0% found this document useful (0 votes)
35 views10 pages

Optimal Selection of Process Mean For A Stochastic Inventory Model

Optimal selection of process mean for a stochastic inventory model

Uploaded by

senthilnathan25n
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
35 views10 pages

Optimal Selection of Process Mean For A Stochastic Inventory Model

Optimal selection of process mean for a stochastic inventory model

Uploaded by

senthilnathan25n
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

European Journal of Operational Research 226 (2013) 481–490

Contents lists available at SciVerse ScienceDirect

European Journal of Operational Research


journal homepage: www.elsevier.com/locate/ejor

Stochastics and Statistics

Optimal selection of process mean for a stochastic inventory model


M.A. Darwish ⇑, F. Abdulmalek, M. Alkhedher
Department of Industrial and Management Systems Engineering, College of Engineering and Petroleum, Kuwait University, P.O. Box 5969, 13060 Safat, Kuwait

a r t i c l e i n f o a b s t r a c t

Article history: It is very common to assume deterministic demand in the literature of integrated targeting – inventory
Received 19 December 2011 models. However, if variability in demand is high, there may be significant disruptions from using the
Accepted 15 November 2012 deterministic solution in probabilistic environment. Thus, the model would not be applicable to real
Available online 29 November 2012
world situations and adjustment must be made. The purpose of this paper is to develop a model for inte-
grated targeting – inventory problem when the demand is a random variable. In particular, the proposed
Keywords: model jointly determines the optimal process mean, lot size and reorder point in (Q, R) continuous review
Quality control
model. In order to investigate the effect of uncertainty in demand, the proposed model is compared with
Targeting problem
Production
three baseline cases. The first of which considers a hierarchical model where the producer determines the
Demand uncertainty process mean and lot-sizing decisions separately. This hierarchical model is used to show the effect of
integrating the process targeting with production/inventory decisions. Another baseline case is the deter-
ministic demand case which is used to show the effect of variation in demand on the optimal solution.
The last baseline case is for the situation where the variation in the filling amount is negligible. This case
demonstrates the sensitivity of the total cost with respect to the variation in the process output. Also, a
procedure is developed to determine the optimal solution for the proposed models. Empirical results
show that ignoring randomness in the demand pattern leads to underestimating the expected total cost.
Moreover, the results indicate that performance of a process can be improved significantly by reducing its
variation.
Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction recycling and reprocessing (Al-Sultan and Pulak, 2000). Thus, the
targeting problem is a trade-off between material cost and the cost
The problem of selecting the optimum process mean (targeting associated with producing nonconforming items.
problem) is one of the areas in economics of quality control, which Traditionally, the process mean selection and production/inven-
has received a lot of interest from researchers in the recent times. tory decisions are determined separately. However, the choice of
Process targeting is often a difficult decision in production process mean affects the probability that a given produced item
processes such as canning/filling, metal plating, grinding, glass is nonconforming. Hence, the process mean determines the
industry, fiber industry, and steel industry (Park et al., 2011; Shao production yield rate which, in turn, influences other important
et al., 2000; Hariga and Al-Fawzan, 2005). Usually, producers set production and inventory decisions, in particular, the production
specification limits on a quality performance measure (weight, vol- lot size and reorder point. Consequently, the process mean and
ume, concentration, thickness, length, etc.) and an item is classified production/inventory decisions should be determined jointly in or-
as conforming when the quality performance measure is within der to control the total cost associated with production processes.
specification limits. Otherwise, the item is classified as noncon- This integration leads to higher conforming and yield rates,
forming and may be sold at a reduced price, reprocessed or reduction in scrap or reprocessing cost, minimal loss to customer
scrapped (Roan et al., 2000). In USA, federal agencies examined due to the deviation from the optimum target value, as well as,
the practice of process mean selection and reported that it is a and perhaps most importantly, providing better products at
common trend that many manufacturers select high process mean reduced cost for customers (Darwish, 2009).
in order to conform to specifications. This strategy leads to a ‘‘give The organization of the paper is as follows; motivation and con-
away’’ cost (Roan et al., 2000). On the other hand, a tight process tribution is discussed in the next section. In Section 3, we review
setting will have less production cost, but high costs of rejection, the literature. Model assumptions and notation are introduced in
Section 4 followed by model development. In Section 6, solution
method is outlined. Then we investigate some baseline cases in
⇑ Corresponding author. Tel.: +965 38601163; fax: +965 38604426. Section 7. In Section 8, we present the empirical results and Sec-
E-mail address: [email protected] (M.A. Darwish). tion 9 concludes the paper.

0377-2217/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.ejor.2012.11.022
482 M.A. Darwish et al. / European Journal of Operational Research 226 (2013) 481–490

