Vous êtes sur la page 1sur 5

SYSTEMS MODELING AND THE ENVIRONMENT

Dynamic Modeling of
Industrial Ecosystems
Matthias Ruth

sumption processes themselves through changes


in technologies or behaviors, changes in the deManaging industrial ecosystems for sustain- sign of products, and changes in the institutions
ability means, to some extent, having to avoid that govern products life cycles from cradle to
the unintended consequences of production and grave. Again, none of these changes typically
occurs instantly. Substitutconsumption
processes.
ing inputs or redesigning
These consequences are Given the complexity of the susproducts often results in exnot easily grasped by those
tainability challenge, dynamic tended trials, altering prowho extract, convert, use,
cesses requires turning over
and release materials and models may have an importhe existing capital stock
energy associated with all tant role to play in closing the
and influencing the behavtransformations along the
loop that connects information iors of large groups of conentire chain of goods and
sumers, and changing inservices. Some of these capture, knowledge generation,
stitutions means reshaping
consequences
manifest system intervention, observation
the rules and procedures
themselves as environof
outcomes,
and
generation
of
that govern their functionmental insults, human
new
knowledge
that
helps
inform
ing and waiting for the
health impacts, and legal
new rules and procedures to
liabilities as well as broader the next round of system interpenetrate often large, hiereconomic and social costs.
ventions.
Dynamic
modeling
archical organizations that
And because these consethemselves exhibit interquences are typically not capturing the dynamics of sysnal inertia and are disfelt instantly, gaining a tems and making the process of
tributed across the globe.
full understanding of the
modeling itself a dynamic one
lags between causes and
All this, of course, hapeffects across space and becomes then an integral part of
pens in a world of incomtime is difficult, if not adaptive management.
plete information at each
impossible.
step along the way. All too
The means to shape the behaviors of industrial often, there are fundamental uncertainties about
ecosystemsthe assemblages of producers and the complexity of interactions among the many
consumers and their interrelationshipsinclude microdecisions made in the social and economic
changes in inputs into production and consump- realms, the vast number and frequently nontion processes, alterations of production and con- linear nature of environmental processes that
influence and are influenced by these decisions, and the emergence of novel phenomena in each of these realms. New technologies advance the measurement and monitoring
Supplementary material is available
of substances that were hitherto not on the
on the JIE Web site

c 2009 by Yale University
radar even of experts. New concerns emerge
DOI: 10.1111/j.1530-9290.2009.00186.x
among the public. New laws and regulations are
instituted.
Volume 13, Number 6

The Challenge

www.blackwellpublishing.com/jie

Journal of Industrial Ecology

839

SYSTEMS MODELING AND THE ENVIRONMENT

On the one hand, all this change could


give one pause, if not paralyze decision making.
On the other hand, people do make management decisions and policy interventions to promote sustainability. Assuming that these achieve
their goal presumes some level of knowledge
about the causeeffect relationships that connect behaviors of individual components of the
environmental, social, and economic spheres
with each other and across system hierarchies
from a population to the larger ecosystem,
from the household to entire communities, or
from the machine to the firm and industry, for
example.

Assembling the Pieces of the


Puzzle
Of course, all one can ever do is make the
best decisions possible on the basis of the limited
knowledge one has. Modeling has always played
a role in enhancing understanding and in providing a basis on which to choose alternative system
interventions. Historically, much of that modeling was mental and verbal in nature, but as the
complexity of decisions has increased, increasing
emphasis has been placed on analytical and quantitative methods (Pagels 1988). Because efforts to
move industrial ecosystems toward sustainability
must be cognizant, to the best degree possible,
of unintended consequences in the economy, society, and the environment, these efforts need
to draw on formal modelsparticularly on those
that connect environmental, social, and economic processes across system hierarchies, space,
and time. And because of the richness of data and
relationships, computers are used to develop and
solve those models.
The study of dynamic systems has evolved
from analysis of relatively simple electromechanical devices to consideration of interdependent social, economic, political, and environmental systems that include highly uncertain
human perception and behavior. Objectives in
dynamic modeling include the identification of
system conditions associated with degraded performance or instabilities, the design of control
systems to ensure stable performance, characterization of possible transitions between multiple
stable states for a system (i.e., bifurcations or
840

