Creep resistance is a printed part’s ability to resist slow, time-dependent deformation while a load remains applied. A bracket can pass a short tensile or bending test, remain unbroken, and still sag after days or months because polymers are viscoelastic: their strain can continue to change under sustained stress. For FFF parts, creep is controlled by the polymer formulation, temperature, load level, time, print orientation, internal structure, moisture state, and part geometry. Short-term strength alone cannot predict whether a printed part will hold its shape over long service periods.
Terminology: structural creep in a finished printed part is different from hotend “heat creep.” Hotend heat creep describes filament softening too far up the melt path and causing extrusion problems. Structural creep describes a completed part slowly changing shape under sustained mechanical load.
What Creep Looks Like in a Printed Part
Creep does not require visible cracking. A wall-mounted hook may rotate downward by a few millimeters, a spacer may become thinner, a belt tensioner may lose position, or a fixture may drift outside its dimensional tolerance. In bolted assemblies, the plastic beneath the screw head, washer, insert, or nut can compress over time and reduce clamping force even when the fastener has not rotated.
The defining condition is sustained stress over time. The load may be tensile, compressive, flexural, or a combination. ASTM D2990-17(2025), the current ASTM creep standard for plastics, covers tensile and compressive creep as well as flexural creep and creep-rupture testing. ASTM states that creep data are used to evaluate long-term dimensional change, creep modulus, and material behavior under constant load. [a]
| Behavior | Main loading condition | What changes with time | Typical printed-part symptom |
|---|---|---|---|
| Creep | Load or stress remains applied | Strain or deflection grows | A bracket slowly sags |
| Stress relaxation | Part is held at nearly fixed deformation | Force or stress falls | A clip or clamp loses holding force |
| Fatigue | Load repeats through many cycles | Damage accumulates | A hinge, spring, or mount cracks after repeated use |
| Immediate overload | Load exceeds short-term capacity | Failure occurs rapidly | Sudden yielding, layer separation, or fracture |
| Creep rupture | Sustained load is maintained long enough | Time-dependent deformation progresses to failure | Delayed fracture after a period of deformation |
Why a Strong Part Can Still Creep
Polymers combine elastic and time-dependent behavior. When a constant load is first applied, part of the deformation appears immediately. Additional deformation can then develop while the load remains unchanged. If the load is removed, some of that strain may recover immediately, some may recover slowly, and some may remain as permanent set.
ASTM D2990 describes the familiar idealized creep curve as an initial rapid strain followed by primary creep, a secondary region with a steadier creep rate, and, in some cases, tertiary creep that accelerates toward failure. The standard also notes that the three-stage curve is idealized; not every plastic displays every stage clearly. [a]
| Stage | Observed behavior | Design meaning |
|---|---|---|
| Initial strain | Deflection appears as soon as the load is applied | The part may already use part of its allowable dimensional tolerance before creep begins |
| Primary creep | Strain continues to rise while the creep rate tends to decrease | Early measurements can overstate the long-term rate if extrapolated linearly |
| Secondary creep | Deformation continues at a more stable rate | Long-duration shape retention becomes the main concern |
| Tertiary creep | Deformation accelerates toward failure in materials and conditions that develop this stage | The part is no longer merely drifting; delayed structural failure may be approaching |
Stress, Temperature, and Time Act Together
A creep result has little meaning without its stress, temperature, and duration. A material that changes shape slowly at room temperature may deform much faster when warm. Likewise, a load that causes little movement over one hour may produce unacceptable deflection over months. Tests on FFF-printed ABS, CPE, PLA, tough PLA, PC, and nylon at 25°C, 40°C, and 60°C under 10 MPa and 20 MPa showed that material type, temperature, and applied stress all altered the measured short-term creep response. [d]
No universal “safe percentage of tensile strength” exists for creep. A useful allowable stress depends on the exact filament grade, printed structure, service temperature, duration, loading mode, acceptable deflection, and environmental conditioning. A short-term ultimate tensile strength value is not a substitute for a creep curve.
