You are on page 1of 37

Accepted Manuscript

Optimal Lot-Sizing under Strict Carbon Cap Policy Considering


Stochastic Demand

Arindam Ghosh , J.K. Jha , S.P. Sarmah

PII: S0307-904X(17)30133-6
DOI: 10.1016/j.apm.2017.02.037
Reference: APM 11625

To appear in: Applied Mathematical Modelling

Received date: 26 November 2014


Revised date: 11 January 2017
Accepted date: 14 February 2017

Please cite this article as: Arindam Ghosh , J.K. Jha , S.P. Sarmah , Optimal Lot-Sizing under Strict
Carbon Cap Policy Considering Stochastic Demand, Applied Mathematical Modelling (2017), doi:
10.1016/j.apm.2017.02.037

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and
all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT

Highlights

 Develop model for a supply chain considering strict carbon cap policy.

 Conceive stochastic demand for this model.

 Both backorder and lost sales have been considered.

 Major sources of emission and cost are taken into account.

T
 Offer managerial insights with numerical examples and sensitivity analysis.

IP
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

Optimal Lot-Sizing under Strict Carbon Cap Policy Considering Stochastic Demand

Arindam Ghosh*, J. K. Jha, and S. P. Sarmah

T
Department of Industrial and Systems Engineering,

IP
Indian Institute of Technology Kharagpur,

CR
Kharagpur - 721 302, India

US
*Corresponding author. Email: arindamiitan@gmail.com
AN
Phone: +91-9734578470

Abstract
M

Due to stringent laws and increasing awareness of customers, organizations are becoming environmental

conscious and looking beyond conventional cost minimization/profit maximization approach and striving
ED

to curb carbon emission. In this article, we have considered strict carbon cap policy to determine the

optimal order quantity, reorder point and number of shipments in a two-echelon supply chain under
PT

stochastic demand considering partial backorders. All the major sources of emission such as production,
CE

inventory, and transportation have been considered. The objective is to minimize the total expected

supply chain cost while satisfying the carbon emission constraint. A numerical example is given to
AC

illustrate the solution procedure. Further, an extensive sensitivity analysis is carried out to get insight

about how with operational adjustment cost and emission could be controlled.

Keywords: Supply chain; carbon emission; strict carbon cap; stochastic demand.

1. Introduction
ACCEPTED MANUSCRIPT

Environment pollution is a burning issue in recent era. Continuous emissions of greenhouse gases

(GHGs) into the atmosphere have raised the threat to environment and existence of human civilization.

Scientists have revealed that over the past 100 years, the average temperature of the earth has raised by

0.75°C. It is anticipated that between 2030 and 2050, approximately 250000 deaths will take place

annually due to weather change. Melting glaciers, rising sea levels, extreme weather conditions are the

T
results of global warming [1]. Therefore, governments, business houses, NGOs, and even customers pay a

IP
great deal of attention to reduce GHGs associated with industrialization. The Kyoto protocol, an

CR
international agreement linked to the United Nations framework convention on climate change, targeted

to reduce emission by 8% by 2012 from 1990 levels. Scientists have suggested more aggressive reduction

US
target of 25% of emission from 1990 level by 2020 [2]. Among all GHGs, carbon dioxide (CO2) is

considered as most important because it contributes 60% of enhanced the greenhouse effect [3]. As CO2
AN
is the predominant component of GHG, and so GHG emission is calculated in units of tones of CO2

equivalent, expressed as tCO2e [4].


M

To reduce carbon emission, governments and different regulatory bodies have started

implementing different carbon policies and initiated different carbon trading schemes. The existing
ED

carbon policies can primarily be grouped under three categories: carbon cost/tax, carbon cap-and-trade,

and strict carbon cap. In this paper, we consider a strict carbon cap policy. Under this policy, regulatory
PT

bodies allow organizations to emit CO2 to a certain threshold limit which is known as cap, and the penalty

for exceeding the cap is infinitely large [5]. Therefore, organizations are forced to manage their emission
CE

within the given limit. According to environmentalists, it is most effective policy to curb carbon emission.

Different countries, which are major emitters of CO2, are seriously considering adopting this policy across
AC

different industries. Barack Obama, the president of the United States of America set a strict limit for car

industry [6]. Similar type of policy is also likely to be extended for the power plants [7]. Some serious

discussion is going on that China may also implement this policy [8]. Apart from environmentalists and

governments, this policy has also become popular among the common people. A recent poll conducted by

Yale University found that 64% people support imposing strict carbon cap on existing power plant [9].
ACCEPTED MANUSCRIPT

Due to stringent laws, customer awareness, and stiff competitions, companies are also striving to

bring sustainable practices into their operational planning. Earlier, all the environmental initiatives were

organization-centric. But in the current scenario, companies cannot compete solely as individual entities

as today's businesses are supply chain oriented. All the big business houses are part of one or more supply

chains, and 50% of the industrial added values are derived from supply chain [10]. In recent years, the

T
number of standalone business houses is getting eroded [11]. So incorporation of environmental aspects

IP
into supply chain management (SCM) has become essential for sustainable business. Organizations like

CR
Walmart, Tesco, and Unilever, which run supply chain oriented businesses, are taking serious initiatives

to implement sustainable practices across their supply chains [12-14]. Integration of environment oriented

US
thinking into SCM including product design, material sourcing, manufacturing processes, delivery of the

product to the customers, and end-of-life management of the product after its use can be defined as green
AN
supply chain management (GSCM) [15]. GSCM can be considered as combination of different activities

such as green purchasing, green manufacturing, green distribution/marketing, and reverse logistics [16].
M

The perceptive of green supply chain has changed from a burden to a competitive unique selling

proposition [17]. Saadany et al. [11] confirmed with their model that incorporation of environmental
ED

related initiatives can even increase the profitability of the firms and also conceived that environment

quality of products do influence demand. Battini et al. [18] proposed a sustainable EOQ model
PT

considering the environmental impact of transportation and inventory. Recently, Jawed et al. [19]

proposed an exergetic version of the EPQ model to compute the cost of inventory in terms of energy units
CE

instead of monetary units using the thermodynamic concepts to derive a new exergetic sustainability

indicator for a production-inventory system.


AC

Production, inventory, and transportation contribute a major percentage of carbon emission in

supply chain. Therefore, the issue of carbon emission should not be ignored while taking operational

decisions. In recent years, many authors considered different carbon policies for optimizing production-

inventory models in supply chains. Bonney and Jaber, Hua et al., Bouchery et al., Toptal et al., and Chen

et al. [20-23, 5] derived optimal ordering/production decisions in a single-stage inventory model under
ACCEPTED MANUSCRIPT

different carbon policies. All of them considered constant demand and redesigned classical EOQ models

under different settings. Bouchery et al., Wahab et al., and Jaber et al. [22, 24-25] developed models

considering deterministic demand to minimize the total cost of the supply chain considering various

carbon policies. Wahab et al. [24] incorporated screening and holding cost of defective items in their

model. Bouchery et al. [22] redesigned classical EOQ type model as a multi-objective problem

T
considering sustainability criteria. Jaber et al. [25] considered different carbon policies and possible

IP
combination of these policies to develop mathematical models. Dobos, Absi et al., Li and Gu, Li, and

CR
Benjaafar et al. [26-30] all considered single stage inventory models in a finite planning horizon under

different carbon policies. Benjaafar et al. [30] considered constant demand, while others considered time

US
dependent deterministic demand. Dobos [26] reckoned production as the source of carbon emission and

expressed emission rate as a function of production rate. Absi et al. [27] formulated lot sizing problems
AN
considering periodic, cumulative, global and rolling carbon emission constraints separately. Li and Gu

[28] incorporated carbon banking and borrowing option under carbon cap-and-trade policy. Li [29]
M

extended the basic Arrow-Karlin model for deteriorating items with tradable emission permit. Benjaafar

et al. [30] developed four EOQ like models under different carbon policies and also extended their work
ED

for a serial supply chain.

