Information and Management Sciences

Volume 16, Number 1, pp.83-93, 2005

The Relation Between Cost and Yield Rate

in a JIT Production System
Jia-Chi Tsou

Jen-Ming Chen

National Central University

National Central University

R.O.C.

R.O.C.

Abstract
In this paper we consider production economic and cost related yield rate in a JIT (Just
in Time) system. Based on the traditional EPQ (Economic Production Quantity) model, we
establish the relation between cost and yield rate. The geometric programming approach is
applied in our model to solve this optimizing posynomials problem. A numerical example is
carried out to verify the proposed model. Through this paper, we connect the cost and yield
rate in a JIT production system. This study demonstrates the power and efficiency of the
geometric programming approach for the particular type of EPQ problem.

Keywords: JIT, EPQ, Geometric Programming, Cost Related Yield Rate.

1. Introduction
The economic production/order quantity (EPQ/EOQ) model proposed by Harris [1]
has been widely applied in different kinds of situation. Researchers tried to improve it
from different viewpoints. Among these, the quality of inventory products is one of the
major concerns in the past decade. In a traditional EPQ/EOQ model, there is no defect
on the quality of the inventory or production line. However, this assumption is not true
in the real world.
The relationship between quality and economics of production has been diversely
studied over the past decade. Porteus [2] is one of the earliest researchers who provided
a mathematical model which characterizes optimal simultaneous investment in setup cost
reduction and quality improvement. Rosenblatt and Lee [3] formulated and analyzed a
similar model that considers investment in process improvements. In a subsequent paper,
Lee and Rosenblatt [4] considered the use of process inspection during the production
run so that the shift to out-of-control state can be detected and restoration made earlier.
Received April 2004; Revised August 2003; Accepted May 2004.
Supported by ours.

84

Information and Management Sciences, Vol. 16, No. 1, March, 2005

Tapiero [5] linked optimal quality inspection policies and the resulting improvements in
manufacturing costs. Fine [6] used a stochastic dynamic programming model to characterize optimal inspection policies, and added the quality-based learning effects in the
model. Fine and Porteus [7] refined Porteus original work to allow smaller investments
over time with potential process improvement of random magnitude. Chand [8] validated
Porteus model when learning effect is present in setups and process quality. In a series of
papers, Cheng [9, 10] involved the production process reliability into a classic economic
order quantity model. Hong et al. [11] established the relationship between process quality and investment. Salameh and Jaber [12] considered a special production/inventory
situation where items, received or produced, are imperfect quality. Ram Ganeshan et
al. [13] brought Taguchis cost of poor quality into the economic production quantity
model, and linked lot-size determination to the loss-based quality accounting system.
In this paper, we consider a cost related yield rate in a JIT production system. JustIn-Time is a Japanese production philosophy developed in 1970s. It was first adopted
by Toyota manufacturing plants by Taiichi Ohno. The principle of JIT is to eliminate
sources of manufacturing wastes by getting right quantity of raw materials and producing
the right quantity of products in the right place at the right time. Since then, the JIT
production system has been diversely applied around the world.
In a JIT production system, the inventory will approach to zero and the holding cost
can be neglected. Under this assumption, the traditional EPQ model can be reduced
to the summation of the setup cost and production cost. Hence, we set production cost
and setup cost as the decision variables in our model. These two decision variables will
affect the yield rate of the production line.
Because of the nature of the relationship between production cost, setup cost and
yield rate, optimization of the target cost function by calculus often leads to a non-linear
equations, which is difficult to be solved in general optimization method. Optimizing
this kind of EPQ problem, we can call it an optimizing posynomials problem. To solve
this problem, the geometric programming approach can be applied, which provides an
efficient solution method for solving this kind of problem. Geometric programming (GP)
is a special type of convex optimization problem, which has been known and used since
the late 1960s [14]. The application of geometric programming approach in operations
research is not new. Papers like Federowicz and Mazumder [15], Jefferson [16], Passy
[17] and Petropoulos [18] all have applied this approach in their studies. However, there

The Relation Between Cost and Yield Rate in a JIT Production System

85

are only a few papers concerned with the solution of inventory problems by using the
geometric programming approach [9, 10].
In the next section, we introduce the mathematical model in this paper. Afterward,
the geometric programming approach is applied to solve the model. Then, a numerical
example is carried out to verify the model. Afterward, we perform sensitivity analysis.
We summarize and conclude the paper in the last section.

