Because the development of digital twin in the construction industry has just started at this stage, the construction enterprise personnel have limited knowledge of this technology. Therefore, in addition to interviews with 27 relevant practitioners of construction enterprises, this study also combines a large number of relevant domestic and foreign literature and some government policies and industry reports in the process of data collection. The data from these three sources complement and verify each other, which not only ensures the comprehensiveness and accuracy of the extraction of driving factors, but also ensures that the driving factor system constructed in this study is in line with the actual situation of China’s construction enterprises, making the research results more practical.

1) The collection and collation of journal literature is based on keyword searches in CNKI and Web of Science databases, such as “Digital Twin”, “Architecture”, “DT” and “Construction”. By browsing the title and abstract of the literature, eliminating duplicate literature, radiation reading and full-text reading, 128 journal articles related to this research problem were retained as the initial data of grounded theory research, including 45 Chinese articles and 83 English articles.

2) Government policies and industry reports are obtained by searching relevant websites. A total of 10 research materials were included, including the “14th Five-Year Plan for Digital Economy Development” (Guo Fa No. 29), the “14th Five-Year Plan for Construction Industry Development” (Jian Shi No.11), the official website of Guanglianda Company: “Digital Twins and Lean Construction Deep Integration Boosts the High-quality Development of the Construction Industry,” PwC Report (2020): “Digital Twins in Smart Cities,” IDC Report (2023): “China’s Smart City Digital Twin Technology Assessment” and so on.

3) In order to ensure the effectiveness of the interview data collection, first of all, the successful cases and project data are obtained through the official websites of many domestic digital twin technology software development companies, so as to determine the construction enterprises with strong representativeness as the research objects. Secondly, the interview outline is designed and pre-surveyed based on actual project cases, and the interview questions are further improved to ensure the validity of the outline. From May to July, the author interviewed 27 relevant personnel of construction enterprises. The interview time was about 40 minutes. The main interview questions included the adoption plan and adoption of digital twin technology, and the driving factors of adopting the technology. In addition, it is worth noting that the construction enterprises studied in this paper refer to all kinds of enterprises participating in engineering construction, management and digital application in the whole industrial chain of the construction industry (Table 1).

Table 1. Descriptive statistics of interviewees.

According to the data processing flow of grounded theory three-level coding, through open coding, spindle coding and selective coding [11], the initial category is mined and the organic relationship between each category is identified. In order to further improve the transparency and traceability of grounded theory research and ensure the scientific and accurate analysis of data, the qualitative analysis tool Nvivo14 software was used to select 4/5 initial data for coding. In the coding process, following the idea of continuous comparison, experts are invited to jointly check and further discuss after each round of coding, so as to determine the final coding results and effectively guarantee the reliability and validity of the research. By creating free nodes and tree nodes in the software, the final refinement is identified 26 key initial categories, including core technologies of digital twins, development of digital twin platforms and related research advancements, digital twin fidelity, cross-network data security and privacy, and technical compatibility and further categorized them into six main dimensions: technological level factors, technical advantage factors, organizational management factors, economic environment factors, policy environment factors, and market environment factors. Through selective coding, the “driving factors for construction enterprises to adopt digital twin technology” were identified as the core category, and a narrative framework was constructed based on this core category, along with its main and subcategories (Table 2).

Table 2. Rooted theory coding results.

Theoretical saturation test is a key step in grounded theory analysis. In the process of data analysis, saturation is usually judged based on whether there is a new code, that is, when no new code appears by adding additional data, it can be regarded as reaching saturation [12]. According to this test method, the 1/5 sample data reserved before (including 26 journal literature, 2 industry report data and 5 interview data) are encoded by repeating the above steps again, and compared with the existing analysis results. It is found that the categories extracted by the new round of coding can be included in the existing categories, and no new categories or main categories are found.

ANP method is a super matrix based on feedback system, so the analysis of the correlation between the driving factors is the first step in the construction of ANP model. The research collects data through expert questionnaires, and invites a total of 16 experts with rich professional knowledge and industry practical experience in the field of digital transformation and digital twin technology research in the construction industry. The correlation between the above 26 driving indicators is judged and expressed by 0 - 1 points (influence = 1, no influence = 0). Because the evaluation of index relationship involves more professional knowledge, in addition to inviting experts from construction enterprises, experts are also invited from universities and technology research and development enterprises. Most of these experts are involved in the research field of digital transformation and digital twin technology in the construction industry, and have certain industry practical experience, which can make better feedback on the research content of this study.

