Sei sulla pagina 1di 23

International Journal of Production Research

ISSN: 0020-7543 (Print) 1366-588X (Online) Journal homepage: http://www.tandfonline.com/loi/tprs20

A supply chain model with vendor-managed


inventory, consignment, and quality inspection
errors

Hesham K. Alfares & Ahmed M. Attia

To cite this article: Hesham K. Alfares & Ahmed M. Attia (2017): A supply chain model with
vendor-managed inventory, consignment, and quality inspection errors, International Journal of
Production Research, DOI: 10.1080/00207543.2017.1330566

To link to this article: http://dx.doi.org/10.1080/00207543.2017.1330566

Published online: 26 May 2017.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=tprs20

Download by: [The UC San Diego Library] Date: 27 May 2017, At: 09:44
International Journal of Production Research, 2017
https://doi.org/10.1080/00207543.2017.1330566

A supply chain model with vendor-managed inventory, consignment, and quality inspection
errors
Hesham K. Alfares* and Ahmed M. Attia

Systems Engineering Department, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia
(Received 10 May 2016; accepted 1 May 2017)

This paper considers the integration between quality control and production inventory control in supply chain
management. Specifically, we study the effect of inspection errors on the costs incurred in a supply chain system with a
single vendor and multiple buyers. In this system, the vendor enters into a vendor-managed inventory (VMI) and a
consignment stock (CS) partnership with several buyers. We assume that the items made by the vendor are not in perfect
quality, but they contain a given proportion of defective units. We also assume that quality inspection of these items by
the buyers is subject to sampling errors. Three cases indicating to different levels of supply integration are considered:
VMI–CS system, traditional system and integrated system. For each case, a mathematical model is formulated, an
optimum solution is developed, and a numerical example is solved.
Keywords: supply chain management; quality control; vendor managed inventory; consignment stock; inspection

1. Introduction
Companies around the world are seeking to gain competitive advantage through increased collaboration with their
supply chain partners. According to Khan, Jaber, and Ahmad (2014), cooperation strategies between vendors (suppliers
or manufacturers) and buyers (retailers) include quantity discounts, common cycle times and quality inspections.
Consignment stock (CS) and vendor-managed inventory (VMI) are also effective strategies that are used for higher
integration and information sharing between suppliers and buyers.
Blackstone and Cox (2008) defined consignment stock as ‘the process of a supplier placing goods at a customer
location without receiving payment until after the goods are used or sold’. Similar to VMI, CS is an effective supply
chain management (SCM) cooperation strategy, indicating a higher degree of collaboration among supply chain partners.
Ben-Daya et al. (2013) described VMI as a vendor–buyer agreement in which ‘the vendor is responsible for managing
the inventory for the buyer, including initiating orders on behalf of the buyer, so the vendor in return gets more
visibility about the product’s demand’. The main difference between the two terms, according to Gümüş, Jewkes, and
Bookbinder (2008), is that the ordering decisions and costs are taken by the buyer in the CS case and by the vendor in
the VMI case.
Ben-Daya et al. (2013) examined a single-vendor, multiple buyer supply chain under an integrated partnership policy
that combines VMI and CS. They concluded that this policy is beneficial to both the vendor and the buyers, if the ven-
dor has a flexible capacity and a low set-up cost, while the buyers have high ordering costs. Khan, Jaber, and Ahmad
(2014) considered the manufacturer’s imperfect quality and the buyer’s inspection errors in a two-echelon supply chain
model. In addition to minimising the total supply chain costs, their model can be used to improve relationship manage-
ment, product and process design and employee training.
This paper combines and extends the models of Ben-Daya et al. (2013) and Khan, Jaber, and Ahmad (2014), in
order to integrate production, inventory and quality control in a collaborative supply chain framework. Specifically, this
paper presents a single-vendor, multiple buyer supply chain model that integrates VMI and CS policies with imperfect
quality and inspection errors. Subsequent parts of this paper are organised as follows. Relevant literature is reviewed in
Section 2. The model is formulated and the solution is developed in Section 3. A numerical example is solved in
Section 4. Finally, conclusions and suggestions are provided in Section 5.

*Corresponding author. Email: alfares@kfupm.edu.sa

© 2017 Informa UK Limited, trading as Taylor & Francis Group


2 H.K. Alfares and A.M. Attia

2. Literature review
In this section, we review the inventory management literature in relevant supply chain models related to the following
topics: (1) single-vendor, multiple buyers models, (2) CS and VMI policies in single-vendor, multiple buyers models,
(3) profit sharing in vendor–buyer partnership, (4) integrated quality control and inventory management and (5) compar-
ison of pertinent vendor–buyer models.

2.1 Single-vendor, multiple buyers models


Financial commitments, such as investments, penalties and discounts, strengthen vendor–buyer cooperation. Woo, Hsu,
and Wu (2001) developed a model to estimate the amount of investments in information technology required to reduce
the joint total cost of the vendor and buyers. The total cost is minimised by reducing the ordering cost, through estab-
lishing an electronic data interchange-based inventory control system. Viswanathan and Piplani (2001) considered a part-
nership policy where the buyers have to replenish only at specific periods determined by the vendor. The vendor offers
a price discount as an incentive for the buyers to have common ordering times. Darwish and Odah (2010) considered a
VMI policy where the vendor pays for the ordering cost and excess inventory. Assuming a supply chain with capacity
constraints and contract-specified bounds on inventory levels, the vendor pays a penalty for exceeding these bounds.
Matching supply and demand rates is also an effective technique for supply chain cooperation. Siajadi, Ibrahim, and
Lochert (2006) examined the effect of different combinations of a single vendor with multiple buyers on the joint total
cost. A significant reduction in the joint total cost is achieved when demand and production rates are close to each
other. Chan and Kingsman (2007) synchronised the production and delivery cycles by aligning the buyers’ delivery
dates with the vendor’s production cycle, while leaving the lot size and order cycle decisions to the buyers. Interest-
ingly, the found this policy to be more economical than the common order cycle policy considered by Viswanathan and
Piplani (2001).
Varying the order quantity size has been used to improve supply chain performance. Hoque (2008) considered three
policies of supply: (1) equal order quantity sent when the vendor finishes production; (2) equal order quantity sent when
the buyer finishes consumption; and (3) unequal order quantity that is proportional to the difference between the produc-
tion rate and the buyers’ total consumption (demand) rate. Comparative analysis indicated the superiority of the third
policy, with unequal order quantities. Hoque (2011) extended the models to incorporate restrictions on transportation
and storage capacities as well as lead times and batch sizes. Hoque assumed the ordering cost is paid by the buyer only
once, regardless of the number of orders.

2.2 CS and VMI in SCM


Dong and Xu (2002) evaluated the effects of VMI on buyer’s profit, supplier’s profit and total supply chain costs. They
found VMI to be a beneficial strategy in an integrated supply chain for both the short-term and the long-term time hori-
zons. Yao, Evers, and Dresner (2007) analysed the effect of parameters, such as the vendor–buyer order cost ratio and
the vendor–buyer holding cost ratio, on VMI benefits to both the vendor and buyer. Wang et al. (2008) also evaluated
the effect of important supply chain parameters on the cost savings resulting from implementing VMI agreements. They
found that VMI tends to increase the ordering frequency, decrease the shortage quantity and decrease the inventory cost.
Zhang, Wei, and Zuo (2008) used genetic algorithms to compare two supply chain models, one with VMI and one with-
out, for a single vendor and multiple buyers.
Wang (2009) studied a supplier–buyer SC system with uncertain demand and random yield under two set-ups. In
the first set-up, the supplier makes the production decisions and the buyer makes the inventory decisions independently
of each other. The second set-up is a VMI collaboration in which the supplier is the decision-maker, who is responsible
of production and inventory decisions. Rouibi and Burlat (2010) simulated a multi-echelon supply chain with one ven-
dor, three manufacturers, three transport companies and a final customer. Considering the (s, S) inventory system, two
VMI policies were analysed, showing greater advantage for less effective supply chains.
Braglia and Zavanella (2003) presented a CS model to for a single buyer and a single vendor, considering uncer-
tainty in delivery lead time and market demand. Using Braglia and Zavanella (2003) model, Min-li and Xiao-hong
(2006) developed a CS model with stochastic demand and uncertain item quality, to minimise the costs of inventory
and defective items.
Valentini and Zavanella (2003) highlighted the benefits of the CS by studying a case study in the automotive indus-
try. They found out that CS benefits are not limited to cost reduction, but also include non-financial gains such as
greater flexibility, improved service and better relationship between suppliers and buyers. Gümüş, Jewkes, and
International Journal of Production Research 3

