In this paper, we analyze relevant literature to construct a conceptual framework of the barriers to the development of solar energy in Taiwan (Figure 1). This framework includes 3 dimensions and 13 criteria for the development of solar energy (Table 1). We applied the Decision Making Trial and Evaluation Laboratory (DEMATEL) [1] integrated with the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) [2] to determine the cause—effect relationships among these dimensions and criteria. Our findings serve as a reference for decision-makers regarding green energy policy.

Sorenson [3] defines renewable energy as energy replenishment from natural processes is as fast as consumption. Renewable energy sources such as solar, wind, geothermal, hydropower, and biomass energies feature different energy densities, location requirements, and physical processes [4]. In particular, solar energy is an increasingly popular choice for power generation across the world [4], and it is this renewable source that is favored to fill the electricity gap in Taiwan [3].

There exists a large body of research into solar energy, including its impacts and requirements. [5] recommended that the photovoltaic industry should act to maintain the environmental friendliness of solar energy as a long-term environmental strategy, which could be achieved through the establishment of effective recycling policies. [6] pointed out that the use of conventional energy sources contributes to the sustainable development of human activities. However, their large-scale deployment inevitably exerts adverse environmental impacts, such as greenhouse gas emissions as well as soil and water pollution. [7] acknowledged that the benefits of electricity supply (including that generated from renewable resources) must be weighed against health costs. For example, the cadmium cells used in solar panels represent a well-researched occupational hazard. However, given the estimated 30-year life cycle of photovoltaic units, the risk seems acceptable. [4] investigated the legislation and policy development encouraging research and development in the photovoltaic industry. [8] considered the solar energy industry as a whole. [9] declared that the end-of-life management of obsolete photovoltaic modules should avoid landfills, emphasizing that the cost of the recovery process is a key variable of recycling. [10] discussed obstacles to the development of solar energy with an eye to solutions for both research and practice. [11] pointed out that although many solar panel recycling-related activities and processes have been initiated around the world, the majority of solar panels will be abandoned in about 25 years. Therefore, the question arises as to whether the current solution to our energy problem is leading to further environmental heritage waste.

Many researchers have examined the challenges specific to individual countries. [12] considered the dilemma faced by India, which is an emerging economy with a severe shortage of electricity. They proposed that the barriers to solar energy installations are comprised of 6 dimensions and 13 subdimensions [13] considered the same problem but identified 7 dimensions and 28 subdimensions. [14] presented an overview of the challenges faced by Barbados in the deployment of alternative energy. [15] identified and ranked obstacles in Nepal to renewable energy use. [16] pointed out that while Pakistan has a large amount of renewable and sustainable energy resources, it lacks the capacity to use them effectively; indeed, [17] supported this with a ranking of key barriers. [18] further confirmed their findings with their evidence supporting 7 main barriers and 29 sub-barriers. Then, research strategies to overcome these barriers. [19] explored the interrelationships among factors affecting renewable energy in China. [20] assessed Ghana’s renewable energy policy goals and built a framework to evaluate each obstacle to these goals. [21] outlined the potential driving force of solar power generation in Vietnam, obstacles to further use of solar energy, and suitable strategies for the next phase of development. In this paper, we draw on this considerable body of research to integrate Taiwan’s experience in solar energy.

Each of the dimensions is comprised of subdimensions which represent the factors influencing solar energy development. There are four economic and financial factors, four political and regulatory factors, and five geographical and ecosystem factors. Details are provided in Table 1. We consider each of these factors in turn.

1) Economic and financial barriers (D₁)

Economic and financial aspects are important for every country. We found four specific financial factors were commonly cited in the literature.

● High initial capital cost (C₁₁)

Manufacturing and installing solar systems entail high investment costs. For example, a silicon solar panel system costs around US$750 per square meter to install [26]. For developers without sufficient capital, this can represent an insurmountable obstacle. This factor was considered by [4] [10] [12] [14] [15] [16] [18] [20] [21] [25].

● Long return on investment (C₁₂)

Solar systems are expensive to set up, and they only generate a moderate amount of power on a daily basis. Over time, the return on investment is significant. However, it takes about 20 to 30 years [27] for a solar panel system to manifest some level of profit. This point was made by [8] [12] [13] [14] [18].

