Computer-implemented system and method for the structural integrity assessment of an additively-manufactured component through detected anomalies

Abstract

The computer-implemented method (1) for the structural integrity assessment of an additively-manufactured component through detected anomalies, comprises the following steps: executing a Finite Element (FE) simulation of a component for obtaining a detailed stress analysis in each point of said component; running a model for determining the relationship between a stress or load applied to the component and a detected anomaly size for a predefined target life and at any location of the component; acquiring data related to the component by means of at least an acquisition device and processing the obtained data to determine a detected anomaly size in correspondence of each detected anomalies of the component; using a series of previously acquired data of samples or test articles relating the detected anomaly size, detected in conditions similar to the ones adopted in the step of acquiring data related to the component, and a series of real anomalies size detected with a more precise device, for determining the sizing error of the acquired data by the acquisition device; determining an acceptable anomaly size according to a deterministic or probabilistic criterion for the target life of the component; verifying the presence of detected anomalies sizes exceeding the acceptable anomaly size for the target life of the component; if at least one detected anomaly size exceed the acceptable anomaly size, signaling that the component does not fulfill the requested condition of target failure probability.

Description

[0001] COMPUTER-IMPLEMENTED SYSTEM AND METHOD FOR THE STRUCTURAL INTEGRITY ASSESSMENT OF AN ADDITIVELY-MANUFACTURED COMPONENT THROUGH DETECTED ANOMALIES

[0002] Technical Field

[0003] The present invention relates to a computer-implemented system and method for the structural integrity assessment of an additively- manufactured component through detected anomalies.

[0004] Background Art

[0005] Finite element analyses (FEA) are widely used to investigate structural behavior of additively manufactured (AM) parts and components, e.g. to compute stress and strain distributions in failure and damage analyses. Although there are many investigations on the fatigue behavior of laser powder bed fusion (L-PBF) parts, a broad acceptance of AM processes is limited due to the lack of knowledge concerning fatigue assessment and qualification of complete components (M. Seifi, M. Gorelik, J. Waller, N. Hrabe, N. Shamsaei, S. Daniewicz, J. J. Lewandowski, Progress Towards Metal Additive Manufacturing Standardization to Support Qualification and Certification, JOM 2017 69:3 69 (3) (2017) 439-455. doi:10.1007 / S11837-017-2265-2; Michael Gorelik. Additive manufacturing in the context of structural integrity. International Journal of Fatigue, 94:168-177, 1 2017).

[0006] Several studies have shown that fracture-based analyses are successful because fatigue properties are largely controlled by the presence of anomalies like internal pores that are due to the manufacturing process (Beretta, S., Patriarca, L., Gargourimotlagh, M., Hardaker, A., Brackett, D., Salimian, M., Gumpinger, J. and Ghidini, T., 2022. A benchmark activity on the fatigue life assessment of AlSilOMg components manufactured by L-PBF. Materials & Design, 218, p.l 10713). As a general statement there are two type of anomalies: material anomalies and process anomalies:

[0007] As for material anomalies, such as gas / keyhole porosities, unconsolidated powder, LOFs, stop / start flaws etc., they can be characterized by different methods (metallography, fractographies of specimens, CT scans) that can be easily applied to the test specimens required for the material qualification (see for example ECSS-Q-ST-70-80C).

[0008] Process anomalies refers to those process-induced anomalies which cause evident quality issues, e.g., build stop, build line skipped, cracking or deformation caused by residual stresses during cooling.

[0009] This kind of anomalies can be potentially detected by OM (Online Monitoring), CT scan or other suitable NDE methods (US, eddy current, fluorescent particles) or a combination of the methods.

[0010] As for the assessment in presence of anomalies, NASA recently suggested that probabilistic analyses are the preferential tool for the analysis of anomalies whose size cannot be detected by NDI or other non destrective technique (W. Tilson and D. Wells and C. McClung. Developing Approaches for Certification of Un-inspectable Fracture Critical AM Components. ICAM2022, 2022) and deterministic legacy methods for defects detectable by NDI.

