Comprehensive Analysis of As-Printed Tolerances in Multi Jet Fusion (MJF)

1. Introduction: The Tolerance Misconception in Additive Manufacturing

The industrialization of additive manufacturing (AM), specifically through HP’s Multi Jet Fusion (MJF) technology, has fundamentally shifted the production landscape from rapid prototyping to functional, short-run manufacturing, with multi jet fusion 3D printing now widely used for end-use production parts. However, this transition has exposed a critical skills gap in the engineering workforce: the misunderstanding of “as-printed” tolerances. Unlike the subtractive logic of CNC machining, where precision is a function of tool stiffness and kinematic positioning, or the volumetric logic of injection molding, where pressure and mold steel define dimensions, MJF is a thermal process governed by the stochastic behavior of polymer powder fusion.

The prevailing misconception among mechanical engineers is that the digital CAD file is a deterministic instruction set that the printer executes with absolute fidelity during an MJF print. This “what you see is what you get” (WYSIWYG) mindset leads to the erroneous assumption that a machine specification—often cited as ±0.2 mm or ±0.3% of the nominal dimension—is a universal guarantee applicable to every feature on a complex geometry. This assumption is the primary source of tolerance failures in MJF production. Accuracy in MJF is a dynamic variable, heavily influenced by the thermal history of the specific build, the orientation of the part within the powder bed, and the mass distribution of the geometry.

This white paper provides a comprehensive technical deconstruction of dimensional accuracy in MJF 3D printing. It argues that “tolerance” in the context of Powder Bed Fusion (PBF) is not a static machine attribute but a complex function of thermodynamics, fluid dynamics, and polymer crystallization kinetics. By analyzing the dissociation between XY and Z resolution, the physics of thermal bleed, and the crystallization behavior of Polyamide 12 (PA12), this report elucidates why “as-printed” tolerances are inherently variable and why standard machining tolerances (ISO 2768) are fundamentally incompatible with the process. Furthermore, it proposes robust strategies for geometric dimensioning and tolerancing (GD&T) that align with the thermal realities of high-energy sintering processes. In production environments, MJF 3D printing is increasingly treated not as prototyping, but as a repeatable manufacturing process where dimensional predictability directly affects assembly performance.

2. The Physics of Tolerance Generation in MJF

To truly understand why tolerances vary, engineers must look beyond the datasheet and understand the fundamental physics of the MJF printing process. It is a process governed by the interaction of chemical agents and infrared energy on a voxel level, subject to the laws of thermodynamics. Unlike filament-based processes, MJF printing relies on volumetric energy distribution, which makes thermal management the primary driver of dimensional accuracy.

2.1 The Agent Interaction Mechanism and Voxel Definition

Multi Jet Fusion distinguishes itself from Selective Laser Sintering (SLS) by replacing the single-point laser source with a wide-area inkjet array and a planar infrared energy source. The process relies on the precise deposition of two proprietary chemical agents: the Fusing Agent (FA) and the Detailing Agent (DA).

The Fusing Agent is typically a carbon-black-based fluid that dramatically increases the optical absorption of the polymer powder in the infrared spectrum. When the print carriage’s energy lamps traverse the bed, the voxels (volumetric pixels) coated with FA absorb significantly more energy than the surrounding virgin powder, raising their temperature above the melting point of the polymer (approx. 187°C for PA12).1 This selective absorption defines the core geometry of the part.

However, heat transfer in a powder bed during multi jet fusion 3D printing is not digital; it is diffusive. Thermal energy naturally conducts from the molten part into the surrounding loose powder. Without intervention, this “thermal bleed” would cause the part to grow indefinitely, resulting in poor dimensional accuracy and fused surface artifacts. This is where the Detailing Agent becomes critical. The DA is a fluid deposited at the exact boundary of the part—the interface between the molten core and the loose powder. Its primary function is evaporative cooling. By absorbing thermal energy to undergo a phase change (evaporation), the DA creates a steep thermal gradient at the part boundary, effectively inhibiting fusion and sharpening the edge definition.3

The interplay between these agents defines the “digital” dimension of the part. The printheads deliver these agents with a resolution of 1200 dpi, translating to a droplet placement precision of approximately 21 microns.6 However, the physical resolution of the part is not 21 microns. It is limited by the particle size of the powder (typically 50–60 microns) and the thermodynamic “fuzziness” of the thermal bleed zone.8 The final dimension of a feature is determined by the equilibrium reached between the energy input of the FA, the cooling power of the DA, and the thermal diffusivity of the surrounding powder cake.

2.2 Thermal Bleed and the Boundary Layer Phenomenon

“Thermal bleed” is the defining mechanism of dimensional error in PBF processes. It creates a transition zone—a “fuzzy” boundary where the polymer particles are partially sintered or “caked” onto the fully dense part. The thickness of this boundary layer is not constant; it is a function of the local thermal mass. This is one of the reasons why two geometrically similar features can behave differently within the same MJF 3D print, depending on their surrounding thermal mass.

