Research on Optimization Method of Integrated Energy System Network Planning

Abstract

The development of an integrated energy system (IES) is conducive to promoting the transformation of the energy system and helping to achieve the ‘double carbon’ goal in China. The IES integrates cooling, heating, electricity, gas, and other energy resources, which is significantly more difficult than single energy network planning. This paper systematically sorts out the process of IES network planning and proposes an improved methodology. Firstly, the bottom-up dynamic multiple-load forecasting method of 8760 h a year is proposed as the basis of system configuration and capacity selection. Subsequently, a planning method for energy station location and route optimization using the Dijkstra algorithm is constructed by applying the P-median optimization model. Finally, when optimizing the capacity allocation of the IES, the complementary characteristics of natural gas, electricity and heat, as well as the corresponding energy demand characteristic, have been fully considered, so that the optimization objectives can be reasonably determined. Through the actual calculation, it is found that the optimization method proposed in this paper can reduce the construction cost of the network by 41%. The work combines the process of energy network planning and capacity configuration of IES, which provides a simple, easy and economical solution for IES planning in new areas.

1. Introduction

With the development of the world economy and the progress of urbanization, the global energy demand is increasing dramatically. Around 83% of global energy consumption is supported by the combustion of fossil fuel, which directly leads to an increase in greenhouse gas emissions and contributes to global warming. As reported by IPCC Sixth Assessment Report [1]: Climate Change 2022: Mitigation of Climate Change, global average annual greenhouse gas emissions were at the highest level in human history from 2010 to 2019, although the rate of growth has slowed. The report also argues that if global warming is to be controlled to less than 1.5 °C compared with the level before industrialization, global greenhouse gas emission needs to reach the peak value before 2025 and achieve net zero carbon dioxide emissions in the early 2050s. At the 75th session of the UN General Assembly in 2020, China put forward the goals of ‘carbon peaking’ in 2030 and ‘carbon neutrality’ in 2060, which is also known as the ‘dual carbon goals’. The ‘dual carbon goals’ are set to control greenhouse gas emission and slow down global warming potential.

As an important initiative to achieve the ‘dual carbon goals’, the development of integrated energy system (IES) has been repeatedly mentioned in relevant Chinese policy documents. For example, in China’s ‘14th Five-Year Plan’ modern energy system plan, ‘14th Five-Year Plan’ renewable energy development plan and other documents, the development of IES with integrated supply of cooling, heat, water, electricity and gas, relying on smart distribution networks, urban gas networks, heat pipe networks and other energy networks, is proposed. By making comprehensive use of renewable energy, energy storage, flexible network and other advanced energy technologies and interconnection communication technologies, the IES can contribute to the realization of efficient and flexible access to distributed renewable energy (RE), as well as the energy transition goal of integrating RE production and consumption [2,3].

The development of the IES is an inevitable choice for the transformation and upgrading of China’s energy system, considering the many advantages of IES. To start with, by integrating distributed and centralized energy production, the IES can comprehensively utilize a variety of energy resources such as cooling, heating, electricity and gas, which breaks the traditional single-energy usage mode. Furthermore, it comprehensively optimizes the progress of supply, transmission, consumption and conversion, and effectively improves energy efficiency and operational reliability through the complementarity and coordination of multiple energy sources. As a result, the emission of greenhouse gases (GHG) and other pollutants is reduced, the life cycle costs of users can be reduced by 20% [4], and sustainable energy development can be realized. The IES consists of energy stations and energy transmission networks [5,6]. Currently, the research on IES planning mainly focuses on the research into load prediction methods, optimization modeling and solution methods, and technical economy research. In terms of the research on load prediction, according to the research of Swan and Ugursal [7], the research results of the load prediction method of IES are summarized and divided into two types: macro prediction (top-down method) and micro prediction (bottom-up method). The micro prediction method mainly focuses on the actual energy demand of each user unit in the region, and finally the prediction results of total regional energy consumption can be obtained. Compared with the macro prediction method, the investigation process of the micro prediction method is more complex; as a result, more accuracy can be gained. Jiang et al. [8] use the least squares support vector machine to solve the problems of high dimensionality, small samples and nonlinearity effectively and to build an electric-load prediction model. What is more, the authors uses the k-mean to improve the bat algorithm to optimize the least squares support vector machine, which effectively improves the prediction accuracy. Based on the shortcomings of traditional grey correlation analysis, Wu [9] established a similarity day selection method for comprehensive similarity of distance similarity and trend similarity and used the vertical and horizontal optimization deep belief network to predict the cooling and heating power load of the community.

