Problem Chosen 2024 Team Control Number

MCM/ICM
B Summary Sheet
18

Photovoltaic Power: China’s New Journey towards "Carbon Peak" and

"Carbon Neutrality"
Summary

To achieve the Chinese government’s strategic goal of carbon peaking and carbon neutrality by
2060, Power mix transformation is a must for China. We examine the potential for photovoltaic power
generation in China and analyze related policies to determine whether China can achieve its goals
through the development of clean energy generation.
To build the model, we gather data on China’s social, economic, environmental, and power devel-
opment from authoritative websites, which is reliable. Then, we establish a power supply prediction
model based on decision tree regression, a PV plant construction suitability evaluation model
based on analytical hierarchy process(AHP), a power generation potential calculation model, and
a 0-1 integer planning model with PV plant site selection.
For Model 1, we analyze the relationship between four factors: GDP, population, industrial structure,
and urbanization rate, and their interaction with electricity supply using a decision tree model. Based
on the trend of the gradual withdrawal of thermal power and the development of clean energy, we
establish an expression for total electricity supply. The trend of power supply can be determined by
regressing the indicator with each power using the decision tree regression algorithm.
For Model 2, based on the raster data of China’s DEM, resources & environment, the relevant
indicators are analyzed hierarchically to obtain the suitability of PV power plant construction in the
region. Based on four conditions: cost, benefit, geography, and light, the selected indicators are: solar
radiation, slope, aspect, mountain shadow, and distance to the road network. The restriction zones
for the construction of photovoltaic power plants are set as follows: nature reserves, arable land, and
water areas. This is in line with the principle of not damaging the ecology and agricultural production.
By analyzing the suitability of Tibet, it can be seen that the total suitability of Tibet is 8.89, and it is
suitable for large-scale construction of large-scale photovoltaic power plants.
For Model 3, the power generation potential is calculated by the PV module area, the total horizontal
irradiation together with the PV module coefficients. The area of the PV module is related to the slope,
slope direction, and land type. The total annual power generation potential of China was finally
calculated to be 5839.4348∗108 kW h.
For Model 4, a 0-1 integer planning model is developed to maximize the total return and the
investment is greater than the total generation cost. The total return is the amount of electricity
generated multiplied by the price of electricity in the region. The change in unit cost is consistent
with the learning curve model, which essentially explains the phenomenon that a decrease in cost is
the result of an increase in appropriateness. This model involves the suitability of Model 2 and the
generation potential of Model 3, and integrates investment, cost, and revenue for 0-1 integer planning
of whether or not to build a PV plant in a unit area. The optimal siting results of PV plants on a
China-wide map are finally obtained.
Finally, the calculation of power generation potential is justified by sensitivity analysis. We believe
that by replacing coal-fired power generation with cleaner energy sources, the Chinese government’s
goals of carbon peaking and carbon neutrality can be achieved.

Keywords: Photovoltaic power generation; Grid method; AHP; 0-1 Integer planning; DTR
Team # 18 Page 1 of 24

Contents
1 Introduction 3
1.1 Problem Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Problem Restatement and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Our Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Assumptions 5

3 List of Notation 6

4 Data Overview 6
4.1 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
4.2 Data Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

5 Problem One 7
5.1 Problem analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
5.2 Preparation of the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
5.2.1 Influencing factors analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
5.3 Establishment of the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
5.3.1 Electricity supply equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
5.3.2 CART regression tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
5.4 Model solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.5 Results analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

6 Problem 2 12
6.1 Problem analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
6.2 Preparation of the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
6.2.1 Land policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
6.2.2 Quantity of solar radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
6.2.3 Distance from road & railroad . . . . . . . . . . . . . . . . . . . . . . . . . . 12
6.2.4 Slope & Aspect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
6.2.5 Mountain Shadow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
6.3 Model establish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
6.4 Site selection model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
6.4.1 Restricted area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
6.4.2 Hierarchical analysis to determine indicator weights . . . . . . . . . . . . . . 14
6.4.3 PV plant construction suitability model . . . . . . . . . . . . . . . . . . . . . 15
6.5 Solution of the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
6.6 Results analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

7 Problem three 17
7.1 Generation Potential Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
7.2 PV plant construction planning model . . . . . . . . . . . . . . . . . . . . . . . . . . 17
7.2.1 Photovoltaic power plant construction target . . . . . . . . . . . . . . . . . . . 18
7.2.2 Investment-cost constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Team # 18 Page 2 of 24

7.3 Result analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

8 Problem Four 19

9 Sensitivity analysis of the model 20

10 Strength and Weakness 21

10.1 Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
10.2 Weakness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

11 Letter 21

12 Appendix 24
Team # 18 Page 3 of 24

1 Introduction
1.1 Problem Background
China’s electricity mix can be categorized into traditional energy generation, renewable energy
generation and other forms of electricity.According to the latest figures, China has the world’s highest
total power generation capacity, exceeding 2 billion kilowatt-hours. The current power generation
model is capable of meeting the country’s significant power demand.The electric power and energy
industry is closely related to various factors, including the economic situation, population consumption,
urbanization rate, and degree of marketization. Additionally, electric power serves as the foundation
for economic development and social progress.As China’s economy expands and living standards
improve, the demand for electricity increases. The Chinese government has proposed peak carbon and
carbon-neutral targets, necessitating a shift in the electricity mix.On the premise of meeting the demand
for electrical energy, gradually reduce the reliance on traditional energy generation and increase the
proportion of renewable energy generation.
Photovoltaic (PV) power generation, as a type of renewable energy, is developing rapidly around
the world and has become an important means of promoting clean energy transition.China has a
significant amount of non-cultivated land resources that can be utilized for constructing PV power
plants. Deserts cover an area of about 13,000 square kilometers, and saline land covers an area
of about 23,300 square kilometers, providing enormous potential in the field of photovoltaic power
generation.Photovoltaic (PV) power generation currently faces challenges in terms of capacity, technical
feasibility, and sustainability.Conversion efficiencies of 25%more are known to be achieved with current
photovoltaic cells.China aims to reach its carbon peak by 2030 and achieve carbon neutrality by 2060.

