Conformal Cooling for AM Tooling: Design Rules, Channel Geometry, and Thermal Analysis

Conformal cooling is one of the best-supported business cases for metal AM. The cycle time and quality improvements are measurable, the ROI is calculable, and the application doesn't require aerospace-level qualification. For injection moulding and die casting insert manufacturing, it's also where LPBF tooling has displaced conventionally drilled tooling at most major Tier 1 moulders.

This article covers the engineering detail: how to design the channels, the thermal shortcuts you can apply before committing to full FEA, and the LPBF-specific constraints that determine what's actually buildable.

Why conformal cooling outperforms straight-drilled channels

Conventional injection mould cooling uses straight-drilled channels — usually 8–12 mm diameter, positioned as close to the cavity surface as the drill can reach without breaking through. The geometry is constrained by the requirement that a drill bit must travel in a straight line.

The consequence is thermal non-uniformity: cooling is efficient near the channel and poor at the midpoint between two channels, particularly around complex 3D cavity geometries. This non-uniformity causes:

• Warpage and sink marks in the moulded part (differential shrinkage between hot and cool zones)
• Longer cycle times because the hold phase must compensate for the hottest zone
• Higher scrap rates in tight-tolerance parts

Conformal cooling follows the cavity surface at a consistent depth, regardless of surface curvature. The thermal uniformity benefits are quantifiable:

The 20–35% cycle time figure is commonly cited and generally achievable. The actual reduction depends on the part geometry, polymer, and existing baseline — complex 3D parts with hot corners benefit more than flat-ish parts with adequate straight drilling coverage.

Channel geometry design rules

Diameter

Minimum channel diameter: 4 mm. LPBF can print internal channels smaller than 4 mm, but channels below 4 mm with circular cross-sections require internal supports that cannot be removed. The self-supporting limit for an internal circular bore in LPBF is approximately 4–6 mm diameter (the exact threshold depends on the material and parameter set). Use a teardrop cross-section if you need to go below this.

Maximum recommended diameter: 12–16 mm for single-circuit cooling. Above this, the pressure drop advantage of a larger channel is outweighed by reduced thermal uniformity — the heat transfer coefficient drops and the temperature gradient within the channel increases. Use parallel circuits rather than a single large channel for large inserts.

Pitch (centre-to-centre spacing)

Rule: pitch = 2–3× diameter.

For a 6 mm channel:

• Minimum pitch: 12 mm (heat transfer is optimal but the structural wall between channels is thin)
• Standard pitch: 15–18 mm (good thermal uniformity, adequate wall thickness)
• Maximum pitch: 20 mm (cycle time benefit begins to diminish for most part geometries)

Below the minimum pitch, the ligament between channels becomes thin enough to create structural risk in the insert — particularly during thermal cycling. The minimum ligament (wall between two adjacent channels) should be at least 3 mm for H13 steel.

Distance from cavity surface (shell distance)

Rule: surface distance = 1.5–2× diameter.

For a 6 mm channel, the centre should be 9–12 mm from the cavity surface.

Closer channels cool more uniformly but increase the risk of thermal fatigue cracking at the cavity surface. For tool steels in high-cycle injection moulding (>500,000 shots), keep the distance at 2× or greater unless the cycle time benefit specifically justifies the reduced margin.

Channel cross-section

Round (circular): Preferred for hydraulic efficiency — equal pressure drop in all directions, easy to analyse. The geometric constraint is that rounds above 4–6 mm require the teardrop modification.

Teardrop (pointed apex at top): The standard solution for LPBF conformal cooling. The pointed top is self-supporting regardless of channel size. The bottom and sides are circular; only the top ~60° arc is modified to a point. Hydraulic diameter is approximately 92–95% of the equivalent round, so pressure drop calculations for round channels are a good approximation.

Elliptical: Useful for tight spaces where the vertical clearance is limited but horizontal span is available. Hydraulic diameter is calculated from perimeter: D_h = 4A/P. Ellipses with aspect ratio >2:1 have significantly higher pressure drop for the same flow area.

Rounded square / rounded rectangle: Sometimes used to maximise surface area in a given wall volume. Higher thermal transfer area per unit flow, but higher pressure drop than round equivalent. Requires careful hydraulic balancing.

Manifolding and flow: achieving turbulent conditions

Laminar flow (Re < 2,300) in cooling channels is thermally inefficient — heat transfer coefficient is substantially lower than in turbulent flow (Re > 4,000).

