A Starship reaches low Earth orbit with its ascent propellant essentially spent; the receiver vehicle begins any aggregation campaign at near-dry state. A Mars departure burn at maximum payload requires a substantial fraction of Starship's full propellant capacity, which SpaceX announced in April 2024 as 1500 metric tons on Block 2 (1170 t liquid oxygen plus 330 t liquid methane, up from approximately 1200 t on Block 1). The gap defines an aggregation campaign: four to more than twenty tanker flights depending on mission class and margin, rendezvousing with the receiver in LEO and transferring propellant ship-to-ship. Ship-to-ship cryogenic transfer at this scale has not flown. IFT-3 demonstrated an internal propellant shift between Starship's header and main tanks in March 2024 under thrust settling; per NASA HLS Deputy Manager Kent Chojnacki (November 2024), that internal transfer was measured to approximately 5 percent accuracy, though the mass actually transferred has not been publicly disclosed as a fraction of the 10-ton Tipping Point contract goal. The Tipping Point ship-to-ship cryogenic-transfer milestone remains outstanding. A multi-hundred-ton aggregation campaign running across weeks to months in LEO raises six coupled engineering problems, each absent from SpaceX's public patent record and each derivable from first principles. The first and largest is boil-off. A bare polished-stainless Starship in LEO with an external tank surface area of approximately 950 square meters absorbs approximately 130 kilowatts of radiative heat load on its methalox tanks at central assumptions (solar absorptance 0.35, IR emittance 0.12, orbit-averaged absorbed flux 137 W/m²), ranging from approximately 95 kilowatts for a freshly polished sun-avoiding vehicle to approximately 190 kilowatts for an aged sun-facing vehicle. At central assumptions the heat load drives approximately 2.9 percent per day methane boil-off and 2.5 percent per day LOX boil-off, emptying the tanks in approximately 34 to 40 days. Aged-skin cases shorten that to 23 to 27 days. Multi-layer insulation at one watt per square meter residual heat flux reduces the rate by approximately three orders of magnitude. Active zero-boil-off on top of the MLI demands approximately 34 kilowatts of continuous electrical power driving cryocoolers against the Carnot limit at 90 K and 111 K, falling to approximately 25 kilowatts with a cooler radiator at 250 K. The remaining five pain points (zero-g liquid acquisition, slosh at transients, zero-g mass gauging, a cryogenic androgynous docking interface, and attitude coupling across a 104-meter mated stack whose combined mass grows from approximately 430 tonnes at the first transfer to approximately 1830 tonnes at the fifteenth) each set independent bounds on the operating envelope. This note works each pain point from first principles and gives quantitative bounds at mission-relevant scale. A companion technical disclosure publishes the integrated architecture under 35 U.S.C. §102 as prior art of record.

The Mars architecture arithmetic

The public Starship plan is a two-vehicle mission. A receiver Starship launches into LEO carrying mission payload and a partial propellant load. One or more tanker Starships then launch, rendezvous with the receiver, and transfer propellant. When the receiver is full, it departs for Mars.

The arithmetic sets the scale. Starship Block 2 carries 1500 tons of methalox at full propellant load (1170 tons LOX plus 330 tons methane at the Raptor stoichiometric mixture ratio), per SpaceX's April 2024 announcement. Block 1 vehicles carried approximately 1200 tons. A Starship reaches LEO with its ascent propellant essentially spent: the vehicle enters parking orbit at near-dry state, and the receiver for a Mars mission begins aggregation from there. The aggregation target is close to the full 1500-ton Block 2 capacity for a maximum-payload Mars departure, and correspondingly less for reduced-mission profiles. Public estimates of tanker-flight count span from four to more than twenty. NASA's 2021 HLS Source Selection documentation cited sixteen launches for a single lunar mission; Musk public statements between 2021 and 2022 suggested between four and eight; and the NASA Office of Inspector General in March 2026 reported "more than 10" for a single HLS mission, with SpaceX targeting a one-tanker-per-6-days cadence across a roughly 200-day launch window.

During aggregation, the receiver sits in LEO with a partially full tank for however long the campaign takes. SpaceX has not published a committed aggregation cadence. NASA HLS Deputy Manager Kent Chojnacki stated in November 2024 that the target tempo was "about once a week per pad," with the aggregation phase running across weeks. The receiver, meanwhile, is exposed to solar radiation, Earthshine, and a 3 K deep-space sink, and is losing propellant the entire time. When the receiver departs, the residual propellant determines the available mission mass. Any tonnage lost to boil-off during aggregation has to be replaced by an additional tanker flight, and that tanker itself pays its own ascent boil-off tax. At low enough cadence and high enough loss rate, the campaign does not close: tankers arrive with less propellant than the receiver has lost since the last tanker, and the aggregate curve goes negative.

