The presentation provides an overview of the AFGROW enhancements/changes implemented over the past year and outlines the planned updates that will be included in the upcoming AFGROW version 5.6.

• A new classic single oblique crack-at-hole solution developed using the most recent wide-plate tabular database.
• A new bearing part-through crack width correction.
• Support for spectra with three independent load channels.

This work introduces a fully automated, repeatable methodology for fitting crack growth rate curves by formulating the traditional Harter-T method as a constrained convex optimization problem. To eliminate subjective manual tuning and handle experimental noise, the framework minimizes the Mean Squared Error (MSE) directly on the experimental crack growth rate data while simultaneously shaping the Da/DN curve and m-curves according to the mathematical constraints and visual intuition of the Harter-T method. This robust pipeline replaces user bias in the labor-intensive fitting procedure with a mathematically rigorous optimization framework, providing repeatable results and significant speedup to the backbone of fracture mechanics.

As part of an effort to conduct through-bushing NDI testing, SwRI has been attempting to grow cracks with an a/c < 1 from corner flaws in holes (or as close as can be achieved). This is a challenge because the natural stress concentration factor of a hole typically results in faster growth down the bore versus the surface, and an a/c > 1. The NDI group wished to grow the flaw without applying bending moments to the coupon, and instead devised a setup using a notched lip and interference fit pin to influence the growth of the crack. The structures group then modelled this setup in BAMpF using a contact solution in StressCheck, across a range of interference levels, lip/undercut geometries, and applied stress levels to predict the aspect ratio of the crack that would emerge. The models predict that this approach can modify the aspect ratio, but the effect is modest, and while the aspect ratio was reduced in some models, it was never less than 1, when accounting for final ream requirements.

Attempts to model short crack growth using standard “long” crack growth rate data have long been shown to be highly unconservative. In response, many have proposed the use of “short” crack growth rate data that extend Region II curves backward into Region I, creating a lower “short” crack growth rate threshold. This presentation will review recent work to explore alternative methods that are based on new data and quantifiable parameters.

The Fighter Aircraft Loading STAndard For Fatigue Evaluation (FALSTAFF) spectrum was collaboratively developed in the 1970s to provide a standardized, flight-by-flight load history representing the severe, maneuver-dominated environment of tactical military aircraft. It traditionally served as a normative benchmark for fatigue testing, material comparison, and evaluating load interaction effects with regards to the early years of Damage Tolerance Analysis (DTA). However, does FALSTAFF have more to offer? Can FALSTAFF be used as an approximate spectrum when a location specific spectrum is not immediately available

To answer these questions, this study evaluated the comparative impact of the FALSTAFF spectrum on calculated fatigue life. Using a standardized set of A-10 critical location geometries, life calculations were performed in AFGROW utilizing established spectra from the F-15, F-16, T-38, and A-10. These baseline lives were then directly compared against parallel runs utilizing the FALSTAFF spectrum. Analyses were conducted both with and without crack growth retardation methods. Generally the FALSTAFF spectrum provided a conservative approximation for a wing root bending moment spectrum. In situations where FALSTAFF did not provide a conservative approximation, the spectra were evaluated to identify the attributes that primarily contributed to the differences. Further comparisons were completed with the Mini- Transport Wing Standard (Mini- TWIST) spectrum to evaluate it’s applicability in scenarios where the FALSTAFF spectrum was unconservative.

Information on the latest research and development efforts and plans beyond AFGROW Release 5.6

While reviewing damage tolerance analysis ground rules for the A-10 and B-52 programs, a disparity was found in the assumed load transfer at a fastener after the ligament failed in a continuing damage analysis. The A-10 program has always assumed the bearing load goes to zero, therefore BSF = 0 and ASF = 1. This is mostly due to early limitations in AFGROW and logically supported with the argument that the cracked ligament reduces stiffness such that bearing load becomes negligible. This was never evaluated through FEM tools however. The B-52 program assumes the opposite; the bearing load doesn’t change at all for the last phase of a continuing damage analysis. This is the more conservative approach but is likely inaccurate. The purpose of this work was to start looking at what an appropriate load transfer assumption should be for this final phase of continuing damage.

