Additive manufacturing metrology

In-Situ Metrology for Metal Additive Manufacturing: Why Measurement Is Becoming the Qualification Bottleneck

Updated: 13 September 2026

Short answer: Metal additive manufacturing can create complex parts layer by layer, but qualification remains difficult because manufacturers cannot rely only on the machine settings or a final visual inspection. NIST identifies the lack of reliable and validated in-situ measurements as a major barrier to wider adoption of metal AM. Current work focuses on measuring thermal behaviour, melt-pool activity, microstructure evolution, residual stress, powder-layer condition and defects during the build, then validating those measurements against controlled reference data and final part properties.

Additive manufacturing changes one of the basic assumptions of conventional production.

In machining, a manufacturer starts with material and removes it. In metal additive manufacturing, the machine creates the part while many thermal and material events occur repeatedly across hundreds or thousands of layers.

A defect may begin inside a layer long before it can be found by final inspection.

That creates a difficult quality question:

Can the manufacturing process measure enough of what is happening during the build to support part qualification?

In 2026, that question became the focus of a dedicated NIST roadmapping effort on in-situ metrology for metal alloy additive manufacturing.

Why in-situ metrology is a major 2026 AM topic

NIST held its Roadmapping In-Situ Metrology for Metal Alloy Additive Manufacturing workshop from 29 April to 1 May 2026.

NIST states that metal AM has significant potential in aerospace, defence, energy and medical devices, but wider adoption remains limited by challenges in process understanding, qualification and certification.

One barrier identified by NIST is the lack of reliable, validated in-situ measurements that can monitor structural and microstructural evolution during the AM process.

The workshop addressed measurements of:

• phase evolution;
• microstructure development;
• residual stress;
• thermal behaviour;
• optical process signatures;
• process monitoring and sensing;
• data and AI;
• model validation;
• qualification and standards.

That combination shows why AM metrology is different from simply installing another camera in the machine.

What does in-situ metrology mean?

In-situ metrology means measuring important process or material behaviour while manufacturing is occurring.

For metal additive manufacturing, this can include measurements made during powder spreading, laser interaction, melting, solidification or layer formation.

Examples include:

• melt-pool size or thermal behaviour;
• temperature distribution;
• spatter behaviour;
• powder-bed condition;
• layer geometry;
• phase changes;
• residual stress development;
• signals associated with process-induced defects.

The goal is not simply to collect more data.

The goal is to collect measurements that are sufficiently reliable to help explain, predict or control final part quality.

In-situ monitoring is not automatically metrology

This distinction matters.

A monitoring system can produce images, intensity values or machine signals without providing a calibrated measurement of a physical quantity.

For example, a camera may detect a bright region around a melt pool. That does not automatically mean the pixel intensity is a traceable temperature measurement.

Metrology requires stronger questions:

• What quantity is being measured?
• How is the instrument calibrated or characterized?
• What is the measurement model?
• What influences the result?
• What is the uncertainty?
• How has the measurement been validated?

Useful distinction: process monitoring tells you that something changed. Metrology aims to tell you what changed, by how much, and with what confidence.

Why laser powder bed fusion is difficult to measure

Laser powder bed fusion of metals, often abbreviated PBF-LB/M, creates a challenging measurement environment.

The laser interacts with metal powder over very small spatial scales and short time scales. Material melts and solidifies rapidly. Vapor, plume behaviour and ejected particles can interfere with optical measurements.

Meanwhile, the final part can be influenced by:

• laser power;
• scan speed;
• beam characteristics;
• powder properties;
• layer thickness;
• thermal history;
• gas flow;
• spatter;
• machine condition;
• part geometry.

The process is therefore dynamic. A single set of machine parameters does not fully describe everything that happened during the build.

What is melt-pool monitoring?

The melt pool is the small region of material melted by the energy source during metal AM.

Its behaviour can provide useful information about the process.

Monitoring systems may examine:

• apparent melt-pool dimensions;
• radiated intensity;
• thermal signatures;
• shape changes;
• temporal behaviour;
• relationships with process parameters.

However, translating an optical or thermal signal into a reliable physical measurement is difficult.

Emissivity, camera response, viewing geometry, wavelength, exposure, optical path and rapidly changing material state can all influence the result.