Journal of Industrial Ecology

catastrophes), and the proper use and enhancement of monitoring data often plagued by missing
or inaccurate observations. Dynamic modeling
methods are generally lumped or aggregate; they
use simultaneous nonlinear differential equations
to predict the coevolution of system variables. In
recent years, however, increases in computational
power have enabled the development of disaggregate models, in which individual agents (e.g.,
companies or consumers) and their interactions
are modeled explicitly.1
Formal models have many purposes. They can
help structure what is known about causeeffect
relationships. They help organize that knowledge, and, by doing so, they help articulate questions about what is not known. Models thus not
only are repositories of information but can direct
further collection of data and guide research.
If oriented toward support of investment and
policy decisions, models ideally are amenable
to experimentation for and by decision makers.
They become the flight simulators by which the
implications of alternative actions are explored
before interventions in the more complex settings of real-world investment and policy making
are deployed. They can help train the minds and
responses of the pilots who fly entire companies,
industriesin some sense, our globe. To be effective, such models capture feedbacks across space
and time and across the hierarchies of the systems of interest. Many examples of such models
are found in the various issues of this journal (e.g.,
Rejeski 1998) and, for example, in the work by
Ruth and Davidsdottir (2008, 2009).

What New Can Come From


Dynamic Modeling?
Usually, models capture what is known and,
on the basis of that knowledge, project into the
future the behavior of a system. If the system
of interest is large and diverse, experts from different disciplines often collaborate to integrate
their data and information into a single model,
which allows them to capture what is usually assumed to be exogenous from their own narrower
viewpoints. The investment and policy decisions
then affect parameters in one or more parts of the
system and thus drive a systems dynamics. Such
decisions are typically made with an eye toward a

SYSTEMS MODELING AND THE ENVIRONMENT

particular goal, such as waste minimization, profit


maximization, or social welfare maximization.
The tendency for system interventions here
is largely informed by past experiences. For example, taxes are introduced to raise the cost of
an input, and, on the basis of historical experiences of pricequantity relationships, changes
in resource use and emissions are observed. Such
interventions then result in gradual movement
away from the status quo. There is usually little
opportunity for true surprise both about system
behavior and about options for interventions, and
the value of these models largely lies in quantifying the outcomes in cases in which the individual
pieces of the larger puzzle are fairly well understood, whereas the interaction of the pieces with
each other may be less apparent.
Whether interventions that manipulate individual parameters are sufficient to promote sustainability is often at question. Of course, one can
always ratchet up one parameter or another and
arrive at more drastic system responses, but one
will largely be tied to the existing structures and
the limited opportunities for thinking and action
they purport.
There are other ways of using models as guides
for investment and policy decisions. One of
them begins with what is known and then explores which parameter choices lead to undesirable system outcomes. This kind of approach
sometimes referred to as anticipatory failure determinationawakens the mischievous behavior in
people, which can free their thinking and promote a different kind of creativity, albeit an initially destructive one. Really exciting surprises
may be had. A better understanding of the constellations of actions that result in system failure
can then be used to put in place safeguards (monitoring, controls, policies) that keep parameters
outside the ranges that promote disaster. This
approach is particularly useful for risk minimization in complex settings. Researchers have used a
similar approach to evaluate alternative scenarios for future climate change and infrastructure
planning.
Yet another approach begins with a description of a desirable end state and then runs the
model back to the present to see which decisions
one needed to make to arrive at the end state.
Such backcasting, too, helps modelers and deci-

sion makers to break out of thinking that is overly


shaped by the status quo and challenges them to
envision and formalize an end state that is typically richer than the one portrayed by traditional
optimization of a specific target function. Backcasting frees the mind and model of the intellectual straightjacket imposed on them by history.
Really new knowledge may be generated by this
method.