Bookbinder (2008) analysed the use of CS alone and in combination with VMI in a supply chain with deterministic
demand. They concluded that a price change could be used as an incentive for both sides in the CS partnership, the ven-
dor and the buyer. Chen and Liu (2008) emphasised the mutual benefits for suppliers and buyers resulting from adopting
CS policy in SCM. They noted that it is not easy for suppliers to motivate buyers to accept the CS policy under a
fixed-fee and per-unit commission. Battini et al. (2010) developed a single-vendor, multi-buyer model considering the
buyer’s warehouse space limitations and risks associated with stock-outs and material obsolescence.
Hariga et al. (2013) formulated a mixed-integer non-linear programme to minimise joint inventory costs for a single
vendor and multiple buyers in a VMI contacts with storage-level restrictions. Assuming the vendor has an unlimited
supply and a lower unit holding cost than the buyers, an efficient heuristic is used to solve the model. Ben-Daya et al.
(2013) studied the effect of three partnership policies for a single vendor and multiple buyers: (1) no coordination agree-
ment, (2) VMI and CS agreements and (3) vertical integration with a single decision-maker. The three policies were
compared, and the benefits and features of the integrated VMI–CS policy were highlighted.
Choudhary et al. (2014) analysed the benefits to suppliers and buyers gained from shifting from buyer-managed
inventory (BMI) to VMI and CS. They concluded that buyer cost savings increase under CS policy if the ordering cost
and demand variability increase over time. However, the supplier would prefer the CS policy if the set-up and shipment
costs are very high. Zanoni, Mazzoldi, and Jaber (2014) included taxes imposed on greenhouse gases emissions in the
system total cost. Zanoni and Jaber (2015) considered a case where the buyer’s customer demand depends on the
buyer’s stock level. Shortages are not allowed, and are prevented by enforcing a minimum stock level and instantaneous
replenishment policy.
Hong et al. (2016) analysed a multi-supplier, multi-buyer supply chain where the buyers’ stochastic demand follows
a uniform distribution. Considering shortages as lost sales, the VMI system was shown to have a lower total cost than
the traditional supply chain system. Zahran, Jaber, and Zanoni (2016a, 2016b) considered a new dimension in applying
the CS policy, which is the effect of payment time. Their models examined three payment time scenarios: on time, delay
without interest and delay with interest. Zahran, Jaber, and Zanoni (2016a, 2016b) concluded that a contract allowing
delayed payments with interest is the best for the system. Recently, Lee, Wang, and Chen (2017) considered the restock-
ing level at the buyer’s side as a decision variable. For a single-vendor, single-buyer supply chain with stock-dependent
demand, they found out that profit is maximised when the buyer adopts a minimum reorder level policy. Hemmati,
Fatemi Ghomi, and Sajadieh (2017) assumed a linear relation between the buyer’s customer demand and both the stock
level and the market price.

2.3 Profit sharing in vendor–buyer partnership


In some situations, the vendor–buyer partnership achieves system-wide savings, but one or more partners are worse off.
Ben-Daya et al. (2013) and Hu, Li, and Govindan (2014) considered situations where the cost of the vendor or one/all
of the buyers increased after the partnership, and they proposed applicable solutions. If the vendor is worse off, some of
the buyers’ savings can be transferred to the vendor through a unit price increase. The maximum price increase is the
one that makes at least one buyer indifferent to go for the partnership. The minimum price increase is the one that
makes the vendor no worse off than without partnership. Similarly, when some buyers (or all of them) are worse off,
the vendor can offer a price discount to these buyers as an incentive to accept the partnership.

2.4 Quality control and inventory management


Collins, Case, and Kemble Bennett (1973) classified inspection errors into either type I (a good item is classified as
defective) or type II (a defective item is classified as good). Figure 1 depicts the different types of errors that occur dur-
ing the inspection process. Salameh and Jaber (2000) proposed an economic production quantity (EPQ) model assuming
the items received or produced are not of perfect quality. Non-perfect quality items are not necessarily useless, but may
be used as lower grade products. Huang (2004) developed a similar vendor–buyer model to minimise the expected total
cost, assuming a deteriorating production process and a constant order lot size.
Min-li and Xiao-hong (2006) minimised the average total cost, including the cost of defective items, in a CS model
with uncertain demand and a variable lead time. Khan, Jaber, and Ahmad (2014) developed a mathematical model to
minimise total supply chain cost, considering the buyer’s errors in inspection and the vendor learning in production.
Chiu et al. (2013a, 2013b) considered the existence of defective items and rework in a supply chain. Their objective
was to determine the optimal lot size and the optimal number of shipments in order to minimise the average total costs
of the vendor and the buyer. Chiu et al. (2013a) enhanced the work of Chiu et al. (2013b) by considering failure in
rework and determining a common cycle time for multiple item production. Jaber, Zanoni, and Zavanella (2014a)
4 H.K. Alfares and A.M. Attia

Figure 1. Possible outcomes of the inspection process.

assumed that the supplier is distant from the vendor, making it impractical to send back defective items to the supplier
for reworking. Therefore, the vendor has to compensate for the defective parts from the nearest supplier. Bazan et al.
(2014) tried to solve the problem imperfect production (i.e. production of defective parts) at the vendor’s side. By apply-
ing minor setups, they considered rework and interruption of the production run to restore the process to the in-control
condition.
Jaber, Zanoni, and Zavanella (2014b) formulated a single-vendor, single-buyer supply chain network under CS stock
policy. The buyer is responsible for production, remanufacturing (repair) and waste disposal. Two scenarios are consid-
ered: inspection at the vendor’s side and inspection at the buyer’s side. The model assumes that newly produced prod-
ucts and repaired products have the same quality. Defective items produced during production and remanufacturing are
ignored. Darwish, Odah, and Goyal (2015) examined two VMI policies for single-vendor multiple buyers SC systems.
The inspection for non-conforming units is assumed to be error-free and to be done at the vendor’s side. Along the
same line, Khan et al. (2016) analysed the effect of the defective items received from the vendor, considering two alter-
native disposal schemes for defective items. Assuming imperfect items are either scrapped or reworked, Akbarzadeh,
Taleizadeh, and Esmaeili (2016) considered a more realistic case where the rework process of the defective (imperfect)
items is not perfect but may still produce imperfect items. Hariga, As’ad, and Khan (2017) considered non-accepted
items as parts that need remanufacturing. They proposed a model that optimises the order quantity and the number of
shipments for both newly manufactured parts and remanufactured parts. Giri, Chakraborty, and Maiti (2017) considered
unequal shipment sizes, assuming inspection is done by the buyer and defective items are classified as either scrapped
or repaired.
Quality inspection can be performed on the vendor’s side as in Goyal, Huang, and Chen (2003) and Bazan et al.
(2014), on the buyer’s side as in Salameh and Jaber (2000), Khan, Jaber, and Ahmad (2014) and Giri, Chakraborty, and
Maiti (2017), or on both sides as in Sana (2011). Goyal, Huang, and Chen (2003) considered the case where the quality
inspection is done on the vendor’s side with no errors in the classification results. The cost of producing bad items is
charged to the vendor, who will sell lower quality items at a discounted price to the buyer. Khan, Jaber, and Ahmad
(2014) formulated a more comprehensive model assuming that quality inspections are performed by the vendor. They
considered that the defective items could be either remanufactured or disposed of at given costs, in charge of the vendor.
Sana (2011) analysed a three-layer supply chain consisting of a supplier, a manufacturer and a retailer. Items available
either at the supplier or the manufacturer are sent to the next stage in unequal shipments. At each stage, received items
are fully inspected, and defectives are returned to the previous stage in a single shipment.
In this paper, quality inspections are assumed to be performed by the buyers because this is the practice in many
real-life situations. This is also the assumption made by many researchers in the field, including Salameh and Jaber
(2000), Jaber, Zanoni, and Zavanella (2014a, 2014b), Khan, Jaber, and Ahmad (2014), Khan et al. (2016), and Giri,
Chakraborty, and Maiti (2017).

2.5 Contribution and comparison with previous models


Table 1 compares and classifies the features of relevant vendor–buyer models. Additional classification is provided by
Ramanathan (2014). Table 1 also summarises the contribution of this paper in comparison to previous models. This
paper builds on the integration of concepts from Ben-Daya et al. (2013) and Khan, Jaber, and Ahmad (2014) and adds
several extensions, including the cost of correctly rejected items and three cases of supply chain integration. Ben-Daya
International Journal of Production Research 5

Table 1. Classification of relevant vendor–buyer models.

[1] [2] [3] [4] [5] [6] [7] [8] [9]