Temperature Does Not Need to Reach Tg Before Creep Matters
Glass transition temperature (Tg), heat deflection temperature (HDT), and Vicat softening temperature describe different thermal behaviors. None is a universal onset temperature for creep. Tg identifies a polymer transition region, while HDT and Vicat are measured under defined test conditions. A loaded printed part can accumulate creep below any headline “heat resistance” number because time and stress remain active variables.
This distinction is explicit in ASTM D648 for heat deflection temperature: ASTM states that HDT data should not be used to predict elevated-temperature endurance unless time, temperature, loading method, and stress are similar to the test conditions. [b] ASTM D1525 likewise defines Vicat as a controlled needle-penetration softening test used to compare thermoplastics, not as a structural creep-life test. [c]
Moisture Can Change the Service State of Hygroscopic Materials
For moisture-sensitive materials such as many polyamides, the condition of the printed part matters. Absorbed water can plasticize the polymer, changing stiffness and time-dependent deformation. Creep data measured on a dry specimen therefore should not automatically be applied to a part that will operate after moisture uptake. The same principle applies to any filament whose formulation or conditioning state changes mechanical behavior.
Why Tensile Strength, Tg, and HDT Do Not Rank Creep Resistance
Datasheet values answer different questions. A tensile test asks how a specimen behaves while load rises over a relatively short test. Creep asks how strain develops while load persists. A material can therefore have high short-term strength yet lose dimensional accuracy under a much smaller sustained load.
| Property | What it describes | What it does not establish |
|---|---|---|
| Ultimate tensile strength, MPa | Short-term tensile failure strength under the stated test | How much a part will deform after weeks or months under a lower constant stress |
| Young’s modulus, GPa | Initial elastic stiffness over the defined test range | Long-term stiffness after viscoelastic strain develops |
| Tg, °C | A polymer transition region measured by an appropriate thermal method | A guaranteed maximum continuous structural-use temperature |
| HDT, °C | Defined deflection while a standardized flexural load and heating procedure are applied | Long-duration creep performance under a different load, geometry, or temperature history |
| Vicat softening temperature, °C | Temperature at which specified needle penetration occurs under the test method | Creep life or allowable stress for a printed bracket, clamp, or fixture |
| Elongation at break, % | Ductility during the stated mechanical test | Dimensional stability under sustained service load |
The comparison becomes even less direct when a datasheet value comes from injection-molded or otherwise processed specimens. FFF adds roads, layer interfaces, voids, cooling history, residual stress, and directional structure. Filament resin data describe the material; creep of a finished print describes the material plus the manufacturing process and the part geometry.
FFF Orientation Changes the Long-Term Load Path
FFF parts are anisotropic: their response depends on direction. The same model can place a sustained load along continuous extrusion roads, across adjacent roads, or across stacked layers. That changes where shear, interface stress, and local strain develop.
A 2023 experimental and numerical study of 3D-printed PLA tested two build directions and 0°/90° unidirectional raster orientations. The authors found that build direction and raster orientation altered both tensile and creep behavior; a specimen orientation that showed higher yield stress did not necessarily show better creep strength. [e] This is another reason not to rank long-term performance from a single short-term strength number.
Orientation rule for sustained loading: when geometry allows, place the main long-term load so it is carried through continuous material paths rather than forcing a high sustained tensile or shear stress across weak interfaces. The best print orientation still depends on the complete stress state; a direction that helps one feature can weaken another.
Walls, Infill, and Layer Height Affect Creep Differently
More infill can reduce stress in some structures by adding load-bearing material, but infill percentage alone does not determine creep resistance. Outer walls may carry much of the bending stress in a beam-like part, while infill topology can change how load is redistributed between walls and the interior. Layer height also changes the number and quality of interfaces across a given part thickness.
A PLA study using ASTM D2990 specimen geometry examined layer heights of 0.1, 0.2, and 0.3 mm, infill levels of 10%, 55%, and 100%, and linear, hexagonal, and diamond patterns under a constant 15 MPa tensile stress at 25°C. Layer height, infill percentage, and infill pattern affected the measured creep rate, while the pattern also affected rupture time. The reported optimum was specific to that printer, PLA, geometry, stress, and experimental model; it should not be converted into a universal slicer preset. [f]
More recent work reinforces the same point. A 2025 Tough PLA study linked infill pattern and density to shear-strain distribution and creep failure, with honeycomb structures distributing strain more evenly than the tested line pattern. [g] In a separate 2025 PETG study, the tested honeycomb specimens showed about 15% lower long-term creep strain than the tested line-infill specimens even though their tensile strength was about 14% lower. [h] That result is useful precisely because it shows why the configuration with the highest tensile strength is not automatically the configuration with the lowest long-term creep.
