Additive manufacturing process

Abstract

An additive manufacturing process comprising: obtaining a command suitable for use by a machine to manufacture an object by selective laser melting and comprising elementary commands, each elementary command comprising: location data enabling the machine to locate a line having a starting point and an ending point, laser parameters of the machine for melting powder grains along the line from the starting point to the ending point; for each elementary command, constructing (102) an associated elementary 3D model representing a melt pool; constructing (104) a 3D model representing the object by union of the elementary 3D models; checking (106) the 3D model to verify whether the object represented by the 3D model meets quality criteria; and if so, instructing the machine to manufacture (108) the object from the command. Figure for the abstract: Fig. 4

Description

Title of the invention: Additive manufacturing process FIELD OF INVENTION

[0001] The present invention relates to the field of additive manufacturing. STATE OF THE ART

[0002] Selective laser melting is an additive manufacturing technique for an object using, as its name suggests, a laser that fuses powder grains stacked layer by layer.

[0003] Generally, an object to be manufactured using this technique is modeled in 3D by computer. A user can view, via appropriate software, a 3D model of the object before starting its manufacture.

[0004] A laser fusion additive manufacturing machine cannot directly interpret such a 3D model. It is first necessary to convert such a 3D model into a command that the machine can interpret. Conventionally, such a command indicates to the machine a very large number of small lines along which the laser must fuse powder grains. These small lines are distributed in different layers of powder.

[0005] However, this conversion operation results in a loss of information, since it transforms a three-dimensional shape into a series of small lines. This loss of information is likely to cause defects in the manufactured object. Description of the invention

[0006] One problem to be solved is that of reliably anticipating the appearance of defects during additive manufacturing by selective laser melting.

[0007] This problem is solved by an additive manufacturing process comprising: • obtain an additive manufacturing command suitable for use by a machine to manufacture an object by selective laser melting, the additive manufacturing command including: • Basic commands specific to be used by the machine to manufacture different parts of the object, each basic command comprising: • Location data enabling the machine to locate in space a line having a starting point and an ending point, • laser parameters intended to be used by a machine laser to fuse powder grains along the line from the starting point to the ending point, • For each elementary command, construct an elementary 3D model associated with the elementary command representing a melt pool resulting from the melting of powder grains along the line from the starting point to the ending point, by a laser using the laser parameters, • construct a 3D model representing the object by uniting the elementary 3D models respectively associated with the elementary commands, • Check the 3D model to verify if the object represented by the 3D model meets quality criteria, • when the object represented by the 3D model meets the quality criteria, command the machine to manufacture the object by selective laser melting from the additive manufacturing command.

[0008] Unlike analysis tools that rely solely on an ideal 3D model of the object from which the command is generated, this approach proposes to leverage information that closely reflects what actually occurs during selective laser melting, by exploiting the content of the command. With this level of information, it is possible to make deterministic local predictions at scales fine enough to capture relevant information for predicting the material health or surface condition of the object to be manufactured.

[0009] The method may also include the following optional features, taken alone or in combination where technically feasible.

[0010] Preferably, constructing an elementary 3D model associated with an elementary control includes: determining a profile of the melt pool in a cutting plane normal to the line, and generating the elementary 3D model from the profile of the melt pool under the assumption that the profile of the melt pool evolves along the line according to a predefined law.

[0011] Preferably, the elementary 3D model is generated from the profile of the melt pool under the assumption that the profile is constant along the line.

[0012] Preferably, determining the estimated profile of the melt pool includes a search, in a database associating reference commands and reference profiles, for a reference profile associated with a reference command satisfying a criterion of correspondence with the elementary command.

[0013] Preferably, when the object represented by the 3D model does not meet the quality criteria, a new additive manufacturing command is generated specifically for use by the machine to manufacture the object by selective laser melting, the new manufacturing command being different from the additive manufacturing command.

[0014] Preferably, controlling the 3D model includes: determining a volume of the object represented by the 3D model, and comparing the volume and a volume of the object represented by an original 3D model from which the command was previously generated.

[0015] Preferably, controlling the 3D model includes: determining a roughness of the object represented by the 3D model, and comparing the roughness with a predefined roughness threshold.

[0016] Preferably, controlling the 3D model includes: comparing a maximum intersection arity between the elementary 3D models and a predefined arity threshold.