The architecture choice that keeps the aggregate curve positive is the unstated part of the Starship program.

What has flown

Ship-to-ship orbital fluid transfer has flown once at operational scale. DARPA's Orbital Express mission (March to July 2007) transferred approximately 14 kilograms of storable hydrazine between two free-flying spacecraft, ASTRO (Boeing) and NextSat (Ball Aerospace), across four autonomous mate-and-demate cycles and seven refueling scenarios including pressure-fed and pump-fed transfer, with client fill fractions up to 95 percent. Orbital Express remains the only autonomous spacecraft-to-spacecraft fluid transfer demonstrated in space. Hydrazine is storable at ambient temperature; Orbital Express did not address cryogenic management, thermal-cycle seal life at 90 K, or aggregation campaigns.

At Starship, IFT-3 in March 2024 performed an internal transfer of liquid oxygen between the header tank and the main tank under thrust settling during a short ballistic coast. The transfer was carried out against a 10-ton contract target under NASA's Tipping Point program ($53.2 million, October 2020 award). Per NASA HLS Deputy Manager Kent Chojnacki's November 2024 Spaceflight Now interview, the measured transfer mass was reported to approximately 5 percent accuracy. The demonstration exercised the internal plumbing and propellant-management architecture at coast-g. It did not demonstrate ship-to-ship transfer, orbital hold at the scale of a full propellant load, or propellant management across timescales longer than the short ballistic coast. Every subsequent flight through IFT-11 (October 2025) has executed other objectives: booster catch, Raptor relight, first payload deploy. None has extended the propellant-transfer demonstration. NASA OIG (March 2026) lists the Starship ship-to-ship cryogenic transfer demonstration as outstanding against the Artemis III critical path. Public estimates place a first attempt in late 2026 or 2027. No component of the full aggregation architecture has flown at Starship scale.

Upstream prior art at smaller scale is extensive. Centaur-class upper stages have flown settling-and-ullage architectures since the 1960s. NASA's SHIIVER (Structural Heat Intercept, Insulation, and Vibration Evaluation Rig) campaign tested a 4-meter-diameter stainless propellant tank with multi-layer insulation and structural vapor cooling from August 2019 through January 2020, publishing the canonical modern MLI performance data at propellant-tank scale (NTRS 20205008233). NASA's MHTB (Multipurpose Hydrogen Test Bed) characterized ground cryogenic MLI (NTRS 20110009964 and 20100034929). The CRYOTE concept carried a small-scale passive cryogenic hold experiment toward flight as an Atlas V secondary payload (NTRS 20090037680), with ground articles built but flight hardware never launched. NASA Glenn's zero-boil-off test corpus across two decades (Plachta and Kittel 2003 NTRS 20030067928; Hedayat et al. 2002; Johnson 2017 NTRS 20170001537; Plachta and Hartwig 2018 NTRS 20180004709) published the canonical architecture for cryocooler-plus-MLI ZBO at laboratory scale.

In-space demonstrations at small scale have accumulated more recently. Radio-frequency mass gauging (Zimmerli et al. NTRS 20070031548) flew on the Intuitive Machines Nova-C lunar lander in February 2024, producing the first microgravity flight data on cryogenic LOX and methane mass measurement (NTRS 20240014338). Chung, Dong, Wang et al. (npj Microgravity 10, article 65, 2024) reported charge-hold-vent and no-vent-fill demonstrations in a simulated propellant storage tank during tank-to-tank cryogen transfer in microgravity.

Boeing US 7,568,352 B2 (Grayson, Hand, and Cady, filed 2006, granted 2009) discloses a thermally coupled LO2/LCH4 storage vessel with liquid acquisition devices in each tank and a vapor-cooled shield surrounding the combined assembly, together with a cryocooler on the LOX side. The patent is the closest prior art to the dual-cryogen thermal storage stack used by Starship-class vehicles and is a ground-concept disclosure without a demonstrated flight architecture. Boeing US 9,395,048 B1 (Grayson and Cady, 2016) discloses a thermally protected liquid acquisition device for cryogenic fluids with integrated pressure and temperature control. United Launch Alliance US 8,196,869 B2 (Kutter, Zegler, Willey, Lin, Ragab, and Dew, granted 2012) discloses a cryogenic propellant depot with deployable sunshield and whole-vehicle axial-rotation centrifugal acquisition, addressing thermal management and bulk-liquid positioning at depot scale. MacDonald Dettwiler US 9,260,206 B2 (Allen, Lymer, Spring, and Ravindran, granted 2016) discloses a propellant transfer system for on-orbit storable-fluid resupply, a direct descendant of the Orbital Express demonstration. All four patents disclose components at sub-Starship scale and at single-vehicle operational scope. None discloses the integrated six-subsystem operational architecture required to aggregate propellant across a multi-flight campaign at the thousand-ton level.

