See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/338281769

BIM-based decision support system for automated manufacturability check of

wood frame assemblies

Article in Automation in Construction · December 2019

DOI: 10.1016/j.autcon.2019.103065

CITATIONS READS
40 556

4 authors:

Shi An Pablo Martinez

University of Alberta Northumbria University
3 PUBLICATIONS 74 CITATIONS 56 PUBLICATIONS 1,049 CITATIONS

SEE PROFILE SEE PROFILE

Mohamed Al-Hussein Dr Rafiq Ahmad

University of Alberta University of Alberta
361 PUBLICATIONS 6,218 CITATIONS 175 PUBLICATIONS 2,246 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Pablo Martinez on 09 December 2021.

The user has requested enhancement of the downloaded file.

 BIM-based decision support system for automated
manufacturability check of wood frame assemblies
Shi Ana, Pablo Martineza, Mohamed Al-Husseina, Rafiq Ahmadb*

a
Department of Civil and Environmental Engineering, University of Alberta, Canada.
b
Laboratory of Intelligent Manufacturing, Design and Automation (LIMDA), Department of Mechanical
Engineering, University of Alberta, Canada.
*Corresponding author. Email: [email protected]

Abstract
As offsite construction is increasing in popularity, an increasing number of construction products
are fabricated in a controlled factory environment. Due to the complexity of construction products
and the rising amount of automation used in the industry, productivity has reached a peak because
the process planning of manufacturing activities is still done manually, for example, building
information models (BIM) do not provide manufacturing information for construction products.
Knowing whether a machine can manufacture a construction product defined by the BIM model
is a critical prerequisite for new products. This paper proposes a BIM-based framework for
automating the evaluation of machine capabilities for the manufacturing of construction-oriented
products. By identifying intersections of the building elements of the product, feasible
manufacturing operations are determined, and manufacturing locations are calculated. These
locations are then compared to the manufacturing capabilities of the machine. The proposed
approach is validated using two wood frame assemblies. The results show that the system
accurately determines whether a user-selected machine can manufacture a construction product
pre-designed using BIM software.
Keywords: building information modeling, ontologies, intelligent manufacturing, wood framing,
construction automation, computer numerical control, mass customization.

1. Introduction
1.1. Background

Recent technological advancements in offsite construction have resulted in frame panels becoming
increasingly popular in North America. Traditionally, frames are constructed manually on-site and
assembled on-site. These frame construction processes are labor intensive, time-consuming, prone
to error, weather dependent, and lack precision. Using offsite construction methods, wall frames
are prefabricated in a factory environment and shipped to the construction site. Due to the
controlled factory environment, offsite construction dramatically increases the productivity,
quality, and deliverability of the products [1]. Through the use of industrial automated machines
in the framing processes, the productivity of frame assembling increases greatly. An example of a
partial wood frame residential house along with a typical wood frame is shown in Figure 1. Figure
1 (a) shows a typical house structure that consists of multiple modular frames known as “frame
assemblies”. As shown in Figure 1 (b), each wood frame includes rectangular lumber with various

1
lengths. Before framing the wood panel, timbers are first cut to length based on the manufacturing
drawing. Then the building elements are placed in the correct locations based on the drawing.
Finally, the intersections between these components are secured using nailing or screw fastening
operation.

Figure 1: (a) BIM model of a partial residential house, and (b) typical wood frame used for
residential building construction.
To produce frame assemblies, panel structures must be fully described. The detailed specifications
of frame assemblies, such as overall dimensions and composed components, is commonly given
by the building information model (BIM). According to the National Institute of Building
Sciences, BIM is “a digital representation of physical and functional characteristics of a facility”
[2]. While BIM provides detailed specifications of construction-oriented products, the integration
of BIM with manufacturing systems is still under development [3]. As an example, quantity take-
off for construction oriented light-frame buildings may be automatically obtained from BIM
information [4]. Similarly, the machinery, time and expenditure required to fabricate a
construction-oriented product could be described in the BIM model and, therefore, would be
instantly obtained in the early stage of design. This approach would ease the process planning for
construction product manufacturing, which is an essential step in optimizing the sequence of
operations “where one-of-a-kind products or the same product are made infrequently” [5]. Wood
panels, considering the diversity in the options given by designers, are one-of-a-kind products in
most cases.
Presently, the lack of integration between manufacturing and BIM is overcome by expert
knowledge and experience in different domains; however, this approach can be time-consuming,
subjective, and relies heavily on the knowledge of the experts. Besides, this lack of integration
creates a barrier that inhibits information exchange between product designers and manufacturers
and adds to the cost to produce a product due to increased communication overhead. To overcome
certain aforementioned limitations and ease the otherwise manual process, commercial options are
available and listed in Table 1.

2
 Table 1. Commercial options for timber BIM softwares.

Name Producer Outputs

2D drawings, cut lists, shop floor drawings, exports
ArchiFrame ArchiFrame
CNC code for specific machinery
2D drawings, cut lists, shop floor drawings,
AGACAD Wood
AGACAD structural analysis, experts CNC code for specific
Framing
machinery
WoodStud_Frame Tekla 2D drawings, structural analysis
2D drawings, cut lists, shop floor drawings,
Timber Framing Autodesk structural analysis, experts CNC code for specific
machinery
Metal Wood 2D drawings, material lists, exports CNC code for
StrucSoft Solutions
Framer specific machinery
2D drawings, material lists, exports CNC code for
Vertex DB Vertex Systems
specific machinery

As observed, these solutions focus on automatizing the information transfer from the design stage
to the production process. From a source BIM model, drawings and material lists can be obtained.
On top of that, machine manufacturers may provide information to software developers so data
can be exported directly from the BIM environment to the shop floor. However, the engineering
implications for these software are limited when providing analysis regarding manufacturability.
Independently of the capabilities of the offsite facility and the machinery available, shops drawings
and material lists can be generated. As such, potential issues may arise from non-manufacturable
panels, namely production delays and increase in manual work. Consequently, an automated, cost-
effective, and real-time decision-support system is in demand in offsite construction facilities that
specialize in panelized construction to support design by linking floor capacities, namely the
machinery capabilities, and the selected panel design.
Frame assemblies are complex products because of their diverse structural compositions and
heterogeneous materials. As a result, the manufacturing activities required to make such products
are also complex and the interactions between the products and manufacturing systems present
many challenges. Due to the complex nature of these domains, it is unrealistic to develop, initially,
a system that accounts for every aspect of the BIM-manufacturing relationship. As a first step, this
paper aims to integrate BIM and manufacturing systems by developing an automated decision-
support system for the manufacturing of wood-frame assemblies. Since wood frames are composed
of 2D manufacturing features, a 2D framework is chosen to be developed.
1.2. Related Work

Offsite construction provides an efficient, productive, safe, and less labor intensive construction
environment thanks to a controlled factory environment [6]. Implementation of BIM for offsite
construction further enhances the productivity and quality of construction-oriented products in the
manufacturing stage. For example, Malik et al. extracted product information from the BIM model
and generated near-optimized tool paths for automated light-gauge steel framing, increasing

