Modeling the Impact of Product Portfolio on the Economic
and Environmental Performance of Recycling Systems
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Dahmus, J.B. et al. “Modeling the impact of product portfolio on
the economic and environmental performance of recycling
systems.” Sustainable Systems and Technology, 2009. ISSST
'09. IEEE International Symposium on. 2009. 1-6. © 2009 IEEE
As Published
http://dx.doi.org/10.1109/ISSST.2009.5156765
Publisher
Institute of Electrical and Electronics Engineers
Version
Final published version
Accessed
Thu May 26 19:16:14 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/58993
Terms of Use
Article is made available in accordance with the publisher's policy
and may be subject to US copyright law. Please refer to the
publisher's site for terms of use.
Detailed Terms
Modeling the Impact of Product Portfolio on the
Economic and Environmental Performance of
Recycling Systems
Jeffrey B. Dahmus, Elsa A. Olivetti, Susan A. Fredholm,
Jeremy R. Gregory, and Randolph E. Kirchain
Abstract— Through the development of a general model of
electronics recycling systems, the effect of product portfolio
choices on economic and environmental system performance is
explored. The general model encompasses the three main
functions of a recycling system – collection, processing, and
system management – and allows for the effect of both contextual
and architectural inputs – including product scope – to be
explored. Overall model results indicate that collecting a broader
portfolio of products can be economically favorable, even for
cases in which lower-value products are added to a recycling
system. In these cases, the higher total mass throughputs that are
realized by the collection of additional product types can help to
drive down the cost per unit mass collected. Expanding product
scope can also yield improvements in environmental
performance, as the energy per unit mass collected can also
decrease with higher mass throughputs.
Index Terms—electronics recycling systems, electronic waste,
process-based cost models, product scope
I. INTRODUCTION
W
ITH an increasingly broad scope of electronic
products being produced, along with diminishing
lifetimes for such products, the amount of
consumer
electronics reaching end-of-life has grown considerably over
the past decade [1, 2, 3]. In response to this growth in
electronic waste, recycling systems to handle such waste have
become more widespread, with various systems currently
implemented in Europe, North America, Asia, and elsewhere.
As these existing systems mature, and as more such systems
come online, understanding the economic and environmental
performance of such systems becomes critical, both to enable
improvement of existing systems and to inform the design of
new systems.
Previous work by the authors has focused on collecting and
analyzing empirical data from existing electronics recycling
systems [4].
While such information is important to
understanding system performance, the sheer number of
variables involved with each system – from architectural
inputs such as product scope, number of collection points, and
location of collection points, to contextual inputs such as
population, population density, and factor costs – make direct
comparison of performance results difficult. Moreover, using
such comparisons to identify system variables that could be
changed in order to improve system performance is virtually
impossible. Given these difficulties, a modeling framework
has been previously proposed and developed to analyze the
economic and environmental performance of both existing and
prospective recycling systems [5, 6].
This model, which comprehends the three main functions
common to most recycling systems – collection, processing,
and system management – can be used to explore the effect of
different architectural and contextual variants. In this work,
the architectural decisions around product portfolio selection
are explored. Product scope, which refers to the types of
products collected within a recycling system, is an important
decision that system designers and system operators must
make. It is also a decision in which there is variation, as
shown in Fig. 1. This variation in product scope is particularly
apparent in the state-based systems in the US, where different
recycling systems collect various combinations of IT,
telecommunication, and consumer equipment. Understanding
the effect of such product portfolio decisions, on both the
economic and environmental performance of a recycling
system, is the aim of this work.
II. MODEL DEVELOPMENT
J. B. D. is a Postdoctoral Associate in the Materials Systems Laboratory at
the Massachusetts Institute of Technology (MIT), Cambridge, MA, 02139
(e-mail: [email protected])
E. A. O. is a Postdoctoral Associate in the Materials Systems Laboratory
and at MIT (e-mail: [email protected]).
S. A. F. was a Research Assistant in the Materials Systems Laboratory at
MIT. She is now a Consultant at PE Americas (e-mail: [email protected]).
