(Haitao Et Al 2013) Lean Transformation in A Modular Building Company A Case For Implementation
(Haitao Et Al 2013) Lean Transformation in A Modular Building Company A Case For Implementation
(Haitao Et Al 2013) Lean Transformation in A Modular Building Company A Case For Implementation
Haitao Yu1; Mohamed Al-Hussein, M.ASCE2; Saad Al-Jibouri3; and Avi Telyas4
Abstract: Encouraged by the remarkable productivity improvements in the manufacturing sector, the construction industry has a long
history of trying to garner the benefits of manufacturing technologies. Whereas industrialized construction methods, such as modular
and manufactured buildings, have evolved over decades, core techniques used in prefabrication plants vary only slightly from those employed
in traditional site-built construction. The objective of this research was to develop and implement a production system for the effective
application of lean tools in building components prefabrication. To overcome the prevalent skepticism among middle management, the lean
journey started with a pilot project involving one production line. Over a six-month period, lean tools such as 5S (sort, straighten, shine,
standardize, and sustain), standardized work, takt time planning, variation management, and value stream mapping were implemented to a
communication shelter production line. The implementation successfully won the support of the middle managers and established the foun-
dation for expanding lean practices to other parts of the factory and applying relevant lean tools and techniques. DOI: 10.1061/(ASCE)ME.1943-
5479.0000115. © 2013 American Society of Civil Engineers.
CE Database subject headings: Manufacturing; Lean construction; Productivity; Case studies; Construction industry.
Author keywords: Manufacturing; Lean construction; Production management; Productivity.
research in this area were plant layouts optimization based on car size and style, body type, engine size, and transmission method.
lean principles and lean implementation plan development using If colors and combinations of various options were included, it was
simulation (Senghore et al. 2004; Mehrotra et al. 2005; Jeong rare to see two completely identical cars (Ohno 1998).
et al. 2006). Moreover, an automobile, like a building modular, is made of
This paper describes a lean implementation initiative in a thousands of parts; the number of processes involved is immense. It
U.S.-based modular building company, Kullman Building Corpo- is extremely difficult to apply just-in-time (JIT) to a production plan
rate (KBC). In collaboration with the University of Alberta, a of every process in an orderly way. An upset in prediction, a
modular production line that was relatively easy to control was mistake in the paperwork, defective products and rework, trouble
selected as a lean pilot project, and a number of lean production with the equipment, absenteeism—the problems that could arise
techniques were tailored and implemented. The lean journey started are countless. A problem early in the process always results in a
with 5S and standardized work. This was followed by using the defect product later. Stops and changes in the production process
value stream mapping technique to analyze current practice, formu- will happen regardless of planning.
late a lean production model, and develop a kaizen plan. The
implementation results were documented in detail to verify the
effectiveness of the lean production model. Through the lean trans- Background of Case Study
formation of the production line, middle management was con-
vinced that lean production was workable and could remarkably The case study company, KBC, is one of the leading modular build-
improve productivity and reduce waste. The success also led to ing manufacturers in the U.S. The company was founded in 1927
the lean implementation in other parts of the factory and manage- by a young salesperson to provide turn-key, portable diners that
ment’s long-term commitment to a lean culture. served a market seeking fast, low-cost, home-cooked meals. After
80 years of growth, KBC now has over 200 employees and has
expanded its market to produce a variety of building types, includ-
Lean Production and the Modular Building Industry ing equipment shelters, schools, dormitories and multistory resi-
dential buildings, correctional facilities, healthcare facilities, and
Although in the past 30 years lean production has been studied and U.S. embassies. The company coined the term accelerated con-
introduced into numerous workplaces regardless of industrial field struction to describe a building process free from uncertainties
or scale, there have always been arguments by the construction of weather, site conditions, and contractor relations. While modu-
practitioners that construction is distinct from auto manufacturing larization provides KBC significant competitive advantages in
and that lean production is not applicable. A high level of custom- terms of site construction time, quality control, and predictability,
ization in building design makes building modules mostly one-of- the company has not yet realized its full potential. Considering the
a-kind products. The need for variety has been regarded by many as cost of transportation and installation, KBC’s products were typi-
a major reason that a lean production system is not feasible cally 10–20% more expensive than their counterparts built on-site.
