WO2023229561A1

WO2023229561A1 - An optimization method

Info

Publication number: WO2023229561A1
Application number: PCT/TR2023/050463
Authority: WO
Inventors: Ahmet Can GUNAYDIN; Ali Rıza YILDIZ
Original assignee: Tusas- Turk Havacilik Ve Uzay Sanayii Anonim Sirketi
Priority date: 2022-05-27
Filing date: 2023-05-23
Publication date: 2023-11-30

Abstract

The present invention relates to at least one part model (2), which is created in order to produce the part (P) with additive manufacturing by making the three-dimensional design and analysis of the part (P) in a virtual environment; a table (3) in the virtual environment, on which the part model (2) is created; a reference position (I) at which the part model (2) is created on the table (3); at least one processor unit (4) that allows design and analysis of the part model (2) in the virtual environment; a plurality of unit cells (5), each with the same geometrical shape, which are designed virtually in the processor unit (4) and used to create the part model (2); an overhang angle (6) between the production direction and the unit cell (5); at least one overhang surface (7) on the part model (2), which has an overhang angle (6) greater than a reference angle predetermined by the user; at least one support element (8) that supports the production of the overhang surface (9) on the table (3).

Description

AN OPTIMIZATION METHOD

The present invention relates to an optimization method used for additive manufacturing technology.

Additive manufacturing, or more commonly known as three-dimensional printing, is a production method that makes it possible to produce three-dimensional parts and/or prototypes by laying metal, ceramic or polymer layers on top of suitable powders or fine wires and subjecting them to heat treatment by means of a printer tip. An angle and position of the part at which it is placed on the table is called the build orientation. If an angle formed between a surface of the part to be produced and the table where the production is initiated exceeds a certain value, the relevant surface is called the “overhang surface”, which requires a support structure for the production thereof. The amount of support structure of the build orientation has a great influence on the part quality, production time and production cost. Support structures are usually removed from the structure following the production of the part, thus becoming a waste material. Both raw materials and energy are consumed to produce the support structures together with the part. Extra labor and time are spent for cleaning the support structures. This increases production time and cost. In addition, the surface quality of the part is adversely affected as the support structures are removed from the part surface. Production cost mainly depends on scanning speed, scanning strategy, amount of support structure, production length and post-production processes. The amount of support structure and the production length are directly dependent on the build orientation.

Optimization studies on processes used to produce functional metal parts in the aviation industry, such as Laser Powder Bed Fusion, have increased in number, because the initial investment and process costs of such processes are quite high. In the build orientation, different objective functions for different geometries may conflict with each other. Therefore, it is important to evaluate these objective functions simultaneously without suppressing each other. As a result of multi-objective optimization, a series of alternative solutions that do not suppress each other are obtained. Many methods have been developed to calculate objective functions. STL file type is widely used in additive manufacturing technology. According to the STL file, an outer surface of the geometry is approximately represented by dividing it into triangles. Since STL is a predictive model using planar triangles, operations such as rotation, slicing and overhang surface detection are performed faster. Some methods have been developed for estimating the amount of support structure, surface roughness and production time by geometric investigations over the STL data type. In one of the methods, the total area of the overhang surfaces is determined. In another similar method, the total reflected area of the overhang surfaces is determined. However, in this method, the surface areas are projected onto the table and the total area reflected on the table represents the amount of support structure. In areabased approaches, the height of the surface relative to the production table is ignored. In the volume-based method, on the other hand, the support structures intersect with the geometry of the part. In this case, more support structure requirements are calculated for the overhang surface.

The United States patent document US20160085882A1 , which is included in the known- state of the art, comprises a method that receives a solid model. The method comprises analyzing the solid model to determine an orientation that minimizes a build height or a support volume.

Thanks to an optimization method according to the present invention, two objective functions that may conflict during the optimization of the build orientation are evaluated simultaneously without suppressing each other.

Another object of the invention is to ensure that the support element volume variable calculated in each iteration of the part is taken into account while calculating the production time.

