Professional Documents
Culture Documents
ISSN No:-2456-2165
Abstract:- A Group Acceptance Sampling Plan (GASP) observing the lifetime of the products until it fails is not
is designedto study the truncated life testplan when the possible. The truncated life test is intentionally used to save
lifetime of an item follows a new compound distribution the time and cost of the experiment in such a way that life
called as Generalized Exponential-Poisson (GEP). In this test can be studied at the specified time period.
article the design parameters are developed for the
group size, its acceptance numbers, OC curve, minimum In an attribute single sampling planbased on the
number of groups are determined through the specified truncated life testa decision of acceptance or rejection of the
consumer’s confidence level and test termination ratio. lot is made based on the single sample which is the
Two points on the OC curve approach is incorporated to traditional procedure in sentencing the lot. Epstein. B (1954)
design the proposed plan. The OC values are calculated discussed truncated life test in the exponential case. Goode
when the ratio of specified average life and the actual H.P and Kao J.H.K (1961) have studied sampling plan based
average life is given. The minimum mean ratio for the on the weibull distribution. Balamuarali S. and Lee S.H
proposed plan are determined at the fixed producer’s (2006) discussed variable sampling plans for Weibull
risk. The obtained plan parameters are illustrated with distribution under sudden death testing. Kaviyarasu and
areal time example with the simulation study which are fawaz (2007) has studied reliability sampling plan to ensure
exhibited in the tables. percentiles through Weibull Poisson Distribution.
Kaviyarasu, V. and Sivasankari, S. (2020) studied the Single
Keywords:- Generalized Exponential-Poisson distribution, sampling plan for life testing under the Generalized
consumer’s confidence level, producer’s risk, operating Exponential-Poisson Distribution. Every single item in the
characteristic function, truncated life test. sampling units are required single tester however in practice
a tester may accommodate multiple number of items
I. INTRODUCTION simultaneously hence it saves more time and the cost of the
experiment. Here the items in a tester can be regarded as a
In the competitive global market, quality product group and the number of items in the group is called as
always seeks more attention and demand to meet the group size such as study is called as Group Acceptance
standards prescribed by the manufactures. In industry Sampling Plan (GASP). In this method many items can be
outgoing or incoming products are widely inspected to tested on the basis of few items are tested from the lot size
control the quality of the products which are essential of infinite. Hence this GASP elevate the ordinary plan to
activities in industries. Statistical quality control may inspect many items with multiple tester. Also it improves the
categorized into process control and product control. precision of the testing because various sampling units are
Product control plays a vital role when the product is in distributed to multiple testers. In a life test experiment, a
finished mode and helps to identify the reliable product and sample of size n is tested from a lot of products is put on the
eliminate the manufacturing errors.The statistical techniques test when the corresponding acceptance number is fixed
are usually employed to remove the defective products in with the test assigned time. Probability of rejecting a good
the production process as an offline product control lot is called the producer’s risk and probability of accepting
techniques at any stage of the manufacturing process as an a bad lot is called the consumer’s risk. Here the confidence
incoming raw materials, semi-finished products or a finished level is p* then the consumer risk will be 1-p*. The main
products can be tested.Product control is equally important objective of any acceptance sampling plan procedure is to
techniques however sampling plans for attributes and reduce both the risk simultaneously.
variables are widely studied, however sampling plan by
attributes is easy to perform in industrial shop floor Therefore many researchers prefers the GASP than any
conditions. Acceptance sampling plan is one of the other plans and have done their researchwith various
important techniques adopted in quality control towards distributions such as Aslam and Jun (2009)designed the
inspection and testing the sampling units in which decision group acceptance sampling plans based on the truncated life
about the lot can be made. In the acceptance sampling test when the life time of products follows an Gamma
procedure the life test plan is carried out when the quality Distribution and Weibull distributions. Rao (2011)
characteristics of the product is defined by its lifetime. In introduced a hybrid group acceptance sampling plans for
particular, truncated life test is adopted at which the test will lifetimes based on generalized exponential distribution and
terminated at a certain point of time in the sense that log logistic distributions. Aslam et al.(2010) introduced an
Here our interest in determining the number of group’s The Cumulative Distribution Function of the GEP is
