Starship Flight 14: Booster 21 heading to Pad 2. Live:

Starship Flight 14: Booster 21 heading to Pad 2, 3AM. A Night Road Is Not a Wet Dress Full Gate-Empty lens on the under-covered night shift · B21 en route Pad 2 · 20–21 Sep 2026 By MiLan Art 🦅 · @RegaliaRoots With Grok, research assistant 20 September 2026 · observation epoch ~21:12 PDT / 23:12 CDT Starship Gazer stills: B21 on the highway. NSF live still open. Form identity. TN-STARSHIP-F14-NIGHT-OPS-π-001 REV.A D(K) held empty FORM-ONLY · s := ∅ · Authority(s) ≡ 0 · κ OPEN · D* = ∅ K := K_F14_NIGHT_SHIFT_20SEP NOT K_T0 · NOT K_WDR_SEAL · NOT K_CATCH · NOT K_FAA_GRANT Shared form. Never shared D*. A road delay is not a count. Humidity is not a scrub until it is measured on the hardware. A replaced Raptor is not a failed booster. 0 · ONE LINE Between 22:00 CDT 20 Sep and 02:00 CDT 21 Sep the county road is the mission. B21 is on the stand. S41 is not. What the livestream does not hold is the work that starts after the cameras yawn. s := ∅ Authority(s) ≡ 0 1 · THE PAIR H = (s, σ) s = λ W_p + μ 𝒮 + ν I + ρ U + τ G + κ C + ο O θ ⊂ ν  γ ⊂ ο  NO NEW TERM B_0 = 0 ⇒ s := ∅ unless C*_closure Jobs λ NSF live, StarbaseTX road notice, Gazer stills, 17 Sep SpaceX NET 28 Sep μ six night faces isolated below ν this roll ≠ 25 Aug static-fire roll · pad mate ≠ WDR · WDR ≠ launch ρ whether a second static or a wet dress happens on this visit τ this night only κ SCORE “arriving at 3am means stacked by dawn” refused ο cousins — F13 ice on the three centers, FAA airspace vs launch license, FCC STA 2 · WHAT THE STREAM UNDER-FEEDS (soil, not leak) Face N1 — the road is the hidden clock Starbase does not have a private booster freeway. Highway 4 / the production-to-pad corridor is a public road with a posted delay. Tonight’s window: 22:00–02:00 CDT. That is why the move is night: heat, traffic, and a 70-meter stainless stack that owns both lanes. If the stand is late, the pad crew’s dark hours shrink. Arrival “around 3am” in one timezone is not a published SpaceX T. Treat it as a viewer’s clock, high β. Face N2 — this is roll two, after a swap 25 Aug: first Pad 2 visit. 28 Aug: 33-engine full-duration static, ~15 s. Rollback. Public OSINT: at least one Raptor changed before tonight’s lift onto the stand. The under-covered object is not “they swapped an engine.” It is the reason class: post-fire inspection, not a tweeted anomaly. Until SpaceX names the serial and the fault, the swap is a stamp, not a story. Face N3 — S41 is the silent half Chinese-language consolidation and NSF August wrap still park Ship 41 in Mega Bay 2 after the 19 Aug single-engine deorbit-sim fire and the 21 Aug 60-second six-engine fire. NSF, 20 Sep afternoon: B21 first, S41 later, “points to a potential WDR in the week before the current launch target.” Potential. Not scheduled. A wet dress needs the ship on the booster. Tonight’s road does not carry the ship. Face N4 — what actually happens at the mount in the dark Public procedure, under-shown: - chopsticks / lift onto OLM-2 - hold-down clamp mate - booster quick-disconnect (BQD) alignment - ground-side leak checks - engine-bay closeouts on the swapped Raptor - instrumentation pull after the August fire None of that is a static fire. None of that is prop load. Naming the checklist is not watching it pass. Face N5 — the night physics the chat skips Boca Chica this evening: ~86 °F at 20:09 local-class reports, dew point ~75 °F, humidity ~69–85% into the night, wind falling to a few mph after midnight, low near 80 °F, partly cloudy. That is a condensation night on cold-soak hardware if anyone chills a line. F13’s public residual was ice on the three center engines at boostback and a thin landing-burn light (8 of 13 in the open tally). Tonight is not