Sie sind auf Seite 1von 11

FORUM

Why Do Cities Grow? Insights from


Nonequilibrium Thermodynamics at the
Urban and Global Scales
David Bristow and Christopher Kennedy

Keywords:
cities
complexity
energy
industrial ecology
thermodynamics
urban metabolism

Supporting information is available


on the JIE Web site

Summary
This forum article explores thermodynamic understanding of the growth of cities, including
theoretical foundations, observations, and analysis. The general theory of nonequilibrium
thermodynamics is reviewed, recognizing differences in interpretation between Prigogine
and Schneider and Kay as well as discussing the hypothesis of maximum entropy production.
Calculations of exergy gradients in a few cities and settlements, along with measures of
anthropogenic heat loss in further cities, support the notion that cities are dissipative
structures. The observation that primary energy use per capita increases in Singapore and
Hong Kong as they grow is further evidence to support the thermodynamic understanding
of the growth of cities, indicative of an increasing rate of entropy production. At the
global scale, the strong linear relationship between global urban population and total global
energy use, and the distribution of city sizes according to Zipfs law, can be understood as
emergent results based on thermodynamics. Parallel results might be derived from models
that represent underlying microscale processes, several of which are reviewed. Issues for
future research include: development of nonequilibrium thermodynamic models specific to
city growth; further study of exergy flows of cities with consistent methodology, including
attention to solar energy exchanges in cities; and further exploration of links between
thermodynamic and economic models of urban growth.

Background
Much recognition has been given to the increasing urbanization of the planet, along with the environmental and social
challenges that it poses (Grimm et al. 2008; Kennedy et al.
2009; Hoornweg et al. 2011; Baynes and Wiedmann 2012).
Some have proposed the emergence of a science of cities, giving
work on scaling laws and networks as examples (Batty 2012.,
2013; Bettencourt et al. 2007; Bettencourt and West 2010;
Bettencourt 2013). A fundamental question for such a science
must be: Why do cities grow? Some insights are provided by
the social sciences, discussed below (Jacobs 1969; Glaeser et al.

1992; Krugman 1996). In this article, we tackle the question


from an industrial ecology (IE) perspective, drawing upon
understanding of the thermodynamics of citiesand the global
system in which they nest.
Several researchers in the fields of IE and ecological economics have been inspired by the work of Georgescu-Roegen
(1971, 1979), who noted that economies are subject to the
laws of thermodynamics. Much of the early work in ecological
economics was concerned with trying to understand the implications of Georgescu-Roegens observation (Cleveland and
Ruth 1997; Rpke 2004). On the face of it, stating that a system
obeys the laws of thermodynamics says nothing at allfor all

Address correspondence to: Christopher Kennedy, Department of Civil Engineering, University of Toronto, 35 St. George Street, Toronto, Ontario, Canada, M5S 1A4.
Email: christopher.kennedy@utoronto.ca
2015 by Yale University
DOI: 10.1111/jiec.12239

Editor managing review: John Ehrenfeld

Volume 19, Number 2

www.wileyonlinelibrary.com/journal/jie

Journal of Industrial Ecology

211

FORUM

systems are subject to the laws of physics. The challenge is really


to understand how the laws of thermodynamics manifest themselves and how such understanding can help us better address
global issues of concern. Progress has been made in the work of
Ruth (1993), for example, in which he developed mathematical
models integrating ecological, economic, and thermodynamic
concepts. Ayres and Warr (2005) and Ayres and Voudouris
(2014) have also shown how exergy can explain economic
growth in the United States and other national economies (see
also Proops [1985], Faber and Proops [1985], and the review by
Hammond and Winnett [2009], among others).
Surprisingly, there has been little recognition of the role of
cities in thermodynamic understanding of how economies work.
In terms of mass and energy flows, it is pertinent to understand
the world as a large network of citiesdense concentrations of
people and infrastructuredrawing upon resources from global
hinterlands. Large quantities of energy are used for heating,
electricity use, transportation, industry, and construction in
cities, as well as for transportation of food, water, goods, and
other materials to cities (Kennedy et al. 2007, 2009, 2011).
Further energy is embodied in these material flows into cities (as
reflected in indirect greenhouse gas studies of cities, e.g., Larsen
and Hertwich [2009], Schulz [2010], or emergy analysis, e.g.,
Huang and Hsu [2003]). In recent years, industrial ecologists
have conducted many studies quantifying energy in the urban
metabolism, but here we aim to go further by investigating
theories and observations that more formally link the growth of
cities to their energy supply.
Perspectives on the growth of cities have been provided
by thinkers from the social sciences. To Jacobs (1969), cities
are centers of innovation and creativity, which grow by the
process of epigenesis as the internal work of companies replaces
imports. Glaeser (2011) focuses on knowledge as being the
primary determinant of urban growth, whereas Kennedy (2011)
describes how physical infrastructure also plays a fundamental
role in creating market power and demands. Based on selforganizing principles, Krugman (1996) shows that trade-offs
between agglomeration benefits and transportation costs can
drive the formation of population centers. Anthropologists such
as White (1959, 1969), Adams (1975), and Morris (2010) have
described how access to energy supplies was important to the
development of cities.
Our objective in this forum article is to piece together theoretical concepts, observations, and further analysis to initiate a
fuller understanding to the thermodynamic basis of the growth
of cities. In the next section, we review the general theory of
nonequilibrium thermodynamics from the works of Prigogine
and Schneider and Kay on dissipative structures, to developments of the principle of maximum entropy production (MEP).
This sets up the section on Observations and Analysis at Urban
and Global Scales, where we present empirical data and applied
analysis pertaining to energy and exergy flows in cities at local
and global scales. These observations, including both new and
previously published findings, enable us to demonstrate just how
the laws of thermodynamics play out in cities. The strength of
thermodynamics is its ability to describe aggregate properties on
212

Journal of Industrial Ecology

Entropy

Senvironment

Scity

Increasing
Disorder

Increasing
Complexity

Time

Figure 1 Entropy budget of a city and its environment. As a


dissipative structure, the city maintains relatively lower internal
entropy through the export of entropy to the environment
(adapted from Figure 6.4 in Lineweaver 2005).

a macro scale that emerge from complex micro-scale processes.


