Skip to main content

Design a synthetic constellation

Difficulty: Beginner. Time: about 15 minutes. Needs: a browser; guest mode works (each assistant request spends a guest query).

You will have the assistant synthesize a Walker constellation, verify the satellites are first-class scene assets rather than chart-only estimates, and run a coverage analysis whose numbers come from the same element sets the globe renders.

1. Ask for the constellation​

Open chat and ask, in plain language:

Create a Walker constellation, 24 satellites in 6 planes at 550 km, 53 degrees inclination.

What happens under the hood is the point of this tutorial: real TLEs are synthesized from your mission intent, planes spaced evenly in RAAN, satellites evenly in mean anomaly per plane, with the Walker inter-plane phase offset applied. The satellites are injected through the exact pipeline uploads use.

Expected outcome: 24 satellites animate on the globe, and the Assets panel shows them under the Synthetic group with a synthetic provenance badge.

2. Verify they are first-class​

Click one of the new satellites:

  • The full detail card opens: orbital elements, period, sub-satellite point, upcoming passes.
  • Press play; the satellite propagates through the same SGP4 worker and interpolation as the operator fleet.
  • Ask the assistant "show me the ground track of one of the synthetic satellites".

The only permanent difference from a real satellite is the label. Every answer that touches these assets will disclose that they are synthetic; they never masquerade as real hardware.

3. Score coverage​

Now ask a design question:

What coverage would this constellation give a ground station at Longyearbyen with a 10 degree elevation mask?

The coverage engine computes from the same element sets the rendered assets use, so the numbers match the picture exactly. Expect the answer to report:

  • Combined coverage: the fraction of the window with at least one satellite visible.
  • Per-satellite coverage and pass intervals.
  • The full set of chosen and derived assumptions: window, step, elevation mask.

Expected outcome: an insight card with the coverage figures, its scope and method attached, and satellites without propagable elements (if any) counted as excluded rather than silently skipped.

4. Iterate like a trade study​

Ask a variant and compare:

Same constellation but at 8 planes of 3. How does Longyearbyen coverage change?

Because the engines are deterministic, both runs score on identical windows and masks, and the delta is an honest subtraction. If a comparison is worth keeping, capture it into a report (Build reports) or save the scenario (Run simulations).

5. Clean up, or keep it​

Synthetic assets are yours: rename, hide, or delete them from the Assets panel. Scenario resets restore the unmodified scene; keeping the constellation is equally valid if you want it around for follow-up sessions.

Troubleshooting
  • "I ran out of assistant queries": guest mode includes 10; sign in for unlimited queries on a paid plan. See Plans and billing.
  • Coverage numbers look too low: check the elevation mask in the answer's assumption list; 10 degrees at high latitude with a low-inclination constellation can honestly be near zero.
  • Satellites did not appear: check the Synthetic group's visibility toggle in the Assets panel.
  • The answer says an asset was excluded: that asset had no propagable orbital elements; the engine excludes and counts it rather than fabricating visibility.

Where next​