[0011] However, if probabilistic tools are available for calculating the failure probability in presence of inherent anomalies lower than NDI limits of AM components (Wormsen A, Sjbdin B, Harkegard G, Fjeldstad A. Nonlocal stress approach for fatigue assessment based on weakest-link theory and statistics of extremes. Fatigue Fract Eng Mater Struct 2007;30(12): 1214-27; Wormsen A, Fjeldstad A, Harkegard G. A postprocessor for fatigue crack growth analysis based on a finite element stress field. Comput Methods Appl Meeh Engrg; 2008; 197(6-8): 834-45; Enright M, Hudak S, McClung R. Application of probabilistic fracture mechanics to prognosis of aircraft engine components. AIAA J 2006;311 — 6;Li P, Warner D, Phan N. Predicting the fatigue performance of an additively manufactured Ti-6A1-4V component from witness coupon behavior. Addit Manuf 2020;35: 101230; Romano S, Miccoli S, Beretta S. A new FE post-processor for probabilistic fatigue assessment in the presence of defects and its application to AM parts. Int J Fatigue 2019;125:324-41) based on the statistical distribution of the anomalies, the assessment in presence of discrete process anomalies cannot follow the same procedure. This is due to the fact that process anomalies occur as random combination of events / factors during processing and therefore it is not possible to have a statistical distribution describing their size.

[0012] Oppositely, their size can be detected: i) by XCT (X-ray computed tomography) or another suitable NDE method; ii) an Online Monitoring signal collected during the component manufacturing.

[0013] A possible scenario for the qualification / assessment of a component is depicted in Figures 1 and 2, where Figure 1 shows the difference between material anomalies and discrete process anomalies, while Figure 2 shows an example of detection of anomalies through XCT.

[0014] Description of the Invention

[0015] The main aim of the present invention is to provide a computer- implemented system and method that allow the structural integrity assessment of a component in presence of random anomalies detected by a suitable system.

[0016] The above mentioned objects are achieved by the present computer- implemented method for the structural integrity assessment of an additively-manufactured component through detected anomalies, according to the features of claim 1.

[0017] The above mentioned objects are also achieved by the present computer- implemented system for the structural integrity assessment of an additively-manufactured component through detected anomalies, according to the features of claim 9.

[0018] Brief Description of the Drawings

[0019] Other characteristics and advantages of the present invention will become better evident from the description of a preferred, but not exclusive embodiments of a computer-implemented system and method for the structural integrity assessment of an additively-manufactured component through detected anomalies, illustrated by way of an indicative but nonlimiting example in the accompanying Figures, in which:

[0020] Figures 1 and 2 show a possible scenario for the qualification / assessment of a component, where Figure 1 shows the difference between material anomalies and discrete process anomalies, while Figure 2 shows an example of detection of anomalies through XCT ;

[0021] Figure 3 is a general diagram that illustrates the computer-implemented method according to the invention;

[0022] Figure 4 shows a FE analysis of a component manufactured by AM;

[0023] Figure 5 shows a detail of Figure 4;

[0024] Figure 6 shows schematics of a relationship between anomaly size and applied stress at a given target life Ntarget;

[0025] Figure 7 shows anomalies detected by CT in proximity of the fracture surface of a component;

[0026] Figure 8 shows an example of correlation between detected size and real size of anomalies;

[0027] Figure 9 shows a scheme for the calculation of the failure probability;

[0028] Figure 10 shows Map of the acceptable anomaly size along the fracture surface of the bracket for a target life of 80,000 cycles;

[0029] Figure 11 shows a comparison between detected anomaly sizes and acceptable anomaly sizes for a target life of 80,000 cycles;

[0030] Figure 12 shows an example of detected anomalies compared with the acceptable defect size;

[0031] Figure 13 shows an example of determined sizing error through an analysis of data of the real size of anomalies;

[0032] Figure 14 shows an example of detected anomalies compared with the acceptable defect size (load 5kN, Ntarget=20,000 cycles);

[0033] Figure 15 shows an example of the probabilistic assessment for the largest anomaly.

[0034] Embodiments the Invention

[0035] With particular reference to general diagram of Figure 3, globally indicated with reference 1 is a computer-implemented method for the structural integrity assessment of an additively-manufactured component through detected anomalies.

[0036] The computer-implemented method 1 comprises a step 2 of executing a Finite Element (FE) simulation of a component for obtaining a detailed stress analysis in each point of said component (as showed as an example in Figures 4 and 5).

[0037] Particularly, the FE simulation is executed considering component geometry, material elasticity and applied loads.

