ARTICLE

pubs.acs.org/est

Decision Support for Green Supply Chain Operations by Integrating

Dynamic Simulation and LCA Indicators: Diaper Case Study
Arief Adhitya,† Iskandar Halim,† and Rajagopalan Srinivasan†,‡,*
†
Institute of Chemical and Engineering Sciences, A*STAR (Agency for Science, Technology and Research), 1 Pesek Road, Jurong Island,
Singapore 627833
‡
Department of Chemical and Biomolecular Engineering, National University of Singapore, 4 Engineering Drive 4, Singapore 117576
bS Supporting Information
ABSTRACT: As the issue of environmental sustainability is becoming an important
business factor, companies are now looking for decision support tools to assess the
fuller picture of the environmental impacts associated with their manufacturing
operations and supply chain (SC) activities. Lifecycle assessment (LCA) is widely
used to measure the environmental consequences assignable to a product. However,
it is usually limited to a high-level snapshot of the environmental implications over the
product value chain without consideration of the dynamics arising from the multi-
tiered structure and the interactions along the SC. This paper proposes a framework
for green supply chain management by integrating a SC dynamic simulation and LCA
indicators to evaluate both the economic and environmental impacts of various SC
decisions such as inventories, distribution network configuration, and ordering policy.
The advantages of this framework are demonstrated through an industrially moti-
vated case study involving diaper production. Three distinct scenarios are evaluated to
highlight how the proposed approach enables integrated decision support for green SC design and operation.

1. INTRODUCTION in recent surveys conducted by Deloitte4 and McKinsey.5 In

Coined by the Brundtland Commission Report,1 sustainable
development refers to development that meets the needs of the
present without compromising the ability of future generations
to meet their own needs. Sustainable development is then about a
balancing act between human activities and the carrying capacity
of the natural ecosystem. A sustainable enterprise can therefore
be defined as one that strives to meet its economic goals while
taking into account the environmental and social responsibilities,
the so-called triple bottom line of sustainability.2 The importance
of sustainability to businesses has been highlighted in a global
survey conducted by KPMG in 2008 which indicates that 80% of
the world’s largest companies now publish their environmental
and social initiatives in their annual reports.3 Among the
initiatives disclosed were issues involving corporate governance,
climate change, and supply chain.
A supply chain (SC) can be described as a network of suppliers,
manufacturers, warehouses, and distribution channels organized
to acquire raw materials, convert them to finished products and
distribute these products to customers. A broad range of support-
ing services, such as sourcing, contracting, planning, schedul-
ing, monitoring, and financing needs to be managed to ensure
that the SC performs smoothly and optimally, these are termed
supply chain management. With sustainability increasingly becoming
a dominant issue, companies are now re-examining their manufac-
turing processes and SCs not just in terms of their economic
viability but also their environmental impacts. The important
role of sustainability in supply chain management is highlighted
principle, a strategy to drive SC sustainability would require an
understanding of the environmental impact of the product
throughout its lifecycle, ranging from the upstream suppliers to
the disposition of obsolete products.6 Such integration of the
product lifecycle and supply chain management is termed as
green supply chain management with its scope ranging from green
purchasing to product design, material sourcing and selection,
benign manufacturing, packaging, delivery of the final product to
the customers, as well as end-of-life management of the product.7
In literature, the growing interest in green supply chain
management is reflected by the increasing number of publications
in that field. In the area of SC planning and design, one common
approach is to formulate a mathematical optimization for max-
imum economic benefits and minimum environmental impacts.
In this case, the target of optimization can be selection of
appropriate raw materials, suppliers, technologies, or transporta-
tion routes.8
11 The economic performance can be measured
through indicators such as profit or customer satisfaction. For
environmental indicators, metrics such as waste reduction
(WAR) algorithm12 and lifecycle assessment (LCA) based
metrics such as CML 200113 have been applied. While much
literature exists on the design of green SC, less attention has
Received: June 6, 2011
Accepted: October 26, 2011
Revised: October 24, 2011
Published: October 26, 2011

