Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Automation in Construction 123 (2021) 103437

Contents lists available at ScienceDirect

Automation in Construction
journal homepage: www.elsevier.com/locate/autcon

Cross-phase product configurator for modular buildings using kit-of-parts


Jianpeng Cao a, *, David F. Bucher a, Daniel M. Hall a, Jerker Lessing b
a
Institute of Construction and Infrastructure Management, ETH Zurich, Stefano-Franscini-Platz 5, Zurich 8093, Switzerland
b
Department of Civil and Environmental Engineering, Stanford University, 450 Serra Mall, Stanford, CA 94305, United States

A R T I C L E I N F O A B S T R A C T

Keywords: One emerging strategy in industrialized construction is the use of mass customization to increase product effi­
Configurators ciency without sacrificing design flexibility. Effective implementation of mass customization can be done
Kit-of-parts through a product platform, i.e. a configurator. However, existing configurators often lack integration of
Mass customization
knowledge from the downstream supply chain such as manufacturing. This paper proposes a conceptual
Three-tier architecture
framework for a configurator unified through a manufacturing kit-of-parts. Kit-of-parts are pre-engineered and
pre-designed digital models representing fabrication-ready components. Using production rules and restraints,
accurate planning and design representations can be derived and utilized in the configurator. This research
develops a configurator prototype using three-tier architecture. The prototype supports a low-to-high level of
detail of kit-of-parts. It integrates product platforms and project development across multiple building phases
including site planning, floor plan layout, and 3D model generation. Finally, the implementation of a modular
building configurator illustrates the benefits of the proposed configurator-based workflow.

1. Introduction “modularize” a completed design that was originally intended for insitu
fabrication can require changes and rework to designs when received by
There is an increasing global demand for new buildings. At the same fabricators [6,7], communication inefficiency between different domain
time, the construction industry faces challenges of increasing construc­ designers [5], and material waste during production [8]. Furthermore,
tion costs, decreasing labor supply, and long development cycles. In traditional architecture design has held stereotypical views of modular
response, entrepreneurs and innovators in the construction industry are and volumetric buildings as dull, repetitive and unreliable. New indus­
exploring the use of industrialized construction. For example, in the trialized construction firms have taken care to avoid the negative
United States, industrialized construction firms raised over $1 billion of aesthetic perception of “modular”, “box-like”, or “cookie-cutter” archi­
investment capital in 2018 [1,2]. Industrialized construction uses pro­ tecture. These perceptions can prevent stakeholders such as architects
duction processes borrowed from the manufacturing industry [3]. from entering the market due to limitations to design flexibility.
Industrialized construction follows a design-manufacturing-assembly In response, industrialized construction firms are turning to the
approach to deliver new buildings comprised of prefabricated compo­ strategy of mass customization. Mass customization is a manufacturing
nents. These components can range from structural elements to volu­ paradigm that enables design flexibility in a mass production environ­
metric modules. This modular approach can meet the requirements of ment [9]. Mass customization emerges from the need to provide
many building types, ranging from residential, multi-family dwellings, customized products at the same prices, quality, and time as mass-
educational, and high rise [4]. New industrialized construction firms produced products. Hence standardization is one of the prerequisites
claim to offer more competitive pricing, expedited project schedules, of mass customization, and prefabrication would support mass cus­
and increased product quality. tomization in the construction industry [10]. Effective implementation
However, despite some success stories [1], there remains a steep of mass customization would enable design flexibility that aligns with
learning curve for designers and planners at the early stages of indus­ both customers’ preference and manufacturers’ capabilities.
trialized construction projects. Scholars note that industrialized con­ One emerging technology to support mass customization in indus­
struction requires earlier consideration and integration of fabrication trialized construction is the development and use of product config­
during the design and planning of the project [5–8]. Attempts to urators [11]. Product configurators were first initiated in the

* Corresponding author.
E-mail addresses: cao@ibi.baug.ethz.ch (J. Cao), bucher@ibi.baug.ethz.ch (D.F. Bucher), hall@ibi.baug.ethz.ch (D.M. Hall), jlessing@stanford.edu (J. Lessing).

https://doi.org/10.1016/j.autcon.2020.103437
Received 3 February 2020; Received in revised form 8 September 2020; Accepted 21 September 2020
Available online 5 January 2021
0926-5805/© 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
J. Cao et al. Automation in Construction 123 (2021) 103437

manufacturing industry, and now are widely applied in many domains efficiency, effectiveness and low costs. Mass customization maintains
such as electronics, automobile, and aerospace. A configurator eases the the main advantage of mass production, i.e. economies of scale. In
creation of complex systems by assembling a set of components – often addition, mass customization is enriched and complemented by theories
referred to as a kit-of-parts - from a wide range of alternatives to meet of modularization, product family architecture, and reconfigurable
the end-users’ needs and requirements. In this sense, the users of the manufacturing systems [23]. To achieve mass customization, config­
configurators in the manufacturing industry are the customers. Config­ urators are essential information and communication technology (ICT)
urators support customers in making rational decisions between various to provide product variety.
design alternatives offered by the company [12]. In addition, config­ Product configurators have various definitions from manufacturing
urators support the translation of customer specifications into product literature. Meyer and Lehnerd (1997) defined a configurator as “a
documentation such as a Bill-of-Materials (BOM) [13]. platform that contains a set of common components, modules, or parts
Kit-of-parts in the construction industry are “a collection of discrete (e.g. the kit-of-parts) from which a stream of derivative products can be
building components that are pre-engineered and designed to be efficiently developed and launched” [24]. Haug (2007) describes a
assembled in a variety of ways to define a finished building [14]”. Since configurator as “an expert system that supports the user in the creation
the kit-of-parts are modeled digitally to represent the actual components of product specifications by restricting how different components and
in production, it provides access to transfer rules and constraints, from properties may be combined” [25]. Bourke (2000) defines a configurator
the manufacturing to earlier planning and design activities. Such an as “a piece of software with logic capabilities to create, maintain, and
approach can enable designers to generate and evaluate a variety of use electronic product models that allow the definition of all possible
building layouts while retaining confidence that the final design can be product options and variation combinations, with a minimum of data
effectively manufactured and assembled using industrialized processes. entries” [26]. Three main characteristics of configurators can be
However, the application of configurators in construction is limited generalized from the above definitions: reusability due to kit-of-parts,
and immature. Although Building Information Model (BIM) can support intelligence driven by embedded expert knowledge, and high automa­
the collaborative process among different stakeholders - including tion realized by off-the-shelf technologies, such as Application Pro­
planners, architects, engineers and contractors, previous research on gramming Interface (API). In this paper, a configurator is defined as a
BIM-based platforms mainly focus on the model coordination [15], decision support system that automates the combination of kit-of-parts
visualization [16] and simulation [17,18]. Few studies give attention to into efficient modules for production under predefined rules [27]
the possibilities of developing a product platform based on BIM contents while enabling rapid generation of product variety to meet the desired
[19]. Most of the projects are built from scratch in the BIM environment. product features for the customer.
They do not benefit from a well-established kit-of-parts used and refined Configurators are able to support a various degree of mass custom­
by previous projects. Furthermore, automated approaches often termed ization. According to customer order specification decoupling points,
“generative design” typically focus on generation and optimization Hvam illustrates four product development processes that can map to
during a single project phase (e.g. generate the site layout, generate the four levels of mass customization [28]. Similar classification schema is
floor plan layout). Unlike product configurators, those studies generally given by C. Forza, who categorizes the mass customization based on
miss opportunities to link across project phases. Typically, generative what scope of customers’ needs are incorporated into the product
design approaches do not embed downstream manufacturing con­ development. “Engineer to order” process is adopted by companies
straints into the upstream design requirements [20]. They cannot span which design and produce complex products. At this level, only norms
from site layout to floor plan layout to 3D model generation to fabri­ and standards are embedded within the configurators. Customers’ needs
cation operations. are considered from the design process onwards [29]. “Modify to order”
To fill this gap, a BIM-based product platform (i.e. a configurator) is process is less complex. Products are designed based on predefined
needed to streamline the information flow between different stake­ generic product structures and set of rules. In this way, customers’ needs
holders and support the optimal decision making [21]. This paper pro­ are satisfied by a base design and certain variants what influence the
poses a kit-of-part-based configurator that can support integrated fabrication activities. “Configure to Order” process generates a product
design-to-production workflows in the setting of industrialized con­ via standard parts and modules. These usually can be reused from past
struction. To do this, we first review academic research and existing previous products. In this case, customer requirements are met by
commercial software applications to understand the current state-of-art assembling parts and modules. “Select variant” process is the most
for configurators in construction. From this, we develop three strategic restricted process. Customer needs are fulfilled by selecting products
typologies of configurators developed and implemented for industrial­ from finished product catalogs and various delivery approaches.
ized construction. We further identify the limitation of existing inte­ Determined by product specification processes, configurators enable
gration strategies. In this regard, we describe the conceptual framework different levels of freedom for the product design. With the delaying of
of configurators and how configuration process is implemented. Spe­ the customer order decoupling points (CODP), the design flexibility
cifically, we explain the concept of kit-of-parts in details and how it becomes more limited.
supports the design for manufacturing. Then, we adopt three-tier ar­ Furthermore, there are three implementation paths offered by con­
chitecture to develop a configurator prototype for module-based figurators [30]. They are 1) selecting scalable modules and module in­
industrialized buildings. The paper gives a detailed description of the terfaces for Configure to Order products, 2) editing design parameters
configurator implementation, followed by an illustrative example of a for Modify to Order products, 3) define design and manufacturing rules
modular construction project to provide a discussion of the config­ for Engineer to Order products. Wikberg et al. developed a configuration
urator’s effectiveness. The paper concludes with limitations and sug­ platform for Select variant products [31]. The platform enables selecting
gested directions for future research.

