Runnax facts
- What it is
- An AI photometric engine for outdoor lighting. For a parking lot it searches pole layouts against your illuminance targets and returns one that passes, or says why none does.
- Company
- Runnax, founded by Kyan Gao. First customer: Auvolar, a commercial LED manufacturer in California. Contact: kyan@runnax.com.
- Stage
- Early access. Runs are done on request and returned within one business day. A REST API, an MCP server and a CLI ship together with the API release; they are not public yet.
- Calculation method
- Point method, the same formula as IES RP-8: E = I(γ, C) · cos θ / d² · LLF, summed over every luminaire head, with intensities interpolated from the IES file. Direct light only: no inter-reflection, obstructions or shadowing.
Parking lots · available
- Inputs: lot boundary, drive aisle, the areas where poles may stand (islands, curbs), a pole-height range, one or more IES files, and your targets for minimum illuminance and Max:Min uniformity.
- Search: regular arrays of up to 16 poles in up to 3 rows. Any layout with a pole outside the allowed areas is rejected.
- Checks: horizontal illuminance on a 10 ft grid, plus vertical illuminance at 5 ft on both sides of the drive aisle.
- Selection: the fewest poles that pass; with several IES files, the luminaire that passes with the fewest poles.
- Outputs: pole table, horizontal and vertical point grids, isolux sheet, and a summary with every input and search counter.
Roadways · calculation
- Validated: the single-side arrangement, area averages against AGi32 and DIALux on real projects.
- Calculates but not yet validated: opposite and median arrangements. Check those numbers in AGi32 or DIALux.
- Not supported: staggered arrangements, three or more roadway lanes in one cross-section, luminance-based criteria.
- Point values and minimums are not validated, so the engine issues no roadway pass/fail.
- BUG rating is computed from the IES file.
Accuracy
| Comparison | Result |
|---|---|
| Parking lot, 271 points, vs an AGi32 direct-light run of the same lot | Min, average and max within 0.5 % |
| Same lot, vs the AGi32 design delivered to the customer | Engine average 1.7 % higher |
| Roadway area averages, 9 zones from real projects, vs AGi32 and DIALux | Within 3.6 % |
| Roadway point values vs AGi32 | Not validated (median about 4 %, worst points over 30 %) |
Method and every sample: Verification.
What a run is not
- Not an RP-8 or code-compliance statement. The engine compares results with the targets you give it; it does not look targets up in a standard.
- Not a construction approval. The engineer of record signs the design.
- Not field-verified. No comparison with measured light levels yet; every result is a simulation.
Your files
Files are used for your run, deleted after 30 days, and never used for training.