J. R. G. is a Research Scientist in the Materials Systems Laboratory at MIT
(e-mail: [email protected]).
R. E. K. is an Associate Professor in Materials Science and Engineering
and in the Engineering Systems Division at MIT (e-mail: [email protected]).
As mentioned above, the system model developed to
analyze recycling systems focuses on the three main functions
in a recycling system, namely collection, processing, and
system management [7]. It should be noted that while the
model inputs described here are specific to electronics
recycling, the broader model framework can be more
generally applied, and can thus be adapted to model recycling
systems for other consumer durable goods, in addition to
electronics.
Jurisdictions
CA
CT
MD
United States
ME
MN
OR
WEEE Category 3: IT and telecommunications equipment
Monitor
x
x
x
x
x
Laptop Computer
x
x
x
x
x
Desktop Computer
x
x
x
Other (e.g. printers)
x
x
x
x
WEEE Category 4: Consumer equipment
Television
x
x
Other (e.g. VCRs)
x
x
x
x
x
Canada
BC
SK
TX
WA
AB
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
NS
BE
CH
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Europe
NL
NO
x
x
x
x
SE
x
x
x
x
x
x
x
x
x
x
x
x
Fig. 1. Comparison of product scopes for current electronics recycling systems. United States: CA = California, CT = Connecticut,
MD = Maryland, ME=Maine, MN = Minnesota, OR = Oregon, TX = Texas; Canada: Canada: AB = Alberta, BC = British Columbia,
SK = Saskatchewan, NS = Nova Scotia; Europe: BE = Belgium (Recupel), CH = Switzerland (SWICO), NL = Netherlands (ICT Milieu),
NO = Norway (Elretur), SE = Sweden (El-Kretsen).
Fig 2. shows a schematic overview of the recycling system
analyzed here, and identifies the key model components.
Consumer households, some with end-of-life electronics ready
for disposal, and some without, are shown on the left side of
Fig. 2. End-of-life electronics travel, typically by car, from
the residential consumers to a local collection facility. Here,
various types of end-of-life electronics, and perhaps other endof-life products and materials, are aggregated. From here,
these larger collections of end-of-life electronics are
transported, typically by truck, to a processing facility. At
such a facility, product dismantling and separation occurs,
with resalable materials, resalable components, and waste
leaving the facility. Providing operational and financial
oversight of the process, from the collection facility to the
processing facility, is typically a system management function.
Each of the main model functions are described briefly here,
and important characteristics of each function are highlighted.
Additional details regarding each of these models can be
found in previous works [5, 6].
A. Collection
As shown in Fig. 2, collection focuses on three main
activities: the transport of end-of-life products from their point
of final use to a collection facility, the collection facility itself,
and the transport of aggregated end-of-life products from a
collection facility to a processing facility. In modeling these
activities, two distinct but integrated models are used: a
logistics model, which accounts for the transportation of
products, and a facility model, which accounts for the
operation of the collection facility.
System Management
Materials
Components
Transport
(car)
Collection
Facility
Transport
(truck)
Collection
Logistics Model and Facility Model
Processing
Facility
Waste
Processing
Facility Model
Fig. 2. An overview of the recycling system analyzed. The three
main functions – collection, processing, and system management
– are shown. Specific models are noted in grey text.
Logistics Model.
The logistics model is primarily
concerned with calculating the distances that end-of-life
electronics must travel and the total amount of electronic
waste collected. While such information is obtainable when
analyzing actual systems, when modeling alternative
hypothetical scenarios, such logistics information must be
generated.
The starting locations of end-of-life electronics, which is
typically the home of the final consumer, can be modeled
using demographic data along with population distribution
models [8]. Once such a characterization of locations takes
place, collection facilities can then be sited based on this data.