2. Motivation and contribution the amount of material used. Golhar (1987) considered the optimal
determination of the optimum process mean when conforming
It is very common to make simplifying assumptions in model- items are sold at a fixed price and nonconforming cans are emptied
ing real-world phenomenon. This is important in order to render and refilled at a reprocessing cost. Boucher and Jafari (1991) ex-
the mathematics tractable. On the other hand, we cannot make tended this line of research by introducing a sampling plan as op-
too many simplifying assumptions, for then our conclusions, ob- posed to 100% inspection; in this model the effect of single
tained from the mathematical model would not be applicable to sampling plan on the optimal set point of a filling operation is
real world situations (Ross, 2007). In integrated targeting-inven- examined. Al-Sultan (1994) extended the model developed by
tory models, it is usually assumed that demand is deterministic. Boucher and Jafari (1991) for two machines in series using a sampling
In reality however, deterministic models are only approximations inspection plan. Tang and Lo (1993), Lee and Jang (1997) and Lee
and the goodness of these approximations depends on the degree et al. (2001) used a surrogate variable in inspection and determined
of uncertainty in the demand pattern (Nahmias, 2005). Thus, there the optimum process mean and the screening limits. Duffuaa and
exists a need to develop a model that addresses the limitations Siddiqui (2003) discussed a model for a multi class screening when
imposed by this assumption. measurement error existed and Darwish and Duffuaa (2010) devel-
Unlike the models developed in the integrated targeting-inven- oped a model that determines the optimal process mean and sam-
tory literature, in this paper, the targeting problem is integrated pling plan’s parameters. The model maximizes producer expected
with inventory decisions when randomness is introduced into profit while protecting the consumer through a constraint on the
the demand pattern. The proposed model simultaneously deter- probability of accepting lots with low incoming quality. Moreover,
mines the optimal decisions regarding process mean, production Elsayed and Chen (1993) determined optimum levels of process
lot size, and reorder point in (Q, R) continuous review model. parameters for items with multiple characteristics. Arcelus and
In general, there are three levels of decisions related to produc- Rahim (1994) provided joint optimal settings for variable and attri-
tion processes: (1) long-term (strategic) (2) Intermediate-term bute target means. Chen and Chung (1996) considered a model for
(tactical), and (3) short-term (operational). Usually, these decision determining the optimal process mean and measuring precision le-
levels are optimized independently due to the nature of their vary- vel for a production process. Further, Hong and Elsayed (1999)
ing time horizon. However, these decision levels affect and are af- developed a model for jointly determining the optimum process
fected by each other. For instance, the selection of process mean mean and cutoff value on the observed characteristic when mea-
(which is operational decision) affects the conformance rate which surement error was present. Further, Pfeifer (1999) used electronic
determines number of conforming/nonconforming items pro- spreadsheet program to find the solution. Hong et al. (1999) exam-
duced. Therefore, the determination of production lot size (which ined the case when there are several markets with different price/
is tactical decision) should compensate for the nonconforming cost structures. Recently, the optimal target value and variance
items produced during a production run. Consequently, integrating using Taguchi loss function is found by Rahim and Shaibu (2000)
the process mean targeting and production planning and inventory and Lee et al. (2004). The effect of variance reduction and process
control decisions is expected to reduce the total cost. Another capability on the optimal target value is explored by Kim et al.
important aspect of this integration is that it jointly addresses (2000). Moreover, Williams et al. (2000) examined process
quality control issues through process targeting and production improvement alternatives for a container-filling process. They con-
planning and inventory control issues. sidered reducing the process setup cost, reducing the frequency of
After deriving the proposed model, it is compared with its the out-of-control signals and reducing the process variation. Teer-
deterministic counterpart by studying three baseline cases. The avaraprug and Cho (2002) designed the most economical process
first of which considers a hierarchical model where the producer target levels for multiple performance variables. Rahim et al.,
determines the process mean and lot-sizing decisions separately. 2002) obtained the optimum target mean and variance for a contin-
This hierarchical model is used to show the effect of integrating uous production process. Lee and Elsayed (2002) determined the
the process targeting with production/inventory decisions. The optimum process mean and screening limits of a surrogate variable
other baseline case is for deterministic demand which is used to associated with product quality under a two-stage screening proce-
explore the effect of variation in demand on the optimal solution. dure. Shao et al. (2000) developed strategies for determining the
Then we use the baseline case where variation in the filling amount optimal process mean for industrial processes when rejected goods
can be neglected. This case shows the sensitivity of the total cost can be held and sold to other customers in the same market at a la-
with respect to the variation in the process output. The proposed ter time. Recently, Bowling et al. (2004) studied the targeting prob-
models show that ignoring randomness in demand may lead to lem in the context of multi-stage serial production process. Lee
underestimating the expected total cost significantly. Furthermore, et al., 2005) and Hong et al. (2006) investigated the optimal process
a procedure is devised to find the optimal solution of the model. mean for variety of production processes. Al-Sultan and Pulak
(1997) presented a model for finding the optimal mean of a filling
process under rectifying inspection. Rahim and Al-Sultan (2000)
3. Literature review considered the problem of joint determination of the optimal target
mean and variance. Other researchers including Gibra (1974), Arce-
The optimal determination of process mean has been investi- lus and Banerjee (1985), Rahim and Banerjee (1988), and Al-Sultan
gated and discussed for more than 60 years. Springer (1951) was and Al-Fawzan (1997), considered the targeting problem when the
the first to consider the problem of process targeting, in which process mean is time dependent. A reverse programming routine
the process mean that minimizes the total cost is obtained. Bettes that identifies the relationship between the process mean and the
(1962) presented a similar model by determining the optimal settings within an experimental factor space was established by
process mean and upper specification limit simultaneously. In this Goethals and Cho (2011). Chen and Lai (2007) developed a model
model, the nonconforming items are scrapped with no salvage va- that finds the economic process mean based on quadratic quality
lue. This model is extended by Hunter and Kartha (1977) by consid- loss function and rectifying inspection plan. Also, Based on quality
ering the problem of selecting the optimum target value when loss function, Chen and Kao (2009) determined the optimal process
nonconforming items are sold at a secondary market. Then, Bisg- mean and screening limits. Hong and Cho (2007) found the optimal
aard et al. (1984) modified Hunter and Kartha’s model to a situation process mean and tolerance limits with measurement errors under
where nonconforming products are sold at a price proportional to multi-decision alternatives.
M.A. Darwish et al. / European Journal of Operational Research 226 (2013) 481–490 483

Integrating the targeting problem with inventory/production tinuously and that the policy is to start producing a lot of sizeQ
decisions has also attracted the attention of many researchers for when the inventory level drops to a reorder point R. It is worth
years. For example, Gong et al. (1988) developed integrated target- mentioning that the probability of producing a conforming item
ing-inventory model where the process mean is constant during (p), yield rate (k), and lot size (Q) are not random variables because
the production cycle. This model is generalized by Al-Fawzan and they depend on process mean (l) which is not a random variable.
Hariga (2002) by considering time-dependent process mean. Roan Assume that the demand in any given interval of time is a random
et al. (2000) integrated the issues of production lot size and raw variable whose probability distribution is stationary. Let s be the
material procurement policy with the targeting problem. Hariga production lead time and Y be a random variable represents the
and Al-Fawzan (2005) determined simultaneously the optimal pro- lead time demand with probability distribution f(y). In addition,
duction lot size and process mean for container-filling processes let A denote the fixed setup cost and h denote the cost of carrying
with multiple markets. Lee et al. (2007) considered the targeting- a finished item for one unit of time. In developing the proposed
inventory problem for a production process where multiple prod- model, the following assumptions are used:
ucts are processed. Moreover, Darwish (2004a) and Darwish
(2004b) integrated the targeting problem with single-vendor sin- (1) Shortage is backordered, this situation corresponds to a cap-
gle-buyer problem. He assumed an equal size shipment policy tive market and at the wholesale-retail link of some distribu-
and nonconforming items are scraped with no salvage value. Chen tion systems (Silver et al., 1998, p. 234).
and Lai (2007) developed a model that finds the optimal process (2) There is never more than a single order outstanding and the
mean, specification limits, and manufacturing quantity under rec- mean rate of demand (E[K]) is constant over infinite hori-
tifying inspection plan. Chen and Khoo (2009) addressed targeting- zon. It is shown in (Hadley and Whitin, 1963, p. 178) that
inventory for serial production systems under quality loss and rec- the expected number of orders per year is E[K]/Q.
tifying inspection plan. Also, Darwish (2009) developed an inte- (3) The expected number of orders per year and the expected
grated targeting-inventory model for a two-layer supply chain. In number of backorders incurred per unit time are both inde-
this model, unequal size shipment policy is used and nonconform- pendent of the demand pattern provided that the stochastic
ing containers are reprocessed at a certain cost. Recently, Alkhed- process generating demand does not change with time. In
her and Darwish (2013) developed a model that determines the other words, using typical cycle approach is valid in this
optimal process mean for a stochastic inventory model under ser- case, that is
vice level constraint. Park et al., 2011) established a profit model E½K
that determined the optimal common process mean and screening Expected cost per unit time ¼ Expected cost per cycle 
Q
limits for a production process with multiple products.
(4) The demand pattern is modeled by a normal probability dis-
4. Model assumptions and notation tribution. In fact, this is a common assumption in stochastic
inventory models. This is because empirically the normal
Consider a manufacturer that orders raw material from a sup- distribution is adequate for most situations (Silver et al.,
plier and uses it to produce a product. The quality characteristic X 1998, p. 273).
is a measure of the amount of raw material used in the manufactur-
ing of the end item. It is assumed that the performance variable X is The following notation will be used in developing the proposed
a ‘‘large-is-better’’ variable. That is an item is classified as conform- model:
ing if X P L where L is a lower specification limit, otherwise, the
item is rejected and scrapped with no salvage value. The practice
X A random variable represents the amount of raw
of setting only a lower specification limit on a quality characteristic
material an item receives
can be found in many industries. For example, in glass manufactur-
l Process mean
ing, raw materials are melted and rolled into sheets with certain
r Standard deviation of the amount of raw material
thickness. It is common that customers set a lower specification
used in an item
limit for the thickness of the sheet and any excess thickness is
L Lower specification limit
acceptable (Al-Sultan and Pulak, 2000). Another example is gold
K A random variable represents the demand per unit
plating of silver which is used in the manufacture of jewelry. The
time
time until the surface of an item is tarnished depends on the thick-
E[K] Expected value of the demand per unit time
ness of the gold layer. Evidently, the expectation of a customer is
rK Standard deviation of the demand per unit time
exceeded when the thickness of gold layer is larger. In this case,
D Constant component of lead time
the quality of the item is defined in terms of a lower specification
Q Lot size
limit. In the literature, one can find more practical examples for
R Reorder point
industries that require only lower specification limit on a quality
r Production rate
characteristic (for instance, steel galvanization industry (Shao
p Probability of producing a conforming item
et al., 2000) and fiber industry (Hariga and Al-Fawzan, 2005)).
k Yield rate of process (k = rp)
It is also assumed that the performance variable X follows a nor-
s Production lead time (s = Q/k)
mal distribution with an adjustable mean, l, and a constant vari-
Y A random variable represents the lead time demand
ance, r2. We have to mention that l is not time dependent, that
(Y = K(s + D))
is, drift in process mean is not considered. Let / denote the stan-
f(y)dy Probability that the lead time demand is between y
dard normal probability density function, thus the probability of
and y + dy
producing a conforming item is given by
E[Y] Expected value of lead time demand
  Z
Ll 1 (E[Y] = E[K](s + D))
p ¼ P½X P L ¼ P Z P ¼ /ðzÞdz rY Standard deviation of lead time demand
r Ll
r
/ Standard probability density function of the normal
and the yield rate of the process isk = rp, where r is the production
rate. Suppose that the inventory level of the item is monitored con- (continued on next page)
484 M.A. Darwish et al. / European Journal of Operational Research 226 (2013) 481–490

