I decided to model this phenomenon of the "energy trap" by using a small computer program, which I wrote in python. Any reader can download the python interpreter for free and run the simulation on his computer (the source code is posted in the comments below).
For the simulation, I made the following assumptions:
- Civilization gets all of its energy as electricity, generated from burning fossil fuels
- All fossil fuels peak on the same day and decline immediately according to the right-hand side of a Gaussian curve
- Fossil fuels start declining immediately without warning, and without any kind of production plateau
- The Gaussian decline curve has a standard deviation of 30 years which is a very rapid decline. As a result, there is a 50% decline in all fossil fuel production in only 34 years.
- There are no "unconventional" fossil fuels which will allow us to delay the decline or extend the decline curve
- No preparation has been made. The investment in renewables beforehand was zero.
- Investors and decision-makers do not begin investing in renewables until 7 years after the declines in fossil fuel production have begun, because it takes time to realize what is happening and ramp up PV production.
- Investors use a very naive formula for determining how much PV to build. Once they realize what is happening, they start investing about 5% of electricity production per year to building renewables, later increasing the investment to 1/ERoEI.
Please note that these assumptions are all incredibly pessimistic. These were by far the most pessimistic assumptions which I could imagine but which were still at least somewhat plausible.
If I run my simulation with those parameters, what results do I get? Here are the results in tabular format:
| year | gross_ff | gross_pv | gross_total | net_total | invest_pv | invest_ff | fraction_original_net |
| 0 | 1.0000 | 0.0000 | 1.0000 | 0.9000 | 0.0000 | 0.1000 | 1.0000 |
| 2 | 0.9978 | 0.0000 | 0.9978 | 0.8978 | 0.0000 | 0.1000 | 0.9975 |
| 4 | 0.9912 | 0.0000 | 0.9912 | 0.8912 | 0.0000 | 0.1000 | 0.9902 |
| 6 | 0.9802 | 0.0000 | 0.9802 | 0.8802 | 0.0000 | 0.1000 | 0.9780 |
| 8 | 0.9651 | 0.0167 | 0.9817 | 0.8449 | 0.0500 | 0.0869 | 0.9388 |
| 10 | 0.9460 | 0.0500 | 0.9960 | 0.8608 | 0.0500 | 0.0851 | 0.9565 |
| 12 | 0.9231 | 0.0833 | 1.0064 | 0.8734 | 0.0500 | 0.0831 | 0.9704 |
| 14 | 0.8968 | 0.1167 | 1.0135 | 0.8828 | 0.0500 | 0.0807 | 0.9809 |
| 16 | 0.8674 | 0.1500 | 1.0174 | 0.8894 | 0.0500 | 0.0781 | 0.9882 |
| 18 | 0.8353 | 0.1833 | 1.0186 | 0.8934 | 0.0500 | 0.0752 | 0.9927 |
| 20 | 0.8007 | 0.2167 | 1.0174 | 0.8953 | 0.0500 | 0.0721 | 0.9948 |
| 22 | 0.7642 | 0.2500 | 1.0142 | 0.8954 | 0.0500 | 0.0688 | 0.9949 |
| 24 | 0.7261 | 0.2833 | 1.0095 | 0.8941 | 0.0500 | 0.0654 | 0.9935 |
| 26 | 0.6869 | 0.3167 | 1.0036 | 0.8918 | 0.0500 | 0.0618 | 0.9908 |
| 28 | 0.6469 | 0.3500 | 0.9969 | 0.8887 | 0.0500 | 0.0582 | 0.9874 |
| 30 | 0.6065 | 0.4000 | 1.0065 | 0.8519 | 0.1000 | 0.0546 | 0.9466 |
| 32 | 0.5662 | 0.4667 | 1.0328 | 0.8819 | 0.1000 | 0.0510 | 0.9799 |
| 34 | 0.5261 | 0.5333 | 1.0595 | 0.9121 | 0.1000 | 0.0474 | 1.0134 |
| 36 | 0.4868 | 0.6000 | 1.0868 | 0.9429 | 0.1000 | 0.0438 | 1.0477 |
| 38 | 0.4483 | 0.6500 | 1.0983 | 0.9580 | 0.1000 | 0.0403 | 1.0644 |
Which raises the question: what will we actually do? Will we decide to forgo 5% of our electricity consumption now, as I assume above, in order to avert the gradual collapse of civilization over the next few decades? Or will we take the short-term view, and decide to "eat our seed corn" (so to speak) and cannibalize our energy infrastructure, leading to a small increase in our energy consumption now but the destruction of our civilization later?
