Juan Ocampo

Development of a probabilistic linear damage methodology for small aircraft

Arisk assessment evaluation of the continued operational safety of the general aviationfleet canprovide important insight to the criticality/severity of a potentially serious structural issue. As such, the methodology and a computer code, SMART|LD, were developed to address risk assessment and risk management of general aviation structural issues. This information will provide a proactive approach to enable a nonbiased review of data to assure airworthiness. To address variability and uncertainties in loading and material properties, a probabilistic methodology was developed and implemented that considers the random loading and probabilistic stress-life material behavior developed from constant amplitude tests. The stress severity factor is used to account for the geometry effects (notches, fastener holes, etc.). Failure is determined using the Miner damage index with the index calibrated from simulations of variable amplitude tests. Monte Carlo sampling is used to calculate the structural probability of failure, or mean and standard deviation of flights and hours to failure, the hazard function, and sensitivity analysis. A numerical example is presented to demonstrate the methodology.

Distributed computing for probabilistic structural integrity analysis of aircraft structures

Probabilistic risk assessment of aircraft structures is inherently time consuming as multiple probabilistic structural integrity analyses are required to account for the airplane-to-airplane and ight-to-ight variations and each probabilistic analysis requires a sucient amount of Monte Carlo samples (MCS) to insure convergence. In this paper, high performance computing tools, OpenMP and Message Passing Interface (MPI), are investigated with an aim to signicantly reduce the computational time required for risk assessment. Although the inherit independence of MCS is favorable, ecient distributed computing demands additional coding work and tuning. OpenMP and MPI directives are implemented to parallelize the risk assessment analysis and tested with a dierent number of processors. The results show a 6:87 times speedup for OpenMP implementation on 8 processors and 288 (50; 000 samples) and 312 (200; 000 samples) times speedup for MPI implementation on 512 processors.

Simulation‐based crack growth mitigation through optimum laser peened residual stress

Purpose – The purpose of this paper is to develop an approach to optimize the cycles‐to‐failure of a peened component with respect to laser peening (LP) variables: pressure magnitude, mid‐span, and spot size when the component is subject to a variable amplitude loading.Design/methodology/approach – To optimally design an LP process, an experimentally validated 3D finite element simulation of the LP process, a cycles‐to‐failure estimation capability incorporating residual stress, and a particle swarm optimization strategy were developed and employed to maximize the cycles‐to‐failure of a component of a titanium turbine disk.Findings – The most critical finding of this research is that a minor difference in the residual stress profile can lead to a large difference in the cycles‐to‐failure. This finding implies that selecting the optimization objective to be the cycles‐to‐failure is a better option as compared to the residual stress profile.Research limitations/implications – The LP‐induced residual stresse...

High Performance Computing Implementation on a Risk Assessment Problem

Risk assessments of engineering structures are notoriously time consuming. Due to the aging of current structures, risk evaluations are needed more often, and in many cases decision makers need the results in almost real time. This work aims to evaluate the impact of using compiler optimizations, Message Passing Interface (MPI), and OpenMP directives for a Monte Carlo sampling fatigue code used in risk assessment of General Aviation. Compiler optimization permits high performance speed with accurate solutions without additional coding effort. OpenMP and MPI provide directives and functions to parallelize programs but demand more coding work and sometimes do not give large speed up results. These two methods can meet the growing need for real time risk assessments.

Probabilistic Damage Tolerance Analysis for General Aviation

The General Aviation (GA) fleet includes about 150,000 airplanes that were certificated with no fatigue evaluation requirements. The average age of these airplanes is about 40 years, and many are high-time. To mitigate the aging effects on the GA fleet, a probabilistic damage tolerance analysis (PDTA) program has been developed. A PDTA approach also provides a mechanism whereby inspection and maintenance operations can be included into the simulation, thus providing engineers the opportunity to assess the benefits of maintenance actions. This paper describes the probabilistic methodology to be utilized in a computer software program (SMART|DT) that performs risk assessment of small airplanes employing NASGRO® or a user selected code as the crack growth engine. The methodology can assess a range of random variables, calculate the extreme value distribution (EVD) of maximum stress per flight from a general aviation (GA) spectrum, and generate a surrogate model for accurate and fast calculations of crack grow. The main objective is to develop a comprehensive probabilistic methodology such that engineers can conduct a risk assessment of GA structural issues in support of policy decisions.

