We are currently experiencing a new paradigm shift that began in early 2011. The change is based on advanced digitisation within factories, using Internet-based information technologies and intelligent cyber-physical systems. This paradigm is known as Industry 4.0.
Additive manufacturing (AM), more commonly known as 3D printing, comprises several manufacturing technologies. These can build three-dimensional objects of virtually any geometrical shape from a digital model and use a layer-by-layer fabrication technique. AM has been successfully used as a new technology to produce complex components with improved properties compared to traditional manufacturing.
Despite its promising advantages, AM is hindered by various inherent scientific and technological challenges, limiting the potential exploitation of their many possible industrial applications. For example, certain defects such as dimensional nonconformance, porosity, cracking, distortions, residual stresses, and low mechanical properties result from an inadequate or absence of monitoring and control strategy. The scientific community has recently placed a particular interest in monitoring and controlling the LMD (Laser Metal Deposition) processes to overcome these defects.
The challenges
In LMD technology, the raw material (powder) is injected into the head via a coaxial or off-axial nozzle with an inert carrier gas (e.g., argon) during the cladding process. The powder flow and the laser are focused on the same substrate area where the deposition or coating will take place. Both are moved to the specified coordinates by a robotic arm or a computer numerical control (CNC) system. On the other hand, molten pool commonly refers to the portion where the raw material (in solid-state) has reached its melting point (in liquid-state). The molten pool is central to the success of the cladding process. The molten pool must be carried along the deposited track in a consistent width and depth. The geometry, gradient temperature, and fluid dynamics directly affect the quality of the cladding track.
Figure 1: Monitoring meta-sensor at CIDESI. The set-up comprises a Phantom VEO 710 high-speed camera (on the right), infrared camera, and two-color pyrometer. The acquired data is used to characterise and control the LMD process.
It is well-known that the interaction between the molten pool, the powder stream, and the substrate (including the influence of their respective properties) is crucial in LMD processes. The study of these interactions can only be performed using special sensors capable of digitalising in high-acquisition rates. For instance, many phenomena determine the quality of the printed component. Several of them occur at high-speed rates. There are two features of interest in real-time monitoring of LMD:
Powder flow dynamics
Another key-process parameter is the powder flow. Related to the amount of raw material provided per unit of time (g/min, for example). Keeping the mass flow constant and well distributed between the nozzle and the substrate is one of the most recurrent problems in any LMD experiment. The scientific community has shown interest in developing systems that allow in-process inspection of the amount of material supplied in LMD processes. With the information obtained from high-speed monitoring, it is more feasible to guarantee the repeatability of the processes (for example, the same characteristics in deposited tracks).
Molten pool dynamics.
The size (e.g., area and diameter), shape, temperature (average or cumulative), and temperature distribution (1D temperature profiles or 2D maps in a defined area) determine the molten pool behavior. These characteristics influence porosity, lack of fusion of particles in the manufactured material, residual stresses, cracks, and layer delamination. As a result, molten pool monitoring represents a challenge in real-time (in-situ) process monitoring as well as in the development of nondestructive AM evaluation techniques. A robust monitoring system must measure the molten pool characteristics in a very small area (a few hundred microns in diameter) and with a sufficient spatial resolution to describe any process variations. Also, a very high sampling rate is required since the various phenomena driving the LMD processes are very fast (e.g., solidification, phase transformations).
The Center for Engineering and Industrial Development (CIDESI), in Querétaro, Mexico, is developing frontier science to solve many of the technological milestones that MA has today. In its Additive Manufacturing and Materials Development (FADMAT) department several technologies are deployed (WAAM, LMD, DIW, FDM, Stereolithography, to name a few) that cover 5 of the 7 categories defined by the American Society for Testing and Materials (ASTM). This makes it the only AM-HUB in LATAM. A research team at FADMAT is related to studying the phenomena of laser-matter interaction in AM processes through active monitoring and control. Dr. Ángel Iván García Moreno leads this group. The incorporation of the Phantom VEO 710 camera into the monitoring tasks has made it possible to observe and understand the behavior of the dynamics of the molten pool and powder flow. Figure 1 shows our 3-port LMD nozzle, it was instrumented with several sensors.
Monitoring Meta-sensor
While the use of AM has been growing, a number of challenges continue to impede its more widespread adoption, particularly in the aerospace, automotive, and biomedical industry sectors. To address these technical issues, high-speed cameras like the Phantom VEO manufactured by Vision Research Inc. (a division from Ametek), have a leading role by facilitating the in-process visualisation of the behavior of all physicochemical phenomena that occur during the additive manufacturing processes.