r 2011 American Chemical Society 10178 dx.doi.org/10.1021/es201763q | Environ. Sci. Technol. 2011, 45, 10178–10185
Environmental Science & Technology
been given to operational aspects. Some relevant work in this area
includes a simulation approach to evaluate a SC operation by con-
sidering transport pollution, costs, time-to-market, and energy
usage.14 Another approach to green SC, called reverse logistics,
relies on collection of obsolete products or their components
(after use) for recycle.15 This paper describes an approach for
holistically evaluating the SC design and operation by using a
dynamic simulation and LCA indicators. The benefits of the
proposed approach are demonstrated using the industrially
relevant case of diaper production SC.
2. DIAPER SUPPLY CHAIN
Disposable diapers are a huge business in developed countries.
In the U.S, it is estimated that up to $2000 are spent on
disposable diapers per baby.16 Diaper manufacturing requires a
highly automated process involving significant capital invest-
ment. Recently, high energy prices have put pressure on operat-
ing costs. This, coupled with the focus on sustainability, has led
big players such as Kimberly-Clark to reorganize their SC to be
more efficient.17
The diaper SC involves suppliers, manufacturer, distributor,
and retailers/customers. The manufacturer and distributor be-
long to a focal enterprise and have different departments
performing various SC functions: receiving order, scheduling,
procurement, storage, operations, and delivery. The manufac-
turer procures raw materials from different suppliers, manufac-
tures the diaper products, and sends them in a bulk packaged
form to the distributor. The distributor operates in a push-mode;
it keeps a certain level of inventory to fulfill customer (retailer)
orders. Different procurement policies for replenishment of
products can be adopted by the distributor. For example, under
the fixed interval policy, procurement is done at regular intervals
to bring its inventory back to a certain top-up level. The
distributor places orders to the manufacturer, which operates
in a pull-mode, that is, the products are manufactured upon
receiving the distributor’s orders. The manufacturer keeps in-
ventory of raw materials and has its own procurement policy for
ordering raw materials from suppliers.
While modern disposable diapers come in a variety of styles,
their basic raw materials are essentially the same consisting of fluff
paper, superabsorbent polymer (SAP), plastic components such
as low density polyethylene (LDPE), polypropylene (PP), as
well as adhesives and elastics.18 Several LCA studies of the
environmental impacts of disposable diapers have been under-
taken previously.18,19 Starting from raw material extraction to
production process, point of use, and final disposal, these LCAs
account for the consumption of natural resources and release of
pollutants into air, water and soil. The overall conclusion from
these studies is that with the introduction of SAP in the 1980s
there has been a significant reduction in the use of raw materials
and energy. As shown in Table 1, the average diaper weight has
been reduced by almost 40% in a period of 18 years from 1987 to
2005.18 With this, the disposable diaper shows no significant
difference from home and commercially laundered reusable cloth
diaper in terms of environmental impacts.19
While the LCA technique can be effectively applied to
measure the environmental consequences associated with bring-
ing a product to market, it offers limited assistance when the SC
of any existing product has to be improved and issues related to
policies along the manufacturer
distributor
customer chain
come to the fore. LCA impacts have been derived from a
10179
ARTICLE
Table 1. Diaper Compositions Adapted from EDANA18
components 1987 (g) 2005 (g)
fluff pulp 54.94 14.57
superabsorber (SAP) 0.74 13.65
polypropylene (PP) 4.36 7.22
polyethylene (PE) 4.29 2.69
adhesive 1.34 1.76
elastic 0.20 0.21
others 1.14 1.85
total 67.01 41.95
product-centric perspective without considering the effects of
different logistics options, inventories, distribution network
configuration, and ordering policy, although these can be sig-
nificant contributors to the overall environmental impacts. For
example, LCA calculations are typically based on a static, average
truckload level and number of trips. In reality, these two variables
are dynamic and would vary during the actual SC operation based
on various exogenous factors (e.g., demand) and SC policies. As
anecdotal evidence of the significance of these, ExxonMobil’s
2010 Corporate Citizenship Report20 mentions that they “seek
to fill truckloads and optimize packaging to reduce the number of
trips while providing associated fuel and cost saving.” LCA alone,
therefore, does not provide useful insights to a SC manager
intending to green a company or product’s SC. The sustainable
SC approach, on the other hand, takes a supply chain-centric
perspective by considering the scope of influence and the various
decisions that can be made by the SC manager. In this paper, we
propose a decision support approach for such sustainability
assessment of SC operations. This is done by evaluating the
LCA environmental indicators in each stage of the SC model
using a dynamic simulator. The scope of the SC of interest, as
shown in Figure 1, includes all the stages in the diaper manu-
facture starting from the procurement of raw materials until
delivery of the product to customers (retailers).
3. DYNAMIC MODELING OF THE DIAPER SUPPLY
CHAIN
Dynamics of SC operation are complex due to the multitiered