2. Model and Assumptions

Notation
P : unit production cost.
C: setup cost per production run.
Q: lot size.
R: annual demand in units.
y(C, P ): yield rate of products
T C(C, P ): total annual cost of a JIT production system. This includes production and
setup cost.
We consider a JIT production system without shortages or backorders. There is an
inspection site at the end of the production line and the inspection cost has been added
into the production cost, P , of each product. Since this is a JIT system, the holding cost
for inventory is zero (no inventory). C is the setup cost per production run; Q is the lot
size of each batch. The annual demand, R, for the product is constant. The yield rate
in this production line is y. The yield rate is the ratio of good quality products in the
production. In inspecting, the product out of the specification limits will be detected
and scraped. All the products within the specification limits will be kept, and be sold.
We make the following basic assumptions about our model:
1. No shortage is allowed.
2. All the variables are strictly positive.
3. The yield rate of products, y(C, P ), is directly related to setup cost and unit production cost according to the following power function:
y(C, P ) = aC b P c

where y(C, P ) 1.

(1)

86

Information and Management Sciences, Vol. 16, No. 1, March, 2005

The higher yield rate of a production line means more good products will be available, and the process capability is improved. According to Porteus [2] study, investing
in process quality will yield better output quality (fewer defects). Following researchers
also demonstrated this kind of relationship [2, 3, 4, 5, 8, 9, 10]. To increase the yield rate
of a production line, investment in improving the reliability of the production process is
necessary which gives rise to higher unit production cost and setup cost. We use equation
1 to present this kind of relationship. In equation 1, a, b and c are real numbers chosen
to provide the best fit of the yield rate function. To find the values of the coefficients, we
can use the regression analysis in the production line. We should record the yield rate of
production and the money that we spent on setup and production for a period. Then,
we can use these data to find the value of a, b and c.
To develop our model, let us start from the traditional EPQ model [19].
T C(Q) = P R +

CR HQ(p r)
+
Q
2p

(2)

Equation (2) is the traditional EPQ model. In the model, total annual cost T C(Q), is a
function of production quantity in each batch, Q. H is the holding cost per unit product
per year. p is the production rate, and r is the demand rate. In our case, we consider a
JIT production system. Hence, the inventory is zero, and the production rate equal to
the demand rate. We can reduce the traditional EPQ model to the new form.
T C(Q) = P R +

CR
Q

(3)

Furthermore, the lot size, Q, is constant, and the product quality is imperfect. We
assume that the yield rate of the production line depends on the setup cost and unit
production cost. We can further develop the model to equation (4).
T C(C, P ) =

CR
PR
+
y(C, P ) y(C, P )Q

which subject to
y(C, P ) 1.

(4)

Our object is to minimize the total annual cost, after substituting (1) into (4), it becomes
min T C(C, P ) =

CR
PR
+
b
c
aC P
aC b P c Q

subject to
y(C, P ) 1.

(5)

The Relation Between Cost and Yield Rate in a JIT Production System

87

3. Geometric Programming Solution

To solve the cost optimization problem above, we first define P , C and T C =
T C(C , P ) to denote the optimal value of P , C and the minimum total cost. We let
PR
R
= ( )C b P 1c ,
b
c
aC P
a
CR
R
X1 =
=(
)C 1b P c ,
b
c
aC P Q
aQ

X1 =

(6)
(7)

and
X3 = y(C, P ) = aC b P c .