The siting of collection facilities for electronics recycling falls
into a more general class of facility-siting problems, in which
a limited number of supply points – in this case collection
facilities – must be sited to serve a given population of
demand points – in this case consumers with end-of-life
electronics. Such a problem is common in location-allocation
modeling; in fact, such problems arise in siting decisions for a
broad range of different facilities, from hospitals to
warehouses to retail outlets [9]. One common approach to this
siting problem is the p-median model, an optimization
approach that, when given a set of demand points, selects a set
of supply points, p, such that the total sum of the distances
from each demand point to its closest supply point, is
minimized [9, 10].
The p-median problem has been shown to be NP-hard; thus,
heuristic methods are often instead used to address this
location-allocation problem [9, 10]. For the collection facility
siting problem faced here, a k-means clustering algorithm is
used. This approach locates a number of collection facilities,
k, such that the variance in the distances that each customer
must travel to reach a collection facility, is minimized. While
this objective differs somewhat from that of the p-median
problem, the two approaches appear to yield comparable
location decisions for collection facilities. For siting of
processing facilities, given a set of collection facility
locations, similar clustering algorithms can again be used. It
is important to point out that the k-means algorithms used here
for facility siting, do not consider existing infrastructure; thus,
siting is done assuming a blank slate.
With locations for the population, collection facilities, and
processing facilities, the travel distances for each of these
transportation segments can then be calculated. Since these
two different transportation segments are typically completed
by different stakeholders using different classes of vehicles,
the economic and environmental impacts of these segments
can be quite different.
Once locations for the population and facilities are
established, the amount of mass collected must be determined.
The amount of mass collected is dependent on a range of
factors, from product scope to the penetration level of
electronic products to the distance that a consumer must travel
to reach a collection facility. The effect of distance in
consumer participation rates is of particular interest in this
work, as this effect can have profound impacts on the amount
of mass collected at a given collection facility. Much of the
recycling literature assumes participation rates to be constant
over a certain distance; beyond that distance, participation
rates drop to zero. However, the effect of distance on
participation rates has been shown to follow more of an
exponential decay – known as distance decay – with
consumers close to a facility participating at high rates and
those farther from a facility participating at considerably lower
rates. This behavior has been observed in numerous cases,
from shopping to commuting to courtship [11, 12, 13].
The effect of distance decay can be modeled using an
exponential function:
R = qe − λx ,
(1)
where R is the distance-adjusted participation rate, q is the
participation rate, λ is the decay rate with distance, and x is the
distance from the customer to the collection facility [14, 15].
The participation rate, q, convolves a variety of factors,
including the saturation of electronic products, the availability
of other end-of-life disposal options, and customer awareness
of collection centers, among others.
Using this distance-adjusted participation rate, R, along with
information about product masses and product scope, the
amount of mass collected in total, along with the breakdown
of that mass by product type, can be estimated.
Facility Model. In modeling collection, the cost of
operating a collection facility is also included. The collection
facility is modeled using a process-based cost model (PBCM),
which maps engineering requirements to process descriptions,
process descriptions to facility requirements, and finally
facility requirements to facility investments [16, 17]. The
costs of the collection facility include capital costs, such as
buildings, equipment, and other infrastructure, as well as
operating costs, such as labor, electricity, and packaging
material; all of these costs can vary with context. In this
model, it is assumed that collection facilities are nondedicated.
B. Processing
During processing, electronic waste that has been collected
and consolidated is broken down into resalable components,
resalable material streams, and waste streams. The processing
facility model, which captures this process, focuses on both
the operations within the facility itself, as well as on the
material flows into and out of the facility. Like the model for
the collection facility, the model for the processing facility is
also a PBCM, built up from engineering requirements, in this
case requirements for end-of-life electronics processing. This
PBCM again takes in a number of contextual inputs, including
factor costs such as capital costs, labor costs, and electricity
costs. However, it also takes in information from the
collection logistics model, both regarding the total mass of
electronics waste collected, as well as regarding the
breakdown of that mass total by product type.
C. System Management
The system management component of the model accounts
for the management and oversight of the entire system. These
costs are largely administrative, and are heavily dependent on
the fee structures and oversight mechanisms that are put in
place; as fee structures become more complex, and as
oversight increases, system management costs increase. In
general, these costs can be modeled simply, with labor and
related expenses often the dominant costs.