structure and numerous interactions among the entities. The
impact of an entity’s action on the overall SC performance, both
economic and environmental, may not be immediately obvious.
An integrated analysis of the impacts of decisions on the overall
system is thus required. This motivates the use of SC simulation
models, which capture the behavior of the entities and their
interactions, and indicate their direct and indirect effects on the
overall SC performance. In this paper, we integrate LCA indica-
tors into the dynamic simulation modeling scheme of Adhitya and
Srinivasan.21
The SC model uses a discrete-time representation, where one
day is divided into a predefined number of time ticks t. The
location of various SC entities such as the plant, distributor, or
customer is represented through a pair of coordinates (x, y). Raw
material arrival at the plant is modeled as shown:
RA r ðtÞ ¼ RPr ðt
LTr Þ ð1Þ
where RAr(t) is the amount of raw material r arriving at the plant
at time t, RPr(t) is the amount of raw material r ordered from the
dx.doi.org/10.1021/es201763q |Environ. Sci. Technol. 2011, 45, 10178–10185
Environmental Science & Technology
Figure 1. Scope of diaper lifecycle and supply chain of interest.
supplier, and LTr is the lead time between order and delivery of
the raw material r.
Processing in the plant is modeled through the following:
RMreqr ¼ RCPr 3 CJamt
RUr ðtÞ ¼ RMreqr
where RMreqr is the amount of raw material r required for
processing a job CJ, which depends on RCPr, the recipe specify-
ing the amount of raw material r required to make one unit of
product, and CJamt, the amount of product to be made in job CJ.
RUr(t) is the amount of raw material r needed for production at
time t.
The material balance on raw material inventory at the plant is
given by
IR r ðt þ 1Þ ¼ IR r ðtÞ þ RA r ðtÞ
RUr ðtÞ
where IRr(t) is the inventory of raw material r at time t.
The complete model equations as summarized in Table 2
consider all the key aspects of raw material procurement,
logistics, manufacturing, and product distribution for the man-
ufacturing plant and the distributor. Using this model, different
SC policies, configurations, and also uncertainties can be simu-
lated and their impacts on overall performance, both economic
and environmental, quantitatively evaluated. Economic perfor-
mance is measured through indicators such as manufacturer
profit and distributor profit (Table 3). Environmental perfor-
mance is measured through 11 indicators, namely abiotic re-
source depletion (ARD), global warming (GWP), ozone layer
depletion (ODP), photochemical oxidation (PO), acidification
(ACD), eutrophication (EUT), human toxicity (HT), fresh
water aquatic ecotoxicity (FWAE), terrestrial ecotoxicity (TE),
water usage (WU), and energy consumption (EC), evaluated at
each stage of the SC (Table 4). These indicators are adapted
from a report by the UK Environmental Agency;19 more details
are provided in the Supporting Information. The model has been
implemented in MATLAB/Simulink,22 where flows of material,
information, and finance are depicted by various mathematical,
logical, and algorithmic operation blocks. Such a model can be
ARTICLE
used for making decision to drive a SC operation to be greener as
illustrated by the following case studies.
ð2Þ 4. RESULTS AND DISCUSSION
Our base-case scenario is a SC network with a manufacturing
ð3Þ plant producing disposable diapers of year 2005 composi-
tion (see Table 1) and a single distributor, whose ordering policy
is 1 day procurement interval. The breakdown of the overall envi-
ronmental impact to each stage in the SC is shown in Table 5. It
can be seen that SC stages other than raw material and manu-
facturer (the two stages that are typically the focus of LCA
studies) have a significant contribution in some impacts. For
example, the distributor contributes 62% of ozone layer deple-
tion, and the three transportation stages further add up to 20%
of ozone layer depletion. The following three scenarios con-
ð4Þ sider effects of product composition, SC network configura-
tion, and SC operation policy on economic and environmental
impacts.
4.1. Scenario 1: Changing Diaper Composition. The first
scenario analyzes the impact of changes to the diaper composi-
tions. Figure 2 compares the overall environmental impacts per
carton diaper from the SCs of the two diaper compositions listed
in Table 1. It shows a significant reduction in the water usage for
the 2005 diaper as compared to the 1987 one. This is mainly due
to the reduction in the quantity of fluff pulp used (14.57 g vs
54.94 g). However, the figure also highlights an increase in the
other seven environmental indicators (global warming, photo-
chemical oxidation, acidification, human toxicity, terrestrial eco-
toxicity, energy consumption, and abiotic resource depletion) for
the 2005 diaper. This is mainly caused by the increased amount
of SAP in the 2005 diaper (13.65 g vs 0.74 g). The increased
amount of polypropylene in the 2005 diaper (7.224 g vs 4.355 g)
also contributes quite significantly to the increase in the environ-
mental impacts for these categories.
The new composition also leads to an eight times increase in
manufacturer profit since it requires less raw material (41.95 g vs
67 g). Although the old composition uses less SAP, it uses much
more fluff pulp. The results also reveal that the environmental
impacts from raw material transportation are 37% less in the
10180 dx.doi.org/10.1021/es201763q |Environ. Sci. Technol. 2011, 45, 10178–10185
Environmental Science & Technology ARTICLE