(8)

For a giving set of positive real numbers, the arithmetic mean is greater than or equal
to the geometric mean. This statement can be written as the following equation:
 X 1  X 2

T C(C, P ) = X1 + X2

where
1 + 2 = 1;

and

1 , 2 > 0.

(9)

Based on the limitation of our initial assumption, y(C, P ) 1, We can define y(C, P ) as:
1 y(C, P ) = X3 X33
and
3 {(1, ) if X3 < 1;

(, )

if X1 = 1}.

(10)

Multiplying X33 into equation (9), we can get equation (11).

T C(C, P ) = X1 + X2

 X  1  X  2
1

X33 = (1 , 2 , 3 ).

(11)

In equation (11), T C(C, P ) and () are what we call the primal and dual functions.
(1 , 2 , 3 ) is a vector of dual variables, where 1 + 2 = 1; and 1 , 2 > 0 called the
normality condition [14].
Substituting (6), (7), (8) into (11), we can get
(1 , 2 , 3 ) =

 R 1 

a1

R 2 3 b1 +(1b)2 +b3 (1c)1 c2 +c3

(a) C
P
aQ2

(12)

We select the value of (1 , 2 , 3 ) depending on conditions of 1 + 2 = 1, 1 , 2 > 0

and 1 y(C, p) = X3 X33 . We also want the exponents of C, P are zero that makes

88

Information and Management Sciences, Vol. 16, No. 1, March, 2005

the right hand side of equation (12) independent of the decision variables. Hence, we
can get equations (13) and (14) as follows:
b1 + (1 b)2 + b3 = 0

(13)

(1 c)1 c2 + c3 = 0.

(14)

Solving (13) and (14) with normality condition, we have:

c
b+c
b
2 =
b+c
1 =

and
3 =

(15)
(16)

c+b1
.
b+c

(17)

With the value of (1 , 2 , 3 ), equation (11) can be reduced to:

T C(C, P ) (1 , 2 , 3 ) =

 R  
1

R 2
(a) 3
aQ2

a1
 R(b + c)  c  R(b + c)  b
c+b1
b+c
b+c
(a) b+c
=
ac
aQb

(18)

min T C(C, P ) = max (1 , 2 , 3 )(1 , 2 , 3 )

(19)

In equation (18), T C(C, P ) is equal to ( 1 , 2 , 3 ) when

Equation (5) has been transformed into the problem above that evaluates the value of
(1 , 2 , 3 ) and be solved by equation (18).
In the next step, we want to find the values of decision variables, C and P . The
optimal solution requires equality in equation (11) which is possible if and only if (X 1 /1 )
and (X2 /2 ) in equation (11) are equal, and X33 = 1. Hence in equation (19), T C(C, P )
will be minimal when (X1 /1 ) and (X2 /2 ) are equal, and X33 = 1. We define:
X 
1

X 
2

=G

and

X33 = 1.

(20)

Substituting (20) into equation (11), it can be obtained that

G1 +2 = G = (1 , 2 , 3 )

(21)

Combining equations (20) and (11), yields

X1 = 1 (1 , 2 , 3 ),

(22)

X2 = 2 (1 , 2 , 3 ).

(23)

The Relation Between Cost and Yield Rate in a JIT Production System

89

Substituting equation (6), (7) into (22), (23) we get linear equations as follows:
c 
b
c+b1
R(b + c)  b+c
c  R(b + c)  b+c
(a) b+c
a
b+c
ac
aQb
 b  R(b + c)  c  R(b + c)  b
 R 
c+b1
b+c
b+c
(a) b+c
C 1b P c =
aQ
b+c
ac
aQb

R

C b P 1c =

(24)
(25)

Solving equations (24) and (25), we can find the optimal values of the decision variables
C and P as

C =

1bc

(1 , 2 , 3 )Q1c ab1c cc
R(b + c)

(26)

c  1bc (1 , 2 , 3 )Q1c ab1c cc

P =
b+c
R(b + c)


(27)

