The architecture, engineering, and construction (AEC) industry is increasingly exploring the potential of mass customization and its impact on digitalization. However, developing digital tools can be challenging in terms of defining, delimiting, and structuring a construction product platform. To address this, a suitable information model is crucial to translate the information from the real world into a subset of data that a configurator can handle. This research aims to identify the common characteristics of construction product platforms to enhance their deployment into an information model, the so called product variant master (PVM) model. The study adopts a case methodology approach, typifying product platforms in three construction companies, and evaluates the applicability of the PVM model. Based on the findings, a systemic framework is proposed for depicting construction product platforms within the PVM model. he research concludes that by adopting this framework, the industry can streamline the modeling process, facilitate collaboration, and pave the way for efective digitalization in the AEC sector.
ConfWS’23: 25th International Workshop on Configuration, Sep 6–7, 2023, Málaga, Spain * Corresponding author. $ ircag@dtu.dk (I. Campo-Gay); lahv@dtu.dk (L. Hvam)
0000-0002-8962-5386 (I. Campo-Gay); 0000-0002-7617-2971 (L. Hvam)
© 2023 Copyright for this paper by its authors. Use permitted under Creative Commons License studies, including the development of a generic systemic CPWrEooUrckReshdoinpgs IhStpN:/c1e6u1r3-w-0s.o7r3g ACttEribUutRion W4.0oInrtekrnsahtioonpal (PCCroBYce4.0e).dings (CEUR-WS.org) framework for construction product platforms, the application of the PVM model in the construction industry, and the diferences in the application of the PVM in the construction industry compared to the manufacturing industry. Finally, Section 5 discusses the results and concludes the paper with implications for future research and practical applications.
Defining product platforms is crucial in fostering mass customization and digitization in the AEC industry. In order to develop configurators that enable this level of customization, it is necessary to transform the knowledge of industry experts into a manageable subset of information [6]. The first step is the construction of a descriptive model that captures both explicit and tacit knowledge of the product. This knowledge is often dispersed across various departments within the organization. Such phenomenon model is collaboratively built with inputs from diferent domain experts and holds significant importance as it sets the foundation of the product platform architecture since it comprehensively defines the structure, functions, and properties of the product, encompassing its entire lifecycle. The next step involves formalizing the model to enable integration and modeling within an IT tool, such as a configurator. Formalization ensures that the knowledge represented in the phenomenon model can be efectively utilized in the development of a computer model tool. Figure 1 illustrates this process. One of the mass customization principles is modularization, which relates to using and arranging modules in a product architecture. There are many definitions of modularity and modules. However, one can describe a module as a definite object of a product with a distinct function and a defined interface to the other modules [2, 7]. The interface function is a crucial part of a modular product, and it should remain unchangeable as much as possible to grant the upgrade of modules over time [8].
The main types of modularity are depicted in Figure 2
[
1. Component-sharing modularity entails sharing modules across the product platform: E.g., the same engine used in diferent tools. 2. Component swapping modularity implies exchanging parts in a product: E.g., a phone with diferent case color options. 3. Cut-to-fit modularity concerns objects with parametric designs. E.g., a curtain cut with diferent lengths. 4. Sectional modularity involves the association without the restriction of modules: E.g., LEGO brick games. 5. Bus modularity (platform) means having the same interfaces for a base element. E.g., an Arduino board is a platform for electronic components.
an exploratory investigation where research has yet not developed a theory. In this case, we opt for a multiplecase study approach to augment external validity. Nevertheless, we keep the number of cases to three to allow an in-depth analysis suitable for theory-building studies.
Hence, we seek to achieve the generality of the conclusions while conceiving robust knowledge for the academic world [16, 17].
We developed and analyzed three diferent product platforms in three diferent companies. Our primary collection methods were semi-structured interviews, interaction with the various domain experts, and observations.
