Urban Metabolism and other Topics

Vaclav Smil's book Growth is wide-ranging enough that this isn't a comprehensive review. Rather, I decided to mainly explore just a few topics from it.

Introduction

Growth was one of my five 500+ page books from last year (Lord of the Rings and Journey to the West have already made it into some posts, and I still have some pending writing ideas related to City of God and Champlain's Dream). It also fits right into the theme of industry, technology, and adjacent topics that I've tried to focus some reading and writing in 2024 around. This is the second book by Vaclav Smil that I've read; one chapter had quite a bit of overlap with Harvesting the Biosphere so I won't spend much time on that part of the content in this post.

Growth is about how things {organisms, cities, technologies} get bigger {larger in size, more powerful, more numerous} over time. It features numerous graphs, the majority of which show S-shaped curves. A recurring theme is that a lot of phenomena which appear exponential at some stage in their lifecycle eventually turn out to be logistic.

This book ranges widely, with the main unifying theme being the dynamics of growth. It provides a good high-level overview of the history of certain technological artifacts—especially those related to energy transformations (engines, motors, turbines, etc.) as they are a prime interest of Smil's. It also has case studies of the population history of selected cities and countries. And lots of discussion of growth and post-growth stagnation or decline applied to various domains. For this post, I'll look at what Growth says about certain technological artifacts (with a short section on windmills broken out to tie in to some previous books I've discussed on this blog), what it says about cities, and what it says about industry. I'll end with a section of miscellaneous notes, because part of the value of reading a book by Vaclav Smil is the wealth of information it provides that can be very useful in back-of-the-envelope estimates on various questions.

Windmills

Windmills are one of the many power sources that Smil considers in Growth. He traces their, well, growth, from medieval times to the present day. Over this timespan, they've increased in the amount of power they can provide by a factor of ~1000x, with much of that coming in the past half-century:

The useful power of medieval windmills was just 2-6 kW, comparable to that of early waterwheels. Dutch and English mills of the 17th and 18th centuries delivered usually no more than 6-10 kW, American mills of the late 19th century rated typically no more than 1 kW, while the largest contemporary European machines delivered 8-12 kW, a fraction of the power of the best waterwheels.

Their ratings rose from 55 kW in 1981 to 500 kW a decade later, to 2 MW by the year 2000, and by 2017 the largest capacity of Vestas onshore units reached 4.2 MW.

Similar stories are told in Growth for many other power sources, from water power to jet engines. I chose to highlight windmills because it gave a chance to refer back to (including re-reading key chapters) a couple of other books I've reviewed on this blog: Water on Sand and Designed for Dry Feet.

In Designed for Dry Feet by Dr. Robert Hoeksema, it was the third chapter that was most directly about windmills so that's the part I re-read. The north of Amsterdam in the 16th and 17th century was the main setting for this chapter. When I was in the Netherlands, I enjoyed biking through part of that area and seeing some of the preserved windmills at Zaanse Schans. Windmills were used for powering proto-industry such as grinding grain and also for land reclamation via the draining of lakes; experience with the former helped with the latter. To accomplish this land reclamation, dikes and ring canals were built around lakes to be drained and windmills were used to pump water from the interior out to the ring canal. Windmills are thus only one piece of an overall system which is useless without all the other parts. This lake draining "was possible due to the prosperity of the seventeenth century and was necessary due to the growing demand for food"—it required raising a lot of capital. Dr. Hoeksema discusses improvements that were made to windmills over this era. There were two major types: post mills (in which the whole structure can rotate to face into the wind) and tower mills (in which only the top cap rotates). To catch variable amounts of wind, windmills could stretch or retract sailcloth over a lattice frame. Another important technical feature for tower mills was having the central power shaft coupled with cogs instead of being a continuous solid member. According to this chapter, an average windmill in this period could lift 0.9 cubic metres of water per second by 1 metre; they had a typical sail span of 25-27 metres and typical wind speeds were 6-11 m/s. This is sufficient information to calculate the power and efficiency. The output works out to 8.8 kW ( = 0.9 m³/s x 1000 kg/m³ x 9.81 m/s² x 1 m) which is in the range that Smil states for Dutch windmills in the 17th century. The total wind energy (taking the midpoint for sail span and wind speed) is 195 kW ( = 1/2 x 1.2 kg/m³ x 530 m² x (8.5 m/s)³), giving an efficiency of 4.5%. In a case study on the draining of the Beemster polder, we learn that the contract included a performance test to make sure enough windmills were installed to do the job. This shows how legal/institutional innovations need to go hand-in-glove with technological developments. Also of note is that later on, 2 diesel pumps replaced 50 windmills at the job of keeping Beemster polder dry.