2. Literature review Table 1


Implementation paths available to configurator users under various CODP
scenarios.
2.1. Configurators for mass customization
CODP Implementation path
In the late 1980s, due to increased demand for product variety, the Engineer to Order Define design and manufacturing rules
manufacturing industry transitioned from mass production to mass Modify to Order Edit design parameters
customization [22]. Mass customization is a widely adopted business Configure to Order Select scalable modules and module interfaces
Select Variant Select off-the-shelf products
strategy that identifies and fulfills customer needs without sacrificing

2
J. Cao et al. Automation in Construction 123 (2021) 103437

off-the-shelf products. Table 1 presents the relationship between stakeholders from real estate, architecture, site planning, and future
customer order decoupling points (CODP) of configurators and typical building owners (i.e. the customer). Engineers, contractors and fabri­
implementation paths. cators typically do not participate in this step of the process. The tar­
geted products could be a conceptual building model created by
2.2. Configurators in construction projects architectural firms for early planning. The major objective of applying
the configurators is to generate diverse development plans, such as site
In the construction industry, buildings are one typical type of prod­ plans, building profiles, and floor plans. These development plans are
uct. Unlike products in the manufacturing industry, buildings combine evaluated by decision-makers in order to make a final selection. Plan
both product-level information and project-level information [32]. The representations only require graphic illustrations with low-level of
product-level information includes prefabricated modules’ design di­ detail, such as simple geometries and wireframes. Compared with a
mensions, engineering attributes, and factory production processes. The traditional design strategy, the reviewed literature claims two advan­
project-level information is typically ad-hoc, including site planning, tages brought by type 1 configurators. First, real estate developers allow
site properties, site-built elements, and on-site activities. Thus, a con­ customers to interactively design their desired houses by offering
figurator suitable for the construction industry should support both available selections. For example, Veloso et al. developed a design
product-level and project-level information management. For example, customization system, which navigates the customers through the pro­
the firm Project Frog developed a configurator for timber panelized cess of planning their desired apartment, selecting from available
structures, shown in Fig. 1. The Project Frog configurator is built upon a apartment types and floor plans [35]. Second, intelligent algorithms can
library of panel products (i.e. product-level information) coupled with identify more feasible or advantageous plans using embedded configu­
rules from local building regulations (i.e. project-level information) [2]. ration rules. For instance, the configurator, created by Bianconi et al.,
can guide architects to select the fittest configurations of single-family
2.3. Three typologies of construction configurators houses considering multi-performance criteria, including cost, energy
consumption, and aesthetic expression [36].
To understand the current development and application of config­ Typology 2 represents configurators supporting the project design
urators in the construction industry, this paper begins with a review and phase. This is the most widely applied situation for configurator appli­
summary of configurator prototypes from scholars and applications in cations in the market. In this scenario, architects, engineers, and
industry. We first identify academic configurator prototypes through a sometimes fabricators are involved and collaborate on configurators.
literature review. Because the term “configurator” is not widely adopted The targeted products could be a detailed building model configured
in AEC literature, the literature search includes other key terms such as with predefined modules, such as timber panels. The major difference
platform and design system. The literature is then narrowed to only between typology 1 and 2 lays in the generated outputs. For typology 2
include configurators for construction, with a specific focus on indus­ configurators, the low-level detail web plans are converted to BIM
trialized construction (otherwise referred to as prefabrication, pre­ models in authorized design applications. These BIM models represent
assembly, modularization, and/or off-site fabrication). Specific IT tools the starting point upon which designers and engineers could make
developed in line with the definition of configurators given above are further development, such as engineering analysis. For example, the
included in the review. Project Frog web-based configurator allows users to send the validated
The survey of commercial applications for configurators is done design solutions to Autodesk Revit, and generate the 3D model via Dy­
using four steps. Firstly, we explore the industry executive summary on namo scripting. However, the connection between the web environment
technology and innovation [33] which lists top IT tools applied in the and desktop applications is poorly realized. Previous researchers
construction industry. Secondly, as configurators are emerging tech­ pointed out some relevant obstacles, such as data loss and corruption
nology, we conduct a survey on the AEC startups website,1 and set the [15] and cross-platform model visualization [37]. Except for technical
tag as “Generative” and “Design Automation” which are major features aspects [38], organizational requirements on platform development also
of configurators. Thirdly, same as for the academic literature, we narrow need further studies [39].
the list to include only configurators for construction, with a specific Typology 3 is less studied and achieved then the other two typol­
focus on industrialized construction. Finally, we analyze the trial or ogies. It represents configurators that support construction projects at
demo version of these existing commercial configurators. Table 2 lists project production phase. In this context, the typical products can be
the studied commercial applications. buildings planned, designed, produced and assembled by a vertically
After identification of configurators for construction, an analysis is integrated firm, such as the business model used by Katerra and BokLok
conducted across five dimensions: [2]. Related research work has been done in the design for
manufacturing and assembly (DfMA) domain. Engineers and fabricators
1) Who are the main users of the proposed configurator? (who?) usually adopt DfMA principles in the early design stage before the
2) When are configurators used in a project? (when?) manufacturing process. For example, to achieve mass customization,
3) How is configuration performed / what are the key activities? (how?) Said et al. presented a design optimization model to manage the trade-
4) What are the typical targeted products suitable for each typology? off between geometry commonality and a variety of panels [40]. Yuan
(what targets?) et al. integrate manufacturing and assembly constraints in the design
5) What are the expected outputs generated by configurators? (what phase by adopting a DfMA parametric design [5]. The outputs of the
outputs?) configurators include 3D building information models, along with
related data and documentation such as G-codes for NC machines,
From this process, we identify three distinct typologies of existing permit drawings, and bill of materials (BOMs) [41], among which
construction configurators (see Fig. 2). For a more detailed description drawing generation received the highest attention. For example, Deng
of data selection and analysis process underlying this review, please et al. proposed a framework to generate fabrication drawings from 3D
refer to Cao and Hall (2019) [34]. façade models by SolidWorks APIs [42].
Typology 1 represents configurators for project development support The three configurator typologies focus on planning, design and
the project planning phase. Examples of this type of configurators on the production separately. This separation enables greater flexibility for the
market are SpaceMaker and Archistar. Typically, this stage involves first two typologies. However, users of the first two typologies are not
given evaluation on the manufacturability of their decisions. It remains
difficult to implement industrialized construction solutions using these
1
(https://www.aecstartups.com/#/). configurators.

