Conjugate heat transfer CFD analysis of LGA1700 CPU water block with jet impingement cooling and porous media fin array modeling.
Geometry Source: EK-CPU Lignum on GrabCAD
CFD analysis demonstrating conjugate heat transfer modeling of a high-performance CPU liquid cooling block for LGA1700 socket. Features jet impingement cooling, porous media representation of microchannel fin arrays, and orthotropic thermal conductivity for directional heat transfer.
Key Capabilities Demonstrated:
- Conjugate heat transfer (CHT) between solid copper and liquid coolant
- Porous media modeling with directional thermal and flow resistance
- Realistic thermal interface resistance (TIM contact modeling)
- Energy balance validation and convergence monitoring
The analysis began with a detailed CAD model of the EK-CPU Lignum water block from GrabCAD. The original geometry included full mechanical detail with mounting hardware, aesthetic features, and complex internal fin structures.
Original imported geometry showing full mechanical assembly
Internal fin array structure from CAD model
The geometry was simplified for CFD analysis while preserving critical thermal-hydraulic features:
Simplifications Made:
- Removed mounting hardware and cosmetic features
- Extracted fluid domain from internal channels
- Simplified complex fin array geometry to porous media representation
- Created uniform CPU block representing IHS (Integrated Heat Spreader)
- Defined inlet/outlet boundaries for flow simulation
Retained Critical Features:
- Jet impingement plate geometry
- Overall flow path and channel dimensions
- CPU contact surface area and geometry
- Thermal mass distribution of copper cold plate
Final CFD-ready geometry with fluid domain and simplified features
- CPU Block (IHS representation): 45mm × 38mm × 4mm (LGA1700 standard)
- Fin Spacing: 0.314 mm
- Fin Thickness: 0.371 mm
- Calculated Porosity: 0.458 (0.5 used for this simulation)
Geometry available: CFD_Geo_STP.stp
Overall mesh structure with cut plane showing internal regions (outer shell transparent)
- CPU (IHS): 54,368 cells
- Cold Plate: 1,698,419 cells
- Fluid Inlet: 1,003,017 cells
- Porous Medium: 4,335,825 cells
- Fluid Outlet: 642,756 cells
- Total cell count: 7,734,385 cells
Refined mesh on CPU block to capture thermal gradients
Details:
- Uniform refinement throughout solid volume
- Fine resolution at IHS/cold plate interface for CHT coupling
Cold plate copper base mesh
Porous zone mesh representing microchannel fin array
Complete fluid domain mesh showing inlet/outlet regions
- Solver Type: Segregated Flow/Energy
- Analysis Type: Steady-state
- Turbulence Model: K-Omega SST
- Heat Transfer: Conjugate Heat Transfer (CHT) with solid-fluid coupling
Cold Plate (Copper)
- Density: 8,940 kg/m³
- Specific Heat: 386 J/kg·K
- Thermal Conductivity: 398 W/m·K
CPU Block (Silicon)
- Density: 2,329 kg/m³
- Specific Heat: 702 J/kg·K
- Thermal Conductivity: 124 W/m·K
Coolant (Liquid Water)
- Temperature-dependent properties
- Reference temperature: 26.85°C
The fin array is modeled as a porous medium with anisotropic properties to represent the directional nature of heat transfer and flow resistance through the microchannels.
