Skip to content

💬 "The Cost of Decarbonization"

The Energy Transition Show

Photo by Sharon McCutcheon / Unsplash

Table of Contents

Host: Chris Nelder
Guest: Dr. Matthew Ives | Senior Researcher | Oxford University
Category: 💬 Opinion

Podcast’s Essential Bites:

[4:40] “How can we model technological progress? […] It's actually possible to model the improvement rate for any given technology reasonably well, if you have at least five years of experience on that technology. […]. Most people have probably heard of Moore's Law, which is applied to the exponential cost declines of computing power. So we've applied something fairly similar to that. It's another empirical law known as Wright’s Law, which relates the cumulative production of a particular technology as a proxy for experience to cost declines. These are consequently known as experience curves or learning rates. […] Most technologies don't actually improve much through time when they're inflation adjusted. But there are those that do like computing and genomics and what we're finding with renewables. So they can be modeled fairly well by a variant of either Wright’s Law or Moore's Law.”

[5:59] “One reason why we're using Wright’s Law is that it has some fairly obvious policy implications that the more you deploy a technology, the cheaper it becomes. So once it's established, the technology appears to be quite persistent in their approvement rates. But one of the other key points is that technologies change at very different rates […] through time as you deploy more. […] We've provided the ability to estimate a probability distribution associated with those future technology costs. So we go back in time and look at how well we could have predicted what happened through time on a particular technology. And from that, we can come up with a probability distribution associated with that cost. And found that to be that the past record of a particular technology is the best forecast for the future cost declines of that technology.”

[7:05] “We tested this method out for about 50 different technologies, from chemicals to energy to hardware, optical fiber, transistors. And the results confirm pretty well that the predictions are robust and reliable. So we've come up with a methodology that can make good predictions for long term growth of established technologies. Finally, the first author on the paper Rupert Way, led the application of this methodology to the energy transition. So modeling endogenous technological growth in energy systems has actually been around for quite some time. But what we've done is novel in that we've applied empirically grounded probabilistic technological forecasts that I've just described, to that global energy system model. And it was a model that was custom built to really focus on the key role that technological progress can make in our energy transition.”

[8:46] “There's quite a few technologies out there that cost haven't changed through time […], which is materials and minerals and what we exploited out of the ground like coal and fossil fuels in general. They've bubbled around at the same costs when you adjust for inflation for the past 100 years. […] There's an explanation for that in terms of the fact that you've got to find more and more resources and go to more and more trouble and improve your technologies to be able to bring those harder to reach resources down to the cost that you used to in the past. […] In contrast, technologies, like solar, wind and batteries have been dropping at roughly 10% per year for the last 30 years. Whereas nuclear power for instance, it's remained fairly stagnant or even risen in a number of countries mainly around safety concerns. So contrast that with solar prices that have fallen by a factor of 1,000 over that same time”

[10:40] “The International Energy Agency […] get a bit of a beating in our research, but it's somewhat justified. And it's not just us, quite a few people have kicked up a fuss about their forecasts because they've been consistently high in terms of the cost that they predicted, and also the subsequent deployment of these technologies for at least 20 years. So there seems to be a systematic bias in what they're doing. They've never been over optimistic about the prices […]. And as I said, they've been fairly well criticized already. But that doesn't seem to change their methodology, which to me is somewhat surprising. They have updated their costs to what they are now. And so now they're starting to say, solar is some of the cheapest electricity in history. But in terms of their modeling, because they don't include as much cost declines and as much deployment in their models, their long term predictions for these technologies have systematic bias in them, that prevents them from seeing the kind of low cost future that we're seeing in our modeling.”

[12:51] “[The cost of energy transition] is fairly clear from the work that we've done, which is based on probabilistic technological forecasts. So there is a probability distribution associated with each of the technologies in the model. And our modeling where we've taken two scenarios. […] One in which we keep the energy system in a similar sort of energy mix as we have it today, dominated by fossil fuels, which is what we call the no transition or stalled transition scenario, […] which is also now called the worst case scenario. […] Interestingly, when you look at it from a technological point of view, that scenario seems close to impossible, because it relies on us actually unlearning most of what we've already learned about these new technologies, which seems highly unlikely.”

[13:56] “Contrast that to the decisive transition scenario, very fast transition scenario, where we basically just take the current deployment trends, which are exponential in their growth of these renewable technologies, progress them forward at that same rate for the next 10 years, and then just taper them off to the background kind of economic growth that we assume in the model. And then we take the probability of the costs of those systems and compare those two scenarios and find that the decisive transition, the mean of the probability distribution, has it about $14 trillion cheaper over the course to 2070 that we model it. But the median because the distribution is a log normal, that's probably a better test of the cost difference, that's about $26 trillion cheaper. So […] it makes economic sense to do this transition.”

[18:55] “We compare that fast transition, where we take the rates of change of these key technologies, solar wind batteries, and electrolyzers and progress them forward at the same rate for the next 10 years before we taper off and they start to dominate the system. And that fast transition basically decarbonizes the entire energy system within 25 years, and over the period to 2070 ends up saving us trillions of dollars. So we get to, to go to enormous lengths in terms of close to 80% of our emissions dealt with and we save money doing it. I mean, it's an absolute no brainer. There's going to be a lot of work done in ramping up our capabilities of bringing renewables and electrifying everything. But in terms of the economics of it, it's a bit of a no brainer based on this analysis that we've done comparing that fast transition to the no transition scenario.”

Rating: ⚡⚡⚡

🎙️ Full Episode: Apple | Spotify
🕰️ 23 min | 🗓️ 11/10/2021
✅ Time saved: 21 min

Additional Links:
Paper: “Empirically Grounded Technology Forecasts and the Energy Transition”