Table 2. Model Equations for Distributor and Manufacturing Plant

no. equation explanation

1 DOrdList r DSPðCO, DOrdListÞ the distributor receives a customer order CO and inserts it into its
order list DOrdList following its scheduling policy DSP, e.g. first-
come-first-serve, priority-based, etc.
2 DCO ¼ DOrdList1 order to be processed by the distributor, DCO, is the first customer
order in the order list DOrdList.
3 DPDðtÞ ¼ DCOamt if DIPðtÞ g DCOamt product of amount DCOamt will be delivered from the distributor
to the customer as DPD(t) if there is sufficient distributor inventory
at time t. DIP(t) is the product inventory at the distributor at time t.
4 DCOst ¼ t DCOst is the time at which the order is sent out from the
distributor.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
5 2 DCOtt is the transportation time for order DCO from the
DCO ¼ DTS ðDx
DCOxloc Þ2 þ ðDy
DCOyloc Þ
tt
distributor to the customer. DTS is the transportation speed.
The distributor is located at coordinates (Dx, Dy) and the
customer is at (DCOxloc, DCOxloc). Straight line distance is
assumed.
6 DCOdt ¼ DCOst þ DCOtt Product reaches the customer at time DCOdt.
(
7
DRPðtÞ ¼
DRT
DIPðtÞ
∑ DRWðtÞ, if DRPðtÞ > 0, t ¼ c 3 DPC, c ¼ 1, 2, 3 the distributor employs the fixed interval procurement policy. It
0 , otherwise places orders with the manufacturer, DO, of amount DRP(t) at a
regular interval DPC to bring the inventory up to a certain top-up
8 DOtime ¼ t
level DRT. ∑DRW(t) is the amount of product which has been
9 DOamt ¼ DRPðtÞ ordered but is yet to arrive at the distributor, and c is the
procurement cycle index. DOtime is the time at which order DO
is placed with the manufacturer. DOamt is the amount ordered in
DO.
10 DPAðtÞ ¼ CJamt at t ¼ CJdt the distributor order is processed by the manufacturer as job CJ.
CJdt is the time at which order CJ arrives at the distributor. DPA(t)
is the quantity that arrives at the distributor at time t. CJamt is the
amount in job CJ.
11 DIPðt þ 1Þ ¼ DIPðtÞ þ DPAðtÞ
DPDðtÞ material balance on product inventory at the distributor
12 JobSch r SPðDO, JobSchÞ the manufacturer receives an order DO from the distributor and
inserts it as a job into its job schedule JobSch following its
scheduling policy SP.
13 CJ ¼ JobSch1 if IR r ðtÞ g RMreqr and PSðtÞ ¼ 0 the first job in JobSch will be assigned as CJ, the job to be processed,
if there is sufficient raw material inventory and the plant is not
currently processing any other job. IRr(t) is the inventory of raw
material r at time t. RMreqr is the amount of raw material r required
for processing CJ. The processing status PS(t) indicates the amount
of time remaining to complete the current job. The job is
completed when PS(t) reaches 0.
14 RMreqr ¼ RCPr 3 CJamt RCPr is the recipe specifying the amount of raw material r required
to make one unit of product. CJamt is the amount of product to be
made in job CJ.
15 CJst ¼ t CJst is the processing start time of job CJ.

