So you are doing useful work by identifying a serious potential problem and trying to get the rest of us to take it seriously. As a neural circuit in the global brain it is a good thing that the Peak Oil movement exists.
I’m not quite sure how to approach this because you are making a conceptual mistake with this argument and I want you to actually see what it is. And I think this is a case where there is a clear truth of the matter that we can both get to and agree on.
But since there was also an argument you had in the comments on your google doc with someone pointing out the same thing I am here, it is clear that there is something about this issue that is hard for your mind to jump to seeing. At the same time it is perhaps is a bit hard for me to explain it, since my mind immediately sees it intuitively.
First I am making a narrow point.
If my point is correct, it is still totally possible that peak oil is the correct model.
I am begging you, try to just pay attention to the point, and decide if you think it is correct, and only afterwards ask if it has any broader implications.
The purpose of my arguing here, is to help you improve your economic model on this single point, and not to change your broader point of view.
With that long preface, my simple point is this: The EROI is not enough to tell you what portion of civilization’s real resources go to energy production. You need more information.
An EROI of 2:1 is not enough to tell you if the energy system requires 1 percent of GDP or 10 times the world’s total GDP. You need more information than just the EROI.
I think you already know this, since you were trying to point at evidence from historical recessions and economic performance to figure out what the economic impact of changes in EROI would be, since just saying EROI of 2:1 does not actually say ’50%’, the 50% comes from using additional information to figure out the economic impact of that number.
To establish that EROI alone does not tell you anything about the percent of GDP that goes to it, I am now going to describe a fake, fictional, toy model of a world. This is not the real world. This is a model. But this sort of model is useful for understanding constraints that exist in the actual real world. Telling me that the extreme cases in this fake, fictional, not real world are in fact fake is not an argument against what I’m saying. What I am trying to establish is that we need at least three parameters to figure out what portion of real resources go to energy production.
EROI is only one of them . I am not saying anything about what the actual value of the other parameter is here, just that any positive EROI is consistent with any GDP % depending on what the other parameters are.
In the following argument, we are going to assume the stated EROI includes all energy costs that are physically necessary to produce energy producing equipment. So it includes the costs of roads, the machines to build the roads, the machines to maintain the roads, and the machines used to build the machines. Otherwise it isn’t the actual EROI.
So to start, what we want to figure out is what part of GDP is required to produce electricity.
A start point could be this equation:
Cost of energy system = Amount of energy producing equipment required * resource cost to make each unit of energy producing equipment
Where does EROI come into the cost of the energy system? It isn’t yet here. Let’s try breaking down one of the terms:
Amount of energy producing equipment required = produced total energy / energy produced per unit of equipment.
A further break down of the equation
Produced total energy = Produced net energy + Produced waste energy
Now EROI is the ratio of total energy to waste energy (EROI = total energy/ total waste energy). So an EROI of 2:1 imples that for every two units of energy produced, there will be 1 unit of waste energy produced and one unit of net energy.
So inserting this into the equation after doing a bit of algebra to get rid of waste energy we get that:
1 = net energy/ total energy + 1/EROI = net energy/ total energy + 1⁄2 ==>
1⁄2 = net energy /total energy ==> total energy / 2 = net energy ==> 2*net energy = total energy.
So now that we’ve incorporated EROI into this term in the equation, we can go back to the orignal equation:
In the case of an EROI of 2 is:
Cost of energy system = 2* net energy produced / energy produced per unit of equipment * cost per unit of equipment.
What is the information that we do not have at this point?
We don’t know how much net energy is produced—that is still a parameter. We don’t know how much equipment we need to make that amount of energy. And we don’t know how much the equipment costs.
What we do know is that an EROI of 2 means that we need twice as many pieces of equipment as we would need with an obviously impossible infinite EROI (if EROI is infinite, then there is no waste energy term, so net energy = total energy). But that is all the EROI tells us.
This is the end of the section that I actually care about you understanding and agreeing with. The rest of it is arguing, not trying to correct a clear mistake. I do think it is interesting and relevant so I’ll leave it.
I did some googling, and it seems that the total consumption of all sources of energy in the world is around 150 million gigawatt hours a year. Some other googling says that 370 watts of installed solar capacity in California or Arizona produces around 2.5 kilowatt hours of energy per day on average. Lets assume the average efficiency of solar panels is somewhat lower than that, and 370 watts installed will produce 2 kilowatt hours in the average location they are actually used.
So if we had an EROI of 2, to get 150 million gigawatt hours of energy per year, we’d need to produce a total of 300 million gigawatt hours of energy per year. This could then be done with 150,000 gigawatts of installed solar capacity. This would probably require 1⁄500 of the world’s surface to be covered with solar panels. So now we know how much equipment is needed to replace the current global energy system with solar. Or one part of it at least, since there is also the storage systems and conversions systems.
Would doing this require 50 percent of the world’s gdp?
The answer is: It still depends on how much the solar panels cost.
Currently it seems that utility scale solar has a price of around 1 dollar/ watt installed. At that price this would be a 150 trillion dollar investment, assume the panels only last twenty years on average, and you have this system costing 7.5 trillion usd a year to maintain. That is less than 10% of global GDP, and I seriously doubt that pumped hydro storage systems and the need to figure out some way to get high temperature metallurgy and jet fuel are going to get you a vastly highly levelized cost.
