Evaluating the Suitability of Ground-Mounted Photovoltaic System Selection and the Differences Between Expert Assessments and Firm Location Preferences: A Case Study of Tainan City

Abstract

Responding to the challenges of global climate change and domestic air pollution, Taiwan revised its energy policy in recent years, introducing an energy transition strategy focused on low-carbon and clean energy. However, if photovoltaic installations are not properly sited, they may have negative impacts on the local environment. Previous research on renewable energy has primarily focused on policy evaluation, with limited attention given to case studies that examine the suitability of site selection for PV system installations. Thus, this study incorporates the Fuzzy Delphi Method (FDM) and the Analytic Hierarchy Process (AHP) to explore the criteria for evaluating site suitability for ground-mounted PV systems. This study considers existing sites with completed ground-mounted PV systems in Tainan City as case study subjects. The results indicate that the most important factor, as prioritized by experts, is the distance from Class I environmentally sensitive areas, followed by the duration of insolation, proximity to the electrical grid, and distance from residential areas. The evaluation model developed in this study provides a valuable reference for future site selection of ground-mounted PV systems. Establishing dedicated PV energy parks also may offer a viable solution to mitigate disputes related to the deployment of ground-mounted PV systems.

1. Introduction

Excessive carbon dioxide emissions due to human activities have been shown to cause climate change [1], and countries worldwide increasingly focus on developing renewable energy sources. The Glasgow Climate Pact, adopted at the 2021 United Nations Climate Change Conference, upholds the goals set by the Paris Agreement, aiming to limit the global temperature rise to within 1.5 °C and gradually reduce coal use to lower carbon emissions. Overall, international environmental protection is urgent and requires everyone to reduce carbon emissions by using renewable energy, reducing pollution and promoting green investment [2]. In 2009, Taiwan implemented the “Renewable Energy Development Act” as a pioneering policy to promote renewable energy and enhance Taiwan’s sustainable development. In response to the challenges brought by global climate change and domestic air pollution, Taiwan then revised its policy in 2016, proposing an energy transition strategy centered on low-carbon and clean energy. In 2022, the National Development Council of Taiwan officially announced its 2050 net-zero carbon emissions roadmap and policies, further accelerating the development of renewable energy [3,4].

The photovoltaic (PV) industry is one of the principal components in Taiwan’s ongoing industrial development. Supported by a series of government policies, the installation of PV systems has increased recently. Currently, PV systems are categorized into three types: roof-mounted, ground-mounted, and building integrated (

Table 1. Advantages and limitations of common renewable energy sources.

Type of Source	Advantage	Disadvantage
Hydropower	Abundant, clean, and safe. Water can easily be stored in reservoirs. Relatively inexpensive. Provides recreational benefits, such as boating and fishing.	Areas surrounding the reservoir may be at risk of flooding due to dam failure. The construction of reservoirs has significant impacts on local ecosystems and hydrology [1]. Can only be used in areas with abundant water resources and suitable terrain.
Solar energy	Has the potential for unlimited supply. The power generation does not produce air, water, or noise pollution [5,6,7].	An energy storage system is needed to store the collected energy for variable demands [8]. Reliability depends on sunlight exposure [9].
Wind power	Land surrounding wind farms can be used for other purposes. The power generation process does not produce air or water pollution [8].	An energy storage system is needed to store the energy for variable demands [10]. Many areas with strong winds are located far from densely populated regions, making power transmission an important consideration. There is noise and the risk of wind turbine blades killing birds.
Biomass	Can burn waste and reduce landfill use Can support local agricultural and forestry industries [10]. Compensates for the instability of intermittent output of other renewable energy sources.	The continuing expense for it to be collected and transported to power plants. High construction costs are involved, and a sufficient amount of waste must be burned to achieve economies of scale [11].
Geothermal	Has the potential for unlimited supply. The power generation process is relatively safe and harmless to the environment [9].	High development costs. Corrosion issues incur substantial maintenance costs [8]. The amount of power generated is relatively small.
Ocean energy	An ideal option for island nations [12]. Captures energy that would otherwise be impossible to collect.	High construction costs. Has potential impacts on marine ecosystems [10].

). Of these, ground-mounted PV systems have been the most debatable, due to the involvement of diverse land uses. Moreover, before their construction a comprehensive environmental impact assessment is necessary, often leading to an extended planning period [5]. Taiwan’s government has also attempted to encourage symbiotic collaboration between electricity generation and industries such as agriculture, fisheries, and husbandry. Looking ahead, achieving the long-term power capacity goal of 40 GW to 80 GW will require the utilization of additional land resources [6]. On the other hand, current research on solar energy in Taiwan tends to focus on overall technology or industry, and rarely analyzes a single type of photovoltaic device. This creates a gap both in academia and policy. Therefore, this study aims to provide a comprehensive understanding of current experiences in PV system implementation, categorize and prioritize evaluation criteria for site selection through expert knowledge, and analyze the alignment between theoretical site suitability and actual firm deployment. These objectives are crucial for bridging the gap between spatial energy planning and real-world practices, ensuring that future solar PV projects are both technically feasible and institutionally grounded.

Accordingly, the contents of this study are as follows:

• Review the history of PV system applications across various countries for a comprehensive understanding of the experiences and challenges encountered during their implementation. The study also categorizes the evaluation criteria for PV system site selection, in order to develop an evaluation model for site selection.
• Examine existing PV sites, utilizing the established evaluation model to analyze the discrepancies between the theoretical model and real-world applications. Based on these findings, the study provides recommendations and solutions for future deployment of PV system.