A bright signal should not be interpreted as an exact temperature without a validated measurement model.

Why thermography is important

Infrared thermography can provide information about thermal behaviour during AM.

NIST's AM Bench programme includes in-situ thermography in highly controlled laser powder bed fusion benchmark experiments.

These data help researchers test simulation models against experimental measurements.

This is important because a thermal simulation may look physically reasonable while still failing to reproduce the real process.

Controlled benchmark measurements provide evidence for model validation.

What is AM Bench?

NIST's Additive Manufacturing Benchmark Test Series, or AM Bench , provides controlled measurement data and challenge problems so researchers can test additive-manufacturing simulations against experimental evidence.

A July 2026 NIST publication explains that AM Bench data include controlled laser scan experiments with measurements such as:

• in-situ thermography;
• surface topography;
• melt-pool cross-sections.

The value is not simply that the data are available.

The experiments are designed as benchmark measurements so different models can be compared against a common physical reference.

Why model validation matters in metal AM

Additive manufacturing increasingly uses simulation to estimate:

• temperature;
• melt-pool behaviour;
• residual stress;
• distortion;
• microstructure;
• defect formation.

A model can help reduce trial-and-error, but only if its predictions are tested against suitable measurements.

A digital model should not become a substitute for physical evidence.

The stronger approach is:

1. measure the process using controlled metrology;
2. compare the model with the measurements;
3. identify disagreement;
4. improve the model or measurement understanding;
5. validate again across relevant conditions.

Spatter is a measurement problem, not only a visual defect

Spatter consists of particles ejected from the laser interaction region during laser powder bed fusion.

These particles can land elsewhere on the build area and contribute to defects or process disturbance.

In July 2026, NIST published high-fidelity measurements of spatter sizes, shapes and landing locations inside a laser powder bed fusion machine.

The researchers used a capture method that preserved particles at their landing locations, allowing high-resolution measurement of particle position, size and morphology.

This matters because conventional in-situ imaging can miss particles or measure their size imprecisely, while ex-situ collection can lose the connection between particle morphology and where it landed.

What did the 2026 spatter study show?

The NIST-linked study reported that larger particles tended to be non-spherical agglomerates and that some of the largest particles travelled substantial distances downstream in the machine.

More importantly for metrology, the work demonstrates why process monitoring needs reference measurements.

An in-situ camera system may detect only part of the actual spatter population.

High-fidelity physical measurements can be used to understand what the monitoring system sees, what it misses and whether a proxy signal is suitable for process control.

Why defect detection is harder than anomaly detection

Anomaly detection asks whether the process looks different from normal.

Defect detection asks whether a physically meaningful flaw exists in the part.

Those are not the same question.

An unusual thermal signal may not create a harmful defect. A serious defect may also occur without producing an obvious signal in a particular sensor.

To connect in-situ signals to part quality, researchers need correlation with post-process evidence such as:

• X-ray computed tomography;
• metallography;
• surface measurements;
• mechanical testing;
• dimensional inspection;
• microstructure characterization.

That relationship is essential if in-situ metrology is eventually going to support qualification.

What is qualification-ready metrology?

A measurement can be scientifically interesting without being ready for an industrial qualification decision.

Qualification-ready metrology needs stronger evidence.

The measurement should be:

• defined clearly;
• repeatable enough for the intended use;
• validated against suitable reference measurements;
• connected to relevant part properties or defects;
• supported by known calibration or characterization;
• accompanied by uncertainty or performance limits where appropriate;
• robust across the operating conditions in which it will be used.

This is why NIST's 2026 workshop focused not only on sensors but also on what measurements are actionable for process control and qualification.

Where measurement uncertainty enters AM monitoring

Uncertainty can arise at several stages.

Measurement stage	Possible influences
Optical imaging	Pixel scale, focus, perspective, exposure, optical distortion and field of view
Thermal measurement	Emissivity, wavelength, detector response, optical transmission and calibration
Spatter detection	Particle visibility, frame rate, segmentation threshold, depth and overlap
Powder-layer measurement	Surface reflectivity, spatial resolution, powder movement and sensor angle
Post-process validation	CT resolution, sectioning location, dimensional uncertainty and sampling
Model comparison	Timing alignment, coordinate mapping and assumptions in the physical model

A monitoring dashboard may show precise numerical values, but those values still require metrological interpretation.