Water on Sand has multiple authors but it was the second chapter, "History and Animal Energy in the Arid Zone" by Richard Bulliet that I wanted to revisit for this post. He contrasts Europe, where population growth led to expensive feed (because animal feed was competing with human food) and draft animals becoming a luxury with the MENA region, where there was lots of non-arable grazing land and thus plentiful/cheap animal power (because it didn't compete with human food). His thesis in this chapter is that animal power is cheapest where it doesn't compete with human food and such places have less incentive for developing non-muscle power. In contrast, places like late medieval or early modern Europe where draft animals are expensive to keep recognize a need for other sources of power, leading to investment and innovation in wind- and water-mills. The MENA region had mill technology earlier or at the same time as Europe—some of it as a legacy from the Roman Empire—but it took off in Europe in the 12th century and not there. There's a debate across various books I've read this year (stay tuned for the next entry in my Industry series that I hope to get posted before the end of the year) about whether technological development drives social change or social conditions instead create an opening for technological development? Bulliet is clearly taking a position on one side of this debate. The contingent (as Bulliet sees it) development of a tech-savvy and entrepreneurial miller class in Europe would be influential in later technological developments like the Industrial Revolution. I actually see this as an example of technology and society influencing each other in turn, though: environmental/economic features gave an impetus to wind and water power; the prevailing power technology led to the rise of a miller class; the presence of millers contributed to industrialization. Of course, this is only one proposed explanation for what's known as the Great Divergence. Other factors I've seen suggested include: cultural/religious factors promoting or stifling science and innovation; specific institutions and their impacts (e.g. Cistercian monasteries exchanging best practices in agriculture, Waqfs tying up capital); arms races between small and nearby kingdoms (i.e. numerous peer competitors) making technological advantages very salient; easily accessible coal deposits.

Windmills were just one small topic in Growth, and like most of the items that book discusses, they're dealt with in a very zoomed-out fashion, showing improvements over centuries. But I hope this section shows that such an overview can be a good starting point to explore further, whether on technical details or social impacts and interactions.

Other Artifacts

In this section, I'll briefly cover some of what Growth has to say about a few technologies related to power, transportation, and audio-visual communication/media. The book is replete with statistics and graphs on many more technological artifacts (not used in the archaeological sense) than I can cover here (e.g. air travel, the internet), but at least this section of this review can provide a sample. I've chosen a handful of graphs from the book that I found interesting and also representative.

Although individual technologies see their growth rate taper off, a relay of different technologies has resulted in a more-or-less steady increase in the capacity of the biggest engine or turbine at any time. Of course, if more power is needed, using more units rather than simply bigger units is also an option.

The next sets of graphs show adoption rates of various appliances and communication/entertainment devices. The rates differ, with things like radio and cellphones spreading much more quickly than stoves or dishwashers, but most of these show the same general trend of an S-curve towards saturation.

Near the end of Growth, Smil also considers decline. This is seen in the following graphs of successions of different technologies for steelmaking (from Bessemer to Basic Open Hearth to Basic Oxygen Furnace to Electric Arc Furnace) or music-playing.