To achieve these objectives, FDM is employed to understand the spatial characteristics of Taiwan and identify the conditions that should be considered in the development and deployment of ground-mounted PV systems. Moreover, AHP is used to assign weights to each evaluated factor through AHP analysis. The results are then compared with existing PV system sites in Tainan for discussion.

2. Literature Review

2.1. The Rise of Renewable Energy and Solar Energy

Due to the depletion of fossil fuels and the environmental issues caused by greenhouse gas emissions, it is essential to develop and transition towards alternative energy sources that are sustainable, safe and economically competitive. The goal is for renewable resources to replace fossil fuels in the long term [7]. Renewable energy is derived from natural resources that replenish faster than they are consumed [8]. The most common renewable energy sources include hydropower, solar, wind, biomass, geothermal, and ocean energy. Among these, solar and wind power account for most of the growth in generation of renewable energy, and each type of renewable energy has inherent limitations [9]. Thus, new renewable energy technologies must be thoroughly assessed and studied before they are deployed to prevent potential social, environmental, and economic issues.

Of the various types of renewable energy, hydropower currently accounts for the largest share of global electricity generation. However, its heavy reliance on stable rainfall makes it vulnerable to the impacts of climate change, and droughts can significantly reduce its production. Moreover, extreme whether events, such as El Niño and La Niña phenomena, are causing even traditionally stable regions to experience abnormal rainfall patterns, further affecting hydropower generation. In contrast, solar energy can generate power from both direct sunlight and diffuse radiation, so it can produce power even in cloudy days, and countries with varying solar resources can effectively harness solar energy [10]. Wind energy, another prominent renewable energy resource, generates power through wind turbines. Since many regions have strong wind resources, offshore wind power, in particular, this resource has enormous potential for large-scale deployment [8]. The technologies for other renewable energies, such as biomass, geothermal, and ocean energy, are still the earlier stages of development. Moreover, not all countries possess the necessary conditions to develop the aforementioned types of renewable energy [11]. Overall, solar energy clearly stands out as the most promising renewable energy sources and has been widely advocated for in many countries (Table 1).

Although many studies address PV site suitability using multi-criteria evaluation frameworks, few directly compare expert-based suitability assessments with real-world PV siting decisions made by firms. This research contributes to the literature by empirically examining this divergence, offering insight into how planning evaluations may be adjusted to reflect practical decision-making in renewable energy deployment.

2.2. Current Status of PV Development and Key Factors in Site Selection

According to a report by REN21 in 2021, as of the end of 2020, the three countries with highest total power generation from PV systems were China, the United States, and Japan. These countries have had distinct operational experiences and have faced different challenges. In China, to avoid utilizing arable land and densely populated areas, most large-scale PV systems were installed in the Gobi Desert or outside the eastern regions. These regions, however, suffer from relatively low grid load capacity, preventing them from accommodating the generated power, which leads to issues related to excess electricity generation [13,14]. As a result, the development and application of PV systems have been increasingly focused on integrating PV with agricultural greenhouses. Despite this shift, most of the PV companies continued to prioritize power generation over crop production [15]. In the United States, utility-scale solar energy (USSE) systems has been concentrated in the states which benefit from vast land areas and abundant sunlight. The direct and indirect environmental impacts of USSE include effects on biodiversity, water resources, soil, human health, land use and land cover changes. These issues span multiple disciplines and necessitate further research and attention [16]. Considering optimal scale and suitable site selection, the U.S. government began to encourage integration of land, energy, and ecology factors in policy development. On the other hand, Japan, which faces insufficient natural resources and a heavy reliance on energy imports, became aware of the importance of energy security after the previous two oil crises [17]. Therefore, the Japanese government began to invest more in the PV industry. However, along with the increase of ground-mounted PV systems, interference by the reflected sunlight, such as glare-related visual disturbances and indoor temperature increase, have been reported from certain regions. In response to these concerns, local governments have started to review and revise inadequate policies and regulations to address the impact of PV systems on landscapes and the environment.

Thus, numerous factors influence the selection of sites for PV systems. A review of the recent literature indicates that many countries are working to systematize the review process in order to mitigate the uncertainties associated with PV system deployment. It is important to note that locations with high potential for PV systems may not necessarily be the most suitable, as the factors involved are often complex [18]. Accordingly, this study organizes and categorizes these multifaceted factors into three major types: physical location conditions, environmental and climate conditions, along with political, economic, and social conditions. The following table outlines how each of these evaluation factors influences PV site selection (

Table 2. Factors affecting site selections for PV installation.

