The good news today comes from three surprising sources: a consumer strike in Pakistan, agrivoltaics trials in Africa, and AI weather forecasting from tech monster Google.
“Declining solar panel prices, coupled with skyrocketing grid electricity tariffs that have increased by 155% over three years, are fuelling a rush in renewable energy adoption in Pakistan, with solar power leading the way. The country is now the world’s sixth-largest solar market. Interestingly, this shift toward solarization has happened largely without active political will, driven instead by external pressures. China's overproduction of solar panels has lowered costs, making Pakistan the third-largest destination for Chinese exports. Industrial, agricultural and residential sectors have embraced solar, with imported Chinese modules totalling 13 gigawatts (GW) in the first half of the year, and forecasts reaching 22GW by year-end.”
For reference, Pakistan’s total solar market in 2023 was only1.3 GW, India’s 7.5 GW. the USA’s 26 GW. The tidal wave wasn’t planned, and has thrown the expensive, badly run and coal-dependent legacy grid into a financial tailspin. It is in effect a mass consumer strike. Blocking new solar imports would be politically dangerous, disconnecting existing panels suicidal. The disruption surely wasn’t planned by China either. Pakistan has been a key target for the soft imperialism of the Belt and Road initiative, but initially it took the form of selling high-cost coal and lignite power stations. It clearly also opened the door to the solar exporters, and now Chinese banks are stuck with a lot of bad debt from the coal plants.
For once, I’m with the neoliberal Davos fundies on this. You no longer need competent government policy for solar to take over. In fact to stop it you would need a hostile government both exceptionally competent and exceptionally determined. FLOP-29 doesn’t matter much.
An important side-issue is installation. I haven't seen any reporting on how this is going in Pakistan, but we can make an educated guess. Assume that the 1.3 GW of 2023 was put up by conventionally trained labour. This cannot have been available for the 22 GW of 2024. Most of it must have been put up by handymen – especially skilled workers in a vaguely related trade - without special training. Perhaps they relied on locally printed “Solar for Dummies” pamphlets in Urdu, perhaps the Chinese suppliers provided wordless IKEA-type instructions in the box. Doubtless there have been a few fires and electrocutions, but there is no sign that the problem has reached major scale. Solar PV installation in rich countries has been over-engineered and over-regulated.
The Pakistani grid will of course need large amounts of storage as it shifts messily from baseload coal to daytime solar. Luckily, the country has vast areas of high mountains including the world’s second highest, the 8,611m and strangely named K2, or, as locals say, Ketu. http://www.jameswimberley.es/Blog%20posts/2008/ketu.php.htm
Agrivoltaics in Africa
A high-quality controlled field study by researchers from Sheffield University at two sites in Kenya and Tanzania has confirmed that agrivoltaics works fine in semi-arid tropical conditions. https://pdf.sciencedirectassets.com/271969/1-s2.0-S1364032124X00167/1-s2.0-S1364032124007925/main.pdf (via PVMagazine). Their two near-identical model plots have 50% shading, N/S orientation, a 3m high modular steel canopy framework, and a 10% panel slope for rainwater runoff and harvesting. One of the two test sites was connected to a battery and microgrid, the other to the regular grid. Yields of beans doubled on the PV plots against the control ones. Unless you are an agronomist, you don’t need to know about the other six crops tested. Only a tiny fraction of total farmland will ever be put under panels, and farmers will simply plant the crop that works best for them.
Can this be generalised? Both sites were at agricultural colleges, with exceptional access to skilled workers for maintenance and data gathering. The trials used selected inputs of natural fertilisers and biopesticides, perhaps not widely available, but that should not affect the critical comparison with the control lots. These are not showstopping difficulties.
The authors do not draw one radical conclusion, but I will. Look at the overhead photos: https://scontent-mad2-1.xx.fbcdn.net/v/t39.30808-6/467643582_2758702030958562_6850032698025508563_n.jpg IMHO they have unintentionally hit on a simple design that lends itself to easy replication by semi-skilled workers. Why not just copy this? Forget about bespoke designs and site-specific optimisation, go with the good-enough 50% shading, jump to scale and secure quick economies.