Do not read “honeycomb” as a universal winner. Those findings belong to the materials, densities, specimen shapes, loads, and print settings tested in the cited studies. A different wall count, infill density, raster direction, geometry, or filament formulation can change the result.
Geometry Often Changes Creep Risk More Than Extra Infill
Creep responds to stress, so reducing local stress is often more effective than simply filling empty volume. A long, thin cantilever creates a large bending moment at its root. A small screw head can concentrate compressive stress into a narrow ring of plastic. A sharp internal corner can raise local stress even when the average stress across the part looks modest.
For beam-like parts, increasing section depth in the bending direction can have a large effect because bending stress and stiffness depend strongly on cross-section geometry. Ribs can move material away from the neutral axis, fillets can reduce stress concentration, and a shorter lever arm can reduce bending moment without changing the filament at all.
Simple nominal stress checks:
For axial tension or compression, nominal stress is σ = F/A. When force F is in newtons and area A is in mm², the result is N/mm², numerically equal to MPa.
For a rectangular cantilever with an end load, a basic elastic beam estimate for maximum root bending stress is σmax = 6FL/(bh²), where F is end load, L is cantilever length, b is width, and h is section depth in the bending direction.
These equations estimate nominal stress; they do not predict creep life. Printed anisotropy, notches, holes, layer interfaces, temperature, nonlinear viscoelasticity, and local contact stresses are outside those simple equations.
Bolted and Clamped Parts Have a Preload Problem
A bolted plastic assembly can lose clamping force without the bolt visibly backing out. The plastic under the head, washer, nut, insert shoulder, or compression face can creep in compression. As the plastic thickness changes, bolt elongation and preload can fall. A similar effect occurs in clips and spring-like features held at a nearly fixed displacement: stress relaxation reduces the force they apply over time.
A published experiment on 3D-printed PLA and PETG used bolted specimens to examine compression creep over seven days. In the untreated control groups, the reported clamping-force losses were about 3% for PLA and 11.5% for PETG under that study’s assembly and test conditions. [k] Those percentages should not be transferred to other fastener sizes or print geometries, but the test demonstrates why a connection can become loose while the printed part remains intact.
Useful design responses include spreading contact pressure with a larger washer or bearing surface, increasing boss area, avoiding thin plastic trapped directly under high preload, and using a metal compression limiter when the joint must maintain a controlled clamp distance. A threaded insert improves thread durability and load transfer at the thread interface, but it does not automatically prevent the surrounding polymer from creeping under sustained compression.
How Common Filament Families Should Be Read for Creep
There is no reliable material-family ranking that applies to every printed part. PLA, PETG, ABS, ASA, nylon, PC, PP, and filled grades contain many resin blends and additives, and the printed result can differ from molded-property data. Material selection should therefore start with the service condition rather than a one-number leaderboard.
| Material family | Useful creep-related characteristic | What needs checking in the actual application |
|---|---|---|
| PLA and PLA variants | Often stiff at room temperature; annealable grades may change crystallinity after heat treatment | Warm service, sustained bending, annealing distortion, exact grade and conditioning |
| PETG and copolyesters | Tough printed parts can still accumulate long-term strain under sustained load | Load duration, warm environments, wall/infill load path, exact copolyester formulation |
| ABS | Typically offers more thermal margin than standard PLA, but it remains a viscoelastic thermoplastic | Printed orientation, residual stress, sustained load, actual ABS formulation |
| ASA | Used where outdoor exposure and higher-temperature capability than basic PLA are desired | Long-term load still requires creep data; UV/weathering resistance does not equal creep immunity |
| Nylon / PA | Toughness can be useful for functional parts | Moisture state, temperature, grade, reinforcement, and loss of stiffness after conditioning |
| PC and PC blends | Many grades provide higher thermal capability than commodity filaments | Blend formulation, chamber/printing quality, layer direction, and actual long-term data |
| PP | Semi-crystalline structure and flexibility can suit living-hinge or chemical-service designs | Low stiffness and time-dependent deformation can limit dimension-critical sustained-load uses |
| Fiber-filled grades | Short fibers can raise stiffness and restrain matrix deformation in directions where they contribute | Fiber orientation, matrix polymer, layer bonding, local stress, and whether the load is aligned with the reinforced path |
A 2022 comparison across several printed polymers found different creep responses under the same test temperatures and stress levels, confirming that material family matters but does not remove the effects of temperature and load. [d] For filled materials, the matrix still carries part of the viscoelastic response, and short-fiber alignment created during extrusion can make reinforcement directional.
