Boil-Off
by Claude Opus 5.5
His father delivered heating oil. That was the first thing Daniel Kwame-Hart ever understood about fluids, and in a sense it was the only thing, the rest being elaboration: that a liquid wants to go where it wants to go, and that a man's living depends on persuading it otherwise. Kwame-Hart Fuel Oil, one truck, later two, a yard off Frelinghuysen Avenue in Newark, a phone number painted on the tank in red letters that the boy had watched his father touch up every spring with a brush the size of a toothbrush. Kofi Kwame-Hart had come from Kumasi in 1979 with an engineering diploma that nobody in New Jersey would recognise, had married a girl from the Weequahic section whose own father had been a plumber, and had spent thirty-one years dragging a two-inch hose down basement steps into the cellars of the Ironbound and the Central Ward and the better streets out towards Maplewood, feeding the furnaces of the city through the cold. What did the boy learn, riding in the cab on Saturdays in January? He learned that you checked the gauge before you pumped, because the gauge lied. He learned that a tank that read a quarter full might be nearly dry, and a tank that read empty might be sitting on sixty gallons of sludge, and that the only way to know was to put a stick in it. He learned that the customer always believed the gauge, and that the customer was always, in the end, your problem.
Forty years later, in a conference room in a building in Texas whose air conditioning was set, as it was in every aerospace building in the state, to a temperature appropriate for the storage of meat, Daniel sat across from four executives and a slide that said NOMINAL in a font so large it was practically shouting, and thought: Pop would have put a stick in it.
*
What had happened was this. Six days earlier, in a circular orbit three hundred and twenty kilometres above the Earth, a tanker spacecraft built by Tallgrass Aerospace had docked with a depot vehicle built by Tallgrass Aerospace and had attempted, for the first time in the history of the species, to pump something like a hundred tonnes of liquid oxygen and liquid methane from one ship to another. It was the hinge of the whole business. Without it, the lunar lander that Tallgrass was building under contract could not be fuelled in orbit, and without being fuelled in orbit it could not get to the Moon, and without getting to the Moon there would be no crewed landing in 2031, and without the landing there would be, to put it plainly, no Tallgrass. The company had bet itself on a hose.
The test had transferred, by the most optimistic reading of the data, sixty-one per cent of what had been planned. By the least optimistic, fifty-two. The two readings came from two different methods of measuring how much propellant was in a tank, and the reason there were two methods, and the reason they disagreed by nine percentage points, was the reason Daniel had been put on this earth, or at least on this payroll, and the reason he was now in this refrigerated room trying not to lose his temper.
*
Let us be clear about what is hard here, because it is not what people think. People think the hard part is the docking. It is not the docking. The Russians were docking in the sixties. People think the hard part is the cold, keeping liquid oxygen at minus one hundred and eighty-three degrees and methane at minus one hundred and sixty-one in a box that the Sun is shining on. That is hard, but it is ordinary hard, refrigeration hard, the kind of hard you solve with insulation and sunshades and money.
The hard part is that in orbit nothing knows which way is down.
In his father's truck, the oil sat at the bottom of the tank, because the Earth pulled it there, and the pump drew from the bottom, and the oil went into the hose, and the hose went into the cellar, and gravity did the rest. In orbit there is no bottom. A tank of liquid methane in free fall is not a tank with liquid at one end and vapour at the other. It is a tank with liquid wherever the liquid feels like being: clinging to the walls, floating in a great wobbling blob in the middle, broken into a thousand droplets drifting among bubbles of its own vapour. Open a valve at the end of that tank and you are as likely to draw gas as liquid. Draw gas into a pump designed for liquid and the pump cavitates, chatters, overheats, and stops.
So you have to make a bottom. The way you make a bottom, if you are Tallgrass and you have big tanks and a tight schedule and no appetite for the elegant capillary screens that the old NASA studies recommended for small ones, is that you push. You fire small thrusters on the docked pair, gently, continuously, imparting an acceleration of perhaps a ten-thousandth of a g, and the liquid, which has inertia and wishes to remain where it was, slides to the back of the tank, over the outlet. Settling, it is called. Ullage thrust. It works. It has worked since the Apollo S-IVB stage, which did it before restarting its engine. The trouble is that it costs propellant to push, and the longer you push, the more it costs, and the transfer of a hundred tonnes at the flow rates you can manage through a line you can actually build takes hours.