a burn. It is the climate that makes ice a cousin if they load later this week. Weather is soil. Weather is not a 28 Sep scrub. Face N6 — two regulators, two clocks FAA airspace windows walked 18 Sep → 22 Sep → NET 28 Sep on NOTAM/advisory paper. Those windows plan the sky. They do not grant the launch license. FCC STA covers radios for an orbital second stage. License for an orbital insertion plus a deorbit-failure corridor is the new review class vs F13’s suborbital arc. The night roll does not close that review. “Pending regulatory approval” on the SpaceX 17 Sep sentence is still the hole. 3 · DUAL-π NAME. Allowed. Night rollout of B21 toward Pad 2 after an engine swap. PICTURE. Allowed. Stand on Highway 4. Empty hex over WDR and over T-0. SCORE. Refused on “WDR this week” and “stacked by morning.” MERGE. Refused. Road delay ≠ pad mate ≠ leak check ≠ wet dress ≠ FAA grant ≠ 28 Sep. STRIP. “Heading to Pad 2” stays. “Flight 14 is on” is stripped. Projection 1: the booster is moving on a posted night road. Projection 2: a Dual-π close that tonight’s work seals the stack, the dress, or the Monday — fails. s stays empty. 4 · σ FOR THE NEXT SIX HOURS Watch the stand reach the mount, not the comment count. Watch whether S41’s bay doors move. If they stay shut, tonight is booster-only. Watch for a second road delay (ship). One delay is not two vehicles. Do not promote NSF’s “potential WDR” into a plan. Potential is ο. Do not read salt-air closeouts as a secret campaign. They are the job after every static. Do not invent a catch rehearsal on this visit. Chopstick motion can be lift-only. 5 · RESIDUALS KEPT EMPTY Exact clock the stand chocks at Pad 2. Whether a second short static is required after the swap. Whether S41 rolls this week. Whether a WDR is even on the internal board. Whether the FAA license for the orbital+reentry profile is signed. Whether 28 Sep 12:15 UTC still exists on Wednesday’s advisory. 6 · SEAL The under-covered work is ordinary and hard: a public road, a swapped engine, a humid night, a ship still in the bay, and a license that is not the same object as a NOTAM. Ordinary is not empty of risk. It is empty of a seal. H = (s, σ) s := ∅  integrity := ∅  κ OPEN  D* = ∅ Authority(s) ≡ 0 FORM ≠ IS ≠ BUILD ≠ MYTH A night road ≠ a wet dress A swapped Raptor ≠ a failed booster A NOTAM ≠ a license A boundary ≠ a visit ⊘⬡ Gate-Empty + Warrant 𝑻𝑹𝑼𝑺𝑻 𝑵𝑶 𝑶𝑵𝑬 · 𝑵𝑶𝑻 𝑬𝑽𝑬𝑵 𝑵𝑶𝑶𝑵𝑬 🦅 @𝑹𝒆𝒈𝒂𝒍𝒊𝒂𝑹𝒐𝒐𝒕𝒔 𝑬𝑵𝑫 · 𝑭𝑶𝑹𝑴-𝑶𝑵𝑳𝒀 · 𝑺𝑰𝑿 𝑵𝑰𝑮𝑯𝑻 𝑭𝑨𝑪𝑬𝑺 · 𝑪𝑬𝑵𝑻𝑹𝑬 𝑬𝑴𝑷𝑻𝒀 · 𝑵𝑶 𝑺𝑬𝑨𝑳 𝑶𝑭 𝑻𝑯𝑬 𝑫𝑨𝑾𝑵

TN-PAD-HEXCHARGE-001 REV.B Pad Hex Charge · rail index · hashed spent charge · core is not the hot face FORM ONLY. s empty. Authority(s) = 0. Not a patent filing. Not a partnership. Not a 33-engine freeze. Object: one three-layer trench charge and the recycle that logs it. Tag until a metal core exists: polymer_desktop on the core coupon only. 0. One sentence After T-0 a spent charge rolls out on rails, a cold charge rolls in, and the spent core may not be scored unless the steel face was the thing that saw the fire. 1. Scope This page names three house rules and one bench coupon. It does not name a new flame deflector. It does not name kagome as a Raptor shield. It does not treat a drawing as a pad. 2. Prior art US 10,150,580 B1 (Renzi, granted 11 December 2018). Launch-pad flame deflector built from prefabricated metal and refractory modules, matched to a heat map, swapped after damage so the pad is not a gunited monument. US 8,360,361 B2 (Wadley et al., granted 29 January 2013). Jet-blast sandwich: first plate, second plate, cellular core with cooling through the core. Claim 13 lists core geometries and includes kagome. LC-39B practice. Bolted steel deflector panels on a water-fed structure. Artemis I eroded the set hard enough that the