3
productivity and reducing cycle times [7]. As well, Martinez et al. proposed a vision-based real-
time inspection system for steel frame manufacturing. The proposed framework successfully
improved the accuracy of the framed panels by comparing the real and the nominal geometries
obtained from the BIM model and providing automatically generated instructions to correct
defective frames to the operator [8].
BIM has been argued to be a key element in the development of offsite construction for buildings.
Its successful application has proven to reduce delays and monitor project schedules [9], a more
streamlined design and production process [10], or provide other benefits and improvements
[11,12]. Specifically, in the timber industry, BIM was investigated as an integrator across the entire
value chain [13]. Several areas have been previously identified where BIM can enhance timber
offsite construction, from the design stage to the final environmental performance of the building
[14].
Although intensive research has been conducted on BIM, timber manufacturing, and offsite
construction, integration between the two fields is still under development [3]. Nonetheless, some
limitations were presented on the current integration of BIM for timber construction [15]. The
information required from the BIM model to enable timber offsite construction is yet not fully
available, namely the manufacturing instructions for the construction products are still not linked
within the BIM environment [16]. The existing approach in the manufacturing of construction
products is a sequential process that includes an important amount of manual work; the overall
process is presented in Figure 2. All the necessary 2D manufacturing drawings are created based
on the BIM model and manual process planning must be completed before manufacturing. This
one-way process has no instant feedback mechanisms, i.e. any necessary changes required in the
manufacturing stage will need to be manually communicated to the design personnel after time
has been used to manually analyze the proposed design.
Several disadvantages are observed in the existing approach: 1) both the creation and interpretation
of 2D drawings are time consuming and error-prone; 2) manufacturing engineers must be familiar
with both the products and the manufacturing resources to perform proper process planning; 3)
manufacturability of frame assemblies is not transparent due to the bottleneck in the flow of
information; and 4) the manufacturing system is extremely sensitive to product changes as updates
to the process planning are required for even minor changes applied to the products. As the global
market becomes increasingly competitive, mass customization of products is in demand compared
to standardized ones [17]. Mass customization requires information transparency and decentralized
decision making of manufacturing systems [18].

Figure 2: Current BIM-based frame assemblies manufacturing.

In manufacturing, computer-aided process planning (CAPP) is the use of computer technology to
assist the process planning for the construction of a part or a product [5]. CAPP is a crucial activity
to bridge and integrate Computer-Aided Design (CAD) and Computer-Aided Manufacturing

4
(CAM). It converts various requirements (such as functional and mechanical requirements) of a
product into manufacturing instructions [19]. As a critical intermediate step, CAPP has attracted
an increasing amount of research interest in the past few decades [20]. As reviewed by Xu et al.,
feature-based technology, among various categories in CAPP research, has been a major topic for
CAD/CAM integrations as well as for CAPP systems [20]. In feature-based approaches,
topological and geometrical features of a part are interpreted and are translated into manufacturing
operations. While CAPP has been developed to provide detailed manufacturing instructions of a
mechanical part, limitations are encountered when applying CAPP to construction-oriented
products. In contrast to sophisticated geometries of mechanical parts, the geometries of the parts
of frame assemblies are primitive. For example, the building elements of wood frames are
rectangular prisms. Challenges arise when assembling these building elements into frames.
Manufacturing operations are rarely needed for individual parts, instead, securing all the different
elements of the frame in the correct location, orientation, and sequence is required. Consequently,
planning of the manufacturing processes is needed for construction-oriented products.
Robust process planning is built using the knowledge of domain experts [20][21]. Knowledge-
based systems are also needed as construction engineering is heavily governed by experience.
Ontology, as defined by Gruber, is “an explicit and formal specification of a conceptualization”
[22]. As reviewed by An et al., ontologies can be used for knowledge modeling because: 1) they
offer interoperability of knowledge from different domains, and 2) they support consistency
checking [9]. Jardim-Goncalves et al. proposed the knowledge framework “funStep” using
ontologies to improve the interoperability of manufacturing systems [23]. Lemaignan et al.
proposed the framework MASON (MAnufacturing’s Semantics ONtology) to manage the
knowledge in the manufacturing environment using ontologies [24]. In construction, specifically,
ontologies have been proven useful in extracting information from BIM for practical use. Zhang
et al. extracted BIM information of construction materials and related job hazards involved in
construction activities using ontology formulation [25]. By modeling tasks, methods, and the job
hazards involved in construction activities using an ontology, the developed system provides
automated job hazard analysis, which significantly improves the efficiency of project management
tasks [25]. Liu et al. proposed an ontology-based semantic approach that extracts quantity take-off
information of construction-related activities [4]. Using the proposed framework, construction
practitioners can readily obtain and visualize the materials needed for construction activities [4].
Recently, and following the ontological model MASON, a knowledge model was built by An et
al. that effectively determines manufacturing operations based on the BIM model of construction-
oriented products [16].
To summarize, using BIM as the source of information with current expert knowledge organized
using ontologies, a system is needed to automatically determine the machine capabilities for
manufacturing 2D wood frame assemblies and to facilitate process planning in offsite construction.
The remainder of this paper is organized as follows: Section 2 illustrates the proposed system
framework in detail; Section 3 presents the experimental setup used for validation of the proposed
methodology; Section 4 presents the simulation results; Section Error! Reference source not
found. discusses the implications of the results and the limitations of the proposed framework;
and, finally, Section 6 concludes by presenting tangible results and discusses the future directions
of the proposed system.

5
2. System framework: Machine Eligibility Determination System (MEDS)
The system presented aims to determine if any wood frame assembly can be manufactured by a
machine using the product information pre-generated in the building information model (BIM)
and the given machine specifications. The proposed system is presented in Figure 3, where the
architecture can be divided into four modules: 1) BIM data input; 2) mating plane detection
algorithm; 3) ontology formulation; and 4) machine eligibility determination. The proposed
framework is developed and implemented using Python programming language. Python is chosen
for the following reasons: 1) Python is an open environment that allows rapid programming, 2)
allows the user to read, process and modify data quickly, and 3) has graphical libraries that allow
complex simulation and visualization.

Figure 3: MEDS system framework overview.

In the proposed framework, the four modules are applied sequentially. First, the BIM data input
module involves retrieving the relevant geometric information of the modeled construction-related
product from a pre-designed 3D BIM model software, such as Autodesk Revit. The input required
for geometry interpretation are the coordinates of each building element (such as studs and plates).
The completion of this stage generates coordinates of points that define the planes of all building
elements and the unit normal of each plane. A wood frame is re-constructed in a simulation
environment based on BIM data.
Once planes and their directions are known, these data are fed into the second module named
mating plane detection algorithm (MPDA). Mating planes refer to areas where building elements
are in contact with each other. While mating planes are directly related to any frame product
manufacturing operations, they are currently not available in the BIM model. After running MPDA
on the generated wood frame, all the mating planes are detected and generated.
The mating planes are then mapped to predefined formulations stored in the ontology model.
Ontology stores the knowledge that construction personnel has gained throughout their

6
experiences. It takes the types of mating planes and determines the feasible operations required to
fabricate such frame assemblies.
Having determined the manufacturing operations, systems that can carry out such operations are
determined. A machine that possesses such systems will be then selected. The machine eligibility
determination module is responsible for determining if a particular device can perform the required
manufacturing operations. Upon completion of this stage, the user of the proposed framework will
have a clear understanding as to which manufacturing operations the selected machine may apply
to the designed wood frame.
The subsections that follow will provide details about the proposed system framework.
2.1. BIM Data Input

BIM models provide a thorough set of information about the designed construction product. The
3D geometry of a construction-oriented product, for example, is an essential subset of BIM. In
Autodesk Revit, each component of the model is referred to as a building element. A building
element is a 3D geometry model that is part of the building. Examples of building elements include
walls, windows, doors, and roofs, etc. [26]. Figure 1 shows the BIM model of a residential house
and a randomly selected wood frame from the presented facility.
Initially, for any wood panel, detailed geometrical data are required to deeply understand the
manufacturing process that would be required to fabricate it. As investigated by Liu et al., ‘Element
ID’, element ‘MaterialSet’ and ‘LocationPoint’ coordinates can be extracted from every building
element [4]. The relevant information adapted from the proposed Revit API unified modeling
language (UML) diagram is shown in Figure 4 [4]. Using these location points from the BIM
model, the geometry of each building element of the panel, such as faces, can be reconstructed in
a simulation environment. As only a tiny fraction of the information that BIM can provide is useful
for this study, the relevant building element location points are downloaded into a Python-based
simulated environment to downsize the computational power required, although programming the
proposed software in the Revit application programming interface (API) environment should
ultimately be done.

