AbstractDigital fabrication is growing in adoption within the architecture, engineering, and construction industry. Because digital fabrication often requires a systematic rethinking of the design process, research emphasizes new understanding is needed for digital systems across multiple areas, including technical development, technological systems, organizational contexts, contractual provisions, and business models. Despite the importance of a system view, the current body of knowledge does not yet provide a consistent identification nor a comprehensive evaluation of factors and their complex interdependencies with one another. To fill this knowledge gap, this work categorizes and identifies the industry needs and perceived benefits under five areas. Next, the authors conducted a questionnaire survey, which received effective responses from 114 industry stakeholders. The authors analyzed the responses using Rao–Scott chi-squared tests of independence to identify 70 pairs of correlated categories. This work further investigates the variables’ correlations using heat map visualization. Two variable-correlation mappings are presented as follows: (1) a multiaspect mapping of 292 variable correlations under 10 categories of the industry needs of fabrication information and aspects in design modeling, review, and documentation; and (2) a single-aspect mapping of 26 correlations regarding the industry needs for various benefits of digital systems to adopt digital fabrication. Thereby, this work proposes seven strategy propositions to achieve the benefits of digital systems when adopting digital fabrication in design. The consistent identification of the needs and their interdependencies constitute an integral part of knowledge in construction management.IntroductionDigital fabrication—or data-driven production—is growing from a nascent state toward large scale adoption in the architectural, engineering, and construction (AEC) industry. This growth requires an understanding of technical advancements in computing methods and automated workflows for practitioners (Bock 2015). In practice, digital fabrication technologies often still operate within customized or stand-alone software packages. Digital fabrication is not yet understood as a standardized process or an integrated digital system. As a contrasting example, building information modeling (BIM) platforms are now built upon industry standard practices and are currently used across diverse project teams for coordination and information exchange in design modeling, design review, and design documentation to deliver tasks. Information systems for digital fabrication do not exist in the same way.Correspondingly, there is an increasing number of studies that investigates the industry needs and strategies for the adoption of digital fabrication (e.g., Pan and Pan 2020; Law et al. 2022; Kim et al. 2022). Scholarship explores the use of digital systems, such as BIM platforms, that can help stakeholders coordinate the management data including 3D models and algorithms that link to digital fabrication (e.g., Chen et al. 2018; García de Soto et al. 2019; Hall et al. 2020). However, the various studies are not yet consistent regarding the industry needs and perceived benefits for adopting digital fabrication using digital systems. Digital fabrication requires a systematic rethinking of not only the new construction process but also the design process in construction projects (Bock and Linner 2015). The lack of consistency is especially true regarding the early design phase, where digital fabrication requires more detailed knowledge, participation, and computation than traditional craft production. The lack of a consistent and holistic view of industry needs for digital fabrication during the design phase hinders the wider and successful adoption of digital fabrication.Overview of This WorkTo address the knowledge gap and create a consistent and systematic framework for adopting digital fabrication projects, this work seeks to identify the industry needs of using digital systems and establish potential strategies to adopt digital fabrication in the design process by answering the following research questions: 1.What are the industry needs for adopting digital fabrication through digital systems?2.How are the industry needs correlated to one another to foster the adoption in practice?This work firstly presents the variables of the industry needs and perceived benefits identification through a questionnaire survey of 144 industry stakeholders, which is later reduced to a sample of 114 after controlling for quality of responses. To analyze the correlations between the needs, this work presents the correlated categories of variables derived through Rao–Scott chi-squared tests of independence. The variable correlations are then further identified using heat map visualization. Thereby, this work presents two variable-correlation mappings and elaborates on the potential strategies of using digital systems for the adoption. Specifically, this work advances the state-of-the-art in the field of construction management and digital fabrication as follows: •This work examines the industry standard practices of BIM usage for adopting digital fabrication in the early design phase through a complex survey with data-rich findings.•A workflow to map correlations of the needs is presented for the strategical investigation.•This work draws feasible strategy propositions based on the identified industry needs in current practice for stakeholders to consider for adoption immediately.The remainder of this work is organized as follows. The next section presents the current state of literature regarding digital systems and digital fabrication. From this knowledge base, the work identifies variables of interest and categorizes them using five thematic areas. This is followed by the explanation of the research methodology of this work. Subsequently, the findings of the variables, the correlated categories, and correlation mappings through chi-squared tests and heat map analysis are presented and explained in three sections, respectively. In the discussion section, this work elaborates on the potential strategies based on the findings. This work concludes with contributions to theory and practice, limitations, and potential future research directions.Literature ReviewDigital Fabrication Using Digital Systems in DesignDigital fabrication refers to data-driven production based on digital design information (Bock and Linner 2015). It includes robotics and automation in construction processes of additive manufacturing such as 3D extrusion printing, subtractive manufacturing with computer numeric control (CNC) machines, robotic assembly, and operation systems. While the majority of scholarship on digital fabrication focuses on technical implementation, a growing number of recent studies have started to explore the specific management processes required to adopt digital fabrication in construction (Bock 2015). Literature such as Bock and Linner (2015) and Ng et al. (2021) suggest Robot-Oriented Design and Design for Digital Fabrication (DfDFAB) to foster a rethinking of the design process and integrate process, organization, and information for adopting digital fabrication.Scholars also found that digital systems can foster digital fabrication adoption through integration and integrated design approaches such as design for manufacture and assembly (DfMA) (Chen et al. 2018; Hall et al. 2020). It is well established that digital systems such as BIM platforms can improve quality and communication amongst project stakeholders by enabling design coordination and integrating design and construction information when used correctly (Succar 2009). More recently, examples show how Industry Foundation Classes (IFC) can be used for information exchange for digital fabrication adoption (He et al. 