Methods and limits

Read these before relying on anything

Relevance is not safety.

The score orders investigations by how closely they relate to a scenario. It says nothing about whether a material or a cabin is safe, and it is not a probability.

Missing is not zero.

An empty value means PSI's table does not report it. It does not mean the quantity was zero or that it did not matter.

Close on the map is not the same test.

Two tests at the same oxygen and pressure can differ in airflow, material, shape, sample size and how they were ignited. The map shows two conditions out of many.

A gap is relative to this archive.

No tests near a scenario means none in the PSI experimental tables read here. Drop-tower and ground tests, and results published only in papers, are outside it.

A matching quote is not a review.

Every curated statement carries a quotation that the build checks, character for character, against NASA's text. That proves where a statement comes from. It does not prove the statement is right, and no combustion scientist has reviewed this site.

All flight tests here ran in microgravity.

PSI holds no test at lunar or Martian gravity. Microgravity and normal gravity can only bracket what happens in between.

Some pressures are assumed.

Several tables do not list pressure. Where an experiment ran in the station cabin or in the Cygnus vehicle, cabin pressure is used and the value is marked as assumed and drawn hollow.

Where the data comes from

All data is read from NASA's Physical Sciences Informatics repository, which NASA releases under CC0. The 24 investigations in the Combustion Science research area hold 13.5 TB of raw files. This site does not read the raw video or sensor files. It reads the experimental table that each team published: 25 tables, 2,143 rows, 28,316 values.

NASA names PSI as the data source for this challenge. For background it points to its Researcher's Guide to International Space Station Combustion Science and its Combustion Science pages. This site does not read either of them yet.

One more NASA file is read: the ReadMe that PSI holds for BASS-II. It says what the table's Air display column is (“flow speed measured in the duct in cm/s”), and that sentence is the stated source of every BASS-II air speed on this site.

1. 1

   Raw

   Exact copies of what PSI served: the index of investigations, each investigation record, and every experimental table. Each file is stored with its fetch time and SHA-256 checksum and is never edited.

2. 2

   Staged

   The raw files parsed into tidy tables. PSI's own column names and values are kept as they are.

3. 3

   Curated

   Every column of every table mapped onto one dictionary of variables, with units converted. Each value keeps the raw cell it came from and a quality label.

4. 4

   Serving

   The curated files loaded into a Postgres database, which is what this site reads.

Where each value comes from

Every one of the 28,316 values carries one of five labels. Click any row in the data to see its label, the raw PSI cell, and the note or quotation behind it.

What was repaired, left out or flagged

The tables were typed by hand over two decades and contain slips. Nothing is changed silently. This is the complete log.