Simonini, Dreyer, Urbano et al. (2024), "Cryogenic propellant management in space: open challenges and perspectives," npj Microgravity 10, article 34, is the most recent published review of the field. The review notes explicitly that ship-to-ship cryogenic transfer between independent spacecraft has never been demonstrated and that Starship-scale aggregation is an open problem. The NASA Marshall Cryogenic Fluid Management Portfolio Project released "Guidelines for In-Space Cryogenic Propellant Transfer" in April 2025 (NTRS 20250004625; AIAA ASCEND 2025-4122), a government-authored guidance document without architectural commitment. A two-phase-flow modeling paper co-authored by SpaceX engineers H. Tani, L. Blackmore, and G. Thome and published as NTRS 20250003325 discloses analytical work relevant to in-space cryogenic transfer without a corresponding patent filing, consistent with SpaceX's broader posture of trade-secret protection on the operational architecture.

None of this prior art operates at the thousand-ton scale, the ten-plus-flight aggregation cadence, or the mated-vehicle inertial scale that Starship requires. The component building blocks exist in the open literature. The integrated operational architecture at Starship scale does not.

The first of six: boil-off

Every pain point in the aggregation architecture compounds with time. Boil-off is the one that sets the time-budget ceiling. The first-principles calculation is straightforward.

A Starship in LEO sees three incident heat fluxes on its tank walls: direct solar at 1361 watts per square meter during the sun-lit fraction of each orbit, Earth albedo plus outgoing infrared at approximately 240 watts per square meter on the Earth-facing side, and a 2.7 K cosmic background on the deep-space-facing side. Averaged over an orbit and accounting for tank attitude, the absorbed flux on a polished stainless surface with solar absorptance 0.35 and IR emittance 0.12 (polished-new central values from NASA thermal handbook data) is approximately 137 watts per square meter. The external tank surface area of a Block 2 Starship with a common bulkhead between the methane and LOX tanks is approximately 950 square meters: cylindrical sections of 14.7 m (LOX) and 10.9 m (methane) plus two external oblate ellipsoidal domes, with the intertank bulkhead internal. Total heat load absorbed at central assumptions: approximately 130 kilowatts, allocated 72 kilowatts to the LOX side (area 529 m²) and 58 kilowatts to the methane side (421 m²).

Surface-finish aging and vehicle orientation widen the band. Across polished-new (α = 0.25, ε = 0.08) through aged-in-orbit (α = 0.50, ε = 0.25) cases, the absorbed flux ranges from approximately 100 to 200 watts per square meter, and the total heat load from approximately 95 to 190 kilowatts, a factor of two band. A sun-avoiding attitude with a dedicated radiator cools below the central case; a sun-facing attitude with aged skin runs at the high end.

That heat has to go somewhere. With no active cooling and no insulation, the only sink is boil-off: incoming heat vaporizes propellant, the vapor is vented, and the liquid approaches equilibrium with its vapor phase. Latent heat of vaporization at saturation and 1 atm is 213 kilojoules per kilogram for liquid oxygen and 511 kilojoules per kilogram for liquid methane. At central heat-load assumptions, boil-off runs at approximately 2.5 percent per day on the LOX tank and 2.9 percent per day on the methane tank (the LOX tank absorbs more heat because it has more surface area, but has proportionally more mass to vaporize; the methane tank has less area but a smaller inventory, making the two percent-per-day rates similar). Across the α-ε sensitivity band, LOX boil-off ranges from approximately 1.8 to 3.7 percent per day, methane from 2.1 to 4.4 percent per day.

At these rates, a bare-stainless Starship in LEO empties its LOX tank in approximately 27 to 40 days and its methane tank in approximately 23 to 34 days depending on skin aging. No realistic launch cadence closes an aggregation campaign inside that horizon.

Insulation buys orders of magnitude. Multi-layer insulation (MLI) in spaceflight applications has a standard performance benchmark around one watt per square meter of residual heat flux through the blanket after accounting for penetrations and mounting hardware. NASA's SHIIVER campaign produced the canonical modern data for cryogenic MLI on representative propellant tanks. At one watt per square meter residual, the Starship tank heat load drops from 130 kilowatts to approximately 0.95 kilowatts. Boil-off drops from 2 to 3 percent per day to approximately 0.02 percent per day. The horizon extends from weeks to multiple years. MLI at benchmark performance solves the aggregation-campaign problem if it can be installed and maintained intact at Starship tank geometry.

