Bending the models
How the world moves from one energy vector to the next and why we are departing from the prevailing substitution cycles.
In hindsight, technological evolution and even revolutions feel like a predictable sequence of events. It is as if every discovery and development along the way was designed to unlock the subsequent one until we could substitute an old way of getting something done for a new one. This applies to energy generation, as the world partially replaced coal for wood and hydrocarbons for coal; to war as we substituted guns for bows and arrows, tanks for horses; and even to narrower applications such as replacing water-based paints for oil-based paints, detergents for soap or synthetic fibers for natural ones.
Because there are multiple technological dependencies at work and often in parallel, predicting and timing technological evolution is much harder.
In the 1970s, J. C. Fisher and R. H. Pry built a simple model to forecast technology-based substitution processes. Their paper, titled A Simple Substitution Model of Technological Change, stated that man has a few broad basic needs to be satisfied such as food, clothing, shelter, transportation, communication, education, and the like and technological evolution consists mainly of substituting a new form of satisfaction for the old one and therefore could easily be predicted by a simple model. The model’s fundamental assumptions were that technical advances that are competitive substitutions, in economic terms, will continue to replace the old method at a rate of fractional substitution proportional to the remainder that is yet to be substituted, considering that:
- When a new method is first introduced, it is less developed than the older method with which it is competing. It therefore is likely to have greater potential for improvement and for reduction in cost.
- If a substitution has progressed as far as a few percent, and therefore has proven economic viability, even without improvement and cost reductions, it will proceed to completion.
- The fractional rate of fractional substitution of new for old is proportional to the remaining amount of the old left to be substituted.
Expanding upon this model, in the late 1970s, Cesare Marchetti demonstrated that primary energy sources exhibited regular long-term trends. He used logistic curves to fit the trends in the world’s mix of primary energy sources and calculate the market fraction (f) of a new technique (or a new energy source) and express it as f/1−f. When that quotient is plotted on a semilogarithmic graph, it appears as a straight line, for which he claimed that “the whole destiny of an energy source seems to be completely predetermined in the first childhood … these trends … go unscathed through wars, wild oscillations in energy prices and depression. Final total availability of the primary reserves also seems to not affect the rate of substitution.”
The precision and simplicity of this model describe a few things about how the world worked at the time. Economic viability was the driving force for technological advances while production and consumption shared coordinated access to natural resources and technology.
Predicting the emergence of substitutes
From a pure technological perspective, it is helpful to use patent application statistics to assess trends and developments in a given field to analyze its rate of progress and possible correlation between technology advancements and gain in market share. By doing so, we assume that patent applications, as exclusive rights that can only be granted for inventions that are novel and inventive, whose owners want to protect their commercial value from being used by competitors, are a proxy for technological advancements.
LNG was latest energy source to progress in market share above a few percentage points. It emerged in the late sixties as a competitive alternative for moving natural gas across long distances, with plenty of room for improvement. Although Carl von Linde first patented a reliable process for air liquefaction in 1916, we take its starting point in the substitution model almost half a century later when Chiyoda Corporation’s patent for Direct re-gasification of liquefied natural gas by heat exchange with water (Patent No. JP48003884B Corp, 1969 ) in 1969. 688 patents were published in the field in the subsequent years, but it wasn’t until the year 2000 when the number of applications for storage, re-gasification, and transportation increased drastically, peaking in 2018 at 67. In the last 40 years, natural gas’ share of the energy market remained at 20–25% while LNG trade grew on average at 11% per year. From 2.6 MT in 1971 to 356.1 MT in 2020, the growth was steadily positive (except 1980–81 and 2012) with 40% of the time double-digit growth rates until 2012, whenLNG shipments reached ~50% of all natural gas exports. The correlation between the number of patent applications and the growth of LNG is strong. It doesn’t necessarily mean causation, but it is definitely a self-reinforcing feedback loop.