• BREColumn 'Oxygen Conc (percent)': '3400%' read as '34' in 1 row(s). A percent format was applied to the number 34 in the PSI table; the column is already in percent.
• BREColumn 'Oxygen Conc (percent)': '2100%' read as '21' in 1 row(s). A percent format was applied to the number 21 in the PSI table; the column is already in percent.
• CLD FlameColumn 'Fuel Bottle': 26 cell(s) were damaged by spreadsheet drag-fill (for example 40% CH5, 40% CH6, 40% CH7, 40% CH8, 40% CH9) and were restored to the first value of each run.
• E-FIELD Flames11 continuation row(s) were folded into the test day they belong to (they list further values of Fuel Conc).
• s-FlamePSI's server cut this record off before the end. Every field up to the cut is used; later fields were filled from PSI's search index.
• s-FlameColumn 'Atmosphere*': 6 cell(s) were damaged by spreadsheet drag-fill (for example ~21% O2 with N3, ~21% O2 with N4, ~21% O2 with N5, ~21% O2 with N6) and were restored to the first value of each run.
• s-FlameColumn 'Fuel Bottle': 2 cell(s) were damaged by spreadsheet drag-fill (for example 50% C2H4 with N3, 45% CH4 with N3) and were restored to the first value of each run.
• BASS-IIOxygen at end of test: 3 value(s) fall outside the expected range 0.05 to 1 fraction (1.6, 0.6, 0.9). They are kept exactly as published, flagged, and left out of charts.
• BASS-IIOxygen at end of test: 2 value(s) are more than 2 points above the oxygen at the start of the same test. Burning uses oxygen up, so this cannot be a reading of the same air. They are kept exactly as published, flagged, and left out of charts.
• BASS-IIM17, How the test ended: NASA's file PSI-25_Reports_BASS-II Summary Report.pdf gives "flow terminated"; NASA's file PSI-25_Experimental table_BASS-II.csv gives "extinction" (row 43, column Air display). The first is the one used.
• BASS-IIM17, How the test ended: NASA's file PSI-25_Reports_BASS-II Summary Report.pdf gives "flow terminated"; NASA's file PSI-25_Science Documents_BASS-II_Test.Matrices.xlsx gives "extinction" (Miller!H18). The first is the one used.
• BASS-IIP5, How the test ended: NASA's file PSI-25_Reports_BASS-II Summary Report.pdf gives "flow terminated"; NASA's file PSI-25_Experimental table_BASS-II.csv gives "extinction" (row 79, column Air display). The first is the one used.
• BASS-IIP5, How the test ended: NASA's file PSI-25_Reports_BASS-II Summary Report.pdf gives "flow terminated"; NASA's file PSI-25_Science Documents_BASS-II_Test.Matrices.xlsx gives "extinction" (Fernandez-Pello!H7). The first is the one used.
• BASS-IIP13a, How the test ended: NASA's file PSI-25_Reports_BASS-II Summary Report.pdf gives "flow terminated"; NASA's file PSI-25_Experimental table_BASS-II.csv gives "extinction" (row 85, column Air display). The first is the one used.
• BASS-IIP13a, How the test ended: NASA's file PSI-25_Reports_BASS-II Summary Report.pdf gives "flow terminated"; NASA's file PSI-25_Science Documents_BASS-II_Test.Matrices.xlsx gives "extinction" (Fernandez-Pello!H13). The first is the one used.
• BASS1 row(s) are not tests and were left out: A cleaning step
• DAFTDAFT-3 Sample Bag #1 (GMT 233, 2006), Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 17236.3 particles/cm³; NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 17314.3 particles/cm³ (Voice Down Data Summary!B21). The first is the one used.
• DAFTDAFT-3 Sample Bag #1 (GMT 233, 2006), Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 17236.3 particles/cm³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_0611zgy time hist.xls gives 17314.3 particles/cm³ (Data!C23). The first is the one used.
• DAFTDAFT-3 Sample Bag #1 (GMT 233, 2006), Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 17236.3 particles/cm³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_06 urban.xls gives 17314.3 particles/cm³ (Data!B21). The first is the one used.
• DAFTDAFT-3 Sample Bag #2 (GMT 233, 2006), Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 16589.4 particles/cm³; NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 17256.2 particles/cm³ (Voice Down Data Summary!E21). The first is the one used.
• DAFTDAFT-3 Sample Bag #2 (GMT 233, 2006), Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 16589.4 particles/cm³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_0611zgy time hist.xls gives 17256.2 particles/cm³ (Data!G23). The first is the one used.
• DAFTDAFT-3 Sample Bag #2 (GMT 233, 2006), Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 16589.4 particles/cm³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_06 urban.xls gives 17256.2 particles/cm³ (Data!E21). The first is the one used.
• DAFTDAFT-3 Sample Bag #3 (GMT 233, 2006), Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 26578 particles/cm³; NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 24350 particles/cm³ (Voice Down Data Summary!H21). The first is the one used.
• DAFTDAFT-3 Sample Bag #3 (GMT 233, 2006), Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 26578 particles/cm³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_0611zgy time hist.xls gives 24350 particles/cm³ (Data!K23). The first is the one used.
• DAFTDAFT-3 Sample Bag #3 (GMT 233, 2006), Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 26578 particles/cm³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_06 urban.xls gives 24522.2 particles/cm³ (Data!H21). The first is the one used.
• DAFTDAFT-3 Sample Bag #1 (GMT 233, 2006), Particle mass concentration (DustTrak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 45.6343 mg/m³; NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 43.15 mg/m³ (Voice Down Data Summary!C21). The first is the one used.