2021; Smarsly et al. 2021). In theory, an open IFC format provides no restriction and consistent data mapping regardless of which digital systems are being used (Teizer et al. 2018). Still, digital systems for digital fabrication are not as developed as that of BIM. There does not yet exist something like a common data environment (CDE) for digital fabrication as an online place for information sharing and management that could integrate design and construction information during the design process to enhance values added throughout the building life cycle (Succar 2009). In brief, there is not yet a consolidated understanding of how digital systems can be used to support the adoption of digital fabrication in the design process.Five Areas Concerning Digital Fabrication Adoption Using Digital SystemsRecent research uses various empirical methods including surveys, interviews, and case studies to explore the industry stakeholders’ perspectives and needs regarding applications and adoption of digital fabrication during the construction process. Through a close scan of these research advancements, the identified industry perspectives and needs can be categorized under five areas: technical development, technological systems, organizational context, contractual provisions, and business models. A summary of this literature across the five areas is provided in Table 1. For example, Pan and Pan (2019) compares and analyzes digital fabrication cases regarding technological, organizational and environmental contexts. Pan and Pan (2020) explores the task areas and influencing factors of robotic applications in Hong Kong through a questionnaire survey. Law et al. (2022) studies perceived benefit and barrier factors that influence robotic adoption in Hong Kong through a questionnaire survey. In a similar vein, Kim et al. (2022) categorizes stakeholders’ perception of robotization derived from interviews into usefulness and output quality (technical), ease of use (technological) and job opportunity (business model) etc. Hwang et al. (2022) conducts surveys and interviews to explore challenges and opportunities for adopting Industry 4.0 in Singapore. They include data information sharing (technological), data ownership (contractual), and the need for workforce training (organizational). Moreover, Hall et al. (2020), Pan and Pan (2019), and Ng et al. (2021) compare various digital fabrication adoption cases in current practice through multiple-case-study methods and explore the circumstances of adopting digital fabrication.Table 1. Summary of industry perceptions, needs, and strategies for adopting digital fabrication in the literatureTable 1. Summary of industry perceptions, needs, and strategies for adopting digital fabrication in the literatureAreaExample of industry perceptions need perceived benefitsReference IDA. TechnicalTechnical sophistication mature technology in practiceHall et al. (2020) and Pan and Pan (2019)Compatibility of robotics with existing practiceNg et al. (2021) and Pan and Pan (2019)Enable design freedomLaw et al. (2022) and Ng et al. (2021)Enhance productivity, efficiency, cost-effectiveness, and reduce workforceBock (2015), Law et al. (2022), Pan and Pan (2020), and García de Soto et al. (2019)Enhance workforce site safetyKim et al. (2022) and Law et al. (2022)Collaboration in a common virtual environmentWeng et al. (2021)Process-oriented design development modelingNg et al. (2021) and Teizer et al. (2018)Configuration kit-of-partsHall et al. (2020)Digital twin for DFAB process simulation and real-time monitoringBock (2015) and Ravi et al. (2021)Task-technology integrative & interactive design for roboticsPan and Pan (2020)Prefabrication, standardization, and modular constructionBock (2015), Kim et al. (2022), Ng et al. (2021), Pan and Pan (2020), and Teizer et al. (2018)B. TechnologicalCommon data environment linked dataBock (2015)Enable visual recognition techniques visual programming interfaceLaw et al. (2022) and Weng et al. (2021)Ensure high accuracy level of developmentLaw et al. (2022)Uptake of information and communication technology (ICT)Hwang et al. (2022), Pan and Pan (2020), and García de Soto et al. (2019)Information integration in BIM for DFAB processes (e.g., prefabrication) and testsRavi et al. (2021) and Teizer et al. (2018)BIM-based integrable environmentLaw et al. (2022) and Ng et al. (2021)Open data formatsTeizer et al. (2018)DFAB code integration to link 3D or BIM modeling platformFardhosseini et al. (2019) and Ravi et al. (2021)Ease of use & interactive availability of robotsKim et al. (2022) and Pan and Pan (2020)Data & information sharingBock (2015), Hall et al. (2020), Hwang et al. (2022), and Ng et al. (2021)C. OrganizationalStrong support from the top managementPan and Pan (2019)Organizational integration enhancementHall et al. (2020), Ng et al. (2021), Pan and Pan (2019, 2020), and García de Soto et al. (2019)Willingness to change work structure and managementPan and Pan (2019, 2020) and García de Soto et al. (2019)Collaboration with DFAB developers, suppliers, & engineersPan and Pan (2019, 2020) and García de Soto et al. (2019)Promote knowledge sharingHall et al. (2020) and Pan and Pan (2020)Promote workforce training or accept new jobsHwang et al. (2022), Kim et al. (2022), Pan and Pan (2019), and García de Soto et al. (2019)High level of trust in technologyKim et al. (2022) and Ng et al. (2021)D. ContractualProvision of incentives reward-sharingLaw et al. (2022), Ng et al. (2021), and Pan and Pan (2019)Regulatory compliance & liaising to advance technology developmentHwang et al. (2022) and Pan and Pan (2019, 2020)Specifying open-data formats in BIM-based design and DFAB processTeizer et al. (2018)Understanding of data format requirements and data ownershipChong et al. (2017), Celoza et al. (2021), and Hwang et al. (2022)E. Business modelMarket-oriented & enable market growthHall et al. (2020), Law et al. (2022), and Pan and Pan (2019)Promote economic performance (e.g., profit-making, cost reduction)Kim et al. (2022), Law et al. (2022), Pan and Pan (2019, 2020), and Teizer et al. (2018)Promote sustainable development (e.g., materials saving)Law et al. (2022) and Pan and Pan (2020)Promote innovation adoptionHall et al. (2020) and Pan and Pan (2019, 2020)Save time and improve efficiency (e.g., delivery time)Kim et al. (2022), Law et al. (2022), Pan and Pan (2019), and García de Soto et al. (2019)Enable design freedomLaw et al. (2022) and Ng et al. (2021)Improve accuracy and qualityKim et al. (2022), Law et al. (2022), and Teizer et al. (2018)Create new jobsLaw et al. (2022), Kim et al. (2022), and García de Soto et al. (2019)Enable localized adaptionsHall et al. (2020) and Pan and Pan (2020)Leverage technology advantagesPan and Pan (2020)Variables of Digital Fabrication under the Five AreasWithin each of the five categorized areas, past scholarship notes many potential variables of the industry needs for the design for digital fabrication. Here, a variable can be defined as either a requirement, a motivation, a role, or a consequence (e.g., a benefit or a risk) identified in the literature as consequential for digital fabrication adoption.For technical development of digital fabrication, literature discusses fabrication information, motivation for digital fabrication, digital twin in digital fabrication provision, as well as the requirements in the design process for design modeling, design review, and design documentation. For example, Ng et al. (2021) identifies that specific fabrication information (e.g., codes and specifications) is required during the design process that includes modeling, review, and documentation in the studied cases; Bock (2015) also reveals that the industry adopts digital fabrication for various motivations such as productivity and cost-effectiveness. Hall et al. (2020) presents digital systems integration that enables