Figure 4: UML diagram of Autodesk Revit building elements [4].

7
The pseudo code to generate all the faces of each element is as follows.
1. Get all the unique identifiers (such as ‘P01’ for plate) from the BIM model. This provides
the total quantity of building elements.
2. Select one identifier and filter the data using this identifier. The output should be 8 rows of
data as there are 8 points in a building element.
3. Select one row and get the coordinate 𝑃1 = (𝑥1 , 𝑦1 , 𝑧1 ).
4. Select the rest of the points with the same x-coordinate (𝑥1 ). This should output the
coordinates of 4 points, which define one face 𝐹1 of the element.
5. Repeat step 4 with y and z-coordinates (𝑦1 and 𝑧1 , respectively). This will give 2 additional
planes 𝐹2 and 𝐹3 .
6. Select diagonal point 𝑃2 with respect to 𝑃1 .
7. Repeat steps 4 and 5 with 𝑃2 . This should give us 3 mutually perpendicular planes 𝐹4 , 𝐹5 ,
and 𝐹6 .
8. Repeat steps 2 to 7 for the rest of the building elements.
After generation of faces, the generated faces of each element are arranged as follows:

[identifier, material, [𝐹1 𝐹2 𝐹3 𝐹4 𝐹5 𝐹6 ]𝑇 ]

𝐹𝑖 = [𝑃𝑖1 𝑃𝑖2 𝑃𝑖3 𝑃𝑖4 ]𝑇 , 𝑖 = 1, 2, 3, 4, 5, 6
𝑃𝑖𝑗 = (𝑥𝑖𝑗 , 𝑦𝑖𝑗 , 𝑧𝑖𝑗 ), 𝑗 = 1, 2, 3, 4

where 𝐹𝑖 is the 𝑖th face of a building element, 𝑃𝑖𝑗 is the 𝑗th point of face 𝑖, and (𝑥𝑖𝑗 , 𝑦𝑖𝑗 , 𝑧𝑖𝑗 ) is the
3D coordinate of point 𝑃𝑖𝑗 . Once all the faces of each building element are identified, the
coordinates of points are ordered counter-clockwise. This is required for successive tasks, such as
constructing two local mutually perpendicular vectors in one face.
2.2. Mating Plane Detection Algorithm (MPDA)

Since construction-oriented product manufacturing involves assembling building elements, the

assembly of building elements needs to be permanently connected. A mating plane is defined as
the common area of two building elements in contact with each other. Mating planes of a product
are where hard connections happen. For the proposed system to decide whether a machine can
manufacture a product by securing the connections, the system first needs to know the locations
of all the mating planes. Automatic detection of the mating planes is therefore needed for each
product.
Mating plane detection algorithm (MPDA) consists of two parts: plane intersection detection and
mating plane determination. Plane intersection detection is used to find out if two planes have the
potential to intersect. Mating plane determination is used to calculate the exact location of each
mating plane. Note that MPDA only applies to mating planes that are in parallel to 𝑥𝑦 plane, 𝑦𝑧
plane, or 𝑥𝑧 plane.

8
2.2.1. Plane Intersection Detection

Let A and B be two planes of two different building elements. If they intersect, knowing their
direction and the distance between them is needed. The direction of each face is represented by a
unit normal vector 𝑢⃗ and is calculated using Equation (1):

𝑎 × 𝑏⃗
𝑢
⃗ = (1)
|𝑎 × 𝑏⃗|
where 𝑎 and 𝑏⃗ are two non-parallel vectors in the face constructed using the points that define the
face as shown in Figure 5. The distance between A and B, 𝑑𝐴𝐵 , is calculated using Equation (2):

𝑑𝐴𝐵 = 𝑣 ∙ 𝑢⃗ (2)
where 𝑣 is a vector formed by connecting 2 points, one from each plane, as shown in Figure 5.

Figure 5: Graphical representation of plane intersection detection.

For any two intersecting planes A and B of two different building elements, the conditions in
Equations (3) and (4) must be satisfied. Equation (3) forces planes A and B to be parallel to each
other and in opposite direction, and Equation (4) ensures that the distance between both faces, A
and B, is zero. Equations (3) and (4) are shown below:

𝑢
⃗𝐴 ∙𝑢
⃗ 𝐵 = −1 (3)
𝑑𝐴𝐵 = 0 (4)
Note that Equations (3) and (4) are necessary, but not sufficient to guarantee an intersection
between planes A and B. Section 2.2.2 calculates the potential overlap between planes A and B.

9
2.2.2. Mating Plane Determination

Once two planes, A and B, from different building elements satisfy Equations (3) and (4), possible
relationships could be established: 1) containment, 2) partial intersection, 3) full intersection, and
4) no intersection. These relationships are illustrated in Figure 6. For the intersection between
planes A and B to generate a mating plane, Equation (5) must be satisfied:
𝐴∩𝐵 ≠∅ (5)

Figure 6: Potential relationships between any two faces: (a) containment, (b) partial
intersection, (c) full intersection and (d) no intersection.
To validate each intersection generated by every pair of planes within a product, several sequential
steps are required. As stated in Section 2.1, points in every plane are ordered counter-clockwise.
A local coordinate system of each plane is generated by constructing two mutually perpendicular
vectors using three consecutive points in that plane using Equations (6) and (7):

𝑢
⃗ 𝐴 = 𝑛⃗𝐴,12, 𝑣𝐴 = 𝑛⃗𝐴,23 and 𝑢
⃗ 𝐵 = 𝑛⃗𝐵,12 , 𝑣𝐵 = 𝑛⃗𝐵,23 (6)
𝑃𝑗 − 𝑃𝑖 (𝑥𝑗 − 𝑥𝑖 , 𝑦𝑗 − 𝑦𝑖 , 𝑧𝑗 − 𝑧𝑖 )
𝑛⃗𝑋,𝑖𝑗 = =
|𝑃𝑗 − 𝑃𝑖 | 2 2 2 (7)
√(𝑥𝑗 − 𝑥𝑖 ) + (𝑦𝑗 − 𝑦𝑖 ) + (𝑧𝑗 − 𝑧𝑖 )

where 𝑢 ⃗ 𝐴 and 𝑣𝐴 are mutually perpendicular and form a local coordinate system 𝑆𝐴 of plane A,
and 𝑢⃗ 𝐵 and 𝑣𝐵 are mutually perpendicular and form a local coordinate system 𝑆𝐵 of plane B. Let
𝑆𝐴 be the reference coordinate system. Three possible relationships exist between 𝑆𝐴 and 𝑆𝐵 : a) 𝑆𝐵
is parallel to and in the same direction as 𝑆𝐴 , b) 𝑆𝐵 is parallel to but in the direction opposite to that
of 𝑆𝐴 , b) 𝑆𝐵 is perpendicular to 𝑆𝐴 counter-clockwise, and d) 𝑆𝐵 is perpendicular to 𝑆𝐴 clockwise.

