You are on page 1of 4

Effect of Confdence Interval on Bottleneck Identifcation via Simulation

C. Kasemset', V. Kachitvichyanukuf
I
Department ofIndustrial Engineering, Chiang Mai University, Chiang Mai, Thailand
2
Industrial and Manufacturing Engineering, Asian Institute of Technology, Pathumthani, Thailand
(ckjane@hotmail.com , voratas@ait.ac.th)
Abstract - This
p
a
p
er
p
resents the a
pp
lication of
simulation-based
p
rocedure to identify bottleneck in a job­
sho
p
environment. Ten jobs ten machines job-sho
p p
roblem
(lOxlO JSP) is used to test the simulation-based
p
rocedure
for identifing the system bottleneck. The result from the
case shows that the choice of confdence interval level (CL)
used in through
p
ut mean com
p
arison has efect on the
bottleneck selection. Two scenarios are
p
resented with two
different CL of through
p
ut mean, 75% CL and 90% CL.
The result from the ex
p
eriment shows that when % CL is
increased, the judgment of bottleneck identifcation can be
changed. Thus, it is necessary to use a
pp
ro
p
riate number of
re
p
lications to match with the required %CL used in the
simulation. In general, more re
p
lications are necessary when
higher
p
ercentage of confdence interval level is required.
Keyords - Simulation, Bottleneck identifcation,
Theory of Constraints (TOC), Confdence interval level (CL)
I. INTRODUCTION
Theory of Constraints (TOC), developed by Goldratt
and Cox in [1], concentates on how to manage the
constraint of the system, the bottleneck resource or
capacity constraint resource (CCR), to drive more income
and for company to survive in the real world business.
In order to implement this policy, there are fve steps
introduced in [1]; (i) bottleneck identifcation, (ii)
bottleneck exploitation, (iii) bottleneck subordination, (iv)
bottleneck elevation ad (v) retur to the frst step but
prevent inertia from being the next constraint.
The bottleneck identifcation is the key to TOC
implementation because the bottleneck is the only
machine tat can limit the overall system performance. If
the tue system bottleneck ca be identifed, the
succeeding steps to improve system performance can
defnitely work.
To locate bottlenecks, many researchers prposed the
bottleneck identifcation methods that utilized only static
data. In [2], [3] and [4], throughput rate was used to
identit the bottleneck. In [5], the expected load was
considered and compared with available load to fmd the
CCR.
Under job-shop environment, it is difcult to identit
the real system bottleneck because different products have
their own routes and usually share machines. In the case
that the bottleneck machine is located behind another
machine that operates at a slow rate, it may be difcult to
978-1-4244-8503-111 0/$26.00 ©2010 IEEE
identit the bottleneck simply fom te utilization or te
number in queue.
H [6], simulation technique was applied to collect
data and used tese data to calculate two factors, the
number of jobs in queue for a resource and resource
criticality factor. These two factors are used to identit
bottlenecks in a job-shop. In [7], [8], [9] and [10], bot
static data ad data collected fom simulation were used
to classit bottleneck candidates ad identifed te real
system bottlenecks via simulation.
The objective of this study is to present the
application of simulation-based procedure proposed in [7]
when it is applied to ten jobs ten machines job-shop
problem ( l Oxl 0 JSP) in the process of bottleneck
identifcation. The test, illustates the effect on the
bottleneck selection when the confdence interal level
(CL) used in throughput mean comparison is changed.
The organization of this paper is as follows. The
based method, simulation-based procedure for bottleneck
identifcation proposed in [1] is introduced in Section 2.
The experiment on changing confdence interval level in
order to identit the system bottleneck is presented in
Section 3. Conclusion and Discussion are addressed in
Section 4.
II. BASED METHOD:
SIMULATION-BASED PROCEDURE FOR
BOTTLENECK IENTIFICATION
Simulation-based procedue for bottleneck
identifcation proposed in [7] is applied with l Ox ¡0 JSP in
this study. The following sections are te explanation of
the procedure.
A. Procedure Assumptions
This procedure is applied based on te following
assumptions;
1) The target demand exceeds the system capacity.
2) All operation times and set up times ae