Woo, Hsu, and Wu (2001) f, K, n SV-MB C NC VMI D NC D NC


Viswanathan and Piplani (2001) T, n SV-MB NC NC VMI D C D NC
Valentini and Zavanella (2003) Q SV-SB NC NC VMI S NC – NC
Braglia and Zavanella (2003) Q SV-SB NC NC VMI D NC D NC
Siajadi, Ibrahim, and Lochert (2006) T, n SV-MB NC NC VMI D NC D NC
Min-li and Xiao-hong (2006) Q SV-SB NC C VMI D NC D NC
Chan and Kingsman (2007) T, n SV-MB NC NC NP, VMI D NC D, GA NC
Gümüş, Jewkes, and Bookbinder (2008) Q, T SV-SB NC NC NP, VMI D C D NC
Hoque (2008) Q, T, n SV-MB NC NC VMI D NC D NC
Chen and Liu (2008) Q SV-SB NC NC VMI S NC AA NC
Darwish and Odah (2010) Q, n SV-MB NC NC VMI D NC KKT NC
Battini et al. (2010) Q SV-MB C NC VMI D NC AA NC
Hariga et al. (2013) T SV-MB NC NC VMI D NC D NC
Ben-Daya et al. (2013) T, n SV-SB NC NC NP, VMI D NC D NC
Chiu et al. (2013a) Q, n SV-SB NC C VMI D NC D NC
Chiu et al. (2013b) Q, n SV-SB NC C VMI D NC D, AA NC
Zanoni, Mazzoldi, and Jaber (2014) Q, n SV-SB NC NC VMI D NC D NC
Khan, Jaber, and Ahmad (2014) Q, n SV-SB NC C VMI D NC – C
Hu, Li, and Govindan (2014) Q, P, r SV-MB C NC VMI S C AA NC
Jaber, Zanoni, and Zavanella (2014a) Q SV-SB NC C BMI D NC D, AA NC
Jaber, Zanoni, and Zavanella (2014b) Q, n SV-SB NC NC VMI D NC D NC
Bazan et al. (2014) Q, n SV-SB NC C VMI D NC CS NC
Zanoni and Jaber (2015) Q, n CV-SB NC NC VMI D NC AA NC
Zahran, Jaber, and Zanoni (2016a) Q, n SV-SB NC NC BMI D NC D NC
Zahran, Jaber, and Zanoni (2016b) Q, n 3-stage NC NC NP, BMI D NC AA NC
Akbarzadeh, Taleizadeh, and Esmaeili (2016) Q SV-SB C C VMI D NC D, AA NC
Khan et al. (2016) Q, n SV-SB NC C VMI D NC AA NC
Hong et al. (2016) Q, T, r MV-MB C NC VMI S NC D, AA NC
Hariga, As’ad, and Khan (2017) Q, T, n SV-SB NC NC BMI D NC AA NC
Lee, Wang, and Chen (2017) Q, n, r SV-SB NC NC VMI D NC CS NC
Hemmati, Fatemi Ghomi, and Sajadieh(2017) Q, n CV-SB NC NC VMI D NC D NC
Giri, Chakraborty, and Maiti (2017) Q, n SV-SB NC C VMI D NC AA NC
This paper Q, T, n SV-MB NC C VMI D NC D, AA C

Notes: – = the information is not provided by the author.


[1] Decision variables: T = cycle time for the buyer; Q = order quantity; r = reorder point; P = item price; n = number of batches;
K = investment in set-up cost reduction; f = backlogging fraction of cycle time, [2] SC structure: SV = single vendor; MV = multiple
vendors; SB = single buyer; MB = multiple buyers, [3] Shortage for the vendor: C = considered; NC = not considered, [4] Defective
items: C = considered; NC = not considered, [5] Inventory management system: NP = no partnership; VMI = Vendor managed inven-
tory; BMI = Buyer managed inventory, [6] Demand type: S = stochastic; D = deterministic, [7] Price discount: C = considered;
NC = not considered, [8] Solution technique: D = Differential Calculus; GA = Genetic Algorithms; KKT = Karush–Kuhn–Tucker con-
ditions; AA = Algebraic Approach; CS = Commercial Software, and [9] Inspection errors: C = considered; NC = not considered.

et al. (2013) assume a multi-buyer SC system, with VMI and CS but no quality or inspection issues. On the other hand,
Khan, Jaber, and Ahmad (2014) assume a single-buyer system with quality defects and inspection errors, but no CS.
The features of Ben-Daya et al. (2013) and Khan, Jaber, and Ahmad (2014) models, combined in this paper, fre-
quently occur together in many real-life supply chain systems. For example, Giri, Chakraborty, and Maiti (2017) cite
the use of the CS policy for new products or new sales channels in the automotive and health care industries. Jaber,
Zanoni, and Zavanella (2014a) discuss the role of inspection in quality assurance in global supply chains, where incom-
ing material is purchased from distant suppliers. Different item categories such as die cast aluminium components, cast
iron parts and printed circuit boards tend to have different proportions of defective items. Finally, Khan et al. (2016)
describe the benefits of combining the VMI policy with quality inspections to mitigate the effects of demand uncertainty
and supply defects. This policy is expected to reduce warranty and repair costs in many real-life situations, such as the
supply chains of Walmart, Proctor & Gamble, and Johnson & Johnson.
In this paper, the models of Ben-Daya et al. (2013) and Khan, Jaber, and Ahmad (2014) are combined and extended
into a more general model with a wider range of realistic features. The new SCM model, described below, is the first to
6 H.K. Alfares and A.M. Attia

consider simultaneously the following realistic aspects: multiple buyers, multiple shipments, VMI–CS, quality defects
and inspection errors.

3. Model development
In this section, steps used to formulate and solve the VMI–CS model with inspection errors are described.

3.1 Notation
The following notation is mainly adapted from Ben-Daya et al. (2013) to describe the integrated VMI–CS model, and
from Khan, Jaber, and Ahmad (2014) to represent errors in the inspection process:

Abpi the cost of placing an order by the ith buyer ($/order), i = 1, …, N


Abri the cost of receiving a shipment by the ith buyer ($/order)
hboi vendor’s financial cost of tied-up capital for one unit in stock held with the ith buyer for one unit of time
($/unit/unit time)
hbsi ith buyer’s physical storage cost for one unit of stock held for one unit of time ($/unit/unit time)
Avs vendor’s set-up cost ($/order)
Avri vendor’s shipment release cost to the ith buyer ($/order)
hv vendor’s total cost of holding one unit in stock for one unit of time ($/unit/unit time)
n equal number of shipments that is sent to each buyer during a cycle
N number of buyers
di demand from buyer i (units/unit
P time)
D total demand of buyers = Ni¼1 di (units/unit time)
P vendor’s production rate (units/unit time)
qi shipment size for buyer i P
Q total shipment quantity to all buyers = Ni¼1 qi
t time to produce one shipment to all buyers = Q/P
T replenishment cycle length
TCk total cost for supply chain party k, where k = v (vendor) and k = bi (buyer i)
TC total cost per unit time for the supply chain system
e1 probability of a good item to be classified as defective (type I error)
e2 probability of a defective item to be classified as good (type II error)
fa cost of falsely accepting a defective item, i.e. cost of producing a lower quality item
fr cost of falsely rejecting a non-defective item
cr cost of correctly rejecting a defective (imperfect) item
p proportion of defective (imperfect) items supplied by the vendor
pe proportion of rejected items, either correctly or falsely rejected
Sr buyer’s screening rate (units /unit time)
tsi screening time of buyer i (time/order) = qi/Sr

3.2 Assumptions
(1) The vendor is responsible for ordering costs Abpi, Avs and Avri and holding costs hboi and hv; while the buyers
are responsible for ordering costs Abri and holding costs hbsi.
(2) The vendor continuously produces nQ units during the first nt time units of each cycle.
(3) During each cycle, all buyers receive the same number of shipments (orders), n.
(4) For the same buyer i, all orders are of equal size, qi, but order sizes of different buyers vary in proportion to
their demand rates (qi/di is constant for all i).
(5) A cyclic delivery policy is used, where a shipment is sent to each buyer and then this cycle is repeated until all
shipments are delivered.
(6) The vendor’s production rate is higher than the total demand for all buyers (P > D).
(7) The screening (inspection) rate is equal for all buyers; this rate is greater than the consumption (demand) rate of
any buyer (Sr > di, i = 1, …, N).
International Journal of Production Research 7

(8) All items are inspected by the buyers, and only accepted items may be consumed. Falsely accepted units are
used to produce low-quality, lower value items.
(9) All rejected items are discarded together at the end of the inspection process, as assumed by Salameh and Jaber
(2000) and Khan, Jaber, and Ahmad (2014).
(10) The proportion of defective items, p, and the probabilities of type I and type II errors, e1 and e2, are all given
constants, which are the expected values of random variables with probability density functions f(p), f(e1) and f
(e2), respectively.
(11) The model is distribution-free, as the results do not depend on the specific forms of the functions f(p), f(e1)
and f(e2).

3.3 Inventory profile and average inventory


Figure 2 depicts the inventory profile for a single vendor and two buyers. The vendor ships the first lot q1 to the first
buyer and then the first lot q2 to the second buyer. After a time t, each buyer receives another shipment, before the
buyer consumes the previous one. This causes a build-up of inventory at the buyer’s end, increasing the inventory hold-
ing cost. However, the buyer’s inventory still belongs to the vendor according to the assumed consignment policy.
Figure 3 depicts the inventory pattern for one shipment where I1 represents the beginning inventory level after each
shipment. During inspection, the items are screened at the rate Sr, and then consumed at the rate di, which reduces the
inventory to level I2. After discarding the rejected items, the inventory level becomes I3. Before receiving the next lot,
each buyer has I4 items.
Due to inspection errors, both the accepted and the rejected items can be either defective or non-defective. The
effective rejection proportion, pe, is the sum of correctly rejected items, p(1 − e2), and the falsely rejected items, (1 − p)
e 1:

pe ¼ pð1e2 Þ þ ð1pÞe1 (1)


Since the proportion of rejected units is pe, then only qi(1 − pe) accepted units will be used out of each order of size qi.
Therefore, in each cycle T, the amount of accepted items used, nqi(1 − pe), must be equal to the demand for buyer i dur-
ing the cycle, Tdi. Therefore:

Figure 2. Inventory profiles for vendor and buyers.