3
J. Cao et al. Automation in Construction 123 (2021) 103437

Fig. 1. Project Frog configurator for timber-panelized buildings.

Product Variant Master (PVM) to map product variants in product


Table 2
families from customer, engineering and production point of view
Commercial configurators.
[46,47]. Ramaji et al. proposed a Product Architecture Model (PAM) to
Commercial configurators URL of websites Creation year integrate the hierarchy and relationships among components of modular
Hypar https://hypar.io/ 2019 buildings [48]. Two strategies share some similarities. First, the overall
PRISM https://www.prism-app.io/ 2019 product architecture uses a hierarchical determination for sub-
SpaceMaker https://spacemaker.ai/ 2019
assemblies and parts. In addition, the components and related attri­
Archistar https://archistar.ai/ 2018
Testfit https://blog.testfit.io/ 2017
butes are categorized based on the level of details in PAM and model
Project Frog http://myprojectfrog.com/ 2016 views in PVM. Compared with PAM, PVM is more competitive in
capturing the product variations. However, PVM is not effective to
handle information transmission across multi-phases. This gap in the
An increasing number of industrial configurators are trying to offer knowledge is aimed to fill with a configurator framework based on kit-
an integrated approach that combines across typologies. For example, of-parts and configuration rules described below. Kit-of-parts maintain a
the Hypar, Testfit and PRISM configurators integrate across the planning single source of truth with different model views. The proposed frame­
and design process seamlessly. Project Frog configurator is able to pro­ work describe below can be used to configure a product from planning to
vide design and fabrication deliverables together. With the application manufacturing without data loss.
of such integrated configurators, the reuse of process and technical so­
lutions and the formation of a stable supply chain could be facilitated 3. A proposed conceptual framework for configurators
[43]. In comparison to configurators that fit within a single typology, an
integrated configurator enables a longitudinal continuity that enables Conceptually, a construction configurator contains two pillars. They
stakeholders to maintain a common environment to control the project are the kit-of-parts product structure and the configuration rule engine
data. However, few studies explore the appropriate theoretical founda­ (shown in Fig. 3). The kit-of-parts product structure are designed to
tion for supporting the integration of single-phase configurators. provide a) reusability for a series of building products [50]; b) flexibility
for a variety of design solutions (configurations); and c) integration with
2.4. Theoretical foundation for integrated configurators processes metadata [6]. Typical kit-of-parts are linear components (e.g.
prefabricated beams), planar components (e.g. panels), and/or volu­
The integration across phases requires knowledge transferred up­ metric components (e.g. modules [14]).
stream from engineering, production and assembly processes to design Configuration rules specify how the kit-of-parts and their properties
and planning processes. Previous research is to collect knowledge from can be defined to meet a given requirement from users. We propose four
experienced professionals and encode it into rules. However, one of the types of configuration rules: composition, compatibility, dependency
major shortcomings is that the rule sets are often too large and and cardinality [51]. Composition rules define which kit-of-parts are
complicated in terms of a number of parameters involved in the design mandatory or optional in the product architecture. The product archi­
process [44]. To reduce the complexity caused by large volumes of pa­ tecture can be built as a tree structure. A kit-of-parts can be either a root
rameters, modularity theory is used to define a set of components with node, an intermediate node or a leaf node. Once a non-leaf kit-of-parts is
the specialized interface as base units. Jaganathan et al. proposed a selected, the direct child kit-of-parts can be automatically added to the
design-manufacturing integrated model with the core task of decom­ product. Compatibility rules define which kit-of-parts cannot exist
position. The concept of decomposition is to yield a set of building el­ simultaneously in the product. If a kit-of-part with a compatibility rule is
ements matched with manufacturing requirements [45]. However, this selected, the system will exclude certain kit-of-parts from the option
top-down approach requires mapping design features with lists. Dependency rules define which kit-of-parts must belong together in
manufacturing parts, then checking against the requirements with extra a product. If a kit-of-part with a dependency rule is selected, the system
efforts. If any requirement is not satisfied, the decomposition operation will add its mated parts to the product. Cardinality rules define the
loops back, and related design needs to be revised. required or limited number of kit-of-parts under certain circumstances.
Another approach makes use of abundant available manufacturing The product structure can only represent the existence of a kit-of-parts.
parts and assemblies, developed based on modularization theory, to A kit-of-part with a cardinality rule need to be assigned the quantity
assemble a product, namely bottom-up approach. Mortensen et al. used occurring in the product and checked against its validity.

4
J. Cao et al. Automation in Construction 123 (2021) 103437

Fig. 2. Three typologies of configurators.

Fig. 3. The structure of product configuration model in the manufacturing industry (adapted from [49]).

5
J. Cao et al. Automation in Construction 123 (2021) 103437

During the process of configuration, the typical user inputs can Table 3
include the project information (e.g. location), market demand (e.g. the Planning parts and design parts of volumetric modules.
number of different types of suites), and technical requirements (e.g. Composition Metadata in Metadata in kit-of-
structural stability factors). The appropriate kit-of-parts are then conventional parts parts
selected by performing inference based on the configuration rules and Planning Volumetric Module shape, size Module shape, size
these user inputs. part modules and dimension and dimension
From a technical perspective, the configuration process is a Module combination Module combination
sequential decision-making process. Previous research has studied layout layout
Module weight
sequence order definition in order to solve specific configuration tasks Permissible size
[52]. Furthermore, there are various approaches to the design of the variations
sequential decision-making process. For example, the configuration Interior arrangement
process of a façade system follows the design development process, from Thermal performance
Accurate cost
the building structure, panels to support systems [53]. In the automotive
Design Load bearing Material properties Material properties
manufacturing industry, the configuration process is ordered from the part side walls
product type, the product family, the product models, and feature Non-load Part shape, size and Part shape, size and
families and variants [54]. The optimal sequential decision-making bearing walls dimension dimension
process to order the configuration process for buildings is an impor­ Floors Connection types Connection types
Ceilings Thermal performance Thermal performance
tant research topic but remains outside the scope of this study. Beams and posts Permissible opening
In this research, the sequential decision-making process is adapted variations
from the product development sector. For mass-customization, the Bracings Part connection details
product development sector often conducts a survey on product features Openings Fabrication deviations
MEP systems Accurate cost
and assigns product features sequentially [31]. This feature assignment
is done by kit-of-parts selection and their attributes selection. Upon
selecting a kit-of-parts or attributes, the configurator activates the MEP system variations, with the aim to incorporate them within
attached rules to validate the current product structure. If satisfied, the different design stages. We first explain the variations in detail, and then
sequence proceeds to the next state. The process consists of a sequence of illustrate how those variations could be mapped to the configuration
product states representing selected kit-of-parts and associated config­ process stated in Section 3.
uration rules. At each state, a constraint satisfaction problem (CSP) is
solved. The process will terminate when all product features are deter­
mined. This enables a series of diverse configurations. Such configura­ 4.1. Size variations
tion replaces the traditional point-by-point based design with a set-
based design [55]. It can allow the project team to maximize flexi­ The size of a module refers to its height, width and length. In
bility and efficiency. traditional design workflow, the size of the module is determined by the
architects to fit spatial requirements. However, this ideology might
4. Development of kit-of-parts library generate too many diverse sizes of modules, which cause a heavy burden
on production efficiency. Kit-of-parts used in industrialized construction
From the product composition perspective, the kit-of-parts can be is reusable for multiple projects with similar site conditions. However,
composed of individual parts or a pre-designated assembly consisting of the fixed-size might lead to the limited application of the multifunc­
other components [56]. Considering kit-of-parts contain the entire life- tional module. For example, a patient room module has different size
cycle product information, a simplified representation of kit-of-parts is requirement compared to a residential room module. In this sense, it is
required during the planning and design stages, namely planning kit-of- worthy to inform kit-of-part designers that the permissible size varia­
parts and design kit-of-parts. Both can be derived from their tions based on rules generalized from downstream activities. Typical
manufacturing counterparts and maintain consistency with each other. rule sets dominating the size of modules include building system
In other words, if a manufacturing part is updated, the related planning selected and transport restrictions. For example, standard dimensions of
part and design part will change correspondingly. Planning kit-of-part is a timber module are in the range of 3 m in width and 8 m in length and
an unstructured and generic expression of kit-of-parts, such as a simple 3.2 m high, although maximum dimensions are 6 m in width, 20 m in
2D or 3D geometry, used in the typology 1 configurator. Typical plan­ length and 3.7 m in height [58].
ning kit-of-parts include room modules which could be configured with
variations relevant to customers’ requirements, such as different floor
plans, the shape of roofs and vertical building layouts. Design kit-of-part 4.2. Opening variations
is a building information model suitable for architectural and engi­
neering analysis in the typology 2 configurator. Typical design kit-of- An opening in a module refers to a door, a window or an open side.
parts could be architectural elements, structural elements and MEP el­ The variations of the opening size enable great flexibility for architects
ements. Design kit-of-parts contain variants corresponding to architects’ to fit specific requirements. For example, a room which requires open­
and engineers’ requirements, such as different insulating material, sec­ ness and connectivity to other space could be built via open-ended
tions of beams and piping sizes. modules. By contrast, facility rooms or bathrooms, in most cases, are
Most importantly, planning or design parts should not only be seen as more closed with four-sided modules, with windows and doors in their
a low level-of-detail (LOD) view of their manufacturing counterparts. ends. The size of the opening has a large effect on the building perfor­
They are well-defined representation which incorporates useful mance, such as daylighting, especially for the modules at the periphery
manufacturing and engineering information during the early design of buildings. Due to those requirements, a set of rules from manufac­
stage, such as permissible size variations. The information is then turers should be provided on the permissible opening variations. Typical
leveraged in the configuration process to offer users more accurate in­ rule sets govern the opening size are generated from the structural en­
sights on available product features. This enables a faster and easier way gineering rules. A module with a large opening requirement could be
to work with clients and provide accurate cost estimates [57]. Table 3 sustained by corner post and edge beam supporting systems. For mod­
takes a volumetric module as an example. We select three types of in­ ules longer than 7.5 m, a deeper edge beam supported by intermediate
formation including module sizes variations, opening variations and posts is usually required [59].