exponentially distributed.
1592
D. Adopted Procedure
H order to identit the real system bottleneck, three
factors are introduced in [7];
1) Machine/process utilization: these data can be
directly collected from the simulation model. Machines or
Proceedings of the 2010 IEEE IEEM
processes having hig utilizations ae selected to be
bottleneck candidates.
2) Utilization factor (p): From [11], the utilization
factor, based on queuing theory, M/MIl /K, is calculated
from p = A / þ where A is the average number of units
added to the time bufer per unit time and þ is the average
service rate unit per unit time. Both A and þ can be
obtained fom simulation model. Bottleneck cadidates
are selected fom machines or processes with high
utilization factors.
3) Bottleneck Rate (Rb): Rb is the rate (pats per uit
time or jobs per unit time) of the workstation having
highest long-term utilization from Basic Factory Dynamic
from [12]. Machines or processes with Rb are considered
as bottleneck candidates.
The three factors above can be used to identif
potential bottleneck candidates.
Among all bottleneck candidates, the station that
meets following three criteria is identifed to be real te
system bottleneck;
1) High value of the machine/process utilization
2) High value of the process utilization factor
3) Low value of the product bottleneck rate
Then, simulation is re-run again to confrm that (i) the
true bottleneck is selected and (ii) no other bottlenecks
remain in the system (in case of multiple bottlenecks).
Real bottlenecks are identifed when the simulation
yields improving in system throughput by relaxing that
machine constraints.
III. THE APPLICATION Í JSP CASE
By applying the method proposed in [13] to evaluate
the performance measure of the system, the mean
confdence interval of system throughput at 75% CL is
constructed and compaed with based situation. The
improvement in the system is observed when no overlap
of the CL's between the CL from the based scenario ad
those fom the test cases (curently used 30 replications).
In tis paper, 10xlO JSP is adopted. The detail of this
problem can be found in [14].
Applying the simulation-based procedure fom [7],
the three factors, the utilization of machine, the utilization
factor or p and the bottleneck rate or Rb are show in
Table I.
From this problem, machine number 7 meets all three
criteria addressed previously to qualif as the system
bottleneck. Then, the simulation is used to confrm that
the true bottleneck is selected and no other bottlenecks
that still remain in the system (in case of multiple
_Cwwa|C.I.0man T.mt
Obsewatkn lntewðh
bottleneck). The result fom te simulation is shown in
Fig. l .
TABLE Ï
I°'ÍPÅP bÚÍPÅÍLP ÅLLÍ PPÍ PPPÍÀZLÍ ÍPÅP
Machine no.
% K
º
Utilization
p
(part/min)
0 40.35 1.052 0.111 (P.5, P. lO)
1 41.68 0.940 0.111 (P.6)
2 51.90 1.058 0.111 (p.1 aad9)
3 41.42 0.946 0.111 (P.5 and 6)
4 45.04 0.977 0.111 (P.4)
5 48.37 1.109
0.111
(P.l, 7 and 10)
6 43.74 1.100 0.111 (P.2)
7 50.51 1.147
0.125 (p.1) aad
0.111 (.6)
8 47.60 1.047
9 33.76 1.094 0.111 (P.8)
Note: 1) Bold letters mean that one of thee cntena ¡S met H order to
identif bottleneck cadidate.
2) Example for machine no. 3 K
º
0.111 (P.5 and 6) means for
product 5 ad product 6 bottleneck rate is 0.111.
From Fig. 1, the result shows that when increasing the
seventh machine's capacity in the test scenario, te
system throughput mean cannot be improved when
comparing at 75% CL (There is some overlapped area of
te confdence interval). It means that oter bottlenecks
may exist in the system. Thus, the simulation should be
re-run again to fnd more system bottleneck candidates.
By relaxing the capacity of machine number 7, the
three factors used to identif the bottleneck are collected
and compared again (See Table II)
TABLE ÏÏ
2"ºÍPÅP 1bLN bÚÍPÅÍLP ÅLLÍ PPÍ PPPÍÀZLÍ ÍPÅP
Machine no.
% K
º
Utilization
p
(part/min)
0 45.11 1.067 0.111 (P.5, P.1 0)
1 43.20 0.954 0.111 (P.6)
2 52.52 0.964 0.111 (p.1 aad9)
3 43.41 0.997 0.111 (P.5 and 6)
4 46.81 1.067 0.111 (P.4)
5 47.89 1.132
0.111 (P.1, 7 ad
10)
6 40.36 1.056 0.111 (P.2)
7 25.00 1.046
8 47.14 1.040
9 38.08 1.206 0.111 (P.8)
Note: 1) Bold letters mean that one of three cntena ¡S met H order to
identif bottleneck candidate.
2) Example for machine no. 3 K
º
0.111 (P.5 and 6) means for
product 5 and product 6 bottleneck rate is 0.111.
BuscdIusv
-0
TotaIÆ 00 -----------

