I solved the hard part of an impossible problem

The geometry looked plausible on paper. The mass balance had already ruled out a passive cartridge before I started drawing. And that still wasn't the number that killed it.

Everybody would like to catch the carbon dioxide coming out of a car. The gas leaves through one pipe, and one pipe is easy to put something on.

There's a reason you don't see a passive CO₂ cartridge hanging under ordinary cars. It isn't the reason I assumed, and it turned out there were two of them.

The hard part

If you put a filter across an exhaust pipe, you choke the engine.

An engine is a pump. It has to push its exhaust out against whatever you've placed in the way, and that resistance — back-pressure — costs power and fuel, and enough of it will damage valves and turbochargers. Put a solid plug of sorbent across the pipe and you've built an effective way to stop a car.

So: capture needs contact, contact needs obstruction, obstruction kills the engine.

My answer was an annular cartridge.

Instead of filling the pipe, line it. The sorbent sits in a ring against the wall and the middle stays open — a clear throat straight through. Gas moves slowest at the wall, so that's where you'd get contact time. The fast core goes through the hole.

I worked on it for months. Drew it, sized it, costed the materials, worked out the mounting. It's in my design record with sections and part numbers.

I never built it. I want that on the record before I say anything else, because everything below is about a drawing.

What the drawing actually buys

Writing this issue forced me to look at my own geometry properly, and it's less impressive than I'd been carrying around in my head.

A ring of sorbent still narrows the pipe, and pressure drop climbs steeply when a pipe narrows — roughly with the fifth power. Squeeze a 50 mm pipe down to a 35 mm throat and the drop goes up about six times, before you count what the flow loses squeezing in and spreading out again. An engine may well tolerate that. But "acceptable, and far better than a plug" is not "solved", and only the first is a claim I've earned, because I never put it on a flow bench.

And the boundary layer argument, which I was proud of, is backwards.

Gas really is slowest at the wall. But in a turbulent exhaust that slow film is a fraction of a millimetre thick, and it isn't a reservoir of gas presented to the sorbent — it's the thing standing between the sorbent and the rest of the flow. Capture is limited by how fast CO₂ can cross that film. My sentence "the fast core goes straight through and never knows you're there" is a description of low capture efficiency. I had it filed as a feature.

This is why catalytic converters are honeycombs with hundreds of tiny channels, and why particulate filters force gas through the medium and accept the back-pressure. A single ring has close to the worst contact area of any geometry you could pick. I bought a low pressure drop by buying very little contact.

The check I didn't run

At no point in those months did I ask how much carbon dioxide a car actually produces.

Not roughly. Not to an order of magnitude. Never at all — because the problem I'd put in front of myself was back-pressure, and I was busy solving it.

Here's the whole check. It takes a minute.

Burning fuel doesn't destroy matter. It combines it. Every carbon atom picks up two oxygen atoms on the way to becoming CO₂, and oxygen is heavier than carbon.

1 kg of fuel → 3.1 kg of CO₂

Petrol is 84 to 86 percent carbon by mass, and CO₂ weighs 44 against carbon's 12. Multiply it out: 3.08 for iso-octane, about 3.15 for real petrol.

And "capture" means "keep". That's the part I never made myself say out loud. You aren't filtering something out and letting it go. You're taking custody of it. It rides in the car, gaining weight, until someone removes it.

Reversible sorbents that work near room temperature hold somewhere between a tenth and a sixth of their own weight in CO₂. Take the best of that range — about 17 percent — and one tank's worth still needs 670 kilograms of sorbent. At the more typical tenth, 1,150 kilograms. Either way it is emptied and replaced at every fill-up.

A realistic cartridge — five kilograms, already heavy for something under a car — holds about half a kilogram. At a modest cruise you fill it in under three minutes. At a thirstier one, under two.

The people actually trying to make this work get the same kind of answer. A Swiss study of onboard capture for a delivery truck calculated that holding a day's CO₂ in the sorbent would mean carrying about 1,055 kg of adsorbent. Their better alternative regenerates the sorbent on the move, compresses and liquefies the CO₂ at around 75 bar, and stores it in a pressure vessel. That cuts the weight of sorbent sharply — at the cost of compressors, heat exchangers, a regeneration cycle and a large tank. A passive cartridge has none of those. That's what makes it passive.

