Creep is time-dependent plastic deformation under constant load, usually at elevated temperature. If you work with metals that get hot—turbine blades, boiler tubes, soldered joints, even lead roof flashing on a sunny day—you need to understand creep. We are not talking about instantaneous yielding. We are talking about a material that slowly, silently changes shape over hours, days, or decades, until it either no longer fits or it fractures. In failure analysis, creep is one of the most misunderstood mechanisms because it does not leave a dramatic fracture surface like a brittle overload. Instead, it leaves subtle clues: elongated grains, intergranular voids, and a story of time and temperature.
This article is for engineers, metallurgists, and serious makers who want a concrete, evidence-based way to think about creep. We will cover the physical mechanism, the classic creep curve, the key variables (stress, temperature, time), how to spot creep in a failure, and how to design against it. No hand-waving. No falling asleep.
What Creep Actually Is (and What It Is Not)
Creep is not just “slow stretching.” It is a thermally activated process where atoms move by diffusion or dislocation motion under a stress well below the yield strength. The defining feature: strain increases with time at constant load and constant temperature. If you remove the load, the material does not spring back completely. The deformation is permanent.
We need to separate creep from other time-dependent behaviors:
- Stress relaxation: The total strain is constant, but stress drops over time (think of a bolt losing preload at high temperature).
- Anelasticity: Time-dependent but recoverable strain (like a spring that slowly returns to shape).
- Viscoelasticity: Common in polymers, where both recoverable and permanent time-dependent strains occur. In metals, we usually call the permanent part creep.
For metals, creep becomes significant above about 0.3 to 0.5 of the absolute melting temperature (Tm in Kelvin). For lead, with Tm = 600 K, room temperature (about 300 K) is already 0.5 Tm. That is why lead deforms under its own weight. For nickel superalloys, Tm is around 1600 K, so 0.5 Tm is about 800 K (527 °C). That is why jet engine blades can run at 1000 °C only with careful alloy design and cooling.
The Creep Curve: Three Stages You Can Actually Use
When we run a constant-load creep test at fixed temperature, we plot strain versus time. The curve has three distinct regions. Understanding these regions helps you predict remaining life and diagnose failures.
Primary Creep (Transient Creep)
Strain rate starts high and decreases with time. This is the material “settling in.” Dislocations move and multiply, but they encounter obstacles and the material work-hardens. The decreasing rate reflects increasing dislocation density and entanglement. In design, primary creep is usually small and often ignored for long-term predictions, but it matters for tight-tolerance parts.
Secondary Creep (Steady-State Creep)
Strain rate is constant. This is the most important region for engineering design. The constant rate reflects a dynamic balance between work hardening and thermal recovery (dislocation climb, annihilation). We use the steady-state creep rate to estimate how much a component will elongate over its service life. The relationship is often described by the Norton-Bailey or power-law creep equation:
έss = A σn exp(-Qc / RT)
Where:
- έss = steady-state creep rate (s-1 or %/hour)
- A = material constant
- σ = applied stress (MPa)
- n = stress exponent (typically 3–8 for dislocation creep, ~1 for diffusional creep)
- Qc = activation energy for creep (J/mol, often close to self-diffusion activation energy)
- R = universal gas constant (8.314 J/mol·K)
- T = absolute temperature (K)
If you double the stress, the creep rate increases by a factor of 2n. For n=5, that is 32 times faster. Small stress changes matter enormously. Temperature is even more dramatic because of the exponential term. A 20 °C increase can halve the remaining life.
Tertiary Creep
Strain rate accelerates until fracture. This is not a new mechanism; it is the result of damage accumulation. Internal voids nucleate and grow (cavitation), the cross-sectional area decreases, so true stress increases, and necking or cracking accelerates. In a creep test, tertiary creep signals the end is near. In service, you want to catch the component before this stage.
Creep Mechanisms: What Is Happening Inside the Metal
Creep is not one thing. The dominant mechanism depends on stress, temperature, and grain size. We can map these on a deformation mechanism map (pioneered by Ashby). For most engineering alloys at high stress and moderate temperature, dislocation creep (power-law creep) dominates. At lower stress and high temperature, diffusional creep takes over. Grain boundary sliding can also contribute, especially in fine-grained materials.