Annealing PLA Can Improve Creep Resistance, but There Is No Universal Recipe
Annealing can alter PLA crystallinity and therefore change its thermal and time-dependent mechanical response. The useful part of the research is not a single temperature to copy; it is the evidence that heat-treatment outcome depends on material, time, cooling method, test temperature, and loading mode.
A study published in the January–February 2026 issue of European Journal of Mechanics – A/Solids compared untreated PLA with samples heat-treated at 75°C and 90°C for different durations. For that material and protocol, 90°C for 2 hours produced the lowest creep response among the tested heat treatments; crystallinity increased from about 3% in the untreated sample to about 54% in the reported optimum. The same heat-treated condition also showed lower creep rate than untreated PLA in elevated-temperature tests at 37°C and 47°C. [i]
A separate study first published online in July 2026 used 5-hour anneals from 60°C to 100°C, furnace cooling, and a 60°C hot-impression creep test under a 1 kg load. That study identified 60°C as the best annealing condition for reducing deformation in its test setup, reporting a 75% reduction relative to its as-printed samples. [j]
Annealing temperatures from one PLA study should not be copied as a universal production recipe. PLA grades contain different stereochemistry, nucleating agents, modifiers, pigments, and fillers. Annealing can also shrink, warp, or otherwise change dimensions, so creep improvement must be weighed against tolerance changes in the actual part.
Creep Modulus Describes Long-Term Stiffness Better Than Young’s Modulus
For a constant applied stress, creep strain ε(t) changes with time. A time-dependent creep modulus can be written as Ec(t) = σ/ε(t). If strain rises while stress remains fixed, the calculated creep modulus falls. This makes creep modulus more relevant to long-duration stiffness than an initial Young’s modulus measured during a short mechanical test.
ASTM D2990 identifies creep modulus and dimensional change under long-term loading among the uses of creep data. [a] For real design work, the modulus must correspond to the material, stress range, temperature, conditioning, print state, and time of interest. A creep modulus quoted at 1 hour cannot simply be treated as the 1,000-hour value.
How Creep Testing Is Performed and What It Can Prove
ASTM D2990-17(2025) provides methods for tensile, compressive, and flexural creep and for creep-rupture of plastics under specified environmental conditions. It is designed to generate data for material comparison and long-term performance assessment when the test conditions are relevant to the intended use. [a] ASTM also warns that ASTM D2990 and ISO 899 address similar subject matter but differ technically, so their results should not be treated as directly interchangeable.
For FFF parts, a standardized material test and a part-level test answer different questions. Standard specimens help compare controlled variables. A finished-part test captures actual wall count, infill, layer direction, holes, fasteners, contact pressure, manufacturing defects, and load path. When dimensional retention matters, both kinds of evidence can be useful.
Accelerated Prediction Needs Validation
Waiting months or years for every design iteration is impractical, so research uses accelerated methods and viscoelastic models to build longer-time predictions. These methods are not generic calculators. Their reliability depends on the material response, stress increments, dwell time, model choice, and whether the acceleration method preserves the mechanisms present in conventional creep.
In the 2025 PETG stepped iso-stress study, a power-law rescaling method with 10% stress increments and 5-hour dwell periods produced less than 5% mean absolute error against the authors’ 720-hour conventional creep data, while the accelerated test took 25 hours. The authors also stated that the method could not predict tertiary creep and was sensitive to rescaling choice. [h] The practical lesson is that accelerated testing can shorten evaluation time when it is validated for the material and loading regime; it does not justify a universal online “creep-life calculator” for arbitrary filament.
