Decision Support for Green Supply Chain Operations by Integrating Dynamic Simulation and LCA Indicators: Diaper Case Study

Published: October 26, 2011

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

ARTICLE

pubs.acs.org/est

Decision Support for Green Supply Chain Operations by IntegratingDynamic Simulation and LCA Indicators: Diaper Case StudyArief 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

1. INTRODUCTION

Coined by the Brundtland Commission Report,1 sustainabledevelopment refers to development that meets the needs of thepresent without compromising the ability of future generationstomeet their own needs. Sustainable development is then about abalancing act between human activities and the carrying capacityof the natural ecosystem. A sustainable enterprise can thereforebe defined as one that strives to meet its economic goals whiletaking into account the environmental and social responsibilities,the so-called triple bottom line of sustainability.2 The importanceof sustainability to businesses has been highlighted in a globalsurvey conducted by KPMG in 2008 which indicates that 80% ofthe world’s largest companies now publish their environmentaland social initiatives in their annual reports.3 Among theinitiatives 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 organizedto acquire raw materials, convert them to finished products anddistribute 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 ensurethat the SC performs smoothly and optimally, these are termedsupply chainmanagement.With sustainability increasingly becominga dominant issue, companies are now re-examining their manufac-turing processes and SCs not just in terms of their economicviability but also their environmental impacts. The importantrole of sustainability in supply chain management is highlighted

in recent surveys conducted by Deloitte4 and McKinsey.5 Inprinciple, a strategy to drive SC sustainability would require anunderstanding of the environmental impact of the productthroughout its lifecycle, ranging from the upstream suppliers tothe disposition of obsolete products.6 Such integration of theproduct lifecycle and supply chain management is termed asgreen supply chain management with its scope ranging from greenpurchasing to product design, material sourcing and selection,benign manufacturing, packaging, delivery of the final product tothe customers, as well as end-of-life management of the product.7

In literature, the growing interest in green supply chainmanagement is reflected by the increasing number of publicationsin that field. In the area of SC planning and design, one commonapproach 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 ofappropriate raw materials, suppliers, technologies, or transporta-tion routes.8�11 The economic performance can be measuredthrough indicators such as profit or customer satisfaction. Forenvironmental indicators, metrics such as waste reduction(WAR) algorithm12 and lifecycle assessment (LCA) basedmetrics such as CML 200113 have been applied. While muchliterature exists on the design of green SC, less attention has

Received: June 6, 2011Accepted: October 26, 2011Revised: October 24, 2011

ABSTRACT: As the issue of environmental sustainability is becoming an importantbusiness factor, companies are now looking for decision support tools to assess thefuller picture of the environmental impacts associated with their manufacturingoperations and supply chain (SC) activities. Lifecycle assessment (LCA) is widelyused to measure the environmental consequences assignable to a product. However,it is usually limited to a high-level snapshot of the environmental implications over theproduct value chain without consideration of the dynamics arising from the multi-tiered structure and the interactions along the SC. This paper proposes a frameworkfor green supply chain management by integrating a SC dynamic simulation and LCAindicators to evaluate both the economic and environmental impacts of various SCdecisions 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 tohighlight how the proposed approach enables integrated decision support for green SC design and operation.

10179 dx.doi.org/10.1021/es201763q |Environ. Sci. Technol. 2011, 45, 10178–10185

Environmental Science & Technology ARTICLE

been given to operational aspects. Some relevant work in this areaincludes a simulation approach to evaluate a SC operation by con-sidering transport pollution, costs, time-to-market, and energyusage.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 forholistically evaluating the SC design and operation by using adynamic simulation and LCA indicators. The benefits of theproposed approach are demonstrated using the industriallyrelevant 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 ondisposable diapers per baby.16 Diaper manufacturing requires ahighly 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 ledbig players such as Kimberly-Clark to reorganize their SC to bemore 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 departmentsperforming 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 packagedform 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 ofproducts can be adopted by the distributor. For example, underthe fixed interval policy, procurement is done at regular intervalsto bring its inventory back to a certain top-up level. Thedistributor places orders to the manufacturer, which operatesin a pull-mode, that is, the products are manufactured uponreceiving the distributor’s orders. The manufacturer keeps in-ventory of raw materials and has its own procurement policy forordering raw materials from suppliers.