16 PSðt þ 1Þ ¼ CJ pt
the plant processing status is updated after a job is started. CJpt is
the total processing time for job CJ.
(
17 PSðtÞ
1 if PSðtÞ > 0 as time progresses, the processing status is updated to keep track of
PSðt þ 1Þ ¼
0 otherwise the amount of time remaining to complete the current job.

18 RUr ðtÞ ¼ RMreqr RUr(t) is the amount of raw material r used for production at time t.

19 CJ ¼ PR 3 CJ
pt amt
CJpt is the processing time for job CJ, based on the processing time
per unit product PR.
20 CJpkgt ¼ PkgR 3 CJamt CJpkgt is the packaging time for job CJ, based on the packaging time
per unit product PkgR.

10181 dx.doi.org/10.1021/es201763q |Environ. Sci. Technol. 2011, 45, 10178–10185

Environmental Science & Technology ARTICLE

Table 2. Continued
no. equation explanation
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi tt
21 2 CJ is the transportation time for job CJ from the plant to the
CJtt ¼ TS ðPx
DxÞ2 þ ðPy
DyÞ
distributor. TS is the transportation speed. The plant is located at
coordinates (Px, Py) and the distributor is at (Dx, Dy). Straight line
distance is assumed.
22 CJdt ¼ CJst þ CJpt þ CJpkgt þ CJtt product reaches the distributor at time CJdt.
(
23
RPr ðtÞ ¼ ∑
RTr
IR r ðtÞ
RW r ðtÞ, if IR r ðtÞ þ ∑ RWr ðtÞ < RR r the plant employs the reorder point procurement policy. Raw
0 , otherwise material will be purchased when its inventory falls below the
reorder point RRr. RPr(t) is the amount of raw material r
ordered by the plant to its supplier, RTr is the inventory top-up
level for raw material r, and ∑RWr(t) is the amount of raw material r
which has been ordered but is yet to arrive at the plant.
24 RA r ðtÞ ¼ RPr ðt
LTr Þ RAr(t) is the amount of raw material r arriving at the plant at time t,
LTr is the lead time between purchase and arrival for raw material r.
25 IR r ðt þ 1Þ ¼ IR r ðtÞ þ RA r ðtÞ
RUr ðtÞ material balance on raw material inventory at the plant

Table 3. Calculation of Economic Indicators

no. equation explanation

1 DCO rev
¼ DPrice 3 DCO amt
DCO rev
is the distributor’s revenue from order DCO. DPrice is the
price per unit product charged by the distributor.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2 DCOtc is the transportation cost for order DCO. DCostT is the
DCOtc ¼ DcostT 3 DCOamt 3 ðDx
DCOxloc Þ2 þ ðDy
DCOyloc Þ
transportation cost per unit distance per unit product from the
distributor to the customer.
0 1
3
B
∑ DCOtc
C
DProfit is the profit of the distributor, Price is the price per unit product
B
B
þ ∑
Price 3 DPAðtÞ C
C
charged by the plant and paid by the distributor, DCostI is the inventory
DProf it ¼ ∑ DCO
B
rev
B
B þ
t
∑
C
DcostI 3 DIPðtÞ C
C
cost per unit product per time tick, and DCostOF is the fixed operating
@ t A cost (charged at each time tick), D is the number of days in the
þ ðDcostOF 3 D 3 TÞ simulation horizon, and T is the number of time ticks in one day.

4 CJ rev
¼ Price 3 CJ amt
CJrev is the plant’s revenue from job CJ. Price is the price per unit
product charged by the plant.
5 CJpc ¼ CostOV 3 CJpt CJpc is the variable processing cost for job CJ. CostOV is the processing
cost charged when the plant is processing a job.
6 CJpkgc ¼ CostPkg 3 CJamt CJpkgc is the packaging cost for job CJ. CostPkg is the packaging cost
per unit product.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2
7 CJtc ¼ CostT 3 CJamt 3 ðPx
DxÞ þ ðPy
DyÞ2 CJtc is the transportation cost for job CJ. CostT is the transportation
cost per unit distance per unit product from the plant to the distributor.
0 1
8
B
∑ ðCJpc þ CJpkgc þ CJtc þ CJpen Þ
C
profit is the profit of the manufacturing plant, CostRr is the price of raw
B
B
þ ∑∑ CostR r 3 RA r ðtÞ C
C
material r, CostI is the inventory cost per unit raw material per time tick,
Prof it ¼ ∑ CJ
B
rev
B
B þ
t
∑∑
r
CostI 3 IR r ðtÞ
C
C
C
and CostOF is the fixed operating cost (charged at each time tick
@ t r A regardless plant is processing or idle).
þ ðCostOF 3 D 3 TÞ

