Beruflich Dokumente
Kultur Dokumente
net/publication/245310331
CITATIONS READS
19 1,557
2 authors:
Some of the authors of this publication are also working on these related projects:
Channel design and mechanisms design in a reversed supply chain View project
All content following this page was uploaded by Jen-Ming Chen on 04 May 2014.
In this paper, we develop a quality improvement model based on the structure of classical
economic production quantity (EPQ) model. The asymmetrical truncated loss function is used
to evaluate the cost of poor quality in a production system. A practical quality improvement
case which follows the 6-sigma DMAIC method in a car seat assembly line is discussed to
verify the proposed model. Our model provides a fundamental structure for studying the cost
of quality improvement in a production system. Based on this model, the management
can evaluate the affect of quality investment to generate significant financial return in their
production line.
quality and the cost of prevention. Banker et al. (1998) production quantity (EPQ) model. The total cost,
developed formal models of oligopolistic competition TC(Qi, I), can be written as:
to investigate whether equilibrium level of quality
increases as competition intensifies. Krishnan et al. TCðQi, I Þ ¼ production cost þ holding cost
(2000) examined the relationship between life-cycle þ setup cost þ poor quality cost
productivity and conformance quality in software pro- þ cost on quality investment.
ducts. Salameh and Jaber (2000) considered a special
production/inventory situation where items, received The notation in our model is summarised as follows.
or produced, are of imperfect quality. Ganeshan et al.
P: production cost of unit product.
(2001) analysed the interaction between the economics
I: investment on quality improvement during
of production and process quality with Taguchi’s
the N period planning horizon.
perspective of poor quality. Chen and Tsou (2003)
CU: rejection cost as the quality character above
developed a static financial model on quality investment
the upper specification limit/unit.
with Taguchi’s perspective of poor quality in a produc-
CL: rejection cost as the quality character below
tion system. Lloréns-Montes et al. (2004) investigated
the lower specification limit/unit.
the impact that implementing quality improvement
S: setup cost per period.
processes has on manufacturing flexibility. In this
Qi: production lot size in period i.
paper, based on the study of Chen and Tsou (2003),
h: holding cost per unit per period.
we propose a quality improvement model with asymme-
Gi1: inventory carried from period i to i þ 1.
trical truncated loss function to analyse the cost on
di : demand in period i.
quality improvement in a production system. A practical
N: production period.
case of a car seat assembly line is used to verify this
D(X,I ): quality distribution function.
model.
(I ): population mean of quality characteristic.
This paper makes three significant contributions.
(I ): population standard deviation of quality
First, a quality improvement model is developed that
characteristic.
can be used to evaluate the total cost in a poor quality
L(X ): loss function of poor quality per unit product.
production system. In this model, we focus on the
K1, K2: Taguchi loss parameter.
financial return of quality improvement, but not the
USL: upper specification limits.
lot-size optimisation. This task is very difficult for a
LSL: lower specification limits.
traditional production-quality model. Second, the
X: value of the quality characteristic.
asymmetrical truncated loss function has been incorpo-
rated into our model, increasing the generality of this Each term in our objective function is defined as below:
model. Third, a practical case in a car seat assembly XN
line, which obeys the 6-sigma DMAIC methodology, Production cost ¼ Qi P ð1Þ
is discussed to verify the proposed model. i¼1
In the next section, we introduce the model of this The first term in our model is the production cost.
paper. Then, a quality improvement case in a car seat The production cost is the cost to produce the product
assembly line is used to verify this model, following and includes both the material cost and manufacturing
the 6-sigma DMAIC methodology in the process of cost. The production cost during the N period can be
problem solving. Then, we use this case to verify written as the summation of the lot size multiplied by
the proposed model and compare the change after production cost of unit product in every period.
quality improvement. The last section is the conclusion XN
of this paper. Holding cost ¼ h½Gi1 þ Qi di ð2Þ
i¼1
The second term in this equation is the holding cost,
2. The model which is the cost to store the product in the warehouse.
This includes the cost to rent the warehouse, salary pay
We assume a production line over N periods. The to the storekeeper and other management cost. The
primary goal of this model is to find the total cost of holding cost on the production line is the summation
a production system. The total cost, TC(Qi, I), is the of the holding cost of the inventory in the N period.
combination of the production cost, the holding cost, In equation (2), Gi1 is the inventory carried from
the setup cost, the poor quality cost and the money previous period, Qi is the production lot size in period
spent on improving quality. This cost structure is i, and di is the demand in period i. [Gi1 þ Qidi] is the
based on the hypothesis of the classical economic inventory in period i and h[Gi1 þ Qi di] is the holding
Case study: quality improvement model in a car seat assembly line 683
Subject to
Qi , I, P 0 Figure 3. The seat assembly line in Rica Auto Parts Co. Ltd.