Active zero-boil-off buys another order of magnitude. A cryocooler removes heat from the propellant faster than it arrives, keeping the tank below vapor-equilibrium and driving boil-off to effective zero. The power budget is set by Carnot, with real cryocoolers operating at 5 to 8 percent of Carnot efficiency at LOX and methane temperatures. For the residual heat load after MLI at Starship scale, active ZBO demands approximately 25 kilowatts of electrical input for the LOX tank at 90 K and approximately 9 kilowatts for the methane tank at 111 K. Total power budget: approximately 34 kilowatts of continuous electrical draw at 300 K radiator reject, falling to approximately 25 kilowatts at 250 K shadowed-radiator reject. That is a manageable bus loading on a Starship-class solar array.

The architecture choice runs across four regimes:

SpaceX has disclosed none of this. The tanker thermal architecture, the residual heat flux specification, the cryocooler power budget, and the fallback strategy for campaign extension are absent from the public record. Public Starship refilling diagrams do not resolve the thermal architecture to the level at which the architecture decision is visible.

The second: zero-g liquid acquisition

Propellant in zero gravity does not sit at the bottom of the tank. In the absence of gravitational settling, the liquid-volume shape is set by surface tension and by whatever residual acceleration the vehicle carries from attitude thrusters and drag. Cryogenic methalox has a surface tension of roughly 10 to 15 millinewtons per meter, two orders of magnitude weaker than water. In a nine-meter-diameter tank, surface tension forces are overwhelmed by any residual acceleration above the 10⁻⁶ g range. For most of the aggregation campaign, the propellant mass is a single, roughly amorphous volume sitting against the tank wall on whichever side the residual-g vector points.

An outlet pipe at an arbitrary tank location sees, in general, a random alternation of liquid and vapor at the outlet face. Any transfer operation through that outlet ingests vapor. Vapor through a pump cavitates the pump. Vapor into a transfer line breaks the flow continuity. The outlet must be preceded either by a settling maneuver that collects liquid at a known location in the tank, or by a passive propellant-management device (PMD) that uses surface tension to channel liquid preferentially to the outlet.

Centaur-class upper stages solve this with ullage-motor settling. A small thrust pulse at 10⁻³ to 10⁻⁵ g for a few seconds before engine ignition collects liquid at the aft bulkhead. This works for a one-time engine ignition. It scales poorly to continuous-flow propellant transfer at Starship scale. At 10⁻³ g across a 30-minute transfer on a mated stack ranging from approximately 430 tonnes (transfer 1, receiver near-dry) to approximately 1830 tonnes (transfer 15, receiver near-full), the reaction-mass cost at autogenous warm-gas methalox specific impulse (Isp approximately 300 s on scavenged CH4/O2, per the ULA Integrated Vehicle Fluids architecture U.S. Patent 10,717,550) is 2.6 to 11 tonnes per transfer. Across a fifteen-tanker aggregation campaign, the transfer-burn settling-mass budget integrates to approximately 100 tonnes. Duty-cycled coast-settling at 10⁻⁵ g across a 30-day aggregation adds approximately 8 tonnes at an 11 percent duty cycle (10 minutes per 90-minute orbit). Grand total settling-propellant budget for a 15-flight, 30-day campaign: approximately 110 tonnes, scavenged from tanker ullage (no dedicated cryogenic RCS propellant).

PMDs, well-validated for storable propellants in geostationary-satellite tanks, have not been scaled to Starship dimensions. Surface-tension forces at a 9-meter tank radius are overwhelmed by even small residual accelerations, so a passive PMD at Starship scale cannot hold liquid against arbitrary attitude. The workable PMD architectures are hybrid: PMDs at the outlet region, settling thrust during active transfer, and low-impulse continuous settling during coast to keep the propellant positioned.

The specific SpaceX architecture is not disclosed. The candidate space is bounded by physics (surface tension scaling with tank radius, residual-g levels available from the attitude-control system, reaction-mass budget constraints) but the selection within that space is proprietary.

The third: sloshing at transients

A step change in propellant flow rate at the start or end of transfer introduces momentum. The propellant mass inside the tank responds by sloshing. In zero-g, slosh has no gravity restoring force; damping comes from viscosity, from baffle drag, and from wall friction. Cryogenic methalox in smooth-walled tanks has very low damping.