2005 case, because less raw materials need to be transported backup to the other, leading to higher customer satisfaction level.
resulting in fewer trips. Table 6 shows the changes in the environmental impacts per
4.2. Scenario 2: Changing Distribution Network Config- carton of diaper in the 2-distributor scenario as compared to the
uration. The second scenario evaluates the effects of the base case. Overall, there is an 85% increase in all environmental
distribution network by comparing a single distributor (the base impacts other than water use for the plant-to-distributor trans-
case) versus two distributors for a given geographical market. portation. The reason is that the second distributor is located
While the single distributor channel could be more cost-efficient farther from the manufacturing plant than the existing one.
and easier to manage, two distribution channels would have the However, this is partially offset by lower distributor-to-customer
benefit of being at closer proximity to customers. Another transportation impacts (other than water use), which decrease by
advantage of the two distributor channels is that robustness 24%, since customers are served by the nearer distributor
increases since in the event of disruptions one can serve as a resulting in less travel distance. The table also shows comparable
10182 dx.doi.org/10.1021/es201763q |Environ. Sci. Technol. 2011, 45, 10178–10185
Environmental Science & Technology ARTICLE

Table 4. Calculation of Environmental Indicators

no. equation explanation

1 TEIRMi ¼
2 TEITSPi ¼
3 TEIMi ¼
4 TEIPkgi ¼
5 TEITPDi ¼
6 TEIDi ¼
7 TEITDCi ¼
8 TEIi ¼ TEIRMi þ TEIPkgi þ TEIMi þ TEIDi
þ TEITSPi þ TEITPDi þ TEITDCi
∑t ∑r EIRMi, r 3 RAr ðtÞ TEIRMi is the total environmental impact i from raw materials, where i = ARD, GWP,
ODP, PO, ACD, EUT, HT, FWAE, TU, WC, or EC. EIRMi,r is the environmental
impact i per unit raw material r.
∑t ∑r EITRMi 3 RAr ðtÞ 3 SPDistðtÞ TEITSPi is the total environmental impact i from raw material transportation from
suppliers to the plant. EITRMi is the environmental impact i from raw material
transportation per unit raw material per unit distance. SPDist(t) is the distance
covered for raw material transportation from suppliers to the plant.
∑t EIMi 3 DPAðtÞ TEIMi is the total environmental impact i from manufacturing. EIMi is the
environmental impact i from manufacturing per unit product.
∑t EIPkgi 3 DPAðtÞ TEIPkgi is the total environmental impact i from packaging. EIPkgi is the environmental
impact i from packaging per unit product.
∑t EITPi 3 PDDistðtÞ TEITPDi is the total environmental impact i from product transportation from the plant
to the distributor. EITPi is the environmental impact i from product transportation per
unit distance. PDDist(t) is the distance covered for product transportation from the
plant to the distributor.
∑t EIDi 3 DPAðtÞ TEIDi is the total environmental impact i from the distributor operation. EIDi is the
environmental impact i from distributor operation per unit product.
∑t EITPi 3 DCDistðtÞ TEITDCi is the total environmental impact i from product transportation from the
distributor to customers. DCDist(t) is the distance covered for product transportation
from the distributor to customers.
TEIi is the total environmental impact i from the whole supply chain.