Rosič and Jammernegg [31] extended the dual sourcing model based on the basic newsvendor
PT

model incorporating the environmental impact of transportation. Zhang and Xu [32] developed the multi-

item newsvendor production planning problem with carbon cap and trade mechanism assuming a
CE

common capacity constraint and carbon emission quota to produce multiple items. Swami and Shah [33]

and Dong et al. [34] developed single period models in two-echelon supply chain considering stochastic
AC

demand. Swami and Shah [33] focused on coordination between two players in a green supply chain,

where both the manufacturer and retailer put efforts for the greening of their operations. Dong et al. [34]

presented profit maximization models for both centralized and decentralized supply chain with carbon

cap-and-trade mechanism. Hammami et al. [35] formulated a deterministic multi-period model

considering both carbon cap and carbon tax policies with lead time constraint in a supply chain. Koca et
ACCEPTED MANUSCRIPT

al. [36] developed a stochastic lot sizing problem with controllable processing time and service level

constraint in a finite planning horizon and discussed the usefulness of convex compression cost function

with respect to carbon emission. Zanoni et al. [37, 38] developed deterministic models with different

emission policies. While Zanoni et al. [37] considered both emission cost and penalty, but Zanoni et al.

[38] considered only penalty due to emission. Bazan et al. [39] conceived the idea of multi-level emission

T
taxes and considered emissions from production and transportation while modifying two optimization

IP
models for a two-echelon vendor-buyer supply chain with deterministic demand and showed that

CR
environmental orientation can also be financially beneficial. Gurtu et al. [40] considered a deterministic

model for analyzing the impact of variation in fuel prices and infliction of carbon tax on batch sizes and

supply chain costs.


US
AN
< Table 1 is supposed to come here>
M

In Table 1, we have summarized the research articles dealing with production-inventory models

in infinite planning horizon considering carbon policies to highlight recent advances in this area. It
ED

confirms that papers with stochastic demand are sparse, and in spite of our sincere efforts we could not

find any paper which has considered both stochastic demand and carbon cap policy. In the seminal work
PT

of Benjaafar et al. [30], it is nicely pointed out that strict emission cap can be considered on a supply

chain if it were possible for firms to share their carbon emission caps. Further, they stated that such
CE

sharing might be possible either the firms are different divisions of a single large firm, carbon trading

between members of the same supply chain is allowed, or the cap is voluntary and the objective for the
AC

supply chain is to eventually certify that the end-product has a carbon foot print that does not exceed a

certain threshold.

Taking the cue from the work of Benjaafar et al. [30], we have considered that the buyer and

vendor are two separate divisions of a single organization. In this research, we investigate a strict carbon

cap policy in a two-echelon vendor-buyer supply chain under stochastic demand with partial backorder.
ACCEPTED MANUSCRIPT

The parent organization takes centralized decisions to optimize the total expected cost of the supply chain

under the strict carbon cap imposed on the parent organization. The stochastic demand during lead time

follows normal distribution. It is frequently conceived that the demand for the fast moving item

constitutes a large number of smaller demands from individual customers, so from the central limit

theorem, the resulting demand follows normal distribution [41]. When shortage takes place at the buyer,

T
some of the loyal customers may wait till the orders get fulfilled in the next replenishment cycle while the

IP
other customers may go somewhere else to satisfy their demand. In reality, it will be more generic to

CR
assume a combination of these two situations, where some of the demands are backordered and the rest

are lost during the stock out period [42]. Emissions from production, inventory, and transportation have

US
been taken into account to develop the model. We assume that the vehicle used for the transportation is

owned by the parent organization. Therefore, all the emissions we have considered as either Scope 1
AN
emissions or Scope 2 emissions. Scope 1 emissions are the emissions that arise directly from the sources

that are owned by the organization and Scope 2 emissions come from consumption of purchased
M

electricity (electricity consumption for warehousing the inventory). In reality, organizations can measure

and mitigate only Scope 1 and Scope 2 emissions. The structures for modeling of emissions from
ED

inventory and production are assumed similar to Chen et al. [5] and Benjaafar et al. [30]. Intuitively, it

seems that there is a convex relationship between transportation emissions and vehicle velocity. However,
PT

if the item is transported using a truck, for which the velocity varies within a normal operating range of

50-100 km/hour, the convex relationship may not be reflected. In this study, the emission from
CE

transportation is modeled based on the work of Bonney and Jaber [20] and Hua et al. [21]. Bonney and

Jaber [20] considered that the emission cost from transportation is directly proportional to transportation
AC

time, which is expressed as the ratio of distance and velocity. On the other hand, Hua et al. [21]

considered emissions from loaded and empty vehicles differently. We adopted these two distinct features

for modelling the emission from transportation.

The rest of the paper is organized as follows. Section 2 presents notation and assumptions.

Section 3 deals with the model formulation and solution technique. Section 4 presents a numerical
ACCEPTED MANUSCRIPT

example and sensitivity analysis. Finally, conclusions and some potential future directions are outlined in

Section 5.

2. Notation and assumptions

The following notation is used to develop the model.

T
Notation

IP
D average demand rate on the buyer

CR
S the vendor’s setup cost per production setup

P the vendor’s production rate

L
the buyer’s ordering cost per order

lead time of the buyer US


AN
b fraction of the demand during the stockout period that will be backordered, 0  b 1

η(r) expected demand shortage at the end of each cycle of the buyer
M

𝜃 the buyer’s shortage cost per unit short

𝜃0 the buyer’s profit per unit


ED

 standard deviation of demand per unit time

x the lead time demand which follows normal distribution with finite mean DL and standard
PT

deviation  L , where  L denotes the standard deviation of demand during lead time,  L   L
CE

d distance between the vendor and buyer

v velocity of the vehicle (truck)


AC

p production cost per unit item

hb the buyer’s holding cost per unit item per unit time

hv the vendor’s holding cost per unit item per unit time

t0 transportation cost per unit time when the vehicle is empty

tQ transportation cost per unit item per unit time when the vehicle is loaded
ACCEPTED MANUSCRIPT

f fixed carbon emission per production setup

 carbon emission per unit item due to production

b carbon emission per unit item per unit time due to inventory at the buyer

v carbon emission per unit item per unit time due to inventory at the vendor

0 carbon emission per unit time due to transportation when the vehicle is empty

T
Q

IP
carbon emission per unit item per unit time due to transportation when the vehicle is loaded

Ĉ cap (maximum limit) on carbon emission per unit time

CR
Decision Variables

r US
number of shipments from the vendor to the buyer per production cycle (a positive integer)

reorder point of the buyer


AN
Q the buyer’s ordering quantity per order

Assumptions
M

(i) There is a single vendor and single buyer, and they deal with a fast moving item.
ED

(ii) The buyer orders a lot of size Q and the vendor produces mQ units with a finite production rate P

(P>D) in one production setup but ships in quantity Q to the buyer over m times as shown in Fig. 1.
PT

The buyer reviews inventory continuously, and an order is placed whenever the inventory level drops

to the reorder point r and there are no orders outstanding.


CE

(iii) Any demand not met from stock is partially backordered during the stock out period.

(iv) Production, inventory, and transportation are the sources of emissions.


AC

(v) Transportation emission is assumed inversely proportional to truck velocity and the truck will not

speed.

