i. What CCECC is (and isn't)

CCECC stands for Cradle to Cradle Energy Conceptual Calculations. It is a fast, first-principles energy ledger for any project that makes, moves, stores, or dissipates energy.

The one rule: every number in a CCECC evaluation is expressed in joules, watts, or their multiples (kJ, MJ, GJ, TJ; Wh, kWh, MWh, GWh, TWh). Not kilograms of CO₂. Not dollars. Not qualitative hand-waves. If you cannot state it in energy units, it does not belong in the ledger.

Why this strictness? Because carbon footprints, cost figures, and vibes all have escape hatches — discount rates, boundary games, offsets, marketing. A joule does not. A joule delivered is a joule; a joule embodied is a joule; a joule recoverable is a joule. The energy ledger is the one accounting system a physicist, an engineer, and a policymaker all have to agree on.

What it is

• An energy-balance worksheet you can fill in over a single afternoon using public data and rough estimates.
• A tool for engineers, founders, planners, and procurement officers at the feasibility stage of a project.
• A communication device — the output is two or three energy ratios and a short verdict, not a 200-page report.

What it is not

• Not a replacement for ISO 14040 life-cycle assessment.
• Not a carbon accounting tool. (Carbon follows from energy and grid mix; CCECC deliberately stops at the energy layer.)
• Not a certification scheme.
• Not a substitute for detailed techno-economic modelling once a project is real.

CCECC aims at the ~10% of LCA work that drives ~80% of early energy decisions.

ii. When to use it

Use CCECC when you need a first-principles answer to one of these questions:

• What is the energy payback time, roughly?
• What is the approximate EROEI of this project?
• How many joules does this system consume per unit of service delivered, vs. the alternative?
• What fraction of the energy flow is fresh vs. waste-stream?
• How much of the embodied energy comes back at end of life?

Don't use CCECC when you need auditable carbon accounting for compliance, detailed techno-economic cost modelling, or product certification — it is not designed for any of those.

iii. The method in four moves

Each move asks an energy-balance question. Answer each one with a number in joules or watt-hours (or the corresponding multiple — MJ, GJ, kWh, GWh, TWh). Rough order-of-magnitude estimates are fine; precision is not the point, the ledger is.

Move 1 — Upstream cradle: E_in

What energy goes in before the project delivers a single useful joule?

• Embodied energy of the hardware (E_emb, in MJ or GJ). Use handbook figures per kg of each material, multiplied by mass.
• Annual operating input energy (E_op,in, in MJ/yr or GWh/yr) — fuel, feedstock, or electricity drawn during normal operation.
• Fraction of operating input that is fresh vs. waste-stream. A joule of waste heat you would otherwise vent is not equivalent to a joule of fresh grid electricity; flag which is which.

Move 2 — Use phase: E_out

What energy does the system deliver as useful work, heat, or service?

• Annual useful energy output (E_out, in kWh/yr, GWh/yr, or GJ/yr).
• Conversion efficiency (η) — useful E_out divided by E_op,in.
• Expected operating lifetime (yr). Lifetime output = E_out × years.
• For systems that do not produce energy but consume it to deliver a service (a server, a tire, a building), record the service-unit and the energy per service unit (J/prompt, MJ/km, kWh/m²·yr).

Move 3 — Downstream cradle: E_recover

How much of the embodied energy comes back at end of life?

• Recoverable material energy — the fraction of E_emb avoided by recycling vs. virgin production.
• Calorific energy — if combustible, heating value × mass (MJ/kg).
• Decommissioning energy cost (subtract from recovery).
• Net: E_recover = recovered material + calorific − decommissioning.

Move 4 — Systemic loop check: EROEI and EPBT

Two headline ratios tell you whether the ledger closes:

• EROEI (Energy Return on Energy Invested) = (lifetime E_out + E_recover) / (E_emb + lifetime E_op,in − waste-stream contribution)
• EPBT (Energy Payback Time) = E_emb / annual net E_out, in years.
• Also record: what fraction of E_op,in is genuinely waste-stream, not fresh extraction. This is the circularity question the ratios alone don't capture.

iv. The one-page worksheet

Copy this for each project. Every blank is filled with a number in joules, watt-hours, or their multiples. Use a plain unit suffix (MJ, GJ, kWh, GWh) next to each figure.