distribution
U Standard cumulative probability function of the
normal distribution
A Setup cost

Net inventory level of


Ar Ordering cost of raw material

finished items, I(t)


h Holding cost for the vendor per item per unit time λ
Q
hr Holding cost for the raw material per unit per unit
R
time
p Fixed penalty cost
b Fixed production cost (b > 0)
a Value-added factor (a P 1)
c Unit material cost S
T Inventory cycle length
Δ τ
NL(u) Normal loss function Time
S A random variable represents safety stock T
s Expected safety stock (s = E[S])
N A random variable represents number of shortages

Inventory level of raw


n Expected number of shortages (n = E[N])
TC Expected total cost per unit time of the integrated

material, Ir(t)
model -rμ
TCHR Expected total cost per unit time of the hierarchical
model
TCPP Expected total cost per unit time of the perfect filling
process Time
TCDD Expected total cost per unit time of the deterministic
demand model Fig. 1. Evolution of inventory over time for finished items and raw material.

   
Q E½K
HC ¼ h 1þ þs ð2Þ
2 k
5. Model development

In this section, we present an integrated targeting-inventory We have to point out that the computation of the expected safety
model where the process mean and inventory decisions are deter- stock s in Eq. (2) depends on the model assumption regarding
mined jointly. In deriving the costs in this model, we use typical whether the shortage is eventually satisfied, partially satisfied or
cycle approach as indicated in previous section. Thus, the costs lost. When the shortage is completely backordered, the safety stock
associated with this model are itemized as follows: S, which is a random variable (because it is a function of Y), is unre-
stricted in sign, that is S = R  Y. Hence, the expected value of the
1. Setup Cost: A cost A is incurred each production run. On the safety stock is R – E[Y], where E[Y] is expected lead time demand
average, the number of setups per unit time is E[K]/Q. There- and is equal to (D + Q/k)E[K]. Thus,
 
fore, the setup cost per unit time is Q
s ¼ E½S ¼ R  D þ E½K ð3Þ
AE½K k
SC ¼ ð1Þ
Q
Hence, the holding cost per unit time can be found from Eqs. (2) and
2. Holding Cost: Fig. 1 depicts the evolution of net inventory over (3) as follows
   
time. In depletion period [0, T], the expected net inventory at Q E½K
HC ¼ h 1 þ R  DE½K ð4Þ
the beginning of a cycle is S + Q and S at the end of a cycle, 2 k
where S is the safety stock. Note that these are also the expected 3. Shortage Cost: A shortage can occur only if the demand during a
values of the on hand inventory when the expected number of lead time exceeds the reorder point. Thus, the shortage quan-
backorders can be neglected. In this situation, since the tity, N, at the end of a cycle is a random variable given by
expected demand rate is constant, the expected on hand inven-
tory changes linearly from S + Q to S, and is Q/2 + S, hence the

Y R y>R
expected inventory cost per unit time is h(Q/2 + s) where s is N¼
0 y6R
the expected safety stock. We have to indicate that this approx-
imation can be found in many textbooks (for example, Hadley and the expected shortage quantity per cycle, n(R), is as follows
and Whitin (1963)) and it is included here so that the paper is Z 1  
R  E½Y
self-contained. This approximation depends on the assumption nðRÞ ¼ ðy  RÞf ðyÞdy ¼ rY NL ð5Þ
R rY
that there is no overshooting (crossing the reorder point) at the
beginning of inventory cycle (see for example Montgomery and where NL(u) is the standardized normal loss function and is given
Johnson, 1974, p. 60). In addition, in the inventory build-up per- by
iod [T  s, T], the inventory level increases at a rate k, building Z 1
inventory, until Q conforming units are produced which then NLðuÞ ¼ ðz  uÞ/ðzÞdz ¼ /ðuÞ  uð1  UðuÞÞ
u
will be used to fulfill demand (Darwish et al., 2012). The
expected inventory cost per unit time in the inventory build- The function NL(u) can be solved numerically and its tables can be
up period during production is hQE[K]/2k. Hence, the total found in many texts, for example (Nahmias, 2005). Therefore, the
expected holding cost per unit time is given by expected shortage cost per unit time can be setout as
M.A. Darwish et al. / European Journal of Operational Research 226 (2013) 481–490 485

 
prY E½K R  E½Y tive is to find the values of Q, R and l that minimize TC(Q, R, l). It is
SHC ¼ NL ð6Þ
Q rY important to indicate that the quantities p and k in Eq. (10) are not
constants but function of the process mean l.
4. Inventory Control Costs Related to Raw Material: The costs under The value of Q which minimizes TC (obtained by differentiating
this category include raw material ordering and holding costs. TC with respect to Q and setting the result to 0) is
When the producer orders raw material from a supplier, he vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
u RE½Y
incurs an ordering cost Ar. In addition, since the vendor is u A þ A þ pr NL
u r Y r Y
charged for this raw material as soon as it is stocked in his Q ¼ t2E½K  ð11Þ
warehouse, he incurs a holding cost for this raw material. The h 1  E½kK þ hr lkpE½K
inventory profile of the raw material at the vendor premises
as a function of time is depicted in Fig. 1. It can be easily verified Similarly, the optimal value of R is given by
that the inventory of raw material at a given time t (t 6 s) is
 