| Property | XX Direction | YY Direction | ZZ Direction |
|---|---|---|---|
| Viscous Resistance (kg/m³·s) | 100,000 | 100,000 | 1.0×10⁸ |
| Inertial Resistance (kg/m⁴) | 1.5 | 1.5 | 100 |
| Thermal Conductivity (W/m·K) | 217 | 217 | 1.0 |
Porosity: 0.5 (50% open volume for flow)
Directional Behavior:
- XX, YY (Flow directions): Low resistance, high effective thermal conductivity (copper-dominated parallel conduction)
- ZZ (Blocked by fins): High resistance, low thermal conductivity (water-limited serial conduction)
Porosity Calculation:
ε = (fin spacing) / (fin spacing + fin thickness)
ε = 0.314 mm / (0.314 mm + 0.371 mm) = 0.458 → rounded to 0.5
Viscous Resistance (Flow Directions): Based on Darcy flow through parallel plate channels:
1/α ≈ 12μ/h²
where h = hydraulic diameter (fin spacing)
Thermal Conductivity (Orthotropic):
- Flow directions (XX, YY): Volume-weighted arithmetic mean (parallel conduction through copper fins)
- k_eff = ε·k_water + (1-ε)·k_copper ≈ 217 W/m·K
- Cross-flow direction (ZZ): Harmonic mean (serial resistance through water gaps)
- k_eff ≈ 1.0 W/m·K (water-limited)
CPU/Cold Plate Contact:
- Interface resistance: 2.5×10⁻⁴ m²·K/W
- Represents high-quality thermal paste (e.g., Arctic MX-4, Noctua NT-H1)
- Applied at IHS/cold plate interface
- Type: Mass flow inlet
- Mass flow rate: 0.01 kg/s (10 g/s)
- Temperature: 26.85°C (300 K)
- Turbulence: 1% intensity, 1 mm length scale
- Type: Pressure outlet
- Gauge pressure: 0 Pa (atmospheric reference)
- Backflow temperature: 26.85°C
- Type: Volumetric heat generation
- Power: 250 W
- Distribution: Uniform across CPU block volume
- Represents: Intel Core i9-12900K at max turbo (PL2)
- External surfaces: Adiabatic (no heat loss to ambient)
- Internal interfaces: Coupled (conjugate heat transfer)
- Reference pressure: 101,325 Pa
- Gravity: Disabled (forced convection dominated)
Temperature distribution through centerplane (26.8°C to 74.9°C)
Temperature distribution capped at 40.1°C to highlight fluid heating
Fluid Thermal Behavior:
- Coolant enters at 26.85°C
- Progressive heating as flow moves radially outward from impingement zone
- Hottest fluid regions at fin array exit (approaching outlet)
Flow streamlines from inlet to outlet colored by temperature
Flow Characteristics:
- Jet impingement creates radial flow pattern from center
- Coolant temperature increases along flow path through porous fin array
- Uniform flow distribution through microchannels
- Exit temperature varies by flow path length
| Parameter | Value |
|---|---|
| CPU Peak Temperature | 75.5°C |
| CPU Average Temperature | ~73.2°C |
| Inlet Temperature | 26.85°C |
| Outlet Temperature | ~32.9°C |
| Temperature Rise (ΔT) | ~6°C |
| Heat Input | 250 W |
| Heat Removed | ~249 W |
| Energy Balance | 99.6% |
Heat Input:
- CPU volumetric heat generation: 250 W
Heat Removed (from CFD):
Q = ṁ × Cp × (T_out - T_in)
Q = 0.01 kg/s × 4186 J/kg·K × (32.9 - 26.85)°C
Q = 0.01 × 4186 × 6.05
Q = 253.3 W
Energy closure: 253.3/250 = 101.3%
Note: Slight over-prediction (~1.3%) is within acceptable numerical accuracy for steady-state CFD.
Predicted outlet temperature (from first law of thermodynamics):
ΔT = Q / (ṁ × Cp) = 250 W / (0.01 kg/s × 4186 J/kg·K) = 5.97°C
T_out_theory = 26.85 + 5.97 = 32.82°C
CFD result: 32.9°C
Difference: 0.08°C (0.24%) Excellent agreement
Case-to-coolant thermal resistance:
R_case-coolant = (T_CPU_avg - T_fluid_in) / Q
R = (73.2°C - 26.85°C) / 250W = 0.185 K/W
Peak thermal resistance:
R_peak = (75.5°C - 26.85°C) / 250W = 0.195 K/W
Typical high-performance CPU water blocks: 0.10-0.20 K/W (case to coolant)
Assessment: Simulated thermal resistance of 0.185-0.195 K/W falls within the expected range for a quality water block with thermal paste interface (2.5E-4 m²·K/W contact resistance). Results are consistent with real-world performance of premium liquid cooling solutions.