In order to start looking at this, a simple model of two plates connected with idealized 3D fasteners was created. Loading was developed to get approximately 100 lbs. of load transfer at each hole. The thickness of the plates was varied to get an understanding for the influence on the load transfer value. Additional work on looking at different fastener diameters and different materials is planned but not yet accomplished. The results of the initial work will be presented.

Fatigue crack growth modeling of corner cracks at a hole considering continuing damage has often been analyzed using a sequential approach, where the primary (0.05”) and secondary (0.005”) cracks grow independently, without affecting one another. Previous evaluations have shown that this approach can be unconservative relative to a simultaneous approach, where both primary and secondary cracks grow simultaneously, since the growth of the secondary crack can be influenced by the primary crack. This work evaluated both sequential and simultaneous approaches for a wide variety of B-52 fatigue details/locations.

When creating a material fit for newly tested 13-8PH material, it was found that there are large differences in Kic values from testing, ranging from 40 ksi/sqrt-in to 100 ksi/sqrt-in. This presentation will discuss this issue with the intent to facilitate a discussion on appropriate paths forward in deciding an appropriate Kic value to use for analysis.

This work investigates machine-learning-based modeling as a complementary approach for representing multidimensional K-solution data used in AFGROW. Multiple nonlinear regression methods, including kernel-based models and multilayer perceptron neural networks, were evaluated for learning high-dimensional mappings between crack-geometry and loading parameters and dimensionless Beta values. The study progressed beyond conventional random train-test validation to include curve-based assessment, interpolation behavior, sparse and structured holdout testing, and evaluation of smoothness across interacting dimensions. Radial-basis-function kernel methods demonstrated smooth nonlinear approximation capability, while recent MLP models achieved strong predictive accuracy and continuous multidimensional response behavior, showing potential to reduce manual curve-fitting effort, support residual-based identification of anomalous data points, and provide compact surrogate representations for increasingly large solution spaces. The work remains exploratory for AFGROW’s existing interpolation framework, with recent ONNX export and independent inference testing providing an initial pathway toward portable model deployment and future integration studies.

This presentation walks through the updated process for preparing new tabular datasets for AFGROW implementation. Previous datasets were approached individually and often modified manually. This approach results in variations in the methods used and a low level of repeatability. The goal of this project was to consolidate existing methods, keep better records of changes to each dataset, and reduce the barrier to adding new datasets. A single processing tool was created to take raw crack front data from FEM analysis all way to a clean dataset ready for AFGROW. The tool was developed on a corner crack at hole dataset for first infinite, then finite plates. Main functions of the program include extracting vertices from crack front data, and smoothing the resulting dataset for clean interpolation. The process for extracting the vertices can be used very generally, while the cleaning currently must be tuned to the specific dataset. Throughout the process various measurements are taken and outputs are recorded to ensure a cohesive picture of the data is available, and no errors are missed. This new workflow will help ensure datasets can meet a high-quality standard in less time, makes it easier to expand and improve tabular models, and adds a previously missing component of record keeping and repeatability.

Closing the gap between FEM results and AFGROW crack growth analysis is not always trivial. This presentation covers the process of integrating finite-element results and live analysis into an AFGROW plugin with both a tabular and live approach, while also highlighting tools that can be used to make the process easier.

A process that uses finite element models to automatically generate β tables for AFGROW has been developed. This capability is not restricted to specific geometries and loading configurations and thus gives the ability to improve KI solution accuracy by using local models that are more representative of the structure or even full component level models.

The process is implemented using a python script which calls the underlying 3D fracture mechanics modeling functionality to automatically insert cracks and perform solutions for a set of predefined crack sizes and shapes. Currently, the solution can be performed using Ansys, Abaqus or CalculiX as solvers.

KI verification benchmarking, using an exact 3D solution as a reference, is used to demonstrate solution convergence and accuracy for various elliptical crack aspect ratios.

The presentation summarizes the ongoing progress in developing the AFGROW FEA Handbook, highlights the benefits of on demand finite element analysis for crack growth analysis, and outlines the project roadmap along with key implementation challenges.

This presentation will cover the details of the upcoming release; bug reports and all things related to BAMpF!