PROJECT NAME:
DATE / ANALYST:
SYSTEM BOUNDARY (what's in, what's out):

[1] UPSTREAM CRADLE — energy in
    Embodied energy of hardware (E_emb) .......... ___ GJ
    Annual operating input (E_op,in) ............. ___ GWh/yr
        of which waste-stream (not fresh) ........ ___ %
    Expected lifetime ............................ ___ yr

[2] USE PHASE — energy out
    Annual useful output (E_out) ................. ___ GWh/yr
    Conversion efficiency η = E_out / E_op,in .... ___ %
    Lifetime output = E_out × lifetime ........... ___ GWh
    Service-unit (if non-energy output) .......... ___ J per unit

[3] DOWNSTREAM CRADLE — energy recoverable
    Recoverable material energy .................. ___ GJ
    Calorific energy (if combustible) ............ ___ GJ
    Decommissioning energy cost .................. ___ GJ
    Net E_recover ................................ ___ GJ

[4] SYSTEMIC LOOP — ratios
    EROEI = (lifetime E_out + E_recover)
            / (E_emb + lifetime fresh E_op,in) ... ___
    EPBT = E_emb / annual net E_out .............. ___ yr
    Fresh-joule fraction of input ................ ___ %

VERDICT (≤100 words, anchored in the ratios above):

Scoring thresholds

Scores follow from the ratios, not from intuition. Typical cut-offs for energy-producing systems:

For systems that consume energy to deliver a non-energy service (servers, vehicles, tires, buildings), EROEI is undefined. State this explicitly, and evaluate on joules per service-unit compared to the next-best alternative, using the same Strong/Medium/Weak bands relative to that alternative.

v. Chaotan One

sCO₂ waste-heat power · Liupanshui · China

Project: Chaotan One — 2 × 15 MW supercritical CO₂ waste-heat power plant, Liupanshui, Guizhou, China. Developer: Nuclear Power Institute of China (NPIC), under China National Nuclear Corporation (CNNC). Commercial operation from December 20, 2025. Coupled to a Shougang Shuicheng steel sinter line.

Move 1 — Upstream: E_in

• E_op,in: ~250 GWh/yr of thermal waste heat at sinter-exhaust temperatures (back-calculated from the announced ~70 GWh_e/yr output at an assumed ~28% thermal-to-electric efficiency for a Brayton sCO₂ cycle on a sinter heat source; CNNC has not publicly disclosed measured cycle efficiency).
• Waste-stream share: ~100%. The sinter line dumps this heat to atmosphere without Chaotan One; the fresh-joule fraction of input is effectively 0%.
• E_emb of hardware: 30 MW × ~3 GJ/kW for high-alloy sCO₂ plant (nickel superalloys, printed-circuit heat exchangers, >73-atm vessels) ≈ 90 GJ = 25 GWh embodied. About 2× a steam Rankine plant of the same capacity (~1.5 GJ/kW), reflecting exotic metallurgy.

Move 2 — Use phase: E_out

• Annual E_out: ~70 GWh_e/yr = 252 TJ/yr (CNNC / Shougang figure).
• η (thermal → electric): ~28% assumed, vs. ~18–22% for conventional sinter-waste steam cycles — claimed by CNNC to be best-in-class at this heat source and scale.
• Assumed lifetime: 20 yr (with the caveat in Move 3 below).
• Lifetime E_out: 70 × 20 = 1,400 GWh = 5,040 TJ.

Move 3 — Downstream: E_recover

• Nickel-alloy recovery: high-grade superalloys have embodied energy ~200 MJ/kg. Assuming ~80 t of superalloy in the hot end and ~80% recovery efficiency at end-of-life → ~13 GJ recoverable as avoided-virgin energy.
• Steel structure: ~500 t structural steel at ~25 MJ/kg embodied, ~90% recovery → ~11 GJ.
• Decommissioning cost: ~2 GJ (crane, transport, cutting).
• Net E_recover ≈ 22 GJ, or about 24% of E_emb. Sits in the Medium band. The uncertainty: if hot-end creep or sCO₂ corrosion halves the 20-year lifetime, E_recover is unchanged but EROEI roughly halves — this is the real open question for the technology.