R  E½Y hQ
U ¼1 ð12Þ
Ir ðtÞ ¼ r lðs  tÞ ð7Þ rY pE½K

Using Eq. (7) and s = Q/k, the holding cost of raw material per cycle where U RE½Y
is the probability that the lead time demand is less
rY
is
than the reorder point. The procedure is started by using Q0 = EPQ,
Z s pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
hr l 2 where EPQ ¼ 2E½KA=hð1  E½K=rÞ. We then find R0 from Eq.
HC r ¼ hr Ir ðtÞdt ¼ Q
0 2kp (12). The values Q0 and R0 are used in the model in Eq. (10) to find
Hence, the average total cost related to inventory control of raw the optimal value of process mean l0. That value of l0 is used to
material per unit time, CR, is given by compute p and k which are substituted into Eq. (11) to find Q1,
which is then substituted into Eq. (12) to find R1, and so on. The
Ar E½K hr lE½K computations should be continued until the difference in two suc-
CR ¼ þ Q ð8Þ
Q 2kp cessive values of TC(Q, R, l) are within a certain level of accuracy
d. Convergence generally occurs within six iterations. The statement
5. Direct Production Cost: This category involves cost of producing
of the algorithm is as follows:
items. As in Al-Fawzan and Hariga (2002), Roan et al. (2000))
and Gong et al. (1988)), we assume that the direct production pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Step 1:
Let i = 1, TC0 ? 1, and Q 1 ¼ 2E½KA=hð1  E½K=rÞ.
cost is a linear function of the item’s material cost and is given hQ i
Step 2:
If pE½K P 1, find Ri from Eq. (12). Otherwise, Ri = E[Y].
by g(X) = b + acX, where a P 1. We have to indicate that a is
Step Determine the value of li which minimizes TC in Eq. (10).
3:
included in g(X) so that the model is more general, however, a
Step 4:
Let TCi = TC.
can be set to 1 whenever appropriate. Consequently, the per-
Rs Step If TCi < TCi1, set Q⁄ = Qi, R⁄ = Ri, and l⁄ = li. Otherwise, go
5:
cycle cost of producing Q items, DPC, is r 0 gðlÞdt which yields
to Step 6.
Q Step 6: If jTCi  TCi1j > d, then increase i by 1, go to Step 7. Other-
DPC ¼ ðb þ claÞ wise, go to Step 8.
p
Step 7: Determine Qi from Eq. (11) and go to Step 2.
As shown in Fig. 1, the amount of raw material in the vendor’s ware- Step 8: Stop.
house at t = 0 is Ir(0). Thus, the amount of raw material that the ven-
dor receives from the supplier in one cycle is Ir(0), it follows therefor It is important to point out that in Step 3, a one-dimensional di-
that the acquisition cost,AC, per cycle is cIr(0) and can be setout as rect search procedure, such as the uniform, Dichotomous, or gold-
follows en-section search method, can be used to find the optimal process
mean. In this paper, interval halving method is utilized because of
cQ l
AC ¼ its simplicity and efficiency (Bazaraa et al., 1993). The range for the
p
search for l is [L, L + zr], where z is a predetermined real number.
Hence, the total production cost per unit time is as follows: The reason L is used as the lower bound is that, when l = L, the
probability that the process produces a conforming item is 50%,
E½K E½K
CP ¼ ðAC þ DPCÞ ¼ ðb þ cða þ 1ÞlÞ ð9Þ which is very low in most realistic applications (Roan et al., 2000).
Q p
The expected total cost is the sum of SC, HC, SHC, CR, and CP in Eqs. 1, 7. Baseline Cases
4, 6, 8 and 9 respectively and is established as follows
   
ðA þ Ar ÞE½K Q E½K 7.1. Hierarchical model
TCðQ; R; lÞ ¼ þh 1 þ R  DE½K
Q 2 k
  In order to study the advantages of using the integrated model
prY E½K R  E½Y hr lE½K E½K
þ NL þ Qþ ðb þ cða þ 1ÞlÞ presented in (10), a hierarchical model is developed where the pro-
Q rY 2kp p
ð10Þ ducer sets the process mean independently of lot sizing decisions.
This hierarchical model is used as a baseline case to show the im-
pact of integrating the process mean with the production/inven-
6. Solution method tory decisions.

Usually models for targeting problem are highly nonlinear and 7.1.1. Determination of optimal process mean for the hierarchical
difficult to solve analytically and the proposed model is not an model
exception. This difficulty is mainly due to the probability of confor- When the process mean is set in isolation of other decisions, the
mance (p) which depends on the decision variable l. In this sec- probability of producing a conforming item increases for high pro-
tion, a brief description of the computational method is outlined. cess mean and consequently the loss due to scrap is reduced. On
The decision variables in the model presented in Eq. (10) are the the other hand, a large process mean will increase the raw material
process mean l, the lot size Q, and the reorder point R. The objec- requirement for producing an item. Thus, the optimal process
486 M.A. Darwish et al. / European Journal of Operational Research 226 (2013) 481–490

   
mean is a balance point between raw material costs and the cost of ðA þ Ar ÞE½K Q E½K
TC PP ðQ ;RÞ ¼ þh 1 þR
producing non-conforming items. Thus, the costs included in this Q 2 r
 
model are the cost of producing items, and raw material costs. prY E½K R  E½Y hr LE½K
These costs are presented in Section 5, therein; we referred to their þ NL þ Q þ E½Kðb þ cða þ 1ÞLÞ
Q rY 2r
sum as the total production costs CP. Using Eq. (9), the average total
cost per unit time is vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
u RE½Y
u A þ A þ pr NL
u r Y rY
E½K Q ¼ t2E½K  ð18Þ
C 1 ðlÞ ¼ ðb þ cða þ 1ÞlÞ ð13Þ h 1  E½rK þ hr LE½ K
p r

The decision variable in C1 is the process mean l and its optimal va- Similarly, the optimal value of R is given by
lue is determined numerically because it is not easy to find it in
 
R  E½Y hQ
closed form expression (note that the quantity p, in Eq. (13), is U ¼1 ð19Þ
rY pE½K
not a constant, rather, it is a function of l). After determining opti-
mal value of l, the production lot size and reorder point are found Eqs. (18) and (19) can be solved iteratively to find the optimal lot
accordingly, this is shown in the next section. size and reorder point.