Adaptive and Anticipatory


Management
Although dynamic models can be powerful
tools to structure debate, they usually are only
one of many contributors to decision making. Often, they play no role at all other than perhaps in
supporting decisions that would have been made
regardless. Sometimes, the information they generate enters into the decision making process
alongside competing pieces of information, and,
if done right, models can then help provide a
structured basis on which to evaluate and reflect
on the information used for decision making.
Given the complexity of the sustainability
challenge, dynamic models may have an important role to play in closing the loop that connects
information capture, knowledge generation, system intervention, observation of outcomes, and
generation of new knowledge that helps inform
the next round of system interventions. Dynamic
modelingcapturing the dynamics of systems
and making the process of modeling itself a dynamic onebecomes then an integral part of
adaptive management (Hannon and Ruth 2001).
Used in combination with such methods as anticipatory failure determination or backcasting,
it can support anticipation of and preparation for
unintended consequences.
Decisions about whether something is indeed
an unintended consequence require some form
of involvement by those groups that may be affected. Similarly, the choice among actions cannot meaningfully be made on the basis of modeled
system behavior alone. Value judgments, ethical
criteria, and norms need to enter the picture. It
is therefore essential to make stakeholders an integral part of the dynamic modeling process.

Ruth, Dynamic Modeling of Industrial Ecosystems

841

SYSTEMS MODELING AND THE ENVIRONMENT

What We Do, and What We


Dont Do but Should
The scientific and environmental management literature is replete with descriptions of
databases, simulations, and decision support
tools. A growing number of case studies tell us
how to leverage involvement of stakeholders.
Theoretical and analytical approaches are being developed to explore behaviors across system hierarchies, space, and time. But researchers
have not fully made the connections across all
these essential elements in part because we lack
commonly agreed-on and followed protocols for
dynamic modeling that span system hierarchies,
space, and time and because we have not yet
developed a sufficiently adequate suite of approaches and tools for model effectiveness assessments. As industrial ecologists, we have not yet
defined an equivalent, for example, to the National Energy Modeling System (NEMS) used
in the United States to monitor and guide energy related developmentsneither at the local
nor the national scale. Nor do we have in place
institutional structures that enable longer term
stakeholder-guided adaptive management of industrial ecosystems. What we do have by now,
though, is an impressive arsenal of data, tools,
and people to elevate dynamic modeling to the
next level, and with that we can help to drive the
sustainability agenda.

Note
1. References on dynamic modeling and related literature can be found in Supplementary Appendix S1
on the Web.

References
Hannon, B. and M. Ruth. 2001. Dynamic modeling.
Second edition. New York: Springer-Verlag.
Pagels, H. 1988. Dreams of reason. New York: Simon
and Schuster.
Rejeski, D. 1998. Learning before doing: Simulation
and modeling in industrial ecology. Journal of Industrial Ecology 2(4): 2944.
Ruth, M. and B. Davidsdottir, eds. 2008. Changing
stocks, flows, and behaviors in industrial ecosystems.
Cheltenham, UK: Edward Elgar.
Ruth, M. and B. Davidsdottir, eds. 2009. The dynamics of regions and networks in industrial ecosystems.
Cheltenham, UK: Edward Elgar.

About the Author


Matthias Ruth holds the Roy F. Weston Chair
in Natural Economics at the University of Maryland in College Park, Maryland, where he serves
as the director of the Center for Integrative Environmental Research.

Address correspondence to:


Prof. Matthias Ruth
Center for Integrative Environmental Research
University of Maryland
Suite 2202, Van Munching Hall
College Park, MD 20742
mruth1@umd.edu
www.publicpolicy.umd.edu/faculty/ruth/

Supplementary Material
Additional Supplementary Material Information may be found in the online version of this
article:
Supplement S1. This supplement contains an appendix with references to the research literature
related to this columns discussion of dynamic modeling.
Please note: Wiley-Blackwell is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be
directed to the corresponding author for the article.

842

Journal of Industrial Ecology

Copyright of Journal of Industrial Ecology is the property of Blackwell Publishing Limited and its content may
not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.

Vous aimerez peut-être aussi