´
¡

'
¸

¡
���

·

¸
)

¡
------ '
)1
1est Sccti:u1o: ìI tnc ¯

TotaI²
-
·
)) -----------

)1
��
1

)

'
+0

)

1
---
-
'
¯¨

Fig. 1. 75% CL comparison of througput mean for 1 ¯ test
1593
Proceedings of the 2010 IEEE IEEM
From Table II, there is no machine tat meets all
three criteria to qualif as the system bottleneck.
Machines number 2 and 9 are identifed to be the
bottleneck candidates. The simulation test is carried out
to fmd the second system bottleneck. H this case, two
scenaios are set up. H each scenario, each bottleneck
candidate capacity is added in addition to the increase
of the capacity of machine number 7.
A. First Scenario: Apply 75% CL
The result in Fig. 2 shows that the throughput of
scenarios 2 and 3 show signifcant improvements over
the based case. Thus, machines number 2 and 9 are also
identifed as system bottlenecks. Thus, for this scenario,
the conclusion is that there are thee bottleneck
machines: machines number 2, 7 and 9.
D. Second Scenario: Apply 90% CL
In this scenario, 90% CL is used instead of 75%
CL. When te percentage of CL is increased, the
decision of bottleneck identifcation can be changed as
in Figure 3.
From this study, with 10xl0 JSP, when % CL is
increased to 90%, only machine number 7 and machine
number 9 are identifed to be the system bottleneck (
when 75% CL is adopted, there are three bottleneck
machines (machine number 2, 7 and 9)).
IV. CONCLUSION AND DISCUSSION
This paper aims to show the application of the
simulation-based procedure for bottleneck identifcation
in job-shop problem. 10xlO JSP is used as a test
_CmwolC| Omoo 1dt
0bæatk InteaIs
problem. The efect of changing confdence interval
level when this procedure is applied is observed as
when % CL used to compare system throughput mean is
increased, the judgment of bottleneck identifcation may
be changed.
In the based method, te simulation test is conducted
with the based scenario at 75% CL based on 30
replications. In case of small number of runs with the
high %CL, the different in throughput mean sometimes
canot be clearly observed so te bottleneck may not be
corectly identifed. Thus, te selected %CL should
match with the number of rs used. Where higher
%CL is required, more replications are necessary.
However, this decision on bottleneck identifcation
has efect on the bottleneck exploitation and
subordination phase because te schedule must be
considered frst at the bottlenecks. Although, tis
situation occurs, the fnal (5
t
) step of TOC
implementation can be applied to handle tis issue
because when the TOC system is adopted, the system
performance must be evaluated to confrm that there is
no hidden bottleneck remaining in the system. The 5
th
step of TOC implementation works as iterative
improvement step that makes the system under TOC
become self-adjusted when the system changes in the
bottleneck location.
Furthermore, the advantage of this procedure when
it is applied with large-scale size problems is to reduce
the number of potential candidates by considering three
criteria as mention earlier. Thus, the number of
simulation test can be reduced in order to identif the
system bottleneck.
B.tsc|t .tsc
z . .003
IotaIV +ºº ------------
z

-

i
.

