Production Planning and Control Model of Technology Migration for
DRAM Industry
Ying-Mei Tu
Department of Industrial Management, Chung Hua University, Hsin Chu, Taiwan, R.O.C.
Keywords: Technology Migration, DRAM Industry, X-factor, Production Planning and Control.
Abstract: Due to product life cycle has been shortened rapidly, it forces the product generation and technology should
be enhanced quickly. When technology generation change occurred, DRAM manufacturers always used the
past experiences to handle the change process. However, the issues are totally different and it made the
companies suffered many difficulties. In this work, a production planning and control model is developed.
The production planning focuses on CCR (Capacity Constraint Resources) to define the complete wafer
release schedule and apply X-factor to schedule the production processes during the migration period.
Regarding to the shop floor control, there are two control mechanisms to control and monitor the migration
process, real time control and predicting control. WIP status is the important factor to decide whether the
production planner needs to launch the rescheduling module or not in the real time control portion. Besides,
a foresee function is performed by predicting control portion which firing the rescheduling module by the
bias between the loading and capacity curves.
1 INTRODUCTION
DRAM industry is a capital intensive, high-tech
industry with complex processes. Nevertheless,
product generation and technology had been quickly
enhanced due to short product life cycle. When new
technology emerges, it reveals a lower cost and more
effective operation model (Cainarca, 1989).
Simultaneously, it also means the current
competitive advantages of company will be
jeopardized (Hastings, 1994). Under this
circumstance, manufactures have to launch new
technology and retrofit generation equipment to
meet the market demand and reduce manufacturing
cost. Chou et al. (2007) pointed out the technology
life cycle of semiconductor manufacturing usually
won’t be over three years. Therefore, the
semiconductor manufacturers always face the
dilemma of new technology migration. Generally,
the major competition factor of DRAM industry is
the manufacturing cost. That is why the frequency of
technology migration is higher than foundries.
When migration occurred, DRAM manufactures
always used the past experiences to handle the
migration. However, the issues are totally different
that caused the manufactures suffered many
unknown difficulties. Generally, the production
planning of technology migration should take the
planning result of high-level strategy into account,
such as the start time of migration, output target of
new technology…etc., to set the migration tempo
and capacity switching schedule. Nevertheless, the
uncertainties and dynamic factors of shop floor (ex:
machine breakdown, schedule delay for new
generation equipment or equipment retrofit…etc.)
can not be taken into consideration in the high-level
strategy. Besides, the high-level decision is based on
the prediction of technology roadmap, there will be
some changes and biases between the setup of high-
level strategy and the execution of technology
migration process. In order to guarantee a smooth
and successful migration process, a robust and
effective production planning and control model of
shop floor for technology migration is very
important.
Many researches have proposed some methods
for production planning and shop floor control of
semiconductor manufacturing. Regarding to the
production planning, queuing theory, linear
programming and mean value analysis are usually
applied to estimate the capacity requirement of
workstations and wafer release quantity (Iwata et al.,
2003); (Walid and Gharbi, 2002); (Chou and You,
2001). Nevertheless, the system uncertainty and the
599
Tu Y..
Production Planning and Control Model of Technology Migration for DRAM Industry.
DOI: 10.5220/0003995005990603
In Proceedings of the 9th International Conference on Informatics in Control, Automation and Robotics (OMDM-2012), pages 599-603
ISBN: 978-989-8565-22-8
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
risk of investment are not taken into account.
Besides, many researches focused on release policy
(Glassey, and Resende, 1998a, 1998b); (Wein,
1988), (Lou, 1989a, 1989b); (Spearman, 1989,
1992); (Bowman, 2002); (Hung and Leachman,
1996). Either opened-loop or closed loop policy is
based on the normal production situation and does
not think of the events of products generation
changes, equipment retrofit and new equipment
move-in. According to the shop floor control, many
dispatching rules were developed to fulfil the
purpose of higher production performance (Dabbas
and Flowler, 2003); (Lee and Kim, 2011); (Louw
and Page, 2004); (Hsieh and Hou, 2006); (Hung and
Chang, 2002); (Uzsoy et al., 1992). Nevertheless,
the issues of process migration were not considered
either in the release policy nor shop floor control
rule. In general, the production system will be more
complicated during the technology migration period,
such as the instability of products mix, the changes
of capacity. Therefore, the proposed methods won’t
be satisfied the requirements. Moreover, the
experiences of semiconductor management showed
that the production management will be extremely
complicated when there are over three generation
products produced in the same time. System
performance will be difficult to keep in such a
circumstance. Hence, an efficient and effective
planning and shop floor control model for a varied
system can not only solve the technology migration
issues but also be applied to the foundry with
multiple generation products.