Reynolds number:

Re = (ρ × v × D_h) / μ
   = (Q × D_h) / (A × ν)

Where:

• Q = volumetric flow rate (m³/s)
• D_h = hydraulic diameter (m)
• A = cross-sectional area (m²)
• ν = kinematic viscosity (m²/s; for water at 20°C: 1.004×10⁻⁶; at 60°C: 0.474×10⁻⁶)

For a 6 mm circular channel targeting Re = 5,000 with water at 40°C (ν ≈ 0.66×10⁻⁶ m²/s):

v = Re × ν / D_h = 5,000 × 0.66×10⁻⁶ / 0.006 = 0.55 m/s
Q = v × A = 0.55 × π(0.003)² = 1.55×10⁻⁵ m³/s = 0.93 L/min

That's achievable with standard tooling cooling circuits (0.5–5 L/min typical).

Parallel circuit balancing: When multiple parallel channels run from a common manifold, balance flow to within 5% between circuits. Imbalanced flow creates hot zones in the under-cooled circuit. Balance by:

1. Making all parallel circuits the same length and cross-section (ideal)
2. Using flow restrictors in shorter circuits
3. Using simulation to verify flow distribution before printing

Pressure drop: For turbulent flow in a smooth circular channel, use the Darcy-Weisbach equation with Moody friction factor f ≈ 0.02–0.04 for turbulent flow in typical conditions. Internal channel Ra in as-built LPBF is 15–30 µm — this is hydraulically rough and increases friction factor. Specify electropolishing if pressure drop is critical.

Baffle and bubbler alternatives

In cores and deep boss features, conformal channels may not be geometrically feasible. The standard alternatives are:

Baffle plate: A flat metal plate inserted into a drilled hole, dividing it into an inlet and return passage. Simple to install, no AM required. Works well for core diameters above ~15 mm. Below that, the baffle plate is too thin to be structurally reliable.

Fountain bubbler: A spiral or internal tube forces water to the tip of a blind hole and allows it to return around the outside. Good for cores 8–25 mm diameter. Can be AM-integrated by printing the bubbler tube as part of the insert.

Spiral insert: A helical channel machined into or AM-printed around a core pin. Higher surface area than a straight bubbler, better heat transfer. AM enables spiral geometries that are impossible to machine on small-diameter cores.

AM-integrated bubbler: The most effective solution for complex cores — print the cooling channel geometry directly into the LPBF insert rather than inserting a separate fitting. Eliminates joint leakage risk and allows conformal routing around core geometry.

LPBF-specific design constraints

Minimum wall between channel and cavity: 1.5 mm minimum; 2 mm recommended for cyclic loading applications. Below 1.5 mm, the AM surface roughness and microstructural variability create unpredictable mechanical behaviour.

Channel support structures: Circular channels above 8 mm diameter in LPBF require internal support structures unless converted to teardrop or other self-supporting cross-sections. Internal supports cannot be removed from closed channels — always use teardrop for channels that will carry fluid.

Lattice infill for non-functional volume: Large insert bodies with minimal functional features benefit from sparse lattice infill in non-cooling zones. This reduces:

• Overall build mass (lower material cost, shorter build time)
• Residual stress and distortion during build
• Risk of delamination in large solid sections

Use a gyroid or TPMS lattice with 20–30% density in non-load-bearing zones. Ensure the cavity-side wall is solid (minimum 5 mm solid skin before lattice begins).

Orientation: Print inserts with the cavity face down-facing only if absolutely necessary — the down-facing surface has higher Ra (15–40 µm vs. 8–15 µm up-facing). Cavity surfaces should be up-facing or vertical, with post-machining planned in any case.

Powder evacuation: Include a drain hole or removable plug at the lowest point of every closed cavity. Trapped powder in cooling channels causes blockage and, in high-humidity environments, corrosion. Minimum 3 mm drain hole is adequate for depowdering with compressed air.

Thermal analysis shortcuts

Full transient thermal FEA is the gold standard, but it's time-consuming and requires detailed material property data. Two analytical shortcuts cover most early-design decisions:

Biot number check

The Biot number determines whether lumped capacitance (simple thermal model) is valid:

Bi = h × L_c / k

Where:

• h = convective heat transfer coefficient (W/m²·K); turbulent water in 6 mm channel ≈ 8,000–15,000 W/m²·K
• L_c = characteristic length = Volume/Surface area (m)
• k = thermal conductivity of insert material (W/m·K); H13 ≈ 28 W/m·K

For Bi < 0.1, the temperature gradient within the solid is negligible compared to the convective resistance — lumped capacitance is valid and full FEA is not required for a first-order estimate.

In practice for conformal cooling inserts, Bi is typically 0.3–2.0, meaning full 3D thermal analysis provides value, but the Biot number check tells you quickly whether your cooling is convection-dominated (Bi < 0.1) or conduction-dominated (Bi > 1).