Suppose we run out of key metals, and the substitutes are equally expensive, and then the solar panels cost way more than they do now, and we end up having to use concentrated solar for 5x as much money as current photovoltaics (conentrated solar does not depend upon any exotic metals that there is any chance we will run low on, aluminum and steel are sufficient). Then this costs 5 dollars per watt, it turns into a 750 trillion investment, and costs thirty five trillion a year to maintain, and around 1⁄3 of global gdp—that would be pretty bad.
Or we have way better general automation and techniques, the production cost curve continues to go down, and we have lots of cheap solar panels, and they only cost .20 cents per installed watt—and then you have the whole thing accomplished for 30 trillion dollars, and the system globally doesn’t take more money than the US defence department.
If solar panels end up at 100 dollars per watt due to resources running out, it would be impossible to support the current energy consumption with an investment of 100% of gdp. etc, etc.
All of this goes back to the point: You need to know the cost, not the EROI. If the cost is small, doubling it doesn’t matter. If the cost is big, you already are in trouble before you double it.
But saying ’50%′ due to EROI of 2 is nonsense. EROI only increases or decreases the amount of resources needed to get a given amount of net energy, it doesn’t tell you anything about what percent of society’s total resources are needed to produce that amount of net energy.
I happen to chance upon this discussion while browsing around, and decided to create an account to reply to this discussion because it is a topic of great interest to me.
I think the main reaon why you believe that Corentin’s argument on EROI affecting percent of GDP required to maintain energy production is a conceptual mistake, is because you have assumed that cost of production (of energy producing equipment) is not linked to energy use.
However, the basis of the EROI argument stems from biophysical economics, and is based on the key assumption that the vast majority of economic activity and economic value are in fact embodiement of energy. One may or may not choose to agree with this assumption, but if you do take this assumption to be true, then Corentin’s point that for e.g. a 2:1 EROI needs roughly half of society’s resources is correct.
So in the simple equation that you described:
let:
C be the cost of entire energy system
E be the total energy produced/demanded
eout,eq be the energy produced per unit of equipment
ceq be the cost per unit of equipment
ein,eq be the energy used to produce each unit of equipment
Then, C=E/eout,eq∗ceq
Because we assume that economic cost of production of anything is directly related to energy, then ceq=αein,eq, where α is some factor describing the economic cost in terms of energy.
Substituting it in the energy cost equation, we get C=αE/eout,eq∗ein,eq
eout,eq/ein,eq is exactly the definition of EROI of the energy producing equipment, and thus C=αEEROI.
Furthermore, with the same key assumption, the total economic output, in other words GDP can be also be expressed in terms of total energy produced or demanded by the economy, i.e. GDP=βE.
We finally obtain that:
C=α⋅GDPβ⋅EROI
If the scaling factor α and β between economic cost and energy is roughly similar for the particular case of energy producing equipment, and for the general case across the whole economy, then EROI approximately determines the proportion of the cost of operating and maintaining the energy system against GDP.
The key assumption put forward by the biophysical economists has been argued both through first principles and empirically (well explored in this textbook of energy and biophysical economics[1]).
I had trouble putting this into mathematical terms, so this is helpful.
I’m trying to read more stuff about EROI in order to explain it better. It’s a good concept but if we have a disagreement about how to use it, then it’s really hard to agree on something.
I hope you managed to find some interesting stuff in this post ! Feel free to share it if you found it useful.
Thank you for your excellent posts summarizing multiple sources of information across domains of energy and material limits of human development, ecological economics etc. I am still reading through your in depth 3-parts works as I speak, and I am finding many useful sources of information for my further reading.
Hi, thanks for the thougful response. You spent quite some time to put things down clearly, and I appreciate that.
I think i can accept your conclusion, for the most part. Saying “a EROI of 2:1 means half your resources go to energy production” is indeed a big simplification on my part, which is based on several simplifications I have made and didn’t detail :
Currently, energy makes about 6,5% of global GDP (well, that was 2021. For 2022, it’s about 13%). So between 1/10th and 1/20th (closer to 1/20th). This means for every point of GDP invested in energy, between 10 and 20 points of GDP are created.
Currently, the global EROI of energy is between 20:1 and 10:1 (closer to 20:1, but depends on whether you take final or out of the mine well). So for every unit of energy, between 10 and 20 units of energy are created.
From this, I make the overal simplification of “EROI is representative of the share of energy in global GDP, roughly”
“Half of resources” translates roughly to “Half of GDP” (since there is a 99% correlation between energy consumption and GDP on a year by year basis, even if this gets bigger over 50 years)
That the current relationship, for fossil fuels, still stands with solar
These are indeed huge simplifications I made in my head, but I can get why you don’t see them as valid. I unfortunately didn’t really understand your algebra bit—I am not very good at reasoning with equations, it doesn’t really “click” with the way my brain works. But I understand your overall point.
So ok, let’s drop the assumption that a 2:1 EROI requires half of society’s resources. I indeed don’t really know the exact percentage. This wasn’t really my main point, so I removed references to this assumption in the full doc.