configuration and kit-of-parts for product optimization. Ravi et al. (2021) studies digital twin and virtual commissioning for managing design for digital fabrication.In addition, the literature discusses technological systems that include the use of programming languages, machine code support, cloud-based information, and supporting function tools for designing using the digital systems. For example, Weng et al. (2021) emphasizes visual programming interfaces such as Grasshopper in Rhino 3D for designing and coding for for digital fabrication. Fardhosseini et al. (2019) describes the interaction and workflow between Rhino, Python, and G-code necessary during the design phase of a digitally fabricated concrete formwork. Ravi et al. (2021) presents supports for codes integration to link the BIM model for the robotic path-planning. Bock (2015) specifies data sharing and linked data for implementing digital fabrication.The literature explains the organizational context that includes the participation of various new roles in digital fabrication projects. García de Soto et al. (2019) emphasizes the following five new digital fabrication (DFAB) roles. The “DFAB manager” defines strategies to adopt digital fabrication and manage the technologies in design and construction processes. The “DFAB programmer” works on programming to control the digital fabrication process with the technologies. The “DFAB engineer” specifies the engineering work of digital fabrication technologies for the desired fabrication work. The “DFAB design coordinator” ensures that project design and digital fabrication products align. Finally, the “DFAB BIM coordinator” ensures that BIM data and digital fabrication align.Past literature also discusses contractual provisions that enforce legal accountability in design for digital fabrication. The topics include data and documentation formats, as well as data ownership and management models in the literature. For example, Chong et al. (2017) and Celoza et al. (2021) study contract provisions including data format and requirements, data management, and BIM usage for project success. McNamara (2020) proposes different networks regarding intelligent contractual requirements of digital systems for industry 4.0. The authors summarized the following six models for data management to contract for digital fabrication using digital systems: •In a centralized model, all data within the platform are owned and managed by one firm throughout the project.•The semi-centralized model means that everything within the platform is owned and managed by one firm that leads the stage, but each stage in the design process can have a different owner.•The decentralized model describes cases where firms are organized in clusters and each cluster has a firm to own and manage the data among the firms in that cluster.•In a discrete model, each firm owns and manages its data within the platform.•In a distributed model, all firms own everything shared within the platform.•In an exclusive model, only one set of data owned and managed by the project client is allowed to be shared.Furthermore, past literature suggests reconsideration of business models of digital systems that include benefits and risks in projects and how the digital systems can be financially funded. For example, Teizer et al. (2018) explores the benefits and risks of BIM platforms for 3D printing technology. Hall et al. (2020) finds three types of management for digital fabrication, namely digital systems integration, vertical integration, and relational strategy with a spin-off factory. Bock and Linner (2015) studies the adapted business models for adopting digital fabrication in countries such as Japan. Pan and Pan (2019) and Law et al. (2022) identify that the adoption in Hong Kong requires market-oriented planning to foster market growth. A better understanding of funding a sustainable development of the digital systems for the adoption is needed.Knowledge GapAlthough digital fabrication has been promoted in various countries, stakeholders’ perspectives and research advancements in addressing the industry needs remain inconsistent and diversified. It is unclear how the different industry perceptions of needs are correlated to one another as the interdependencies of the needs create considerable complexities for stakeholders to learn and adopt digital fabrication projects. This could further hinder the adoption of digital fabrication to a larger scale. To create such consistency in the understanding of the essential variables and to provide a systematic framework of strategies for the adoption, it is needed to synthesize and categorize the industry perception of needs and the perceived benefits as variables. This is proposed previously under five areas (see Fig. S1): technical development, technological systems, organizational contexts, contractual provisions, and business models. Also, there is need to understand from industry practitioners their needs for digital fabrication adoption and how these needs are related to one another.MethodologyThis research is conducted across four stages: a literature review to identify the variables and five thematic areas (presented earlier), a questionnaire survey, Rao–Scott chi-squared tests, and heat map analysis. The specific research steps are described in more detail below.Questionnaire SurveyBuilding upon a content review of the existing research, the authors conducted successive brainstorming and discussion sessions with five peers in order to compose an online questionnaire with 21 questions under the five areas. To make sure that the question descriptions, concepts, and terminology are legible and well-explained for the survey participants, the key terminologies including the nine design stages, “digital fabrication,” “BIM,” “platforms,” “BIM-based platforms,” “platform-based integration,” the five new DFAB roles, and the six different management models are clearly defined in the questionnaire. The questionnaire consists of 21 questions, where each question includes 7 to 12 multiple-response answers for selection. The survey used a complex survey design approach, allowing more than one response per question, which is considered a common and reliable approach to elicit expert opinions (Hsieh and Shannon 2005). For most of the questions, participants could choose a minimum of one and a maximum of three choices.In Stage 2, the questionnaire was published and shared online to selected groups of industry stakeholders via emails and multiple social media platforms in May 2020. Since this survey aims to examine digital fabrication adoption using digital systems in design, seven key stakeholder groups, i.e., architects, project managers, project owners, design engineers, BIM specialists, and contractors (general contractors and digital fabrication trade contractors) and consultants, were invited to participate.The survey had 144 initial participants. An analysis of the participant data revealed several demographic characteristics about the sample. First, participants noted different overlapping roles amongst different groups. For example, some participants are also working as digital fabrication researchers and platform-related professionals. The sample also included few policy makers; it is possible that the technical nature of the questions incentivized more participation by the day-to-day practitioners and less by high-level policymakers or the upper management. Moreover, the participants’ market foci cover different continents to ensure the survey results cover the needs of industry stakeholders to adopt digital fabrication in different regions. Amongst all, 66% of indicated market foci are in Asia, 46% in Europe, and 23% and 22% in North America and the Middle East, respectively. It is important to note that the survey measures how industry participants perceive the importance of certain variables, but