<Fig. 1 is supposed to be inserted here>


ACCEPTED MANUSCRIPT

3. Model formulation

In this section, a mathematical model is formulated to find the optimal order quantity and reorder point of

the buyer, and the number of shipments between the vendor and buyer by minimizing the total expected

cost of the supply chain while satisfying the carbon emission constraint. The model is derived based on

approximate approach which parallels several such models in the literature including Hadley and Whitin

T
[43]. The buyer places an order after every Q demands at the average interval of time Q/D. On the other

IP
hand, the vendor produces mQ units in each production cycle in order to reduce the total production setup

CR
cost, and so the expected production cycle length of the vendor is given by mQ/D.

Fig.1 depicts the production, shipment and inventory variation patterns at the vendor and the

US
buyer. The buyer orders a lot of size Q, the vendor produces the item in a lot of size mQ with a constant

production rate P in each production cycle and ships to the buyer in m lots each of size Q. The first lot of
AN
size Q is ready for shipment after Q/P units of time after the start of the production, and then the vendor

continues the delivery on average every Q/D units of time. The total expected cost per unit time for the
M

supply chain comprises of the expected costs per unit time for the buyer and the vendor.

The total expected cost per unit time of the buyer is the sum of ordering cost, inventory holding
ED

cost, shortage cost and lost profit, which can be expressed as

AD Q  D  (1  b) D
TCb  Q, r    hb   r  DL  (1  b)  r     r   0  r  . (1)
PT

Q 2  Q Q

When the lead time demand follows normal distribution, the expected shortage per cycle can be obtained
CE

as [44]


 r  DL   r  DL 
 (r )    x  r  f  x  dx = L f     DL  r  G 
AC

, (2)
r  L   L 

where f(x) and G(x) are the standard normal density function and the complimentary cumulative

distribution function of the standard normal distribution, respectively.

Next, the total expected cost per unit time of the vendor comprises of setup cost, production cost,

inventory holding cost and transportation cost. The vendor’s average inventory is evaluated from Fig.1 as
ACCEPTED MANUSCRIPT

the difference of the vendor’s accumulated inventory and the buyer’s accumulated inventory, which is

developed based on the approach proposed by Joglekar [45]. Therefore, the average inventory of the

vendor can be written as

  Q Q  m 2Q 2   Q 2   D Q   D 2D 
          1  2    m  1     m  1   1
P 
mQ m 1 .
  P D  2P   D   mQ 2   P

T
Since the expected production cycle length of the vendor is mQ/D, and so the total expected cost

IP
per unit time for the vendor can be expressed as

CR
SD Q  D 2D  t0 dD tQ dD
TCv  Q, m    pD  hv  m 1    1    . (3)
mQ 2  P P  vQ v

Therefore, the total expected cost per unit time for the supply chain can be expressed as

TC  Q, m, r  
Q
AD
 pD 
mQ
Q
2
SD US  Q  D
 r  DL  1  b   r    hv m  1    1 
 hb 
 2  P
2D 
P 
AN
(4)
t dD tQ dD  D  (1  b) D
 0    r   0   r .
vQ v Q Q

Now, the major sources of carbon emissions are considered from production, inventory at the
M

buyer and vendor, and transportation. Accordingly, the total expected carbon emission per unit time from
ED

these sources can be derived as

Q  Q  D 2D 
  D   b   r  DL  1  b   r    v m 1   1 
fD
TE  Q, m, r   
mQ 2  2  P P .
PT

(5)
 dD  Q dD
 0 
vQ v
CE

As we are considering strict carbon cap policy, so the total expected carbon emission per unit

time should not exceed the specified limit Ĉ . Then, the carbon emission constraint can be written as
AC

Q  Q  D 2D   0 dD  Q dD ˆ .
  D  b   r  DL  1  b   r    v
fD
m 1   1    C (6)
mQ 2  2  P P  vQ v

Thus, the problem is to find the optimal order quantity Q, reorder point r and the number of

shipments m in each production cycle that minimize the total expected cost in Eq. (4) and satisfies the

carbon constraint given by Eq. (6).


ACCEPTED MANUSCRIPT

3.1. Solution methodology

The problem formulated in the previous section is a cost minimization constrained non-linear

programming problem. Before solving the problem, we have first proved the convexity of the total

expected cost equation TC(Q, m, r) with respect to Q, m and r. To do so, we have shown that the Hessian

T
matrix of TC(Q, r) is positive definite at point (Q, r) for fixed m (see Appendix A). Next, in order to

IP
check the convexity of TC(Q, m, r) with respect to m, we temporarily relax the integrality requirement on

CR
m and take the first and second derivatives of TC(Q, m, r) with respect to m for fixed (Q, r). We get

TC Q, m, r  SD Q D
  2  hv  1   (7)
m mQ 2 P

and

 2TC Q, m, r  2 SD
US
AN
 3 0. (8)
m2 mQ

Therefore, TC(Q, m, r) is convex with respect to m.


M

Now, initially we ignore the carbon constraint and derive the optimal value of Q and r for fixed m

as (see Appendix B)
ED

1/2
  S t0 d 
 2D  A      r    0 1  b   r   
Q0    m v  (9)
PT

   D 2D  
 hb  hv  m 1   1   
   P P  
CE

and

 r  DL  Qhb
G   . (10)
  L  Qhb 1  b    D  0 1  b  D
AC

Thus, r can be found using the following equation (please see Eq. (B.7) in Appendix B).

  Qhb 
r  DL   NORMSINV 1     L . (11)
  Qhb 1  b    D  0 1  b  D  
ACCEPTED MANUSCRIPT

To get the optimal value of Q and r, we have used the iterative procedure of Hadley and Whitin [43],

which has been discussed in the proposed algorithm.

So far, we have not considered the carbon constraint for determining the order quantity Q0 and r,

and this may violate the carbon constraint. So we need to determine an optimal value of Q and r for fixed

m that satisfies the carbon constraint. For the purpose, we have adopted the method proposed by Chen et

T
al. [5]. Now, the carbon constraint in Eq. (6) can be written as

IP
fD
mQ
  D   b   r  DL  1  b   r    v  m 1 

2
Q


Q
2
D
P
1  
2 D   0 dD  Q dD ˆ

P  vQ

v
C  0. (12)

CR
The roots of the corresponding quadratic equation of the above inequality are given by

1/ 2


US  dD ˆ  
2

  b  r  DL  1  b   r     D  Q C 
 Q dD   
  b  r  DL  1  b   r    
v
Cˆ   D  
 2     m 1  D   1  2 D   fD   0 dD  
 
v
  b v    
P    m v  
AN
   P
Q1  (13)
  D 2D 
 b   v  m 1    1 
  P  P 
M

and

1/ 2
  Q dD ˆ  
2

  b  r  DL  1  b   r     D 
ED

C 
 Q dD   
  b  r  DL  1  b   r    
v
Cˆ   D  
 2      m 1  D   1  2 D    fD   0 dD  
v
  b v 
 

P
 
P    m v  
Q2  
PT

. (14)
  D 2D 
 b   v  m 1    1 
  P P 
CE

Where Q1 and Q2 are the lower and upper bounds, respectively, for the feasible range of Q, and they are

positive (see Appendix C).


AC

Now, for fixed m, the optimal Q will be obtained at Q̂ as given below in Eq. (15), which will

satisfy the carbon constraint.

Qo , if Q1  Q0  Q2

Qˆ  Q1 , if Qo  Q1 . (15)

Q2 , if Qo  Q2
ACCEPTED MANUSCRIPT

The above condition can easily be explained using Fig. 2.