A village handyman (say the local moped mechanic) should be able to copy most of the setup. The one operation that requires trained labour is the electrical wiring of the microgrid. PV panels output DC not AC current, and are live as soon as they are in position: somewhat risky for amateurs, but the Pakistani example tells us the risk is manageable. Much of the work – cutting cables to length, fixing them to the mounts, guttering, building a shed for the inverters, etc - is within the scope of the village handyman, or you and me. The electrician needs I would guess a day at the office in town to design the system and print the bill of parts, and a second day on site to hook everything up.
The study does not include an economic analysis, but with panels at 10c per watt wholesale, a solid finding from many studies that agrivoltaic crop yields are at least equal for at least some crops, and massive unmet demand for electricity in rural Africa, it would be amazing if the scheme were not profitable over a very wide area.
Adam Smith tells us that the self-interest of African farmers, Chinese suppliers and global bankers can do the rest. I suggest Western do-gooders like us concentrate on two critical bottlenecks: pressing policymakers to emulate fair regulations (as in Australia) for feeding in decentralised electricity to the grid, and accelerated training of solar electricians.
Google’s AI weather forecasts
The whiz-kids at Google’s DeepMind skunk works in London have been busy.
“GenCast provides better forecasts of both day-to-day weather and extreme events than the top operational system, the European Centre for Medium-Range Weather Forecasts’ (ECMWF) ENS, up to 15 days in advance.”
This is a probabilistic global forecasting model. It does not need a supercomputer like ECMWF’s (it urgently needs a better name), and runs on cheap cloud computing services like, um, Google’s. The code is patented but anyone can download it, and there is no hint of monetizing the forecasts. Of course Google expect to make money from other related AI products.
The good news today comes from three surprising sources: a consumer strike in Pakistan, agrivoltaics trials in Africa, and AI weather forecasting from tech monster Google.
World Economic Forum on Pakistan:
https://www.weforum.org/stories/2024/11/pakistan-solar-power-energy-transition/ ;
“Declining solar panel prices, coupled with skyrocketing grid electricity tariffs that have increased by 155% over three years, are fuelling a rush in renewable energy adoption in Pakistan, with solar power leading the way. The country is now the world’s sixth-largest solar market. Interestingly, this shift toward solarization has happened largely without active political will, driven instead by external pressures. China's overproduction of solar panels has lowered costs, making Pakistan the third-largest destination for Chinese exports. Industrial, agricultural and residential sectors have embraced solar, with imported Chinese modules totalling 13 gigawatts (GW) in the first half of the year, and forecasts reaching 22GW by year-end.”
For reference, Pakistan’s total solar market in 2023 was only1.3 GW, India’s 7.5 GW. the USA’s 26 GW. The tidal wave wasn’t planned, and has thrown the expensive, badly run and coal-dependent legacy grid into a financial tailspin. It is in effect a mass consumer strike. Blocking new solar imports would be politically dangerous, disconnecting existing panels suicidal. The disruption surely wasn’t planned by China either. Pakistan has been a key target for the soft imperialism of the Belt and Road initiative, but initially it took the form of selling high-cost coal and lignite power stations. It clearly also opened the door to the solar exporters, and now Chinese banks are stuck with a lot of bad debt from the coal plants.
For once, I’m with the neoliberal Davos fundies on this. You no longer need competent government policy for solar to take over. In fact to stop it you would need a hostile government both exceptionally competent and exceptionally determined. FLOP-29 doesn’t matter much.
An important side-issue is installation. I haven't seen any reporting on how this is going in Pakistan, but we can make an educated guess. Assume that the 1.3 GW of 2023 was put up by conventionally trained labour. This cannot have been available for the 22 GW of 2024. Most of it must have been put up by handymen – especially skilled workers in a vaguely related trade - without special training. Perhaps they relied on locally printed “Solar for Dummies” pamphlets in Urdu, perhaps the Chinese suppliers provided wordless IKEA-type instructions in the box. Doubtless there have been a few fires and electrocutions, but there is no sign that the problem has reached major scale. Solar PV installation in rich countries has been over-engineered and over-regulated.