● Lack of bank credit and loan mechanisms (C₁₃)

Solar energy projects are usually funded by local banks, but they have limited financing capacity. The inability to obtain credit can create difficulties in the financing of new projects [4] [8] [12] [15] [16] [18] [19] [21] [25].

● Lack of government incentive and subsidy policy (C₁₄)

Lack of or limited government incentives and subsidy policies, insufficient regulatory framework, and lack of coherent renewable energy policies create obstacles to solar power generation [4] [8] [12] [13] [14] [15] [16] [18] [19] [20] [21] [25]. In many cases the subsidy policy and related laws and regulations are in place, but there is no substantive implementation of these frameworks.

2) Political and regulatory barriers (D₂)

We identified four factors within the dimension of political and regulatory barriers.

● Lack of waste disposal regulations (C₂₁)

A lack of policies and regulations regarding global solar waste disposal remains an obstacle to solar power generation [11]. Furthermore, once these policies are in place, they need to be adapted to local conditions.

● Lack of waste recycling systems (C₂₂)

There are few waste recycling systems that enable companies to sort, recycle, and reuse solar panel waste. In particular, the safe disposal of batteries and CdTe solar panels is important and requires a reliable waste management system. Proper supervision of the recycling of solar panels is imperative, as pollution is generated during the recycling process, and the energy required for collection, transportation, and recycling must be quantified according to local conditions. The above points were made by [4] [5] [6] [9] [10] [11].

● Political turbulence caused by party rotation (C₂₃)

Unstable politics and government intervention in the domestic market are the main obstacles to renewable energy. This political factor includes corruption, nepotism, and favoritism. In addition, party rotation can lead to incoherence in policies and regulations, resulting in confusion. This factor was considered by [8] [12] [15] [16] [18] [25].

● Lack of political commitment (C₂₄)

There is a shortage of political commitment to solar power generation. This is aggravated by corruption and nepotism. It adversely affects the planning process through lengthy regulatory approval and permit procedures, and through weak and incoherent policy implementation. Renewable energy development would also benefit from the creation of renewable energy zones, and this has yet to be implemented. The above was reported in [12] [14] [15] [16] [19] [20] [21] [25].

3) Geographical and ecosystem barriers (D₃)

Geographical and ecosystem barriers are unique to the landscape and climate of each country, but the nature of renewable resources means all are affected to some degree by this dimension. We discuss four important factors.

● Air pollution (C₃₁)

Although the solar energy industry has no direct impact on the environment, solar cells are toxic to the environment. Discharging them is a complex task for manufacturers and consumers, bringing in considerations of environmental protection [6] [10] [19].

● Water pollution (C₃₂)

Environmental impacts related to solar energy include land use and habitat loss, water pollution, and the use of hazardous materials in manufacturing processes [4]. An in-depth study of the potential pollution of a lagoon in Vietnam was conducted by [21]. This factor has also been considered by [4] [10] [19] [25].

● Land pollution (C₃₃)

Land use and loss of biological habitat are potential environmental impacts of solar power [4]. In many cases, ecosystems, plants, and fauna are destroyed [18]. Furthermore, substances dissolved by heat from the sun seep into the soil [8]. Solar cells are usually made of various chemicals that are toxic to the environment and discharging them is a challenging social responsibility for manufacturers and consumers [17]. The factor of land pollution was considered by [6] [10] [19] [25].

● Human health (C₃₄)

Cadmium batteries are a known occupational hazard [11]. The dangers associated with the transfer of heated fluids (i.e., water and oil) are also a concern [18]. These health hazards were also discussed by [6] [7].

● Different geographical and topographical influences result in different sunlight exposure times (C₃₅)

The performance of solar panels is greatly affected by the intensity of sunlight [17] and solar energy is often intermittent, with limited daytime hours. The geographical distribution of solar energy resources is also uneven [8]. Pakistan, for example, is an arid tableland surrounded by dry mountains [4]. The effects of this landscape on the scope of solar power generation are further impacted by scattered households [15] [20] [28]. In the work of [8] [10] [18] [25], this factor was discussed in detail.

According to Table 1, we constructed a conceptual framework express as Figure 1.

DEMATEL was applied to create a visual model of our conceptual framework and to study the interrelations among its components. This approach was proposed by the Battelle Memorial Institute of the Geneva Research Center between 1972 and 1976 [1]. Its aim is to identify the factors that exert the greatest influence on a selected phenomenon. We conducted DEMATEL and integrated our results using TOPSIS to obtain findings which we present in a traditional narrative-type format. Our research flow is presented in Figure 4.