these perceptions may differ from the actual importance of variables in the adoption process. To ensure that perception matches reality as much as possible, all survey participants were asked to share their work experience in digital fabrication to different extents. Amongst all, one-third and one-fourth of the participants have more than five years of digital fabrication experience in design and in construction, respectively. The full questionnaire and the participants’ background information can be found in the appendix of this work.Rao–Scott Chi-Squared Tests of IndependenceIn Stage 3, the findings of the survey were then further analyzed through Rao–Scott chi-squared tests of independence to investigate the correlations of the variables derived from the questionnaire survey. Due to multiple possible responses for each question, the contingency table or “the cross-tabulation of responses” from the survey data represents multiple marginal independence. Instead, this work aims to investigate industry needs, perceived benefits, and strategies to adopt digital fabrication using digital systems in the design process. Thus, the Rao–Scott test was chosen over the classical Pearson chi-squared test, which suits more for analyzing surveys regarding single marginal independence. The Rao–Scott test resizes the expected values by introducing a correction factor. It provides good control for multiple categories of variables. The Rao–Scott test statistic, i.e., the adjusted chi-squared test statistic, χRS2, is expressed in Eq. (1), and the correction factor can be obtained from Eq. (2). Also, it should be noted that this research is exploratory and does not use structural equation modeling. Therefore, the test for goodness of fit, such as Cronbach’s alpha test, is not needed for the preprocessing of the survey data (1) (2) where χ2 = standard Pearson chi-squared test statistic; δ = correction factor; m = total count of the multiple responses; n = total number of participants; and C = number of response levels (i.e. the number of columns) in the contingency table.To minimise bias and errors, the authors did not select responses from the participants, for example, whose professions were not clearly provided (e.g., those who refused to fill in their background information regarding their roles, experiences, and market focus, or those with zero experience in any indicated aspect). Amongst all, the responses from 114 participants were selected for the statistical analysis. Since the effective respondents were selected on purpose to serve the goal of this research, the sample of 114 participants is intended to represent the larger population target of industry stakeholders characterized by sufficient experience in digital fabrication. This can ensure some homogeneity of the sample while acknowledging that some possible variations according to the geographical location are not accounted for in this study.Heat Map Analysis for Variable-Correlation MappingsIn Stage 4, this work uses an extensive and exhaustive testing process known as heat map analysis to investigate the correlations between the variables surveyed. Heat map analysis has been widely adopted in statistical literature using a hierarchical cluster structure coupled with a graphical display of a data matrix. The corresponding element’s value is represented by a colour scale in each tile (Wilkinson and Friendly 2009). This method helps to visualize data correlations in each pair of variables and visualize the most significant variables in terms of the percentages of the responses. An example with the data used in this work is presented in Table S1.To better present the variable correlations indicating merely the most significant variables for studies in this work, the authors mapped the variable correlations in multipartite diagrams. For example, to map the variable correlations under the correlated pair of A1 (fabrication information) and A6 (digital twin provision), each provided multiple-choice answer of A6 is displayed in rows, while each provided answer of A1 is displayed in columns. Through heat map visualization, the authors identified the A6 variable “digital twin strategy should be included for project tendering” is correlated to the A1 variable “project implication.” Hence, the variable correlation is mapped as shown in the tripartite diagram in Fig. 2.Descriptive Analysis of Survey ResponsesThis section highlights the industry needs in each of the five areas and the resultant industry needs through statistical methods to draw variable-correlation mappings. This shows how the analysis can assist industry stakeholders to adopt digital fabrication using digital systems in design.The variables from the questionnaire survey with the highest percentage of response are presented in Table 2. A full set of all responses in percentage to each question can be found in the appendix. Further details about the descriptive statistical results can be found in Ng et al. (2020).Table 2. Categories of variables, questions in the questionnaire, and their most significant variables from the surveyTable 2. Categories of variables, questions in the questionnaire, and their most significant variables from the surveyNo.Category of variablesQuestion in the questionnaireMost significant variable from the survey 1ProfessionPlease select the role(s) that best describe your experience and background?Architects 2ExperiencePlease indicate your experience in the following field(s)- Digital fabrication in designMore than 5 years of experience- Digital fabrication in constructionNo experience- BIMMore than 5 years of experience- Software developmentNo experience 3DemographicsPlease indicate the market(s) in which you have experience?Asian market A1Fabrication informationWhich fabrication-related information is the most essential to the design process?Design specification and configuration A2Motivation for DFABWhat are the primary motivations for you to design for digital fabrication over traditional fabrication processes?Increase geometrical complexity and architectural form A3Design modelingWhat aspect(s) in design modeling is/are the most essential for successful digital fabrication?Collaboration in a common virtual environment A4Design reviewWhat aspect(s) in design review is/are the most essential for successful digital fabrication?Process visualization and simulation A5Design documentationWhat content(s) in design documentation is/are the most essential for successful digital fabrication?Model ontology and configuration to match the fabrication approach A6Digital twin provisionTo what extent should a digital twin in fabrication be prepared for digital fabrication?A digital twin strategy should be included for project tendering B1Programming languageWhat software/program language(s) do you usually use for digital fabrication programming?Rhino 3D plugins B2Machine code supportHow should the machine code (e.g., G-code) be supported for editing on the platform?Adopted in an integrated visual programming language B3DFAB information on cloudWhich digital fabrication information, if shared on an online cloud platform, would most benefit the design process?3D modeling interface B4Supporting functionIf a collaboration platform was created, what supporting function(s) for communication and knowledge sharing would be most useful to you?Drawing-related communication tool C1DFAB roles participationIn your opinion, how should following roles participate in the design process of a digital fabrication project?