Publications

Corrosion remains a critical issue in various industrial sectors, often being underestimated until significant damage occurs and heavy maintenance operation has to be carried out. Early detection and intervention can drastically extend asset life and reduce maintenance costs. Non-destructive testing (NDT) methods, widely used in the aeronautical industry [1], have proven effective in detecting structural damage, though they typically require human involvement. Recent advances in ultrasonic techniques, particularly the use of piezoelectric transducer (PZT) networks, offer promising alternatives to shift towards a condition-based maintenance approach. Building upon earlier work [2] that developed an ultrasonic Lamb wave technique for detecting electrochemically induced localized corrosion pit, this study expands the approach by integrating both electrochemical and mechanical analyses of crack initiation and propagation on Aluminum 2024 alloy. In our work, PZT discs generate and receive ultrasonic signals to monitor the evolution of a single corrosion and mechanical fatigue damage. This talk presents the results of our experiments, demonstrating how the combination of electrochemical control and ultrasonic monitoring provides a reliable method for tracking corrosion and crack growth. Embedding these sensors directly into strategic point of structures for continuous monitoring will reduce human intervention/error while generating data that could train machine learning algorithms for predictive maintenance strategies in the future. [1] S. Benavides, Corrosion in the aerospace industry, 2009. [2] C. Nicard, M. Rébillat, O. Devos, M. El May, F. Letellier, S. Dubent, M. Thomachot, M. Fournier, P. Masse, N. Mechbal, In-situ monitoring of µm-sized electrochemically generated corrosion pits using Lamb waves managed by a sparse array of piezoelectric transducers, Ultrasonics, Volume 147, 2025, 107527, ISSN 0041-624X, https://doi.org/10.1016/j.ultras.2024.107527

Due to the effects of fatigue and excessive vibration, structures may present a different health condition than initially observed. The change in system behavior due to damage also causes the system to demand more energy, which can saturate the actuators and make the system unstable. In this context, this work proposes to estimate in real-time the dynamics of a structure and control the undesired effects of vibration and damage. A data-driven model will be developed based on the Koopman operator, analyzing only data from sensors and actuators already installed by the control system. The information obtained by this updated model can be used to monitor changes in structural health and adapt a controller to meet performance specifications, even if the system dynamics vary over time. Changes in the spectral characteristics of the Koopman operator can help identify damage in the structure. In addition, an adaptive model predictive controller can incorporate the possible changes in dynamics in real time, adjusting the optimization problem according to the current estimated model. Therefore, our results have demonstrated the benefits and limitations of this online monitoring and control strategy based on data already measured by the controller. REFERENCES [1] N. Mechbal and E. G. O. Nóbrega, Damage tolerant active control: Concept and state of art. IFAC Proceedings Volumes, vol. 45, no. 20, pp. 63–71, 2012. [2] M. Korda and I. Mezić, Linear predictors for nonlinear dynamical systems: Koopman operator meets model predictive control. Automatica, vol. 93, pp. 149–160, 2018.

Corrosion is a major threat in the aeronautic industry, both in terms of safety and cost. Efficient, versatile, and cost affordable solutions for corrosion monitoring are thus needed. Ultrasonic Lamb Waves (LW) appear to be very efficient for corrosion monitoring and can be made cost effective and versatile if emitted and received by a sparse array of piezo-electric elements (PZT). A LW solution relying on a sparse PZT array and allowing to monitor µm-sized corrosion pit growth on stainless 316L grade steel plate is here evaluated. Experimentally, the corrosion pit size is electrochemically controlled by both the imposed electrical potential and the injection of a corrosive NaCl solution through a capillary located at the desired pit location. In parallel, the corrosion pit growth is monitored in-situ every 10 seconds by sending and measuring LW using a sparse array of 4 PZTs bonded to the back of the steel plate enduring corrosion. As a ground truth information, the corrosion pit volume is estimated as the dissolved volume balancing the electronic charges exchanged during corrosion. The corrosion pit radius is additionally checked post-experiment precisely with an optical measurement. Measured LW signals are then post-processed in order to compute a collection of synthetic damage indexes (DIs). After dimension reduction steps, obtained DI values correlates extremely well with the corrosion pit radius. Using a linear model relating those DI values to corrosion pit radius, it is demonstrated that corrosion pit from 30 µm to 150 µm can be reliably detected, located, and their upcoming size extrapolated. Two independent experiments were achieved in order to ensure the repeatability of the proposed approach. LW managed by a sparse PZT array thus appears to be reliable and efficient to monitor growth of µm-sized corrosion pits on 316L steel plates.

