# Connected research / V3 scientific and interaction review

## Sources checked

1. Gil-Esmendia, Mansourifilestan, Flores & Brouwer. *Feasibility of Zero-Emission Cruise Ships: A Novel Hydrogen Tri-Generation System for Propulsion and Hotel Loads.* JMSE 14, 431 (2026). DOI: https://doi.org/10.3390/jmse14050431. Supplied full article `jmse-14-00431.pdf`; Figure 1 and §§2.2, 2.5–2.8, Table 5 and the seven-day scenario were inspected. The original Figure 1 remains accessible in the workbench.
2. Flores et al. *Experimental performance of a first-of-a-kind fuel cell hybrid fast harbor craft.* IJHE (2026), 156680. DOI: https://doi.org/10.1016/j.ijhydene.2026.156680. Supplied author proof `HE156680_proof-redline_v8_acrobat_comments.pdf`, Figure 2 and §§2.1–2.4. The original schematic remains accessible. “Miami Vice Paper” is the owner's display name; the studied vessel is FCV Vanguard.
3. US DOE, PEM fuel-cell parts and operation: https://www.energy.gov/cmei/fuels/parts-fuel-cell and https://www.energy.gov/cmei/fuels/fuel-cell-animation-text-version .
4. US NRC, neutron absorbers: https://www.nrc.gov/reading-rm/basic-ref/glossary/poison-neutron ; DOE nuclear history/fission: https://www.energy.gov/ne/history .
5. Owner's California green-ammonia manuscript `Pasted text(2).txt`, §4.1 / Figure 8, for candidate-location names only. Profile values on this website are newly generated synthetic placeholders.
6. Owner's academic CV and earlier portfolio records for ASUCI Engineering Senator activity. R58-120: https://asuci.uci.edu/senate/legislation/print/?legislation_number=R58-120 ; R58-22: https://internal.studentgov.uci.edu/wp-content/uploads/formidable/35/legislation-r58-22.pdf . Press coverage is reproduced in Exhibit D of https://uci.edu/hydrogen/uci-h2-project-report.pdf (Brooke Staggs / Southern California News Group, March 2023). Grant and hackathon fundraising totals are the owner's account, not independently audited amounts.

## Cruise thermal topology: what was corrected

Figure 1 uses a simplified combined exhaust-gas line through the EHX and the block named “H₂ boiler.” The prose is more specific: §2.5 identifies **moist cathode exhaust** as the EHX heat source; §2.7 identifies **residual anode hydrogen** as catalytic-burner fuel. The new interactive drawing separates those streams instead of suggesting that the cathode-air stream itself is hydrogen fuel.

EHX → space-heating duty; coolant HX → domestic-hot-water duty; catalytic burner → hot thermal-fluid duty. Their useful heat is aggregated only for the secondary thermal-energy balance. The drawn heat links are **energy accounting**, not a claim that air, stack coolant and domestic water share a physical pipe. The TESS is described in §2.7 but not drawn as a tank in the original Figure 1. The added TESS/thermal-load connections are the reduced-order companion to that text.

The burner is labeled **Catalytic burner**, with 4 MW useful-heat limit and 95% HHV efficiency. `voyage-model.js` uses H₂ HHV = 141.8 MJ/kg and LHV = 120 MJ/kg. Electrical stack efficiency (default 55%) multiplies reacted-H₂ LHV. The overall chemical-energy ledger uses HHV, including unrecovered latent heat in losses. This avoids applying the paper's HHV burner efficiency inside a mismatched LHV conservation claim.

The 35% exhaust / 65% coolant share of recovered fuel-cell heat is an **assumed split**, not an experimentally measured division. Recovery = 65% of the LHV stack rejection by default. DC conversion 97%, motor 96%, ±0.2 MW electric buffer power and ideal, lossless 0.3 MWh electric/12.5 MWh thermal storage are retained reduced-model assumptions. The thermal store is 45 GJ, with a 24 MW exchange cap. The paper has inconsistent battery/capacity descriptions; the website uses the declared 300 kWh scenario and does not claim to reconcile its component-volume table.