Calibration of AM monitoring sensors

NIST's real-time AM monitoring programme explicitly includes development of calibration and characterization techniques for monitoring tools.

That is important because AM sensors operate in harsh conditions and may be integrated directly into proprietary equipment.

Calibration questions can include:

• Can the sensor be calibrated in its installed geometry?
• Does the optical path alter the response?
• How often should performance be checked?
• What reference source or artefact is appropriate?
• How is drift identified?
• Does calibration remain valid across the actual operating range?

These are familiar metrology questions in a difficult new environment.

Can AI solve the AM monitoring problem?

AI can help process large quantities of image, thermal and sensor data.

It may identify patterns that are difficult for a human operator to detect across millions of data points.

But AI still needs trustworthy reference data.

A model trained on incorrect defect labels or poorly characterized sensor data can learn the wrong relationship.

This is one reason the NIST 2026 roadmapping workshop included data, AI and digital infrastructure alongside measurement science.

The stronger sequence is:

1. develop reliable measurement;
2. establish reference data;
3. connect measurements with real part outcomes;
4. train or evaluate AI using those data;
5. validate the AI under independent conditions.

In-situ metrology vs post-process inspection

In-situ measurement does not automatically eliminate post-process inspection.

In-situ metrology	Post-process metrology
Observes the build while it happens	Examines the completed part
Can provide layer-by-layer process information	Can directly evaluate final geometry or internal structure
May detect precursor signals	Can confirm whether a defect actually exists
Useful for control and early warning	Useful for validation and acceptance
Requires correlation with final part quality	May not reveal exactly when or why the defect formed

The two approaches are strongest when used together.

Why qualification is expensive without trustworthy in-process data

If manufacturers cannot trust what happened during the build, they may need extensive post-process testing to demonstrate part quality.

For high-value applications, that can include:

• destructive testing;
• non-destructive testing;
• coupon testing;
• microstructure analysis;
• dimensional inspection;
• process requalification after significant changes.

Reliable in-situ evidence could eventually reduce some uncertainty about the build process.

It should not be assumed that one monitoring signal can replace all final inspection.

What should manufacturers measure first?

The answer depends on the failure mode that matters.

A good measurement strategy begins with the engineering question rather than the available sensor.

For example:

• If lack of fusion is the concern, identify process signatures that have a demonstrated relationship with that defect.
• If residual stress is critical, thermal imagery alone may not provide enough evidence.
• If powder spreading is unstable, layer-wise surface or powder-bed measurements may be more useful.
• If spatter contamination matters, imaging performance should be validated against physical particle measurements.

The measurement should be selected because it supports a decision, not simply because it is easy to collect.

A practical validation workflow for AM process monitoring

1. Define the defect or process characteristic. State what the monitoring system is intended to detect or measure.
2. Define the physical measurand. Avoid vague variables such as "process health" unless they are tied to measurable quantities.
3. Characterize the sensor. Understand resolution, response, calibration and environmental influences.
4. Collect synchronized process data. Preserve timing, machine parameters and spatial location.
5. Collect independent reference evidence. Use post-process metrology or destructive analysis where appropriate.
6. Establish the relationship. Determine whether the in-situ signal reliably predicts the property or defect of interest.
7. Quantify performance. Evaluate false positives, missed defects, repeatability and uncertainty as appropriate.
8. Validate on independent builds. Do not rely only on the data used to develop the monitoring method.
9. Control changes. Reassess performance if the material, machine, optics, process parameters or software change materially.

What should quality managers ask about AM monitoring systems?

• What physical quantity does the system actually measure?
• How is the monitoring sensor calibrated or characterized?
• What is the spatial and temporal resolution?
• What defects can it detect reliably?
• Which defects can it miss?
• How was performance validated?
• What independent reference method was used?
• How is sensor drift monitored?
• How are software and threshold changes controlled?
• What evidence links the in-situ signal to final part quality?