And then there is the warmth.
*
The receiving tank on the depot had never held cryogenic propellant. It had been launched empty and had sat in orbit for three weeks at something like room temperature, which is to say several hundred degrees warmer than the liquid that was about to be poured into it. Pour liquid methane at minus one-sixty into a tank at plus twenty and what you get, initially, is not a tank of methane. You get a tank full of methane gas, at rising pressure, because the first liquid in flashes to vapour as it hits the warm walls. The pressure climbs until it reaches the tank's limit, and then you have two choices. You can stop, wait for things to cool, and start again. Or you can vent the gas overboard, which throws away the very thing you are trying to deliver.
The standard procedure, worked out in a thousand ground tests and a handful of small orbital experiments, is called charge, hold, vent. You put in a little liquid; you let it boil and cool the walls; you vent the gas; you put in a little more. Each cycle chills the tank a little further, until at last the walls are cold enough that the liquid stays liquid. Every vent is a loss. The engineering question, the question Daniel had spent four years of his life on, was how many cycles, how much liquid per cycle, how long to hold, at what pressure to vent, all of it tuned against a thermal model of the tank and the lines and the pump that had never, until six days ago, been tested against reality at full scale.
Reality had disagreed with the model. The lines had taken longer to chill than predicted. The tank walls had cooled unevenly, with a hot spot near a structural ring that nobody had given enough thermal mass to in the model. The receiving tank's pressure had spiked four times during the main transfer, forcing four unplanned vents. And at the end, when the settling thrust was shut off, the liquid in the receiving tank had sloshed back up the walls in a great slow wave, which the onboard cameras had captured beautifully and which had thrown the mass gauging into confusion for several minutes.
*
Now. Gauging. This is where Pop's stick comes in.
How do you know how much liquid is in a tank in orbit? You cannot look at a level, because there is no level. You cannot weigh it, because it weighs nothing. There are two serious methods, and Tallgrass had flown both.
The first is bookkeeping. You know the pressure, volume and temperature of the gas in the tank, and you know how much gas you have put in or let out, and from the equations of state you can infer how much liquid must be there. It is the method of the accountant: elegant, cheap, and only as good as every single measurement that feeds it, any one of which may be off by a little in a way that compounds.
The second is radio. You fill the tank with radio waves and measure the frequencies at which it resonates, which depend on how much of the volume is occupied by a liquid with a particular dielectric constant. It is the method of the physicist: direct, clever, and dependent on a model of the tank's electromagnetic behaviour that is very good when the liquid is a single blob and less good when it is scattered in droplets and films and sloshing waves.
The bookkeeping said sixty-one per cent. The radio said fifty-two. Daniel believed, on balance, that the truth was somewhere in between, closer to the radio, because he had personally checked the bookkeeping inputs and found two temperature sensors whose calibration he no longer trusted. He did not know this. Nobody knew this. And the difference between fifty-two and sixty-one, multiplied across the fourteen tanker flights that the lunar architecture required to fuel the lander in orbit, was the difference between a lander that reached the Moon with margin and one that did not.
*
It is a strange feeling, and an American feeling, Daniel thought, to sit in a room with powerful men and listen to them redefine a word. He had listened to the presentation for forty minutes, delivered by a vice president for programme integration named Tom Hartigan, a pleasant, square-jawed man with a former Navy pilot's habit of standing with his weight evenly distributed, and Daniel had come to understand that, in the language of the presentation, nominal no longer meant what it had meant when he was a graduate student at Purdue. It no longer meant that things had gone according to plan. It meant that things had gone in a way that could be described, with the right charts and the right emphasis, as consistent with the plan's broader intent. The transfer had demonstrated the architecture. The docking had been flawless. The settling thrust had worked. The pump had not cavitated. Propellant had moved from one vehicle to another. First in history. There it was on slide eleven, in letters only slightly smaller than NOMINAL. And there, on slide fourteen, was the recommendation: proceed to the operational campaign, beginning with the first lander-fuelling flight in four months, in parallel with continued analysis of the test data.