- DFAB managerAs member of the architect team- DFAB programmerAs member of the engineer team- DFAB engineerAs member of the engineer team- DFAB design coordinatorAs member of the architect team- DFAB BIM coordinatorAs member of the architect team D1Platform and data formatHow should software/platform(s) and data format(s) be included in the project contract?A list of software/platforms are explicitly specified as options in a basic clause since a project begins D2Documentation formatWhat documentation format(s) should be included in the project contract for digital fabrication in the design process?Specification text with the general fabrication intents D3Management modelsWith what management model(s) should the following be owned and managed within the platform?- Design model and dataCentralized management model- Fabrication informationSemi-centralized management model- Intellectual properties of all dataDiscrete management model E1Benefit of digital systemsWhat would be the key benefits if a collaboration platform for BIM-based design projects existed?Digitization and computational advancements E2Platform business modelWhat is/are ideal for a platform to be accessed and financially funded?Partially open-source; no charge for general usage, but charging for upgraded services E3Risk of digital systemsWhat are the key risks of a platform in projects?Uncertainties in digital fabrication implementation in the industryThe six questions from A1 to A6 evoke the answers to the technical needs in terms of technical information flows in digital fabrication. The technical needs indicate that digital fabrication information would be included in the design process and influence design development to different extents. For example, process visualization and simulation and model ontology to match the fabrication approach are the variables that the industry perceived as most needed for design review and design documentation, respectively. Also, a digital twin strategy is perceived in the design process for project tendering. The four questions from B1 to B4 evoke the answers to technological needs in terms of digital platforms in digital fabrication. The technological needs indicate that visual programming languages are needed for digital fabrication adoption using digital systems while communication tools are still needed to deal with conventional drawings-related functions. The questions under the C1 tier evoke the answers to organizational needs in digital fabrication. The organizational needs indicate that the most significant digital fabrication roles would be members of either the architect team or the engineer team. Regarding contractual needs, digital systems should be mentioned in a basic clause when a project begins. Fabrication intents as a kind of fabrication-related information should be specified in contracts. Relatively centralized management models are perceived as needed for design models and data and fabrication information in digital systems while intellectual properties are perceived to be managed with a discrete model. Computational advancements are perceived by survey participants as the benefits of digital systems. This echoes with the increased geometrical form complexity as the motivation for digital fabrication based on question A2. The risks of digital systems are perceived as a concern because digital fabrication implementation has uncertain profit and value for money in the current market.Correlated Categories of Variables in the Design ProcessAmongst all, the categories of variables in the design process, including A1, Fabrication information; A3, Design modeling; A4, Design review; and A5, Design documentation, represent the needs in the design process. To investigate the categories of variables correlated to those in the design process, the authors first paired all variables under the 21 categories of variables derived from the 21 questions in the questionnaire as mentioned earlier with those under the categories of variables A1, A3, A4, and A5 to form 100 pairs. Then, the authors analyzed the percentages of the responses in the variables in the contingency table for each pair using the Rao–Scott chi-squared tests with Python (Scott 2007). Through the Rao–Scott tests, the pairs with their p-values below 0.01 means that there is a significant dependency relationship. The resultant p-values for each pair are shown in Table 3. For example, all p-values on row 1, the profession category of variables, have p-values below 0.01. This indicates that this category has a dependency relationship and is correlated to the categories (A1, A3, A4, and A5) in the design process. While on row 2, the experience category of variables, merely the p-value under column A5, design documentation is below 0.01. This indicates that this category only has a dependency relationship with design documentation category of variables in the design process.Table 3. Resultant p-values using Rao–Scott chi-squared tests with the 100 pairs of categories of variablesTable 3. Resultant p-values using Rao–Scott chi-squared tests with the 100 pairs of categories of variablesNo.Category of variablesA1A3A4A51Profession0.00000.00000.00000.00002Experience0.01720.1690.15080.00663Demographics0.69740.26740.85390.7235A1Fabrication information—0.00000.00000.0000A2Motivation for DFAB0.98050.34550.95430.3986A3Design modeling0.3124—0.00000.0008A4Design review0.06720.0000—0.0001A5Design documentation0.07830.00000.0179—A6Digital twin provision0.00000.00000.00000.0000B1Programming language0.00000.00000.00000.0000B2Machine code support0.00000.00000.00000.0000B3DFAB information on cloud0.00000.00000.00000.0000B4Supporting function0.63710.05140.19600.0352C1DFAB roles participation- DFAB manager0.00000.01350.00520.0028- DFAB programmer0.00000.00000.00000.0000- DFAB engineer0.00000.00000.00000.0000- DFAB design coordinator0.49430.00030.23430.0016- DFAB BIM coordinator0.00000.00000.00000.2620D1Platform and data format0.00000.00000.00000.0000D2Documentation format0.00040.00000.44910.0000D3Management models- Design model(s) and data0.00000.00160.76210.0001- Fabrication information0.00000.00000.00000.0000- Intellectual properties of all data0.00120.00000.00790.0000E1Benefit of digital systems0.00120.00320.38340.0000E2Platform business model0.00130.00000.03210.0000E3Risk of digital systems0.06860.00040.23450.0410To better illustrate the correlations between categories of variables, this work further illustrates the 70 pairs of correlated categories in a bipartite diagram as shown in Fig. 1. Based on the findings from the Rao–Scott chi-squared tests, this work identifies the correlated categories as follows.Regarding technical industry needs, the digital twin provision category is correlated to all four categories in the design process while fabrication information is correlated to the rest of the three categories. Motivation for digital fabrication is not correlated to any category. Within technological needs, the programming language, machine code support, and digital fabrication information on cloud categories are correlated to all four categories in the design process; while the supporting function category is not correlated at all. For organizational needs, the “DFAB programmer” and “DFAB engineer” participation categories are correlated to all four categories in the design process. Participation of all roles except “DFAB manager” are correlated to the design modeling category; while that of all roles except “DFAB BIM coordinator” are correlated to design documentation. In regards to contractual needs, the platform and data format category, as well as management models for the fabrication information and intellectual properties categories, are correlated to all four categories in the design process. Within the needs for business model, the benefit of digital systems category is correlated to the fabrication information, design modeling, and design documentation categories in the design process while the risks of digital systems category is merely correlated to the design modeling category.In regards to backgrounds, the profession category shows its correlations to the four categories in the design process, namely the categories of fabrication information, design modeling, design review, and design