Jeremy S. Knopp

Investigation of a Model‐Assisted Approach to Probability of Detection Evaluation

This paper presents a model‐assisted probability of detection (MAPOD) study for inspection of a two‐layer airframe structure. Eddy current measurements for varying crack length around fastener holes in a two‐layer aluminum structure are studied using both experimental and model‐generated data. New statistical algorithms are used to calculate the probability of detection. Good agreement was achieved between empirical and model‐assisted approaches.

MODEL‐ASSISTED PROBABILITY OF DETECTION EVALUATION FOR EDDY CURRENT INSPECTION OF FASTENER SITES

A model‐assisted approach for the design and execution of probability of detection (POD) studies is proposed. General agreement was achieved between experimental and full‐model assisted results for eddy current inspection of cracks at fastener sites located at both the first and second layers. The accuracy of the POD results was found to be dependent upon the NDE model and assumptions in the model‐assisted POD evaluation. Insight is presented for improving the quality of future studies.

Aging Aircraft NDE: Capabilities, Challenges, and Opportunities

US Air Force aircraft are managed by Damage Tolerance Assessment (DTA) and Nondestructive Evaluation (NDE) plays a critical role in insuring their flight safety. This paper provides an overview of current NDE capabilities for aircraft structures. The potential migration to Condition‐Based Maintenance (CBM) will require NDE to evolve from detection of defects to characterization of their location and size. This represents multiple significant challenges as aircraft have many material and geometry factors that complicate this analysis. This paper provides a strategy to resolve these factors.

Computational methods in eddy current crack detection at fastener sites in multi-layer structures

Reliable detection of both surface and subsurface cracks around fastener sites continues to be a need for maintaining ageing aircraft structures. In this work, a comprehensive overview of both experimental and modelling work addressing eddy current crack detection around fastener holes in multi-layer structures is presented. This overview consists of a strategy for the application of computational methods along with case studies that demonstrate the role of modelling for this particular class of problems. The state-of-the-art on eddy current modelling applicable to low-frequency eddy current problems is reviewed. The role of modelling in feature extraction algorithm development, crack characterisation, and probability of detection evaluation is highlighted.

Ultrasonic plate waves for fatigue crack detection in multi-layered metallic structures

A representative area of concern for fatigue crack growth in aircraft occurs in multi-layered metallic structures. Ultrasonic plate waves are currently being investigated by multiple initiatives to detect these types of flaws with a minimal number of sensors to enable Structural Health Monitoring (SHM). Previous work has focused on structures with one or two layers, coupled with modeling of the wave propagation within these representative samples. However, it is common for multi-layered structures to have more than two layers in many areas of interest. Therefore, this study investigates ultrasonic wave propagation and flaw detection in a multi-layered sample consisting of 2 to 4 total layers with fatigue cracks located in only one layer. The samples contain fastener holes configured as would be expected to find on typical aircraft structure. The flaws in this study are represented by electric discharge machined (EDM) notches. Preliminary measurements show that EDM notches can be detected by the guided ultrasonic waves, but that the sensitivity to EDM notch location is dependent on the boundary conditions of each layer. The boundary conditions are changed by applying various loads on the surface of each layer by tightening and loosening the fasteners that hold the sample together. This variation depicts representative conditions found of aircraft. The experimental results are supplemented by modeling of the guided wave propagation within the structure using the Finite Element Method. The primary parameter studied in the modeling effort is the effect of the changes in the boundary condition on the mode and amplitude of the guided wave. The results of this investigation establish some guidelines for the use of guided waves in multi-layered structures, plus challenges that exist for their use in SHM applications and strategies to address these challenges.

CASE STUDIES FOR MODEL‐ASSISTED PROBABILISTIC RELIABILITY ASSESSMENT FOR STRUCTURAL HEALTH MONITORING SYSTEMS

This paper describes progress on the development of a protocol and presents several case studies for model‐assisted probabilistic reliability assessment of structural health monitoring (SHM) technologies. The design and progress of an experimental demonstration study is presented highlighting the protocol and test plan. A simulation‐based study for an SHM system incorporating vibration methods is also introduced to demonstrate the protocol process. Probability of detection and probability of correct characterization curves were generated for different transducer locations providing key insight on sensor placement and expected detection and characterization performance.

Bayesian methods in probability of detection estimation and model-assisted probability of detection evaluation

In this paper, the application of Bayesian methods for probability of detection (POD) estimation and the model-assisted probability of detection methodology is explored. A demonstration of Bayesian estimation for an eddy current POD evaluation case study is presented and compared with conventional approaches. Hierarchical Bayes models are introduced for estimating parameters including random variables in physics-based models. Results are presented that demonstrate the feasibility of simultaneously estimating model calibration parameters, model random variables and measurement error.

Demonstration of model-assisted probability of detection evaluation methodology for eddy current nondestructive evaluation

A model-assisted POD (MAPOD) evaluation approach is explored for the eddy current inspection of fatigue cracks in titanium. Experimental results are presented comparing crack and EDM notch responses and different eddy current measurement systems. The MAPOD evaluation process is performed using a simulated case study, highlighting the benefit of leveraging better models to mitigate variation in the model fit and minimize test sample requirements. The benefit of validating NDE techniques that use inverse methods to estimate uncontrolled measurement conditions is also studied.

Demonstration study for reliability assessment of SHM systems incorporating model-assisted probability of detection approach

This paper presents the results of a demonstration featuring the application of a validation protocol to a vibration-based structural damage sensing system. The results of the full validation study highlight the general protocol feasibility, emphasize the importance of evaluating key application characteristics prior to the POD study, and demonstrate an approach to quantify varying sensor durability on the POD performance. Challenges remain to properly address long time-scale effects with accelerated testing and large testing requirements due to the independence of the inspection of each flaw location.

Comparing the Effectiveness of a90/95 Calculations

Most practitioners see a90/95 as a static, single‐point summary of an entire inspection’s capability. It purports to be the size of the target having at least 90% probability of detection in 95 of 100 POD experiments under nominally identical conditions. But in some situations the actual coverage is closer to 80%, rather than 95%, with 50% coverage being the median POD(a) curve itself. This paper discusses the two philosophies, the Wald Method, and the Loglikelihood Ratio Method, for constructing lower bounds on POD(a) curves (and therefore determining a90/95) and compares the effectiveness of each as functions of other experimental realities such as sample size and balance.