Masterplan
7. Performance
Unit N+1 on every fleet we sized. Dry most of the year. Remaining PE studies are in §7.5.
27 of 31 design checks pass. Failures and out-of-scope items are listed below.
7.1Reliability and redundancy
Unit N+1 on every fleet we sized. Availability target 0.995. Each island is 0.997 internally, so the chance all 3 are up at once is 0.991. Expected IT on a given hour is 349.3 MW. Losing one section drops you to 233.6 MW, not to zero. That is still not Uptime Tier III.
Three numbers, three questions. Probability all 3 sections are up at once: 0.9910. Expected IT available on any given hour: 349 MW, which is 99.7% of contracted. SLA target we will sign: 0.995. The first falls as the campus grows and the second does not, because losing one section of 3 costs 33% of IT rather than the campus. Sign against the second, disclose the first.
Uptime Institute Tier III requires concurrently maintainable paths. Split islands without closed ties are not that. We do not claim Tier III or Tier IV. A tenant SLA that copies those words will be false until the sections are tied or ringed.
N+1 at the block. 3 independent 34.5 kV sections at 117 MW IT each: losing one drops 33% of IT, not the campus. No closed ties, so Uptime Tier III is not claimed. Independent sections, each internally N+1 at 0.997 working. Industry block figure, not a parts list, and not a Markov model. A real RBD needs component MTBF from the switchgear and PCU vendors.
7.2Code compliance status
| Result | Check | Detail |
|---|---|---|
| FAIL | Storage covers all three duties | 140 MW / 75 MWh specified against 175.2 MW / 13.0 MWh required: 175.2 MW GPU transient, a 34.5 MW unit trip, and a 20 min bridge |
| SCOPE | Islanded dynamics | the studies above are first-cycle and steady-state. What they do not cover is machine decrement over the following seconds, inverter fault behaviour under its own control law, and transient stability after a trip, which on an island is where protection actually gets decided |
| ACTION | Transformer procurement | 0 in stock · 160 of 160 to order at 40 weeks |
| ACTION | PCU procurement | 0 in stock · 372 of 372 to order at 40 weeks |
Failures exist because we refused to mark islanded dynamics as passed, and because procurement is not an electrical pass. Interrupting, arc-flash (worst 17.8 cal/cm² at LVBUS-HOUSE-001), voltage drop, and N-1 transfers passed on the island estimator (26.5 kA MV / 87.1 kA LV). pandapower IEC 60909 on the same section: 28.6 kA ikss (+7.9% vs the closed-form island estimate), peak 72.3 kA. Catalog 63 kA still clears. A PE still stamps ETAP before the turbine PO.
Code checks C01–C26
| ID | Rule | Cls | Status | Finding |
|---|---|---|---|---|
C01 | Transformer loading within nameplate less headroom NEC 220; manufacturer | CI | Calculated pass | worst 3,710 kVA of 3,750 kVA (99%) |
C02 | Bus and busway loading UL 891/1558; IEC 61439 | CI | Calculated pass | worst 92.8% on C-MECH-BRANCH-0001 (1 x 4000 A busway) |
C03 | Interrupting rating ≥ available fault NEC 110.9 | PS | Calculated pass | 26.5 kA collector · 87.1 kA LV bus · 5.3 kA PCU-limited · 6 of 18 machines on the worst of 3 collector section(s) at 0.16 pu plus 46.6667 MW of grid-forming inverter; subtransient, decaying within cycles |
C04 | Series-rating policy NEC 110.9; UL | PS | Selected n/a | Radial plant; listed series combinations are not used. |
C05 | Continuous load 125% NEC 210.20(A); 215.2(A)(1) | CI | Calculated pass | every low-voltage trip sits at or below conductor ampacity |
C06 | Ampacity after derating NEC 310.15, 310.16 | CI | Calculated pass | worst 92.8% on C-MECH-BRANCH-0001 (1 x 4000 A busway) |