Tom Murphy has this to say about it:
"Politically, the Energy Trap is a killer. In my lifetime, I have not witnessed in our political system the adult behavior that would be needed to buckle down for a long-term goal involving short-term sacrifice."
I disagree with that remark. These decisions are not made by our political system, but by investors in energy markets. Those investors routinely make short term sacrifices for larger payouts later. That is what investment means. For example, investors routinely carry out long-term planning and buy capital equipment (such as power plants) which will pay out over 30 years, but which require an up-front investment now. That is why we have power plants. Investors could always eat their seed corn and spend the money now rather than investing in the future. In general, they don't do that.
When fossil fuels start declining, the price of energy will skyrocket. Even a modest decline of a few percent of energy, could lead to a tripling of prices or more. At that point, the financial return of investing in renewables would be enormous and nearly certain. Any investment in renewables would promise vast payouts down the line, far higher than are obtained by any other investments. As a result, investors will transfer money from other investments in to this one. Investors are capable of outbidding consumers for that 5% of yearly electricity which is necessary to invest for the transition.
The energy trap is actually a fairly mild problem. Even using the incredibly pessimistic assumptions I outlined above, we will never face more than a 6.12% deficit of energy. The deficit starts decreasing right away and almost vanishes within 9 years after it begun. The energy trap is easy to overcome, with only modest and temporary sacrifices.
Furthermore, the deficit of 6.12% is almost certainly higher than what we will face in reality. We have begun transitioning to renewables decades before fossil fuels have begun declining. Furthermore, we get a large fraction of our energy now from sources other than fossil fuels (like nuclear and hydro-electric). What's more, the decline in fossil fuel production will be far more gradual than I modeled above. Also, there will be a production plateau lasting decades before fossil fuels start declining. Furthermore, investors will use a more sophisticated algorithm when determining how much PV to build, rather than just suddenly increasing PV investment from 0% to 5% (as I modeled above) which briefly worsens the energy deficit. When I run my model with more realistic assumptions that aren't so incredibly pessimistic, I find an energy deficit of less than 0.4% at its worst point.
In conclusion, the energy trap is easy to overcome with only modest adjustments. It requires modest planning--the kind which investment markets routinely carry out. As a result, the energy trap will be a minor problem which will impose only temporary and insignificant reductions in energy, in my opinion. It is also possible that civilization will transition to renewables before we reach peak fossil fuels, in which case the energy deficit will be zero.
(NOTE: The python source code is posted in the comments below)
(NOTE: I made minor changes to the wording of this article two days after initial publication. The values from the table have not changed.)
Sorry. I don't understand well your tables.
ReplyDeleteAre you proposing a fixed percentage investment over GDP?
One of the problems of the PV expansion is that takes time. The infrastructure (factories) grows exponentialy, but at manageable percentage.
For example, taking the BP data (see my old discussion in the other post) now renewable is growing at 15%, while PV is growing at 32%.
Perhaps renewable could grow slightly faster, but it's not realistic to expect over 50% growth at worldwide scale when the numbers are not very small.
So the PV deployment speed is mainly driven by inteslf, not GDP.
Perhaps, only, if this energy trap reach the numbers of PV growth, then they could limit the PV expansion.
More things. Although I think that the Hubbert curve of your assumption is realistic, it would be interesting to try a pessimistic one to check the results in a "very bad, near worst scenario", like a Seneca cliff. The Seneca cliff is becoming popular on peak oil movement (you know... always expecting the worst), but there is reasons to admit as a valid scenario altough not the most probable.
Last, I didn't find the Phyton code. Do you linked it?
Hi Oatleg,
Delete"One of the problems of the PV expansion is that takes time. The infrastructure (factories) grows exponentialy, but at manageable percentage."
I include a 7-year delay in the simulation for the amount of time necessary to realize FF production has peaked and to ramp up PV factories. This number is adjustable.
"it would be interesting to try a pessimistic one to check the results in a "very bad, near worst scenario", like a Seneca cliff. The Seneca cliff is becoming popular on peak oil movement"
I am already making extremely pessimistic assumptions for rates of decline. Using the assumptions I've made, all fossil fuels decline by half in 34 years. That decline is at least twice as fast as any that would actually occur.
I just don't see any reason at all why FF production would follow a Seneca cliff. Statistically, it just doesn't make any sense. Presumably, the reason FF production follows a bell curve is because discoveries are normally distributed. As a result, there is just no chance that all fossil fuel production from all wells (gas and oil) in all fields would be synchronized, and all drop off a cliff at the same time. That is like flipping coins and getting heads a million times in a row. The chances of that are so close to being zero that I don't think it's a serious possibility.
