Jeremy Keller (@_jqkeller) header
Jeremy Keller (@_jqkeller) profile picture

Today: sabbatical In the past: Frontier CDR, Thermal Batteries, Efficiency, Waste Water, Biogas, H2, Industrial Heat Pumps, Biomass, HVAC, Solar, BESS, Etc.

Seattle, WA
Joined January 2017
Photo by Jeremy Keller (@_jqkeller)Photo by Jeremy Keller (@_jqkeller)
As an energy transition founder: we need permitting reform now. Electrification is colliding with huge demand from AI. We need to make it easier to build the power infrastructure required. To my friends on the climate left: don’t wait for a better deal. Pass BAAJA.
5
14
1
82
5,173
THE WORST PROJECT OF ALL People always ask me, "Mikey, what's the worst project you've ever had to work on?" and I always answer, I'm glad you asked that question. Good news: my company got the NASA contract to provide liquid hydrogen ("LH2") to fuel the Space Shuttle. So, we needed to build a LH2 plant. We already had one called NOLA LH2, which was in New Orleans. Now, NOLA LH2 had some, let's say, "troubles." What kind of troubles, you might ask. Well, I'll tell you what kind of troubles. Consider hydrogen for a moment. This substance has the smallest, tiniest, teensiest molecules. You leak check a system using helium, but test it with hydrogen? Leaks like a sieve. Consider also that hydrogen is gonna leak no matter what. It will pass through steel like a ghost through a wall. It's called "permeation." And hydrogen, when it hits normal air, lights itself on fire (probably from static electricity). And a hydrogen fire is invisible. So at the truck loading area, there are so many instances of hydrogen leaks that operators walking in the area to do their duty would have to wave a straw broom in front of them. Straw broom explodes when hydrogen hits it. So, if your broom blows up, don't walk there. Now, before the liquid hydrogen goes into a truck, it passes through a filter. It's a cylinder about the size of one of those 5-gallon containers you see at Lowes or Home Depot. It's got to be cleaned before loading, see. But you can't just open it up because if you do, all those 5 gallons of hydrogen and the volume in the piping to the filter will catch fire and it will be bad. In point of fact, it would explode. So first, you purge the system with nitrogen, the inert gas that composes 78% or so of normal air. That nitrogen sweeps through the piping and filter and takes all that volume that used to have hydrogen and puts it out to an elevated vent pipe (a "flare"), where it burns off harmlessly. After the nitrogen purge, you can safely open up the top lid of the filter to get the basket out to clean it. So it's a normal Tuesday. The operator, let's call him Bub, purges out the filter with nitrogen, then opens the lid. Now, the filter is at nipple-level. And when the lid opens up, Bub gets a facefull of hydrogen. And thanks to skin oil, hydrogen + air's oxygen + skin oil = explosion. After emergency medical treatment, Bub no longer had a face. It was all scar tissue. They put in one tube where his nostrils used to be and one tube in where his lips used to be. Now he could breathe and get nutrition from a liquid diet. But he had no eyes, no eyebrows, no nose, no lips, no chin. Just this mass of blotchy red and black scar tissue. He would have made a great monster in a horror movie. I was told that the lawsuit settlement was rather large. How much is your face worth? Over ten million clams, I heard. First, though, he had to assert he'd correctly purged out the filter, but the company took one look at where his face used to be and correctly concluded that even with a perfect legal case, any jury would award Bub far more than a settlement would. So, here's ten million smackers - hand over your face. There were other troubles. NOLA LH2 was leaking badly everywhere. It was at EOL - that means end-of-life. It would limp on while we built our LH2 project in Pace, Florida. Now, Pace, Florida is a few houses and a Taco Bell. We would build on a chemical plant that our company owned. But the chemicals guys (1) hated the gas guys and (2) reported to different management. To their credit, the chemicals guys - acting as an internal client - were exacting about safety. They were safety first and last. The hosting plant site, a few blocks over, had bullets of high pressure toxic liquid that once caused another monster to be created when a relief valve leaked and they called for a volunteer to shut its isolation valve and stop the leak. See, this chemical was so toxic that the vapor cloud would kill the neighbors. This stuff was nasty. You know when Aunt Maude had that dead rat trapped behind the stove? And it made the kitchen smell of death and decay? Same smell this stuff had. One drop could be smelled 20 miles away. When the volunteer - who had no children or pets and whose parents had passed on - went to isolate the valve, it destroyed most of his central nervous system and melted off most of his skin (there had been no time to get him into a steam suit). This guy didn't sue - he just wanted a job for life. I first met him when he lurched into the office I'd been given and croaked, groaned, moaned and growled. No one had prepared me for this guy, who was the mail delivery guy. Terrifying. So, justifiably, the chemicals guys were safety Nazis. And proud of it. But they stupidly tried to force their chemical plant standards on us. Just one example out of a hundred - when they had a piping flange, you know, to connect one pipe to another, they'd put in a "spectacle blind." This is a figure-8 made of steel. One circle of the 8 has the center part of the steel cut out to match the inner diameter of the pipe. The