7.3. Deterministic demand


7.1.2. Determination of optimal lot size and reorder point for the
hierarchical model In this case, the randomness in the demand pattern is ignored,
The costs in this case are setup cost, holding cost of finished that is, rK = 0. The expected total cost per unit time is obtained
items, and raw material ordering and holding costs. For a fixed va- by substituting R = rY = 0 into expression (10)
lue of l, one can use Eqs. 1, 4, 6, and 8 to find the expected total  
cost per unit time C2 as follows ðA þ Ar ÞE½K hQ E½K hr lE½K
TC DD ðQ; lÞ ¼ þ 1 þ Q
Q 2 k 2kp
   
ðA þ Ar ÞE½K Q E½K E½K
C 2 ðQ ; RÞ ¼ þh 1 þR þ ðb þ cða þ 1ÞlÞ ð20Þ
Q 2 k p
 
prY E½K R  E½Y hr lE½K
The optimal Q for a given l can be found by solving @TCDD/@Q = 0.
þ NL þ Q ð14Þ
Q rY 2kp The result is stated as follows.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
The optimal values of lot size and reorder point are respectively gi- u 2ðA þ A ÞE½K
u r
ven as follows Q ¼t  ð21Þ
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi h 1  E½kK þ hr lkpE½K
u 
u RE½Y
u A þ A r þ pr Y NL rY
Q ¼ t2E½K  ð15Þ Substituting Q given by (21) into (20), the following expression for
h 1  E½kK þ hr lkpE½K the expected total cost is obtained.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
   
E½K hr lE½K
and TC DD ðlÞ ¼ 2ðA þ Ar ÞE½K h 1  þ
k kp
 
R  E½Y hQ E½K
U ¼1 ð16Þ þ ðb þ cða þ 1ÞlÞ ð22Þ
rY pE½K p
One can now find the average total cost per unit time of the hierar- It is worthwhile to indicate that the total cost in (22) is a function of
chical model, TCHR, as follows single variable l. As a result, a one dimensional search method can
be used to find the optimal process mean.
TC HR ¼ C 1 þ C 2 ð17Þ
8. Sensitivity analysis
where C 1 and C 2 are the minimum values of C1 and C2 respectively.
One obvious drawback of the hierarchical model is that the produc-
The effects of filling process variation, demand variation, ex-
tion/inventory decisions are not considered in determining the pro-
pected demand rate, raw material cost, holding cost of end items,
cess mean, and consequently it may lead to infeasible solution if the
and process setup cost are investigated in this section. The benefit
production yield rate is smaller than the demand rate (k < D). In or-
of using the integrated model in (10) is investigated by comparing
der to find the optimal solution for the hierarchical model, we first
its performance with that of the hierarchical model where the ven-
determine the optimal process mean l by minimizing C1 in Eq. (13),
dor sets the process mean in isolation of lot sizing decisions. We
we then determine the value of k. If k P D, we find the optimal val-
define the percent benefit (saving) of the integrated model over
ues of Q and R from Eqs. (15) and (16) respectively. On the other
the hierarchical model as follows:
hand, if k < D, then there is no feasible solution.
TC HR  TC
e¼  100%
TC HR
7.2. Perfect filling-process
The data in Table 1 is used as the basis for all results unless specified
In this case, the variation in the amount of material in a pro- otherwise.
duced can is neglected, that is r = 0. In this situation, it is optimal
that the amount filled in the cans is equal to the lower specification 8.1. Sensitivity analysis on process parameters
limit (l⁄ = L) and the probability of conformance is 1. This reduces
the raw material requirement and at the same time no cost associ- 8.1.1. Effect of process variation
ated with the nonconforming items is incurred because all contain- In order to investigate the effect of variation of filling process on
ers are conforming. The model in (10) becomes the optimal solution, the expected total cost of the integrated
M.A. Darwish et al. / European Journal of Operational Research 226 (2013) 481–490 487

Table 1 to consider the filling process variation in determining the


Basic data. inventory decisions and the integrated model is more appropriate
Lower specification limit L = 1.00 than the hierarchical model.
Standard deviation of the amount of raw material used in an r = 1.00
item
8.1.2. Effect of demand variation
Expected value of the demand per unit time E[K] = 300.00
Standard deviation of the demand per unit time rK = 100.00 In order to study the effect of demand standard deviation on the
Production rate r = 500.00 total cost of the integrated model, we compare the performance of
Setup cost A = 50.00 the integrated model presented in (10) with that in (22) where the
Ordering cost of raw material Ar = 20.00
demand is deterministic. The impact of the randomness in demand
Holding cost for the vendor per item per unit time h = 4.00
Holding cost for the raw material per unit per unit time hr = 1.00
is quantified by the parameter w which is defined as follows:
Fixed penalty cost p = 7.00
TC  TC DD
Fixed production cost b = 1.00 w¼  100%
Value-added factor a = 1.00 TC
Unit material cost c = 0.50
The value of w measures the magnitude of the underestimation
Constant component of lead time D = 0.01
in the expected total cost if the randomness in demand is ignored.
From Table 3, one would observe that if demand standard devia-
tion (rK) is high, w is significantly high. For example, in run num-
model when variation exists is compared with perfect filling pro- ber nine (Table 3), if the actual standard deviation rK = 200, then
cess where variation is neglected. We, therefore, define the per- the actual expected total cost is TC = 2840.10. However, when
centage increase in the expected total cost due to variation as ignoring the variation in demand, the total cost, TCDD = 1762.69 is
follows: obtained from Eqs. (20)–(22). Therefore, ignoring uncertainty in
demand can cause the total cost to be underestimated by approx-
TC  TC PP imately 37.9%. Hence, there is significant disruption from using the
X¼  100%
TC deterministic solution in probabilistic environment and therefore,
Table 2 shows that as the filling process standard deviation (r) adjustment must be made.
increases, the benefit of using the integrated model (e) increases. Moreover, it is well known that high demand variation results
This is true because for high values of r the likelihood of producing in high shortages and consequently high penalty cost. Also, when
a nonconforming item increases, this leads to high cost of scrap the variation in demand rK rises, the production lot size Q must
resulting in high total cost. To overcome the high cost of scrap, be high enough to cover fluctuations in demand which gives high
the process mean may be set at a high target value, however, this inventory cost. Therefore, increasing demand standard deviation
leads to a higher material cost. Therefore, high variation in the fill- leads to increased total cost in both hierarchal and integrated mod-
ing process leads to higher total cost and the need for integration is els. In addition, the benefit of the integrated model over hierarchal
more crucial. For example in the seventh run when r = 1.2, the to- model (e) is significant for high values of rK as shown in Table 3.
tal cost for the integrated model is almost $2,116 whereas the total For instance, when rK = 140, the total cost for the integrated model
cost for the hierarchical model is approximately $2,303 with a ben- is roughly $2,224 while the total cost for the hierarchical model is
efit of almost 8%. This trend is evident because in the integrated approximately $2,785 with a reduction in total cost of almost 20%.
model the decisions are taken simultaneously with respect to l, The results also show that demand variation affects the process
Q, and R as oppose to the hierarchical model where l is found first mean (l). This is evident because shortages depend not only on de-
and then Q and R are determined accordingly. It is also important mand uncertainty but also on the probability of producing a con-
to mention here that when r is larger than 1.6, the solution given forming item. Therefore, increasing demand standard deviation
by the hierarchical model is infeasible. This is true because the leads to higher process mean as shown in Table 3. Finally, we have
probability of producing a conforming item (p) at that level is to point out that for rK more than 160, the hierarchical model
not high enough to give high yield rate (k) that is sufficient to sat- gives no feasible solution since k < E[K].
isfy demand, that is k < E[K].
Finally, it can be also observed in Table 2 that X increases with 8.1.3. Effect of expected demand rate
r. For example, ignoring the variation in the filling process when r To study the effect of expected demand rate, we obtained opti-
is actually 1, can cause the total cost to be underestimated by mal solutions for some selected values of E[K] ranging from 200 to
approximately 30%. This indicates that, when the effect of variation 400 with an increment of 20. The results are reported in Table 4. It
in filling process is ignored, the producer significantly underesti- is expected that, as the expected demand increases, the total cost
mates the actual cost of the produced items. Thus, it is important in both integrated and hierarchal model will increase. This is due

