Automation in Construction 119 (2020) 103355

Contents lists available at ScienceDirect

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

Adopting lean thinking in virtual reality-based personalized operation T

training using value stream mapping
⁎
Peng Wanga,b, Peng Wuc, Hung-Lin Chid, , Xiao Lie,f
a
College of Engineering and Technology, Southwest University, Chongqing 400715, China
b
Australasian Joint Research Centre for Building Information Modelling, Curtin University, Perth 6102, Australia
c
School of Design and the Built Environment, Faculty of Humanities, Curtin University, Perth 6102, Australia
d
Department of Building and Real Estate, The Hong Kong Polytechnic University, Hong Kong, China
e
Department of Real Estate and Construction, The University of Hong Kong, Hong Kong, Hong Kong, China
f
Department of Future Cities, Qianhai Institute of Innovative Research, Shenzhen, China

A R T I C LE I N FO A B S T R A C T

Keywords: Lean thinking has been proven effective in helping practitioners identify and eliminate wastes during en-
Lean gineering operations. However, systematic instructional mechanisms and training protocols based on individual
Value stream mapping trainee's performance are insufficient in existing training to define value-added activities for further productivity
Virtual reality improvement in a training environment. This study aims to investigate how value stream mapping (VSM), as a
Personalized training
lean tool, can be applied to help improve operation training performances through an immersive virtual reality
Productivity
(VR)-based personalized training program. A before–after experiment based on a virtual scaffolding erection
scenario is established to simulate the training process. The training performance resulting from the VSM-based
VR approach is compared with conventional VR training. Comparative results indicate that the waste time and
errors reduce significantly. Compared with the conventional method, the overall productivity improvement of
the erection process using VSM-based VR training is 12%. This demonstrates that integrating lean thinking into
the operation training process can be a more effective approach for VR-based personalized operation training,
provided that appropriate instructions are implemented.

1. Introduction training, if integrated with the lean method, can help employees learn
how to eliminate waste effectively and achieve efficiency in construc-
Lean principles have been successfully adopted in the architecture, tion operations.
engineering, & construction (AEC) fields to increase profitability and Lean training aims to educate employees regarding operational
productivity [1,2]. Lean techniques and tools, such as value stream processes more effectively, which is key to lean manufacturing. For
mapping (VSM), 5S, and just-in-time (JIT) offer new methods for example, Deros et al. [8] reported that the understanding level of
identifying customer values and eliminate non-value-added activities trainees improved significantly when a lean training course is provided.
[3]. Heravi and Firoozi [4] used VSM, which is a lean technique to Therefore, it is believed that the lean approach can be adapted ac-
systematically describe and investigate the production processes and cordingly for employees to implement process improvements more ef-
further help identify wastes that can be removed from the process, in fectively. In addition, VSM, a lean tool, has been typically used in the
prefabricated construction. And they discovered that VSM is effective education sector. Ahmad et al. [9] demonstrated that the integration of
for time reduction and cost saving. In addition, workplace productiv- VSM into a project-based engineering curriculum can not only help
ities in construction industries can be improved using 5S, especially in students learn lean theorems, but also enable them to use VSM for
working areas [5,6]. Ezema et al. [7] reported that JIT provided better problem solving.
work motivation and operation in manufacturing plants. Anderson and Although lean training can be beneficial, most AEC training courses
Kovach [1] demonstrated that lean methods could help reveal the un- are conventionally conducted in classrooms, using examples and video
derlying links of activities in each phase of maintenance projects to clips from previous construction projects. Lean training through videos
identify value-adding activities and waste. Construction industry can assist trainees in visualizing construction tasks and activities;

⁎
Corresponding author.
E-mail addresses: [email protected] (P. Wang), [email protected] (P. Wu), [email protected] (H.-L. Chi), [email protected] (X. Li).

https://doi.org/10.1016/j.autcon.2020.103355
Received 5 March 2020; Received in revised form 8 July 2020; Accepted 11 July 2020
0926-5805/ © 2020 Elsevier B.V. All rights reserved.
P. Wang, et al. Automation in Construction 119 (2020) 103355

however, trainees cannot interact with the video environment. In recent a few of these lean tools have been used in the education sector. The
years, researchers have adopted VR-related technology via building current research gap can be discovered accordingly. Although the im-
information modelling (BIM) [10,11], game technologies [12,13], and plementations have been proven useful in eliminating waste and im-
smart devices [14] to improve construction training performance. The proving productivity, studies focusing on how the related lean tools will
advantages of adopting VR technologies in training compared with contribute to construction training and education are limited [38,39].
other means include enriched intractability, intuitive replicate of the In addition, applying lean concepts in construction operations is still
reality, cost-saving, and safety guarantees [10,15]. Li et al. [13] in- new, demonstrating strong research needs in this area [21].
dicated that the VR-based training can help the trainees simulate safety
hazards under the virtual work environment. This study demonstrates 2.2. VSM for engineering education and training
the weaknesses of the trainees who even have already passed the tra-
ditional field training processes and a VR-based training can further Among all available lean tools for engineering education and
improve the understanding of safety hazards. Although these studies training, VSM has proven effective for process improvement. It can
have proven that VR is effective for students or trainees, it is note- provide a better understanding of value-adding and non-value-adding
worthy that these VR training programs adopted traditional one-size- activities from materials and information flows and deliver a product
fits-all training methods that rarely consider the diversity of learning that satisfies customer requirements. Engineering and technology cur-
needs among individual trainees. According to Jeelani et al. [16], better riculum with VSM can be taught in classes to achieve learning objec-
training performances can be personalized owing to the knowledge tives [42]. Lobaugh [43] used VSM to analyze the information flow of
gaps and learning needs of individuals. Jeelani et al. [17] stated that a manufacturing processes in engineering fields for waste elimination.
more effective personalized training experience for construction Steinlicht et al. [44] used VSM to map the educational process of a
workers can be provided through a virtual training environment. manufacturing engineering curriculum. The results showed that the
Given the current gaps identified above, this study aims to in- course can be improved to better satisfy learning outcomes regarding
vestigate how VSM, as a lean tool, can be applied to help improve the understanding of manufacturing processes and related technical
operation training performances through an immersive VR-based per- information and skills. However, there are limited studies investigating
sonalized training program. The specific objectives of this study are as the direct implementation of VSM in construction-related operation
follows: 1) to develop an immersive VR-based personalized training training for productivity improvement. It is expected to address the
system to enhance training productivity for onsite workers; 2) to design productivity issues of the related training practices through identifying
and implement a systematic VSM-embedded training protocol to en- operation wastes and eliminating these wastes.
hance training performance by adopting VSM; 3) to evaluate the overall
performance of the training system. 2.3. Virtual construction personalized training
This paper is organized as follows. Section 2 primarily discusses
previous studies conducted in the areas of lean for education, VSM, and Virtual training has been widely adopted in construction operations
virtual and personalized training. Section 3 describes the proposed because it can improve the training outcomes and create a good op-
training protocol in this study. Section 4 presents the experimental portunity for trainees to practice before they perform actual construc-
process for evaluating the performance of the training system. Section 5 tion work. As an example, Li et al. [45] developed a multiuser virtual
presents the results and discussions, followed by the conclusions and safety training system for a virtual tower crane operation process. The
details on future studies being presented in Section 6. results showed that training performance improved significantly com-
pared with the traditional training approach. In addition, the use of VR
2. Literature review technologies in engineering design based on personalized learning has
been demonstrated by Adas et al. [46]. This virtual learning environ-
2.1. The lean concept ment can provide students with step-by-step instructions for interac-
tions to achieve training objectives. Jeelani et al. [17] developed an
The lean philosophy originated from the Toyota Production System immersive VR-based safety training environment that provided custo-
[18], which maximizes value and reduces waste. Lean implementation mized training elements for construction workers. It was demonstrated
in the manufacturing industry typically focuses on productivity im- to be effective compared with the traditional “one-size-fits-all” method
provement by reducing wastes and delivering the maximum value to for knowledge and skill improvement. The summarized research effort
customers. Wastes in the manufacturing industry are generally cate- can be seen in Table 3. The current research gap for VR in training is
gorized into eight categories: overproduction, waiting, transportation, also identified. The previous VR-personalized training scenarios were
overprocessing, motion, inventory, defects, and unused talent [19]. preset and lacked unified measurements to recognize individual
Lean education is the adaption of lean thinking to identify and solve training performances. The instructions delivered by the trainer were
educational problems and improve learning and teaching activities. based on their own knowledge and experience. A systematic judgment
Antony et al. [20] translated eight wastes of lean manufacturing to the approach for providing sufficient guidance based on trainee perfor-
education sector, as shown in Table 1. Lean implementation in educa- mance is lacking. To improve training performance in operation pro-
tion can reduce cost and educational cycle time, as well as increase the ductivity, VSM may be integrated into the processes of personalized
satisfaction level for students, overall learning process [21], student training. Nevertheless, research regarding the integration of VSM in
academic experiences, and productivity [22]. In addition, engineers such a training scenario is non-existent, to the best knowledge of the
with lean knowledge are crucial to the development of modern lean authors.
enterprises, where employers are increasingly expecting the necessary
engineering knowledge and competency levels. Lean education adopted 3. Training protocol design
in engineering processes may provide leading-edge approaches to
content and competency mastery for workplace preparation [23]. 3.1. Overview of the proposed training protocol
As shown in Table 1, the lean concept has been successfully used by
construction companies to reduce project costs and waste on con- The research approach of this study is to propose a new VR perso-
struction sites [1,35]. Over 40 lean techniques and tools have been nalized training protocol for integrating lean concepts in training gui-
adopted in lean construction [36]. According to the study by Bajjou dance, and validating its usefulness through a user experiment by
et al. [37], the most typically adopted lean tools for the construction comparing it with the conventional VR training. As shown in Fig. 1,
industry include 5S, (JIT), poka-yoke, and VSM (Table 2). Furthermore, trainees usually go through a task briefing and lecturing session to