This research focuses on the structural health monitoring (SHM) of a foreign object damage (FOD) composite panel substructure of an aircraft engine fan blade equipped with an architecture array of novel screen printed piezoelectric sensors and is being carried out within the purview of the MORPHO – H2020 project. The state of the art printing technology ensures that architecture network of printed sensor is not only non-intrusive and lightweight but can also be printed during the manufacturing process before the structure goes into service. The FOD panel in this work is made of 3D woven composite, measuring approximately 800 mm x 350 mm, with a stainless-steel leading edge bonded to one of the longer edges and hosts a network of 5 arrays of 5 printed sensors each. The printed sensors can potentially be used in multiple ways to analyse the health of the host structure. Since the fabrication process of the sensors is an on-going research, first the electromechanical behaviour of the sensors is analysed with the help of impedance measurements. It is observed that the printing process ensures repeatability. Secondly, the performance of the printed sensors in case of impact loading is discussed here, as bird impact is one of the leading causes of engine fan blade failure. The impact response measured by the printed sensors is studied to detect the impact location on the FOD panel. Next, the ability of the printed sensors to sense ultrasonic guided wave responses generated by standard ceramic piezoelectric disc actuators is demonstrated. Finally, the health of these printed sensors upon undergoing multi-load multi-cycle bending tests is also discussed here. The ultimate goal of this project is to create diverse diagnostic and prognostic techniques for estimating the remaining lifespan and Structural Health Monitoring (SHM) of the FOD panel based on the range of measurements gathered using the printed sensors.

Structural Health Monitoring (SHM) has been gaining increased attention over the past decades as an important step towards Condition Based Maintenance (CBM). Measurements from the SHM systems provide the necessary information to monitor the condition of a (sub)component or structure and enable the use of this knowledge for maintenance task when needed, increasing availability and safety while reducing downtime related costs [1]. On the final level of SHM lie diagnostics and prognostics [2, 3], whose output inform about the current and future state of the (sub)component and their accuracy impact the effectiveness of the CBM decision making, and hence a capable sensor network is important. In this research our focus lies with monitoring the structural integrity of composite aircraft engine blades, through a capable network of SHM technologies. Engine blades are an important part of any aircraft and their integrity is imperative for its safe operation. A common yet critical damage case is impact damage (cause by hail or bird strikes) which can significantly reduce the load bearing capabilities of the blade. The aim is to demonstrate the feasibility, effectiveness and usefulness of different SHM systems in identifying the existence of damage, monitoring the damage and degradation growth and eventually use robust and reliable indicators to estimate the remaining useful life. To accomplish that task, a subpart of the engine blade is manufactured from 3D-woven CFRP preforms via the injection molding technique. The panels are curved and their length is 800mm while their width is 350mm. A secondary adhesively bonded steel edge is also adhered to the entire length of the panel with a width of 50mm. The panels are subjected to a repeated 4-point bending load-unload scheme with increased severity to simulate low frequency fatigue at ~0.02 Hz and introduce controlled and gradual degradation. First, one panel is subjected to 4-point bending quasi-static loading to determine the failure load which was approximated at 22kN through finite element analysis. Experimental collapse was reached at 28 kN and with this a guide the load envelope was decided. The loads include [4, 8, 12, 14, 16, 18, 20, 22, 24, 26, 28] kN and each load is applied for 400 cycles. Impact damage is also introduced to most of the panels in order to create a damage area to monitor. Regarding the SHM systems employed, the panel is equipped with state of the art FBG sensors and piezoelectric sensors. In this work, the data from the optical fibers are studied more in depth. The fibers can either be surface mounted (tensiled side of the panel) or embedded during the layup process or a combination of both. The majority of optical fibers run across the length of the panel at different width locations, two focused in the middle section, and one close to each of the edges. The sensors are used to monitor the strain field and it is attempted to correlate damage formulation and overall degradation with alterations to the strain field [4]. The first indication of degradation can be observed by analyzing the data collected from the hydraulic machine. By processing the load and displacement data, the experimental stiffness can be calculated as the slope