Move 4 — Systemic: EROEI and EPBT

• EROEI = (lifetime E_out + E_recover) / (E_emb + fresh E_op,in) = (5,040 + 0.022 TJ) / (0.090 TJ + 0 TJ) ≈ 56.
• EPBT = E_emb / annual E_out = 90 GJ / 252,000 GJ·yr⁻¹ ≈ 0.36 yr (~4 months).
• If lifetime halves to 10 yr, EROEI drops to ~28 — still solidly Strong.
• Fresh-joule fraction of input: ~0%. This is what makes the ratios so favourable.

Verdict

On a pure energy ledger, Chaotan One is the best kind of CCECC project: a ~4-month energy payback and an EROEI of ~56 because its fuel side contributes zero fresh joules. The embodied energy of the alloy-heavy hot end (~25 GWh) is only ~0.5% of lifetime output (~1,400 GWh). The single open question is lifetime — if sCO₂ corrosion and creep shorten hardware life to under ten years, EROEI halves but remains comfortably Strong. Downstream recovery at ~24% of E_emb is Medium and has room to improve with better alloy-recycling streams.

vi. One AI prompt vs. the sulphur in one winter tire

Comparing a digital service to a physical material on common cradle-to-cradle terms.

Why this odd pairing? CCECC is most useful when it forces us to compare unlike things on a common cradle-to-cradle basis. Setting one AI prompt — a digital, ephemeral service — against the sulphur in one winter tire — a physical, durable material input — reveals something that per-kilogram carbon comparisons miss: whether the impact is happening inside a loop or outside one.

Worked example 2 — side A · One typical AI prompt

A median-length query to ChatGPT, Claude, or Gemini in 2026.

Note: a prompt consumes energy but produces no joules of useful energy output. EROEI is therefore undefined. CCECC evaluates this on joules per service-unit (J/prompt) against alternatives, not on an energy-return ratio.

Move 1 — Upstream: E_in per prompt

• End-to-end inference electricity per median prompt: 0.24–0.34 Wh = 860–1,225 J.
  • Google Gemini median text prompt: 0.24 Wh, full-stack measurement including accelerator, host CPU/RAM, idle reserve, and data-centre overhead. AI accelerators alone account for ~58% of this; the remaining ~42% is host machine, idle backup, and cooling/conversion. (Google Cloud / arXiv 2508.15734, Aug 2025, data from May 2025.)
  • OpenAI/ChatGPT average query: ~0.34 Wh, per Sam Altman's blog post; methodology not disclosed. (Jun 2025.)
  • Epoch AI modelled estimate for GPT-4o: ~0.3 Wh. (Feb 2025.)
  • Note: do not apply a separate PUE multiplier on top of the Google figure — it already includes data-centre overhead.
• Amortised GPU embodied energy: a data-centre GPU has ~1–2 GJ embodied and serves in the low millions of prompts over 3–5 yr life. Adds roughly 300–1,000 J per prompt. (Inherited estimate, not independently re-verified in this pass.)
• Total E_in per median prompt: ~1,150–2,225 J, i.e. ~0.3–0.6 Wh.
• Waste-stream share: ~0%. Every joule is freshly drawn from grid generation, except where hyperscalers contract dedicated renewables (a market-based accounting trick, not a thermodynamic one).

Move 2 — Use phase: J per service-unit

• E_out (useful joules): 0 J. The output is bits.
• Alternative comparisons (J per "equivalent answer"):
  • Reading a well-chosen Wikipedia article: ~200–500 J (≈0.05–0.14 Wh end-to-end).
  • Driving 1 km to a library: ~2,000,000 J (~2 MJ), ~1,000× the prompt.
  • Asking a domain expert by phone for 3 min: ~5,000–20,000 J (phone + network).
• A prompt is an efficient J/service-unit compared to physical-movement alternatives, and comparable to or modestly worse than a pre-existing web page. Hence Medium on this move — context-dependent.