In the research into optimization modeling and solving methods, Ge et al. [10] depicted the dynamic characteristics of energy in the IES through the energy network theory and described the network characteristics by using the energy quality characteristics. Chen et al. [11,12] constructed a general branch power flow model of grid, gas network and heat network according to the conservation of mass, conservation of energy and Kirchhoff equations. For the electric-heating-gas regional IES, Wu et al. [13] summarized four modes based on different CHP unit operation modes, grid-connected conditions and whether there is electricity-to-gas energy storage, and research on the multi-energy power flow calculation algorithm and process. Soliman et al. [14] proposed a two-stage IES model. The day-ahead stage optimizes renewable energy use, manages battery charging and determines power demand. The real-time scheduling stage regulates power flow with an optimal power dispatch algorithm. Al Alahmadi et al. [15] developed an IES management model with a non-integer order controller and hierarchical management control strategy. Sahri et al. [16] proposed a fuzzy logic controller-based IES model. The model regulates power flow based on the predetermined control rules. These methods optimize renewable energy use and manage power demand in microgrids, increasing system efficiency while reducing operating costs.

In terms of technical economy research, the main focus is on economic benefits, energy conservation and emission reduction. Zhang et al. [17] used a Particle Swarm Optimization algorithm to optimize a solar-powered cold/electricity cogeneration prototype for an insular tropical setting, which employs hydrogen storage and recovers waste heat to increase overall efficiency, while identifying the economic feasibility of hydrogen as a storage medium based on load profile characteristics, and providing insights into the potential benefits of incorporating a thermochemical system to the cogeneration process. Fu [18] used the POME distribution of temperature, the β distribution of solar radiation, and the Ville distribution of wind power to predict the weather in the planning, introduced the Monte Carlo method to generate random samples, and introduced the concept of information entropy to assess the cost of uncertainty. Soderman [19] established an optimization model of electric-heat distributed energy systems based on economic cost and proposed an optimal layout method for a heating network. Morvaj [20] took the minimum system construction and operating cost and carbon emission as the objective functions, established the optimal configuration of equipment capacity and system economic operation model of the distributed energy system, and discussed the optimal layout of the district heating network.

Single energy system planning has developed a very mature methodology. However, the simple combination of single energy system planning cannot meet the need for coupling and coordination between IES networks. There are several difficulties, described as follows:

• Multivariate load prediction. The accuracy of load prediction is directly related to whether the planning and operation are reasonable and effective, so it is an important prerequisite of IES planning. The traditional energy system load prediction method is limited to the independent prediction of various forms of energy. Due to the neglect of the coupling and interaction effect between multiple energies, they cannot meet the need for the development of IES, although they are more established and proven. For the load prediction of IES, electricity, gas, heating, cooling and other multiple types of loads, as well as their mutual coupling, should be taken into consideration, which is significantly more difficult than single-form energy system prediction. On one hand, there are many effect factors, such as climate, social economy, the regional layout structure, architectural design characteristics, etc., which makes prediction difficult. On the other hand, multi-dimensional and multi-time scales increase the difficulty of solving load prediction models.
• Model characterization and optimization solving. When describing the model, the coupling mode of electricity, gas, heating and cooling networks should be considered to include the physical relationships and constraints between variables in the model as accurately and comprehensively as possible. When solving physical models, gas networks and heating networks have the characteristics of delay and energy storage. The partial differential form needs to be used for nonlinear optimization, which belongs to non-deterministic polynomial problems (NP-hard). Regarding the solution of mixed integer nonlinear programming, researchers have conducted extensive research [21,22,23,24] but still cannot establish whether the current solution is the global best solution.

Overall, existing energy network optimization studies are more about prediction, modeling and optimization of single energy systems. For the actual project, there are fewer considerations about the network planning affected by the property of the site and the municipal roads, and there is a lack of research on the site selection of energy stations, the optimization of network paths, and the optimization of capacity allocation of IES. This study proposes a planning method that includes collaborative forecasting of cooling, heating and electricity loads, site and pipe network planning, and determination of energy station equipment capacity for IES planning in new areas. The proposed method can fill the research gap in the pre-system planning of IES.

2. General Process of IES Planning

The energy network planning process of IES mainly contains four steps: basic data collection, demand prediction, optimization analysis, and decision-making, as shown in Figure 1.