However, what empirical data seems to indicate is that society still requires a high EROI to function. As said in another comment :
In the meantime, what I’m basing myself on is that most of the past societies had quite a high EROI (>10), including hunter-gatherers and agrarian societies (source, page 42). This surplus would allow for the many things a society needs (taking care of children and the ederly, providing for non-productive elites and administration and armies). So it’s uncertain we can really go below that, especially as we are a much more complex society.
For instance, according to this paper, you’d need a minimum of a 3:1 EROI to have transportation, when you take into account its energy needs (making and maintaining roads and trucks). An even higher EROI would be required if we add the needs for food, education, administration, healthcare and stuff like that. I made some changes to the section on EROI in the full document The great energy descent—Full Version, including how the 3:1 measure was calculated, you may find that interesting.
Of course, it may be theoretically possible that a complex society can work out with a EROI<10 or less. I’m not saying it’s cannot happen. I just think that it’s risky to make this assumption, since the historical record seems to point out that having a high energy surplus was needed in most societies.
On your second section : I do find the calculations interesting. This is well structured.
However, estimating future prices is notoriously tricky. As you put forward, on the short term prices have been decreasing in a quite impressive way, so in this time scale, and for electricity, it should go down.
I could see many reasons, however, that prices will not do that forever, and solar panels could get less affordable in the future. For instance, your calculation does not include:
The cost of upgrading the electric grid (getting the grid in deserts with a lot of sun)
The cost of switching transportation systems to electric (especially as hydrogen requires building much more infrastructure)
The cost of storage, especially seasonal (pumped hydro is good but geographically limited. Batteries, although improving, are much more expensive, and our main options depend a lot on finite materials like lithium. More in the storage section)
Metal smelting relies on coal and gas—it’s far from certain we’ll switch to electrified fast enough (or how)
China could increase its prices (80%+ of solar panels are made there)
High-grade silicon and other materials can get scarcer (as you underline)
Solar is not a good option for say Poland or Canada
So far, the best and cheapest spots have been taken, but at a large scale land is going to get expensive, especially in rich countries
I personally do not attempt to calculate prices (as seen with oil prices, it’s really hard), but it sounds likely to me that it will be more expensive than today. This doesn’t mean solar is useless—it’s just that I have trouble seeing how it can be cheap enough to support an “infinite growth” economy.
It doesn’t really defend the concept of minimum EROI as a thing that actually makes sense. My whole point is that minumum EROI of creating the seperate pieces of an energy system makes no concept.
A very bad EROI where the components are extremely cheap in terms of other resources is fine, a very high EROI where the components are extremely expensive in terms of other recources can’t be used.
Imagine a completely automated robot that is building solar panels in space, and beaming the excess energy to earth. It doesn’t matter to us right now if it used 1 (ie an EROI of 100 to 1) percent of the energy to maintain the system, or 99 percent (an EROI of 1.01 to 1), because it isn’t using any terrestial resources.
On its own, minimum EROI is a nonsense phrase. It only makes sense once you’ve specified the whole technological package and environmental context.
You have an equation with multiple terms in it. EROI is only one term, and sufficiently large changes in the other terms can compensate for changes in EROI.
Oh, ok, I get a bit better what you’re saying. (yeah, it’s tough arguing on EROI, people usually have very different views on it).
I agree that the cost of unit equipement matters a lot too.
However, I’d argue that these costs are increasing when EROI is declining. The simple reason is that you need more stuff to do the same thing. This is not a 100% correlation of course, the cost of labor matters too, but there’s a general trend, I think.
For oil with 50:1 EROI at the Ghawar field in Saudi Arabia, you just had to put a drill in place and get the oil. Shale oil, in the other hand, with an EROI between 5 and 10, requires complex chemical compounds, horizontal drilling, hundreds of trucks transporting water, and a lot of financial investment. If the EROI of shale oil was 50:1, then you’d get back 10 times more oil, so it’d be much cheaper, you’d need less materials, and you’d have more resources to power the rest of the economy.
Since there is a strong coupling between GDP and material use and resource use (at a global level), it would make sense that an increasing material and energetic cost translates to an increased financial cost.
There can be improvements of course—like solar panels getting a higher EROI and being much cheaper at the same time.
Now, let’s take the automated robot that sent solar energy back to Earth (a purely theoretical prospect with not relevance to the problem of energy depletion as will exist for the next decades, of course). With an EROI of 1.01:1 instead of 100:1, then it would need to depoly 10 000 more solar panels for the same thing. You’d need 10 000 times more solar panels, so 10 000 more materials, more rockets, more robots to build these, more factories, more maintenance… Not talking about the fact that all the computing stuff would require specialty metals that are in a finite amount.
Also, the process would be 10 000 times longer, which is of great importance.
SoI have a hard time seeing how this wouldn’t multiply the price by at least several orders of magnitude.
The main point about EROI, and I don’t think we disagree on this, is that the raw amount of produced energy that needs to be put in is only one factor. You also need to know how much human labor has to be put in, and how much physical stuff has to be put in.