A large, solid block retains heat for a significantly longer period than a thin wall or a fine strut. This retained heat continues to conduct outward into the surrounding powder even after the energy source has passed. Consequently, a massive feature will drive the thermal bleed deeper into the powder bed, causing the feature to “grow” beyond its nominal CAD dimensions. Conversely, a thin feature cools rapidly, halting the thermal bleed early. This phenomenon explains why a 50 mm solid cube will often measure larger than a 50 mm hollow cube printed in the same build; the solid cube has a higher thermal inertia that overwhelms the detailing agent’s boundary control.9

Manufacturers attempt to compensate for this through “scaling factors” and “erosion/dilation” algorithms in the slicing software, but these are often global corrections applied to the entire build volume. They cannot account for the localized thermal complexity of every unique geometry, leading to the variance that engineers observe between different features on the same part.11

2.3 Crystallization Kinetics and Volumetric Shrinkage

While thermal bleed defines the outer boundary during the heating phase, the final dimensions are ultimately determined by the shrinkage that occurs during the cooling phase. This is driven by the crystallization kinetics of the semi-crystalline polymer.

PA12, the standard material for MJF, undergoes a significant volumetric reduction as it transitions from a molten amorphous state to a solid semi-crystalline state. This shrinkage is typically in the range of 2–3%.13 The crystallization process is governed by the cooling rate:

• Slow Cooling: Promotes the formation of large, perfect crystals and higher density. This results in greater volumetric shrinkage but improved mechanical properties.
• Fast Cooling: Suppresses crystallization, leading to an amorphous-rich structure with less shrinkage but higher internal stresses and potential for warping.15

The critical misunderstanding is assuming this shrinkage is uniform. It is highly anisotropic and geometry-dependent. The core of a thick part cools much more slowly than the skin, creating a differential shrinkage gradient. The skin solidifies first, forming a rigid shell. As the core subsequently cools and shrinks, it pulls on the already-solid skin, creating compressive stresses in the skin and tensile stresses in the core. This manifests as “sink marks” on the surface (depressions in flat areas) or as macroscopic warping of the part.18

Because the crystallization rate is temperature-dependent, the position of the part in the build chamber matters. Parts deep in the powder bed (the bottom of the bucket) stay hot for the entire duration of the print job (often 12+ hours), resulting in a different crystalline structure and shrinkage profile than parts at the top of the bed, which cool sooner.20 This variability means that “tolerance” is not just a function of geometry, but of specific build placement and thermal history.

3. The Anisotropy of Accuracy: Why XY ≠ Z

A persistent myth in the AM industry is that powder bed fusion technologies like multi jet fusion printing are isotropic. While MJF parts exhibit impressive mechanical isotropy compared to Fused Deposition Modeling (FDM) or SLS (with Z-strength often reaching 90-95% of XY-strength) 21, they are inherently dimensionally anisotropic. The mechanism creating dimensions in the XY plane is fundamentally different from the mechanism creating dimensions in the Z axis, leading to distinct tolerance capabilities.

3.1 XY Dimensions: The Digital Plane

In the XY plane (the horizontal plane of the print bed), dimensions in an MJF print are defined by the digital firing of the inkjet printheads.

• High Resolution: The 1200 dpi resolution allows for extremely precise placement of the fusing and detailing agents. The digital definition of the part boundary is accurate to within microns.7
• Constraint: The limitation in XY is not the printhead, but the fluid dynamics of the agents wicking into the powder and the subsequent thermal expansion/contraction of the entire powder cake.
• Accuracy: Typically, XY accuracy is high because the printhead gantry position is strictly encoded. The primary source of error is the scaling factor mismatch—if the software predicts 2% shrinkage but the part shrinks 2.2%, the XY dimensions will be off.23

3.2 Z Dimensions: The Mechanical Plane

In the Z axis (vertical height), dimensions are defined by a mechanical process: the recoater blade and the piston movement.

• Layer Discretization: MJF typically prints in discrete layers of 80 microns (0.08 mm).7 This creates a quantization error; a vertical feature can theoretically only be a multiple of the layer height. A 10.05 mm vertical wall is impossible to print perfectly with 0.08 mm layers; it will be approximated to the nearest layer count.
• Z-Growth and Thermal Bleed: As new layers of fusing agent and energy are applied on top of the build, heat conducts downward into the previously printed layers. If the thermal control is not perfect, this energy can re-melt the top surface of the previous layer or cause the fusing agent to penetrate deeper than 80 microns. This results in “Z-growth,” where parts measure taller than designed.24
• Mechanical Compression: Conversely, the weight of the powder in a tall build can compress the lower layers while they are still in a semi-molten state, a phenomenon known as “Z-squish.” This creates a density gradient and dimensional deviation that varies with the Z-height of the part within the build.20

3.3 The Stair-Stepping Effect

The discretization of the Z-axis leads to the “stair-stepping” effect, which is a critical driver of tolerance failure on angled surfaces. While XY features are smooth curves defined by agent droplets, Z features are stacks of plates.

