You are on page 1of 7

ARTICLE IN PRESS

Int. J. Production Economics 118 (2009) 501–507

Contents lists available at ScienceDirect

Int. J. Production Economics


journal homepage: www.elsevier.com/locate/ijpe

Trading off between heijunka and just-in-sequence


Andreas Hüttmeir a, Suzanne de Treville a,, Ann van Ackere a, Léonard Monnier a,
Johann Prenninger b
a
Faculty of Business and Economics, University of Lausanne, 1015 Lausanne-Dorigny, Switzerland
b
BMW AG, Munich, Germany

a r t i c l e in fo abstract

Article history: The concept of heijunka—controlling the variability of the job arrival sequence to permit
Received 25 October 2006 higher capacity utilization—plays an integral role in lean production theory. In
Accepted 11 December 2008 situations where the customer defines the delivery sequence, however, scheduling
On-line el 19 January 2009
production to maximize utilization becomes more challenging and requires a
Keywords: subsequent reordering. The cost of the extra work and space required by this reordering
Just in sequence needs to be traded off against the value of the higher utilization. We present the results
Heijunka of a stylized simulation-based model of the two approaches inspired by a BMW engine
Lean production plant.
Simulation
& 2009 Elsevier B.V. All rights reserved.
Scheduling

0. Introduction that the producer has a choice concerning the amount of


variability in the job arrival sequence to accept. Rather
Production systems must always balance inventory, than naively accepting all jobs that arrive in the order
capacity utilization, and system variability according to received, the producer can choose to sort orders so that
the laws of ‘‘Factory Physics’’ (Hopp and Spearman, 1996). the arrival sequence seen by production is relatively
de Treville and Antonakis (2006) took the concept a step smooth.
further, proposing to define lean production according to As product variety increases, however, the practice of
its Factory Physics. Lean production thus defined calls for heijunka becomes more challenging. Also, eliminating
high capacity utilization combined with relatively low variability is not always strategically desirable: A supply-
inventories, requiring that system variability (concerning ing company can increase competitiveness through
both arrival and service rates) be minimized. Many lean being able to respond to variations in customer needs
production practices serve to either control inventory (variability that can be exploited to gain a competitive
buildup (kanban) or to reduce system variability (through edge is referred to as strategic variability, Suri, 2003). An
defects, unplanned downtime, exposure to worker absen- alternative approach, referred to as just-in-sequence (JIS)
teeism, etc.). The lean production practice that protects scheduling, has begun to play an increasing role in lean
the producer from variability in the sequence of jobs to be production, even though production lines operating JIS
processed is heijunka, in which production is scheduled will clearly be less lean than those operating according to
such that the production line produces the same sequence heijunka. To what extent should a company be willing to
of products throughout a given time period, with that compromise its leanness to take advantage of such
sequence alternating between demanding and less de- strategic variability?
manding products. The assumption underlying heijunka is The choice between heijunka and JIS scheduling
illustrates the tension between leanness and agility that
increasingly must be sorted out if manufacturing is to be
 Corresponding author. Tel./fax: +41 21 692 3341. competitive (Prince and Kay, 2003; Yao and Carlson, 2003;
E-mail address: suzanne.detreville@unil.ch (S. de Treville). Narasimhan et al., 2004; Browning and Heath, 2008). In

0925-5273/$ - see front matter & 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijpe.2008.12.014
ARTICLE IN PRESS
502 A. Hüttmeir et al. / Int. J. Production Economics 118 (2009) 501–507