I’d note a lot of the complaints here that in a bad EROI environment with needing to build more stuff, you are also running out of key metals is double counting. The reason that the EROI is 2 to 1 in that scenario, instead of 10 to 1 is because we’ve run out of the easy sources of those metals, so pointing out that the metals are also hard to acquire in that context doesn’t say anything new.
“Since there is a strong coupling between GDP and material use and resource use (at a global level), it would make sense that an increasing material and energetic cost translates to an increased financial cost. “
I don’t know if I really want to dig into this very deeply, since it involves a familiarity with economics that you clearly don’t have, but theings like this claim, and the ’99 percent correlation between energy use and gdp growth’ simply do not mean what you think they mean.
For example, you might get a correlation that is nearly as strong between gdp growth and fast food purchases, or clothes purchases, or home improvement purchases, or almost anything except for medical and government spending.
That is what a recession literally is: It is when people buy less of stuff that can be cut back on easily. And booms are when people buy more of that stuff. You are going to find extremely high correlations between gdp growth and any variable consumption good if you are looking for that, but it is meaningless in terms of saying what is causally important for allowing continued economic growth.
In a similar way that recessions usually follow very high energy prices (which is causal), does not actually mean that the economy cannot deal with energy taking up that big of a proportion of total resources without going into a permanent recession. It means that if in a given year everyone has to spend way more money on energy, they won’t have as much money left to spend on everything else they want, so they will buy less of it, so the economy will enter a recession.
But if the energy prices stayed high, this would be a one time thing, where improvements in productivity through out the economy would allow higher profits and wages again, and thus with the fixed high energy price, they would be able to purchase more non energy things in year two of high energy prices than in year one—ie the economy would be growing.
Having ten percent of the economy go to a sector simply doesn’t mean the other sectors can’t keep increasing total output per capita over time. For example, in most countries the health care sector has been becoming bigger and bigger relative to the total economy over time. In the US it is around 20 percent of the economy now, while it was 7 or 8 percent (I think) in the 80s. Despite this, the non health care sectors have consitently been getting bigger at the same time. Of course the giant allocation of resources to health care does cause bad things, and we are poorer than we would be if all health care happened by magic and didn’t cost anything, and it possibly has crowded out capital investments that would have led to growth elsewhere, and thus we are poorer in dynamic terms in addition to static terms due to health care costs. But it has not, and will not, cause a permanent recession (ie the rest of the economy makes fewer things per person each year until at the limit nothing is ever made) if it gets sufficiently big.
You also need to know how much human labor has to be put in, and how much physical stuff has to be put in.
Ok, I can agree with that.
It’s just that today, human muscles are such a small part of the labor produced (one barrel of oil = 4.5 years of manual labor, after conversion losses) that I didn’t though of including it.
For the metals, I understand that it’s extraction is already in the theoretical 2:1 figure. I just mentioned them to point out that we don’t really know how energy costly it is to get specialty metals of electronics in a “sustainable” way (from either extremely abundant ores or from common ground). My personal impression on the topic is that, except for iron and aluminium and maybe a few others (rare earths, ironically?), getting stuff like indium, tellurium or molybdenum from common ground (for electronics) would be so ridiculously expensive that we’d give up before that.
For example, you might get a correlation that is nearly as strong between gdp growth and fast food purchases, or clothes purchases, or home improvement purchases, or almost anything except for medical and government spending.
I agree here that just using the energy/GDP correlation is not enough. This is why I tried to make a section listing the scientific papers that study this correlation, and conclude that it is more serious than, say, the relationship between GDP and tomatoes.
Here is one account that you might find of interest:
“While the classical economists Adam Smith and David Ricardo generally thought that it was human labor that was the principal generator of wealth, natural resources, in particular land, still played a major role as a source of value and as a constraint to unlimited economic growth. Later Karl Marx, while still seeing human labor as the source of value, removed this constraint by referring to the evolution of the ‘means of production’ (that is, technology) that only depended on (principally unlimited) human ingenuity.”
“In the twentieth century, the explanation of wealth left natural resources behind and focused on capital and labor only (see production functions by Cobb and Douglas, 1928 and Solow, 1956). As in mathematical calculations there remained a large ‘residual’, this was attributed to technological innovation [the Solow residual] (that could not, however, be properly measured). Authors like Cleveland et al. (1984a), Cleveland (1991), Ayres and Warr (2005) and Hall and Klitgaard (2012), in contrast, attributed this residual to energy (or exergy) input into the economy and were able to provide convincing empirical evidence. Unexplained residuals disappeared.”
So we are not dealing with a random commodity here. We are dealing with a factor of production.
If we look at a biophysical standpoint, the economy is the production of goods and services. Energy is what allows to produce these goods and services (and the food/transport/housing of the workers). It seems unlikely that we can produce ever more and more goods and services using less and less energy. Maybe for a short period as we use the lowest-hanging appels, but not in a sustained way.
The historical record seems to indicate that less energy and more GDP at a global level is a very strong departing of the current trend, and unlikely to happen. Maybe not impossible (for how long?), but we shouldn’t assume this as he default scenario.
