Skip to main content

Defects, residual stress and distortion

📍 Where we are: Part I · The Physics of the Melt — Chapter 2 of 19. Chapter 1 showed how the liquid pool behaves; here the pool freezes, and we ask what it leaves behind: voids, cracks, and stress.

The simple version

Imagine building a wall by laying down a hot, wet strip of clay, one strip at a time. If a strip does not quite stick to the one below, you get a hidden gap. If the clay traps air, you get bubbles. As each strip cools it shrinks, but the cooler wall underneath holds it back, so the strip is left stretched and the wall quietly bends. Metal AM has all three problems at once: gaps, bubbles, and a part that is internally pulled tight like a bowstring.

What this chapter covers​

We separate the defects that come from not melting enough (lack of fusion), from melting too violently (keyhole pores), from gas, and from solidification stress (cracks). Then we turn to residual stress and distortion, whose cause is thermal and which can be reduced but not removed. Each claim is tagged as lab-scale, production or contested.

Problem​

A part can look perfect and still contain pores of tens of micrometres, cracks along grain boundaries, or stress at a significant fraction of yield strength. Pores and cracks act as sites where fatigue cracks start; residual stress distorts the part off the plate, can crack it during removal, and shifts the stress state in service. A review of flaws in powder bed fusion catalogues these flaws and their effect on performance [1]. The problem for industry is that these flaws are often invisible until inspection (CT, destructive testing) and the process cannot guarantee their absence.

State of the art​

Pores and lack of fusion​

DefectCauseTypical shapeEvidence level
Lack of fusionPool too shallow or hatch spacing too wide, so tracks or layers do not overlapIrregular, flat voids, often with unmelted powderProduction; geometric criteria are predictive [2]
Keyhole poresCavity collapse at high energy densityRound, deeper in the trackLab-scale imaging; seen on production parts (Chapter 1)
Gas poresGas in powder or shielding atmosphereSmall, roundProduction, hard to attribute
Surface-connected poresRough, partly fused layersOpen to the surfaceProduction

Lack of fusion is the most predictable, because it follows from geometry: if the melt-pool cross-section, layer thickness and hatch spacing do not overlap enough, porosity follows. A simple model built on melt-pool dimensions predicts the onset of lack-of-fusion porosity for powder bed fusion [2]. Process maps that mark lack-of-fusion, keyhole and other regimes against power and speed organise the trade-offs, with the caution that they are built per machine and alloy [3].

Pores are not necessarily permanent. Work on pore dynamics showed that bubbles can be swept out of the melt pool by fluid flow before it freezes, depending on conditions [4], and X-ray studies of directed energy deposition (DED) show pore evolution with mechanisms of their own [5]. Hot isostatic pressing (HIP) closes internal pores after the build but does not heal surface-connected ones, which we return to in Chapter 3.

Cracking​

Two cracking families matter in metals that crack: solidification cracking (the mushy zone tears while it is still partly liquid) and liquation or grain-boundary cracking in the heat-affected region. A widely used criterion links crack susceptibility to how steeply temperature changes with solid fraction near the end of solidification [6]. In practice:

  • Many high-strength aluminium alloys that crack under LPBF were made printable by adding nucleant particles that refine the grains and break up the columnar structure; the demonstration on 7075 and 6061 was published in 2017 [7]. Demonstrated at lab scale.
  • In a nickel-base superalloy considered non-weldable, cracking in electron beam melting was traced to grain boundaries and liquid films [8].
  • Alloy-design reviews group the remedies into crack-mitigation-oriented, microstructure-manipulation-oriented and machine-learning-assisted strategies [9]. A 2022 Nature Communications study used thermodynamic and kinetic reasoning to redesign the hot-cracking-prone superalloy IN738LC for printing, as an example of the approach [10]. Alloy design is taken up in Chapter 8.

Residual stress and distortion​

The mechanism has a standard name, the temperature gradient mechanism: a laser heats the top layer, which wants to expand but is restrained by cooler material below, so it yields in compression; on cooling it contracts and the layer ends in tension [11]. Stacking layers then produces a stress field with tension at the surface and compression inside, and removal from the baseplate releases it as distortion. Scan strategy, island size, preheating and part geometry all change the outcome [12]; overviews of residual stress in powder bed fusion cover the measurement methods and the influence of parameters [13][14].

