The first time a 3D-printed metal part was used on a commercial aircraft was back in 2014. At that time, the Airbus company produced a titanium bracket on a 3D printer and installed it on an Airbus test aircraft. Thanks to that successful rapid prototype created by Airbus, the entire industry begun to evolve.
Ensuring reliable product quality in sensitive industries like aerospace is crucial. For example, in metal-based 3D printing, the entire process must be monitored from end-to-end. Keeping track of the stock material, the melt pool, the changes in temperature, object’s dimensions, etc. is vital to ensure quality. In this article we will discuss how Lanner Intelligent Edge platforms are used in remote monitoring solutions for 3D printers.
The process of industrial 3D printing takes time and requires personnel to be onsite, continually monitoring for failures and job completion. Although video surveillance may help with remote monitoring, it does come with many limitations. Video data can be resource-demanding; it generates massive data, requires large bandwidth, and can be intensive to process. Plus, with video, there is no way to automate monitoring. There still needs to be a set of human eyes behind a monitor.
We will be reviewing a 3D printing monitoring solution that helps resolve manufacturers’ challenges. It uses cameras, intelligent edge hardware (Lanner LEC-2290), and monitoring and analysis software.
The solution presented in this brief meets the needs of different industries using additive manufacturing, from turbine and engine construction, orthopedics, aerospace, production parts, tools, etc.
It is common to see product defects and dimensional alterations while 3D printing. But these failures must be minimized or even eliminated when manufacturing parts for sensitive industries like aerospace and defense.
Monitoring industrial 3D printing can be challenging, especially serial additive manufacturing processes that deal with challenging materials like metal or polycarbonate.
The following are two common challenges when monitoring industrial 3D printing processes:
Monitoring the entire printing process in real-time is challenging
The 3D printing must be monitored throughout its entire process to avoid wasting printing material, operator’s time, and ensuring product quality. Although some 3D printers come with simple built-in monitors, there are currently no advanced 3D printer real-time remote control and monitoring systems. Keeping an eye on failures and job completions during 3D printing processes needs to be done on-site, either by operators or video cameras.
Monitoring industrial additive manufacturing can be resource-demanding
One solution to monitor the entire 3D printing process in real-time is with remote video surveillance. This solution works well for small productions, but it can be challenging to scale for larger and more sensitive productions. The amount of video data generated by more than six 3D serial printers can be overwhelming to transmit and analyze.
Additionally, sending all this data for external processing at the cloud would require large bandwidth and low-latency communications.
The solution presented in this brief helps resolve the common challenges of monitoring industrial additive manufacturing processes.
It continuously tracks the external conditions of the 3D printing production using cameras or sensors. The solution uses Machine Learning or Vision algorithms to improve the monitoring precision and automate the entire process. When a normal value deviates from a predefined threshold, the process may be stopped and the problem quickly notified to the manufacturer. The solution may also employ thermal imaging to help with the monitoring of the melting process of metal-based 3D printing.
The solution encompasses the following different elements.
Cameras and Sensors
Most of today’s 3D printers come with a built-in monitoring system. These printers have an integrated sensor located around the nozzle to detect a lack of feedstock material or nozzle’s failures— not much different from 2D printers. But to expand to broader forms of monitoring, a 3D printer may be well combined with sensors or external cameras that collect data about the 3D printing process.
For example, thermal imaging cameras may help gather different metal powders’ melting temperatures during the printing process. Or comprehensive optical scanning systems with multiple cameras can provide an object’s dimensional and geometrical analysis with the help of machine vision algorithms.
Hardware – Intelligent Edge Box
Lanner’s LEC-2290 is the brain of the monitoring operation. It is a crucial component for monitoring, analyzing, and controlling industrial 3D printing processes.
Lanner’s LEC-2290 is a compute-intensive intelligent edge computer that runs on-site. It executes resource-demanding workloads and is optimized for imaging analytics. The LEC-2290’s capacity is perfect for executing computer vision algorithms. It can even be expanded with additional GPUs, VPU, and FPGA for powerful video processing.
LEC-2290 Additional Highlights:
- Support for Intel® Core™ i7-8700T/i7-8700. LEC-2290 is based on Intel Core i7-8700T/i7-8700 to provide higher energy efficiency and execute applications faster.
- GPU expansion. Includes an integrated graphics card, Intel® UHD Graphics 630, to improve graphics performance. But GPU capacity may also be expanded with more GPUs, or also VPUs and FPGA.
- Fast memory and storage. LEC-2290 comes with two DDR4 2133/2400 SO-DIMM with a maximum capacity of 32GB. This type of memory helps the CPU and GPUs to change quickly between image execution processing.
- Support for the Intel OpenVINO toolkit and Movidius inferencing engine. OpenVINO (Open Visual Inference and Neural network Optimization) allows the rapid development of applications that model the human eye. The Movidius is a Vision Processing Unit (VPU) that enables demanding edge computer vision and AI workloads to run efficiently.
Monitoring and Analytics Software
The other vital component in the additive manufacturing monitoring solution is the software. Today, most industrial 3D printing monitoring software runs on the cloud. The on-premises cameras feed their data to a cloud-based system that run the intelligent workflows.
But keeping the data for processing on-site is critical for certain industries, where low latency monitoring and security is concerned. To run most processing on site, the software runs on top of the intelligence edge appliance. The 3D printing optimized software may perform the following (although not limited to):
- Provide the intelligence. Machine learning and machine vision algorithms.
- Provide deep insights into the production parts and prototypes in real-time.
- Raise alarms, alerts, or execute operations that pause/stop the printing process.
- Provide video surveillance of the printing process remotely and in real-time.
Collecting 3D printing process data processing and analyzing it on-site provides fantastic benefits. These benefits include real-time and remote monitoring, improving processes and ensuring quality, predictive maintenance of equipment, and improved security.
- Real-time remote monitoring. Real-time communications over remote distances require one thing: low latency. Fewer data traveling back and forth will make real-time communications more “real”.
With the 3D printing monitoring solution, the processing and analysis of the data collected by cameras and sensors are performed on-site. So, there is no need to send out tremendous amounts of data to an offshore data center.
- Security. The intelligent edge appliance processes and analyses a significant percentage of data on-site. This solution reduces the amount of data that needs to be sent to a cloud-based monitoring server, eliminating the risk and exposure of Internet-based transmissions.
- Ensure product quality and improve manufacturing processes. In additive manufacturing assembly lines (especially for sensitive industries like aerospace), meeting high-quality production standards is often carefully controlled by compliance.
Applying Machine Learning or Machine Vision (MV) algorithms will help ensure compliance and product quality when manufacturing with 3D printers. These algorithms allow the printer to take intelligent corrective actions through an automated monitoring process. For example, in serial metal additive manufacturing, using thermal imaging plus MV helps provide comprehensive monitoring of the materials’ melting reaction.
- Predictive Maintenance. The intelligent 3D printer monitoring solution also helps employ predictive maintenance on the printers. Predictive maintenance helps personnel proactively keep their machines working. It predicts when the equipment might fail so that the right maintenance procedure can be performed beforehand. Predictive maintenance is based on the data collected by sensors or cameras, and the evaluation performed by the intelligent edge box.