Of course, it’s possible to decouple GDP from producing goods and services. This may be what the finance sector is doing, generating money (8% of US GDP) while not contributing much to the well-being of society. I’d be tempted to see something similar with healthcare in the US—it has quite a reputation for being extremely expensive compared to what you get for the same price in Europe. I’m tempted to ask, is growing GDP any use if it doesn’t contribute to society ?
I agree with the example of the robot in the space. There the EROI doesn’t matter so much. Until we have this solution in place, we would have to analyze the whole technological package and environmental context, as you very well said.
I would be very interested to know what your assumptions about this whole technological package and environmental context are, especially when it comes to a fast transition to replace a declining amount of energy from fossil sources. Have you ever done this exercise for your country or the world? I would love to see the results.
So you are doing useful work by identifying a serious potential problem and trying to get the rest of us to take it seriously. As a neural circuit in the global brain it is a good thing that the Peak Oil movement exists.
I’m not quite sure how to approach this because you are making a conceptual mistake with this argument and I want you to actually see what it is. And I think this is a case where there is a clear truth of the matter that we can both get to and agree on.
But since there was also an argument you had in the comments on your google doc with someone pointing out the same thing I am here, it is clear that there is something about this issue that is hard for your mind to jump to seeing. At the same time it is perhaps is a bit hard for me to explain it, since my mind immediately sees it intuitively.
First I am making a narrow point.
If my point is correct, it is still totally possible that peak oil is the correct model.
I am begging you, try to just pay attention to the point, and decide if you think it is correct, and only afterwards ask if it has any broader implications.
The purpose of my arguing here, is to help you improve your economic model on this single point, and not to change your broader point of view.
With that long preface, my simple point is this: The EROI is not enough to tell you what portion of civilization’s real resources go to energy production. You need more information.
An EROI of 2:1 is not enough to tell you if the energy system requires 1 percent of GDP or 10 times the world’s total GDP. You need more information than just the EROI.
I think you already know this, since you were trying to point at evidence from historical recessions and economic performance to figure out what the economic impact of changes in EROI would be, since just saying EROI of 2:1 does not actually say ’50%’, the 50% comes from using additional information to figure out the economic impact of that number.
To establish that EROI alone does not tell you anything about the percent of GDP that goes to it, I am now going to describe a fake, fictional, toy model of a world. This is not the real world. This is a model. But this sort of model is useful for understanding constraints that exist in the actual real world. Telling me that the extreme cases in this fake, fictional, not real world are in fact fake is not an argument against what I’m saying. What I am trying to establish is that we need at least three parameters to figure out what portion of real resources go to energy production.
EROI is only one of them . I am not saying anything about what the actual value of the other parameter is here, just that any positive EROI is consistent with any GDP % depending on what the other parameters are.
In the following argument, we are going to assume the stated EROI includes all energy costs that are physically necessary to produce energy producing equipment. So it includes the costs of roads, the machines to build the roads, the machines to maintain the roads, and the machines used to build the machines. Otherwise it isn’t the actual EROI.
So to start, what we want to figure out is what part of GDP is required to produce electricity.
A start point could be this equation:
Cost of energy system = Amount of energy producing equipment required * resource cost to make each unit of energy producing equipment
Where does EROI come into the cost of the energy system? It isn’t yet here. Let’s try breaking down one of the terms:
Amount of energy producing equipment required = produced total energy / energy produced per unit of equipment.
A further break down of the equation
Produced total energy = Produced net energy + Produced waste energy
Now EROI is the ratio of total energy to waste energy (EROI = total energy/ total waste energy). So an EROI of 2:1 imples that for every two units of energy produced, there will be 1 unit of waste energy produced and one unit of net energy.
So inserting this into the equation after doing a bit of algebra to get rid of waste energy we get that:
1 = net energy/ total energy + 1/EROI = net energy/ total energy + 1⁄2 ==>
1⁄2 = net energy /total energy ==> total energy / 2 = net energy ==> 2*net energy = total energy.
So now that we’ve incorporated EROI into this term in the equation, we can go back to the orignal equation:
In the case of an EROI of 2 is:
Cost of energy system = 2* net energy produced / energy produced per unit of equipment * cost per unit of equipment.
What is the information that we do not have at this point?
We don’t know how much net energy is produced—that is still a parameter. We don’t know how much equipment we need to make that amount of energy. And we don’t know how much the equipment costs.
What we do know is that an EROI of 2 means that we need twice as many pieces of equipment as we would need with an obviously impossible infinite EROI (if EROI is infinite, then there is no waste energy term, so net energy = total energy). But that is all the EROI tells us.
This is the end of the section that I actually care about you understanding and agreeing with. The rest of it is arguing, not trying to correct a clear mistake. I do think it is interesting and relevant so I’ll leave it.
I did some googling, and it seems that the total consumption of all sources of energy in the world is around 150 million gigawatt hours a year. Some other googling says that 370 watts of installed solar capacity in California or Arizona produces around 2.5 kilowatt hours of energy per day on average. Lets assume the average efficiency of solar panels is somewhat lower than that, and 370 watts installed will produce 2 kilowatt hours in the average location they are actually used.