Probabilistic damage tolerance for small airplanes using a linear-elastic crack growth fracture mechanics surrogate model

Most general aviation (GA) aircraft are designed for safe-life based upon a crack initiation type failure mechanism, e.g., Miner’s rule. However, newer GA aircraft have fatigue crack growth as a design option. In addition, it may be necessary to evaluate a field event such as a cracked structure to ascertain the remaining life. Therefore, a risk based probabilistic damage tolerance analysis (PDTA) program is needed in several aerospace situations. A comprehensive probabilistic damage tolerance method requires a combination of deterministic crack growth, inspection methods, probabilistic methods, and random variable modeling to provide a single probability-of-failure, cumulative probability-of-failure, and hazard rate with and without inspection. In this work, a general methodology to conduct probabilistic crack growth based damage tolerance methodology for small airplanes will be developed and incorporated in a computer software. Random variables can be included in the model using Monte Carlo Sampling (MCS) and efficient numerical integration algorithms. Probabilistic damage tolerance analysis involves mathematically complex models and computational expensive simulations, which makes these analyses very inefficient. In this work the computational weight will be reduced using an error based adaptive surrogate model; the surrogate model will include the most influential random variables. The surrogate model will be used as a temporary substitution for the original crack growth model. An example problem will be presented to demonstrate the methodology.

Investigation of magnetostrictive sensor capabilities for crack growth prediction in second-layer structure

In this study, magnetostrictive sensor (MsS) technology was used to detect cracks and predict crack length in double lap joint specimens representative of aircraft substructure. It was shown that the MsS system was successful in detecting crack growth, and could predict the second-layer crack length to within 10% of the measured crack length using a simple geometry-based model. Inconsistent sensor placement was determined to be a large source of error; however, this can be easily addressed in the future by bonding the sensor to the test specimen.

Sensitivity Analysis for General Aviation Risk Assessment

General Aviation structural risk assessment is of vital importance to ensure airworthiness and safety. However, it is even more important to understand the role of all the different variables and their importance or sensitivity over the airplane life in order to reduce the risk. Based on this idea, different sensitivity methods (scatter plots, parallel box plots, and global sensitivity) were applied to the software SMART (Small Aircraft Risk Technology). SMART is probabilistic fatigue linear damage software developed by the University of Texas at San Antonio (UTSA) to conduct risk assessment in general aviation (GA) airplanes. SMART computes the airplane structural life considering variables such as: maneuver and gust load limit factors, ground stress, one-g-stress, flight length-velocity matrix, flight length-weight matrix, exceedance curves, Miner’s damage coefficients, and stress life curves, considering randomness in some of the variables. The results from the sensitivity methods indicate that the variables one-g stress, gust and maneuver loads, PSN curve and damage coefficient play an important role in the airplane life and more caution should be focused on these variables.

Extreme value modeling and parametric investigations of gust and maneuver loads for General Aviation

An extreme value distribution (EVD) of the maximum load per flight of a load spectrum is critical for a probabilistic damage tolerance analysis of a General Aviation aircraft. The EVD parameters are important because the structural integrity of the aircraft depends upon the maximum load seen by the structure during a specified number of flights. It is well known that the load spectrum that an aircraft experiences depends upon a large number of variables including number of flights, type of usages, number of usages, percentage of each usage, maneuver and gust load limit factors, aircraft velocity, flight duration, ground stress, one-g-stress, exceedance curve, and randomness in these variables. This research investigates the effect of three selected variables (type of usage, exceedance curve, and flight length-velocity and flight length-weight matrices) on the maximum load per flight EVD. A computer code (load module) capable of generating a realistic load spectrum for a given set of loading parameters was developed. A generalized extreme value approach was developed to estimate the EVD of the maximum load per flight. A number of parametric investigations were performed to determine the effect of load spectrum variables on the EVD parameters. The preliminary results indicate that exceedance curves and type usage have the largest effect on the EVD parameters.

Probabilistic risk assessment for small airplanes

*† ‡ § ** †† ‡‡ §§ This paper describes the development of a probabilistic methodology that can perform risk assessment of small airplanes. The objective was to develop a comprehensive probabilistic methodology to allow Federal Aviation Administration (FAA) engineers to conduct a risk assessment of general aviation (GA) structural issues in support of policy decisions. Requisite-supporting technology and data issues, in particular, probability distributions of relevant inputs, were investigated and developed so that a realistic risk assessment could be obtained. Example problems are presented to demonstrate the methodology that includes the calculation of flights (or hours)-to-failure and the probabilityof-failure for a specified number of flying hours. Representative sensitivity studies were also conducted to determine significant variables.