this paper, we combine a stylized simulation model with a might increase firm competitiveness when operating in
case study of a BMW engine plant to gain insights into the configuration with other routines or capabilities.
nature of the trade-off and explore ways to combine McLachlin (1997, p. 287) listed ‘‘uniform plant loading’’
limited heijunka with partial JIS production. as a JIT flow element, operationalized as ‘‘(a) there is a
The paper is structured as follows: Sections 1 and 2 fixed and level schedule; (b) they produced the same mix
describe heijunka and JIS and provide a comprehensive of end items or families each day, and possibly each hour
literature review. Section 3 describes the issues the BMW (to match daily demand rates); (c) there is a reduction in
engine plant was facing in 2003 as it worked through the upstream inventory swings and panic reactions to
question of how to schedule production. The model and (changes in) schedule(d) demand; and (d) there is little
its results are presented in Section 4. Finally, the main or no expediting.’’ In six JIT implementations evaluated
findings and implications are briefly summarized in using a Likert scale ranging from 1 to 6, two rated their
Section 5. level of uniform plant loading at 4, two at 3, one at 2, and
one at 1, indicating that for these plants leveling the
production schedule was not a top priority.
1. Heijunka Shah and Ward (2003) did not include heijunka or
leveling the production schedule in their extensive list of
The objective of heijunka is to avoid peaks and valleys lean manufacturing practices. Production smoothing was
in the production schedule. Consider, for example, a mentioned in the context of bottleneck removal rather
workstation that produces two products, A and B, with A than schedule leveling, and there was a general reference
requiring 1.5 min, and B requiring 1 min of processing, to planning and scheduling strategies, which could refer
respectively. Suppose that the company receives an order to either heijunka or JIS scheduling.
for 100 units of both A and B. A naı̈ve schedule would Rinehart (1997) however, described difficulties faced
be to produce 100 units of one product and then 100 by one lean auto assembly plant in leveling the production
units of the other, resulting in a situation in which the schedule. Although heijunka was a stated company policy,
demand faced by the workstation would vary consider- variation in demand meant that it was constantly violated.
ably. Transport this workstation to a production line with Unfortunately, production was scheduled and capacity
a cycle time of 1.4 min, and the workstation is overloaded allocated under the assumption that heijunka was
(and a bottleneck for the entire line) for 100 cycles and functioning correctly, so surges of ‘‘overburdened jobs’’
underloaded for another 100 cycles. Accommodating this resulted in extreme worker stress, repetitive strain
schedule requires increasing the cycle time for the entire injuries, and potential quality problems. In labor disputes
line, at least during the period when A is being produced. that eventually emerged, overburdened jobs resulting
On many production lines it is not realistic to change the from a failure to maintain heijunka were a major point
cycle time to accommodate such workload fluctuations of conflict.
(this would not be possible at the BMW engine plant De Smet and Gelders (1998) noted that implementa-
described in this paper, for example), hence this work- tion of heijunka was only possible in situations where
station might well be obliged to operate at relatively low there were few schedule disturbances, that is, demand
average capacity utilization. was relatively stable and predictable. Cusumano (1994) in
Heijunka calls for distributing the jobs requiring more his article summarizing the limits to lean production
labor input throughout the production schedule to permit observed that production smoothing through techniques
higher average utilization assuming that the cycle time is like heijunka reduces manufacturing flexibility (see also
held constant over time. In our simple example, products Hines et al., 2004). As mentioned in the introduction,
A and B would be alternated, so that the workstation other authors have called for an addition of agility to
could either work in lots of one unit of A and one unit leanness, indicating a need to explore the trade-off
of B, with cycle time determined based on the combined between leanness and responsiveness (Prince and Kay,
work content of 1.5+1.0 ¼ 2.5 min, or allowing the work- 2003; Yao and Carlson, 2003; Narasimhan et al., 2004;
station to get a bit behind during the cycle when A is Browning and Heath, 2008). Such a trade-off has implica-
produced, catching up during the cycle when B is tions for the implementation of heijunka. The purpose of
produced. Monden (1983) suggested a simple algorithm heijunka is to protect the production line from demand
for heijunka scheduling that has been used in practice. volatility, but when that demand volatility is strategic,
Heijunka (also referred to as production smoothing or there may be a need to relax or even eliminate heijunka. In
leveling the production schedule) has played an integral cases where manufacturing flexibility must be added to
role in just-in-time and lean production since its inception leanness, heijunka may be replaced by a JIS production
(Schonberger, 1982; Hall, 1983; Monden, 1983; Womack et planning and control approach, which we describe in the
al., 1990; Chase, 1993; Hopp and Spearman, 1996; following section.
Fujimoto, 1999). Abdulmalek and Rajgopal (2007), for
example, described use of heijunka (implemented using
Monden’s algorithm) as one of several lean production 2. JIS production scheduling
practices implemented on the cold end of an integrated
steel mill. JIS occurs according to a sequence defined by the
Teece et al. (1997) referred to heijunka as an example customer throughout the supply chain. It can be imple-
of a dynamic manufacturing capability or routine that mented in delivery, production, and distribution. Delivering
ARTICLE IN PRESS
A. Hüttmeir et al. / Int. J. Production Economics 118 (2009) 501–507 503