<Fig. 2 is supposed to be inserted here>

Selecting the appropriate value of Q̂ from Eq. (15) and putting in Eq. (11) as Q  Qˆ , one can get the

T
value of r for fixed m. Since Q and r are interdependent, so it will be difficult to derive the value of one

IP
variable explicitly, without knowing the other one. Therefore, we assume deterministic demand for

CR
initialization and set  L  0 in Eqs. (9), (13) and (14) and get an initial value of Q0, Q1, and Q2 as

1/ 2
  S td 
 2D  A   0  
Q 
0
 m v   ,
 h  h  m 1  D  1  2 D  
 b v   P 


P  
US (16)
AN
1/2
 dD   dD ˆ  2 D    fD  0 dD  
2
   D
Cˆ   D  Q    D  Q  C   2   b  v  m 1   1    
v  v     P P    m v  
Q1 
 (17)
M

  D 2D 
 b  v m 1   1 
  P P 
ED

and

1/2
 Q dD   Q dD ˆ  2 D    fD  0 dD  
2
ˆ    D
C  D     D   C   2   b  v  m 1   1    
P    m v  
PT

v  v     P 
Q2  . (18)
  D 2D 
 b  v m 1   1 
  P P 
CE

Using the value of Q0, Q1 and Q2 obtained from Eqs. (16), (17) and (18), respectively, we select an initial
AC

value of Q̂ satisfying the condition given in Eq. (15). Next, using the value of this Q̂ in Eq. (11), we find

the initial value of r. This r in turn is used in Eqs. (9), (13) and (14) to determine the new value of Q0, Q1,

and Q2, and then the condition given in Eq. (15) to find the new value of Q̂ . We repeat this iterative

procedure until the convergence of Q̂ and r [43].


ACCEPTED MANUSCRIPT

Till this point, we have discussed the procedure to find the optimal value of Q and r for fixed m.

Further, it has been shown that TC(Q, r, m) is convex with respect to m by relaxing the integrality

requirement on m. Based on these properties, the following algorithm is developed to determine the optimal

Q, r and m.

T
Algorithm

IP
Step 0: Set m=1

CR
Step 1: Compute initial Q0 , Q1 and Q2 from Eqs. (16), (17) and (18), respectively.

Step 2: Select appropriate value of Q̂ satisfying the condition given in Eq (15).

Step 3: Compute r using Q̂ in Eq. (11).


US
AN
Step 4: Find Q0, Q1 and Q2 using Eqs. (9), (13) and (14), respectively. Select the appropriate value of Q̂

using the condition in Eq. (15).


M

Step 5: Repeat Steps 3 and 4 until no change occurs in the value of Q̂ and r.


Step 6: Set Qˆ ( m)  Qˆ and r m  r . Thus, Qˆ ( m ) , r( m)  is the optimal solution for fixed m and compute
ED

 
TC Qˆ ( m) , r( m) , m using Eq. (4).
PT


Step 7: Set m  m  1 and repeat Steps 1 to 6 to get new TC Qˆ( m) , r( m) , m . 
   
Step 8: If TC Qˆ m , r m , m  TC Qˆ m 1 , r m 1 , m  1 go to Step 7; otherwise go to Step 9.
CE

 
Step 9: Set TC  Q* , r * , m*   TC Qˆ( m1) , r( m1) , m  1 , then  Q* , r * , m*  is the optimal solution.
AC

Using the above stepwise procedure, the optimal order quantity, reorder point and number of

shipments per production cycle can be obtained. Further, the total expected cost and the total expected

carbon emission can be determined using Eqs. (4) and (5), respectively. In the next section, we have

illustrated the model through a numerical example and examined the impact of some key parameters.
ACCEPTED MANUSCRIPT

4. Numerical example and sensitivity analysis

To illustrate the proposed model and solution procedure, a two-echelon supply chain is considered with

the following data [24, 40, 30, 46]: D = 600 units per year, P = 2000 units per year, A = $200 per order, S

= $1500 per setup, θ = $50 per unit, θ0 = $65 per unit, d =100 km, v =50 km per hour, p = $200 per unit,

T
hb = $5 per unit per year, hv = $2 per unit per year, t0 = $10 per hour, tQ = $5 per unit per hour,  b = 0.80

IP
Ton per unit per year,  v = 0.80 Ton per unit per year, f = 20 Ton per setup,  = 2 Ton per unit,  0 = 0.03

CR
Ton per hour,  Q = 0.004 Ton per unit per hour, b = 0.6,  = 1 unit per day, L = 7 days, Ĉ = 1550 Ton.

(Assume one year = 365 days).

US
Now, using the proposed solution procedure, we determine the optimal value of Q, r and m.

Subsequently, the corresponding value of TC and TE are found by putting the optimal value of decision
AN
variables in Eqs. (4) and (5), respectively. The procedure for finding the optimal solution for the base case

numerical example is summarized in Table 2. For the base case, we get m* = 4, Q* = 241.56, r* = 16.64,
M

TC =128692.23, and TE = 1550.


ED

<Tables 2 is supposed to be inserted here>

Next, sensitivity analysis is carried out with respect to some key parameters to draw certain

managerial insights. In most of the cases (except sensitivity analysis for Ĉ), we have varied the value of
PT

each key parameter within a range such that the carbon constraint remains active, i.e. TE is always equal
CE

to Ĉ. In the sensitivity analysis, we have not shown emission from production, as it is independent of the

decision variables and remains constant.


AC

4.1. Effects of  b and  v

<Tables 3(a) and 3(b) is supposed to be inserted here>

<Fig. 3 is supposed to be inserted here>

Fig. 3 indicates that as  v or  b increases, TC increases at faster rate with respect to  v compared to

 b . Further, Tables 3(a) and 3(b) show that the buyer’s expected cost decreases first and then increases as
ACCEPTED MANUSCRIPT

long as the number of shipments (m) during each production cycle remains unchanged with increase in

 v or  b , whereas the vendor’s cost increases. When these inventory emission parameters increase,

order quantity decreases as long as m remains unchanged and the vendor is required to produce in smaller

lot size (mQ) in order to reduce the average inventory level. Due to smaller production lot size, the

production setup and transportation related costs increase significantly, which in turn increases the total

T
expected cost of the vendor, so the total expected cost.

IP
4.2. Effects of hb

CR
< Table 4 is supposed to be inserted here>

< Fig. 4 is supposed to be inserted here>

US
The results for the variation of hb are summarized in Table 4, and Fig. 4 shows the carbon emissions from

inventory at the buyer’s end and at the vendor’s end with hb. When there is no change in m with hb, we
AN
find a negligible change in Q, and so the emissions due to inventory at the buyer and at the vendor remain

unchanged. With increasing hb if m increases, the buyer should order in a smaller lot size and the average
M

inventory at the buyer reduces, which decreases the emission due to inventory at the buyer. On the other

hand, production lot size (mQ) increases with increase in hb, which results in higher average inventory at
ED

the vendor, and so increased emission at the vendor due to inventory. It can be observed from Table 4 that

the total expected cost and the expected cost of the buyer increase with hb. Also, the vendor’s expected
PT

cost increases if the number of shipments (m) increases with hb, but decreases slightly if m remains

constant.
CE

4.3. Effects of velocity

< Table 5 is supposed to be inserted here>


AC

< Fig. 5 is supposed to be inserted here>

Fig. 5 depicts how the total expected cost (TC) and the emission from transportation change with velocity

(v) of the vehicle (truck). As discussed earlier, we have varied the value of v in such a way that the total

expected emission (TE) remains constant and always equals to Ĉ. It is clearly observed from Fig. 5 that

both TC and emission from transportation will come down with increase in v, as both of them are
ACCEPTED MANUSCRIPT

inversely related to v. The variations in decision variables with v are shown in Table 5. We observe that

when v increases, order quantity and the buyer’s expected cost do not change much, but the vendor’s cost

comes down significantly as the transportation cost decreases and this leads to decrease in TC.