C07 | Voltage drop practice 3%/5% | CI | Calculated pass | 1.46% PCU to rack against a 3% limit |
C08 | Ground-fault protection of equipment NEC 230.95; 215.10 | PS | Selected n/a | 415Y/240, not 480Y/277 ≥1,000 A service GFPE trigger. |
C09 | Working space NEC 110.26 | PS | Pending must_ask | NEC 110.26 clearances are a layout PE check, not this compile. |
C10 | Selective coordination NEC 700/701/708; IEEE 242 | ND | Calculated pass | 5 adjacent device pairs, every one selective by at least its coordinating interval (0.1 s low voltage, 0.25 s medium) |
C11 | Redundancy N-1 invariant IEEE 3002.x | CI | Calculated pass | 3 reserve unit picks up 3,710 kVA through 2 x 3200 A busway |
C12 | Spare excluded from capacity practice | CI | Calculated pass | Checked 3 transformer classes. Installed counts exceed required on every one, so the reserve unit is genuinely spare and is not carrying load: 137 installed vs 134 required at 3750 kVA, 17 installed vs 14 required at 3750 kVA, 6 installed vs 3 required at 3750 kVA |
C13 | Zone and facility reconciliation practice | CI | Calculated pass | 3 zones sum to 350 MW against a 350 MW target (gap 0). Facility 372.74 MW equals PCU input plus mechanical plus house to 0.004 MW. |
C14 | Single-point-of-failure scan (2N) IEEE 3002.x | PS | Selected n/a | Unit N+1, not 2N. Sections are independent, not dual-path. |
C15 | Phase balance practice | CI | Calculated pass | Three-phase NVIDIA power shelves. Aggregate rack imbalance is unpublished. |
C16 | Transformer inrush vs upstream trip IEEE 242 | ND | Estimate scope | the studies above are first-cycle and steady-state. What they do not cover is machine decrement over the following seconds, inverter fault behaviour under its own control law, and transient stability after a trip, which on an island is where protection actually gets decided |
C17 | Cable short-circuit withstand ICEA; IEEE 242 | PS | Calculated pass | low voltage 89.5 kA X/R-adjusted against 100 kA · medium voltage 40.3 kA against a 1.6 x 31.5 = 50.4 kA close-and-latch capability |
C18 | Arc-flash incident energy IEEE 1584-2018 | ND | Calculated pass | worst 17.8 cal/cm2 at LVBUS-HOUSE-001, PPE category 3, boundary 7.3 m, cleared in 0.05 s by C-SECONDARY-0155 |
C19 | DC bipolar balance OCP Diablo 400 | CI | Selected n/a | 800 VDC path deferred (Alternative D). |
C20 | DC ground-fault detection IEEE 946 | PS | Selected n/a | 800 VDC path deferred (Alternative D). |
C21 | DC arc-fault detection OCP | ND | Selected n/a | 800 VDC path deferred (Alternative D). |
C22 | DC isolation monitoring IEC; IEEE 946 | PS | Selected n/a | 800 VDC path deferred (Alternative D). |
C23 | DC-rated interruption UL/IEC emerging | PS | Selected n/a | 800 VDC path deferred (Alternative D). |
C24 | BESS separation and limits UL 9540; NFPA 855 | PS | fail fail | 140 MW / 75 MWh specified against 175.2 MW / 13.0 MWh required: 175.2 MW GPU transient, a 34.5 MW unit trip, and a 20 min bridge |
C25 | Occupancy consistency IBC F-1/S-1 | PS | Calculated pass | IBC F-1/S-1 selected for halls. Not a sealed occupancy study. |
C26 | Ground-grid touch and step IEEE 80 with IEEE 81 ρs | PS | Calculated pass | IEEE 80 screen: ρ=100.0 Ω·m, Rg=0.058 Ω, island GPR 553.0 V vs Etouch 595.0 V. Wenner still required to seal. |
Codes: NEC 2023 through TAS 2012. King County will not be the electrical AHJ; TDLR will.
7.3Efficiency: what can be guaranteed
- PUE 1.064 Calculated from the cooling mix. Offer a design PUE of 1.26 with a 0.1 annual deadband. 2024 at this fence: 4.3–110.1 °F, ~110.1 °F 0.4%-class cooling design. Do not write 1.064 into a lease.