G0 ¼ 0
and (
0 if Qi ¼ 0
ðQi Þ ¼ ð7Þ quality for original equipment manufacturers (OEMs)
1 if Qi > 0 and aftermarket vehicles. According to market feed-
back, customers complain about the operating noise of
their power seats. In order to decrease warranty cost and
increase customer satisfaction, the production manager
3. Quality improvement case in a car seat assembly line
has been instructed to reduce warranty claims and
evaluate the production cost of their car seat assembly
In this section, we consider a quality improvement case
line. The 6-sigma DMAIC method has been used in this
in a car seat assembly line to verify the proposed model.
assignment. Rica Auto Parts has its own seat assembly
The 6-sigma DMAIC methodology has been applied
line and its quality operation system has obtained the
on this quality improvement case, and its roadmap is
TS16949 certification. The appearance of Rica’s seat
shown in figure 2.
assembly line is shown in figure 3.
To give a clear picture, figure 4 is the flow chart of
the manufacturing process of the power seat. There are
3.1. Define
seven major steps in the manufacturing process. These
include:
In the define phase things which are critical to con-
sumers (CTC) are defined such as: What are the most . wrapping cushion,
important things to your consumer? What do they care . assembling sliding tracks,
about? The purpose and goal of our 6-sigma project also . assembling seat back,
are clarified in this phase. . wrapping seat back,
Rica Auto Parts Co. Ltd is a local supplier of car . functional testing,
seats. This company produces seat units of premium . packing and delivering to the warehouse.
Case study: quality improvement model in a car seat assembly line 685
All factors that will affect the result of each step have
also been listed in figure 4.
In order to find the root causes of warranty claims,
all the warranty parts were recalled for analysis. After
study, it was found that 87% of the warranty claims
were caused by operating noise, 10% were due to
customer abnormal use and 2% were from other issues.
Based on the 80/20 rule, the operating noise problem
was recognized as the major sources of quality issues.
3.2. Measure
Figure 5. Measuring the distance between the front edges of
In the measure phase, the performance standard of the the right and left seat frame.
process is verified and the measurement system is estab-
lished to obtain a baseline for future improvements.
The production manager gathered a group of Table 1. Distance between the front edges of the right and left
experts to determine the root causes of operating seat slide before improvement.
noise. The C&E matrix and production/design failure No. No. No.
mode and effects analysis (P/D FMEA) was used to
analyse the problem. After analysis, it was found that 1 402.9 11 402.4 21 402.8
the root cause of the operating noise problem was 2 403.4 12 403.2 22 403.6
3 403.1 13 402.4 23 402.9
caused by the lack of parallel alignment between right 4 401.8 14 401.8 24 403.4
and left seat slides. To control the alignment, the 5 402.3 15 402.7 25 401.8
distance between the front edges of the right and left 6 402.3 16 402.9 26 403.2
seat slide was measured, as shown in figure 5. Based 7 402.8 17 403.5 27 402.9
on this figure, the upper specification limit (USL) and 8 402.9 18 402.0 28 403.1
9 402.1 19 402.8 29 402.4
the lower specification limit (LSL), of this distance is 10 402.9 20 403.5 30 402.9
404 mm and 400 mm respectively. When the distance SPEC 402 2 mm
exceeds the specification limits and operating noise
is too large, the inspector will adjust the slides to control
the alignment. The rework cost is $1 per unit.
We measured the distance between the front edges of The computer software, Minitab, was used to analyze
the right and left seat slide for a 30 unit sampling from the collected data, and the results are shown in figure 6.
the production line to evaluate the process capability The mean of these sampling data is 402.757 mm and
of seat slides assembly. The data are shown in table 1. standard deviation is 0.599 mm. In addition, the process
686 J.-C. Tsou and J.-M. Chen
3.3. Analyse
the fixture, the mean of the distance between the front 3.5. Control
edges have been centred to be consistent with the
target, and the standard deviation has been decreased. In the control phase, the control plan to maintain
The mean and standard deviation after improvement progress and verify the measurement system in the
are 402.103 mm and 0.43 mm. The process capability future is also done. All corrective actions are feedback
has been improved from 0.69 Cpk value to 1.47 Cpk to the quality operation system.
value and the operating noise also clearly improved. After this improvement, we identified the parallelism
Our process has been improved to a 4.38 sigma process. between right and left seat slide as the high impact
characteristic (HIC) and decide to drew the X-bar R
chart to control the stability of this key dimension.