Perspective	Evaluation Factor
Physical Location Conditions	1-1 Population density: The demand of electricity is closely related to population size. PV systems installed in areas with high population densities can serve a large number of people [19]. Thus, population density and its potential of energy demand is a significant evaluation factor.
	1-2 Distance from residential areas: Large PV systems are unsuitable near residential zones, as they may interfere with urban development [18]. In cases where PV systems need to be installed in densely populated areas, buffer zones or other strategies should be considered to minimize the impact on residents [20].
	1-3 Proximity to road networks: Because PV systems require access during installation and for subsequent maintenance activities, it is generally advisable to locate them in areas with easy access to transportation [21].
	1-4 Proximity to power grid: The distance between PV systems and the power grid can hinder the efficient supply of renewable energy, due to energy losses occur in transmission [7]. Thus, the proximity of PV systems to the power grid is a critical factor for PV site selection.
	1-5 Distance from protected areas: Installing artificial infrastructure in the regions with abundant natural resources can disrupt bird migration and lead to deforestation, which further threatens wildlife [7]. Accordingly, such sensitive environmental areas are typically considered inappropriate for the development of solar power systems in many countries [20].
Environmental & Climate Conditions	2-1 Air temperature: While abundant sunlight is a favorable condition to PV systems, excessive sunlight can lead to overheating, which reduces the efficiency and output of the systems [22].
	2-2 Relative humidity: When the water vapor in the air reaches saturation and condenses on the surface of solar panels, the resulting dew enhances the accumulation of dust, which reduces the efficiency of power generation [22].
	2-3 Solar radiation: The amount of solar radiation required to meet the threshold for developing ground-mounted PV systems depends on the evaluation criteria in different countries. This factor influences the subsequent cost-effectiveness calculations for deployment [23].
	2-4 Sunlight duration: Sunlight is an intermittent energy resource, and the total power generated by PV systems is determined by the daily duration of sunlight exposure [22].
	2-5 Wind speed: Strong wind conditions may result in the misalignment of photovoltaic panels from their originally designed orientation [19]. In more severe cases, strong winds may damage the panels [22].
	2-6 Elevation: Installing PV systems in high-altitude areas not only increases construction difficulties but also raises the construction costs and affects power transmission [7,24].
	2-7 Slope: The land slope can affect the efficiency of power generation from PV systems. When solar panels are installed on steep slopes, they may block each other, reducing the overall power output.
	2-8 Orientation: For optimal radiation absorption, panels should be aligned with the direction of sunlight. In the Southern Hemisphere, panels should face geographic north, while in the Northern Hemisphere, they should face geographic south [25].
Political, Economic, and Social Conditions	3-1 Land use: Land use is one of the most influential factors for PV site selection [26]. Areas with certain land uses are unsuitable for the installation of PV system.
	3-2 Low agricultural productivity: Land with potential for agricultural use should be dedicated to crop production, so these areas are not appropriate for the installation of PV systems [27].
	3-3 Visual impact: PV systems are visually detrimental to natural landscape, reducing quality of life in the region [25].
	3-4 Land cover: Sites for PV system installations should avoid areas that have already been utilized [28]. The most suitable locations have sparse vegetation [7].
	3-5 Scale: It is generally necessary to find a location sufficient to accommodate large PV systems [27]. An upper limit for the installation area can help minimize environmental impact and other effects on surrounding areas.
	3-6 Land cost: Ground-mounted PV systems typically require large areas of land, making land acquisition a challenge due to high costs [25]. As land costs increase, the overall costs of installation rise significantly [19].

).

2.3. Research Methods for PV Site Selection

Establishing a stable power supply system and ensuring sufficient power supply for economic and public needs are key government objectives. However, each energy source has its inherent limitations. To minimize the negative impacts of these systems, selecting suitable sites for power generation is essential. Geographic Information Systems (GIS) are widely applied across various fields, including the energy industry. The data derived from analyzing the collected information on various aspects are used for planning and deploying energy systems. Over the past 15 years, the application of GIS in site selection for PV systems has been significantly developed and widely adopted [29].

In addition to the growing integration with GIS, the increasing complexity of energy planning means that considering only a few objectives is insufficient for energy planning. Thus, Multi-Criteria Decision-Making (MCDM) methods have become increasingly popular for site selection in renewable energy projects. Among these methods, AHP, proposed by Saaty in 1987 [30], decomposes complex issues from a levels to lower levels, gathers evaluations from decision makers, and identifies the advantages and limitations of each project [31]. AHP is widely applicable, from small-scale decisions, such as site selection for private companies, to large-scale decisions for public infrastructure. For instance, Wang et al. (2009) integrated the real conditions of a research area with its economic factors, calculated standard weights using AHP, and developed a site selection model for a landfill in Beijing, China [32]. More than half of the related studies have employed traditional AHP as the primary research method [33]. The combination of FDM and AHP is particularly suitable for this research context. FDM allows for handling linguistic uncertainty in expert judgments, while AHP enables structured pairwise comparisons and the derivation of weights. Compared to other MCDM approaches like TOPSIS or ELECTRE, this hybrid method balances qualitative expert input with quantitative rigor, which is crucial when site selection involves subjective and spatially heterogeneous criteria.

3. Research Design

3.1. Research Methods

This study employs the Fuzzy Delphi Method (FDM) and the Analytic Hierarchy Process (AHP) as the research methods to establish an evaluation model for Tainan City. It further examines the existing ground-mounted PV systems in the city, and the objective is to review current site selection procedure and improve future site selection. The detailed process is displayed in Figure 1.

3.2. Delphi Method (DM)

The Delphi Method (DM) integrates expert opinions through a series of anonymous, multi-round questionnaires, ultimately reaching a consensus among the experts. The Fuzzy Delphi Method (FDM) combines Fuzzy Theory with the traditional DM to resolve the limitation of consensus formation by considering not only the range within which expert opinions fall, but also the fuzziness within that range. Since the indicators considered, such as distance, proximity, and density, have fuzzy evaluations, this study uses FDM. Moreover, a double triangular fuzzy function is adopted to identify the degree of convergence in expert questionnaires. The study further examines whether a consensus has been reached among the expert opinions using the grey zone test method. The detailed procedures are outlined as follows [34]:

Step 1: Statistical analysis of the “maximum acceptable value” and “minimum acceptable value”: Design the expert questionnaire, form an appropriate expert panel, and request each expert to assign values for evaluation factors, including minimum acceptable value, the optimal value, and maximum acceptable value.