other one is uncut, just a circular plate. The idea is, if you see a circle? That means the plate is bolted into the pipe flange to isolate the pipe. If you see a plate, that means the pipe is fully connected. Simple and elegant, yes? So they insisted our prefabricated piping lengths (called "spools") have, at their connection points to connect one spool to another, a spectacle blind. We tried to tell them that LH2 is within a couple dozen degrees of absolute zero. If you ran it past a spectacle blind, you'd form an iceball the size of a Mercedes van and block the pipe. It would form its own "freeze seal." Our spools connected with bayonets. In LH2 piping, it's a pipe within a pipe. Between the inner stainless steel pipe and the outer carbon steel pipe (the "jacket") is a vacuum. You put a vacuum pump on it and suck out all the air. That vacuum keeps the stuff inside the inner pipe insulated. You connect one spool to another spool by having the outer jacket neck down and make a cone about 6' long. That cone ("bayonet") gets inserted into a matching shape in the next spool. This way, you minimize heat entry to the pipe. The chemicals guys refused to allow our VJ (vacuum jacketed) pipe to use bayonets and insisted on spectacle blinds with conventional flanges. We ignored them and went about our business, and then they forced a stop work order. We essentially had an internal "lawsuit" to involve the CEO himself to explain to the VP Chemicals that his division was being stupid. That became a month delay, and all the while management was breathing down our throats about being behind schedule. Oh yeah, whoever came up with the schedule should have been whacked with a big stick. And budget, too. In fairness, how do you originate the schedule and budget for something you've never built and no one else has built? Same reason the moon shot and the Space Shuttle were over budget. We were a year late and our cost forecast was double the budget. So, yeah, every monthly meeting with senior management was just a whipping session. I got sent to the plant as field engineer (I was the number 2 guy on the job, working for Rocket Ron, the project manager). Rocket Ron was so named because he was volatile as hell and when given bad news would blast off until only his sneakers showed below the ceiling tiles. People were terrified of Rocket Ron, but he confided to me that was always just pretending to intimidate people. He acted in his small town's professional theater with leading man roles. Anyway, due to cost and schedule pressure, we were building a plant that had new technology ("frontier technology") and no pilot plant. Look, you want to try a new process technology? Fine. Build a pilot plant that demonstrates it'll fookin work. Oh no, we skipped the whole proof of concept part. So, in our "cold box" that had plate/frame aluminum heat exchangers and aluminum piping, the heat exchangers were to be manufactured by a very famous Japanese company in Kobe, Japan. There were about 10 of them, each the size of a car. We had monthly meetings with them over the phone, and they were halfway through the contract with 5 units built and 5 still being manufactured. Good news, right? Wrong. Headline news: KOBE, JAPAN HIT WITH MEGA EARTHQUAKE. Oh fuck. It took 10 days before we could get the Japanese on the phone. When we did, we just assumed they would declare "force majeure," which is the excuse that they couldn't fulfill their contract because they suffered an act of God. Rocket Ron: How bad is it? Japanese PM: Bad. All of the completed work is destroyed. The factory was leveled. It is all ruins. All the raw material is, in technical terms, tits-up. Rocket: What will that do to your schedule? (They were originally supposed to ocean ship in 6 months.) JPM: Sir, we will deliver on time. We will skip ocean shipment and air freight them in. That will save 6 weeks. Rocket: Air freight? That's gonna cost half a million dollars. JPM: We know. We will call it an insurable loss. Rocket: Are you seriously telling me you won't be late. JPM: Yes, Rocket-san. You see, you did nothing wrong. You should not be made to suffer. Rocket: Have I told you lately that I love you? We got less lucky on the MR compressor. This was part of the frontier technology, the use of a mixed refrigerant cycle instead of using liquid nitrogen on the front end. We'd basically use a big refrigeration cycle to pre-cool and intercool the hydrogen. By mixed refrigerant, we'd use isopentane and some other exotic chemicals to come up with a mixture with a very specific aggregate atomic number. The trouble? The only 4-stage centrifugal machine that would do the job was made by our good friends in Germany. This was in the pre-Euro days, so we had to purchase this $2m machine with Deutsche Marks. Now, here's where the accountants enter the chat. At some point before delivery we have to pay with German Marks. That means we need to exchange dollars for Marks. And exchange rates change all the time. Now, you can buy a bond, see, that will allow you to make the exchange at a particular rate. Depending on the amount of risk you want to take on the exchange rate. To simplify, I'm carrying $2m for that compressor in my cost forecast. But it could blow up to $4m or even 5 if the exchange rate for Marks tanks. Or, I'd get lucky and that $2m machine might only cost half a mil, right? So we had to make a decision on the exchange rate. We made the wrong one. That $2m machine ended up costing us 5, dammit. Hence, yet more abuse from the bosses. But it got worse, see? Remember I told you about frontier technology and how we went ahead without a pilot plant for proof of concept? Yeah. On