Equipment: Models energy consumption and carbon capture capacity to reduce emissions, directly contributing to SDG 13.
5. Multi-Energy Loads: Models controllable electric and heat loads with constraints on building temperatures.

Energy Characterization Models

• Fuel Balance: Ensures reliable fuel supply and accounts for storage and losses.
• Heat Balance: Analyzes heat input, output, and losses for operational efficiency.
• Electric Energy Balance: Accounts for generation, consumption, export, and purchase of electricity.

Carbon Intensity Model

The carbon emission intensity of the MES is modeled based on fuel combustion and carbon capture operations, enabling quantification of emissions per unit of electricity output. This model supports carbon accounting aligned with SDG 13.

Energy Optimization Model of Multi-Energy System

Objective Functions

1. Operating Costs: Includes costs of CHP units, electric and heat recovery boilers, carbon capture equipment, purchased electricity, storage, fuel consumption, and load regulation.
2. Carbon Emission Costs: Based on carbon allowances, actual emissions, and a reward-penalty mechanism incentivizing emission reductions, supporting SDG 13.

The overall optimization aims to minimize both operating and carbon emission costs, balancing economic and environmental objectives.

Constraints

• Electric power balance ensuring supply meets demand.
• Heat power balance maintaining thermal energy equilibrium.
• Capacity limits on energy supply equipment outputs.
• Adjustable ranges for controllable electric and heat loads within comfort constraints.

Energy Optimization Model Based on Deep Reinforcement Learning

Markov Decision Process Formulation

• State Set: Includes renewable output, uncontrollable loads, electricity cost, carbon quota, building temperatures, and time.
• Action Set: Decisions on CHP output, boiler heat output, storage charging/discharging, carbon capture operation, and controllable loads.
• Reward Function: Negative of combined operating and carbon emission costs, guiding optimization.
• State Transition: Controlled by actions and influenced by uncertainties such as renewable output.
• Value Function: Expected cumulative discounted rewards under policy.

Solution via Deep Deterministic Policy Gradient (DDPG)

The DDPG algorithm is employed to solve the optimization problem, training policy and value networks using historical operational data. The approach enhances the system’s ability to learn optimal energy management strategies, promoting efficient and low-carbon operation consistent with SDG 7 and SDG 13.

Model Data Inputs

• Operational parameters of energy equipment.
• Historical electric and heat load data.
• Renewable energy output data (e.g., PV).
• Carbon emission data from MES operations.
• Cost data including operation and carbon emission costs.

Solution Process

1. Initialize algorithm parameters and networks.
2. Iteratively input system states and compute optimal actions.
3. Update state based on actions and calculate rewards.
4. Store experiences and update networks using prioritized experience replay.
5. Repeat until convergence and output optimized energy schedules.

Simulation Analysis

Simulation Setup

A simulation model based on a Northeast China power plant MES is developed, incorporating rooftop PV and other energy equipment. Parameters are set according to real operational data, supporting SDG 7 by modeling renewable integration.

Results

• The MES meets power and heat demands through coordinated energy supply and load management.
• Carbon capture operation increases during peak CHP output to meet emission constraints.
• Battery storage discharges primarily in the evening, charging during daytime PV generation.
• Controllable loads adjust consumption to optimize costs and emissions.
• Heating temperatures are maintained within comfort ranges while optimizing energy use.

Scenario Analysis

1. Scenario 1: No coordination of controllable grid-side loads.
2. Scenario 2: Coordination with controllable grid-side loads using deep convolutional neural networks.
3. Scenario 3: Coordination with controllable grid-side loads using the proposed DDPG algorithm.

Scenario 3 achieves the lowest total energy optimization cost and highest carbon emission reduction, with cost reductions of approximately 19-20% and carbon reductions of 22-24% compared to Scenario 1, demonstrating the effectiveness of coordinated source-grid-load management in line with SDG 7 and SDG 13.

Conclusion and Research Outlook

1. The proposed energy optimization model effectively characterizes MES energy input, output, and conversion processes, integrating distributed PV coordination for carbon reduction.
2. Coordination between MES energy supply resources and controllable grid-side loads enhances operational flexibility, reduces costs, and lowers carbon emissions, advancing SDG 7 and SDG 13.
3. The deep reinforcement learning-based optimization approach significantly reduces total energy costs and carbon emissions in MES operations.