JIS requires either an inventory of parts that can be picked During the late 1990s, BMW management moved
in the right sequence, or producing and/or assembling parts responsiveness to a top priority, implementing a com-
in the order needed (Sayer, 1986; Kempfer, 2005). JIS pany-wide initiative entitled customer-oriented sales
delivery is viewed as playing an increasing role in effective processing to achieve delivery of a car within 10 working
manufacturing (Liker and Wu, 2000), and is now required days of a customer placing an order (Selto et al., 1995;
by an increasing number of companies in addition to BMW Prenninger, 2001; Pietsch, 2002). At the same time,
such as DaimlerChrysler, Saturn, and Dell (Prenninger, strategic variability (variability which an organization
2001; Feare, 2003; Trebilcock, 2006). uses to maintain its competitive edge) was increasing,
One of the earliest success stories of JIS production and such that today there are 1032 possible configurations of a
delivery is told in the Toyota Motor Manufacturing case BMW 7 series car: A car with a given configuration comes
(Mishina and Takeda, 1994), describing how the seat down the production line on average once every 4
supplier developed the capability to produce and deliver months. The number of possible configurations continues
seats in the order required. Toyota had originally sug- to increase, and BMW management has observed that the
gested that the seat supplier produce to stock and deliver profit potential in responding to this variability in demand
JIS (as was done for all other Toyota plants). The seat is substantial. For example, only a few cars per hundred
supplier argued, however, that the production and have head-up displays (in which key information is
transportation lead times for the seat were short enough projected onto the windshield), but these cars are
to permit seat production to begin after the car left considerably more profitable than standard models.
painting (the point at which the final sequence was As delivery lead times decrease and product variability
established due to variability in painting operations) and increases, the area around the assembly line is not
be delivered to the factory floor in time to be installed. sufficient to store the entire array of supplier parts at
Also, the number of seat variations was sufficiently BMW’s assembly plants. Furthermore, reordering the parts
high to require a quite large warehouse, and the number into the correct production sequence at the assembly plant
of variations was increasing over time. The seat supplier takes considerable space. Therefore, in the fall of 2003,
was located 30 min away from the Toyota plant, making three major BMW auto assembly plants in Germany began
JIS production easier. The case also describes, however, to require JIS delivery of parts such as engines.
some of the problems arising with JIS when quality Responding to this request required (a) a warehouse to
problems occurred, as JIS leaves much less latitude for permit engines to be resorted prior to shipment along
rework. JIS also forced the seat supplier and Toyota with labor to do the sorting of engines, (b) a decoupling
to work together to fix the quality problem much point in the production process to allow standard engines
more aggressively than would have occurred otherwise, to be produced to stock and then finalized in the right
so that the disruptions led to organizational learning. sequence, or (c) that the engine plant move to a
This is consistent with Narasimhan et al.’s (2006) results production schedule that matched the required sequence.
that showed that lean producers outperformed agile Although the idea of JIS production was attractive, there
producers in terms of cost, but not in terms of con- was great concern that its implementation would result in
formance quality. In other words, just as removal of unacceptably low capacity utilization. Furthermore, the
inventory has been demonstrated to lead to learning 300- to 600-km distance from the engine plant to
under the right conditions (Suri and de Treville, 1986), so customer plants and the complexity of the product
JIS production can also create powerful motivation for implied that JIS implementation would be relatively
improvement. difficult. The question came down to where in the process
Some authors have described use of a postponement engines should be put into sequence, that is, at what point
strategy to permit JIS delivery and assembly from a in the process did need for flexibility become greater than
decoupling inventory, allowing an emphasis on high need for efficiency?
utilization upstream, and responsiveness downstream Although many companies attempt to achieve JIS
(Feitzinger and Lee, 1997; Lambert et al., 1998; van Hoek, delivery without implementing JIS production through
2001; Abdulmalek and Rajgopal, 2007). combining postponement with a final sorting operation in
a warehouse, experience shows that this solution becomes
unmanageable as companies look to strategic variability
3. BMW to increase their margins. For example, as BMW exploits
new possibilities to customize engine and car options, it
Along the lines discussed in the previous two sections, will require an enormous warehouse to produce according
BMW faced a choice between heijunka and JIS production to a heijunka-defined schedule yet deliver JIS. A rethinking
in one of their European engine factories in 2003. The of the balance between heijunka and JIS production is
engine plant had historically been organized around the clearly required.
concept of heijunka, permitting extremely stable produc-
tion and a high utilization (ranging between 95% and
99%). The smoothing of the production schedule was 4. Modeling the heijunka-JIS trade-off
carried out by the advanced planner and optimizer (APO)
module of the SAP/R3 system. Engines were delivered to In order to gain insight into the difference between
the customer (the assembly plants) approximately in the heijunka and JIS production we constructed a 10-work-
order in which they were produced. station simulation model using Extend version 6
ARTICLE IN PRESS
504 A. Hüttmeir et al. / Int. J. Production Economics 118 (2009) 501–507