[0017] A second object of this disclosure is a computer program product comprising program code instructions for performing the steps of the process described above, when this program is executed by a computer.

[0018] A third subject of this disclosure is a system comprising a selective laser melting additive manufacturing machine, and a control unit configured to control the additive manufacturing machine by implementing the process described above. DESCRIPTION OF THE FIGURES

[0019] Other features, objectives and advantages of the invention will become apparent from the following description, which is purely illustrative and not limiting, and which should be read in conjunction with the accompanying drawings on which:

[0020] Fig. 1 schematically illustrates a system according to one embodiment.

[0021] Figure 2 schematically represents layer-by-layer additive manufacturing successive.

[0022] Fig. 3 schematically represents a trajectory followed by a laser on the surface of a powder layer.

[0023] The [Fig.4] is a flowchart of steps of a process according to an embodiment.

[0024] Figure 5 represents a strategy for scanning a layer with a laser, according to a method of implementation.

[0025] Figure 6 details one embodiment of one of the steps shown in the [Fig.4],

[0026] Fig. 7 represents a profile model of a melt pool in a cutting plane perpendicular to a laser scanning direction.

[0027] Fig. 8 represents a profile model of a melt pool in a cutting plane parallel to a laser scanning direction.

[0028] Fig. 9 is a view of a 3D model reconstructed during an implementation of the process of Fig. 4.

[0029] Fig. 10 is another view of the 3D model of Fig. 9, superimposed with another model, and also shows a difference between the two aforementioned models.

[0030] The [Fig. 11] is a view of a set of elementary 3D models that are superimposed on each other.

[0031] Throughout the figures, similar elements bear identical references. DETAILED DESCRIPTION OF THE INVENTION

[0032] With reference to [Fig.1], a system comprises an additive manufacturing machine 1 by selective laser melting and a control device 2.

[0033] The additive manufacturing machine 1 by selective laser melting is known per se. The machine comprises a support 10, an applicator 12 for applying superimposed layers of powder onto the support so as to form a stack of layers, and a laser 14 suitable for emitting a beam directed towards the stack of layers.

[0034] With reference to [Fig.2], the machine 1 is configured to stack layers of powder to be fused in a stacking direction parallel to a z-axis. Each layer extends in a plane parallel to x and y axes. The x, y and z axes form a Cartesian coordinate system with origin O.

[0035] The additive manufacturing machine 1 further includes an input interface 16 suitable for receiving an additive manufacturing command, and a controller 18 for controlling the applicator 12 and the laser 14 from the additive manufacturing command, so as to manufacture an object.

[0036] The control device 2 includes a processor 20, a memory 22, and a control interface 24.

[0037] The device may further include or be connected to a display screen 26.

[0038] Memory 22 stores a computer program comprising instructions for code executable by processor 20, this execution causing the implementation of a process whose steps will be described later.

[0039] The processor 20 is configured to execute the aforementioned computer program. It can have any structure, in particular having multiple cores (to perform calculations in parallel) or a single core. The processor 20 can be a CPU, a GPU, an ASIC, an FPGA, or any other type of circuit.

[0040] In one embodiment, the processor 20 has access to a database that associates elementary reference commands with melt bath profiles. This database will be described in more detail later. This database can also be stored in memory 22 or in memory external to the control device 2.

[0041] The control interface 24 is intended to communicate an additive manufacturing command to the additive manufacturing machine 1. Additive manufacturing order format

[0042] The additive manufacturing control comprises a plurality of elementary controls respectively associated with different portions of the object to be manufactured.

[0043] Each elementary command includes location data and laser parameters.

[0044] The location data enabling the machine to locate in space a line having a starting point A and an ending point B. An example of a line forming a segment (straight line) is shown in [Fig.3].

[0045] The location data includes a starting point position A and an ending position B. These two positions can, for example, be expressed by Cartesian coordinate triplets, for example in the (x, y, z) coordinate system discussed previously. In this case, points A and B will have the same z-coordinate, since points A and B belong to the same layer.

[0046] In a simple implementation, the location data consists of the position of the starting point A and the position of the ending point B, and the machine is configured to trace, using the laser, a segment between the starting point A and the ending point B, in other words, a straight line. In other embodiments where the line is not necessarily a segment but may, for example, be curved, the location data may include other descriptive data about the line between the starting point and the ending point.

[0047] The laser parameters are intended to be used by the laser 14 of the machine 1 to fuse powder grains along the line from the starting point A to the ending point B.

