Case Study

A Discrepancy Analysis of BIM-Based Quantity Take-Off for

Building Interior Components
Seongah Kim, Ph.D.1; Sangyoon Chin, Ph.D.2; and Soonwook Kwon, Ph.D.3

Abstract: The more detailed building information modeling (BIM) is, the more accurate are the quantities that can be extracted in BIM-
based quantity take-off. However, the time and cost typically allowed for modeling are insufficient to extract accurate quantities throughout a
Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

project’s life cycle. In addition, variations in quantity interfere with estimating accurate construction costs. Despite the recognition that
extracted quantities can differ depending on the details of building interior components, the number and extent of discrepancies, or whether
these would differ by material, is unknown. Therefore, this study identifies and analyzes the approximately 6–9% quantity discrepancy (QD)
in quantities extracted from different building interior components, in order to improve the accuracy of cost estimation using BIM. The dis-
crepancy was analyzed by comparing the extracted quantities via two different model types, specifically a compositely modeled object and an
individually modeled object. These models were applied to four cases of interior work in residential units. Results indicate that the QD occurs
non-sequentially and exceeds the loss rates for some interior materials. Furthermore, this study identifies the causative factors of the QD by
common features observed in the particular materials and suggests modeling and measurement strategies for minimizing QD. DOI: 10.1061/
(ASCE)ME.1943-5479.0000684. © 2019 American Society of Civil Engineers.
Author keywords: Building information modeling; Component; Quantity take-off; Discrepancy; Compositely modeled object; Individually
modeled object.

Introduction for accurately estimating the construction cost. Architects and esti-
mators have questioned who should be responsible for validating that
As the architecture/engineering/construction (AEC) industry adopts BIM built during the design phase contains the necessary information
building information modeling (BIM) in its construction projects, for cost estimation (McCuen 2015).
stakeholders such as owners, architects, and general contractors are In response to this problem, the Royal Institution of Chartered
paying more attention to ways to reduce time and cost, including Surveyors (RICS), the Quantity Surveyor Association in the UK,
cost estimation using BIM (Ghoilzadeh et al. 2018; Jin et al. 2018). defined a BIM guideline intended for cost managers, and an associ-
Research projects have shown that estimation using BIM reduces ated procedure for validation (RICS 2015). The guideline includes a
working time and errors, in addition to improving the estimator’s process to confirm whether the quantities suitable for the measure-
performance when compared to traditional estimation practices ment rule can be obtained from BIM when checking the quantities
(Staub-French et al. 2003; Shen and lssa 2010; Peterson et al. of BIM generated in the design stage. Monteiro and Martins (2013)
2011). However, the use of BIM estimation poses a number of chal- suggested modeling and measuring building elements to extract
lenges: (1) insufficient estimator knowledge and understanding of quantities suitable for a BIM the measurement rule. Nevertheless,
BIM; (2) implementing data exchange between various applications there are constraints in validating the quantities measured from
such as the BIM authoring tool and estimation programs; and (3) BIM owing to quantity differences dependent on the details of BIM
limitations to the maintenance of the relationships between cost model components. Researchers have mentioned the discrepancy in
data and building elements modeled in three-dimensional (3D) an extraction quantity caused by the modeling composition and
objects (Aibinu and Venkatesh 2014; Wu et al. 2014). Furthermore, detail of the included objects (Monteiro and Martins 2013), and the
it is difficult to trust the quantity data automatically extracted from concern that variations in quantities extracted from BIM can inter-
BIM, because BIM generated by architects lacks sufficient properties fere with accurate estimation of construction costs.
In element-based estimating, a discrepancy in the quantity
extracted from building components can hinder the consistency of
1
Research Fellow, School of Civil Architectural Engineering &
Landscape Architecture College of Engineering, Sungkyunkwan Univ.,
quantities. Because the quantities used for generating a bill of quan-
Seobu-ro 2066, Suwon 16419, South Korea. Email: [email protected] tities (BOQ) in the design stage serves as the basis for determining
2
Professor and Chair, School of Civil Architectural Engineering & the tender price and comparing the appropriateness of the construc-
Landscape Architecture College of Engineering, Sungkyunkwan Univ., tion cost when concluding a general contract, it is crucial to measure
Seobu-ro 2066, Suwon 16419, South Korea (corresponding author). accurate quantities that have less risk of increasing or decreasing
ORCID: https://orcid.org/0000-0002-4639-9649. Email: [email protected] the total building cost during construction (Ashworth and Perera
3
Professor, School of Civil Architectural Engineering & Landscape 2015; Hyari 2016). In addition, measured quantities can be reflected
Architecture College of Engineering, Sungkyunkwan Univ., Seobu-ro in subcontracts with other companies. Therefore, it is important to
2066, Suwon 16419, South Korea. Email: [email protected]
estimate quantities consistently in the tendering stage. A BIM
Note. This manuscript was submitted on May 26, 2018; approved on
September 10, 2018; published online on January 16, 2019. Discussion pe- model divided into architecture, structure, and mechanical, electri-
riod open until June 16, 2019; separate discussions must be submitted for cal, and plumbing (MEP) can be generated at different levels of de-
individual papers. This paper is part of the Journal of Management in velopment (LOD) for each element by consultation among partici-
Engineering, © ASCE, ISSN 0742-597X. pants (BIMForum 2017). Thus, the models and components that

© ASCE 05019001-1 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

 comprise BIM can be expressed in various ways that can cause dis- Many studies have tried to extract the required QT information
crepancies in the resultant quantities. automatically using the ontology and have tried various approaches to
This study aims to identify the cause of the quantity discrepancy overcoming the limit of manual data connection as well as data
(QD) in models with different details, and to find a solution to exchange with another domain (Abanda et al. 2017; Liu et al. 2016;
ensure the consistency and reliability of quantity take-off (QT) Lee et al. 2014). Wei et al. (2017) developed a method to calculate the
using BIM. To examine discrepancies in the quantities extracted quantity of auxiliary materials required for a timber formwork using
from models with different details, this study defines two ways of BIM. In addition, using the extracted BIM information, measuring
modeling a component and compares the quantities extracted from the cash flow to simulate a financial plan according to a contract type
these; one approach is to model a group of interior components in a (Lu et al. 2016), and analyzing the daily labor productivity of struc-
single object, and the other is to model interior components in sepa- tural formworks based on properties of 3D objects (Lee et al. 2017)
rate objects. This study also suggests a solution for minimizing the have been studied. Furthermore, Niknam and Karshenas (2015) pro-
discrepancies in measured quantities from the models used for the posed a web-based construction cost estimation application based on
case studies. sharing BIM data among participants to reduce human involvement
in repetitive estimation work.
Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