Our proposal is a multi-sensor monitoring platform (see Fig. 2). Composed of sensors (cameras and pyrometers) with the ability to acquire data in the visible and infrared wavelengths (both 1D and 2D). The acquisition platform will be composed of multiple networked sensors with communication capabilities where the whole net can be thought of as a meta-sensor that can be controlled. Each node (sensor) will have certain degrees of freedom that will allow the system to self-calibrate. Depending on the task or query, it is desirable for the monitoring system to control the data acquisition process so as to acquire the “most informative data” for the specific task or query. This platform seeks to implement techniques and methodologies that address sensor-selection, modality selection, and active observation for real-time assessment and improvements of sensing performance.
Figure 2: Meta-sensor scheme. The LMD process is monitored by a multi-sensor platform (meta-sensor). All data is processed in a high-performance computing cluster (HPC). An intelligent system analyses the critical patterns found in the data, and with this information controls the peripherals in our LMD cell to improve the deposition task. The goal is to ensure a stable and repeatable process.
Phantom VEO camera and LMD
Powder flow dynamics: The powder flow in LMD process is developed by a carrier gas that allows the flow of metallic powder to a focal point, this zone is known as the convergence distance. This convergence distance depends on several factors related to raw material such as morphology, density, and mass flow.
The study of the convergence zone allows better usage of metallic powder during the LMD process and its interaction with the laser beam to be more effective. For this reason, by means of a high-speed camera capable of following the powder flow at the discharge of the nozzle at the speed that this phenomenon develops, it is possible to determine through computational algorithms (e.g., image analysis and PIV) the tracking of the powder particles when they leave the nozzle and with this, the speed that they develop. This information makes it possible to identify the specific convergence zone for the nozzle by the powder stream behavior. See Fig. 3 top row, data was taken with a Phantom VEO 710 camera at a frequency of 8000 Hz at acquisition resolution 1280 x 720.
Melt pool dynamics: The interaction between the powder flow and laser beam allows a cladding process on substrate with molten particles. In this interaction, the molten pool of LMD is generated which triggers the formation of cladding and with them the ability to develop 3D printing. Information gathered through a high-speed monitoring system can be used, among other purposes, for detecting unusual dynamics in the molten pool and its impact on the manufacturing process. For example, data provided by the Phantom camera is used to identify geometrical patterns on the surface of additively manufactured parts, which were subsequently correlated to lack of fusion. On the other hand, the molten pool dynamics and its relation with the edge of the track as well as with the penetration depth can be determined. See Fig. 3 bottom row, data was taken with a Phantom VEO 710 camera at a frequency of 4000 Hz at acquisition resolution 1280 x 720.
In laser welding processes, it is of equal interest to be able to monitor the behavior of the welding pool.
Figure 3: CIDESI's technological capabilities in LMD processes include many deposit nozzles, the powder is dosed in a coaxial way. During the flight of the particles, it is interesting to know the trajectory they follow and the velocity they develop, which are relevant variables for the cladding process. Also is of interest to analyse the influence between the powder flow, the laser beam, and the molten pool dynamics with the mechanical properties. Therefore, the first step is monitoring and understanding the fluid-flow behaviour on the molten pool, in our case, using a Phantom VEO 710.
Biography
Dr. Ángel Iván García Moreno is a researcher/professor at the Center for Engineering and Industrial Development (CIDESI). Member of the Cátedras-CONACYT initiative from the National Council on Science and Technology (CONACYT). He leads the Computer Vision and Artificial Intelligence research lines applied to Additive Manufacturing, specifically for Laser Metal Deposition technology. His research projects seek to understand the physical laser-matter interactions in DED processes by developing new image analysis algorithms and Machine Learning models. Also has interest in developing new Non-Destructive Evaluation techniques to validate the quality and performance of the additively manufactured components. García is member of several Consortiums (e.g. Additive Manufacturing Consortium from CONACYT) and National Laboratories (e.g. National Laboratory of Thermal Projection). He is member of the National System of Researchers (SNI) of CONACYT, level 1.
Contact: angel.garcia@cidesi.edu.mx
MSc Aldo López-Martínez is a PhD candidate in Additive Manufacturing by Laser Metal Deposition (LMD). He works as an Associate Researcher at the Center for Engineering and Industrial Development (CIDESI) in Additive Manufacturing and Materials Development Area (FADMAT). He has extensive experience in the technical and administrative execution of technology development and innovation projects in several industrial sectors. He is working on characterisation of the LMD process by exploring the powder stream behavior at the nozzle discharge and the evolution of the molten pool, in addition, he is developing various numerical models through Computational Fluid Dynamics (CFD) of the LMD process, validating them by the experimental data obtained.
Contact: aldo.lopez@cidesi.edu.mx
Jorge Villanueva , Solutions integrator and multicultural project leader with more than 20 years of experience. Computer Engineer Bachelor Degree, Graduated at Engineering Faculty UNAM. CEO and Founder of SLAM Solutions. Exclusive distributor of Vision Research Phantom cameras for Mexico, Central America and the Andean region since 2008.
Contact: jv@corpslam.com
Meet Jorge at the Testing & Simulation Live Event in Mexico. - November 2022, visit the Live Events section.
Comments