of a linear equation between the load and displacement during the loading part. What was observed, is that at first the experimental stiffness slightly increases after the first load case, attributed to the increased robustness of the panel after significant bending. A somewhat constant stiffness follows, until the time close to failure, where rapid stiffness drop can be seen. This is accompanied by the first visible mode of damage in the form of skin tears and fiber cracking at the top surface close to the loading pins. Final collapse is dominated by matrix and fiber breakage close to the loading pin locations which extend across the entire width of the panel. These results are summarized in Figure 1 and Figure 2. FBG data are dependent on the type of FBG. Embedded FBG sensors display a mixture of tension and compression behavior while surface mounted show predominantly tension strains with increased intensity as the max load increases. An example of strains from embedded and surface mounted FBG sensors can be seen in Figure 3. The end goal of analyzing the FBG data is to extract capable indicators, similar to [5], and correlate the online SHM measurements to the degradation, as observed by the stiffness reduction, in a semi-quantitative way. In this research the use of strain based SHM sensors for degradation monitoring is demonstrated. Strain based indicators were used in an attempt to capture degradation evolution in large curved composite panels representative of aircraft engine blades. An unprecedented experimental campaign was launched on engine blade panels, which are subjected to fatigue-like 4-point bending load, and are mounted with state of the art SHM systems. The raw strains are transformed into an informative measure of degradation, demonstrating the feasibility, effectiveness and usefulness of such sensors for diagnostic and prognostic tasks, a first step towards a CBM paradigm. 1. Kessler, S.S. and S.M. Spearing, Design of a piezoelectric-based structural health monitoring system for damage detection in composite materials. Smart Structures and Materials 2002: Smart Structures and Integrated Systems, 2002. 4701: p. 86-96. 2. Ling, Y. and S. Mahadevan, Integration of structural health monitoring and fatigue damage prognosis. Mechanical Systems and Signal Processing, 2012. 28: p. 89-104. 3. Loutas, T., N. Eleftheroglou, and D. Zarouchas, A data-driven probabilistic framework towards the in-situ prognostics of fatigue life of composites based on acoustic emission data. Composite Structures, 2017. 161: p. 522-529. 4. Broer, A., et al., Fusion-based damage diagnostics for stiffened composite panels. Structural Health Monitoring-an International Journal, 2022. 21(2): p. 613-639. 5. Galanopoulos, G., et al., A novel strain-based health indicator for the remaining useful life estimation of degrading composite structures. Composite Structures, 2023. 306.

sustainable environments. Hydrogen, as a vector of energy, offers a high energy density and produces only water when used in fuel cells, making it an environmentally friendly alternative to fossil fuels. Hydrogen storage in gaseous form is typically achieved in type IV composite tanks. These tanks feature a thermosetting matrix, usually an epoxy reinforced with carbon fibers, which provides mechanical strength. Additionally, a polymer liner, onto which the composite is deposited via filament winding, ensures the hydrogen barrier function. A metal base is incorporated to facilitate the introduction of hydrogen into the tank. Type V hydrogen tanks are made from thermoplastic polymers and consist of a monolithic structure including an internal layer of unfilled polymer and a multilayer composite external layer made of the same thermoplastic matrix reinforced with continuous carbon fibres. The interface between the composite and the liner is thus assumed to be perfect. These tanks will be subjected to the different types of thermomechanical loads as type IV tanks: nominal pressure of 700 bars, burst pressure of 1750 bars, temperature ranging from -60°C to 85°C, fill-drain cycles, shocks, and impacts (figure 1). However, in terms of service life, the difficulty of damage detection and measurement and monitoring of physical quantities within composite structures still requires the development of specific and reliable technological tools and methods, such as the integration of sensors into the composite structure. The interest in integrating such a function lies in the potential for in-situ health monitoring of the structure from its implementation (Process Health Monitoring - PHM) to its service life (Structural Health Monitoring - SHM). Mastering this technology in all its aspects could significantly reduce maintenance costs and ensure better durability while optimizing the manufacturing process. The integration of sensors into hydrogen storage structures is essential to advancing the use of hydrogen as an energy carrier in transportation. Integrated sensors can provide an easy and effective means to monitor manufacturing quality and mechanical performance of the tanks during their service life. Indeed, these sensors contribute to several