However, not all quantities of materials needed for construction

Literature Review
cost estimation can necessarily be extracted from the BIM model
(Aibinu and Venkatesh 2014; Monteiro and Martins 2013; Zhiliang
Traditionally, cost estimation in a construction project is largely di-
et al. 2011). Shortfalls of 14% were reported in BIM estimation of
vided into two processes: (1) QT for materials; and (2) cost and gen-
the costs of concrete, drywall, and door work in a project that
erating a BOQ. QT is conducted with measurement rules, such as
adopted BIM at Sutter Medical Center, Castro Valley, and the short-
the UK RICS’s “New Rules of Measurement” or estimation man-
fall was retrieved with traditional estimation. The reason that BIM
uals offered by the owners and government as a standard to calcu-
estimation was performed along with the traditional estimation was
late the required quantities of construction resources (Abanda et al.
that aspects of the construction could not be modeled as a 3D object,
2017). Since the 1990s, QT has been determined by importing
dimensions needed for mensuration formulas from computer-aided nor the quantities calculated from the properties of the objects
design (CAD) drawings (Saleh 1999; Seeley and Winfield 1999). (Denerolle 2013). This made it necessary to use traditional practices
More recently, QT has been determined by directly extracting quan- in QT in parallel with BIM. RICS divided the BIM measurement
tities that comply with existing measurement rules from 3D objects into three categories: automated, derived, manual take-off. The last
in BIM (Staub-French et al. 2003; Jrade and Alkass 2007). As the is a traditional estimation practice that does not use BIM at all (Wu
property values of components and materials are assigned in the et al. 2014).
process of modeling the objects, the quantities can be extracted Regarding the LOD of the BIM model, RICS mentioned a need to
from the property values of the BIM objects. This is different from verify the extracted quantities from a BIM model and to check the
traditional estimation where mensuration formulas are calculated quantity of materials required from the perspective of the quantity
and the dimensions are imported into CAD drawings (McCuen surveyor (QS), excluding the designer’s consent and approval. There
2015; Shen and lssa 2010). The process of determining QT and the is a BIM execution plan (BEP) that includes all BIM tasks, processes,
estimation that follows using BIM is called model-based estimating and information determined by an agreement of stakeholders and
(Monteiro and Martins 2013; Wu et al. 2014; Borhani et al. 2017). deals with the minimum scope of BIM needed for data exchange
Since the traditional QT approach takes 50–80% of the overall (BIMForum 2017). The QT result is completely unacceptable when
working time in estimations, extracting data from the model means automatically extracted from the unsatisfactory BIM and information
that a QT estimated with BIM can enhance the productivity and ac- that does not comply with the QS requirements; clear criteria and
curacy of an estimator’s work (Lawrence et al. 2014; Peterson et al. ways to validate the BIM from the QS viewpoint have not been estab-
2011). Estimation performance has improved because BIM enables lished yet (Wu et al. 2014).
estimators to check in advance for errors and omissions in a design Although different LODs are applied throughout a project’s life
(Shen and Issa 2010). cycle, it is necessary to find a way to at least determine, and prefer-
To increase the use of BIM, some preceding studies identified entially to minimize, the discrepancy caused by different modeling
activities as connections between 3D objects and cost data and pro- approaches before determining the LOD of BIM. This is because
posed an integration of data with regard to cost and time (Lu et al. building a BIM with a high LOD requires 2–11 times more effort
2016; Staub-French et al. 2003). Another study proposed a system than building one with a low LOD (Leite et al. 2011). Modeling
driven by queries based on industry foundation classes extensible with a high LOD offers the advantage of producing clear and reli-
markup language (IFC xml) rather than a user’s manual designation able results from BIM but causes a concomitant increase in working
of cost data connected with an object. This allows for easy estima- hours and costs, which obstructs the swift decision-making required
tion by design revisions (Lawrence et al. 2014). Some studies throughout a project. As such, Nassar (2012) noted that it is neces-
focused on approximate estimation, an approach that connects the sary to have an estimation framework that can extract or derive
quantities extracted from a space model to the cost data of a para- quantities from relatively simple schematic models. Some preced-
metric estimation model. Thus, the construction cost in either the ing studies proposed minimum BIM modeling criteria required for
initial or the design stages of a project can be estimated (Jrade and measurement rules (Monteiro and Martins 2013; Wu et al. 2014).
Alkass 2007; Choi et al. 2015). Despite these studies and the various However, the differences in the quantities extracted from objects
uses and benefits of BIM, QT with BIM is yet to be positively with different LODs have not yet been clarified, and there is no evi-
applied (Ghoilzadeh et al. 2018; Boktor et al. 2014) due to a lack of dence to project the variations in quantity and cost as the BIM
experience with BIM and universal established estimation prac- model becomes more detailed. Nassar (2012) noted differing QTs
tices. Currently, QT with BIM still requires a manual process to using BIM results by work type, and Monteiro and Martins (2013)
integrate 3D object data and cost data, and review is necessary to found that quantity discrepancies may occur following the use of a
make sure that the QT from BIM complies with a measurement rule composite object. Motivated by those studies, this study: (1) deter-
(McCuen 2015; Olatunji and Sher 2014). mines the discrepancy of the quantities extracted from models

© ASCE 05019001-2 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

 generated with different details and components, and (2) aims to units. The 59- and 84-m2 units are typical housing sizes, compris-
determine causative factors of quantity discrepancies and to suggest ing approximately 75% of apartments and 65% of the total hous-
a way to maintain consistency in the quantity of BIM. The results ing supply in Korea as determined in 2015 (Statistics Korea
can be utilized to evaluate and better comprehend the changes in 2016). KLHC is a government-owned enterprise that has built
extracted quantities from models with different LODs by reviewing approximately 50,000 public housing units per year for the past
the consistency of the quantities. four years and is operating 800,000 public rental houses. Notably,
this study selected apartment units with different floor area sizes
to determine whether there is any relationship between quantity
Research Scope and Methodology discrepancies and floor areas.