Figure 1. Structural diagram of the ANP network.

The final results of the recovered questionnaires were calculated by geometric mean, and the threshold value was set to 0.5 when the index relationship was determined. The mean value ≥ 0.5 was judged to be related (establishing ANP connection), and the mean value < 0.5 was judged to be unrelated (not establishing connection). The correlation results of the driving factors are shown in Figure 1.

Enter the correlation results of each driving factor shown in Figure 1 into the Super Decision software to construct an ANP network model of the driving factors for construction enterprises adopting digital twin technology. After model construction, pairwise importance comparisons must be conducted among interconnected driving factors to calculate their weight values, ultimately forming a judgment matrix. The research determines the relative importance of each driving factor through the 1 - 9 scale method, and assigns the importance of the index by forming an expert evaluation team. The experts are invited from the relevant personnel who have carried out the correlation evaluation of the driving index in the previous round. Because there are many judgment matrices involved in this stage and the scoring process is more complicated, in order to improve the efficiency of data collection, this study formed an expert evaluation group of 10 people. The scoring results were judged according to the group discussion. After the final discussion, the consensus was retained as the data source to construct the judgment matrix. Then, the normalized weight of the judgment matrix is solved by Super Decision software and the consistency test is carried out. The test standard is the CR value of each judgment matrix. If the CR value is less than 0.1, it passes the consistency test. For the judgment matrix that does not pass the consistency test, it is re-discussed by the expert group until it passes the test.

For example, by comparing the factors influencing D1 within the A2 technical advantage factor group under the sub-criteria of D1’s core digital twin technologies, the resulting judgment matrix is presented in Table 3.

Table 3. Example of a judgment matrix.

The core work of ANP network analysis is to solve the super matrix, including solving the unweighted super matrix, the weighted super matrix and the limit super matrix. The judgment data is input into the Super decision software, and the super matrix is convergent. Finally, the weight and ranking results of each driving factor are obtained as shown in Figure 2.

Figure 2. Weight distribution results and comprehensive ranking of key performance indicators for construction enterprises adopting digital twin technology.

Based on ANP network analysis method, the importance of driving factors for construction enterprises to adopt digital twin technology is quantitatively evaluated. It is found that the development of D2 digital twin platform and related scientific research progress are the most important factors to promote the technology adoption of construction enterprises. This is because the application of digital twin technology mainly relies on digital twin platform. The progress of related scientific research and the development of digital twin platform for construction enterprises have promoted the progress of technology and improved the usefulness and ease of use of technology. The second is the policy support and guidance of the D23 government, which ranks second and has a high weight. This also fully shows that the policy environment factor is not unimportant. On the contrary, government policies play an important leading and guiding role in the adoption of digital twin technology by construction enterprises. The three driving factors in the group of organizational management factors are all included in the scope of particularly important factors, ranking ninth, fourth and third respectively, which once again confirms that construction enterprises are the main body of digital twin technology adoption. The training of D18 digital twin technology professionals, the strategic planning of D19 construction enterprises for digital twin implementation, and the technical recognition and support of D20 senior managers all play an important role in the process of technology adoption. The driving factor of D21 “improve the cost-effectiveness of enterprises” ranks fifth, which is consistent with the actual situation. Cost control is the key to the survival and development of construction enterprises. Whether the adoption of new technologies can bring significant cost-effectiveness directly affects the profitability and return on investment of the project. D24 customer satisfaction ranks sixth, which corresponds to the driving factor of D6 “continuous improvement of products and services” ranked tenth, is also a key factor that enterprises need to consider when adopting digital twin technology. The importance of the driving factor of D12 improving sustainability is ranked seventh, and this factor ranks first among all the driving factors of technical superiority, which reflects the long-term pursuit of construction enterprises in saving resources and improving efficiency, further improving economic, environmental and social benefits, and realizing sustainable development. In addition, China’s construction enterprises are usually market-oriented, which means that the business decision-making and business development of enterprises are based on market demand. Therefore, the market competition environment of D26 construction industry has also become a particularly important driving factor for construction enterprises to adopt digital twin technology, ranking eighth. The weights of the above 10 factors are all above 0.038, which are all greater than the average weight level. Therefore, they are regarded as the key driving factors, and the future is the key direction that construction enterprises need to pay attention to when adopting digital twin technology.