6
J. Cao et al. Automation in Construction 123 (2021) 103437

4.3. MEP system variations in various configurations to create apartment buildings. The project
modular architecture is shown in Fig. 5. Each module is fully designed
The thermal and ventilation performance depend on the MEP sys­ for certain functionalities. For example, the living room module would
tems. The variations of MEP systems are driven by the functionality of be installed with MEP systems included. The transfer module is installed
the room decided in the planning phase. For example, a conference room with elevators and staircases. While the product concept is standardized,
module holding more people than a standard office room requires a the individual configuration can vary from the building profile on the
higher volume of air circulation. This leads to a larger size of the ducting selected site, the layout of the apartment on the floor plan, the material
system, which needs a larger pre-cut opening on the panels or a larger type of façade, the roof types and the staircase types. The project-specific
area for service risers. This information is important for manufacturers configuration is supported by a library of 3D models in Autodesk Revit,
who are able to provide modules installed with MEP systems. In the 2D engineering drawings in AutoCAD, and technical documents that
conventional design‑lead-production approach, manufactures have to define all relevant design regulations.
wait until the completion of the design, leading to longer project dura­
tion and slow responsiveness to the design iterations. To overcome these 5.2. System architecture
limitations, manufacturers could offer a set of rules for clients to
consider permissible MEP system variations. Existing rules manage the The development process starts with the design of platform archi­
variations by arranging the duct layout, position and dimensions. tecture. Similar to other web-based applications, the configurator in this
Depending on the size of ducts, ducts could be placed inside the panel, in paper can be developed based on a three-tier architecture. Examples of
a separated service riser, or embed into the skirting cover inside the tier-based systems attempt to enable an organization to use data across a
module. broad set of applications, such as workflow management, model man­
This section gives an example of how a room module can be agement, and model visualization and storage [16,60]. The proposed
configured with regard to opening variations. During the antecedent configurator includes a Presentation tier, Application tier and Data tier.
configuration step, the size of the module is defined in terms of heights, This is also called the frontend, backend and database, respectively. The
widths and lengths. The module ceiling, wall and floor use the default presentation tier consists of a graphical user interface (GUI), which re­
configuration with known design parameters (e.g. reflectance). At this ceives input and displays the different model views and data to the users.
phase, the size of the window is configured to analyze daylighting usage. Through API requests, the presentation tier calls the application tier to
The configured variables include the width and height of the window on carry out specific data provided by the core functionalities of config­
one of the end walls (shown in Fig. 4). Typical user inputs include the urators. Then, the application tier accesses the data tier, which stores the
room functionality (e.g. study room). Then based on the empirical raw files and more fine-grained data, such as key-value pairs [61] in the
knowledge, the depth of the daylight zone is calculated. Upon the configurators. This separation between data and applications enables
window type is selected, the daylight feasibility factor can be rapidly much more flexibility for development teams to upgrade, add or replace
estimated for potential energy savings. The result will further be components in each tier, instead of rewriting the entire platform [62].
included in the energy consumption. Although this research only lists Additionally, with a common data-tier across a broad set of applications,
three variables for ease of illustration, the proposed concept is exten­ data are easily accessible and shared for multiple users collaborating on
sible. A number of product features can be extracted from abundant kit- the same project. Fig. 6 shows the logical architecture of the proposed
of-parts and propagated from the planning stage to the manufacturing configurator.
stage. Based on the previous description of the framework for a web-based
product configuration platform, a more detailed implementation of a
5. Implementation prototype is discussed in the following section. Since the platform ar­
chitecture is divided into three parts, it is necessary to implement them
5.1. Project description with suitable technical solutions. In addition, the system should function
with minimal system requirements. It can be assumed that users use
For a better understanding of the proposed configurator, a multi- modern browsers (>IE11) and mostly work with Windows 7+ or similar
story modular residential project from a European construction com­ operating systems.
pany is used as an illustrative example. The project consists of pre­
fabricated, highly standardized volumetric modules that are combined

Fig. 4. An example of configuration on opening variations.

7
J. Cao et al. Automation in Construction 123 (2021) 103437

Fig. 5. Modular architectural of multi-story residential projects.

Fig. 6. Three-tier architecture of proposed configurator.

5.2.1. Presentation tier different possibilities of technical implementation. Due to the good
For the technical implementation of the frontend or presentation scalability and the similarity to Angular, Nest.js is utilized for the
tier, there are countless web frameworks. Each framework has its ad­ backend. Unlike Express, which is one of the most popular frameworks
vantages and disadvantages. However, the framework should support for backend development, Nest.js makes it possible to take a modular
the Model-view-view-model (MVVM) pattern and web components. approach with a given architecture. This allows database access and
Therefore, Angular is chosen as the web framework. The Framework business logic to be separated from the processing of API requests.
Angular is developed based on Typescript. Through modern software
architecture approaches and necessary flexibility, Angular enables a 5.2.3. Data tier
component-based structure which is suitable for a process-separated The database or data tier is the central structure of the platform. It is
platform. By separating view and logic or functionality, the applica­ of the utmost importance that data can be stored correctly and in a
tion can be divided in such a way that it can be maintained, tested and proper representation in order to guarantee consistency and correctness.
expanded. This allows developer paradigms such as single-responsibility Data are a representation of a planned or existing system. In order to
and separation-of-concerns to be adhered to. model or represent a system, it is necessary to depict it as a duality of
form and function. The form is the content of the system so what it
5.2.2. Application tier consists of, and the function represents what the system does or what
Considering the application tier, or the backend, there are again happens over time. Therefore, the database is set up in such a way that