(see www.imaginethatinc.com). All workstations have a cycle times of 50, 51 and 53 s. Moving from an interarrival
fixed processing time of 50 s with the exception of rate of 50–51 s (which equals an increase of capacity of
workstations 4 and 8, whose processing time is either 2%) improves the situation already significantly. The small
40 or 60 s depending on whether an engine’s work content peaks can be compensated for by small buffer areas on the
is low or high. Under heijunka, high and low work content assembly line to permit engines to wait without going out
(difficult and easy) engines are alternated. Under JIS, the of sequence, assuming that the wait was not too long.
sequence of difficult and easy engines is randomly Large engines go with large cars, so we also considered
determined. In both cases, 50% of engines are modeled the possibility that the engine sequence would already
as being difficult, 50% as being easy. Engines arrive to the have an element of smoothing completed by the auto
line at a rate of one engine every 50 s. There is no assembly plant which applies heijunka. This scenario
processing time or interarrival time variability beyond the suggested a possibility to us of a sequence that would be
difference in processing times for difficult and easy largely heijunka, but would allow a few engines to be out
engines at workstations 4 and 8. Each model run of sequence. In other words, the production sequence
represents 1 million seconds (approximately 4 weeks of would follow heijunka discipline for the majority of
production time at the BMW engine plant), requiring engines, but a small number of outliers would be
around 8 h on an average personal computer. The tolerated. We modeled this sequence as a series of 20
structure of the model is shown in Fig. 1 below. All results engines following heijunka rules (easy-difficult-easy-
retrieved from this simulation are based on the specific difficult), followed by a break in the heijunka sequence
setup described above. They give insight into the behavior in the form of a second difficult engine, before restarting
of the system, but cannot be considered as generalizable. the heijunka sequence. The interarrival rate was set to
Model results comparing the buildup of work in 51 s. The results, shown in Fig. 2c, suggest that partial
process inventory (WIP) under JIS and heijunka sequen- heijunka allows some reduction of the required capacity
cing are shown in Fig. 2a. The system has sufficient buffer compared to the 53-s interarrival rate shown in Fig.
capacity for the heijunka scenario, but becomes unstable 2b; in Fig. 2b an increase of interarrival time by 3 s still
with the extra variability of JIS (i.e. WIP continually leads to the formation of some WIP peaks whereas partial
increases over time). heijunka as illustrated in Fig. 2c allows an increase of
One approach to regaining system stability under JIS is interarrival time by only 1 s to fully eliminate WIP
to extend the cycle time by increasing the number of buildup.
seconds between engines. Increasing the interarrival rate If BMW wants to take advantage of strategic variability
to 54 s completely eliminates the inventory buildup. This it has to eliminate re-work under both heijunka and JIS.
represents a loss of capacity on the order of 7%, with Fig. 3 illustrates how re-work disturbs the production and
production during an 8-h shift decreasing from 576 to 533 how inventory starts to build up. In this model we assume
engines. The choice between JIS and heijunka does not that an engine has a probability of 1% at every station to
have to be completely binary, however, and can instead be get re-worked and 99% of going through. At an interarrival
seen as a continuum between these two extremes. The rate of 50 s, the system is unstable and WIP becomes
actual sequencing of orders can be any compromise infinite under both heijunka and JIS because of the extra
between the required delivery sequence and a sequence time required for rework. Increasing the interarrival time
that makes the schedule as level as possible. by two seconds (i.e., a capacity buffer of 4%) is enough to
The choice between heijunka and JIS implies a cost stabilize the system and bring WIP under control; or in
trade-off: as we have just observed, the responsiveness other words, 1% rework requires 4% additional capacity to
and flexibility of JIS comes at the price of a capacity buffer avoid building up of WIP.
(i.e. higher production costs), while pure heijunka requires How do these insights compare to the situation at
building responsiveness at the end of an inflexible BMW’s engine plant? We began by collecting one year
production line (i.e. resorting costs). Therefore, we used (representing millions of records) of production data from
the simulation model to explore some intermediate points the SAP system to determine how many engines changed
on the heijunka-JIS continuum. place, and by how many positions. The production data
We began by considering how much inventory buildup indicated that well over half of the engines changed
could be tolerated in the line, increasing the interarrival places, but the majority remained within 60 places of
time enough to stop the instability, but less than required their original position. A scant 15% of engines changed
to eliminate WIP buildup. Fig. 2b shows the results with more than 60 places. Only a very small percentage of

