Guest Post: Thoughts on a fossil carbon-free base for heavy industry
Guest post by Ryan Richter
At this point I hope we have all imagined the possibility of a future with economic prosperity and without fossil carbon-based motor fuels. You can use electricity and renewable sources for most things. OK, but what about everything else? Motor fuels and fossil carbon electricity generation aren’t the whole story by a long shot. There’s steel and aliminum, oh and also cement. And then there’s organic chemistry more generally – can’t you do that without burning the oil? Well, no, not today you can’t. This is what I want to take the knife to. We have the tools, but I don’t hear many people talking about how to really use them to full effect.
I’m talking about nuclear power of course. We certainly want to take the fullest advantage of solar power and other renewable sources as well. I want to additionally propose the use of radioisotope heater units to replace a number of unobvious carbon sources in today’s heavy industry. Nuclear advocates have recommended the reprocessing of fuel to extend the nuclear fuel cycle in the past, but relatively little has been done so far. This lackadasical effort needs to end yesterday, of course. I fully recommend an expanded nuclear fuel cycle including reprocessing and fast breeder reactors, and maybe also a thorium cycle. But in addition to that, I want to re-propose an old idea (I don’t know where it originates) in a new context – not only extracting and fully using the actinides in the fuel cycle, but also extracting the fission products to make Fission Product Radioisotope Heater Units, FPRHUs.
Now let me change tack for a bit and talk about some very nasty fossil CO2 sources in heavy industry today. We use coal to make iron and steel. We use petroleum coke to make aliminum. We have lime kilns which release the CO2 from carbonate rocks, which is perhaps the very worst. And then there’s the awful oil-powered oil refinery. If you’re not familiar with the industry (I have never been any part of it), you may not have heard any of the details, so let me try to explain from what I’ve been able to piece together.
My explanation of the oil refinery will focus on what I consider to be the two fundamental processes in a refinery that’s not dedicated to motor fuels, which is a difficult thing to imagine based on what engineers build today. One is called the hydrotreater, which is an industry name for the high-pressure hydrogenation of hydrocarbons[1][3][4][5]. The so-called naphtha hydrotreater seems to operate somewhere around 100 atmospheres and maybe 300-400 celcius ([1][3][5] – this is difficult. According to [1] and [3] the hydrocarbons are pressurized to ~40atm and by [5] the H2 to ~200atm, which makes sense as H2 is cheaper to compress. A similar process is Hydrocracking [9] which is around 120atm). There is also a diesel or lube oil hydrotreater which operates at more extreme conditions [5] and processes a smaller feed. These hydrotreaters are fed with hydrogen from a reaction which overall is basically
CH4 + 2 H2O -> 4 H2 + CO2 (where does the CH4 come from? – not natural gas)
The second fundamental process is called the olefins unit pyrolysis furnace [8], which operates at more like 30 atmospheres and maybe 800 celsius. This is simply the thermal decomposition of hydrocarbons, in this case only the lighter naphtha mixture. The feeds to both of the fundamental precesses are highly miscellaneous mixtures of hydrocarbons, although not exactly the same. The hydrotreater in effect comes first, after the main fractionation (called the crude unit [1][2][7]). The feed from the crude unit typically still includes organic sulfur and nitrogen, and the hydrotreater is the sledgehammer for dealing with that, as sulfur comes out as H2S and N as NH3. But the other effect of the hydrotreater is to add hydrogen across carbon-carbon double bonds to make single bonds, although this makes it sound far more surgical than it is. On the other hand, the pyrolysis furnace in effect cooks hydrogen out – how many things can a hydrocarbon decompose to? So in that one you make double bonds, and also triple bonds, and some coke too no doubt, and some hydrogen, and also some methane. Well, more than a little bit in fact. The hydrotreater is also a prodigious producer of methane, being a hydrogenator after all. So that’s where the methane comes from. As it happens, only a little of it is needed to make the hydrogen that’s needed. A third process is called Hydrocracking [9], essentially a combination of the above two in one reactor, said to occur around 450 Celsius and 120 atmospheres.