The catalytic burner recovers available residual fuel, subject to its rating. Surplus useful heat charges the store; insufficient useful heat discharges it; excess after storage is full is rejected. The sign convention is positive storage power OUT. The diagram's base path points INTO the store, so a discharge reverses its particles. The same applies to the battery branch. With unmet duty, powers shown are a preview and animated delivered flows stop.

The LH₂ vaporizer supplies a **cold-recovery duty**, not the ship's complete A/C system. Qcold = ṁ × 3.07 MJ/kg × 0.7 (§2.8); at 0.7 kg/s it is about 1.5 MW. Cold duty draws on cryogenic enthalpy and is kept separate from the chemical HHV ledger. No additional absorption chiller is invented.

## One cruise state, not two unrelated demos

The voyage and powertrain read one object: elapsed time, H₂ mass, consumed mass, distance, electrical charge and thermal charge. Starting hydrogen = 312 tonnes, an explicit scenario (approximately 240 tonnes × 1.30 reserve), not a tank mass measured on an operating cruise ship. An editable 1–1000 tonne initial fill applies only when “Refill & restart” is pressed. No boil-off or tank-pressure solver is included.

The illustrative replay starts at the assumed first departure, day 1 at 18:00. Leg distances, durations and speeds follow the source itinerary; departure scheduling and sinusoidal hotel/heat/cold loads are assumptions. The approximate power–speed curve is retained from the earlier digitization. Editing a demand exits itinerary-following mode; speed then means the equivalent speed on that assumed curve. Displayed distance is integrated from **delivered** shaft-power equivalent, never simply awarded by the timeline.

Each step checks demand feasibility before committing time or inventories. A blocked step consumes no fuel and does not advance distance or storage. This requested all-or-nothing teaching gate is not the ship's actual protective controller. Default playback is one model second per wall-clock second; accelerated modes are marked. Off-screen or hidden-tab time is not caught up. At 3600×, integration still uses short bounded substeps and stops at the first infeasible step. A changing voyage demand can therefore stop the replay until supply is rebalanced.

## Vanguard verification

The redrawn Figure 2 topology explicitly includes two fuel-cell modules with DC/DC converters feeding a common HV DC bus, bidirectional 23 kWh battery, inverter/motor, HV loads and a DC/12 V branch. A cooling-system link is retained; its flow rates and temperatures are not calculated. BoP is a combined electrical duty, not a second independent energy source.

The original pure Vanguard model in `powertrain-models.js` is unchanged: finite 9.7 kg H₂ inventory, 23 kWh battery, 10–90% operating SOC (assumed), 250 kW shaft-demand limit. The source text reports 85 kW automotive module ratings and marine cooling enabling up to 100 kW per module; its specification table says 90 kW. The website retains independently editable 0 or 30–100 kW commands as a manual operating-envelope illustration, **not** the actual staged controller. Conversion, auxiliary and battery-power-limit assumptions remain in `POWERTRAIN_METHODS.md`.

The introductory achievements (217 NM, nine-month demonstration, average 35% HHV / 41% LHV) describe the paper's results. They are not outcomes newly calculated or validated by this browser model.

## Fuel-cell → ATLAS → operating cell

The seven-layer cut names anode flow plate, anode GDL, anode catalyst, proton-exchange membrane, cathode catalyst, cathode GDL and cathode flow plate. Plates are thick and channeled, GDLs porous, catalyst coatings very thin, membrane distinct. Visual widths are deliberately exaggerated and are not dimensional CAD. Microporous layers are bundled into GDLs; perimeter gaskets are described but not separately modeled.

H₂ enters the anode. H⁺ crosses the membrane; electrons follow the external circuit. O₂ enters the cathode. Product water exits the cathode in the simplified view; actual cells have internal water transport and can discharge vapor and/or liquid depending on conditions. O₂ and H₂ symbols differ in size for readability; they are not calibrated molecular-radius measurements.

ATLAS deposits Pt/C-containing ink on a diffusion substrate, creating a catalyst-coated electrode (GDE). It does not spray fission products. Typical owner-confirmed target: 0.30 mg Pt/cm² over one 5×5 cm electrode = 7.5 mg deposited Pt; 0.60 mg Pt/cm² = 15 mg over that area is another reported example, not an upper limit. The comparison curves, interpolated values and their source attribution are unchanged. The operating PEM-cell engine remains illustrative, not a calibration to those ATLAS data.