8 H.K. Alfares and A.M. Attia

Figure 3. Single shipment profile for buyers.

nqi ð1  pe Þ nQð1  pe Þ
T¼ ¼ (2)
di D

di T
qi ¼ (3)
nð1  pe Þ
Since the production rate is P, the time to produce one order of size qi is equal to qi/P, or diT/nP(1 − pe). Therefore, the
time to produce one order each for all buyers is given by:
DT
t¼ (4)
nPð1  pe Þ
Average vendor inventory Iv is calculated as follows. In the vendor’s inventory profile shown in Figure 2, each trian-
gle has a height equal to qi and a base equal to qi/P. Therefore, using (3), the area of each triangle for buyer i is given
by:
T 2 di2
TRIi ¼ (5)
2Pn2 ð1  pe Þ2
During any cycle, n orders are
N sent to each buyer. Summing n triangular areas for each of the N buyers and dividing
P
by the cycle time T gives Iv ¼ n TRIi =T , or:
i¼1
T X
N
Iv ¼ di2 (6)
2Pnð1  pe Þ2 i¼1

For each buyer, the average inventory is equal to the sum of the areas under the inventory profile (Figures 2 and 3)
divided by the cycle time. The total area is the sum of the areas of n pairs of trapezoids (TRP1 and TRP2) and one tri-
angle (TRI) at the end of the cycle. Therefore, the average inventory for buyer i is given by:
" ! #
1 Xn
Ib;i ¼ TRP1i;j þ TRP2i:j þ TRIi (7)
T j¼1

where
I1i:j þ I2i:j
TRP1i;j ¼ tsi (8)
2

I3i;j þ I4i;j
TRP2i;j ¼ ðt  tsi Þ (9)
2

I42i;n
TRIi ¼ (10)
2di
International Journal of Production Research 9

I1i;j ¼ jqi ðj1Þ½pe qi þ di t  (11)

I2i;j ¼ I1i;j di tsi (12)

I3i;j ¼ I1i;j ½pe qi þ di tsi  (13)

I4i;j ¼ I1i;j ½pe qi þ di t  (14)


In the above equations, the subscript i represents the buyer number, while the subscript j represents the order number
in the cycle. After substituting the values from (8) to (14) into (7) and performing simplification steps shown in
Appendix 1, we obtain the following expression for the average inventory of buyer i:
  
Ib;i ¼ nqi tdi ð1  pe Þð1  nÞ þ qi n  2npe þ np2 þ 2di pe (15)
e
2di T Sr
Replacing qi and t by T using (3) and (4), respectively, and simplifying (Appendix 1) gives:
 
Ib;i ¼ di T 2Ppe di
aþ (16)
2Pnð1  pe Þ2 Sr
where
a ¼ ð1  pe Þ½D þ nðP  Ppe  DÞ (17)

3.4 Mathematical model


3.4.1 VMI–CS system: vendor cost
According to the assumed VMI–CS policy, the vendor is responsible for deciding the order time and quantity for each
buyer. The total cost for the vendor TCv is composed of the following components:
(1) Vendor’s set-up cost: this is simply the set-up cost, Avs, at the start of each cycle.
(2) Vendor’s shipment release cost: this is the vendor’s cost per order to buyer i, Avri, multiplied by the number of
orders, n, and summed up for all buyers (i = 1, …, N).
(3) Vendor holding cost: this cost is the vendor’s unit holding cost, hv, multiplied by the vendor’s average inventory
level, Iv .
(4) Vendor’s share of the buyers’ ordering costs: this cost is charged to the vendor because of the VMI–CS partner-
ship. This is the cost per order, Abpi, multiplied by the number of orders, n, and summed up for all buyers
(i = 1, …, N).
(5) Vendor’s share of the buyers’ holding costs: also due to the VMI–CS partnership, this is the unit holding cost for
buyer i, hboi, multiplied by the buyer’s average inventory level, Ib;i and summed up for all buyers (i = 1, …, N).
(6) Cost of inspection errors (false rejection): due to the consignment policy assumed here, false rejection is charged
to the vendor and false acceptance is charged to the buyer. The proportion of false rejection is (1 − p)e1, while the
proportion of used items is (1 − pe). Therefore, the vendor’s cost due to inspection errors is the unit false rejection
cost, fr, multiplied by the number of falsely rejected items for all buyers, which is D(1 − p)e1/(1 − pe).
(7) Cost of defective items (correct rejection): this cost is charged to the vendor. The total correct rejection cost is
obtained by multiplying the unit correct rejection cost, cr, by the number of correctly rejected items for all buyers,
which is Dp(1 − e2)/(1 − pe).
Adding the above components, the vendor’s total cost per unit time is given by:
P P
Avs n Ni¼1 Avri  n Ni¼1 Abpi X N
TCv ðn; T Þ ¼ þ þ hv I v þ þ hboiIb;i
T T T i¼1 (18)
fr Dð1  pÞe1 cr Dpð1  e2 Þ
þ þ
1  pe 1  pe
 
Substituting the values of Iv and Ibi from (6) and (16) into (18) leads to:
10 H.K. Alfares and A.M. Attia
PN
Avs þ n i¼1 ðAvri þ Abpi Þ
fr Dð1  pÞe1 þ cr Dpð1  e2 Þ
TCv ðn; T Þ ¼ þ
T 1  pe
PN 2 PN h i (19)
2Ppe di
hv i¼1 di þ i¼1 hboi di a þ Sr
þ T
2Pnð1  pe Þ2

3.4.2 Traditional system: Buyer’s cost


For each buyer i, the relevant costs include the following components:
(1) Cost of order receipts: this cost is simply the receipt cost per order, Abri, multiplied by the number of orders, n.
(2) Cost of inventory holding: this cost is calculated by multiplying the unit physical storage cost, hbsi, by the aver-
age inventory level, Ibi .
(3) Cost of false acceptance of defective items: this cost is obtained by multiplying the false acceptance unit cost,
fa, by the rate of false acceptance.
Adding the above cost components, the total cost for each buyer i is given by:
 
nAbri fa pe2 di hbsi di 2Ppe di
TCbi ðn; T Þ ¼ þ þ a þ T (20)
T 1  pe 2Pnð1  pe Þ2 Sr

3.4.3 Integrated System: Total SC cost


Adding the costs of the vendor and all buyers from (19) and (20), the total cost for the supply chain system is given
below.
PN
Avs þn ðAvri þAbpi þAbri Þ
TCðn; T Þ ¼ i¼1
þ D½fr ð1pÞe1 þc r pð1e2 Þþfa pe2 

PT N 2 PN 1pe
2Ppe (21)
hv i¼1 i
d þ i¼1 ðhboi þhbsi Þdi ½aþ Sr di 
þ 2Pnð1p Þ 2 T
e

4. Solution procedure and example


4.1 VMI–CS system: minimising the vendor’s cost TCv(n, T)
Equation P(19) is convex in terms of T, since its second partial derivative with respect to T is equal to:
2½Avs þ n Ni¼1 ðAvri þ Abpi Þ=T 3 ; which is positive for all T > 0. Taking the first partial derivative of (19) with respect to
T and equating it to zero, we obtain the following equation for the optimum value of T as a function of n.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P ffi
u
u2Pnð1  pe Þ2 ½Avs þ n Ni¼1 ðAvri þ Abpi Þ
T ¼t P P h i (22)
hv Ni¼1 di2 þ Ni¼1 hboi di a þ 2Pp e
S r di

Substituting the value of T from (22) into (19) leads to:


fr Dð1  pÞe1 þ cr Dpð1  e2 Þ
TCv ðn; T Þ ¼
1  pe
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u h P i (23)
u2 Avs þ n Ni¼1 ðAvri þ Abpi Þ  X XN  
t N 2þ 2Ppe
þ h v i¼1 i
d h
i¼1 boi i
d a þ d i
Pnð1  pe Þ2 Sr
It is clear that the minimum TCv(n, T) is obtained by minimising the term under the squared root sign in (23). Since
the unknown variable is n, we can ignore the constant multiplier 2/P(1 − pe)2. Therefore, the minimum vendor cost is
obtained by minimising the following function:
 
PN
Avs þ n ðAvri þ Abpi Þ XN XN  !
i¼1 2Pp e
bðnÞ ¼ hv di2 þ hboi di a þ di (24)
n i¼1 i¼1
Sr
International Journal of Production Research 11

Since β(n) is a function of n, which is integer, the first-difference approach is used to minimise β(n) and
subsequently TCv(n, T). This approach is based on finding limits on the optimum value of n that satisfy the following
conditions:
bðn þ 1Þ [ bðnÞ

bðn1Þ [ bðnÞ
As shown in Appendix 2, applying this approach leads to the following optimal value for n:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h P  i
u P
u 1 þ 4Avs hv Ni¼1 di2 þ Ni¼1 hboi di D  Dpe þ 2Ppe di
u Sr
n¼t P hP
i (25)
N N
4ð1  pe ÞðP  Ppe  DÞ i¼1 h boi d i i¼1 A vri þ A bpi

Solution procedure to minimise TCv(n, T):


(1) Use (25) to solve for n. If n is not integer, round it to the two nearest integer values, down to bnc and up to
dne.
(2) Use (23) to calculate TCv(n, T) for each of the two possible values of n. Choose the value of n that gives the
minimum total cost for the vendor, TCv(n, T).
(3) Substitute the optimum value of n into (22) to calculate T.