All these actions were fed back to the quality control
plan of car seat assembly line. We also fed back these To compare the change in production line after
improving activities to the PFMEA of the car seat improvement, we use Mathematica again to find the
manufacturing process. To sustain the benefit of quality total cost of the production line after improvement.
improvement over the long run, we have set a control The results are listed in table 5. The total cost for
plan to track all correcting activities in the future. these ten production periods after improving activities
is $62 075.8, which is lower than the cost before quality
improvement. This is because improving quality can
reduce the cost of poor quality and the saving is the
3.6. Analysing the total cost in the car seat assembly
line before and after improvement
Cost of poor quality
In this section, we use the quality improvement case 1
of the car seat assembly line to verify the quality
improvement model we proposed. We assume that 0.8
there are 10 production periods and inventory in the
beginning is 0. The production cost is $12 for each 0.6
car seat, and the setup cost per production run is
0.4
$100. The holding cost is $0.05 per unit per period.
The supply and demand condition in the ten production 0.2
periods are listed in table 3. From the data, it can be
seen that the volume is different in every period. Distance
400 402 404 406
In business practice, production lot sizing is affected
by the size of the order. Figure 10. Cost of poor quality of the car seats.
The poor quality cost of the car seat is a special case
of the asymmetrical truncated loss function, L(X). In
the car seat, CU and CL are the rework cost to adjust
Table 4. Inventory and cost change with time before
the parallelism of seat slides which is $1 per unit. Based improvement.
on this, the L(X) can be written as:
( Period Supply Demand Inventory Cost
$1 404 mm < X; X < 400 mm
LðXÞ ¼ 2
if 1 500 450 50 6218.98
0:25 ðX 402Þ 400 mm X 404 mm 2 510 520 40 6340.81
3 490 510 20 6095.15
The drawing of L(X) is shown in figure 10. 4 510 480 50 6341.31
With the above information, we can find the total 5 490 470 70 6097.65
cost of the production line before quality improvement. 6 500 510 60 6219.48
To calculate the total cost, the computer software 7 550 530 80 6832.13
8 490 510 60 6097.15
Mathematica is used to do the calculation. The results
9 500 490 70 6219.98
are listed in table 4. It should be noted that the inventory 10 510 500 80 6342.81
in every period is different and it is affect by the Total 5050 4970 80 62805.5
supply and demand status in every period. The total
cost for these 10 production periods is $62 805.5.
Table 5. Inventory and cost change with time after
improvement.
Table 3. Supply and demand during ten periods.
Period Supply Demand Inventory Cost
Period Supply Demand
1 500 450 50 6326.94
1 500 450 2 510 520 40 6246.93
2 510 520 3 490 510 20 6004.95
3 490 510 4 510 480 50 6247.43
4 510 480 5 490 470 70 6007.45
5 490 470 6 500 510 60 6127.44
6 500 510 7 550 530 80 6730.88
7 550 530 8 490 510 60 6006.95
8 490 510 9 500 490 70 6127.94
9 500 490 10 510 500 80 6248.93
10 510 500 Total 5050 4970 80 62075.8
Case study: quality improvement model in a car seat assembly line 689
6400
Before
6200
After
6000
5800
5600
5400
1 2 3 4 5 6 7 8 9 10
Period
Figure 12. Behaviour of the total cost to the quality
Figure 11. Behaviour of the total cost of production line investment.
before improvement to the total cost after improvement in
every period. The initial value of the variance 02 is 0.36 mm2 and
the initial mean 0 is 402.757 mm. These numbers are
taken from the previous case before quality improve-
cost on quality investment (modifying the fixture). From ment. The target value of the mean and the variance
the data, we also can notice the inventory in every per- are 402 mm and 0 mm2. According to the previous
iod is the same as the inventory in the production line case, the mean and the variance after we invest $200
before improvement. This is because quality improve- on quality improvement are 402.103 mm and
ment will not affect the supply and demand status in 0.185 mm2. Hence, we can find ¼ 0:00998 and
every period, only affecting the cost of poor quality. b ¼ 0.00333.
To compare the total cost in every period, figure 11 The supply and demand condition in the 10
shows a trend chart to trace the change of the total cost production periods is the same as the condition in
of production line before and after improvement. previous case. Then we achieve the relationship between
We notice in the first period that the total cost of the quality investment and the total cost. The behaviour of
production line after improvement is higher than the quality investment to the total cost of the production
total cost before improvement. This is because the cost line is drawn in figure 12.
to modify the fixture occurs in the first period, and this From figure 12, there is a minimal total cost, $62 075,
quality investment increases the total cost for the as we invest $205 on quality improvement. After that,
improved production line. After that, the cost of poor the total cost increases with quality investment. This
quality will be reduced. The reduction in the cost of result verifies the assumption that there is an optimal
poor quality will decrease the total cost in the last nine balance between quality investment and production
periods after quality improvement. This is why only cost.
in the first period is the total cost in the improved
production line higher than the original production line.