for modular production, but ironically, variety is in fact the soil As a result, its customers were limited primarily to wireless prov-
in which the lean production system was cultivated and the very iders, education institutes, and government, which are less cost
trades and materials went to the building in sequence, similar to the standard modules with similar configurations, standardization of
on-site construction. the production process was relatively easy. Meanwhile, as a major
Both the KBC chief executive officer (CEO) and the chief production line generating 40% of the overall revenue of the com-
operating officer (COO) came from the semiconductor industry pany, it was highly visible. Backed by the top management, a top-
and had profound knowledge in lean manufacturing; both of them down approach was adopted, and the objectives of the project were
believed that modularization would be the future of construction defined as synchronizing the production line to takt time, reducing
and had a passion to transform KBC from a traditional construction average labor hours per module by 20%, and controlling overtime
company to a lean manufacturer. The company had two vice pres- under 10% of total labor hours. The duration of the pilot project
idents (VP), including VP of personnel, accounting, and project was scheduled as six months, and the first step was planned to be
management, and VP of sales. the process documentation of the communication shelter line and a
The KBC organization had a conventional hierarchical struc- 5S campaign.
ture, with a manager for each functional area. The head of produc- The 5S plan has been recommended by many lean experts as the
tion was a plant manager who reported to the CEO. The production starting point of lean transformation (Productivity Press 2006), be-
planning and workforce on the floor were managed by a production cause compared to other lean tools, 5S, which focuses on cleaning
control manager, assisted by five line managers. A logistics man- and organizing the workplace, is easier to get workers to buy-in and
ager was responsible for the procurement and storage of all materi- it produces immediate visible results. More importantly, effective
als and tools. All middle managers in the production department cleaning and efforts to sustain a better organized workplace involve
were company veterans, having been with the company for more many key lean principles and methods, such as standardized work
than ten years. Although most of them had taken Lean 101 training, and visual management. The 5S plan can help people that have no
they did not have any experience in manufacturing and showed lean production experience build teamwork, discipline, and a cul-
little interest in lean implementation at the beginning of the project. ture of continuous improvement, which are the cornerstones of
The frontline workforce of KBC consisted of 110 full-time em- lean implementation.
ployees and temporary workers from two agencies. These workers After shop floor observation, a series of workshops were held
represented a wide variety of trade disciplines: carpentry, welding, with frontline workers to review the findings. At each workshop,
electrical, painting, plumbing, rigging, and computer numerical the basic concepts of 5S and seven types of wastes were explained,
control machine operation. The KBC employees were unionized, pictures like Fig. 2(a) were presented, and examples of workplace
with an average of four years of tenure. Frontline workers had also management at some world-class lean enterprises were introduced.
taken Lean 101 training provided by an external consultant, but Following these, a brainstorming session was conducted to identify
they were generally reluctant to change. Skepticism was prevalent three top areas of waste and to develop a team action plan with
because of concerns that eliminating waste might result in in- possible solutions, completion deadline, and persons responsible.
creased work intensity and workforce reduction. The production manager and line managers also attended these
workshops to show their commitment and support to the initiatives.
In two weeks, workers at the communication shelter line were
Lean Pilot Project and 5S organized into eight workgroups, each with its own 5S action plan
and biweekly 5S meetings. The working condition of the shop floor
Lean transformation requires complete commitment from top man- was remarkably improved, as shown in Fig. 2(b). One advantage of
agement, but the biggest challenge and key success factor is to let starting a lean journey with 5S was that people could see the results
a natural choice, because a given task should always be performed job, they did not have to refer to standard worksheets during their
at a designated location so that required materials and equipment operation. However, the combination table and standard work pro-
could be put next to that location. Comparing the two states shown cedures were posted at each station to provide a visual reference
in Fig. 2, the second one was not only much cleaner, but held all the for management to check adherence to the standard. Any deviation
materials and equipment required by the operations at station 3. from the standard meant an abnormal situation, usually caused by
problems. The role of management was to recognize the deviation,
uncover the root causes, ensure that they were corrected quickly,
Standardized Work and Variation Management and reestablish the standardized work.