2
P. Wang, et al. Automation in Construction 119 (2020) 103355

Table 1
Waste translation from manufacturing to the education sector.
Waste categories Definition in the manufacturing sector Explanation in the education sector Explanation in the construction sector

Overproduction Waste from making more products than Course content or additional knowledge exceeds the Construction task is completed faster than scheduled or
customers demand requirement for the current learning process [24] before it is required in the process [25]
Waiting Time spent on idling for the next process Knowledge acquired by students must be retained until Typically occurs when a worker is ready, but the
step to occur the following subject in the learning process [26] materials required for work have not been delivered, or
the previous task has not been completed [27]
Transportation Transportation waste is the unnecessary Movement of knowledge from one subject to another, Materials, equipment, or workers are moved from one
movement of products and materials that which must be retaught, and the movement of materials job site to another before they are required [28]
do not create value related to the curriculum [24]
Overprocessing Waste owing to more work than that Excessive inappropriate teaching and learning processes Overprocessed construction activities that have no
required by the customer for students [29] value to the customer [30]
Motion Waste time and effort related to Movement from one subject to another that is lacking Unnecessary movements by workers to accomplish
unnecessary movements by people, the coherent streaming of curriculum, or the their work, which do not add value to the customer
machine, or equipment misunderstanding of the previous subject [26] [31]
Inventory Waste resulting from excess unprocessed Knowledge must be retained for the future subject, Materials stored on the construction site that are not
products and materials which tends to be forgotten or becomes obsolete [24] required immediately [32]
Defects Waste from a product or service fails to Shallow learning and failing to understand the related Defects in construction are incorrect work requiring
satisfy customer expectations subject matter [29] rework or repair [33]
Unused talent Waste owing to failing to utilize human Failing to recognize the ideas and suggestions of Workers who have extensive experiences or skills are
talents, skills, and knowledge teachers and students for improvement [26] not matched to the right jobs on construction sites [34]

know the details of the operational tasks. Then they will be immersing erection in the virtual environment individually. Next, the first round of
themselves in the VR environment and performing exercises to imple- exercises for trainees to complete the virtual scaffolding tasks is per-
ment what they learned before. The performance will be recorded and formed. Their performances during the operational processes are re-
further used by the trainers to give trainees feedbacks to address their corded, including the value-added time, number of errors, and lead
specific weaknesses during the exercises. Afterward, trainees start to time. The trainee performance baseline is hence identified (see Section
conduct exercise by referring back to the guidance and attempt to im- 3.3 for more details). Subsequently, all trainees are randomly categor-
prove performance. This is a basic process for conventional VR training. ized into two groups. The first group focuses on conventional perso-
The proposed training approach further introduce VSM as a tool to nalized guidance that provides instructions based on observations and
assist the achievement of three objectives: 1) performance analysis and trainee performances through the exercises. The second group is coa-
waste identification on trainees' exercise results; 2) performance pro- ched to provide instructions through VSM-based personalized guidance.
filing based on performance analysis and waste identification for trai- The detailed procedure of the guiding process is detailed in Section 3.4.
ners to provide personalized coaching; and 3) productivity estimation Subsequently, all trainees must reproduce the scaffolding erection op-
for identifying trainees' potential improvement. eration under the same scenario. As discussed in Section 3.5, the per-
In this study, the training task was set to be a scaffolding erection formance of the second-round exercise for each trainee is assessed using
mission, which is typically performed prior to inspections under turn- the same indicators. Finally, the training productivity is estimated for
around maintenance (TAM) in plant scenarios. The liquefied natural gas further performance comparison.
industry uses TAM to increase the reliability of plant facilities. In TAM,
plants must be shut down periodically for inspections and repairs to
maximize production capacity and ensure the reliable and safe opera- 3.2. Lecture and practice session
tions of all equipment. A few weeks of TAM may incur a year's main-
tenance expense in terms of the direct cost of TAM and lost production During the lecture, trainers introduce the general guide of scaf-
[50]. Hence, temporary scaffolding works must be performed to address folding work to all trainees, including the scaffolding erection proce-
the special needs of repairing production equipment as well as schedule dure shown in Fig. 3. Moreover, the policy for safe scaffolding erection
and process controls efficiently. Additionally, the related training pro- and use has been established [49,50]. The lecture kits include instruc-
cess must be performed effectively. tional handouts and video demonstrations. A two-story scaffolding
VSM-based personalized VR training is proposed herein and com- erection task for plant tank inspection was selected as an example from
pared with conventional personalized VR training, based on a virtual the lecture materials in this study. In such a scaffolding erection, step
scaffolding erection scenario in a before-after experiment. The archi- one involves the appropriate preparation for scaffolding operators to
tecture of the comparative training protocols is shown in Fig. 2. The define the work area and verify the availability of scaffolding compo-
proposed training framework comprises three modules. First, a general nents. Steps two to five comprise standard procedures for scaffolding
scaffolding erection procedure is delivered to all trainees by lecturers. foundation erection. Steps six to thirteen describe the process of the
Subsequently, the trainees must familiarize themselves with the VR- ground-floor scaffolding erection. The following steps until step 20
based equipment under training scenarios and exercise scaffolding pertain to the first-floor scaffolding erection. As most of the steps were
repetitive, the process was simplified into seven steps from the

Table 2
Definitions of related lean tools and applications in education.
Lean tools Descriptions

5S 5S represents sort, simplify, sweep, standardize, and self-discipline, which refers to the effective management of production factors, such as
personnel, machinery, and materials in construction sites [40]
Just-in-time (JIT) JIT is a methodology for reducing waste in production, in which products are manufactured based on need, time of need, and the amount needed
[36]
Poka-yoke Poka-yoke, a Japanese term, is a mechanism to help operators avoid errors [41]
Value stream mapping (VSM) VSM is a lean management method for visually analyzing and improving workflow circulation [37]

3
P. Wang, et al. Automation in Construction 119 (2020) 103355

Table 3
Summary of previous studies in VR operation training.
Research Description Finding