[0048] The laser parameters may include a laser power to be used to scan the line and / or a beam scanning speed along the line (i.e., a speed at which a spot projected by the laser beam moves on a top layer of a layer stack deposited on the support 10 of the machine 1.

[0049] The table below summarizes the content of an elementary command according to one embodiment. Location data Laser parameters Starting point coordinates A Ending point coordinates B *A yA ^A XB yB PV

[0050] In the additive manufacturing control, the different elementary commands can be ordered so as to indicate to the machine 1 the order in which they should be processed.

[0051] We will now describe a process implemented by the system. Additive manufacturing order generation

[0052] It is assumed that the control device 2 has previously received an input 3D model representing the object to be built. The 3D model was, for example, created using computer-aided design (CAD) software.

[0053] With reference to [Fig.4], the processor 20 generates in a step 100 the additive manufacturing command from the input 3D model.

[0054] This step 100 is known from the prior art. Step 100 may include the following substeps. The input 3D model is meshed to obtain a 3D mesh. The 3D mesh comprises a set of points, called vertices, whose respective coordinates are known, and includes edges that connect the points to form faces. Each face helps to define the object. The mesh is then sliced ​​into parallel sections, these sections representing the layers of powder that the additive manufacturing machine 1 will stack to manufacture the object. Each section has a position in the stacking direction z. This slicing is a discretization in the stacking direction, inducing a loss of information (information about the object between the interlayers is lost). Then, for each section, elementary commands relating to the section are generated.Here again, we can consider that there is discretization, since a "full" slice is transformed into a set of lines without thickness. The elementary commands are ordered in a defined direction according to a scanning strategy known in advance.

[0055] Figure 5 illustrates an example of a scanning strategy, which consists of dividing a slice into a checkerboard pattern, with the squares of the checkerboard being alternated between laser scans in a direction parallel to the x-axis and laser scans in a direction parallel to the y-axis. It can also be seen that the order in which these squares are merged is not continuous, but that the laser beam is made to merge squares of the slice that are not adjacent.

[0056] Generation of a reconstructed 3D model from the command

[0057] Returning to [Fig. 4], the processor 20 constructs, in step 102, from one of the elementary commands, an associated elementary 3D model representing a melt pool resulting from the fusion of powder grains along the line indicated by the elementary command, assuming that the laser parameters of the elementary command are used to perform this fusion. As is known, the term "melt pool" refers to a set of grains that fuse together under the effect of a laser. The melt pool has an elongated bead shape. It is understood that step 102 is a simulation step, which does not involve the additive manufacturing machine 1; this step merely estimates in advance the three-dimensional shape that the melt pool would take if it were generated by the machine 1 and applied the elementary command.This step is essentially equivalent to restoring thickness to the line indicated by the elementary command.

[0058] During step 102, an elementary 3D model is constructed for each elementary command. At the end of this step, the processor 20 therefore obtains a plurality of elementary 3D models of melt baths, which are respectively associated with elementary controls.

[0059] An embodiment of construction step 102, comprising the following substeps, is shown in [Fig.6].

[0060] In a substep 200, the processor 20 determines a profile of the melt pool in a cutting plane normal to the line indicated by the elementary command.

[0061] To this end, the processor 20 searches the previously discussed database for a reference command that satisfies a predefined matching criterion with the elementary command. Once such a reference command is found, the processor 20 selects the reference profile associated with the found reference command as the melt pool profile.

[0062] The melt pool profile determined by the processor 20 can be represented by data having various formats.

[0063] In an advantageous embodiment, the molten pool profile is delimited by two joined semi-ellipses, as shown in [Fig. 7]: an upper semi-ellipse delimiting an apparent portion of the molten pool, and a lower semi-ellipse delimiting a remelted area. Such a profile provides a good approximation of an actual molten pool profile generated during selective laser melting. Such a profile can be represented by three parameters. For example, these three parameters are as follows: • e“PP: width of the melt pool in a direction perpendicular to the line and perpendicular to the direction of stacking of the powder layers; this width is twice half the axis of the upper semi-ellipse and half the axis of the lower semi-ellipse. • Happ; height of the apparent part of the molten pool in the direction of layer stacking; this height is equal to the other half-axis of the upper half-ellipse. • ^zr: height of the remelted zone in the stacking direction of the layers; this height is equal to the other half-axis of the lower half-ellipse.