10
These relationships are depicted in Figure 7. To find the boundaries of a potential mating plane,
both coordinate systems, 𝑆𝐴 and 𝑆𝐵 , must be aligned in the same direction as in case (a).

Figure 7: (a) 𝑺𝑩 and 𝑺𝑨 are parallel and in the same direction, (b) 𝑺𝑩 is 90° counter-clockwise
from 𝑺𝑨 , (c) 𝑺𝑩 and 𝑺𝑨 are parallel and in the opposite direction, and (d) 𝑺𝑩 is 270° counter-
clockwise from 𝑺𝑨 .

The goal is to have 𝑆𝐴 and 𝑆𝐵 parallel and in the same direction (case (a)). Therefore, for all the
other mentioned relationships between 𝑆𝐴 and 𝑆𝐵 , points in plane B need to be re-ordered using
Equation (8):
[𝑃𝐵4 𝑃𝐵1 𝑃𝐵2 𝑃𝐵3 ]𝑇 , 𝑐𝑎𝑠𝑒 (𝑏)
∗ ∗ ∗ ∗ 𝑇
[𝑃𝐵1 𝑃𝐵2 𝑃𝐵3 𝑃𝐵4 ] ∶= { [𝑃𝐵3 𝑃𝐵4 𝑃𝐵1 𝑃𝐵2 ]𝑇 , 𝑐𝑎𝑠𝑒 (𝑐) (8)
[𝑃𝐵2 𝑃𝐵3 𝑃𝐵4 𝑃𝐵1 ]𝑇 , 𝑐𝑎𝑠𝑒 (𝑑)

The ordered points are represented in Figure 8 (a). Once the boundaries of planes A and B are
found, the points that define the construction lines of the boundaries of their intersecting area can
be calculated using Equation (9):
∗ ∗
𝑃𝐵1 𝑖𝑓 (𝑃𝐵1 − 𝑃𝐴1 ) ∙ 𝑢
⃗𝐴 ≥ 0
Left: 𝐿𝐼 = { ∗ (9)
𝑃𝐴1 𝑖𝑓 (𝑃𝐵1 − 𝑃𝐴1 ) ∙ 𝑢
⃗𝐴 < 0

11
 ∗ ∗
𝑃𝐵2 𝑖𝑓 (𝑃𝐵2 − 𝑃𝐴2 ) ∙ 𝑢
⃗𝐴 ≥ 0
Right: 𝑅𝐼 = { ∗
𝑃𝐴2 𝑖𝑓 (𝑃𝐵2 − 𝑃𝐴2 ) ∙ 𝑢 ⃗𝐴 < 0
∗
𝑃 𝑖𝑓 (𝑃𝐵3 − 𝑃𝐴3 ) ∙ 𝑣𝐴 ≥ 0
Top: 𝑇𝐼 = { 𝐴3
∗ ∗
𝑃𝐵3 𝑖𝑓 (𝑃𝐵3 − 𝑃𝐴3 ) ∙ 𝑣𝐴 < 0
∗ ∗
𝑃𝐵2 𝑖𝑓 (𝑃𝐵2 − 𝑃𝐴2 ) ∙ 𝑣𝐴 ≥ 0
Bottom: 𝐵𝐼 = { ∗
𝑃𝐴2 𝑖𝑓 (𝑃𝐵2 − 𝑃𝐴2 ) ∙ 𝑣𝐴 < 0

The construction lines generated using Equation (9) are shown in Figure 8 (b). Intersecting points
must satisfy Equation (10):

𝑃1 = 𝐿𝐼 + ((𝐵𝐼 − 𝐿𝐼 ) ∙ 𝑣𝐴 ) ∗ 𝑣𝐴
𝑃2 = 𝑅𝐼 + ((𝐵𝐼 − 𝑅𝐼 ) ∙ 𝑣𝐴 ) ∗ 𝑣𝐴
(10)
𝑃3 = 𝐿𝐼 + ((𝑇𝐼 − 𝐿𝐼 ) ∙ 𝑣𝐴 ) ∗ 𝑣𝐴
𝑃4 = 𝑅𝐼 + ((𝑇𝐼 − 𝑅𝐼 ) ∙ 𝑣𝐴 ) ∗ 𝑣𝐴

where 𝑃1 , 𝑃2 , 𝑃3 , 𝑃4 are vertices of the potential mating plane (as shown in Figure 8 (c)) and ‘∗’ is
element-to-element multiplication.

Figure 8: Determination of intersecting area: (a) re-ordering points in plane B, (b)

construction lines of the mating plane for planes A and B, and (c) mating plane of planes A
and B.
Note that points may still be generated even if there is no intersecting area as shown in Figure 6
(d). Therefore, it is necessary to verify that these points are in both planes A and B. Using the unit
normal determined using Equation (1), coordinates of relevant directions 𝜉 and 𝜂 are checked. As
the wood frame elements are forcibly aligned with an orthogonal frame due to its inherent
geometry and to simplify the following equations, only coordinates within the 𝑥𝑦, 𝑦𝑧 or 𝑥𝑧 planes
will be checked. The containment relationship between the points 𝑃𝑖 obtained from Equation (10)
and plane 𝐴 and plane 𝐵 are checked using Equation (11):
∀𝑃𝑖 ∈ 𝐴 ∧ ∀𝑃𝑖 ∈ 𝐵, 𝑖 = 1, 2, 3, 4
(11)

12
If Equation (11) is satisfied, it can be concluded that the mating plane exists between planes A and
B. As a result, Equation (5) is satisfied. To summarize the process, Figure 9 shows the flowchart
for determining all the mating planes of all the building elements contained in a BIM model.

START

Select 2 different
elements

Select one plane from

each element: plane A
and B

Calculate unit vectors of Calculate the distance

Determine potential
plane A and B using between A and B using
mating plane
Equation (3) Equation (4)

YES YES

Equation (5) Equation (6) Equation (7)

satisfied? satisfied? satisfied?
NO NO
NO YES
NO
Store the detected
mating plane

Last combination?
NO
YES

Last 2 elements?
NO
YES

END

Figure 9: Intersection plane detection flowchart.

Even if the MPDA criteria are satisfied, exceptions are made in the cases where 1) double-plate or
double-stud is encountered, or 2) two side-by-side parallel studs with different dimensions are
detected. Double-plate is formed when two identical plates are placed next to each other. Similarly,
double-stud refers to two identical studs that are in contact with each other. While it is required to
secure the connection between double-plate/double-stud, standard practice is to treat the two
elements as one and secure the combined element with the building elements it is in contact with.
Therefore, if a double-plate is detected in MPDA, the possible mating plane between the two plates
is ignored, and both plates are merged into one plate.
In the case where two parallel studs that are side-by-side with different dimensions are detected,
the mating plane between the studs is ignored. Similarly, in the case of a double-plate/double-stud
connection, it is not required to secure the two studs as they will be connected to other building
elements. The two studs, however, are not merged due to the previous assumption that every
building element is a rectangular prism.