While modern disposable diapers come in a variety of styles,their basic rawmaterials are essentially the same consisting of fluffpaper, superabsorbent polymer (SAP), plastic components suchas low density polyethylene (LDPE), polypropylene (PP), aswell as adhesives and elastics.18 Several LCA studies of theenvironmental impacts of disposable diapers have been under-taken previously.18,19 Starting from raw material extraction toproduction process, point of use, and final disposal, these LCAsaccount for the consumption of natural resources and release ofpollutants into air, water and soil. The overall conclusion fromthese studies is that with the introduction of SAP in the 1980sthere has been a significant reduction in the use of raw materialsand energy. As shown in Table 1, the average diaper weight hasbeen reduced by almost 40% in a period of 18 years from 1987 to2005.18 With this, the disposable diaper shows no significantdifference from home and commercially laundered reusable clothdiaper in terms of environmental impacts.19

While the LCA technique can be effectively applied tomeasure the environmental consequences associated with bring-ing a product to market, it offers limited assistance when the SCof any existing product has to be improved and issues related topolicies along the manufacturer�distributor�customer chaincome to the fore. LCA impacts have been derived from a

product-centric perspective without considering the effects ofdifferent logistics options, inventories, distribution networkconfiguration, and ordering policy, although these can be sig-nificant contributors to the overall environmental impacts. Forexample, LCA calculations are typically based on a static, averagetruckload level and number of trips. In reality, these two variablesare dynamic andwould vary during the actual SC operation basedon various exogenous factors (e.g., demand) and SC policies. Asanecdotal evidence of the significance of these, ExxonMobil’s2010 Corporate Citizenship Report20 mentions that they “seekto fill truckloads and optimize packaging to reduce the number oftrips while providing associated fuel and cost saving.” LCA alone,therefore, does not provide useful insights to a SC managerintending to green a company or product’s SC. The sustainableSC approach, on the other hand, takes a supply chain-centricperspective by considering the scope of influence and the variousdecisions that can be made by the SC manager. In this paper, wepropose a decision support approach for such sustainabilityassessment of SC operations. This is done by evaluating theLCA environmental indicators in each stage of the SC modelusing a dynamic simulator. The scope of the SC of interest, asshown in Figure 1, includes all the stages in the diaper manu-facture starting from the procurement of raw materials untildelivery of the product to customers (retailers).

3. DYNAMIC MODELING OF THE DIAPER SUPPLYCHAIN

Dynamics of SC operation are complex due to the multitieredstructure and numerous interactions among the entities. Theimpact of an entity’s action on the overall SC performance, botheconomic and environmental, may not be immediately obvious.An integrated analysis of the impacts of decisions on the overallsystem is thus required. This motivates the use of SC simulationmodels, which capture the behavior of the entities and theirinteractions, and indicate their direct and indirect effects on theoverall SC performance. In this paper, we integrate LCA indica-tors into the dynamic simulationmodeling scheme of Adhitya andSrinivasan.21

The SC model uses a discrete-time representation, where oneday is divided into a predefined number of time ticks t. Thelocation of various SC entities such as the plant, distributor, orcustomer is represented through a pair of coordinates (x, y). Rawmaterial arrival at the plant is modeled as shown:

RArðtÞ ¼ RPrðt � LTrÞ ð1Þ

where RAr(t) is the amount of raw material r arriving at the plantat time t, RPr(t) is the amount of raw material r ordered from the

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



supplier, and LTr is the lead time between order and delivery ofthe raw material r.

Processing in the plant is modeled through the following:

RMreqr ¼ RCPr 3CJamt ð2Þ

RUrðtÞ ¼ RMreqr ð3Þwhere RMreqr is the amount of raw material r required forprocessing a job CJ, which depends on RCPr, the recipe specify-ing the amount of raw material r required to make one unit ofproduct, and CJamt, the amount of product to be made in job CJ.RUr(t) is the amount of raw material r needed for production attime t.

The material balance on raw material inventory at the plant isgiven by

IRrðt þ 1Þ ¼ IRrðtÞ þ RArðtÞ � RUrðtÞ ð4Þwhere IRr(t) is the inventory of raw material r at time t.