-e

ei
��
z

s
.

ee

i
-----¬
--
eei
Scvnm1o Î. Adtmtg mc¯ otuy
z �eei
IotaI1P see----------- � ------- eei

·
s -eei • �ei
Scvnm1o 2: Addatg mc¯ ¬ mc2
� "
IotaIV see------------- �� �-¬+-·eei
z sz+ ei i z+eei
Scvnm1o 3: Adtm:g mc¯ ¬ mc9
\ i· �·"ei J
IotaI1P . --eeii-------� --eei
'
.s --

i --eei
Fig. 2. 75%CL comparison of througput mean for 2nd test
_Oo|C.I.Omoo D.ot _
0bænatknInteab
��IP::l 1sr

eei-----

----��
'

·



e

e

�----- eei
z i--·eei z.ss--eei
Scmtm1o Í. Adtmtg mc 7 otuy
/03
Iota|IP see -----------
z

¸ ¸
e

e
,
i

«
¸¸
e

e
,
---
°°¯
eei
Scrnm1o 2: Addatg mc ¯ * mc2
J 0 -e0J
Iota|V see ------------ � �� ----eei
z s---eei
V
i
«
--ºei
Scrnm1o 3: Adtm:g mc 7 * mc9
� '
Iota|IP ;. ·ee ----- �
-4•.eei
zs
.
e

.is
.
eei
Fig. 3. 90OoCL comparison of througput mean for 2n
d
Test
1594
Proceedings of the 2010 IEEE IEEM
REFERENCES
[1] E. M. Go1dratt and J. Cox, Te Goal: A Process of
Ongoing Improvement, Revised Edition, Nort River
Press, 1986.
[2] R. A. Reid, "Applying the TOC fve-step focusing
process in the service sector", Managing Service Qualit,
vol. 17, no. 2, pp. 209-234, 2007.
[3] S. Taj and Í. Bero, "Application of constrained
maagement and lea manufacturing in developing best
practices for productivity improvement in a auto­
assembly plat", Interational Joural of Productivit
and Perormance Management, vol. 55 , no.3/4, pp. 332-
345, 2006.
[4] C. C. Pegels ad C. Watous, "Application of te theory
of constaints to a bottleneck operation in a
maufacturing plat", Journal of Manufacturing
Technolog Management, vol. 16, no. 3, pp. 302-311,
2005.
[5] S. X. Wu, J. S. Morris and T. M. Gordon, "A simulation
Analysis of te efectiveness of Drum-Buffer-Rope
scheduling in friture manufacturing", Computers
Industrial Engineering, vol. 26, noA, pp. 757-764, 1994.
[6] V. B. Gageya, "Resource Constraint Measures in a Dual
Constrained Job Shop", Omega, vol. 22, no. 6, 659-668,
1994.
[7] C. Kasemset and V. Kachitvichyaukul, "Simulation­
based procedue for bottleneck identifcation", in
Proceedings of AsiaSim 2007, Asia Simulation
Conference 2007, Seoul, Korea, Springer, Berlin, pp. 47-
55, 2007.
[8] C. Kasemset and V. Kachitvichyaukul, "Simulation­
based tool for implementing theory of constraints", in
Proceedings of the 6th APEMS Conference, The 6th
APEMS Conference, Manila, Philippines, Article 4426,
2005.
[9] C. Kasemset and V. Kachitvichyaukul, "Simulation­
based procedure for implementing teory of constraints:
Extension for cases with multiple bottlenecks", in
Proceedings of the 9th APIEMS Conference, The 9th
APEMS Conference, Bali, Indonesia, pp. 1811-1819,
2008.
[10] C. Kasemset and V. Kachitvichyanul, "Simulation
Tool for TOC implementation", in Proceedings of
ASIMMOD 2009, ASIMMOD 2009, Bangkok, Thailad,
pp. 86-97, 2009.
[11] Z. D. Radovislsky, "A Quantitative Approach to Estimate
the Size of the Time Bufer in te Theory of Constraints",
Interational Joural Production Economics, Vol. 55,
pp. 113-119, 1998.
[12] W. J. Hopp ad M. Í. Speara, Factor Physic, Second
Edition, McGraw-Hill, Inc., 1999.
[13] C. Kasemset, "Simulation-based Tool for Theory of
Constraints (TOC) Implementation", Master research
study Industrial System Engineering, School of
Advaced Technology, Asian Institute of Technology,
Bagkok, Thailad, 2005.
[14] C. Kasemset, "TOC based Procedure for Job-shop
Scheduling", Doctoral Dissertation, Industial
Engineering and Maagement, School of Engineering
ad Technology, Asian Institute of Technology,
Bagkok, Thailad, 2009.
1595