Li et al. [45] A multiuser virtual safety training system was developed for a virtual tower The results showed that the VR training performance improved
crane operation process significantly compared with the traditional training approach
Adas et al. [46] The use of VR technologies in engineering design based on personalized This virtual learning environment can provide students with step-by-step
learning was demonstrated instructions for interactions to achieve training objectives
Cheng and Teizer [47] A framework that involved real-time data collection and visualization for The operation information can be tracked and visualized for construction
construction worker was developed workers to improve their situational awareness
Le et al. [48] A framework was proposed for construction site training based on mobile Using mobile VR training would improve construction site safety
VR
Jeelani et al. [17] An immersive VR-based safety training environment that provided It was demonstrated to be effective compared with the traditional “one-
customized training elements for construction workers was developed size-fits-all” method for knowledge and skill improvement

scaffolding erection process and adopted for the training experiments. [55]. According to Hattie and Timperley [56], four levels of feedback
The detailed steps of the process are shown in Fig. 4. It should be noted exist: the task, processing, regulatory, and self-levels. Feedback at the
that steps two to six are related to the essential production process, processing level is beneficial to help trainees reject erroneous hy-
while other steps are non-essential (e.g., safety precaution or hazard potheses and improve an individual's training performance.
avoidance). Therefore, only steps two to six were considered in calcu- Hence, the video-assisted feedback method [57] was used as the
lating the trainees' productivity performance in the virtual training conventional guiding approach for each trainee in the first group
environment. After the lecture session was performed, the trainees (Group 1). In addition, the trainers provided instructions based on their
could familiarize themselves with the related VR equipment and scaf- observation of the trainees' performance, including waste categoriza-
folding erection scenario through a practice session. tion during the operation.
Unlike gathering the observational data to provide guidance for
3.3. Training exercise for baseline Group 1, VSM was used as a personalized guidance tool in Group 2.
According to Rother and Shook [51], VSM can improve the process flow
To evaluate the potential benefits through the adoption of perso- through four steps. The first step is to select the product family, which is
nalized training with or without VSM, a baseline of the trainees' per- the virtual scaffolding components to be erected in this study. The
formance must be established. As the first round of the official virtual second step is to construct a current state map (CSM) for waste iden-
operation exercise, each trainee participates in the virtual training tification in the value stream of the erection process and describe waste
scenario to perform the scaffolding operation based on what they have in detail. It is the map that the trainers can use to guide the trainees
learned from the lecture. The Value-Added Time (VAT), number of systematically to identify potential productivity issues from the trai-
errors, Cycle Time (CT), and waste categories in each step were re- nees' performance. For instance, each step of the scaffolding erection
corded; the Lead Time (LT) and Processing Time (PT) of the scaffolding process was drawn on the map as a chain connected by blocks. The
processes were calculated for further productivity evaluation. The PT is trainee's performance (e.g., VAT, LT, error, WT) at each step was listed
the duration between the start and finish time of the entire scaffolding on each block for trainers to identify if there are any significant wastes
erection process. The VAT is the processing time when the value-adding on a specific step of the erection process. The third step is to construct a
activities are performing during the scaffolding erection. VAT is the future state map (FSM) based on waste elimination suggested by the
part of processing time, excluding Waste Time (WT) and non-value- trainers to set up an ideal goal for the individual trainee to follow. The
adding time in the experiments. And it can be measured through ex- final step is to achieve the future state, which is, in the study, to guide
cluding wastes and no-value-but-necessary behaviors during the op- the trainees based on CSM/FSM evaluation results and allow them to
eration, including picking up the wrong scaffolding components, car- perform a post-exercise to assess whether the identified wastes can be
rying scaffolding components to be in position, assembling adjustment, prevented. The trainers provide instructions and suggestions based on a
unnecessary traveling, idling, and performing rework. The details are CSM that allow the trainees to visualize the sources of waste at each
given in Table 4. scaffolding step. Furthermore, the FSM shows the proposed changes in
the scaffolding operation for each trainee in Group 2 for further im-
provement.
3.4. Guiding session through conventional method and VSM

After baseline identification, all the trainees were randomly cate- 3.5. Post-exercise and improvement evaluation
gorized into two groups: Group 1 and 2. The feedback of the scaffolding
erection process was provided to each group by the trainers. The out- After Groups 1 and 2 have been trained through the conventional
comes of the training baselines enable the trainers to assess the per- personalized and VSM-based personalized guidance, respectively, all
formances of the trainees based on their training tasks or processes the trainees reproduced the VR-based scaffolding erection process.

VSM
Performance analysis
and
waste identification Productivity estimation

Performance profiling
results

VR Exercise Personalized Repeated VR Trainee

Start Task briefing End
training exercise performance coaching training exercise improvement

Fig. 1. The proposed VSM-based VR personalized training process.

4
 P. Wang, et al.

5
Fig. 2. Architecture of training protocols under a before–after training scenario.
Automation in Construction 119 (2020) 103355
P. Wang, et al. Automation in Construction 119 (2020) 103355

Fig. 3. Scaffolding erection process.

Fig. 4. Steps selected for VR training experiments.

These two groups were compared to demonstrate the benefits and dif- learn how to perform the scaffolding erection. To evaluate the training
ferences between lean-based VR training and traditional training. LT, efficiency, the VAT (min), WT (min), errors, and PT (min) to accom-
PT, VAT, and the number of errors during scaffolding erection were plish the training tasks were used to evaluate the trainees in the two
assessed for process improvement. Furthermore, training productivity groups.
was measured. In this study, the productivity index was considered in
all the activities performed from steps two to six in Fig. 4, in which the
trainees had to operate scaffolding components and place them at the 4.1. VR-based scaffolding erection scenario and participants
correct positions.
The design of the virtual scaffolding erection scenario is shown in
Fig. 5. The virtual scenario was modeled using Unity3D, which is a
4. Before–after training experiment game engine to create a virtual interactive environment. The virtual
models, including the scaffolding components, foundations, and tanks
To evaluate the performance of the designed training protocol, the to be inspected, were created using Autodesk Revit 2018, a BIM soft-
before–after training experiment was conducted. The participants of ware; they were exported in the FBX format and imported to this virtual
Group 1 used traditional personalized training (video and lecture), environment. The components of the scaffolding included 22 base
whereas those in Group 2 used VSM-based personalized training to plates, 22 standards, 62 transoms and ledgers, and 10 diagonal

Table 4
Indicators of baseline identification.
Indicators Description

Lead time (LT) LT is the time consumed from the beginning to the end of the scaffolding (specifically, steps 2 to 6 in Fig. 4) [51]
Processing time (PT) PT is the time consumed for scaffolding erection (specifically, steps 1 to 7 in Fig. 4) [52]
Value-added time (VAT) VAT is the processing time associated with value-adding activities (e.g., actual installation work) during the scaffolding erection [53]
Cycle time (CT) Frequency of scaffolding erection completion by every step [54]
Number of errors During the operation of steps 2–6 (in Fig. 4), the incorrect construction of each scaffolding component is recorded as one error (e.g., components
misplaced with different dimensional requirements)
Waste time (WT) WT is the time consumed by the trainee to perform the non-value-adding activity (as mentioned in the waste category)
Waste category A: Taking wrong scaffolding components
B: Unnecessary traveling
C: Thinking (idling)
D: Rework

6
P. Wang, et al. Automation in Construction 119 (2020) 103355

Fig. 5. Overview of the virtual scaffolding erection scenario.

bracings, as shown in Fig. 5.

The participants in the experiment were 32 male undergraduate Monitor
students who have no experience related to VR operation and scaf- Desk
folding construction. They were from the School of Engineering and
Technology at Southwest University, China. The average age of them Simulated scaffolding
was 21.3 years old, with a range of 20 to 22. The 32 students were erection area
randomly assigned to Groups 1 and 2. Each group comprised of 16
students. As shown in Fig. 6, all the trainees had to perform the scaf-
observation
folding erection in the virtual environment using the VR equipment.
The hardware includes a head-mounted display device, computer
monitor, and game controller. Detailed information regarding the en-
vironmental setting of the experiment is shown in Fig. 7. Detailed
configurations of the hardware are provided in Table 5. In the experi-
ment, the student participants wore the VR headset and held the con- observation 6m
troller to perform simulated erection activities in the virtual scenario. A
facilitator, a researcher who is also the trainer, monitored the behaviors
Trainer Trainee
(act as Desk (with VR
of the participants by viewing the monitor that displayed the projected
information of the virtual scenario. The facilitator was responsible for facilitator headset and
recording the performances of the participants, identifying errors, and and controller)
conceiving effective instructions for personalized guidance. observer)

4.2. Evaluation of the training performance

3m
The training performance of the participants was evaluated from
four aspects: time, error, productivity, and satisfaction. For time re- Fig. 7. Environmental settings of the simulated scaffolding erection scenario.
cording, it was recorded in seconds and then converted to minutes,
rounding to one or three decimal places. The detailed evaluations in- validate whether a significant difference occurred between the perfor-
clude comparing the significance between the different performance of mances of the two groups. According to Johnson [58], if p < 0.05, then
the two groups in terms of time and error, assessing corresponding their performances are significantly different; if p ≥ 0.05, they are not.
productivities, and evaluating trainees' confidence in undertaking the In addition, to adequately compare the training performances of the
individual training through a questionnaire survey. two groups, baselines were identified to ensure that the participants
A t-test for identifying statistical significance was adopted to (trainees) in the two groups have similar prior knowledge and abilities

Fig. 6. Example of simulated scaffolding erection operation in the virtual environment.