1.2 Problem Restatement and Analysis

We are required to establish a model to predict a sustainable energy mix dominated by photovoltaic
(PV) power generation under peak carbon and carbon neutral targets.The energy mix must meet
the demand for electricity and gradually increase the share of renewable energy generation while
minimizing economic input.In addition,the impacts of technology, policy,and other relevant factors
must be considered.
In order to solve those problems, we will proceed as follows:

• Probem One. We need to analyze the factors that influence the supply of electricity, explain
their impact, and develop a forecasting model to predict the trend in China’s electricity supply
from 2024 to 2060.

• Problem Two. When setting up a photovoltaic (PV) plant, four main aspects must be considered:
cost, revenue, geography, and lighting.The task is to create a model that evaluates the suitability
of an area for a PV power plant based on four indicators: cost, revenue, topography, and light.
The model should provide the possibility of constructing a PV power plant in the area.

• Problem Three. Based on (question 2 model), add considerations of geographic resources and
investment capacity, cost and revenue factors, build multiple power plants in China.Calculate the
maximum amount of photovoltaic power that can be generated in China.
Team # 18 Page 4 of 24

• Problem Four. Replacing coal-fired power generation with clean energy to achieve the Chinese
government’s strategic goal of peak carbon and carbon neutrality by 2060, and giving a strate-
gic plan for the sustainable development of photovoltaic power generation in China under the
feasibility of this idea.

• Sensitivity analysis. With the criteria defined before, we evaluate the reliability of our model
and do the sensitivity analysis.

• Strength and Weakness.

1.3 Our Work

To avoid complicated description,intuitively reflect our work process,the flow chart is show as the
following Fig 1.

Figure 1: The Framework of Our Work

Team # 18 Page 5 of 24

2 Assumptions
To simplify the problem , we make the following basic assumptions, each of which is properly
justified.

• Forecasting trends in electricity supply trends does not take into account the occurrence of
contingencies.
Reason: Sudden events such as wars, financial crises, and global outbreaks of infectious diseases
can cause major disruptions in the trend of electricity supply. Sudden time is a small probability
event and is beyond the scope of our study. We study the development trend of electricity supply
in general.

• Large-scale photovoltaic power plants are equal in value to distributed photovoltaic power
plants.
Reason: Distributed photovoltaic (PV) power plants are suitable for small-scale installations,
such as homes, and enable greater utilization of solar energy. It has as much value as a
large photovoltaic power plant. When calculating the feasibility of building a power plant, the
difference between large and small power plants is not taken into account.

• The amount of electricity generated is only related to the amount of solar irradiation
received by the PV module and the solar energy conversion efficiency of the module.
Reason: The study does not consider other losses in the PV power generation process, such as
the loss of electricity during grid transportation. This loss is related to the length of the grid,
which is difficult to calculate and falls outside the scope of our research.

• The cost of building a PV plant and the feasibility of building the plant are consistent with
a learning curve model.
Reason:The higher the feasibility of building the power station, the lower the cost of building
the station. The decrease in reference cost is the result of accumulated experience, and the cost
of building a station and the feasibility of building a station are also consistent with the learning
curve model.
Team # 18 Page 6 of 24

3 List of Notation

Table 1: Notations

Symbol Description Unit

Ps Total electricity supply 108 kW h
h Maximum height difference KM
U( x, y) Available area 1KM 2
Hi,t Mountain Shadow 1KM 2
A Pixel area 1KM 2
X Restricted area 1KM 2
θ Slope degree
Note: Symbols not listed in the table will be stated as they first appear.

4 Data Overview
4.1 Data Collection
Our study collected data on China’s socio-economic aspects, including GDP, urbanization rate,
and industrial structure, from 2007-2022, as reported in the China Statistical Yearbook. Additionally,
we obtained data on China’s annual power generation from 2007-2022 from the China Electric Power
Yearbook. To further analyze China’s environmental factors, we collected data on China’s DEM, solar
irradiation, and land use types from the Resource and Environmental Science Data Registration and
Publication System. Table 2 displays the specific data sources. The remaining data is obtained from
the related literature, and the references are described in the text.
Due to the large amount of data, it is not convenient to list them all, so visualizing the data for
display is a good method.

Table 2: Date and Date Sources

Date Data Sources

social& economic https://www.stats.gov.cn/sj/ndsj/
power generation capacity https://www.stats.gov.cn/
resource & environment https://www.resdc.cn/Default.aspx

4.2 Data Visualization

The visualization of the collected data related to China’s resources & environment is shown in the
figure below. Meanwhile, Social & economic data and power generation data from 2007 to 2022 are
shown in the annex .
Team # 18 Page 7 of 24

Figure 2: DEM Figure 3: Solar Radiation Figure 4: Railway

Figure 5: Land Types Figure 6: nature preserves

5 Problem One
Question 1 asks us to analyze the relevant factors affecting the supply of electricity and to forecast
the trends in the supply of electricity from 2024 to 2060.