Dislocation Creep
Dislocations move by glide and climb. Climb is the rate-limiting step because it requires vacancy diffusion. This gives the characteristic stress exponent n = 3–8 and activation energy close to self-diffusion. Sub-structures form: subgrains, dislocation networks. Alloying elements that pin dislocations (like precipitates in nickel superalloys) increase creep resistance dramatically.
Diffusional Creep
Atoms diffuse from grain boundaries under compression to those under tension. This elongates grains along the stress axis. Nabarro-Herring creep occurs through the lattice; Coble creep occurs along grain boundaries. Stress exponent n = 1. This mechanism is important in fine-grained materials at high temperature and low stress, like in some electronic solder joints or nuclear fuel cladding.
Grain Boundary Sliding
Grains slide past each other, accommodated by diffusional flow or dislocation motion. This can lead to wedge cracks at triple points or rounded cavities on boundaries. In failure analysis, extensive grain boundary sliding often leaves a distinctive “cobblestone” appearance on the fracture surface.
How to Spot Creep in a Failed Part
You receive a fractured turbine blade or a ruptured superheater tube. How do you know creep was involved? We look for a cluster of evidence, not just one feature.
Macroscopic Clues
- Localized necking or swelling: A tube that has “ballooned” or a blade that has stretched against the casing. Measure dimensions and compare to original drawings.
- Multiple cracking: Creep often produces numerous small cracks on the surface, especially in heat-affected zones of welds.
- Oxide patterns: Thick, multi-layered oxide scales that crack perpendicular to the stress direction can indicate long-term high-temperature exposure.
Microscopic Clues (Optical and SEM)
- Intergranular voids: Rounded or wedge-shaped cavities on grain boundaries, especially those oriented perpendicular to the applied stress. This is the hallmark of creep cavitation.
- Elongated grains: In the direction of stress, grains stretch. In extreme cases, you see “creep fissures” along boundaries.
- Precipitate coarsening: Carbides or intermetallic phases may have coarsened, indicating time at temperature. Compare to known aging curves for the alloy.
- Subgrain formation: Under TEM, you may see well-defined subgrains, evidence of recovery during creep.
Fracture Surface Features
Creep rupture surfaces are often dull and oxidized, not shiny like ductile overload. At high magnification, you may see intergranular facets with microvoids. In some alloys, creep rupture leaves a “rock candy” appearance. However, be cautious: oxidation after fracture can obscure features. Always look for corroborating evidence.
Designing Against Creep: Practical Rules
If you are selecting a material for a high-temperature application, you need creep data. The most common parameter is the creep rupture strength: the stress that causes rupture in a specified time at a specified temperature (e.g., 100,000-hour rupture strength at 600 °C). This data comes from long-term testing and is often extrapolated using the Larson-Miller parameter.
The Larson-Miller Parameter (LMP)
LMP = T (C + log tr)
Where T is temperature in Kelvin, tr is rupture time in hours, and C is a material constant (often ~20 for many steels). This parameter allows you to correlate short-term, high-temperature tests with long-term, lower-temperature service. But be careful: extrapolation beyond the data range can be misleading if the creep mechanism changes.
Material Selection Strategies
- High melting point: Creep is diffusion-controlled, so higher Tm generally means better creep resistance. Nickel-based superalloys, cobalt alloys, refractory metals.
- Stable microstructure: Avoid overaging. Use alloys with stable precipitates (e.g., gamma prime in Ni superalloys, MX carbonitrides in creep-resistant steels).
- Large grain size: For dislocation creep, larger grains reduce grain boundary sliding and cavitation. But for diffusional creep, fine grains are worse. Know your dominant mechanism.
- Solid-solution strengthening: Molybdenum in steel, tungsten in nickel alloys slow diffusion.
- Oxidation resistance: Creep and oxidation interact. A protective oxide scale (Cr2O3, Al2O3) prevents section loss and surface cracking.
Design Margin and Life Assessment
We typically allow 1% creep strain for the design life of a component. For steam turbine blades, that might be 100,000 hours. We use the Monkman-Grant relationship to estimate rupture life from minimum creep rate. Regular in-service inspections look for dimensional changes, surface replicas to detect cavitation, and hardness testing to track microstructural degradation.
