Quality control: how to link the Hardware and the Software ?

The uptake of Laser Powder Bed Fusion across many industries as the favored metal additive manufacturing technology highlights the need for an industrial oriented solution for quality control of metal 3D printed parts. DMP Monitoring and DMP Inspection tools bridge the gap between data process collection and relevant interpretation for the end-user. Those state-of-the-art solutions shorten iterations at the design and validation phases while ensuring a cost efficient quality control process during the production phase.


Laser Powder Bed Fusion is a manufacturing process in which a laser selectively melts metal powder, layer-by-layer, thus reconstructing in three dimensions a part. During the laser-metal powder interaction, many physical phenomena occur (melting, vaporization, solidification…), some of which may stochastically induce defects in the part being built (porosity or warping for instance). The extent of such defects, detrimental to the part mechanical properties, needs to be assessed for quality control purposes. Currently, the options available are either too expensive to be widely adopted, not precise enough or destructive and thus applicable to only a limited number of parts. This challenge, DMP Monitoring and DMP Inspection can help address in a systematic and cost-effective way on the DMP 350 and DMP 500 platforms.


The quality control of manufactured parts is usually divided between Destructive and Non-Destructive technics.

The former, of which tensile tests, hardness tests, metallographic tests are part of, cannot by nature, be applied to all manufactured parts. Conversely, relevant Non-Destructive technics (Archimedes density measurements, CT-scan, FPI) for additively manufactured parts, are too expensive to be utilized on a large scale production (CT-scan) or not sensitive enough to detect a low number of defects (Archimedes measurements). In this context, performing a cost-effective, systematic and performant quality control on parts remains a challenge.


The combination of DMP Monitoring for the in-situ process data collection and DMP Inspection for the data analysis and interpretation offers a quality control solution viable on a large scale while remaining economical and performant.

The solution envisioned to manage the quality of parts in a seamless way relies on the combination of DMP Monitoring for the data process collection and DMP Inspection for the data treatment. The result, a merge between the CAD part and the events detected, is readily accessible in 3DXpert®, the all in one additive software solution. The data collection through DMP Monitoring is made possible by two sensors based systems : DMP Vision and DMP Meltpool. The former, focusing on the powder bed quality, is implemented using a high speed and high-resolution industrial camera. The latter relies on the use of photodiodes recording meltpool data at 50 kHz. For a given layer, DMP Monitoring will capture two sorts of data :

  1. Powder bed images captured after powder deposition and after the laser scanning of the considered layer, for the purpose of assessing the homogeneity of the powder deposition and quality of the scanned surface.
  2. Spatially integrated light emission from the meltpool during the powder bed scanning. The emphasis here is on the detection of meltpool abnormalities following a defined set of rules (e.g. abrupt fluctuation in the signal recorded from a single scan vectors).

The process data recorded are then consolidated and analyzed by DMP Inspection, an add-on module to 3DXpert®. The integration of DMP Inspection extends 3DXpert® long list of features (CAD, part positioning, supports creation, lattices design, process simulation, CAM functions…) by providing the user with a seamlessly integrated root cause analysis tool. This solution ensures that all required functions from design to part quality control can be performed in a single environment, thus avoiding the hurdles associated with files transfer between softwares.

Figure 1 : Hardware implementation of the DMP Vision and DMP Meltpool solutions
Figure 2 : The DMP Monitoring interface as seen by the user during the manufacturing process. The far left window displays a live imaging of the printing process while central windows display the powder bed imaging (left) and meltpool imaging (right).

Beyond, the user-friendly workflow provided, the use of such solutions allows for a systematic part inspection, a procedure currently hardly applicable to a large production series, given the cost and time required for the data collection. Combined, the DMP Monitoring and DMP Inspection state-of-the-art tools, yield results highly correlated with that of a CT-scan (Figure 3). This versatile tool, which benefits are clear for mass series production of part requiring quality control procedure, is also applicable to the development of novel process parameters, often tedious and heavily relying of parts cross sections inspection for porosity characterizations.

Figure 3 : CT-scan and DMP Inspection porosities detection comparison. In this specific case (printing of test cubes in Ti-6Al-4V), a perfect match is exhibited between both technics for porosities larger than 200μm.

In order to further ease the data interpretation and the resulting decision regarding the considered printed part, an intuitive visualization of the results can be performed in 3DXpert® by overlaying the defects position to the original CAD part. In the example of a printed bracket (Figure 4), the CT-scan results in terms of porosity sizes and locations are presented in blue, the DMP Inspection results are presented in red, exhibiting the use of the system on an industrial part. This relevant formatting of the data eases the analysis of the defects severity, and help assess whether the defects detected could be detrimental to the part integrity because of their size or position.