our attempts to start up the completed plant? We couldn't get the MR compressor to come up to speed. The mixed refrigerant was too dense, too heavy, to allow the unit to come up to full RPMs. So we had to dump the entire $50k charge of refrigeration chemicals (with no designed-in way to recover them) and put in the lighter elements to get the machine to spin up to full speed, then slowly bleed in the heavier components, and to get this procedure to work, we went through 10 charges (half a million clams worth) and even then, the machine would only run for a day or two, and then you'd have to dump and recharge and kiss good-bye another $50k. The bosses were big mad at the process engineers. We couldn't get the plant to run. So we scrapped the MR technology and went back to using liquid nitrogen on the front end. But that meant we had to design and build one. Figure $10m and another year of delay to a project that was already $50 over budget and a year late. And guess what? NASA got mad and terminated the fuel contract. "You get nothing! Good day, sir!" Which made the plant, which now featured a front end of liquid nitrogen, a white elephant (a useless thing). So it all got demolished. If you go to google Earth and look for it at the chemical plant in Pace? It ain't there. Imagine how you feel when the project you sweated blood over for 3 years doesn't work and gets taken down by bulldozers, with the whole company looking at you like you're a failure. Now, add the pain of that to Divorce 1.0 + Paternity Lawsuit and you have a pretty good recipe for the precursors to self-yeeting. Which I didn't do. I can prove it. I wrote this, didn't I?
19
19
8
178
14,310
Jeremy Keller retweeted
Business model density all deep tech companies that survive end up in the same place: they can put dollars to work and get back a multiple of those dollars from the market, with a product that beats incumbents on cost, margin, and performance. but in the early years, the business model is usually an afterthought. the product is the focus, and the model is treated as a pricing wrapper. that is a mistake. your business model is your survival plan, financing plan, and product plan expressed as a single choice: who holds which risks, in what order, and what kind of evidence comes back per dollar. so the real question, at least at the early stage, is not “how do we maximize economic value from our model?” but “what do we spend the next dollar proving?" underneath any business model sit four inescapable risks: - market risk (does anyone really want this) - product maturity risk (can it run without heroics) - economic risk (do unit economics converge) - capital risk (can this structure ever be financed at scale) for deep tech, the model choice is usually: sell equipment, offer raas, or become a prime. equipment pushes asset risk to the buyer but assumes real market pull and product maturity. raas keeps asset and operating risk on your balance sheet in exchange for smoother revenue. prime internalizes almost everything so you can manufacture demand and control outcomes. for an early stage founder, the binding risk is almost always capital led. there is effectively one pool of risk capital to raise from: venture. you are asking vcs to underwrite unproven, capital intensive technology, market, and asset risks at the same time. they can only tolerate that if each dollar buys very dense proof. that is what business model density is, how much proof you get per unit of capital and time on two questions: - is there repeatable demand for this - could this ever be financed by something cheaper than venture once you see it this way, there are only two kinds of early deep tech companies: 1. there are “evidence machines,” which treat the current business model as scaffolding. every deployment is designed to tighten the story for both customers and future capital. they pick a dominant risk, aim the model at it, and adjust the structure as proof accumulates. 2. there are “fiction machines,” where the structure exists to look good in a fundraise but produces very little proof in the real world. they mix equipment, raas, and prime in ways that feel big but leave market, maturity, economics, and capital all half proven. business model density is the distance between “we raised and shipped a lot” and “we can point to a narrow wedge of the world where this clearly works and could be financed at scale.”
1
2
7
2,070
Jeremy Keller retweeted
Another big milestone for industrial decarbonisation: @rondoenergy has just powered up what it calls the world’s largest industrial heat battery — a 100 MWh system delivering continuous high-temperature steam from renewable power. Industrial heat makes up roughly a quarter of global final energy use, yet remains one of the hardest sectors to decarbonise. Most of that heat still comes from fossil fuels, largely because reliable, cost-effective clean alternatives have been lacking. Projects like this one are encouraging because they show that renewable energy can provide not only electricity, but also the very high-temperature heat industries depend on — 24 hours a day. Using simple, durable materials and integrating directly with existing steam systems, Rondo’s “brick-and-wire” design is an interesting step toward electrifying heat at scale. It’s early days, but this kind of innovation could be an important piece of the puzzle for decarbonising cement, chemicals, food processing, and other energy-intensive sectors. rondo.com/news-press/rondo-p…
8
82
6
221
14,261
It used to be about building power and load far apart. Then it was about building power where load is. Now it’s about building load where power is.