4. An Example
In order to verify the proposed model, a numerical case will be used to demonstrate
the usefulness of our model. Horsy Industry Co. is a manufacturer of automotive seat
assy. There is an inspection site at the end of its JIT production line, and the inspection
cost is added into the production cost of each product.
The annual demand is 100000 units. The lot size is 500 units per batch. The setup cost
and production cost are decision variables which affect the yield rate of the production
line. The relationship between yield rate of products with production cost and setup
cost can be written as the following power function:
y(C, P ) = C 0.12 P 0.3
Annual demand in units, R = 100000 units/year.
Lot size, Q = 500 units/batch.
Setup cost per production run, C: decision variable.
Unit production cost, P : decision variable.
The decision variables are C and P whose optimal values can minimize the total
annual cost. Substituting the power function of yield rate into equation (5), we can
write our objective function.
min T C(C, P ) =
subject to y(C, P ) 1.

PR
C 0.12 P 0.3

CR
C 0.12 P 0.3 Q
(28)

90

Information and Management Sciences, Vol. 16, No. 1, March, 2005

Substituting the value of a, b and c into equation (18) we can find the minimal annual
cost as
min T C(C, P ) = (1 , 2 , 3 ) = 30810 dollars/year

(29)

Using equations (26) and (27), the optimal values of C and P can be found as
C = 44 dollars/batch,
P = 0.22 dollars/unit.

(30)
(31)

5. Sensitivity Analysis
In our model, the coefficients of the yield rate function, a, b and c have the most
frequency to change. In this section, we will perform the sensitivity analysis on these
parameters in previous numerical example. The mathematical software, Mathematica
4.1, is used to solve the problem.
As the coefficient a changes from 0.5 to 2 the minimal cost of the JIT production
system can be found by equation (18). The relationship between coefficient a and minimal
cost, T C(C, P ), has been drawn in Figure 1.

Figure 1. The behavior of the minimal cost to the change of coefficient a.

From Figure 1, we notice that the minimal cost increases with the coefficient a.
Then, we consider the change in coefficient b. As the coefficient b decreases from
0.01 to 0.5 the behavior of the minimal cost can be drawn in Figure 2.

The Relation Between Cost and Yield Rate in a JIT Production System

91

Figure 2. The behavior of the minimal cost to the change of coefficient b.

From Figure 2, we can find that the minimal cost decreases with the coefficient b.
Afterward, we consider the change in coefficient c. As the coefficient c decreases from
0.1 to 0.8 the behavior of the minimal cost can be drawn in Figure 3.

Figure 3. The behavior of the minimal cost to the change of coefficient c.

From Figure 3, we notice that the minimal cost increases as the coefficient c decreases
from 0.1 to 0.8.

6. Summary and Conclusion

In this paper we have considered a cost related yield rate in a JIT production system.
We have proposed a power function to model the relationship between cost and yield
rate. The decision variables in our model include the production cost and the setup
cost. The geometric programming approach has been applied to solve this optimizing
posynomials. Through this paper, we have demonstrated the power and efficiency of
geometric programming approach for this particular type of EPQ problem. We also

92

Information and Management Sciences, Vol. 16, No. 1, March, 2005

connected the cost and the yield rate in a JIT production system.
Acknowledgements
We would like to thank the editors and the two anonymous referees for their valuable
and constructive comments, which have led to a significant improvement on this paper.