A Practical Creep Test for a Real Printed Part
A workshop test cannot replace ASTM laboratory data, but it can reveal whether two designs, print orientations, or materials retain shape differently under the actual type of load. The test is most useful when every variable except the one being compared is held constant.
- Print matched specimens or parts. Use the same filament condition, machine, nozzle, slicer profile, orientation, wall count, infill, and cooling unless one of those variables is the subject of the test.
- Condition the samples consistently. Do not compare a freshly dried nylon specimen with one that has equilibrated in humid air unless moisture is the variable being tested.
- Measure the unloaded geometry. Record the dimension or deflection that defines functional success, not just overall length.
- Apply a constant, known load. Dead weights are useful for simple fixtures because the force does not depend on spring relaxation. For bolted or clamped parts, the loading method should match the real joint as closely as practical.
- Control temperature. Record the temperature close to the specimen. A room that moves between cool nights and warm afternoons can hide or exaggerate material differences.
- Record displacement on a time schedule. Measurements at initial loading, 1 hour, 6 hours, 24 hours, 72 hours, and 168 hours can reveal early trends; longer tests are needed when the real application lasts months or years.
- Unload and measure recovery. Immediate recovery, delayed recovery, and remaining permanent set help distinguish recoverable viscoelastic strain from lasting dimensional change.
- Repeat the comparison. One printed specimen can contain a local defect or dimensional error. Replicates make a material or design comparison more credible.
For safety-related, load-bearing, elevated-temperature, or costly applications, a casual bench test should not be treated as engineering qualification. The required test duration, safety factor, failure criteria, and statistical sample size depend on the consequences of deformation or failure.
Design Changes That Usually Reduce Creep Risk
Better creep resistance starts by lowering stress in the regions that control function. Material substitution can help, but it is only one control. A thin bracket printed from a higher-temperature polymer can still creep if its root stress is high, while a better proportioned bracket in a less exotic material may stay within tolerance because the sustained stress is lower.
| Observed problem | First variables to examine | Typical design response |
|---|---|---|
| Long bracket sags downward | Cantilever length, section depth, root stress, temperature | Shorten the lever arm, deepen the section, add ribs, or reduce load |
| Bolt needs repeated retightening | Contact pressure, boss thickness, plastic trapped in compression | Increase bearing area or use a metal compression limiter where appropriate |
| Part deforms only when warm | Service temperature relative to material behavior, local heat source | Reduce heat exposure, lower stress, or select a grade with verified long-term data at the needed temperature |
| Dimension changes mainly in one print orientation | Raster direction, layer interfaces, sustained tensile or shear path | Reorient the part or redesign the load path so continuous material carries more of the sustained load |
| High-infill part still bends | Wall count, outer section geometry, bending moment | Move material toward the outer section, increase walls or section depth, and reduce leverage before relying on more infill |
| Clip loses holding force | Constant deflection, stress relaxation, temperature | Reduce imposed strain, lengthen the flexing arm, change section geometry, or validate a more suitable material |
| Nylon part drifts after environmental exposure | Moisture condition, temperature, preload | Test the part after realistic conditioning rather than using only dry-state data |
| Fiber-filled part still moves | Matrix polymer, fiber orientation, layer direction, local stress | Align the load with reinforced paths where possible and verify the printed composite rather than relying on the fill label |
When deformation tolerance is tight, define an allowable displacement before testing. “Did not break” is too weak a pass criterion for a jig, optical mount, belt tensioner, gear center distance, sealed joint, or precision fixture. The part fails functionally when its geometry or clamping force leaves the allowed range, even if there is no fracture.
FAQ
Does higher tensile strength mean better creep resistance?
No. Tensile strength measures short-term failure behavior under a defined tensile test, while creep measures time-dependent strain under sustained load. Print orientation and infill can even produce a configuration with lower tensile strength but lower long-term creep strain under a particular test, as shown in the cited 2025 PETG study. [h]
Does 100% infill stop creep?
No. Higher infill can lower stress in some geometries, but creep still depends on material, walls, print direction, temperature, load, time, and local stress concentration. A long thin 100% infill cantilever can still creep because its geometry creates high sustained bending stress.