The complete model equations as summarized in Table 2consider all the key aspects of raw material procurement,logistics, manufacturing, and product distribution for the man-ufacturing plant and the distributor. Using this model, differentSC policies, configurations, and also uncertainties can be simu-lated and their impacts on overall performance, both economicand environmental, quantitatively evaluated. Economic perfor-mance is measured through indicators such as manufacturerprofit and distributor profit (Table 3). Environmental perfor-mance is measured through 11 indicators, namely abiotic re-source depletion (ARD), global warming (GWP), ozone layerdepletion (ODP), photochemical oxidation (PO), acidification(ACD), eutrophication (EUT), human toxicity (HT), freshwater aquatic ecotoxicity (FWAE), terrestrial ecotoxicity (TE),water usage (WU), and energy consumption (EC), evaluated ateach stage of the SC (Table 4). These indicators are adaptedfrom a report by the UK Environmental Agency;19 more detailsare provided in the Supporting Information. The model has beenimplemented 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

used for making decision to drive a SC operation to be greener asillustrated by the following case studies.

4. RESULTS AND DISCUSSION

Our base-case scenario is a SC network with a manufacturingplant producing disposable diapers of year 2005 composi-tion (see Table 1) and a single distributor, whose ordering policyis 1 day procurement interval. The breakdown of the overall envi-ronmental impact to each stage in the SC is shown in Table 5. Itcan be seen that SC stages other than raw material and manu-facturer (the two stages that are typically the focus of LCAstudies) have a significant contribution in some impacts. Forexample, 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-sider effects of product composition, SC network configura-tion, and SC operation policy on economic and environmentalimpacts.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 percarton diaper from the SCs of the two diaper compositions listedin Table 1. It shows a significant reduction in the water usage forthe 2005 diaper as compared to the 1987 one. This is mainly dueto the reduction in the quantity of fluff pulp used (14.57 g vs54.94 g). However, the figure also highlights an increase in theother seven environmental indicators (global warming, photo-chemical oxidation, acidification, human toxicity, terrestrial eco-toxicity, energy consumption, and abiotic resource depletion) forthe 2005 diaper. This is mainly caused by the increased amountof SAP in the 2005 diaper (13.65 g vs 0.74 g). The increasedamount 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 vs67 g). Although the old composition uses less SAP, it uses muchmore fluff pulp. The results also reveal that the environmentalimpacts from raw material transportation are 37% less in the

Figure 1. Scope of diaper lifecycle and supply chain of interest.



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 DCO¼DOrdList1 order to be processed by the distributor, DCO, is the first customer

order in the order list DOrdList.

3 DPDðtÞ¼DCOamt ifDIPðtÞ 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.

5DCOtt ¼DTS

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðDx�DCOxlocÞ2 þ ðDy�DCOylocÞ2

qDCOtt is the transportation time for order DCO from the

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 ¼DCOst þDCOtt Product reaches the customer at time DCOdt.

7DRPðtÞ¼ DRT�DIPðtÞ � ∑DRWðtÞ, if DRPðtÞ > 0, t¼ c 3DPC, c¼ 1, 2, 3

0 , otherwise

(the distributor employs the fixed interval procurement policy. It

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

level DRT. ∑DRW(t) is the amount of product which has been

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.

8 DOtime ¼ t

9 DOamt ¼DRPðtÞ

10 DPAðtÞ¼CJamtat t¼CJdt 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 CJ¼ JobSch1 if IRrðtÞ 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 rawmaterial 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 ¼RCPr 3CJamt 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 CJpt the plant processing status is updated after a job is started. CJpt is

the total processing time for job CJ.

17PSðt þ 1Þ ¼ PSðtÞ � 1 ifPSðtÞ > 0

0 otherwise

(as time progresses, the processing status is updated to keep track of

the amount of time remaining to complete the current job.

18 RUrðtÞ ¼ RMreqr RUr(t) is the amount of rawmaterial r used for production at time t.

19 CJpt ¼ PR 3CJamt CJpt is the processing time for job CJ, based on the processing time

per unit product PR.

20 CJpkgt ¼ PkgR 3CJamt CJpkgt is the packaging time for job CJ, based on the packaging time

per unit product PkgR.