8) Note: 1) Bold letters mean that one of three cntena IS met III order to identify bottleneck candidate. e .8) Note: 1) Bold letters mean that one of three cntena IS met III order to identify bottleneck candidate. 7 and 10) 0.5.058 0.1 and 9) 0. l .111. P. 2) Example for machine no.11 43. Bottleneck candidates are selected from machines or processes with high utilization factors.111 (P. 61 +003 Fig. Among all bottleneck candidates.Proceedings of the 2010 IEEE IEEM processes having high utilizations are selected to be bottleneck candidates. Applying the simulation-based procedure from [7]. Both A and J.5 and 6) 0. The result from the simulation is shown in Fig.37 43. Then.52 43. Thus. By relaxing the capacity of machine number 7.1 where A is the average number of units added to the time buffer per unit time and J.111 (p. the mean confidence interval of system throughput at 75% CL is constructed and compared with based situation.047 1. 1. +003 8001------------03 ---: 69 -== -:c :. the utilization of machine. The improvement in the system is observed when no overlap of the CL's between the CL from the based scenario and those from the test cases (currently used 30 replications).6) Rb 52.89 40. the three factors.111 (P.111 (P.04 48.5 and 6) means for III.111 (P.76 1. 10xlO JSP is adopted.111 (P. 7 and 10) 0.111 (p.940 1. 75% CL comparison of throughput mean for 1st test 1593 . 3 -:: +O ::: 03 ---':" 07 -:c :---l 4 e+0 2 .20 p 1.4) 0.5 and 6) means for Based Cas!' Totallll Totallll T!'st SCeIl>U10: AddiII2 IDC - 4oo 1----------------: :-+O 3� -:c -== � � 8. TABLE I 1ST DATA SIMULATION TOOL AND ANALYZED DATA % Utilization 40. is calculated from p A / J.90 41.1 is the average service rate unit per unit time.l.60 33.052 0.74 0.5 and 6) 0.2) 1.51 47. 3) Bottleneck Rate (Rb): Rb is the rate (parts per unit time or jobs per unit time) of the workstation having highest long-term utilization from Basic Factory Dynamic from [12].00 47.946 0.111 (P. M/MIl /K. the result shows that when increasing the seventh machine's capacity in the test scenario. simulation is re-run again to confirm that (i) the true bottleneck is selected and (ii) no other bottlenecks remain in the system (in case of multiple bottlenecks).36 25. 3 product 5 and product 6 bottleneck rate is 0.1) and 0. THE APPLICATION IN JSP CASE By applying the method proposed in [13] to evaluate the performance measure of the system. Real bottlenecks are identified when the simulation yields improving in system throughput by relaxing that machine constraints. On Mean Observation Intervals Th.132 1. 0 1 2 3 4 5 6 7 8 9 % Utilization 45.1 can be obtained from simulation model.067 1.1 and 9) 0.03� � -. It means that other bottlenecks may exist in the system.977 1.954 0.81 47. . the system throughput mean cannot be improved when comparing at 75% CL (There is some overlapped area of the confidence interval).1. the utilization factor or p and the bottleneck rate or Rb are shown in Table I. From this problem.111 (P.964 0.14 38. machine number 7 meets all three criteria addressed previously to qualify as the system bottleneck.2) 50.111 (P.046 1. In this paper.067 0.111 (P. the simulation should be re-run again to find more system bottleneck candidates.10) 0.6) Rb 51. Rb 0.056 1. e 2 .4) 0.111 (P.111 (P. lO) 0.125 (p. 1) High value of the machine/process utilization 2) High value of the process utilization factor 3) Low value of the product bottleneck rate Then.6) 0.109 1. 2) Utilization factor (p): From [11].08 0.35 41.111 (p. Machines or processes with Rb are considered as bottleneck candidates.147 1. the three factors used to identify the bottleneck are collected and compared again (See Table II) 2nd DATA FROM TABLE SIMULATION TOOL AND ANALYZED DATA II Machine no. the station that meets following three criteria is identified to be real the system bottleneck.111.206 0. 0 1 2 3 4 5 6 7 8 9 p 1. 2) Example for machine no.111 (P.68 Machine no.I. 3 product 5 and product 6 bottleneck rate is 0.7" 3 -----------l 4 e+1l 03 of:-: +O = 0 2 43 e o .100 (part/min) 0.111 (P.42 45. 7 2.+1l . Rb 0.094 0.dat From Fig. 1. P. based on queuing theory.040 (part/min) 0.111 (P. = bottleneck). The three factors above can be used to identify potential bottleneck candidates. the utilization factor. The detail of this problem can be found in [14]. the simulation is used to confirm that the true bottleneck is selected and no other bottlenecks that still remain in the system (in case of multiple J ClasSlCdl C.111 (P.997 1.111 (P.5.41 46.