[0064] In another embodiment, the profile of the molten bath is a square or rectangular profile, preferably inclined at 45 degrees with respect to the stacking axis of the layers. Such a profile can be represented by one parameter (side length of the square) or by two parameters (lengths of the longer and shorter sides of the rectangle).

[0065] In other embodiments, the profile can be defined by a set of points which together delimit the contour of the melting bath in the cutting plane considered.

[0066] The database was previously established. The reference profiles for the melt pools may have been determined experimentally, for example by metallography, and / or by numerical simulation. The experimental and numerical approaches are complementary: the database may consist of data from both physical and numerical measurements. Furthermore, it is possible to obtain a continuous database from a set of discrete points, by interpolating existing points or by defining a response surface with a meta-model.

[0067] Returning to [Fig. 6], the processor 20 generates in a substep 202 the elementary 3D model representing the molten pool from the molten pool profile, and under the assumption that the molten pool profile evolves along the line according to a predefined law. This step can be considered as a virtual extrusion using a die having the molten pool profile.

[0068] In a computationally inexpensive embodiment, it is assumed in this step 202 that the profile is constant along the line from the starting point A to the ending point B. By this, we mean that for any point on the line indicated by the elementary command, the profile of the elementary 3D model in a cutting plane normal to a tangent to the line at that point is the same. If the line considered is a segment, then the 3D model has a cylindrical shape whose cross-section corresponds to the profile determined in the previous step.

[0069] In another, more complex but also more realistic embodiment, the predefined law is not a constant law, but rather an expansion law. In other words, the further one moves along the line from the starting point to the ending point, the larger the area of ​​the profile becomes. As an example, a simulation showing the evolution of the height of a melt pool is shown in [Fig. 8]. It can be seen that this melt pool has an increasing height from left to right (which corresponds to the direction of scanning of the laser beam).

[0070] In one embodiment, the line can be considered to consist of three portions: an initial portion starting from the starting point A, an intermediate portion that extends the initial portion, and a final portion that extends the intermediate portion and terminates at the arrival point B. Different evolution laws can be used for these three portions. In particular, the profile can be considered constant in the intermediate portion, but not constant in the other two portions.

[0071] At this stage, the processor 20 has ultimately built as many elementary 3D modules as there are elementary commands. It should be noted that the elementary 3D models intersect. This reflects the fact that during additive manufacturing, a grain is generally fused several times. Not only can such a grain be fused by the laser as soon as it is deposited, i.e., when it is part of the top layer of the layer stack, but it can also be brought to be refused after being covered by a new layer of powder. Indeed, we see that a melt pool extends vertically, and therefore includes buried grains. Similarly, a melt pool extends in the plane of the last layer deposited: a grain of the last layer can therefore be fused during several passes of the laser over this layer, in close positions.

[0072] Returning to [Fig. 4], the processor 20 applies a Boolean union operation to the various 3D model elements obtained in step 104. This Boolean union operation is known from the prior art. The result of this Boolean operation is an overall 3D model representing the object to be manufactured.

[0073] The reconstructed overall 3D model represents the same object as the input 3D model, but these two models have slightly different shapes. These differences are due to information loss during the additive manufacturing command generation step (discretization). This information loss can cause imperfections to appear during the machine's manufacturing of the object. However, these imperfections are modeled by the reconstructed 3D model.

[0074] The processor 20 can control the display of the reconstructed 3D model on a display screen, so that a user can observe such imperfections. The 3D model constitutes an information medium that is easy for a user to interpret, unlike the additive manufacturing command, which is unreadable to the naked eye.

[0075] Figure 9 shows an example of a reconstructed 3D model M1 obtained via step 104, from a set of elementary commands relating to lines forming parallel segments. In this example, the weld pool profiles are of the two-half-ellipse shape discussed previously. Bumps can be observed on the surface of this reconstructed 3D model; these bumps originate from the half-ellipses.

[0076] In a step 106, the processor 20 checks the 3D model to verify whether the object represented by the 3D model meets predefined quality criteria. This check 106 results in two possible alternative outcomes: an OK check result, indicating that the object meets the quality criteria, or a KO check result, indicating that the object does not meet the quality criteria.

[0077] When the control result is OK, the processor 20 can command the machine 1, via the control interface 24, to manufacture the object by selective laser melting from the additive manufacturing command (step 108).