essential functions of so-called "smart tanks" currently in development. However, the integration of sensors is complex and poses several challenges such as type of sensors, the methodology of integration of the sensors. To overcome these challenges, the proposed methodology aims to establish a structured approach for selecting and/or optimizing the Process-Material-Sensor (PMS) trio to ensure the reliability of the mechanical and electronic functions integrated within hydrogen tanks. The proposed approach involves combining an experimental method with a numerical one. Additionally, it leverages cross-disciplinary skills, including: • The characterization and modeling of the Mechanical behavior of materials and the optimization of tank performance. • The processes involved in the manufacturing of hydrogen tanks (plastics engineering). • The integration and use of sensors, particularly the analysis of electronic signals to correlate with the assessment of the tank's degradation state. In this paper the development of a highly transversal global approach that called Mechaplastronic has been proposed. Indeed, it can be considered that optimizing the integration of sensors within a tank involves studying different types of intrusiveness: A) Intrusiveness of the Fabrication Process on Electronic Functions: Firstly, when sensors are integrated within the materials or at their interfaces, they must be compatible with the fabrication process of the structure (Automated Fiber Placement (AFP) / Filament Winding). Therefore, it is necessary to study the impact of the process on the operating mechanisms of the sensors and the reliability of its electrical and magnetic properties. This involves optimizing the process parameters and the choice of materials. B) Mutual Intrusiveness of Electronic and Mechanical Functions: The presence of sensors creates interfaces and stress concentrations that can weaken the structures. Therefore, it is important to study the effect of the presence of sensors on the mechanical properties of the structural materials (especially the composite). Conversely, the various thermomechanical stresses endured by the structure can affect the operation of the sensors. The proposed methodology, aiming to control the different potential intrusiveness and their interactions within a hydrogen tank, can be schematized as illustrated in figure 2. Thus, this study aims to enhance the understanding of the performance of critical areas in H2 tanks. The entirety of this study's results, along with the understanding of the phenomena involved, particularly damage, constitutes an important experimental and numerical foundation.

The study of the effect of the environment on the fatigue strength of metallic materials has been, and still is, the subject of a great deal of research. Although the phenomena of crack initiation and propagation in fatigue-corrosion appear to be well understood, they remain highly dependent on several parameters linked to the environment studied and the material. Numerous dimensioning approaches exist to consider the effect of corrosion on fatigue life. Some treat the two problems (fatigue and corrosion) separately, while others attempt to model the complex effects of cyclic loading/corrosion synergy. A multi-physics and multi-scale experimental approach was developed to better understand crack initiation and propagation mechanisms in fatigue/corrosion coupling situations. This approach is based on the multi-instrumentation of fatigue-corrosion tests, combining electrochemical and optical microscope measurements. More recently, lambs-wave monitoring has been introduced to track the evolution of damage and to propose a self-health monitoring based on physical evidence. Based on the results of in-situ electrochemical measurements and optical observations, the mechanisms and laws of fatigue/corrosion coupling have been identified: • Corrosion pitting initiation is controlled by an original experimental setup developed to generate a controlled size of a single corrosion pits in the middle of the fatigue specimen equipped with four piezoelectric sensors to track the corrosion fatigue damage, • The size of the corrosion pit is estimated from the measured electrochemical current using the faraday law, • The first phase of crack propagation (short crack regime) is highly dependent on corrosion. Material dissolution and applied stresses contribute to lowering the crack initiation threshold and increase the kinetics of short fatigue crack propagation, • In the Paris fatigue crack propagation regime, crack growth kinetics are higher than in air. This detailed understanding of damage mechanisms has enabled us to: • identify the parameters and laws governing fatigue-corrosion crack initiation and propagation, • determine the pit-to-crack transition step using in-situ lambs-wave monitoring and optical microscope observation, • and propose an analytical model for estimating fatigue-corrosion life. Microstructure/cyclic loading/corrosion interactions and the role of hydrogen, introduced during fatigue-corrosion processes, in crack initiation and propagation mechanisms remain to be studied.