I'm interested in the ocean and shipping, so I enjoyed the parts of Growth that dealt with ships as artifacts. Smil presents long term trends in both size and speed. From a description in The Odyssey, he estimates Odysseus' ship would have weighed up to 12 tonnes (and total displacement would be around 40-50% above that). In the Classical era, Greek triremes would have displaced up to 50 tonnes, and the Romans' typical cargo ships had displacements up to 200 tonnes with capacities of around 70 tonnes (although apparently they built some giant ships with capacities of 1,000 tonnes for special purposes). Viking ships were similar in size to Odysseus' from the Bronze Age, while those of Columbus and Magellan from the age of exploration were around 120 - 150 tonnes in terms of displacement. By the 19th century, however, a ship of the line might weigh over 3,000 tonnes (with a displacement correspondingly higher). Smil puts that in context as requiring 6,000 oak trees to build. When it comes to sailing speeds, these increased from around 7 km/h in the Renaissance (scarcely improved from Classical antiquity) to 15 km/h for the fastest clippers to 60 km/h for aircraft carriers. However,

The typical speed of large container ships (30-40 km/h) is not radically higher than the typical speed of the 19th-century clippers; of course, their cargo capacities are orders of magnitude apart, but there has been no hyperbolic growth of speeds in ocean shipping, and there is no realistic prospect that this fundamental mode of transport that enabled modern economic globalization will enter a new era of radically increased speeds.

My understanding is that trying to push vessel speed runs into Froude number constraints.

When it comes to electrification of transportation, he makes this observation:

Since 1900, the maximum battery energy densities rose from 25 Wh/kg for lead-acid units to about 300 Wh/kg for the best lithium-ion designs in 2018, a 12-fold gain that fits a logistic curve predicting about 500 Wh/kg by 2050. We must hope that new discoveries will vault us onto a new logistic trajectory as even 500 Wh/kg is not enough for battery-powered machines to displace all liquid derived from crude oil: the diesel fuel used to power heavy machines, trains, and ships has energy density of 13,750 Wh/kg. 

I'd consider the most defining artifacts of the present to be cars and cellphones, and Smil has plenty to say about both of those. Apparently, enough mobile phones are now made for every person on earth to get a new one every 4 years! And 1.5 billion phones are thrown out annually (they have a short life span so disposal trends don't lag production trends by much). Smil makes an attention-getting comparison between the energy footprint of manufacturing cars versus phones. He notes that a car has around 10,000x the mass of a smartphone. The embodied energy (of all the materials and processing) is estimated at 100 GJ for the former and 0.25 GJ for the latter, though: only a 400x difference. Furthermore, way more phones are sold (almost 2 billion per year, versus 72 million vehicles) and they are replaced more often. So Smil estimates that the energy footprint of manufacturing portable electronic devices (he adds in laptops and tablets to make the comparison work) approaches the order of magnitude of manufacturing vehicles (although using them is another matter).

The book is 500+ pages of this kind of thing, so this section of the review should give you a good idea of whether it would be one you'd like to read.

Urban Metabolism

The growth of any organism or of any artifact is, in fundamental physical terms, a transformation of mass made possible by conversion of energy

 Growth spends a chapter on the growth of biological organisms (an interesting thing there was scaling rules for various metrics proportional to [body mass]^0.75). It also applies this notion of metabolism—explicitly as such—to cities:

But already during antiquity some cities engaged in long-distance trade and today the energy and material metabolism of large modern cities rests on truly global foundations, with fuels, foods, raw materials, and finished products traded worldwide. In turn, cities have been repaying this growing energy and material dependence by providing first manufacturing and now, most importantly, commercial, financial, managerial, educational, and recreational services that generate significant income and extend their reach and appeal on levels ranging from regional and nationwide to continental and global. Cities have also had a disproportionate influence on creating culture, setting commercial priorities, influencing political development, supplying an educated labor force, determining tastes and fashions, and fostering innovation.

This excerpt gives a function-focused perspective on cities, considering their inputs and outputs. As a process engineer, this kind of perspective appeals to me. The outputs, especially recently, are not only material goods but also services and intangibles like culture and governance.

As it does with artifacts, Growth takes a long historical view towards cities:

The subsequent history of civilization can be seen as a quest for ever higher reliance on extrasomatic energies. The process began with the combustion of phytomass (chemical energy in wood, later also converted to charcoal, and in crop residues) to produce heat (thermal energy) and with small-scale conversions of water and wind flows into kinetic energy of mills and sails. After centuries of slow advances, these conversions became more common, more efficient, and available in more concentrated forms (larger unit capacities), but only the combustion of fossil fuels opened the way to modern high-energy societies.