Fourier number and cooling time estimate

The Fourier number estimates how long it takes for thermal diffusion to penetrate a given wall thickness:

Fo = α × t / L²

Where:

• α = thermal diffusivity = k/(ρ·c_p) (m²/s)
• t = time (s)
• L = half-thickness of the mould wall (m)

For H13 steel: α ≈ 6.8×10⁻⁶ m²/s

A moulded polymer part typically requires the cavity surface to drop below the ejection temperature (~80–120°C for most thermoplastics). Setting Fo ≈ 1 and solving for t:

t ≈ L² / α = (0.010)² / 6.8×10⁻⁶ ≈ 14.7 s

For a 10 mm half-wall mould, the minimum thermal soak time is ~15 seconds. If your current cycle includes a 25-second cooling phase, there is geometric improvement potential. Use this estimate before commissioning FEA to determine whether the thermal analysis investment is warranted.

Material selection for LPBF tooling inserts

H13 is the default choice. Its printability, post-machining characteristics, and tempering response are well-characterised. For high-pressure die casting inserts (up to 700 bar cavity pressure), consider M300 maraging steel for its superior toughness after aging.

Copper-chromium-zirconium (CuCrZr) inserts are used for extreme thermal loading — gate areas, thin features near hot runners. Thermal conductivity is 10× H13, but LPBF printing of copper alloys requires specific laser parameter development and careful oxygen control.

Post-machining and finishing

Cavity surface: Plan for EDM (sinker or wire) to finish cavity surfaces after LPBF. EDM removes the AM surface layer and achieves Ra 0.4–1.6 µm before hand polishing to Ra <0.1 µm for optical/cosmetic surfaces. Do not apply aggressive EDM settings immediately after LPBF — the brittle re-melt layer (white layer) from EDM on as-built AM material is prone to micro-cracking. Apply a light stress relief anneal at 600°C/2 hours before EDM finishing.

Internal cooling channels: Electropolish cooling channels to reduce Ra from 15–30 µm to <3 µm. Benefits:

• Lower pressure drop and more predictable flow distribution
• Reduced bacterial growth risk (food and medical moulding)
• Better corrosion resistance in water-cooled circuits
• Easier inspection and cleaning

Electropolishing aluminium LPBF (AlSi10Mg) is straightforward. H13 and maraging steel electropolishing requires specific electrolyte selection — consult your post-processing supplier before designing tolerances that depend on post-EP dimensions.

Worked example: cycle time calculation

Part: 150 g PA6 component, 3 mm wall section, mould surface at 80°C melt, target ejection at 60°C.

Insert: H13 LPBF, conformal cooling, 6 mm teardrop channels at 10 mm from cavity, 16 mm pitch, water at 15°C, Re = 6,000.

Convective h ≈ 12,000 W/m²·K (turbulent water in 6 mm channel)

PA6 thermal diffusivity ≈ 1.1×10⁻⁷ m²/s. Half-wall thickness L = 1.5 mm = 0.0015 m.

Cooling time estimate from polymer side:

t ≈ (L²/α) × ln(π/4 × ΔT_initial/ΔT_final)
  ≈ (0.0015²/1.1×10⁻⁷) × ln(π/4 × (80–15)/(60–15))
  ≈ 20.5 × ln(1.13) ≈ 20.5 × 0.12 ≈ 2.5 s

Actual cooling cycles add safety margin and gate solidification time — expect 8–15 s total cooling phase for this part. Without conformal cooling (assuming ΔT across cavity surface is +15°C), the equivalent straight-drilled insert requires 30–40% longer cooling phase. This calculation, done analytically in 5 minutes, gives sufficient basis to proceed to LPBF quote before FEA is commissioned.

Real case study values

Published results from BMW Werkzeugbau (tooling division) for a B-pillar interior trim moulding tool:

• Baseline (straight drilled): 48-second cycle, 3.8% scrap rate, ΔT cavity 34°C
• Conformal cooling LPBF insert: 31-second cycle (35% reduction), 1.2% scrap, ΔT cavity 9°C
• AM insert cost premium over straight-drilled: €8,400
• Annual savings (3 shifts, 250 days): ~€62,000 cycle time reduction + €11,000 scrap reduction
• ROI payback period: 1.4 months

This case is at the high end of benefit (complex 3D geometry, significant thermal non-uniformity in baseline). For simpler parts, 15–20% cycle time reduction is more representative.

Further reading

• Conformal cooling calculator — channel geometry, pitch, and flow rate optimisation
• Thermal distortion tool — distortion risk in LPBF tooling
• Surface roughness guide — as-built Ra and post-processing targets
• DfAM checklist — pre-build design review
• Cost per part calculator — LPBF insert cost vs. conventional tooling