4.2 Traditional System: minimising individual buyer’s cost TCbi(n, T)


Equation (20) is obviously convex with respect to T. Taking the partial derivative of (20) with respect to T and equating
it to zero, the buyers’ optimum cycle time is given by:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u2Pn2 ð1  pe Þ2 Abri
T ¼t h i (26)
hbsi di a þ 2PpSre di

Substituting the value of T from (26) into (20) leads to:


vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h i
u
u2Abri hbsi di a þ 2Ppe di
fa pe2 di t Sr
TCbi ðn; T Þ ¼ þ (27)
1  pe Pð1  pe Þ2
In order to minimise TCbi(n, T), we need to minimise the term under the squared root sign. From (17), α is an
increasing function of n as long as P(1 − pe) > D. Practically, this condition is always satisfied when P > D and pe ≅ 0.
Therefore, (27) is an increasing function of n, and the minimum value of TCbi(n, T) is obtained with the minimum value
of n (i.e. n = 1).
Solution procedure to minimise TCbi(n, T):
(1) Setting n = 1, use (26) to solve for T.
(2) With n = 1, use (27) to calculate TCbi(n, T).

4.3 Integrated System: Minimising total SC cost TC(n, T)


The total cost function (21) is convex is terms of T. Differentiating (21) with respect to n and setting the derivative
equal to zero, we obtain the optimum cycle time for the supply chain system:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P ffi
u
u2Pnð1  pe Þ2 ½Avs þ n Ni¼1 ðAvri þ Abpi þ Abri Þ
T ¼t P P h i (28)
hv Ni¼1 di2 þ Ni¼1 ðhboi þ hbsi Þdi a þ 2Pp Sr
e
d i

Substituting the above value of T into (21) leads to:


12 H.K. Alfares and A.M. Attia

D½fr ð1  pÞe1 þ cr pð1  e2 Þ þ fa pe2 


TCðn; T Þ ¼
1  pe
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
iffi
u P
 P P h (29)
u2½Avs þ n Ni¼1 Avri þ Abpi þ Abri  hv Ni¼1 di2 þ Ni¼1 ðhboi þ hbsi Þdi a þ 2Ppe di
t Sr
þ 2
Pnð1  pe Þ
To minimise (29), we only need to minimise the term under the squared root sign. This can be done by a following
a process similar to that described in Appendix 2 to minimise β(n) in (24), as. Applying the first-difference process to
minimise the squared-root term above leads to the following optimum value for n:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h P  i
u PN
u 1 þ 4A h N
d 2þ ðh þ h Þd D  Dp þ 2Ppe
d
u vs v i¼1 i i¼1 boi bsi i e Sr i
n¼t hP ihP i (30)
N N
4ð1  pe ÞðP  Ppe  DÞ i¼1 ðh boi þ h bsi Þd i i¼1 ðA vri þ A bpi þ A bri Þ

Solution procedure to minimise TC(n, T):


(1) Use (30) to solve for n. If n is not integer, round it to the two nearest integer values, down to bnc and up to
dne.
(2) Use (29) to calculate TC(n, T) for each of the two possible values of n. Choose the value of n that gives the
minimum total cost for the system, TC(n, T).
(3) Substitute the optimum value of n into (28) to calculate T.

4.4 A numerical example


The numerical example presented in this section is adapted from Ben-Daya et al. (2013) and Khan, Jaber, and Ahmad
(2014). The following data is given for a supply chain consisting of a vendor and two buyers:

p = 3200 items/year D = 1500 items/year

f(p) = U(0, 0.04) hv = $4 /item/year


Sr = 175,200 units/year cr = $100 /item
fr = $50/unit fa = $200 /unit
d1 = 500 items/year d2 = 1000 items/year
f(e1) = U(0, 0.04) f(e2) = U(0, 0.04)
hbo1=2.5 hbo2 = 2
hbs1 = 2.5 hbs2 = 3
Abp1=15 Abp2 = 50
Abr1 = 10 Abr2 = 25
Avr1 = 0/shipment Avr2 = 0 /shipment
Avs = $400/set-up

The first step in solving the above example is to calculate the expected values of the uniformly distributed parame-
ters (p = 0.02, e1 = 0.02, e2 = 0.02) in order to use them in all subsequent steps.

4.4.1 VMI–CS system solution


Starting with the objective of minimising the vendor’s total cost, using (25) produces n = 3.482. Substituting n = 3 into
(23) leads to TCv(n, T) = 6401.42, while substituting n = 4 leads to TCv(n, T) = 6400.79. Therefore, the optimum solu-
tion is obtained with four shipments per cycle (n = 4). Finally, substituting n = 4 into (22) leads to T = 0.729. For this
solution, the corresponding total cost for the supply chain system is 7700.11. Table 2 shows solutions based on the min-
imum vendor cost objective obtained with different values of n (n = 1, …, 4). Highlighted cells indicate the optimum n
and the optimum TCv(n, T) values.

4.4.2 Traditional system solution


To minimise the cost of buyer 1, using (26) and (27) with i = 1 and n = 1 leads to: T = 0.126, TCb1(n, T) = 199.77 and
TC = 8976.59. For the objective of minimising TCb1, the solutions corresponding to different values of n are shown in Table 3.
International Journal of Production Research 13

Table 2. Solutions for different values of n to minimise TCv(n, T).

Table 3. Solutions for different values of n to minimise TCb1(n, T).

Table 4. Solutions for different values of n to minimise TCb2(n, T).

Table 5. Solutions for different values of n to minimise TC(n, T).

To minimise the cost of buyer 2, following the same approach but using i = 1 and n = 2 leads to: n = 1, T = 0.129,
TCb2(n, T) = 470.66 and TC = 9059.86. Solutions with minimum TCb2(n, T) corresponding to different values of n are
shown in Table 4, where the optimum n and TCbi(n, T) values are highlighted.
If the objective is to minimise the total cost of the two buyers, then the optimum solution has n = 1, T = 0.128,
TCb(n, T) = 670.44 and TC = 9201.55.

4.4.3 Integrated system solution


Assuming the objective is to minimise the total cost of supply chain system, then applying (30) leads to n = 2.361. Sub-
stituting both n = 2 and n = 3 into (29), the optimum value is found to be n = 2, corresponding to a minimum total cost
TC = 7511.59 for the supply chain system. For this solution, the cycle time is T = 0.429. Table 5 shows solutions result-
ing from different values of n if the objective is to minimise the total cost for the entire supply chain system.
In the spirit of collaboration, the vendor and the two buyers should adopt the policy that gives the minimum total
cost for the entire supply chain. This policy, with total SC cost, TC = 7511.59, is outlined in the second row of Table 5.
This solution involves some sacrifice from both sides, as they give up their individual minimum cost policies for the
sake of minimising the total cost for the system. The vendor’s cost TCv(n, T) increases by 2.1%, from 6401.42 to
6534.01. On the other hand, the cost of buyer 1, TCb1(n, T), increases by 46% from 199.77 to 291.03, while the cost of
buyer 2, TCb2(n, T), increases by 46% from 470.66 to 686.55. The cost of the inspection errors is equal to 1654.88, or
22% of TC, divided into 1529.98 for the vendor, 41.63 for buyer 1 and 83.26 for buyer 2.
14 H.K. Alfares and A.M. Attia

4.4.4 Sensitivity analysis and managerial insights


Sensitivity analysis was performed on the solution shown in Table 5, in order to compare the impact of changing input
parameter values on total system cost TC(n, T). The main objective of sensitivity analysis is to gain managerial insights
in order to guide decision-making. Individually, each input parameter was decreased by 20% and then increased by
20%, while fixing all other parameters at original values. From the results shown in Table 6, two main observations can
be made on the sensitivity of the total system cost TC(n, T). First, all input parameters are positively correlated with the
total cost, thus increasing any input value increases TC(n, T) and vice versa. This conclusion is in line with Khan, Jaber,
and Ahmad (2014) results, where the total cost of the system increases with higher values of type I and type II errors.
The second observation from Table 6 is that the demand di has the highest impact on TC(n, T), while the cost of receiv-
ing a shipment Abri has the least impact. The high influence of demand on total cost TC is expected, since two terms of
(21) are multiplied by either D or d2. On the other hand, the low impact of Abri on TC is logical because Abri is only a
minor component in one term of (21).
Impact of parameter changes on other solution variables has also been observed. The demand di has the highest
impact on the vendor’s cost TCv(n, T), while the buyer’s physical storage cost hbsi and the demand di have the highest
impact on buyers’ cost TCbi(n, T). Moreover, the buyer’s ordering cost Abpi has the highest impact on the order size qi,
while the vendor’s set-up cost Avs has the highest impact on the cycle time T. Finally, the number of shipments per cycle
n increases with higher values of di and Avs and with lower values of P and Abpi. Considering overall impact, the
demand di is the most important input parameter, distantly followed by parameters p, cr, ei and Avs. Contrary to intuitive
expectations, minimising the given individual input cost parameters is not the most effective tool for managers to min-
imise total cost. In order to achieve the lowest total cost, managers should focus first on increasing the demand D, and
then on reducing the proportion of defects, p. Moreover, to reduce the high cost of inspection errors, managers should
find ways to minimise type I and type II errors probabilities, e1 and e2.
For additional insights, the effect of changing certain critical ratios has been analysed for both the VMI–CS system
and the integrated system. As suggested by Ben-Daya et al. (2013) and Bazan et al. (2014), these ratios are: the capacity
ratio, D/P, the order cost ratio, Avs/Ab and the holding cost ratio, hv/hb. The results for the three ratios are shown in
Tables 7–9, where shaded cells indicate the VMI–CS system, and clear cells indicate the integrated system. These ratios
are adapted for our multi-buyer, multi-cost model by adopting the following definitions:

Table 6. Sensitivity analysis for the objective of minimising system cost TC(n, T).