What process monitoring should not be allowed to hide

• poor machine qualification;
• uncalibrated or poorly characterized sensors;
• uncontrolled powder variation;
• changes in material lot;
• weak post-process validation;
• unknown measurement uncertainty;
• changes in software algorithms;
• an assumption that every anomaly is a defect;
• an assumption that no visible anomaly means the part is defect-free.

Why this matters for aerospace, defence and medical manufacturing

These sectors are repeatedly associated with metal AM because additive methods can produce geometries that are difficult to manufacture conventionally.

They also demand stronger evidence about material and part performance.

That makes process monitoring attractive but also raises the standard for validation.

A monitoring system that is adequate for process development may not automatically be adequate for a qualification or acceptance decision.

The intended use determines how strong the measurement evidence must be.

Why this is a metrology opportunity for Indian manufacturing

Indian manufacturers working in aerospace, defence, tooling, energy and precision engineering are increasingly exposed to metal additive manufacturing.

The metrology opportunity is broader than selling a single sensor.

It includes:

• sensor characterization;
• thermal measurement;
• optical calibration;
• dimensional validation;
• CT and NDT correlation;
• measurement uncertainty;
• data validation;
• reference artefacts;
• qualification support.

As AM moves from prototyping toward production, measurement capability becomes part of the manufacturing capability.

The likely future: measurement-rich additive manufacturing

Metal AM machines are likely to become more measurement-rich.

Future systems may combine:

• thermal sensing;
• high-speed imaging;
• layer topography;
• machine-state data;
• powder-bed measurements;
• AI-based anomaly detection;
• physics-based models;
• closed-loop process control.

The difficult part will not be collecting the data.

The difficult part will be proving which data are reliable enough to support engineering decisions.

That is why in-situ metrology is becoming one of the central qualification problems in metal additive manufacturing.

Related Multitek reading:

• Browse the Multitek Metrology calibration and measurement knowledge base
• Reading Calibration Certificates: As-Found vs. As-Left Data Explained
• Multitek Technologies

Frequently asked questions

What is in-situ metrology in additive manufacturing?

In-situ metrology is the measurement of important process or material behaviour while the additive-manufacturing build is taking place. It can include thermal, optical, structural, dimensional or other measurements used to understand process behaviour and part quality.

What is melt-pool monitoring?

Melt-pool monitoring uses optical, thermal or other sensing methods to observe the small region of material melted by the energy source during metal additive manufacturing. The signals must be validated before they are treated as reliable physical measurements or defect indicators.

Why is spatter important in laser powder bed fusion?

Spatter particles are ejected during the laser-material interaction and can land elsewhere in the build area. They can contribute to process disturbance and defects, so their size, shape, transport and landing locations are useful process-monitoring variables.

What is AM Bench?

AM Bench is a NIST-led additive-manufacturing benchmark programme that provides highly controlled measurement datasets and challenge problems so researchers can test and validate manufacturing simulations against physical experiments.

Can in-situ monitoring replace final inspection?

Not automatically. In-situ monitoring can reveal process behaviour and potential defect signals, while post-process inspection can provide direct evidence about the completed part. The two approaches often need to be correlated and used together.

Why is measurement uncertainty important in AM monitoring?

Monitoring systems are affected by sensor response, optical conditions, emissivity, spatial resolution, timing, calibration and data-processing choices. Understanding these influences is necessary before measurements are used for qualification or automated process control.

Sources checked

• NIST: Roadmapping In-Situ Metrology for Metal Alloy Additive Manufacturing , April 29 to May 1, 2026.
• NIST: Enabling Modelers to Test Their Simulations Against Rigorous, Highly Controlled Additive Manufacturing Benchmark Measurement Data , published 6 July 2026.
• NIST: Captured: High-Fidelity Measurements of Spatter Sizes, Shapes, and Landing Sites within a Laser Powder Bed Fusion Machine , published 5 July 2026.
• NIST: Fundamental Measurements for Metal Additive Manufacturing .
• NIST: Metrology for Real-Time Monitoring of Additive Manufacturing .

Editorial note: Additive-manufacturing monitoring technologies and qualification methods are evolving. This article explains public measurement-science information available as of 13 September 2026. A monitoring signal should not be treated as a validated defect or qualification metric unless its relationship to the relevant physical property has been demonstrated for the intended application.