Smil estimates that hunter-gatherer societies had population densities up to 0.002 ppl/ha (1 person per 5 sq. km) or as low as 0.0001 ppl/ha. This rises to 1 ppl/ha for early agriculture (e.g. Ancient Egypt and Mesopotamia) and 5 ppl/ha in the most intensive farming locales (mainly in Asia) prior to the mechanization of agriculture and the Green Revolution. Modern agriculture can support 10 ppl/ha (1000 people per sq. km).

The Roman empire has some of the best data from the ancient world on the metabolism of a city and the infrastructure needed to support it. Growth contains statistics that during the 2nd century A.D. the empire needed 12 megatonnes of grain per year (for a population of 60 million people), much of which was grown in Egypt, and that the city of Rome had 1 billion litres of water per day supplied via aqueducts. Aside from the aqueducts, the infrastructure that the Roman empire is well known for was their roads:

By the beginning of the 4th century, their total length reached about 85,000 km and I have calculated that the task required as much as 1.2 billion labor days, or an annual equivalent of some 20,000 full-time workers during 600 years of expansion and maintenance.

Smil also provides estimates for the energy use in the Roman empire, which increased only slowly over the subsequent millennium:

My calculations indicate that annual use of all forms of energy averaged at least 18 GJ/capita during the height of [Roman] imperial rule—while Galloway et al. (1996) estimated London's average fuel consumption at about 25 GJ/capita in 1300.

Whether millennia ago or today, concentrating people in cities has a cost and requires on-going sustenance to maintain:

Cities are our civilization's most complex and most intensive dissipative (entropy-producing) structures.  Unprecedented agglomerations of people and economic activities must be supplied incessantly with enormous, and rising, amounts of energy in order to satisfy their high power-density demand, food to sustain record-size concentrations of humanity, and raw materials to build, maintain and renew many required infrastructures, ranging from water supply and wastewater treatment to roads and garbage collection.

I would also note that cities throughout history have often needed to be supplied with people as well, as they have tended to be population sinks (whether due to disease or reduced fertility). I shared some excerpts from Growth on the related issue of aging populations in an earlier post this year. Even in developing countries cities apparently also have higher per capita use of water and energy compared to small towns or rural areas.

However, cities have some compelling advantages to go with their high metabolic needs. Regarding the benefits of spatial concentration Smil refers to Marshall's law, von Thünen's law, and Tobler's law:

1. Marshall's: "Growing cities enjoy disproportionate increases in productivity, income, and accumulated wealth that can be attributed to the phenomenon of agglomeration whose importance—in terms of transport cost savings, customer-supplier interactions, labor pooling, and exchange of knowledge allowing companies, regardless of their size, to take advantage of economies of scale and scope" (Silicon Valley, Wall Street, and Hollywood are offered as examples)
2. von Thünen's: "But the principle of a nonlinear decline of urban attributes with distance or transportation cost from a city's core or central business district remains readily discernible as far as declining population densities and average rents are concerned." (see here for more on this topic)
3. Tobler's: "everything is related to everything else, but near things are more related than distant things"

For reasons such as these, some modern cities have grown to unprecedented sizes, with the energy footprints and other metabolic needs to go with them:

And high concentrations on the order of 50,000 people/km² are commonly found in the most densely populated parts of Asian megacities… Assuming an age- and sex-weighted conservative mass mean of 45 kg/capita, densities of more than 50,000 people/km² correspond to a live weight human biomass (anthropomass) of more than 2 kg/m², a rate that is unmatched by any other social mammal and that is three orders of magnitude higher than the peak seasonal biomass of large herbivorous ungulates (antelopes, wildebeest, giraffes, zebras) grazing on East Africa's richest grasslands.

Densities like these far outpace energy—and remember that food is a form of energy—and material that can be sourced locally:

The energy demand of modern cities requires unprecedented amounts of fuels and electricity delivered with the very high reliability needed to cover not only household and industrial demands but also to power vital infrastructures that operate incessantly (electricity, supply, sewage pumping) or are needed for more than 18 hours a day (subways and commuter trains typically shut down only for a few early morning hours). What sets modern cities apart are the scale and highly concentrated nature of their energy demand. Power density consumption (energy flow per year per unit of area, usually expressed as W/m²) of overall energy in modern buildings (most of it being electricity and natural gas) ranges from less than 10 W/m² for energy-efficient detached single-story houses in mild climates, to post-1990 (that is better-built) American two-story houses averaging 30-40 W/m² of their foundation area, while older (pre-1980) 20-story office buildings in temperate climates will need 800 W/m² (Smil 2015c). A city-wide mapping of New York showed that some mid-Manhattan blocks with high numbers of high-rises have power densities up to 900 W/m², while 400-500 W/m² is common in the financial district, Greenwich Village, Midtown South, and East Side (Howard et al. 2012).