• Angle Sensitivity: Surfaces angled less than 15-20° from the horizontal plane exhibit visible stepping contours.25 This is not just an aesthetic issue; it is a dimensional one.
• Tolerance Implication: If an engineer specifies a tight tolerance (e.g., ±0.1 mm) on a sloped surface or a chamfer, the stair-stepping roughness (which can exceed 50-80 microns in peak-to-valley height) effectively consumes the entire tolerance band. Standard calipers measuring over the peaks of the steps will read a different dimension than a CMM probe measuring the average surface.
• Mitigation: For high-precision interfaces, engineers must orient the part so critical features lie in the XY plane or orthogonal to it, avoiding shallow angles that induce stepping errors.27

3.4 Comparative Tolerance Bands

The divergence in physics leads to a divergence in achievable precision. Research and manufacturer data indicate a distinct difference in tolerance capabilities between the planes:

Dimension Plane	Typical Tolerance	Dominant Driver of Variance
XY (Horizontal)	±0.2 mm or ±0.2%	Thermal shrinkage, agent wicking, calibration scaling
Z (Vertical)	±0.3 mm or ±0.4%	Layer thickness discretization, recoater variance, thermal Z-growth

Data aggregated from.23

This table illustrates why orientation is the single most important decision in MJF build preparation. Critical fits, such as bearing bores or seal glands, should almost always be oriented in the Z-axis (so the circular feature is drawn in the high-resolution XY plane) to maximize circularity and dimensional control.27

Understanding this distinction is essential when designing precision interfaces for MJF printing, where orientation directly determines achievable tolerances.

4. Machine Specifications vs. Real-World Part Specifications

A critical source of friction between design engineers and service bureaus is the difference between what the machine is capable of (Machine Spec) and what a specific geometry achieves (Part Spec).

4.1 The “Calibration Artifact” Fallacy

HP and service bureaus typically quote tolerances based on “calibration artifacts”—standardized geometries printed under ideal conditions. These are often specific test parts (like the torture test pieces or dimensional accuracy bars) described in white papers.11

• Ideal Conditions: These artifacts are printed in the center of the build volume, with balanced thermal load, specific orientation, and often in a build with low packing density to prevent thermal interference from neighbors.
• The Disconnect: Real-world production builds are packed densely (often 12-15% packing density) to minimize cost. Real parts have varying cross-sections, entrapped heat zones, and complex topologies. A thin-walled manifold with a heavy mounting flange will not behave thermodynamically like a standardized calibration bar. It will warp, shrink differentially, and exhibit thermal bleed that the artifact does not. Therefore, the machine spec is a “best case” scenario, not a guarantee for every feature.

4.2 The “±0.3% with a lower limit of ±0.2 mm” Rule

The industry-standard tolerance for MJF PA12 is frequently cited as ±0.3% of the nominal dimension, with a minimum variance floor of ±0.2 mm (or ±0.3 mm depending on the provider).1 Understanding the two components of this rule is vital.

• The Lower Limit (The Floor): The ±0.2 mm floor is non-negotiable. For small features (e.g., a 10 mm boss), applying the ±0.3% rule would imply a tolerance of ±0.03 mm (30 microns). This is physically impossible for a powder-based process where the particle size itself is approx. 60 microns.8 The particle size sets the “resolution floor.” You cannot hold a tolerance tighter than the grain size of the material you are using. Furthermore, surface roughness (Ra ~10 µm) introduces measurement uncertainty that makes sub-50-micron tolerances unverifiable with standard tools.
• The Percentage (The Scaling): As parts get larger (e.g., >100 mm), thermal shrinkage becomes the dominant error source. A 300 mm part shrinking by 2% involves a movement of 6 mm. Even a small error in predicting that shrinkage (e.g., predicting 2.0% when it actually shrinks 2.1%) results in a 0.3 mm deviation. Thus, larger parts are governed by the percentage rule, which accounts for the accumulation of shrinkage error over distance.1

4.3 Process Capability (Cpk) and Statistical Control

For mass production, engineers look at Cpk (Process Capability Index). MJF has demonstrated Cpk values >1.33 for specific dimensions when the process is tightly controlled.11 However, achieving high Cpk in MJF requires a frozen process:

1. Fixed Orientation: The part must be printed in the exact same orientation every time. Rotating a part 90 degrees changes the layer slicing and the thermal profile, altering the dimensions.
2. Fixed Position: The part should ideally be restricted to specific regions of the build volume (e.g., avoiding the extreme corners where thermal gradients are highest).
3. Dimensional Tuning: The CAD model must be adjusted (biased) based on measurement data from initial runs to center the process mean.

Without these strict controls—which are rarely applied in “on-demand” manufacturing—the “as-printed” variance is much higher, and Cpk values will naturally be lower.

5. How Geometry Drives Tolerance Variance

Designing for MJF 3D print accuracy, therefore, means designing for thermal balance, not just geometric intent. In MJF, geometry is not just shape; it is thermal information. The volume, topology, and distribution of material determine how much heat is absorbed, how it conducts, and how slowly it releases. This thermal history directly dictates the final dimensions of the part.