13
2.3. Manufacturing Rules

In the knowledge model proposed by An et al., product, operation, and resource ontologies are
formulated in Protégé (an open source ontology editor) [9]. The objective of using the knowledge
model is to obtain the possible manufacturing methods given the mating planes detected by the
MPDA. To achieve this, the detected mating planes must be mapped to those previously defined
and stored in the knowledge base. Therefore, the definitions of wood frame intersections, or mating
planes, need to be retrieved from the knowledge model.
Extending this proposed approach, mating planes determined in Section 2.2.2 are formulated and
categorized, as shown in Figure 10 (a). In the lowest level of the class hierarchy shown in Figure
10 (a), the naming rules are as follows: 1) the first two letters represent two building elements
involved in the connection, and 2) the integer in the third place refers to the number of the building
element represented by the second letter. For example, the class ‘SP2’ represents a single stud and
a double-plate connection. All the possible types of connections are listed in the ontology model
and determined automatically from the panel geometry and wood member’s information. Typical
components in a wood frame and their respective symbols are shown in Figure 10 (b).

Figure 10: (a) Product ontology formulation of mating planes (wood frame intersections),
and (b) building elements in a wood frame.
Based on the mating plane definitions, feasible manufacturing operations can be determined using
the manufacturing rules defined in the ontology model. As proposed by An et al., possible
manufacturing operations for the wood frame shown in Figure 10 (b) are nailing and screw
fastening operations 16]. Nailing is a popular operation for securing wood framing members due
to the relative speed of the operation. For the purpose of this study, the manufacturing operation
is, therefore, manually overruled to be the nailing operation.

14
2.4. Manufacturing Plane Determination

Once the mating planes are detected using MPDA, they will be matched to the ontology
formulation to determine the possible type of operations that are required to secure the connection.
A manufacturing plane is a region where manufacturing operation occurs. The locations of all the
manufacturing planes are needed to simulate where the selected manufacturing operation will
occur. The manufacturing plane is not only determined by the type of manufacturing operation,
but also by the connection type. For example, a welding operation most likely would happen at the
location of the mating plane whereas a nailing operation would take place at the outermost surface
of the building element that the mating plane can be projected onto. In this study, nailing operations
are applied to secure the wood frame elements. For a typical SP1 (single-stud-single-plate)
connection, the manufacturing plane is calculated using Equation (12):

𝑃𝑗 = 𝑃𝑖 + (𝑥𝑚 , 𝑦𝑚 , 𝑧𝑚 ) ∗ 𝑛⃗, 𝑖, 𝑗 = 1, 2, 3, 4 (12)

where 𝑃𝑗 is a point that defines the manufacturing plane, 𝑃𝑖 refers to a point in the mating plane, 𝑛⃗
is the normal vector of mating plane, and 𝑥𝑚 , 𝑦𝑚 , 𝑧𝑚 are the dimensions of the manufacturing part
in 𝑥, 𝑦 and 𝑧 directions, respectively. Figure 11 shows the procedure followed to generate a
manufacturing plane from a mating plane.

Figure 11: Mating plane and manufacturing plane.

In addition to the offset required based on the particular manufacturing operation, some types of
connections also influence the location of the manufacturing plane. Based on the ontology
formulation, SP2 connections occur when studs are connected to a double-plate, commonly two
identical plates next to each other. As stated in Section 2.2.2, mating planes generated between
two plates are ignored, and both are merged into one plate. To prevent false projections of

15
manufacturing planes, they are calculated based on the merged plate. This procedure is illustrated
in Figure 12.

Figure 12: Predicted and actual manufacturing plane of a double-plate connection.

The procedure is different when two parallel studs with different dimensions are encountered as
the studs cannot be merged into a rectangular prism, such as king studs and jack studs located on
doors. If any generated manufacturing plane is in contact with two building elements, it is relocated
further to the outer surface of the building element.
2.5. Region of Effect of 2D Systems

Manufacturing operations are carried out by machine systems. The region in which a system can
perform manufacturing activities is defined as the Region of Effect (ROE). Several factors can
affect the ROE of a system and can be summarized into two categories: machine configurations
and machine logic. The physical infrastructure of the system dictates machine configurations.
Once a machine is installed, physical parameters such as range of motion and work direction are
fixed and cannot be changed unless important changes to the machine are made, such as upgrade
of the actuators. Machine logic, however, dictates how the machine should operate. Although
machine logic can be a limiting factor of the ROE of a system, changes to the machine logic are
much easier to make. In addition, machine logic is subjected to how the machine is set up and the
limitations of the hardware. As a result, machine configurations control the ROE of a system and
machine logic is ignored in this study.
The ROE of a 2D system that is subjected to linear motion is limited by the range of motion of the
actuator that drives the system. Linear motion can be achieved by directly using a linear actuator
or converting a rotary motion driven by a rotary actuator to linear motion. The range of motion of

16
the system can be determined using a similar approach. To determine the range of motion of an
actuator 𝑖, we need: 1) the coordinates of each mounting position, (𝑥𝑖0 , 𝑦𝑖0 , 𝑧𝑖0 ); 2) the direction
cosine of each linear actuator, 𝑛⃗𝑖 ; 3) the starting location of each actuator, (𝑥𝑖1 , 𝑦𝑖1 , 𝑧𝑖1 ); and 4)
the stroke of each actuator, 𝐿𝑖 . The mounting position is needed to locate where the actuator is,
while the direction cosine represents the orientation of the actuator in the reference frame. The
direction cosine is defined by Equation (13):
𝑛⃗𝑖 = (𝑐𝑜𝑠𝜙𝑖 , 𝑐𝑜𝑠𝜃𝑖 , 𝑐𝑜𝑠𝜓𝑖 )
(13)
where 𝜙𝑖 , 𝜃𝑖 , and 𝜓𝑖 represent the rotation angles about the 𝑥, 𝑦, and 𝑧 axes, respectively. In this
study, the linear actuators are aligned with 𝑥, 𝑦, or 𝑧 axes only. The starting location of each
actuator is obtained from its specifications and is determined using Equation (14).

(𝑥𝑖1 , 𝑦𝑖1 , 𝑧𝑖1 ) = (𝑥𝑖0 , 𝑦𝑖0 , 𝑧𝑖0 ) + 𝐷𝑖 ∙ 𝑛⃗𝑖 (14)

The end position of each actuator is calculated by adding the stroke of each actuator to the start
position as shown in Equation (15):

(𝑥𝑖2 , 𝑦𝑖2 , 𝑧𝑖2 ) = (𝑥𝑖1 , 𝑦𝑖1 , 𝑧𝑖1 ) + 𝐿𝑖 ∙ 𝑛⃗𝑖 = (𝑥𝑖0 , 𝑦𝑖0 , 𝑧𝑖0 ) + (𝐿𝑖 + 𝐷𝑖 ) ∙ 𝑛⃗𝑖 (15)
The above characteristic parameters of the linear actuators are represented in Figure 13.

Figure 13: Range of motion of a linear actuator.

For a system 𝑆 subjected to linear motion that is driven by a linear actuator 𝑖, the ROE of the
system 𝑆 is a line segment defined by two points. This is given by Equation (16).