Value Original New n t T q1 q2 TCv TCb1 TCb2 TC

P 3200 2560 3 0.106 0.521 90 181 6418 293 691 7402


3840 2 0.086 0.424 110 221 6539 300 708 7547
d1, d2 500, 1000 400, 800 2 0.092 0.473 98 197 5416 266 629 6311
600, 1200 3 0.092 0.471 98 196 7522 329 777 8628
hv 4 3.2 2 0.106 0.435 113 226 6491 293 692 7475
4.8 2 0.103 0.424 110 220 6577 289 682 7548
cr 100 80 2 0.105 0.429 112 223 5922 291 687 6900
120 2 0.105 0.429 112 223 7146 291 687 8124
fr, fa 50, 200 40, 160 2 0.105 0.429 112 223 6228 283 670 7181
60, 240 2 0.105 0.429 112 223 6840 299 703 7843
p 0.02 0.016 2 0.104 0.429 111 222 5912 283 670 6865
0.024 2 0.105 0.429 112 224 7161 299 703 8164
e1 = e2 0.02 0.016 2 0.104 0.429 111 222 6223 283 670 7176
0.024 2 0.105 0.429 112 224 6848 299 703 7850
hbo1, hbo2 2.5, 2 2, 1.6 2 0.109 0.446 116 232 6405 297 702 7404
3, 2.4 2 0.101 0.414 108 215 6656 286 673 7615
hbs1, hbs2 2.5, 3 2, 2.4 2 0.110 0.452 118 235 6509 257 604 7370
3, 3.6 2 0.100 0.409 107 213 6561 323 763 7646
Abp1, Abp2 15, 50 12, 40 3 0.079 0.485 84 168 6405 308 729 7442
18, 60 2 0.107 0.438 114 228 6583 294 695 7572
Abr1, Abr2 10, 25 8, 20 2 0.103 0.424 110 221 6540 280 659 7479
12, 30 2 0.106 0.434 113 226 6528 302 714 7544
Avs 400 320 2 0.097 0.399 104 208 6376 280 662 7318
480 3 0.086 0.527 91 183 6598 321 759 7677
International Journal of Production Research 15

Ab ¼ buyers ordering cost; i.e. cost of placing and receiving an order for all buyers

¼ Ri Abpi þ Abri

hb ¼ buyers holding cost; i.e. cost of capital and physical storage for all buyers
¼ Ri ðhboi þ hbsi Þ
From the capacity ratio (D/P) sensitivity analysis results shown in Table 7, the following managerial insights are
obtained.
• The vendor’s cost is inversely related to the capacity ratio, under both the VMI–CS policy and the integrated sys-
tem policy.
• The total system cost is inversely related to the capacity ratio under the integrated system policy.
• Since the costs of the buyers are not directly affected by the capacity ratio, they should cooperate with the vendor
to decrease the capacity ratio (D/P).
• The best case for the vendor is higher D/P under the VMI–CS policy, and the best case for the SC system is
higher D/P under the integrated system policy.
Sensitivity analysis results for the order cost ratio (Avs/Ab) are shown in Table 8. For the given example, buyer order-
ing costs are given by: Ab = 15 + 10 + 50 + 25 = 100. Based on the results in Table 8, the following managerial insights
are obtained:
• As the order cost ratio increases, all costs (vendor, buyer and total cost) increase, under both the VMI–CS policy
and the integrated system policy.
• All parties, i.e. vendor and buyers, should try to decrease the order cost ratio (Avs/Ab) to achieve cost savings.
• The best case for the vendor is lower Avs/Ab under the VMI–CS policy, and the best case for the buyers and the
SC system is lower Avs/Ab under the integrated system policy.
Sensitivity analysis results for the holding cost ratio (hv/hb) are shown in Table 9. For the given example, buyer
holding costs are given by: hb = 2.5 + 2.5 + 2 + 3 = 10. Based on the results in Table 9, the following managerial
insights are obtained:

Table 7. Sensitivity analysis of capacity ratio D/P.

Note: Shaded cells: VMI–CS system, Clear cells: integrated system.

Table 8. Sensitivity analysis of order cost ratio Avs/Ab.

Note: Shaded cells: VMI–CS system, Clear cells: integrated system.


16 H.K. Alfares and A.M. Attia

Table 9. Sensitivity analysis of order cost ratio hv/hb.

Note: Shaded cells: VMI–CS system, Clear cells: integrated system.

• As the holding cost ratio increases, the vendor cost and the total system cost increase, under both the VMI–CS
policy and the integrated system policy.
• As the holding cost ratio increases, buyers’ costs decrease under the integrated system policy.
• The best case for the vendor is lower hv/hb under the VMI–CS policy, the best case for the buyers is higher hv/hb
under the integrated system policy, while the best case for the SC system is lower hv/hb under the integrated
system policy.
• The effect of the holding cost ratio on buyers’ costs is negligible. Therefore, for best overall performance, the
buyers should agree to lowering this ratio and operating under the integrated system policy.
Dealing with production quality defects and quality inspection errors can be quite challenging for multiple buyers in
a VMI–CS agreement with a single supplier. The models of Ben-Daya et al. (2013) and Khan, Jaber, and Ahmad
(2014) included subsets of the issues that must be considered in these situations. Integrating these issues together, the
model presented in this paper can help in making coordinated decisions to maximise the benefits of all SC partners. This
model allows concerned VMI–CS partners to make optimum decisions on shipment sizes and frequencies to minimise
the total cost for the supply chain system. Comparing the results of the three cases of supply chain integration allows
each partner to know their optimum ‘individual’ solution, and the amount of tradeoff or sacrifice they are contributing
to the system. Considering the three cases, along with the sensitivity analysis results, is essential to compare alternative
scenarios in order to determine the most appropriate course of action for each partner as well as for the entire system.

5. Conclusions
This paper considered an integrated production inventory control and quality control problem in supply chain manage-
ment. The supply chain system consists of one vendor and several buyers, collaborating in a combined vendor-managed
inventory and consignment stock (VMI–CS) agreement. Two types of inspection errors were taken into consideration:
accepting a defective unit and rejecting a good unit. Three levels of supplier–buyer collaboration were investigated. A
mathematical model was constructed and optimum solution procedures were developed to minimise the total costs for
either the vendor, the individual buyers or the entire supply chain system. An example was solved to illustrate the solu-
tion process and properties. While considering inspection errors significantly increase the costs of both the vendor and
the buyers, the impact is more substantial on the vendor’s cost.
This paper contributed to the theory by combining and expanding two previous models and formulating a new
model with a unique set of realistic assumptions, as well developing an effective optimum solution algorithm for this
new model. The paper also contributed to practice by discussing the managerial insights of the new model and consider-
ations for its practical applications. Obviously, the model presented in this paper is limited by the set of assumption on
which it is built. However, the main limitations are the assumptions on the size, number and sequence of shipments to
the multiple buyers.
Removing the model’s limiting assumptions is one approach among several promising directions which are possible
for extending this work. These directions include the consideration of the following aspects: (1) uncertainty in market
demand, (2) a stochastic proportion of errors in the inspection process, (3) vendor’s capacity limitations, (4) shortages
and late deliveries, (5) shipment lead time variability, (6) unequal shipment sizes qi for each buyer, (7) a different num-
ber of shipments ni for each buyer, (8) a different cycle time Ti for each buyer and (9) environmental performance as in
(Bazan, Jaber, and Zanoni 2016).
International Journal of Production Research 17

Disclosure statement
No potential conflict of interest was reported by the authors.