Figure 3: Basic notation of the PVM model. On the other hand, we conducted data representation and documentation tasks mainly employing the PVM. Finally, we analyzed the information models under an iterative
The tool relies on three theoretical domains [3]. First, observation process of the PVM. object-oriented modeling [11] makes it suitable for further developing digital tools. Second, the systems theory 3.1. Case description [12] provides the structure of the PVM. Third, modeling mechanical products [13], which is one of the reasons Companies 1 and 3 are medium enterprises with over 350 for this research to investigate the validity of using the and 450 employees, respectively, while Company 2 is a PVM in the AEC industry. micro-enterprise with less than five employees. All com
The PVM technique, also named by some researchers panies operate in Scandinavian countries, Sweden and as Product Family Master Plan (PFMP) [14, 15], provides Denmark, and have embedded digital tools in their roua holistic, systemic representation of the information tine tasks to a certain extent, but only the third company from three dimensions: the customer, the engineering, has experience employing configurators. Moreover, each and the part view. First, the customer view reflects the company performs in a diferent stage of the construccustomer’s desire to buy the product. Second, the En- tion value chain and experiences a particular obstacle gineering view contains the functions and principles to regarding a fragmented specification process. Table 1 configure a solution. Third, the part view presents all the provides an overview of the main distinctive features of physical objects that can integrate the final product. the companies.
Moreover, the PVM is divided into two general sections. Company 1 pursues delivering more sustainable soOn the one hand, the left side of the PVM illustrates the lutions to private investors to fulfill new governmental generic structure or part-of structure with the diferent regulations. However, no digital tools can support them objects organized in a hierarchical structure. On the other in developing environmental declarations, and they must hand, the right side of the PVM represents the variants resort to technical consultants to generate certified envior kind of structure, which describes the alternatives of ronmental declarations. the objects to the left. Company 2 seeks to speed the generation of quotes
Additionally, the generic structure is organized into and bills of material to provide a faster response to priclasses further described by a cardinality property and vate investors and agilely decide on the contractor by a set of attributes and constraints. Finally, classes relate benchmarking. to instance connections on the left side of the PVM to Company 3 aims to speed up the design generation represent when a class needs another class to fulfill its process. Even if they use digital tools to support diferent responsibility. tasks during the process, no one can co-generate this
The PVM model is primarily used as a data collection design with the designers and potential customers and method to retrieve information from the real world. Be- additionally include environmental assessment currently sides, it has a significant role as a communication tool done in a separate operation. to exchange and validate data with diferent knowledge All three companies have a shared approach when it experts. Building the PVM consists of multiple iterations comes to the configurator tool, which is seen as a decision that refine the model. Figure 3 presents the basic notation support tool utilized by designers to adopt a proactive of the PVM model. approach to design rather than a reactive one. This proactive approach helps prevent potentially high costs in subsequent project phases. Additionally, in all cases, the conifgurator is integrated without the need for connecting external data or undergoing extensive reengineering processes. Therefore, the configurator successfully fulfills its primary objective of automating processes, relieving the workload on human resources, and speeding up lead times.
in parallel for 30 months until we produced functional and testable configuration system prototypes. On the other hand, we developed the case study in Company 3 separately over seven months. In all cases, we gather the product information through modeling sessions with the relevant domain experts in each case. The sessions were an hour long, and we held them mostly individually.
In Company 1, we had 35 sessions with the project leader, 24 sessions with an environmental assessor, and three sessions with the domain expert. Additionally, we held a testing workshop with the external project committee.
In Company 2, we held 115 sessions with the project leader and 36 sessions with technicians. Additionally, we evaluated the prototype with the potential users through a testing session followed by a semi-structured interview.
In Company 3, we held 20 sessions with the project leader and 20 sessions with the architectural firm in charge of developing the product platform design. We used open-ended questions to gather the data, and later, we reflected it in an ontology model, the PVM, which at the same time served as a communication tool with the domain experts.
Finally, we correlated the three PVM models through an observation analysis. Based on the discussions held in the research group, we developed the study findings under an iterative process to refine the results.
rived from an advanced stage of research. Due to conifdentiality reasons, the specific PVM models utilized by each company cannot be disclosed. However, the subsequent sections describe the outcomes based on the aforementioned research.
Currently, both Company 1 and Company 2 have successfully adopted the PVM model, leading them to incorporate configuration systems into their work environments. These companies have integrated configurator tools using standard configuration systems as supplementary resources to alleviate the burden on human resources. In Company 1, the configurator tool is undergoing final validation, where engineers employ it to make more informed design choices. Similarly, Company 2 has reached a comparable stage, where the configurator replaces previously manual tasks, reducing lead time from weeks to hours. Importantly, these configurators do not interfere with additional software, such as CAD systems, as they are employed at diferent stages and outputs of the construction value chain. Additionally, Company 3 has also achieved success in developing a configurator tool, which has been operational for the past three years. This tool serves as a decision support resource for architects, providing assistance during early design phases of projects. In this case, the tool enhances early design phases of the project. It is worth highlighting that the PVM model played a strong role, drawing attention to various modular design components and assemblies on the platform that required redesign to facilitate the subsequent development of the configurator.
panies despite their stage in the construction value chain. For this purpose, we analyzed three product platforms in three companies with entirely diferent characteristics: company size, construction stage, product, digitalization aim, and on-site or prefabrication construction.
