Additive Manufacturing CT Porosity Release Case Study

This case study follows the disposition of internal porosity found by x-ray computed tomography in an additively manufactured metallic bracket. The engineering problem is not simply whether the total porosity percentage is low. The real question is whether the pore size, cluster location, detection capability, fatigue-critical stress field, and validation evidence support release.

Additive manufacturing can create useful geometry, but it can also create process-specific defect populations: lack-of-fusion defects, keyhole pores, gas pores, rough internal surfaces, anisotropy, residual stress, trapped powder, and support-removal damage. CT evidence is powerful when it is qualified, but it can be misleading if voxel size, segmentation, artifact control, and acceptance criteria are treated as administrative details.

Case Context

A lightweight bracket is produced by a powder-based additive manufacturing route and then stress relieved, machined at interfaces, and inspected before release. The bracket has two lugs, a web, and a curved fillet where finite-element stress review shows high cyclic stress.

The design intent is limited production for a fatigue-sensitive assembly. The bracket is not pressure-retaining, but a crack from a pore cluster near the lug fillet would shorten life and could create loss of alignment in service.

The release package includes:

• material certificate and powder lot traceability;
• build parameter record and build orientation;
• stress-relief and machining record;
• hardness map and witness tensile coupon;
• x-ray CT scan of the fatigue-critical region;
• acceptance criteria for global porosity, local pore size, and pore location;
• engineering disposition for nonconforming indications.

CT Inspection Data

Quantity	Value
inspected region of interest volume	V_{ROI}=8200\ \text{mm}^3
total segmented pore volume	V_p=28\ \text{mm}^3
global porosity limit	0.50\%
local critical-zone volume	V_{cz}=260\ \text{mm}^3
local critical-zone pore volume	V_{p,cz}=5.8\ \text{mm}^3
local critical-zone porosity limit	1.0\%
projected area of largest pore	A_p=0.140\ \text{mm}^2
critical-zone equivalent pore diameter limit	d_{limit}=0.30\ \text{mm}
low-stress-zone equivalent pore diameter limit	0.50\ \text{mm}
pore sizing expanded uncertainty	U_d=0.05\ \text{mm}
CT voxel size	v_x=30\ \mu\text{m}
reliable feature span requirement	4 voxels
critical defect size from fatigue review	a_c=0.55\ \text{mm}
required CT detection margin	M_a\geq0.20\ \text{mm}
distance from largest pore to fillet surface	1.2\ \text{mm}
fatigue-critical zone depth from fillet	3.0\ \text{mm}

The CT method was calibrated on reference inserts and known pore-like features in a representative material block. Reconstruction settings and segmentation thresholds were fixed before reviewing the production scan.

Step 1: Check Global Porosity Fraction

Global porosity fraction is:

Substitute:

Convert to percentage:

The global porosity requirement is:

The global porosity screen passes:

Engineering Comment

Global porosity passing is not enough. Fatigue is often controlled by the largest defect in the worst location, not by average pore volume across the full region. A low global porosity value can hide a dangerous cluster near a fillet, hole, surface, or load-transfer feature.

Step 2: Check Local Critical-Zone Porosity

Local critical-zone porosity fraction is:

Substitute:

Convert:

Compare with the local limit:

The critical-zone local porosity screen fails.

Engineering Comment

The location changes the decision. The same pore volume in a low-stress web region might be accepted or monitored. A cluster inside the fatigue-critical fillet zone is much more severe because it can interact with stress concentration, surface roughness, residual stress, and cyclic loading.

Step 3: Convert Largest Pore Area to Equivalent Diameter

Equivalent circular diameter is:

Substitute:

Apply sizing uncertainty in the conservative direction:

The pore lies within:

of the fillet surface, inside the:

fatigue-critical zone.

Critical-zone acceptance limit:

Comparison:

The guarded pore size fails the critical-zone limit.

Engineering Comment

The low-stress-zone limit of 0.50 mm is not applicable because the pore is in the critical zone. Acceptance criteria must be location-specific when stress gradients are steep. Using the wrong zone limit would turn a rejectable defect into a false pass.

Step 4: Check CT Resolution Against Critical Defect Size

Convert voxel size:

Minimum reliable feature size from the four-voxel rule:

Detection margin against the critical defect size:

The required margin is:

The CT resolution screen passes:

Engineering Comment

The method appears capable enough to detect features below the critical defect size, but that does not make the part acceptable. Method capability and part conformance are different decisions. The scan can be good evidence and still show a rejectable defect.

Step 5: Review Stress-Critical Relevance

The largest pore is near the fatigue-critical fillet. The stress review identifies this zone as the likely crack-initiation region because it combines:

• geometric stress concentration at the fillet;
• cyclic tensile stress;
• local machining and surface-finish sensitivity;
• build-direction anisotropy;
• possible tensile residual stress before stress relief;
• internal pore cluster close enough to interact with the local stress field.

The engineering issue is not only pore size. It is the combination of pore size, cluster density, surface proximity, local stress, and fatigue mechanism.

Step 6: Decide Disposition

Evidence summary:

Evidence	Result	Disposition effect
global porosity	0.341\%, passes 0.50\% limit	not sufficient for release
local critical-zone porosity	2.23\%, fails 1.0\% limit	blocks release
largest guarded pore diameter	0.472\ \text{mm}, exceeds 0.30\ \text{mm} critical-zone limit	blocks release
CT resolution margin	0.430\ \text{mm}, passes method screen	supports confidence in finding
pore location	inside fatigue-critical fillet zone	increases severity

The engineering decision is:

Do not release the bracket as-is. Hold the lot, segregate similar builds, review build records, and disposition by rework route, hot isostatic pressing plus reinspection, destructive confirmation from a sibling coupon, redesign, or rejection.

Hot isostatic pressing may reduce internal porosity, but it is not an automatic rescue. It may not close surface-connected lack-of-fusion defects, rough internal passages, or defects that remain connected to an oxygen-contaminated surface. If HIP is chosen, the part needs post-HIP CT, mechanical evidence, and confirmation that dimensional, heat-treatment, and surface-finish requirements are still met.

Failure Mode Review

Failure mode	Cause	Effect	Initial rating
fatigue crack starts at internal pore cluster near lug fillet	AM pore cluster accepted using only global porosity	crack initiation, stiffness loss, possible alignment failure	S=8,\ O=4,\ D=5

Initial risk priority number:

After process review, CT requalification, critical-zone acceptance limits, post-HIP reinspection, and lot segregation:

Engineering Comment

RPN is a prioritization tool, not a release criterion. The severity remains high. The value of the review is that it shows which controls reduce occurrence and improve detection: process window control, representative CT qualification, critical-zone limits, and action rules for clustered pores.

Transferable Lessons

Porosity percentage is a weak standalone acceptance metric. Engineers should also review pore size, shape, clustering, location, surface distance, stress field, material orientation, residual stress, CT resolution, segmentation uncertainty, and validation evidence.

CT is strongest when the inspection plan is written before looking at the image. The plan should state voxel size, reconstruction settings, artifact controls, reference features, segmentation threshold, critical zones, zone-specific limits, uncertainty, and what happens when a defect is found.

For additively manufactured fatigue-critical parts, release evidence should connect process parameters, heat treatment, surface finishing, machining, CT inspection, mechanical witness data, and nonconformance response. A part can pass global porosity and still fail engineering release because the wrong defect is in the wrong place.