The four steps in Figure 1 are explained in detail as follows:

(1)

Basic data collection. In order to make sure the planning of IES meets the local demand, the information of the area to be planned should be collected comprehensively, including overall planning (such as population, site use, energy structure, economic level, layout and development planning of each functional zone), municipal planning (such as power structure and capacity, gas and heating supply capacity and network layout), resource conditions (such as solar energy, wind energy, geothermal energy), etc. In terms of historical data and system parameters, it is necessary to fully investigate the demand characteristics of different types of local buildings, the energy consumption characteristics of operation, and the operating status of electricity, gas and heating systems.

(2)

Demand prediction. Through the prediction of the load data of electricity, cooling and heating, the installed capacity of energy stations and the design of the network can be better planned. Demand is influenced by many factors, including the building energy consumption characteristics, the indoor environment, regional climate characteristics, and the economic level of the site [25]. IES should predict various load data such as electricity, cooling, heating and gas. According to the classification of the objects of study in the prediction, it can be divided into macro load prediction and micro load prediction. The macro load prediction method focuses on the overall results, through the regression analysis of the relationship between the historical data and total regional energy load, the future regional energy consumption can be acquired. Commonly used macro load prediction methods include the single consumption method, elastic coefficient method, extrapolation method, etc. The single consumption method predicts the total electricity consumption through the unit output value and total output value [26]. The elastic coefficient method uses the elasticity coefficient of energy consumption growth rate in relation to the growth rate of GDP to predict the total regional energy consumption [27]. The extrapolation method predicts future energy consumption trends based on historical energy consumption data [28]. The micro load prediction method predicts the energy consumption demand of a certain building or a certain type of building through the historical hourly load of the building combined with building characteristics, environment, user energy characteristics and other information, such as using the engineering model method for regression analysis of historical data at the building level [29]. With the development of technology and the assistance of professional software such as SketchUp Pro, EnergyPlus, Openstudio, etc., accurate prediction of hourly loads on a single building can be achieved. Combined with the distribution of building types in the region and the coefficient of simultaneous use of buildings, the energy consumption demand of the region can be obtained.

(3)

Optimize analysis. The structure of the IES is expressed according to the conservation of mass, conservation of energy and Kirchhoff’s equations. To establish a multi-energy coupling network model, in the model description, firstly, the coupling method of the multi-energy network should be considered. Secondly, the physical relationships and constraints between variables in the model should be described as accurately and comprehensively as possible. Finally, with the integration of new energy sources, the simulation of uncertainty is also important. The performance indicators such as ‘economy’, ‘energy saving’, and ‘environmental protection’ are set as the optimization goals and decision-making variables. Considering the energy supply benefits and space constraints, the construction site of the energy hub is selected in each sub-region, and the layout form path of the network is planned. In terms of system form and capacity configuration, constraints such as capacity, state variables, and system balance are fully considered to make the model obtain a unique solution or an optimal solution. Some mature commercial linear solvers, such as Gurobi, IBM CPLEX^® Optimizer, FICO Xpress, MOSEK and others, can usually be used.

(4)

Decision-making. According to the optimization results, the site selection of the energy station, equipment capacity, network parameters, and connection relationship can be determined.

3. Methodology

In this study, a collaborative IES planning method is proposed. Firstly, the bottom-up dynamic multivariate load forecasting method of 8760 h per year is used to forecast the cooling, heating and electrical loads of the system. Then, the P-median model is applied, and the Dijkstra algorithm is used for energy station location and line planning. Finally, the capacity configuration of the integrated energy system is optimized at the selected energy station site according to the optimization objective.

3.1. Multivariate Load Prediction Method

In order to improve the accuracy of load prediction, a bottom-up estimation method is used. Firstly, according to the function of the site, construction area and load characteristics, the dynamic load model of the site is established, and the hourly load simulation analysis of electricity, gas, heating and cooling is carried out for different types of buildings for 8760 h a year, as shown in Equations (1) and (2). Regional load prediction is the hourly summation of loads across all sites, considering certain simultaneous use factors.

$$E_{i,t}^m=A_i\cdot a_j{\displaystyle \sum _{k=1}^{10}{\alpha }_k^j{e}_{k,t}^m}$$

(1)

$$E_i^m=\max (E_{i,t}^m)$$

(2)

$E_{i,t}^m$: the total load of the energy form m through the time t in the site i, kW;

$A_i$: the area of block i, m²;

$a_j$: the floor area ratio of the block, m²; subscript j refers to the function of the block;

${\alpha }_k^j$: The proportion of the floor area of building type K in the site with function J;

$e_{k,t}^m$: Unit area load of energy form m of time t in the building type k, kW/m²;

$E_i^m$: Static load of the energy form m in the site i, kW.