Creep in Common Alloys: What You Need to Know
Different alloy families have different creep strengths and failure modes. Here is a quick reference for the metallurgist or engineer in the field.
Low-Alloy Steels (e.g., 2.25Cr-1Mo, Grade 91)
Used in power plant piping and pressure vessels. Creep strength comes from solid-solution strengthening and carbide precipitates. Grade 91 (9Cr-1Mo-V) relies on fine MX carbonitrides. After long-term service, these coarsen and the creep rate increases. Type IV cracking in the fine-grained heat-affected zone of welds is a notorious failure mode.
Stainless Steels (e.g., 304H, 316H, 347H)
Used in superheater tubes and reformer tubes. The “H” grades have higher carbon for creep strength. Sigma phase embrittlement can occur after long exposure at 600–900 °C. Watch for sensitization and intergranular corrosion that accelerates creep cavitation.
Nickel-Based Superalloys (e.g., Inconel 718, Hastelloy X)
These are the champions of high-temperature creep resistance. Strengthened by gamma prime (Ni3(Al,Ti)) or gamma double prime precipitates. Inconel 718 is limited to about 650 °C because the strengthening phase coarsens. For higher temperatures, use alloys like Inconel 617 or single-crystal CMSX-4.
Aluminum Alloys
Aluminum has a low melting point (660 °C), so creep is a concern even at moderately elevated temperatures (150–300 °C). Precipitation-hardened alloys (e.g., 2219, 6061) lose strength as precipitates coarsen. For better creep resistance, use Al-Sc or Al-Mg-Sc alloys, or metal matrix composites.
Creep Testing: How We Get the Data
Standard creep tests follow ASTM E139 or ISO 204. A specimen is heated to a uniform temperature, a constant load is applied (dead weights or a servo-controlled machine), and strain is measured over time. Tests can run for 10,000 hours or more. Accelerated tests at higher temperature are common, but you must ensure the mechanism does not change.
For remaining life assessment, we sometimes use small-punch or impression creep tests on tiny samples extracted from service components. These are not standardized for all alloys, so interpret with caution.
FAQ: Creep Questions Engineers Actually Ask
What is the difference between creep and stress rupture?
Creep is the time-dependent deformation under constant load. Stress rupture is the final fracture that ends the creep process. A creep test measures strain vs. time; a stress rupture test often only records time to failure. In practice, the terms overlap, but creep focuses on deformation, rupture on life.
Can creep happen at room temperature?
Yes, in materials with low melting points. Lead, tin, and some solders creep at room temperature. Even in steel, very high stresses (near yield) can cause measurable creep at room temperature, but it is usually negligible for engineering purposes. The rule of thumb: significant creep above 0.3–0.5 Tm.
How do I estimate remaining creep life of a component in service?
You need the operating history (temperature, stress, hours), material creep data, and ideally a non-destructive evaluation. Surface replication can detect cavitation. Hardness measurements can track softening. Then use the Larson-Miller parameter or a damage accumulation rule (like Robinson’s rule) to sum life fractions. When the sum reaches 1, the component is at end of life. This is a probabilistic assessment, not a guarantee.
Why do welds fail in creep before the base metal?
Welds have a heterogeneous microstructure: coarse-grained heat-affected zone (HAZ), fine-grained HAZ, and weld metal. The fine-grained HAZ often has lower creep strength because of grain refinement and carbide dissolution. This leads to Type IV cracking, a classic creep failure in welded steam pipes. Post-weld heat treatment and careful filler metal selection can mitigate this.
Next Steps for Your Creep Knowledge
If you are dealing with a creep failure right now, start with the operating conditions. Get the exact temperature and stress history. Then look at the microstructure. If you see intergranular voids, you are likely looking at creep. If you are designing a new component, get the creep rupture data for your candidate alloys and apply a reasonable design margin. Remember that small changes in temperature or stress can have huge effects on life.
This article is part of our series on high-temperature failure mechanisms. In a future post, we will cover thermal fatigue and how it interacts with creep to produce complex cracking patterns. If you have a specific creep failure you would like us to analyze, send us a question through the contact page. We read every one.
Note: All Pexels images used in this article are for illustrative purposes and do not depict specific proprietary equipment or failures.