7
P. Wang, et al. Automation in Construction 119 (2020) 103355

Table 5
Detailed configurations of the VR hardware equipment.
Hardware Purpose Description

Headset Provide immersive virtual scaffolding erection scenario to the participants The HTC VIVE headset used has a refresh rate of 90 Hz, 110° field of view,
and display resolution of 1080 × 1200
Wireless controller Grasp and release scaffolding components in the installation positions in the The HTC VIVE wireless controller includes a trackpad, grip buttons, and
virtual scaffolding erection scenario dual-stage trigger
Monitor (with PC) Project participants' views and actions in the virtual scaffolding erection Dell S2340L 23-Inch screen LED-Lit monitor was used and synchronized
scenario to the trainer and perform videotaping with the headset

in performing the simulated operations without personalized instruc- qualitative results obtained from the trainees.
tions. To perform the adjustments, the performance of the first round
exercise (baseline) under each indicator before training in Group 1 was 5. Experimental results
used, and the others were adjusted to the corresponding baselines. The
adjustments were computed through a standardization method. It was The experimental results, including the before–after exercise per-
assumed that the performance before the training exhibited an identical formance, CSM, FSM, and productivity indices, were gathered. The
statistical distribution, which was characterized by the mean and average duration of the lecture session was about 40 mins, and that of
standard deviation in the standardization method. In the adjustment the practice time for each trainee to familiarize himself with the VR
approach, X1, 0 and X1, u were the performances before and after environment was about 30 mins. About 20 mins were spent on perso-
training for Group 1, respectively, and X2, 0 and X2, v were those for nalized guiding sessions in the case of the VSM-based VR training
Group 2, respectively. The performances before training in Group 2 was process.
adjusted as follows:
X ′2,0 = g (X2,0 ) sd (X1,0 ) + mean (X1,0 ), (1) 5.1. Performance in terms of time and error

where X′2, 0 is the adjusted performance for X2, 0, g() is the standardi- As for the first group (Group 1) related to the traditional training
zation function, sd() is the standard deviation, and mean() is the mean guidance, the before–after training performance is as shown in Table 6.
value. The standardization function is calculated as follows: The average VAT before and after the training was 17.4 and 16.1 mins,
X2,0 − mean (X2,0 ) respectively; the WT was 11.9 mins and 9.3 mins, respectively; the error
g (X2,0 ) = was 33.8 and 29.8, respectively; and the PT was 29.2 and 25.4 mins on
sd (X2,0 ) (2)
average, respectively.
Consequently, the adjusted performance X′2, 0 has the identical The second group (Group 2) underwent a VSM-based personalized
statistical distribution parameters, mean, and standard deviation va- training, and the before–after training performance of Group 2 is shown
lues, which ensures the same baseline and comparativeness of the two in Table 7. Similar to the trend of Group 1, the average VAT before and
groups through observations. Correspondingly, the performance after after the training was 16.8 and 15.2 mins, respectively; the WT was
training in Group 2 was adjusted as follows: 12.4 and 8.0 mins, respectively; the error was 37.2 and 24.0, respec-
X ′2, v = h (X2, v ) sd (X1,0 ) + mean (X1,0 ), (3) tively; and the PT was 29.3 and 23.2 mins on average, respectively.
To identify the waste during the scaffolding erection process for
where X′2, v is the adjusted performance for X2, v, and h() is a modified each trainee in Group 2, a CSM was first constructed based on the
standardization function for X2, v according to the baseline of X2, 0, trainee's performance baseline and videotaping to determine the ap-
which is calculated as follows: propriate strategy for improvement. Fig. 8 shows an example of a CSM
X2, v − mean (X2,0 ) based on the performance of a trainee. Once the CSM was shown to the
h (X2, v ) = trainee, the trainee and trainer discussed the metrics that require im-
sd (X2,0 ) (4)
provement. The waste types of each activity that contributed the most
By adopting the adjustment Eqs. (1)–(4), performance in terms of were listed, such as the waste categories, WT, number of errors, and PT.
time and error in Group 2 are normalized to those of Group 1 with The trainer first determined the maximum errors and WT in each ac-
identical statistical distributions for further comparisons. tivity. For example, as shown in the figure, most of the errors occurred
To further evaluate the overall training productivity between the in Activity 2 (10 times), but only five errors in Activity 3 caused more
two training approaches, the productivity index [59] was calculated as time waste. Finally, the trainer discussed with the trainee to determine
follows: possible improvement approaches to transform a CSM to an FSM, i.e., to
P = (Qa − Qb)/(Tf − Ts ) improve the scaffolding erection performance. Subsequently, the ideal
(5)
FSM was created by the trainer and trainee. As shown in Fig. 9, all
where Qa denotes the number of processing activities, Qb is the number sources of waste were expected to be eliminated adequately for the
of non-value-adding activities, Tf and Ts are the finish and start time of trainee's reference.
operation. In terms of baseline adjustment, the result of the t-test before the
To evaluate the trainees' confidence qualitatively, questionnaire training is as shown in Table 8. The p-values of the VAT, WT, error
surveys were adopted. Five questions were posed as follows: (1) the task count, and PT were 0.562, 0.660, 0.309, and 0.901, respectively. The p-
and instruction were easy to understand and helped in learning-related values for all indicators were much higher than 0.05, implying that the
information; (2) I can easily and quickly identify the waste when re- two groups of participants (Group 1 and normalized Group 2) did not
producing the training tasks; (3) I can effectively complete the desig- differ significantly before the training.
nated training tasks; (4) the training approach was helpful and effec- As for the performance improvement after the personalized training
tive; (5) overall, I was satisfied with the training process. The approaches, the t-test is further conducted, as shown in Table 9. The
participants were requested to assign a rating from 0 to 10 (0: com- VSM-based personalized training is significantly better than the con-
pletely disagree; 10: complete agree) to each question. A paired-sample ventional personalized training in terms of WT elimination
test was used to compare the effectiveness of the lean-based VR per- (t = − 4.066; p = 4.72E − 04 < 0.05), error reduction (t = − 5.957;
sonalized training with the traditional training, according to the p = 3.68E − 06 < 0.05), and PT improvement (t = − 3.945;

8
P. Wang, et al. Automation in Construction 119 (2020) 103355

Table 6
Before–after training performance of Group 1.
Group 1

Trainee No. VAT before VAT after WT before WT after Error count before Error count after PT before PT after training
training (min) training (min) training (min) training (min) training (times) training (times) training (min) (min)

1 19.7 18.5 16.4 10.1 35 28 36.1 28.6

2 15 16.1 9.4 6.2 30 27 24.4 22.3
3 15.5 17.1 14 10.3 35 31 29.5 27.4
4 14.8 14.7 7.7 8.1 27 28 22.5 23.3
5 21.5 19.6 14.5 12.1 47 39 36 31.7
6 13.8 13.9 7.7 8.4 22 26 21.5 22.3
7 20.2 18.5 12.2 9.7 47 39 32.4 28.2
8 19.9 18.1 15.2 10.3 44 31 35.1 28.4
9 11.4 11.6 9.4 9.7 31 31 20.8 21.5
10 17.5 16.3 11.0 9.0 33 30 29.0 24.8
11 18.1 13.8 15.2 5.7 31 20 33.3 19.5
12 15.7 14.6 8.3 8.8 22 23 24 23.4
13 17 15.4 9.5 7 22 20 26.5 22.4
14 20.6 17.6 16.9 14.5 52 44 37.5 31.1
15 17.9 14.3 8.8 7.2 22 20 24.7 21.5
16 20 18.2 13.9 12.5 40 40 33.9 29.7
Average 17.4 16.1 11.9 9.3 33.8 29.8 29.2 25.4