Standard mitigations are demonstrated in production: baseplate or chamber preheating, rotated scan vectors between layers, supports that anchor the part, and stress-relief heat treatment before the part is cut off the plate. They reduce stress; they do not eliminate it.

Left: a block on a baseplate showing the thermal-gradient mechanism, with a hot top layer in compression and a cooled layer in tension; right: a catalogue of four defects (lack of fusion, keyhole pore, gas pore, crack) with their cause. Defects map onto too little energy (lack of fusion), too much (keyhole pores), trapped gas, and stress-driven cracking; residual stress arises from the heating and cooling mismatch between top layer and substrate. Original diagram by the authors, created with AI assistance.

Why it is hard​

  • The mechanisms trade off. Raising energy density to avoid lack of fusion pushes toward the keyhole regime; lowering it for stability invites lack of fusion. The safe window may be narrow for some alloys.
  • Stress is a property of the whole part. It depends on geometry and the entire thermal history, so a parameter validated on a coupon does not transfer cleanly to a large part with thin and thick regions.
  • Measurement is limited. Residual stress methods (diffraction, contour method, hole drilling) each see different depths and scales; none maps stress through a complex part non-destructively and cheaply [13].
  • Cracking depends on chemistry and grain structure, so it is not captured by generic process maps.

Open questions​

  1. How well can distortion be predicted before printing on a complex part? Part-scale simulation exists and is discussed in Chapter 4, but blind benchmark exercises compare predictions against measured residual strain and deflection [15], and we treat accuracy on complex geometries as contested until those comparisons are favourable and repeated.
  2. Can a crack-susceptibility criterion be turned into a quantitative design rule for new alloys, rather than a post-hoc explanation?
  3. What pore size and shape actually matter for a given property? Size, shape, location (surface versus interior) and the surrounding microstructure all enter, which Chapter 3 pursues.
  4. Can in-situ sensing flag lack-of-fusion and keyhole events early enough to correct them (Chapters 6 and 7)?

Industry impact​

Distortion and stress are why many metal parts are built with generous supports and machining allowance, why failed builds are common for new geometries, and why a build-preparation step often involves a distortion pre-compensation or a trial build. Porosity and cracking determine which alloys can be printed at all, and which parts need HIP. The cost is paid in scrap, rework and qualification effort; Chapter 12 on qualification and certification describes how regulators handle it.

Why it matters​

For any safety-critical part (a turbine bracket, a spinal implant) the question is not "is it dense on average" but "what is the largest flaw, and where is it". Defect and stress control sit between the physics of Chapter 1 and the properties in Chapter 3.

In the real world​

  • Production service providers routinely apply stress-relief heat treatment before removing parts from the plate and HIP for critical parts.
  • Alloy designers modify chemistry, or add nucleants, specifically so that crack-prone alloys can be printed [7][9]; how widely this has reached routine production varies by alloy.
  • Open benchmark datasets from NIST include residual strain and part deflection measurements against which modellers can test their predictions [15].

Key terms​

  • Lack of fusion — voids caused by insufficient melting so that adjacent tracks or layers do not bond.
  • Keyhole pore — a round pore from collapse of a deep vapour cavity.
  • Solidification cracking — tearing of the partly liquid mushy zone as it shrinks during freezing.
  • Liquation cracking — cracking where grain boundaries are re-melted in the heat-affected region.
  • Residual stress — stress locked in the part without any external load.
  • Temperature gradient mechanism — the yielding of a heated top layer against cooler material below, which leaves tension after cooling.
  • Distortion — change of shape when stress is released, for example on cutting from the baseplate.
  • Hot isostatic pressing (HIP) — high-temperature, high-pressure gas treatment that closes internal pores.
  • Process map — a diagram of defect regimes against laser power and scan speed.
  • Stress relief — a heat treatment that lowers residual stress without a full change of microstructure.

Where this leads​

Grain structure, texture and cellular substructure decide what the part can carry: Microstructure, anisotropy and process–structure–property links.