3.2. Site Selection and Network Layout Optimization

3.2.1. P Median Model

Network optimization is based on certain optimization goals, selecting several energy stations among all alternative energy stations so that the distance from the energy station to the load center through the alternative route is minimized. This type of problem is called the P median problem. The P median problem is a complex linear optimization problem with NP-hard solution complexity. The common solution methods include the enumeration method, scattered search method, genetic algorithm, greedy algorithm, ant colony algorithm, etc. In this study the problem is solved using a modified enumeration method [30].

3.2.2. Optimization Methods and Steps

The planning area is divided into sub-areas according to the type of land use; the center of the site is the load center. The location of the energy station is selected in each sub-area based on a comprehensive consideration of energy supply benefits and space limits, and the layout form and laying path of the network are planned at the same time. The algorithm flowchart is shown in Figure 2:

M is defined as the set of load centers. N is defined as the collection of energy stations. Q is defined as the collection of intersections of the road network. The vertex set containing the energy station, all load centers and the intersection of alternative lines can be obtained: V = M ∪ N ∪ Q. The alternative lines make up an edge set, E, that represents the connectivity between points. The weight set, W, can be obtained according to the length of the alternative line and other actual situations. Vertice set V, edge set E and weight set W constitute the empowerment network of a regional energy system layout G = (V, E, W).

According to the established weighted network, the Dijkstra algorithm is used to calculate the minimum weight matrix from the energy station to each load center and the corresponding path matrix. The specific algorithm is as follows:

Step 1: The edge matrix E is solved according to the correlation between the vertice set V in the network G. The specific calculation is:

$$e(i,j)=\left\{\begin{array}{c}1\hspace{1em}\mathrm{i}\mathrm{f}\hspace{0.17em}i,j\hspace{0.17em}\mathrm{a}\mathrm{r}\mathrm{e}\hspace{0.17em}\mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\hfill \\ 0\hspace{1em}\mathrm{i}\mathrm{f}\hspace{0.17em}i,j\hspace{0.17em}\mathrm{a}\mathrm{r}\mathrm{e}\hspace{0.17em}\mathrm{n}\mathrm{o}\mathrm{t}\hspace{0.17em}\mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\hfill \\ 0\hspace{1em}\mathrm{I}\mathrm{f}\hspace{0.17em}i\hspace{0.17em}\mathrm{a}\mathrm{n}\mathrm{d}\hspace{0.17em}j\hspace{0.17em}\mathrm{a}\mathrm{r}\mathrm{e}\hspace{0.17em}\mathrm{t}\mathrm{h}\mathrm{e}\hspace{0.17em}\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{e}\hspace{0.17em}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{e}\mathrm{x}\hfill \end{array}\right.$$

(3)

In the equation, $e(i,j)\in \mathbf{E};i,j\in \mathbf{V}$.

Step 2: According to the position information of each node in the network G and the edge matrix E, the weight matrix W is solved. The specific calculation method is:

$$w(i,j)=\left\{\begin{array}{c}{C}_{tun,ij}\hspace{1em}\mathrm{i}\mathrm{f}\hspace{0.17em}e(i,j)=1\hfill \\ \infty \hspace{1em}\hspace{1em}\hspace{0.17em}\mathrm{i}\mathrm{f}\hspace{0.17em}e(i,j)=0,\hspace{1em}i,j\hspace{0.17em}\mathrm{a}\mathrm{r}\mathrm{e}\hspace{0.17em}\mathrm{n}\mathrm{o}\mathrm{t}\hspace{0.17em}\mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\hfill \\ 0\hspace{1em}\hspace{1em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\mathrm{i}\mathrm{f}\hspace{0.17em}e(i,j)=0,\hspace{1em}i,j\hspace{0.17em}\mathrm{a}\mathrm{r}\mathrm{e}\hspace{0.17em}\mathrm{t}\mathrm{h}\mathrm{e}\hspace{0.17em}\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{e}\hspace{0.17em}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{e}\mathrm{x}\hfill \end{array}\right.$$

(4)

In the equation $w(i,j)\in \mathbf{W};i,j\in \mathbf{V}$, $C_{tun,ij}$ is the unit price of the network between i and j, including power lines, natural gas pipelines, and heating and cooling energy transmission pipelines.