5.1 Thermal Mass and Heat Retention

Thick sections of a part absorb more IR energy and retain it longer than thin sections. This leads to two distinct and opposing tolerance issues:

1. Oversize Growth (Thermal Bleed): Massive parts act as heat reservoirs. They stay hot long after the print pass, causing the detailing agent barrier to break down or the surrounding powder to sinter onto the surface. Consequently, a 50mm solid block will typically measure larger than a 50mm hollow block because the solid one bleeds heat outward.32
2. Sink Marks: Just like in injection molding, thick sections in MJF can suffer from sink marks. As the large internal core crystallizes and shrinks, it pulls the outer skin inward. This results in dished surfaces and dimensional undersizing on the faces of thick blocks. This is particularly prevalent if the “Fast Cool” profile is used, exacerbating the skin-core temperature differential. 18

Design Implication: Engineers must hollow out thick sections (shelling) to maintain a uniform wall thickness (typically 2–3 mm). This normalizes the thermal mass, ensures predictable shrinkage, and prevents both thermal bleed and sink marks. 33. This is one of the reasons why two geometrically similar features can behave differently within the same MJF 3D print, depending on their surrounding thermal mass.

5.2 Aspect Ratio and Warpage

Long, thin parts (high aspect ratio) are susceptible to non-uniform cooling gradients. If a part is 200 mm long but only 2 mm thick, the differential cooling between the bottom layers (which stay hot longer) and the top layers (which cool faster) creates internal residual stress.

• Curling: This stress releases as mechanical deformation, typically warping or curling the corners of the part upwards.25
• Tolerance Impact: While the linear dimension along the neutral axis might be correct, the bowing causes the effective end-to-end measurement to fail flatness, straightness, and overall length tolerances. A warped part is an out-of-tolerance part.

Mitigation: High aspect ratio parts should be oriented at an angle in the build to reduce the cross-section printed per layer. Additionally, adding stiffening ribs or flanges can increase the moment of inertia and resist the thermal stresses, inducing warpage.19

5.3 Feature Specifics: The Hole vs. Pin Asymmetry

A classic misunderstanding in PBF processes like multi jet fusion printing is that holes and pins behave symmetrically (i.e., if the process adds 0.1 mm, holes get smaller and pins get bigger). While partially true, the magnitude is often different due to the geometry of heat dissipation.

• Holes (Internal Diameters): Holes almost always print undersized. As the molten polymer ring surrounding the hole cools, it shrinks inward toward its own center of mass (hoop shrinkage). Additionally, thermal bleed from the surrounding wall mass can effectively “grow” the wall into the negative space of the hole. A 5.0 mm hole in CAD typically prints as 4.7 or 4.8 mm. 35
• Pins (External Diameters): Pins tend to print oversized or near nominal. The thermal bleed expands the pin outward into the loose powder. However, because a pin is surrounded by loose powder (insulator), it may cool slightly faster than a hole surrounded by solid part mass, making the expansion less aggressive than the hole’s contraction.

Implication: “Windage” or CAD compensation is almost always required for holes. A standard rule of thumb is to increase hole diameters by 0.15–0.30 mm in CAD to achieve the nominal diameter in the physical part.28

6. Detailed Analysis of “As-Printed” Anomalies

Beyond simple linear tolerances, MJF exhibits specific geometric anomalies driven by the process physics that engineers must anticipate.

6.1 The “Elephant Skin” and Surface Artifacts

If the energy input is too low, or if the powder quality is degraded (high refresh rates with aged powder), parts can exhibit “elephant skin”—a rough, wrinkled surface texture. This destroys surface profile tolerances. Conversely, too much energy leads to extreme thermal bleed and loss of fine detail.38 The “Balanced” print mode attempts to mitigate this, but geometry dictates the local thermal load. A part with a sudden transition from thin to thick geometry may show surface defects at the transition line due to the abrupt change in thermal absorption.

6.2 Variable Density and Lattice Structures

Lattice structures challenge the tolerance limits of MJF.

• Powder Removal Limits: In dense lattices, the thermal bleed can fuse the loose powder inside the lattice cells, making it impossible to clean. This effectively turns a lattice into a solid block, completely violating the dimensional intent. Research suggests that for high-density lattices (relative density > 0.4), powder removal is limited to a depth of only 1.5–2 layers from the surface.39
• Thin Struts: Lattice struts below 0.5 mm are fragile and subject to warping. The “as-printed” dimension of a 0.5 mm strut is highly variable due to the dominance of the surface roughness (Ra values of 10–15 microns) relative to the feature size. A strut may measure 0.45 mm at a valley and 0.55 mm at a peak.14

6.3 Position in the Build Chamber

Research indicates that dimensional accuracy is spatially dependent within the build volume.