5.1 Problem analysis

Problem 1 we classified power supply into five categories: hydro-power, thermal power, nuclear
power, wind power, and solar power. The literature review[1] reveals that GDP, population, output
value of primary, secondary, and tertiary industries, and urbanization rate are the factors affecting its
development. The data from 2012 to 2022 are used as a training set to construct a decision tree for
predicting the future development of China’s electricity supply from 2024 to 2060.

5.2 Preparation of the model

5.2.1 Influencing factors analysis
Electricity supply in China is affected by a variety of factors, and we compile statistics on the
historical data and future trends of socio-economic indicators that are closely related to the development
of electricity supply, such as GDP, urbanization rate, population size, and industrial structure. We study
their relationship with electricity supply.
Team # 18 Page 8 of 24

• GDP:The economy cannot develop without the supply of electricity. The proportion of electric
energy in terminal energy consumption is increasing. And GDP growth is an important driving
force for the growth of electricity consumption, thus promoting the development of electricity
supply.

• Urbanization level :The electricity demand of urban population is very different from that of
rural population. Moreover, in the process of urban development, the level of electrification
will continue to increase, thus increasing electricity consumption. However, as the economy
develops, the difference between urban and rural electricity demand will gradually decrease.
Therefore, we need to study their specific relationship with electricity supply.

• Size of population:An increase in population will inevitably lead to an increase in the demand
for electricity. This in turn affects the development of electricity supply. Population is the most
direct cause of changes in electricity consumption.

• Industrial structure :Different types of industries have different energy demand structures. We
use the ratio of secondary industry to tertiary industry to study the relationship between industrial
structure and electricity supply.
The trends of the above indicators were obtained through the study of China Energy Outlook
2060 (2024 Edition)[2], as shown in Figure 7.

Figure 7

5.3 Establishment of the model

5.3.1 Electricity supply equation
Electricity supply is divided into five main categories: hydro-power, thermal power, wind power,
nuclear power, and solar power.

Ps = p h + p t + p w + p n + p s . (1)

where Ps represents total electricity suppy. ph , pt , pw , pn represent hydro-power, thermal power, wind
power, nuclear power and solar power respectively.
Team # 18 Page 9 of 24

pt = P s − ph − pw − pn − ps . (2)
The Chinese government’s "14th Five-Year Plan for a Modern Energy System"[3] indicates that
it will vigorously develop non-fossil energy sources and promote the construction of a new type of
energy system.It aims to accelerate the development of wind and solar power generation, actively
develop nuclear power in a safe and orderly manner, and reduce reliance on traditional fossil energy
combustion for power generation.That is, the trend of power supply of thermal power will be in a
declining region, which is significantly different from the other four, so we first forecast the total power
supply, and then subtract the four components of hydro-power, wind power, nuclear power and solar
power respectively from the total to get the development trend of thermal power.Thermal power is
valued as total electricity supply minus, in descending order, hydro-power, nuclear power, wind power,
solar power.

5.3.2 CART regression tree

Use tree modeling to do the regression problem of each indicator with electricity supply. It is used
to analyze the relationship between the above indicators and power supply, and it is used to predict the
trend of power supply. The decision tree is based on the principle of entropy minimization as the basis
for selecting features and splitting. Using the structure of the decision tree, classification and decision
making is done based on the features of the training data and finally an electricity supply is output.
GDP, urbanization rate, population, secondary industry share, and tertiary industry share are
normalized. The normalization formula is shown in Eq. (3).
x − xmin
xnormalize = , (3)
xmax − xmin
where x is the original data,xmin xmax are the minimum and maximum values of the data respectively.
Decision Tree Regression is an effective data mining algorithm that learns and predicts quantitative
variables quickly and universally. Decision tree regression can effectively learn information about
quantitative variables and use it to predict and model future trends. As a classification algorithm based
on the principle of entropy minimization, it uses the structure of a decision tree to classify and make
decisions based on the features of the training data, and finally outputs a definite result.
Data preparation:The data were divided into training set and test set data, the training set data were
used to construct the model and the test set data were used for model evaluation. Combine the existing
data [4], harmonize the time series of the data, and select the data from 2012-2022 for integration.
Feature selection:Based on the Gini coefficient, the best feature is selected as the current node’s
segmentation criterion. This selection will be done based on the impurity of the dataset.
Constructing a decision tree: Based on the selected features, the dataset is partitioned into subsets,
each corresponding to a branch. This process proceeds recursively until termination conditions are
met, e.g., maximum depth is reached, minimum number of samples is reached, etc.
Recursive Segmentation: for each subset, repeat steps 2 and 3 to find the best segmentation
features and continue building the decision tree. This process will keep recursing down until all leaf
nodes are assigned categories or predicted values.
Pruning: the constructed decision tree may overfit the training data, resulting in poor generalization
ability. Therefore, pruning is needed to improve the generalization ability of the model. Common
pruning methods are pre-pruning and post pruning.
Team # 18 Page 10 of 24

Prediction and Evaluation: Use the constructed decision tree to predict new samples for clas-
sification or regression. The performance of the decision tree model is evaluated by calculating the
RMSE,relative error.

5.4 Model solution

A total of five decision tree regression models are solved to obtain GDP, urbanization rate, popula-
tion, secondary industry share, and tertiary industry share with total electricity, hydro, nuclear, wind,
and solar power generation, respectively. In particular, hydro-power is capped at 21300.
Taking the total power as an example, the decision tree structure obtained by solving is shown in
Fig8.

