You are on page 1of 5

The Impact of Costs of Quality: A Simulation Approach

Mohamed K. Omar1, H. K. Sim2, Sharmeeni Murugan1and M. R. Muhamad1


1
Faculty of Manufacturing Engineering, Technical University Malaysia (UTeM), Melaka, Malaysia
2
Faculty of Engineering & Technology, Multimedia University (MMU), Melaka, Malaysia
(mkhaled@utem.edu.my, hksim@mmu.edu.my, mohdrazali@utem.edu.my)

model and the process described by Son and Hsu [6].


Abstract - In this research an analytical model developed by Their model was modified and a simulation model was
Son and Hsu to compute the cost of quality was improved to developed as detailed in the next sections of this paper.
include the acceptance sampling costs associated with
inspection of incoming materials. The modified model was II. THE QUALITY COST MODEL
then used to carry out comprehensive simulation work that
investigated a set of quality control strategies practiced by
the manufacturing community. The results indicate that In this section, we briefly explain the model developed by
measuring and understanding cost of quality is very Son and Hsu [6] and the way that it was modified to be
important. Moreover, the results show that some of the suitable for this research work.
quality control strategies have a significant impact on cost of
quality. A. Notations
t sampling interval
n sample size per sampling
Keywords – Cost of quality, simulation, quality control  average number of assignable cause per hour
strategies decisions d1 mean time spent to identify an assignable cause
d2 mean time spent to correct an assignable cause
g time required to sample a product
I. INTRODUCTION to time inspection of the acceptance sampling
Product quality is one of the important and critical cas cost of investigating acceptance sampling
factors for achieving competitiveness. But quality comes  probability of process in control state 2
with cost, and this leads to the fact that measuring and  Probability of process in control state 2
reporting cost of quality is very important. One must  time taken for assignable cause to occurs
stress the fact that the objective of understanding and T1 in control period
measuring cost of quality is not just to reduce the costs, T2 time period until the assignable is detected for the
but to ensure that the costs associated with quality are the first time
right kind of costs that maximizes benefits. In the T3 time required to investigate the true alarm
literature, one can classifies costs of quality models into T4 time to correct the assignable cause
four groups of generic models. These are: P-A-F model,
opportunity cost models, process cost models and ABC A Description of the model
(activity based costing) models. Feighenbaum [1] Son and Hsu [6] considered a small manufacturing
categorized operating quality costs into two major system which consists of a machining area (sampling
components: cost of control and cost of failure of control. inspection) and a final inspection area (100%
Cost of control includes appraisal and preventive costs inspection).However; they assume that 100% inspection
and cost of failure includes internal and external costs. is not always possible. They explained that a complete
Chavez and Beruvides [2] incorporate opportunity loss check of a component part may require the part removal
into traditional P-A-F quality expenses. According to from the fixture; the removal makes it difficult to realign
them, opportunity losses are: underutilization of installed the part to its original position. Therefore, they assume
capacity, inadequate material handling and poor delivery that sampling inspection is assumed for component parts
of service. They express costs of quality as revenue lost (in-process inspection), and 100% inspection for finished
and profit not earned. The process costs model developed parts (final inspection). In their modeling, first they
by Ross [3] and first used for quality cost by Marsh [4]. It calculate the cycle time, and then quality cost (prevention
focuses on process rather than products or service and it and failure) per cost was determined. In the next section,
defines process cost as the total cots of conformance and equations for all components are stated, obviously,
non-conformance for a particular process. The active- readers interested in the detail steps of deriving these
based costing (ABC) was developed by Cooper and equations should referred to the original article.
Kaplan [5].Under this approach, accurate costs for various
costs objects are achieved by tracing resource costs to B Cycle Time Calculations
their respective activities and the cost of activities to cost The assignable cause occurs according to the Poisson
objects. This paper takes a special interest in the cost process with mean rate . The period of the process is in

978-1-4244-4870-8/09/$26.00 ©2009 IEEE 1327


Proceedings of the 2009 IEEE IEEM

control is represented by T1. T1 follows an exponential


distribution with mean 1/ Where E(CP) is representing the prevention cost, E(CF) is
representing the failure costs and E(CA) is representing
The expected length of a cycle; the acceptance sampling costs.