p = 5.08E − 04 < 0.05). However, the VAT (t = − 1.899, be minimized through the normalizations of the two groups' results. In
p = 6.79E − 02 > 0.05) between the two groups was not significantly general, the mean VAT of the baseline exercise was 17.41 mins, and the
different after the training. All the comparative results related to t-tests standard deviation was 2.95 mins. After the conventional VR training
have been validated through Cohen's d benchmark [60]. The value of exercise (Group 1), the mean VAT was 16.13 mins, a 7% improvement
Cohen's d for 95% confidence interval was tested on a scale of medium of the VAT compared with the baseline. On the other hand, due to the
to large size effect, which is 0.71 for VAT, 1.54 for waste time, 2.25 for accurate training guidance and more confidence in operations with the
errors, and 1.49 for processing time. CSM and FSM, a 14% improvement of the VAT, with the mean VAT of
In terms of the comparative summary between the conventional and 14.96 mins, was observed.
VSM-based personalized training, the average performances of each The training effects quantified by the WT are shown in Fig. 10(b);
indicator are presented. Fig. 10 shows the training effectiveness be- the mean WT of the baseline was 11.94 mins, and the standard devia-
tween conventional and VSM-based personalized training after baseline tion was 3.36 mins. After the conventional VR training, the mean WT
adjustment. The effectiveness is presented from three aspects, including was 9.37 mins, which was reduced by approximately 21%, when
the statistical summaries shown in boxplots, mean value variations of compared with the baseline in Group 1. In addition, by using the VSM-
the two groups, and the mean values of confidence after the training. based personalized training, the mean WT was now 5.95 mins, which is
The training effects quantified by the VAT, as shown in Fig. 10(a), a 50% reduction. In other words, using VSM-based personalized
indicate that the VAT can be reduced critically by the personalized training can reduce unnecessary travel, rework, and errors more ef-
training approach. It was assumed that the effect of familiarity was fectively when compared with conventional personalized training.
reduced or eliminated through the practice session. Furthermore, there The training effects quantified by the error count are shown in
are two different groups of participants who perform the exercises by Fig. 10(c), showing a significant difference between the two groups.
using different training approaches individually, instead of using the The mean error count of the baseline is 33.8 times, and the standard
approaches sequentially. So, the effect of the familiarity issue can also deviation is 10.2 times. After the conventional VR training, the mean

Table 7
Before–after training performance of Group 2.
Group 2

Trainee No. VAT before VAT after WT before WT after Error count before Error count after PT before PT after training
training (min) training (min) training (min) training (min) training (times) training (times) training (min) (min)

1 18 15.3 12.2 8.8 36 21 30.2 24.1

2 13.9 13.8 7.6 4.4 27 12 21.5 18.2
3 18.5 16.6 14.2 8.7 44 28 32.7 25.3
4 21 18.2 14 8.2 48 28 35.1 26.4
5 15.7 14.6 11.5 7.1 40 23 27.2 21.7
6 18.5 15.8 12.8 8.6 34 23 32.3 24.4
7 16.6 15.3 13.8 8.7 34 26 30.4 24
8 18 15.8 8.8 6.1 27 20 26.8 21.9
9 15.6 15.2 12.0 8.1 36 24 28.7 22.9
10 13.6 12.3 8.2 6.2 23 21 21.8 18.7
11 17.4 15.4 12.2 7.2 37 19 29.6 23.6
12 18.3 14.3 15.1 9 48 24 33.4 22.3
13 15.8 15.7 16 10.2 44 35 32.1 25.9
14 17.1 15.5 12.8 7.5 35 22 29.9 23
15 15.9 16.1 13.8 9.6 42 30 29.7 26.2
16 14.7 13.7 13.3 9.9 39 28 28 23.6
Average 16.8 15.2 12.4 8.0 37.2 24.0 29.3 23.2

9
P. Wang, et al. Automation in Construction 119 (2020) 103355

Fig. 8. Example of CSM based on a trainee's performance in Group 2.

errors in Group 1 were 29.8 times, which was a 12% decrease. The Table 8
mean errors of Group 2 were 15.9 times (approximately a 50% de- Performance comparison before the personalized training using t-test.
crease), which was a more significant reduction compared with that of Indicator Mean of group 1 Mean of group 2 t value p value
Group 1 after the VSM-based personalized training was introduced. As
the error details were provided to the individuals for each step of the VAT (min) 17.407 16.867 0.588 0.562
scaffolding erection task through VSM, a significant reduction was ex- WT (min) 11.940 12.420 −0.445 0.660
Error count (times) 33.800 37.200 −1.038 0.309
pected.
PT (min) 29.880 29.380 0.255 0.801
The training effects quantified by the PT are shown in Fig. 10(d); the
mean PT of the baseline was 29.8 mins, and the standard deviation was
6.5 mins. After the conventional VR training exercise, the mean PT was Table 9
reduced by 15% to 25.4 mins compared with the baseline. In the VSM- Performance improvement comparison after personalized training using t-test.
based personalized training group, the mean PT was 19.6 mins, which
Indicator Improvement mean Improvement mean t value p value
was reduced by approximately 34% due to the overall error times and for Group 1 for Group 2
WT reduction.
VAT (min) 1.273 2.443 −1.899 6.79E−02
WT (min) 2.567 5.993 −4.066 4.72E−04
5.2. Training productivity evaluation Error Count 4.000 17.863 −5.957 3.68E−06
(times)
PT (min) 4.460 10.264 −3.945 5.08E−04
An example of the productivity estimation can be seen in Table 10.
It is based on one trainee's performance after he obtained the perso-
nalized training. As can be seen from the table, the total number of 5.3. The trainees' confidence evaluation
scaffolding erection activities is 166, the number of non-value-adding
activities is 28, the start timestamp is 0.9, and the finish timestamp is All the trainees were instructed to complete the designed ques-
28.6. The productivity index can be thus estimated to be 3.30. tionnaire, and the results are shown in Fig. 11. As shown, adopting the
Tables 11 and 12 show the productivity indexes of the two groups VSM-based personalized training over the conventional personalized
after the training. Because the training efficiency improved in both training is advantageous in terms of waste identification (from 5.75 to
groups, the average productivity of Group 2 was higher than that of 6.94), effectiveness (from 5.56 to 6.63), helpfulness (from 5.75 to 6.56),
Group 1. The productivity improved by approximately 12%, from 3.76 and satisfaction (from 5.81 to 6.81). However, the two groups did not
to 4.24. This was primarily attributable to a significant error and WT differ much in terms of knowledge acquisition (from 6.06 to 6.31).
reduction in Group 2.

Fig. 9. Example of ideal FSM based on a trainee's performance in Group 2.

10
P. Wang, et al. Automation in Construction 119 (2020) 103355

Fig. 10. Comparison of conventional and VSM-based personalized training: (a) Value-added time; (b) Waste time; (c) Error count; and (d) Processing time.

6. Discussion Table 11
Productivity index of Group 1 after the conventional personalized training.
Based on the experiment results and observations, some discussions No. Qa Qb Ts(min) Tf(min) P
are provided as follows:
1 116 28 0.9 27.6 3.30

• As can be seen from Tables 8 to 12, the improvement of VR-based 2

3
116
116
27
31
1
0.7
21.7
26.7
4.30
3.27
personalized operation training by VSM includes time reduction, 4 116 28 1.5 22.8 4.13
error elimination, and productivity improvement. The average PT of 5 116 39 0.4 30.6 2.55
the two groups was reduced after the individual training ap- 6 116 26 1 21.5 4.39
proaches, but VAT has not been improved significantly compared 7 116 39 0.8 27.4 2.89
8 116 31 1.3 27.6 3.23
with other performance indicators of the two groups. This shows
9 116 31 1.4 20.6 4.43
that there is no significant difference between the two personalized 10 116 30 1 23.8 3.77
training approaches for trainees to learn what essential tasks (value- 11 116 20 0.5 19 5.19
adding activities) are and how to complete the scaffolding opera- 12 116 23 0.8 22.5 4.29
tions. The advantages of the VSM-based personalized training from 13 116 20 0.7 21.6 4.59
14 116 44 1 30.3 2.46
the conventional VR approach is that VSM helps the trainees sys- 15 116 20 0.8 20.8 4.80
tematically understand their wastes during the scaffolding erection 16 116 40 1.2 29.2 2.71
processes through CSM, and further develop the strategy to elim- Average = 3.76
inate them through FSM. This is also reflected in the results of time
reduction, error elimination and productivity improvement.
• As all participants in the experiment did not have any prior lean trainees in the creation of the CSM and FSM could help them clearly
identify wastes that could be eliminated during the training. The
knowledge and related scaffolding construction background, further
benefits of adopting VSM in the operation training were identified. overall productivity (Tables 11 and 12) through the VSM-based
Graphically describing the operational process and involving the training improved from 3.77 to 4.23, approximately a 12% increase

Table 10
An example of trainee's productivity profile.
Exercise after the personalized training – Trainee 1

Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 Step 7 Total

Number of activities N/A Place 22 base Insert 22 base plates into Insert 31 bottom Installation of 31 top Installation of 10 N/A 166
plates standards transoms transoms bracings
VAT (min) 0.9 2.5 3.5 5 3.8 1.8 1 18.5
WT (min) 0 1.7 1.6 3.8 2 1 0 10.1
Waste category N/A D D ABD ABCD CD C N/A
Error count (times) N/A 3 5 11 6 3 N/A 28
PT (min) 0.9 4.2 5.1 8.8 5.8 2.8 1 28.6
Productivity 3.30

Waste category: A: taking wrong scaffolding components; B: unnecessary traveling; C: thinking; D: rework.