Figure 8: Modeling process

5.5 Results analysis

Calculate the difference between the electricity supply predicted by the electricity supply model
and the actual electricity supply through RMSE, and quantify the accuracy of the electricity supply
model we constructed through the relative error.
RMSE is used to measure the degree of difference between predicted and actual observations.
v
u n
u1 X
RM SE = t (yi − ŷi )2 (4)
n i=1

where n represents the number of observation samples.yi represents the actual observed value of theith
observation sample,and ŷi is the predicted value of the it h observation sample.
Team # 18 Page 11 of 24

Table 3: RMSE

PS ph pn pt pw
RMSE 963.87 261.8 66.71 871.24 312.85

Relative error is a common method of assessing the difference between the measured or estimated
value and the true value, which quantifies the accuracy of the measurement or estimate.
measured or estimated value - true value
relative error = . (5)
true value
The relative error in total electricity supply for 2020-2060 is shown in the table below:

Table 4: The relative error in total electricity supply

2025 2030 2035 2040 2045 2050 2055 2060

relative error 0.0089 0.0035 0.0047 0.0032 0.0025 0.016 0.012 0.016

From the above table, it can be seen that the error between the predicted value and the actual value
of the power supply prediction model is almost close to 0, so the accuracy of the model is very high.
The structure and development trend of electricity supply in 2020-2060 is shown in Fig9.

Figure 9: Modeling process

From the figure12, we can see that the total power supply is a continuous growth trend, hydro-power
in the development of the threshold 21,300 after the threshold has always remained in the threshold,
thermal power power supply is reduced, nuclear power power supply is a small and stable increase in
the supply of electricity, wind power, Solar power power supply has increased significantly and rapidly,
gradually replacing thermal power to become the main source of electricity, the structure of power
Team # 18 Page 12 of 24

supply to reduce reliance on the traditional combustion of fossil fuels to generate electricity, increase
reliance on renewable energy, and vigorously develop clean energy power generation. The power
supply structure to reduce reliance on traditional fossil fuel combustion power generation, increase
reliance on renewable energy, and vigorously develop clean energy power generation. The overall
development trend is in line with the "14th Five-Year Plan for a Modern Energy System".

6 Problem 2
6.1 Problem analysis
Question 2 asks us to consider the siting of a PV system in terms of costs and benefits, as well as
geography and lighting conditions, and to provide an evaluation model that can be used to assess the
suitability of an area for a PV system.

6.2 Preparation of the model

In analyzing the feasibility of constructing a PV plant in a region, we considered policy, cost,
revenue, geography, and lighting conditions[5].In terms of policy, the Chinese government has regula-
tions on the construction of photovoltaic power plants that must not affect ecological and agricultural
production, which requires us to analyze the type of land in the region.Geographically, photovoltaic
plants are suitable to be established on flat spreading terrain and gently sloping hills. The terrain can be
determined by the gradient and direction of the slope.In terms of revenue, for photovoltaic power plants,
the richer the solar resource, the greater the revenue of power generation in the region.In terms of cost,
the initial construction and maintenance costs of a power plant are highly related to its transportation
costs, which requires us to consider the distance of the power plant from the road network.In terms of
lighting conditions, PV modules should not be obscured by shadows during operating hours[6].

6.2.1 Land policy

Construction of photovoltaic power plants should avoid impacts on ecology and agricultural pro-
duction.According to the land use types given by the Resource and Environment Science and Data
Center of the Chinese Academy of Sciences [6], paddy fields and dry lands in the arable land types,
rivers and canals, lakes, reservoirs and pits and ponds in the watershed types, as well as nature preserves
are restricted areas.

6.2.2 Quantity of solar radiation

The abundance of solar energy resources directly affects the power generation revenue of PV power
plants. Classification of total annual solar radiation according to the National Technical Committee for
Standardization of Meteorological Disaster Prevention and Reduction.

6.2.3 Distance from road & railroad

The closer the site of a power plant is to human settlements and the road network represents the
smaller its cost to build and maintain the plant, with denoting the straight-line distance between pixel i
Team # 18 Page 13 of 24

and its nearest county, and denoting the distance between pixel i and its nearest railroad-road network.

6.2.4 Slope & Aspect

The aspect of slope of the PV plant site is oriented to the south, and the aspect should not be too
high. Using the plane slope algorithm.Calculate the slope θi at pixel i according to Eq. (6).
∆h
tan−1 θi = (6)
∆d
where ∆h is the maximum height difference between the raster point at i and the neighboring raster
point in each direction; ∆d is the distance difference between the raster point at i and the raster point
with the maximum height difference.aspect is determined by the calculated slope value.

6.2.5 Mountain Shadow

It is not desirable for the PV module to be shaded by the mountain shadow within the winter solstice
time 9:00-15:00. Calculate the mountain shadow value of pixel i at each whole point within the winter
solstice time 9:00-15:00 according to Eq. (7) If the mountain shadow value of pixel i at time t is 0, it
means that it is located in the mountain shadow at time t.

Hi,t = 225 ∗ [cos αt cos βi + sin αt sin βi cos(γt − δi )] (7)

where αt and γt are the solar altitude angle and solar azimuth angle at time t, respectively, and βi and
δi are the slope and aspect at pixel i, respectively.
Based on the above principles, the indicators were graded as shown in the table 5.