So if we had an EROI of 2, to get 150 million gigawatt hours of energy per year, we’d need to produce a total of 300 million gigawatt hours of energy per year. This could then be done with 150,000 gigawatts of installed solar capacity. This would probably require 1⁄500 of the world’s surface to be covered with solar panels. So now we know how much equipment is needed to replace the current global energy system with solar. Or one part of it at least, since there is also the storage systems and conversions systems.
Would doing this require 50 percent of the world’s gdp?
The answer is: It still depends on how much the solar panels cost.
Currently it seems that utility scale solar has a price of around 1 dollar/ watt installed. At that price this would be a 150 trillion dollar investment, assume the panels only last twenty years on average, and you have this system costing 7.5 trillion usd a year to maintain. That is less than 10% of global GDP, and I seriously doubt that pumped hydro storage systems and the need to figure out some way to get high temperature metallurgy and jet fuel are going to get you a vastly highly levelized cost.
Suppose we run out of key metals, and the substitutes are equally expensive, and then the solar panels cost way more than they do now, and we end up having to use concentrated solar for 5x as much money as current photovoltaics (conentrated solar does not depend upon any exotic metals that there is any chance we will run low on, aluminum and steel are sufficient). Then this costs 5 dollars per watt, it turns into a 750 trillion investment, and costs thirty five trillion a year to maintain, and around 1⁄3 of global gdp—that would be pretty bad.
Or we have way better general automation and techniques, the production cost curve continues to go down, and we have lots of cheap solar panels, and they only cost .20 cents per installed watt—and then you have the whole thing accomplished for 30 trillion dollars, and the system globally doesn’t take more money than the US defence department.
If solar panels end up at 100 dollars per watt due to resources running out, it would be impossible to support the current energy consumption with an investment of 100% of gdp. etc, etc.
All of this goes back to the point: You need to know the cost, not the EROI. If the cost is small, doubling it doesn’t matter. If the cost is big, you already are in trouble before you double it.
But saying ’50%′ due to EROI of 2 is nonsense. EROI only increases or decreases the amount of resources needed to get a given amount of net energy, it doesn’t tell you anything about what percent of society’s total resources are needed to produce that amount of net energy.
I happen to chance upon this discussion while browsing around, and decided to create an account to reply to this discussion because it is a topic of great interest to me.
I think the main reaon why you believe that Corentin’s argument on EROI affecting percent of GDP required to maintain energy production is a conceptual mistake, is because you have assumed that cost of production (of energy producing equipment) is not linked to energy use.
However, the basis of the EROI argument stems from biophysical economics, and is based on the key assumption that the vast majority of economic activity and economic value are in fact embodiement of energy. One may or may not choose to agree with this assumption, but if you do take this assumption to be true, then Corentin’s point that for e.g. a 2:1 EROI needs roughly half of society’s resources is correct.
So in the simple equation that you described:
let:
C be the cost of entire energy system
ceq be the cost per unit of equipment
Then, C=E/eout,eq∗ceq
Because we assume that economic cost of production of anything is directly related to energy, then ceq=αein,eq, where α is some factor describing the economic cost in terms of energy.
Substituting it in the energy cost equation, we get C=αE/eout,eq∗ein,eq
eout,eq/ein,eq is exactly the definition of EROI of the energy producing equipment, and thus C=αEEROI.
Furthermore, with the same key assumption, the total economic output, in other words GDP can be also be expressed in terms of total energy produced or demanded by the economy, i.e. GDP=βE.
We finally obtain that:
C=α⋅GDPβ⋅EROI
If the scaling factor α and β between economic cost and energy is roughly similar for the particular case of energy producing equipment, and for the general case across the whole economy, then EROI approximately determines the proportion of the cost of operating and maintaining the energy system against GDP.
The key assumption put forward by the biophysical economists has been argued both through first principles and empirically (well explored in this textbook of energy and biophysical economics[1]).
Hall, C. A., & Klitgaard, K. A. (2011). Energy and the Wealth of Nations. New York: Springer.
Thank you for this !
I had trouble putting this into mathematical terms, so this is helpful.
I’m trying to read more stuff about EROI in order to explain it better. It’s a good concept but if we have a disagreement about how to use it, then it’s really hard to agree on something.
I hope you managed to find some interesting stuff in this post ! Feel free to share it if you found it useful.
Thank you for your excellent posts summarizing multiple sources of information across domains of energy and material limits of human development, ecological economics etc. I am still reading through your in depth 3-parts works as I speak, and I am finding many useful sources of information for my further reading.
Hi, thanks for the thougful response. You spent quite some time to put things down clearly, and I appreciate that.
I think i can accept your conclusion, for the most part. Saying “a EROI of 2:1 means half your resources go to energy production” is indeed a big simplification on my part, which is based on several simplifications I have made and didn’t detail :
Currently, energy makes about 6,5% of global GDP (well, that was 2021. For 2022, it’s about 13%). So between 1/10th and 1/20th (closer to 1/20th). This means for every point of GDP invested in energy, between 10 and 20 points of GDP are created.
Currently, the global EROI of energy is between 20:1 and 10:1 (closer to 20:1, but depends on whether you take final or out of the mine well). So for every unit of energy, between 10 and 20 units of energy are created.