ORCID
Hesham K. Alfares http://orcid.org/0000-0003-4040-2787

References

Akbarzadeh, M., A. A. Taleizadeh, and M. Esmaeili. 2016. “Developing an Economic Production Quantity Model with Scrap,
Rework and Backordering under Vendor-managed Inventory Policy.” International Journal of Advanced Logistics 1–16.
Battini, D., A. Gunasekaran, M. Faccio, A. Persona, and F. Sgarbossa. 2010. “Consignment Stock Inventory Model in an Integrated
Supply Chain.” International Journal of Production Research 48 (2): 477–500.
Bazan, E., M. Y. Jaber, S. Zanoni, and L. E. Zavanella. 2014. “Vendor Managed Inventory (VMI) with Consignment Stock (CS)
Agreement for a Two-level Supply Chain with an Imperfect Production Process with/without Restoration Interruptions.”
International Journal of Production Economics 157: 289–301.
Bazan, E., M. Y. Jaber, and S. Zanoni. 2016. “A Review of Mathematical Inventory Models for Reverse Logistics and the Future of
Its Modeling: An Environmental Perspective.” Applied Mathematical Modelling 40 (5–6): 4151–4178.
Ben-Daya, M., E. Hassini, M. Hariga, and M. M. AlDurgam. 2013. “Consignment and Vendor Managed Inventory in Single-vendor
Multiple Buyers Supply Chains.” International Journal of Production Research 51 (5): 1347–1365.
Blackstone, J. H., and J. F. I. Cox. 2008. APICS Dictionary. 12th ed. Alexandria, VA: Falls Church, APICS.
Braglia, M., and L. Zavanella. 2003. “Modelling an Industrial Strategy for Inventory Management in Supply Chains: The
‘Consignment Stock’ case.” International Journal of Production Research 41 (16): 3793–3808.
Chan, C. K., and B. G. Kingsman. 2007. “Coordination in a Single-vendor Multi-buyer Supply Chain by Synchronizing Delivery and
Production Cycles.” Transportation Research Part E: Logistics and Transportation Review 43 (2): 90–111.
Chen, S.-L., and C.-L. Liu. 2008. “The Optimal Consignment Policy for the Manufacturer under Supply Chain Co-ordination.”
International Journal of Production Research 46 (18): 5121–5143.
Chiu, S. W., L.-W. Lin, K.-K. Chen, and C.-L. Chou. 2013a. “Determining Production–Shipment Policy for a Vendor–Buyer
Integrated System with Rework and an Amending Multi-delivery Schedule.” Economic Modelling 33: 668–675.
Chiu, S. W., F.-Y. Pai, and W. K. Wu. 2013b. “Alternative Approach to Determine the Common Cycle Time for a Multi-item
Production System with Discontinuous Deliveries and Failure in Rework.” Economic Modelling 35: 593–596.
Choudhary, D., R. Shankar, P. K. Dey, H. Chaudhary, and L. S. Thakur. 2014. “Benefits of Retailer–Supplier Partnership Initiatives
under Time-varying Demand: A Comparative Analytical Study.” International Journal of Production Research 52 (14):
4279–4298.
Collins, R. D. Jr., K. E. Case, and G. Kemble Bennett. 1973. “The Effects of Inspection Error on Single Sampling Inspection Plans.”
International Journal of Production Research 11 (3): 289–298.
Darwish, M. A., and O. M. Odah. 2010. “Vendor Managed Inventory Model for Single-vendor Multi-retailer Supply Chains.” Euro-
pean Journal of Operational Research 204 (3): 473–484.
Darwish, M. A., O. M. Odah, and S. K. Goyal. 2015. “Vendor Managed Inventory Models for Single-vendor Multi-retailer Supply
Chains with Quality Consideration.” International Journal of Industrial and Systems Engineering 20 (1): 22–57.
Dong, Y., and K. Xu. 2002. “A Supply Chain Model of Vendor Managed Inventory.” Transportation Research Part E: Logistics and
Transportation Review 38 (2): 75–95.
Giri, B. C., A. Chakraborty, and T. Maiti. 2017. “Consignment Stock Policy with Unequal Shipments and Process Unreliability for a
Two-level Supply Chain.” International Journal of Production Research 55 (9): 2489–2505.
Goyal, S. K., C.-K. Huang, and K.-C. Chen. 2003. “A Simple Integrated Production Policy of an Imperfect Item for Vendor and
Buyer.” Production Planning & Control 14 (7): 596–602.
Gümüş, M., E. M. Jewkes, and J. H. Bookbinder. 2008. “Impact of Consignment Inventory and Vendor-managed Inventory for a
Two-party Supply Chain.” International Journal of Production Economics 113 (2): 502–517.
Hariga, M., M. Gumus, A. Daghfous, and S. K. Goyal. 2013. “A Vendor Managed Inventory Model under Contractual Storage
Agreement.” Computers & Operations Research 40 (8): 2138–2144.
Hariga, M., R. As’ad, and Z. Khan. 2017. “Manufacturing-remanufacturing Policies for a Centralized Two Stage Supply Chain under
Consignment Stock Partnership.” International Journal of Production Economics 183: 362–374.
Hemmati, M., S. M. T. Fatemi Ghomi, and M. S. Sajadieh. 2017. “Vendor Managed Inventory with Consignment Stock for Supply
Chain with Stock-and Price-dependent Demand.” International Journal of Production Research: 1–18.
Hong, X., W. Chunyuan, L. Xu, and A. Diabat. 2016. “Multiple-vendor, Multiple-retailer Based Vendor-managed Inventory.” Annals
of Operations Research 238 (1–2): 277–297.
Hoque, M. A. 2008. “Synchronization in the Single-manufacturer Multi-buyer Integrated Inventory Supply Chain.” European Journal
of Operational Research 188 (3): 811–825.
18 H.K. Alfares and A.M. Attia

Hoque, M. A. 2011. “An Optimal Solution Technique to the Single-vendor Multi-buyer Integrated Inventory Supply Chain by
Incorporating Some Realistic Factors.” European Journal of Operational Research 215 (1): 80–88.
Hu, W., Y. Li, and K. Govindan. 2014. “The Impact of Consumer Returns Policies on Consignment Contracts with Inventory
Control.” European Journal of Operational Research 233 (2): 398–407.
Huang, C.-K. 2004. “An Optimal Policy for a Single-vendor Single-buyer Integrated Production–Inventory Problem with Process
Unreliability Consideration.” International Journal of Production Economics 91 (1): 91–98.
Jaber, M. Y., S. Zanoni, and L. E. Zavanella. 2014a. “Economic Order Quantity Models for Imperfect Items with Buy and Repair
Options.” International Journal of Production Economics 155: 126–131.
Jaber, M. Y., S. Zanoni, and L. E. Zavanella. 2014b. “A Consignment Stock Coordination Scheme for the Production, Remanufactur-
ing and Waste Disposal Problem.” International Journal of Production Research 52 (1): 50–65.
Khan, M., M. Y. Jaber, and A.-R. Ahmad. 2014. “An Integrated Supply Chain Model with Errors in Quality Inspection and Learning
in Production.” Omega 42 (1): 16–24.
Khan, M., M. Y. Jaber, S. Zanoni, and L. Zavanella. 2016. “Vendor Managed Inventory with Consignment Stock Agreement for a
Supply Chain with Defective Items.” Applied Mathematical Modelling 40 (15): 7102–7114.
Lee, W., S.-P. Wang, and W.-C. Chen. 2017. “Forward and Backward Stocking Policies for a Two-level Supply Chain with Consign-
ment Stock Agreement and Stock-DEPENDENT Demand.” European Journal of Operational Research 256 (3): 830–840.
Min-li, X., and C. Xiao-hong. 2006. “Consignment Stock Policy with Defective Items.” 2006 International Conference on Manage-
ment Science and Engineering, IEEE, 540–544.
Ramanathan, U. 2014. “Performance of Supply Chain Collaboration–A Simulation Study.” Expert Systems with Applications 41 (1):
210–220.
Rouibi, S., and P. Burlat. 2010. “The Impact of the Vendor Managed Inventory on Supply Chain Performance.” 2010 40th Interna-
tional Conference on Computers and Industrial Engineering, (CIE), IEEE, 1–6.
Salameh, M. K., and M. Y. Jaber. 2000. “Economic Production Quantity Model for Items with Imperfect Quality.” International Jour-
nal of Production Economics 64 (1–3): 59–64.
Sana, S. S. 2011. “A Production-inventory Model of Imperfect Quality Products in a Three-layer Supply Chain.” Decision Support
Systems 50 (2): 539–547.
Siajadi, H., R. N. Ibrahim, and P. B. Lochert. 2006. “A Single-vendor Multiple-Buyer Inventory Model with a Multiple-shipment
Policy.” The International Journal of Advanced Manufacturing Technology. 27 (9–10): 1030–1037.
Valentini, G., and L. Zavanella. 2003. “The Consignment Stock of Inventories: Industrial Case and Performance Analysis.”
International Journal of Production Economics 81–82: 215–224.
Viswanathan, S., and R. Piplani. 2001. “Coordinating Supply Chain Inventories through Common Replenishment Epochs.” European
Journal of Operational Research 129 (2): 277–286.
Wang, C. X. 2009. “Random Yield and Uncertain Demand in Decentralised Supply Chains under the Traditional and VMI Arrange-
ments.” International Journal of Production Research 47 (7): 1955–1968.
Wang, C., S. Ji, J. Shen, and W. Wei. 2008. “Supply Chain Model in Vendor Managed Inventory.” IEEE International Conference on
Service Operations and Logistics, and Informatics, 2008, IEEE/SOLI 2008, IEEE, 2110–2113.
Woo, Y. Y., S.-L. Hsu, and S. Wu. 2001. “An Integrated Inventory Model for a Single Vendor and Multiple Buyers with Ordering
Cost Reduction.” International Journal of Production Economics 73 (3): 203–215.
Yao, Y., P. T. Evers, and M. E. Dresner. 2007. “Supply Chain Integration in Vendor-managed Inventory.” Decision Support Systems
43 (2): 663–674.
Zahran, S. K., M. Y. Jaber, and S. Zanoni. 2016a. “The Consignment Stock Case for a Vendor and a Buyer with Delay-in-payments.”
Computers & Industrial Engineering 98: 333–349.
Zahran, S. K., M. Y. Jaber, and S. Zanoni. 2016b. “Comparing Different Coordination Scenarios in a Three-level Supply Chain
System.” International Journal of Production Research 1–21.
Zanoni, S., and M. Y. Jaber. 2015. “A Two-level Supply Chain with Consignment Stock Agreement and Stock-Dependent Demand.”
International Journal of Production Research 53 (12): 3561–3572.
Zanoni, S., L. Mazzoldi, and M. Y. Jaber. 2014. “Vendor-managed Inventory with Consignment Stock Agreement for Single
Vendor–single Buyer under the Emission-Trading Scheme.” International Journal of Production Research 52 (1): 20–31.
Zhang, L., H. Wei, and X. Zuo. 2008. “Research on Vendor Managed Inventory in the Condition of Multiple Retailers in
E-commerce.” 2008 Fourth International Conference on Natural Computation.