Hi oatleg,
Delete"Sorry. I don't understand well your tables. Are you proposing a fixed percentage investment over GDP?"
No. The numbers are fractions of 1, where 1 is the original gross energy available. So a value of 0.05 for "invest_pv" means that 5% of initial energy is invested in building PV panels that year.
Hi Tom,
ReplyDeleteI would be interested to see the source code of your program, but a few thoughts spring to mind:
Your table shows (I think) that a 0.05 energy investment in PV would yield an immediate 0.0167 return in energy each year. That doesn’t seem right to me, given that most of the energy for PV is invested up-front. If we accept an EROEI of 5 for PV (some will think this harsh, some generous), then we can expect that a 0.05 energy investment to pay off as 0.25 units of energy over the lifetime (25 years [1]) of the system. This would imply 0.01 annual energy units per year return on the 0.05 annual energy unit investment. This is also an overestimate of the energy return, because such a huge increase in renewables production would also require a commensurate increase in manufacturing capacity — we would not just be building more PV panels and turbines, but more factories, supply chains, mining equipment, etc — all of which would be an up-front energy cost.
That’s the most glaring technical problem I can see, the other problems are of a systems nature. You assume that you can change fuel availability in such a significant way, but that nothing else will change — this is an error.
The first problem I can see, is that the reduction in availability would be mostly borne by the developed world. What might be a 6% global energy deficit, would be more like 15% in the developed world (which uses about 40% of global energy). This is because developing countries derive more value from their (much smaller) energy spending, and hence can out-bid developed countries for the limited supplies that were available (this process is already underway: the energy consumption of the developed world is decreasing).
Clearly, the developed world would not tolerate a 15% reduction in its liquid fuel supply, so we would likely see a marked increase in conflict. This would lead to a reduction in global trade, and reduced investment, which would negatively affect the roll-out of renewable energy.
This is a thought-experiment only, not a prediction. I’m just thinking about how global systems might react to a decrease in energy supply of “only” 6%. The point is to illustrate that it is not reasonable to expect no flow-on effects of sharply decreasing energy supplies.
Cheers, Angus
[1] I realise that a solar PV system will still work after 25 years, but this is typically the lifetime that is assumed when working out the stats
Hi Angus,
ReplyDelete"That doesn’t seem right to me, given that most of the energy for PV is invested up-front. If we accept an EROEI of 5 for PV ... then we can expect that a 0.05 energy investment to pay off as 0.25 units of energy over the lifetime (25 years [1]) of the system. This would imply 0.01 annual energy units per year return..."
I assume an ERoEI of 10 and a lifetime of 30 years. The formula (0.05*10)/30 = 0.0167
I assume that PV has an ERoEI of 10. That is well below what is indicated in the most recent studies published. It is also below the studies cited by the National Renewable Energy Labratory: http://www.nrel.gov/docs/fy04osti/35489.pdf
Those studies are more than 15 years old and so are almost certainly an underestimate of PV ERoEI, since ERoEI has been improving.
I do not wish to cherry-pick the few EROI studies produced by what appears to be a doomsday group. There are only two recent studies showing PV with an ERoEI lower than 5, but those are extreme outliers and have serious problems. I'm using a figure which is representative of the results that researchers generally obtain.
"we would not just be building more PV panels and turbines, but more factories, supply chains, mining equipment, etc — all of which would be an up-front energy cost."
Most ERoEI studies include energy costs of mining, installation, and so on. Perhaps they don't include the energy cost of building the mining equipment and factories. However, if fossil fuels peaked and we started transitioning to renewables, we could STOP investing so much energy in new gas pipelines, new coal rail lines, new factories for turbines, and so on. These are upfront costs for ALL sources of energy.
It's not possible to find reliable information on this, because nobody does ERoEI studies which are so detailed that they include the energy cost of building mining equipment or trains. However, with both fossil fuels and PV, things like that are an upfront investment in both cases, and there is no a priori reason to assume that PV requires more transportation or mining equipment than coal (perhaps less, since PV panels are transported only once during their lifetime).
"You assume that you can change fuel availability in such a significant way, but that nothing else will change — this is an error. "
I do not assume that. There would be all kinds of changes in the economy right away. The economy would sacrifice the LEAST important uses of energy first. It would suddenly become expensive to blast your air conditioner all day long.
However, that's just beside the point. It is not relevant to the "energy trap".
