CEWT 20 MW CRT Trigeneration

CEWT 20 MW CRT Trigeneration Latrobe Valley AI Data Centre — Resilience & Storage Summary Preliminary engineering/investor note • 7 October 2026 1. Executive summary CEWT’s proposed Latrobe Valley configuration combines renewable electricity, a modest short-duration BESS/UPS layer and Carbon Recycling Technology (CRT) using recycled methane (RSNG) as the long-duration firming medium. The central proposition is: renewables supply the primary energy; batteries provide speed; CRT provides endurance. For a constant 20 MW AI data-centre load, three days of autonomous generation corresponds to 1,440 MWh of electricity. On the frozen 45.2% gas-engine electrical-efficiency basis, this requires approximately 229.4 t of methane. Using a preliminary liquid-methane density basis and 20% storage allowance, the corresponding gross RSNG/LNG storage envelope is about 650 m³. 2. Proposed three-layer resilience architecture Layer 1 — UPS/BESS: millisecond-to-minute continuity for GPU transients, switching, and ride-through. Layer 2 — Short-duration BESS: renewable smoothing and engine-transition support; a preliminary 2–4 hour range (about 40–80 MWh at 20 MW) should be evaluated. Layer 3 — CRT/RSNG: hours-to-days firming, with recycled methane acting as the strategic energy reserve. Normal operation should preferentially route renewable electricity directly to the data centre. PEM electrolysis and methanation should use allocated/surplus renewable energy to replenish the RSNG inventory. CRT generation is then dispatched during prolonged renewable shortfalls. 3. BESS versus RSNG — physical storage comparison Firming duration Electricity delivered Indicative BESS installed* CH₄ required Gross RSNG/LNG storage** 24 h 480 MWh 613 MWh 76.5 t ~216 m³ 48 h 960 MWh 1,227 MWh 152.9 t ~432 m³ 72 h 1,440 MWh 1,840 MWh 229.4 t ~648 m³ * Screening basis used in the working comparison: 90% BESS round-trip efficiency plus 15% energy reserve. ** Preliminary liquid-methane basis with 20% storage allowance; final tank sizing must include usable capacity, heel, boil-off, composition, and code requirements. 4. 72-hour CRT reserve calculation Continuous electrical load 20 MW Electrical energy for 72 h 1,440 MWh Frozen engine electrical efficiency 45.2% Required methane thermal input ≈3,186 MWhₜₕ Methane inventory ≈229.4 t CH₄ Simple Sabatier H₂ required to regenerate that inventory ≈114.7 t H₂ PEM electricity at 53 kWh/kg H₂ ≈6.08 GWh The 6.08 GWh figure is a reserve-regeneration calculation, not a claim of battery-like round-trip efficiency. The simple electricity→H₂→CH₄→engine pathway is inherently less efficient than a lithium-ion battery, which is why renewable electricity should feed the data centre directly whenever available and CRT should be used for long-duration firming rather than daily cycling. 5. PEM capacity sensitivity PEM capacity H₂ production @ 53 kWh/kg 72-h reserve recharge time Approx. days 20 MW 377 kg/h 304 h 12.7 40 MW 755 kg/h 152 h 6.3 60 MW 1,132 kg/h 101 h 4.2 80 MW 1,509 kg/h 76 h 3.2 84.4 MW 1,593 kg/h 72 h 3.0 Recommendation: retain the 20 MW PEM as the present demonstration basis rather than oversizing the electrolyser solely to refill an emergency reserve rapidly. At 20 MW, a completely depleted 72-hour reserve requires about 12.7 equivalent full-load days to regenerate. If reserve depletion is infrequent, gradual replenishment can be commercially preferable to installing 80+ MW of PEM capacity for a 20 MW data-centre demonstration. 6. Market comparison and positioning The emerging 20–30 MW AI data-centre segment provides a useful comparator for CEWT. Renewable+BESS projects demonstrate that developers are considering modular initial phases rather than only very large hyperscale campuses. CEWT’s differentiation is not that CRT should replace batteries: the design deliberately retains BESS for fast response while using chemical energy storage for prolonged firming. Investor positioning: do not compare CRT with a four-hour battery on $/kWh alone. Compare complete 24/7 resilience architectures. As firming duration extends from hours toward days, the physical storage requirement for a battery-only solution becomes very large, while the RSNG inventory remains compact. 7. Design basis to retain • AI data-centre load: constant 20 MW. • Generation: six gas engines, 20 MW net aggregate, modelled at 45.2% electrical efficiency. • PEM: retain 20 MW for the current demonstration basis; larger modules remain a sensitivity case. • Strategic RSNG reserve: approximately 229 t usable CH₄ for 72-hour autonomy; preliminary gross storage envelope ~650 m³. • BESS: size for UPS/transients and short-duration smoothing, not multi-day energy storage. • Cooling: retain liquid-cooling integration and evaluate dry/low-water heat rejection alongside useful recovery of engine and methanation heat. • Start-up: external RLNG can establish the initial carbon/methane inventory; CRT subsequently recycles captured carbon. 8. Important boundary conditions The 229.4 t CH₄ / 114.7 t H₂ calculation uses the simple CO₂ + 4H₂ → CH₄ + 2H₂O relationship to size the strategic reserve. It is not the final integrated CRT process mass balance. The detailed 20 MW Trigen model must continue to use the agreed CRT stoichiometry and H₂-rich syngas design basis. The 20 MW Latrobe Valley Trigen project must remain separate from the CEWT Green Iron / Western Australia case, including the much larger ~1.15 GW electrolyser requirement derived for that project. Values must not be transferred between the two designs. 9. Working investment proposition “Renewables supply the energy. Batteries provide speed. CRT provides endurance. Recycled methane provides the strategic energy reserve.” Next engineering step: close the full 20 MW CRT mass and energy balance, confirm CO₂ capture and methanation duties, select the short-duration BESS/UPS rating, and obtain vendor-based CAPEX/OPEX so the hybrid architecture can be compared with renewable+BESS-only alternatives on an equivalent availability basis.