[0078] When the control result is KO, the processor 20 generates a new additive manufacturing command suitable for use by the machine to manufacture the object by selective laser melting, the new manufacturing command being different from the additive manufacturing command.

[0079] To generate this new command, the processor 20 can repeat step 100, starting again from the input 3D model, after modifying a parameter involved in the implementation of step 100 (for example, the scanning strategy used or the orientation of the stacking axis, i.e., the direction in which the input 3D model is sliced). Alternatively, the processor 20 generates the new command directly from the existing command, by directly modifying all or part of the elementary commands that constitute it. For example, the processor 20 can modify only the elementary commands that selectively relate to a particular region of the object, which will have been selected by an operator. For example, the processor can modify the laser parameters (in particular the laser power and / or the laser scanning speed).

[0080] Next, the process steps are repeated based on the new order. This step repetition is implemented until the OK control result is obtained. Examples of quality controls

[0081] During quality control step 106, one or more quality criteria may be examined, each quality criterion relating to a specific property of the object as represented in the reconstructed 3D model. Volume

[0082] A first quality criterion concerns the volume of material of the object. To evaluate this first criterion, the processor 20 determines the volume of the object as represented by the 3D model, and compares this volume with the volume of the object as represented by the input 3D model, which was originally used to generate the additive manufacturing command.

[0083] This comparison can, in particular, be accompanied by the calculation of a volume difference between these two volumes. This volume difference can indicate that the reconstructed 3D model has excess material compared to the input 3D model. In this situation, the processor 20 considers that if the object were manufactured according to the command, then this manufacturing process would introduce excess material compared to what was initially desired. Alternatively, the aforementioned volume difference can indicate a material defect.

[0084] In one embodiment, the processor 20 applies a Boolean subtraction operation between the reconstructed 3D model and the input 3D model, and can then display the resulting differential model. This operation makes it possible to retain only the aforementioned excesses or defects of material, and thus to detect them more effectively. As For example, [Fig. 10] represents the reconstructed D3 model M1, already shown in [Fig. 9], superimposed on a portion of a straight-edged input 3D model MO. The differential model M1-M0, shown to the right of [Fig. 10], highlights the fact that there is a material defect in the reconstructed 3D model M1. Therefore, there is a risk of a material defect in the object if it is manufactured using the command.

[0085] Furthermore, the presence of closed cavities within the reconstructed 3D model Ml can be detected. Such cavities are generally undesirable in additive manufacturing, since they constitute areas where powder is deposited but not fused by the laser. Consequently, once the object is fully manufactured, these areas trap unfused powder, which cannot be removed. In particular, if such cavities are present in the reconstructed 3D model Ml but not in the original model MO, there is a risk that the object will have degraded mechanical properties if it is subsequently manufactured using the additive manufacturing command. Surface condition

[0086] A second quality criterion concerns the surface finish of the object. To evaluate this second criterion, the processor 20 determines the roughness of the object represented by the reconstructed 3D model M1 and compares this roughness with a predefined roughness threshold. If the determined roughness is greater than the threshold, then the object is considered to be too rough if it were manufactured based on the additive manufacturing command generated in step 100.

[0087] The evaluated roughness is, for example, the Rz roughness, also called the maximum profile height. The Rz roughness indicates the absolute difference between a maximum peak height and a maximum valley depth over a given baseline length on the surface of the object. Alternatively or in addition, the Rp roughness (valley depth) or the Rv roughness (peak heights) can be evaluated, knowing that Rz - Rp + Rv. Number of refusions

[0088] We have seen previously that the elementary 3D models on the basis of which the 3D model is reconstructed intersect in such a way as to model refusions.

[0089] A third quality criterion concerns the maximum number of reflows at any point on the object to be manufactured. To evaluate this third criterion, the processor 20 determines a maximum intersection arity N between the elementary 3D models. As is known, the parity of an intersection denotes the number of sets present in the intersection. In the context of the process of this disclosure, the The sets are the elementary 3D models. N actually designates the maximum number of reflows that will occur during the manufacturing of the object.