• DAFTDAFT-3 Sample Bag #1 (GMT 233, 2006), Particle mass concentration (DustTrak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 45.6343 mg/m³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_0611zgy time hist.xls gives 43.15 mg/m³ (Data!D23). The first is the one used.
• DAFTDAFT-3 Sample Bag #1 (GMT 233, 2006), Particle mass concentration (DustTrak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 45.6343 mg/m³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_06 urban.xls gives 43.15 mg/m³ (Data!C21). The first is the one used.
• DAFTDAFT-3 Sample Bag #2 (GMT 233, 2006), Particle mass concentration (DustTrak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 33.8377 mg/m³; NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 34.0875 mg/m³ (Voice Down Data Summary!F21). The first is the one used.
• DAFTDAFT-3 Sample Bag #2 (GMT 233, 2006), Particle mass concentration (DustTrak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 33.8377 mg/m³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_0611zgy time hist.xls gives 34.0875 mg/m³ (Data!H23). The first is the one used.
• DAFTDAFT-3 Sample Bag #2 (GMT 233, 2006), Particle mass concentration (DustTrak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 33.8377 mg/m³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_06 urban.xls gives 34.0875 mg/m³ (Data!F21). The first is the one used.
• DAFTDAFT-3 Sample Bag #3 (GMT 233, 2006), Particle mass concentration (DustTrak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 62.6509 mg/m³; NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 63.1833 mg/m³ (Voice Down Data Summary!I21). The first is the one used.
• DAFTDAFT-3 Sample Bag #3 (GMT 233, 2006), Particle mass concentration (DustTrak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 62.6509 mg/m³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_0611zgy time hist.xls gives 63.1833 mg/m³ (Data!L23). The first is the one used.
• DAFTDAFT-3 Sample Bag #3 (GMT 233, 2006), Particle mass concentration (DustTrak): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 62.6509 mg/m³; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_06 urban.xls gives 63.2889 mg/m³ (Data!I21). The first is the one used.
• DAFTDAFT-3 Sample Bag #1 (GMT 233, 2006), Particle diameter from the mass per particle (assumed density): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.71639 µm; NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.68213 µm (Voice Down Data Summary!B25). The first is the one used.
• DAFTDAFT-3 Sample Bag #1 (GMT 233, 2006), Particle diameter from the mass per particle (assumed density): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.71639 µm; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_0611zgy time hist.xls gives 1.68213 µm (Data!C27). The first is the one used.
• DAFTDAFT-3 Sample Bag #1 (GMT 233, 2006), Particle diameter from the mass per particle (assumed density): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.71639 µm; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_06 urban.xls gives 1.68213 µm (Data!B25). The first is the one used.
• DAFTDAFT-3 Sample Bag #2 (GMT 233, 2006), Particle diameter from the mass per particle (assumed density): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.57347 µm; NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.55674 µm (Voice Down Data Summary!E25). The first is the one used.
• DAFTDAFT-3 Sample Bag #2 (GMT 233, 2006), Particle diameter from the mass per particle (assumed density): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.57347 µm; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_0611zgy time hist.xls gives 1.55674 µm (Data!G27). The first is the one used.
• DAFTDAFT-3 Sample Bag #2 (GMT 233, 2006), Particle diameter from the mass per particle (assumed density): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.57347 µm; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_06 urban.xls gives 1.55674 µm (Data!E25). The first is the one used.
• DAFTDAFT-3 Sample Bag #3 (GMT 233, 2006), Particle diameter from the mass per particle (assumed density): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.65121 µm; NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.70491 µm (Voice Down Data Summary!H25). The first is the one used.
• DAFTDAFT-3 Sample Bag #3 (GMT 233, 2006), Particle diameter from the mass per particle (assumed density): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.65121 µm; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_0611zgy time hist.xls gives 1.70491 µm (Data!K27). The first is the one used.
• DAFTDAFT-3 Sample Bag #3 (GMT 233, 2006), Particle diameter from the mass per particle (assumed density): NASA's file PSI-47_Analyzed Data_DAFT Retrieved Data Summary.xls gives 1.65121 µm; NASA's file PSI-47_Analyzed Data_ISS_Daft_Aug21_06 urban.xls gives 1.70186 µm (Data!H25). The first is the one used.
• DAFTSession D1, Particle number concentration (P-Trak): NASA's file PSI-47_Analyzed Data_DAFT Flight Data GMT053.xls gives 1.43189 particles/cm³; NASA's file PSI-47_Reports_DAFT_FinalReport.doc gives 2 to 5 particles/cm³ (section II.D.1 (DAFT-1 & DAFT -2, Session 1, GMT 54, 2005)). The first is the one used.
• FLEXSubscripts were lost in the PSI file: the columns "O", "N" and "CO initial ambient composition" are O2, N2 and CO2 (carbon dioxide is one of the suppressants FLEX tested).
• FLEXPressure: 1 value(s) fall outside the expected range 10 to 700 kPa (0). They are kept exactly as published, flagged, and left out of charts.
• Saffire-I2 row(s) are not tests and were left out: Pre-test flow visualization video
• Saffire-ISample S1, Sample width: the experimental table gives 94 cm; NASA's file NTRS-20170008805.pdf gives 40.6 cm (p. 29, Table I, row 1-1, column Width). The table's value is the one used.