• Center of Build: Parts in the center of the build volume typically stay hottest the longest. They are most prone to thermal bleed (oversizing) but often have the most uniform crystallization (best mechanical properties).20
• Perimeter of Build: Parts near the walls of the build chamber cool faster due to conduction through the metal walls. They are more prone to warping and may exhibit slightly different shrinkage rates than centered parts, leading to dimensional variation. 20
• Z-Height Variation: Parts at the very bottom of the build (first layers printed) experience a different thermal history than parts at the top. The bottom layers are kept at elevated temperatures for the entire duration of the print job (often 12–15 hours), while top layers might only be hot for a few hours. This can lead to a gradient in crystallinity and shrinkage from the bottom to the top of the bucket.41

7. The Incompatibility of ISO 2768 and the Need for New Standards

A major friction point in MJF engineering is the habitual application of ISO 2768-mk tolerances. ISO 2768 is a standard designed for machining (subtractive manufacturing), where precision is a function of machine tool stiffness and position control. It is fundamentally unsuited for the thermal realities of PBF.

7.1 Why ISO 2768 Fails for MJF

• Assumption of Stiffness: ISO 2768 assumes the workpiece is rigid during manufacturing. MJF parts are semi-molten, rubbery, and thermally dynamic during the process.
• Assumption of Tool Path: Machining tolerances are tighter for small features and looser for large features, but the progression is linear and based on tool deflection. MJF tolerances are dominated by thermal shrinkage (percentage-based), which diverges from ISO 2768 classes at larger sizes. For example, ISO 2768-m might allow ±0.8 mm for a 400 mm part, but MJF’s ±0.3% would allow ±1.2 mm. Holding a 400 mm printed plastic part to ±0.8 mm is extremely difficult without post-machining.42
• Flatness Requirements: ISO 2768 imposes strict flatness requirements that are difficult to hold in PBF due to the natural tendency of nylon to warp during cooling. A printed flat plate will almost always exhibit some degree of bow that exceeds ISO 2768-m flatness specs.

Applying “ISO 2768-m” (medium) to an MJF drawing often sets the manufacturer up for failure, or necessitates expensive post-machining (CNC) of the printed part to meet the spec.

7.2 The Better Standards: ISO 20457 and DIN 16742

Engineers should pivot to standards designed for molding, as MJF physics (melting, cooling, shrinking) closely mirror injection molding.

• ISO 20457 (formerly DIN 16742): This standard defines tolerances for plastic molded parts. It categorizes tolerances based on the material’s shrinkage characteristics (TG grades). PA12 is typically categorized similarly to a TG6 material.
• Applicability: These standards account for the fact that a plastic part moves, shrinks, and warps. They provide realistic tolerance bands for “non-rigid” parts, acknowledging that long dimensions in plastic cannot be held to metal-machining precision. 44
• ISO/ASTM 52910: This standard specifically addresses design guidelines for additive manufacturing, advocating for “function-based tolerancing” rather than blanket default tolerances. It encourages engineers to tolerate only what matters for the assembly, rather than applying a blanket block tolerance that increases cost and reject rates.47

8. Compensation Strategies: Engineering the Delta

Since variance is inevitable in a thermal process, engineering for MJF requires active compensation strategies. Accuracy is not found; it is engineered. Effective dimensional control in MJF printing depends more on proactive design compensation than on post-build correction.

8.1 Global Scaling Factors

The printer software allows for global scaling factors (e.g., X=1.025, Y=1.025, Z=1.020). Service bureaus dial these in based on their specific machine calibration and material lot. However, this is a blunt instrument. It corrects the average shrinkage of the average part in the build. It cannot correct for the specific thermal mass of a unique geometry. If a user relies solely on the service bureau’s global scaling, features with atypical thermal mass (very thick or very thin) will likely be out of tolerance.12

8.2 Local Geometric Compensation (CAD Offsets)

Engineers must apply “Windage” or local offsets to critical features in the CAD model. This is the manual addition or subtraction of material to counteract known thermal behaviors. This “pre-distortion” of the CAD model is the most effective way to hit tight tolerances.

Feature Type	Thermal Behavior	Recommended CAD Offset (Windage)
Vertical Hole (Z-axis)	Shrinks heavily / Undersized	Increase diameter by 0.15 – 0.30 mm
Horizontal Hole (XY)	Shrinks / Ovalizes	Increase diameter by 0.20 – 0.40 mm
Post / Pin	Expands / Oversized	Decrease diameter by 0.05 – 0.10 mm
Mating Faces	Interference Risk due to roughness	Add 0.2 – 0.4 mm gap between faces
Snap Fits	Tightening due to shrinkage	Design clearances of 0.3 mm minimum
Threads (Internal)	Shrinkage / Powder Residue	Model at max tolerance (e.g., H9), use 0.5mm offset

Data synthesized from.28

8.3 Post-Processing Effects

Engineers must also account for the dimensional changes induced by post-processing of MJF 3D printed parts.