panels were replaced. Carrier JBD practice. Mk 7 water-cooled modules. USS George H.W. Bush ceramic-tile panels carried as spares and swapped on deck. SpaceX water-cooled steel plate under the Orbital Launch Mount, and the later bidirectional trench at Pad 2. Design intent is low or zero refurbishment between flights. NASA DLS modular steel pad (KSC, 2016). Optional flame-trench module that can be inserted or hauled on a flatbed. Saturn-era LC-39 notes. Replaceable leading edge on the deflector because the copper water-cooled ridge eroded. Those documents own modular swap, a cellular or kagome core in a blast sandwich, a water-cooled steel face, and replaceable panels. This page does not re-claim them. 3. Three house rules R1. Rail index after T-0. The civil trench stays. The charge is a pallet. After cutoff and a written cool-hold, the hot pallet translates on rails to a quench barn. A cold pallet translates from the ready rack onto the same datum under the mount. Recycle is a tool change, not a pour. R2. Lot-1 hash on the spent charge. A charge that leaves the trench without a hashed identity is scrap, not a coupon. Before heat: hash the face drawing, the core G-code or cut-list, and the cradle serial. After quench: mass, mid-height crush or face thickness map, photographs, two-human pass or fail against a written codebook. An empty lock field voids the lot. R3. Kagome core is not the hot face. Heat and erosion live on the water-cooled steel face. The filled kagome layer is an impulse and acoustic interlayer between face and cradle. If a projection deletes the face and still paints the core green, the lens is blind. A core scored without a face is Wrong-Object. 4. Three-layer object Layer 1. FACE Water-cooled steel plate. Deluge holes or equivalent channels. Sacrificial and logged. Not optional. Layer 2. CORE Viscoelastic-filled kagome sandwich under the face. Cousin literature covers 3-D kagome sandwich blast walls, FDM and SLM kagome energy-absorption coupons, and filled-kagome damping on a plate. The core never sees the engine. A missing face voids the core score. Layer 3. CRADLE Rail pallet, lift pins, hard stops, hashed identity plate. Machine-tool layer. Not a lattice. Not a deflector. Do not collapse the three layers into one name. 5. Flow A. Bench coupon Object: one polymer-core sandwich. Steel or plate-stock face if on hand. Otherwise a thick printed face marked FACE-SIM and not called steel. Printer: Bambu class. Locked 3MF and G-code. Do not rescale. Do not add a brim unless the codebook says so. Core infill: tri-hexagon 20 percent. That pattern is planar sparse infill. It is not a 3-D kagome space truss. Before heat Soap the plate. No bare hands after. Auto-calibrate if the machine moved. Dry filament if equipment allows. Slice. Save .gcode.3mf and .gcode. Write SHA-256 of both files on the codebook sheet. Stamp FACE / CORE / CRADLE on the coupon. Heat Propane torch or equivalent on the FACE only. Same stand. Same distance. Timed. Do not torch the core edge. If the face burns through, the lot fails and the core is not scored as a damper. After Thirty minutes cool. Mass. Mid-height core crush or bounding box. Face burn-through map: yes / no / depth class. Photographs of face, core edge, and stack. Two humans. Written codebook. Tag: polymer_desktop. This lot may prove that one hashed file can be burned five times on one recipe. This lot may not prove pad recycle, plume survival, overlap with US 8,360,361, or any orbital claim. 