• Saffire-ISample S1, Sample width: the experimental table gives 94 cm; NASA's file NTRS-20180005168.pdf gives 40.6 cm (p. 30, Table 1, row I-1, column Width). The table's value is the one used.
• Saffire-ISample S1, Sample thickness: the experimental table gives 3.7 mm; NASA's file NTRS-20170008805.pdf gives 0.37 mm (p. 29, Table I, row 1-1, column Thickness). The table's value is the one used.
• Saffire-ISample S1, Sample thickness: the experimental table gives 3.7 mm; NASA's file NTRS-20180005168.pdf gives 0.37 mm (p. 30, Table 1, row I-1, column Thickness). The table's value is the one used.
• Saffire-ISample S1, Sample length: the experimental table gives 40.6 cm; NASA's file NTRS-20170008805.pdf gives 94 cm (p. 29, Table I, row 1-1, column Length). The table's value is the one used.
• Saffire-ISample S1, Sample length: the experimental table gives 40.6 cm; NASA's file NTRS-20180005168.pdf gives 94 cm (p. 30, Table 1, row I-1, column Length). The table's value is the one used.
• Saffire-ISample S2, Sample width: the experimental table gives 94 cm; NASA's file NTRS-20170008805.pdf gives 40.6 cm (p. 29, Table I, row 1-2, column Width). The table's value is the one used.
• Saffire-ISample S2, Sample width: the experimental table gives 94 cm; NASA's file NTRS-20180005168.pdf gives 40.6 cm (p. 30, Table 1, row I-2, column Width). The table's value is the one used.
• Saffire-ISample S2, Sample thickness: the experimental table gives 3.7 mm; NASA's file NTRS-20170008805.pdf gives 0.37 mm (p. 29, Table I, row 1-2, column Thickness). The table's value is the one used.
• Saffire-ISample S2, Sample thickness: the experimental table gives 3.7 mm; NASA's file NTRS-20180005168.pdf gives 0.37 mm (p. 30, Table 1, row I-2, column Thickness). The table's value is the one used.
• Saffire-ISample S2, Sample length: the experimental table gives 40.6 cm; NASA's file NTRS-20170008805.pdf gives 10 cm (p. 29, Table I, row 1-2, column Length). The table's value is the one used.
• Saffire-ISample S2, Sample length: the experimental table gives 40.6 cm; NASA's file NTRS-20180005168.pdf gives 10 cm (p. 30, Table 1, row I-2, column Length). The table's value is the one used.
• Saffire-II2 row(s) are not tests and were left out: Pre-test flow visualization video
• Saffire-IIColumn 'Material': 2 blank cell(s) filled from the row above.
• Saffire-IIThe micro sign was lost in the PSI file: the columns "?-g Burn Length" and "?-g Spread Length" are the microgravity results.
• Saffire-IIThe column "?-g Spread Length" holds spread rates in mm/s, so it is mapped to flame spread rate.
• Saffire-IISample 2-1, Oxygen: the experimental table gives 0.215 fraction; NASA's file NTRS-20170008805.pdf gives 0.221 fraction (p. 29, Table I, row 2-1, column Δ %O2). The table's value is the one used.
• Saffire-IISample 2-1, Oxygen: the experimental table gives 0.215 fraction; NASA's file NTRS-20180005168.pdf gives 0.221 fraction (p. 30, Table 1, row II-1, column Δ %O2). The table's value is the one used.
• Saffire-IISample 2-6, Flow speed: the experimental table gives 20 cm/s; NASA's file NTRS-20170008805.pdf gives 25 cm/s (p. 29, Table I, row 2-6, column Flow). The table's value is the one used.
• Saffire-IISample 2-6, Flow speed: the experimental table gives 20 cm/s; NASA's file NTRS-20180005168.pdf gives 25 cm/s (p. 30, Table 1, row II-6, column Flow). The table's value is the one used.
• Saffire-IIColumn 'Samle Thickness': '037 mm' read as '0.37 mm' in 3 row(s). Decimal point missing in the PSI table (the other samples are 0.27 to 1.03 mm).
• Saffire-IIISample S1, Sample thickness: the experimental table gives 3.3 mm; NASA's file NTRS-20180005168.pdf gives 0.37 mm (p. 30, Table 1, row III-1, column Thickness). The table's value is the one used.
• Saffire-IIISample S1, Flow speed: the experimental table gives 30 cm/s; NASA's file NTRS-20180005168.pdf gives 25 cm/s (p. 30, Table 1, row III-1, column Flow). The table's value is the one used.
• Saffire-IIISample S2, Sample thickness: the experimental table gives 3.3 mm; NASA's file NTRS-20180005168.pdf gives 0.37 mm (p. 30, Table 1, row III-2, column Thickness). The table's value is the one used.
• Saffire-IIISample S2, Sample length: the experimental table gives 94 cm; NASA's file NTRS-20180005168.pdf gives 10 cm (p. 30, Table 1, row III-2, column Length). The table's value is the one used.
• Saffire-IIISample S2, Flow speed: the experimental table gives 30 cm/s; NASA's file NTRS-20180005168.pdf gives 25 cm/s (p. 30, Table 1, row III-2, column Flow). The table's value is the one used.
• SAME4 row(s) say 'same conditions as' other rows; they are kept as envelopes listing those conditions.
• SPICEFlow speed: 8 value(s) fall outside the expected range 0 to 200 cm/s (-0.28, -5.92). They are kept exactly as published, flagged, and left out of charts.
• SPICEDay 45 17:27:48, Flow speed: the experimental table gives 17.59 cm/s; NASA's file PSI-107_Raw Data_SmokePointReview-0g-3.xlsx gives 17.587 cm/s (Smoke Point Datasheet (2)!AP10). The table's value is the one used.
• SPICEDay 173 12:39:29, Flow speed: the experimental table gives 38.27 cm/s; NASA's file PSI-107_Raw Data_SmokePointReview-0g-3.xlsx gives 38.2718 cm/s (Smoke Point Datasheet (2)!AP30). The table's value is the one used.
• SPICEDay 174 11:04:43, Flow speed: the experimental table gives 29.81 cm/s; NASA's file PSI-107_Raw Data_SmokePointReview-0g-3.xlsx gives 29.8098 cm/s (Smoke Point Datasheet (2)!AP44). The table's value is the one used.
• SPICEDay 174 12:46:17, Flow speed: the experimental table gives 31.69 cm/s; NASA's file PSI-107_Raw Data_SmokePointReview-0g-3.xlsx gives 31.6902 cm/s (Smoke Point Datasheet (2)!AP54). The table's value is the one used.