The main findings of the research are presented in the following three subsections. First, we describe the suggested systemic approach for developing product platforms in the AEC industries. Second, we illustrate how to use the PVM model in AEC projects to depict construction information. Finally, we highlight the main diferences between the application of the PVM in the manufacturing industry compared to the AEC industry.
PVM models, we have identified a generic model applicable to any modular construction product platform embracing mass customization. The systemic framework comprises three layers: site, construction, and product.
1. This site depicts the place in which the construction is located. These can have relevance, for example, in terms of the transportation distance of the products from the factory to the working site, calculating the maximum structural load in the roof based on the average snowfall level, or knowing the accommodation capacity of construction machinery such as trucks or cranes, among other features. Moreover, the site layer can have more than one level, for instance, in renovation projects where both location and previous construction need to be considered. 2. The construction represents the volumetric shell in which the company’s products are installed.
In most cases, the construction might be broken up into construction parts. For example, the roof can be one of the construction parts of a building construction. Figure 4 illustrates the generic systemic framework The predominant modularity type in this level is using UML notation. “cut-to-fit,” which has the property of parametrization and, hence, describes the volumetric object. 4.2. PVM in the construction industry 3. The products layer illustrates the actual commercialized products. This layer is composed of mul- The generic systemic framework facilitates the modeling tiple instances, and its total number depends on process in the PVM model by providing a better underthe project’s complexity. There are two defined standing of the construction product platform. Layer types of products: 1, site, and layer 2, construction, are described in the Customer View since they directly depend on customer preferences and choices. Likewise, layer 3, products, is a) Predefined products are predominant in prefabricated construction and are mainly deifned by “component sharing” and “component swapping” modularity. The module interface is significant in predefined products and needs to be particularly welldefined. A frequent example of predeifned products is windows and doors. Another example of a predefined product could be a modular room in which “cut-to-fit” modularity might also be present but in which “component sharing” and “component swapping” modularity have a more significant influence. b) Volumetric products are predominant in onsite construction, and they are mainly deifned by “cut-to-fit” and “sectional” modularity. Hence, the module interface has limited significance, and its principal characteristic is its parametric design. An illustrative example of a volumetric product could be the concrete used to build a wall. depicted in the Part View as it represents all the physical components of the project. The three layers are closely related and utterly dependent on one another.
Figure 5 illustrates the applicability of the PVM information model in the construction industry. Besides, the generic systemic framework is reflected to envision its use in construction product platforms.
This reinterpretation of the PVM model can assist construction companies in portraying their product range, particularly in the early design phase of the information model. Hence, the PVM description could potentially reduce the time and resources invested in designing and organizing the modules and their relationship.
Notable distinctions between the application of the PVM model in industrial manufacturing companies and its application in the construction industry have become evident. The following outlines the unique characteristics and novel approaches of the PVM model specifically tailored for the construction industry, in contrast to previous PVM applications, focused on mass-customized products in manufacturing: which can present challenges in mapping out the production process. • Production variability: Construction products exhibit higher tolerances compared to manufactured products, necessitating allowances and adaptations due to site-specific conditions and project-specific requirements. • Production volume: Mass customized products in manufacturing companies usually target customization at higher volumes. Conversely, the construction industry typically operates at lower volumes and on a project basis. • Product life-cycle: AEC industry products are primarily designed for long lifespans, and consequently, maintenance and renovation processes have a significant influence on the overall product.