The classical Abramson result for the fundamental lateral slosh mode of a cylindrical tank gives the slosh period as a function of effective gravity and tank radius. For a 9-meter Starship tank at deep fill, the period varies inversely with the square root of the effective g:

At pure coast-g, the slosh mode persists across the entire transfer. The fundamental frequency is low enough to interact resonantly with attitude-control thrusters, with feed-pressure oscillation, and with transfer-line pressure modes. Without active slosh suppression, coast-g transfer is not a viable operating point.

At 10⁻³ g settling thrust, slosh periods fall into the minute scale and active damping becomes tractable. Model-predictive control with slosh-mode estimation is the standard approach. The tradeoff is the settling-thrust mass budget noted above.

Active damping architectures extend the tractable envelope. Internal baffles, staged flow-rate transitions, dedicated slosh-suppression thrusters firing at the slosh frequency in anti-phase, and magnetic-bearing-driven momentum wheels acting on liquid-magnetic-fluid sub-volumes have all been proposed in the literature. None has flown at Starship scale. The specific suppression architecture chosen is a trade-secret.

The fourth: zero-g mass gauging

Mission planning for a Mars departure burn requires knowing how much propellant is on board. On Earth, the answer is trivial: measure liquid surface height, compute tank volume from geometry. In zero-g, there is no surface.

Flight precedent is limited. Radio-frequency mass gauging (Zimmerli et al., NASA Glenn) flew on the Intuitive Machines Nova-C lunar lander in February 2024 and produced the first microgravity flight data on cryogenic LOX and methane mass measurement (NTRS 20240014338). Reported accuracy at the Nova-C tank scale (a few cubic meters) is in the 1 to 3 percent band under the specific in-flight conditions. At Starship tank scale the same method faces a different trade: the larger tank volume improves some error terms and degrades others. Ground-tested systems that achieve sub-1 percent accuracy on simulated cryogenic tanks under microgravity conditions have not yet flown. Public SpaceX disclosure on the Starship mass-gauging architecture is limited to general acknowledgment of the problem.

Candidate gauging methods and their approximate accuracy at Starship scale:

A tanker delivering 150 tons with 15 tons of uncertainty is a 10 percent per-transfer error. Stack fifteen transfers and the aggregate uncertainty compounds. A Mars departure burn plan tolerates sub-one-percent error on total propellant mass; at Starship scale that is a 15-ton envelope across a 1500-ton load. No flown system meets this specification.

The candidate architectures are sensor-fusion. No single method crosses the 1 percent threshold at Starship scale, but weighted-average estimates from PVT, RF, and fiber-optic methods in combination can plausibly reach 1 percent or better. The fusion architecture, the sensor layout, and the in-flight calibration procedure are unpublished.

The fifth: cryogenic androgynous docking

The original framing of this pain point was the mating-seal leak rate. That framing was wrong. Heritage cryogenic seal architectures (metal C-ring, double-ring with purge, elastomer-backed metal face) meet the required leak rate with orders-of-magnitude margin. At Starship-scale interface dimensions, the cumulative leak over a 30-minute transfer is sub-microgram. Seal leak rate is not the hard problem.

The hard problems are mechanical. A 15-tanker aggregation campaign means 15 docking cycles per receiver Starship. Across a Mars program, the mate-demate count rises into the thousands per fleet. The interface must survive:

1. Repeated mate-demate cycling without surface wear or gasket degradation.
2. Thermal shock on every mate from ambient hardware contacting 90 K fluid-side surfaces in seconds.
3. Docking loads on the order of several kilonewtons from final-approach capture and axial separation.
4. Misalignment tolerance in the centimeter range during autonomous docking.
5. Cryo-compatible seal material that stays compliant at 90 K. Most elastomers glass below 180 K and require metal-backed designs or exotic polymer formulations.
6. Contamination management. Ice forms on the seal face during ambient-hold. Debris accumulates from repeated docking. Seal-face cleaning or self-wiping architectures become service-life constraints.

No public component combines all six. The Common Berthing Mechanism used on the ISS is ambient-temperature only. The various orbital-docking mechanisms (APAS, NDS, IDSS) are mechanical-capture devices without cryogenic fluid transfer. The candidate space is a cryogenic androgynous berthing mechanism with integrated fluid transfer, and the specific SpaceX implementation is unpublished.

Candidate architectures include metal C-ring seals with cryogenic-resilient plating plus spring-loaded misalignment capture; double-ring seals with purge and self-cleaning wipe action; and scaled-up NDS-class androgynous mechanisms hardened for cryogenic methalox service. The integration of thermal management, structural capture, and fluid transfer into one mechanism is the engineering package whose details are trade-secret.