Figure 4 : Overlay of detected defects during the printing process over the original geometry for a user-friendly data analysis


The DMP Inspection tool allows the user to access process-induced defects based on the melt pool signature or the powder bed quality.

Based on both metrics, the user can identify an extensive number of possible defects (lack of fusion, dross, warping, spatters, powder lumps, and coater lines). In order to ease the severity assessment of such defects, a color-coded scale and the location of the considered defects are displayed. With such built-in functions, DMP Inspection eases the quality assessment of L-PBF-induced defects and allows for a fast assessment of the quality of printed parts.

Appendix: a use case


The in situ process monitoring tools integrated into the DMP technology assist the quality control operations of additively manufactured parts.

The use case presented here details, step-by-step, the procedure used when analyzing the process data using DMP Inspection, a tool integrated into 3DXpert™. The parts considered here are 20 tensile bars printed in Ti-6Al-4V with a 60 µm layer thickness (Figure 1). For the purpose of the use case, the argon flow was altered in order to artificially induce defects in the parts.

After the four hours and twenty minutes required for the printing, the 10.2 GB process monitoring data acquired during printing are transfered and ready for analysis. The melt-pool data is stored under a proprietary format, while the powder bed deposition quality data is stored in PNG format. In order to perform the analysis, the user will follow the procedure explained below.

Figure 1 : 20 tensile bars as vizualized in 3DXpert during the file preparation
Although not detailed here, it is possible to extract from CT-scan data the defect positions and load them into DMP Inspection in order to compare the results of both technics.

Import and treat data

In this operation, the user acesses the Build Inspection Analysis option (Figure 2) to import melt-pool and powder-bed-related data. The options available at this stage are to perform a full or partial analysis (i.e., include melt-pool and powder bed quality data, or only one of the two), and decide which data sensor should be used when analyzing melt-pool data (left sensor, right sensor or both depending on the position of the part considered). The data treatment by the algorithm and the consolidation of relevant information took in this case less than six minutes on a regular computer, and the user now has access to the relevant process data.

Figure 2 : Import and analyze data procedure

Analysis of melt-pool data

During the data treatment run by the Build Inspection Analysis tool, a point cloud of abnormal melt-pool signatures is constructed. The high number of such signatures in a restricted area leads the algorithm to identify those areas as ones in which the likelihood of defect formation is high. The result is the creation of an object in the identified location, which the user sees under the form of a dot on the part (Figure 3), the scale of which can be chosen to ease the visualization of such events. This dot indicates a possible lack-of-fusion defect. In order to assist in the defect severity analysis, a color-coded scale indicates an estimation of the defect dimensions; the visualization of the defect position will help the user decide whether the defect size or location is critical to the part’s function.

Figure 3 : Tensile bars and high-probability defects’location as identified by DMP Inspection. Note the creation of defects as volume objects, which size is indicated by the color coded bar. In this case, lack-of-fusion defects are smaller than 0.056mm3.

The melt-pool data is also used to assess regions in which dross might have occurred. In this case, the user might notice a high number of events in over-hanging areas. The geometry and orientation of the parts here prevent dross from happening. For illustration purposes, Figure 4 shows a part prone to dross and the associated point cloud visua-lization. The high number of melt-pool-related events in an overhanging area is indicative of a high probability of dross formation, later visually confirmed on the printed part.

Figure 4 : For the purpose of illustration: a part prone to dross and the process monitoring data treatment by DMP Inspection

Analysis of powder bed quality data

The powder bed quality assessment during the printing leads to the identification of different types of defects, similarly using a point-cloud approach as for the melt-pool data analysis. Spatters, induced by the L-PBF process, can be identified by small clusters spread over the build platform. Poor deposition quality, due for example to the filling of the platform edges during the first layers, is easily identified (Figure 5) by combining the position (i.e., platform edges) of the defect with the associated severity. Powder lumps are identified by looking for bigger clusters, while coater induced line defects, are identified by looking for discontinued lines parallel to the recoating direction. Finally, warping is identified through large flagged areas inside parts. Figure 5 shows examples of such powder bed quality-related defects, with powder lumps and warping not being present on this build job.

Figure 5 : Powder-bed-related defects as displayed in 3DXpert™
©2020 by GF Machining Solutions and 3D Systems Inc, All rights reserved. The technical data and illustrations are not binding. They are not warranted characteristics and are subject to change.