Technical Manual · Thermal Lab
Liquid-to-Chip System Modelling Lab — Methodology & Formulas
Every input, equation, constant, output, and reference behind the LTC steady-state solver. All math is executed by the deterministic engine js/ltc-system-modelling-lab.js (function computeModel) — no back-solved constants, no random values. This is a SIMULATED / educational model; it is not a substitute for site-specific hydraulic or thermal commissioning data.
01 Purpose & engineering basis
The lab sizes and balances a liquid-to-chip AI campus: coolant flow and pump power, CDUs, the heat rejection for the liquid loop, the CRAHs, chilled water, chillers and towers for the heat liquid cooling does not capture, the cooling of UPS and transformer losses, and the PUE, WUE, CUE, energy, water and cost that follow.
Since v3.17.0 every physical number comes from the workbench’s shared plant engine, js/ltc-plant.js, through js/ltc-lab-plant.js. The engine is a steady-state energy balance that closes: electricity in equals heat out on every path, asserted to within 0.5 % on 202 test plants (it closes to about 10⁻¹⁴). The lab’s design scores, risk index and confidence remain its own screening layer. The same plant is the one every LTC Workbench page reads, so a change here moves the Tier simulator, ASHRAE, ISO, TIA and NFPA tools, and the reverse.
02 Inputs
Three tiers. Most readers touch only the first. Everything else is auto: derived, or a stated default, until changed.
| Tier | Inputs | Default |
|---|---|---|
| Design | IT load · liquid capture · TCS supply (return follows at ΔT 10 K) · plant configuration (4 presets) · CDU feed single/dual · redundancy | 100 MW · 85 % · 35 °C · dry coolers + water-cooled chillers · single · N+1 |
| Site | Climate (illustrative tropical / temperate / arid bins, or bins pasted from the site’s EPW) · grid emission factor · renewable share · heat reuse · electricity and water tariffs | tropical · 0.72 kg/kWh · 0 % · 0 % · 0.10 /kWh · 1.5 /m³ |
| Advanced | 55 engine inputs: unit sizes, approaches, heads and efficiencies, chiller COP and staging, electrical losses, hall geometry, thermal masses | stated in LTCPlant.DEFAULTS; ASSUMED values are typical design values, not a vendor selection |
The lab’s own panel still carries its 61 controls under Advanced · auto; the six primary sliders and the site fields write the shared plant on every run. A plant chosen on another workbench page is read first, so the lab’s defaults never overwrite it.
03 Calculation methodology
One operating point = IT at a utilisation, ambient dry and wet bulb. The design day sizes the plant; the year is the same plant run over the climate bins.
Liquid path
Air path — the heat liquid cooling does not capture
Electrical chain and metering points (ISO/IEC 30134-2)
Annual figures
04 Constants & data sources
STANDARD: water cp 4.186 kJ/kg·K, ρ 997 kg/m³; air cp 1.006 kJ/kg·K, ρ 1.18 kg/m³; h_fg 2,430 kJ/kg; 1 RT = 3.517 kW. ASSUMED (each editable under Advanced): CDU 2,000 kW, CRAH 250 kW, chiller 4,000 kW_th at rated COP 6.5, SFP 1.2 kW per m³/s, TCS head 25 m at η 0.70, FWS head 30 m at η 0.75, UPS 97 %, transformer 99 %, cord 0.5 %, PDU 1.5 %, LV 0.5 %, tower cycles 4, cooler minimum approaches dry 4 K / adiabatic 1.5 K / fluid 2 K / tower 3 K. The climate bins are illustrative; replace them with bins built from the site’s EPW file.
05 Outputs
| Output | Derivation | Note |
|---|---|---|
| PUE (annual / design) | Σ facility·h / Σ IT·h; design-day facility / IT | The lab’s headline PUE is annual; the design day sizes the plant |
| PUE categories 1/2/3 | M0 / M1, M0 / M2, M0 / M3 | Same facility, three numbers: state the category |
| Flow, pump power, CDUs | TCS loop | Pump power = TCS + FWS pumps |
| System COP | IT / all cooling power | Pumps, fans, chillers, trim, towers |
| WUE, water, CUE, CO₂ | Over the bins | Site layer sets grid factor and renewable share |
| Energy, water cost, net OPEX | M0 × tariff, water × tariff − reuse credit | Tariffs from the site layer |
| Scores, risk, confidence | Lab screening model | Screening, not a compliance finding |
06 Worked example
Defaults: 100 MW IT in 5 halls, capture 85 %, TCS 35/45 °C, dry coolers + water-cooled chillers, tropical bins (design 35 °C dry bulb / 28 °C wet bulb), utilisation 85 %. Every number is what the engine returns; tools/test-ltc-plant.mjs recomputes the first ones by hand.