In the section on Underlying Processes, then we briefly review
parallel theories from economics and complexity theory that
describe underlying microscale processes. Key questions for further research are then given from each of the topics discussed.
In examining the growth of cities in thermodynamic terms, it
is important to recognize that two systems are under discussion
here. At the smaller scale, there is the city, which is an open
thermodynamic system, exchanging energy and matter with its
surroundings. Cities grow as complex structures, but do so with
relatively low internal entropy at the expense of increasing the
entropy of the surrounding environment (figure 1). At the larger
scale, there is the earth system, in which all cities are embedded.
The earth is a closed system in that it has negligible exchange
of matter with its surroundings, but it is not an isolated system,
because it has an exchange of energy with space. Both the earth
system and the city systems within it have access to energy
reservoirs and exist in nonequilibrium states.

Review of General Theory of


Nonequilibrium Thermodynamics
Prigogine and colleagues (1972) introduced the concept of
dissipative structures in attempting to describe the thermodynamics of living systems. Biological systems are highly complex
and ordered. Their growth and evolution cannot be explained
by the second law of thermodynamics in the context of an
isolated system, which tends toward an equilibrium state with
maximum disorder. Rather, living systems are nonequilibrium,
open systems that require exchanges of energy and matter with
their environment. In the context of a far from equilibrium system, the second law of thermodynamics allows for the creation
of ordered structures, with increasing disorder outside the system. Examples of these structures include both living organisms
and nonliving phenomena, such as the formation of thermal
convection cells and the coherence of light in a laser. Because
maintenance of such structures requires dissipation of energy

FORUM

(and destruction of exergy) they have been called dissipative


structures.
Priogine developed a version of the second law, which applies to both isolated and open systems. The variation of entropy, dS, over some time period, dt, is given by equation (1):
dS
dSI
dSE
=
+
dt
dt
dt

dSI = 0

(1)

where dSE denotes the flow of entropy from exchanges with the
surroundings and dSI is the entropy production resulting from
irreversible processes inside the system, for example, diffusion,
chemical reactions, and heat (Prigogine et al. 1972; Kleidon and
Lorenz 2005). For an isolated system, dSE is zero, but an open
system (or a closed, but not isolated, system) can be maintained
at a steady state (dS = 0) by a sufficient influx of negative
entropy (equation 2):
dSE = dSI 0

(2)

Thus, an ordered state, such as a living system, can be maintained and even grow under nonequilibrium conditions.
Prigogine noted how fluctuations play an important role in
the appearance of the ordered state in nonequilibrium systems,
as can be seen from experiments using Benard cells. This experimental apparatus consists of a working fluid between two
plates held at different temperatures. At small temperature differences, conduction predominantly transfers heat through the
fluid, though small fluctuations from this average motion can
emerge in the form of convection cells that are then quickly
damped. At a critical temperature difference, the fluctuations
are amplified and an ordered pattern of stable convective cells
form to dissipate the large temperature gradient.
Schneider and Kay (1994) performed an extended analysis
of the Benard cell to explore the dynamics of energy dissipation,
entropy production, and exergy destruction, which occur as the
fluid develops a new structure in response to the temperature
gradient. Their findings indicate that these rates increase with
temperature difference, initially at a nearly linearly rate in the
case of dissipation and faster than linearly in the case of entropy
production and exergy destruction. At the critical temperature
difference, there is a jump in all of these rates as the formation
of convective cells add an additional mode of heat dissipation (see, e.g.: www.youtube.com/watch?v=UhImCA5DsQ0).
Thus, these cells, and structures like them, are dubbed dissipative structures, their form disappearing if the gradient is
removed.
Based on their examination of the Benard cell, Schneider
and Kay (1994) have a different interpretation of dissipative
structures to that of Prigogine (see also Kay 2003). They suggest that Prigogines mathematical formulations are formally
limited to near equilibrium conditions. Moreover, they emphasize the importance of gradient destruction in understanding
dissipative structures. Examples of gradients include temperature gradients and differences in chemical potential. Differing
from Prigogine, they instead use the reformulated second law
expressed as follows:

The thermodynamic principle which governs the behaviour


of systems is that, as they are moved away from equilibrium,
they will utilize all avenues available to counter the applied
gradients. As the applied gradients increase so does the systems ability to oppose further movement from equilibrium.
(Schneider and Kay 1994, 6)

Schneider and Kay, moreover, avoid using entropy extremum principles to describe nonequilibrium thermodynamics.
In interpreting measurements from Benard cells, they observe
that the principle governing these systems is not one of maximum entropy production but rather one of entropy production
change being positive semi-definite as you increase the gradient (p. 11). They conclude that the term dissipative structure
means more than just increasing dissipation of energy and matter; it also entails destruction of gradients.
Although inspired by the work of Schrodinger (1944), much
of the study of nonequilibrium thermodynamics has occurred
over the past 40 years with particularly interesting results at
the global scale (Kleidon 2010; Nicolis and Prigogine 1977;
Prigogine 1978; Martyushev and Selesznev 2006). The earth
system has an exchange of energy with space, receiving incoming solar radiation, balanced by outgoing terrestrial long-wave
radiation. This radiative exchange maintains numerous earth
system processes in nonequilibrium statesthe composition of
the atmosphere, the hydrologic cycle, and the carbon cycle,
to name a few (Kleidon 2010). Some of these processes have
been studied using the proposed principal of MEP, which states
that thermodynamic processes in nonequilibrium systems assume steady states at which their rates of entropy production are
maximized (Sawada 1981; Ziegler 1983; Dewar 2003, 2005; Zupanovic et al. 2010; Kleidon 2010). MEP has been demonstrated
for a variety of phenomena, including atmospheric circulation,
ocean circulation, convection in the earths mantle, and decomposition of organic matter in ocean sediments, among others (Martyushev and Selezner 2006; Kleidon 2009). Whether
or not MEP holds for growing systems, not at steady state, such
as cities, is uncertain.