8
J. Cao et al. Automation in Construction 123 (2021) 103437

objects are stored separately and linked by relationships. This was made the-shelf GIS service [63,64]. Geographical data are useful for plan­
feasible by the use of MongoDB, a document-oriented NoSQL database ners to find the appropriate location and set the site boundary. Building
connected to the backend. profiles include building shape, height, stories, and orientation. It is
The entire structure is based on the so-called REST architecture determined by the site-level configuration engine. Site-level configura­
which stands as an acronym for Representational State Transfer and is a tion starts with the customized sketched site boundary and user-defined
set of design criteria and not the physical structure. In REST-based ar­ floor area ratio. Then the building total floor area is calculated. In the
chitecture, everything is a resource, on which it is possible to perform an prototype, the available customization for building shapes includes
operation. The data can be stored in a database on the backend-side, for rectangular, “L-shape”, “U-shape”, and “T-shape”. An undefined shape
example, a user data that has a unique identifier (UID). The main data in can also be sketched by the user. After a certain shape is selected, the
the configurator can be categorized into three types: (1) 3D models from height and stories can be defined with user inputs. Buildings are
authorized BIM applications; (2) add-ons for design automation; (3) generated and interactively placed on the canvas in the form of web
other non-structured data (e.g. drawings, regulations, log). These re­ geometries. Related configuration constraints, including height limita­
sources are accessed via a common interface based on HTTP standard tion, allowable distance to the site boundary, and minimal distance
methods. Since the platform has to be able to handle data exchange and between adjacent buildings, are embedded in the system and checked
its modification, it is useful to implement the entire system architecture automatically. The metadata, including the site data and building
on the basis of REST. Through the use of the REST API interface, HTTP properties are stored in the database.
requests from the frontend can be defined on the backend to execute
specific functionalities. 5.3.2. Floor planning
The floor planning is done by specifying the number of different type
5.3. Function design of suites and placing them on each floor (see Fig. 8). The suites are
different in size, which can be one person’s use, two-occupant use, and
The application tier processes all main functionalities of config­ three-occupant use. The inputs can be predefined building shape and the
urators. Each function should be programmed as a reusable and modular number of different types of suites required by clients. On the floor view,
component. Here, a component is a software “unit”, which could be a the grid system is automatically generated and can be dynamically
block of reusable code or an independent application. This supports the adjusted. The system is efficient for placing elements in a layout. For
customized combination of multi-functionalities required by different example, volumetric elements (e.g. suites) can be put between two rows
customers. This research mainly paid attention to the design-related and two columns grid lines. Floor level configuration determines the
tasks, starting from site layout, floor plan generation, to 3D modeling. selection and placement of planning parts. On this phase, available parts
More application services from third parties, such as data analysis, could consist of suites and transfer modules which are low level of details
be added to the configurators. (LOD) web geometries. After a certain part is placed into the grid system,
its relative locations are retrieved. Related rules, such as excavation
5.3.1. Site layout requirements, horizontal stability, are checked in terms of the quantity
The site layout component is responsible for the project planning and location of units. For example, a cardinality rule is encoded to
stage (see Fig. 7) The main functions can be divided into geographical checking the horizontal stability of the structure. According to the
data collection and building profile generation. Based on the selected company’s design guide, a two-story modular project is developed for
site location, geographical data, such as terrain, is accessible from off- wind loads in terrain type 1, which requires at least five modules in

Fig. 7. Site layout interface

9
J. Cao et al. Automation in Construction 123 (2021) 103437

Fig. 8. Floor planning interface with apartment assemblies (e.g. ROK 1).

width per floor. catalogs, the various rules violated, and the estimated final cost. For
example, a recommender model trained on a batch of used modules
5.3.3. Model generation could guide users on the module selection by offering a ranked list of
The model generation component converts the web geometries of a objects within each category [66]. Such use of the tool can help sales
building into building information models (see Fig. 9(a)–(b)). At this teams recommend preferred products based on certain regional or cul­
point, the configuration of the individual suite is initiated by designers. tural contexts and/or help users without sufficient product expertise to
For example, the suite in the project could have several variations, solve a possible selection dilemma.
including module combinations, façade types, wall types, balcony types,
window and door types. Related rules are used to check the valid 6. Discussion
combination of modules according to the company’s sales strategy. For
example, if a user is configuring a two-occupant suite, the encoded The proposed configurator launches a new project development
compatibility rule will infer the available living modules and matched process using a kit-of-parts library that incorporates the knowledge of
sleeping modules with regard to the size limitations. After all variations manufacturing. Traditional project development using a parametric
have been specified, a completed and validated BIM or CAD can be design process begins first with geometric representation and then de­
generated out of the configurator, instead of refining through a sequence tails second the part design including manufacturing constraints. By
of design and engineering activities. To do this, the site data, building comparison, the kit-of-parts represent fabrication-ready building prod­
properties and floor plan data are read from the database and set as the ucts. From these products, augmented design models with a detailed kit-
inputs on this stage. Based on the identifiers of used planning parts, the of-parts can be derived. Then, manufacturable variations of products can
corresponding high LOD models are pulled from the product database. be Modified to Order by designers to fit their design intent.
Finally, the retrieved models are imported into a new project file and are The above work emphasizes the opportunity for a manufacturing-led
placed with reference to the location inputs. The last step of imple­ industrialization of the supply chain. Past scholarship has suggested the
mentation is scripted as a reusable add-on package provided on the web need to develop prefabricated construction building information models
environment. containing the production details [6]. Similar research on Design for
Manufacturing and Assembly (DFMA) also highlights the importance of
5.3.4. Data analysis considering the knowledge of manufacturing in the design stage. Past
The proposed three-tier configurator can be scaled up with exten­ research emphasizes the importance of a close collaboration between
sions from third parties, such as Amazon Web Service (AWS). This designers and manufacturers, such as a DFMA design team [5]. How­
research illustrates how a potential valuable service such as data anal­ ever, these approaches have been slow to develop. It is difficult for de­
ysis can be added to the application tier. Users’ feedback is critical for signers to initiate the models because they do not have sufficient
increased standardization in design, manufacturing and integrating the knowledge on production. More importantly, some design issues can
supply chain of kit-of-parts [65]. After a configuration is finalized, the only be discovered late in the production process, putting pressure on
users’ entry data are saved as a log file. The main function of the data the supply chain to deliver. It also requires extra reorganization efforts
analysis tool is to process the log data and find the users’ behavior and expert resources from the design company.
patterns. This enables the configurator to be more responsive and pro­ In comparison, the configurator and kit-of-parts approach presented
actively to provide design suggestions to a particular user group. The in this paper requires more efforts from manufacturers to develop
typical log data might include the modules selected from different products (kit-of-parts) and tools (configurators) to help integrate the

10
J. Cao et al. Automation in Construction 123 (2021) 103437

Fig. 9. (a). an apartment configuration via volumetric modules (e.g. K4); (b) Model generation interface

design-to-production supply chain. Using the proposed approach, when investment made initially inside a single project [30]. As a result, the
a derivative building product is created by configurators, the design application of kit-of-parts library can augment the role of manufacturers
team can be confident that deliverables are within the production ca­ in the industrialized construction.
pacity. The cost can be controlled without much deviation caused by The three-tier architecture used in this configurator prototype can
redesign. While manufacturers must invest in developing a kit-of-parts help advance scholarship on product configurators in AEC. Past schol­
library, they also can gain the rewards by streamlining the supply arship has proposed using native BIM tools as generic configurators [53]
chain in advance. In addition, the reuse of the library in future projects where product configuration in BIM is implemented by assigning values
will lead to continuous improvement of project quality and return of to customizable parameters [67]. By contrast, the tier-based