Research Design and Scope Definition of Modeling Interior Components

To compare the discrepancies in quantities between models with dif- Most BIM authoring tools represent various objects, such as a slab,
ferent details, this study confines the scope of QT to interior building wall, column, roof, stair, and zone, by object-based parametric
Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

works. Nassar (2012) showed that interior components, such as modeling. These objects have various properties to represent com-
wood, plastic, protection, and finishes, are prone to less accurate ponents of buildings (Eastman et al. 2008). When modeling build-
estimates than other components, such as concrete, masonry, doors, ing components, they can be represented as a composite object or
and windows. In addition, the frequency of interior work modeling an individual object depending on the materials. Fig. 1 presents two
is lower than other work, while still offering high potential value waterproofed walls modeled in different ways, with cement mortar
(Jones and Laquidara-Carr 2016; Kim et al. 2017). If a quantity dis- and tiles. These walls can be represented differently according to
crepancy is found after the contract is underway, it poses the risk of the composition of materials for the object, regardless of the func-
increased construction costs and reduced quality. Thus, this study tion supported by the BIM authoring tool. This study defines the
limits the scope of QT to interior works, including masonry, wood, components for modeling interior materials in two ways, as pre-
thermal and moisture protection, insulation, and finishes. sented in Fig. 1. The first is a compositely modeled object (CMO)
Residential units in apartment housing were selected as cases that represents a group of materials with a single BIM object, and
for modeling and QT in this study because one in four residential the second is an individually modeled object (IMO) that represents
buildings built annually in Korea is apartment housing. The materials with separate BIM objects.
design and spatial layouts for residential units are almost standar- BIM users can select either of these modeling methods accord-
dized in Korea, and construction companies have internal estima- ing to their strategy and a given time period for modeling interior
tion manuals for cost estimation and budgeting. Four residential components, because CMO is a more time-efficient method than
unit types built by the Korea Land and Housing Corporation IMO. However, CMO has difficulty extracting quantities from BIM
(KLHC) were selected for study: 59-, 74-, and two types of 84-m2 for particular components since all the components of CMO are
assumed to have identical length and area, while IMO makes it pos-
sible to model and extract quantities for components installed in a
particular area or partially installed components. Therefore, consid-
eration must be given to adjusting quantities extracted from CMO
for materials whose quantities are over- or underestimated due to
their installation characteristics in a specific aspect of interior work.
To identify interior components whose quantities are over- or
underestimated, and suggest a way to reduce any excessive discrep-
ancy, this research focusses on the discrepancy of quantities
extracted from the two different models.

Data Collection and Analysis

Fig. 1. Definitions of modeling interior components: (a) CMO; and Two types of 3D models (CMO and IMO) were built for the four
(b) IMO. cases, and the quantities of materials for the interior works were
extracted from these models (Fig. 2). The interior components were

Fig. 2. Research procedure.

© ASCE 05019001-3 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

 modeled by considering the construction sequence and installation issue of quantity types arising in data exchange between BIM
methods of the interior materials of each unit. The QT for the inte- authoring tools. The same quantities can be obtained from other
rior works was determined according to the KLHC estimation man- BIM authoring tools producing similar QT results to those of this
ual (KLHC, “Cost Estimating Manual,” unpublished internal data, study if the interior materials are modeled in the same way as
2009). Quantity discrepancies were verified by comparing the quan- defined in this study.
tities extracted by the CMO and IMO models. As reported from a BIM tool to Microsoft Excel, the QT table
It is worth noting that even if two models were created in the was used to link the materials and quantities that comply with
same way in BIM, the types of quantity data extracted from the KLHC’s cost estimation manual, and then it was imported to MS
models could differ depending on the BIM tools used (AutoDesk Access. The imported data was arranged for sorting QT results, and
Inc. 2017; Graphisoft 2017). The buildingSMART alliance defined R and R Studio software was used to analyze the quantity discrepan-
the data and properties that are required in their Information cies. The quantities were assumed to be extracted from the model
Delivery Manual (IDM) for seamless data exchange when estimat- without any error. Those quantities extracted from the IMOs were
ing a conceptual design (NIBS 2015). Along with the required prop- set as the target variable because the IMO is more detailed than the
erties for QT, IDM defines 17 data types, such as beam, column, CMO, and therefore more closely representative of the actual con-
Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

curtain wall, door, equipment, opening, slab, stair flight, wall, and struction status. The QD was defined as the ratio of the difference
window. It defines the elements of the floor, roof, and landing as a between the target variable of each material in the IMO and CMO
slab type, but it does not include the ceiling. By modeling the inte- models. Through this data collection, the patterns and causative fac-
rior materials placed on the floor and ceiling with a slab object, this
tors of the QDs found in the four cases were analyzed. The causes
study reflects the quantity types required by IDM, as listed in
of the QDs were validated by refining the data collected from R and
Table 1, to compare the quantities extracted from different models.
R Studio and removing the outliers. Notably, the cases that
The focus is on identifying the discrepancies between native quanti-
exceeded the typical loss rates for construction materials (i.e., the
ties measured from BIM models, and the study does not address the
range of QD acceptance) were surveyed in detail. By synthesizing
the findings from these cases, a modeling guideline was proposed to
minimize the quantity discrepancy caused by differences in the
Table 1. Required quantities in Information Delivery Manual
methods of setting up components and maintaining consistency in
Quantity type Description extracted quantities using BIM.
Slab thickness Length measure along the extrusion axis
Slab perimeter Aggregate length measure of all faces parallel to
Modeling and Quantity Take-Off
the extrusion axis
Slab gross area Area measure of the top face of the slab
Wall height Length measure of the wall along its extrusion Typical Interior Materials for Modeling a Residential Unit
axis (base to top)
Typical interior materials of a residential unit in Korea are catego-
Wall length Length measure of the wall along its reference-
rized by room type, with types such as entrance, living room,
line axis
kitchen, bedroom, walk-in wardrobe, bathroom, and terrace. The in-
Wall gross side area Area measure of the two vertical (or near-vertical)
terior materials installed in these rooms can then be further catego-
faces of the wall that are parallel to its reference
rized again as floor, wall, or ceiling materials. The interior materials
axis
listed in the KLHC cost estimation manual are similarly classified

Material Family Walk-in Non-

Element Entrance Kitchen Bedroom Bathroom Terrace
Type Room wardrobe Assigned
Cement Cement
Mortar
Rough EPS/EPP, Floor Heating Waterproofing, Waterproofing,
Bed
Floor Mortar Bed Mortar Bed -
Stone
Finish Laminated Wood Porcelain Tile Porcelain Tile
Tile
Waterproof Masonry,
Gypsum Cementitious Fixed
Gypsum Board/ Gypsum board/ Hand trowel/
Rough Board/ Waterproofing, Partition
Cement Plaster Cement Plaster Cement plaster
Cement Mortar Bed
Wall Plaster
Finish Paper Tile, Paper Paper Ceramic Tile Paint
-
Baseboard Medium-density fiberboard (MDF) - Paint
Suspension Ceiling System Suspended
Rough Hand trowel
Gypsum Board Ceiling
Ceiling Finish Paper SMC Tile Paint -

Molding Polyvinyl chloride (PVC) panel - -

Work Rough Thermal and Moisture Protection, Plaster Partitions

Type Finish Paper covering, Tile, Wood Tile Painting -

Fig. 3. Classifications of interior materials for residential unit.