4.4. Effects of carbon cap

<Table 6 is supposed to be inserted here>

T
<Fig. 6 is supposed to be inserted here>

IP
Fig. 6 shows the variation of TC and TE with Ĉ. In strict carbon cap policy, organizations cannot emit

CR
carbon beyond the allotted carbon limit. But if the constraint due to carbon cap is relaxed, organizations

can minimize their cost by optimizing their decisions without bothering about the carbon emissions.

US
Therefore, when the carbon constraint is active with increase in Ĉ, TC can be reduced by fully exhausting

the allotted carbon limit. But beyond a certain threshold level of Ĉ, the constraint becomes inactive and
AN
organizations do not need to emit more to reduce the cost. Now, to explain the behavior of the model

when the carbon constraint remains active, the results are presented in Table 6 for the various values of Ĉ.
M

Table 6 shows that the number of shipments increases with increase in Ĉ, and the vendor tends to ship in

larger lot size if m remains unchanged with Ĉ. As far as emission is concerned, with increasing Ĉ initially,
ED

emissions at the buyer and the vendor increase as the order quantity Q increases and leads to increase in

emission due to inventory holding at the buyer and at the vendor. If m increases with further increase in
PT

Ĉ, Q decreases, and so the emission at the buyer decreases. On the other hand, production lot size (mQ)

increases with Ĉ, so the emission at the vendor increases.


CE

4.5. Effects of production rate

< Table 7 is supposed to be inserted here>


AC

< Fig.7 is supposed to be inserted here>

The results for effect of production rate is presented in Table 7, and Fig. 7 shows the variation patterns of

carbon emissions from inventory at the buyer and at the vendor with increase in P. When P increases, m

remains initially unchanged with decrease in Q, which decreases the emission from the inventory at the

buyer. But as P increases, the vendor produces the production lot quickly, and the average inventory and
ACCEPTED MANUSCRIPT

the emission from the inventory at the vendor increase. At a threshold level, increase in P leads to

decrease in m and mQ but with increased value of order lot size Q. Due to increase in Q, the emission

from inventory at the buyer will increase, but the emission at the vendor from the inventory decreases as

the production lot size mQ decreases. This pattern of sudden change in emissions from the inventory at

the buyer and the vendor repeats whenever m decreases with P. But when m remains constant, the

T
emissions from the buyer and the vendor will exhibit the earlier pattern.

IP
Table 7 shows that with increase in P when m remains constant, Q decreases, and so the average

CR
inventory level at the buyer falls down and thereby decreases the expected cost of the buyer. But the

average inventory level at the vendor and the expected cost of the vendor increase when m remains

US
constant with increase in P. Now, if m decreases with increase in P, the buyer needs to order in a larger

lot size and thereby it increases the total expected cost of the buyer mainly due to increased inventory
AN
holding cost of the buyer. But when m reduces with increase in P, the expected cost of the vendor

increases due to increase in production setup cost, though it reduces inventory holding cost of the vendor.
M

4.6. Effects of emission from production setup

< Table 8 is supposed to be inserted here>


ED

< Fig. 8 (a) and Fig. 8 (b) are supposed to be inserted here>

From Table 8, it can be seen that the order quantity Q decreases with increase in emission from setup f.
PT

As a result, the emission from the inventory both at the buyer and at the vendor decreases. The emission

from transportation is inversely related with order quantity Q, and so the decrease in Q with increase in f
CE

also increase the transportation related emission. But due to increase in setup related emission at the

vendor, there is overall increase in emission at the vendor including transportation (See Fig. 8(a)).
AC

Further, Fig. 8(b) shows that TC increases with f, because the ordering, transportation and setup cost

components increase with decrease in the order quantity Q.

5. Conclusions
ACCEPTED MANUSCRIPT

In this paper, a two-echelon supply chain consisting of a single-vendor and a single-buyer is considered.

In practice this model is useful for a firm having two separate divisions that are trying to optimize the

total expected cost by integration of their operations considering carbon emission constraint. We have

considered infinite planning horizon and emissions from all the major sources in the two-echelon, namely

production, inventory, and transportation. The total expected cost of the vendor–buyer integrated system

T
is minimized by simultaneously optimizing the ordering quantity, reorder point and the number of

IP
shipments from the vendor to the buyer in each production cycle in a stochastic demand environment

CR
under strict carbon cap policy.

This model will help organizations to determine the optimal production-inventory policy.

US
Sensitivity analysis has revealed that organization needs to be more careful about emission due to

inventory at the vendor’s end as compared to emission due to inventory at the buyer’s end, since latter has
AN
less effect on TC. Therefore, organization should take more precautionary measures at the vendor level to

check emission. The sensitivity analysis also reveals how with changing different parameters, emissions
M

from different sectors change and balance each other. We have shown the effect of allotted carbon cap (Ĉ)

on TC and TE. It is shown that after a certain level, carbon constraint becomes inactive. Therefore, at the
ED

time of allocating carbon cap, regulatory bodies should take this thing into account. This work will help

both industries and regulatory bodies in implementing strict carbon cap policy quite effectively and
PT

efficiently. We have also shown that how TC and emission varies with v. Both TC and emission from

transportation come down with increasing velocity, so an investment to improve vehicle’s velocity is
CE

worthy for the organization. Our study also reveals that emission from set up has a sound impact on TC,

so organizations need to be cautious to check the setup related emission.


AC

In future, the paper can be extended in many directions. We have considered a two-echelon

supply chain. A more complex model with multi-echelon or reverse supply chain can be explored. In this

paper, we have considered normal items. But as emissions from different sources are product dependent, a

significant change can be witnessed in sensitivity analysis if one considers perishable product. Further,
ACCEPTED MANUSCRIPT

one can consider finite planning horizon problem. A reverse supply chain with defective and waste item

can be a good direction for future study.

Appendix A

To prove the convexity of TC(Q, m, r) with respect to Q and r, we need to show that the Hessian matrix

T
of TC(Q, m, r) is positive definite at point (Q, r) for fixed m. The Hessian matrix H is given by

IP
  2TC    2TC  
 
Q 2 Qr 
H 2 , where TC   TC Q, m, r  .

CR
  TC    2TC  
 
 rQ r 2 

 2TC  
Q 2

2D 
3 
Q  m v

A   0    0 1  b   ( r )   0 ,
S td
 US
AN
 2TC      r  DL 
  hb 1  b  
D
  0 1  b   f 0
r    L 
2
Q
M

and

 2TC    2TC   D  r  DL 
2  0
   1  b     G  .
Qr rQ  L 
ED

Also, we have
PT

 2TC     2TC     2TC     2TC   


  
Q 2
r 2
Qr r Q
 r  DL   S t0 d    
 hb 1  b      0 1  b     hb 1  b     0 1  b   (r ) 
2D D
CE

 f   A   
Q3   L   m v  Q  
D2   r  DL    r  DL   
2

 4    0 1  b   2 (r ) f 
     0
2
   G 
AC

Q          
 L L

2
 r  DL    r  DL    r  DL 
since 2 (r ) f     G     0 for  (r )  0 and f    0 (see Ouyang et al. [46]) .
 L    L   L 

Thus, TC(Q, m, r) is convex in Q and r for fixed m.


ACCEPTED MANUSCRIPT

Appendix B

Putting the expression of  (r ) from Eq. (2) in Eq. (4), we get

AD SD Q   r  DL   r  DL    Q  D 2D 
TC  Q, m, r    pD   h   r  DL  (1  b)   L f     DL  r  G      hv m 1    1  
Q mQ b  2    L    L    2   P  P . (B.1)
t0 dD tQ dD  D   r  DL   r  DL   0 (1  b) D   r  DL   r  DL  
    f     DL  r  G     L f     DL  r  G  
vQ v Q  L   L    L   Q    L    L  

T
Taking the first order derivatives of Eq. (B.1) with respect to Q and r for fixed m, we get

IP
TC  Q, m, r  AD SD hb hv   D  2 D  t0 dD
    m 1    1  
Q Q2 mQ 2 2 2   P P  vQ 2 (B.2)

CR
  D  (1  b) D    r  DL   r  DL  
 2  0 2   L f     DL  r  G  
Q Q   L    L 

and

TC  Q, m, r 
r

 hb   hb 1  b  


US
 D 0 1  b  D    r  DL  
  G     .
 