- Liquid heat:
100,000 × 0.85 =85,000 kW; TCS ΔT 10 K →85,000 / (4.186 × 10) =2,030.6 kg/s = 2,036.7 L/s - TCS pumps:
997 × 9.81 × 25 × 2.0367 / 1000 / 0.70 =711 kW; CDU duty 85,711 kW → 10 CDUs per hall (9 + 1) - Facility water 32/42 °C at 2,033 L/s. At 35 °C the dry coolers leave 39 °C: the trim chiller carries 70 % on the design day and runs 3,160 h a year; 65 coolers including the spare, sized for the cooler hours
- Air path: 15,000 kW IT + 857 kW CDU skin + 500 kW cords + 1,530 kW PDUs, fan heat closed at 8.42 % → 3,907 kW per hall, 17 CRAHs (16 + 1)
- Chillers: CRAHs + 5,005 kW electrical rooms + CHW pumps = 25,327 kW; 8 machines at 79 % load, COP 6.81
- Losses: UPS 3,178 kW, LV 530 kW, transformers 1,298 kW; mechanical (MCC) 22,004 kW; facility 129,750 kW
- PUE 1.298 design-day, 1.189 annual; categories 1/2/3 = 1.157 / 1.183 / 1.189 (M0 885,295 · M1 765,016 · M2 748,323 · M3 744,600 MWh)
- WUE 0.67 L/kWh (499,745 m³/yr, the trim chiller’s condenser towers); CUE 0.856 kgCO₂/kWh at 0.72 kg/kWh; energy balance closes to 1.2 × 10⁻¹⁴
A Legacy screening model (comparison only)
Shown only when Legacy screening model is ticked under Run Model. This was the lab’s engine until v3.17.0: one deterministic pass with heuristic multipliers (density stress, control index, future factor) that no component can be traced to, and no separate air, chilled-water or electrical-room path. Kept for comparison; its equations follow, as they are in computeModelLegacy.
Steady-state energy balance (conceptual)
Supply temperature rise & mass flow
Hydraulic pump power (affinity-law corrected)
fanKw = itKw × (fanPower%) × (1 − effectiveCapture/100 × 0.55) × (1 − controlBonus × 0.18) — residual fan demand after liquid-capture and control reductions.HX Institute · pump affinity laws · inline computeModel
HX effectiveness & liquid cooling COP
PUE impact (facility + CDU pump)
System COP
WUE and annual energy
07 References & standards
- ASHRAE TC9.9 Liquid Cooling Design Guide — Liquid cooling system design, coolant selection, temperature ranges, flow sizing methodology for data centers.
- HX Institute Standards — Heat exchanger thermal performance, approach temperature, effectiveness-NTU method, and hydraulic sizing.
- Pump affinity laws (centrifugal) — Flow ∝ speed; head ∝ speed²; power ∝ speed³. Used in fan-power modelling (fanKw ∝ flow³ scaling with control reduction).
- ISO/IEC 30134-2 / 30134-9 — PUE and WUE metric definitions for data centres. (ISO 14644, cited here before, is the cleanroom air-cleanliness standard and defines neither.)
- ANSI/BICSI 002 — Data center design and implementation best practices; redundancy configurations.
- Uptime Institute Tier Standard — Availability tiers (N, N+1, 2N) and their concurrently-maintainable requirements.
- ASHRAE 90.4 / Green Grid — PUE and system COP definitions and reporting methodology.
- Engine source —
js/ltc-plant.js(physics, shared with every workbench page) throughjs/ltc-lab-plant.js; the legacy screening model iscomputeModelLegacyinjs/ltc-system-modelling-lab.js. Gates:tools/test-ltc-plant.mjs,tools/test-ltc-workbench.mjs.
08 Assumptions & limitations
Steady state at each operating point; annual figures are hour-weighted over climate bins at a mean utilisation, which captures ambient-driven effects (trim hours, chiller lift, fan energy) but not load dynamics. Fault responses on the workbench (DH01 map, Tier sweep) use a stated first-order thermal-mass model checked against its analytic solution, and the DH01 map is a thermal approximation, not CFD. Unit sizes and efficiencies are typical design values until a vendor selection replaces them; climate bins are illustrative until replaced with the site’s EPW bins.
This manual and the lab are engineering education and pre-design aids. Outputs are labelled SIMULATED; scores are screening, not compliance findings.