DEMATEL consists of 10 stages: 1) collecting the results of the questionnaire to

construct matrix D; 2) establishing a direct relationship matrix to calculate initial matrix D; 3) obtaining normalization matrix N; 4) assigning identify matrix I; 5) expressing extension matrix I − N; 6) computing inverse matrix [ I − N ] − 1 ; 7) calculating total influence relationship matrix T = N [ I − N ] − 1 ; 8) calculating d, r, d + r, and d − r (defined below); 9) drawing a causal diagram; and 10) determining key criteria.

We describe these steps using mathematical notation as follows:

Step 1) Set the number of elements (criteria) n. Each criteria is evaluated on a five-point scale as follows: (0) no impact, (1) low impact, (2) medium impact, (3) high impact, and (4) extremely high impact.

Step 2) Establish a direct relationship matrix to compute initial matrix D, then i is notation column elements and j is notation row elements, using degree of interaction to obtain matrix = [ d i j ] , where d i j represents the degree of effect on the ith criteria. When the elements ofi have a direct effect on the elements of j, then d i j ≠ 0 and inverse [ d i j ] = 0 .

[ d i j ] = [ d 11 ⋯ d 1 j ⋯ d 1 n ⋯ ⋯ ⋯ d i 1 ⋯ d i j ⋯ d i n ⋯ ⋯ ⋯ d n 1 ⋯ d n j ⋯ d n n ] (1)

Step 3) Obtain normalization matrix N:

N = [ n i j ] = [ d 11 / d 11 ⋯ d 1 j / d j j ⋯ d 1 n / d n n ⋯ ⋯ ⋯ d i 1 / d 11 ⋯ d i j / d j j ⋯ d i n / d n n ⋯ ⋯ ⋯ d n 1 / d 11 ⋯ d n j / d j j ⋯ d n n / d n n ] (2)

Step 4) Assign identify matrix I:

I = [ 1 ⋯ 0 ⋯ 0 ⋯ 1 ⋯ 0 ⋯ 0 ⋯ 1 ⋯ 0 ⋯ 0 ⋯ 1 ⋯ 0 ⋯ 0 ⋯ 1 ] (3)

Step 5) Express extension matrix I − N:

I − N = [ 1 − n 11 ⋯ − n 1 j ⋯ − n 1 n ⋯ ⋯ ⋯ − n i 1 ⋯ 1 − n i j ⋯ − n i n ⋯ ⋯ ⋯ − n n 1 ⋯ − n n j ⋯ 1 − n n n ] (4)

Step 6) Compute inverse matrix ( I − N ) − 1 :

( I − N ) − 1 = [ ( 1 − n 11 ) − 1 ⋯ ( − n 1 j ) − 1 ⋯ ( − n 1 n ) − 1 ⋯ ⋯ ⋯ ( − n i 1 ) − 1 ⋯ ( 1 − n i j ) − 1 ⋯ ( − n i n ) − 1 ⋯ ⋯ ⋯ ( − n n 1 ) − 1 ⋯ ( − n n j ) − 1 ⋯ ( 1 − n n n ) − 1 ] (5)

Step 7) Compute total relationship matrix T = N ( I − N ) − 1 :

T = [ ( d 11 d 11 ) ( 1 − n 11 ) − 1 ⋯ ( d 1 j d j j ) ( − n 1 j ) − 1 ⋯ ( d 1 n d n n ) ( − n 1 n ) − 1 ⋯ ⋯ ⋯ ( d i 1 d 11 ) ( − n i 1 ) − 1 ⋯ ( d i j d j j ) ( 1 − n i j ) − 1 ⋯ ( d i n d n n ) ( − n i n ) − 1 ⋯ ⋯ ⋯ ( d n 1 d 11 ) ( − n n 1 ) − 1 ⋯ ( d n j d j j ) ( − n n j ) − 1 ⋯ ( d n n d n n ) ( 1 − n n n ) − 1 ] (6)

Step 8) Calculate d, r, d + r, and d − r.

Step 9) Draw a causal diagram by mapping the state of (d + r, d − r).

Step 10) Determine key criteria.