## Reactor labels and moving rods

The six reservoirs depict delayed-neutron **precursor populations**, not water. Labels G1–G6 and half-lives identify the groups; dashed equilibrium reference and the release pathway are explained. The period is the e-fold time: positive for growth, negative for decay, infinity for steady power. The power column is relative and logarithmic (0.01–100, four decades); each decade is a factor of ten. The ribbon shows the recent two model minutes.

The control-rod illustration reads actual `R.rho`, requested `R.cmd`, and the existing trip state. Withdrawal corresponds to increasing reactivity; insertion to more neutron absorption. Ghost outlines show the request, solid rods the rate-limited state (40 pcm/s in this model). The percent-insertion mapping is a **conceptual linear drawing scale**, not calibrated rod worth, geometry, a real TRIGA controller, or instructions for operating a reactor. The engine and its delayed-neutron kinetics were not replaced.

The previously disclosed 250 kW thermal reference and hypothetical home-equivalent remain comparisons, not real telemetry. UCI's research reactor does not supply household electricity. The owner is described as an operator **in training**, not as licensed.

## Renewable profiles replace the economics UI

The active page now only compares MOJ-PV, SAC-PV, TEH-LW, ALT-LW, MEN-OW, SC-MIX and NC-MIX. Site membership matches the supplied study; all 8,760 hourly values are deterministic synthetic examples. Solar is zero outside the constructed daylight window; wind fluctuates within [0,1]. Outputs normalize to 1 MW total installed capacity. Mixed cases are 50/50 installed capacity, not the manuscript's energy-matched/optimized sizing. Displayed means are synthetic, not reported study capacity factors. No transmission losses, meteorological measurements or cost forecast is implied. Old lifecycle source is retained but not loaded or presented. `ResourceProfiles.get(id)` is the clean replacement point for actual data later.

## Home film and contact

The film is a visual research narrative: U-235 fission → match-cut to ATLAS material deposition → PEM-cell assembly/reaction → cruise ship → Vanguard overtaking → solar electrolysis and N₂/H₂ ammonia synthesis → water boiling. One balanced fission channel, 235U + n → 141Ba + 92Kr + 3n, is an example, not the only outcome. Fission fragments are not represented as hydrogen or platinum. The fission-to-catalyst connection is explicitly a cinematic cut.

Fuel-cell stoichiometry is 2H₂ + O₂ → 2H₂O, with separate electron/proton paths. Green ammonia explicitly receives hydrogen from electrolysis and nitrogen from air separation; synthesis is not portrayed as electrolysis itself. The flood transition enters a water-boiling illustration, not an ammonia release or an actual reactor/fuel-cell plant. Vessel scale, particle scale and time are composed for storytelling. Animation can be paused, respects reduced motion, and suspends off-screen. Contact is part of Home, without the retired network animation or a separate navigation tab.

## Verification boundary

Automated browser, conservation and input-bound tests check consistency and behavior. They do not independently validate component efficiencies, operating safety, CFD, reactor calibration, synthetic resource representativeness, or the papers' complete solvers. Phone tests are touch-enabled Chromium/WebKit viewports, not physical-device certification.

Manual rod input now exits the independent one-neutron cold-start visualization before setting a rate-limited reactivity command. The cold-start picture freezes the point-kinetics integrator by design; leaving it frozen after a user command was an interaction defect. The integrator equations, rod drive limit and trip logic are unchanged.


## Home-film audit correction (8 September 2026)
Section 2.4.4 explicitly calls 76.1 MW the net installed electrical output after balance-of-plant losses. The active cruise model now caps **net** output at 76.1 MW; its assumed 97% conversion block implies a gross cap of 76.1 / 0.97 MW. The prior code capped gross at 76.1 MW and deducted losses twice relative to this stated net rating. Other efficiency and heat-split assumptions remain uncalibrated, user-adjustable browser parameters. See HOME_SCIENCE.md for independent benchmark scope.