This paper investigates the technology migration
of DRAM industry from manufacturing point of
view. In this work, a production planning and
control model of technology migration was
developed. There are two portions in this model
including production planning and shop floor
control. The production planning focused on CCR to
define the complete wafer release schedule and
applying X-factor to schedule the production
processes and equipment retrofit during the
transition period. Regarding to the shop floor
control, there are two control mechanisms to control
and monitor the migration process, which are real
time control and predicting control module.
2 PRODUCTION PLANNING
MODULE
As mentioned above that the migration process has
to fulfil the target of high-level strategy. The major
decision factors of high-level strategy include the
fluctuation of future demand, technology
development and company financial situation. The
complication and variation of production system are
difficult to take into account in the strategy level.
Therefore, a robust planning and control model not
only can help to a successful migration process but
also to find out various migration problems in
advance. In production planning module, the major
target is to transfer the output targets of new
generational products to execution plan. The plan
includes the wafer release plan of new/old
generational products, the release plan of new
generational equipment and equipment retrofit plan.
Generally, the placement of new/old generational
products will be progressed step by step. Hence, the
migration period is divided into several time periods
for planning. Furthermore, X-factor is applied to the
scheduling process. The following is the procedure
of production planning.
Step 1. Set up the time unit
It can be defined as a day, three hours…etc.
Step 2. Plan wafer start schedule
In this step, the wafer start schedule of new/old
generational products should be planned by referring
the output target of new generational products.
Generally, top management will hope to keep the
total output of factory as before. However, the
manufacturing complexity of new generational
products may be higher than old one and it will
result to the total output decreasing. Therefore, the
total output during migration period should be
planned in this step. The sub-steps are as follows.
1) Identify Capacity Constraint Resources (CCR)
Generally, the CCR will be only one of equipment in
a factory. However, due to the heavy investment of
equipment, several workstations are highly utilized.
If we assign the equipment with the highest
utilization to be the CCR and based on this CCR to
make all plans, the issue of bottleneck shifting will
be occurred. Hence, multiple CCRs are suggested
and can be the equipment with the utilization rate
being higher than the predefined value.
2) Calculate capacity consumption rate of CCRs by
new and old generational products
Because the new/old generational products will be
processed by the same equipment, the capacity
consumption rate should be decided for the
calculation of migration plan. The equations are
show as follows.
M
M
N
O
M
C
C
CR =
(1)
ICINCO2012-9thInternationalConferenceonInformaticsinControl,AutomationandRobotics
600
∑∑
==
=
x
q
n
i
MiNNN
qqM
PTRC
11
(2)
∑∑
==
=
y
k
m
j
MjOOO
kkM
PTRC
11
(3)
=
=
x
k
k
q
N
q
R
1
λ
λ
(4)
=
=
y
l
l
k
O
k
R
1
λ
λ
(5)
Where
M
CR
:
The capacity consumption rate of new to old
generational product in CCR M
M
N
C
:
The average required capacity for the new
generation product in CCR M
M
O
C
:
The average required capacity for the old
generation product in CCR M
q
N
R
:
The ratio of product q in new generational
products
k
O
R
:
The ratio of product k in old generational
products
p
λ
:
Arrival rate of product p
MiN
q
PT
:
The ith processing time of product q in CCR M
MjO
k
PT
:
The jth processing time of product k in CCR M
3) Compute the reducing quantity of old
generational products.
Based on the capacity consumption rate, the
reducing quantity of old generational products can
be calculated by the following equation.
MN
CRQQ
×
=Δ
O
(6)
Where
O
QΔ
:
The reducing quantity of old generational products
N
Q
:
The required quantity of new generational products
4) Release new and old generational products by
uniform distribution.
Step 3. Apply X-factor to pre-schedule all
production processes.
In this step, the concept of X-factor will be
applied to schedule all production process including
WIP and new release products, and calculate the
loading of CCRs in all time periods. The definition
of X-factor is as equation (7) and it has to be defined
by new/old generational products and equipment.
Regarding to the detailed calculation equations of
workstations for the wafer fabrication, please refer
to Tu et al. (2009).
pm
pm
m
RPT
CT
FactorX =
(7)
Where
mp
CT
: The cycle time of product p in equipment m
mp
RPT
: The raw processing time of product p in
equipment m
Step 4. Plan equipment retrofit schedule.