For comparison, even in Dubai the annual mean of incoming solar radiation is about 215 W/m². The only way to meet such a highly concentrated energy demand in a reliable fashion is to electrify everything. 

So when cities go much above 200 W/m² (200 MW per sq. km) for their total energy needs or above 1000 people per sq. km for a population density, they will need to bring in energy and food from beyond their boundaries. This obviously requires infrastructure. Smil points out that pipelines (a 1 m diameter pipeline can transport 50 megatonnes of oil annually) and high-voltage transmission lines are reliable ways to enable energy to be used far from its source. The challenge is that infrastructure requires maintenance and maintenance is easy to defer or neglect:

Societies have usually had much better success in building the essential infrastructures than maintaining them, and regular American assessments clearly demonstrate that chronic shortcoming.

I thought it would be worthwhile to engage more deeply on this topic of urban metabolism so I decided to do some ballpark-level analyses on the city of Winnipeg. I've never been there (yet) but it seemed a fitting choice as it is where Professor Smil lives and teaches. Also, cities on the prairies have geometry that simplifies this kind of exercise.

According to per Wikipedia, Winnipeg has a population of 750,000 and an area of 462 sq. km for a population density of 1430/km² (for the metro area, these rise to 5285 sq. km and 158/km² but I'll consider the city itself in this analysis); it is surrounded by a beltway (not the exact city boundary, but a decent proxy for it) with an approximate 90 km perimeter. I'll use these values along with the 10 ppl/ha that modern agriculture can support and a few other statistics listed below for some Smil-style calculations about Winnipeg and its "metabolism":

• Per capita total energy consumption in Canada is 7.4 tonnes of oil equivalent, or 310 GJ annually (17x ancient Rome)
• Per capita (but this includes non-domestic uses) drinking water volumes in Manitoba are 280 L/d (less than the national average of 400 L/d). We can assume roughly equal volumes of wastewater.
• Solid waste generation rates in Canada are 2.33 kg/d. This puts a hard floor on the amount of physical material brought in, because: there are other waste streams (e.g. recycling, composting), there are durable goods, there are raw materials and intermediate products going to businesses in Winnipeg (many of whom send products outside of the city), and Winnipeg is a logistics hub for a pretty large radius. So it wouldn't surprise me if the total amount of physical goods moved in and out of the city on a daily basis was an order of magnitude or more above this value.

Using these values, here are some interesting results and comparisons I came up with:

• The total estimated annual energy usage for Winnipeg is 232.5 million GJ. This is 7370 MW on a continuous basis (but this is all energy, not just electric) or 16 MW/km² normalized by area. Comparing power densities cited by Smil and land uses in Winnipeg, this seems reasonable. It is low enough that widespread rooftop solar could make a big dent in it. Using tonnes of oil equivalent units, this energy usage is 5.55 megatonnes per year (which a 1 m diameter pipeline could apparently supply in ~6 weeks). Of course, in reality, this energy demand is met by a mix of liquid and gaseous fuels transported in various ways, electricity, and (based on other parts of Canada I'm more familiar with) probably at least some biomass (i.e. firewood and/or wood pellets) for domestic heating.
• At 10 ppl per hectare, Winnipeg requires 75,000 ha (750 km²) to feed the city. This is an area ~60% larger than the city itself (but needs to be devoted to agriculture, not competing land uses) and can be met with ease in its adjacent hinterland (with imports from further afield providing more variety in residents' diets). Indeed, this area is a significant agricultural exporter.
• I'd estimate the city needs to supply in the ballpark of 210 ML/d of water, and remove roughly the same amount of wastewater.
• For solid waste, my estimate is 1750 tonnes per day. As discussed above, the total quantity of physical materials being brought in or out of the city could easily be 10x or more of this quantity. For context, a 53' transport trailer can carry up to 22 tons so this is like 80 truckloads per day just for solid waste. A railcar can carry 2-4 times as much as a transport trailer (and a long train might have 200 railcars +/-).