[0038] The computer-implemented method 1 comprises a further step 3 of running a model for determining the relationship (by points or analytically) between a stress or load applied to said component and a detected anomaly size for a predefined target life (Ntarget) and at any location of said component (see for example Figure 6).

[0039] Particularly, the step 3 of running a model is executed using one of the following tools:

[0040] - models based on explicit crack growth calculations considering a 'shortcrack' effect;

[0041] - simplified descriptions of a S-N diagram as a function of defect size;

[0042] - ML models based on a training datasets of fatigue data on specimens containing defects.

[0043] The acceptable defect at a given target life can be also obtained as an interpolation of data obtained at different number of cycles Nl, N2, N3.

[0044] As an example, Figure 6 shows schematics of a relationship between anomaly size a and applied stress AS at a given component location for the target life Ntarget for a given stress level.

[0045] Furthermore, the computer-implemented method 1 comprises a step 4 of acquiring data related to the component by means of at least an acquisition device D and a step 5 of processing the obtained data to determine a anomaly size in correspondence of each of the detected anomalies of the component.

[0046] According to a preferred embodiment, the acquisition device D comprises a Computerized Tomography scanner, the step 4 of acquiring data comprises performing a Computerized Tomography scan of the component, and the step 5 of processing the obtained data comprises processing the obtained scan to determine a anomaly size in correspondence of each of the detected anomalies of the component.

[0047] As an example, Figure 7 shows detected anomalies from the fracture surface of a component obtained by processing a CT scan of the component.

[0048] However, different types of acquisition devices D are not excluded.

[0049] Particularly, the acquisition device D is selected from at least one of: an NDE (Nondestructive Evaluation) acquisition system or device (for example Eddy current, FPI, US scan); an Online Monitoring signal system.

[0050] Subsequently, the computer implemented method 1 comprises a step 6 of using a series of previously acquired data of samples or test articles relating the detected anomaly size, detected in conditions similar to the ones adopted in the step 4 of acquiring data related to the component, and a series of real anomalies size detected with a more precise device (a much higher resolution scanner or another more precise NDE or by metallographic sections) for determining the sizing error of the acquired data by the acquisition device D by a correlation between said previously acquired detected anomaly size and said previously acquired real anomaly size (see Fig. 8).

[0051] Particularly, the correlation between the previously acquired detected anomaly size and the previously acquired real anomaly size is analyzed with a suitable statistical model to derive the sizing error (or the uncertainty about the size estimated by the CT scan or other NDE adopted) (as an example of analysis see ASTM E3023-21).

[0052] Furthermore, the computer-implemented method 1 comprises:

[0053] - a step 7 of determining an acceptable anomaly size or probabilistic criterion for a target life Ntarget of the component, where the target life has possibly included a deterministic safety factor.

[0054] - a step 8 of verifying the presence of detected anomalies sizes exceeding the acceptable anomaly size for the target life (Ntarget) of the component.

[0055] If at least one detected anomaly size exceed the acceptable anomaly size, the computer-implemented method 1 comprises a step 9 of signaling that said component does not fulfill the requested condition of target failure probability.

[0056] However, the measurements with CT-scan (or other NDI or OM method) is affected by the reviously determined in Step 6.

[0057] Unless the sizing error is negliglible (CVO.Ol), the problem of deciding if the detected anomalies that are exceeding the critical size are acceptable cannot be a simple YES / NO but it has be solved in a probabilistic format, which is the STEP 10 shown in Figure 9, wherein is showed a scheme for the calculation of the failure probability for a given target life Ntarget. In details referring to a given anomaly the acceptance criterion can be set as an acceptable failure probability for the component Pf,accept that is predefined by some standards (e.g. AC33.70-1 and AC33.70-2 issued by FAA, or EurocodeO), that can be taken as a reference for defining it.

[0058] Particularly, the failure probability that pertain to a given detected anomaly Cl, in presence of its sizing error, is calculated (with different methods as algebra of gaussian variables, Monte Carlo simulations or other methods) as:

[0059] Pf a- = Fr[.S > B] where S is the applied stress (either deterministic or with a scatter or distribution) and R is the prospective resistance distribution as a function of anomaly size and its sizing error distribution, that can be calculated from the relationship between defect size and applied stress at the life Ntarget as shown in Figure 9.