The sixth: attitude coupling between mated vehicles

Two Starships mated end-to-end at Block 2 geometry form a stack approximately 104 meters long (2 × 52.1 m). During each transfer, mass moves continuously from the tanker to the receiver. At 100 tonnes delivered per transfer over a 30-minute window, the flow rate is approximately 55 kilograms per second. Per-transfer center-of-mass drift ranges from approximately 12 meters (transfer 1, small mated mass) to approximately 3 meters (transfer 15, large mated mass). The donor dry-mass assumption (publicly stated target near 100 tonnes, telemetry-derived central estimate 110 to 165 tonnes; this note uses 140 tonnes as central) feeds both the CoM drift and the moment-of-inertia calculations.

The moment of inertia of the mated stack about its transverse midpoint axis varies across the aggregation campaign. Under a uniform-rod approximation on the 104-meter mated stack, at transfer 1 with the tanker carrying approximately 130 tonnes of usable residual and the receiver at dry-plus-ascent-residual (160 tonnes total), mated-stack mass is approximately 430 tonnes and MoI is approximately 3.9 × 10⁸ kilogram square meters. At transfer 15, with the receiver near-full at approximately 1560 tonnes of propellant plus 140 tonnes dry, mated-stack mass reaches approximately 1830 tonnes and MoI reaches approximately 1.67 × 10⁹ kilogram square meters, roughly four times the transfer-1 value. The attitude-control authority must maintain pointing tolerance for thermal management, communications, and settling-thrust continuity while compensating for a continuously shifting center of mass and internal fluid momentum from slosh.

The slosh period at coast-g (16.5 minutes) is comparable to the total transfer duration (30 minutes). The slosh mode cannot be treated as a fast transient and filtered out of the control loop. It must be modeled explicitly in the attitude controller. Standard spacecraft attitude-control approaches assume rigid-body dynamics with small perturbations. A mated-tandem stack with continuous mass transfer, asymmetric propellant distribution, and slow-mode slosh is not a rigid body.

Candidate architectures are model-predictive control with full-order dynamics including slosh-mode estimation, fluid-flow-rate measurement, and real-time moment-of-inertia tracking. The control law replans the center-of-mass trajectory every cycle. Allocation to thruster clusters accounts for the changing geometry. No public control architecture at this scale has flown. The specific SpaceX implementation is unpublished.

The integrated architecture

The six pain points compound. Boil-off sets the time-budget ceiling. Liquid acquisition and slosh set the lower bound on settling thrust. Settling thrust drives reaction-mass consumption which compounds the time budget. Mass gauging sets the precision floor on the departure burn plan. The cryogenic docking interface sets the cycle-life budget of the tanker fleet. Attitude coupling sets the control-authority budget of both vehicles. An architecture that closes one pain in isolation buys nothing if another pain re-opens the same time budget.

The integrated architecture closes all six coupled pain points with one commitment per pain. The companion technical disclosure gives the full quantitative specification at patent-application grade; the commitments summarized here are:

1. Thermal. MLI blanket sized for 1 watt per square meter residual heat flux over the methalox tank surface, supplemented by active cryocooler loop sized for approximately 34 kilowatts electrical at steady state. Solar array sized for cryocooler duty cycle plus bus loads. Sun-shield configuration maintained during aggregation-hold attitude.
2. Liquid acquisition. Hybrid architecture. Surface-tension PMD at the outlet handles start-stop transients and low-rate polish. Light continuous settling thrust at approximately 10⁻⁵ g during steady transfer keeps the propellant bulk positioned.
3. Slosh. Staged flow-rate ramps at transfer start and stop. Model-predictive attitude control with explicit slosh-mode estimation at coast-g, with a 10⁻³ g settling thrust fallback available on demand.
4. Mass gauging. Multi-sensor fusion. PVT as the baseline, supplemented by RF tank level and fiber-optic distributed temperature. Target fusion accuracy 1 percent or better at Starship scale.
5. Cryogenic docking. Metal face seal with autogenous-methane purge. The seal is purged at preload with gaseous methane; methane condenses on cold surfaces as a secondary leak barrier, with no helium required. Spring-loaded misalignment capture. Cryo-compatible plating selected for 90 K thermal-cycle compatibility.
6. Attitude coupling. Model-predictive controller with real-time CoM tracking, mass-flow-rate feedback, and slosh-mode state estimation. Control law replans every cycle.

No individual component of this architecture is novel; the contribution of this note and its companion disclosure is the integration. The integrated architecture closes the six coupled pain points across a ten-plus-flight aggregation campaign, at 1000-ton cryogenic scale, with ship-to-ship mating at 90 K and attitude coupling across a 104-meter mated stack carrying 1500 tons of live mass flow. The Starship-scale operational integration is the unpatented, undocumented part of the public record that this disclosure places into the public domain.

