ORNL’s New AI Controller Could Solve One of Composite 3D Printing’s Biggest Problems

Updated on June 01, 2026 • 7 min read

Large-format composite 3D printing has made enormous progress over the last decade. Manufacturers can now print boat molds, aerospace tooling, construction panels, shipping containers, and even structural composite components measured in meters rather than centimeters.

Yet one challenge continues to hold the technology back:

Print failures caused by poor layer bonding and temperature variations.

When a desktop 3D printer fails, the loss might be a few dollars of material.

When a large-format composite printer fails, the consequences are far more serious:

• Hundreds of kilograms of wasted composite material
• Dozens of hours of machine time lost
• Significant energy consumption
• Production delays
• Higher manufacturing costs

Researchers at Oak Ridge National Laboratory (ORNL) believe they may have found a solution.

The U.S. Department of Energy laboratory has developed a new intelligent control system capable of automatically detecting and correcting printing errors in real time during large-scale composite additive manufacturing.

The development could become an important milestone in the industry’s transition from operator-dependent printing to autonomous composite manufacturing.

Why Large-Format Composite 3D Printing Is Difficult

Large-area additive manufacturing (BAAM) and other large-format extrusion systems work by depositing molten composite material layer by layer.

The concept sounds simple.

The reality is far more complex.

Every deposited layer must satisfy two conflicting requirements:

1. Stay hot enough to bond properly with the next layer
2. Cool quickly enough to support the structure above

If the material cools too much:

• Interlayer adhesion weakens
• Delamination risk increases
• Mechanical performance drops

If the material remains too hot:

• Distortion occurs
• Dimensional accuracy suffers
• Surface quality deteriorates

Maintaining the perfect balance becomes increasingly difficult as part size increases.

A small test coupon may take minutes to print.

A full-scale aerospace mold or marine component may require many hours or even days.

Throughout that time, environmental conditions continuously change.

Even minor temperature fluctuations can create defects.

ORNL’s Solution: A Self-Correcting Printing System

Instead of relying on operators to constantly monitor the process, ORNL developed a controller that can supervise the print automatically.

The system combines several technologies:

Sensor Network

The platform monitors:

• Nozzle position
• Printing speed
• Extrusion conditions
• Material temperature

Thermal Camera Ring

Researchers added a ring of low-cost thermal cameras around the print head.

These cameras continuously observe the deposited composite material as it cools.

Computer Vision

Computer vision algorithms analyze thermal images in real time.

The software identifies:

• Hot spots
• Cold spots
• Temperature deviations
• Process instability

Automatic Process Control

When the system detects an issue, it automatically adjusts process parameters.

The goal is simple:

Maintain ideal layer temperature without requiring human intervention.

How the Technology Works

According to ORNL researchers, the controller behaves much like an experienced operator.

It continuously observes the process.

When it notices conditions drifting away from target values, it responds immediately.

Project lead researcher Kris Villez explained that the controller essentially watches the process and makes small corrections until desired conditions are restored.

Unlike traditional quality inspection systems that identify defects after manufacturing is complete, ORNL’s approach actively prevents defects from developing.

This distinction is critical.

Finding a defect after printing may be too late.

Preventing the defect altogether creates far greater value.

The Demonstration Test

To validate the technology, researchers printed a large hexagonal component.

The printed part was larger than a truck tire.

The experiment deliberately began under challenging conditions.

Researchers started with a print speed that resulted in material temperatures significantly below the desired range.

The controller quickly detected a problem.

Material temperatures were approximately:

30% lower than target values

Normally, this temperature deficit would create weak interlayer bonding.

Instead, the controller automatically increased print speed.

This adjustment restored temperatures to the desired range and maintained proper bonding conditions.

The correction occurred without operator intervention.

The system successfully demonstrated real-time adaptive control.

Why This Is Different From Existing Monitoring Systems

Process monitoring is not new.

Several companies already provide advanced monitoring solutions for additive manufacturing.

Examples include:

• Sigma Additive Solutions
• Additive Assurance
• Fraunhofer IAPT

However, most existing systems focus on:

• Laser Powder Bed Fusion (LPBF)
• Metal additive manufacturing
• Aerospace metal components

Large-format polymer composite printing has received far less attention.

Yet many industrial composite prints are equally expensive to replace when failures occur.

ORNL’s work specifically targets this underserved area.

No Retraining for Every New Part

One of the most impressive features of ORNL’s system is flexibility.

Many AI-based manufacturing systems require retraining whenever:

• Geometry changes
• Materials change
• Process conditions change

That creates significant deployment challenges.

ORNL’s controller was designed differently.

Researchers report that it can operate across:

• Different geometries
• Different polymers
• Different composite formulations
• Different printer configurations

without requiring extensive retraining.

This significantly lowers implementation barriers for manufacturers.

Digital Twin Technology Adds Another Layer

The system also incorporates machine-learning-driven digital twin technology.

A digital twin is a virtual representation of the manufacturing process.

This allows engineers to:

• Test new geometries
• Simulate process changes
• Evaluate material behavior
• Predict thermal performance

before printing actual parts.

Because experiments occur virtually, manufacturers can reduce:

• Development costs
• Material waste
• Machine downtime

while accelerating process optimization.

Why Composite Manufacturers Should Pay Attention

Composite additive manufacturing continues expanding into industrial production.

Applications include:

Aerospace

• Tooling
• Assembly fixtures
• Trim molds
• Layup tools

Marine

• Boat molds
• Hull tooling
• Structural components

Construction

• Architectural elements
• Building panels
• Modular structures

Transportation

• Truck bodies
• Shipping containers
• Lightweight structures

As component sizes increase, process consistency becomes increasingly important.

An intelligent controller capable of automatically correcting errors could dramatically improve production reliability.

The Bigger Trend: Autonomous Manufacturing

The ORNL project reflects a broader trend within advanced manufacturing.

For years, additive manufacturing innovation focused primarily on hardware:

• Larger printers
• Faster deposition rates
• New materials

Today, attention is shifting toward intelligence.

Manufacturers increasingly want systems that can:

• Monitor themselves
• Detect problems
• Correct errors
• Document quality automatically

This approach reduces dependence on operator experience and improves scalability.

The future factory will likely combine:

• AI
• Machine vision
• Digital twins
• Process automation

into integrated production environments.

ORNL’s controller represents an important step in that direction.

🔒 Content Transparency & Editorial Integrity

This article is developed based on real engineering experience, machine testing data, and practical production knowledge from Jota Machinery’s work in advanced composite manufacturing.

All technical explanations—including material structure, processing methods, and performance characteristics—are reviewed and verified by our engineering team to ensure accuracy and real-world relevance.

To improve clarity and structure, AI-assisted tools may have been used during content organization and language refinement. However:

• All key technical insights originate from first-hand industrial experience
• All data and claims are manually reviewed and validated
• The content is created with the primary goal of educating engineers, manufacturers, and buyers

We do not publish content solely for search ranking purposes. Every article is designed to provide practical, experience-based value to professionals in the composite materials industry.

Editorial Perspective

The most important aspect of ORNL’s research is not the thermal cameras.

It is not the machine learning model.

It is not even the automatic speed adjustment.

The real significance lies in moving additive manufacturing from error detection to error prevention.

Many current systems can tell manufacturers something went wrong.

Far fewer can prevent the problem before the part is lost.

As large-format composite printing moves toward production applications in aerospace, maritime, defense, transportation and construction, this distinction becomes increasingly valuable.

A future where composite printers continuously monitor, adjust and optimize themselves is no longer theoretical.

ORNL’s latest work suggests that future may be arriving faster than many expected.