III. CASE STUDY
To explore the implications of product portfolio decisions
on the economic and environmental performance of recycling
systems, a case study was conducted using the models
described above. The contextual inputs and architectural
inputs – outside of those related to product scope – are based
loosely on the demography, electronics recycling system
structure, and factor costs for the state of Maine.
The electronics recycling system analyzed here focuses on
the collection of end-of-life monitors and printers from
residential consumers. The total population covered by this
system is approximately 1.3 million people, distributed over
an area of roughly 80,000 square kilometers. Given this
population, and a population distribution based on recent
census data for Maine [18], collection facilities were sited
using the k-means clustering algorithm described earlier.
Participation rates at these facilities were determined using the
exponential form of distance decay, as shown in (1), with the
distance decay calibrated such that 75% of the participants at a
given collection facility live within 10 km of that facility. For
the processing facility, the outgoing material streams were
priced using current market data [19]. System management
costs, based on literature on the Maine electronics recycling
system, were a fixed amount, and did not scale with either the
amount of mass collected or the number of collection sites [7].
The three product portfolios considered in this analysis
were: 1) monitors only, 2) printers only, and 3) both monitors
and printers. While both monitors and printers incur a net cost
for recycling, monitors, with a relatively high metal content,
represent a more-valuable waste stream; printers, typically
dominated by plastics, represent a less-valuable waste stream.
6
Printers
Monitors
Monitors and Printers
Total Mass Collected (million kg)
5
4
3
2
1
0
0
50
100
150
200
250
300
Collection Facilities
Fig. 3. Total mass collected as a function of the number of
collection facilities for three systems with different product
portfolios: 1) printers, 2) monitors, and 3) monitors and printers.
1.6
Total System Cost per Unit Mass Collected ($/kg)
A. Economic Performance Results
Product scope plays an important role in the amount of
mass collected in each scenario. As shown in Fig. 3, an
electronics recycling system based on collecting only monitors
– estimated to be 25 kg each – collects significantly more
mass than a recycling system based on collecting only printers
– estimated to be 10 kg each. Not surprisingly, a system that
includes both monitors and printers is able to collect
significantly more mass than if either product is collected
separately. It is also clear from Fig. 3 that the number of
collection facilities has a profound effect on the amount of
mass collected, particularly at lower numbers of collection
facilities.
Fig. 4 shows the total system cost per unit mass collected
for the three different product portfolios. These costs include
the collection facility, transportation by truck, the processing
facility, and system management. The cost borne by the
consumer, in transporting the end-of-life electronics from their
home to the collection facility, is not included in Fig. 4.
Overall, each of the portfolios follows the same general
curve. The cost per unit mass collected at very low numbers
of collection facilities, is largely driven by the costs of
processing and system management.
In the case of
processing, the low mass totals at low numbers of collection
facilities, lead to limited economies of scale, and thus higher
costs; in the case of system management, the amortization of
system management costs over a relatively small amount of
mass, leads to high costs per unit mass collected. As the
number of collection facilities increases, the overall cost of
operating all these facilities also increases. At higher numbers
of collection facilities, this proves to be the main system cost
driver.
The product portfolio selected also has an important impact
on the cost per unit mass collected of the electronics recycling
system. For a system focused on recovering only printers, the
cost per unit mass collected is quite high, as can be seen in
1.4
1.2
1
0.8
0.6
0.4
Printers
Monitors
Monitors and Printers
0.2
0
0
50
100
150
200
250
300
Collection Facilities
Fig. 4. Total system cost per unit mass collected as a function of
the number of collection facilities for three systems with different
product portfolios: 1) printers, 2) monitors, and 3) monitors and
printers.
Fig. 4. This is due in part to the fact that printers, with a large
mass fraction of plastics, do not yield particularly valuable
outgoing material streams from the processing facility.
