How can AI overcome 3D printing flaws?
A growing number of academic and industrial research groups see the integration of artificial intelligence-based algorithms into the 3D printing process as a promising approach to improving the quality and efficiency of 3D printing technology.
Image Credit: Vladimir Panchenko/Shutterstock.com
While industrial additive manufacturing, known as 3D printing, has reached technological levels never seen before, quality control and monitoring of manufacturing defects is still a daunting challenge. .
Additive manufacturing plays an important role in a wide range of industrial sectors, such as automotive, aerospace, biomedical and others, by enabling the manufacture of parts with complex geometry, internal structural details and custom designs to low cost and in small production. lots.
However, at present, the ability of additive manufacturing technology to produce components with complex shapes with minimal material waste cannot be fully exploited when manufacturing the high quality parts required to meet demanding requirements. exacting standards. Depending on the type of 3D printing process involved in the production cycle, different manufacturing defects can compromise the consistency, reliability, and performance of 3D printed materials.
The factors that govern the quality of the 3D printing process are diverse. These can be related to the quality of the source materials used (plastic filaments, metal powder or liquid photopolymers) or related to the process, such as over- and under-extrusion, gas pockets in the sintered material, separation diapers (lack of adhesion), and others. In most cases, these defects result in increased porosity and lower mechanical properties of the fabricated part.
Establish process control
In the first place, ensuring that 3D printed components meet the required standards depends on the quality of the material used. Quality control of raw materials is an ongoing challenge for most additive manufacturing companies.
Additionally, a wide range of variables that can affect the quality of the final product cover the entire additive manufacturing workflow, from initial design to 3D printing and post-processing. For example, it may be the path and intensity of the sintering laser (in a direct metal sintering process) or the filament extrusion rate (in a fused filament manufacturing process). Other factors may include the design of the support structures or even the number of times the metal powder has been collected and reused.
Currently, trial and error is the most common approach to optimizing the 3D printing process to achieve consistent and repeatable component quality. This approach typically involves repeating manufacturing steps multiple times and extensive testing of the final part.
The result is an expensive and inefficient manufacturing process that negates the main benefits of additive manufacturing – cost-effective small batch manufacturing.
Reduce human error
Most additive manufacturing processes still require additional human intervention. This may involve initial component design, post-processing (removal of supports and finishing) and manual testing of final products to meet requirements. Thus, reducing the probability of human error is of crucial importance for the elimination of 3D printing defects.
Additive manufacturing experts recognize the need for more sustainable and feasible methods to process and control quality. One of the most promising ways to achieve this is to use artificial intelligence (AI) algorithms to automate the most critical steps in the 3D printing process.
Efficiency in the prefabrication phase
AI-based software packages, like Netfabb from Autodesk and Agile Metal Technology from Sculpteo (a subsidiary of BASF), can evaluate and optimize design files for 3D printing using machine learning algorithms in the generative design approach. Manufacturers can enter desired design parameters and AI analyzes design requirements to find the most efficient production path.
Automated fault detection and closed-loop control
The development of closed-loop control systems has been a long-standing key goal for additive manufacturing engineers, becoming a possibility in recent years due to the rapid development of advanced AI applications.
Researchers at General Electric’s Additive Research Laboratory in Niskayuna, New York, have developed a proprietary machine learning platform that uses high-resolution cameras to monitor the layer-by-layer printing process and detect streaks, pits, voids and other defects often invisible to the naked eye.
Data is compared in real time to a pre-stored defect database using computed tomography (CT) imaging. Using high-resolution imaging and CT scan data, the AI system can be trained to predict issues and detect defects during the printing process.
A similar integrative machine learning approach is being used by Ai Build, a London-based company specializing in the development of AI-based automated 3D printing technology, to create a smart extruder for additive manufacturing.
It is a high-precision accessory for industrial robotic arms capable of 3D printing large objects at high speed with high precision. By combining advanced artificial intelligence algorithms with real-time sensor data processing, the smart 3D printing extruder can detect any issues and make autonomous decisions to achieve the best possible print quality.
AI creates new 3D printing materials
A University of Cambridge spin-off company called Intellegens has used machine learning algorithms in its Alchemite platform to develop new materials for 3D printing. The company has successfully used the AI platform to create a new nickel-based alloy suitable for the direct laser deposition manufacturing process. Alchemite’s deep learning capabilities allow a large database of material properties to be used to identify the optimum alloy composition for the best processability and final product quality.
So far, the use of AI in additive manufacturing has focused on improving design, improving the efficiency of 3D printing processes, and enabling autonomous manufacturing. Soon, advanced AI solutions can help reduce design complexity, lower the knowledge threshold for additive manufacturing industries, and improve cybersecurity in the field.
References and further reading
Zhu, Z. et al. (2021) Multifunctional materials 3D printed using artificial intelligence-aided manufacturing technologies. Nat Rev Mater 6, 27–47. Available at: https://www.nature.com/articles/s41578-020-00235-2
Paraskevoudis, K., et al. (2020) Remote detection of real-time 3D printing defects (rope) with computer vision and artificial intelligence. Process 81464. Available at: https://www.mdpi.com/2227-9717/8/11/1464
K. Sertoglu (2020) Argonne scientists use machine learning to predict defects in 3D printed parts[Online] www.3dprintingindustry.com Available at: https://3dprintingindustry.com/news/argonne-scientists-use-machine-learning-to-predict-defects-in-3d-printed-parts-174544 (Accessed November 23, 2021)
C. Valdivieso (2020) Why combine artificial intelligence and 3D printing? [Online] www.3dnatives.com Available at: https://www.3dnatives.com/en/artificial-intelligence-and-3d-printing-060120204 (Accessed November 23, 2021)
C. Valdivieso (2019) Ai Build implements AI to detect and correct 3D printing errors in real time [Online] www.3dnatives.com Available at: https://www.3dnatives.com/en/ai-build-3d-printing-errors-110620195 (Accessed November 23, 2021)