One unique characteristic of construction is the high level of
Standardized work is regarded as the backbone of lean processes customization; it is rare to see two identical buildings. The produc-
and the basis for continuous improvement and quality. If a process tion of communication shelters faced the same challenge. Some of
is always shifting, then any effort for improvement just creates one the modules were so different that workloads in one or two stations
more variation that is occasionally used and mostly ignored (Liker changed dramatically. For example, compared to a standard 3.7 ×
2004). One common problem in construction is that most construc- 9.1 m (12 × 30 ft) module with a single interior gypsum board and
tion tasks are done by trades people based on their skill and expe- exterior stenni finish, a module with double-layer interior and
rience. Although the production of communication shelters in KBC exterior boarding, waterproofing, and hand-laid brick almost
was on an assembly line, the production process was highly unpre- doubled the carpenters’ workloads at station 2 and station 3.
dictable. The line manager dynamically assigned workers to tasks The extra workload for one task might have a remarkable impact
on a daily or hourly basis, and the line was moved when most of the on the overall work sequence. When there was only one layer of
tasks had been done. As a result, the production process varied and interior gypsum board, the interior finishing crew started its work
people did not know which state was normal. one hour after the start of the installation of interior boarding so
The first step of standardized work was to determine takt time, they could both complete their job within takt time. However, if
the maximum time allowed for a modular to stay in a station. there were two layers of interior boarding, the interior finishing
According to its definition, takt time can be calculated using the
crew could not start until the entire first layer and 20% of the sec-
following formula:
ond layer of interior boards had been completed. To maintain the
Ta synchronization of the production process, measures had to be
T¼ ð1Þ taken to accommodate workload fluctuation.
Td
The two most commonly used methods for this purpose are
where T = takt time; T a = net time available to work; and T d = overtime and workforce pool, but they need to be used in a system-
customer demand. Based on a time series analysis of the historical atic way to minimize waste. In KBC, different standard work
data of customer orders and demand forecast provided by custom- sequences for each station were developed to deal with the different
ers, takt time was determined as seven hours in months two to four module types and takt time requirements. For instance, the work
and six hours in months five to seven. Because the average produc- combination table shown in Fig. 3 was an 8-h takt time work
tion cycle time in the current month (month one) was eight hours, to sequence for a 3.7 × 9.1 m (12 × 30 ft) module with double
synchronize the production line to the takt time target meant a 25% interior and exterior gypsum board, which accounts for 35% of
improvement in production capacity in three months. modules passing through the station. There were separate work
combination tables for the 3.7 × 9.1 m (12 × 30 ft) modules with VSM team to handle. A high-level value stream map like the
single or no boarding and the 3.7 × 6.1 m (12 × 20 ft) modules. one shown in Fig. 5 provided a big picture of the process, but
Float workforce, which did not belong to any work station, was showed little detail of the operations; it could not be used for root
used to deal with the extra workload for different module types, cause analysis and future map formulation. In contrast, a station
schedule delays, workers’ vacations, and absenteeism. In normal could be seen as a value stream with the preceding station as
situations, the float workforce worked on an off-line module, which the supplier and the following station as the customer, as shown
was usually a special module with significant workload variation in Fig. 6. The problem with mapping the process at this level
that could not be accommodated at the production line. was that the value stream was not stable, because it might be nec-
essary to move a task from one station to another to optimize the
production line or to achieve a different takt time. In this research,
Value Stream Mapping maps at two levels were used simultaneously for waste identifica-
tion and solution development.