documentation. By contrast, the experience category is only correlated to design documentation category in the design process, and the demographics category is not correlated to any categories in the design process.Variable-Correlation MappingsFurther to the correlated categories identification, this work further identifies the correlated variables within the corresponding categories by investigating the percentages of responses using heat map analysis (Wilkinson and Friendly 2009). An example of the heat map analysis data in contingency tables can be found in Table S1. This work maps the variable correlations so as to provide insights for industry stakeholders regarding how the variables can be considered to various extents in the design process. In this work, two variable-correlation mappings are presented to investigate, based on Fig. 1: (1) the correlations of variables under the identified 10 categories of the industry needs that are correlated to fabrication information, design modeling, design review, and design documentation in the design process; and (2) the variable correlations under the benefits of digital systems category that is correlated to fabrication information, design modeling and design documentation in the design process. A workflow of correlation mappings can be derived based on the findings from the questionnaire survey and the Rao–Scott chi-squared tests in the previous sections. With the provided findings, further mappings can be developed upon researchers’ interests in the study of digital fabrication adoption using digital systems in the design process as explained in the following paragraphs.Multiaspect Mapping: Variable Correlations in the Design ProcessFrom the preceding sections, there are 10 categories of variables that are correlated to the four categories in the design process. This work elaborates the findings further to present one example of a multiaspect variable mapping that indicates the most significant variables under 10 categories and illustrates the results in a tripartite diagram as shown in Fig. 2. Also, to further differentiate the variable correlations, this work categorizes the correlations under six aspects of who, what, when, where, why, and how. The following paragraphs explain the variable correlations in each aspect.The who aspect of variable correlations includes variables under the profession, DFAB engineers participation, and DFAB programmer participation categories. The correlations belong to the who aspect because the needs concern people. In general, the variables are significantly correlated to design specification andconfiguration as the digital fabrication information as the industry needs in design. They are also highly correlated to modeling follows fabrication approach for design modeling, process visualization and simulation for design review, and model ontology and configuration for design documentation. Amongst all, some findings indicate unconventional practices possibly adopted by the survey participants. For example, only digital fabrication contractor amongst all roles is highly correlated to demonstration of the fabrication intents for design review. Also, only DFAB engineers participation in the client team has a high correlation with connections with computer-aided manufacturing/process planning (CAM/CAPP) system for design modeling.The what aspect includes variables under the programming language category. The correlations belong to the what aspect because the needs what to be used for the adoption. Many of the variables are correlated to project implication and design specification and configuration as digital fabrication information needed in design. They are also highly correlated with common virtual collaborative environment for design modeling, process visualization and simulation for design review, and model ontology and configuration for design documentation. Besides, the usage of different programming languages is correlated to various variables for design documentation.The when aspect includes variables under the digital fabrication information on cloud category. The correlations belong to the when aspect because the needs concern the adoption when sharing on cloud (or not). Overall, most of the variables are correlated to design specification and configuration as digital fabrication information needed in design, common virtual collaborative environment and modeling follows fabrication approach for design modeling, and model ontology and configuration for design documentation. All variables are correlated to process visualization and simulation for design review. Also, the most significant variable of information shared on cloud (as per Table 2)—3D modeling interface—is the only variable, amongst all, that is correlated to supply-chain information and matrix for design documentation.The where aspect includes variables under the platforms and data formats in contracts category. The correlations belong to the where aspect because the needs concern where the platform(s) and format(s) are included. Overall, many of the variables correlated to design specification and configuration as digital fabrication information needed in design, modeling follows fabrication approach for design modeling, and process visualization and simulation for design review. All variables are correlated to model ontology and configuration for design documentation. The single platform in basic clause variable seems to have different correlations in the design process amongst other variables under the same category.The why aspect includes variables under the digital provision category. The correlations belong to the why aspect because the needs concern the reasons why the provisions are needed (e.g., to acquire a digital twin-ready model or physical mockup). Overall, most of the variables are correlated to design specification and configuration as digital fabrication information needed in design, common virtual collaborative environment and modeling follows fabrication approach for design modeling, process visualization and simulation for design review, and model ontology and configuration for design documentation. Surprisingly, the most significant variable—digital twin strategy preparation for tendering—is the only one that is correlated to project implication as fabrication information in design.The how aspect includes variables under the machine code support, management model for fabrication information, and management model for intellectual properties categories. The correlations belong to the how aspect because the needs concern ways models (e.g., how machine code is being supported) for the adoption. Overall, many of the variables are correlated to design specification and configuration and project implication as digital fabrication information needed in design, modeling follows fabrication approach for design modeling, and model ontology and configuration for design documentation. Almost all how variables are correlated to process visualization and simulation for design review. It is worth noting that only exclusive management model for fabrication information variable, amongst others, is correlated to supply chain information and matrix for design documentation.Single-Aspect Mapping: Benefits of Digital Systems Variable Correlations in the Design ProcessThis work further applies the workflow to investigate another example of a single-aspect variable correlation regarding the benefits of digital systems for adopting digital fabrication in the design process. This is undertaken by mapping the variables under the benefit of digital system category with those under the correlated categories of variables in the design process, namely the fabrication information, design modeling and design documentation categories as per the Rao–Scott chi-squared tests. The results of the variable-correlation mapping are illustrated in a bipartite