A further object of the invention is to control intersection of the support structures with the geometry of the part by a ray-triangle intersection algorithm, thus providing more realistic results.

The optimization method realized to achieve the object of the invention, which is defined in the first claim and other claims dependent thereon, comprises a part model which is designed and analyzed in three dimensions in a virtual environment before the part is produced with the additive manufacturing method. It comprises a table that allows the part model to be produced with a reference and represents the table at the time of manufacture. The part model is placed at a reference position on the table. There is a processor unit that enables the design and analysis of the part model in a virtual environment. The part model consists of a plurality of unit cells in the processor unit, each of which has the same geometrical shape. There is an overhang angle, which is the angle between the production direction perpendicular to the table and the unit cell. The part model has at least one overhang surface with an overhang angle that is greater than a reference angle predetermined by the user. It comprises at least one support element that enables the overhang surface to be supported while being produced on the table.

The optimization method according to the invention calculates a plurality of support element volumes required for the production of the support elements, which is calculated for each angle depending on whether the initial position of the part model is placed on the table with angle ranges predetermined by the user or with angle ranges created by an optimization algorithm predetermined by the user. The processor unit uses each calculated support element volume to calculate the times required to manufacture the part.

In an embodiment of the invention, in the optimization method, only the three-axis coordinate and surface normal data of the unit cells are input to the processor unit by the user. The processor unit provides the calculation of the optimum time merely by this information.

In an embodiment of the invention, the optimization method comprises a first support element volume, which is a volume requiring support element between the unit cell and the table when the projection of the unit cell is reflected on the table. When the projection of the unit cell is reflected on another unit cell, there is a second support element volume, which is a volume requiring support element between the unit cells. The support element volume, which is the sum of the first support element volume and the second support element volume, is calculated by the processor unit.

In an embodiment of the invention, the optimization method comprises at least one laser used for melting powder during additive manufacturing. There is at least one outer surface volume on the outer surface of the part and at least one inner surface volume that represents the remaining volume when the outer surface volume is subtracted. The processor unit provides individual calculation of the times required for scanning the distance of the inner surface volume and the outer surface volume by the laser. The sum of these times is multiplied by the number of layers and the time required for re-laying, and the calculation of the optimum time required for production is provided by the processor unit.

In an embodiment of the invention, the optimization method comprises a volume ratio that is the ratio of the total volume of the support element predetermined by the manufacturer. By multiplying the volume ratio with the support element volume, the optimum support element volume required during production is calculated by the processor unit.

In an embodiment of the invention, in the optimization method, the processor unit provides the calculation of support element volume and time data by using the NSGA-II algorithm, one of the multi-objective optimization methods.

In an embodiment of the invention, in the optimization method, the processor unit applies Pareto analysis for the user to use any of the support element volume and time data created at the end of the NSGA-II algorithm as a priority. As a result of the Pareto analysis, selection is made according to the prioritization determined by the user.

In an embodiment of the invention, the optimization method comprises the processor unit that provides a single solution by taking the ratios of the support element volume and time data input by the user, by using Compromise Programming Method and Weighted Sum Method. For example, the support element volume is determined as 0.8, and the time thereof as 0.2. According to this prioritization, a single solution is obtained.

In an embodiment of the invention, the optimization method comprises the processor unit that enables artificial intelligence to process the support element volume and time data input by the user using the compromise programming and weighted sum methods. In an embodiment of the invention, in the optimization method, the processor unit provides the calculation of the optimum time by taking into account the time for removing the support element from the part by the user, which is predetermined by the user.

In an embodiment of the invention, the optimization method comprises the unit cell in a triangle form; the processor unit having a ray-triangle intersection algorithm and using Moller Trumbore algorithm as ray-triangle intersection algorithm. It is used to calculate the intersection of a three-dimensional ray and a triangle without a need for a pre-calculation of the plane equation of the plane containing the Moller T rumbore triangle.