In order to fulfill the manufacturing requirements of
new generational products, some kinds of equipment
should be retrofitted. During the equipment
refurbishment period, it cannot work and the
capacity will lose. Furthermore, it may hurt the
factory throughput if the loss belongs to the
bottleneck machine. In this step, the equipment
loading from schedule result of step 3 has to apply to
compare to the provided capacity. The equipment
retrofit can be scheduled when the loading is under
capacity.
Step 5. Come back to step 2 and recalculate X-
factor when the product mix of new/old generational
products changed.
3 SHOP FLOOR CONTROL
MODULE
Regarding to the shop floor control, there are two
control mechanisms to control and monitor the
migration process including real time control and
predicting control module.
3.1 Real Time Control Module
Generally, WIP status is an important and sufficient
information to reflect the production situation. If
WIP level in front of workstation is too high, it
reveals the capacity of this workstation is
insufficient or there is something wrong in
dispatching. Contrarily, low WIP level indicates
some problems occurred in upstream workstations or
wrong dispatching. Both situations cannot achieve
the target of plan. In the real time control module,
actual WIP level is taken as an indicator to judge the
rescheduling mechanism should be launched or not.
The buffer management concept of TOC is applied
to control CCRs. Besides, the queuing theory and
the capability of factory management are used to
ProductionPlanningandControlModelofTechnologyMigrationforDRAMIndustry
601
define the high and low control limits. When WIP
level is over these limits, the response module will
be triggered. The control limits are defined as the
following equations.
((0) )(1)
jj j
HL P W EW
λ
α
>× ×+
(8)
((0) )(1)
jj j
LL P W EW
λ
α
>× ×
(9)
(0)
j
WEW
η
>=×
(10)
22
2(1 )/( )
j
jajsj
mCC
ηρ
=− +
(11)
211
22
()
2(1)
j
m
aj sj j j
j
jj
CC
EW
m
τρ
ρ
+−
+
(12)
Where
j
λ
:
Arrival rate of workstation j
j
m
:
Parameter of capability of factory management0~1
j
EW
:
Expected waiting time of workstation j
sj
C
:
Number of machines for workstation j
j
ρ
:
Utilization rate of workstation j
2
aj
C
:
Squared coefficients of variation (SCV) of inter-arrival
time of workstation j
2
sj
C
:
SCV of service time of workstation j
3.2 Predicting Control Module
As mentioned above, the real time control module is
based on current shop floor information to diagnose
the plan can be achieved or not. However, current
shop floor status is the execution result. If the result
is far away from the plan, the most possible action is
to revise the plan. It seems behinds manager’s
expectation. Therefore, a predicting control function
is needed in the shop floor control module. In
predicting control module, a foresee function will be
performed which will trigger the response module
when the bias between loading and capacity curves
is over the predefined deviation tolerance (DT). The
major task of the foresee function is to predict the
production situation in the future. The deterministic
simulation is applied to this function. Based on the
deterministic simulation, the loading curves of CCRs
by time can be defined. As to the capacity curves of
CCRs, they can be derived from current capacity,
the move-in schedule of new generational equipment
and equipment retrofit plan. Fig. 3 is an example of
equipment capacity curve and loading curve.
Besides, as everyone knows that the accuracy of
prediction will decrease as the time increasing.
Therefore, the time factor should be considered into
the bias tolerance. The equation for defining the
deviation tolerance is as follows.
n
DT n C
β
×
(13)
Where
β
:
Parameter of capacity deviation
n
:
The time period
n
C :
The capacity of period n
4 CONCLUSIONS
Technology migration is imperative for
semiconductor manufacturing, particularly for
DARM industry. The migration of technology will
result in dramatic decreases in manufacturing cost
and significantly increases competitive advantage.
Nonetheless, how to guarantee a smooth and
successful migration is very crucial. Therefore, the
solution of the production planning and control of
technology migration for DRAM industry is
proposed in this work. There are two major modules
developed in this model, one is production planning
and the other is shop floor control. The production
planning module is based on the output plan of new
generational products to come out the wafer start
schedule of new/old generational products,
equipment retrofit schedule and move-in schedule of
new generational equipment. The shop floor control
module includes three sub-modules, real time
control, predicting control and response module.
Through the shop floor control module, the
execution can be monitored and controlled to meet
the plan target.
Regarding to the future works, the response
module can be enhanced. An ideal response module
should provide the detailed action items instead of
direction when the abnormal situation occurred.
Therefore, an intelligent system should be set up in
this module.
ACKNOWLEDGEMENTS
The author would like to thank the National Science
Council of the Republic of China for financially
supporting this research under Contract No.NSC 100-
2628-E-216-002-MY2.
