Constellation shape

A 648-satellite polar shell, the OneWeb pattern

A 648-satellite polar shell at 1200 km as idealised geometry: what a dense high-LEO constellation delivers at every latitude, including above the Arctic Circle.

How the planes are arranged, and how this site arranges them. Every figure here comes from a Walker-delta shell: planes spread evenly around the whole circle of right ascension. The real fleet flies a star pattern, and its twelve planes sit 15.2 degrees apart in the catalogue snapshot, spanning half the circle rather than all of it. Those are not the same arrangement. At this inclination, planes half a turn apart trace circles only 4.2 degrees apart over the ground, so spreading them around the whole circle spends half of them twice. This shell is dense enough to close under either arrangement at the mask it is published with, so nothing on this page is distorted by the choice. It costs a great deal at a harsher mask, where the shell stops closing. The measured comparison quantifies it against the real fleet.

648 satellites in 12 planes of 54, at 1200 km and 87.9 degrees. Density and altitude together. At 1200 km each satellite covers a wide footprint, and at 648 of them the shell closes everywhere rather than only where the orbit lingers.

At latitude 55 with a 25 degree elevation mask, this shape closes: no gap at any sampled longitude, a full 1440 service minutes a day. The service band reaches the poles.

It holds that everywhere it reaches: continuous at the equator and at 70 degrees alike. A shell this dense stops being a coverage question and starts being a capacity one.

Scenario

SatellitesWalker 648/12/1

Orbital planes

Altitude1200 km

Inclination87.9°

Min elevation25°

Service latitude55° N/S

Latitude only. Results are worst-cased around the whole latitude circle, so this is the band a city sits on rather than the city itself.

Walker delta 648/12/1 · 48 h window · 40 s step · worst of 8 longitudes at 55°

Worst outage

none

continuous service

Average wait

no gaps

Windows / day

one unbroken window

Service / day

1440 min

of 1440 min

Coverage

100.0%

time in view

Service timeline at your latitude

48 h · worst-case longitude at 55°

Continuous service. No gaps at this latitude.

Outage vs constellation size

Worst outageAverage wait

At 55° latitude, min elevation 25°, 1200 km / 87.9°. Log scale. Labels mark the worst outage.

Worst outage vs latitude

648 sats · 1200 km · 87.9° · min elevation 25°

Latitude sweep is sampled at 60 s steps across 4 longitudes, so it is coarser than the headline figures.

Numbers

Constellation	Planes	Worst outage	Avg wait	Windows/day	Service min/day	Coverage
3 sats	3 × 1	3.7 h	1.9 h	12.0	88	6.1%
12 sats	4 × 3	2.8 h	43.8 min	27.1	242	16.8%
22 sats	11 × 2	37.0 min	32.2 min	29.5	495	34.4%
48 sats	8 × 6	4.0 min	2.6 min	96.1	1195	83.0%
90 sats	10 × 9	2.7 min	79 s	82.2	1333	92.6%
200 sats	20 × 10	continuous	0	1	1440	100.0%
648 sats	12 × 54	continuous	0	1	1440	100.0%

Same sampling as the headline tiles: 8 longitudes, 20 s steps (40 s above 400 satellites). Every row except your own uses the automatic plane rule.

Model and assumptions

Geometry: spherical Earth (R = 6371 km), circular orbits, Walker-delta constellation with evenly spaced planes and phasing F = 1, no J2 drift or drag. Service means at least one satellite above the minimum elevation angle.
Sampling: headline numbers and the table simulate 48 h (96 h for fleets of 12 or fewer) at 20 s steps, worst-cased across 8 longitudes at your latitude. The latitude chart uses 60 s steps and 4 longitudes.
Planes: "Auto" spreads satellites across the divisor of N nearest above the square root of N, a revisit-friendly default. Real constellations may choose otherwise: a single-plane test block clusters its passes.
Fidelity: planning-grade, for sizing intuition and commercial conversations. Contractual coverage commitments need full-fidelity tooling (STK, GMAT) with real ephemerides, beam patterns and link budgets. This tool models geometry only, not capacity or link margin. Full method and validation anchors.

Density plus altitude

Two levers pulled at once. At 1200 km each satellite sees a wide stretch of ground, and at 648 of them there are enough to keep every stretch occupied. The result is a shell that closes almost everywhere rather than only where the orbit lingers: the worst outage at 70 degrees is continuous service and at the equator continuous service.

The buildout is the hard part

A finished shell is the easy thing to model. What a business plan has to survive is the years before it exists. A quarter of this fleet, 162 satellites in 6 planes, gives 1.7 min at this latitude. Half of it, 324 in 12 planes, gives continuous service. Coverage does not arrive proportionally with hardware, and the last stretch before the shell closes buys the least visible improvement per launch.

What geometry does not tell you

A shell this size is not capacity-limited by visibility, it is capacity-limited by spectrum, beam count and gateway throughput. Everything on this page says a satellite is overhead. None of it says there is a beam pointed at you, that the beam has capacity left, or that the service is licensed where you are standing. For a dense shell those are the binding constraints, and they live outside this model entirely.