Step 2: Eliminate outliers beyond two standard deviations: Conduct a statistical analysis based on the “minimum acceptable values” and “maximum acceptable values” given by the experts. Exclude the outliers beyond two standard deviations.

Step 3: Establish double triangular fuzzy numbers: A double triangular fuzzy function is established based on the previously defined values. The “most conservative triangular fuzzy function (ai)”, where $i=(li,mi,ui)$, is formed with the smallest value (li), the geometric mean (mi) and the largest value (ui) of the “minimum acceptable values.” The “most optimistic triangular fuzzy function (Ai)”, where $i=(Li,Mi,Ui)$, is established using the smallest value (Li), the geometric mean (Mi), and the largest value (Ui) of the “maximum acceptable values.”

Step 4: Test for expert consensus: Once the triangular fuzzy functions are established, the grey zone test is employed to assess whether the experts have reached a consensus. The equation for the difference in geometric mean is $Ci=Mi-mi$, whereas that for the grey zone value is $Gi=ui-Li$. The formula for the grey zone test is $Zi=Ci-Gi$. The testing can be categorized into three scenarios as displayed in the table below. If the experts’ opinions do not reach a consensus or if they diverge, the previous steps are repeated with a new round of questionnaires until convergence is achieved for all evaluation factors.

Step 5: Test for consensus among expert opinions: Once convergence among expert opinions has been achieved, the decision-maker determines the threshold values based on the research objectives and decides whether any particular evaluation factor should be eliminated.

3.3. Analytic Hierarchy Process (AHP)

AHP gathers evaluations from relevant decision-makers and assesses the advantages and disadvantages of each project as follows [31]:

Step 1: Establish the hierarchical relationship: First, a hierarchical framework is established for a complex evaluation problem. The first level defines the goal to be achieved. The second level shows approaches to attain the goal, while the third level specifies the criteria used to evaluate the degree the goal is accomplished.

Step 2: Construct the pairwise comparison matrix: After the hierarchical framework is established, pairwise comparisons are made. For a hierarchy with n components, a total of n (n − 1)/2 numbers of pairwise comparisons need to be conducted. These comparisons are usually performed using a 9-point scale rating system, with values ranging from 1 to 9, reflecting increasing levels of importance.

Step 3: Calculate the weights and verify consistency: After completing the pairwise comparisons, the eigenvalue method is adopted to calculate the eigenvectors, which represent the weight (w) of each factor in the hierarchy. In some cases, discrepancies may arise in the evaluation of the relative importance levels provided by experts. To ensure the reliability of the survey data, the consistency of the pairwise comparison matrix is usually tested using the following Equation:

$$CI={\displaystyle \frac{{\lambda }_{max}-n}{n-1}}$$

To assess the consistency of the evaluation results, the consistency ratio (CR) is employed as a criterion for determining the consistency of the pairwise comparison matrix. The CR is defined as follows:

$$CR={\displaystyle \frac{CI}{RI}}$$

The RI value represents the random index (RI), and according to Saaty (1987), when n is 1 to 10, the value of RI will be between 0.00 and 1.49 [30].

If $\mathrm{C}.\mathrm{R}.\le 0.1$, the evaluation values in the pairwise comparison matrix are considered acceptable, while $\mathrm{C}.\mathrm{R}.\ge 0.1$ indicates that the evaluation values in the pairwise comparison matrix exceeds the allowable margin. In such cases, the evaluation should be revisited, or the data should be discarded.

Step 4: Calculate the overall weight of each project: After passing the consistency test, the overall weight of each project is calculated and ranked by its overall weight to identify the optimal choice.

The hierarchical aggregation of weights was applied to combine criteria across dimensions, following the multiplicative synthesis rule common in AHP applications [30]. This method preserves the relative importance within and across evaluation layers, enabling consistent priority computation throughout the decision hierarchy.

4. Case Study

4.1. Establishment of the Evaluation Model

In the initial stage of this study, a total of 17 expert questionnaires were distributed using FDM. These questionnaires were sent to experts across various fields, including 5 from the public sectors, 1 from a legal-entity research institutes, 6 from the green energy industries, and 5 academic experts specializing in industrial research. The background information of these experts, representing relevant fields within government, industry and academia, is summarized in

Table 3. Background information of expert survey respondents.

Number	Field of Expertise	Organization
1	Public Sector	Industrial Park Management Unit, Economic Development Department, Hsinchu County Government
2		Industrial Development Administration, Ministry of Economic Affairs
3		Energy Administration, Ministry of Economic Affairs
4		Bureau of Urban Development, Tainan City Government
5		Southern Taiwan Science Park Bureau, National Science and Technology Council
6	Foundation	Industry Service Center, Industrial Technology Research Institute
7	Industry	Y. KU Planning Intl. Co.
8		Sinotech Engineering Consultants, Ltd.
9		Kai Yuan KYC Co., Ltd.
10		Sheng Yang Engineering Consultants Ltd.
11		Ri Xi Energy Ltd.
12		WINCHAIN CONSULTANTS CO., LTD.
13	Academic Community	Department of Land Economics, National Cheng Chi University
14		Department of Real Estate Management, National Pingtung University
15		Program in Interdisciplinary Studies, National Sun Yat-sen University
16		Local Creative Development Center, Southern Taiwan University of Science and Technology
17		Maste’sr Degree Program for Intelligent City and Community Planning, Chung Hua University

. The response rate for the questionnaires was 100%.