© ASCE 05019001-4 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Different IMOs by installation sequence.

by room and installed part and sorted into seven functional work Table 2. Result of modeling cases
types: partition, thermal and moisture protection, plaster, carpentry,
Num. of interior components (object)
covering, tile, and painting (Fig. 3). These work types do not include File size
components of the necessary interior work, such as doors, windows, (KB) Floor Wall Ceiling Total
miscellaneous metals, and furnishings, because the work types are Unit CMO IMO CMO IMO CMO IMO CMO IMO CMO IMO
modeled by a product with particular specifications that are not sig-
nificantly different to the quantity extracted from the model gener- 59AL 4,270 5,065 10 34 42 340 10 25 62 399
ated at LOD 200–300. Nassar (2012) has found, depending on how 74A 4,669 6,509 14 37 68 385 11 38 93 460
the components were modeled, estimated quantities of concrete, 84E 4,022 5,002 13 49 50 375 13 39 76 463
84EH 7,848 9,332 15 41 104 553 15 40 134 634
door, and window work components to be more accurate and with-
out discrepancies than other works.
In Fig. 3, the interior materials are classified into two types: Fig. 2. The file size of the modeled CMO was counted about 4 MB
rough and finish. The finish types of interior materials are used for in each residential unit, but the file size of 84E-type was 8 MB with
finishing and decorating a space, while the rough types contain 134 objects (Table 2). For IMO, the file size was counted 1–2 MB
items installed before the finish materials are applied to ensure the larger than CMO in all types of units, and the number of objects was
quality of a space. Rough items include mortar, waterproofing, and approximately 300 to 400 more, as listed in Table 2. The file size
insulation. In Korean residential spaces, the floor is generally cov- has increased depending on the number of objects. Despite the dif-
ered with floor heating systems and laminated wood, while the ference of 10 m2 in area from 74A and 84E, the reason for little dif-
walls are covered with wallpaper. Ceilings are generally finished ference in the number of objects is a difference in the plan design of
with a suspension ceiling system that consists of gypsum board and the residential unit, as presented in Fig. 2.
paper. Bathroom and terrace floors are covered with waterproofing As mentioned in the section Data Collection and Analysis, the
and ceramic tiles in preparation for frequent water usage and result- interior materials were modeled using wall and slab objects to sat-
ant damp conditions. Thus, the interior materials vary according to isfy the quantity type required for the IDM, as listed in Table 1. The
space, usage, and function, and one or more materials are applied to interior materials placed on the wall were modeled using a wall
an interior component, such as the floor, wall, or ceiling. object, and the materials placed on the floor and ceiling were mod-
The interior materials in the CMO model are represented by eled using a slab object. The quantities extracted from the CMO and
three objects corresponding to the floor, wall, and ceiling for each IMO models were applied to calculate the QD, which was defined
room. Conversely, the interior materials in the IMO model are rep- by the ratio of the difference in the quantities extracted from the
resented as separate objects, where quantity discrepancies can be CMO and IMO, as given in Eq. (1):
observed depending on the sequence of interior materials installed,
QCMO  QIMO
as illustrated in Fig. 4. Fig. 4(a) gives a typical case where a gypsum QD ¼ (1)
board is installed after the floor heating system, while Fig. 4(b) QIMO
gives a case where a gypsum board is installed after the expanded
polystyrene (EPS). The area of the gypsum board varies according The total number of QDs calculated from the CMO and IMO
to the installation sequence, thickness of screed, and floor heating. models for four units was 535, and this is illustrated in the histo-
This study selected type (a) of Fig. 4 for use in a quantity compari- grams given in Fig. 5. The QDs occur non-sequentially over a range
son between the CMO and IMO models. of −1 to 1, and about a half of the QDs approach 0. This means that
a QD takes place for a fraction of the interior materials depending
on the modeling method. Some QDs show large values, and others
Modeling and Quantity Take-Off for Cases
approach 0 or 61 (6100%). A QD close to 0 means that the
This study built two types of models of each residential unit type for extracted quantities of the interior material do not differ between
collecting data on the QDs that occurred between the CMO and the CMO and IMO models, whereas a QD close to 1 or −1 means
IMO by using a BIM authoring tool, ArchiCAD, as presented in that the model could not extract the quantity of material from the

© ASCE 05019001-5 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Histogram of quantity discrepancy.