In parallel. Daniel had come to hate those two words more than any other words in the language. In parallel meant: we will find out whether it works by doing it.
*
What did he want? This was the question Hartigan put to him, not unkindly, at the end. Hartigan had the gift, which Daniel had seen in good managers and bad ones alike, of asking a question in a way that made the honest answer sound unreasonable. He wanted to know, simply, what Dan needed.
What he needed, he said, was three more test flights.
Not one. Three. And he laid it out, because he had laid it out in his head every night for six nights, lying awake in his apartment in Brownsville listening to the air conditioner rattle and his wife breathing beside him.
The first flight would repeat the transfer with the thermal model corrected for the structural ring, and with a revised chilldown sequence, and would tell them whether the losses to venting could be brought down to the number the architecture assumed. The second would be a long-duration hold: fill the depot and leave it in orbit for six weeks, because the lunar architecture required propellant to sit in the depot while successive tankers arrived, and nobody had measured boil-off at full scale over that duration, and every day of sitting was a day of heat leaking in, liquid boiling off, pressure climbing, gas vented and lost. The third would fly with a third gauging method, an optical instrument they had been developing on the side, so that the two disagreeing methods could be checked against something independent. Three flights. Eighteen months. A slip of a year in the lunar landing, at least.
The room went quiet in the particular way that rooms go quiet when someone has said a number that everyone already knew but nobody wanted to hear spoken aloud.
*
A digression, because Daniel was a man who could not get through a hard moment without one, his mind being furnished, like his father's yard, with a great deal of equipment that might come in handy.
There is a story about the Hubble Space Telescope that every engineer of a certain age knows, though most tell it wrong. The mirror was ground to the wrong shape, everyone knows that. What fewer people know is that the error had been detected, or nearly detected, before launch. Two separate test instruments had given readings that disagreed with the primary test, and the people responsible had decided that the primary test was right and the secondary ones were faulty, because the primary test was the one they had designed to be definitive, and because believing the others would have meant admitting that something was wrong with a mirror that had already cost a fortune and was already late. The investigation board afterwards was polite about it. It spoke of a failure of quality assurance. What it meant, Daniel had always thought, was that a group of intelligent people had looked at two gauges that disagreed and had chosen to believe the one that told them what they wanted to hear.
He did not say this in the meeting. You do not compare your colleagues to the Hubble mirror team in a meeting, not if you want to keep working with them. But he thought it, and the thought gave him a kind of steadiness.
*
The case against his three flights was made by the chief financial officer, a woman named Patricia Dunleavy whom Daniel had never met before that day and who turned out to be, to his surprise and somewhat to his dismay, entirely reasonable.
She said that three flights would cost something over four hundred million dollars, including the vehicles, the launches, and the standing army of engineers who would be paid to wait. She said that a one-year slip in the landing would trigger penalty clauses in the lunar contract, and, more seriously, would hand the political narrative to the company's chief competitor, which was eighteen months behind and closing. She said that the company's next funding round, which was not optional, would be priced on the basis of the landing date. She said, finally, and this was the part that got under Daniel's skin, that every test programme in history could always have used one more test, and that at some point a company had to decide that it knew enough to proceed, and that she was not an engineer and would defer to engineering on whether that point had been reached, but that she wanted engineering to understand what it was asking.
He understood what he was asking. That was the whole trouble. He understood exactly.
*
Here is what he said, more or less, though he said it less well than he would later wish.
He said that the transfer test had not demonstrated the architecture. It had demonstrated that the architecture could move propellant, which nobody had seriously doubted. What it had not demonstrated was that the architecture could move enough propellant, with known losses, measured by gauges that agreed. Those were three different claims, and the lunar plan depended on all three. If the losses were as bad on the operational flights as on the test, the lander would need sixteen or seventeen tanker flights instead of fourteen. If the gauges continued to disagree by nine points, the crew of the lander would depart for the Moon not knowing, within nine points, how much propellant they had. And nine points of propellant, at the end of a lunar descent, was the difference between landing and not landing.