Table 4. Results of Fuzzy Delphi Method analysis.

		The Most Conservative Triangular Fuzzy Function			The Most Optimistic Triangular Fuzzy Function			Test Value	Expert Consensus Value	Optimal Value
		li	ui	mi	Li	Ui	Mi	Zi	Hi	Min	Max	GM
Physical Location Conditions	Population Density	1	7	3.02	2	9	5.95	−2.07	4.49	2	8	4.62
	Distance from Residential Areas	2	7	4.33	6	9	7.63	2.30	6.38	4	8	6.06
	Proximity to Road Networks	1	7	3.27	3	9	6.35	−0.93	4.89	2	8	4.98
	Proximity to Power Grid	2	8	5.17	7	9	8.59	2.42	7.36	3	9	6.78
	Distance from Environmentally Sensitive Areas	3	8	5.57	6	9	8.57	1.01	7.03	5	8	7.08
Environmental and Climate Conditions	Air Temperature	1	6	2.97	3	9	6.07	0.10	4.51	2	7	4.47
	Relative Humidity	1	6	2.37	2	8	5.09	−1.28	3.84	2	7	3.98
	Solar Radiation	4	8	5.66	6	9	8.29	0.63	6.99	3	8	6.60
	Sunlight Duration	5	7	6.14	9	9	9.00	4.86	9.00	6	9	7.60
	Wind Speed	1	5	2.18	2	9	4.82	−0.36	3.50	2	7	3.69
	Elevation	1	6	2.93	3	9	5.81	−0.13	4.43	2	7	4.44
	Slope	1	7	3.17	3	9	6.14	−1.02	4.80	2	8	4.75
	Orientation	1	8	4.12	4	9	7.12	−0.99	5.78	2	9	5.60
Political, Economic, and Social Conditions	Land Use	3	7	5.14	6	9	7.97	1.82	6.51	3	8	6.24
	Low agricultural productivity	3	8	4.71	5	9	7.72	0.01	6.36	2	8	5.85
	Visual Impact	1	6	3.49	4	9	6.60	1.11	5.02	3	8	5.15
	Land Cover	1	6	3.49	4	9	6.07	0.58	4.90	3	8	4.94
	Scale	3	7	5.14	5	9	7.77	0.63	6.20	2	8	6.06
	Land Cost	3	7	5.43	7	9	8.51	3.08	7.00	4	8	6.61

Note: Grey areas did not pass the threshold of expert consensus value.

indicates the results of the expert questionnaires analyzed by FDM. The threshold of determining whether evaluation factors are retained or excluded is based on the expert consensus values (Hi), which are decided in various ways. The selection of methods directly impacts the final evaluation results. While many studies employ the same research method based on subjective decisions made by the decision-maker, this study seeks to prevent excessively limiting the final evaluation factors. Additionally, experts encountered difficulties in reaching a consensus on some factors even after several rounds of questionnaires. Some experts also expressed concerns about the excessive complexity of certain factors. Consequently, this study uses a threshold based on the arithmetic mean of the Hi values. The average consensus value across the 19 evaluation factors is 5.74, with 9 factors failing to meet the threshold due to the concerns about their complexity. Considering the concept of FDM, the calculation of average consensus value, and the impact on the research timeline due to the complexity of certain factors, 10 evaluation factors were retained. For the results of expert questionnaires, when consensus was not reached for specific factors, multiple rounds of questionnaires were distributed until consensus was achieved. In addition, due to the small sample size, test values for some factors may exhibit negative values or inconsistencies. The arithmetic means of the Hi values has been adopted in line with previous applications of FDM for expert consensus filtering. Nonetheless, we acknowledge that this approach may not account for sampling variability. Future studies may consider Modified Fuzzy Delphi Models with bootstrap resampling to increase statistical robustness or implement a two-round Delphi process to enhance stability.

Among the three primary perspectives, environmental and climate conditions account for the highest number of exclusions, with five factors eliminated. Some changes in the evaluation framework for the deployment of ground-mounted PV systems were also made. Specifically, regarding the factor “distance from environmentally sensitive areas,” and in consideration with the classification of environmentally sensitive areas, prohibition of development in level one areas is unequivocal. However, areas classified as level 2, such as groundwater recharge zones and flood-prone areas, may be suitable for deploying PV systems. Additionally, due to varying governing authorities for different environmentally sensitive areas, the availability and accessibility of relevant information and data are inconsistent, complicating accurate assessment of the actual impacts. Currently, relatively complete information is only available in the “Query Platform for Environmentally Sensitive Areas.” Hence, this study redefines the evaluation factor “distance from environmentally sensitive areas” to a stricter definition: “distance from level one environmentally sensitive areas.” The adjusted framework is presented in Figure 2.

In the second stage, expert questionnaires employing the AHP were distributed to the same target group as for the FDM questionnaires. A total of 17 responses were collected, a 100% response rate.

To verify whether the questionnaire results can be adopted, the threshold of consistency test for AHP is suggested to be 0.1. Consequently, only factors with C.R. values below 0.1 were included in the weight calculation for this study. The results of weight calculation are summarized in

Table 5. Results of factor weight calculations.