Move 3 — Downstream: E_recover

• The prompt itself leaves no recoverable artefact. E_recover per prompt = 0 J.
• The serving GPU has ~1–2 GJ embodied; precious-metal recovery at end-of-life returns perhaps 5–10% of that as avoided-virgin energy — but this is pooled across millions of prompts, not attributable to any one.

Move 4 — Systemic

• EROEI: undefined (no energy output).
• Aggregate AI draw (Schneider Electric SRI scenarios, 2025): global AI electricity demand ~100 TWh in 2025; modelled 2030 outcomes range from 510 TWh ("Limits to Growth" scenario) to 880 TWh ("Abundance Without Boundaries"). European subset: ~15 TWh in 2025 → 45–145 TWh by 2030 across the same four scenarios. For context, the IEA projects total data-centre electricity demand (not AI-only) to roughly double from ~415 TWh in 2024 to ~945 TWh by 2030.
• Fresh-joule fraction of input: effectively 100% globally (slightly lower where hyperscalers contract dedicated renewables, but grid-equivalent accounting still attributes the fresh draw).

Worked example 2 — side B · The sulphur in one passenger-car winter tire

~150 g of elemental sulphur in a ~10 kg tire.

A tire doesn't produce energy either. But the sulphur's role is to avoid a tenfold increase in tire consumption — so the meaningful energy ledger is "joules of future tire manufacturing avoided" per joule of sulphur invested.

Move 1 — Upstream: E_in for 150 g of sulphur

• Embodied energy of recovered Claus sulphur: ~0.5–1 MJ/kg (by-product of oil & gas desulphurisation; the energy is almost entirely attributable to the host refining process).
• 150 g × ~1 MJ/kg = 0.15 MJ = 150 kJ per tire.
• Waste-stream share: ~95%. The refining industry has to remove sulphur regardless; the tire industry absorbs it productively.
• Energy to distribute the sulphur through the compound during vulcanisation: ~30–50 MJ for the whole tire, but only a fraction attributable to sulphur. Net marginal energy of "using sulphur" vs. "not using sulphur": essentially the 150 kJ above.

Move 2 — Use phase: functional leverage in joules

• Embodied energy of a 10 kg tire: ~750 MJ (rubber compound ~75 MJ/kg including synthetic rubber, carbon black, steel belt, processing).
• Without sulphur cross-linking, the tire would fail in ~1/10 the distance (soft in summer, brittle in winter — exactly the 1839 problem Goodyear solved).
• Avoided manufacturing energy: 9 × 750 MJ = 6,750 MJ over the same driving distance.
• Rolling-resistance energy over tire life (~60,000 km): ~3,000 MJ dissipated as heat (~50 J/m × 60 × 10⁶ m).

Move 3 — Downstream: E_recover

• Calorific energy: 10 kg tire × ~32 MJ/kg heating value = ~320 MJ recoverable as tyre-derived fuel.
• Material recycling of the crosslinked rubber is essentially zero — vulcanisation is irreversible by design.
• E_recover / E_emb ≈ 320 / 750 = 43%. Medium, all of it as low-grade fuel rather than material.

Move 4 — Systemic: leverage ratio

• Energy-leverage ratio (the sulphur equivalent of EROEI, for a material that enables rather than produces energy) = avoided manufacturing energy / energy invested in the sulphur = 6,750 MJ / 0.15 MJ ≈ 45,000.
• Fresh-joule fraction of input: ~5%.
• Caveat: as oil & gas peak, Claus sulphur supply shrinks. If the tire industry has to mine virgin sulphur (from Frasch-process deposits), the input-side embodied energy rises roughly 10× — still excellent leverage, but worth watching.

Verdict — the comparison in joules

Per service-unit, an AI prompt costs ~1,150–2,225 J and produces zero joules of return. The sulphur in one tire costs ~150 kJ and enables ~6,750 MJ of manufacturing to be avoided — a leverage ratio near 45,000. CCECC's verdict is stark: a 19th-century material intervention has an energy-leverage ratio four orders of magnitude higher than a 21st-century digital service. This is not an argument against AI — it is an argument that AI's energy ledger only closes when a prompt leads to an action that itself saves joules (a better route, a correct part number, a trip avoided). If the prompt's output is discarded or replaces nothing, every one of its ~1,500 J is a pure net draw on fresh energy. The pedagogy: joules-per-service-unit is the only unit in which unlike systems can be honestly compared.

vii. Second-life EV batteries

Repurposed for grid storage · a cradle-to-cradle archetype.