Observations and Analysis at Urban and


Global Scales
The notion that cities are dissipative structures, in the thermodynamic sense, is evident from studies of exergy loss and
anthropogenic heat rejection in cities. The number of studies
of exergy loss in cities is quite limited, and there are issues
of methodological consistency (discussed below). Nonetheless,
the results show the expected finding that cities are locations
of intensely concentrated degradation of anthropogenic exergy
gradients. This can been seen by contrasting the small Mexican
village of Cheranatzicurin against the cities of Vienna and Beijing (table 1). Cheranatzicurin, which is composed of a small
population of under 2,500 persons and their surrounding farmlands, has been the subject of two studies (Masera and Dutt
1991; Martnez-Negrete et al. 2013). Exergy destruction for the
community was 0.84 watts per square meter (W/m2 ) in 1986,
Bristow and Kennedy, Why Do Cities Grow?

213

FORUM

Table 1 The intensity of exergy destruction in Beijing, Vienna, and the Mexican village of Cherantzicurin
Location and year

Source

Exergy loss
(PJ)

Area
(km2 )

Intensity
(W/m2 )

Beijing (1996)
Beijing (2006)

Liu and colleagues (2012)

1,520
2,920

1,368

35.1
67.7

Cheranatzicurin (1986)
Cherantzicurin (2012)

Masera and Dutt (1991)


Martnez-Negrete and colleagues (2013)

0.0313
0.0351

37.1

0.84
0.95

Vienna (2007)

Grubler (2012)

163

414

12.5

Note: km2 = square kilometers; PJ = peta joules; W/m2 = watts per square meter.

Table 2 Regression analysis of changing global energy use and populationa

Intercept
 urban popn.
 rural popn.

Coefficients

Standard error

t stat

p value

Lower 95%

Upper 95%

0
0.002938
0.000971

#N/A
0.000717
0.001348

#N/A
4.0958
0.7205

#N/A
.004597
.494583

#N/A
0.001242
0.002216

#N/A
0.004635
0.004159

= 9; changes, over 5 years, for population in 000s, and for global energy use in millions of barrels of oil equivalent.
popn. = population; stat = statistic; #N/A = not applicable.

an

rising to 0.95 W/m2 in 2012. This is smaller than the exergy


destruction of 12.5 W/m2 for Vienna and 35.1 to 67.7 W/m2 for
Beijing that we have determined based on the exergy analysis in
Liu and colleagues (2012) and Grubler and colleagues (2012).
Generally speaking, we would expect larger, more densely
constructed cities to have greater intensities of anthropogenic
exergy destruction. This would be consistent with what Schneider and Kay (1994) observed for natural ecosystems, where the
exergy gradientalso measured in W/m2 increased from a
quarry and clear-cut land through to a mature forest of 400-yearold Douglas fir. There have been insufficient exergy studies to
show this for cities; however, quantification of anthropogenic
energy transformation in urban metabolism studies provides
some intuitive support. As shown in figure 2, the intensity of
direct energy use in cities increases with the urbanized population density. We know from urban metabolism studies that
energy use in cities is also dependent upon climate (Kennedy
et al. 2009), so it is also not surprising to see the variation about
the linear trend shown in figure 2. A relationship of similar
direction can be expected for the intensity of exergy destruction in cities. (To relate this to exergy destruction in natural
ecosystems, it is useful to work in units of energy or power per
m2 , but we now move to a different analysis using per capita
energy use.)
Further to the studies of exergy use over time in Beijing and
Cheranatzicurin, there have also been a small number of longtime series analyses of direct urban energy use. Kim and Barles
(2012) study per capita energy consumption in Paris from 1730
to 2000, showing a particularly rapid increase from 1950 to
1990. Similar findings are shown by Reiter and Marique (2012)
for Liege (Belgium), whereas Baynes and Bai (2012) assess the
doubling of energy consumption in Melbourne between 1973
and 2005.

214

Journal of Industrial Ecology

We can gain further insights by analyzing data on primary


energy consumption for the cities of Hong Kong and Singapore
(Bristow and Kennedy 2013). To describe the growth in the

rate of energy use, E(t),


relative to population, N(t), over time,
we use the model shown in equation (3):

E(t)
= E o N(t) E

(3)

where E o is a normalization constant. Note that although equation (3) is of the same form as that used by Bettencourt and
colleagues (2007), we are applying it to specific cities with
time-series data, whereas Bettencourt and colleagues applied it
to cross-sectional data of many cities; hence, parameter values
are not necessarily comparable. For the scaling parameter in
equation (3), we determined values of E = 2.24 (95% confidence interval [CI]: 1.02 to 3.46) for Hong Kong and E = 2.39
(95% CI: 1.42 to 3.36) for Singapore (for statistical analysis, see
the Supporting Information on the Web). The observation that
primary energy use per capita increases in Singapore and Hong
Kong as they grow is consistent with our thermodynamic understanding of the growth of cities: Energy is dissipated as heat at
increasingly higher rates as the cities grow, thereby destroying
exergy and increasing the rate of production of entropy, as per
Schneider and Kay.
The understanding that cities are dissipative structures suggests that there should be a strong relationship between the
growth of cities and total energy consumption at the global
scale. This is clearly the case as shown in figure 3. From 1965
to 2010, global urban population grew by 200%, from 1.184 to
3.558 billion, thereby surpassing global rural population. Over
the same time period, total annual global energy use increased
from 3,767 million tonnes oil equivalent (mtoe) to 12,002 mtoe,
a rise of 219%. Table 2 shows regression results for a simple linear model of change in total global energy use versus changes