17
 𝑥 ,𝑦 ,𝑧 𝑅
𝑅𝑂𝐸𝑆 = [ 𝑖1 𝑖1 𝑖1 ] = [ 𝑖1 ] (16)
𝑥𝑖2 , 𝑦𝑖2 , 𝑧𝑖2 𝑅𝑖2

Note that all the ROE coordinates are aligned with the frame coordinates. If the system 𝑆 is
collaborating with the system driven by linear actuator 𝑗 that can move in the direction
perpendicular to that of 𝑆, the ROE of system 𝑆 is a rectangle defined by four points, which are
formulated by Equation (17).
𝑅𝑖1
𝑅𝑖2
𝑅𝑂𝐸𝑆 = 𝑅 + 𝑅 − 𝑅
𝑖1 𝑗2 𝑗1 (17)
[𝑅𝑖2 + 𝑅𝑗2 − 𝑅𝑗1 ]

Similarly, if the system 𝑆 is collaborating with the third system driven by linear actuator 𝑘 that can
move perpendicular to both actuators 𝑖 and 𝑗, the ROE of system 𝑆 is a rectangular prism defined
by eight points, which are represented by Equation (18).
𝑅𝑖1
𝑅𝑖2
𝑅𝑖1 + 𝑅𝑗2 − 𝑅𝑗1
𝑅𝑖2 + 𝑅𝑗2 − 𝑅𝑗1
𝑅𝑂𝐸𝑆 =
𝑅𝑖1 + 𝑅𝑘2 − 𝑅𝑘1 (18)
𝑅𝑖2 + 𝑅𝑘2 − 𝑅𝑘1
𝑅𝑖1 + 𝑅𝑗2 − 𝑅𝑗1 + 𝑅𝑘2 − 𝑅𝑘1
[𝑅𝑖2 + 𝑅𝑗2 − 𝑅𝑗1 + 𝑅𝑘2 − 𝑅𝑘1 ]

2.6. Machine Eligibility Determination

Machines that can perform the required manufacturing operations are determined by: 1)
determining if the machine contains a system that can perform the necessary manufacturing
operation suggested by ontology formulation; 2) comparing the generated manufacturing planes
with the ROEs of all available systems; and 3) comparing the directions of each manufacturing
plane with the system tool directions. Equation (19) describes a system ROE, 𝐴𝑅𝑂𝐸 , that contains
a manufacturing plane, 𝐴𝑚𝑓𝑔 :

∀𝑃 ∈ 𝐴𝑚𝑓𝑔 ⊂ 𝐴𝑅𝑂𝐸 (19)

where 𝑃 refers to any point inside the manufacturing plane 𝐴𝑚𝑓𝑔 . The direction of the
manufacturing plane must be opposite to the direction of the tool used for the manufacturing
activity required as formulated in Equation (20):

𝑛⃗𝑚𝑓𝑔 ∙ 𝑛⃗𝑡𝑜𝑜𝑙 = −1 (20)

where 𝑛⃗𝑚𝑓𝑔 and 𝑛⃗𝑡𝑜𝑜𝑙 represent the unit vectors of the manufacturing plane and tool, respectively.

18
3. Experimental Setup and Case Studies
Two wood frames are studied to validate the proposed framework: one simple wood panel with
only vertical studs, and a more complex panel with a window component. Both panels are built in
the 3D modeling software Revit using FrameX, a Revit add-on developed at the University of
Alberta, Canada [27]. Relevant specifications of wood panels obtained from BIM are presented in
Table 2.
Table 2: Summary of relevant panel specifications.
Panel Design Panel 1 Panel 2

Shop Drawing

Frame Dimensions X = 3048, Y = 2438.4, X = 3048, Y = 3048,

(mm) Z = 139.7 Z = 139.7
SP2, CH1, SH1, ST1,
Connection Type(s) SP1
JT1, TS2, CT1, SP1

The wood-framing machine prototype (WFMP), as shown in Figure 14, is used as the
manufacturing resource for this study. The WFMP is a semi-automated wood-framing machine
designed and built by the modular construction group at the University of Alberta, Canada. The
purposes of the machine are to increase the productivity and accuracy of the wood framing process.
A recipe file containing all the manufacturing instructions is first imported into the programmable
logic controller (PLC). An operator then places the top and bottom plates and the first stud into
place. The machine will secure the first stud by nailing both plates to the stud. Once nailing is
finished, a dragging system will grab the frame and drag it to the next operating location defined
in the recipe file. The above processes are repeated until the frame structure is finished. It has been
proved that the WFMP produces fast and accurate wood frame assemblies, which are critical in
residential buildings. Relevant machine specifications of the WFMP used in this study are
summarized in Table 3.

19
Figure 14: (a) Systems of Wood Framing Machine Prototype, and (b) details of nailing
system.

Table 3: Summary of WFMP machine specifications.

Work Work Range Collaborative Tool
System
Direction (mm) Systems Direction
Nailing System
𝑧 0 – 150 Dragging System (0, 1, 0)
(Left)
Nailing System Dragging System,
𝑧 0 – 150 (0, −1, 0)
(Right) Table Movement
Dragging System 𝑥 0 – 3200 - -
Table Movement 𝑦 2438 – 3658 - -
Cutting System
𝑧 0 – 150 Dragging System (0, 0, 1)
(Left)
Cutting System Dragging System,
𝑧 0 – 150 (0, 0, 1)
(Right) Table Movement
Drilling System 𝑦 0 – 100 Dragging System (0, 1, 0)

To secure the connections in a typical wood frame panel, nails are needed parallel to 𝑥𝑦 plane at
different height (along 𝑧-axis). This requires the machine to have 3 degree-of-freedom (DOF):
translational motion in 𝑥, 𝑦, and 𝑧 axes. In the WFMP, both left and right nailing system can move
along 𝑧-axis. At the same time, both nailing systems are working with dragging system that moves
the work piece (wood frame) in 𝑥 direction. As shown in Table 3, right nailing system is also
collaborating with the table that accommodates frame height in 𝑦 direction. As such, the WFMP
can achieve the DOF as needed.

20
4. Results
The previously mentioned wood frame panels are studied to validate the proposed framework. The
validation process can be broken down into four stages: 1) the locations of mating planes detected
by MEDS are inspected visually and compared with expected locations of contact area in actual
products, 2) the locations of manufacturing planes generated by the system are then compared to
knowledge obtained based on common practice, 3) the calculated ROEs of the WFMP are verified
with relevant machine specifications, and 4) the machine eligibility of the WFMP predicted by
MEDS is compared with knowledge held by design personnel who work with that machine. If the
output of every stage yields a positive result, it can be confirmed that the proposed approach is
validated.
4.1. Case Study 1

The expected mating planes for Panel 1 are the cutting faces of all the studs as they are in contact
with the plates. Based on expert knowledge, nailing operations are the most feasible manufacturing
operations to secure the building elements in Panel 1. As a result, all the manufacturing planes of
Panel 1 are expected to be parallel to the mating planes and on the outside surface of both plates.
Figure 15 (a) shows Panel 1 with all the automatically generated mating planes and Figure 15 (b)
additionally shows the location of all the manufacturing planes detected by MEDS. It can be
observed from Figure 15 that locations and geometry of both mating planes and manufacturing
planes are as expected.