In an embodiment of the invention, in the optimization method, the processor unit uses the table size data input by the user to calculate the position of the part on the table according to the table size.

In an embodiment of the invention, in the optimization method, the processor unit provides artificial intelligence processing of the optimum support element volume data required during production by multiplying the volume ratio with the support element volume.

In an embodiment of the invention, the optimization method comprises the processor unit that allows the user to input parameters as the printer's scanning speed, layer thickness, scanning interval, number of lasers and re-lay time, before production.

In an embodiment of the invention, in the optimization method, the processor unit enables the coordinates of the part model to be rotated by the quaternion method to change the reference position of the part model.

In an embodiment of the invention, the optimization method comprises the processor unit that provides calculation of the optimum reference position using the buy-to-fly ratio. The buy-to-fly ratio is defined as a material used to manufacture the part I a resulting part. For example, if a 50 kg metal billet is used to produce a part, and a part weighing 10 kg is obtained by processing it, the buy-to-fly ratio is 5. In additive manufacturing, the buy-to-fly ratio is calculated as (Part volume + support element volume) I part volume. Optimization is made by taking this ratio into account during multi-objective optimization. The optimization method realized to achieve the object of the invention is illustrated in the attached drawings, in which:

Figure 1 is a schematic view of the processor unit.

Figure 2 is a schematic view of the additive manufacturing.

Figure 3 is a schematic view of the support element, reference position, overhang angle and overhang surface.

Figure 4 is a schematic view of the first support element volume and the second support element volume.

Figure 5 is a schematic view of the inner surface volume and the outer surface volume.

All the parts illustrated in figures are individually assigned a reference numeral and the corresponding terms of these numbers are listed below:

1. Optimization method

2. Part model

201. Inner surface volume

202. Outer surface volume

3. Table

4. Processor unit

5. Unit cell

6. Overhang angle

7. Overhang surface

8. Support element

9. Support element volume

901. First support element volume

902. Second support element volume

(P) Part

(I) Reference position

(L) Laser

The optimization method (1) comprises at least one part model (2), which is created in order to produce the part (P) with additive manufacturing by making the three-dimensional design and analysis of the part (P) in a virtual environment; a table (3) in the virtual environment, on which the part model (2) is created; a reference position (I) at which the part model (2) is created on the table (3); at least one processor unit (4) that allows design and analysis of the part model (2) in the virtual environment; a plurality of unit cells (5), each with the same geometrical shape, which are designed virtually in the processor unit (4) and used to create the part model (2); an overhang angle (6) between the production direction and the unit cell (5); at least one overhang surface (7) on the part model (2), which has an overhang angle (6) greater than a reference angle predetermined by the user; at least one support element (8) that supports the production of the overhang surface (9) on the table (3) (Figure 1 , Figure 2, Figure 3).

The optimization method (1) according to the invention comprises a plurality of support element volumes (9) required for the production of the support elements (8), which are calculated for each angle depending on whether the reference position (I) of the part model (2) is changed by angle ranges predetermined by the user or by angle ranges created by a user predetermined optimization algorithm; the processor unit (4) which calculates the times required to manufacture the part (P) using each calculated support element volume (9) (Figure 1).

Before the part (P) is produced with additive manufacturing, a three-dimensional part model (2) is designed in the virtual environment in order to analyze it. In the same virtual environment, there is a table (3) that represents the table during production. The part model (2) is referenced on this table (3) so as to be created. There is a reference position (I) at which the part model (2) is created on the table (3). The design of the part model (2) is carried out by the processor unit (4). The part model (2) consists of unit cells (5), each of which has the same geometrical shape. The angle of the unit cells (5) between the table and the axis perpendicular to the table is called the overhang angle (6). Surfaces with an overhang angle (6) greater than the reference angle predetermined by the user are defined as the overhang surface (7). In order for the overhang surfaces (7) to be produced on the table (3), they must be supported by the support element (8). The support element (8) is generally removed from the part (P) after the production, which is then called waste material. The support element (8) affects the production time and cost.