IMO or CMO models. If a material component for measuring is not Table 3. Analysis of quantity discrepancy
represented in the CMO, the extracted quantity of the material is Data Case Min. 1st Qu. Median Mean 3rd Qu. Max. SD
zero and the QD becomes −1. In all four units, more than 10 types
of materials with QDs of −1 were discovered. A negative QD indi- QD 59AL −0.379 −0.003 0.008 0.059 0.083 0.758 0.154
cates that the quantity extracted from the CMO is less than that — 74A −0.313 −0.006 0.007 0.065 0.082 0.807 0.183
extracted from the IMO, and conversely, a positive QD indicates — 84E −0.723 −0.007 0.013 0.065 0.097 0.878 0.237
that the quantity extracted from the CMO is greater than that — 84EH −0.390 −0.009 0.003 0.088 0.121 0.830 0.212
extracted from the IMO. |QD| 59AL 0 0.004 0.028 0.090 0.119 0.758 0.138
— 74A 0 0.006 0.037 0.110 0.149 0.807 0.160
— 84E 0 0.011 0.044 0.135 0.173 0.878 0.205
Analysis of Quantity Discrepancy
— 84EH 0 0.006 0.032 0.126 0.157 0.830 0.191
As listed in Table 3, the mean of the QDs was calculated to range
Note: Max. = maximum; Min. = minimum; Qu. = quartile; and SD =
between 5.9 and 8.8% for the four cases, excluding 48 items with
standard deviation.
QDs of −1 (Fig. 5). The mean and standard deviation (SD) of the
QD increases as the floor area of the residential units increases. The procurement, whereas a QD exceeding the loss rate could cause a
mean calculated for the QDs implies that the quantities estimated cost variation. Therefore, this study presumed that a QD exceeding
by the CMO model are approximately 6 to 9% higher than those the loss rate would vary the cost. It was reanalyzed by adding the
generated by the IMO model, which could imply that the quantity data of the loss rate and work type to the QD results, as presented in
extracted from the CMO has more contingency quantity than that Fig. 6, with the added loss rate listed in Table 4.
from the IMO. As presented in Fig. 5 and listed in Table 3, the QDs are widely
The mean of the QDs in Table 3 is a sum of the negative and pos- distributed between 1 and −1, deviating from the first and third
itive QDs. Because the QDs summed to obtain the mean were offset quartiles. Considering that the loss rate of construction materials in
by the negative QDs, the mean of the absolute values of the QDs Korea is between 3–10% depending on the material, 157 QDs are
was recalculated as 9–14%, as given in Table 3. In particular, the distributed between zero and their loss rates, while the rest of the
difference between the means of the QDs and absolute QDs was the QDs (378) are distributed outside the range between zero and their
highest at 7% in the 84-m2-type unit because the SD of the QD for loss rates. Among 378 QDs, 48 QDs are −1, which means the corre-
this type is larger than that for the other units. In addition, the me- sponding material is not represented in the CMO so that no quantity
dian between the absolute QDs and the QDs was similar for all four can be extracted from BIM. Materials common to four units with
units, showing a difference of only 2–3%. QDs of −1 were analyzed in order to determine any common fea-
Not all the QDs found in this study affected a variation of cost tures. The materials whose QD were not −1 were sorted by material
and quantity. When procuring construction materials, general or and work type and used to determine a causative factor impacting
subcontractors purchase more than the measured quantity of the the quantity variation when using BIM. In summing up these inves-
materials while considering the potential loss. A QD less than the tigations, a solution is suggested for minimizing the discrepancy of
loss rate of materials does not raise the quantity of materials for measured quantities from the model (Fig. 6).

© ASCE 05019001-6 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

 Fig. 6. Analysis procedure of quantity discrepancy.
Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

Table 4. Loss rate of interior materials Causative Factors of Quantity Discrepancy

Most QDs are caused by the property and representation of compo-
Work type Materials Rate (%) nents in the CMO and IMO. For example, Fig. 8 gives the difference
Thermal and moisture Styrofoam insulation, EPS/EPP 10 between the CMO and IMO representations in the example case of
protection Gypsum board with bonding 5 insulation installed around an entrance door. The quantity of insula-
Gypsum board with nail 3 tion for the entrance door is extracted from both the CMO and IMO
Cement waterproofing, mortar bed 5 models. However, there are differences in the insulation quantity
Plaster Lightweight concrete for floor heating 2 extracted from the CMO and IMO because there is an area not pre-
Cement plaster 15 sented in the CMO. If the insulation quantities around the door were
Partitions Brick, gypsum board 3 not extracted from the CMO model at all (that is, if QD is −1), the
Tile Ceramic, porcelain, SMC 3 difference can be addressed by manually estimating the entire quan-
Stone 5 tity of insulation. However, in a case such as that in Fig. 8, the quan-
Wood Laminate floor, baseboard 10 tity of the partly not-expressed insulation can be easily ignored
Painting Water-based paint 5 because the missing part cannot be visually identified in the model.
Paper Paper covering 10 It is not easy to derive the quantity of material not expressed in the
Metal Suspended ceiling 10 CMO and only partially installed on an element.
Etc. (etcetera) — 3 Fig. 8 depicts a case of a negative QD generated owing to the
representation of the model. However, the QD according to the rep-
resentation of the model does not always result in a negative num-
Quantity Discrepancies Outside the Range of Zero to ber. For example, the height of paint estimated for the balcony wall
Loss Rates by the CMO is 2,620 mm, whereas the IMO estimates it as
Among the materials for which the QDs are distributed outside the 2,490 mm, as given in Fig. 9. In the CMO, all materials on the wall
range between zero and their typical loss rates, there were five types have the same heights. By contrast, each material on the wall can
of materials with QDs of −1 in all four cases. These materials were: have different heights in the IMO according to the installed area.
(1) the insulation installed on the wall and ceiling, (2) the mortar CMO estimates can be higher than IMO estimates, depending on
bed on the floor for water tightness, (3) the bullnose tiles partially how the interior components are modeled, especially on the wall. It
installed on the entrance floor along with ordinary tiles, (4) the bull- is also possible to apply different heights for each material on the
nose trim tile on the floor of the bathroom, and (5) the curtain mold- wall in the CMO, but this requires additional work to remove
ing on the ceiling. These materials are typically installed in a spe- the area of the door and window from consideration. Therefore, the
cific area and on a specific element and could not be represented QDs produced by the models’ representations can be observed
effectively in the CMO because the CMO can represent only the simultaneously as negative and positive.
materials that are evenly applied to an area and/or element. As such, Moreover, KLHC’s estimation manual specifies that the quanti-
although the materials were not represented in the CMO, they could ties of materials for the interior finish type should not be calculated
be represented and were extractable in the IMO because they are on areas covered by a bathtub or built-in furniture. Fig. 10 presents
modeled with a separate object. a kitchen in a residential unit that is modeled by the CMO. Here, the
For the 330 materials with QDs > −1, the QDs were grouped quantity of the finish-type material on the wall should be the area of
by work and interior surface type, as presented in Fig. 7. Here, the kitchen wall excluding the area of the sink. However, the quan-
the negative QDs are distributed mostly around work types of tity of the finish-type materials extracted from the CMO includes
thermal, moisture protection, and plaster, while the positive QDs the unnecessary area that is covered by the sink, which the manual
are distributed around work types of wood, tile, paper, and manu- requires excluding, resulting in an overestimation of the required
factured products (related to molding and baseboard) and are quantity of porcelain paver tiles. Therefore, if the area that the built-
mostly of the finish type. Because a positive QD means that the in furniture covers is not specifically excluded from the quantity
quantity extracted from the CMO is larger than that from estimation, the unnecessary area for the QT in the CMO model will
the IMO and the mean of the QD is positive in all the units, if the lead to greater estimated quantities than required.
quantity from the CMO is reflected in the construction costs, an Last, it is found that the QDs caused by the thickness of rough-
appropriate budget can be determined. However, with regard to type materials can occur in the lengths and areas of finish-type ma-
thermal, moisture protection, and plaster work, there exists a risk terial. A QD may occur because of a difference in lengths caused by
of going over budget. To find out why the QD varies by work and the thickness of an object on top of which it was modeled. In the
interior type, the causative factors of the QD in these three types CMO, all types of materials contained in objects have the same
are analyzed. quantities extracted from the object, whereas in the IMO they have