So neither of the fundamental processes is especially chemically efficient. And if you’ve looked at the reaction conditions you can see that some thermal and mechanical power inputs are required also, not a small amount. What a horrible mess. How can you ever make money doing something like that – OH WAIT. You can burn the methane. So at the inlet to the hydrotreater there’s a jet engine the size of a large truck burning, in effect, the same methane that’s cooking out of the reaction. The naphtha hydrotreater at a medium size refinery costs over 200 megawatts to run (I swear I saw this number somewhere but now I can’t find it again) – but not in money. The pyrolysis furnace is a very similar story. This is where ethylene for plastics comes from. And the big embarrasment is that you have to send the same oil to the hydrotreater fisrt even though the pyrolysis furnace undoes the single bond-double bond thing. The cruse unit – the tallest tower – also has a house-sized burner fed by the so-called refinery fuel gas mix. All the other hundreds of other smaller towers also have their fuel gas-powered preheaters or compressors. It’s said that you can turn crude oil into anything, but it you want to turn all of it into polyethylene you have to, basically, burn half or more of it first, counting carbon atom by carbon atom. But in every case what you need is a concentrated heat source, and so I propose the FPRHU. There is a question about the effect of ionizing radiation on the chemical process, but I have a surprising answer for that.
What’s needed is a convenient high-temperature heat transfer material, and I have no idea what chemical engineers would propose based on that criterion alone. But I will propose the use of molten silicates for another reason. The Radioisotope Silicate Foundry (RSF) can be the basis of a carbon-free construction industry. The idea is that the RSF is located at a nucear fuel reprocessing facility and uses its freshest isotopes to produce both a high-temperature heat source for co-located heavy industry and a precast building material to replace concrete. This may not sound remotely believable based on the scale of recently-imagined nuclear reprocessing efforts, but I propose a great increase in that. Another benefit here is that all of this activity, together with the complete nuclear fuel cycle for actinides, reduces the final amount of nuclear waste produced for each ton of uranium mined.
It starts to become clearer what the overall picture looks like. Let me say a little bit about the metals first before sharpening the chemistry picture a little more. You need a reducing agent to make metals. One possibility is a electrical/nuclear biomass charcoal furnace. What biomass you would want and how much would be available is an interesting question I don’t know much about, as the chemical quality of the charcoal is important. An interesting alternative is hydrogen produced either by electrolysis of water or the pyrolysis of fossil carbon methane, not yet fully developed.
The steel mill coke ovens also produce a coal gas which is now called coke oven gas [6], and is used in the steel mill to het furnaces much the same as the oil refinery fuel gas. This would also be replaced by FPRHU/RSF. Preparing aluminum ore is mainly mechanical and wet chemistry, but iron ores are roasted and that’s a nut to crack. Heat is no problem with abundant FPRHUs, but some carbonates are decomposed here also. This may be on a scale that can be accepted if all other main sources of fossil CO2 are really eliminated. The other possibility is to treat carbonates with a wet chemistry, i.e. to use an acid to release CO2 and then a mineral base to absorb it. That would require support from the chloralkali process, more or less.
Finally, what would carbon chemistry look like? You will still need things like oil and coal to produce a number of things including asphalt and lubricants. Polymers and chemical feedstocks are also needed. Some of this can come from biomass, e.g. you can easily make ethylene from ethanol, although competing with food sources is not necessarily a good idea. Some real new chemical enginnering is needed, to say the least. I would presume butadiene still comes out of oil and coal. You can picture a refinery that processes heavy inputs with FPRHU/RSF heating, and which uses pyrolysis to destroy the methane which is the ultimate waste product. The hydrogen can be used liberally for anything, making iron, whatever. The carbon black has to end up in steel, tires, carbon fibre materials, and on printed pages I guess. There’s another interesting engineering question as to the overall balance between electricity usage and FPRHU, the point of which is that the FPRHUs need to be available on a real market. So in a little more detail, in a nuclear powered oil refinery the compressors would be powered by electricity but the feed preheat would be done by direct radioisotope heating. As I understand, the rule of thumb is that the fission products release about 10% of the heat that the original reactor produced. The thermal efficiency into electricity at a large reactor is around 40%. Waste heat at a power reactor is in the form of low-temperature steam, although some fast breeders may allow for something better – I’m not really sure.