Table 2 Table 3
Effect of filling process variation. Effect of demand standard deviation.

Run # r la p TC TCHR e X Run # rK la p TC TCHR e w


1 0.0 1.00 1.00 1566.50 1578.33 0.75 0.00 1 20 2.05 0.85 1806.45 1842.08 1.93 2.42
2 0.2 1.40 0.98 1675.51 1705.85 1.78 6.96 2 40 2.07 0.86 1855.52 1901.40 2.41 5.00
3 0.4 1.66 0.95 1775.12 1821.71 2.56 13.32 3 60 2.10 0.86 1909.19 1969.81 3.08 7.67
4 0.6 1.87 0.93 1867.77 1931.96 3.32 19.23 4 80 2.13 0.87 1968.98 2052.07 4.05 10.48
5 0.8 2.03 0.90 1954.84 2041.77 4.26 24.79 5 100 2.15 0.88 2037.36 2158.77 5.62 13.48
6 1.0 2.15 0.88 2037.36 2158.77 5.62 30.06 6 120 2.18 0.88 2118.93 2324.29 8.84 16.81
7 1.2 2.24 0.85 2116.17 2302.99 8.11 35.09 7 140 2.21 0.89 2224.19 2785.33 20.15 20.75
8 1.4 2.29 0.82 2192.16 2609.84 16.00 39.94 8 160 2.25 0.89 2394.95 –a –a 26.40
9 1.6 2.30 0.79 2266.64 –a –a 44.69 9 180 2.26 0.90 2709.64 –a –a 34.95
10 1.8 2.26 0.76 2342.14 –a –a 49.51 10 200 2.28 0.90 2840.10 –a –a 37.94
a a
Denotes that the solution given by the hierarchical model is not feasible Denotes that the solution given by the hierarchical model is not feasible
because k < E[K]. because k < E[K].
488 M.A. Darwish et al. / European Journal of Operational Research 226 (2013) 481–490

Table 4 Table 5
Effect of expected demand rate. Effect of raw material unit cost.

Run # E[K] la p TC TCHR e Run # c l⁄ p TC TCHR e


1 200 2.13 0.87 1368.97 1418.92 3.52 1 0.2 2.450 0.93 1407.02 1448.00 2.83
2 220 2.13 0.87 1498.86 1558.72 3.84 2 0.3 2.310 0.90 1575.63 1632.98 3.51
3 240 2.14 0.87 1630.39 1701.93 4.20 3 0.4 2.207 0.89 1738.29 1812.40 4.09
4 260 2.14 0.87 1763.84 1849.16 4.61 4 0.5 2.125 0.87 1896.93 1989.14 4.64
5 280 2.15 0.87 1899.43 2001.15 5.08 5 0.6 2.058 0.85 2052.61 2165.14 5.20
6 300 2.15 0.88 2037.36 2158.77 5.62 6 0.7 2.001 0.84 2206.04 2342.18 5.81
7 320 2.16 0.88 2177.92 2323.18 6.25 7 0.8 1.951 0.83 2357.71 2522.24 6.52
8 340 2.16 0.88 2321.36 2495.98 7.00 8 0.9 1.907 0.82 2507.98 2708.03 7.39
9 360 2.17 0.88 2467.96 2680.28 7.92 9 1.0 1.869 0.81 2657.09 2904.43 8.52
10 380 2.17 0.88 2618.04 2926.36 10.54 10 1.1 1.833 0.80 2805.30 3121.86 10.14
11 400 2.18 0.88 2771.93 –a –a
a
Denotes that the solution given by the hierarchical model is not feasible
because k < E[K].

in part to the need for a higher yield rate to cover the increasing Table 6
demand which will directly influence the production cost and con- Effect of end items holding cost.
sequently the total cost, this observation is shown in Table 4. Run # h l⁄ p TC TCHR e
Moreover, when expected demand rate is large, the process mean
1 4.2 2.132 0.871 1923.82 2028.62 5.17
increases to raise the process yield rate and, in turn, satisfying the 2 4.4 2.140 0.873 1950.77 2070.09 5.76
demand. Another important observation is that the benefit of 3 4.6 2.147 0.874 1977.82 2114.15 6.45
integration (e) is larger as expected demand rate is closer to pro- 4 4.8 2.154 0.876 2005.05 2161.66 7.24
5 5.0 2.161 0.877 2032.52 2213.92 8.19
duction rate. As a result, the importance of the integrated model
6 5.2 2.168 0.879 2060.29 2272.96 9.36
is manifested as one gets close to the production rate (r = 500). It 7 5.4 2.174 0.880 2088.48 2342.23 10.83
is also worth mentioning that when the demand rate is larger than 8 5.6 2.181 0.881 2117.22 2427.96 12.80
400, the solution given by the hierarchal model is infeasible be- 9 5.8 2.188 0.882 2146.56 2542.69 15.58
cause the yield rate is less than the expected demand rate. 10 6.0 2.194 0.884 2176.80 2715.56 19.84

8.2. Sensitivity analysis on cost parameters


Table 7
At the outset, we derive the rate of change of the total cost with Effect of setup cost.

respect to different cost parameters. The partial derivatives of the Run # A l⁄ p TC TCHR e
cost function with respect to raw material unit cost, holding cost 1 40.0 2.140 0.873 1970.13 2076.73 5.13
of end items, and process setup cost are as follows: 2 50.0 2.152 0.875 2037.36 2158.77 5.62
3 60.0 2.163 0.878 2099.92 2236.88 6.12
@TC E½Kða þ 1Þl 4 70.0 2.173 0.880 2158.79 2312.33 6.64
¼ ð23Þ
@c p 5 80.0 2.181 0.881 2214.58 2386.24 7.19
   