To some degree, Fisher and Pry’s and Marchetti’s model predicted LNG’s progress, from a few percentage points in market share to double-digit penetration, propelled by technological advancements.
But when we compare the last 20 years of LNG’s advancements with those in the green hydrogen and ammonia field, something seems off. The report Innovation trends in electrolyzers for hydrogen production shows that patent applications for hydrogen production technologies have grown on average by 18% each year since 2005. From 2005 to 2020, 10,894 patent families related to the electrolysis of water were published worldwide, with an average annual increase of 18%. In 2016, the number of patent families related to water electrolysis surpassed the number of patents related to solid or liquid coal- and oil-based hydrogen sources. By 2020, it was double that number. During the same period, 332 patents related to ammonia co-firing for power generation were published worldwide.
These numbers leave LNG in the dust. Yet, in contrast to what the model predicted, technological advancements came before any progress in market share or economic viability.
Departing from the model
Marchetti’s model captured the gradual nature of energy transitions, with primary resource substitutions taking decades before claiming significant shares of the overall market.
But somewhere in the 1990s, the model started to lose track. It was to deliver these global shares by the 2010s — less than 5% for coal, about 25% for oil, and just over 60% for natural gas — while the actual shares were, respectively, 29, 34, and 24%. When trying to discern the future of natural gas, the model’s greatest error was in highly overestimating its future rise. At roughly 60% of global primary energy use, natural gas would have delivered about 300 EJ of primary energy in 2010 — while in reality, it provided no more than 120 EJ.
For many reasons, the worldwide transition to natural gas has proceeded slower than in the two preceding shifts. It will have taken natural gas more than 80 years to go from 5 to 25% market share, while coal took 35 years and crude oil 40 years to reach those shares.
Today, for the first time since the Industrial Revolution there is no single dominant fuel.
The departure from Marchetti’s model is a function of multiple factors: First, as the modern global primary energy supply expanded (in 2021, it was about 595EJ; in 1950, it was about 80 EJ; in 1900, it was only 22 EJ), it is more difficult to claim an additional share of the market. Second, the model assumes that the world moves in complete sync from one vector to the next as it did through the industrial revolution. But as new productive regions and economies emerge on top of access to different resources, flexible pollution regulations, and low labor costs, the decline in the share of fuels that would have been left behind in a more concentrated world assumed in the models gets postponed.
Third, Fisher and Pry’s and Marchetti’s model was based on the assumption that the substitute source would first become economically viable to then capture a few percentage points, and grow to completion with room for improvements and scale. But they did not contemplate environmental aspects to weigh as much as economic ones.
Bending the model
CO2 fuel substitutes such as green hydrogen/ammonia are bending the model upside down. Even though they are not yet economically viable, and haven’t claimed even a percentage point in market share, leading nations like Japan and Korea are forcing (accelerating?) their adoption curve to such a scale that would make it economically viable and ready to then get on track to substitute existing sources, per the substitution models.
Japan and Korea are currently targeting 20% ammonia co-firing in existing coal power plants by 2030, while Jera, the Japanese power company, aims to reach 50% co-firing rates across their coal power plants by that time. Coal today powers roughly ~25% of those economies and account for roughly ~45% of their CO2 emissions. Through ammonia co-firing strategies, both nations could reduce up to 10% CO2 equivalent tons and, by reaching that scale, make green ammonia co-firing economical even in comparison to natural gas or coal. To support this transition, we estimate that carbon pricing requirements would need to be US$40-US$120/t of carbon dioxide equivalent for 20% and 60% co-firing shares, respectively, while ammonia delivered costs to Japan could fall to US$500/t or lower in the long term.
Green ammonia alone could grow to become a US$100 billion market in the coming decades, according to Wood Mackenzie, for the sole reason of being CO2-free.
In this sense, the rise of green hydrogen and ammonia would occur in the opposite direction of how the models were supposed to work. It is yet to be seen how far they can progress in terms of market share, though its crusade would be studied for decades.