the final (5 ) step of TOC implementation can be applied to handle this issue because when the TOC system is adopted..003 Sepnmio 3: Adding me 7 + me 9 \ 3.. 2..."�.. +003 6 ..39 +003 d Fig.=... machines number 2 and 9 are also identified as system bottlenecks._---1 . Machines number 2 and 9 are identified to be the bottleneck candidates..+ .---"::"":--== 3-----1 '. IV.o03 2._..003 • �03 Sepnmio 2: Adding me 7 + me 2 Total TH 800f---------------::-:±r::o:o�. the number of simulation test can be reduced in order to identify the system bottleneck. Total TH . only machine number 7 and machine number 9 are identified to be the system bottleneck ( when 75% CL is adopted.. When the percentage of CL is increased. Furthermore.+003 . ----i 4HOC3 1.. In the based method. Thus. �003 Total TH 800f--------------:-:7""'C=""+"'�-= _::__:_:::--------1 4. . 75%CL comparison of throughput mean for 2nd test J Classicdl C.. Total TH 800'1-----------------:.6 e '00 31 1----------::-:-:-:..l1Se Sepnmio Fig.92& ..003 2. The 5 step of TOC implementation works as iterative improvement step that makes the system under TOC become self-adjusted when the system changes in the bottleneck location. the selected %CL should match with the number of runs used..�F.. Thus.1�. the simulation test is conducted with the based scenario at 75% CL based on 30 replications. more replications are necessary.84e+003 Seenmio 3: Adding me 7 + me 9 IV 3.9 1 +00".31e+003 2..-:� .-----. 2.�..:-.�.""-':":' -='. 'oo 1---------------:.. Observatkm Intervals . 1"..:=>+�.:.::-:-.lI Se SeNlmio 1: 2.. Although. +003 ' . A...�. with 10xl0 JSP.. the conclusion is that there are three bottleneck machines: machines number 2.dat _ ���IP:::' \. In case of small number of runs with the high %CL.... +003 . B.� :: �"r:. The simulation test is carried out to fmd the second system bottleneck.09.�.0'3 3.. = ��.+00� '\ 1.35e. Thus.-T:"..... In each scenario...::-: ::.003 2 . ::.. the advantage of this procedure when it is applied with large-scale size problems is to reduce the number of potential candidates by considering three criteria as mention earlier.n03 Total TH 800'1-------------.----14e+O03 '-1. On f'\ean Observation Intervals rh. this decision on bottleneck identification has effect on the bottleneck exploitation and subordination phase because the schedule must be considered first at the bottlenecks. 7 and 9.. 90O/oCL comparison of throughput mean for 2n Test 1594 . 88 /... this th situation occurs.6e+0031-----------:-:+-'� � .89e+003 2.. each bottleneck candidate capacity is added in addition to the increase of the capacity of machine number 7. there are three bottleneck machines (machine number 2. the judgment of bottleneck identification may be changed. the different in throughput mean sometimes cannot be clearly observed so the bottleneck may not be correctly identified.----1 'e'003 2. 90% CL is used instead of 75% CL..+:.. ' 5. the system performance must be evaluated to confirm that there is th no hidden bottleneck remaining in the system. 2 shows that the throughput of scenarios 2 and 3 show significant improvements over the based case..69 • .-/' . 6 . 7 and 9)).003 2. 10xlO JSP is used as a test J ClassIcal C I On Mean Th ddt problem.. the decision of bottleneck identification can be changed as in Figure 3. First Scenario: Apply 75% CL The result in Fig.2'&... From this study.. O03 3 1 Seenmio 2: Adding me 7 + me 2 3.__ . However. Thus... 8.. CONCLUSION AND DISCUSSION This paper aims to show the application of the simulation-based procedure for bottleneck identification in job-shop problem.1 5.Proceedings of the 2010 IEEE IEEM From Table II.I.9 e+O Total TH 1: Bl1SNI \.::-..61 . 3... .lI-. In this case...00 2 .o03 J .::. Second Scenario: Apply 90% CL In this scenario. when % CL is increased to 90%.-..=_ Total TH -..34e+003 Adding me 7 only 2 .__1 4.003 t. Where higher %CL is required. • .....43 :-c 03 +0 Adding me 7 only 2. there is no machine that meets all three criteria to qualify as the system bottleneck. two scenarios are set up. .�-------I 4 .. 003 .... The effect of changing confidence interval level when this procedure is applied is observed as when % CL used to compare system throughput mean is increased.. for this scenario. 2 .... 001--------------�:-c:. 2.