Perspective			Evaluation Factor
Item	Local Weight (NW1)	Ranking	Item	Local Weight (NW2)	Total Weight (NW1*NW2)	Ranking
Physical location conditions	0.419	1	Distance from residential areas	0.282	0.118	4
			Proximity to power grid	0.298	0.125	3
			Distance from level 1 environmentally sensitive areas	0.421	0.176	1
Environmental and climate conditions	0.380	2	Solar radiation	0.277	0.105	6
			Sunlight duration	0.427	0.162	2
			Orientation	0.296	0.113	5
Political, economic, and social conditions	0.201	3	Land use	0.245	0.049	8
			Low agricultural productivity	0.342	0.069	7
			Scale	0.180	0.036	10
			Land cost	0.231	0.047	9

. The local weight represents the weight of each evaluation factor within a specific level, while the total weight is derived from the calculation across the entire hierarchy. The final ranking is based on the total weight.

Figure 3 compares the weights for each factor. Among the three primary perspectives, physical location conditions have the highest weight, followed by the environmental and climate conditions, while political, economic, and social conditions are assigned the lowest weight. Notably, based on Mohtasham’s study [12], not all societies are suitable for the development of renewable energy as this depends on the availability of natural resources, whose distribution is closely related to the inherent geographical conditions. According to the weight obtained from this study, the expert group appears to prioritize physical location conditions, likely due to the relatively small land area of Taiwan. Thus, physical location conditions are considered to more accurately reflect the advantages and disadvantages of specific locations. The low ranking of the factor “political, economic, and social conditions” may be attributed to the lack of clear explanations for certain evaluation factors. For example, the “land use” factor did not clearly indicate the types of land suitable for ground-mounted PV systems. Moreover, the “scale” factor did not define whether large- or small-scale systems are most appropriate. These unclear instructions likely led to challenges for experts in making decisions, and further affected the overall ranking.

The top four factors in overall weight ranking among the ten evaluation factors are “distance from level one environmentally sensitive areas,” “sunlight duration,” “proximity to power grid,” and “distance from residential areas.” Based on the survey results, there is concern regarding environmental impacts of PV systems, particularly from environmental organizations and residents, as Taiwan actively develops renewable energy and installs ground-mounted PV systems. This concern is reflected in the top ranking of the factor “distance from level one environmentally sensitive areas.” “Sunlight duration,” which directly affects power generation efficiency, is recognized as a significant environmental condition. Given that power generated from solar plants must be connected to the Taiwan Power Company, the factor “proximity to power grid” also ranks high. Last, residents typically have low acceptance of new land uses, and there is a not-in-my-backyard effect in Taiwanese society for the establishment of power generation facilities. The factor “distance from residential areas” indicates the general public’s attitude towards the establishment of power plants.

4.2. Existing PV Sites in Tainan City

To gain a deeper understanding of the critical factors influencing site selection conditions for the deployment of PV systems, this study further discusses the existing ground-mounted PV sites in Tainan City, the county with the highest solar photovoltaic capacity in Taiwan. Unlike typical business locations, most ground-mounted PV sites do not have a street address. This study utilizes the location data and relevant information of approved PV sites sourced from Energy Administration, Ministry of Economic Affairs. The dataset includes records such as names of operators, solar plant names, approval dates, counties and cities, districts or townships, land sections, and land serial numbers. After 2020, two additional fields, including land area and installation capacity, were included in the dataset. These data have been adopted for analysis in the current study.

Figure 4 is a statistical chart illustrating the number of PV projects with setup permits issued from 2017 to 2021. The chart reveals a continuous increase in the number of PV system installation applications both nationwide and Tainan City, with a peak observed in 2020. However, the number of applications declined in 2021. This study extrapolates that the decrease of application may be associated with the policy change in 2020, in which the Ministry of Agriculture revised the regulations concerning PV systems installed in agricultural land. The revision resulted in a backlash from solar energy developers, slowing down the approval rate in PV installations. Despite this, the number of applications in Tainan City, compared to the national figures, highlights its prominence as a preferred site for PV installations.

To investigate the spatial distribution of PV sites in Tainan City, this study integrates the “land serial numbers of the approved PV sites” with the “property data of land numbers belonging to Tainan in 2020,” presenting PV site distribution through the conversion of the latitude and longitude. Figure 5 shows the distribution of ground-mounted PV sites across Tainan City, with the highest density observed in Qigu and ShanShang Districts. The continuous increase in new PV projects in Qigu district reveals its popularity for site selection. PV installations, extending northwards from Qigu District to Jiangjun and Beimen Districts, have formed a corridor of ground-mounted PV sites. In contrast, the increase in new PV projects in Shanshang District mainly occurred from 2020 to 2021. Overall, ground-mounted PV sites are typically located in the southeastern region of Tainan [36].

To enable a comprehensive discussion, this study focuses on regions with highest density of PV installations from 2017 to 2021, considering the geographic conditions of Tainan. A total of 6 ground-mounted PV sites and 1 fishery-solar power system from east to west of the city are selected as research subjects. The evaluation framework for site selection of ground-mounted PV systems developed in this study is applied to evaluate current site selections in Tainan City. Detailed information of the factors is summarized in

Table 6. Summary table of discussion subjects.