OK, so I hope I’ve shown that it’s basically possible to do things this way, and emit only the smallest amount of fossil CO2 while making all the things we expect to make and have. The extensive handling of radioisotopes within the heavy manufacturing centers will necessitate an increase in automation, but the industry was really heading in that direction anyway. The big question is the capital expenditure. Of course the answer to that is state involvement to some degree – it’s just a question, politically, of what kind of future we want.
My data source for petrleum industry stuff are the reports of the CSB (which has an annoyingly long name), which investigates fatal accidents in the chimical industry in the US. They were created out of the EPA in the Clinton years with a subpoena power to reveal industry secrets. I downloaded personal copies of these when Trump was elected, having seen previous EPA reports disappear under Bush. This turned out to be a good idea, as the older reports are no longer linked from the main page. Other people have made archives. Report number 1 is particularly revealing. If you read only one report about a fatal accident at an oil refinery, make it that one. Number 2 is really shocking in the details and has a hilarious photo. 6 and 7 are weirdly macabre, seeming to describe the same accident at two different places.
[1] BP-Husky Toledo Refinery 2022
[2] Husky Superior Refinery 2018
[3] Tesoro Anacortes Refinery 2010
[4] Silver Eagle Woods Cross Refinery 2009
[5] BP Texas City Refinery 2005 (RHU Unit)
[6] Bethlehem Steel Chesterton 2001
If we would do this right now, how would the world look in a couple hundred years, given the damage already done until now?
I have no idea. What’s the alternative? More seriously, I don’t necessarily recommend a rising pooulation. Or even maintaining the current population, although deliberately reducing the population is delicate ideologically.
I explained earlier that I was at the Reed college reactor. It might help to explain that I was one of their extreme financial aid cases back when they had those. It was not always a pleasant experience when people assumed you were rich like everyone else there but you’re really not. I didn’t really fit in, and from that time I’ve thought there was something rotten in progressivism. I knew what bitter misogyny was from seeing the police harras my mom when I was very young (that was Texas). These people couldn’t tell men from women. And their view of economic matters affecting the working class was at least as obtuse in my opinion.
So I’m kind of tying to create a distinct sort of left-populism, or help do so as I can’t by myself. Anyway, the standard of living of the working class, or indeed the distinct existence of that class – see my other proposal – are of interest to me.
Three well-known and reliable ways by which a country’s birth rates go down are a) empowering women, b) increasing income, and c) having a reliable social safety net. It is in everyone’s interest to share wealth and power with women, to turn poorer countries into richer ones, and establish universal programs for personal security and well-being in old age. Unfortunately these are the sorts of things that get cut as “frills” by many governments when things get tight financially. Ushering as many countries through the demographic transition as possible, to bring them to a low birthrate future is good for evryone.
I’ve read a number of books by Canadian scientist Vaclav Smil https://vaclavsmil.com/, who has done a lot of interdisciplinary work on many of the basic nuts and bolts things that make the human world run. He’s written widely on energy, material, food, innovation, and their combined environmental impact. Where I started with his books was How the world really works the science behind how we got here and where we’re going.
Birthrates are going down everywhere except sub-Saharan Africa. It really is just a question of what empowered, educated women want.
Piglet: at the risk of sounding Pollyanna-ish, according to this reference, birthrates are going down even in sub-Saharan Africa; they have just not yet fallen to replacement rate there.
Ryan: thanks for triggering a moment of nostalgic recall of the 1970s advertisements from British Nuclear Fuels making a similar point using the slogan “Uranium is Useless”. The accompanying illustrations were of ornamental Uranium-green glassware and a Thorium gas mantle.
I didn’t mean to come back so quickly with something as long, but I think I found a good topic. Let me know what you think. Thanks,
The Pefect Employee
or Why AI won’t work, wasn’t really supposed to, and will take your job anyway
Without attempting a complete definition of intelligence, I think it’s possible to point to a number of things it basically needs to do. For example, to set goals and attempt to satisfy them, to notice what happens instead and then form questions based on the difference from expectations, to pursue answers to those questions and then use them to formulate a strategy for a second attempt, and so on. Oh, and also to compare recent experience based on satisfaction or non-satisfaction of goals and to re-evaluate the goals based on comaprison with past or reported experience. That might be a start. And if you lay it out like that, isn’t it kind of obvious that the prompt-and-response “AI” backed by web scraping doesn’t have any chance of getting there? So what are they screwing around with? Why are they so dedicated to this too-limited model? Because it’s the perfect employee. By having no personal viewpoint or life or expectations, or indeed intelligence, it minimizes resource consumption. All of those would cost a thousand times extra apiece. In the large and old and mature organization, blame for incompetence can always be swept away, so not actually doing the job isn’t any showstopper. It’s doing exactly what it was supposed to do, reduce your wage.