al components, roughening the outer surface, loosening fasteners, hastening cracking, and facilitating the entry of water into electronic fixtures. In 2016, the combined commercial aircraft fleet operated by European airlines was around 7900 airplanes. The annual corrosion cost for this number of aircraft was estimated by the US National Association of Corrosion Engineers to 2.2 B$, which includes corrosion maintenance at 1.7 B$ and downtime due to corrosion to 0.3 B$. Anticipating corrosive conditions ahead of time can lead to significant cost savings and less aircraft downtime. It is estimated that savings between 15% and 35% of the cost of corrosion could be realized [1]. In order to limit corrosion issues, a typical aeronautic structure is made of qualified steel or aluminium alloys eventually protected from the environment by treating the surface with adequate coatings. It is generally expected that the protective surface is perfectly flawless over the whole structure. But on a large structure undergoing everyday service this cannot be the case. Impact damage during service life is a common occurrence of coating failure. Damage during maintenance (tools and split fluids) can also occur as well as paint cracking at high stress points around joints. From these initial premises, corrosion pits can start and threaten the structural integrity at a global level, thus motivating the need for in-situ pitting corrosion monitoring technologies. Industrially speaking, an ideal corrosion monitoring technology should allow to monitor large-scale structures, to automate measurements, to detect corrosion pits from their premises, and to exhibit a high correlation between sensors measurements and the size and locations of corroded areas. Although various effective non-destructive testing (NDT) methods have been developed to monitor corrosion [2], they remain limited in their ability to detect and assess corrosion premises and to reliably size a given corrosion damage. The first reason for that is technological. Eddy currents allow local monitoring (one side), are slow, difficult to calibrate, and remain a small-scale approach. Standard ultrasonic methods (A-scan and C-scan) are again small scale, rely on coupling fluid & human intervention and are thus rather slow. Optical methods (visual inspection, liquid penetrant) are large scale but can cope only with surface features that can be visually inspected and remain difficult to interpret. Radiography (X-rays, tomography) is small scale and relies on extensive hardware that cannot be used on the field. The second reason is methodological. Actual monitoring procedures are referred as planned maintenance and imply regular parts inspection by means of visual inspection and non-destructive testing through human highly qualified operators. Many of these inspections are (fortunately) unnecessary and as they rely on human intervention, their reliability is then subject to confidence intervals that establish a certain probability of detection. Moving from planned maintenance to condition based maintenance (CBM), i.e. maintenance only when necessary, seems mandatory to solve these issues. Monitoring in real time and autonomously the health state of structures is thus of high interest in this context. Such a process is referred to as Structural Health Monitoring (SHM) [3, 4]. To achieve this goal, structures become “smart” in the sense that they are equipped with sensors, actuators, and algorithms that allow them to state autonomously regarding their own health. One can compare smart structures with the human body which, thanks to its various senses and nerves, is able to assess if it has been hurt, where it has been hurt, and to estimate how severe it is. Following this analogy, the SHM process is classically decomposed into four steps: damage detection, localization, classification, and quantification [5]. There is thus a real need for a large scale, automated, sensitive, low cost and embedded technology to develop corrosion monitoring SHM methodologies of aeronautic parts. Ultrasonic technologies relying on Lamb waves have been shown to be extremely sensitive to corrosion damage [6, 7] and can be easily automated thanks to permanently embedded piezoelectric elements (PZT) networks [8]. In the present work, the focus is thus put on corrosion monitoring methods based on ultrasonics waves applied experimentally to plate-like structures. A literature on that topic survey has shown that ultrasonic inspection method based on spare PZT transducers arrays are the most suitable from a practical point of view but that they have never been validated on small corrosion pits realistic of actual corrosion (from 10 µm to 100 µm) but rather on large generalized corrosion areas (in cm). The reason explaining the fact that it was yet not possible to generate a single corrosion pit in a controlled manner and to simultaneously and in-situ record ultrasonic Lamb waves interacting with such a damage. The objective of this contribution is thus