Site	Year of Application	Power Company Name	Power Plant Name	District	Section	Land Area (m2)
Ground-mounted PV sites
1	2017	Yu Ting Electric Co., Ltd.	Xinying Landfill No. 1 and No. 2 Solar Power Plants	Xinying District	Hou Chen Section	75,760
2	2018	SIN JHONG SOLAR POWER CO., LTD.	Sin Jhong Xuejia Type 1 Solar Power Plant	Xuejia District	Xuejialiao Section Chai Tzu Section	280,211
3	2019	Tien Chin Energy Co., Ltd.	Tien Chin Phase 1 Solar Power Plant	Qigu District	Qigu Section Yucheng Section	100,680
4	2019	CHIMEI Green Energy Co., Ltd.	Chimei Green Energy Phase 2 Shanhua Ground-Mounted Solar Power Plant	Shanhua District	Chiehpa Section 2nd Subsection	158,717
5	2020	Hao Yang Electric Co., Ltd.	Hao Yang No. 2 Solar Power Plant	Beimen District	Xidiliao Section Erchonggang Subsection	17,008
6	2021	Shuo Ming No. 1 Co., Ltd.	Shan Shang Solar Power Plant	Shanshang District	Fengte Section	163,199
Fishery-solar power system
7	2021	Hsueh Tsan Energy Co., Ltd.	Tainan City Xuejia PV Zone Development Project (Phase 1) (Fishery-Solar Hybrid Demonstration Zone)	Xuejia District	Xuejialiao Section	180,937

. Of the selected sites, most are completed, with the exception of site 6, Shan Shang Solar Power Plant, which remains under construction.

To evaluate the distinct inherent advantages of each case, this study develops a scoring system for each factor to objectively evaluate different cases. A value is calculated for each study case to verify which case best aligns with the evaluation framework developed in this study. Based on the results displayed in the following two tables, the actual values derived from each case are employed to calculate the values of the evaluation factors within the category of physical location conditions. The environmental and climate conditions category evaluation factors are all scored using a rating scale for calculations. With the exception of land cost, for which actual values are adopted, other factors within the category of political, economic, and social conditions are also scored by a rating scale, followed by calculations. The ratings and the corresponding calculated results for each case are outlined in

Table 7. Evaluation factor calculations.

Perspective	Evaluation Factors	Explanation for Calculation Values
Physical Location Conditions	Distance from Residential Areas	Total number of households within a 500-m distance around the site.
	Proximity to Power Grid	Distance (km) from the site to the nearest electrical substation.
	Distance from Environmental Sensitive Areas	The total number of “level one environmentally sensitive items” to be examined within the area where the site is located.
Environmental and Climate Conditions	Solar Radiation	Classifying the daily solar radiation of Tainan City into two levels.	1	The one with the best GHI value.
			2	The one with the second-best GHI value.
	Sunlight Duration	The conditions for all sites are identical, so the value is 1 for all.
	1 Orientation	Classifying orientations into three levels.	1	Facing south
			2	Facing east, west, and flat regions
			3	Facing north
Political, Economic, and Social Conditions	Land Use	Classifying land use into three levels.	1	The installation of PV systems does not affect the original land use.
			2	The installation of PV systems coexists with the original land use.
			3	The installation of PV systems replaces the original land use.
	Low Agricultural Productivity	Classifying the agricultural productivity into two levels.	1	Without agricultural use.
			2	With agricultural use.
	Scale	Classifying the areas of PV systems into four levels.	1	Land area ranging from 6 to 15 hectares.
			2	Land area ranging from 1 to 5 hectares and from 16 to 20 hectares.
			3	Land area ranging from 21 to 25 hectares.
			4	Land area greater than 26 hectares.
	Land Cost	The unit price of each site. (priced in New Taiwan dollars, NTD)

¹ The land slope is not always oriented in a single direction. For such land, companies measure the land areas with different orientations and the corresponding sunlight duration. The “Orientation” factor is defined as the main direction with the largest land area or the longest duration of sunlight exposure.

.

4.3. Empirical Results and Discussion

According to the weight calculation described in Table 5 and the use of values outlined in Table 7, smaller values derived from

Table 8. Summary of calculation results for evaluation factors of each study case.

Perspective	Evaluation Factor	Yu Ting (Xinying)	Sin Jhong (Xuejia)	Tien Chin (Qigu)	Chimei (Shanhua)	Hao Yang (Beimen)	Shuo Ming (ShanShang)	Hsueh Tsan Fishery-Solar Power (Xuejia)
Physical Location Conditions	Distance from Residential Areas	28	15	204	9	33	37	12
	Proximity to Power Grid	3.749	10.674	14.969	6.586	18.713	1.129	9.649
	Distance from Level 1 Environmentally Sensitive Areas	6	6	7	5	8	7	6
Environmental and Climate Conditions	Solar Radiation	2	1	1	2	1	2	1
	Sunlight Duration	Due to the inability to obtain the solar radiation hours for individual sites, it is assumed that all installations located in Tainan City receive the same amount of solar radiation hours.
	Orientation	1	2	1	1	3	2	2
Political, Economic, and Social Conditions	Land Use	1	3	3	1	3	3	2
	Low Agricultural Productivity	1	2	1	2	2	1	2
	Scale	1	4	1	1	2	2	2
	Land Cost	640	560	773	9200	570	898	520

are considered more favorable. Based on the calculated results, Yu Ting (Xinying) case, with a value of 0.108, best aligns with the evaluation framework of this study. The cases rank second to fifth are Hsueh Tsan Fishery-Solar Power (Xuejia), Shuo Ming (ShanShang), Sin Jhong (Xuejia), and Chimei (Shanhua), with calculated results of 0.125, 0.130, 0.135, and 0.140, respectively. The values for these four cases are relatively close. Hao Yang (Beimen) case ranks sixth with a value of 0.170, followed by Tien Chin (Qigu) case, which has a value of 0.191. As the ranking decreases, the suitability of the case for this study framework also decreases. However, no significant differences in the values among these cases are observed, demonstrating that the evaluation framework can accurately reflect the advantages of each case in the real-world contexts.

