VIL

A high-potential industrial technique – with caveats

Additive metal manufacturing technology is driving Industry 4.0 in diverse domains, from aerospace and energy to automotive and medical tooling. Powder bed 3D printing melts metal using high power lasers to rapidly create lightweight and complex objects, parts and tools while using less material.

However, the productivity of this manufacturing method as well as specific quality limitations remain key challenges regarding the competitiveness of additive metal manufacturing compared to metalworking techniques. The main quality issues are due to excessive porosity, metal spattering and cracking. Productivity is limited by lack of real-time control loops to avoid scrap and by the long time needed to print variant / new material by finding optimal printing parameters.

Linking visual features with product defects

Overcoming these quality challenges requires a high-speed visual system that is integrated into the additive manufacturing process. It should be capable of accurately monitoring the dynamics of the melt pool and its environment. Additionally, predictive modeling must be used to:

• correlate visual features of the product while it is built and melt pool events with visual defects in the final product; and
• enable an adaptive learning strategy to improve quality monitoring and control approaches.

A monitoring system that self-optimizes the printing process

The VIL consortium consists of 3D printing companies, system integrators and experts in system control, image and video processing, print modeling and X-ray-based industrial quality inspection. They will develop a “vision-in-the-loop” system that can monitor and control the additive manufacturing process in real time. The system will include:

• a high-speed, high-resolution imaging system to monitor the melt pool and its surroundings as well as to detect defects;
• a GPU acceleration platform that supports a high data rate video stream;
• algorithms that analyze visual features in difficult conditions and predictive models that correlate these features with product defects;
• control algorithms based on predictive modeling and simulations that optimize the printing process.

Higher productivity, less waste, higher accuracy

The solution developed by VIL will not just reduce additive manufacturing scrap and shorten production time; it will also improve the accuracy of the printed products. Additionally, the solution will allow end users to automate and improve product quality and gain valuable insights into process control.

“The Vision-in-the-Loop project will improve print quality, reduce waste and cut the cost of additive metal manufacturing by deploying AI to monitor and control the melt pool, product and printing process in real time.”