11
P. Wang, et al. Automation in Construction 119 (2020) 103355

Table 12 Furthermore, the participants gained the knowledge to use VSM for
Productivity index of Group 2 after the VSM-based personalized training. problem-solving. As shown in Fig. 11, the agreement levels for
No. Qa Qb Ts(min) Tf(min) P knowledge acquisition in terms of the scaffolding erection task did
not change significantly. As stated by one of the participants: “I did
1 116 21 0.9 23.4 4.22 not learn more about scaffolding erection through the VSM-based
2 116 12 0.6 17.7 6.08
personalized training; however, this approach allowed me to learn
3 116 28 1 24.5 3.74
4 116 28 1.2 25.4 3.64
about lean, which I was not aware of previously. Furthermore, it
5 116 23 0.6 20.9 4.58 taught me to solve problems in a different approach.”
6
7
116
116
23
26
0.5
1
23.8
23
3.99
4.09
• The design of the experiment and the VSM-based approach can be
further improved by encapsulating a more sophisticated virtual
8 116 20 0.7 21.3 4.66
scenario and more tools for process automation. Although the VSM-
9 116 24 1 21.9 4.40
10 116 21 0.7 17.9 5.52 based personalized training achieved a better performance than the
11 116 19 1.2 24.8 4.11 conventional approach, it was only proven in a simplified scaf-
12 116 24 0.8 21.6 4.42 folding erection scenario. As mentioned by one of the participants:
13 116 35 1 27.3 3.08
“I can maintain high attention during the reproduction operations
14 116 22 0.7 24.5 3.95
15 116 30 1.2 25.5 3.54
for the seven steps of the scaffolding process. However, if the op-
16 116 28 1 24.1 3.81 eration process is more complicated, I may not be able to remember
Average = 4.24 the steps and perform all improvements even if they are identified
by the VSM-based approach.” In reality, the scaffolding process in-
volves more than seven steps, including safety precautions and er-
compared with the conventional approach. As shown in Fig. 10, the gonomic issues, which must be considered. Further arrangements to
significant improvement after the training was based on the reduced split the training sessions to avoid information overloading is ne-
numbers of errors (improved from 12% to 53%) and WT (improved cessary. Moreover, the CSM and FSM mapping processes in the fu-
from 21% to 50%). These were consistent with the characteristics of ture should be automatically generated and displayed in the virtual
VSM used in the manufacturing and construction sectors [9,43], training scenario for reducing guidance time, as recommended by
which were efficient in identifying and removing/reducing wastes most of the participants.
among value streams. As the VAT decreased by 7% and 14% in
Groups 1 and 2, respectively, it was observed that after the VSM-
• The observations of the experiment were aligned with educational
and perceptional loading principles, in that, a trainee would learn
based training, the trainees could not only identify the problems more if the training materials and feedback information were con-
during operation but also could use an efficient strategy to manage densed and systematic [61]. The struggles from trainees who were
wastes and errors for improvement. For instance, one participant taught via the conventional approach could be caused by the per-
claimed that “compared with a previous VR training I attended, in sonalized feedback information from the trainers, which was always
which the comments were always provided, I can identify the mis- diverged and sometimes more or less than required. According to
takes I made during each step of the operation more effectively in LeMahieu et al. [62], trainers must understand how trainees per-
the current training, and I feel more confident in operating after ceive values and how the values can be transferred into the learning
receiving the instructional feedback.” Judging by the similar re- process. As shown from the results, VSM, as a lean tool for training,
sponses from participants, VSM can be regarded as an effective tool can easily help trainers to systematically identify aspects of the
that provides systematic information related to the training process operational improvements that facilitate the learning of the entire
and fills the gap between knowledge required by the trainee and the scaffolding erection process. In addition, VSM processes can classify
expected improvements after the training. A CSM can confirm both wastes and operational errors for individuals with organized
correct and incorrect activities in detail, and an FSM can indicate thinking to the solutions, whereas the conventional approach relies
the directions pursuable by the trainees to achieve higher opera- significantly on the experience of the trainers and their commu-
tional productivity. nication skills.
• Even though the learning content did not increase by using the VSM-
based approach, the CSM and FSM were constructed during the
7. Summary, conclusion and future studies
guidance sessions not only for the training improvements, but also
to facilitate the spread of lean thinking to the participants.
In many studies, VSM has been used as a lean tool to reveal the

Fig. 11. Comparison of mean values of training effects between the conventional and VSM-based personalized VR training groups.