Future research will focus on large-scale pilot implementations in actual thermal power plants to validate and refine the model. Further exploration of policy frameworks such as carbon trading and subsidies will support MES optimization and carbon reduction. Integration of emerging technologies like artificial intelligence and the Internet of Things will enhance MES intelligence and optimization capabilities. Interdisciplinary approaches combining energy engineering, economics, and environmental science will provide comprehensive solutions aligned with SDGs.

1. Sustainable Development Goals (SDGs) Addressed

• SDG 7: Affordable and Clean Energy
  • The article focuses on energy optimization in thermal power plants integrating distributed photovoltaics and renewable energy sources, aiming to improve energy efficiency and promote clean energy.
• SDG 9: Industry, Innovation and Infrastructure
  • The development and application of deep reinforcement learning algorithms for energy optimization represent innovation in industrial processes and infrastructure.
• SDG 11: Sustainable Cities and Communities
  • The MES exports electric and heat energy to meet the demands of residents, contributing to sustainable urban energy supply.
• SDG 12: Responsible Consumption and Production
  • Optimization of energy consumption and carbon emission reduction in multi-energy systems promotes sustainable consumption and production patterns.
• SDG 13: Climate Action
  • The article addresses carbon emission reduction in thermal power plants, contributing directly to climate change mitigation.

2. Specific Targets Under the Identified SDGs

1. SDG 7: Affordable and Clean Energy
   • Target 7.2: Increase substantially the share of renewable energy in the global energy mix.
   • Target 7.3: Double the global rate of improvement in energy efficiency.
2. SDG 9: Industry, Innovation and Infrastructure
   • Target 9.4: Upgrade infrastructure and retrofit industries to make them sustainable, with increased resource-use efficiency and greater adoption of clean and environmentally sound technologies.
3. SDG 11: Sustainable Cities and Communities
   • Target 11.6: Reduce the adverse per capita environmental impact of cities, including by paying special attention to air quality and municipal and other waste management.
4. SDG 12: Responsible Consumption and Production
   • Target 12.2: Achieve the sustainable management and efficient use of natural resources.
   • Target 12.4: Achieve environmentally sound management of chemicals and all wastes throughout their life cycle.
5. SDG 13: Climate Action
   • Target 13.2: Integrate climate change measures into national policies, strategies and planning.
   • Target 13.3: Improve education, awareness-raising and human and institutional capacity on climate change mitigation.

3. Indicators Mentioned or Implied to Measure Progress

• Carbon Emission Intensity of MES
  • Indicator: Carbon dioxide emissions per unit of electric power output (e.g., (gamma_{MES}), carbon emission intensity (E_{MES,g,t}))
  • Used to measure the efficiency of carbon emission reduction in the multi-energy system.
• Operating Costs and Carbon Emission Costs
  • Indicator: Total operating cost (f_{MES,op}) and carbon emission cost (f_{MES,CE}) of the MES.
  • Reflects economic efficiency and environmental impact of energy optimization.
• Energy Balance Constraints
  • Indicators: Electric power balance, heat power balance, and fuel balance within the MES (e.g., equations (6), (7), (8)).
  • Measures the effectiveness of energy optimization and system stability.
• Carbon Allowances and Emission Penalties
  • Indicator: Carbon allowances (CE_{MES,CO2}) and reward/penalty costs (f_{CE,u}) based on actual emissions versus allowances.
  • Measures compliance with carbon emission regulations and incentives.
• Load and Renewable Energy Output Data
  • Indicators: Controllable electric and heat loads, renewable energy output (e.g., photovoltaic power (P_{NEW,t})).
  • Used to assess demand response and renewable integration effectiveness.
• Performance of Deep Reinforcement Learning Algorithm
  • Indicator: Reward function (R_{MES,t}) reflecting minimized operating and carbon costs.
  • Measures the success of the optimization strategy in balancing cost and emissions.

4. Table of SDGs, Targets, and Indicators

Source: nature.com