Standardized work is a great tool to stabilize the process and help Upon drawing up the current-state map, several wastes could be
people identify problems that lead to abnormal situations. However, identified immediately. For a linear assembly line, a basic lean prin-
a standardized procedure is not necessarily the best practice, and ciple is to balance the workload and synchronize the station cycle time
the current process often needs to be adjusted to meet the changed to takt time. This may be common sense, but in practice it is not an
requirements. In the case of KBC, a new production model was easy job for production management. From the data in Fig. 5, it is easy
needed to reduce the takt time of the communication shield line to see that the production cycle times (C/T) of some stations were
from eight hours to six hours. To achieve this goal, value stream much lower than the takt time, which was 480 min in the current-
mapping (VSM), a widely used lean planning tool, was selected state map. That meant the production capacities of those stations were
because of its process view and the ability to link lean initiatives higher than the demand. Conversely, the percentage of completion
into a whole. Based on standardized work, a four-step method was (Completion), which was defined as the percentage of modules with
adopted to develop the future lean production model: (1) current- all tasks completed at the time of being moved to the next station, were
state mapping; (2) existing practice analysis; (3) formulation of a quite low, even at the stations with a much lower cycle time. For in-
future production model; and (4) laboratory testing of the model stance, the average station cycle time of station 3 was 400 min, ap-
using simulation. proximately 17% lower than the takt time, but the percentage of
Prior to the commencement of VSM, two management completion at station 3 was only 80%; one of every five modules
decisions must be made: (1) select a value stream; and (2) decide moved to the next station had unfinished tasks. For stations with a
the level of mapping. In this research, those two decisions were cycle time close to the takt time, the incompletion rates were much
interrelated. When the door-to-door production flow was looked higher. In the case of station 5, almost half of the modules were moved
at as a value stream, the mapping could only be done at the station out of the station unfinished. The ripple effect of unfinished tasks
level, because a single map encompassing all tasks conducted disturbed the production pace and led to significant wastes. For in-
within each station would be too large and cumbersome for a stance, if the electricians at station 4 did not finish the rough conduit
in the equipment room, they needed to continue their job at station 5. Moreover, installing rough conduits at station 5 meant that workers
Then the workers who should pull electrical wires could not start their had to move back and forth between station 4 and station 5 to get
job on time, so they were idle at the beginning and had to hurry at the tools and materials, and the more time they spent at station 5, the less
end to finish their job before the module moved to the next station. time they had for the next module in station 4.
I
QS
Station 1 Station 2 Station 3 Station 4 Station 5 Station 6 Station 7 Station 8
QS QS QS QS QS QS QS
Interior drywall FRP & plywood Roof termination Cable tray VCT
Final check & ship
Insulation Ceiling, light & uni- Stenni Paint door frame Finish wiring Hang doors
loose
Erection Ceiling, light & uni- strut (E-room) Rough conduit A/C unit Electrical Paint base frame
Electrical testing
strut (G-room) Exterior J-channel Start grounding Louver & cable termination Paint floor
Load trailor
Exterior J-channel for stenni port Ground holes Finish grounding
2 Exterior drywall Wireway 9 Trim/finish 6 RW 3 2
5 RW 5 RW C/T = 300 min
Electrical wiring
C/T = 290 min
C/T = 300 min Plumbing C/T = 330 min C/T = 320 min
C/T = 360 min Completion=95% 10 RW Completion=100%
Completion=100% C/T = 320 min Completion=95% Completion=100%
# Defects = 0 Completion=90% Completion=90% # Defects = 0 C/T = 360 min # Defects = 0 # Defects = 0 # Defects = 0
# Defects = 0 # Defects = 0 Completion=90%
# Defects = 0
Statistics:
Push Arrow Load
Outside OXOX Processing Time = 2610 min
Data Box Leveling
Sources Production Lead Time = 2915 min
Kaizen
Information Flow Burst T/O: Station Cycle Time
Completion: Percentage of
Process Inventory Completion
Shipment Arrow QS: Quality at the Source
RW: Restructuring Work
A major reason for the low percentage of completion was the of products being produced over a specific time period. In modular
variation of workload caused by different module configurations. production, the major method of load leveling is to establish an
The average cycle time of station 2 was 405 min, but 35% of the optimized production sequence to ensure that the delay in the com-
modules were 3.7 × 9.1 m (13 × 30 ft) with double interior and pletion of one module does not lead to the delay of the next module,
exterior gypsum boards that needed 480 min to install. There was and that crews shared by multiple tasks at different stations have
no time buffer in this case, and any delay would result in unfinished enough time to perform all assigned tasks. As a part of lean pro-
tasks. Although flexible workforce and overtime were effective tools duction planning, the responsibility of deciding production se-
to handle the workload variation, as explained in the previous sec- quence was moved from sales to production line managers. Every
tion, a further measure was necessary to reduce the variation at the Wednesday, the sales department sent a tentative 2-week schedule
source. Quality problems were another cause of high variation in to production based on the sequence of orders and customers’ de-
cycle time. It was quite common that a worker had to stop his or manding dates, and the line manager adjusted the module sequence
her work and go to a downstream station to fix defects. in the schedule based on a set of predefined rules. For instance,
The focus of future-state mapping was to eliminate the root if there was a 3.7 × 9.1 m (12 × 30 ft) module with hand-laid
causes of wastes and to link the value stream in a smooth flow. bricks (Nailite) as exterior finishing in the schedule, two extra
As shown in Fig. 7, three measures, including workload-leveling, workers would be needed at station 4, because the labor hour of
restructuring work, and in-station quality, were used to increase brick installation was almost 60% more than that of stenni instal-
process reliability and achieve a 6-h takt time. The basic idea of lation. According to the production design, these two workers were
workload leveling is to meet varying customer demand (a mix crew 1 of station 2. The optimal production sequence was as shown
of modules with variations) without workload fluctuation in the in Fig. 8.