12
P. Wang, et al. Automation in Construction 119 (2020) 103355

wastes, inefficiencies, and non-value adding activities for productivity influence the work reported in this paper.
improvement in manufacturing and construction sectors [63,64]. Given
its benefits, the effectiveness of VSM in construction education and Acknowledgment
training, especially when integrated with other advanced technologies,
such as VR, should be investigated. VR has proven to be effective in This study was funded by the Australian Research Council Discovery
providing better understanding and visualization capabilities. Pedro Project (No. DP18010402) and scientific research fund project of
et al., [14] argued that many studies on VR training focus on the iso- Southwest University, China (SWU 1908037).
lated application of VR to address a specific training need, it is im-
perative to understand how to improve the training productivity in- References
dividually. Studies [45,46] have demonstrated that VR can help
improve individual operation training performance, but these training [1] N.C. Anderson, J.V. Kovach, Reducing welding defects in turnaround projects: a
platforms do not systematically provide effective instruction and feed- lean six sigma case study, Qual. Eng. 26 (2) (2014) 168–181, https://doi.org/10.
1080/08982112.2013.801492.
back for individuals. Compared with prior studies, this study provides [2] S. Mostafa, S.-H. Lee, J. Dumrak, N. Chileshe, H. Soltan, Lean thinking for a
an example of how VSM, as one of the lean tools, can help VR perso- maintenance process, Production & Manufacturing Research 3 (1) (2015) 236–272,
nalized operation training to improve training productivity. https://doi.org/10.1080/21693277.2015.1074124.
[3] L. Zhang, X. Chen, Role of lean tools in supporting knowledge creation and per-
To determine the benefits of lean thinking in operation training, a formance in lean construction, Procedia Engineering 145 (2016) 1267–1274,
VR scaffolding erection scenario, as an example, was conducted in this https://doi.org/10.1016/j.proeng.2016.04.163.
study to evaluate the training effectiveness between conventional and [4] G. Heravi, M. Firoozi, Production process improvement of buildings’ prefabricated
steel frames using value stream mapping, Int. J. Adv. Manuf. Technol. 89 (9–12)
VSM-based VR personalized training. The former requires conventional (2017) 3307–3321, https://doi.org/10.1007/s00170-016-9306-9.
training guidance, in which the trainers educate trainees directly [5] A. Bayo-Moriones, A. Bello-Pintado, J. Merino-Díaz de Cerio, 5S use in manu-
through video recordings and observations. The latter implies a lean- facturing plants: contextual factors and impact on operating performance,
International Journal of Quality & Reliability Management 27 (2) (2010) 217–230,
based training approach, in which the trainee is trained systematically
https://doi.org/10.1108/02656711011014320.
using VSM tools. From the results identified in the experiment, both [6] B.D. Gratiela, Study case: yellow tag vs quality management, Procedia Soc. Behav.
approaches could effectively enhance the VR training performance. Sci. 62 (2012) 313–318, https://doi.org/10.1016/j.sbspro.2012.09.051.
However, adopting VSM-based personalized guidance demonstrated [7] C.N. Ezema, E.C. Okafor, C.C. Okezie, Industrial design and simulation of a JIT
material handling system, Cogent Engineering 4 (1) (2017) 1292864, https://doi.
better productivity improvement than adopting the conventional per- org/10.1080/23311916.2017.1292864.
sonalized guidance. Furthermore, participants in conventional training [8] B.M. Deros, N. Saibani, B. Yunos, M.N.A. Rahman, J.A. Ghani, Evaluation of
demonstrated significantly higher error and WT reduction. Hence, training effectiveness on advanced quality management practices, Procedia Soc.
Behav. Sci. 56 (2012) 67–73, https://doi.org/10.1016/j.sbspro.2012.09.633.
VSM-based personalized training was more efficient compared with the [9] R. Ahmad, C. Masse, S. Jituri, J. Doucette, P. Mertiny, Alberta learning factory for
conventional VR training approach, which relied significantly on the training reconfigurable assembly process value stream mapping, Procedia
experience of trainers. The overall training productivity improved by Manufacturing 23 (2018) 237–242, https://doi.org/10.1016/j.promfg.2018.04.
023.
12% compared with conventional training. This demonstrates that [10] C.S. Park, Q.T. Le, A. Pedro, C.R. Lim, Interactive building anatomy modeling for
VSM, as a lean tool, is more effective in reducing waste during the experiential building construction education, J. Prof. Issues Eng. Educ. Pract. 142
teaching and learning processes and offers a good example of how lean (3) (2015) 04015019, , https://doi.org/10.1061/(ASCE)EI.1943-5541.0000268.
[11] C. Clevenger, C. Lopez del Puerto, S. Glick, Interactive BIM-enabled safety training
thinking can facilitate VR operation training. The contribution of this piloted in construction education, Advances in Engineering Education 4 (3) (2015)
study is that a systematic VSM-based VR personalized training protocol n3 https://advances.asee.org/publication/interactive-bim-enabled-safety-training-
is developed to enrich the learning tools of operation training by in- piloted-in-construction-education/ Accessed data:10 Nov.2019.
[12] D. Nikolic, S. Lee, S.E. Zappe, J. Messner, Integrating simulation games into con-
tegrating lean thinking into the training process. Compared with con-
struction curricula-the VCS3 case study, Int. J. Eng. Educ. 31 (6) (2015) 1661–1677.
ventional training processes, VSM-based training can effectively im- [13] H. Li, G. Chan, M. Skitmore, Visualizing safety assessment by integrating the use of
prove training productivity, especially in waste identification and error game technology, Autom. Constr. 22 (2012) 498–505, https://doi.org/10.1016/j.
reduction. autcon.2011.11.009.
[14] A. Pedro, Q.T. Le, C.S. Park, Framework for integrating safety into construction
However, this study has certain limitations. All the test participants methods education through interactive virtual reality, J. Prof. Issues Eng. Educ.
were civil engineering undergraduates aged 20 to 22, without any ex- Pract. 142 (2) (2015) 04015011, , https://doi.org/10.1061/(ASCE)EI.1943-5541.
perience in VR operation, and all were males. The actual onsite workers 0000261.
[15] A.Z. Sampaio, O.P. Martins, The application of virtual reality technology in the
may come from different countries and have different cultures, which construction of bridge: the cantilever and incremental launching methods, Autom.
could have a certain impact on the effectiveness of training. Differences Constr. 37 (2014) 58–67, https://doi.org/10.1016/j.autcon.2013.10.015.
of the participants in age and gender during the training could be fur- [16] I. Jeelani, A. Albert, R. Azevedo, E.J. Jaselskis, Development and testing of a per-
sonalized hazard-recognition training intervention, J. Constr. Eng. Manag. 143 (5)
ther discussed, too. They can be verified in future studies. The process (2016) 04016120, , https://doi.org/10.1061/(ASCE)CO.1943-7862.0001256.
of scaffolding erection was simplified in the VR training scenario, in [17] I. Jeelani, K. Han, A. Albert, Development of immersive personalized training en-
which only seven steps were designed, and only the essential proce- vironment for construction workers, Computing in Civil Engineering 2017 (2017)
407–415, https://doi.org/10.1061/9780784480830.050.
dures of the operation were considered. The sense of weight, safety
[18] J.F. Krafcik, Triumph of the lean production system, MIT Sloan Manag. Rev. 30 (1)
precautions, and working posture were not evaluated in the VR training (1988) 41, https://doi.org/10.1023/A:1013933407680.
scenario. Moreover, the training scenario was only suitable for a single- [19] P. Myerson, Lean Supply Chain and Logistics Management, McGraw-Hill New York,
NY, 9780071766265, 2012.
person training process. In reality, the tasks may be completed by
[20] J. Antony, J. Douglas, A. Douglas, Waste identification and elimination in HEIs: the
multiple workers with the potential of cooperation. Future studies will role of lean thinking, International Journal of Quality & Reliability Management
focus more on addressing the abovementioned limitations by further (2015), https://doi.org/10.1108/IJQRM-10-2014-0160.
investigating the ergonomics and safety indicators and extending the [21] S. Vukadinovic, M. Djapan, I. Macuzic, Education for lean & lean for education: a
literature review, International Journal for Quality Research 11 (1) (2017), https://
scenario to a multiuser cooperative scenario, which can yield a more doi.org/10.18421/IJQR11.01-03.
realistic operation simulation. Approaches to automatically generate [22] D.R. Simmons, G. Young, Improving the student academic experience through lean
CSMs and FSMs will be considered for future improvements. engineering principles, 2014 IEEE Frontiers in Education Conference (FIE)
Proceedings, IEEE, 2014, pp. 1–4, , https://doi.org/10.1109/FIE.2014.7044326.
[23] A.C. Alves, S. Flumerfelt, F.-J. Kahlen, Lean Education: An Overview of Current
Issues, Springer, 2016, https://doi.org/10.1007/978-3-319-45830-4.
Declaration of competing interest [24] A.F.U. Mansur, F.C. Leite, H.P.P. Bastos, Lean Education for Applied Science
Universities: A Proposal by Federal Institutes of Applied Sciences in Brazil, Lean
Education: An Overview of Current Issues, Springer, 2017, pp. 25–40, https://doi.
The authors declare that they have no known competing financial org/10.1007/978-3-319-45830-4_2.
interests or personal relationships that could have appeared to [25] M.S. Bajjou, A. Chafi, A. En-Nadi, A comparative study between lean construction