• Bead Blasting: Standard parts are bead blasted to remove powder. This is an abrasive process that typically removes 10–20 microns of surface material. While small, this reduction can affect tight press fits. Engineers should clarify if the tolerance spec applies to the “as-printed” or “as-finished” state.
• Vapor Smoothing: This process uses chemical solvents to melt the surface, sealing porosity. It typically reduces dimensional volume by smoothing peaks, potentially reducing wall thickness by 0.01–0.03 mm depending on exposure time. It significantly improves airtightness and surface smoothness (Ra ~2 µm) but rounds sharp edges.8
• Tumbling: Extended vibratory tumbling for smoothness acts as a radius operation, rounding off sharp corners and potentially affecting the fit of square mating parts.

9. Material Variations: PA12 vs. PA11 vs. Glass Beads

The base material selection fundamentally alters the tolerance landscape, as each material has unique thermal and crystallization properties.

9.1 PA12 (Polyamide 12)

The industry standard. PA12 offers the best balance of properties and the most established process parameters.

• Tolerance: Standard ±0.3% / ±0.2 mm.
• Behavior: Predictable shrinkage. Most service bureaus have highly tuned scaling factors for PA12, making it the safest choice for general accuracy.1

9.2 PA11 (Polyamide 11)

Derived from renewable castor oil, PA11 is more ductile and has higher impact strength than PA12.

• Tolerance: Generally slightly looser than PA12 due to higher elasticity and different crystallization rates. It is more difficult to control thermally.
• Warpage: PA11 is notably more prone to warping on large flat surfaces than PA12. It requires more aggressive ribbing and support in the design phase.
• Behavior: Requires different scaling factors. Engineers switching from PA12 to PA11 must re-evaluate their fits and clearances.48

9.3 PA12 Glass Beads (GB)

PA12 filled with 40% glass spheres. This is a composite material designed for stiffness.

• Stiffness: Much higher rigidity (Young’s Modulus) than unfilled nylon.
• Shrinkage: Significantly lower shrinkage and warpage. The glass beads act as a dimensional stabilizer, inhibiting the movement of polymer chains during cooling.
• Tolerance: This is the material of choice for large, flat parts, housings, or fixtures where dimensional stability is paramount. It holds flatter tolerances than unfilled PA12 and resists warping effectively. However, it is more brittle and has lower impact strength.50

10. Conclusion and Recommendations

The successful deployment of MJF parts in engineering applications requires a paradigm shift. Engineers must move away from the expectation of deterministic, “machine-like” precision and embrace the probabilistic, thermodynamic nature of the process. “As-printed” tolerances are not defects; they are the physical signatures of the material’s phase change.

Key Recommendations for Engineering Success:

1. Abandon the CNC Mindset: Do not apply ISO 2768 machining tolerances to MJF prints. Adopt ISO 20457 or functional tolerancing strategies that account for polymer flexibility and shrinkage.
2. Design for Thermal History: Recognize that thick sections shrink more and retain heat. Aggressively shell parts to a uniform wall thickness (2–3 mm) to stabilize tolerances and prevent sink marks.
3. Orient for Accuracy: Place critical dimensions in the XY plane. Avoid printing critical circular fits in the Z-direction if possible, or accept larger tolerances.
4. Compensate Locally: Do not rely on the machine’s global scaling to fix specific features. Manually offset holes, pins, and mating surfaces in the CAD model (Windage) to account for thermal bleed and shrinkage.
5. Contextualize Specifications: A tolerance of ±0.2 mm is a guideline, not a guarantee. It applies to well-behaved geometries. Complex parts require iterative prototyping to dial in the dimensions.

By understanding the disparate mechanics of XY and Z formation, the impact of thermal mass on shrinkage, and the necessity of CAD-level compensation, engineers can bridge the gap between digital intent and physical reality, leveraging MJF as a reliable, high-performance production technology.

