Many engineers have been told that 3D printing frees them from design-for-manufacture constraints. As most have discovered through practice, it replaces one set of rules with another. Multi Jet Fusion (MJF), like any thermal process, is governed by heat flow, material crystallization, and geometry; all of which determine how close a printed part stays to its intended shape. Ignoring these fundamentals leads to the same issues seen in injection molding: warpage, sink, distortion, and dimensional drift.

This paper outlines how to maximize geometric stability in PA12 parts printed with HP’s Multi Jet Fusion technology. It focuses on the variables that matter most before a build: geometry, wall thickness, feature transitions, and orientation—and explains how design choices influence heat distribution and shrink behavior. The recommendations draw from MJF process data, injection molding design standards, and polymer crystallization studies.

While post-processing methods such as thermal annealing can fine-tune accuracy, the primary control point is design. By applying these guidelines, engineers can achieve tighter, more repeatable tolerances and reduce the need for corrective machining or secondary operations.

The goal is to provide a clear, data-backed framework for producing dimensionally stable MJF parts that meet engineering intent on the first run.

Designing for Maximum Dimensional Stability in MJF PA12

The rules below are the fastest levers to reduce warp, curl, and drift before you touch an anneal cycle.

Understanding the tolerances and limitations of MJF

What the Platform Can Hold

On the HP Jet Fusion 5200 running HR PA12, HP reports a Cpk 1.33 dimensional capability of approximately ±0.25 mm in the XY plane for features under 30 mm, and ±0.42–0.60 mm in Z depending on build height. With the hardware-specific dimensional profile enabled and HP 3D Process Control active, these tolerances tighten to ±0.17–0.23 mm (XY) and ±0.25–0.37 mm (Z) for parts up to about 80 mm in size [1]. These figures reflect measured statistical capability under optimized conditions—controlled powder refresh rates, verified calibration, and stable chamber temperature uniformity.

HP also lists a more general accuracy specification of ±0.20 mm or ±0.038 mm per 25 mm for the Jet Fusion 5200 series. In practical terms, that means a 100 mm feature would typically hold ±0.38 mm under ideal conditions. However, most experienced service bureaus—who deal with part variability across geometries, orientations, and environmental conditions—recommend using about looser tolerances, around ±0.30 mm or ±0.05 mm per 25 mm.

If you’re outsourcing production, always confirm your vendor’s published tolerance guidelines rather than relying solely on HP’s nominal specifications. Each bureau develops its own statistical process control data based on their machine calibration, powder handling practices, and build volume utilization. Some apply conservative tolerancing to account for part nesting density, thermal gradients, or powder reuse variation—factors that can easily exceed HP’s lab conditions.

As a rule of thumb: tolerance drawings to what your supplier can actually hold, not what the printer brochure claims. HP’s data defines the upper bound of process potential; field data defines the realistic capability.

Orientation Matters

Build orientation has a direct impact on dimensional accuracy and surface consistency in MJF. HP’s process data shows that the XY plane, parts laid flat in the build, consistently delivers the highest dimensional precision. Features printed in XY benefit from uniform heat exposure, better powder packing, and minimal layer-to-layer stepping compared to the vertical (Z) direction, where cumulative layer errors and shrinkage gradients can stack up.

As a rule, place your most critical dimensions, holes, pins, and mating features in the XY plane whenever possible. HP explicitly recommends co-planarity for features that must align or assemble, such as mounting bosses, fastener holes, or datum pads. Keeping these features in the same horizontal layer reduces cross-plane variation and simplifies downstream inspection and machining.

Features oriented vertically—such as tall posts, pins, or narrow walls—are more sensitive to build height variation and heat flux. Expect slightly looser tolerances in Z, often 30–50% higher than comparable XY features, even within the same build. If vertical geometry is unavoidable, compensate with additional thickness, add fillets at the base, or plan for light post-machining.

In short, treat the XY plane as your precision plane. Design assemblies so critical datums and interfaces live there, and reserve the Z axis for non-critical features or secondary operations. This single choice can cut dimensional error and rework by more than half in most MJF builds.

One caveat however is that best surface finish comes from the side and bottom faces, often the best surface finish also comes from parts printed at an angle. There are many tradeoffs in selecting the right orientation for your parts. Sometimes improving dimensional accuracy will reduce finish quality. These competing factors should be weighed diligently and can be dialed in through small batch prototyping. This is where engineering support with this technology is key, whether from your in house team or via your service bureau.

If you want a manufacturer who is competent with understanding and maximizing the benefits of orientation, work only with HP certified service bureaus, like RapidMade, Inc.