manufacturing process. In typical manufacturing, load leveling The purpose of restructuring work is to balance the production
is done through a heijunka box showing the quantity of a mix line so that the overall cycle time of each station can be as close to
improvement in terms of production throughput, productivity, modeling for the manufactured housing industry.” J. Urban Plann.
and labor cost was observed. Dev., 132(4), 217–225.
The success of the pilot project won the support of middle man- Koskela, L. (1992). “Application of the new production philosophy to
agement, and research on lean implementation at three other pro- construction.” Technical Rep. #72, Center for Integrated Facility Engi-
duction lines in KBC is ongoing. The communication shelter line is neering, Dept. of Civil Engineering, Stanford Univ., CA.
unique in modular building production. The lean production model Koskela, L. (2000). An Exploration towards a Production Theory and
developed in the pilot project cannot be used for other production Its Application to Construction, VVT, Technical Research Centre of
lines, but the lean implementation approach and key strategies pre- Finland, Espoo, Finland.
sented in this paper could be generalized for the modular building Liker, J. (2004). The Toyota Way, McGraw-Hill, New York.
industry and tailored for any particular modular production line. Manufactured Housing Research Alliance (MHRA). (2005). Getting lean:
The fundamental implementation of this research is that the current Assessing the benefits of lean production in factory built housing, U.S.
Dept. of Housing and Urban Development, Affordable Housing
practice of modular building production has large potential for
Research and Technology Division, Washington, DC.
improvement through the application of lean production principles
Mehrotra, N., Syal, M., and Hastak, M. (2005). “Manufactured housing
and techniques. production layout design.” J. Archit. Eng., 11(1), 25–34.
Milberg, C., and Tommelein, I. (2003). “Role of tolerances and process
capability data in product and process design integration.” Proc.,
References 2003 Construction Research Congress, ASCE, Reston, VA.
Arbulu, R. J., Ballard, G., and Harper, N. (2003). “Kanban in construction.” Nasereddin, M., Mullens, M., and Cope, D. (2007). “Automated simulation
Proc., 11th Annual Conf. of the Int. Group for Lean Construction, development: A strategy for modeling modular housing production.”
Blacksburg, VA. Autom. Constr., 16(2), 212–223.
Ballard, G. (1993). “Lean construction and EPC performance impro- Ohno, T. (1998). Toyota production system: Beyond large-scale produc-
vement.” Lean construction, L. Alarcon, ed., Balkema, Rotterdam, tion, Productivity, New York.
Netherlands, 79–91. Productivity Press. (2006). Visual tools: Collected practices and cases,
Ballard, G. (2000). “The last planner system of production control.” Ph.D. Productivity, New York.
dissertation, Univ. of Birmingham, Birmingham, U.K. Salem, O., Solomon, J., Genaidy, A., and Minkarah, I. (2006). “Lean con-
Ballard, G., and Howell, G. (1994a). “Implementing lean construction: struction: From theory to implementation.” J. Manage. Eng., 22(4),
Improving downstream preference.” Lean construction, L. Alarcon, 168–175.
ed., Balkema, Rotterdam, Netherlands, 111–125. Senghore, O., Hastak, M., Abdelhamid, T. S., AbuHammad, A., and Syal,
Ballard, G., and Howell, G. (1994b). “Implementing lean construction: M. G. (2004). “Production process for manufactured housing.”
Stabilizing work flow.” Proc., 2nd Annual Meeting of the Int. Group J. Constr. Eng. Manage., 130(5), 708–718.
for Lean Construction, Santiago, Chile. Winch, G. (2003). “Models of manufacturing and the construction process:
Bertelsen, S. (2003). “Construction complexity analysis.” Proc., 11th The genesis of re-engineering construction.” Build. Res. Inf., 31(2),
Annual Conf. of the Int. Group for Lean Construction, Blacksburg, VA. 107–118.