From this, I make the overal simplification of “EROI is representative of the share of energy in global GDP, roughly”
“Half of resources” translates roughly to “Half of GDP” (since there is a 99% correlation between energy consumption and GDP on a year by year basis, even if this gets bigger over 50 years)
That the current relationship, for fossil fuels, still stands with solar
These are indeed huge simplifications I made in my head, but I can get why you don’t see them as valid. I unfortunately didn’t really understand your algebra bit—I am not very good at reasoning with equations, it doesn’t really “click” with the way my brain works. But I understand your overall point.
So ok, let’s drop the assumption that a 2:1 EROI requires half of society’s resources. I indeed don’t really know the exact percentage. This wasn’t really my main point, so I removed references to this assumption in the full doc.
However, what empirical data seems to indicate is that society still requires a high EROI to function. As said in another comment :
For instance, according to this paper, you’d need a minimum of a 3:1 EROI to have transportation, when you take into account its energy needs (making and maintaining roads and trucks). An even higher EROI would be required if we add the needs for food, education, administration, healthcare and stuff like that. I made some changes to the section on EROI in the full document The great energy descent—Full Version, including how the 3:1 measure was calculated, you may find that interesting.
Of course, it may be theoretically possible that a complex society can work out with a EROI<10 or less. I’m not saying it’s cannot happen. I just think that it’s risky to make this assumption, since the historical record seems to point out that having a high energy surplus was needed in most societies.
On your second section : I do find the calculations interesting. This is well structured.
However, estimating future prices is notoriously tricky. As you put forward, on the short term prices have been decreasing in a quite impressive way, so in this time scale, and for electricity, it should go down.
I could see many reasons, however, that prices will not do that forever, and solar panels could get less affordable in the future. For instance, your calculation does not include:
The cost of upgrading the electric grid (getting the grid in deserts with a lot of sun)
The cost of switching transportation systems to electric (especially as hydrogen requires building much more infrastructure)
The cost of storage, especially seasonal (pumped hydro is good but geographically limited. Batteries, although improving, are much more expensive, and our main options depend a lot on finite materials like lithium. More in the storage section)
Metal smelting relies on coal and gas—it’s far from certain we’ll switch to electrified fast enough (or how)
China could increase its prices (80%+ of solar panels are made there)
High-grade silicon and other materials can get scarcer (as you underline)
Solar is not a good option for say Poland or Canada
So far, the best and cheapest spots have been taken, but at a large scale land is going to get expensive, especially in rich countries
I personally do not attempt to calculate prices (as seen with oil prices, it’s really hard), but it sounds likely to me that it will be more expensive than today. This doesn’t mean solar is useless—it’s just that I have trouble seeing how it can be cheap enough to support an “infinite growth” economy.
You may be interested on this paper:
https://www.sciencedirect.com/science/article/pii/S0301421513006447
It explores the question of whether EROI correlates with several quality of life metrics.
Section “5.1. The concept of minimum EROI” covers your debate with Corentin about what is the lowest EROI for a complex society.
It doesn’t really defend the concept of minimum EROI as a thing that actually makes sense. My whole point is that minumum EROI of creating the seperate pieces of an energy system makes no concept.
A very bad EROI where the components are extremely cheap in terms of other resources is fine, a very high EROI where the components are extremely expensive in terms of other recources can’t be used.
Imagine a completely automated robot that is building solar panels in space, and beaming the excess energy to earth. It doesn’t matter to us right now if it used 1 (ie an EROI of 100 to 1) percent of the energy to maintain the system, or 99 percent (an EROI of 1.01 to 1), because it isn’t using any terrestial resources.
On its own, minimum EROI is a nonsense phrase. It only makes sense once you’ve specified the whole technological package and environmental context.
You have an equation with multiple terms in it. EROI is only one term, and sufficiently large changes in the other terms can compensate for changes in EROI.
Oh, ok, I get a bit better what you’re saying. (yeah, it’s tough arguing on EROI, people usually have very different views on it).
I agree that the cost of unit equipement matters a lot too.
However, I’d argue that these costs are increasing when EROI is declining. The simple reason is that you need more stuff to do the same thing. This is not a 100% correlation of course, the cost of labor matters too, but there’s a general trend, I think.
For oil with 50:1 EROI at the Ghawar field in Saudi Arabia, you just had to put a drill in place and get the oil. Shale oil, in the other hand, with an EROI between 5 and 10, requires complex chemical compounds, horizontal drilling, hundreds of trucks transporting water, and a lot of financial investment. If the EROI of shale oil was 50:1, then you’d get back 10 times more oil, so it’d be much cheaper, you’d need less materials, and you’d have more resources to power the rest of the economy.
Since there is a strong coupling between GDP and material use and resource use (at a global level), it would make sense that an increasing material and energetic cost translates to an increased financial cost.
There can be improvements of course—like solar panels getting a higher EROI and being much cheaper at the same time.
Now, let’s take the automated robot that sent solar energy back to Earth (a purely theoretical prospect with not relevance to the problem of energy depletion as will exist for the next decades, of course). With an EROI of 1.01:1 instead of 100:1, then it would need to depoly 10 000 more solar panels for the same thing. You’d need 10 000 times more solar panels, so 10 000 more materials, more rockets, more robots to build these, more factories, more maintenance… Not talking about the fact that all the computing stuff would require specialty metals that are in a finite amount.