@TC Q E½K 6 90.0 2.189 0.883 2267.87 2459.77 7.80
¼ 1 þ R  DE½K ð24Þ 7 100.0 2.197 0.884 2319.03 2534.28 8.49
@h 2 k 8 110.0 2.203 0.886 2368.42 2611.73 9.32
@TC E½K 9 120.0 2.210 0.887 2416.32 2695.32 10.35
¼ ð25Þ 10 130.0 2.216 0.888 2462.99 2791.34 11.76
@A Q
As can be seen from Eqs. (23)–(25), the rate of change of the total
cost with raw material unit cost depends on the targeting problem
decision (l) and the inventory side of the problem through the ex-
for large values of c. We also observe that the benefit of the inte-
pected demand rate. Further, the rate of change of the objective
grated model increases for high material unit cost.
function with respect the holding cost depends on targeting prob-
lem parameters through k and Q (note that Q depends on l). It also
depends on the production/inventory control side through Q, R, D 8.2.2. Effect of end items holding cost
and E[K]. Similarly, the change in TC with respect to A is function As the holding cost of end items (h) increases, the total holding
of targeting (through Q and E[K]) and production/inventory control cost per unit time becomes a significant part of the total cost. Rel-
(through Q). Thus, the interaction between the targeting and the atively speaking, this makes raw material cost less important and
production/inventory problems exist. Consequently, the choice of consequently an increase in the optimal target value is expected.
one decision will affect the other decision. In the next three subsec- This is demonstrated by Table 6. Also, Table 6 shows the total costs
tions, we numerically study the effect of c, h and A on the optimal of the integrated and hierarchical models increase with h. Further-
total cost. more, the benefits of the integrated model are more apparent
when the holding cost h increases.
8.2.1. Effect of raw material unit cost
We determined optimal solutions for selected values of raw 8.2.3. Effect of process setup cost
material unit cost, c, ranging from 0.2 to 1.1 with an increment When the setup cost A is large, expected total costs of the inte-
of 0.1 and the results are presented in Table 5. When raw material grated and hierarchical models increase. This observation is shown
unit cost is high, the total material cost per unit time becomes a in Table 7. In addition, the optimal process mean increases with the
major component of the expected total cost function TC. Hence, it setup cost. This is because the raw material cost becomes less sig-
is more economical to fill less content into a container to reduce nificant compared to setup cost. Another important observation is
material cost, this leads to a lower optimal process mean as shown that the savings increase due to using the integrated model over
by Table 5. It is evident also that the expected total cost increases the hierarchical model.
M.A. Darwish et al. / European Journal of Operational Research 226 (2013) 481–490 489