TOPSIS was developed in 1981 [2]. It proceeds as follows:

Step 1) Normalize ratings:

r i j = x i j ∑ i = 1 m x i j 2 ,     i = 1 , ⋯ , m ;   j = 1 , ⋯ , n . (7)

Step 2) Weight normalized ratings:

υ i j = w j r i j ,     i = 1 , ⋯ , m ;   j = 1 , ⋯ , n , (8)

where w j is the weight of the jth attribute.

Step 3) Evaluate the positive and negative solutions (i.e., “*” and “−”):

A * = { υ 1 * , υ 2 * , ⋯ , υ j * , ⋯ , υ n * } = { ( max i υ i j | j ∈ J 1 ) , ( min i υ i j | j ∈ J 2 ) | i = 1 , ⋯ , m } (9)

A − = { υ 1 − , υ 2 − , ⋯ , υ j − , ⋯ , υ n − } = { ( min i υ i j | j ∈ J 1 ) , ( max i υ i j | j ∈ J 2 ) | i = 1 , ⋯ , m } (10)

where J 1 denotes benefit attributes and J 2 denotes cost attributes.

Step 4) Obtain separation measures:

The positive solutions are ranked according to

S i * = ∑ j = 1 n ( v i j − v j * ) 2 ,     i = 1 , ⋯ , m . (11)

Similarly, the negative solutions are ranked according to

S i − = ∑ j = 1 n ( v i j − v j − ) 2 ,     i = 1 , ⋯ , m . (12)

Step 5) Find similarities to positive ideal solution:

C i * = S i − S i * + S i − ,     i = 1 , ⋯ , m . (13)

Note that 0 ≤ C i * ≤ 1 , where C i * = 0 when A i = A − and C i * = 0 when A i = A * .

Step 6) Rank preference order:

Select the alternatives with the max C i * , or rank the criteria based on C i * in descending order.

Based on the results of DEMATEL, we set the weights for TOPSIS as follows: 0.2, 0.3, and 0.5. We then applied the steps described above.

Threshold value: 0.152.

Threshold value: 0.121.

Threshold value: 0.13.

Threshold value: 0.042.

Threshold value: 0.043.

Threshold value: 0.034.

Threshold value: 1.854.

Threshold value: 0.054.

Threshold value: 2.04.

Step 1) Normalize ratings.

Step 2) Weight normalized ratings: (Table 16).

Step 3) Evaluate “*” solutions and “−” solutions: (Table 17).

Positive ideal: ( 0.0386 * , 0.0528 * , 0.128 * )

Negative ideal: ( 0.0082 − , 0.0057 − , 0.0065 − )

Step 4) Obtain separation measures:

S i * = ∑ j = 1 n ( v i j − v j * ) 2 , we got S i * = ( 0 , 0.045 , 0.13 ) , rank = C ≻ B ≻ A .

S i − = ∑ j = 1 n ( v i j − v j − ) 2 , we got S i − = ( 0.53 , 0.1 , 0 ) , rank = A ≻ B ≻ C .

Step 5) Find similarities to positive ideal solution: (Table 18).

C i * = S i − S i * + S i − , we got C i − = ( 1 , 0.68 , 0 ) , rank = A ≻ B ≻ C .

Step 6) Rank preference order:

rank = A ≻ B ≻ C , therefore, rank = ( 0.0 386 , 0.0 528 , 0. 128 ) ≻ ( 0.0 152 , 0.0 144 , 0. 111 ) ≻ ( 0.00 82 , 0.00 57 , 0.00 65 ) .

The results of this study can be used as a reference for the Taiwan Energy Development Center and the Urban Development Bureau, as well as for students studying renewable energy.

The limits of this research are based on the actual environment of Taiwan, other areas with similar environments to Taiwan. Welcome to refer to and provide valuable opinions.

The ultimate goal of this paper is to make recommendations to policymakers with limited land resources in Taiwan. Meanwhile, also we can learn from the idea of the Dutch innovative solar bike path [31] [32], create new thinking into the development of solar power to preciously and utility to optimize Taiwan’s land effectiveness, to relieve the dilemma between Taiwan’s land and power developed, and make Taiwan’s economy take a big step toward a milestone. And provide other geographic and topographical environments in the world that are in the same situation as Taiwan to reference. This research method uses DEMATEL to integrate TOPSIS to explore the obstacles to the solar energy development in Taiwan, we will continue to use ingenuity and find innovative ideas and combine the optimal algorithm research method to continue retrieving study.