(B.3)
AN
Q Q

Setting Eqs. (B.2) and (B.3) to zero, we have


1/2
 
M

 S t0 d
 2D  A      r    0 1  b   r   
Q  m v  (B.4)
   D 2D  
 hb  hv  m 1   1   
ED

   P P  

and
PT

 r  DL  Qhb
G  . (B.5)
  L  Qhb 1  b    D  0 1  b  D
CE

Using the inbuilt function “NORMSINV” of Microsoft EXCEL for finding the cumulative distribution

function of standard normal, Eq. (B.5) can be written as


AC

r  DL  Qhb 
 NORMSINV 1   . (B.6)
L  Qhb 1  b    D  0 1  b  D 

Therefore, r can be found as

  Qhb 
r  DL   NORMSINV 1     L . (B.7)
  Qhb 1  b    D   0 1  b  D  
ACCEPTED MANUSCRIPT

Appendix C

From Eqs. (13) and (14), we have

1/ 2
  Q dD ˆ  
2

  b  r  DL  1  b   r     D  C 
 Q dD
  b  r  DL  1  b   r      
v
Cˆ   D 
  2 D    fD  0 dD  
 2   b   v  m  1    1 
v D

T
   
    P P   m v  
Q1  (C.1)

IP
  D 2D 
b   v m 1    1 
  P  P 

CR
and

1/ 2
  Q dD ˆ  
2

 b 
              

US
r DL 1 b r D C
 Q dD
  b  r  DL  1  b   r      
v
Cˆ   D 
     
 2   b   v  m  1    1 
v D 2 D fD dD
   0

    P P   m v  
Q2  . (C.2)
AN
  D 2D 
b   v m 1    1 
  P P 

Now, Eq. (6) can be written as


M

Cˆ 
fD
mQ
  D   b   r  DL  1  b  (r )    v m 1 

2
Q


Q
2
D
P
1 
P  vQ  
2 D   0 dD  Q dD

v
0. (C.3)
ED

 Cˆ   D 
 Q dD
v
  b  r  DL  1  b   r    b   v m 1 
Q
2
Q
2
D
P
1 
P  mQ  
2 D  fD  0 dD

vQ
. (C.4)
PT

Since,  v
Q

2
m 1  
D
P
1 
2D 
P 
 0,for m  1 and P  D .
CE

 Q dD   D 2D 
Therefore, Cˆ   D    b  r  DL  1  b   r  and  b   v  m  1   1 will always be
v   P P 
AC

positive.

 Q dD
2
     D 2 D   fD  0 dD 
Now, if   b  r  DL  1  b   r     D   Cˆ   2   b   v  m  1    1    0
 v     P  P  m v 

holds true, we can conclude that both Q1 and Q2 will be positive.

Again, from Eq. (6), we have


ACCEPTED MANUSCRIPT

Cˆ   D 
 Q dD
v
  b  r  DL  1  b   r    b   v m 1 
Q
2
Q
2
D
P
1
2 D  fD  0 dD

P  mQ

vQ  
. (C.5)

On squaring both sides, we get

 Q dD
2 2
ˆ   Q Q  D 2 D  fD  0 dD 
C   D   b  r  DL  1  b   r    b   v  m 1    1     . (C.6)
 v   2 2  P P  mQ vQ 

T
 Q Q  D 2 D    fD  0 dD 
On subtracting 4  b   v  m 1    1      from both sides, we obtain
 2 2  P P    mQ vQ 

IP
2
ˆ  Q dD   Q Q  D 2 D    fD  0dD 
  b  r  DL  1  b   r     4  b   v  m 1    1 

CR
C   D     
 v   2 2   P  P    mQ vQ 
2
. (C.7)
 Q Q  D 2 D   fD  0dD   Q Q  D 2D    fD  0dD 
  b   v  m  1    1       4  b   v  m 1    1     
 2 2   P  P  mQ vQ   2 2   P  P    mQ vQ 

On simplification, we have

  Q dD ˆ 
2

US   D 2 D    fD  0 dD 
AN
 b  r  DL  1  b   r     D   C  2  b   v  m 1    1    
 v     P P    m v 
2
. (C.8)
 Q Q  D 2 D    fD  0 dD  
  b   v  m  1    1      
 2 2  P P    mQ vQ  
M

2
 Q Q  D 2 D    fD  0 dD  
Since   b   v  m 1    1        0 .
ED

 2 2  P P    mQ vQ  

 Q dD
2
     D 2 D    fD  0 dD 
Hence, b  r  DL  1  b   r     D   Cˆ   2  b   v  m 1    1     0.
P    m v 
PT

 v     P

Thus, it can be concluded that both Q1 and Q2 are positive.


CE

References
AC

[1] World Health Organization. Climate change and health. (2014). Available from <http://www.
who.int/mediacentre/factsheets/fs266/en/>, 2015 (17. 07. 2015).
[2] Kinaxis. (2009). Providing carbon footprint visibility and planning capabilities across the supply
chain: why you need to do it and what you need to do it, Kinaxis Inc, Ottawa.
[3] BBC Weather Centre. Carbon change. (2014). Available from <http://www.bbc.co.uk /climate
/evidence/carbon_ dioxide.shtml>, 2015 (17. 07. 2015).
ACCEPTED MANUSCRIPT

[4] N. Abhishek, M.B. Nidhi, V.M. Pillai, Carbon assessment model for supply chain network.
International Conference on Green Technologies (ICGT), IEEE, Trivandrum, December 18-20,
2012, pp. 329–334.
[5] X. Chen, S. Benjaafar, A. Elomri, The carbon-constrained EOQ, Oper. Res. Lett. 41 (2) (2013) 172–
179.
[6] S. Goldenberg. Obama sets strict limits on car exhaust emissions. The Guardian 2009 May 20.
Available from <http://www.theguardian.com/world/2009/may/19/obama-carbon-emissions-auto-

T
industry>, 2016 (03.05.2016).

IP
[7] J.M. Broder. Obama readying emissions limits on power plants. The New York Times 2013 June 19.
Available from: <http://www.nytimes.com/2013/06/20/science/earth/obama-preparing-big-effort-to-

CR
curb-climate-change.html?pagewanted=all&_r=1>, 2016 (03.05.2016).
[8] A. Vaughan and T. Branigan. China to limit carbon emissions for first time, climate adviser claims

US
The Guardian 2014 June 3. Available from: <http://www.theguardian.com/environment/
2014/jun/03/china-pledges-limit-carbon-emissions>, 2016 (03.05.2016).
[9] L.B. Lopez. Poll: Majority support carbon limits on coal plants. The Hill 2014 May 29. Available
AN
from: <http://thehill.com/policy/energy-environment/207573-poll-majority-support-carbon-limits-on
-coal-plants>, 2016 (03.05.2016).
[10] F. Ciliberti, P. Pontrandolfo, B. Scozzi, Logistics social responsibility: standard adoption and
M

practices in Italian companies, Int. J. Prod. Econ. 113 (1) (2008) 88–106.
[11] A.M.A. El Saadany, M.Y. Jaber, M. Bonney, Environmental performance measures for supply
ED

chains, Manag. Res. Rev. 34 (11) (2011) 1202–1221.