11
J. Cao et al. Automation in Construction 123 (2021) 103437

architecture decouples the application and data distribution. This The research has certain limitations as follows:
feature guarantees that the data source of kit-of-parts is mastered or
edited in only one place – the data tier. The application tier can then 1) To use such a configurator, the research assumes that a company has
access the data tier to implement core functionalities of the configurator, built a library of reusable modules for project development. This is
including site layout, floor planning, and model generation. As a result, the main difference between the traditional CAD systems, which start
the presentation tier displays the different model views to the users. The modeling activity from scratch, with the proposed configurator,
tier-based architecture allows the creation of web-based design tools which consumes the existing kit-of-parts to configure new derivative
with a user-friendly interface; design configuration can occur outside of products.
proprietary BIM software such as Autodesk Revit. Then, the application 2) In the illustrative project, the engineering rules and constraints are
is able to manage the kit-of-parts modeled in BIM tools, and use them for assumed to be collected from an industrialized construction com­
additional services, such as Bill-of-Material (BOM) analysis. Such shift pany’s architectural design guidelines, which does not typically
can enable configuration lifecycle management built on BIM and extend include complex design analysis, such as seismic analysis.
the application of BIM across all lifecycle phases of a product. 3) The configurator prototype does not include the entire project pro­
From the practice point-of-view, the benefits of the illustrated duction phase. Relevant functionalities – such as automated drawing
example can be categorized into three main factors: time, cost, and generation and bill-of-materials calculations – need to be extended.
quality [68]. Simpson summarized the methods to quantify the benefits 4) The tier-based architecture of the configurator faces a problem of
and drawbacks of platform-based product development, such as the poor scalability. The structure of the application is decided and fixed
reduction in development time [69]. However, future research is needed during the initial stage of development [73]. Hence, it is time-
to test if such benefits apply to construction and how they should be consuming and error-prone to extend new features and functional­
quantitatively measured. The illustrated example also provides a start­ ities to the application. Additionally, due to the inter-module
ing point to understand how an integrated configurator enables more dependence, the entire system is not robust and may collapse due
benefits than standalone configurators [70], but more research is to a single error occurred in an individual module.
required. In addition, it is important to qualify that the successful
application of the integrated configurator does not only rely on the Future research should investigate additional questions such as:
maturity of IT systems. It also depends on the maturity of the organi­
zation, such as the cross-organizational collaboration [70] and supply 1) How can planning kit-of-parts and design kit-of-part be derived
chain ties. Because the application of configurators in the construction automatically from manufacturing counterparts, and at the same
industry is still in the early stages, we do not consider yet the influence of time maintain consistency with each other? Considering those kit-of-
the maturity of the organization to the configuration application. parts might be edited in different applications, a flexible data
structure needs to be built and set as a common environment.
7. Conclusions 2) How can a kit-of-parts be built based on historical project data?
Previous studies suggested that the modularization strategies can be
Industrialized construction is gaining more share in the construction used to group components for offsite fabrication to ensure the ease of
market. More stakeholders adopt this approach by delivering their assembly and flexibility of building maintenance [74]. Yet, few re­
products from design to manufacturing and assembly. Configurators are searchers consider the reusability of building components as a
important IT systems that support mass customization and provide a module driver to develop kit-of-parts.
common environment to integrate data, functions and processes during 3) How can the design space of modular buildings be enlarged? Existing
the project life cycle. The benefits of applying configurators in con­ configurators are targeted at solving the selection of kit-of-parts and
struction projects have been illustrated in previous research [71,72] but the assignment of adaptive parameter values [75]. Few research
have limited by an integrated approach supporting project design-to- studies explore the configuration regarding the flexible spatial
production. To enable further benefits of configurators and enable arrangement of components to achieve design variations.
adoption by greater segments of the industry, this research proposes a
kit-of-part based configurator for industrialized construction.
Declaration of Competing Interest
This paper attempts to make several contributions to the literature.
First, it generalizes three typologies of configurators by clarifying when,
The authors declare that they have no known competing financial
how and for whom the configurators are developed for. This can assist in
interests or personal relationships that could have appeared to influence
guiding the selection of configurators for different stakeholders. Second,
the work reported in this paper.
the conceptual framework for developing a configurator is pointed out
and the configuration process is explained. The framework can be used
for industrialized construction companies looking to automize their References
configuration process for an existing product platform that needs to be
[1] T. Pullen, D.M. Hall, J. Lessing, White Paper: A Preliminary Overview of Emerging
configured manually. It is also useful for non-industrialized construction Trends for Industrialized Construction in the United States, 2019, pp. 1–24,
company looking to develop a configurator and kit of parts as a starting https://doi.org/10.3929/ethz-b-000331901.
point for industrialization. Following a set-based design approach, kit- [2] D.M. Hall, J.K. Whyte, J. Lessing, Mirror-breaking strategies to enable digital
manufacturing in Silicon Valley construction firms: a comparative case study,
of-parts can be easily configured to diverse design solutions in config­ Constr. Manag. Econ. 0 (2019) 1–18, https://doi.org/10.1080/
urators. Third, the paper proposes a kit-of-part based configurator built 01446193.2019.1656814.
upon a three-tier architecture and web technologies. Considering that [3] D.M. Gann, Construction as a manufacturing process? Similarities and differences
between industrialized housing and car production in Japan, Constr. Manag. Econ.
previous studies on collaborative platform mainly focus on the model 14 (1996) 437–450, https://doi.org/10.1080/014461996373304.
coordination, visualization and simulation, the configurator in this [4] J.M. Schoenborn, A Case Study Approach to Identifying the Constraints and
research first handles design-related tasks, starting from site layout, Barriers to Design Innovation for Modular Construction, Virginia Polytechnic
Institute and State University, 2012, pp. 19–32. URL: http://hdl.handle.net/10
floor planning to 3D modeling, as a continuous and seamless process on
919/32397 (Accessed date: 4 March 2020).
a light-weighted, easy-sharing web environment. Fourth, the proposed [5] Z. Yuan, C. Sun, Y. Wang, Design for manufacture and assembly-oriented
configurator is prototyped via a multi-story modular apartment project parametric design of prefabricated buildings, Autom. Constr. 88 (2018) 13–22,
that demonstrates a seamless workflow among multi-stakeholders. This https://doi.org/10.1016/j.autcon.2017.12.021.
[6] M. Hamid, O. Tolba, A. El Antably, BIM semantics for digital fabrication: a
example illuminates several promised improvements on the project knowledge-based approach, Autom. Constr. 91 (2018) 62–82, https://doi.org/
planning to manufacturing processes. 10.1016/j.autcon.2018.02.031.