Figure 15: Wood frame with (a) mating planes, and (b) mating planes and manufacturing
planes.
Figure 16 shows the predicted manufacturing planes of Panel 1 and the ROE of the nailing systems
of the WFMP. All the manufacturing planes are on the outside of the top and bottom plates.

21
Manufacturing planes on the bottom plate (left 𝑥𝑧 plane) are covered by the ROE of the left-hand
side nailing system. Manufacturing planes on the top plate (right 𝑥𝑧 plane) are included in the ROE
of right-hand side system. Since all the manufacturing planes are included in the system ROEs,
Equation (19) is satisfied. In addition, manufacturing planes and tool directions are summarized in
Table 4. It can be clearly shown in the table that Equation (20) is satisfied. As a result, the WFMP
is fully capable of manufacturing Panel 1 using its nailing systems.

Figure 16: Manufacturing planes of Panel 1 and ROEs of nailing systems.

Table 4: Manufacturing planes of Panel 1 and tool direction.

Intersection Manufacturing Plane Tool direction
𝑛⃗𝑚𝑓𝑔 ∙ 𝑛⃗𝑡𝑜𝑜𝑙
Type direction (𝑛⃗𝑚𝑓𝑔 ) (𝑛⃗𝑡𝑜𝑜𝑙 )
SP1 (0, −1, 0) (0, 1, 0) -1
SP2 (0, 1, 0) (0, −1, 0) -1

4.2. Case Study 2

Figure 17 shows the mating planes of Panel 2 detected by MEDS. As shown in the figure, Panel 2
comes with various types of mating planes because of the double-plate and window structure.
Expected and detected types and quantities of mating planes are listed in Table 5. It can be
observed in the table that the detected mating planes match the expected mating planes.

22
 Figure 17: Mating planes of Panel 2.
Table 5: Summary of mating planes of Panel 2.
NO. Intersection Type Detected Quantity Actual Quantity
1 SP2 13 13
2 CH1 10 10
3 SH1 4 4
4 ST1 2 2
5 JT1 2 2
6 TS2 2 2
7 CT1 5 5
8 SP1 15 15

Figure 18 represents detected mating planes and predicted manufacturing planes of Panel 2. The
location of all the manufacturing planes are as expected. For SP1 connections shown in Figure 18
(b), manufacturing planes are projected onto the outer surface of the plate as before. For JT1
connections shown in Figure 18 (a), manufacturing planes are not projected between the studs.
Instead, they are relocated to the outermost surface of respective building elements due to
manufacturing plane correction, as stated in Section 2.4. Figure 18 (c) represents the
manufacturing planes of double-plate connections. As expected, manufacturing planes are found
on the outer surface of the plate due to the merging of double-plate as illustrated in Section 2.2.2.
In summary, all of the manufacturing planes calculated using MEDS are as expected.

23
Figure 18: Mating planes and manufacturing planes of Panel 2. (a) JT1 connections, (b) SP1
connections, and (c) SP2 connections.
Figure 19 shows the manufacturing planes and ROE of nailing systems of the WFMP. The
manufacturing operation required to fabricate Panel 2 is nailing operation based on ontology
formulation; therefore, nailing systems are required to perform nailing operations. As shown in
Figure 16 and Figure 19, the ROE of left- and right-hand side nailing systems are identical.
However, only manufacturing planes on the bottom and top plates are covered in the ROE.
Manufacturing planes in the window component are outside the nailing system ROE. That is, only
manufacturing planes on the top and bottom plates satisfy Equation (19) and (20) and can be
fabricated using WFMP nailing systems. Additional effort is required to fully manufacture Panel
2, which is consistent with the current use of the WFMP in the manufacturing procedure of such
panels where window components are pre-assembled.

24
 Figure 19: Manufacturing planes of Panel 2 and ROE of nailing systems.

5. Conclusions and future work

As automated manufacturing of construction-oriented products is trending, process planning is in
demand to reduce error and increase the productivity of manufacturing activities. As a prerequisite,
the proposed BIM-based framework is intended to determine the machine capabilities before
process planning. In the case studies investigated in Section 4, the proposed framework (MEDS)
detects mating planes and calculates manufacturing planes of wood frames accurately. Besides, it
allows the software to decide whether the machine can perform the required manufacturing
operations or not. Implementation of MEDS results in several advantages. First, the proposed
framework is accurate and reliable as it does not require constant involvement of manufacturing
engineers to determine the machine capability. Second, the system provides feedback to building
engineers about if the designed wood frames can be fabricated by the designated machines. Frames
with infeasible manufacturing features will be spotted and corrected in the early design stage. This
saves the time and cost of potential late design changes due to non-manufacturable frames. Third,
the overall framework of the MEDS is transferrable to other BIM-based construction-oriented
products fabrication. That is, the intersection detection, manufacturing feature computation, ROE
calculation, and machine eligibility determination. As most offsite construction manufacturing
activities can be considered assembly processes, the intersecting regions as defined in this study
contain manufacturing features.
Nonetheless, future work is needed to extend the proposed framework to more realistic building
elements, including non-planar geometries and building elements oriented in 3-dimensional space.
Furthermore, MEDS considers linear motion for all the machine actuators, which is a limiting
factor that will need to be addressed for robotic environments, in which tools may be used from
any orientation. For example, header-to-cripple nailing operations are currently performed by
nailing diagonally instead of following standard manufacturing rules due to the excessive nail

25
length required otherwise (see Figure 20). A similar correction is needed for the ROE formulation
as rotational motions are not included in the proposed approach. While rotary actuators are beyond
the scope of this study, rotations in 3-dimensional space are constantly employed in manufacturing
activities as robot manipulators are getting more popular and are starting to be used in offsite
construction facilities. The aforementioned limitations affect the implementation of the MEDS to
real-world applications in different ways, however, once the infrastructure of intelligent
manufacturing in offsite construction is developed, manufacturing activities, such as generating
intelligent toolpaths, can be integrated into the manufacturing of construction-oriented products.

Figure 20: Header-to-cripple nailing operations

Acknowledgements
The authors would like to dedicate thanks to Mahmud Abushwereb, Gurjeet Singh, and all
personnel involved in the development of FrameX and the wood frame machine prototype. The
authors would also like to thank the National Sciences and Engineering Research Council of
Canada (NSERC) for financial support (Grant File No. IRCPJ 419145-15).

References
[1] “Panelized Building Systems.” [Online]. Available:
https://www.nahb.org/en/consumers/home-buying/types-of-home-construction/types-of-home-
construction-panelized-building-systems.aspx. [Accessed: 22-Mar-2019].
[2] N. H. R. Standards, “Frequently Asked Questions About the National Hr,” National
Institute of Building Sciences, 2011. [Online]. Available:
http://www.buildingsmartalliance.org/index.php/nbims/faq/. [Accessed: 22-Mar-2019].