‘g’ with the various values of acceptance number c and the given as
test termination time t0 are assumed to be specified. The 𝛼
1−𝑒 −λ+λexp(−βx)
operating procedure of GASP is as follows, 𝐹(𝑥; 𝜃) = ( )
1−𝑒 −λ
Step 1: Select a random sample of size n from a lot of size -----(2)
N and assign r number of units to each of g groups, so that
n= r*g Where θ (>0) = (α, β, λ), α is the shape parameter, β is
Step 2: Fix the acceptance number c and the experiment the scale parameter of the Exponential distribution and λ is
time t0. the Poisson parameter. When α = 1, Generalized
Step 3: Perform the experiment for the g groups Exponential Poisson reduces to Exponential Poisson
simultaneously and record the number of failures for each distribution. When α = 1 and λ→ 0, Exponential Poisson
group till the specified time t0 𝑡
reduces Exponential distribution with parameter β. Let x =β2
Step 4: Accept the lot if the number of failures from all the
groups together is smaller than or equal to c. , Then consider the CDF of the mean life product quality of
GEP distribution becomes,
Step 5: Reject the lot whenever number of failures more
than c as well as terminate the test before time t0. 𝑡
−λ+λexp(− )
𝛼
1−𝑒 β
The quality of the product is tested with GASP on the 𝐹(𝑥; 𝜃) = ( 1−𝑒 −λ
)
basis of the above procedure when few items are taken from t, λ,α, β > 0 -----(3)
an infinite lot is tested. Here, Group Acceptance Sampling
Plan (GASP) is studied under the proposed probability IV. TWO POINT ON THE OPERATING
distribution on the truncated life test under percentile as a CHARACTERISTIC CURVE
quality parameter when the life time of a product assumed to
follow the Generalized Exponential-Poisson distribution. The two important risks involved in the acceptance
sampling procedure is well known as producer’s risk and the
III. GENERALIZED EXPONENTIAL-POISSON consumer’s risk. The risk happening in the inspection
DISTRIBUTION procedures which exclusively depends with making of
Most of the probabilistic models are studied to describe wrong decision such as rejecting the good lot and accepting
the life time of data follows a certain life time distribution. the bad lot. Hence rejecting the good lot due to inherent
Here the failure time of an inspecting product may follows a nature of random sampling is the producer’s risk and
life time distribution is modelled using a statistical accepting the bad quality lot due to inherent nature of the
distribution. Kus(2007) introduced a two parameter random sample is known as consumer’s risk. Both the risk
distribution called Exponential-Poisson distribution. Later, have to be kept minimum for producing the reliable product.
Wagner Barreto-Souza and Francisco Cribari-Neto (2009) Hence it is considering the two levels Acceptable Reliability
derived a new distribution with three parameters known as Level (ARL) and Limiting Reliability Level (LRL) to
Generalized Exponential-Poisson (GEP) Distribution. This minimize the risks, which are obtained through Producer’
new distribution is a compounding of an exponential and a confidence level (1-α) and the consumer’s level β. The
Poisson distribution. In reality the failure item of a reliability sampling plan is an efficient one when both the
manufacturing product may not follow a particular risks are under control. Here, α ≤ 0.05 and β ≤ 0.10. Thus
distribution which may vary on the design parameters the probability of acceptance can be obtained for the
regardless on the underlying statistical distribution. The incoming quality using the following inequality,
failure rate of the distribution can be decreasing or 𝐿(𝑝1 ) ≥ 1 − 𝛼and𝐿(𝑝2 ) ≤ 𝛽
0.5
0.4
0.3
0.2
0.1
0
1 1.5 2 2.5 3 3.5 4 4.5 5
dq
The probability of acceptance can be regarded as a function of the deviation for the specified values are given to test the
percentile life. The function is called the Operating Characteristics curve of the proposed sampling plan is given. From this one
can obtain the minimum sample size and interested to find the probability of lot acceptance when the quality of the item is
sufficiently good under the study.
VIII. EXAMPLE
Consider an electronic toys manufacturing company Thus the proposed plan is performed as the testing with
wants to adopt the proposed sampling plan for life testing 6 tester (group) with 2 items in each group simultaneously at
the electronic toys. Suppose that the quality testing engineer the exact consumer’s risk 𝛽 = 0.0096. Accept the lot if no
wants to study the lifetime of a product which may follows more than 1 failure in each of all the groups occur or else
the Generalized Exponential-Poisson distribution, it is reject the lot. For this proposed electronic toy testing a
desired to design the GASP to test the actual lifetime is sample size of 12 items are tested with g=6 and r=2 (2*6, 1,
greater than 1000 hours when the test terminated at 800 0.8) with 12 items are tested and one may interested in
hours and 2 items on each tester with allowed number of finding the probability of acceptance for the method from
failures for each group is 2. It leads the ratio 𝑡⁄𝛽 = 0.8 with the Table-4 when the true lifetime of the product is greater
0 𝛽
c=1. From the Table-1theminimum number of groups for the than the specified mean lifetime 𝛽 ≥ β0 𝑜𝑟 β can be
0
consumer’s risk 0.01 is obtained. obtained.