9. Conclusion Chen, Chung-Ho, Khoo, Michael B.C., 2009. Optimum process mean and
manufacturing quantity settings for serial production system under the
quality loss and rectifying inspection plan. Computers and Industrial
In this paper, the classical stochastic continuous review inven- Engineering 57, 1080–1088.
tory control (Q, R) model is integrated with process targeting prob- Chen, Chung-Ho, Kao, Hui-Sung, 2009. The determination of optimum process mean
and screening limits based on quality loss function. Expert Systems with
lem to determine the optimal lot size, reorder point, and process
Applications 36, 7332–7335.
mean. In the proposed model, the performance variable of the Chen, Chung-Ho, Lai, Min-Tsai, 2007. Determination of optimum process mean
product has a lower specification limit, and items that do not con- based on quadratic quality loss function and rectifying inspection plan.
European Journal of Operational Research 182, 755–763.
form to specification limit are scrapped with no salvage value. It is
Chen, S.L., Chung, K.J., 1996. Selection of the optimal precision level and target value
assumed that the performance variable follows a normal distribu- for a production process: the lower specification limit case. IIE Transactions 28,
tion with an adjustable mean and constant variance. 979–985.
A procedure is developed to find the optimal solution of the Darwish, M.A., 2009. Economic selection of process mean for single-vendor
single-buyer supply chain. European Journal of Operational Research 199 (1),
proposed model. The impact of considering the randomness in de- 162–169.
mand pattern is investigated by comparing the proposed model Darwish, M.A., 2004a. Determination of the optimal process mean for a single
with three baseline models. The first of which considers a hierar- vendor-single buyer-problem. International Journal of Operations &
Quantitative Management 10 (3), 1–13.
chical model where the vendor determines the process mean in Darwish, M.A., 2004b. An integrated single-vendor single-buyer targeting problem.
isolation of inventory decisions. This hierarchical model is used In: 2nd International Industrial Engineering Conference, Riyadh, Saudi Arabia.
to show the effect of integrating the process mean with produc- Darwish, M.A., Duffuaa, S.O., 2010. A mathematical model for the joint
determination of optimal process and sampling plan parameters. Journal of
tion/inventory decisions when the demand is a random variable. Quality in Maintenance Engineering 16 (2), 181–189.
The other baseline case is for the deterministic demand. This spe- Darwish, M.A., Goyal, S.K., Alenezi, Abdulrahman, 2012. Stochastic inventory model
cial case is used to explore the effect of variation in demand on the with finite production rate and partial backorders. International Journal of
Services and Operations Management.
optimal solution. The last baseline case is for the situation when
Duffuaa, S.O., Siddiqui, A.W., 2003. Process targeting with multi-class screening and
the variation in the filling amount is neglected. measurements error. International Journal of Production Research 41 (7), 1373–
Empirical results confirm that ignoring the randomness in de- 1391.
Elsayed, E.A., Chen, A., 1993. Optimal levels of process parameters for products with
mand or filling amount may lead to invalid results. In particular,
multiple characteristics. International Journal of Production Research 31, 1117–
using deterministic demand leads to significantly underestimating 1132.
the expected total cost, thus, the model would not be applicable to Gibra, I.N., 1974. Optimal production runs of processes subject to systematic trends.
real world situations and adjustment to stochastic model must be International Journal of Production Research 12 (4), 511–517.
Goethals, P.L., Cho, B.R., 2011. Reverse Programming the optimal process mean
made. Moreover, the results show that the variation in the process problem to identify a factor space profile. European Journal of Operational
is key parameter in controlling the total cost, and the performance Research 215, 204–217.
of a process can be improved considerably by reducing its Golhar, D.Y., 1987. Determination of the best mean contents for a canning problem.
Journal of Quality Technology 19 (2), 82–84.
variation. Gong, L., Roan, J., Tang, K., 1988. Process mean determination with quantity
discounts in raw material cost. Decision Sciences 29 (1), 271–302.
Hadley, G., Whitin, T.M., 1963. Analysis of Inventory Systems. Prentice Hall, New
Acknowledgments Jersey, USA.
Hariga, M.A., Al-Fawzan, M.A., 2005. Joint determination of target value and
The authors are grateful to the anonymous referees for their production run for a process with multiple markets. International Journal of
Production Economics 96 (2), 201–212.
constructive comments and helpful suggestions. They would like Hong, S.H., Cho, B.R., 2007. Joint optimization of process target mean and tolerance
also to acknowledge Kuwait University. limits with measurement errors under multi-decision alternatives. European
Journal of Operational Research 183, 327–335.
Hong, S.H., Elsayed, E.A., 1999. The optimum mean for processes with
References normally distributed measurement error. Journal of Quality Technology 31
(4), 338–344.
Al-Fawzan, M.A., Hariga, M., 2002. An integrated inventory-targeting problem with Hong, S.H., Elsayed, E.A., Lee, M.K., 1999. Optimum mean value and screening limits
time dependent process mean. Production Planning and Control 13 (1), 11–16. for production processes with multi-class screening. International Journal of
Alkhedher, M., Darwish, M.A., 2013. Optimal process mean for a stochastic Production Research 37 (2), 155–163.
inventory model under service level constraint. International Journal of Hong, S.H., Kwon, H.M., Lee, M.K., Cho, B.R., 2006. Joint optimization in process
Operational Research. target and tolerance limit for L-type quality characteristics. International
Al-Sultan, K.S., 1994. An algorithm for the determination of the optimal target Journal of Production Research 44 (15), 3051–3060.
values for two machines in series with quality sampling plans. International Hunter, W.G., Kartha, C.P., 1977. Determining the most profitable target value for a
Journal of Production Research 12 (1), 37–45. production process. Journal of Quality Technology 9 (3), 176–181.
Al-Sultan, K.S., Al-Fawzan, M.A., 1997. An extension of Rahim and Banerjee’s model Kim, Y.J., Cho, B.R., Phillips, M.D., 2000. Determination of the optimal process mean
for a process with upper and lower specification limits. International Journal of with the consideration of variance reduction and process capability. Quality
Production Economics 3, 265–280. Engineering 13 (3), 251–260.
Al-Sultan, K.S., Pulak, M.F., 1997. Process improvement by variance reduction for a Lee, M.K., Elsayed, E.A., 2002. Process mean and screening limits for filling processes
single filling operation with rectifying inspection. Production Planning and under two-stage screening procedure. European Journal of Operational
Control 8 (5), 431–436. Research 138 (1), 118–126.
Al-Sultan, K.S., Pulak, M.F.S., 2000. Optimum target values for two machines in Lee, M.K., Hong, S.H., Elsayed, E.A., 2001. The optimum target value under single and
series with 100% inspection. European Journal of Operational Research 120, two-stage screenings. Journal of Quality Technology 33 (4), 506–514.
181–189. Lee, M.K., Jang, J.S., 1997. The optimum target values for a production process
Arcelus, F.J., Banerjee, P.K., 1985. Selection of the most economical production plan with three-class screening. International Journal of Production Economics 49,
is a tool-wear process. Technometrics 27 (4), 433–437. 91–99.
Arcelus, F.J., Rahim, M.A., 1994. Simultaneous economic selection of a variables and Lee, M.K., Kim, S.B., Kwon, H.M., Hong, S.H., 2004. Economic selection of mean value
an attribute target mean. Journal of Quality Technology 26 (2), 125–133. for a filling process under quadratic quality loss. International Journal of
Bazaraa, M., Sherali, H., Shetty, C., 1993. Nonlinear Programming: Theory and Reliability, Quality and Safety Engineering 11 (1), 506–514.
Algorithms, second ed. John Wiley & Sons. Lee, M.K., Kwon, H.M., Hong, S.H., Kim, Y.J., 2007. Determination of the optimum
Bettes, D.C., 1962. Finding an optimum target value in relation to a fixed lower limit target value for a production process with multiple products. International
and an arbitrary upper limit. Applied Statistics 11, 202–210. Journal of Production Economics 107 (1), 173–178.
Bisgaard, S., Hunter, W.G., Pallesen, L., 1984. Economic selection of quality of Lee, M.K., Kwon, H.M., Kim, Y.J., Bae, J.H., 2005. Determination of optimum target
manufactured products. Technometrics 26, 9–18. values for a production process based on two surrogate variables. Lecture Notes
Boucher, T.O., Jafari, M., 1991. The optimum target value for single filling operations in Computer Science, vol. 3483. Springer, Berlin, pp. 232–240.
with quality sampling plans. Journal of Quality Technology 23 (1), 47–48. Montgomery, D., Johnson, L., 1974. Operations Research in Production Planning,
Bowling, S.R., Khasawneh, M.T., Kaewkuekool, S., Cho, B.R., 2004. A Markovian Scheduling, and Inventory Control. Wiley, New York.
approach to determining optimum process target levels for a multi-stage serial Nahmias, S., 2005. Production and Operations Analysis, fifth ed. Irwin, Homewood,
production system. European Journal of Operational Research 159, 636–650. IL, USA.
490 M.A. Darwish et al. / European Journal of Operational Research 226 (2013) 481–490

Park, T., Kwon, H.M., Hong, Sung-Hoon, Lee, M.K., 2011. optimum common process Ross, S.M., 2007. Introduction to Probability Models, ninth ed. Academic Press,
mean and screening limits for a production process with multiple products. Burlington, MA, USA.
Computers and Industrial Engineering 60, 158–163. Shao, Y.E., Fowler, J.W., Runger, G.C., 2000. Determining the optimal target for a
Pfeifer, P.E., 1999. A general piecewise linear canning problem model. Journal of process with multiple markets and variable holding costs. International Journal
Quality Technology 31 (4), 326–337. of Production Economics 65 (3), 229–242.
Rahim, M.A., Al-Sultan, K.S., 2000. Joint determination of the optimum target mean Silver, E.A., Pyke, D.F., Peterson, R., 1998. Inventory Management and Production
and variance of a process. Journal of Quality in Maintenance Engineering 6 (3), Planning and Scheduling, third ed. John Wiley & Sons, New York, USA.
192–199. Springer, C., 1951. A method for determining the most economic position of a
Rahim, M.A., Banerjee, P.K., 1988. Optimum production run for a process with process mean. Industrial Quality Control 8 (1), 36–39.
random linear drift. Omega 16 (4), 317–351. Tang, K., Lo, J., 1993. Determination of the process mean when inspection is based
Rahim, M.A., Bhadury, J., Al-Sultan, K.S., 2002. Joint economic selection of target on a correlated variable. IIE Transactions 25 (3), 66–72.
mean and variance. Engineering Optimization 34 (1), 1–14. Teeravaraprug, J., Cho, B.R., 2002. Designing the optimal process target levels for
Rahim, M.A., Shaibu, A.B., 2000. Economic selection of optimal target values. Process multiple quality characteristics. International Journal of Production Research 40
Control and Quality 11 (5), 369–381. (1), 37–54.
Roan, J., Gong, L., Tang, K., 2000. Joint determination of process mean, production Williams, W.W., Tang, K., Gong, L., 2000. Process improvement for a container-
run size and material order quantity for a container-filling process. filling process with random shifts. International Journal of Production
International Journal of Production Economics 63, 303–317. Economics 66 (1), 23–31.

You might also like