4. Summary and conclusion
Further study in this field could focus on the quality Ganeshan, R., Kulkarni, S. and Boone, T., Production
investment model of a stochastic dynamic production economics and process quality: A Taguchi perspective.
Int. J. Prod. Econ., 2001, 71, 343–351.
system. Researchers may include more financial factors Goyal, S.K. and Gunasekaran, A., Effect of dynamic process
or different quality evaluation models into their study. quality control on the economics of production, Int. J. Op.
& Prod. Manage., 1990, 10, 69–77.
Goyal, S.K., Gunasekaran, A., Martikainen, T. and
Yli-Olli, P., Integrating production and quality control
Acknowledgements policies: A survey. Euro. J. Op. Res., 1993, 69, 1–13.
Hong, J., Xu, S.H. and Hayya, J.C., Process quality improve-
We would like to thank the editor and the two ment and setup reduction in dynamic lot-sizing. Int. J. Prod.
anonymous referees for their valuable and constructive Res., 1993, 31, 2693–2708.
Krishnan, M.S., Kriebel, C.H., Kekre, S. and
comments which have led to a significant improvement Mukhopodhayay, T., An empirical analysis of productivity
in this paper. and quality in software products. Manage. Sci., 2000, 46,
745–759.
Lee, H.L. and Rosenblatt, M.J., Simultaneous determination
of production cycles and inspection schedules in a produc-
References tion system. Manage. Sci., 1987, 33, 1125–1137.
Lloréns-Montes, F.J., Garcı́a-Morales, V.J. and Verdú-Jover,
Banker, R., Khosla, I. and Sinha, K., Quality and competition. A.J., Flexibility and quality management in manufacturing:
Manage. Sci., 1998, 44, 1179–1192. an alternative approach. Prod. Planning & Control, 2004,
Bryne, D.M. and Taguchi, S., The Taguchi approach to 15, 525–533.
parameter design. ASQC Qual. Con. Trans., 1986, 177, Maghsoodloo, S. and Li, M.C., Optimal asymmetric tolerance
168–173. design, IIE Trans., 2000, 32, 1127–1137.
Chand, S., Lot sizes and setup frequency with learning in Ng, W. and Hui, Y., Interactive quality improvement of a
setups and process quality. Euro. J. Op. Res., 1989, 42, process subject to complete inspection, Int. J. Prod. Res.,
190–202. 1996, 34, 3275–3284.
Chen, J.M. and Tsou, J.C., An optimal design for process Porteus, E.L., Optimal lot-sizing process quality improvement
quality improvement: modelling and application. and setup cost reduction. Op. Res., 1986, 34, 137–144.
Prod. Planning & Control, 2003, 14, 603–612. Rosenblat, M.J. and Lee, H.L., Economic production cycles
Cheng, T.C.E., An economic production quantity model with with imperfect production processes. IIE Trans., 1986, 18,
flexibility and reliability considerations. Euro. J. Op. Res., 48–55.
1989, 39, 174–179. Salameh, M.K. and Jaber, M.Y., Economic production quan-
Cheng, T.C.E., Economic order quantity model with demand- tity model for items with imperfect quality. Int. J. Prod.
dependent unit production cost and imperfect production Econ., 2000, 64, 59–64.
processes. IIE Trans., 1991, 23, 23–28. Taguchi, G. and Wu, Y., Introduction to Off-line Quality
Fine, C.H., A quality control model with learning effects. Control, 1980 (Central Japan Quality Control Association:
Op. Res., 1988, 36, 437–444. Meieki Nakamura-ku Magaya, Japan).
Fine, C.H. and Porteus, E.L., Dynamic process improvement. Tapiero, C., Production learning and quality control.
Op. Res., 1989, 37, 580–591. IIE Trans., 1987, 19, 362–370.
Jen-Ming Chen is a professor at the Institute of Industrial Management at the National Central
University (Taiwan). He received a BS in Industrial Management Science from the National
Cheng Kung University (Taiwan) in 1983, an MS in Industrial Engineering from the
University of Arizona in 1988, and a PhD in Industrial Engineering from the Pennsylvania
State University in 1992. His research interests include inventory and supply chain management,
and pricing and yield management. He is an active member of several professional organisations,
including Informs, DSI, and IIE. Dr. Chen is the recipient of the George B. Dantzig Dissertation
Award from Informs and recipient of the IIE Doctorial Dissertation Award, both in 1994.