diagram as shown in Fig. 3(a). Overall, most identified benefits are correlated to design specification and configuration as digital fabrication information needed in design. Only the enhance sustainable development opportunities and contemplating the future of work variables are correlated to project implication. Also, the variables show very differentiated correlations regarding design modeling; the most significantly correlated variables for design modeling are common virtual collaborative environment and parametric modeling capacity. Surprisingly, all benefit variables are correlated to modeling ontology and configuration.DiscussionStrategy Propositions with Digital Fabrication Adoption Using Digital SystemsBased on the findings from single- and multiaspect mapping, the authors further examined potential strategy propositions that can be derived from the findings from variable-correlation mappings presented in this work. Since the demonstrated variable-correlation mappings in this work were derived based on the industry needs in current practice, they can provide technical, technological, organizational, contractual, and business-model insights to different extents for industry stakeholders to develop potential strategies. In the following paragraph, this work demonstrates one example of strategy propositions to illustrate one possible workflow to develop potential strategies with digital fabrication adoption. In the context of the aforementioned technologies, the use of the word “strategy” in this work refers to a long-term plan for actions or an implementation mechanism that would allow digital fabrication to be successfully adopted in the design process.This work integrates variable-correlation mappings presented in the previous section in a tripartite diagram as shown in Fig. 3. Thus, this presents strategy propositions for benefits of digital systems that involve, for instance, digital twin provision, participation of DFAB engineers, platform and data format in contracts, and management models for fabrication information in the design process. Based on the findings from Rao–Scott chi-squared tests as illustrated in Fig. 1, the benefit of digital systems category of variables is correlated to the fabrication information, design modeling, and design documentation categories in the design process. These correlations belong to the why aspect and they are mapped in Fig. 3(a) while Fig. 3(b) presents four correlations extracted from Fig. 2. The two mappings do not necessarily mean that benefit of digital systems category is correlated to, for example, digital twin provision because this requires further tests to verify the correlations. However, potential strategy propositions can be inspired by combining them with their commonly correlated categories of variables. The strategy propositions provide an instant landscape of what should be done by the stakeholders to bring certain benefits of digital fabrication into real practices. Based on Fig. 3, the authors proposed, amongst others, the following seven highlights of strategy propositions as takeaways: 1.All perceived benefits of digital systems from the survey would be achieved by incorporating model ontology and configuration in design documentation, where digital twin in fabrication is recommended to be provided for tendering to different extents depending on stakeholders’ specific purposes and resources.2.Based on the survey, digitization and computational advancements is the most significant benefit variable of digital systems. To achieve this benefit, the needs of modeling follows fabrication approach in design modeling is highly perceived. Thus, DFAB engineers are suggested to participate as either individual external consultants or as members of the engineer team. The design process should take into consideration constraints from fabrication workflow to eliminate future rework in component manufacturing and installation.3.To enhance sustainable development opportunities using digital systems, parametric modeling capacity is perceived in design modeling to reduce rework and waste before tendering through the use of physical mockup to test the feasibility and soundness of the integrated digital twin models.4.A decentralized management model is suggested for digital fabrication information networks that can manage, in particular, machine constraints information and achieve sustainable development opportunities (e.g., reduce component’s quality defects) through specific digital systems during the design process.5.When project owners would like to include a digital twin strategy for project tendering, project implication information would be needed during the design process. Also, specifying one specific single platform in the basic clause on contracts would be recommended before the project begins. This can simultaneously benefit in clarification of stakeholders’ liabilities and future workforce.6.To maximize profits and return on investment in the business model, a tight feedback loop in design modeling can be provided. It is also recommended that DFAB engineers can participate in the architect team to enable a smooth information flow for the coordination of engineering, fabrication, and design requests.7.To request for a physical mockup to test digital twin technology or virtual commissioning before tendering, the model follows fabrication approach is recommended. This could help project stakeholders integrate supply chains to spur systemic innovations with digital systems.Contributions to Theory and PracticeThis work contributes to the increasing digital fabrication adoption in current practice in the following aspects. Firstly, this work examines the current industry practice used in digital fabrication adoption. Whereas BIM platforms have been widely used nowadays by project teams in many countries in the design and construction processes, digital fabrication and its associated digital system needs are still not standardized. The identification of correlated industry practices can help clarify what it is exactly that digital fabrication requires. Secondly, this work provides data regarding the industry needs in current practice for digital fabrication adoption. The findings can provide information for stakeholders’ reference while adopting digital fabrication in projects, including at which stage in design that specific needs are most present. Thirdly, this work comprehends the correlations amongst various industry needs and thus provides a more in-depth understanding of the needs for the adoption. Fourthly, this work draws feasible strategy propositions based on the identified industry needs in current practice. Hence, stakeholders can adopt and verify the proposed strategies immediately in projects.Overall, this work contributes to the increasing research about digital fabrication adoption in the following aspects. Past scholarship had not yet provided a consistent and comprehensive evaluation of the factors and their complex interdependencies to adopt digital fabrication in current design practice. This work fills the gap by identifying and categorizing the variables possibly relevant to the adoption of digital fabrication using digital systems, in particular, during the design process. Secondly, this work provides data-rich findings of industry needs through a complex survey of wide-ranging industry stakeholders. The findings can help digital fabrication researchers to research for the needs. Thirdly, this work demonstrates a workflow of variable-correlation mapping through statistical methods and a workflow to elaborate the correlations to draw potential strategies for future studies. Referring to the single- and multiaspect variable-correlation mappings in this work, researchers and industry stakeholders can develop any other strategy propositions that interest them. Strategy propositions demonstrated in this work