first to introduce an experimental setup allowing to control electrochemically the growth of a µm-sized corrosion pit and to simultaneously monitor it in situ by means of ultrasonic Lamb Waves emitted and received using a sparse array of PZT transducers. It will secondly be demonstrated that such measurements can be post-processed to compute damage indexes (DIs) that correlate very well with actual corrosion pit radius as estimated electrochemically and that using a linear model relating DI values to corrosion pit radius, corrosion pit size from 10 µm to 150 µm can be reliably detected, located, and their upcoming size extrapolated. [1] S. Benavides, Corrosion control in the aerospace industry, Elsevier, 2009. [2] S. J. Harris, M. Mishon and M. Hebbron, "Corrosion sensors to reduce aircraft maintenance," in Workshop on enhanced aircraft platform availability through advanced maintenance concepts and technologies. , 2006. [3] D. Balageas, C.-P. Fritzen et A. Güemes, Structural health monitoring, vol. 493, Wiley Online Library, 2006. [4] C. R. Farrar et K. Worden, «An introduction to structural health monitoring,» Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, vol. 365, pp. 303-315, 2007. [5] K. Worden, C. R. Farrar, G. Manson et G. Park, «The fundamental axioms of structural health monitoring,» Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, vol. 463, pp. 1639-1664, 2007. [6] C. Jirarungsatian and A. Prateepasen, "Pitting and uniform corrosion source recognition using acoustic emission parameters," Corrosion Science, vol. 52, pp. 187-197, 2010. [7] P. Kudela, M. Radzienski, W. Ostachowicz and Z. Yang, "Structural Health Monitoring system based on a concept of Lamb wave focusing by the piezoelectric array," Mechanical Systems and Signal Processing, vol. 108, pp. 21-32, 2018. [8] S. Grondel, C. Delebarre, J. Assaad, J.-P. Dupuis and L. Reithler, "Fatigue crack monitoring of riveted aluminium strap joints by Lamb wave analysis and acoustic emission measurement techniques," Ndt & E International, vol. 35, pp. 137-146, 2002.

Validation and testing of Lamb wave based SHM algorithms requires numerous simulations that require themselves qualitatively and quantitatively consistent models representative of the physical behavior of the monitored structures and that are in agreement with experimental data. Finite elements models appear as an interesting solution to achieve this goal but are associated with large computational costs and low generalization abilities. On the other hand, data driven machine learning approaches are computationally very efficient and can predict fine details but at the cost of low physically interpretability. Original approaches trying to build physically informed models balancing the advantages and drawbacks of physics-based approaches and of machine learning approaches also exist and will be discussed in the context of Lamb waves based SHM of aeronautical structures.

METHODS : Eighty participants aged 20 to 79 years (nearly even distribution), who self-reported no history of significant health conditions and with no neck pain, were recruited. Two custom apparatus were used to support participants in relaxed lying. Their head was rotated to maximum ROM; applied moment and head-torso motion were recorded. Muscle activation was monitored in real-time to ensure electromyographic signals from agonist muscles remained below a passive threshold. Stiffness was determined from the moment-angle data within each of three zones, with zone boundaries delineated to maximize moment-angle linearity within each zone. The age and sex effects on passive stiffness and ROM were assessed using generalized linear models for flexion and extension, and linear mixed models for lateral bending and axial rotation. RESULTS : Passive neck ROM decreased by 0.2° per year of age in lateral bending and axial rotation for males and females, and extension ROM for males was 5.8° lower than for females. Passive stiffness in lateral bending (zone 1 and 2: 0.9 and 3.5 Nmm/°/year; zone 3: 3%), axial rotation (zone 1 and 2: 1%; zone 3 for males and females: 1.9 and 0.9 Nmm/°/year) and some zones in extension (zone 2: 0.8 Nmm/°/year; males in zone 3: 2.7 Nmm/°/year) increased with age, and males had higher stiffness than females in lateral bending (zone 1 and 2: 22.3 and 43.9 Nmm/°; zone 3: 35%) and axial rotation (zone 1 and 2: 49% and 35%). CONCLUSIONS: Passive neck ROM decreased with age in lateral bending and axial rotation, while passive neck stiffness tended to increase with age in all motions but flexion. Extension ROM was higher for females, and lateral bending and axial rotation stiffness at lower angles were higher for males. CLINICAL SIGNIFICANCE: The neck ROM, stiffness, and moment-angle corridors developed in this study provide benchmarks for clinical assessment of cervical spine function, and can assist the development of surrogate and computational models incorporating minimal muscle activation, for injury simulation and clinical skill training.