Step 3: The Dijkstra algorithm is used to solve the minimum weight between the vertices in the network G. The calculation is shown as follows.

$$d(i,j)=\left\{\begin{array}{c}w(i,j)\hspace{1em}\hspace{1em}\mathrm{i}\mathrm{f}\hspace{0.17em}e(i,j)=1\hfill \\ \mathit{min}{d}_{ij}\hspace{0.17em}\hspace{0.17em}\hspace{1em}\mathrm{i}\mathrm{f}\hspace{0.17em}e(i,j)=0,\hspace{1em}i,j\hspace{0.17em}\mathrm{a}\mathrm{r}\mathrm{e}\hspace{0.17em}\mathrm{n}\mathrm{o}\mathrm{t}\hspace{0.17em}\mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\hfill \\ 0\hspace{1em}\hspace{1em}\hspace{1em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\mathrm{i}\mathrm{f}\hspace{0.17em}e(i,j)=0,\hspace{1em}i,j\hspace{0.17em}\mathrm{a}\mathrm{r}\mathrm{e}\hspace{0.17em}\mathrm{t}\mathrm{h}\mathrm{e}\hspace{0.17em}\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{e}\hspace{0.17em}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{e}\mathrm{x}\hfill \end{array}\right.$$

(5)

$d(i,j)\in \mathbf{D};i,j\in \mathbf{V}$, $d(i,j)$ is the minimum weight between vertices i and j. $l(i,j)$ is the path between vertices i and j when the weight is minimum.

Step 4: The minimum weight matrix D and path matrix L are output, and the construction cost Z of the network is calculated.

3.3. Model Characterization of IESs

The typical energy station model includes the complementarity of the three forms of energy: cooling, heating and electricity. The relationship between energy conversion and transmission is shown in Figure 3.

3.3.1. Coupling Node Energy Conversion Model

Energy conversion takes place in energy conversion equipment, and different physical models of energy conversion equipment are established as follows:

(1) Gas boilers and gas turbines

Natural gas from external gas sources can not only meet the gas load but also be used as fuel for gas turbines and gas boilers to produce other forms of energy to realize coupling between networks. The energy model related to gas boilers and gas turbines is shown in Equation (6):

$$H_b^t==LH{V}_{C{H}_4}{\dot{V}}_{C{H}_4}{\eta }_b$$

(6)

$H_b^t$: the heat energy generated by a gas boiler at time t, kW;

$LH{V}_{C{H}_4}$: the low heating value of natural gas;

${\dot{V}}_{C{H}_4}$: natural gas consumption at time t, $\mathrm{N}{\mathrm{m}}^3/\mathrm{s}$;

${\eta }_b$: the efficiency of heat generation in the gas boiler.

(2) Chiller and heat pump

Electricity can not only provide user load but can also be used to drive equipment that generates heating and cooling, such as heat pump units, absorption units, etc. The energy conversion relationship is

$$C_{echi}=CO{P}_e{P}_{echi}$$

(7)

$C_{echi}$: indicates the heating/cooling capacity of a heat pump unit, kW;

$CO{P}_e$: Energy efficiency ratio of heat pump units;

$P_{echi}$: The input electrical power of the heat pump unit, kW.

3.3.2. Capacity of Energy System Optimization Model

The optimization model of energy network installed capacity is shown in Equation (8)

$$\begin{array}{l}\min f(x)\\ \mathbf{s}\mathbf{.}\mathbf{t}\mathbf{.}\{\begin{array}{l}g(x)=0\\ h(x)\le 0\end{array}\end{array}$$

(8)

f(x) represents the optimized objective function involving economy, energy saving and environmental protection.

g(x) and h(x) are constraints of the optimization model. g(x) represents a series of equation constraints, mainly reflecting the energy conversion relationship in the energy station, and the energy conversion devices involved include gas turbines, gas boilers, heat pumps, etc., as shown in Equations (6) and (7). h(x) represents a series of inequality constraints that constrain the energy output of the energy station to meet the load of the energy station.

4. Case Study Analysis

A regional planning data of a community in Chongqing was selected as a case study. The length of the community is 1.92 km from east to west and the width is 0.85 km from north to south. The total building area is about 1,500,000 square meters, with 18 load demand areas. The building types include residential building, office building, commercial building and hospital. There are 13 locations that can be used to build the energy station. As shown in Figure 4, points 1–13 represent candidate energy stations, blue points represent load centers, and the solid black line shows the network distribution. The system makes full use of the combined cooling, heating and power trigeneration, waste heat cascade energy, renewable energy and energy storage. The multiple energy systems are applied comprehensively and collaboratively.