References
[1] Harris, F., How many parts to make at once. Factory, The Magazine of Management, Vol.10, pp.135136; 152, 1913.
[2] Porteus, E. L., Optimal lot-sizing process quality improvement and setup cost reduction, Operations
Research, Vol.34, pp.137-144, 1986.
[3] Rosenblat, M. J. and Lee, H. L., Economic production cycles with imperfect production processes,
IIE Transactions, Vol.18, pp.48-55, 1986.
[4] Lee, H. L. and Rosenblatt, M. J., Simultaneous determination of production cycles and inspection
schedules in a produc tion system, Management Science, Vol.33, pp.1125-1137, 1987.
[5] Tapiero, C., Production learning and quality control, IIE Transactions, Vol.19, pp.362-370, 1987.
[6] Fine, C. H., A quality control model with Learning Effects, Operations Research, Vol.36, pp.437-444,
1988.
[7] Fine, C. H. and Porteus, E. L., Dynamic process improvement, Operations Research, 37, pp.580-591,
1989.
[8] Chand, S., Lot sizes and setup frequency with learning in setups and process quality, European
Journal of Operations Research, Vol.42, pp.190-202, 1989.
[9] Cheng, T. C. E., An economic production quantity model with flexibility and reliability considerations,
European Journal of Operations Research, Vol.39, pp.174-179, 1989.
[10] Cheng, T. C. E., Economic order quantity model with demand-dependent unit production cost and
imperfect production processes, IIE Transactions, 23, pp.23-28, 1991.
[11] Hong, J., Xu, S. H. and Hayya, J. C., Process quality improvement and setup reduction in dynamic
lot-sizing, International Journal of Production Research, Vol.31, pp.2693-2708, 1993.
[12] Salameh, M. K. and Jaber, M. Y., Economic production quantity model for items with imperfect
quality, International Journal of Production Economics, Vol.64, pp.59-64, 2000.
[13] Ram Ganeshan, Shailesh Kulkarni, and Tonya Boone, Production economics and process quality: a
Taguchi perspective, International Journal of Production Economics, Vol.71 , No.1-3, p.343, 2001.
[14] Federowicz, A. J. and Mazumdar, M., Use of geometric programming to maximize reliability achieved
by redundancy, Operations Research, Vol.16, pp.948-954, 1968.
[15] Jefferson, T., Geometric programming with an application to transportation planning, Ph.D. Thesis,
Northwestern University, Illinois, 1972.
[16] Passy, U., Nonlinear assignment problems treated by geometric programming, Operations Research,
Vol.19, pp.1675-1690, 1971.
[17] Petropoulos, P. G., Optimal selection of machine rate variable by geometric programming, International Journal of Production Research, Vol.11, pp.305-312, 1973.
[18] Tersine, R. J., Principles of Inventory and Materials Management, 4/e, Prentice-Hall, New Jersey,
1994.
[19] Duffin, R. J., Peterson, E. L. and Zener, C., Geometric Programming, John Wiley and Sons, New
York, 1967.

The Relation Between Cost and Yield Rate in a JIT Production System

93

Authors Information
Jia-Chi Tsou is an Assistant Professor in the Department of Business Administration at the Chung Kuo
Institute of Technology, Taiwan. He received the Ph.D. degree in Industrial Management from the National Central University, Taiwan. He graduated with an MBA in Entrepreneurship from the University
of Liverpool, UK. He also gained an MS and a BS in Mechanical Engineering at the National Central
University, Taiwan. His current research interests include quality management and yield management
in the automotive industry. Dr. Tsou is an IRCA (International Register of Certificated Auditors)
registered QMS Lead Assessor. He worked as 6-Sigma Master Black Belt at the Ford Motor Company.
Institute of Industrial Management, National Central University, Chung-Li, Taiwan 32054.
E-mail: jtsou.tw@yahoo.com.tw

Tel : (+886-3) 453-0536.

Jen-Ming Chen is a professor in the Institute of Industrial Management at the National Central University
(Taiwan). He received a B.S. in Industrial Management Science from the National Cheng Kung University
(Taiwan) in 1983, an M.S. in Industrial Engineering from the University of Arizona in 1988, and a Ph.D.
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 organizations, including Informs, DSI, and IIE. Dr. Chen is the recipient of
the George B. Dantzig Dissertation Award from the Informs and the recipient of the IIE Doctorial
Dissertation Award, both in 1994.
Institute of Industrial Management, National Central University, Chung-Li, Taiwan 32054.
E-mail: jmchen@mgt.ncu.edu.tw

Tel: (+886-3) 422-7151 ext. 6553