Each component draws on published prior art at sub-Starship scale. The dual-cryogen thermally coupled storage stack with LAD and vapor-cooled shield is disclosed in Boeing US 7,568,352 B2 (Grayson, Hand, Cady, 2009) at single-vehicle concept level. The cryogenic depot with deployable sunshield and axial-rotation centrifugal acquisition is disclosed in ULA US 8,196,869 B2 (Kutter et al., 2012). The thermally protected LAD for cryogenic fluids with integrated pressure and temperature control is disclosed in Boeing US 9,395,048 B1 (Grayson and Cady, 2016). The on-orbit propellant transfer servicer architecture for storable fluids is disclosed in MDA US 9,260,206 B2 (Allen, Lymer, Spring, Ravindran, 2016). The cryogenic androgynous fluid-transfer coupling with electro-permanent-magnet latching is disclosed in the Altius Space Machines (Voyager Technologies) patent family (NASA SBIR Phase II, 2020; integrated on Eta Space LOXSAT, 2026). Single-spacecraft model-predictive control is disclosed in MERL US 10,967,991 B2 (Weiss, Zlotnik, Di Cairano, 2021). RFMG is published NASA Glenn prior art with a 2024 flight precedent.

The component building blocks exist in the open literature. The integrated operational architecture at Starship scale does not.

Validation protocol

The integrated architecture is testable against the first ship-to-ship transfer flight in the IFT-12 through IFT-15 window. The observables that will score the disclosed architecture against SpaceX's implementation:

1. Aggregation duration versus propellant retention. The ratio of propellant delivered to propellant retained after transfer characterizes effective boil-off plus transfer-loss mass. At benchmark MLI plus active ZBO, the disclosed architecture predicts retention exceeding 98 percent at a 30-day campaign scale (sub-0.1 percent per-day residual loss against tank inventory).
2. Stack attitude excursions during transfer. The pointing-error envelope during each 30-minute transfer characterizes the effectiveness of the attitude controller against CoM drift and slosh. The disclosed MPC architecture predicts pointing retention of approximately 0.5 degree or better at 10⁻³ g settling, with transient excursions under 1 degree at flow-rate ramp-up and ramp-down.
3. Mass-gauge closure across transfer. The difference between donor-measured mass transferred and receiver-measured mass received characterizes sensor-fusion accuracy. The disclosed three-method fusion architecture predicts per-transfer closure of 0.3 to 1 percent and cumulative aggregate accuracy better than 1 percent at 1500-ton scale.
4. Cryogenic seal performance at early mate-demate cycles. Per-cycle leak rate and mate-quality telemetry on the first three to five mate-demate cycles partially validates the docking architecture. The disclosed metal face seal with autogenous methane purge predicts leak rate below 10⁻⁶ mbar·L/s at 90 K across the service envelope. Full cycle-life validation at the hundreds-to-thousands-of-cycles scale requires fleet-scale operating history beyond the IFT-12 through IFT-15 window.
5. Slosh-mode observations during transient. Fill-sensor data at transfer start and stop characterizes the low-g slosh behavior. The disclosed architecture predicts Abramson-scaling slosh periods of approximately 16 minutes at coast-g (10⁻⁵ g) and 1.7 minutes at 10⁻³ g settling, within a tolerance band of approximately ±20 percent for tank-geometry and fill-fraction effects.

Each observable has a first-principles prediction from the architecture above. Each will be scored against the as-flown SpaceX implementation as the aggregation campaign progresses. If the as-flown architecture converges on the disclosed one, the disclosure has derived SpaceX's private design from public physics alone. If it diverges, the physics constrains the divergence to a narrow set of alternatives each of which is disclosed in the companion document.

Why this note is public

The Mars architecture depends on orbital propellant transfer. So does every other public commercial or civil plan that relies on cryogenic fluid management in orbit at scale. Lunar Gateway resupply, nuclear thermal rocket architecture, electric-propulsion tugs, commercial depot concepts, and the full stack of in-space logistics infrastructure all run into the same six pain points.

Coracle's method is to derive trade-secret technical architectures from public patent absences and from first-principles analysis of public operational data. The six pains here meet both conditions. They are absent from SpaceX's patent record, and derivable from the physics of orbital cryogenic fluid handling at Starship scale without reference to any non-public source. The derivation is the method, and publication places the integrated architecture in the public domain as prior art under 35 U.S.C. §102.