Where SI tools were used

SI is NASA's term in this challenge for super intelligence: language models, image and video generators and coding tools.

Building the site

The code, the data pipeline and this text were written with Claude (Anthropic), working with the team.

Reading the investigations

The one-line summaries, the fire-safety level, the prevention, detection and suppression ratings and the short insights were written by the SI after reading each PSI record. Each carries a quotation, and the build stops if a quotation is not found in NASA's text.

Answering questions

In the SI panel beside each feature, the SI answers by running fixed lookups over this database, lists them, and its figures are checked against what they returned. The questions the site suggests use wording prepared ahead of time, whether the live model is on or off; their lookups and the number check still run on every request. Any other question goes to a language model hosted on Amazon Bedrock (gpt-oss-120b, an open model from OpenAI), and an answer it has already given to the same question about the same cabin is shown again for a few hours. The Listen button reads an answer aloud with ElevenLabs where that is set up, and otherwise with the voice built into the browser; the words are the ones on the screen.

Images

Every photograph and every piece of footage is NASA's. The line drawings of cabins, scenes and the three fire-safety jobs, the short animations of them, and the two white models that can be turned (a lunar habitat and a droplet chamber) are concept illustrations made with image and video models through Higgsfield (FLUX.2, Seedance 2.5, Nano Banana Pro, GPT Image 2.6). They show an idea and are not NASA designs or hardware. Each one carries a visible SI-generated mark, in the picture and beside it, and none contains a NASA logo, flag or mission badge. The candle flame on the home page that changes with gravity, and the droplet flame in the Flame Lab, are drawings made in code; the Flame Lab's timing and outcomes come from NASA's tests. None of these pictures is evidence.

The home page

Earth, Mars, the Moon and the space station are NASA's own 3D models (Earth and Mars: NASA/JPL-Caltech, from NASA Science 3D resources; the Moon: NASA's Scientific Visualization Studio, from Lunar Reconnaissance Orbiter data; the station: NASA's ISS model). They are drawn in the browser with three.js. The flame between them is a drawing made in code, and its shape at the Moon's and Mars's gravity is a guess: no test exists there. NASA's downlink recordings in the Flame Lab are played with mpegts.js.

Not used for

The numbers. No value on the map, in the tables or in the ranking was produced or estimated by a model. They come from PSI's tables through code that is in the repository.

Source files

The exact PSI files this site was built from, with their checksums. Database built 2026-11-14.