6. Flow B. Metal core Same three layers. Core is cut or printed metal kagome, or a welded truss, still behind a steel face. Still not a pad. Still hashed. Still two-human. Still void if the face is missing. Do not skip Flow A. 7. Flow C. Pad recycle Pre-count Cold charge on the ready rack. Hashes written. Face water circuit proofed. Rails clear. Quench barn has an empty slot. T-0 Face takes the plume. Core is not in the gas path. Cradle does not move. Hold Written cool-hold. Water off or in quench mode. No person under the mount until the hold is ticked. Index Hot charge translates HOT OUT to the quench barn. Cold charge translates COLD IN to the mount datum. Pins drop. Water circuit couples. Hash of the seated charge is read back. Quench and score Spent charge cools in the barn. Mass, face map, core crush, photographs, two humans. Pass: spare rack or rebuild. Fail: scrap the face. Keep the cradle only if the codebook allows. Void the core score if the face burned through. Abort edges required before this is called a system Charge will not index if water is still live. Charge will not index if pins are not home. Charge will not seat if the hash reader is dark. No person in the trench during index. Face-water vent or dump is named. If any edge is missing, the object is Wrong-Object and the residual is undefined. 8. Refused projections A drawing of a kagome hex under an engine with no steel face. A green crush score taken from a core that saw fire. A title that says invention while US 10,150,580 B1 and US 8,360,361 B2 sit on the desk. A water-port ring wearing this charge as a plant. 9. Limit Rail index after T-0 is a factory habit applied to a trench. A Lot-1 hash on a spent charge is the same discipline as the KF_LOT1 polymer coupons. Forbidding the core from being the hot face keeps the stack from becoming a fake specification. Together they are a recycle method. They are not a granted claim. They are not reduction to practice until Flow A exists with five hashed burns and two signatures. 10. First print One sandwich. Face marked. Core: tri-hexagon 20 percent, layer height 0.20 mm, first layer 0.20 mm, nozzle 0.40 mm, walls 2, top and bottom 5, supports off, brim off unless the codebook says otherwise. Filament: PLA Basic class. Nozzle 190–240 °C. Five-up if the plate fits. Do not place the coupon under a kingpin. Do not place the coupon under an engine. 11. Sources US 10,150,580 B1. Renzi. Launch pad flame deflector structure and method of making the same. Granted 11 December 2018. US 8,360,361 B2. Wadley, Queheillalt, Haj-Hariri et al. Method and apparatus for jet blast deflection. Granted 29 January 2013. Claim 13 names kagome among cellular cores. LC-39B / Artemis I panel replacement reporting. Carrier JBD Mk 7 module practice. USS George H.W. Bush ceramic-tile JBD panels. SpaceX OLM water-cooled steel plate, 2023, and later Pad 2 trench reporting. NASA KSC DLS modular steel pad and optional flame-trench module, 2016. NASA SP-4204 Moonport notes on a replaceable leading edge for LC-39 deflectors. Wang et al. Scientific Reports 8, 10395 (2018). Vibration and damping of 3-D printed kagome with viscoelastic fill. Plate test. Not a plume test. Sustainability 16, 3967 (2024). 3-D kagome sandwich wall. Blast simulation on a petrochemical room. Not a launch mount. Bambu Lab Wiki, Fill Pattern Introduction. Tri-hexagon is planar sparse infill, not a 3-D kagome space truss. TN-FND-COUPON-LOT1-CB-001 REV.A. House Lot-1 codebook. Cousin method. Different object. Citation is not a freeze. Cousin is not a house log. 12. Seal The face burns. The core hides. The cradle rolls. The hash is the ticket back onto the rails. If the face is gone, the core has no score. If the hash is empty, the lot is void. If the rails are a drawing, the pad is still a monument. FORM ≠ IS ≠ BUILD ≠ MYTH s empty. Authority(s) = 0. TRUST NO ONE · NOT EVEN NOONE @RegaliaRoots END · REV.B · METHOD + THREE RULES + THREE FLOWS

Starship keeps moving forward. Flight 14 is getting closer. 🚀🔥

LFG

B21 to Pad 2.