E(Tc) = E(T1) + E(T2) + E(T3) + E(T4) (1) III. TH PROBLEM AND SOLUTION
Which can be obtained using equation 2 METHODOLOGY

= 1/ + (t + t/(1-)-) + (gn + d1) + d2 (2) Unlike the single stage problem considered by Son
and Hsu [6], we consider a manufacturing system that
C Prevention Cost per Cycle consists of two stages. Moreover, rather than representing
There are three types of prevention costs that occurs the appraisal cost by a single component (in-process
at the machining area during a cycle of time. The first cost inspection), we thought it is more appropriate to include
is associated with inspection works and denoted as (Cp1), the incoming inspection (acceptance sampling plan)
the second cost is representing the cost of investigating which practiced by the manufacturing community. So the
the false alarms (Cp2), and the third cost is associated with problem can be described as: a manufacturing system that
adjusting the assignable cause (Cp3). Then, the expected consists of two stages, incoming raw materials inspection
prevention cost per cycle (Cp) can be determined by carried out according to some quality plan and as a result
summing up these three components as in equation 3. an element of appraisal cost in incurred prior to the
commencement of the production activity. Once orders
E(Cp) = E(Cp1) +E(Cp2) + E(Cp3) (3) are realized by the company, raw materials are brought
into the shop floor and at this stage; preventive and failure
D Failure Cost per Cycle costs are incurred. In order to investigate the cost of
The failure cost has three components; the first quality for this manufacturing system, we have adopted
component is the cost of rework per cycle (Cf1), the the four strategies for inspection and removal of
second component is the scrap cost per cycle (Cf2) and the defectives across a range of detection rates reported by
third component is the external failure cost during a cycle Gardner et.al [7]. The strategies are summarized
of time (Cf3). Then, the expected failure cost per cycle The strategies are summarized as:
(CF) can be determined by summing up these three
components as in equation 4. 1-Inspection and removal of defectives based on
acceptance sampling prior to assembly points.
E (CF) = E(Cf1) + E(Cf2) + E(Cf3) (4) 2-Inspection and removal of defectives at completion of
finished product only.
E Acceptance Sample Cost Per Cycle 3-Inspection and removal of defectives prior to assembly
Costs associated with sampling inspections prior points.
entering the manufacturing system are not considered in 4-Inspection and removal of defectives following every
the model developed by Son and Hsu [6]. Our idea operation
consists of the fact that companies often receive a
shipment of material from a supplier and need to ascertain Using the cost model developed earlier with its three
that the quality of the shipment. If it impractical to inspect components, a simulation model was developed for the
every item in the shipment, then a sampling plan is the (n, two stage process. @Risk spreadsheet simulation software
c) is used. In this plan, n items are chosen (without package was used to develop the model and carryout all
replacement) from a batch of shipped material. If c or the simulation works.
fewer of the sample items are defective, then the batch is
accepted; otherwise, the batch is rejected. Therefore, there IV. EXPERIMENT DESIGN
will be a cost incurred during inspection of the sample
before entering the manufacturing system. The expected In designing the experiments, we thought it would be
number of samples entering the manufacturing system is a good idea to consider inspection time as the main factor
n whereas to is representing the time taken to inspect the for calculating cost of quality. In real-life manufacturing
sample. Let (CaS) be the cost of investigating acceptance systems, one of the most import factor that production
sampling during a cycle of time. Then the expected cost line managers pay attention to is the cycle time and many
associated with this activity is given by equation 5 times, quality control activity were sacrificed to keep a
E (CA) = Casnto (5) desired cycle time. Therefore, we have decided to use
fraction detection rate to every operation in our
Hence, the total cost of quality can be represented by manufacturing system. The fractions of detection rate are
summing up all the cost components (1) through (5) and divided into five categories: 0.001, 0.005, 0.010 and
is presented by equation 6. 0.100. Every trail will run for 10000 times of simulation
runs, (see Table 1) Therefore, there will be twenty
E (CQ) = E(CP) + E(CF) + E(CA) (6) combinations of strategies and detection rate. It is worth