Production time in additive manufacturing technology is affected by many parameters depending on the working principle of the process. However, one of the most important factors affecting the production time, regardless of the process and work, is the support element (8). The part model (2) is rotated from the reference position (I) at angle intervals determined by the user. The support element volume (9) is calculated for each angle. For each calculated support element volume (9), the times required for the production of the part (P) are calculated. Production time is calculated by the following formula: t_btMd = (y_P+ v_sr_{vf) +} — Ap ₊

y and represents the part (P) and the support

^m_ner X_hN_L 0contourL_tN_L ^{P S K} M \ / MM element volume (9).

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) that determines the support element volume (11) and the optimum time merely by coordinates and surface normal data of unit cells (5) in three axes, which are input by the user. The algorithm in the processor unit (4) enables the user to calculate the support element volume (8) and the optimum time that can be used for production by inputting only the coordinate and surface normal data of the unit cells (5) in the three axes (Figure 1).

In an embodiment of the invention, the optimization method (1) comprises a first support element volume (901) which requires support element (8) between the unit cell (5) and the table (3) when the projection of the unit cell (5) is reflected on the table (3); a second support element volume (902) which requires support element (8) between the unit cells (5) when the projection of the unit cell (5) is reflected on another unit cell (5); the processor unit (4) that provides the calculation of the support element volume (9) consisting of the sum of the first support element volume (901) and the second support element volume (902). It is checked whether the support element (8) intersects with the geometry of the part (P). In this way, the most realistic results are obtained. The support element volume (9) is calculated with the formula = Z ₌₁ ^₁ cos e₁((₁9_llz+₁9_12z+₁9_13z)) - /i_z). According to the formula, N represents the number of facets in the STL data. If the surface is an overhang surface, K=1 , otherwise

represents the height in the z axis if the support element (8) intersects with the geometry of the part (P). If there is no intersection, it is considered zero (Figure 1 , Figure 4).

In an embodiment of the invention, the optimization method (1) comprises at least one laser (L) used for melting the powder during production; at least one inner surface volume (201) on the inner surface of the part (P); at least one outer surface volume (202) on the outer surface of the part (P); the processor unit (4) which enables individual calculation of the times required for scanning the distance of the inner surface volume (301) and the outer surface volume (202) by the laser (L), and sums these with the product of the data of the number of layers and the time required for re-laying, thereby providing calculation of the optimum time required for production. In the additive manufacturing process, the part (P) is initially sliced and the scanning algorithm is run for each layer, so that the paths to be followed by laser (L) are determined. The sum of these paths is the total distance the laser (L) will move from start to finish of production. This total distance is divided by the laser (L) speed to obtain the time spent scanning during the process. In order to increase the surface quality in advanced additive manufacturing applications, the outer contour of the cross-sectional area, that is, the surface of the part, can be scanned at a different speed than the interior of the part (P). In this case, the distance that the laser (L) will scan on the cross-sectional contour representing the outer surface of the part (P) and the distance that the laser (L) will scan on the inside of the part (P) are calculated separately. In addition, a plurality of lasers (L) are used for the scanning process to reduce the production time. In powder bed processes, powder material is re-laid after the production of each layer. This process is repeated as much as the number of layers. Actual production time is calculated by the formula t_{realistic-build} = +

L_ct_recoat. In the formula, X_inneri is the distance the laser (L) will scan for the inner region of each layer. X_contouri is the distance the laser will scan for the contour of each layer. inner and d_contour are the speed of the laser (L) used to scan the inner region of the laser (L) and contour, respectively. N_L , L_c and t_recoat represent the number of lasers (L), number of layers and re-lay time, respectively. The number of layers is calculated by the formula L_c = ^{hzmax dz}. h_Zmnv represents the highest point in the production line, d_z

fi X represents the space between the part (P) and the table (3), and L_t represents the layer thickness.