Table 5: Hierarchy of levels

level solar radiation dis to road net work slope aspect shadow
1 0-1050 500-3000 49∼90 north(0∼22.5)(337.5∼360) 5∼7
2 1050∼1190 320∼500 30∼40 northeast(22.5∼67.5) 3∼4
3 1190∼1330 250∼320 20∼30 east(67.5∼112.5) 1∼2
4 1330∼1470 180∼250 15∼20 northwest(292.5∼337.5)
5 1470∼1610 140∼180 west(247.5∼292.5)
6 1610∼1750 100∼140 10∼15
7 1750∼1890 60∼100
8 1890∼2030 40∼60 6∼10 southeast(112.5∼157.5)
9 2030∼2170 20∼40 3∼6 southwest(202.5∼247.5)
10 2170∼2500 0∼20 0∼3 south(157.5∼202.5) 0

6.3 Model establish

To solve this problem, we mainly use the raster method, which is a commonly used method in GIS
to represent geographic information by dividing the geographic space into small blocks, each of which
is called an pixel or raster. In the feasibility analysis of PV power plant construction, we take 1km as
Team # 18 Page 14 of 24

the raster unit and rasterize to create a digital elevation model (DEM) composed of raster points, which
represents the geographic information of various factors in the form of raster, and facilitates spatial
analysis and comprehensive evaluation.

6.4 Site selection model

Ux,y = Ax,y − Tx,y − Rx,y − Dx,y − Wx,y − Sx,y , (8)

where (x,y) are the coordinates of the raster points of the target area, and the set at U(x, y)=1 is the
available area; the set at Ax,y=1 is the target area for the construction of the photovoltaic station; the
set at T(x,y)=1 is the area that does not satisfy the requirement of slope; the set at R(x,y)=1 is the area
that does not satisfy the requirement of protected land; the set at D(x,y)=1 is the area that is shadowed
by the mountain; the set at W(x,y)=1 is the area that does not satisfy the requirement of highway and
railroad; the set at S(x,y)=1 is the area that does not satisfy the overall solar radiation amount.

6.4.1 Restricted area

According to the principle of protecting agricultural production and ecology[5] as well as policy
regulations, nature reserves, arable land, rivers, lakes and reservoirs will be set up as restricted areas
for photovoltaic power plants.

0, pixel i is in the restricted area.
Xi = (9)
1, otherwise.

6.4.2 Hierarchical analysis to determine indicator weights

The feasibility decision problem of constructing PV power plants is decomposed into three levels,
the top layer is the target layer, i.e., it is the selection of suitable construction of PV power plants; the
middle layer is the quasi-measurement layer, and the five indexes of the solar radiation, the distance
from the road network, the slope, the direction of the slope, and the shadows are used as the guideline
layer in the hierarchical analysis; and the lowest layer is the scenario layer, i.e., it is the individual pixels
in the region.

Figure 10: Hierarchical Analysis Schematic

Team # 18 Page 15 of 24

Qualitative two-by-two comparison of criterion indicators to get the importance degree of different
criterion indicators. Take Pij to denote the importance degree of the ith indicator compared with the
jth indicator, when Pij is 1, it means that the indicator i is equally important as the indicator j, and
Pij from 1 to 10 means that the importance degree is increasing step by step. The judgment matrix is
shown in table8.

Table 6: Judgement matrix

Facotr Solar radiation Dis to the road net Slope Aspect Shadow
Solar radiation 1 4 6 3 2
Dis to the road net 0.25 1 2 0.5 0.33
Slope 0.17 0.5 1 0.25 0.2
Aspect 0.33 2 4 1 0.5
Shadow 0.5 3 5 2 1

Before calculating the weights of the indicators, the consistency test was performed on the judgment
matrix. The consistency ratio CR = 0.0141 is obtained, and since CR < 0.10, the consistency of this
judgment matrix A is acceptable.

6.4.3 PV plant construction suitability model

According to the above method, the PV power plant construction suitability model is shown in Eq.
(). The flow of PV power plant siting suitability assessment is shown in Fig. (10).
6
X
f easibility = Xi ωj F actorj , (10)
j=1

where ω is weight. Factor are five indicators for solar radiation, distance from road net, slope, aspect,
shadow.

Figure 11: A model for evaluating the suitability of photovoltaic plant selection
Team # 18 Page 16 of 24

6.5 Solution of the model

In order to ensure the robustness of the results, the weights of the indicators were calculated from
the judgment matrix by the three methods of arithmetic average, geometric average, and eigenvalue
method respectively, and then the scores of each pixel were calculated based on the obtained weights.
This avoids the bias arising from the use of a single method, and the conclusions drawn will be more
comprehensive and valid.

Table 7: The weight results obtained by AHP

Weight
Arithmetic mean Geometric mean Eigenvalue method Average value
Solar Radiation 0.4198 0.4207 0.4224 0.4210
Dis to the road net 0.0948 0.0947 0.0938 0.0945
Slope 0.0528 0.0520 0.0521 0.0523
Aspect 0.1666 0.1649 0.1650 0.1655
Shadow 0.2660 0.2676 0.2666 0.2667

6.6 Results analysis

We select Tibet and discuss the feasibility of constructing PV power plants in Tibet. Through Site
selection model combined with hierarchical analysis table, we conclude that the average suitability of
Tibet is 8.89, and the area suitable for constructing PV power plants in Tibet is characterized by a large
area and distribution aggregation, which is suitable for constructing large-scale PV power plants.