Any party contemplating a patent filing in the orbital-refilling architecture space after the publication date above inherits this disclosure as prior art of record. Any engineering team building an orbital-refilling implementation can take the architecture, fit it to their specific vehicle geometry, and proceed. The commissioned work at Coracle is narrower: implementation-specific scoping, parameter fitting to a specific vehicle, trade studies against a specific mission profile. See engagements for the paid scope.

A companion technical disclosure is published alongside this note with composition windows, process parameters, reference-numbered figures, and alternative embodiments at patent-application-grade specificity. Prior-art treatment and the full 35 U.S.C. §102 defensive-publication clause live there.

Sources and further reading

SpaceX Starbase presentation, 4 April 2024 (Block 2 stretch announcement, 1500 t propellant capacity). Ringwatchers, "Bigger is Better: Starship's Extended and Optimized Tanks," ringwatchers.com (updated January 2026). Kent Chojnacki interview, Spaceflight Now, 1 November 2024 (IFT-3 internal-transfer measurement reported to approximately 5 percent accuracy). NASA Office of Inspector General, "NASA's Management of the Human Landing System Contracts," IG-26-004, March 2026 (more than 10 tanker flights per HLS mission; one-tanker-per-6-days cadence target). S. Rotenberger, D. SooHoo, G. Abraham (Northrop Grumman Corp.), "Orbital Express fluid transfer demonstration system," Proc. SPIE 6958, 695808 (15 April 2008), DOI 10.1117/12.783948. G. Grayson, M. Hand, E. Cady (Boeing), "Thermally coupled liquid oxygen and liquid methane storage vessel," US 7,568,352 B2 (granted 2009). G. Grayson, E. Cady (Boeing), "Thermally protected liquid acquisition device for cryogenic fluids," US 9,395,048 B1 (granted 2016). B. Kutter, F. Zegler, C. Willey, J. Lin, M. Ragab, M. Dew (United Launch Alliance), "Cryogenic propellant depot and deployable sunshield," US 8,196,869 B2 (granted 12 June 2012). A. Weiss, D. Zlotnik, S. Di Cairano (Mitsubishi Electric Research Laboratories), "Model predictive control of spacecraft," US 10,967,991 B2 (granted 6 April 2021). F. Zegler (United Launch Alliance), "Integrated vehicle fluids," US 10,717,550 B1 (granted 21 July 2020). A. Allen, J. Lymer, K. Spring, R. Ravindran (MacDonald Dettwiler), "Propellant transfer system and method for resupply of fluid propellant to on-orbit spacecraft," US 9,260,206 B2 (granted 2016). D. Plachta, P. Kittel, "Updated zero boil-off cryogenic propellant storage analysis applied to upper stages or depots in a LEO environment," NTRS 20030067928 (2003). D. Plachta, J. Hartwig, "NASA cryocooler technology developments and goals to achieve zero boil-off and to liquefy cryogenic propellants for space exploration," NTRS 20180004709 (2018). SHIIVER final report, "Demonstration of multilayer insulation, vapor cooling of structure, and mass gauging for large scale upper stages," NTRS 20205008233. Simonini, Dreyer, Urbano et al., "Cryogenic propellant management in space: open challenges and perspectives," npj Microgravity 10, 34 (20 March 2024). J. N. Chung, J. Dong, H. Wang et al., "Demonstration of charge-hold-vent (CHV) and no-vent-fill (NVF) in a simulated propellent storage tank during tank-to-tank cryogen transfer in microgravity," npj Microgravity 10, article 65 (2024), DOI 10.1038/s41526-024-00403-6. "Guidelines for In-Space Cryogenic Propellant Transfer (ISCPT)," NASA MSFC Cryogenic Fluid Management Portfolio Project, April 2025 (NTRS 20250004625; AIAA ASCEND 2025-4122). G. Zimmerli, M. Vaden, "Radio frequency mass gauging of propellants," NTRS 20070031548 (2007). "Results from the Radio Frequency Mass Gauge Technology Demonstration on the IM Nova-C Lunar Lander," NTRS 20240014338 (2024). H. N. Abramson, "The dynamic behavior of liquids in moving containers," NASA SP-106 (1966). R. F. Barron, Cryogenic Heat Transfer, 2nd ed., CRC Press (2016). D. G. Gilmore, ed., Spacecraft Thermal Control Handbook, Vol. 1, 2nd ed., AIAA (2002). B. Wie, Space Vehicle Dynamics and Control, 2nd ed., AIAA (2008). NIST WebBook and REFPROP 10.0 for cryogenic propellant properties at NBP. Full prior-art treatment with per-patent claim distinctions in the companion technical disclosure.

·