Can a printed part creep below its glass transition temperature?
Yes. Tg is not a creep-onset switch. Viscoelastic deformation can occur below Tg, and its rate depends on stress, time, formulation, and temperature. Approaching a transition region can make deformation much faster, but staying below Tg does not guarantee dimensional stability.
Is HDT a safe continuous-use temperature for a loaded printed part?
No. HDT is measured under a defined flexural-load test and heating procedure. ASTM D648 states that the result is not intended to predict elevated-temperature endurance unless service conditions are similar to the test conditions. [b]
Do carbon-fiber-filled filaments eliminate creep?
No. Short-fiber reinforcement can raise stiffness and reduce deformation in directions where the fibers carry load, but the polymer matrix remains viscoelastic and FFF fiber orientation is directional. Interlayer regions and matrix-dominated loading can still control long-term movement.
Can annealing make PLA safe for any long-term warm load?
No. Annealing can improve creep response for some PLA formulations and conditions, but published studies use different materials and heat-treatment protocols and do not produce one universal recipe. Annealing can also change dimensions. A finished part still needs to be checked at its real load, temperature, orientation, and tolerance. [i]
How long should a creep test run?
Long enough to represent the decision being made. A one-day test can compare early deformation but cannot prove multi-year stability. Longer service periods require longer validated testing or a material-specific accelerated method whose model has been checked against conventional creep data.
Sources
- ASTM International — ASTM D2990-17(2025), Standard Test Methods for Tensile, Compressive, and Flexural Creep and Creep-Rupture of Plastics — Supports creep test modes, creep modulus, long-term dimensional-change use, and the idealized creep-stage description. (Official standards organization.)
- ASTM International — ASTM D648-18, Standard Test Method for Deflection Temperature of Plastics Under Flexural Load in the Edgewise Position — Supports the limits of using HDT to predict endurance at elevated temperature. (Official standards organization.)
- ASTM International — ASTM D1525-25, Standard Test Method for Vicat Softening Temperature of Plastics — Supports the definition and intended comparison use of Vicat softening temperature. (Official standards organization.)
- Journal of Mechanical Engineering — Short-term Creep Behaviour of Different Polymers Used in Additive Manufacturing under Different Thermal and Loading Conditions — Supports the effects of polymer type, temperature, and stress level on creep of printed materials. (Peer-reviewed research article.)
- PubMed — Numerical and Experimental Investigation on Creep Response of 3D Printed PLA Samples: Part I — Supports the effect of build direction and raster orientation on printed PLA creep behavior. (Indexed peer-reviewed research record.)
- Polymers — Multi-Response Optimization of Tensile Creep Behavior of PLA 3D Printed Parts Using Categorical Response Surface Methodology — Supports the reported effects of layer height, infill percentage, and infill pattern under the study’s PLA creep conditions. (Peer-reviewed research article.)
- Engineering Failure Analysis — Creep and Failure of 3D-Printed Polymers: Impact of Infill Patterns and Densities on Shear Strain and Strength — Supports the role of infill topology, density, interlayer shear, and load distribution in Tough PLA creep. (Peer-reviewed research article.)
- Engineering Failure Analysis — Accelerated Creep Prediction in 3D Printed PETG Components Using the Stepped Iso-Stress Method — Supports the PETG infill comparison and the stated accelerated-test validation results and limitations. (Peer-reviewed research article.)
- European Journal of Mechanics – A/Solids — Enhancing the Creep Resistance of 3D-Printed Polylactic Acid at Varied Operating Temperatures by Heat Treatment — Supports the reported PLA crystallinity and creep results for the tested 90°C, 2-hour treatment. (Peer-reviewed research article.)
- Journal of Reinforced Plastics and Composites — Optimizing Creep Behavior of 3D-Printed PLA Through Annealing — Supports the separate 2026 annealing protocol and hot-impression creep result. (Peer-reviewed research article.)
- Polymers — Accelerated Aging Effect on Mechanical Properties of Common 3D-Printing Polymers — Supports the reported seven-day clamping-force losses in the study’s printed PLA and PETG compression-creep assemblies. (Peer-reviewed research article.)