The Pakistani grid will of course need large amounts of storage as it shifts messily from baseload coal to daytime solar. Luckily, the country has vast areas of high mountains including the world’s second highest, the 8,611m and strangely named K2, or, as locals say, Ketu. http://www.jameswimberley.es/Blog%20posts/2008/ketu.php.htm
Agrivoltaics in Africa
A high-quality controlled field study by researchers from Sheffield University at two sites in Kenya and Tanzania has confirmed that agrivoltaics works fine in semi-arid tropical conditions. https://pdf.sciencedirectassets.com/271969/1-s2.0-S1364032124X00167/1-s2.0-S1364032124007925/main.pdf (via PVMagazine). Their two near-identical model plots have 50% shading, N/S orientation, a 3m high modular steel canopy framework, and a 10% panel slope for rainwater runoff and harvesting. One of the two test sites was connected to a battery and microgrid, the other to the regular grid. Yields of beans doubled on the PV plots against the control ones. Unless you are an agronomist, you don’t need to know about the other six crops tested. Only a tiny fraction of total farmland will ever be put under panels, and farmers will simply plant the crop that works best for them.
Can this be generalised? Both sites were at agricultural colleges, with exceptional access to skilled workers for maintenance and data gathering. The trials used selected inputs of natural fertilisers and biopesticides, perhaps not widely available, but that should not affect the critical comparison with the control lots. These are not showstopping difficulties.
The authors do not draw one radical conclusion, but I will. Look at the overhead photos: https://scontent-mad2-1.xx.fbcdn.net/v/t39.30808-6/467643582_2758702030958562_6850032698025508563_n.jpg IMHO they have unintentionally hit on a simple design that lends itself to easy replication by semi-skilled workers. Why not just copy this? Forget about bespoke designs and site-specific optimisation, go with the good-enough 50% shading, jump to scale and secure quick economies.
A village handyman (say the local moped mechanic) should be able to copy most of the setup. The one operation that requires trained labour is the electrical wiring of the microgrid. PV panels output DC not AC current, and are live as soon as they are in position: somewhat risky for amateurs, but the Pakistani example tells us the risk is manageable. Much of the work – cutting cables to length, fixing them to the mounts, guttering, building a shed for the inverters, etc - is within the scope of the village handyman, or you and me. The electrician needs I would guess a day at the office in town to design the system and print the bill of parts, and a second day on site to hook everything up.
The study does not include an economic analysis, but with panels at 10c per watt wholesale, a solid finding from many studies that agrivoltaic crop yields are at least equal for at least some crops, and massive unmet demand for electricity in rural Africa, it would be amazing if the scheme were not profitable over a very wide area.
Adam Smith tells us that the self-interest of African farmers, Chinese suppliers and global bankers can do the rest. I suggest Western do-gooders like us concentrate on two critical bottlenecks: pressing policymakers to emulate fair regulations (as in Australia) for feeding in decentralised electricity to the grid, and accelerated training of solar electricians.
Google’s AI weather forecasts
The whiz-kids at Google’s DeepMind skunk works in London have been busy.
Press release https://deepmind.google/discover/blog/gencast-predicts-weather-and-the-risks-of-extreme-conditions-with-sota-accuracy/, incomprehensible Nature paper https://www.nature.com/articles/s41586-024-08252-9 :
“GenCast provides better forecasts of both day-to-day weather and extreme events than the top operational system, the European Centre for Medium-Range Weather Forecasts’ (ECMWF) ENS, up to 15 days in advance.”
This is a probabilistic global forecasting model. It does not need a supercomputer like ECMWF’s (it urgently needs a better name), and runs on cheap cloud computing services like, um, Google’s. The code is patented but anyone can download it, and there is no hint of monetizing the forecasts. Of course Google expect to make money from other related AI products.
Five days more warning of extreme weather events could save a lot of lives as the climate continues to worsen. If the policymakers are listening, which does not seem to have been the case before the floods in Valencia. Lunch or “lunch” came first. https://www.independent.co.uk/news/world/europe/spain-floods-valencia-rain-lunch-b2646278.html