However, the larger cost driver is the fact that a system based
around collecting only printers does not accumulate large
amounts of total mass. Thus, opportunities to amortize costs –
in particular collection facility costs and system management
costs – are not realized.
For a system focused on recovering only monitors, the cost
per unit mass collected is considerably lower, as seen in
Fig. 4. While this is due in part to the fact that monitors have
more valuable materials than in printers – including more bulk
metals and precious metals – it is also due in large part to the
larger total system masses collected. These higher mass totals
lead to significant reductions in the cost per unit mass..
Of the three product portfolios considered, the lowest cost
portfolio is that which includes both printers and monitors.
This combination results in the most mass collected, which
drives down the total system cost per unit mass. The fact that
the portfolio with both printers and monitors offers the lowestcost solution has important ramifications for recycling system
design. It appears that once a system is established, collecting
as much mass as possible – which often means collecting a
portfolio of products – can offer cost advantages over systems
with more limited product scopes. With collection facilitiy
costs and system management costs largely driven by fixed
costs, an effective approach to reducing these costs per unit
mass collected is to simply collect more mass. Also, with
economies of scale in processing, more mass results in lower
processing costs per unit mass processed. Even in a system in
which printers – a generally lower-value product – are added
to monitors, the resulting combination leads to lower costs per
unit mass collected than a system in which only monitors are
collected. Thus, while collecting and processing printers by
themselves can be quite expensive, adding such a product to
existing systems can in fact drive down total costs per unit
mass collected. This is an important consideration when
making product portfolio decisions for electronics recycling
systems.
3500
Total Transportation Energy (GJ)
3000
2500
2000
1500
1000
Printers
500
Monitors
Monitors and Printers
0
0
50
100
150
200
Collection Facilities
250
300
Fig. 5. Total transportation energy as a function of the number of
collection facilities for three systems with different product
portfolios: 1) printers, 2) monitors, and 3) monitors and printers.
Printers
8
Total Transportation Energy per
Unit Mass Collected (MJ/kg)
B. Environmental Performance Results
Combining model outputs with environmental data can
allow the environmental performance of different recycling
scenarios to be evaluated. For example, the energy burden of
transportation, which is a topic of interest in any product takeback system, can be analyzed for the various product
portfolios considered. Fig. 5 shows the total transportation
energy, both from private automobiles – used to transport endof-life electronics from the consumer’s home to the collection
facility – and from trucks – used to transport collected end-oflife electronics from the collection facility to the processing
facility.
Data on the environmental burden of these
transportation modes comes from the ecoinvent database [20].
For all three of the product portfolios examined, there are two
competing trends that determine the total transportation
energy. At low numbers of collection facilities, fewer
consumers participate in the system, but each participant must
travel a longer distance. At high numbers of collection
facilities, a much larger number of consumers participate, but
each participant only needs to travel a short distance. The
reduction in distance travelled per customer eventually
dominates, thus explaining the decline in transportation energy
as the number of collection sites increases. In comparing the
transportation energy across the different product portfolios,
the portfolios with more mass lead to higher energy
expenditures, as additional truck trips are necessary.
However, despite this increase in the total transportation
energy with broader product scopes, as seen in Fig. 5, the
transportation energy per unit mass collected, as seen in Fig. 6,
decreases. Thus, much like in the cost analyses earlier, the
most resource-efficient approach, on a per-unit-mass basis, is
the scenario with the broadest product scope. As before,
larger total masses serve to drive down the impacts – both
9
Monitors
7
Monitors and Printers
6
5
4
3
2
1
0
0
50
100
150
200
250
300
Collection Facilities
Fig. 6. Total transportation energy per unit mass collected as a
function of the number of collection facilities for three systems
with different product portfolios: 1) printers, 2) monitors, and 3)
monitors and printers.
economic and environmental – per unit mass collected.