Works cited

1. Streamlining machinery and production lines with HP Multi Jet Fusion 3D printing, accessed December 3, 2025, https://h20195.www2.hp.com/v2/GetDocument.aspx?docname=4AA7-3038ENW
2. HP Multi Jet Fusion technology – Cimquest Inc., accessed December 3, 2025, https://cimquest-inc.com/wp-content/uploads/2018/04/HP-MJF-Technology-Technical-White-Paper.pdf
3. HP Multi Jet Fusion Handbook – CGS plus, accessed December 3, 2025, https://cgsplus.si/wp-content/uploads/2021/11/CGS-HP-Multi-Jet-Fusion-Handbook.pdf
4. MJF Multi Jet Fusion: How does it work? | Dassault Systèmes®, accessed December 3, 2025, https://www.3ds.com/make/service/3d-printing-service/mjf-multi-jet-fusion
5. Multi Jet Fusion (MJF) of polymeric components:A review of process, properties and opportunities | Request PDF – ResearchGate, accessed December 3, 2025, https://www.researchgate.net/publication/382751473_Multi_Jet_Fusion_MJF_of_polymeric_componentsA_review_of_process_properties_and_opportunities
6. What is the difference between Selective Laser Sintering (SLS) and Multi Jet Fusion (MJF) 3D printing? | Protolabs Network, accessed December 3, 2025, https://www.hubs.com/knowledge-base/hp-mjf-vs-sls-3d-printing-technology-comparison/
7. The Manufacturer’s Guide to MJF from Prototypes to End-Use Parts | All3DP Pro, accessed December 3, 2025, https://all3dp.com/1/multi-jet-fusion-mjf-3d-printing-simply-explained/
8. Multi Jet Fusion – 3Faktur, accessed December 3, 2025, https://3faktur.com/en/technology/multi-jet-fusion/
9. Dialing in an HP Jet Fusion (MJF) 3D Printer – RapidMade, accessed December 3, 2025, https://rapidmade.com/3d-printing/dialing-in-an-hp-jet-fusion-mjf-3d-printer/
10. Example of thermal bleeding phenomenon | Download Scientific Diagram – ResearchGate, accessed December 3, 2025, https://www.researchgate.net/figure/Example-of-thermal-bleeding-phenomenon_fig18_352554515
11. White paper – HP 3D HRPA 12, enabled by Evonik, and PA 12 S, enabled by Arkema for the HP Jet Fusion 5200 Series 3D Printing Solution, accessed December 3, 2025, https://h20195.www2.hp.com/V2/getpdf.aspx/4AA7-7138ENW.pdf
12. HP SmartStream 3D Build Manager User Guide, accessed December 3, 2025, https://h10032.www1.hp.com/ctg/Manual/c06067864.pdf
13. 3D Printing Dimensional Accuracy: Definition & Tips to Achieve It – Raise3D, accessed December 3, 2025, https://www.raise3d.com/blog/3d-printing-dimensional-accuracy/
14. Tolerances & Accuracy in 3D Printing Technologies – Xometry Pro, accessed December 3, 2025, https://xometry.pro/en/articles/3d-printing-tolerances/
15. Non-isothermal crystallization behaviour of polyamide 12 analogous to multi-jet fusion additive manufacturing process | Request PDF – ResearchGate, accessed December 3, 2025, https://www.researchgate.net/publication/355273471_Non-isothermal_crystallization_behaviour_of_polyamide_12_analogous_to_multi-jet_fusion_additive_manufacturing_process
16. Crystallization Kinetics of Polyamide 12 during Selective Laser Sintering – PMC – NIH, accessed December 3, 2025, https://pmc.ncbi.nlm.nih.gov/articles/PMC6415035/
17. Influence of Crystallization Kinetics and Flow Behavior on Structural Inhomogeneities in 3D-Printed Parts Made from Semi-Crystalline Polymers – NIH, accessed December 3, 2025, https://pmc.ncbi.nlm.nih.gov/articles/PMC11008537/
18. Avoiding Sink Marks in Injection Molding: Tips for Flawless Parts – Aprios, accessed December 3, 2025, https://www.aprios.com/insights/avoiding-sink-marks-in-injection-molding-tips-for-flawless-parts
19. Multi Jet Fusion printing tips and tricks – Cimquest Inc., accessed December 3, 2025, https://cimquest-inc.com/resource-center/HP/Tips&Tricks/HP%20MJF%20General%20Part%20quality%20tips%20and%20tricks%20-3D%20Printer%20Series%20.pdf
20. Accuracy and Powder Removal Limits in Multi Jet Fusion 3D Printing – PMC, accessed December 3, 2025, https://pmc.ncbi.nlm.nih.gov/articles/PMC12567854/
21. MJF vs SLS: 3D Printing Technologies Compared | Xometry Pro, accessed December 3, 2025, https://xometry.pro/en/articles/3d-printing-mjf-vs-sls/
22. Producing manufacturing aids with HP Multi Jet Fusion 3D Printing, accessed December 3, 2025, https://h20195.www2.hp.com/v2/GetDocument.aspx?docname=4AA7-2326ENW
23. Multi Jet Fusion (MJF) 3D Printing: A Comprehensive Guide for Engineers – Wevolver, accessed December 3, 2025, https://www.wevolver.com/article/multi-jet-fusion-mjf-3d-printing-a-comprehensive-guide-for-engineers
24. Measuring SLS Dimensional Accuracy and Print Repeatability – Formlabs, accessed December 3, 2025, https://formlabs.com/white-papers/measuring-sls-dimensional-accuracy-and-print-repeatability/