13
P. Wang, et al. Automation in Construction 119 (2020) 103355

and the traditional production system, International Journal of Engineering outcomes, 2010 IEEE Frontiers in Education Conference (FIE), IEEE, 2010, https://
Research in Africa, Vol. 29 Trans Tech Publ, 2017, pp. 118–132, , https://doi.org/ doi.org/10.1109/FIE.2010.5673395 T1H-1-T1H-2.
10.4028/www.scientific.net/JERA.29.118. [45] H. Li, G. Chan, M. Skitmore, Multiuser virtual safety training system for tower crane
[26] R.I. Morien, Legility in Pedagogy: Applying Logistics and Supply Chain dismantlement, J. Comput. Civ. Eng. 26 (5) (2012) 638–647, https://doi.org/10.
Management Thinking to Higher Education, Teacher Education in the 21st Century, 1061/(ASCE)CP.1943-5487.0000170.
IntechOpen, 2019, https://doi.org/10.5772/intechopen.85264. [46] H.A. Adas, S. Shetty, S.K. Hargrove, Virtual and augmented reality based assembly
[27] G. Polat, G. Ballard, Waste in Turkish construction: need for lean construction design system for personalized learning, 2013 Science and Information Conference,
techniques, Proceedings of the 12th Annual Conference of the International Group IEEE, 2013, pp. 696–702. https://ieeexplore.ieee.org/document/6661817 Accessed
for Lean Construction IGLC-12, August, Denmark, 2004, pp. 488–501. http://iglc. date: 15 Nov. 2019.
net/Papers/Details/324/pdf Accessed date:10 Nov.2019. [47] T. Cheng, J. Teizer, Real-time resource location data collection and visualization
[28] S.-H. Lee, J.E. Diekmann, A.D. Songer, H. Brown, Identifying waste: applications of technology for construction safety and activity monitoring applications, Autom.
construction process analysis, Proceedings of the Seventh Annual Conference of the Constr. 34 (2013) 3–15, https://doi.org/10.1016/j.autcon.2012.10.017.
International Group for Lean Construction, 1999, pp. 63–72. http://iglc.net/ [48] Q.T. Le, A. Pedro, C.R. Lim, H.T. Park, C.S. Park, H.K. Kim, A framework for using
Papers/Details/77 Accessed: 11 Nov.2009. mobile based virtual reality and augmented reality for experiential construction
[29] D. Pavlović, M. Todorović, S. Mladenović, P. Milosavljević, The role of quality safety education, International Journal of Engineering Education 31 (3) (2015)
methods in improving education process: case study, Serbian Journal of 713–725. https://www.researchgate.net/publication/276025929_A_Framework_
Management 9 (2) (2014) 219–230, https://doi.org/10.5937/sjm9-5538. for_Using_Mobile_Based_Virtual_Reality_and_Augmented_Reality_for_Experiential_
[30] I. Nahmens, L.H. Ikuma, Effects of lean construction on sustainability of modular Construction_Safety_Education Accessed date:15 Nov. 2019.
homebuilding, J. Archit. Eng. 18 (2) (2011) 155–163, https://doi.org/10.1061/ [49] C. Berry, A guide to safe scaffolding, Division of, https://www.safety.duke.edu/
(ASCE)AE.1943-5568.0000054. sites/www.safety.duke.edu/files/ig38.pdf, (2002) Accessed date:25 Nov.2019.
[31] T. Abdelhamid, S. Salem, Lean construction: a new paradigm for managing con- [50] P. Wang, P. Wu, X. Wang, X. Chen, T. Zhou, Developing Optimal Scaffolding
struction projects, Proceedings of the International Workshop on Innovations in Erection Through the Integration of Lean and Work Posture Analysis, Engineering,
Materials and Design of Civil Infrastructure, 2005, pp. 28–29. https://www. Construction and Architectural Management, 2020, https://doi.org/10.1108/
researchgate.net/publication/242085758_LEAN_CONSTRUCTION_A_NEW_ ECAM-04-2019-0193.
PARADIGM_FOR_MANAGING_CONSTRUCTION_PROJECTS Accessed data:16 Nov. [51] M. Rother, J. Shook, Learning to See: Value Stream Mapping to Add Value and
2019. Eliminate Muda, Lean Enterprise Institute, 2003. https://www.researchgate.net/
[32] A.S.I. Conte, D. Gransberg, Lean construction: from theory to practice, AACE Int. publication/244433983_Learning_to_See_Value_Stream_Mapping_to_Create_Value_
Trans. 10 (1) (2001), https://search.proquest.com/docview/208190113?pq- and_Eliminate_Muda.
origsite=gscholar Accessed date: 21 Nov. 2019. [52] B.K. Jeong, T.E. Yoon, Improving IT process management through value stream
[33] D. Setijono, R. Al-Aomar, A lean construction framework with Six Sigma rating, mapping approach: a case study, JISTEM-Journal of Information Systems and
International Journal of Lean Six Sigma (2012), https://doi.org/10.1108/ Technology Management 13 (3) (2016) 389–404, https://doi.org/10.4301/s1807-
20401461211284761. 17752016000300002.
[34] H. Ismail, Z.M. Yusof, Perceptions towards non-value-adding activities during the [53] S. Tyagi, A. Choudhary, X. Cai, K. Yang, Value stream mapping to reduce the lead-
construction process, MATEC Web of Conferences, Vol. 66 EDP Sciences, 2016, p. time of a product development process, Int. J. Prod. Econ. 160 (2015) 202–212,
00015, , https://doi.org/10.1051/matecconf/20166600015. https://doi.org/10.1016/j.ijpe.2014.11.002.
[35] M.A. Marhani, A. Jaapar, N.A.A. Bari, M. Zawawi, Sustainability through lean [54] R. Barathwaj, R. Singh, G. Gunarani, Lean construction: value stream mapping for
construction approach: a literature review, Procedia Soc. Behav. Sci. 101 (2013) residentials construction, International Journal of Civil Engineering and
90–99, https://doi.org/10.1016/j.sbspro.2013.07.182. Technology 8 (5) (2017), https://www.researchgate.net/publication/317768016_
[36] R.H. Ansah, S. Sorooshian, Effect of lean tools to control external environment risks Lean_construction_Value_Stream_Mapping_for_residentials_construction Accessed
of construction projects, Sustain. Cities Soc. 32 (2017) 348–356, https://doi.org/ date: 26 Nov. 2019.
10.1016/j.scs.2017.03.027. [55] P.C. Earley, G.B. Northcraft, C. Lee, T.R. Lituchy, Impact of process and outcome
[37] M. Bajjou, A. Chafi, A. Ennadi, M. El Hammoumi, The practical relationships be- feedback on the relation of goal setting to task performance, Acad. Manag. J. 33 (1)
tween lean construction tools and sustainable development: a literature review, (1990) 87–105, https://doi.org/10.2307/256353.
Journal of Engineering Science & Technology Review 10 (4) (2017), https://doi. [56] J. Hattie, H. Timperley, The power of feedback, Rev. Educ. Res. 77 (1) (2007)
org/10.25103/jestr.104.20. 81–112, https://doi.org/10.3102/003465430298487.
[38] F. Yalçin Tilfarlioǧ Lu, A new method in education: lean, Electronic Turkish Studies [57] Z. Oseni, H.H. Than, E. Kolakowska, L. Chalmers, B. Hanboonkunupakarn,
12 (6) (2017), https://doi.org/10.7827/TurkishStudies.11489. R. McGready, Video-based feedback as a method for training rural healthcare
[39] E. Sfakianaki, A. Kakouris, Lean thinking for education: development and validation workers to manage medical emergencies: a pilot study, BMC Medical Education 17
of an instrument, International Journal of Quality & Reliability Management 36 (6) (1) (2017) 149, https://doi.org/10.1186/s12909-017-0975-3.
(2019) 917–950 (doi:10.1108.IJQRM-07-2018-0202). [58] D.H. Johnson, The insignificance of statistical significance testing, J. Wildl. Manag.
[40] F.C. Filip, V. Mărăscu-Klein, The 5S lean method as a tool of industrial management (1999) 763–772, https://doi.org/10.2307/3802789.
performances, IOP Conference Series: Materials Science and Engineering, Vol. 95 [59] R. Sacks, A. Esquenazi, M. Goldin, LEAPCON: simulation of lean construction of
IOP Publishing, 2015, p. 012127, , https://doi.org/10.1088/1757-899X/95/1/ high-rise apartment buildings, J. Constr. Eng. Manag. 133 (7) (2007) 529–539,
012127. https://doi.org/10.1061/(asce)0733-9364(2007)133:7(529).
[41] M. Dudek-Burlikowska, D. Szewieczek, The Poka-Yoke method as an improving [60] J. Cohen, Statistical Power Analysis for the Behavioral Sciences, 2nd edition,
quality tool of operations in the process, Journal of Achievements in Materials and Lawrence Erlbaum, Hillsdale, 9780071766265, 1988.
Manufacturing Engineering 36 (1) (2009) 95–102. http://jamme.acmsse.h2.pl/ [61] S.L. Frank, Learn more by training less: systematicity in sentence processing by
papers_vol36_1/36112.pdf Accessed date: 16 Nov. 2019. recurrent networks, Connect. Sci. 18 (3) (2006) 287–302, https://doi.org/10.1108/
[42] K. Rosentrater, R. Balamuralikrishna, Value Stream Mapping–A Tool for QAE-12-2016-0081.
Engineering and Technology Education and Practice, Proceedings of the 2006 ASEE [62] P.G. LeMahieu, L.E. Nordstrum, P. Greco, Lean for education, Qual. Assur. Educ. 25
IL/IN Conference, Fort Wayne, IN, USA, http://www.ars.usda.gov/research/ (1) (2017) 74–90, https://doi.org/10.1108/QAE-12-2016-0081.
publications/Publications.htm?seq_no_115=194841, (2006). [63] H. Yu, M. Al-Hussein, S. Al-Jibouri, A. Telyas, Lean transformation in a modular
[43] M. Lobaugh, The value of value stream mapping to students, Proceedings of the building company: a case for implementation, J. Manag. Eng. 29 (1) (2011)
2008 American Society for Engineering Education Annual Conference & Exposition, 103–111, https://doi.org/10.1061/(ASCE)ME.1943-5479.0000115.
2008, pp. 1–12. http://peer.asee.org/4416 Accessed date: 20 Nov. 2019. [64] W. Shou, J. Wang, P. Wu, X. Wang, H.-Y. Chong, A cross-sector review on the use of
[44] C. Steinlicht, A. Neary, T. LeBrun, K. Sauke, C. Sundermann, J. Weber, Program value stream mapping, Int. J. Prod. Res. 55 (13) (2017) 3906–3928, https://doi.
development work in progress: value stream mapping the educational process org/10.1080/00207543.2017.1311031.

14