Engineering CalculatorsCold Plate DesignerWork in progress
Hydraulic and thermal performance of a liquid cold plate whose coolant channel you build up from straight sections and 45°/90°/180° bends — heat-transfer coefficient, pressure drop, thermal resistance and base temperature for a given heat load. Drop an offset pin-fin field into any section to model a direct-cooled power-module baseplate. Pairs with the Heatsink, Heat Exchanger and Flow-in-Pipes calculators.
Work in progress: this calculator is still being validated and refined — treat its results as indicative and cross-check anything design-critical against CFD or a bench test.
Plan-view sketch — it routes simple serpentine paths well, but more complex bend sequences can lay out inaccurately. Visual aid only; it doesn't affect the calculated results.
New to Liquid Cold Plate Design? Read the guide: Cold Plate Design: Rectangular-Channel Heat Transfer and Pressure Drop — a plain-language explainer of the standard behind this calculator.
Rectangular-channel liquid cold plate, steady single-phase Newtonian flow. Each straight section is solved on its own hydraulic diameter Dh = 2·w·h/(w+h) and per-channel velocity. The Darcy friction factor uses the rectangular-duct Poiseuille number f·Re = 96·(1 − 1.3553α + 1.9467α² − …) for laminar flow (Shah & London, α = short/long side — 56.9 at a square, 96 at parallel plates, so a channel is not the 64/Re of a round pipe) and Swamee-Jain when turbulent, interpolated across the 2300–4000 transitional band. The heat-transfer coefficient h = Nu·k/Dh uses the constant-heat-flux (H1) rectangular-duct laminar Nusselt 8.235·(1 − 2.0421α + …) (3.61 at a square) and Dittus-Boelter when turbulent. Bend losses use ΣK·(ρv²/2) with representative sharp-milled-bend K (45°≈0.3, 90°≈1.1, 180°≈2.0), which vary with bend radius. Thermal resistance from the channel wall to the inlet fluid is R_conv = 1/UA (UA = Σ h·wetted-area over every section and channel) plus R_caloric = 1/(2·ṁ·cp); optional 1-D base conduction t/(k·A) adds the module path. The channel side and top walls are counted as fully-effective heat-transfer area (no fin-efficiency derating), which over-estimates UA for tall/thin channels — treat the thermal result as a first-order estimate. Fluid density and cp reuse this site's coolant presets; ν/k/Pr reuse the Heat Exchanger transport table. Verify against CFD or a bench test before committing a design.
Offset pin-fin sections (⬢, direct-cooled power-module baseplates) are modelled as a staggered bank of circular pins in cross-flow, with all velocities and Reynolds numbers on the maximum velocity through the narrowest gap and the pin diameter as length scale. Heat transfer uses the Zukauskas staggered-bank Nusselt (C·Cₙ·Rem·Pr0.36·(S_T/S_L)0.2, with the tube-row correction Cₙ). Pressure drop uses the Gaddis-Gnielinski (VDI Heat Atlas) staggered drag coefficient per row, ΔP = ½·ζ·N_rows·ρ·u_max², so packing the pins tighter raises the pressure drop sharply. Each pin's contribution is derated by a circular-fin efficiency (tanh(mL)/mL, m = √(4h/k_pin·D)) using the selected pin material. Because the pinned baseplate is itself the sealing surface, only the base side is heated and you pick the pin-tip boundary: an open-cavitytip wetted by coolant (corrected length L + D/4) or an against-housing tip that bottoms out on the sealed housing floor and loses no heat there (adiabatic). The few-row inlet/outlet correction (f_nt) is neglected (valid for ≳10 rows), the wall-to-fluid Pr ratio is taken as 1, and any tip-clearance bypass flow is not modelled — reasonable for a first-order estimate, but confirm a direct-cooled design against CFD or a bench test.
Validated: the rectangular-duct correlations reproduce their Shah & London anchor values exactly — laminar f·Re = 56.92 and Nusselt = 3.61 for a square channel, and 96 / 8.235 for the parallel-plate limit. A hand-worked single 5×3 mm × 100 mm water channel at 1 L/min (Dh = 3.75 mm, v = 1.11 m/s, Re ≈ 4150 turbulent) reproduces h ≈ 6270 W/m²K, a 0.66 kPa section pressure drop, a 1.44 °C coolant temperature rise and a 180° bend loss of 1.23 kPa — all matching hand calculation. The pin-fin path is checked against an independent hand calculation too: a Ø1 mm staggered array at 2 mm pitch in 40 °C water (approach 0.5 m/s → 1.0 m/s in the gaps, Re 1515) gives ζ = 0.482, h ≈ 30,200 W/m²K and a 4.78 kPa loss over 40 mm — turning a 0.035 kPa plain channel into 4.8 kPa while cutting its thermal resistance ~6.6×.