Figure 12: Suitability of PV power plant construction in Tibet

Combined with the figure note, we can see that from red to blue represents a gradual decrease
in suitability. Red, orange area is suitable for the construction of photovoltaic power plants in the
region, high suitability is mainly due to high solar irradiation, as well as is located in the Lhari
Railway, Sichuan-Tibet Railway, Lalin section and the Qinghai-Tibet Railway, Gelar section of the
Tibet T-shaped railroad network together on the main skeleton. The green area is not suitable for the
Team # 18 Page 17 of 24

construction of photovoltaic power plants in the region. Blue areas are restricted areas such as rivers,
lakes and nature reserves.

7 Problem three
The task assigned to us is to determine the maximum potential for photovoltaic (PV) power
generation in China, considering various factors such as geographic resources, investment capacity,
cost, and revenue. Our objective is to calculate the maximum amount of power that can be generated by
PV. To achieve this, we calculated the solar irradiation for each rectangular pixel, slope, slope direction,
land type, and whether it is a restricted area (nature reserve, arable land, river, lake, and reservoir)
based on the rasterization of Problem 2[5].

7.1 Generation Potential Model

Electricity generation potential in a region:
X
Ppotential = Rsolar (i) ∗ Apv (i) ∗ η, (11)
i

where Rsolar (i) is the solar irradiation of the ith pixel, Ap v(i) is the area of PV module in the ith pixel,
and η denotes the utilization factor of PV module.
In this paper, we take monocrystalline silicon PV modules as an example, and take the conversion
efficiency of PV modules as 20%, and the electrical efficiency as80%.So the generation potential is the
product of PV module area, total horizontal irradiance PV module utilization factor is constant 0.16
The photovoltaic module area Apv is obtained by using the effective land area, the coefficient of
glass-similar utilization, the coefficient of glass-degree utilization, and the coefficient of land type
utilization, as shown in Eq(12).
X
i i i
Apv = Xi ∗ A ∗ ηaspect ∗ ηslope ∗ ηland , (12)
i

where Xi denotes whether the ith pixel is within the restricted area, A is the area of the pixel, and
i i i
ηaspect , ηslope , ηland denote the utilization coefficient of the ith pixel in terms of the direction of the
slope, the slope utilization coefficient and the utilization coefficient of the land type, respectively.

7.2 PV plant construction planning model

X
max (Pi ∗ Yi ∗ P ricei )
Y
i
 P

 Cinv ≥ i (Costi ∗ Yi ),


 (13)
−φ
s.t. Costi = Cbase ∗ f easibilityi ,




Y ∈ {0, 1}.


where Pi is the power generation of pixel i.f easibilityi is the suitability of pixel i.Yi is the 0-1 variable,
whether or not to choose to build a PV plant at pixel i.P ricei is the price of electricity at pixel i.
Team # 18 Page 18 of 24

7.2.1 Photovoltaic power plant construction target

Considering that photovoltaic power generation will be the main source of energy supply in the
future, the operation of photovoltaic power plants needs to be long-term and sustainable, and it is
necessary to ensure that photovoltaic power plants have a certain degree of profitability in order to
repay the construction investment in the early stage, as well as to meet the operating expenses and
maintenance costs incurred in the process of future operation.
We aim to maximize PV plant revenues:

Revenue = Electricity generation * Electricity price, (14)

Since we use the province as the unit of division in the regional division, the tariffs refer to the provincial
tariffs.

7.2.2 Investment-cost constraints

X
Cinv ≥ (Costi ∗ Yi ), (15)
i

where Cinv is the amount of investment.

Costi = Cbase ∗ f ea−φ

i , (16)

where Cbase is the base cost when suitability is 1,φ is the coefficient of cost elasticity with respect to
suitability.After comparing the resulting pixels for the four cases where φ is taken as 1.5, 1, 0.85, and
0.5, respectively, it can be concluded that φis most reasonable when taken as 0.85.
Costi ≥ 0,as suitability increases, Costi decreases,the higher the coefficient of elasticity, the faster
the cost decreases with increasing suitability.

7.3 Result analysis

The annual generation potential value of a like dollar ranges from 0 to 164.3 kwh/m2 , with a total
annual generation capacity of 5839.4348 ∗ 108 kW h for the whole region of our country.When φ is
0.85,the cost is 419.5273,the revenue is 1675.6619.
Team # 18 Page 19 of 24

Figure 13: Photovoltaic power station location results

Based on results from Fig13, it can be seen that the suitable areas for the construction of PV
power plants in China are most of Tibet Province, the central and eastern part of Xinjiang Province,
the northern part of Qinghai Province, the northwestern part of Gansu Province, the northern part of
Ningxia, and the western and southern parts of Inner Mongolia. Northwest China is characterized by
year-round drought and low rainfall, low cloud cover, high total solar radiation, and small inter-annual
variations in sunshine duration, making it more suitable for the construction of PV power plants.
Large-scale PV power plants are suitable for construction in the northwest of China, and distributed
PV power plants are suitable for construction in the south of China, which is helpful for meeting the
power supply in most parts of China.