Table 5. Relative Contribution of Environmental Impact in the Base Case

stage ARD GWP ODP PO ACD EUT HT FWAE TE WU EC

raw materials 83% 80% 96% 87% 84% 45% 12% 85% 99% 41%
supplier to plant 4% 1% 1% 1% 1%
manufacturer 7% 12% 15%
2% 4% 5% 33% 57% 10% 29%
packaging 6% 2% 3% 3% 1% 1% 1% 1% 1%
plant to distributor 1% 8% 1% 2% 1% 1% 1%
distributor 3% 5% 62% 5% 4% 5% 18% 27% 5% 26%
distributor to customer 1% 8% 1% 2% 1% 1% 1%

environmental impacts in the other SC stages. Overall, there is a

statistically significant 5% increase in ozone layer depletion and a
statistically insignificant (1%) increase in four of the indicators
(eutrophication, human toxicity, fresh water aquatic ecotoxicity,
and energy consumption) in the 2-distributor case. The profit of
the 2-distributor scenario is 10% lower than the single distributor
case due to higher operating costs. Such increases in the
environmental impacts and costs demonstrate the trade-off
arising from higher resilience and customer satisfaction provided
by the two distributors. This scenario thus highlights the
significant impact of network configuration on the environmen-
tal and economic performances.
4.3. Scenario 3: Changing Distributor’s Ordering Policy. In
the third scenario, the effect of different ordering policies by the Figure 2. Total environmental impacts along the supply chain per
distributor is analyzed. Less frequent ordering in larger batches carton diaper for Scenario 1.
would mean fewer transportation trips and consequently reduc-
tion in transportation impact and cost. In the base-case scenario, involving 2-day procurement interval. Table 7 shows a statisti-
the distributor places an order to the manufacturing plant daily cally significant (6%) reduction in all environmental impacts
and products are delivered daily from the plant. We simulate and other than water use for the plant-to-distributor transportation
compare this base-case scenario with a different ordering policy due to fewer transportation trips (although of higher loads). It
10183 dx.doi.org/10.1021/es201763q |Environ. Sci. Technol. 2011, 45, 10178–10185
Environmental Science & Technology ARTICLE

Table 6. Comparison of Environmental Impact Per Carton Diaper in 2-Distributor and 1-Distributor Scenariosa
stage ARD GWP ODP PO ACD EUT HT FWAE TE WU EC

raw materials ∼ ∼ = ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼
supplier to plant ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ = ∼
manufacturer = = = = = = = = = = =
packaging = = = = = = = = = = =
plant to distributor +85% +85% +85% +85% +85% +85% +85% +85% +85% = +85%
distributor ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼
distributor to customer
24%
24%
24%
24%
24%
24%
24%
24%
24% =
24%
total ∼ ∼ +5% ∼ ∼ +1% +1% +1% ∼ ∼ +1%
a
Legend: = Equal value; ∼ Less than 0.5% difference.

Table 7. Comparison of Total Environmental Impact in 2-Day and 1-Day Procurement Interval Scenariosa
stage ARD GWP ODP PO ACD EUT HT FWAE TE WU EC

raw materials ∼ ∼ = ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼
supplier to plant ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ = ∼
manufacturer
1%
1%
1%
1%
1%
1%
1%
1%
1%
1%
1%
packaging
1%
1%
1% =
1%
1%
1%
1%
1%
1%
1%
plant to distributor
6%
6%
6%
6%
6%
6%
6%
6%
6% =
6%
distributor = = = = = = = = = = =
distributor to customer ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ = ∼
total ∼ ∼
1% ∼ ∼
1% ∼ ∼ ∼ ∼ ∼
a
Legend: = Equal value; ∼ Less than 0.5% difference.