4.1. Load Prediction Data Analysis

Energy consumption simulation software Trnsys was used to simulate the annual load of typical building types. The electrical, heating and cooling load data of different types of buildings on a typical summer or winter day were selected, as shown in Figure 5. The regional load statistics of 8760 h per year are shown in Figure 6. According to the load analysis results, the total regional electricity load was 25,922.4 kW, the total cooling load was 84,654.0 kW, and the total heating load was 32,153.6 kW. That equaled a unit area electrical load of 17.28 W/m², a cooling load per unit area of 56.43 W/m² and a heat load per unit area of 21.44 W/m².

The electricity, heating and cooling loads obtained were clustered by the K-means unsupervised learning method, and the final cluster number was determined by the elbow law and contour coefficient. Figure 7 shows the contour factor and average distance of the electrical load. According to the definition of the contour coefficient, the higher the score, the better the effect, so the result after selecting three classes will have a relatively high contour coefficient. According to the elbow rule, it shows a clear elbow turn in class 3, so the cluster number K = 3 is selected. The typical load characteristics for different seasons were obtained by clustering the load forecast values using K = 3, as shown in Figure 8.

4.2. Energy Station and Network Optimization Analysis

According to the model solution method of Section 3.2, the cost distribution of energy station and network routes in the study area is shown in Figure 9, and the results of energy station site selection and routes optimization are shown in Figure 10. According to the optimization results, energy station 4 is the best site selection among the 13 candidate energy stations, and the yellow path indicates the network path from the energy station to each load center. The construction cost of the pipe network with the best site selection was 0.95 million USD, which was 41% lower than the construction cost of 1.63 million USD for energy station 11 by using the P median model optimization. In other words, by applying the method to the district with an area of 1 km², the construction cost of the pipe networks can be reduced by 0.68 million USD for most.

4.3. Energy System Capacity Optimization Analysis

The equipment to be capacity-optimized includes a combined cooling, heating and power (CCHP) system, ground source heat pumps (GSHP), energy storage (ES), and gas boilers/chillers. The initial capacity of the system was determined according to conventional engineering experience. The CCHP system was based on the principle of bearing the basic power load of the project and ensuring a certain number of opening hours, and 10% of the designed power load of the project was taken as the rated power generation. The ground source heat pump system used 60% of the smaller value of the designed heating and cooling load as the rated capacity. The energy storage used 60% of the cumulative heating load on the typical design day as the total energy storage.

The system capacity configuration was optimized with the single objects including life cycle cost (LCC) and the annual energy consumption (AEC), as well as the multiple objects “economy”, “energy saving”, and “environmental protection” (3E), respectively; the results are shown in

Table 1. System capacity configuration optimized for different goals.

Optimize Goals	CCHP	GSHP	ES	Gas Boiler/Chiller
LCC	11.38%	31.72%	21.73%	35.17%
AEC	4.82%	58.19%	24.92%	12.07%
3E	12.89%	34.93%	24.27%	27.91%

. According to Table 1, with different optimization objects, the configuration of the energy equipment is different. The capacity of GSHP can range from 31.72% to 58.19%, a difference of up to 26.37%.

5. Conclusions

In this paper, the methods of multivariate load prediction, energy station site selection, network layout optimization, and energy system capacity optimization are studied by sorting out the process of energy network planning of IES. Through the analysis of example, the scientificity and effectiveness of the proposed method are verified, and the following conclusions are obtained:

The bottom-up multivariate load prediction method is used to help the capacity selection of the comprehensive energy system through dynamic load simulation for different types of individual buildings for 8760 h and the regional dynamic load accumulation.

When considering the site selection and route planning of the IES, the nature of the site and the conditions of municipal roads should be fully considered. The P median optimization model can significantly reduce the construction cost of the network, and according to the example, it is found that the energy station with the best location can reduce the construction cost of the network by 41%.

When the energy system capacity configuration is optimized with different optimization objectives, there are great differences between the optimization results of the capacity ratio. The capacity ratio difference can be as high as 26%. In engineering applications, the optimization goal of system configuration should be determined according to the actual demand.

Overall, IES involves the integration of various energy systems, such as renewable energy sources, energy storage systems, and demand-side management systems, to achieve improved energy efficiency, reduced energy costs, and decreased environmental impact. Therefore, in the process of conducting energy planning, local governments should fully consider the advantages of synergy and the complementary nature of integrated energy systems and provide sufficient support from a policy perspective.