© ASCE 05019001-7 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Distribution of QDs outside the range of zero to material loss rate except when QDs are −1.

different quantities extracted from the object corresponding to spe- than the mean of the total QDs in the four units (Table 3 and
cific materials, such as the lengths given in Fig. 11. Here, the Fig. 12). Therefore, the QDs caused by changes in object length
lengths of insulation (lR) and wallpaper (lF) in the living room are were found to have a nonsignificant influence on the quantity varia-
shown to be different in the IMO model. Conversely, since the wall- tion, even though some of the materials exceeded the loss rate.
paper and insulation are modeled as one object in the CMO, the The QDs caused by the representation of the model coexist as
length of the wallpaper is the same length (lR) as the insulation. both positive and negative values and are widely distributed in all
However, the QDs caused by the thickness of an object are observed cases (Fig. 12). There were 146 QDs, out of which 123 were
to be lower than those from other causes because insulation is one found to exceed the loss rate (Table 5). The largest cause of QDs
of the thickest interior materials in a residential unit, and the thick- was limited model representation and unnecessary modeling in
ness of insulation on a wall is typically no more than 230 mm in the CMO.
Korea.
The causative factors (CFs) of the QD can be summarized as
falling into one of three categories: (1) representation of a model, Discussion
(2) unnecessary modeling of areas excluded from the QT in a CMO,
and (3) changes in the object length owing to the thickness of previ- This study found that 6–9% of QDs occurring in each residential
ously modeled objects and materials. It is difficult to explain the unit was due to the approach to modeling the interior components,
QD of any material with a single factor because these factors are and that the quantity of some materials could not be extracted by
compounded for each material. Therefore, CFs that exerted the the CMO model. When use of the interior model is close to LOD
greatest influence on the 330 QD results are identified, and the QDs 200, such as in the case of CMO, manual take-off should be
are reanalyzed with consideration of the CFs. The reanalysis of the applied to calculate the quantities of any materials with a QD of
QD results (e.g., Fig. 12) showed that the inclusion of unnecessary −1 in this study, that is, those for which quantity estimates were
areas for the QT is the largest cause of QDs in all cases, followed by not extracted from the CMO model. If the model is used to extract
the way a model is represented. quantities, it is necessary to consider whether 6–9% of QD is ac-
Fig. 12 gives an increasing trend of positive QDs by unnecessary ceptable at the stage of adopting BIM for a housing project. The
area as the area of a residential unit increases, which is thought to be quantities extracted from CMO include the unnecessary area in
caused by the increase in the area of built-in furniture, such as bath- QT, and thus the derived take-off can be lower than the 6–9% of
rooms, entrances, kitchens, and walk-in wardrobes, incorporated QD by the exclusion of the unnecessary area. The QDs caused by
into each unit. The frequency of QDs by an unnecessary area was different modeling methods for an interior component are widely
50, and all of them exceeded the loss rate (Table 5). Of the QDs distributed across positive and negative values, meaning that
caused by changes in the length of an object, 157 out of 291 quantities of some materials are overestimated while others are
exceeded the loss rate. The mean of these 157 QDs was not greater underestimated. This might be not perceived as a risk because the

© ASCE 05019001-8 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

 quantity extracted from a CMO is greater than that from an IMO, Owners and project managers should also recognize that an
on average. However, if the quantity extracted from a CMO was unintended quantity discrepancy can occur depending on the repre-
applied to the BOQ for subcontracting, the quantities of some sentation and detail used in a given BIM. A CMO-based QT does
materials would be overestimated while others would be underes- not necessarily generate greater quantities. This limitation should
timated. When using a low-detail model like the CMO, project be handled in advance through modeling guidelines for general
stakeholders should review their quantities and consider which contractors and subcontractors and applied to a BIM project execu-
materials may be overestimated and underestimated, based on the tion plan in collaboration with the participants. This paper proposes
cases in this study. the following guidelines for preventing QDs caused by modeling
interior components by referring to the BIM measurement of
RICS, as mentioned in the literature review.
1. Manual take-off should be applied in cases where the quantity
required by the measurement rule is not generated and extracta-
ble from the detail and representation of a model. Such cases
mostly include materials installed partly in the element and not
Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

evenly applied to an area and element, such as a bead board for

insulation, molding on a curtain track, and bullnose trim tile in
residential units.
2. Derived take-off is applied in cases where the quantities that
comply with the measurement rule can be obtained by addi-
tional calculation, where such quantities are not directly
extracted from the model. For example, if the estimation man-
ual mentions that the quantity of finish-type materials in a fur-
niture area is not calculated, the quantity of these materials can

Fig. 8. Insulation around entrance door: (a) vertical cutting view in CMO;
(b) vertical cutting view in IMO; (c) horizontal cutting view in CMO; and Fig. 10. Example of unnecessary area for QT. (Adapted from Kim
(d) horizontal cutting view in IMO. (Adapted from Kim 2015.) 2015.)

Fig. 9. Different interior wall heights in (a) CMO; and (b) IMO.

© ASCE 05019001-9 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Different lengths by thickness of previously installed material in IMO.

Fig. 12. Box plot of quantity discrepancy by causative factor.

Table 5. Frequency of quantity discrepancya by causative factor

Causative Factors 59AL 74A 84E 84EH Total (A) Over typical loss rate (B)
1. Constraint by model representation 30 38 50 28 146 123
2. Unnecessary area 10 16 12 12 50 50
3. Change in object length 62 82 89 58 291 157
a
Quantity discrepancies of −1 are not included in Fig. 12 or Table 4.

be measured by subtracting a furniture area from an area 3. If underestimation of a quantity is suspected, objects should
extracted from the interior component. The derived quantity be modeled separately, at least for the materials included in
reduction is then used to prevent over- and underestimation of the work types of plastering, water tightness, and insulation.
the quantity automatically generated from the model. Conversely, when the concern is overestimation, whether the

© ASCE 05019001-10 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

 extracted quantities include what should be removed or unneces- Acknowledgments
sary, parts defined in the estimation manual should be checked.
4. It is necessary to mark the BIM measurement type (automated, This work was supported by a grant from the National Research
derived, or manual take-off) with the measured quantities in the Foundation of Korea (2017R1A2B4007709) and the Ministry
BOQ in order to easily review the QT results and judge the of Land, Infrastructure and Transport (18AUDP-B127891-02)
extent of any over- and underestimation. funded by the Korean government.