Fig. 1. Trading off between heijunka and JIS.


ARTICLE IN PRESS
A. Hüttmeir et al. / Int. J. Production Economics 118 (2009) 501–507 505

80
70
60
WIP [engines]

50
JIS: 50 sec
40
30
20 Heijunka: 50 sec

10
0
0 100 200 300 400 500 600 700 800
Time [in thousands of seconds]

80
50 sec
70
60
WIP [engines]

50
40
30 51 sec
53 sec
20
10
0
0 100 200 300 400 500 600 700 800
Time [in thousands of seconds]

450
400
350
WIP [engines]

300
250 50 sec

200
150
51 sec
100
50
0
0 100 200 300 400 500 600 700 800
Time [in thousands of seconds]

Fig. 2. WIP for different scenarios of JIS, heijunka and partial heijunka: (a) JIS versus heijunka with a cycle time of 50 s, (b) JIS with cycle times of 50, 51
and 53 s, and (c) partial heijunka with cycle times of 50, 51 and 53 s.

engines changed place in the original sequence by more 2003). One could argue that such an investment in a
than 200 places. Furthermore, it was relatively easy to capacity buffer could be warranted given the expense
predict which engines would be delayed by many places: of the warehouse and sorting, and could even be
Most were special engines that required substantial extra seen as a strategic investment as it would permit a
testing procedures. The only engines that unexpectedly higher level of flexibility and customer responsiveness.
faced a delay of several places were those that had quality Nevertheless, such thinking was not readily embraced
problems requiring rework. within the company due to a commitment to leanness and
Initial data analysis indicated that pure JIS would the fact that utilization was a primary performance
require a capacity buffer on the order of 10% (Hüttmeir, measure.
ARTICLE IN PRESS
506 A. Hüttmeir et al. / Int. J. Production Economics 118 (2009) 501–507

450

400

350

300
WIP [engines]

JIS: 50 sec
250

200 Heijunka: 50 sec

150

100 Heijunka: 52 sec


JIS: 52 sec
50

0
0 100 200 300 400 500 600 700 800
Time [in thousandsof seconds]

Fig. 3. The impact of rework on WIP for different scenarios of JIS and heijunka. Rework causes WIP to explode for both the JIS and heijunka scenarios.