Additionally, the government has actively promoted and encouraged the adoption of ground-mounted PV systems by integrating agriculture, fisheries, and livestock industries to achieve hybrid power generation in recent years. Consequently, this study selects a fishery-solar power case to examine using the established evaluation framework. This case demonstrates very good performance, ranking second out of seven cases. However, several controversial issues have been raised, including ecological concerns and land-use challenges. Not all aquaculture species are allowed to employ the fishery-solar power systems. From the perspective of the evaluation framework developed in this study, integration of existing land use with PV systems appears to be a feasible strategy for site selection. Nevertheless, when considering the balance between existing land use and the implementation of PV systems, this evaluation framework does not provide a comprehensive analysis and suggestions for fishery-solar power systems.

5. Discussion, Conclusions and Suggestions

The utilization of renewable energy is a critical factor in achieving the target of 2050 net-zero carbon emission. Though PV systems have experienced rapid development in Taiwan recently, there have been numerous debates regarding environmental concerns. In response, this study employs expert questionnaires with AHP to evaluate the spatial development and further identify conditions that are most important for deploying ground-mounted PV systems in Taiwan. Case studies are then conducted for selected existing PV sites in Tainan City. Among all factors included in the evaluation framework in this study, physical location is considered the most significant by the expert group. The factor of distance from level one environmentally sensitive areas is weighted the highest among the ten factors, followed by sunlight duration, proximity to power grid, and distance from residential areas. This result contrasts with the findings in international literature which emphasize the significance of inherent geographic conditions for the development of renewable energy, but it aligns with Taiwan’s specific characteristics, such as limited land area and high population density. Additionally, the AHP-based calculations of each case’s advantages reveal that Project 7 Yu Ting (Xinying) case best aligns with the evaluation model developed in this study, while Tien Chin (Qigu) case deviates the most from the framework.

The findings suggest that future PV site selections consider the advantages of different cases, particularly the integration of PV systems with existing land uses. However, for hybrid power plants combining PV systems with agricultural, fishery, and livestock industries, further in-depth research and assessment are necessary. Furthermore, establishing designated parks for renewable energy, where related projects can be developed in a concentrated area, may be a viable solution. Furthermore, to mitigate potential controversies related to ground-mounted site selections, the government can guide developers by providing a comprehensive database to identify lands suitable for PV installations. According to this research, areas that are unfavorable for PV installations typically have agricultural production or significant land-use restrictions, such as areas dedicated for forestry. These areas generally receive lower rankings in the evaluation process. Therefore, achieving symbiosis between the natural environment and renewable energy development can be facilitated by focusing on sites with favorable conditions for the deployment of PV systems.

This study identifies the most suitable evaluation factors for PV site selections primarily through expert questionnaires, while also incorporating the characteristics of existing ground-mounted PV sites in Tainan City to obtain a comprehensive understanding of the conditions most valued in current site selection practice in Taiwan. A literature review and evaluation by energy firms are used to determine the evaluation factors. This should provide a different research perspective compared to previous solar photovoltaic research, which mostly focused on single-level factors or comprehensively explored the photovoltaic development of the entire country. Future studies may introduce different evaluation factors tailored by the specific areas of interest or focus. Furthermore, applying spatial information from Tainan City or other counties and cities into GIS software at varying administrative levels, such as towns, cities, districts, and villages, may offer a valuable approach for identifying appropriate areas for PV installations. Calculating the potential land area and installation capacity for ground-mounted PV systems also helps achieve long-term capacity goals. Evaluation factors are classified into three main categories, including physical location conditions, environmental and climate conditions, political, economic, and social conditions. However, the factors need clearer definition within the expert questionnaires, which is a limitation of this study. Future studies can propose more suitable categories and offer detailed explanations to improve the clarity and applicability of the evaluation framework.

Due to the short duration of this study, it has the following limitations and related future research suggestions. While it did not perform a full sensitivity analysis, future research may incorporate such techniques to test the robustness of weight variations on site rankings. Moreover, visualizing suitability through spatial mapping (e.g., GIS-based overlay) can further support the interpretation and policy translation of site selection results. This research advances current knowledge on solar PV planning by integrating fuzzy Delphi and AHP approaches to evaluate site suitability, and by highlighting the divergence between theoretical assessments and private-sector decision-making. It underscores the need for planning models that not only reflect environmental and technical suitability but also accommodate institutional and economic realities. These findings are particularly relevant for improving spatial strategies in sustainable energy transitions. The methodological framework adopted in this study can be adapted to other regional or national contexts, provided that the evaluation criteria are localized through context-sensitive expert engagement. While the FDM–AHP process offers a flexible structure, the origin and domain knowledge of the experts significantly affect the reliability of results. Applying this model in other countries would require revalidation of indicator weights, ideally by convening a new panel with relevant regional experience. The expert panel in our study consisted of 17 respondents from the public sector, industry, and academia. While this sample size aligns with common FDM practices, it falls below the recommended size for complex multi-criteria problems. Future applications may enhance statistical validity through power analysis–based sample determination and improved balance across stakeholder groups to mitigate potential biases. In addition, regarding some AHP calculation problems mentioned in the Case Study chapter, future research will also consider the relevant design more carefully or use other methods such as bootstrap resampling [37]. Another limitation of the current study is the absence of a formal sensitivity or uncertainty analysis to test the robustness of site rankings under varying weights. Such analysis could improve confidence in the model’s predictive reliability and highlight the most influential factors. This remains an important direction for future research.