8 Problem Four
Carbon Peak means that carbon dioxide emissions reach a maximum value in a particular year
and then enter a declining phase. Carbon Neutral refers to a period of time in which carbon dioxide
generated by specific organizations or social activities as a whole is absorbed and offset by natural and
man-made means, such as afforestation, oceanic absorption, and engineered storage, so as to achieve
relatively "zero" carbon dioxide emissions from human activities. The Chinese Government’s strategic
goal of achieving peak carbon and carbon neutrality by 2060 means that China will control its carbon
emissions at a high point before that time and gradually reduce them until they reach net-zero emissions.
Energy transformation is the key to realizing medium- and long-term carbon neutrality, but the
contradiction between energy categories and the requirements of energy system transformation is a
problem that needs to be solved urgently. The realization of the goal of carbon neutrality requires the
matching of the whole process of the energy demand side and the supply side, and for China, it is
especially important to promote the low-carbon transformation of energy sources, i.e., upgrading and
optimizing for the realization of the goal of carbon neutrality in terms of increasing the proportion of
clean energy consumption and building a new type of power system. The energy transition characterized
by clean, low carbon and smart and efficient energy is the main trend of future energy development.
Team # 18 Page 20 of 24

According to the power supply structure development prediction model established by us in question
one, the development trend of power supply structure from 2025 to 2060 is predicted as follows: the
proportion of solar power and wind power will grow rapidly year by year, and the proportion of thermal
power will decrease year by year. Therefore, we can deduce that the clean energy represented by solar
power and wind power is gradually and orderly replacing the traditional fossil energy under the premise
of guaranteeing to meet the demand for electricity. The overall trend of energy structure transition is
in line with Low Carbonization Transition[7].
Photovoltaic power generation as the main part of solar power generation, from the power generation
potential model established in question three we conclude that China’s total annual power generation
capacity of 5839.434888661253 billion kWh for the whole region, which is a huge power generation
potential. If this solar energy resource is utilized, the bright future of the PV power generation industry
is conducive to increasing the proportion of clean energy consumption, promoting the low-carbon
transformation of energy sources, and thus achieving the goal of carbon neutrality.
Factors affecting the construction of photovoltaic power plants:

• Supply of raw materials:Silicon is the raw material for photovoltaic power plants. The relevant
supply chain and PV power plant construction synergistic development, for the supply of materials
to meet the demand for power plant construction.[8].

• Investment allocations:According to the 12th, 13th and 14th Five-Year Plans for a Modern
Energy System issued by the National Development and Reform Commission (NDRC) of China,
it can be seen that the development trend of the Chinese government’s investment is to increase
year by year, and the forecasts for China’s GDP development in question 1 are also on an
increasing trend, which satisfies the constraint that the amount of investment is greater than the
total cost of constructing the plant [3].

• Technological developments:Promoting the construction of power grids, developing new smart

PV modules and improving conversion efficiency [3].

• Green Development:Establishment of a sound economic system of green, low-carbon and

recycling development, promotion of a comprehensive green transformation of economic and
social development, upgrading the proportion of renewable energy utilization, and vigorously
promoting the development of wind power and photovoltaic power generation[4].

9 Sensitivity analysis of the model

The PV plant construction planning model established in this paper based on 0-1 integer planning
is somewhat subjective in the value of the correlation coefficient. Since reality often falls short of
the ideal state we envision, in order to ensure that the evaluation model is more objective, relevant,
independent, practical and scientific, it is necessary to discuss the effect of coefficient of cost elasticity
with respect to suitabilityφ on the suitability of PV plant construction.
To ensure the fairness of the experiment, we only made small changes (up or down no more than
0.05).We revise φ up and down by 5%.The results are shown in the table(8).
Team # 18 Page 21 of 24

Table 8: Adjusted economic value of different φ

φ 0.85 0.8075 0.8925

Cost 419.527344041737 440.4174247 399.9365124
Revenue 1675.661894 1654.771813 1695.252726

Conclusion: The influence ofφ on our model is small, and ourmodel is stable without being affected
by φ.

10 Strength and Weakness

10.1 Strength
• We use decision tree regression, raster method and 0-1 linear integer programming are several
simple methods to solve the problem.These methods have a simple design and are easy to
understand, yet they produce highly accurate prediction results.

• Our model is fairly robust due to our careful corrections in consideration of real-life situations
and detailed sensitivity analysis.

• Our model effectively achieves all of its objectives. A comprehensive model has been developed
to study the strategic planning for the sustainable development of photovoltaic power generation
in China.

• We do a good job of visualization, such as maps and some structured schematics for assessing
the suitability of building PV power plants across China. Dull data may reflect patterns, but they
are not as visual as so many images.

• In addition, our model takes into account the time dimension and has good scalability.

10.2 Weakness
• Having knowing the range of some parameters from others’ essays, we choose a value from them
to apply in our model. Those values may not be reasonable in reality.

• Although we have conducted a large number of surveys on the indicators affecting the suitability
of PV plant construction, they are very complex, and some of them, such as the local land price,
have not been taken into account in the cost, which requires further research.

11 Letter
A Letter to Chinese Government
From：Team 202418
To:Chinese Government
Subject:Strategic planning for sustainable development of photovoltaic power
generation in China
Date:January 21, 2024

Dear Sir or Madam:

The world's energy pattern is changing in the direction of cleanliness, low carbon, greenness, high efficiency and
diversification. China's energy development should take into account the goals of energy security for economic
development and carbon emission reduction, accelerate the construction of a new type of power system system based
on new energy, and actively promote the transformation of China's high-carbon energy structure based on coal to a
low-carbon energy structure based on new energy. Our team has conducted research on this topic on the development
trend of power supply, the siting of photovoltaic power plants, and obtained the following results.
First of all, based on the data provided by the National Bureau of Statistics of China (NBS) on electricity supply,
GDP, urbanization rate and other data, we have built a forecast model of electricity supply. We found that from 2024 to
2060, China's energy structure shows a decrease in fossil energy consumption, non-fossil energy, nuclear power to
maintain the safe and orderly development, hydro-power to develop in a steady and orderly manner, wind and solar
energy to accelerate the development of China's energy consumption gradually become the main trend of development.

