Stratos Hyperscale Data Center
Subsurface Thermal Risk Analysis

1. Site Dossier

The Stratos Project occupies a 40,000+ acre parcel in Hansel Valley, a closed graben within the Basin and Range physiographic province of northwestern Utah. The basin is hydrologically closed, tributary to the Great Salt Lake, with no perennial surface water. The facility is proposed to reach 9 GW IT at full buildout with 16 GW total thermal rejection.

1.1 Key Parameters

Confidence ratings indicate data quality: HIGH = directly measured/stated; MEDIUM = inferred/regional estimate; LOW = unknown/modeled.

1.2 Johansen Profile — Hansel Valley Nominal Stratigraphy

The basin floor is underlain by late Quaternary lacustrine deposits from three major lake cycles (Little Valley ~140 ka, Hansel Valley ~80 ka, Lake Bonneville 30–10 ka). The Bonneville cycle dominates the basin floor surface. Because MAP < 28 cm/yr with PET ~120 cm/yr, the vadose zone above the water table is in a state of permanent moisture deficit. This is the physics-critical site condition — it controls thermal conductivity.

2. Physics Audit — Substrata Module Coverage

All physics modules are implemented in TypeScript in web/app/lib/ and run in the browser via the Next.js app. For this Python case study, they were ported directly from source. The ILS/MILS line-source model is not the primary model here — Stratos uses air-cooled rejection with no ground heat exchanger, so the dominant subsurface loading mechanism is a surface thermal boundary condition.

2.1 Modules Applied

2.2 Gaps vs. This Site's Requirements

3. Simulation Results

3.1 Thermal Input by Scenario

The facility uses closed-loop hydronic cooling with air-cooled dry heat exchangers. All waste heat is rejected to the atmosphere — there is no direct ground heat exchanger. The primary subsurface loading mechanism is the surface temperature anomaly created by the localized heat island. Surface ΔT is computed from an energy balance: ΔT_surface = Q_flux / h_eff where h_eff = 20 W/m²K (composite convection + radiation, desert terrain, average wind 5.6 m/s).

Area basis: 10,000 acres (40.5 km²) active development footprint. Thermal rejection scales linearly with IT load (16 GW at 9 GW IT per Stratos dossier).

3.2 Scenario A — Phase I Cap (1.5 GW IT)

3.3 Scenario B — Phase I Full (3 GW IT)

3.4 Scenario C — Full Buildout (9 GW IT)

3.5 Peclet Number Analysis — Transport Regime

Pe = (q × ρC_p_fluid × L) / λ_eff, where L = 1 m reference length. For saline lacustrine silts with poor hydraulic connectivity in a flat graben: q_Darcy ≈ 1×10⁻⁸ – 2×10⁻⁷ m/s.

4. Key Risk Findings

4.1 Thermal Saturation of Hansel Valley Lacustrine Clays

The permanent moisture deficit (MAP 28 cm/yr ≪ PET ~120 cm/yr) means vadose zone conductivity is near dry-state baseline (0.22–0.59 W/m·K at S_r = 0.10–0.25). Heat accumulates in the near-surface rather than dispersing. This is thermal saturation: the soil cannot shed heat fast enough to prevent a growing thermal anomaly that deepens year over year.

The thermal front reaches the water table within 10 years at 3 GW and within 10 years at 9 GW full buildout. By year 30, the plume extends to 56.8 m depth under full buildout conditions — well into the consolidated lakebed formation.

4.2 Soil Dry-Out Feedback Loop

Quantified sensitivity: If S_r drops from 0.20 → 0.10 (plausible under sustained heat loading), λ_eff drops from 0.56 → 0.22 W/m·K — a 61% reduction in conductivity. This would increase thermal penetration depth by approximately 30–50% at the 30-year horizon.

4.3 Locomotive Springs Hydrological Linkage

The Locomotive Springs wetland complex (~15 km south) is a critical migratory bird habitat with regulatory curtailment triggers: if deep groundwater extraction causes a measurable reduction in spring discharge, pumping must be immediately curtailed. Two risk pathways:

4.4 Seismic + Liquefaction Compound Risk

The 1934 M6.6 earthquake on the Hansel Valley Fault produced documented liquefaction, sand blows, and lateral spreading across the identical Quaternary lake bed sediments. The fault is <5 km from the facility, near-vertical, with 10–16 ky recurrence. Geothermal warming of the saturated zone reduces effective cohesion in clay-silt mixes by approximately 5–15% per 10°C (Mitchell & Soga, 2005). At full buildout, ΔT in the saturated zone reaches 9.86°C at 10m depth at 10 years. Over a 30–50 year facility lifetime, compound event probability is non-negligible.

5. Substrata vs. Industry Standard Methods

What static methods miss here: (1) The 2.5–7× overestimate of conductivity from using saturated k on a permanently dry site. (2) The positive feedback between heat loading and moisture depletion in the vadose zone. (3) The decadal timescale of subsurface heat accumulation — static methods show a snapshot, not a trajectory. (4) The compound seismic + thermal risk in the saturated silt zone.

6. Limitations and Confidence Bounds

7. Next Steps

Data that would materially improve the model

Substrata modules that need to be built for this site class