Project: Lithium-ion battery packs retired from electric vehicles (at ~70–80% of original capacity) repurposed as stationary grid-scale or behind-the-meter storage. Reference pack: 60 kWh NMC, ~400 kg. Real deployments include Nissan/Eaton xStorage, BMW's Leipzig plant storage, Renault/Connected Energy.

Move 1 — Upstream: E_in for second life

• Original E_emb (first-life manufacturing): 60 kWh NMC pack × ~150 MJ/kg × 400 kg = ~60 GJ embodied (equivalent to ~16,700 kWh of manufacturing energy, or ~280 kWh per kWh of pack capacity). Inherited 2017-era figure; recent process studies (Degen et al. 2022/2023) put cell-level production at 147–225 MJ/kWh, which would reduce the pack figure if updated.
• Attributed to second life: 0 GJ. The pack already exists; repurposing is pure deferral of new manufacturing.
• Refurbishment energy: ~0.5–2 GJ per pack (disassembly, cell grading, BMS installation, enclosure). Call it 1 GJ.
• Waste-stream share of E_in: ~98%. The only fresh joules are the refurbishment step.

Move 2 — Use phase: E_out

• Residual usable capacity: 60 kWh × 75% = 45 kWh per pack.
• Second-life duration: ~7 years.
• Assume 1 full-cycle-equivalent per day in grid-storage duty (solar shifting or peak shaving).
• Lifetime second-life throughput: 45 kWh × 365 × 7 = ~115 MWh = 414 GJ delivered.
• Round-trip efficiency: ~82% (slightly below a new pack's ~90% due to degradation).

Move 3 — Downstream: E_recover

• After second life, the pack still enters the same recycling stream as a first-life pack. Second life defers recycling, it does not avoid it.
• Hydrometallurgical recovery (2026 state of the art): ~60–70% of E_emb recoverable as avoided-virgin-material energy.
• E_recover: 0.65 × 60 GJ = ~39 GJ.

Move 4 — Systemic: EROEI for the second-life intervention

• Treating the intervention as "the decision to run a second life instead of scrapping":
• EROEI = second-life E_out / refurbishment E_in = 414 GJ / 1 GJ ≈ 414.
• EPBT for refurbishment energy: 1 GJ / (414 GJ / 7 yr) ≈ 6 days.
• Even if you attribute 20% of the original E_emb (12 GJ) to second life as a crediting approach, EROEI is still ~32 — solidly Strong.
• Fresh-joule fraction: ~2%.

Verdict

The energy ledger for second-life EV batteries is almost unbeatable: a ~1 GJ refurbishment investment returns ~414 GJ of grid storage over seven years before the pack enters recycling at ~65% embodied-energy recovery. EROEI ≈ 400, EPBT ≈ 6 days. This is what "cradle to cradle" looks like when expressed in joules — the original manufacturing energy is amortised over two productive lives plus a recycling tail, rather than one life and a shredder.

viii. How to contribute

This handbook is a living document. The three examples here are seeds, not the canon.

Ways to contribute

1. Submit a worked example using the one-page worksheet above. Aim for 500–800 words.
2. Challenge a score. If you think one of the scores above is wrong, send a revised radar with reasoning.
3. Translate. Danish, German, Spanish, Mandarin, Hindi, and French translations are priorities.
4. Build tooling. Spreadsheets, web tools, API integrations, and batch-compare utilities are all high-leverage next steps.

Target domains for next examples

• Heat pumps replacing gas boilers (building scale).
• Biogas from food waste (agriculture + energy).
• Green hydrogen via electrolysis vs. grey hydrogen from steam reforming.
• Timber construction vs. reinforced concrete.
• Single-use vs. returnable packaging systems.

Core principle for all contributions: Fast, honest, and first-principles. A CCECC that takes a month to produce is not a CCECC.