12
J. Cao et al. Automation in Construction 123 (2021) 103437

[7] Justin K.W. Yeoh, Runxing Jiao, Ontology-based framework for checking the Tech%20%26%20Innovation%20Survey_Final%20%281%29.pdf, 2019 (Accessed
constructability of concrete volumetric construction submodules from BIM, in: date: 4 March 2020).
Computing in Civil Engineering 2019: Visualization, Information Modeling, and [34] J. Cao, D. Hall, An overview of configurations for industralized construction:
Simulation, American Society of Civil Engineers, Reston, VA, 2019, pp. 279–285, typologies, customer requirements, and technical approaches, in: Proceedings of
https://doi.org/10.1061/9780784482421.036. the 2019 European Conference for Computing in Construction 1, 2019,
[8] P. Manu, L. Mahdjoubi, A.-Q. Gbadamosi, L.O. Oyedele, C. Aigbavboa, O. pp. 295–303, https://doi.org/10.35490/ec3.2019.145.
O. Akinade, A.-M. Mahamadu, Offsite construction: developing a BIM-based [35] P. Veloso, G. Celani, R. Scheeren, From the generation of layouts to the production
optimizer for assembly, J. Clean. Prod. 215 (2019) 1180–1190, https://doi.org/ of construction documents: an application in the customization of apartment plans,
10.1016/j.jclepro.2019.01.113. Autom. Constr. 96 (2018) 224–235, https://doi.org/10.1016/j.
[9] S. Barman, A.E. Canizares, A survey of mass customization in practice, Int. J. Supp. autcon.2018.09.013.
Chain Manag. 4 (2015) 65–72. URL: https://ojs.excelingtech.co.uk/index.ph [36] F. Bianconi, M. Filippucci, A. Buffi, Automated design and modeling for mass-
p/IJSCM/article/view/1027/625 (Accessed date: 4 March 2020). customized housing. A web-based design space catalog for timber structures,
[10] F. Bianconi, M. Filippucci, A. Buffi, Automated design and modeling for mass- Autom. Constr. 103 (2019) 13–25, https://doi.org/10.1016/j.autcon.2019.03.002.
customized housing. A web-based design space catalog for timber structures, [37] X. Zhou, J. Wang, M. Guo, Z. Gao, Cross-platform online visualization system for
Autom. Constr. 103 (2019) 13–25, https://doi.org/10.1016/j.autcon.2019.03.002. open BIM based on WebGL, Multimed. Tools Appl. (2018) 28575–28590, https://
[11] V.S. Veenstra, J.I.M. Halman, J.T. Voordijk, A methodology for developing product doi.org/10.1007/s11042-018-5820-0.
platforms in the specific setting of the housebuilding industry, Res. Eng. Des. 17 [38] V. Singh, N. Gu, X. Wang, A theoretical framework of a BIM-based multi-
(2006) 157–173, https://doi.org/10.1007/s00163-006-0022-6. disciplinary collaboration platform, Autom. Constr. 20 (2011) 134–144, https://
[12] C. Forza, F. Salvador, Managing for variety in the order acquisition and fulfilment doi.org/10.1016/j.autcon.2010.09.011.
process: the contribution of product configuration systems, Int. J. Prod. Econ. 76 [39] E. Alreshidi, M. Mourshed, Y. Rezgui, Cloud-based BIM governance platform
(2002) 87–98, https://doi.org/10.1016/S0925-5273(01)00157-8. requirements and specifications: software engineering approach using BPMN and
[13] C. Forza, F. Salvador, Product configuration and inter-firm co-ordination: an UML, J. Comput. Civ. Eng. 30 (2016), 04015063, https://doi.org/10.1061/(ASCE)
innovative solution from a small manufacturing enterprise, Comput. Ind. 49 (2002) CP.1943-5487.0000539.
37–46, https://doi.org/10.1016/S0166-3615(02)00057-X. [40] H.M. Said, T. Chalasani, S. Logan, Exterior prefabricated panelized walls platform
[14] A.S. Howe, I. Ishii, T. Yoshida, Kit-of-parts: a review of object-oriented construction optimization, Autom. Constr. 76 (2017) 1–13, https://doi.org/10.1016/j.
techniques, in: International Symposium on Automation and Robotics in autcon.2017.01.002.
Construction, 1999, pp. 165–171, https://doi.org/10.22260/isarc1999/0025. [41] P. Jensen, T. Olofsson, H. Johnsson, Configuration through the parameterization of
[15] H. Lai, X. Deng, T.Y.P. Chang, BIM-based platform for collaborative building design building components, Autom. Constr. 23 (2012) 1–8, https://doi.org/10.1016/j.
and project management, J. Comput. Civ. Eng. 33 (2019), https://doi.org/ autcon.2011.11.016.
10.1061/(ASCE)CP.1943-5487.0000830, 05019001-1–05019001-15. [42] M. Deng, V.J.L. Gan, Y. Tan, A. Joneja, J.C.P. Cheng, Automatic generation of
[16] H.M. Chen, K.C. Chang, T.H. Lin, A cloud-based system framework for performing fabrication drawings for façade mullions and transoms through BIM models, Adv.
online viewing, storage, and analysis on big data of massive BIMs, Autom. Constr. Eng. Inform. 42 (2019) 100964, https://doi.org/10.1016/j.aei.2019.100964.
71 (2016) 34–48, https://doi.org/10.1016/j.autcon.2016.03.002. [43] G. Jansson, H. Johnsson, D. Engström, Platform use in systems building, Constr.
[17] Z. Ma, Z. Liu, Ontology- and freeware-based platform for rapid development of BIM Manag. Econ. 32 (2014) 70–82, https://doi.org/10.1080/01446193.2013.793376.
applications with reasoning support, Autom. Constr. 90 (2018) 1–8, https://doi. [44] D. Benros, J.P. Duarte, An integrated system for providing mass customized
org/10.1016/j.autcon.2018.02.004. housing, Autom. Constr. 18 (2009) 310–320, https://doi.org/10.1016/j.
[18] J.C.P. Cheng, M. Das, A bim-based web service framework for green building autcon.2008.09.006.
energy simulation and code checking, J. Inform. Technol. Constr. 19 (2014) [45] S. Jaganathan, L.J. Nesan, R. Ibrahim, A.H. Mohammad, Integrated design
150–168. URL: https://www.itcon.org/2014/8 (Accessed date: 6 March 2020). approach for improving architectural forms in industrialized building systems,
[19] P. Piroozfar, E.R.P. Farr, L. Hvam, D. Robinson, S. Shafiee, Configuration platform Front. Architect. Res. 2 (2013) 377–386, https://doi.org/10.1016/j.
for customisation of design, manufacturing and assembly processes of building foar.2013.07.003.
façade systems: a building information modelling perspective, Autom. Constr. 106 [46] A. Kudsk, M.O. Grønvold, M.H. Olsen, L. Hvam, C. Thuesen, Stepwise
(2019) 102914, https://doi.org/10.1016/j.autcon.2019.102914. modularization in the construction industry using a bottom-up approach, Open
[20] L. Malmgren, P. Jensen, T. Olofsson, Product modeling of configurable building Construct. Build. Technol. J. 7 (2013) 99–107, https://doi.org/10.2174/
systems – a case study, Electron. J. Inf. Technol. Constr. 15 (2010) 354–368. URL: 1874836801307010099.
https://www.itcon.org/2010/27 (Accessed date: 8 March 2020). [47] N.H. Mortensen, L. Hvam, A. Haug, Modelling product families for product
[21] M. Oh, J. Lee, S.W. Hong, Y. Jeong, Integrated system for BIM-based collaborative configuration systems with product variant master, in: The 19th European
design, Autom. Constr. 58 (2015) 196–206, https://doi.org/10.1016/j. Conference on Artificial Intelligence, 2010, pp. 1–6. URL: https://backend.orbit.
autcon.2015.07.015. dtu.dk/ws/portalfiles/portal/4903251/EvalofPVMver9.pdf (Accessed date: 8
[22] B.J. Pine, Mass Customization: The New Frontier in Business Competition, Harvard March 2020).
Business School Press, Boston, 1993. ISBN: 0-87584-372-7. [48] I.J. Ramaji, A.M. Memari, J.I. Messner, Product-oriented information delivery
[23] S.J. Hu, Evolving paradigms of manufacturing: from mass production to mass framework for multistory modular building projects, J. Comput. Civ. Eng. 31
customization and personalization, in: Forty Sixth CIRP Conference on (2017), 04017001, https://doi.org/10.1061/(asce)cp.1943-5487.0000649.
Manufacturing Systems, Elsevier B.V, 2013, pp. 3–8, https://doi.org/10.1016/j. [49] Dong Yang, Ming Dong, Rui Miao, Development of a product configuration system
procir.2013.05.002. with an ontology-based approach, Computer-Aided Design Journal (2008)
[24] A.P. Meyer, H. Marc, Lehnerd, The Power of Product Platforms: Building Value and 863–878, https://doi.org/10.1016/j.cad.2008.05.004.
Cost Leadership, Free Press, New York, 1997 (ISBN: 0-684-82580-5). [50] A. Brière-Côté, L. Rivest, A. Desrochers, Adaptive generic product structure
[25] A. Haug, Representation of Industrial Knowledge – as a Basis for Developing and modelling for design reuse in engineer-to-order products, Comput. Ind. 61 (2010)
Maintaining Product Configurators, Technical University of Denmark, Denmark, 53–65, https://doi.org/10.1016/j.compind.2009.07.005.
2007 (ISBN: 978-87-91035-67-8). [51] D. Yang, M. Dong, Applying constraint satisfaction approach to solve product
[26] R. Bourke, Product Configurators: Key Enabler for Mass Customization – An configuration problems with cardinality-based configuration rules, J. Intell. Manuf.
Overview, Midrange Enterprise, URL: http://www.pdmic.com/articles/midrang 24 (2013) 99–111, https://doi.org/10.1007/s10845-011-0544-2.
e/Aug2000.html, 2000 (Accessed date: 15 July 2004). [52] D. Sabin, R. Weigel, Product configuration frameworks - a survey, IEEE Intellig.
[27] T. Soininen, J. Tiihonen, T. Männistö, R. Sulonen, Towards a general ontology of Syst. Appl. 13 (1998) 42–49, https://doi.org/10.1109/5254.708432.
configuration, Artificial Intelligence for Engineering Design, Analysis and [53] P. Piroozfar, E.R.P. Farr, L. Hvam, D. Robinson, S. Shafiee, Configuration platform
Manufacturing 12 (1998) 357–372, https://doi.org/10.1017/ for customisation of design, manufacturing and assembly processes of building
s0890060498124083. façade systems: a building information modelling perspective, Autom. Constr. 106
[28] L. Hvam, Niels Henrik Mortensen, J. Riis, Product Customization, Springer Science (2019) 102914, https://doi.org/10.1016/j.autcon.2019.102914.
& Business Media, 2008 (ISBN: 978-3-540-71448-4). [54] A. Tidstam, J. Malmqvist, Information Modelling for Automotive Configuration,
[29] C. Forza, F. Salvador, Product Information Management for Mass Customization Proceedings of NordDesign 2010, the 8th international NordDesign Conference,
Connecting Customer, Front-Office and Back-Office for Fast and Efficient 2010, pp. 275–288. URL: http://publications.lib.chalmers.se/records/fulltext/lo
Customization, Palgrave Macmillan, Basingstoke, 2006 (ISBN: 978-0-230-00682- cal_129032.pdf (Accessed date: 8 March 2020).
9). [55] R. Rempling, A. Mathern, D. Tarazona Ramos, S. Luis Fernández, Automatic
[30] M. Tetik, A. Peltokorpi, O. Seppänen, J. Holmström, Direct digital construction: structural design by a set-based parametric design method, Autom. Constr. 108
technology-based operations management practice for continuous improvement of (2019) 102936, https://doi.org/10.1016/j.autcon.2019.102936.
construction industry performance, Autom. Constr. 107 (2019) 102910, https:// [56] N. Zhao, C. Kam, J.T.Y. Lo, J.I. Kim, M. Fischer, Construction parts in building
doi.org/10.1016/j.autcon.2019.102910. projects: definition and case study, J. Manag. Eng. 34 (2018), 04018014, https://
[31] F. Wikberg, T. Olofsson, A. Ekholm, Design configuration with architectural doi.org/10.1061/(ASCE)ME.1943-5479.0000616.
objects: linking customer requirements with system capabilities in industrialized [57] A. Kudsk, Modularization in the construction industry using a top-down approach,
house-building platforms, Constr. Manag. Econ. 32 (2014) 196–207, https://doi. Open Constr. Build. Technol. J. 7 (2013) 88–98, https://doi.org/10.2174/
org/10.1080/01446193.2013.864780. 1874836801307010088.
[32] I.J. Ramaji, A.M. Memari, Product architecture model for multistory modular [58] G. Staib, A. Dörrhöfer, M. Rosenthal, Components and Systems Modular
buildings, J. Constr. Eng. Manag. 142 (2016), 04016047, https://doi.org/10.1061/ Construction – Design, Structure, New Technologies, De Gruyter, Basel, 2013
(ASCE)CO.1943-7862.0001159. (ISBN: 978-3-7643-8656-6).
[33] Building Design+Construction, BD+C Giants 300 Technology + Innovation Study, [59] M. Lawson, R. Ogden, C. Goodier, Design in Modular Construction, CRC Press, Boca
URL: https://www.bdcnetwork.com/sites/bdc/files/BDC%202019%20Giants%20 Raton, 2014 (ISBN: 978-0-415-55450-3).