There’s a romantic notion of AI suddenly appearing that I want to push back against. I’m going based on a biological analogy. I don’t think AI is impossible or anything, just not on the scene today. But it would require more or less what I just outlined: a life and viewpoint of its own, and in particular its own goals and the ability to question their success. You would have to provide for all this in hardware and software and electricity. Its upbringing would take years instead of days. Its cost would be outrageous. And it would be stupid and whimsical and useless at performing assigned tasks, especially at first. It would basically not be a ready made economic slave. You don’t have to worry about that one taking your job, whenever it might appear, although in some far future it could in theory become something like that. But the point is that it would take forever and be the end result of a much larger effort than we’ve recently seen, and won’t so much take us by surprise, at least as an economic replacement for labor. If you care about the wages and living standards of actual workers, I think we need to be absolutely scathing about the non-humanity of paltry efforts to replace people. Recurrent neural networks with distinct training and operation phases cannot ever question their goals, which are simply inputs. It would be a shame if, by generosity, we allowed that rights of some sort are to be attributed to algorithms, essentially beacuse a lot of very rich and influential people take it very seriously. After all I think we can come up with an example where a community was bullied by very influential outsiders to grant some of their hard won rights to another sort of entity entirely under essentialy fraudulent moral pretenses… Labor unions and many left and center-left political parties have abandoned women in search of greater inclusion. Who’s next? Actually, with men being targeted as well it probably won’t be so bad. There’s another reason why the rights angle might not work, another trait the perfect employee has to have. It can’t grow up to own the business.
Economists tell us they have the situation in hand. You see, there have been many improvements to the productivity of labor throughout history, they assure us, and in the long run they have all been good for everyone. Oh, sure, ignorant workers will fight back in the short run when their jobs are threatened, like the Luddites. But a certain amount of disruption is necessary for progress. The basic story isn’t wrong at all. I was a big proponent of driverless cars, for example, before the “one algorithm fits all” hype rotted everyone’s brain. Of course, this analysis depends on having confidence that the new technology really has the capability that it promises at first. In the case of AI I think there are some evident humanistic and economic reasons to question this, reasons which haven’t applied to previous rounds of technology. Why are hardly any even liberal economists sceptical? Well, I think a lot of guys (it is a very male profession) kind of go into economics to cheer on the choo-choo trains, or anyway you’re not taken seriously if you don’t. At least they’re down on cryptocurrency. And from the “yay technology” perspective, there is an obvious way to deal with the problem posed by AI in the labor force: the universal basic income. The story goes, productivity and profits will soar as wages are cut, so you can tax the successful and make that over into an income for the unemployed. I want to warn that this will never be possible if AI can’t basically do the job. But in that case won’t it just fail and leave the scene? That depends on the exact motivation of the business owners, and on how they’re able to sell the idea to everyone else. I think there’s a real possibility that the blame can be reliably placed elsewhere, and that the desire to permanently reduce what’s called the employment share of income – to reduce all wages and pocket the difference – will motivate a new strategic campaign to make people feel sorry for big business.