ICINCO2012-9thInternationalConferenceonInformaticsinControl,AutomationandRobotics
602
REFERENCES
Bowman, R. A., (2002). Job Release Control Using a
Cyclic Schedule, Production and Operations
management, 11(2), 274-286.
Cainarca C., (1989). Dynamic Game Results of the
Acquisition of New Technology, Operations
Research, 37(3), 410-425.
Chou, Y. C., Cheng, C. T., Yang F. C. and Liang, Y. Y.,
2007. Evaluating alternative capacity strategies in
semiconductor manufacturing under uncertain demand
and price scenarios, International Journal of
Production Economics, 105(2), 591-606.
Chou Y. C. and You, R. C., (2001). A Resource Portfolio
Planning Methodology for Semiconductor Wafer
Manufacturing, International Journal of Advanced
Manufacturing Technology, 18(1), 12-19.
Dabbas, R. M., & Fowler, J. W., (2003). A new
scheduling approach using combined dispatching
criteria in wafer fabs. IEEE Transactions on
Semiconductor Manufacturing, 16(3), 501–510.
Glassey, C. R. and Resende, M. G. C. (1988a), ‘Closed-
loop job release control for VLSI circuit
manufacturing’, IEEE Transactions on Semiconductor
Manufacturing 1/1, 36-46.
Glassey, C. R. and Resende, M. G. C., (1988b). ‘A
scheduling rule for job shop release in semiconductor
fabrication’, Operations Research Letters 7/5, 213-
217.
Hastings, J., (1994). AmCoEx Five Year Comparsion of
Used Computer Prices, Computer Currents, 9, 20-24.
Hsieh, S. and Hou, K. C., (2006). ‘Production-flow-value-
based job dispatching method for semiconductor
manufacturing ’, International Journal of Advanced
Manufacturing Technology, 30, 727–737.
Hung, Y. F. and Chang, C. B. (2002). Dispatching Rules
Using Flow Time Predictions For Semiconductor
Wafer Fabrications, Journal of the Chinese Institute of
Industrial Engineers, 19(1), 61-74.
Hung, Y. F. and Leachman, R. C., (1996). A production
planning methodology for semiconductor
manufacturing based on iterative simulation and linear
programming calculations, IEEE Transactions on
Semiconductor Manufacturing, 9(2), 257-269.
Iwata, Y., Taji, K. and Tamura, H., (2003). Multi-
objective capacity planning for agile semiconductor
manufacturing, Production Planning & Control,
14(3), 244-254.
Lee, Young Hoon and Kim, Jeong Woo (2011), Daily
stepper scheduling rule in the semiconductor
manufacturing for MTO products, International
Journal of Advanced Manufacturing Technology
(54):323–336
Lou, S. X. C., (1989a), ‘Optimal control rules for
scheduling job shops’, Annals of Operations Research
17, 233-248.
Lou, S. X. C., and Kager, P. W., (1989b), ‘A robust
production control policy for VLSI wafer fabrication’,
IEEE Transactions on Semiconductor Manufacturing
2/4, 159-164.
Louw, L. and Page, D. C., (2004). Queuing network
analysis approach for estimating the size of the time
buffers in Theory of Constraints-controlled production
systems, International Journal of Production
Research, 42(6), 1207-1226.
Spearman, M. L., Woodruff, D. L., and Hopp, W. J.,
(1989), ‘CONWIP: a pull alternative to kanban’,
International Journal of Production Research 28/5,
879-894.
Spearman, M. L., and Zazanis, M. A., (1992), ‘Push and
pull production systems: issues and comparisons’,
Operational Research 40/3, 521-532.
Tu, Ying-Mei, Lu, Chun-Wei and Chang, Sheng-Hung,
(2009). “Model To Determine A General X-Factor
Contribution And Apply To Cycle Time Improvement
For Wafer Fabrication”, Int. J. Services Operations
and Informatics, Vol 4., No. 3, p.272-291
Uzsoy, R., Lee, C. Y. and Martin-Vega, L. A., (1992). A
Review of Production Planning and Scheduling
Models in Semiconductor Industry Part I: System
Characteristics, Performance Evaluation and
Production Planning, IIE Transactions, 24(4), 47-60.
Wein, L. M., (1988). ‘Scheduling semiconductor wafer
fabrication’, IEEE Transactions on Semiconductor
Manufacturing 1/3, 115-130.
ProductionPlanningandControlModelofTechnologyMigrationforDRAMIndustry
603