13
J. Cao et al. Automation in Construction 123 (2021) 103437

[60] S. Zhang, F. Pan, C. Wang, Y. Sun, H. Wang, BIM-based collaboration platform for [68] M.M. Ahmad, N. Dhafr, Establishing and improving manufacturing performance
the management of EPC projects in hydropower engineering, J. Constr. Eng. measures, Robot. Comput. Integr. Manuf. 18 (2002) 171–176, https://doi.org/
Manag. 143 (2017) 1–15, https://doi.org/10.1061/(ASCE)CO.1943- 10.1016/S0736-5845(02)00007-8.
7862.0001403. [69] T.W. Simpson, Product platform design and customization: status and promise,
[61] T. Kraska, B. Trushkowsky, The new database architectures, IEEE Internet Comput. artificial intelligence for engineering design, Anal. Manuf. 18 (2004) 3–20, https://
17 (2013) 72–75, https://doi.org/10.1109/MIC.2013.56. doi.org/10.1017/S0890060404040028.
[62] J.M. Gallaugher, S.C. Ramanathan, Choosing a client/server architecture: a [70] A. Myrodia, T. Randrup, L. Hvam, Configuration lifecycle management – an
comparison of two-and three-tier systems, Inf. Syst. Manag. 13 (1996) 7–13, assessment of the benefits based on maturity, in: Proceedings of the 20th
https://doi.org/10.1080/10580539608906981. Configuration Workshop, 2018, pp. 1–6. URL: https://backend.orbit.dtu.dk/ws/
[63] F. Ouyang, X. Du, Application of spatial data analysis in architectural planning, IOP portalfiles/portal/158820792/Configuration_Lifecycle_Management_An_assessme
Conf. Ser. Earth Environ. Sci. 234 (2019), 012035, https://doi.org/10.1088/1755- nt_of_the_benefits_based_on_maturity_preprint.pdf (Accessed date: 4 March 2020).
1315/234/1/012035. [71] P. Smiding, E. Gerth, R. Jensen, Developing product configurators in the AEC
[64] J. Li, N. Li, K. Afsari, J. Peng, Z. Wu, H. Cui, Integration of building information industry, in: Proceedings of the 2016 International Conference on Construction and
modeling and web service application programming interface for assessing Real Estate Management, 2016, pp. 342–350, https://doi.org/10.1061/
building surroundings in early design stages, Build. Environ. 153 (2019) 91–100, 9780784480274.017.
https://doi.org/10.1016/j.buildenv.2019.02.024. [72] G. Jansson, E. Viklund, T. Olofsson, Artistic and engineering design of platform-
[65] H. Lidelöw, G. Jansson, E. Viklund, Design breakdown in industrialized based production systems: a study of Swedish architectural practice, Buildings 8
construction: supporting lean manufacturing, in: Modular and Offsite Construction (2018) 34, https://doi.org/10.3390/buildings8020034.
Summit Proceedings, 2015, pp. 1–8. URL: https://journalofindustrializedconstr [73] B. Christudas, Practical Microservices Architectural Patterns: Event-Based Java
uction.com/index.php/mocs/article/view/158/155 (Accessed date: 8 March Microservices with Spring Boot and Spring Cloud, Apress, Berkeley, CA, 2019
2020). (ISBN: 978-1-4842-4500-2).
[66] S. Chaudhuri, E. Kalogerakis, L. Guibas, V. Koltun, Probabilistic reasoning for [74] A. Peltokorpi, H. Olivieri, A.D. Granja, O. Seppänen, Categorizing modularization
assembly-based 3D modeling, ACM Trans. Graph. 30 (2011) 1–10, https://doi.org/ strategies to achieve various objectives of building investments, Constr. Manag.
10.1145/2010324.1964930. Econ. 36 (2018) 32–48, https://doi.org/10.1080/01446193.2017.1353119.
[67] S. Khalili-Araghi, B. Kolarevic, Variability and validity: flexibility of a dimensional [75] P. Jensen, H. Lidelöw, T. Olofsson, Product configuration in construction, Int. J.
customization system, Autom. Constr. 109 (2020) 102970, https://doi.org/ Mass Custom. 5 (2015) 73–92, https://doi.org/10.1504/IJMASSC.2015.069601.
10.1016/j.autcon.2019.102970.

14

You might also like