We then considered the scenario in which the SAP/R3 reduction in rework, production downtime, and material
system parameters were set to be less drastic in smooth- shortages.
ing the production schedule, combined with a smaller
capacity buffer (on the order of 2–3%) and racks on the
line to collect and re-sort engines. The production data 5. Summary and conclusions
suggested that this system would work well for the
approximately 75% of engines changing less than 60 Is it better for a manufacturing plant to use heijunka to
places. The remaining 25%, however, would require maximize its leanness, or to use JIS to maximize its
substantially larger buffers (which we estimated at responsiveness? Our work at BMW indicates that the
around 3–8 h of production). answer may lie somewhere in the middle, with heijunka
Heijunka may naturally propagate in the BMW supply used to smooth out the most extreme production values
chain in that large cars require large engines. A down- with the remainder of production carried out JIS. It is
stream producer following heijunka may well result in important to understand this trade-off, as it gives
substantial smoothing upstream. In such a case, the essential insights into the bigger picture of trading off
approach that we have been discussing that is primarily between leanness and agility. We have blindly accepted
heijunka with space for regular exceptions might work lean production practices—and leanness itself—as good
extremely well. For example, the sequence could be for competitiveness, but it is now time to reach a finer
planned so that an extra difficult engine could be added understanding of these dynamics. Over time, we may find
once per time period (say, once per 20 min). This would that the essential dynamic capabilities for competitive
distribute the disruption but with more flexibility than operations will not come from configurations of routines
pure heijunka, and could be achieved with a much smaller that drive leanness, but from configurations of routines
capacity buffer. that permit essential flexibility in spite of leanness. Also, it
Although the main concern within the BMW engine may be that the combination of leanness and agility will
plant concerning the move from heijunka to JIS was create new incentives to improve process reliability, given
with the reduction in capacity utilization, we observed the lack of tolerance for rework under JIS production.
from the production data that the primary problem
came from rework. Under heijunka operating in combina- References
tion with a postponement strategy, an engine is not
dedicated to a given customer until relatively late in its Abdulmalek, F.A., Rajgopal, J., 2007. Analyzing the benefits of lean
manufacturing and value stream mapping via simulation: a process
production (referred to by BMW as late order assignment).
sector case study. International Journal of Production Economics 107,
Under JIS, however, the dedication of the engine to 223–236.
a customer is expected to occur relatively early in Browning, T.R., Heath, R.D., 2008. Reconceptualizing the effects of lean
on production costs with evidence from the F-22 program. Journal of
production. Once the engine is dedicated, any changes
Operations Management, doi:10.1016/j.jom.2008.03.009, 22pp.
in sequence are much more disturbing. Success with Chase, R.L., 1993. Differing perceptions. The Total Quality Management
JIS will require almost complete elimination of rework. Magazine 5 (2), 19–21.
Such sensitivity to rework generates an interesting Cusumano, M.A., 1994. The limits of ‘‘lean’’. Sloan Management Review 1,
27–32.
historical parallel to the original concept of JIT production, De Smet, R., Gelders, L., 1998. Using simulation to evaluate the
in which removal of buffer inventories requires a radical introduction of a kanban subsystem within an MRP-controlled
ARTICLE IN PRESS
A. Hüttmeir et al. / Int. J. Production Economics 118 (2009) 501–507 507