25. The Ultimate HP Jet Fusion (MJF) Design Guidelines – RapidMade, accessed December 3, 2025, https://rapidmade.com/3d-printing/the-ultimate-hp-jet-fusion-mjf-design-guidelines/
26. Study on Geometry, Dimensional Accuracy and Structure of Parts Produced by Multi Jet Fusion – NIH, accessed December 3, 2025, https://pmc.ncbi.nlm.nih.gov/articles/PMC8398662/
27. Multi Jet Fusion printing tips and tricks for the 500 3D printer series – Cimquest Inc., accessed December 3, 2025, https://cimquest-inc.com/resource-center/HP/Tips&Tricks/HP%20MJF%20JF500%20Series%20Part%20quality%20tips%20tricks.pdf
28. MJF Design Guidelines – Multi-Jet Fusion – Forge Labs, accessed December 3, 2025, https://forgelabs.com/design-guides/mjf
29. MJF 3D Printing, Multi Jet Fusion – Strong & Detailed Parts – Sculpteo, accessed December 3, 2025, https://www.sculpteo.com/en/materials/jet-fusion-material/
30. Dimensional tolerancing – Endeavor 3D, accessed December 3, 2025, https://endeavor3d.com/wp-content/uploads/2024/12/Dimensional-Tolerancing-Design-for-HP-MJF-Union-Joints-Design.pdf
31. What level of dimensional accuracy can MJF parts achieve? – CNC machining, accessed December 3, 2025, https://www.newaymachining.com/pt/services/3d-printing/faq-what-level-of-dimensional-accuracy-can-mjf-parts-achieve
32. Designing Fixtures and Manufacturing Aids Using MJF PA12 – RapidMade, accessed December 3, 2025, https://rapidmade.com/designing-fixtures-and-manufacturing-aids-using-mjf-pa12/
33. Designing a part for Multi Jet Fusion – Protolabs, accessed December 3, 2025, https://www.protolabs.com/en-gb/resources/blog/designing-a-part-for-multi-jet-fusion/
34. How MJF Stacks Up: A Comparison of Dimensional Accuracy in 3D Printing – Phasio, accessed December 3, 2025, https://www.phas.io/post/mjf-dimensional-accuracy
35. How to Design for Accuracy with HP Multi Jet Fusion – Tempus 3D, accessed December 3, 2025, https://www.tempus3d.com/post/how-to-design-for-accuracy-with-hp-multi-jet-fusion
36. How to adjust part dimentions based on 3d print material tolerances : r/3Dprinting – Reddit, accessed December 3, 2025, https://www.reddit.com/r/3Dprinting/comments/1i3rw1k/how_to_adjust_part_dimentions_based_on_3d_print/
37. MJF 3D Printing Design Tips: 9 Best Practices | Xometry Pro, accessed December 3, 2025, https://xometry.pro/en/articles/mjf-design-guidelines/
38. New HP 3D Jet Fusion 5210 3D Printing Solution – RE3DTECH, accessed December 3, 2025, https://re3dtech.com/uncategorized/new-hp-3d-jet-fusion-5210-3d-printing-solution-unboxing/
39. Accuracy and Powder Removal Limits in Multi Jet Fusion 3D Printing – ResearchGate, accessed December 3, 2025, https://www.researchgate.net/publication/396754877_Accuracy_and_Powder_Removal_Limits_in_Multi_Jet_Fusion_3D_Printing
40. Accuracy and Powder Removal Limits in Multi Jet Fusion 3D Printing – MDPI, accessed December 3, 2025, https://www.mdpi.com/2073-4360/17/20/2804
41. What is dimensional accuracy in 3D printing and how do you achieve it? – Protolabs Network, accessed December 3, 2025, https://www.hubs.com/knowledge-base/dimensional-accuracy-3d-printed-parts/
42. The General CNC Machining Tolerance: ISO 2768-mk – SogaWorks, accessed December 3, 2025, https://www.sogaworks.com/blogs/mastering-iso-2768-mk-tolerances/
43. Best practices for designing tolerances when parts should fit together? : r/3Dprinting – Reddit, accessed December 3, 2025, https://www.reddit.com/r/3Dprinting/comments/mi45wh/best_practices_for_designing_tolerances_when/
44. Plastic Molding Tolerances | Custom Injection Molded Parts by Jiangzhi, accessed December 3, 2025, https://www.swcpu.com/blog/plastic-molding-tolerances-guide/
45. A Guide to Injection Molding Tolerances | Jiga, accessed December 3, 2025, https://jiga.io/injection-molding/injection-molding-tolerances/
46. TOLERANCES, accessed December 3, 2025, https://www.okw.com/etc/Documents/en/OKW-Tolerances-en.pdf
47. Additive Manufacturing – Design – Requirements, Guidelines and Recommendations | PDF | Radiation | 3 D Printing – Scribd, accessed December 3, 2025, https://www.scribd.com/document/455583774/ISOASTM52910-23216
48. Nylon PA11 vs PA12: Differences in Polyamides for 3D Printing – Fama 3D, accessed December 3, 2025, https://www.fama3d.com/en/materials/pa11-vs-pa12-differences
49. MJF material with the best low-temp resistance – Xometry Pro, accessed December 3, 2025, https://xometry.pro/en-eu/topic/mjf-material-with-the-best-low-temp-resistance/?bbp_reply_to=114405&_wpnonce=429bcff9b9
50. Minimizing Deformations during HP MJF 3D Printing – PMC – NIH, accessed December 3, 2025, https://pmc.ncbi.nlm.nih.gov/articles/PMC10707100/
51. Guide to MJF Nylon[+ Property Comparison] – Unionfab, accessed December 3, 2025, https://www.unionfab.com/blog/2025/05/mjf-nylon