FORUM

70.0
Shanghai

Energy intensity W/m2

60.0

50.0

40.0
Geneva
NYC
Tianjin
Beijing
London

30.0

Prague

20.0

Barcelona

Buenos Aires

Bangkok
Toronto-GTA
Paris-IDF
Denver
Chicago
Los Angeles
Amman

10.0

Rio
Sao Paulo
Dar es Salaam

0.0
-

Jakarta

Cape Town

5,000

10,000

15,000

Urbanized Populaon Density

20,000

25,000

persons/km2

Figure 2 Intensity of energy use in 22 global cities (based on data collected at the city scale from Kennedy et al. 2014). W/m2 = watts per
square meter; km2 = square kilometer.
Urban Populaon

Rural Populaon

Energy Use

4,000,000

14,000

3,500,000

3,000,000

Energy Use
(Million tonnes oil equivalent)

12,000

Populaon (000s)

10,000
2,500,000
8,000
2,000,000
6,000
1,500,000
4,000
1,000,000
2,000

500,000

1965

1970

1975

1980

1985

1990

1995

2000

2005

2010

Figure 3 Global energy use, urban and rural population, 19652010 (population data from UN Statistics; energy data from the
International Energy Agency).

in urban and rural populations, using the 5-year increments for


which population data are available from the United Nations.
Time-differenced data are used to avoid time-series correlation.
The model has an R2 of 0.90, but all of the explanatory power
lies with the urban population growth; the contribution of rural
population growth to changing global energy use is statistically
insignificant (see the Supporting Information on the Web for
further analysis). On average, an increase in urban population

of 1 person corresponds to an increase in annual energy use of


3 tonnes of oil equivalent.
We can infer, at the global scale, that cities grow in a manner that increases the destruction of exergy. In simple terms,
citiesand the infrastructure systems that connect themare
the physical products that result from the human exploitation of energy sources on the planet. But we can go further
here and specifically ask: Why is it that the dissipation of

Bristow and Kennedy, Why Do Cities Grow?

215

FORUM

0.5

ln(Ezipf)-ln(Euniform)

0.4
0.3
0.2
0.1
0
0

0.5

1.5

-0.1
-0.2
E
Figure 4 Difference in energy dissipation versus scaling factor
beta.

anthropogenically accessed energy on the planet occurs through


the formation of large, relatively dense human settlements?
Why, alternatively, does the extraction of energy not result in
a larger number of small dispersed settlements?
To get a first insight into this question, we can conduct a
numerical experiment based on equation (3). Consider a total
population of 1 million people distributed between 25 cities.
The distribution can either be uniform or follow Zipfs law
(i.e., a Pareto distribution) (Zipf 1946, 1949). Zipfs law gives
the frequency of an element of rank k from a population of N
elements to be (equation 4):
1/k s
f (k; s, N) =  N
s
n=1 1/n

(4)

where the exponent s is taken to be 1 (as per classic Zipfs


law). We then determine the total energy dissipation for the 25
cities using equation (3) in both cases. Our time-series analysis
of Hong Kong and Singapore suggests that beta is a constant
with city size, although this may not necessarily always be the
case. Figure 4 shows the difference in total energy dissipation
between the two population distributions over a range of E
values. The figure shows that, for E values greater than 1, the
structure of population that follows Zipfs law has the greatest
dissipation of energy into heat.
Our observation that the scaling parameter in equation (3)
is so much greater than 1 means that each doubling of population is associated with an increase in energy dissipation by
a factor of approximately 5 (using the value of E = 2.3; the
value may vary between cities). This means that city-size distributions that favor concentration of people in big cities (e.g.,
Zipf), as opposed to distributions that do not (e.g., uniform),
are preferential (provided E . > 1). In other words, based on
the energy scaling relationship, one big city and several small
cities together dissipate more energy than a set of medium-sized
cities would (for the same total population).
Turning to a related issue, the distribution of city sizes, according to Zipfs law that is observed under some circumstances,
216

Journal of Industrial Ecology

can also be understood in terms of nonequilibrium thermodynamics. Empirical analysis of population data for many countries, including the United States at various time periods, shows
a linear plot of log(rank) versus log(population) with a gradient of 1 (figure 5) (Zipf 1946, 1949; Gabaix 1999; Rosen and
Resnick 1980). The relationship only holds for cities in similar country contexts (i.e., for coherent samples); it might not
hold for global cities or cities in the European Union (EU), for
example (Cristelli et al. 2012). A variety of explanations have
been offered for Zipfs law, including a simple hypothesis that it
arises, from any starting conditions, if cities are subject to random growth processes independent of city size (Gabaix 1999).
Zipfs law has also been derived as a result of entropy maximization in the context of information entropy (Curry 1964;
Hernando amd Plastino 2012; Visser 2013). There is a relationship between maximization of thermal entropy and information
entropy (Dewar 2005; Tribus and McIntyre 1971), but we are
cautious about using them interchangeably (see the critique by
Hammond and Winnett [2009]). Nonetheless, if it is recognized
that random growth processes require correspondingly related
random quantities of energy input, then Gabaixs explanation
for Zipfs law can be seen in energetic thermodynamic terms
(i.e., using Clausius conception of entropy). Zipfs law, as with
other results in thermodynamics, describes macroscale behavior that emerges from possibly unknown, underlying microscale
processes.