Engineering design guidelines for MJF

1) Avoid geometry that is prone to warpage

HP flags three patterns that warp: long thin parts, abrupt cross‑section changes, and long thin curves. Rule‑of‑thumb: avoid aspect ratios over 10:1 for thin flats unless you stiffen or reshape. Prefer smooth transitions, thicker long walls, and weight reduction via hollowing or lattices to reduce heat accumulation. [3]

2) Keep sections uniform and transition gradually

• Uniformity. Injection molding literature is blunt: non‑uniform walls drive sink and warp. The fix is uniform sections with smooth transitions rather than local mass. DuPont’s design manual is still the most specific: avoid heavy sections and replace bulk with ribs sized correctly (see below). Numeric thresholds for “max allowed wall variance before warp” are not standardized for PA12/MJF. Use ribs or domes to increase stiffness without local heat sinks. [4]

• Blend transitions. Use inside blends and smooth tapers between thicknesses to reduce stress concentration and shrink mismatch. DuPont demonstrates reduced stress with radiused transitions; exact “must use” radii are geometry‑dependent. [4]

3) Use ribs correctly when you need stiffness

• Rib thickness: target 0.5–0.6 × wall thickness at the base. Thicker ribs concentrate shrink and print heat, which promotes warp. [4]

• Rib base fillet: ≈ 0.5 × rib thickness at the base to reduce stress and sink. [4]

• Rib placement: multiple shallow ribs beat a single massive rib. Published numeric spacing rules vary by OEM; explicit spacing guidance for PA12/MJF is not standardized. Treat 2–3× wall as a design heuristic, verify on your parts. Evidence mixed. [4]

4) Design bosses and insert lands to avoid local distortion

• Boss outside diameter: 2.0–2.5 × hole diameter for small bosses, then tie bosses to nearby walls with ribs rather than growing a solid mass. [4]

• Inserts: HP provides insert design callouts for MJF. Follow their hole, boss, and fillet geometry and avoid large solid bosses on thin panels. [3]

5) Control features that heat unevenly

• Large flats: avoid big unsupported planes. Use doming, shallow corrugations, or isogrid patterns to raise bending stiffness with minimal local mass. HP explicitly warns against adding ridges on large flat areas because they act as heat sinks. [3]

6) Respect minimums and clearances that MJF actually resolves

• Minimum walls: 0.3 mm short walls in XY, 0.5 mm in Z. [3]

• Connecting parts: ≥ 0.4 mm nominal interface gap (±0.2 mm per part). [3]

• Moving parts: ≥ 0.7 mm clearance in general; can go lower on very thin‑walled small parts but verify. [3]

• Small details and text: HP lists 0.1 mm minimum feature width and 6 pt minimum text size. For legibility, 1.0 mm emboss or deboss depth/height works well. [3]

7) Manage moisture and thermal history to limit drift

• Hygroscopy. PA12 absorbs moisture slowly but measurably. Arkema’s Rilsamid PA12 datasheet reports ~0.7 percent mass at 50 percent RH per ISO 62 equilibrium conditioning. Dimensional measurements should be made at 23 °C and 50 percent RH per ISO 291 after sufficient conditioning time. [5][6]

• Thermal expansion. Typical PA12 CTE is ~120–140 µm·m⁻¹·K⁻¹; measure per ISO 11359‑2 (TMA) or ASTM E831 on your parts if tight assemblies are temperature sensitive. [7][8]

• Practical tip. If your drawing tolerance is near HP’s XY band and the part sees 30 K swings, expansion alone can eat 3.6–4.2 µm per mm of length. Design your clearance stack accordingly. [1][7][8]

8) Choose build orientation and nesting with stability in mind

• Orientation. Put long, thin walls in XY where layer stepping and Z‑banding have less influence; group critical features co‑planar. [3]

• Spacing and nesting. HP does not publish a single “official” part‑to‑part gap for all cases. Many service guides suggest 2–3 mm for general builds, but that is vendor experience, not an HP standard. Treat as unknown and verify in your workflow. [3]

9) Datum strategy and post‑machining

• Datums. Establish primary datums on stiff, well‑supported surfaces away from heat‑affected transitions.

• Metrology. Condition parts to ISO 291 atmospheres before measurement. Use CMM or structured‑light scanning with a simple R&R: aim for gage %GRR ≤ 10 percent of tolerance band on critical features; this aligns with industry practice, but values are not specified in ISO. For material properties, reference the correct test standards on your drawings: CTE per ISO 11359‑2/E831, DSC per ASTM D3418, and geometric capability checks per ISO/ASTM 52902. [6][8][9][10]

Where annealing fits after good design

Annealing can relax residual stress and raise crystallinity. It is not a substitute for the rules above. What the literature supports:

• Temperatures and mechanisms. PA12’s glass transition is roughly 40–50 °C, melting about 170–180 °C. Annealing below melt allows chain relaxation and growth of crystals that reduce internal stress. Verify temperatures by DSC per ASTM D3418 on your material lot. [7][11]

• MJF‑specific evidence. A high‑temperature anneal near 173 °C has been reported to increase MJF PA12 tensile strength and modulus significantly by raising crystallinity. The study used PA12 and GF/PA12; exact dwell schedule is in the paper. Time parameters are not fully visible in open summaries. Evidence supports benefit; schedule details require full‑text access. [12]

• SLS cross‑evidence. For SLS PA12, 170 °C for 6 h increased crystallinity and improved mechanical properties. This is a reasonable starting point for coupon pilots on MJF, staying below melt. [13]

• Dimensional outcomes. CTE and shrinkage after anneal are seldom reported for MJF PA12. Treat dimensional shift direction as predictable but magnitude unknown for complex parts. Pilot on coupons bracketing your wall thickness and orientations. [7][13]

Process window to pilot on coupons

• Ramp ~1–2 °C/min to a soak between 160–170 °C.