Perhaps the more important environmental issue regarding
product recycling systems is whether or not the environmental
benefits of recycling outweigh the environmental impacts of
collection. Combining the models developed here with energy
data for transportation and material production – both primary
and secondary [21, 22, 23, 24] – can provide preliminary
answers to such questions. In a best-case scenario, assuming
high material recovery rates of both plastic and metals, and
assuming energy savings equal to the difference between the
energy required for primary material production and the
energy required for secondary material production, recycling
any of the three product portfolios examined here makes sense
from an energy perspective. That is, the energy savings from
using secondary material streams far outweighs the burden of
collection logistics. A more in-depth analysis would account
for other potential energy requirements – including other
material preparation and processing steps – as well as actual
recovery yields.
Reasonable scenarios do exist in which the environmental
decision is less clear. For example, with a product such as
printers, which feature a high plastic content, efficient
recovery of these plastics is critical to the overall energy
balance. If plastics cannot be recovered – either for recycling
or for energy recovery – and if the metals in the product
cannot be recovered at high enough yields, scenarios exist in
which a recycling system can transition from providing a net
energy savings to being a net energy burden. Fig. 7 shows
two such scenarios for electronics recycling systems focused
on collecting only printers. For the case in which none of the
plastics and 80% of the metals in a printer are recovered, this
crossover occurs around 40 collection facilities. With fewer
collection facilities, the energy burden from transporting endof-life electronics outweighs the energy savings from
recycling; with more collection facilities, the energy savings
from recycling outweighs the energy burden from
6
Energy Savings or Burden (million MJ)
0% Plastics, 80% Metals
0% Plastics, 50% Metals
5
Transportation
[4]
4
[5]
3
2
[6]
1
[7]
0
0
50
100
150
200
Collection Facilities
250
300
Fig. 7. The energy savings and energy burden of an electronics
recycling systems focused on printers. The energy savings from
material recovery for two scenarios – one with no plastics
recovery and 80% metal recovery, and one with no plastics
recovery and 50% metal recovery – are plotted. The energy
burden from the transportation logistics is also shown, allowing
energy crossover points – in terms of the number of collection
facilities – to be identified.
transportation. With lower metal recovery rates, additional
mass must be collected, in order to break even; this scenario,
with a metal recovery rate of 50%, is also shown in Fig. 7.
Again, it should be noted that these analyses represent bestcase scenarios for recycling. In practice, there are additional
energy burdens that should be considered.
IV. CONCLUSION
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
The development of detailed models to analyze electronics
recycling systems helps to provide important insights into how
such systems can be designed and modified to maximize
economic and environmental performance. The analyses
presented here around product portfolio decisions indicate that
broader product scopes can lead to systems that are both
economically and environmentally more-efficient than
systems with limited product scopes.
[16]
ACKNOWLEDGMENT
[19]
The authors would like to thank Edgar Blanco from the
MIT Center for Transportation and Logistics for his
participation in the development of the collection logistics
model.
[20]
REFERENCES
[1]
[2]
[3]
United States Environmental Protection Agency – US EPA
“Management of Electronic Waste in the United States, Fact Sheet.”
EPA530-F-08-014, July 2008. Available online at
http://www.epa.gov/osw/conserve/materials/ecycling/docs/fact7-08.pdf
(Accessed March 25, 2009).
United States Environmental Protection Agency – US EPA “Electronics
Waste Management in the United States, Approach 1.” EPA530-R-08009, July 2008. Available online at
http://www.epa.gov/osw/conserve/materials/ecycling/docs/app-1.pdf
(Accessed March 25, 2009).
United States Environmental Protection Agency – US EPA
“Management of Electronic Waste in the United States, Approach 2.”
[17]
[18]
[21]
[22]
[23]
[24]
EPA530-R-07-004b, April 2007. Available online at
http://www.epa.gov/osw/conserve/materials/ecycling/docs/app-2.pdf
(Accessed March 25, 2009).
S. A. Fredholm, J. R. Gregory, and R. E. Kirchain, "Characterizing
Architectural Options for Electronic Waste Recycling Systems," IEEE
International Symposium on Electronics and the Environment,
Burlingame, CA, May 19-21, 2008.