Conclusion References

This study addressed how discrepancies in quantity are observed Abanda, F. H., B. Kamsu-Foguem, and J. H. M. Tah. 2017. “BIM—New
for interior materials with different details, using models defined rules of measurement ontology for construction cost estimation.” Eng.
as CMO and IMO. The estimated quantities from the CMO were Sci. Technol. Int. J. 20 (2): 443–459. https://doi.org/10.1016/j.jestch
on average 6–9% greater than those from the IMO. QDs were .2017.01.007.
Aibinu, A., and S. Venkatesh. 2014. “Status of BIM adoption and the BIM
Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

widely distributed and exceeded the typical loss rates for some in-
experience of cost consultants in Australia.” J. Prof. Issues Eng. Educ.
terior materials. For such materials, causative factors were
Pract. 140 (3): 04013021. https://doi.org/10.1061/(ASCE)EI.1943-5541
identified. The largest causes of QD were the model representa- .0000193.
tion and unnecessary modeling, while QDs by differences in Ashworth, A., and S. Perera. 2015. Cost studies of buildings. 4th ed. New
object length were found to have little influence on the quantity York: Pearson, Prentice Hall.
variation. AutoDesk Inc. 2017. “Wall foundation instance properties.” Revit 2017 help.
This study suggested modeling and QT strategies to minimize Accessed March 14, 2018. http://help.autodesk.com/view/RVT/2017
the discrepancy in the quantity of materials caused by inconsis- /ENU/?guid=GUID-615AB37C-044C-4104-85BC-7FA8E29A5AE0.
tencies in the model or components, according to the BIM meas- BIMForum. 2017. “Level of development specification.” Accessed January
urements categorized in RICS’s guidelines. It is necessary to par- 21, 2018. https://bimforum.org/wp-content/uploads/2017/11/LOD-Spec
tially modify the extracted quantity through modeling and the -2017-Part-I-2017-11-07.pdf.
Boktor, J., A. Hanna, and C. C. Menassa. 2014. “State of practice of build-
QT to prevent the risk of cost variation by representation of an
ing information modeling in the mechanical construction industry.” J.
unnecessary model area. If an inaccurate quantity extracted from
Manage. Eng. 30 (1): 78–85. https://doi.org/10.1061/(ASCE)ME.1943
the model is used as a criterion during the pretender or contract -5479.0000176.
process, serious problems may result, including construction Borhani, A., C. S. Dossick, H. W., Lee, and L. Osburn. 2017. “A technical
cost overruns, claims for underestimation, poor-quality construc- rep. on standardization of model-based estimating.” Accessed January
tion, and disputes between the general contractor and subcontrac- 21, 2018. http://cm.be.washington.edu/wp-content/uploads/2017/11/UW
tor. In this respect, this study is also valuable for identifying _MBEst_Report_v2_small.pdf.
work types and materials that might be problematic in estima- Choi, J., H. Kim, and I. Kim. 2015. “Open BIM-based quantity take-off sys-
tions using BIM. tem for schematic estimation of building frame in early design stage.” J.
Results imply that quantity discrepancies caused by a model Comput. Des. Eng. 2 (1): 16–25. https://doi.org/10.1016/j.jcde.2014.11
with different details are an unintended outcome for project partici- .002.
Denerolle S. 2013. Technical Rep. “The application of target value design
pants, including subcontractors. By identifying discrepancies by
to the design phase of 3 hospital projects.” Berkeley: Univ. of
model interior components, this study can assist in early prediction California. Accessed September 6, 2018. http://p2sl.berkeley.edu/wp
of risk of cost variation caused by discrepancy of quantity that can -content/uploads/2016/03/Denerolle-2011-Application-of-TVD-to-the
be detected after the contract or subcontract. Even for the low- -Design-Phase-of-3-Hospital-Projects.pdf.
detail, low-effort model quantity extraction, these results mean that Eastman, C., P. Teicholz, R. Sacks, and K. Liston. 2008. BIM handbook:
the accuracy of such quantities can be increased by eliminating the A guide to building information modeling for owners, managers,
unnecessary areas in QT. Stakeholders can discuss whether QD is architects, engineers, contractors, and fabricators. Hoboken, NJ:
acceptable at their stage of estimation and thereby inform any deci- John Wiley & Sons.
sions about adopting BIM, determining and discussing the quality Ghoilzadeh, P., B. Exmaeili, and P. Goodrum. 2018. “Diffusion of building
of information and the model required to adopt BIM with a focus on information modeling functions in the construction industry.” J.
cost estimation, rather than modeling in detail to get the exact quan- Manage. Eng. 34 (2): 04017060. https://doi.org/10.1061/(ASCE)ME
.1943-5479.0000589.
tities. This study contributes to the evaluation of the quality of BIM
Graphisoft. 2017. “Wall listing parameters.” Graphisoft Help Center.
as well as the body of BIM knowledge from the viewpoint of cost Accessed March 14, 2018. https://helpcenter.graphisoft.com.
estimation. Hyari, K. 2016. “Handling unbalanced bidding in construction projects:
However, this study has a limit in that analysis was performed Prevention rather than detection.” J. Constr. Eng. Manage. 142 (2):
on the extracted quantities from only four case-types of residential 04015060. https://doi.org/10.1061/(ASCE)CO.1943-7862.0001045.
units and did not deal with the risk and impact of QDs on projects. Jin, R., C. M. Hancock, L. Tang, and D. Wanatowski. 2018. “BIM invest-
Future research will aim to identify QDs in other types of buildings ment, returns, and risks in China’s AEC industries.” J. Const. Eng.
or interior designs through more cases and further data. Drawing on Manage. 143 (12): 04017089. https://doi.org/10.1061/(ASCE)CO.1943
the results of this study, future work can predict a range of QDs by -7862.0001408.
level of detail through the analysis of more cases with different Jones, S. A., and D. Laquidara-Carr. 2016. “SmartMarket brief: BIM
advancements no. 1.” New York: Dodge Data & Analytics.
types of buildings and estimate the completeness and reliability of
Jrade, A., and S. Alkass. 2007. “Computer-integrated system for estimating
QT results using BIM for various types of buildings. Validation of the costs of building projects.” J. Archit. Eng. 13 (4): 205–223. https://
the suitability of a generated model can be performed based on this doi.org/10.1061/(ASCE)1076-0431(2007)13:4(205).
study before the construction stage for estimation of the construc- Kim, S. A., S. Y. Chin, J. T. Han, and C. H. Choi. 2017. “Measurement of
tion budget. Moreover, in the future, criteria can be developed to construction BIM value based on a case study of a large-scale building
measure the completeness of a BIM model for cost estimation by project.” J. Manage. Eng. 33 (6): 05017005. https://doi.org/10.1061
evaluating the reliability of the QT. /(ASCE)ME.1943-5479.0000551.