Secondly, based on the study of PV power plant Based on the results of our research, we make the
siting, we mainly focus on the construction suitability following recommendations to the Chinese government:
of PV power plants using AHP to establish a PV 1. Formulate and improve policies and regulations:
power plant construction suitability assessment Introduce a series of policies and regulations to
model. And on this basis, we use 0-1 integer planning support photovoltaic power generation, reduce the
to establish a power generation potential model, cost of photovoltaic power generation, improve the
which results in a total annual power generation efficiency of power generation, and promote
capacity of 5839.43 billion kilowatt-hours for the industrial development.
whole region of China, which is a huge power 2. Increase technology research and development and
generation potential. If this solar energy resource is innovation: focus on the research and development
utilized, it can greatly alleviate the problem of energy and innovation of photovoltaic power generation
shortage in China. technology, promote technological progress and
As can be seen from the PV power plant siting application, and strengthen cooperation between
suitability assessment chart, the western region of scientific research institutions and enterprises.
China has a strong PV power plant siting suitability 3. Development of PV power generation industry
and a large area, which is suitable for the construction chain: create a complete PV power generation
of large-scale PV power plants. In the eastern region industry chain, optimize resource allocation, and
of China, the site suitability of PV power plants is improve the efficiency and sustainability of PV
scattered, and it is suitable for building small PV power generation.
power plants.
Team # 18 Page 23 of 24

References
[1] YAO Li MA Jie WU Chenrui & DUAN Jinhui SIN Baoguo, JI Xingpei. Trends in the evolution
of china’s electricity supply and demand patterns under high-quality energy development. China
Light and Power, 54(1-9+18), 2021. 5.1

[2] Cao Yong Liu Xiaoxia Luo Daqing Ding Xuansheng... & Mari Dai Baohua, Wang Deliang. China
energy industry review 2022 and outlook 2023. Contemporary petroleum and petrochemical,
31(2-9), 2023. 5.2.1

[3] Du Wei & Man Teng. The "14th five-year plan" for a modern energy system and other policies were
introduced to lay out china’s new energy system. International Petroleum Economics, 31(9-10),
2023. 5.3.1, 8

[4] Huang Weiwei Yu Jianning& Huang Haixia Kuang Lichun, Zou Cai Cai. Trends in china’s energy
demand forecast and transition under the vision of carbon peak and carbon neutrality. Petroleum
Science and Technology Forum, 41(9-17), 2022. 5.3.2, 8

[5] Zhou Yushi Zhang Yan Wang Haijin, Tang Ruogasa. Photovoltaic power plant siting evaluation
based on arcgis and multi-factor modeling. Solar Energy Journal, 44(120-130), 2023. 6.2, 6.4.1,
7

[6] Wei Guanxiang Zhao Qing Lei Ming. Research on macro-siting scheme of large-scale photovoltaic
power station based on gis technology. solar energy, (43-48), 2023. 6.2, 6.2.1

[7] Du Ershun Zhang Ning Nielsen Zhuo, Zhenyu. Research on macro-siting scheme of large-scale
photovoltaic power station based on gis technology. solar energy, (43-48), 2023. 8

[8] Shen Guiying. Ministry of industry and information technology and other 3 departments issued
the notice on promoting collaborative development of photovoltaic industry chain supply chain. ,
30(12), 2022. 8
Team # 18 Page 24 of 24

12 Appendix

Total
Urbanization Primary Secondary Tertiary
index GDP(109 yuan) population
rate (%) industry industry industry
(104 people)
2007 270092.3 45.89 132129 27674.1 126630.5 115787.7
2008 319244.6 46.99 132802 32464.1 149952.9 136827.5
2009 348517.7 48.34 133450 33583.8 160169.8 154765.1
2010 412119.3 49.95 134091 38430.8 191626.5 182061.9
2011 487940.2 51.83 134916 44781.5 227035.1 216123.6
2012 538580 53.1 135922 49084.6 244639.1 244856.2
2013 592963.2 54.49 136726 53028.1 261951.6 277983.5
2014 643563.1 55.75 137646 55626.3 277282.8 310654
2015 688858.2 57.33 138326 57774.6 281338.9 349744.7
2016 746395.1 58.84 139232 60139.2 295427.8 390828.1
2017 832035.9 60.24 140011 62099.5 331580.5 438355.9
2018 919281.1 61.5 140541 64745.2 364835.2 489700.8
2019 986515.2 62.71 141008 70473.6 380670.6 535371
2020 1013567 63.89 141175 78030.9 383562.4 551973.7
2021 1149237 64.72 141212 83216 451544 614476
2022 1210207 65.22 141260 88207 473790 642727

Total power
Thermal Nuclear
index generation Hydropower Wind power Solar power
power power
(106 kWh)
2007 32644 4714 27207 629 57 0
2008 34510 5655 28030 692 130 1.7
2009 36811 5716 30116 700 276 1.52
2010 42277 6867 34166 747 494 2.65
2011 47306 6681 39003 872 741 6
2012 49856 8556 39255 983 1030 36
2013 53721 8921 42216 1115 1383 84
2014 56045 10601 42274 1332 1598 235
2015 57399 11127 42307 1714 1856 395
2016 60228 11748 43273 2132 2409 665
2017 64171 11931 45558 2481 3034 1166
2018 69947 12321 49249 2950 3658 1769
2019 73269 13021 50465 3487 4053 2240
2020 76264 13553 51770 3662 4665 2611
2021 83959 13399 56655 4075 6558 3270
2022 86939 13517 57337 4178 7624 4276