manufacturing environment. International Journal of Production Pietsch, D., 2002. Das neue Vertriebs- und Produktionssystem der BMW
Economics 56–57, 111–122. Group-Kundenorientierter Vertriebs- und Produktionsprozeß (KOVP).
de Treville, S., Antonakis, J., 2006. Can lean production be intrinsically ZfAW (BMW), 45–51.
motivating? Contextual, configurational, and levels-of-analysis is- Prenninger, J., 2001. Neue Anforderungen an eine Werksübergreifende
sues. Journal of Operations Management 24 (2), 99–123. Logistik—Innovationen im Bereich der ‘‘Logistikkette Motor’’ bei.
Feare, T., 2003. Just-in-time, just-in-sequence. Modern Materials Hand- Gabler Verlag, BMW.
ling. Prince, J., Kay, J., 2003. Combining lean and agile characteristics: creation
Feitzinger, E., Lee, H.L., 1997. Mass customization at Hewlett Pack- of virtual groups by enhanced production flow analysis. International
ard—the power of postponement. Harvard Business Review 75 (1), Journal of Production Economics 85, 305–318.
116–121. Rinehart, J., 1997. After lean production: evolving employment practices in
Fujimoto, T., 1999. The Evolution of a Manufacturing System. Oxford the world auto industry. American Journal of Sociology, 1212–1214.
University Press, Toyota, NY. Sayer, A., 1986. New developments in manufacturing: the just-in-time
Hall, R.W., 1983. Zero Inventories. Irwin, Homewood, IL. system. Capital & Class 10, 43–71.
Hines, P., Holweg, M., Rich, N., 2004. Learning to evolve: a review of Schonberger, R.J., 1982. Japanese Manufacturing Techniques: Nine
contemporary lean thinking. International Journal of Operations & Hidden Lessons in Simplicity. Free Press, New York.
Production Management 24 (10), 994–1011. Selto, F.H., Renner, C.J., Young, S.M., 1995. Assessing the organizational fit
Hopp, W.J., Spearman, M.L., 1996. Factory Physics, first ed. Irwin, Chicago. of a just-in-time manufacturing system: testing selection, interac-
Hüttmeir, A., 2003. Competitive operations and trade-off analysis. tion, and systems models of contingency theory. Accounting,
Master Thesis, University of Lausanne, Switzerland. Organizations & Society 20 (7/8), 665–684.
Kempfer, L.M., 2005. Speeding inline-vehicle sequencing. Material Shah, R., Ward, P.T., 2003. Lean manufacturing: context, practice bundles,
Handling Management. and performance. Journal of Operations Management 21, 129–149.
Lambert, D.M., Cooper, M.C., Pagh, J.D., 1998. Supply chain management: Suri, R., 2003. QRM and POLCA: a winning combination for manufactur-
implementation issues and research opportunities. The International ing enterprises in the 21st century. Technical Report, QRM Center,
Journal of Logistics Management 9 (2), 1–20. University of Wisconsin, Madison, WI.
Liker, J.K., Wu, Y., 2000. Japanese automakers, US suppliers and supply- Suri, R., de Treville, S., 1986. Getting from just-in-case to just-in-time:
chain superiority. Sloan Management Review, 81–93. insights from a simple model. Journal of Operations Management 6
McLachlin, R., 1997. Management initiatives and just-in-time manufac- (3), 295–304.
turing. Journal of Operations Management 15, 271–292. Teece, D.J., Pisano, G., Shuen, A., 1997. Dynamic capabilities and strategic
Mishina, K., Takeda, K., 1994. Toyota Manufacturing USA, Inc. Case 9-693- management. Strategic Management Journal 18 (7), 509–533.
019, Harvard Business School, Cambridge, MA, 22pp. Trebilcock, B., 2006. Building a new supply chain. Modern Materials
Monden, Y., 1983. Toyota Production System. Industrial Engineering Handling 61 (1), 32.
and Management Press, Institute of Industrial Engineers, Norcross, Van Hoek, R.I., 2001. The rediscovery of postponement: a literature
GA. review and directions for research. Journal of Operations Manage-
Narasimhan, R., Swink, M.L., Kim, S.W., 2006. Disentangling leanness and ment 19 (2), 161–184.
agility: an empirical investigation. Journal of Operations Manage- Womack, J.P., Jones, D.T., Roos, D., 1990. The Machine that Changed the
ment 24 (5), 440–457. World. Harper Collins Publishers, New York.
Narasimhan, R., Talluri, S., Das, A., 2004. Exploring flexibility and Yao, A.C., Carlson, J.G.H., 2003. Agility and mixed-model furniture
execution competencies of manufacturing firms. Journal of Opera- production. International Journal of Production Economics 81–82,
tions Management 22 (1), 91–106. 95–102.

You might also like