Table 4: OC values of (n= 2*6 = 12, c= 1and 𝒕⁄𝜷 = 0.8)under GEP for p*=0.99
𝟎
The minimum mean ratio for this proposed plan efficient plan for studying the percentile life as a quality
referred from the Table-3 is 5.3226 reveals that the product parameter over the other sampling plans. Here the quality
will have an average life of 5 times of the specified average engineer can adopt the proposed sampling plan in the
life of 1000 hours with acceptable probability 0.99. manufacturing sector to reach a decision regarding either to
accept or not to accept the incoming / outgoing quality lots.
IX. CONCLUSIONS To ensure the life quality of the products the pattern of
failure can be occurred using the sampling distribution
This article provides a new statistical probability which protects both the producer and the consumer with
distribution named as Generalized Exponential-Poisson more precision than the specified average life. Suitable
distribution to test the quality of products when acceptance illustrations under electronic toy manufacturing are given
sampling for life test is studied. Numerical table are for ready made reference for the industrial shop floor
developed to obtain the minimum sample size, OC values conditions which provides better discrimination of accepting
and the minimum ratio values are given when producer’s good lots among minimum number of groups.
risk is fixed. The proposed plan was found to be a more
p* c r 𝑡⁄
𝛽0
0.5 0.6 0.7 0.8 0.9 1
0.75 0 2 2 2 1 1 1 1
0.75 1 2 3 3 2 2 2 1
0.75 2 2 5 4 4 3 3 2
0.75 3 2 6 6 5 4 4 3
0.75 4 2 8 7 5 5 5 4
0.90 0 2 3 3 2 2 2 1
0.90 1 2 4 4 3 3 3 2
0.90 2 2 6 5 4 4 4 3
0.90 3 2 8 6 5 5 4 4
0.90 4 2 9 8 7 6 5 5
0.95 0 2 3 3 2 2 2 2
0.95 1 2 5 4 4 3 3 3
0.95 2 2 7 6 5 4 4 4
0.95 3 2 9 7 6 6 5 5
0.95 4 2 11 9 8 7 6 6
0.99 0 2 5 4 3 3 2 2
0.99 1 2 7 6 5 4 4 4
0.99 2 2 9 8 6 6 6 5
0.99 3 2 11 9 8 7 7 6
0.99 4 2 13 11 9 8 7 7
Table 1: minimum number of groups (g) for mean life under GEP Distribution for GASP
0.9 0.5 6 0.5507 0.8298 0.9335 0.9717 0.9868 0.9934 0.9965 0.998
0.9 0.6 5 0.5513 0.7154 0.9265 0.9675 0.9844 0.992 0.9956 0.9975
0.9 0.7 4 0.6273 0.7672 0.9396 0.9729 0.9868 0.9931 0.9962 0.9978
0.9 0.8 4 0.5316 0.6873 0.9071 0.9558 0.9775 0.9879 0.9931 0.9959
0.9 0.9 4 0.4446 0.6077 0.8682 0.9337 0.965 0.9805 0.9887 0.9931
0.9 1 3 0.6734 0.7882 0.9374 0.9697 0.9843 0.9914 0.995 0.997
0.95 0.5 7 0.4266 0.7554 0.8981 0.9549 0.9786 0.9891 0.9941 0.9966
0.95 0.6 6 0.3965 0.5844 0.8766 0.9427 0.9717 0.9851 0.9918 0.9952
0.95 0.7 5 0.4277 0.6051 0.878 0.9423 0.9709 0.9844 0.9912 0.9948
0.95 0.8 4 0.5316 0.6873 0.9071 0.9558 0.9775 0.9879 0.9931 0.9959
0.95 0.9 4 0.4446 0.6077 0.8682 0.9337 0.965 0.9805 0.9887 0.9931
0.95 1 4 0.3684 0.5316 0.8243 0.9071 0.949 0.9708 0.9826 0.9892
0.99 0.5 9 0.2376 0.5994 0.8115 0.9104 0.9554 0.9765 0.987 0.9925
0.99 0.6 8 0.1835 0.3571 0.7540 0.8749 0.9344 0.9642 0.9795 0.9879
0.99 0.7 6 0.2743 0.4548 0.8043 0.9015 0.9484 0.9717 0.9838 0.9903
0.99 0.8 6 0.1847 0.3435 0.7221 0.8493 0.9165 0.9522 0.9717 0.9826
0.99 0.9 6 0.1221 0.2539 0.6359 0.7885 0.8766 0.9265 0.9554 0.9717
0.99 1 5 0.1784 0.3242 0.6878 0.821 0.8969 0.9386 0.9625 0.9761
Table 2: OC values for mean life under GEP distribution when c = 2 and r=2 for GASP