2005 case, because less raw materials need to be transportedresulting in fewer trips.4.2. Scenario 2: Changing Distribution Network Config-

uration. The second scenario evaluates the effects of thedistribution network by comparing a single distributor (the basecase) versus two distributors for a given geographical market.While the single distributor channel could be more cost-efficientand easier to manage, two distribution channels would have thebenefit of being at closer proximity to customers. Anotheradvantage of the two distributor channels is that robustnessincreases since in the event of disruptions one can serve as a

backup to the other, leading to higher customer satisfaction level.Table 6 shows the changes in the environmental impacts percarton of diaper in the 2-distributor scenario as compared to thebase case. Overall, there is an 85% increase in all environmentalimpacts other than water use for the plant-to-distributor trans-portation. The reason is that the second distributor is locatedfarther from the manufacturing plant than the existing one.However, this is partially offset by lower distributor-to-customertransportation impacts (other than water use), which decrease by24%, since customers are served by the nearer distributorresulting in less travel distance. The table also shows comparable

Table 2. Continuedno. equation explanation

21CJtt ¼TS

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðPx�DxÞ2 þ ðPy�DyÞ2

qCJtt is the transportation time for job CJ from the plant to the

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 ¼CJst þCJpt þCJpkgt þCJtt product reaches the distributor at time CJdt.

23RPrðtÞ¼ RTr � IRrðtÞ � ∑ RWrðtÞ, if IRrðtÞ þ ∑ RWrðtÞ < RRr

0 , otherwise

(the plant employs the reorder point procurement policy. Raw

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 rawmaterial r, and ∑RWr(t) is the amount of rawmaterial r

which has been ordered but is yet to arrive at the plant.

24 RArðtÞ¼RPrðt� LTrÞ RAr(t) is the amount of rawmaterial r arriving at the plant at time t,

LTr is the lead time between purchase and arrival for rawmaterial r.

25 IRrðt þ 1Þ¼ IRrðtÞ þ RArðtÞ � RUrðtÞ material balance on raw material inventory at the plant

Table 3. Calculation of Economic Indicators


1 DCOrev ¼DPrice 3DCOamt DCOrev is the distributor’s revenue from order DCO. DPrice is the

price per unit product charged by the distributor.

2DCOtc ¼DcostT 3DCO

amt3

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðDx�DCOxlocÞ2 þ ðDy�DCOylocÞ2

qDCOtc is the transportation cost for order DCO. DCostT is the

transportation cost per unit distance per unit product from the

distributor to the customer.

3

DProf it¼ ∑DCOrev �

∑DCOtc

þ ∑tPrice 3DPAðtÞ

þ ∑tDcostI 3DIPðtÞ

þ ðDcostOF 3D 3TÞ

0BBBBBBB@

1CCCCCCCA

DProfit is the profit of the distributor, Price is the price per unit product

charged by the plant and paid by the distributor, DCostI is the inventory

cost per unit product per time tick, and DCostOF is the fixed operating

cost (charged at each time tick), D is the number of days in the

simulation horizon, and T is the number of time ticks in one day.

4 CJrev ¼ Price 3CJamt CJrev is the plant’s revenue from job CJ. Price is the price per unit

product charged by the plant.

5 CJpc ¼CostOV 3CJpt CJpc is the variable processing cost for job CJ. CostOV is the processing

cost charged when the plant is processing a job.

6 CJpkgc ¼CostPkg 3CJamt CJpkgc is the packaging cost for job CJ. CostPkg is the packaging cost

per unit product.

7 CJtc ¼CostT 3CJamt

3ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðPx�DxÞ

p 2 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.

8

Prof it¼ ∑ CJrev �

∑ ðCJpc þCJpkgc þ CJtc þCJpenÞþ ∑

t∑rCostRr 3RArðtÞ

þ ∑t∑rCostI 3 IRrðtÞ

þ ðCostOF 3D 3TÞ

0BBBBBBB@

1CCCCCCCA

profit is the profit of the manufacturing plant, CostRr is the price of raw

material r, CostI is the inventory cost per unit rawmaterial per time tick,

and CostOF is the fixed operating cost (charged at each time tick

regardless plant is processing or idle).