Underlying Processes
In some applications, thermodynamics is a parallel theory,
which describes emergent macro phenomena based on probabilities and degrees of freedom in systems, without necessarily
representing the underlying processes. Hence, our purpose now
is to review theories or models of the underlying microscale processes, which may provide further understanding of the observations of city growth seen from a thermodynamic perspective.
In economic geography, Krugman (1993, 1996) presents a
general equilibrium model by which the growth and location
of cities emerge on aggregate from the economic behavior of
individual actors. Acting with monopolistic competition, manufacturing firms are assumed to clump together to lower costs
of interacting with suppliers and consumers, while balanced by
a desire to move away from competitors. Households derive
welfare from the consumption of agricultural and manufactured
goods, while tending to move toward locations with aboveaverage real wages. Starting from an initial random distribution
of firm locations, Krugman shows results that are consistent
with both central-place theory (Christallier 1933) and cumulative causation theory (Pred 1966).
There are possible parallels between Krugmans theory and
MEP in that the growth of cities is formulated through a balancing of driving and depleting gradients. Economies of scale
encourage cities to grow, expending more exergy, but agricultural and manufactured goods must also be transported between
cities (and rural locations), requiring an expenditure of energy.

FORUM

Figure 5 Rank-frequency distribution of communities in the United States in 1930 and 1940. The ideal line, A, has a slope of 1 (adapted
from Figure 1 of Zipf [1946]).

Other descriptions of economic processes underlying city


formation might also be given. For example, the generic process
by which capitalists search and invest in opportunities that
deliver the greatest risk-adjusted returns on investment may
be the most general expression of the underlying economic
process. Other economists stress the importance of knowledge
and innovation in the growth of cities. This has connections to
the physical, energetic basis of city growth, because knowledge
is required to develop and apply the technologies that exploit
the earths renewable and nonrenewable energy supplies. Some
have postulated that through face-to-face interactions and
agglomeration effects, cities are centers of knowledge formation
(Jacobs 1969; Glaeser 2011). Hence, there is potentially a
circular relationship at play: The formation of cities arises from
the exploitation of energy (direct and embodied); cities are
centers of knowledge creation; and some of the knowledge they
create leads to ways of living that encourage the generation of
new energy extraction methods.
Complexity theory, and the candidate law of maximization
of the adjacent possible (dubbed here MAP), potentially offers

an alternative underlying theory. MAP, put forth by Kauffman


(2000), states that open systems, on average, continuously strive
to maximize the diversity of what can happen next (the adjacent possible). Kauffman postulates that self-organization of the
MAP variety is as equally important as Darwinian evolution in
the rise of the terrestrial biosphere. His arguments could also
be extended to cities, given that they are open systems. For
example, the work of an urban planner resolving the various
interests of stakeholders to provide a bottom-up solution to development in a part of the city may be part of a self-reinforcing,
autocatalytic process under MAP.

Questions for Further Research


Before concluding, we can highlight three open research
questions that follow from the review of theory and empirical
observations discussed above:
1. Thermodynamic models of cities. Can a specific nonequilibrium thermodynamic theory of city formation be

Bristow and Kennedy, Why Do Cities Grow?

217

FORUM

developed? In this article, we have reviewed the general theory of nonequilibrium thermodynamics, but have
not developed models specific to cities or systems of cities.
To develop a theory specific to cities will require a nonsteady formulation as per equation (1), together with
boundary conditions that possibly may include socioeconomic context. This general equation of nonequilibrium
thermodynamics would be a starting point for mathematical modeling of the thermodynamics of growing cities, in
that it is a non-steady-state equation that explicitly links
growth to changes in entropy over time.
If such models can be developed, an intriguing question to
ask is whether cities grow in ways such that they maximize the
production of entropy. Applying Prigogines interpretation of
the second law of thermodynamics might indicate that they do,
though it may not necessarily follow under Schneider and Kays
interpretation. Dewar (2005) argues that Prigogines minimum
entropy principle is a special case of MEP, whereas Schneider and Kay avoid using entropy extremum principles, as we
discussed above.
Aoki (1995, 2008) has also shown that living systems at
first increase their entropy production and energy dissipation
rates as they grow, but the rates decrease with maturity. We
have previously studied a similar process for cities (Bristow and
Kennedy 2013), but only using energy data; it would be useful
to develop models of the changing entropy for cities as they
mature.
2. Further empirical study of exergy gradients for cities. Although many studies of energy use in cities have been
conducted, with some including embodied energy in
goods and materials, there are relatively few studies that
have gone the extra step of determining exergy gradients. Grubler and colleagues (2012) refer to exergy analyses of London, Geneva, and Malmo, but these have not
been published. Improved understanding of the thermodynamics of city growth requires more empirical studies,
especially with time-series data of changes of exergy gradients over time. In conducting further studies, it will also
be desirable to examine methodologies and assumptions.
It is not entirely clear whether methods used in previous studies have been consistent (e.g., with respect to
assumed second-law efficiencies). Moreover, there is an
intersection between anthropogenic energy supply and
natural solar radiation that occurs in cities. For example,
the energy performance of buildings depends on daily
and seasonally varying solar gains. How such effects are
taken into account in exergy analysis of cities remains a
challenge.
3. Intersection of thermodynamic and economic theories of urban
growth. Further work still remains in answering the challenge laid down by Georgescu-Roegen (1971) of understanding the thermodynamic limits of economic growth.
Insights have been made by ecological economists and
industrial ecologists focused on other scales or sectors,