Winnipeg isn't a seaport, but to illustrate the advantages that coastal cities have when it comes to logistics/metabolism, a single Neopanamax ship can carry 13,000 TEUs (and a TEU has roughly the same capacity as a transport trailer).

For any of these "metabolic" flows coming from outside the city limits, they can be thought of as a flux across the 90 km perimeter. In fact, entry points are even more discrete than that. As the following pictures illustrate, there's 30 or so roads or rail lines coming in and out of Winnipeg. Of the roads, the TransCanada (Hwy 1) and the route to the US (Hwy 75) are the most significant for goods coming from outside Manitoba. Aside from that, there's also an international airport. The Red River flows through Winnipeg and connects to the United States (including Fargo ND) and to Lake Winnipeg and onward to Hudson Bay but as far as I know it doesn't have any significant barge traffic. 30 surface connections plus an international airport is actually really good for a city of this size, though.

Industry

The preceding sections certainly have some relevance to my reading and writing theme for 2024 (although I read Growth in 2023) on industry, technology, and supply chains. In this section, though, I'll deal with the more direct things Smil has to say about how things are made and about some macro trends (specifically around dematerialization) related to industrial production.

Dematerialization is a term for the trend of modern economies using less material "stuff" relative to their size. This is due to growing trade in non-material items (e.g. the service economy and the digital economy) as well as increasing efficiency. I've referred to it at least a couple of times before on this blog. At the same time, it's good to be reminded that we are material beings in a material world (Material World is the title of one of the stand-out books that I read this year and I'm currently working on a draft post that includes it, so stay tuned!) and will always be reliant on the continuation of some mass and energy flows in the real, physical world. Vaclav Smil makes this point clear:

I will look at the fundamental physical necessities and outcomes of growth, that is at energy and material flows that are now commonly overlooked (or deliberately ignored) as if the process could be only about energy-free, dematerialized changes of abstract aggregates quantified so imperfectly in such an inadequate measure of real value as money.
Although economists have a long history of ignoring energy, all economic activities are, in fundamental physical (thermodynamic) terms, simple or sequential energy conversions aimed at producing specific products or services.

 On a similar note, he says,

Decoupling economic growth from energy and material inputs contradicts physical laws: basic needs for food, shelter, education, and employment for the additional billions of people to be added by 2100 will alone demand substantial energy flows and material inputs. True, those inputs will have lower relative intensities … but the absolute totals will keep rising (with continued population growth) or will moderate but remain substantial.

 And a statement underlying this section and also the previous one on urban metabolism is that:  

Modern economies are based on massive linear flows of energy, fertilizers, other agrochemicals, and water required to produce food, and on even more massive energy and material flows to sustain industrial activities, transportation, and services.

I think these are critical points, but I also think there is something to dematerialization. As a synthesis, I'd add information to materials and energy.

Later in Growth, he does have a more balanced statement on dematerialization:

At the same time, there is no doubt that since 1973 (when the unprecedented period of rapid post-WWII growth ended) the world economy has become impressively more energy efficient and relatively less material-intensive--while continuing population growth, further increases in consumption in affluent countries, and fast economic advances in Asia in general, and in China in particular, have translated into relatively strong absolute global growth in both energy and material requirements.

Taking a couple of industries as specific examples, steel and cement production haven't reached an asymptote. By 2030, Smil projects that they will be around 2.5 Gt/yr and 10 Gt/yr, respectively, and still increasing. He writes,

They may dismiss the steel industry as an old industrial undertaking and as a largely spent low-value-added activity—but no segment of the modern economy (including all services, electronics, and data management) could exist without a constant and affordable supply of many varieties of high-quality steel.

I also appreciated this point that it isn't just materials that have enduring significance, but also processes:

Similarly, there are many industrial and manufacturing processes whose efficient operation relies on deploying such ancient inventions as sawing, grinding, polishing, casting, annealing, and welding.

The next post I have planned in my series on Industry will consider the themes in this section (among other topics) further.

Permalink