Every number on a Frontier.Design results page traces back to one of the submodels below. Coefficients have inline citations; the underlying code lives in packages/neemo (Apache-2.0) and is reproducible from a fresh checkout. The Verification & Validation PDF generated for every scenario is the proof-of-math counterpart to this page.
The first five rows (Aquaculture → IAEM) are the published NEEMO framework. The remaining rows are extensions shipped by Frontier.Design on top of the open-core package.
Predicts feed requirement, fish growth, and N/P waste from first principles. Daily mass balance across species-specific energy-partition equations. Tilapia path uses Chowdhury 2013 / Lugert 2014 (corrected from the salmonid carry-over in the manuscript workbook — see corrections below).
Plant production, resource demand, stage-weighted nutrient targets. Yield = annual_crop × processing_efficiency. Per-crop dry-matter fractions (leafy 0.05 / herb 0.10 / fruiting 0.06 / roots 0.12 / microgreen 0.08) drive the N mass balance — replaces the bug where the published model implied fish_lwg > plant_kg for leafy systems.
Couples fish and plants via nitrogen mass balance (Lobanov et al. 2025). Coupled mode: fish biomass is sized to plant N demand. Decoupled mode: both are independent and surplus N is reported. Multi-crop scenarios run aggregate per-crop N-demand via coupled_production_multi.
BMP-based biogas yield + digester economics. Co-substrate blending uses VS-weighted BMP and volume-weighted nutrient concentrations. Defaults: 9.94 kWh/m³ CH₄ LHV, digestate-recovery factors 0.85 / 0.75 / 0.94 for N / P / K. Does NOT implement the full ADM1 kinetic model (left as future work).
Discounted cash flow over the project horizon. 0.75 economy-of-scale exponent, mid-life equipment replacement, terminal salvage, CHP heat + electricity offsets against on-site loads. compute_npv emits a return_details dict that downstream finance + ramp helpers consume verbatim (no 35/65 heuristic).
Auto-attached to every scenario. P&L + balance sheet + cash flow (monthly Y1–Y3, annual Y4–Y10). Debt schedule with capitalised construction-period interest, DSCR / LLCR / PLCR coverage ratios, loan covenants, four bank stress tests (rate +200 bps, revenue −20%, OPEX +15%, energy +30%). Default config: 60/40 D/E, 7% rate, 20-year term, 18-month construction, 1.20× DSCR covenant — overridable from the wizard's optional Finance step.
Logistic ramp (k = 6.0 default) applied to revenue + variable COGS across Year 1 and Year 2; fixed cost runs at full from Year 1. steady_state_from_npv_details() derives the rev/var/fixed split exactly from IAEM's return_details — replacing prior heuristic decomposition.
52-week × bay planting calendar with staggered or sequential rotation, phase-tagged. Fish-batch Gantt computes staggered cohort schedules using aquaculture.simulate_cohort growth math, sized to hit annual LWG target across n_tanks. Renders to the Production Plan PDF.
GWP100, blue water, land per kg sellable. Stage breakdown across electricity / feed / fertilizer / infrastructure. Economic allocation for coupled aquaponics by default (mass allocation is a toggle). ReCiPe2016 midpoint method.
Scalar adjusters per ISO2: feed-in tariffs, CapEx grants, production tax credits, carbon credits, plus a US federal stack (§45Q PTC, §48 ITC, REAP). Each rule cites its statute + effective date. Indicative only — verify with a local tax advisor.
MACRS class-life tables, §168(k) bonus depreciation phase-down (2022→2030 CSV-driven), §45Q carbon PTC, §48 ITC, US state credits, EU per-country linear depreciation. Plugs into finance.FinanceConfig via the TaxAdapter Protocol; open ships FlatRateTaxAdapter, neemo-tax ships MacrsTaxAdapter + EUDepreciationAdapter.
Pre-baked monthly CSVs per ISO2 + curated lat-band water-inlet temps scaffold the seasonal refactor of the flat RAS-heating and supplemental-lighting coefficients in the manuscript. The manuscript's Section 4.6 sketches the right shape (~209 MWh/yr water heating at Gothenburg tilapia; ~955 MWh/yr lighting at 57.7°N); shipping the seasonal calc in code is the next milestone.
Schema-driven runner reads validation_cases.yaml and asserts model output stays within published acceptance windows. Two cases ship at v1: UVI BAPS (Rakocy 2007/2012) and Avgoustaki & Xydis 2020 vertical lettuce. Outliers are acknowledged with case notes for genuine yield-optimisation-vs-demonstration-system gaps. Every release runs test_no_catastrophic_drift_against_published_data.
Per-scenario standalone doc (also 6th doc in the LoanPack.zip bundle). Renders math derivations per submodel, the 30-entry citation provenance table, the 426-test catalogue grouped by category, MC convergence + Q-Q + benchmark residuals charts, reconciliation between IAEM cashflows and finance.financials_extended (1% tolerance), and a reproducibility hash + rerun command. Designed for underwriters, peer reviewers, and model-risk-management.
Corrections layered in code at packages/neemo/src/neemo/species_corrections.py and packages/neemo/src/neemo/baps.py. The manuscript workbook (model_inputs.xlsx) is preserved byte-for-byte so the JCLP reproducibility chain stays intact. A full courtesy errata report is queued for Lobanov et al. prior to journal final acceptance.
Workbook stored the salmonid value (Bureau 1998 Eq. 14) under the tilapia row. Corrected to the tilapia value (Chowdhury 2013 Eq. 20). Raises feed requirement ~165% at given growth rate.
Same source mix-up. Affects dissolved-N budget that drives BAPS fish:plant coupling.
Workbook implied 45% via `1 − Processing Efficiency Filleted = 0.45`; FAO 2018 reports 33–35% for skinless tilapia. Reduces fish revenue 25–65%.
Workbook default reflects pumping-only. Badiola 2018 systematic review median for total system (incl. aeration, UV, filtration) is ~7.5. The original masked energy-cost sensitivity.
Manuscript formula acp × 1000 × N_Assim produced fish_lwg > plant_kg for leafy systems — violates the N mass balance for leafy greens at ~1.5 g N/kg FW. Frontier applies per-crop-group dry-matter fractions (leafy 0.05 / herb 0.10 / fruiting 0.06 / roots 0.12 / microgreen 0.08) before N assimilation. Legacy R parity test is xfail until upstream R model carries the same fix.
equipment_useful_life reattributed to AACE 16R-90 + Engle 2010 (was IRS Pub 946, a tax document). TGC for tilapia switched to Chowdhury 2013 + rainbow trout to Lugert 2014 (was Jobling 2003, less species-specific).
cost_year. Math stays EUR-canonical; conversion is at render time only.packages/neemo/src/neemo/data/policy/rules.json, updated quarterly.packages/neemo-tax (proprietary).The model and rendering pipeline are guarded by 426+ tests + 1 documented xfail across packages/neemo, packages/neemo-tax, apps/api, and apps/worker. The catalogue is tagged by category (numerical / render / plumbing / api / other) and surfaces in §5 of the per-scenario V&V PDF so “X tests passing” doesn't conflate plumbing with numerical correctness.