provide insight as takeaways for researchers and industry stakeholders when adopting digital fabrication using digital systems. Future research can be done to validate and consolidate strategies and verify the effectiveness in practice. For example, qualitative case research on project design could be used to observe the correlated categories in practice. Additional surveys could be conducted with a repeated survey in future research to further refine and validate the findings in this work.ConclusionDigital fabrication has received great attention in the past decade due to its potential in connecting design and construction activities to improve project efficiencies. Advanced research and practical efforts have been seen in various perspectives for promoting the successful uptake of digital fabrication technology and processes. While these perspectives are often inconsistent and a comprehensive view of adoption strategies remains not adequately studied, the adoption of digital fabrication faces great challenges. To address this knowledge gap, this work identifies the industry needs under five areas of technical development, technological systems, organizational contexts, contractual provision and business model through a complex survey of 114 samples of industry stakeholders using an online questionnaire. Based on the survey results, this work pairs the questions as categories of variables with the categories regarding fabrication information, as well as design modeling, design review, and design documentation in the design process and identifies 70 correlated pairs, amongst the 100 pairs, of categories using Rao–Scott chi-squared tests of independence. Moreover, two variable-correlation mappings are presented in this work: (1) a multiaspect mapping of 292 correlations of variables under 10 categories that are correlated to fabrication information, design modeling, design review, and design documentation in the design process; and (2) a single-aspect mapping of 26 correlations under the benefits of digital systems category. Furthermore, this work elaborates a workflow to discuss potential strategy propositions with digital fabrication adoption using digital systems in design based on the variable-correlation mapping. Hence, the authors proposed seven highlights of strategies as takeaways. For example, to maximize profits and return on investment in the business model, tight feedback loop in design modeling can be provided. Hence, DFAB engineers are recommended to participate in the architect team.This work further discusses its seven contributions to theory and practices, including the industry needs identification and strategy propositions. In spite of that, the authors would like to highlight several limitations, amongst others, as follows. The survey was based on a content analysis of all key identified literature and peer discussion, but not a systematic review. The industry needs identified in this work through the survey do not limit to the findings of this work Also, a correlation does not imply absolute causation. Future research can examine the cause-and-effect relationships between the variables of needs. Further research can be conducted to validate and verify the benefits and strategies in quantitative and qualitative ways. For example, a future study can examine how added values (e.g., productivity) and reduced wastes (e.g., CO2) can be achieved due to modeling in a common virtual environment and the specification of digital twin-ready models in project contracts.Data Availability StatementSome or all data, models, or code generated or used during the study are available in a repository online in accordance with funder data retention policies. The code snippets for the tests of independence and all the contingency tables with heat map visualization for this research can be accessed via https://doi.org/10.5281/zenodo.5807842.References Bock, T., and T. Linner. 2015. Robot-oriented design. New York: Cambridge University Press. Chen, Q., B. García de Soto, and B. T. Adey. 2018. “Construction automation: Research areas, industry concerns and suggestions for advancement.” Autom. Constr. 94 (Oct): 22–38. https://doi.org/10.1016/j.autcon.2018.05.028. Fardhosseini, M. S., H. Abdirad, C. Dossick, H. W. Lee, R. DiFuria, and J. Lohr. 2019. “Automating the digital fabrication of concrete formwork in building projects: Workflow and case example.” In Computing in civil engineering 2019: Data, sensing, and analytics, 360–367. Reston, VA: ASCE. García de Soto, B., I. Agustí-Juan, S. Joss, and J. Hunhevicz. 2019. “Implications of Construction 4.0 to the workforce and organizational structures.” Int. J. Construct. Manage. 22 (2): 205–217. https://doi.org/10.1080/15623599.2019.1616414. Hall, D. M., J. K. Whyte, and J. Lessing. 2020. “Mirror-breaking strategies to enable digital manufacturing in Silicon Valley construction firms: A comparative case study.” Construct. Manage. Econ. 38 (4): 322–339. https://doi.org/10.1080/01446193.2019.1656814. He, R., M. Li, V. J. Gan, and J. Ma. 2021. “BIM-enabled computerized design and digital fabrication of industrialized buildings: A case study.” J. Cleaner Prod. 278 (Jan): 123505. https://doi.org/10.1016/j.jclepro.2020.123505. Hwang, B.-G., J. Ngo, and J. Z. K. Teo. 2022. “Challenges and strategies for the adoption of smart technologies in the construction industry: The case of Singapore.” J. Manage. Eng. 38 (1): 05021014. https://doi.org/10.1061/(ASCE)ME.1943-5479.0000986. Kim, Y., H. Kim, R. Murphy, S. Lee, and C. R. Ahn. 2022. “Delegation or collaboration: Understanding different construction stakeholders’ perceptions of robotization.” J. Manage. Eng. 38 (1): 04021084. https://doi.org/10.1061/(ASCE)ME.1943-5479.0000994. Law, K. K., S. Chang, and M.-F. F. Siu. 2022. “Factors influencing adoption of construction robotics in Hong Kong’s industry: A multistakeholder perspective.” J. Manage. Eng. 38 (2): 04021096. https://doi.org/10.1061/(ASCE)ME.1943-5479.0001011. McNamara, A. 2020. “Smart cities and construction technologies.” In Automating the Chaos: Intelligent construction contracts, edited by S. Shirowzhan and K. Zhang. London: IntechOpen. Ng, M. S., M. M. Bonanomi, D. M. Hall, and J. Hackl. 2020. “Design for digital fabrication: An industry needs analysis of collaboration platforms and integrated management processes.” In Proc., 37th ISARC, 318–325. Dubai, UAE: International Association for Automation and Robotics in Construction. Ng, M. S., K. Graser, and D. Hall. 2021. “Digital fabrication, BIM and early contractor involvement in design in construction projects: A comparative case study.” Archit. Eng. Des. Manage. https://doi.org/10.1080/17452007.2021.1956417. Ravi, K. S. D., M. S. Ng, J. Medina Ibáñez, and D. M. Hall. 2021. “Real-time digital twin of robotic construction processes in mixed reality.” In Proc., 38th ISARC, 451–458. Dubai, UAE: International Association for Automation and Robotics in Construction. Scott, A. 2007. “Rao–Scott corrections and their impact.” In Proc., 2007 Joint Statistical Meetings, 3514–3518. Alexandria, VA: American Statistical Association. Smarsly, K., P. Peralta, D. Luckey, S. Heine, and H. M. Ludwig. 2021. “BIM-based concrete printing.” In Vol. 98 of Proc., 18th Int. Conf. on Computing in Civil and Building Engineering, 992–1002. Berlin: Springer. Teizer, J., A. Blickle, T. King, O. Leitzbach, D. Guenther, H. Mattern, and M. König. 2018. “BIM for 3D printing in construction.” In Building information modeling, edited by A. Borrmann, M. König, C. Koch, and J. Beetz, 421–446. Cham, Switzerland: Springer. Weng, Y., N. A. N. Mohamed, B. J. S. Lee, N. J. H. Gan, M. Li, M. Jen Tan, H. Li, and S. Qian. 2021. “Extracting BIM information for lattice toolpath planning in digital concrete printing with developed dynamo script: A case study.” J. Comp. Civ. Eng. 35 (3): 05021001. https://doi.org/10.1061/(ASCE)CP.1943-5487.0000964.