[12] Walmart. Beyond 50 Years: Building a sustainable future. 2012. Global Responsibility Report.
Walmart, USA. Available from <http://c46b2bcc0db5865f5a76-
PT

91c2ff8eba65983a1c33d367b8503d02.r78.cf2.rackcdn.com/d3/35/66be9cc44c2b8d096565166e79f4
/2012-global-responsibility-report_129823695403288526.pdf> 2016, (28.04.2016).
CE

[13] Business in the community. How Tesco is building its sustainability strategy through its supply
chain. Business in the Community, 2015 April 9. Available from <http://www.bitc.org.uk/blog/post
AC

/how-tesco-building-its-sustainability-strategy-through-its-supply-chain> 2016, (28.04.2016).


[14] L. Beavis, Unilever: design once, deploy everywhere. The Guardian, 2015 April 20. Available from
<http://www.theguardian.com/sustainable-business/2015/apr/30/unilever-design-once-deploy-
everywhere> 2016, (28.04.2016).
[15] S.K. Srivastava, Green supply-chain management : A state-of- the-art literature review, Int. J.
Manag. Rev. 9 (1) (2007) 53–80.
ACCEPTED MANUSCRIPT

[16] N. Dheeraj, N. Vishal, An overview of green supply chain management in India, Res. J. Recent Sci.
1 (6) (2012) 77–82.
[17] R.I. Van Hoek, From reversed logistics to green supply chains, Supply Chain Manag. An Int. J. 4 (3)
(1999) 129–135.
[18] D. Battini, A. Persona, F. Sgarbossa, A sustainable EOQ model: Theoretical formulation and
applications, Int. J. Prod. Econ. 149 (2014) 145–153.
[19] H. Jawad, M.Y. Jaber, M. Bonney, M.A. Rosen, Deriving an exergetic economic production

T
quantity model for better sustainability. Appl. Math. Model. 40 (11) (2016) 6026–6039.

IP
[20] M. Bonney, M.Y. Jaber, Environmentally responsible inventory models: Non-classical models for a
non-classical era, Int. J. Prod. Econ. 133 (2011) 43–53.

CR
[21] G. Hua, T.C.E. Cheng, S. Wang, Managing carbon footprints in inventory management, Int. J. Prod.
Econ. 132 (2011) 178–185.

US
[22] Y. Bouchery, A. Ghaffari, Z. Jemai, Y. Dallery, Including sustainability criteria into inventory
models, Eur. J. Oper. Res. 222 (2) (2012) 229–240.
[23] A. Toptal, H. Özlü, D. Konur, Joint decisions on inventory replenishment and emission reduction
AN
investment under different emission regulations, Int. J. Prod. Res. 52 (1) (2014) 243–269.
[24] M.I.M. Wahab, S.M.H. Mamun, P. Ongkunaruk, EOQ models for a coordinated two-level
international supply chain considering imperfect items and environmental impact, Int. J. Prod. Econ.
M

134 (1) (2011) 151–158.


[25] M.Y. Jaber, C.H. Glock, A.M. A. El Saadany, Supply chain coordination with emissions reduction
ED

incentives, Int. J. Prod. Res. 51 (1) (2013) 69–82.


[26] I. Dobos, Tradable permits and production-inventory strategies of the firm, Int. J. Prod. Econ. 108
(2007) 329–333.
PT

[27] N. Absi, S. Dauzère-Pérès, S. Kedad-Sidhoum, B. Penz, C. Rapine, Lot sizing with carbon emission
constraints, Eur. J. Oper. Res. 227 (1) (2013) 55–61.
CE

[28] S. Li, M. Gu, The effect of emission permit trading with banking on firm’s production–inventory
strategies, Int. J. Prod. Econ. 137 (2) (2012) 304–308.
AC

[29] S. Li, Optimal control of the production–inventory system with deteriorating items and tradable
emission permits, Int. J. Syst. Sci. 45 (11) (2014) 2930–2401.
[30] S. Benjaafar, Y. Li, M. Daskin, Carbon footprint and the management of supply chains : insights
from simple models, IEEE Trans. Autom. Sci. Eng. 10 (1) (2013) 99–116.
[31] H. Rosič, W. Jammernegg, The economic and environmental performance of dual sourcing: a
newsvendor approach, Int. J. Prod. Econ. 143(1) (2013) 109–119.
ACCEPTED MANUSCRIPT

[32] B. Zhang, L. Xu, Multi-item production planning with carbon cap and trade mechanism, Int. J. Prod.
Econ. 144 (1) (2013) 118–127.
[33] S. Swami, J. Shah, Channel coordination in green supply chain management, J. Oper. Res. Soc. 64
(3) (2012) 336–351.
[34] C. Dong, B. Shen, P.S. Chow, L. Yang, C.T. Ng, Sustainability investment under cap-and-trade
regulation, Ann. Oper. Res. 240 (2) (2016) 509-531.

[35] R. Hammami, I. Nouira, Y. Frein, Carbon emissions in a multi-echelon production-inventory model

T
with lead time constraints, Int. J. Prod. Econ. 164 (2015) 292–307.

IP
[36] E. Koca, H. Yaman, M. Selim Aktürk, Stochastic lot sizing problem with controllable processing
times, Omega. 53 (2015) 1–10.

CR
[37] S. Zanoni, L. Mazzoldi, M.Y. Jaber, Vendor-managed inventory with consignment stock agreement
for single vendor–single buyer under the emission-trading scheme, Int. J. Prod. Res. 52 (1) (2013)
20–31.
US
[38] S. Zanoni, L. Mazzoldi, L.E. Zavanella, M.Y. Jaber, A joint economic lot size model with price and
environmentally sensitive demand, Prod. Manuf. Res.: An open access Journal. 2 (1) (2014) 341–
AN
354.
[39] E. Bazan, M.Y. Jaber, S. Zanoni, Supply chain models with greenhouse gases emissions, energy
usage and different coordination decisions, Appl. Math. Model. 39 (17) (2015) 5131–5151.
M

[40] A. Gurtu, M.Y. Jaber, C. Searcy, Impact of fuel price and emissions on inventory policies, Appl.
Math. Model. 39 (3-4) (2015) 1202–1216.
ED

[41] S. Axsäter, Inventory Control, third ed., Springer, Switzerland, 2015.


[42] D.C. Montgomery, M.S. Bazaraa, A.K. Keswani, Inventory models with a mixture of backorders and
lost sales, Naval Res. Logist. Q. 20 (1973) 255–263.
PT

[43] G. Hadley, T. Whitin, Analysis of Inventory Systems, Prentice-Hall, Englewood Cliffs, New Jersey,
1963.
CE

[44] A. Ravindran, D.T. Phillips, J.J. Solberg, Operations Research: Principles and Practice, second ed.,
Wiley, New York, 1987.
AC

[45] P.N. Joglekar, Note-Comments on “A quantity discount pricing model to increase vendor profits”,
Manag. Sci. 34 (11) (1988) 1391–1398.
[46] L.Y. Ouyang, K.S. Wu, C.H. Ho, Integrated vendor–buyer cooperative models with stochastic
demand in controllable lead time, Int. J. Prod. Econ. 92 (2004) 255–266.
ACCEPTED MANUSCRIPT

Production cycle
Production and shipment Shipment

mQ
Vendor’s accumulated
inventory level
Vendor’s stock

T
Q

IP
Q
Q

CR
Q
Q Q

Q/P Q/D Q/D

mQ/P USQ/D

mQ/D
Q/D Time
AN
Buyer’s stock

M
ED

Buyer’s accumulated
Q
inventory level

Q
PT

r Q Q Reorder point
Q Q Q Q Q Q
CE

L Q/D
Time
(m-1)Q/D
AC

Fig. 1: The inventory pattern for the vendor and the buyer.
ACCEPTED MANUSCRIPT

T
IP
Total Cost
Total Cost

Total Cost

CR
Q1 Q0 Q2

(i) Q̂  Q0
US Q0

(ii) Q̂  Q1
Q1 Q2 Q1 Q2 Q0

(iii) Q̂  Q2
AN
Fig. 2: Illustration of optimal order quantity Q̂ for fixed m for all possible cases.
M
ED
PT
CE
AC

Fig. 3: Variation of TC with  b and  v .