Correcting my own record

When I closed this branch I wrote in the report that the sorbent needed "would weigh more than the vehicle."

Checking it properly: about 1,150 kilograms against a car of 1,300 to 1,500. Comparable to the car. Not heavier than it. And the ratio barely moves for a smaller car, because the tank shrinks with it.

Mine was the more vivid version and it was the one I reached for. The true numbers are less quotable and considerably worse, because "as much as the car, replaced at every fill-up, forever" is not a weight problem you engineer around.

I ran the rule. It wasn't enough.

After the bicycle — Issue 02 — I gave myself a rule: compute before you build. I'd since sharpened it into something more specific. Write down what the goal requires in units and check the requirement is physically available. Not the mechanism. The budget.

I ran that rule here. I found the mass budget. I stopped.

There was a second boring number underneath it, and I only found it because I was writing this.

Temperature.

The sorbents I'd been sizing are the low-temperature class. They pick CO₂ up near room temperature and give it back when you heat them to somewhere between 100 and 150 degrees. That's how they're regenerated — heat is how you empty them.

Exhaust gas leaves an engine somewhere between 100 and 500 degrees, depending on the engine, the load and where in the pipe you measure.

So raw exhaust isn't where these sorbents pick CO₂ up. It's much closer to the temperature where they let it go. I'd been carefully sizing a bed of low-temperature material for a stream that would first have to be cooled — and there was no cooler in my design, because I didn't know I needed one.

It gets worse in the details. Amine sorbents break down at those temperatures when there's oxygen about, and exhaust is full of oxygen. Sulphur dioxide sticks to them and won't come off. The other option is zeolite, and water vapour takes the sites it needs — exhaust is up to a tenth water.

The serious onboard systems do exactly what I hadn't: cool the exhaust first, take the water out, adsorb at low temperature, then heat the bed to release it. A passive, unheated cartridge in a hot pipe is a single-use item at best.

I ran the check I'd learned to run. I still stopped one number short.

That's the part of this I find hardest to write. The mass balance was the lesson from a previous failure. I applied it correctly. It still wasn't enough — because I stopped at the first fatal number instead of asking what else the goal quietly required.

One check is a habit. It isn't a method.

Difficulty is a poor compass

Two questions were in front of me.

Can you make contact without choking the engine? Genuinely hard, took months, and I got somewhere on paper.

How much gas is there, and how hot is it? Two minutes, and either one is fatal.

I spent all my attention on the first. Not by decision — by instinct, because the hard problem is the one that feels like real work. It has drawings in it. Sitting down with the atomic mass of oxygen does not feel like inventing.

In my own projects the impossible part has almost never been the difficult part. It's usually something short and boring underneath. Nobody checks it, precisely because it's short and boring.

I assumed the hard part was the risky part. It wasn't. The hard part was just the part I was paying attention to.

The gate I never ran

One more thing, and it's the one I like least.

I never did a prior-art search on the ring itself.

Sorbent beds lining an exhaust duct with an open bypass down the middle are not new. That arrangement has been worked on for decades for cold-start hydrocarbon traps — catching the unburnt fuel that comes out in the first minute before a catalytic converter warms up — which is very close to the application I'm about to say the thing might still be good for.

My own second gate asks: is it new, or are you following someone else's path in your own notation? On this device I never asked. I was so occupied with whether the geometry worked that I never checked whether it was mine.

So take everything below as a description of something I'd have to look up before claiming any of it.

What survived

What's true about the cartridge is that gas gets through it easily and barely touches it on the way. Whether it catches enough of anything to be worth fitting is unmeasured, and after this week I'd put the odds lower than I used to.

But there are things in an exhaust that come out in grams rather than hundreds of kilograms — particulates, unburnt hydrocarbons, the fumes that make a garage unpleasant. Against a target measured in grams, "enough" is at least a number I could hit. And the temperature problem belongs to CO₂ sorbents. A filter doesn't care how hot the gas is.

That makes it worth a prior-art search, then a bench, in that order. It does not make it worth claiming, and I won't say anything more about it until there's a measurement.

It is not a carbon capture device, and I will not sell it as one.

And the rule, amended. Write down what the goal requires in units — and then keep going until you've found every unit it requires. Mass. Temperature. Energy. Time. Money. Stopping at the first fatal number feels like diligence. It's just a different place to stop early.

What's the second boring number underneath the thing you're working on?