[0060] Once the failure probability Pf for a i-th detected anomaly is calculated, then the failure probability is calculated as:

[0061] Another bound for the failure probability can be calculated because if the defects were of the same type (and their sizing error is the same), then the failure probability would only correspond to the failure probability of the largest defect.

[0062] Therefore, the failure probability is calculated as: where aniaxis the largest anomaly. The component can be then accepted, even in presence of detected anomalies, if the the failure probability is lower than acceptable limits set for the components.

[0063] The present invention also relates to a computer-implemented system for the structural integrity assessment of an additively-manufactured component through detected anomalies, comprising: at least an acquisition device D for acquiring data related to the component; at least a processing unit for executing the steps of the computer- implemented method disclosed above.

[0064] According to a preferred embodiment, the acquisition device D comprises a Computerized Tomography scan.

[0065] However, different types of acquisition devices not excluded.

[0066] Particularly, the acquisition device D is selected from at least one of: an NDE (Nondestructive Evaluation) acquisition system or device (for example Eddy current, FPI, US scan); an Online Monitoring signal system.

[0067] Example.

[0068] An example of a possible application of the computer-implemented method and system according to the invention is hereby described.

[0069] An additively manufactured bracket made of AlSilOMg has been obtained by L-PBF.

[0070] A set of X-ray CT scans has been conducted on the part at different resolutions, from a voxel size of 105 pm to a value of 10 pm.

[0071] The component has been subjected to testing at a load of 10 kN at load ratio R=0.1, a failure occurred to one of the two lugs of the component after 80,000 cycles.

[0072] The computer-implemented method and system are used to assess the fatigue performance of the material and to infer the cause of failure.

[0073] A FE simulation of the part in service has been performed, extracting the state of stress across the entire bracket; having experimental evidence of the location of the failure, the area has been partitioned to locally improve mesh refinement. Material data from fatigue tests and crack propagation tests are used to set up the model correlating the state of stress from FE analysis to the size of the critical detected anomaly for a target life Ntarget= 80,000 cycles; the simplified description of the S-N diagram as function of the size of the anomaly has been implemented (Romano, S., Briickner-Foit, A., Brandao, A., Gumpinger, J., Ghidini, T., & Beretta, S. (2018). Fatigue properties of AlSilOMg obtained by additive manufacturing: Defect -based modelling and prediction of fatigue strength. Engineering Fracture Mechanics, 187, 165-189. Beretta, S., L. Patriarca, M. Gargourimotlagh, A. Hardaker, D. Brackett, M. Salimian, J. Gumpinger, and T. Ghidini. "A benchmark activity on the fatigue life assessment of AlSilOMg components manufactured by L-PBF." Materials & Design 218 (2022): 110713 ).

[0074] Thus using the state of stress at the nodes of the mesh in the failed area, the acceptable anomaly size is calculated according to a deterministic criterion according to a target life Ntarget of 80,000 cycles also considering the presence of the residual stresses determined by XRD. The nodes of the finite element containing a detected anomaly are used to interpolate the acceptable flaw size onto the position of the detected flaw.

[0075] The detected anomalies are compared to the acceptable anomalies under the size perspective. Figure 10 shows a map of the acceptable anomaly size along the fracture surface of the bracket for a target life Ntarget of 80,000 cycles.

[0076] The same data are shown in Figure 11 in a representation as the one of Figure 6.

[0077] The detected anomalies’ sizes are compared to the acceptable flaws’ size obtained for Ntarget of 80,000 cycles, as showed in Figure 11.

[0078] As it can be seen six anomalies exceed the maximum allowable defect size and this justifies the failure of the component.

[0079] If the component was subjected to an applied load of 5kN, then the map of acceptable defect size for Ntarget=80,000 cycles would become the one shown in Fig. 12. As it can be seen the largest defect is lower than the line of acceptable defect sizes and so if a safety factor of 4 is considered (i.e. ECSS-Q-ST-70-80C) then the component could be accepted for a service life of 20,000 cycles. This kind of analysis represents a 'deterministic analysis'.

[0080] Figure 12 shows an example of detected anomalies compared with the acceptable defect size (load 5kN, Ntarget=80,000 cycles).

[0081] However, instead a simple deterministic analysis it would be better to consider also the sizing error of any of the detected anomalies. In this scenario it is important to evaluate the prospective reliability of the component considering the uncertainties of the detection.