ACCEPTED MANUSCRIPT

T
IP
CR
US
Fig. 4: Variation of emission due to inventory at the buyer and the vendor with hb.
AN
M
ED
PT
CE
AC

Fig. 5: Variation of TC and emission due to transportation with velocity.


ACCEPTED MANUSCRIPT

T
IP
CR
US
Fig. 6: Variation of TC and TE with carbon cap.
AN
M
ED
PT
CE
AC

Fig. 7: Variation of emission due to inventory at the buyer and the vendor with production rate.
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
Fig. 8(a): Variation of emissions with f.
M
ED
PT
CE
AC

Fig. 8(b): Variation of TC with f.


ACCEPTED MANUSCRIPT

Table 1. Production-inventory models with carbon policies in infinite planning horizon.


Demand Decision variable Carbon policy

Carbon cap combined


Number of shipments

with penalty and cost


Carbon cost / Tax

Strict carbon cap


Order quantity

Reorder point
Deterministic

Stochastic
Author(s)
Others

T
IP
Wahab et al. [23] √ √ √ √

CR
Bouchery et al. [21] √ √ √ √
Jaber et al. [24] √ √ Production rate √ √
Zanoni et al. [36]

Zanoni et al. [37]







√ US Production rate

Selling price, Investment to


improve production
√ √
AN
process

Bazan et al. [38] √ √ √ √ √


Gurtu et al. [39] √ √ √ √
M

This paper √ √ √ √ √
ED

Table 2. Summary of the solution procedure for the base case.


PT

m Q0 Q1 Q2 Q r TC TE

1 608.10 37.31 620.32 608.10 15.51 129425.58 1544.03


CE

2 408.54 18.47 408.50 408.50 16.02 128882.46 1550.00


3 321.29 12.30 303.71 303.71 16.38 128727.53 1550.00
4 270.37 9.24 241.56 241.56 16.64 128692.23 1550.00
AC

5 236.41 7.41 200.47 200.47 16.85 128710.58 1550.00


ACCEPTED MANUSCRIPT

Table 3(a). Variation of decision variables and costs with αb.


αb (Q, r, m) TC TCb TCv
0.80 (241.56, 16.64, 4) 128692.23 1131.36 127560.87
0.90 (232.24, 16.68, 4) 128706.06 1128.19 127577.87
1.00 (223.56, 16.73, 4) 128723.41 1126.74 127596.67
1.10 (215.44, 16.77, 4) 128743.99 1126.85 127617.14
1.20 (207.83, 16.81, 4) 128767.53 1128.40 127639.13

T
IP
CR
Table 3(b). Variation of decision variables and costs with αv.
αv (Q, r, m) TC TCb TCv
0.80
0.90
(241.56, 16.64, 4)
(221.13, 16.74, 4)
128692.23
128729.09
US 1131.36
1126.62
127560.87
127602.47
AN
1.00 (203.77, 16.83, 4) 128781.96 1129.85 127652.11
1.10 (243.06, 16.63, 3) 128828.85 1132.02 127696.83
1.20 (227.68, 16.71, 3) 128884.64 1127.24 127757.40
M
ED

Table 4. Variation of decision variables, production lot size, and costs with hb.
PT

hb (Q, r, m) mQ TC TCb TCv

5 (241.56, 16.64, 4) 966.23 128692.23 1131.36 127560.87


CE

6 (200.57, 16.64, 5) 1002.86 128815.89 1236.85 127579.04


7 (200.66, 16.47, 5) 1003.30 128921.13 1342.24 127578.89
AC

8 (200.74, 16.31, 5) 1003.69 129026.30 1447.54 127578.76


9 (171.59, 16.36, 6) 1029.51 129120.66 1524.50 127596.16
10 (171.64, 16.23, 6) 1029.83 129211.21 1615.14 127596.07
ACCEPTED MANUSCRIPT

Table 5. Variation of decision variables and costs with v.


v (Q, r, m) TC TCb TCv
50 (241.56, 16.64, 4) 128692.23 1131.36 127560.87
60 (242.19, 16.64, 4) 127683.20 1131.63 126551.57
70 (242.64, 16.63, 4) 126962.50 1131.83 125830.67
80 (242.98, 16.63, 4) 126422.00 1131.98 125290.02
90 (243.24, 16.63, 4) 126001.62 1132.10 124869.52

T
100 (243.45, 16.63, 4) 125665.33 1132.19 124533.14

IP
CR
Table 6. Variation of decision variables, total cost, and emissions with Ĉ.

1400
(Q, r, m)

(151.97, 17.14, 3)
TC

129513.88
US TE

1400.00
Emission at the
buyer

65.30
Emission at the
vendor and from
transportation
1334.70
AN
1450 (204.85, 16.82, 3) 129000.73 1450.00 86.20 1363.80
1500 (202.92, 16.83, 4) 128785.17 1500.00 85.44 1414.56
M

1550 (241.56, 16.64, 4) 128692.23 1550.00 100.74 1449.26

1600 (232.04, 16.68, 5) 128674.91 1600.00 96.97 1503.03


ED

1650 (236.04, 16.66,5) 128674.44 1606.36 98.55 1507.81


1700 (236.04, 16.66,5) 128674.44 1606.36 98.55 1507.81
PT

1750 (236.04, 16.66,5) 128674.44 1606.36 98.55 1507.81


CE
AC
ACCEPTED MANUSCRIPT

Table 7. Variation of decision variables, production lot size, and costs with P.
P (Q, r, m) mQ TC TCb TCv
1500 (216.95, 16.76, 5) 1084.74 128619.18 1126.72 127492.46
1600 (212.59, 16.78, 5) 1062.95 128641.61 1127.26 127514.35
1700 (249.65, 16.60, 4) 998.612 128657.38 1135.32 127522.06
1800 (246.59, 16.62, 4) 986.378 128670.18 1133.70 127536.48

T
1900 (243.92, 16.63, 4) 975.674 128681.74 1132.40 127549.34

IP
2000 (241.56, 16.64, 4) 966.228 128692.23 1131.36 127560.87

CR
128701.79
2100 (239.46, 16.65, 4) 957.832 1130.51 127571.28

2200 (237.58, 16.66, 4) 950.32 128710.52 1129.82 127580.70

US
Table 8. Variation of decision variables, costs, and emissions with f.
AN
Emission at Emission at the vendor and
f (Q, r, m) TC TCb TCv
the buyer from transportation
10 (246.22, 16.62, 4) 128686.99 1133.52 127553.47 102.59 1447.41
M

15 (243.91, 16.63, 4) 128689.45 1132.40 127557.05 101.67 1448.33

20 (241.56, 16.64, 4) 128692.23 1131.36 127560.87 100.75 1449.25

25 (239.15, 16.65, 4) 128695.36


ED

1130.39 127564.97 99.84 1450.16

30 (236.69, 16.66, 4) 128698.87 1129.51 127569.36 98.92 1451.08

35 (234.18, 16.67, 4) 128702.79 1128.72 127574.07 98.01 1451.99


PT

40 (231.61, 16.69, 4) 128707.17 1128.04 127579.13 97.09 1452.91

45 (228.97, 16.70, 4) 128712.06 1127.47 127584.59 96.18 1453.82


CE

50 (226.27, 16.71, 4) 128717.51 1127.03 127590.48 95.26 1454.74


AC

You might also like