© ASCE 05019001-11 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001

 Kim, S. A. 2015. “Study on modeling strategy of building interior works for Accessed August 11, 2016. http://www.blis-project.org/IAI-MVD/IDM
improving accuracy of quantity based on BIM.” Ph.D. thesis, /GSA-004/ER_GSA-004.pdf.
Sungkyunkwan Univ. Niknam, M., and S. Karshenas. 2015. “Integrating distributed sources of in-
Lawrence, M., R. Pottinger, S. Staub-French, and M. P. Nepal. 2014. formation for construction cost estimating using Semantic Web and
“Creating flexible mappings between building information models and Semantic Web Service technologies.” Autom. Constr. 57 (Sept): 222–
cost information.” Autom. Constr. 45 (Sept): 107–118. https://doi.org 238. https://doi.org/10.1016/j.autcon.2015.04.003.
/10.1016/j.autcon.2014.05.006. Olatunji, O. A., and W. Sher. 2014. “Perspectives on modelling BIM-
Lee, S. K., K. R. Kim, and J. H. Yu. 2014. “BIM and ontology-based enabled estimating practices.” Aus. J. Constr. Econ. Build. 14 (4): 4102.
approach for building cost estimation.” Autom. Constr. 41 (May): 96– https://doi.org/10.5130/ajceb.v14i4.4102.
105. https://doi.org/10.1016/j.autcon.2013.10.020. Peterson, F., T. Hartmann, R. Fruchter, and M. Fischer. 2011. “Teaching
Lee, J., Y. Park, C. Choi, and C. Han. 2017. “BIM-assisted labor productiv- construction project management with BIM support: Experience and
ity measurement method for structural formwork.” Autom. Constr. 84 lessons learned.” Autom. Constr. 20 (2): 115–125. https://doi.org/10
(Dec): 121–132. https://doi.org/10.1016/j.autcon.2017.08.009. .1016/j.autcon.2010.09.009.
Leite, F., A. Akcamete, B. Akinci, G. Atasoy, and S. Kiziltas. 2011. RICS (Royal Institution of Chartered Surveyors). 2015. BIM for cost manag-
“Analysis of modeling effort and impact of different levels of detail in ers: Requirements from the BIM model. 1st ed. London: Parliament Square.
Saleh, M. 1999. “Automation of quantity surveying in construction
Downloaded from ascelibrary.org by Iowa State University on 01/16/19. Copyright ASCE. For personal use only; all rights reserved.

building information models.” Autom. Constr. 20 (5): 601–609. https://

doi.org/10.1016/j.autcon.2010.11.027. projects.” J. Archit. Eng. 5 (4): 141–148. https://doi.org/10.1061
Liu, H., M. Lu, and M. Al-Hussein. 2016. “Ontology-based semantic /(ASCE)1076-0431(1999)5:4(141).
approach for construction-oriented quantity take-off from BIM models Seeley, I. H., and R. Winfield. 1999. Building quantities explained. London:
in the light-frame building industry.” Adv. Eng. Inf. 30 (2): 190–207. Macmillan.
https://doi.org/10.1016/j.aei.2016.03.001. Shen, Z., and R. R. A. Issa. 2010. “Quantitative evaluation of the BIM-
Lu, Q., J. Won, and J. C. P. Cheng. 2016. “A financial decision making assisted construction detailed cost estimates.” J. Inf. Technol. Constr. 15:
framework for construction projects based on 5D Building Information 234–257. https://www.itcon.org/papers/2010_18.content.07578.pdf.
Modeling (BIM).” Int. J. Project Manage. 34 (1): 3–21. https://doi.org Statistics Korea. 2016. “Korean statistical information service.” Housing
/10.1016/j.ijproman.2015.09.004. Supply Ratio and Statistics of Housing Construction. Accessed March 2,
McCuen, T. L. 2015. “BIM and cost estimating: A change in the process for 2017. http://kosis.kr/eng/.
determining project costs.” Chap. 3 in Building information model- Staub-French, S., M. Fischer, J. Kunz, and B. Paulson. 2003. “An on-
ing: applications and practices. Technical Council on Computing tology for relating features with activities to calculate costs.”
and Information Technology of the American Society of Civil Engineers, J. Comput. Civ. Eng. 17 (4): 243–254. https://doi.org/10.1061
edited by R. R. A. Issa and S. Olbina, 63–81. Reston, VA: ASCE. /(ASCE)0887-3801(2003)17:4(243).
https://doi.org/10.1061/9780784413982.ch03. Wei, H., S. Zheng, L. Zhao, and R. Huang. 2017. “BIM-based method calcu-
Monteiro, A., and J. P. Martins. 2013. “A survey on modeling guidelines for lation of auxiliary materials required in housing construction.” Autom.
quantity takeoff-oriented BIM-based design.” Autom. Constr. 35 (Nov): Constr. 78 (Jun): 62–82. https://doi.org/10.1016/j.autcon.2017.01.022.
238–253. https://doi.org/10.1016/j.autcon.2013.05.005. Wu, S., K. Ginige, G. D. Wood, and S. W. Jong. 2014. “How can building
Nassar, K. 2012. “Assessing building information modeling estimating information modeling support the New Rules of Measurement
techniques using data from the classroom.” J. Prof. Issues Eng. Educ. (NRM1)?” Royal Institution of Chartered Surveyors Research Paper.
Pract. 138 (3): 171–180. https://doi.org/10.1061/(ASCE)EI.1943-5541 Accessed June 10, 2014. http://www.rics.org/uk/knowledge/research.
.0000101. Zhiliang, M., W. Zhenhua, S. Wu, and L. Zhe. 2011. “Application and
NIBS (National Institute of Building Sciences) buildingSMART alliance. extension of the IFC standard in construction cost estimating for tender-
2015. Version 3: National BIM Standard—United States. In Chap. 4.5 ing in China.” Autom. Constr. 20 (2): 196–204. https://doi.org/10.1016/j
IDM for design to Quantity Takeoff for Cost Estimating, 22–23. .autcon.2010.09.017.

© ASCE 05019001-12 J. Manage. Eng.

J. Manage. Eng., 2019, 35(3): 05019001