Appendix 1
Derivation of (15), (16) and (17)
The total area of the trapezoids in (7) is defined as follows:
X
n
TRPZi ¼ TRP1i;j þ TRP2i:j (A1)
j¼1
International Journal of Production Research 19

Substituting in the values of TRP1i and TRP2i from (8) and (9) gives:
1X n
TRPZi ¼ tsi ðI1i;j þ I2i;j Þ þ ðt  tsi ÞðI3i;j þ I4i;j Þ
2 j¼1

Substituting in the values of I1i,j, …, I4i,j from (11)–(14) gives:


1X n
TRPZi ¼ tsi ð2I1i;j  di tsi Þ þ ðt  tsi Þð2I1i;j  2pe qi  di tsi  di tÞ
2 j¼1

1X n

TRPZi ¼ 2tsi I1i;j  di tsi2 þ 2tI1i;j  2tpe qi  di ttsi  di t 2  2tsi I1i;j þ 2tsi pe qi þ di tsi2 þ di ttsi
2 j¼1

1 Xn
TRPZi ¼ 2tI1i;j  2tpe qi  di t 2 þ 2tsi pe qi
2 j¼1

X
n
1X n
TRPZi ¼ t I1i;j þ 2tpe qi  di t 2 þ 2tsi pe qi
j¼1
2 j¼1

!
X
n
n

TRPZi ¼ t I1i;j þ 2tpe qi  di t 2 þ 2tsi pe qi (A2)


j¼1
2

However, from (11):

X
n X
n
I1i;j ¼ jqi  ðj  1Þ½pqi þ di t
j¼i j¼1

X
n X
n X
n
I1i;j ¼ qi j  ½pqi þ di t ðj  1Þ
j¼i j¼1 j¼1

X
n
nðn þ 1Þ ðn  1Þn
I1i;j ¼ qi  ½pqi þ di t
j¼i
2 2

X
n
n
I1i;j ¼ ½ðn þ 1Þqi  ðn  1Þðpqi þ di tÞ (A3)
j¼i
2

Substituting (A3) into (A2) gives:


nt n
TRPZi ¼ ½ðn þ 1Þqi  ðn  1Þðpe qi þ di tÞ þ 2tpe qi  di t2 þ 2tsi pe qi
2 2
n
TRPZi ¼ ntqi þ tqi  ntpe qi  ndi t2 þ tpe qi þ di t2  2tpe qi  di t2 þ 2tsi pe qi
2
n
TRPZi ¼ ntqi þ tqi  ntpe qi  ndi t 2  tpe qi þ 2tsi pe qi (A4)
2
Assuming j = n and combining (11) and (14) gives:
I4i;n ¼ nqi n½pe qi þ di t  (A5)
Substituting the above value of I4i,n in (10) gives:

ðnqi  npe qi  ndi tÞ2


TRIi ¼ (A6)
2di
20 H.K. Alfares and A.M. Attia

Substituting (A4) and (A6) into (7) gives:


" #
1 n ðnqi  npe qi  ndi tÞ2
Ib;i ¼ 2
ntqi þ tqi  ntpe qi  ndi t  tpe qi þ 2tsi pe qi þ
T 2 2di

 
Ib;i ¼ n ndi tqi þ di tqi  ndi pe tqi  ndi2 t 2  di pe tqi þ 2di tsi pe qi
2 2 2 2 2 2
2di T þðnqi  2npe qi  2ndi tqi þ 2ndi pe tqi þ npe qi þ ndi t Þ
 
Ib;i ¼ n ½di tqi  di pe tqi þ 2di tsi pe qi 
2di T þðnq2i  2npe q2i  ndi tqi þ ndi pe tqi þ np2e q2i Þ


Ib;i ¼ n tqi ðdi  di pe  ndi þ ndi pe Þ þ q2 ðn  2npe þ np2 Þ þ 2qi di tsi pe (A7)
i e
2di T
Substituting ts = qi/Sr into (A7) gives:
  
Ib;i ¼ n tqi ðdi  di pe  ndi þ ndi pe Þ þ q2 n  2npe þ np2 þ 2di pe (15)
i e
2di T Sr
Substituting out qi and t using (3) and (4) gives:
"  #
2 2 2
Ib;i ¼ n d i DT di T 2 2di p e
ðdi  di pe  ndi þ ndi pe Þ þ n  2npe þ npe þ
2di T Pn2 ð1  pe Þ2 n2 ð1  pe Þ2 Sr

   
Ib;i ¼ T 2 2di pe
ðdi  di pe  ndi þ ndi pe ÞD þ n  2npe þ npe þ di P
2Pnð1  pe Þ2 Sr
 
Ib;i ¼ T 2di2 Ppe
di D  di Dpe  ndi D þ ndi Dpe þ ndi P  2ndi Ppe þ ndi Pp2e þ
2Pnð1  pe Þ2 Sr
 
Ib;i ¼ di T 2 2di Ppe
D  Dpe  nD þ nDpe þ nP  2nPp e þ nPp e þ
2Pnð1  pe Þ2 Sr
 
Ib;i ¼ di T 2Ppe di
a þ (16)
2Pnð1  pe Þ2 Sr
where
a ¼ D  Dpe  nD þ nDpe þ nP  2nPpe þ nPp2e

a ¼ Dð1  pe Þ þ n ðDð1  pe Þ þ Pð1  2pe þ p2e Þ
h i
a ¼ Dð1  pe Þ þ n ðDð1  pe Þ þ Pð1  pe Þ2

a ¼ ð1  pe Þ½ D þ nðP  Ppe  DÞ (17)


International Journal of Production Research 21

Appendix 2
Derivation of (25)
Substituting α from (17) into (24) gives:
 
PN
Avs þ n ðAvri þ Abpi Þ
i¼1
bðnÞ ¼
n (B1)
XN XN  !
2 2Ppe
 hv di þ hboi di ð1  pe Þ½ D þ nðP  Ppe  DÞ þ di
i¼1 i¼1
Sr

Simplifying and rearranging, we obtain:


!
Avs X N
bðnÞ ¼ þ ðAvri þ Abpi Þ
n i¼1
  !
X
N X
N
2Ppe XN
 hv di2 þ hboi di D  Dpe þ di þ nð hboi di Þð1  pe ÞðP  Ppe  DÞ
i¼1 i¼1
Sr i¼1

Multiplying out the above terms leads to:


"  #
Avs XN XN
2Ppe
2
bðnÞ ¼ hv di þ hboi di D  Dpe þ di
n i¼1 i¼1
Sr
XN
þ Avs ð1  pe ÞðP  Ppe  DÞð hboi di Þ
i¼1
"  # (B2)
X
N X
N X
N
2Ppe
þ ðAvri þ Abpi Þ hv di2 þ
hboi di D  Dpe þ di
i¼1 i¼1 i¼1
Sr
" #
X
N XN
þ nð1  pe ÞðP  Ppe  DÞð hboi di Þ ðAvri þ Abpi Þ
i¼1 i¼1

Setting β(n + 1) > β(n), and deleting redundant terms, we obtain:


"  #
Avs XN XN
2Ppe
2
hv di þ hboi di D  Dpe þ di
nþ1 i¼1 i¼1
Sr
" #
XN X N
þ ðn þ 1Þð1  pe ÞðP  Ppe  DÞð hboi di Þ ðAvri þ Abpi Þ
i¼1 i¼1
"  # (B3)
Avs XN XN
2Ppe
[ hv 2
di þ hboi di D  Dpe þ di
n i¼1 i¼1
Sr
" #
X N X
N
þ nð1  pe ÞðP  Ppe  DÞð hboi di Þ ðAvri þ Abpi Þ
i¼1 i¼1

Combining similar terms gives:


" #
X
N XN
ð1  pe ÞðP  Ppe  DÞð hboi di Þ ðAvri þ Abpi Þ
i¼1 i¼1
"  #
Avs XN XN
2Ppe
[ hv 2
di þ hboi di D  Dpe þ di
nðn þ 1Þ i¼1 i¼1
Sr

Dividing by the left hand-side term, the above inequality can be written as follows:

n2 þ n  d [ 0 (B4)
where:
 N  
P PN
Avs hv di2 þ hboi di D  Dpe þ 2Pp Sr
e
di
i¼1 i¼1
d¼ N  (B5)
P
N P
ð1  pe ÞðP  Ppe  DÞð hboi di Þ ðAvri þ Abpi Þ
i¼1 i¼1
22 H.K. Alfares and A.M. Attia

The positive root of the quadratic function (B4) is given by:


pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ 1 þ 4d
n[ (B6)
2
Continuing with the first-difference method, we next set β(n − 1) > β(n). Following the same approach described above leads to:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ 1 þ 4d
n\ (B7)
2
Combining (B6) and (B7), we obtain:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 1 þ 4d
n ¼ (B8)
2
Substituting the value of δ from (B5) into (B8) leads to:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 N ffi
u
u P 2 P N  
u 1 þ 4Avs hv di þ hboi di D  Dpe þ 2Pp e
Sr di
u
n¼u u i¼1 i¼1
N  (25)
t PN P
4ð1  pe ÞðP  Ppe  DÞð hboi di Þ ðAvri þ Abpi Þ
i¼1 i¼1

Potrebbero piacerti anche