"The first problem I can see, is that the reduction in availability would be mostly borne by the developed world. What might be a 6% global energy deficit, would be more like 15% in the developed world (which uses about 40% of global energy). "
If you exclude oil, the vast majority of energy worldwide is from domestic sources. The United States gets almost all its coal and gas from its own territory, and so does China, India, and most other places.
When talking about the energy trap, we are dealing with the energy used to BUILD renewables. That energy is overwhelmingly thermal energy and electricity which would come from coal and gas at the beginning, not from oil.
...Obviously, we could envision all kinds of political possibilities here. Perhaps that decline of 6% for one year would cause President Trump to go crazy and nuke somebody. Maybe there would be panic, chaos. I have no way of predicting those things. However, it is a separate issue from the energy trap. The trap itself is overcome easily enough by basic, automatic market mechanisms.
> I assume that PV has an ERoEI of 10.
Delete...Overbuilt to keep production constant year-round, and with 7 days of battery back-up?
"...Overbuilt to keep production constant year-round, and with 7 days of battery back-up?"
DeleteNo. In an energy trap, I think we would use fossil fuel plants in a load following manner. Even coal plants can be used in that way, to some extent. I don't think there would be any overbuilding or battery back-up until renewable penetration reached a certain level (perhaps 20%), at which point we'd already be well out of the energy trap.
Here is the python source code.
ReplyDelete(NOTE: I replaced tabs with dollar signs, which you can revert by typing "cat yourFileName.py|tr '$' '\t > yourNewFileName.py'" on a Mac command line, or you can do it manually in a text editor. I had to replace tabs with dollar signs because white space is important in python but there is no way to include tabs in a blogger comment. You can run the source code by installing python and typing 'python yourFileName.py' at a command prompt).
The code is as follows:
# This is free and unencumbered software released into the public domain.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
# IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR
# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
# OTHER DEALINGS IN THE SOFTWARE.
import math
#
# You can change these parameters
#
eroei_pv = 10.0
eroei_ff = 15.0
run_years = 30
# Number of years before a PV plant expires
pv_lifetime = 30
# The standard deviation of the Gaussian decline curve
std_dev = 30
# The fraction of FF energy investment which is a recurring cost instead of an
# upfront cost for building FF power plants. Even fossil fuel plants require
# SOME upfront cost, which will cease when FF start declining
fraction_ff_invest_operating = 0.9
# How many rows to print; one every n years
print_every_n_year=1
# How many years before decision-makers discover that FF are on a permanent
# decline so don't invest in new FF plants and start investing in PV
figure_out_year=7
#
# Do not change these
#
orig_net_ff = 1.0 - (1.0/eroei_ff)
invest_pv = 0.0 # initial PV investment is 0
gross_ff = 1.0 # initial FF amount extracted is 1
gross_total = 1.0
old_pv_investments=[] # This is used to keep track of and remove PV panels which have expired due to age
def gaussian_curve(num, std_dev):
$return math.exp(-(math.pow(num, 2))/(2*math.pow(std_dev, 2)))
# Print table header
print "year gross_ff gross_pv gross_total net_total invest_pv invest_ff fraction_original_net"
for year in range(0,run_years):
$gross_ff = gaussian_curve(float(year), std_dev)
$# After it's discovered that FF are on a permanent decline, stop investing
$# so much in new FF plants and extraction
$if (year < figure_out_year):
$$invest_ff = 1.0 / eroei_ff
$else:
$$invest_ff = (gross_ff / eroei_ff) * fraction_ff_invest_operating
$# After it's discovered that FF are on a permanent decline, start investing
$# in PV
$if (year > figure_out_year):
$$if (year < std_dev):
$$$invest_pv = (1.0 / eroei_pv) / 2.0
$$else:
$$$invest_pv = (1.0 / eroei_pv)
$# Old PV panels are expired after pv_lifetime
$old_pv_investments.insert(0,invest_pv)
$if (len(old_pv_investments) > pv_lifetime):
$$old_pv_investments.pop()
$# Add up all PV contribution from all panels in the last n years which
$# are still operating
$gross_pv = 0.0
$for old_pv_inv in old_pv_investments:
$$# PV panels generate this much each year
$$gross_pv += old_pv_inv * eroei_pv / pv_lifetime
$# Sum totals
$gross_total = gross_ff + gross_pv
$net_total = gross_total - invest_ff - invest_pv
$# Print chart
$if (year % print_every_n_year == 0):
$$print "%2d %9.4f %9.4f %9.4f %9.4f %9.4f %9.4f %9.4f" \
$$$% (year, gross_ff, gross_pv, gross_total, net_total, \
$$$$invest_pv, invest_ff, (net_total/orig_net_ff))