- Mechanical PUE 1.0179 Calculated from 52,608 hourly pairs: fans 7 GWh/yr and pumps 49 GWh/yr, 6.3 MW average. The electrical half is the conversion and distribution stack. Total PUE is the product.
- WUE 0.0467 L/kWh Calculated from 116 AF/yr over 357 adiabatic hours. Do not put this in a contract before pump tests.
- Path efficiency 95.6% Calculated PCU 97% times residual transformer and distribution. 800 VDC would claim more; it is Alternative D.
- CUE 0.586 kgCO2/kWh-IT Estimate heat-rate × EPA NG EF. About 1,528,071 tCO2/yr. No methane leakage, no Scope 3, no 24/7 CFE.
7.4Environmental performance
Emissions follow the turbine fleet. First power at 158.5 MW gross sits under the 360 MW TCEQ standard-permit ceiling. The campus at 422.7 MW gross does not, so case-by-case NSR is on the Phase 2 critical path. At 1,528,071 tCO2/yr the campus is 20.4× the 75,000 tpy PSD greenhouse-gas trigger, which makes Title V and PSD/GHG BACT mandatory rather than a contingency. CUE is a heat-rate sketch (0.586 kgCO2/kWh-IT), not a vendor curve. Noise at the fence is modeled: Buy the silencers (~$15M, inside the owned-generation line). Then 50 dBA night is met at ~396 m, inside the gen-island setback. Skip them and the 50 dBA contour jumps to ~1 km and starts eating the pad. OEM octave-band Lw (NEMA TR-1 / ISO 3744) and one 30-day baseline at the fence. Replaces this ±5 dB. Water: MAG 49 AF/yr is ~30.4 gpm, enough for camp domestic and fire but not for 116.1 AF/yr dry-plant demand (~72 gpm). Chemistry is gypsum (TDS thousands). Dry cooling is not optional; the remaining water is an import / GCD-variance / brackish-RO question, not a well-count question. Construction SWPPP under TXR150000.
| Receptor | km | dBA, no extra silencer | dBA, with OEM silencers |
|---|---|---|---|
| Pad edge / CR-117 (working) Closest public road and fence. Night criterion 50 dBA (WHO / typical rural insurer), Texas has no statute. | 0.4 | 64.9 | 49.9 |
| Workforce camp 2.5 mi east of perimeter. Sleeping crews. 45 dBA night inside tents/cabins after façade. | 4 | 34.7 | 19.7 |
| Guthrie (county seat) USGS populated place. Not a noise-sensitive city; still the named town. | 19 | -20.6 | -35.6 |
7.5Studies still required before construction
| Study | Cost | Lead | Blocks |
|---|---|---|---|
| Test wells + 72-hr pumps + TDS/chemistry Hydrogeologist | $150–300k | 8–12 weeks | Well permits, RO, WUE language |
| Geotech borings + karst geophysics + IEEE 81 resistivity Terracon / Collier class | $250–450k | 8–12 weeks | Foundations, C26 ground grid |
| ALTA/NSPS 2021 + Schedule B Texas RPLS | $80–150k | 8–16 weeks | Title insurance, Ω_build legal, plats |
| Mineral runsheet + SUAs Landman + O&G counsel | $40–90k + negotiation | 6–12 weeks | LOI / surface-use close |
| Phase I ESA (ASTM E2247-23 rural) Environmental | $25–45k | 4–6 weeks | Lender; Phase II on two RRC locations |
| H&H 2D (5/25/100-yr) + AJD Civil hydrology PE | $120–220k | 12–20 weeks | FFE, detention, 404 |
| IPaC + habitat assessment; cultural / THC USFWS/archaeologist | $40–120k | 8–16 weeks | Federal nexus filings |
| Islanded protection & stability Power-systems PE | $180–350k | 12–20 weeks | Relay settings, turbine procurement |
| WETT informal screening (POI fault, headroom) WETT / ERCOT SSWG | screening fee + NDA | 8–16 weeks | Sealed short-circuit; Phase 3 filing |
| ISO 9613-2 noise + FM wind/hail envelope Acoustics / insurer engineer | $30–70k | 6–10 weeks | Setbacks, roof spec |