26
[3] X. Yin, H. Liu, Y. Chen, and M. Al-Hussein, “Building information modelling for off-site
construction: Review and future directions,” Automation in Construction, vol. 101, pp. 72–91,
May 2019. DOI: https://doi.org/10.1016/j.autcon.2019.01.010.
[4] H. Liu, M. Lu, and M. Al-Hussein, “Ontology-based semantic approach for construction-
oriented quantity take-off from BIM models in the light-frame building industry,” Advanced
Engineering Informatics, vol. 30, no. 2, pp. 190–207, 2016. DOI:
https://doi.org/10.1016/j.aei.2016.03.001.
[5] P. M. Swamidass, Encyclopedia of Production and Manufacturing Management. Springer
Science & Business Media, 2006. DOI: https://doi.org/10.1007/1-4020-0612-8.
[6] Construction Global, “Top six benefits of offsite construction,” 2014. [Online]. Available:
http://www.constructionglobal.com/majorprojects/311/Top-six-benefits-of-offsite-construction.
[Accessed: 27-Mar-2019].
[7] N. Malik, R. Ahmad, and M. Al-Hussein, “Generation of safe tool-paths for automatic
manufacturing of light gauge steel panels in residential construction,” Automation in Construction,
vol. 98, pp. 46–60, Feb. 2019. DOI: https://doi.org/10.1016/j.autcon.2018.11.023.
[8] P. Martinez, R. Ahmad, and M. Al-Hussein, “A vision-based system for pre-inspection of
steel frame manufacturing,” Automation in Construction, vol. 97, pp. 151–163, Jan. 2019. DOI:
https://doi.org/10.1016/j.autcon.2018.10.021.
[9] C. Z. Li, J. Hong, F. Xue, G. Q. Shen, X. Xu, and M. K. Mok, “Schedule risks in
prefabrication housing production in Hong Kong: a social network analysis,” Journal of Cleaner
Production, vol. 134, part B, pp. 482-494, Oct. 2016. DOI:
https://doi.org/10.1016/j.jclepro.2016.02.123.
[10] N. C. Babic, P. Podbreznik, and D. Rebolj, “Integrating resource production and
construction using BIM,” Automation in Construction, vol. 19, pp. 539-543, Aug. 2010. DOI:
https://doi.org/10.1016/j.autcon.2009.11.005.
[11] F. H. Abanda, J. H. M. Tah, and F. K. T. Cheung, “BIM in off-site manufacturing for
buildings,” Journal of Building Engineering, vol. 14, pp. 89-102, Nov. 2017. DOI:
https://doi.org/10.1016/j.jobe.2017.10.002.
[12] L. Doumbouya, G. Gao, and C. Guan, “Adoption of the building information modeling
(BIM) for construction project effectiveness: the review of BIM benefits,” American Journal of
Civil Engineering and Architecture, vol. 4(3), pp.74-79, Apr. 2016. DOI: https://doi.org/
10.12691/ajcea-4-3-1.
[13] S. Le Roux, F. Bannier, E. Bossanne, and M. Stieglmeier, “Investigating the interaction of
building information modelling and lean construction in the timber industry,” World Conference
on Timber Engineering (WCTE 2016), Aug. 2016. [Online]. Available:
https://www.researchgate.net/publication/307205224_Investigating_the_interaction_of_building
_information_modelling_and_lean_construction_in_the_timber_industry. [Accessed: 14-Oct-
2019].

27
[14] P. Patlakas, A. Livingstone, and R. Hairstans, “A BIM platform for offsite timber
construction,” Education and Research in Computer Aided Architectural Design in Europe
(eCAADe), Sept. 2015. [Online]. Available:
https://www.researchgate.net/publication/323289056_A_BIM_Platform_for_Offsite_Timber_Co
nstruction. [Accessed: 14-Oct-2019].
[15] P. Patlakas, J. Menendez, and R. Hairstans, “The Potential, Requirements, and Limitations of
BIM for Offsite Timber Construction,” International Journal of 3-D Information Modeling
(IJ3DIM), vol. 4, Oct. 2015, pp. 54-70. DOI: https://doi.org/10.4018/IJ3DIM.2015010104.
[16] S. An, P. Martinez, R. Ahmad, and M. Al-Hussein, “Ontology-Based Knowledge
Modeling for Frame Assemblies Manufacturing,” in International Symposium on Automation and
Robotics in Construction, 2019. DOI: https://doi.org/10.22260/ISARC2019/0095.
[17] R. Y. Zhong, X. Xu, E. Klotz, and S. T. Newman, “Intelligent Manufacturing in the Context
of Industry 4.0: A Review,” Engineering, vol. 3, no. 5, pp. 616–630, 2017. DOI:
https://doi.org/10.1016/J.ENG.2017.05.015.
[18] M. Hermann, T. Pentek, and B. Otto, “Design Principles for Industrie 4.0 Scenarios,” in
2016 49th Hawaii International Conference on System Sciences (HICSS), 2016, pp. 3928–3937.
DOI: https://doi.org/10.1109/HICSS.2016.488.
[19] N. K. Jain and V. K. Jain, Computer Aided Process Planning for Agile Manufacturing
Environment. Elsevier Science Ltd., 2007. DOI: https://doi.org/10.1016/b978-008043567-
1/50027-9.
[20] X. Xu, L. Wang, and S. T. Newman, “Computer-aided process planning - A critical review
of recent developments and future trends,” International Journal of Computer Integrated
Manufacturing, vol. 24, no. 1, pp. 1–31, 2011. DOI:
https://doi.org/10.1080/0951192X.2010.518632.
[21] R. Ahmad, S. Tichadou, and J. Y. Hascoet, “A knowledge-based intelligent decision
system for production planning,” International Journal of Advanced Manufacturing Technology,
vol. 89, no. 5–8, pp. 1717–1729, 2017. DOI: https://doi.org/10.1007/s00170-016-9214-z.
[22] T. R. Gruber, “Toward principles for the design of ontologies used for knowledge sharing,”
International Journal of Human - Computer Studies, vol. 43, no. 5–6. pp. 907–928, 1995. DOI:
https://doi.org/10.1006/ijhc.1995.1081.
[23] R. Jardim-Goncalves, J. Sarraipa, C. Agostinho, and H. Panetto, “Knowledge framework
for intelligent manufacturing systems,” Journal of Intelligent Manufacturing, vol. 22, no. 5, pp.
725–735, 2011. DOI: https://doi.org/10.1007/s10845-009-0332-4.
[24] S. Lemaignan, A. Siadat, J. Y. Dantan, and A. Semenenko, “MASON: A proposal for an
ontology of manufacturing domain,” Proceedings - DIS 2006: IEEE Workshop on Distributed
Intelligent Systems - Collective Intelligence and Its Applications, vol. 2006, no. June 2014, pp.
195–200, 2006. DOI: https://doi.org/10.1109/DIS.2006.48.

28
 [25] S. Zhang, F. Boukamp, and J. Teizer, “Ontology-based semantic modeling of construction
safety knowledge: Towards automated safety planning for job hazard analysis (JHA),” Automation
in Construction, vol. 52, pp. 29–41, 2015. DOI: https://doi.org/10.1016/j.autcon.2015.02.005.
[26] Autodesk Inc., “Element Behavior in Revit,” Autodesk Help, 2016. [Online]. Available:
https://knowledge.autodesk.com/support/revit-lt/getting-
started/caas/CloudHelp/cloudhelp/2016/ENU/RevitLT-GetStarted/files/GUID-5BFA499A-
5ACA-4069-852C-9B60C9DE6708-htm.html. [Accessed: 19-Mar-2019].
[27] J. D. Manrique, M. Al-Hussein, A. Bouferguene, and R. Nasseri, “Automated generation
of shop drawings in residential construction,” Automation in Construction, vol. 55, pp. 15–24,
2015. DOI: https://doi.org/10.1016/j.autcon.2015.03.004.

29

View publication stats