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4), which enables the optimum support element (8) volume required during production to be determined by multiplying the support element volume (9) by the ratio of the total volume of the support element (8) predetermined by the manufacturer. Said support element volume (9) is the entire volume requiring the support element (8), but not all of it is used during production. Production time is calculated by the formula t_buad = (y_P+ v_sr_{vf) +} — Ap ₊ t y and represent part (P) and support element

^m_ner X_hN_L 0contourL_tN_L ^{C VeCOat P S K} M \ / MM volume (9). The support element volume (9) represents the total volume required for the support element (8), but this volume is not completely filled during production. Therefore, r_Vf is used as the volume ratio (V) to represent the actual support element (8) volume. A_p represents the total outer surface area of the part (P) and X_h represents the scanning distance (Figure 1 , Figure 5).

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) that provides calculation of support element volume (11) and time data by using NSGA-II algorithm, one of the multi-objective optimization methods. In the build orientation, different objective functions for different geometries may conflict with each other. Therefore, these objective functions should be evaluated simultaneously without suppressing each other. As a result of multi-objective optimization, a series of alternative solutions that do not suppress each other are obtained. Production time and support element volume (9) are objective functions (Figure 1).

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) that applies Pareto analysis so that any of the support element volume (9) and time data created at the end of the NSGA-II algorithm can be used primarily by the user. Pareto analysis provides the best solutions that do not suppress each other (Figure 1).

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) that provides a single solution by proportioning the support element volume (9) and time data input by the user by using compromise programming and weighted sum methods. The weight ratios of the support element volume (9) and time objective functions are provided by the user (Figure 1).

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) that provides artificial intelligence processing of support element volume (9) and time data input by the user by using compromise programming and weighted sum methods. The support element volume (9) and time objective functions, weight ratios and results are recorded in the database by the user. Optimum support element and time are calculated by means of artificial intelligence (Figure 1).

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) which enables the optimum time to be determined by taking into account the user- predetermined time for the removal of the support element (8) from the part (P) after the production of the part (P). The removal time of the support element from the part affects the production time. Taking this time into account ensures the most accurate result (Figure 1).

In an embodiment of the invention, the optimization method (1) comprises the unit cell (5) in triangular form; the processor unit (4) having a ray-triangle intersection algorithm and using the Moller Trumbore algorithm as the ray-triangle intersection algorithm. In additive manufacturing technology, the outer surface of the geometry is divided into triangles in STL (Standard Triangle Language) file format, so that it is represented approximately. Since STL is a predictive model using planar triangles, operations such as rotation, slicing and overhang surface detection are performed faster. With the beam-triangle intersection algorithm, it is checked whether the support element (8) intersects with the geometry of the part (P) (Figure 1).

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) that uses the table (3) size data input by the user, so that the optimum reference position (I) is calculated based on the table (3) size. It is ensured that all alternative build orientations obtained by using the table (3) size data are reproducible (Figure 1, Figure 3).

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) which enables the optimum support element (8) volume required during production to be determined by multiplying the support element volume (9) by the ratio of the total volume of the support element (8) predetermined by the manufacturer (Figure 3).

In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) which allows the printer's scanning speed, layer thickness, scanning interval, number of lasers and relay time parameters to be input by the user before production. In an embodiment of the invention, the optimization method (1) comprises the processor unit (4) which enables the coordinates of the part model (3) to be rotated using the quaternion method, so as to change the reference position (I) of the part model (3).