also shows a statistically insignificant (1%) reduction in envir- ’ REFERENCES

onmental impact during the packaging and manufacturing stage
due to slightly lower diaper production in the 2-day scenario.
The amount of diapers ordered by the distributor and hence the
amount of diapers to be produced by the manufacturing plant
are functions of the inventory top-up point, whose primary role
is to ensure a desired customer satisfaction. Different top-up
points would therefore result in different economic and envir-
onmental performance. Further analysis of this issue would be
the subject of our future study. Overall, the three scenarios serve
to highlight the advantage of the integrated SC simulation and
LCA indicator approach for studying the environmental impacts
arising from the dynamics of the SC network and operating
policies.
’ ASSOCIATED CONTENT
bS Supporting Information. Derivation of environmental
impact indicators and case study parameters. This material is
available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
*Phone: +65 65168041; fax: +65 67791936; e-mail: chergs@nus.
edu.sg.
’ ACKNOWLEDGMENT
We thank Ms. Patricia Petrus of the National University of
Singapore, Department of Chemical and Biomolecular Engineer-
ing for her contribution to the modeling work.
(1) Brundtland, G. H. (Ed.). Our Common Future: The World
Commission on Environment and Development; Oxford University Press:
Oxford, 1987.
(2) Elkington, J. Cannibals With Forks: The Triple Bottom Line of 21st
Century Business; Capstone Publishing: Oxford, 1997.
(3) KPMG. KPMG international survey on corporate responsibility
reporting 2008. http://www.kpmg.com (accessed October 28, 2010).
(4) Deloitte. Sustainability in business today: A cross-industry view.
http://www.deloitte.com (accessed October 28, 2010).
(5) McKinsey. How companies manage sustainability. http://www.
mckinseyquarterly.com (accessed October 28, 2010).
(6) Fiksel, J. Evaluating supply chain sustainability. Chem. Eng. Prog.
2010, May, 28–36.
(7) Srivasta, S. K. Green supply-chain management: A state-of-the-
art literature review. Int. J. Manage. Rev. 2007, 9, 53–80.
(8) Hugo, A.; Pistikopoulos, E. Environmentally conscious long-
range planning and design of supply chain networks. J. Clean. Prod. 2005,
13, 1471–1491.
(9) Bojarski, A. D.; Laínez, J. M.; Espu~ na, A.; Puigjaner, L. Incorpor-
ating environmental impacts and regulations in a holistic supply chains
modeling: An LCA approach. Comp. Chem. Eng. 2009, 33, 1747–1759.
(10) Guillen-Gosalbez, G.; Grossmann, I. A global optimization for
environmentally conscious design of supply chains under uncertainty in
the damage assessment model. Comput. Chem. Eng. 2010, 34, 42–58.
(11) Mele, F. D.; Kostin, A. M.; Guillen-Gosalbez, G.; Jimenez, L.
Multiobjective model for more sustainable fuel supply chains. A case
study of the sugar cane industry in Argentina. Ind. Eng. Chem. Res. 2011,
50 (9), 4939–4958.
(12) Cabezas, H.; Bare, J. C.; Mallick, S. K. Pollution prevention with
chemical process simulators: The generalized waste reduction (WAR)
algorithm—Full version. Comp. Chem. Eng. 1999, 23, 623–634.
(13) Handbook on Lifecycle Assessment. Operational Guide to the ISO
Standards; Guinee, J. B., Ed.; Kluwer Academic Publishers: Dordrecht,
2002.

10184 dx.doi.org/10.1021/es201763q |Environ. Sci. Technol. 2011, 45, 10178–10185

Environmental Science & Technology ARTICLE

(14) Khoo, H. H.; Spedding, T. A.; Bainbridge, I.; Taplin, D. M. R.

Creating a green supply chain. Green. Manage. Int. 2001, 35, 71–88.
(15) Komoto, H.; Tomiyama, T.; Silvester, S.; Brezet, H. Analyzing
supply chain robustness for OEMs from a lifecycle perspective using lifecycle
simulation. Int. J. Prod. Econ. 2011, DOI: 10.1016/j.ijpe.2009.11.017.
(16) Consumer Report. Getting started—Diaper guide. http://
www.consumerreports.org (accessed October 28, 2010).
(17) Jamison, M. Kimberly-Clark’s supply chain network of the
future. http://www.industryweek.com (accessed October 28, 2010).
(18) EDANA (European Disposables and Nonwovens Association).
Sustainability Report: Baby Diapers and Incontinence Products. http://
www.edana.org (accessed October 28, 2010).
(19) UK Environmental Agency. Lifecycle Assessment of Disposa-
ble and Reusable Nappies in the UK. http://www.environment-agency.
gov.uk (accessed October 28, 2010).
(20) ExxonMobil. 2010 Corporate Citizenship Report. http://www.
exxonmobil.com (accessed October 21, 2011).
(21) Adhitya, A.; Srinivasan, R. Dynamic simulation and decision
support for multisite specialty chemicals supply Chain. Ind. Eng. Chem.
Res. 2010, 49, 9917–9931.
(22) MathWorks. Simulink User’s Guide, 2010.

10185 dx.doi.org/10.1021/es201763q |Environ. Sci. Technol. 2011, 45, 10178–10185