[0090] Figure 11 represents a set of five elementary 3D models of melt pools corresponding to parallel lines indicated in five elementary commands. These lines are perpendicular to the plane of Figure 11. The respective profiles of these melt pools are therefore shown in this figure. The 3D model obtained by combining the five elementary 3D models includes regions with different arities: • Regions of arity equal to 1, represented in white, in which only an elementary 3D model is found; • Regions of arity equal to 2, represented in light grey, which are intersections between 2 elementary 3D models; • Regions of arity equal to 3, represented in dark grey, which are intersections between 2 elementary 3D models; • Regions of arity equal to 2, represented in black, which are intersections between 4 elementary 3D models.

[0091] In this example, there is no region of arity equal to 5. Thus, in this example, we have N=4.

[0092] Once the maximum parity N has been determined by the processor 20, the processor 20 compares this maximum parity N with a predefined parity threshold. If the determined maximum parity is greater than the threshold, then the processor 20 may consider that there will be an excessive number of reflows during the manufacturing of the object. In this case, it generates a new additive manufacturing command in which N will be reduced compared to the last generated manufacturing command.

[0093] Reducing the maximum parity is an advantageous corrective measure, as it prevents the laser from passing over the same spot too many times without adding any value to the object. Consequently, by reducing the maximum parity, the total laser melting time will be reduced, and therefore the manufacturing time of the object as well as the amount of energy consumed by the laser of machine 1.

[0094] As we have seen so far, there are several quality criteria that can be evaluated during the quality control step 106.

[0095] Ideally, we would like: • Minimize excess molten material, • Minimize the number of overlaps, • Minimize the number of vectors, • Minimize roughness.

[0096] A multi-criteria decision algorithm can thus be used in step 106 to decide whether the control result is OK or KO. According to a strict approach, one It may be possible to stipulate that the OK result is generated only if all evaluated criteria are simultaneously met. A more flexible approach allows the result to be generated by finding a compromise between these different criteria.

Claims

1. Demands Additive manufacturing process comprising: • obtain an additive manufacturing command suitable for use by a machine to manufacture an object by selective laser melting, the additive manufacturing command including: • Basic commands specific to be used by the machine to manufacture different parts of the object, each basic command comprising: • Location data enabling the machine to locate in space a line having a starting point and an ending point, • laser parameters intended to be used by a machine laser to fuse powder grains along the line from the starting point to the ending point, • for each elementary command, construct (102) an elementary 3D model associated with the elementary command representing a melt pool resulting from the melting of powder grains along the line from the starting point to the ending point, by a laser using the laser parameters, • construct (104) a 3D model representing the object by union of the elementary 3D models respectively associated with the elementary commands, • check (106) the 3D model to verify whether the object represented by the 3D model meets quality criteria, • when the object represented by the 3D model meets the quality criteria, command the machine to manufacture (108) the object by selective laser melting from the additive manufacturing command.

2. A method according to the preceding claim, wherein constructing an elementary 3D model associated with an elementary control comprises: • determining (200) a profile of the melt pool in a cutting plane normal to the line, • generating (202) the elementary 3D model from the profile of the melt pool under the assumption that the profile of the melt pool evolves along the line according to a predefined law.

3. A method according to the preceding claim, wherein the elementary 3D model is generated from the profile of the melt pool under the assumption that the profile is constant along the line.

4. A method according to any one of claims 2 and 3, wherein determining the profile of the melt pool includes searching, in a database associating reference commands and reference profiles, for a reference profile associated with a reference command satisfying a criterion of correspondence with the elementary command.

5. A method according to any one of the preceding claims, comprising: • when the object represented by the 3D model does not meet the quality criteria, generating (100) a new additive manufacturing command suitable for use by the machine to manufacture the object by selective laser melting, the new manufacturing command being different from the additive manufacturing command.

6. A method according to any one of the preceding claims, wherein controlling the 3D model comprises: • determining a volume of the object represented by the 3D model, • comparing the volume and a volume of the object represented by an original 3D model from which the control was previously generated.

7. A method according to any one of the preceding claims, wherein checking the 3D model comprises: • determining the roughness of the object represented by the 3D model, and • compare the roughness and a predefined roughness threshold.

8. A method according to any one of the preceding claims, wherein controlling the 3D model comprises: • comparing a maximum intersection arity between the elementary 3D models and a predefined arity threshold.

9. Product computer program comprising program code instructions for carrying out the steps of the process according to any one of the preceding claims, when such program is executed by a computer.

10. System comprising: • a selective laser melting additive manufacturing machine, • a control unit configured to control the additive manufacturing machine by implementing the process according to any one of claims 1 to 8.

Images