218

Journal of Industrial Ecology

but the emergent phenomena proposed here for cities at


the global scale point to new and significant opportunities
for understanding the thermodynamic basis of economies.
The second law underlies the growth and distribution of
cities, which are the centers of wealth in economies. The
challenge is to understand what limits thermodynamics
puts on the urbanization process, which gives rise to the
generation of wealth. New efforts might look at parallel studies of the growth of cities both in economic and
thermodynamic termsand perhaps eventually couple
them together. Further exploration of Krugmans model
might be particularly useful given that it is grounded in
complex systems theory and can reproduce emergent phenomena similar to thermodynamics. Also important is to
understand the economic and energetic conditions under which Zipfs law hold, including its application to
polycentric cities and large metropolitan regions. If it has
a physical basis to it as we suspect, then perhaps Zipfs
law essentially holds for cities that have access to similar
energy sources and technologies, with similar economic
conditions.

Conclusion
Mumford (1961) defined the city to be a point of maximum
concentration for the power and culture of a community. He
no doubt meant power in the sense of control or influence, but
the definition also holds in the physical sense of energy per
unit time. Cities are literally centers of concentrated energy
conversion, as we have shown through the few available studies
of exergy loss and anthropogenic energy use in cities.
From a thermodynamic perspective, cities grow because they
are dissipative structures where exergy gradients are destroyed.
Each year, approximately 12,000 mtoe of energy (and growing)
are captured, transformed, or released from storage in the Earth
system by humans. This amount (16 terrawatts [TW]) is
smaller than the 65 TW of primary production by photosynthesis on land surfaces and far smaller than the 122,000 TW of
solar flux to the earth (Steger et al. 2005). Nonetheless, this is
the amount of useful energy that human societies use to serve
human needs, and most of it is dissipated into heat during the
construction, operation, and maintenance of cities (including
provision of materials, food, and other goods from rural areas).
Our main contribution here has been to make connections
between relatively disparate theories and observations that collectively begin to form an understanding of the growth of cities
in terms of nonequilibrium thermodynamics. The notion that
cities can be understood as thermodynamic systems, as dissipative structures, is neither new nor remarkable, for all systems
are subject to the laws of thermodynamics. We have shown
how observations and analysis of exergy and energy flows at
the city scale are consistent with the notion that cities are
dissipative structures. More significantly, we observe two aggregate phenomena relating to the growth and energy use in cities
that can be explained as physical manifestations of the laws of

FORUM

thermodynamics; these are a new relationship between global


energy use and urban population and, potentially, Zipfs previously established result on the distribution of city sizes. These
observations, demonstrating that the second law underlies the
growth and the distribution of cities, have potential implications for understanding the thermodynamic basis of economies.
The observation that cities are dissipative structures raises
difficult questions as to whether cities can help solve global sustainability challenges broadly related to energy use or whether
they are the root of the problem (Rees 2012). A possible interpretation of the strong relationship in figure 3 is that policies
aimed at promoting energy efficiency may be undermined by
urban growth (i.e., city-building processes cause an energy rebound effect). The salvation, perhaps, is that the processes at
play inherently involve human agency, and even if the growth
of cities inherently requires more energy, it might be obtained
from more-sustainable sources.

Acknowledgments
This research was supported by the Natural Sciences and
Engineering Research Council of Canada. The authors are
grateful to comments on an early version of the manuscript
from Brad Bass, Axel Kleidon, and Michael Batty. The authors
appreciate the advice on the statistical analysis from Dr. Laurel
Duquette, University of Toronto, Department of Statistics.

References
Adams, R. N. 1975. Energy and structure: A theory of social power. Austin,
TX, USA: University of Texas Press.
Aoki, I. 2008. Entropy law in aquatic communities and the general
entropy principle for the development of living systems. Ecological
Modeling 215: 8292
Aoki, I. 1995. Entropy production in living systemsFrom organisms
to ecosystems. Thermochimica Acta 250(2): 359370.
Ayres, R. U. and B. Warr. 2005. Accounting for growth: The role of
physical work. Structural Change and Economic Dynamics 16(2):
181209.
Ayres, R. U. and V. Voudouris. 2014. The economic growth enigma:
Capital, labour and useful energy? Energy Policy 64: 1628.
Batty, M. 2012. Building a science of cities. Cities 29: S9S16.
Batty, M. 2013. The new science of cities. Cambridge, MA, USA: MIT
Press.
Baynes, T. and X. Bai. 2012. Reconstructing the energy history of a
city: Melbournes population, urban development, energy supply
and use from 1973 to 2005. Journal of Industrial Ecology 16(6):
862874.
Baynes, T. M. and T. Wiedmann. 2012. General approaches for assessing urban environmental sustainability. Current Opinions in
Environmental Sustainability 4(4): 458464.
Bettencourt, L. M. A., J. Lobo, D. Helbing, C. Kuhnert, and G. B. West
Growth. 2007. Innovation, scaling, and the pace of life in cities.
Proceedings of the National Academy of Sciences of the United States
of America 104(17): 73017306.
Bettencourt, L. M. and G. B. West. 2010. A unified theory of urban
living. Nature 467(7318): 912913.

