Engineer
First year computer engineering. External device data, sourced and keyed, two reference devices drawn at real scale in 3D, and from 18 ก.ย. the one who puts the physiological values the medical side finds onto the simulator and the CAD view. On the sheet under Design 1, now the Design team, and the only person in Engineering.
/quest opens with the gap map (nine stations by seven value groups, built from the heart data) and gives Non three quests with deliverable and acceptance criteria: N1 where each device attaches per the patent, its operating and failure conditions, and the pressure bands drawn on /public/heart; N2 pressure, velocity and Reynolds number at the nine points as SOURCED or checked DERIVED values, keyed through the build; N3 a small simulation run at 4 GB with real memory and time figures, a recommendation on continuum against particle, and whether the university CFD machine exists. Text and photos file into /dropin/quest/.
The gate that holds three cards in review. Open the CAD view with Colour by source on, walk the red parts, compare each against the patent figures and the IFU, and either fix heart_models.py or write down why the drawing is right. Then the 3D page and both PDFs can move to done.
Live at /public/cad with parts tree, measure on click, exploded view, section plane, labels, walkthrough and the flow simulation. Done 15 ก.ย.: every number carries its source, a citation key opens the paper or patent, DERIVED opens the checked calculation, a guess is red on red, and the build refuses a value with none. Still to do: the physiological targets drawn as graphs, now that Trio's rows of 21 ก.ย. are on file (trio/ in the repo).
Model A (HeartMate 3) and Model B (BiVACOR rotary TAH), 1:1 in mm, placed on a real heart failure heart from an open CT dataset. Every part named, exploded views, what every tube connects to, blood path animation and a lumped parameter flow simulation. Assumed values are red on the page. In review until Non checks the geometry.
Trio sent her three quests as documents rather than through the form, so her rows were keyed on Ton's side into trio/TRIO-T1-T3-2026-09-21.md (short TRI in nont_tables.py) and the pages quote that file directly. Nothing to re-key. Non's part: where her rows overlap his (right atrial and pulmonary artery pressures, the aortic diameters), check the two agree and say so in his file; and take the model order about the lungs if she sends one in LINE. Her 17 unsourced rows go back to her, never in as ASSUMED.
HeartMate 3 taken apart into 12 numbered parts, what each tube connects to, and how it works. 7 A3 pages. Internal sizes the manufacturer does not publish are marked ASSUMED.
BiVACOR geometry from its patents and papers: exploded, assembled, cutaway, rotor, in the heart, and a part by part table with what each tube connects to. 9 A3 pages.
Volute 71 mm against a 60 mm envelope, rotor and vane datum undefined, 240 um clearance below the 250 um patent minimum. Not yet traced to source.
Sourced dimensions and operating data for a rotary total artificial heart. Every value keyed to a paper, patent or regulatory document.
10 page summary of the sourced research: BiVACOR data, the human side of both designs, and how each number was checked. Dated 10 September, before the 3D model work.