Also, the process would be 10 000 times longer, which is of great importance.
So I have a hard time seeing how this wouldn’t multiply the price by at least several orders of magnitude.
The main point about EROI, and I don’t think we disagree on this, is that the raw amount of produced energy that needs to be put in is only one factor. You also need to know how much human labor has to be put in, and how much physical stuff has to be put in.
I’d note a lot of the complaints here that in a bad EROI environment with needing to build more stuff, you are also running out of key metals is double counting. The reason that the EROI is 2 to 1 in that scenario, instead of 10 to 1 is because we’ve run out of the easy sources of those metals, so pointing out that the metals are also hard to acquire in that context doesn’t say anything new.
“Since there is a strong coupling between GDP and material use and resource use (at a global level), it would make sense that an increasing material and energetic cost translates to an increased financial cost. “
I don’t know if I really want to dig into this very deeply, since it involves a familiarity with economics that you clearly don’t have, but theings like this claim, and the ’99 percent correlation between energy use and gdp growth’ simply do not mean what you think they mean.
For example, you might get a correlation that is nearly as strong between gdp growth and fast food purchases, or clothes purchases, or home improvement purchases, or almost anything except for medical and government spending.
That is what a recession literally is: It is when people buy less of stuff that can be cut back on easily. And booms are when people buy more of that stuff. You are going to find extremely high correlations between gdp growth and any variable consumption good if you are looking for that, but it is meaningless in terms of saying what is causally important for allowing continued economic growth.
In a similar way that recessions usually follow very high energy prices (which is causal), does not actually mean that the economy cannot deal with energy taking up that big of a proportion of total resources without going into a permanent recession. It means that if in a given year everyone has to spend way more money on energy, they won’t have as much money left to spend on everything else they want, so they will buy less of it, so the economy will enter a recession.
But if the energy prices stayed high, this would be a one time thing, where improvements in productivity through out the economy would allow higher profits and wages again, and thus with the fixed high energy price, they would be able to purchase more non energy things in year two of high energy prices than in year one—ie the economy would be growing.
Having ten percent of the economy go to a sector simply doesn’t mean the other sectors can’t keep increasing total output per capita over time. For example, in most countries the health care sector has been becoming bigger and bigger relative to the total economy over time. In the US it is around 20 percent of the economy now, while it was 7 or 8 percent (I think) in the 80s. Despite this, the non health care sectors have consitently been getting bigger at the same time. Of course the giant allocation of resources to health care does cause bad things, and we are poorer than we would be if all health care happened by magic and didn’t cost anything, and it possibly has crowded out capital investments that would have led to growth elsewhere, and thus we are poorer in dynamic terms in addition to static terms due to health care costs. But it has not, and will not, cause a permanent recession (ie the rest of the economy makes fewer things per person each year until at the limit nothing is ever made) if it gets sufficiently big.
Ok, I can agree with that.
It’s just that today, human muscles are such a small part of the labor produced (one barrel of oil = 4.5 years of manual labor, after conversion losses) that I didn’t though of including it.
For the metals, I understand that it’s extraction is already in the theoretical 2:1 figure. I just mentioned them to point out that we don’t really know how energy costly it is to get specialty metals of electronics in a “sustainable” way (from either extremely abundant ores or from common ground). My personal impression on the topic is that, except for iron and aluminium and maybe a few others (rare earths, ironically?), getting stuff like indium, tellurium or molybdenum from common ground (for electronics) would be so ridiculously expensive that we’d give up before that.
I agree here that just using the energy/GDP correlation is not enough. This is why I tried to make a section listing the scientific papers that study this correlation, and conclude that it is more serious than, say, the relationship between GDP and tomatoes.
Here is one account that you might find of interest:
So we are not dealing with a random commodity here. We are dealing with a factor of production.
If we look at a biophysical standpoint, the economy is the production of goods and services. Energy is what allows to produce these goods and services (and the food/transport/housing of the workers). It seems unlikely that we can produce ever more and more goods and services using less and less energy. Maybe for a short period as we use the lowest-hanging appels, but not in a sustained way.
The historical record seems to indicate that less energy and more GDP at a global level is a very strong departing of the current trend, and unlikely to happen. Maybe not impossible (for how long?), but we shouldn’t assume this as he default scenario.
Of course, it’s possible to decouple GDP from producing goods and services. This may be what the finance sector is doing, generating money (8% of US GDP) while not contributing much to the well-being of society. I’d be tempted to see something similar with healthcare in the US—it has quite a reputation for being extremely expensive compared to what you get for the same price in Europe. I’m tempted to ask, is growing GDP any use if it doesn’t contribute to society ?
I agree with the example of the robot in the space. There the EROI doesn’t matter so much. Until we have this solution in place, we would have to analyze the whole technological package and environmental context, as you very well said.
I would be very interested to know what your assumptions about this whole technological package and environmental context are, especially when it comes to a fast transition to replace a declining amount of energy from fossil sources. Have you ever done this exercise for your country or the world? I would love to see the results.