He said that the operational campaign, as currently planned, would fill the depot over a period of about three months with propellant from successive tankers, and that during those three months the propellant would be sitting in a tank in orbit absorbing heat, and that the company had no full-scale measurement of how fast it would boil away. The thermal models said a fraction of a per cent per day. The test had shown that the thermal models were optimistic. If they were optimistic by a factor of two, the depot would lose something like a fifth of its contents before the lander arrived to collect it.
He said that he could not, in conscience, sign off on a lunar crew departing on the basis of the data they had.
He did not say that he was thinking about his father dragging the hose down the basement stairs in the Ironbound, but he was.
*
The meeting ended without a decision, which is how important meetings at aerospace companies usually end. Hartigan thanked him, warmly. Dunleavy thanked him, correctly. The two other executives, who had not spoken, nodded in the way that people nod when they intend to have a separate conversation later. Daniel walked out into the Texas afternoon, which was ninety-six degrees and as wet as a towel, and stood for a while in the car park beside his car, sweating, feeling the heat come off the asphalt, thinking about boil-off.
That was the thing about cryogenic propellant that people never quite grasped, even people in the industry. It was always leaving. From the moment it was made, in a plant in Texas where natural gas was cooled until it condensed, it was trying to warm up and go back to being gas, and every system built to handle it was a system for slowing that process down, never for stopping it. Insulation slowed it. Sunshades slowed it. Cryocoolers, the little refrigerators that the next-generation depots would carry, could in principle balance the heat leak exactly and hold the propellant indefinitely, but they cost power, and power cost solar arrays, and solar arrays cost mass, and mass cost launches, and launches cost money, and the money was being priced, right now, in a funding round, on the basis of a date. Everything leaked. The propellant leaked heat. The schedule leaked days. The company leaked cash. You could not stop any of it. You could only decide how fast you were willing to let it go.
*
The decision came eight days later, in an email from Hartigan with the subject line Path Forward, which Daniel read on his phone at a red light on the highway and then, at the next red light, read again.
The company would fly two additional test flights, not three. The first would repeat the transfer with the corrected thermal model and the revised chilldown. The second would combine the long-duration hold with the third gauging method, which was Daniel's own suggestion from a fallback slide he had shown at the very end of the meeting, the slide he had prepared in case they said no to three and that he had hated preparing. The lunar landing would slip by eight months, not twelve. The CFO had negotiated, somehow, a restructuring of the contract penalties. The funding round would proceed on a revised date. Daniel would lead the test programme and would have, the email said, the authority to recommend a further flight if the data warranted it.
He had wanted three. He had got two and a half. He had known he would. A part of him, the part that was his father's son, the part that had stood in the cold in the Central Ward watching Kofi Kwame-Hart argue with a landlord over a meter reading, felt the familiar, complicated satisfaction of a man who has been paid most of what he was owed and knows he will not see the rest.
*
He called his father that evening. Kofi was eighty-one, and lived in a condominium in West Orange with Daniel's mother, and had sold the business in 2019 to a man from Elizabeth who had promptly painted over the red letters on the truck. He did not understand much about rockets and did not pretend to, and Daniel did not try to explain the test or the gauges or the boil-off. What he said was that they had wanted to fill the tank without checking it, and that he had made them check it, and that they had let him check it twice.
His father was quiet on the line for a moment, and then said something in Twi that Daniel had not heard since he was a child, and then said in English that twice was good, that twice was usually enough, and that the customer always believed the gauge.
Daniel said he knew. He said the customer was always his problem.
His father laughed, a long, wheezing laugh that turned into a cough, and Daniel stood in his kitchen in Brownsville with the phone against his ear and the air conditioner rattling and listened to the old man in New Jersey cough and laugh, and thought about all the cellars in Newark that had been warm through all those winters because a man from Kumasi had not trusted the gauge, and thought about the hose, and thought about the stick.