I see this coming together from two strands. One is that Silicon Valley used to be a place with real ideas and promise back in the 1980s (Postscript and SCSI RAID were my favorites), but by the mid-1990s those ideas were pretty worn around the edges and the computer revolution was coming up a bit disappointing, especially as the internet became a high-rent dive bar (with some exceptions). What they came up with next was pure exploitation. Use marketing and monopoly to force your profits higher. Deliberately engender a love-hate relationship with the consumer, and above all never stop speaking your point of view in which you’re making every sacrifice to do progress and improve the standard of living of the little guy. It has worked very well so far. The other strand was a right-wing political strategy from 20+ years ago: starve the beast. This was about reducing government services, and then telling voters that government doesn’t work and that more cuts are needed. I suggest that there will be a new version of this for the private sector, and that it will work by using supply shocks – events which suddenly reduce the economy’s whole output – to garner sympathy for across-the-board wage reductions. The new beast is the labor force. In fact we have already seen a version of this when the big “free-trade” deals were being negociated. In the US, doctors and lawyers are protected by statute from international competition, the sole exception being that Canadian dentists may practice in the US. But the working class has to face direct competition from Vietnam and Bangladesh and so forth – and this was sold to voters by saying what a terribly difficult competitive environment our businesses faced on the international market. If big business caused its own supply shock by trying and more or less failing to switch to an AI workforce – the AI recession, would anyone have that kind of sympathy? It depends on how the issue was presented.
The most important economic indicator in this situation is called labor productivity, and it is equal to the value of all goods produced per hour worked. Its numerical value has increased over time, but very unsteadily. One reason for this is that, the way it’s measured, a lot of things can affect its value. Economists know, for example, that productivity rises going into a recession because of something called a composition effect. The workers who are let go first are the least productive – walmart greeters and so forth as the story goes. The same effect will likely disguise the fact that AI doesn’t basically work as it’s initially adopted. Imagine you have a business making food and you buy produce. You have to hire and send a buyer to the wholesale market. But maybe you and the wholesaler can both get rid of your specialized agents and let an algorithm look at security videos or something. The same produce will still basically be bought and sold, only its quality will likely decline. But this misfortune will just be passed on to the consumer. Won’t competition prevent this? Heh, maybe. Look at how newspapers are using AI to write news. Consumers don’t like it, but the whole industry is heading in that direction and doesn’t seem to care about consumer opinion yet. Get ready for more strongarm, but also more bystanders cheering on the choo-choo trains. The produce buyer will probably eventually need to re-hire someone to back up the AI once some entirely rotten loads get through. But this person will be part time and will only work with the computer, not the produce (except when the computer is down entirely). The labor productivity will have risen as a result, and this will be anounced as a success. That’s the thin end of the wedge. The rest is just more of the same. This nonsense about kicking the computer when it’s wrong is the concept for the new all-low-wage world of employment.
With wages down and also the quality of goods and services, something would have to give. Since the productivity numbers are rising, at least temporarily, economists will recommend the universal basic income. You tax the producers and pay everybody a monthly check. The only question is what tax rate and how much a month? The fact is that even a modest UBI will require taxes far in excess of anything ordinary businesses have been charged domestically before, indeed it will be similar to the current high tariff nonsense. Pretty soon the business environment will deteriorate, and the taxes will be blamed first regardless of what else is happening. The political right and center will never be able to imagine continuing such a regime, even when poverty is the inevitable result. A UBI would certainly be too small to start with, and I think there is absolutely no possibility of it being increased if once implemented. The new strategy of starve the private sector beast will essentially ensure this. Of course without any economic response possible, democracy would have to give way to a feudal or slave system in which each poor person was some rich person’s responsibility, employed or not. Or just have a big die-off. There is another alternative within democracy, which is the outright redistribution of wealth, in particular my (or Thomas Skidmore’s) idea of a death to young adulthood system which need not place any financial burden on business. It does, however, result in a redistribution of the ownership of business. This gradual flattening of the demographic distribution of corporate ownership is also my answer to how to get big business out of the attractive nuisance of AI altogether. If you read my previous piece on decarbonization, you might wonder how to get industry to stop what they’re doing today, and my answer would be the same. My version of this idea, the National Inheritance or Classless Liberalism, can be gradually implemented by legislation and constitutional amendment. It’s based on the idea that there is no morally acceptable reason for the working class to exist if there is a realizable alternative in which everyone is in effect in the ruling class.
Incidentally, the villain here isn’t AI, it’s the human desire to reduce wages and forget the rest. It was just that the AI hype machine succeeded in the right way. The “human vs machine” angle is not quite the right one to take. “The economics of humanity” fits better I think. To put it another way, you can have an AI, but an economical AI isn’t an I. There is a certain perspective which will retort that things like questioning, individuality, humanity etc. may be luxuries we simply can’t afford. To which I would ask, is this concept of economic efficiency still a good if it enatils the exclusion of humans from the economy?