1
11
838
Customer built a building and included a microgrid covering about 50% of their power. Utility now wants to put "departing load charges" on the microgrid's power. Departing load?? It was never served by the utility in the first place. And we're already paying standby charges.
21
10
4
138
15,221
Jeremy Keller retweeted
Each of these pictures are from Google Maps at the same elevation (zoom) and about 10 miles apart, one of them is a wind farm and the other is an oil and gas field.
6
9
2
46
8,649
Replying to @rondoenergy
@rondoenergy This is a great article on the coming deployment of one of our 100MWh Heat Batteries with GreenLab in Denmark. Powered by their existing wind and solar the system will reduce their natural gas consumption by 80% annually.
3
3
112
Electrified Future: Rondo Turns Renewable Electricity Into Industrial Heat - Newsweek
1
49
Jeremy Keller retweeted
Time to make 2024 updates to my annual “opinions about solar” thread. If you like these, the second edition of my book, Solar Power Finance Without The Jargon, is new this year. A 30% discount code WSQ0437 is valid on publisher website until end of Oct. worldscientific.com/worldsci…
32
163
63
545
116,515
Jeremy Keller retweeted
Money from Sam Altman. A massive reveal party with Leonardo DiCaprio. Sweeping claims about a tech "unlike anything we've seen in history." Except we've seen this story before. Many times. If you read one thing this morning, make it our Exowatt story: bit.ly/3MMXgnl
1
6
10
2,077
Jeremy Keller retweeted
Emergency Conditions 2 years ago, smooth sailing today. Has California’s grid finally caught up to its vision? Last week, temperatures in the west reached up to 115° leading to high loads and reduced availability for both generation & transmission; similar conditions to September 2022.
5
39
13
153
50,621
Okay, I have some commentary and theory on how China hit the 2030 goal 6 years early, because it is indeed uncommon for China to set a target they can smash so easily. Context: In December 2020, Xi announced this 1200 GW target at that year's UN Climate Ambition Summit. 🧵
It’s actually crazy that there exists any large government anywhere that hits major goals (that involve significant mobilization of people and industry) EARLY. Like not just a little early, but many years early. This is beyond “authoritarian” or “Chinese people work really hard.” It’s actually outside of my Overton Window. How is this even possible? What positive feedback loop kicked in that made their planners so off-the-mark that they got there in half the time? It shouldn’t happen. (China doesn’t hit other plans early like this; some plans even fail)
19
251
16
1,549
525,822
Jeremy Keller retweeted
How a small amount of demand elasticity stops wild prices with wind and solar: new pre-print with theory and examples arxiv.org/abs/2407.21409 Co-authors: @fneum_ & @IegorRiepin TL,DR: compare prices with inelastic demand on left to right with -5% elasticity (at average price)
6
47
9
106
33,128
Jeremy Keller retweeted
There are two primary mechanisms of driving institutional change: exit and voice “Exit” refers to “exiting” an incumbent relationship and pursuing competitive offers for provided services “Voice” is the expression of dissatisfaction or the attempt to improve conditions from within existing institutional channels Here’s why they will both ultimately break franchise rights (bear with me)...
1
2
4
1,288
Great overview on all of our recently announced commercial projects at Rondo Energy.
5
85
Jeremy Keller retweeted
Rondo Energy announces €75M project funding with @Breakthrough Energy Catalyst and the European Investment Bank (@EIB) for 3 Rondo Heat Battery installations that will decarbonize food, clean fuel, and chemical production across Europe. rondo.com/press-releases/ron…
1
8
521
@rondoenergy also netted a new investment this week from H&M Group Ventures. Rondo is the first investment from the new strategic investment arm of the global clothing chain, which envisions using heat batteries in their supplier network to cut emissions.
1
1
1
3
1,981
Exciting news: @RondoEnergy & Eastman will partner to deploy thermal batteries to supply decarbonized process heat at Eastman's molecular recycling facility in Longview, Texas. eastman.com/en/media-center/… The project is supported by an award of up to $375M from the Dept of @Energy's Industrial Demonstrations Program, the second Demo program award to feature Rondo heat batteries (Rondo will also deploy heat batteries to decarbonize distilleries run by Diageo with support from DOE; see diageo.com/en/news-and-media…) Disclosure: I'm proud to be an advisor to Rondo Energy and have an equity interest in the company.
7
12
2
90
11,918