Bettencourt, L. M. 2013. The origins of scaling in cities. Science


340(6139): 14381441.
Bristow, D. and C. Kennedy. 2013. The energy for growing and maintaining cities. Ambio 42(1): 4151.
Christaller, W. 1933. Central places in Southern Germany. Englewood
Cliffs, NJ, USA: Prentice-Hall.
Cleveland, C. and M. Ruth. 1997. When, where and by how much
does thermodynamics constrain economic processes? A survey
of Nicholas Georgescu-Roegens contribution to ecological economics. Ecological Economics 22(3): 203223.
Cristelli, M., M. Batty, and L. Luciano Pietronero. 2012. There is more
than a power law in Zipf. Scientific Reports 2: 812.
Curry, L. 1964. The random spatial economy: An exploration in settlement theory. Annals of the Association of American Geographers
54(1): 138146.
Dewar, R. C. 2003. Information theory explanation of the fluctuation
theorem, maximum entropy production, and self-organized criticality in non-equilibrium stationary states. Journal of Physics A
36: 631641.
Dewar, R. C. 2005. Maximum entropy production and non-equilibrium
statistical mechanics. In Non-equilibrium thermodynamics and the
production of entropy: Life, earth, and beyond, edited byA. Kleidon
and R. D. Lorenz. Heidelberg, Germany: Springer-Verlag.
Faber, M. and J. L. R. Proops. 1985. Interdisciplinary research between
economists and physical scientists: Retrospect and prospect. In
KYKLOS 38 (4th quarter) (pp. 599616); reprinted in: A survey
of ecological economics, edited by R. Krishnan, J. M. Harris, and N.
R. Goodwin. Montague, Prince Edward Island, Canada: Island,
1995.
Gabaix, X. 1999. Zipf s law for cities: An explanation. Quarterly Journal
of Economics 114(3): 739767.
Georgescu-Roegen, N. 1971. The entropy law and the economic process.
Cambridge, MA, USA: Harvard University Press.
Georgescu-Roegen, N. 1979. Energy analysis and economic valuation.
Southern Economic Journal 45(4): 10231058.
Glaeser, E. L., H. D. J. Kallal , A.Scheinkman, and A. Shleifer. 1992.
Growth in cities. Journal of Political Economy (centennial issue)
100(6): 11261152.
Glaeser, E. 2011. Triumph of the city: How our greatest invention makes us
richer, smarter, greener, healthier, and happier. New York: Penguin.
Grimm, N., H. Stanley, N. Golubiewski, C. Redman, J. Wu, X. Bai, and
J. Briggs. 2008. Global change and the ecology of cities. Science
319(5864): 756780.
Grubler, A., X. Bai, T. Buettner, S. D. J. Dhakal Fisk, T. J. E. Ichinose
Keirstead, G. Sammer, D. N. B. Satterthwaite Schulz, N. Shah, J.
Steinberger, and H. Weisz. 2012. Urban energy systems. Chapter
18 in Global energy assessmentToward a sustainable future (pp.
13071400), Cambridge University Press (Cambridge, UK, and
New York) and the International Institute for Applied Systems
Analysis (Laxenburg, Austria).
Hammond, G. P., and A. B. Winnett. 2009. The influence of thermodynamic ideas on ecological economics: An interdisciplinary
critique. Sustainability 1(4): 11951225.
Hernando, A. and A. Plastino. 2012. Thermodynamics of urban population flows. Physical Review E 86: 066105.
Hoornweg, D., L. Sugar, and C. L. Trejos-Gomez. 2011. Cities and
greenhouse gas emissions: Moving forward. Environment and Urbanization 23(2): 121.
Huang, S-L. and W.-L. Hsu 2003. Materials flow analysis and emergy
evaluation of Taipeis urban construction. Landscape and Urban
Planning 63(2): 6174.

Bristow and Kennedy, Why Do Cities Grow?

219

FORUM

Jacobs, J. 1969. The economy of cities. New York: Random House.


Kauffamn, S. 2000. Investigations. New York: Oxford University Press.
Kay, J. 2003. About some common slipups in applying Prigogines minimum entropy production principle to living system. Unpublished
manuscript. www.jameskay.ca/musings/mep.pdf. Accessed 1 July
2014.
Kennedy, C. A. 2011. The evolution of great world cities: Urban wealth
and economic growth. Toronto, Ontario, Canada: University of
Toronto Press.
Kennedy, C., J. Steinberger, B. Gasson, T. Hillman, M. Havranek,
Y. Hansen, D. Pataki, A. Phdungsilp, A. Ramaswami, and
G. Villalba Mendez. 2009. Greenhouse gas emissions from
global cities. Environmental Science & Technology 43(19): 7297
7302.
Kennedy, C. A., J. Cuddihy, and J. Engel-Yan. 2007. The changing
metabolism of cities. Journal of Industrial Ecology 11(2): 4359.
Kennedy, C. A., S. Pincetl, and P. Bunje. 2011. The study of urban
metabolism and its applications to urban planning and design.
Journal of Environmental Pollution 159(89): 19651973.
Kennedy, C. A., N. Ibrahim, and D. Hoornweg. 2014. Low carbon
infrastructure strategies for cities. Nature Climate Change 3: 343
346.
Kim, E. and S. Barles. 2012. The energy consumption of Paris and its
supply areas from the eighteenth century to the present. Regional
Environmental Change 12(2): 295311.
Kleidon, A. 2010. A basic introduction to the thermodynamics of
the Earth system far from equilibrium and maximum entropy
production. Philosophical Transactions of the Royal Society of London
365(1545): 13031315.
Kleidon, A. 2009. Nonequilibrium thermodynamics and maximum entropy production in the Earth system: Applications and implications. Naturwissenschaften 96: 653677.
Kleidon, A. and R. D. Lorenz. 2005. Entropy production by earth system
processes. In Non-equilibrium thermodynamics and the production of
entropy: Life, earth, and beyond, edited by A. Kleidon and R. D.
Lorenz. Heidelberg, Germany: Springer-Verlag.
Krugman, P. 1996. The self-organizing economy. Cambridge, MA,
USA:Blackwell.
Krugman, P. 1993. On the number and location of cities. European
Economic Review 37(23): 293298.
Larsen, H. N. and E. G. Hertwich. 2009. The case for consumptionbased accounting of greenhouse gas emissions to promote local climate action. Environmental Science & Policy 12(7): 791
798.
Lineweaver, C. H. 2005. Cosmological and biological reproducibility:
Limits on the maximum entropy production principle. In Nonequilibrium thermodynamics and the production of entropy: Life, earth,
and beyond, edited by A. Kleidon and R. D. Lorenz. Heidelberg,
Germany: Springer-Verlag.
Liu, G. Y., Z. F. Yang, M. R. Su, and B. Chen. 2012. The structure,
evolution and sustainability of urban socio-economic system. Ecological Informatics 10(2): 29.
Martnez-Negrete, M., R. Martnez, R. Joaqun, C. Sheinbaum, and O.
R. Masera. 2013. Is modernization making villages more energy
efficient? A long-term comparative end-use analysis for Cheranatzicurin village, Mexico. Energy for Sustainable Development
17(5): 463470
Martyushev, L. M. and V. D. Seleznev. 2006. Maximum entropy production principle in physics, chemistry and biology. Physics Reports 426: 145.