pp. vol. vol. "Simulation Tool for TOC implementation". Wu. 209-234. International Journal of Productivity and Performance Management. Hopp and M.3/4. 1595 . Kasemset. Kasemset and V. Spearman. Kachitvichyanukul. [3] S. "Simulation­ based procedure for implementing theory of constraints: Extension for cases with multiple bottlenecks". The 9th APIEMS Conference. [9] C. 17. 47- 55. [7] C. pp. 200 5. [13] C. International Journal Production Economics. 659-668. [12] W. [11] Z. Master research study Industrial System Engineering. Berlin. no. Pegels and C. "Simulation­ based tool for implementing theory of constraints". Kachitvichyanukul. Asian Institute of Technology. 332345. vol.Proceedings of the 2010 IEEE IEEM REFERENCES [1] E. Kasemset. 16. "Application of the theory of constraints to a bottleneck operation in a manufacturing plant". ASIMMOD 2009. 55 . "A simulation Analysis of the effectiveness of Drum-Buffer-Rope scheduling in furniture manufacturing". "TOC based Procedure for Job-shop Industrial Scheduling". [4] C. 3. 2007. vol. [8] C. C. Philippines. Gordon. J. 113-119. [2] R. 1986. 200 8. D. Radovislsky. McGraw-Hill. 22. 26. Cox. Factory Physic. 757-764. The 6th APIEMS Conference. pp. L. vol. Article 4426. 200 7. Vol. pp. The Goal: A Process of Ongoing Improvement. Thailand. 1998. Computers Industrial Engineering. North River Press. Asia Simulation Conference 2007. 302-311. A. [5] S. "Application of constrained management and lean manufacturing in developing best practices for productivity improvement in an auto­ assembly plant". Second Edition. in Proceedings of AsiaSim 2007. Kachitvichyanukul. Asian Institute of Technology. Thailand. "Simulation­ based procedure for bottleneck identification". pp. Seoul. 2005. Bali. "Resource Constraint Measures in a Dual Constrained Job Shop". 1994. pp. Springer. Revised Edition. Gargeya. Kachitvichyanukul. 200 9. Managing Service Quality. in Proceedings of the 9th APIEMS Conference. Bangkok. no. pp. M. M. Engineering and Management. Inc. 55. Kasemset and V. [14] C. "A Quantitative Approach to Estimate the Size of the Time Buffer in the Theory of Constraints". in Proceedings of the 6th APIEMS Conference. 6. Korea. pp. 2005. Thailand. "Simulation-based Tool for Theory of Constraints (TOC) Implementation".. Go1dratt and J. Reid. noA. Manila. 1994. [6] V. Kasemset and V. Omega. Morris and T. in Proceedings of ASIMMOD 2009. 1999. Bangkok. Watrous. Kasemset and V. Y. Indonesia. School of Advanced Technology. Journal of Manufacturing Technology Management. S. Taj and L. no. 86-97. "Applying the TOC five-step focusing process in the service sector". Doctoral Dissertation. 1811-1819. 2. Bangkok. School of Engineering and Technology. J. no. Berro. 2006. B. [10] C. 2009.