In this paper, we conducted an analysis of construction product platforms and developed a systemic framework for their depiction using the PVM model. Although construction projects predominantly involve on- figuration project development methods have been used site construction, encompassing numerous dis- in the AEC industry, the suitability of the PVM model for ciplines, manual operations, and coordination, representing construction product platforms has not been thoroughly studied. Previous literature shows limited ap- try can help streamline the fragmented value chain of plication of the PVM model in construction projects and, construction projects, which often rely on siloed specimoreover, it was originally designed for industrial me- fication processes. Documenting construction product chanical products. Therefore, our study aims to analyze platforms using the PVM model can bring similar benethe validity and applicability of the PVM model rather fits to those achieved by manufacturing industries, such than its feasibility for construction product platforms. as easier maintainability and smoother development of
Our research has three main contributions and out- the product platform. Additionally, this approach has the comes: potential to reduce the modeling phase of the configu
Firstly, we developed a generic framework that pro- rator. It is conceivable that other business fields beyond vides a systematic organization of construction product manufacturing or the AEC industry could benefit from platforms into modules. This framework characterizes the same rationale applied in this research. Therefore, the relationship and cardinality of these modules, de- further studies could contribute to the theory of informascribing them based on their modularity and interface tion models, specifically investigating the applicability significance. Implementing this framework can enhance of the PVM model in fields such as logistics, services, or collaboration between knowledge representation experts processes. and domain experts in the construction industry, leading to a better understanding of construction product platforms. Consequently, more robust, logical, and compre- References hensive models can be created, streamlining the modeling processes and providing deeper insights into the prod- [1] C. L. Thuesen, J. S. Jensen, S. C. Gottlieb, Making ucts themselves. Additionally, this development of IT the long tail work: reflections on the development tools, which was previously hindered by the challenge of of the construction industry the past 25 years, Prorepresenting complex structures in AEC products, can be ceedings 25th Annual Arcom Conference (2009) considerably improved. This framework also addresses 1111–20.
RQ 1. [2] L. Hvam, N. H. Mortensen, J. Riis, Product Cus
Secondly, the framework helps answer RQ 2 by demon- tomization, Springer Publishing Company, 2008. strating the applicability of the PVM model in AEC cases. doi:10.1007/978-3-540-71449-1. Despite being initially designed for industrial manufac- [3] N. H. Mortensen, L. Hvam, A. Haug, Modelling turing projects, our observations confirm its suitability in product families for product configuration systems the construction sector. Thus, we can describe the use of with product variant master, Ecai 2010 (2010). the PVM model in the construction industry and validate [4] A. Kudsk, L. Hvam, C. Thuesen, M. O. Grønvold, its applicability beyond manufacturing projects. M. H. Olsen, Modularization in the construction
Thirdly, we address RQ 3 by uncovering that the ap- industry using a top-down approach, Open Conplication of the PVM model in the construction indus- struction and Building Technology Journal 7 (2013) try diverges from its usage in industrial manufacturing. 88–98. doi:10.2174/1874836801307010088. The construction industry has diferent characteristics, [5] A. Kudsk, M. O. Grønvold, M. H. Olsen, L. Hvam, including modularity, digitalization, stakeholder depen- C. Thuesen, Stepwise modularization in the dencies, production processes, variability, production construction industry using a bottom-up apvolume, and product life-cycle. Consequently, applying proach, Open Construction and Building Techthe PVM model in the construction industry requires nology Journal 7 (2013) 99–107. doi:10.2174/ specialized approaches. It is essential to recognize these 1874836801307010099. diferences to efectively use the PVM model in the con- [6] A. Dufy, M. Andreasen, Enhancing the evolution of struction industry. design science, in: Proceedings of ICED, volume 95,
To conduct a comprehensive analysis of the research 1995, pp. 29–35. questions, we deliberately chose a smaller sample size. [7] J. K. Gershenson, G. J. Prasad, Y. Zhang, Product The observed variations among the three company cases modularity: Definitions and benefits, Journal of Enprovide further evidence supporting the generalizability gineering Design 14 (2003) 295–313. doi:10.1080/ of our findings. The validity of the research outcomes 0954482031000091068. is reinforced by the advanced stage of development of [8] A. Ericsson, G. Erixon, Controlling design variants: the configuration tools. However, in order to strengthen modular product platforms, Society of Manufacturthe framework even further, it would be recommended ing Engineers, 1999. to replicate and evaluate the proposed framework in ad- [9] B. Pine, Mass customisation: The new frontier in ditional cases. business competition, Harvard Business School
Furthermore, it is important to emphasize that Press, Boston, MA, USA (1993).
widespread use of the PVM model in the AEC indus- [