[0082] In details, a series of data acquired before the assessment have allowed to determine the 'sizing error' through an analysis of data of the real size of anomalies (detected through a CT with high resolution) detected on specimens that were analyzed for the aim of obtaining a plot such as the one in Figure 8 analyzing the data in log scales (Georgiou, G. A. (2007). PoD curves, their derivation, applications and limitations. Insight-Non- Destructive Testing and Condition Monitoring, 49(7), 409-414. or ASTM E3023-21) for obtaining the sizing error (a) as shown in Figure 13.

[0083] Fig. 14 shows an example of detected anomalies compared with the acceptable defect size (load 5kN, Ntarget=20,000 cycles).

[0084] Once the sizing error has been determined it is possible to obtain the probabilistic analysis depicted in Figure 15 and then draw a conclusion about acceptability of the component if Pf is lower than or equal to Pr.accept-

Claims

CLAIMS1) Computer-implemented method (1) for the structural integrity assessment of an additively-manufactured component through detected anomalies, characterized in that it comprises at least the following steps:- executing a Finite Element (FE) simulation of a component for obtaining a detailed stress analysis in each point of said component (step 2);- running a model for determining the relationship between a stress or load applied to said component and a detected anomaly size for a predefined target life (Ntarget) and at any location of said component (step 3);- acquiring data related to the component by means of at least an acquisition device (D) (step4) and processing the obtained data to determine a detected anomaly size in correspondence of each detected anomalies of the component (step 5);- using a series of previously acquired data of samples or test articles relating said detected anomaly size, detected in conditions similar to the ones adopted in the step (4) of acquiring data related to the component, and a series of real anomalies size detected with a more precise device, for determining the sizing error of the acquired data by the acquisition device D by a correlation between said previously acquired detected anomaly size and said previously acquired real anomaly size (step 6);- determining an acceptable anomaly size according to a deterministic or probabilistic criterion for the target life (Ntarget) of said component (step 7);- verifying the presence of detected anomalies sizes exceeding said acceptable anomaly size for the target life (Ntarget) of said component(step 8);- if at least one detected anomaly size exceed said acceptable anomaly size, signaling that said component does not fulfill the requested condition of target failure probability (step 9).2) Computer-implemented method (1) according to claim 1, characterized in that said acquisition device (D) is selected from at least one of: an NDE acquisition system or device; an Online Monitoring signal system or device.3) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that said acquisition device (D) comprises a Computerized Tomography scan, said step (4) of acquiring data comprises performing a Computerized Tomography scan of said component, and said step (5) of processing the obtained data comprises processing the obtained scan to determine a detected anomaly size in correspondence of each detected anomalies of said component.4) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that the failure probability that pertain to a given detected anomaly CL, in presence of its sizing error, is calculated as:P = Pr[S > P] where S is the applied stress and R is the prospective resistance distribution as a function of anomaly size and its sizing error distribution.5) Computer-implemented method (1) according to claim 4, characterized in that, once said failure probability Pf for a i-th detected anomaly is calculated, then the failure probability is calculated as:6) Computer-implemented method (1) according to claim 5,characterized in that the failure probability is calculated as:P Jf ia^max where amax is the largest anomaly.7) Computer implemented method (1) according to one or more of the preceding claims, characterized in that said step (3) of running a model is executed using the following tools:- models based on explicit crack growth calculations considering a 'shortcrack' effect;- simplified descriptions of a S-N diagram as a function of defect size;- ML models based on a training datasets of fatigue data on specimens containing defects.8) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that said correlation between said detected anomaly size and said real anomaly size is analyzed with a statistical model to derive the sizing error.9) Computer-implemented system for the structural integrity assessment of an additively-manufactured component through detected anomalies, characterized in that it comprises:- at least an acquisition device (D) for acquiring data related to the component;- at least a processing unit for executing the steps of the computer- implemented method (1) according to one or more of the preceding claims.10) Computer-implemented system according to claim 9, characterized in that said acquisition device (D) is selected from at least one of: an NDE (Nondestructive Evaluation) acquisition system or device; an Online Monitoring signal system.11) Computer-implemented system according to one or more of the claims 9 and 10, characterized in that said acquisition device (D) comprises aComputerized Tomography scanner.

Images