environmental impacts in the other SC stages. Overall, there is astatistically significant 5% increase in ozone layer depletion and astatistically 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 ofthe 2-distributor scenario is 10% lower than the single distributorcase due to higher operating costs. Such increases in theenvironmental impacts and costs demonstrate the trade-offarising from higher resilience and customer satisfaction providedby the two distributors. This scenario thus highlights thesignificant 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 thedistributor is analyzed. Less frequent ordering in larger batcheswould mean fewer transportation trips and consequently reduc-tion in transportation impact and cost. In the base-case scenario,the distributor places an order to the manufacturing plant dailyand products are delivered daily from the plant. We simulate andcompare this base-case scenario with a different ordering policy

involving 2-day procurement interval. Table 7 shows a statisti-cally significant (6%) reduction in all environmental impactsother than water use for the plant-to-distributor transportationdue to fewer transportation trips (although of higher loads). It

Table 4. Calculation of Environmental Indicators


1 TEIRMi ¼ ∑t∑rEIRMi, r 3RArðtÞ 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.

2 TEITSPi ¼ ∑t∑rEITRMi 3RArðtÞ 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.

3 TEIMi ¼ ∑tEIMi 3DPAðtÞ TEIMi is the total environmental impact i from manufacturing. EIMi is the

environmental impact i from manufacturing per unit product.

4 TEIPkgi ¼ ∑tEIPkgi 3DPAðtÞ TEIPkgi is the total environmental impact i from packaging. EIPkgi is the environmental

impact i from packaging per unit product.

5 TEITPDi ¼ ∑tEITPi 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.

6 TEIDi ¼ ∑tEIDi 3DPAðtÞ TEIDi is the total environmental impact i from the distributor operation. EIDi is the

environmental impact i from distributor operation per unit product.

7 TEITDCi ¼ ∑tEITPi 3DCDistðtÞ 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.

8 TEIi ¼TEIRMi þ TEIPkgi þ TEIMi þ TEIDi

þ TEITSPi þ TEITPDi þ TEITDCi

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%

Figure 2. Total environmental impacts along the supply chain percarton diaper for Scenario 1.



also shows a statistically insignificant (1%) reduction in envir-onmental impact during the packaging and manufacturing stagedue to slightly lower diaper production in the 2-day scenario.The amount of diapers ordered by the distributor and hence theamount of diapers to be produced by the manufacturing plantare functions of the inventory top-up point, whose primary roleis to ensure a desired customer satisfaction. Different top-uppoints would therefore result in different economic and envir-onmental performance. Further analysis of this issue would bethe subject of our future study. Overall, the three scenarios serveto highlight the advantage of the integrated SC simulation andLCA indicator approach for studying the environmental impactsarising from the dynamics of the SC network and operatingpolicies.

’ASSOCIATED CONTENT

bS Supporting Information. Derivation of environmentalimpact indicators and case study parameters. This material isavailable free of charge via the Internet at http://pubs.acs.org.

’AUTHOR INFORMATION

Corresponding Author*Phone: +65 65168041; fax: +65 67791936; e-mail: [email protected].

’ACKNOWLEDGMENT

We thank Ms. Patricia Petrus of the National University ofSingapore, Department of Chemical and Biomolecular Engineer-ing for her contribution to the modeling work.

’REFERENCES

(1) Brundtland, G. H. (Ed.). Our Common Future: The WorldCommission on Environment and Development; Oxford University Press:Oxford, 1987.

(2) Elkington, J. Cannibals With Forks: The Triple Bottom Line of 21stCentury Business; Capstone Publishing: Oxford, 1997.

(3) KPMG. KPMG international survey on corporate responsibilityreporting 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 chainsmodeling: An LCA approach. Comp. Chem. Eng. 2009, 33, 1747–1759.

(10) Guill�en-Gos�albez, G.; Grossmann, I. A global optimization forenvironmentally conscious design of supply chains under uncertainty inthe damage assessment model. Comput. Chem. Eng. 2010, 34, 42–58.

(11) Mele, F. D.; Kostin, A. M.; Guill�en-Gos�albez, G.; Jim�enez, L.Multiobjective model for more sustainable fuel supply chains. A casestudy 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 withchemical 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 ISOStandards; Guin�ee, J. B., Ed.; Kluwer Academic Publishers: Dordrecht,2002.

Table 6. Comparison of Environmental Impact Per Carton Diaper in 2-Distributor and 1-Distributor Scenariosa


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


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.



(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 forOEMs from a lifecycle perspective using lifecyclesimulation. 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 (EuropeanDisposables andNonwovens 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.

Documents

Decision Support for Green Supply Chain Operations by Integrating Dynamic Simulation and LCA Indicators: Diaper Case Study