220

Journal of Industrial Ecology

Masera, O. R. and G. Dutt. 1991. A thermodynamic analysis of energy


needs: A case study in a Mexican village. Energy 16(4): 763769.
Morris, I. 2010. Why the West rulesFor now: The patterns of history,
and what they reveal about the future. New York: Farrar, Strauss and
Giroux.
Mumford, L. 1961. The city in history. San Diego, CA, USA: Harcourt.
Nicolis, G. and I. Prigogine. 1977. Self-organization in nonequilibrium
systems: From dissipative structures to order through fluctuations. New
York: John Wiley & Sons.
Pred, A. 1966. The spatial dynamics of U.S. urban-industrial growth.
Cambridge, MA, USA: MIT Press.
Prigogine, I. 1978. Time, structure and fluctuations. Science 201(4358):
777785.
Prigogine, I., G. Nicolis, and A. Babloyantz. 1972. Thermodynamics
of evolution. Physics Today 25(11): 2328.
Proops, J. L. R. 1985. Thermodynamics and economics: From analogy to
physical functioning. In Energy and time in economics and physical
sciences, edited by W. van Gool and J. Bruggink. Amsterdam:
Elsevier.
Rees, W. E. 2012. Cities as dissipative structures: Global change and
the vulnerability of urban civilization. In Sustainability science:
The emerging paradigm and the urban environment (pp. 247
275), edited byM. P. Weinstein and R. E. Turner. New York:
Springer.
Reiter, S. and A.-F. Marique 2012. Toward low energy cities. A case
study of the urban area of Liege, Belgium. Journal of Industrial
Ecology 16(6): 829838.
Rpke, I. 2004. The early history of modern ecological economics.
Ecological Economics 50(34): 293314.
Rosen, K. and M. Resnick. 1980. The size distribution of cities: An
examination of the Pareto law and primacy. Journal of Urban
Economics 8(2): 165186.
Ruth, M. 1993. Integrating economics, ecology and thermodynamics. Dordrecht, the Netherlands: Kluwer Academic.
Sawada, Y. 1981. A thermodynamic variational principle in nonlinear
non-equilibrium phenomena. Progress of Theoretical Physics 66(1):
6876.
Schneider, E. D. and J. J. Kay. 1994. Life as a manifestation of the second law of thermodynamics. Mathematical and Computer Modelling
19(6): 2548.
Schrodinger, E. 1944. What is life? Cambridge, UK: Cambridge University Press
Schulz, N. B. 2010. Delving into the carbon footprints of Singapore
Comparing direct and indirect greenhouse gas emissions of a
small and open economic system. Energy Policy 38(9): 4848
4855.
Steger, U., W. Achterberg, K. Blok, H. Bode, W. Frenz, C. Gather, G.
Hanekamp, et al. 2005. Sustainable development and innovation in
the energy sector. Berlin: Springer.
Tribus, M. and E. C. McIntyre. 1971. Energy and information. Scientific
American 225(3): 179188.
Visser, M. 2013. Zipfs law, power laws, and maximum entropy. New
Journal of Physics 15 (2013) 043021.
White, L. A. 1959. The evolution of culture: The development of civilization
to the fall of Rome. New York: McGraw-Hill.
White, L. A. 1969. The science of culture: A study of man and civilization.
New York: Farrar, Strauss and Giroux.
Ziegler, H. J. 1983. Chemical reactions and the principal of maximum
rate of entropy production. Journal of Applied Mathematics and
Physics (ZAMP) 34: 832844.

FORUM

Zipf, G. K. 1949. Human behavior and the principle of least effort. Cambridge, MA, USA: Addison-Wesley.
Zipf, G. K. 1946. On the intercity movement of persons. American
Sociological Review 11(6): 677686.
Zupanovic, P., S. Botric, D. Juretic, and D. Kuic. 2010. Relaxation processes and the maximum entropy production principle. Entropy
12(3): 473479.

About the Authors


David Bristow is a postdoctoral fellow at the Center
for Resilience of Critical Infrastructure at the University of
Toronto, Toronto, Ontario, Canada. Christopher Kennedy
is a professor of civil engineering at the University of
Toronto.

Supporting Information
Additional Supporting Information may be found in the online version of this article at the publishers web site:
Supporting Information S1: This supporting information provides details about the univariate analysis of relationships
between population and global energy use and details about the growth of energy use in Hong Kong and Singapore.

Bristow and Kennedy, Why Do Cities Grow?

221

Das könnte Ihnen auch gefallen