Claims

CLAIMS An optimization method (1) comprising at least one part model (2), which is created in order to produce the part (P) with additive manufacturing by making the three-dimensional design and analysis of the part (P) in a virtual environment; a table (3) in the virtual environment, on which the part model (2) is created; a reference position (I) at which the part model (2) is created on the table (3); at least one processor unit (4) that allows design and analysis of the part model (2) in the virtual environment; a plurality of unit cells (5), each with the same geometrical shape, which are designed virtually in the processor unit (4) and used to create the part model (2); an overhang angle (6) between the production direction and the unit cell (5); at least one overhang surface (7) on the part model (2), which has an overhang angle (6) greater than a reference angle predetermined by the user; at least one support element (8) that supports the production of the overhang surface (9) on the table (3), characterized by a plurality of support element volumes (9) required for the production of the support elements (8), which are calculated for each angle depending on whether the reference position (I) of the part model

(2) is changed by angle ranges predetermined by the user or by angle ranges created by a user predetermined optimization algorithm; the processor unit (4) which calculates the times required to manufacture the part (P) using each calculated support element volume (9). An optimization method (1) according to claim 1 , characterized by the processor unit (4) that determines the support element volume (11) and the optimum time merely by coordinates and surface normal data of unit cells (5) in three axes, which are input by the user. An optimization method (1) according to claim 1 or claim 2, characterized by a first support element volume (901) which requires support element (8) between the unit cell (5) and the table (3) when the projection of the unit cell (5) is reflected on the table

(3); a second support element volume (902) which requires support element (8) between the unit cells (5) when the projection of the unit cell (5) is reflected on another unit cell (5); the processor unit (4) that provides the calculation of the support element volume (9) consisting of the sum of the first support element volume (901) and the second support element volume (902).

4. An optimization method (1) according to any of the above claims, characterized by at least one laser (L) used for melting the powder during production; at least one inner surface volume (201) on the inner surface of the part (P); at least one outer surface volume (202) on the outer surface of the part (P); the processor unit (4) which enables individual calculation of the times required for scanning the distance of the inner surface volume (301) and the outer surface volume (202) by the laser (L), and sums these with the product of the data of the number of layers and the time required for re-laying, thereby providing calculation of the optimum time required for production.

5. An optimization method (1) according to any of the above claims, characterized by the processor unit (4), which enables the optimum support element (8) volume required during production to be determined by multiplying the support element volume (9) by the ratio of the total volume of the support element (8) predetermined by the manufacturer.

6. An optimization method (1) according to any of the above claims, characterized by the processor unit (4) that provides calculation of support element volume (11) and time data by using NSGA-II algorithm, one of the multi-objective optimization methods.

7. An optimization method (1) according to any of the above claims, characterized by the processor unit (4) that applies Pareto analysis so that any of the support element volume (9) and time data created at the end of the NSGA-II algorithm can be used primarily by the user.

8. An optimization method (1) according to any of the above claims, characterized by the processor unit (4) that provides a single solution by proportioning the support element volume (9) and time data input by the user by using compromise programming and weighted sum methods.

9. An optimization method (1) according to any of the above claims, characterized by the processor unit (4) that provides artificial intelligence processing of support element volume (9) and time data input by the user by using compromise programming and weighted sum methods.

10. An optimization method (1) according to any of the above claims, characterized by the processor unit (4) which enables the optimum time to be determined by taking into account the user-predetermined time for the removal of the support element (8) from the part (P) after the production of the part (P).

11. An optimization method (1) according to any of the above claims, characterized by the unit cell (5) in triangular form; the processor unit (4) having a ray-triangle intersection algorithm and using the Moller Trumbore algorithm as the ray-triangle intersection algorithm.

12. An optimization method (1) according to any of the above claims, characterized by the processor unit (4) that uses the table (3) size data input by the user, so that the optimum reference position (I) is calculated based on the table (3) size.

13. An optimization method (1) according to any of the claims 5 to 12, characterized by the processor unit (4) which provides artificial intelligence processing of the optimum volume data of the support element (8) required during production by multiplying the volume ratio (V) with the support element volume (9).

14. An optimization method (1) according to any of the above claims, characterized by the processor unit (4) which allows the printer's scanning speed, layer thickness, scanning interval, number of lasers and relay time parameters to be input by the user before production.

15. An optimization method (1) according to any of the above claims, characterized by the processor unit (4) which enables the coordinates of the part model (3) to be rotated using the quaternion method, so as to change the reference position (I) of the part model (3).