J. B. Dahmus, S. A. Fredholm, E.A. Olivetti, J. R. Gregory, and R. E.
Kirchain, "Modeling the Economic and Environmental Performance of
Recycling Systems,” IEEE International Symposium on Electronics and
the Environment, Burlingame, CA, May 19-21, 2008.
J. B. Dahmus, S. A. Fredholm, E.A. Olivetti, J. R. Gregory, and R. E.
Kirchain, "Modeling the Economic and Environmental Performance of
Recycling Systems,” TMS Global Symposium on Recycling, Waste
Treatment and Clean Technology, Cancun, Mexico, Oct. 12-15, 2008.
J. R. Gregory and R. E. Kirchain, “A Framework for Evaluating the
Economic Performance of Recycling Systems: A Case Study of North
American Electronics Recycling Systems," Environmental Science and
Technology, 42(18), pp. 6800-6808, 2008.
T. Rand-Nash, R. Roth, and R. Kirchain, Massachusetts Institute of
Technology, Personal Communication, February 6, 2008.
Current, J., M. Daskin, and D. Schilling, “Discrete Network Location
Models,” in Facility Location: Applications and Theory, Z. Drezner and
H. Hamacher , Eds. Heidelberg, Germany: Springer-Verlag, 2002,
pp. 81-118.
Megiddo, N. and K. Supowit, “On the Complexity of Some Common
Geometric Location Problems,” SIAM Journal of Computing, 13(1),
pp. 182-196, February 1984.
Farhan, B. and A. T. Murray, "Distance decay and coverage in facility
location planning," The Annals of Regional Science, vol. 40, pp. 279295, 2006.
Hurst, M. E. E. A Geography of Economic Behavior: An Introduction.
North Scituate, Massachusetts, USA: Duxbury Press, 1972.
Taylor, P. J. “Distance Transformation and Distance Decay Functions,”
Geographical Analysis, vol. 3, pp. 221-238, 1971.
Sheppard, E. "Modeling and Predicting Aggregate Flows," in The
Geography of Urban Transportation, S. Hanson, Ed. New York, New
York, USA: The Guilford Press, 1995, pp. 100-128.
Fotheringham, A. S. and M. E. O'Kelly. Spatial Interaction Models:
Formulations and Applications. Dordrecht, The Netherlands: Kluwer
Academic Publishers, 1989.
Field, F., R. Kirchain, R. Roth. "Process cost modeling: Strategic
engineering and economic evaluation of materials technologies." JOM
Journal of the Minerals, Metals and Materials Society, vol. 59(10),
pp. 21-32.
Kirchain, R.E. “Cost Modeling of Materials and Manufacturing
Processes,” in Encyclopedia of Materials: Science and Technology,
K.H.J. Buschow, et al., Eds. Oxford, England: Elsevier, 2001, pp. 17181727.
US Census Bureau, Maine -- Place and County Subdivision, 2000.
Available online at http://factfinder.census.gov. (Accessed March 29,
2009).
"Recycler's World," Available online at www.recycle.net. (Accessed
March 29, 2009).
Ecoinvent database 2.0, Swiss Center for Life Cycle Inventories, 2009.
Available online at http://www.ecoinvent.org. (Accessed March 29,
2009).
Atlee, J. R. Operational Sustainability Metrics: A Case of Electronics
Recycling. MS Thesis, Massachusetts Institute of Technology, June
2005.
Habersatter, K. Ecobalance of packaging materials: state of 1990.
Environmental Series No. 132. Berne, Federal Office of Environment,
Forests, and Landscape (BUWAL), 1991.
Keolian, G. A., K. Kar, M. Manion, and J. Bulkley. Industrial Ecology
of the Automobile: A Life Cycle Perspective. Warrendale, Pennsylvania,
USA: Society of Automotive Engineers, 1997.
Graedel, T. E. and B. Allenby. Industrial Ecology and the Automobile.
Upper Saddle River, New Jersey, USA: Prentice-Hall, Inc., 1998.
Download

Modeling the Impact of Product Portfolio on the Economic