• Soak 2–6 h.

• Cool still air in oven to avoid drafts.
  These ramp and dwell suggestions are derived from polymer annealing practice and SLS PA12 studies; you must confirm on your geometry. Evidence mixed for optimal MJF schedule. [13][11]

Post annealing is a Band-Aid

• HP capability data already consumes most of your tolerance budget. Geometry that evens out heat and shrink gets you inside those bands without extra cost or days of thermal cycling. [1][3]

• Injection molding rules generalize well to MJF because the dominant failure mode is the same: non‑uniform shrinkage in semi‑crystalline nylon. Ribs, bosses, and transitions sized per nylon design handbooks reduce local shrink mismatch, which reduces warp. [4]

• Annealing is a scalpel, not a hammer. It helps when the part is close and the geometry is sound. It does not rescue poor section design. [12][13]

Inline design checklist you can drop into drawings

• Walls: aim for uniform sections; avoid abrupt jumps. Use ribs sized 0.5–0.6 × wall, base fillet 0.5 × rib thickness. [4]

• Flats: avoid >10:1 aspect ratio without doming or corrugation. [3]

• Bosses: OD 2.0–2.5 × hole; tie into nearby walls with ribs. [4]

• Clearances: connecting parts ≥ 0.4 mm, moving parts ≥ 0.7 mm. [3]

• Text/details: min feature 0.1 mm, text ≥ 6 pt, emboss/deboss ≈ 1.0 mm. [3]

• Orientation: keep critical features co‑planar in XY. [3]

• Tolerance planning: for 0–30 mm features, plan to ±0.25 mm XY and ±0.42 mm Z at Cpk 1.33 unless your vendor provides a hardware‑specific profile. [1]

• Conditioning and measurement: condition at 23 °C, 50 percent RH; list test standards on drawings. [6][8]

References

[1] HP. “White paper: HP 3D HR PA 12 for the HP Jet Fusion 5200 Series — Dimensional Capability.” 2019. Cpk tables for XY and Z tolerances across 0–80 mm features. 3DPrint

[2] HP. “White paper: HP 3D HR PA 12 W — Dimensional Capabilities for HP Jet Fusion 5420W.” 2023. Similar methodology adapted to PA12 W on 5400W. HP Support

[3] HP. “HP Multi Jet Fusion Handbook: Design for HP MJF” rev. 2019. Minimum walls, aspect‑ratio warpage guidance, clearances, text, and design‑for‑accuracy notes. Website Files

[4] DuPont. “General Design Principles for DuPont Engineering Polymers.” Classic nylon design rules for ribs, fillets, bosses, uniform walls. See rib‑to‑wall 0.6× and fillet 0.5× rib guidance; boss OD 2–2.5× hole. distrupol.com

[5] Arkema. “Rilsamid PA12 MED grades — summary datasheet.” Moisture absorption ~0.7 percent at 50 percent RH, ISO 62. Also lists shrinkage examples by internal method. 2021. NCC – National Chemical Company

[6] ISO. “ISO 11359‑2: Plastics — Thermomechanical analysis — Determination of coefficient of linear thermal expansion and glass transition temperature.” References ISO 291 for conditioning atmospheres. 2021. ITeh Standards

[7] NETZSCH Polymers. “PA12: Polyamide 12 — property overview.” Typical CTE 120–140 µm·m⁻¹·K⁻¹, Tg and Tm ranges. 2023. For design awareness; verify your lot by ISO 11359‑2. NETZSCH Polymers

[8] Intertek Testlopedia. “Coefficient of Linear Thermal Expansion — ASTM E831, ASTM D696, ISO 11359.” Summary of accepted CTE methods for plastics. 2024. Intertek

[9] ISO/ASTM. “52902:2023 — Additive manufacturing — Test artefacts — Geometric capability assessment of AM systems.” For capability checks with standardized artefacts. ITeh Standards

[10] ASTM. “D3418‑21 — Transition Temperatures and Enthalpies of Fusion and Crystallization of Polymers by DSC.” Use to verify PA12 Tg, Tm, and crystallinity changes pre/post anneal. Antpedia

[11] De Gruyter. “Crystal transition and thermal behavior of Nylon 12.” E‑Polymers, 2020. Explains α/γ forms and thermal transitions that underpin annealing response. De Gruyter Brill

[12] Additive Manufacturing (Elsevier). “Enhancing the mechanical strength of MJF‑printed PA12 via high‑temperature annealing.” Reports ~173 °C anneal raising UTS and modulus through increased crystallinity for MJF PA12 and GF/PA12. Time not visible in open abstracts. 2021. ScienceDirect+1

[13] Polymers (MDPI). “Effects of Laser Power and Hatch Orientation on Final Properties of SLS PA12.” Demonstrates 170 °C for 6 h anneal increasing crystallinity and mechanical properties; mechanism generalizes to semi‑crystalline PA12. 2022. MDPI+1