1328
Proceedings of the 2009 IEEE IEEM

nothing that when designing the experiment for the resulted from considering each of the four strategies and
sampling acceptance, we have assumed that the company treating process one and two independently from each
would receives a batch of 10,000 items which require be other.
assembled using processes one and two. Moreover, we
have assumed that the company quality engineers have The results in Table 2 reveal that adopting the
developed a quality plan which is the (n, c) system. For strategy that calls for carrying out quality works only at
example, in process one, a sample of 90 items will be the end of process (third row for process one and ninth
inspected and the entire 10000 items will be rejected if 0, row for process 2) have produced the largest cost of
1, 2, 3, 4, or 5 defective items are found in the sample, quality for each process when compared with other
otherwise, the batch is rejected. On the other hand, for quality strategies. This finding is expected since
process 2, a sample of 40 items will be inspected and the undetected quality problems will increase the rework and
entire batch will be accepted if 0, 1, 2, or 3 defects are scraps which will cause the failure cost to increase.
found, otherwise, the batch will be rejected. Therefore, Moreover, the results in table 2 indicate that the
the idea here is to set the Acceptable Quality level (AQL) inspection time has a significantly impact on cost of
equal to 1% and the Lot Tolerance Percentage Defective quality for all of the four strategies considered. For
(LTPD) equal to 5% and then carry out the simulation example, in the case of process one and when considering
works to calculate the costs associated with quality when the strategy to conduct quality control upon completion of
a set of management strategies which were explained finished product, increasing the fraction detection rate
earlier. Moreover, since verification of the simulation from 0.001 to 0.1 will reduce the cost of quality by about
model is an important issue that has to be done, we have 26 times for process one and about 3.6 times for process
used the data provided by Son and Hsu [6] and run the two. It is true to conclude that increasing the time window
simulation model without the acceptance inspection cost for quality control activities will increase the cycle time, a
component and then compared the results provided by problem that many production line managers strive very
Son and Hsu [6] analytical method and that obtained by hard to avoid. However, the consequences of reducing
the simulation model. Both results exactly matched and detection time have a significant impact of failure costs.
that indicates that the simulation model is free from any
illogical errors. TABLE 2 Cost of Quality for process 1 and 2

TABLE 1 Experiment design for Simulation Run Cost of Quality for process 2
Inspection Fraction Detection Rate per Operation
Inspection and Fraction Detection per Operation and defective 0.001 0.005 0.01 0.05 0.1
defective removal 0.001 0.005 0.01 0.05 0.1 removal
strategies strategies
Acceptance sampling 10000 10000 10000 10000 10000 Acceptance
run run run run run sampling 1295.07 312.57 180.31 72.41 49.41
Completion of finish 10000 10000 10000 10000 10000 Completion 1298.23 313.25 180.66 72.49 49.42
product run run run run run of finish
Prior to assembly 10000 10000 10000 10000 10000 product
run run run run run Prior to 1072.03 265.28 159.31 65.60 49.33
Following every 10000 10000 10000 10000 10000 assembly
operation run run run run run Following 922.97 227.84 134.27 57.93 41.66
every
operation
Cost of Quality for process 2
V. RESULTS AND DISCUSSIONS Inspection Fraction Detection Rate per Operation
and defective 0.001 0.005 0.01 0.05 0.1
When considering the manufacturing system removal
described in this research, and since simulation as a tool strategies
Acceptance
allows the user to create different scenarios that can be sampling 3,550.00 1,824.78 1,471.26 1,151.72 974.50
used to investigate cost of quality, we thought it would be Completion 3558.95 1829.28 1474.85 1154.48 976.81
a good idea to create there simulation scenarios for the of finish
case considered here. The first scenario is to consider product
Prior to 3073.71 1577.40 1271.42 994.84 841.87
process one alone, the second scenario is to consider
assembly
process two and the third scenario is to consider the two Following 3062.25 1574.73 1269.92 994.40 841.60
processes combined. For each scenario, the four strategies every
were used to investigate the impact of each decision on operation
cost of quality. A two-fold rational for such thought: first,
more insights can be revealed and secondly, it allows the To provide a graphical illustration of the behavior of
user to check for model consistency. Model results cost of quality with the detection rates for processes 1 and
consistency is important factor in simulation works when 2, we have created figures 1 and 2 at the end of this paper.
the user lacks a real application to carryout model It is very clear from figures 1 and 2 that cost of quality is
validation activities. Table 2 shows the cost of quality that it its maximum value when the detection time is zero, and

1329
Proceedings of the 2009 IEEE IEEM

then steadily decreased as the window for inspection Secondly, failure costs are very large and managers
increases. should completely avoid these costs since there is no
tradeoff that exists with these costs. Another important
Considering simulation scenario three that calls for issue is the fact that using simulation to measure and
combining the two processes provides a near situation to understand cost of quality will provide managers with
what is happening in real-life industry. In this case, one opportunity to rank their process in terms of cost of
has to determine the possible quality strategies quality and the cost consequences that resulted from
combinations that could be considered for simulation adopting a specific set of strategies .
runs. If we let R be the possible strategies combination
outcome, then we would have:

R = Si, j (8)
VI. CONCLUSIONS
Where
i = A strategy in process 1 i = 1, 2, 3, 4
In this paper, an analytical model reported in the
j = A strategy in process 2 j = 1, 2, 3, 4
literature for cost of quality computations was
`
considered and modified to include an important
Table 3 shows all possible quality strategies
component of cost of quality. The model is then used to
combinations with their respective cost at each selected
develop a simulation model for a two-stage
fraction detection rate for each operation.
manufacturing system. Moreover, quality control
strategies common in the manufacturing community
TABLE 3 Total Cost of Quality for process 1
were used in the simulation works to investigate their
Inspection Fraction Detection Rate per Operation
impact on cost of quality. The results indicate that some
and 0.001 0.005 0.01 0.05 0.1
of these practiced strategies when combined with the
defective detection periods will significantly increase cost of
removal quality. Furthermore, the results indicate simulation
strategies
S1,1
works can be used to understand and measure cost of
4,845.07 2,137.34 1,651.57 1,224.13 1,023.91
S1,2
quality. An interesting venue for further research is to
4,854.02 2,141.85 1,655.16 1,226.89 1,026.21
S1,3
use simulation to investigate a real-industrial application
4,368.78 1,889.97 1,451.73 1,067.25 891.27
S1,4
to investigate the cost of quality; the authors are
4,357.32 1,887.30 1,450.23 1,066.82 891.01
S2,1 4,848.23 2,138.03 1,651.92 1,224.21 1,023.93
currently considering such idea.
S2,2 4,857.18 2,142.53 1,655.51 1,226.97 1,026.23
S2,3 4,371.95 1,890.65 1,452.08 1,067.32 891.30
S2,4 4,360.48 1,887.98 1,450.58 1,066.89 891.03 REFERENCES
S3,1 4,622.03 2,090.06 1,630.57 1,217.32 1,023.83
S3,2 4,630.98 2,094.56 1,634.16 1,220.08 1,026.13 [1] A. V. Feigenbaum, “Total quality control”, 3rd. ed.,
S3,3 4,145.74 1,842.68 1,430.73 1,060.44 891.20 McGraw-Hill, New York, NY. 1991.
S3,4 4,134.28 1,840.00 1,429.23 1,060.01 890.93 [2] D.A. Sandoval-Chavez and M. G. Beruvides, “Using
S4,1 4,472.97 2,052.62 1,605.53 1,209.65 1,016.16
opportunity costs to determine the cost of quality: A case
S4,2 study in a continuous-process industry, Engineering
4,481.93 2,057.12 1,609.12 1,212.41 1,018,.47
S4,3 Economist, vol. 43, pp. 107-109. 1998.
3,996.69 1,805.25 1,405.69 1,052.76 883.53
[3] D.T. Ross, “Structured analysis (SA): A language for
S4,4 3,985.23 1,802.57 1,404.19 1,052.33 883.26 communication”, IEEE Trans. On Software Eng., Vol. SE-
3, No. 1, pp. 16-19.1977.
As it could be seen from table 3, the combination of [4] J. March, “Process modeling for quality improvement”,
{2, 2} strategies have caused the system to incur the Proceedings of the second international conference on total
largest cost of quality. This is the case when the quality management, London, IFS publication, pp. 111-121
management decided to use the quality strategy in which 1988.
quality control inspection works is only to be carried at [5] R. Cooper and R. S., Kaplan, “Measure costs right: Make
the end of the production runs. On the other hand one can the right decision”, Harvard Business Review, Vol. 66, No.
5 pp. 96-99.1988.
see from table 3 that that adopting combination of {4, 4} [6] Y. K. Son and L. F. Hsu, “A method for measuring quality
strategies have caused the system to incur the least cost of costs”, International journal of production research, vol. 29,
quality. As it could be easily seen that {4, 4} strategies no. 9, pp. 178-179. 1991.
calls for quality control works to be carried after each [7] L. L. Gardner, M. E. Grant and L. J. Rolston, “Using
operation and as a result, completely eliminating the simulation to access costs of quality”, Proceedings of the
failure costs. IEEE 1995 winter simulation conference, pp. 945-950.
There are many important management implications
illustrated in all the examples presented in this paper.
First, quality costs are very large when quality activities
time window is sacrificed to reduce the total cycle time.

1330
Proceedings of the 2009 IEEE IEEM

Total cost of quality versus the detection rate per operation

$1,400,000.00
Total cost of quality

$1,200,000.00

$1,000,000.00
1. Acceptance sampling

$800,000.00 2. Completion of finished


product
$600,000.00 3. Prior to assembly

$400,000.00 4. Following every operation

$200,000.00

$0.00
0 0.05 0.1 0.15
Detection rate per operation

FIGURE 1 Cost of quality for process 1

Total cost of quality versus detection rate per


operation
$4,000,000.00

$3,500,000.00
Total cost of quality

$3,000,000.00
1. Acceptance sampling
$2,500,000.00
2. Completion of finished
product
$2,000,000.00
3. Prior to assembly
$1,500,000.00
4. Following every operation
$1,000,000.00

$500,000.00

$0.00
0 0.02 0.04 0.06 0.08 0.1 0.12
Detection rate per operation

FIGURE 2 Cost of quality for process 2

1331

You might also like