A novel approach for detecting bolt tightness loss is presented, employing real-time measurements from washer-shaped piezoresistive nanocomposite sensors. The sensor is composed of natural rubber embedded with carbon nanotubes (CNT) and carbon black (CB), which together introduce conductivity within the rubber matrix. The electrical conductivity of nanocomposites with different CNT and CB contents was first evaluated under static conditions and subsequently investigated through compression tests to assess their electrical response under varying loading levels, confirming the piezoresistive behavior of the nanocomposites. The washer-shaped sensor was used to monitor bolted joints in two configurations: bolts arranged in one dimension and in two dimensions. The shape of the sensor allows it to be installed along with the washer and receive the compression load that the bolt applies to the structure. In the two bolted joints, specific bolts were subjected to loosening at different levels to assess the detection of damage in the structure. The sensor reliably detected torque variations, including in bolts that were not directly loosened, demonstrating its capability to monitor entire joints without instrumenting every bolt, because of the redistribution of loads caused by the loosening of one of the bolts. Tests conducted on the bolted joint under dynamic excitation and different bolt tightening levels demonstrated that the sensor did not significantly affect the dynamic properties of the structure, except when the bolts were extremely loose, a condition under which structural damping tends to increase considerably. The proposed approach demonstrates the capability to identify bolt loosening, with the sensor providing qualitative measurements that indicate the presence of damage and showing the potential to detect torque variations on the order of 0.2 Nm. Therefore, the proposed sensor is capable of identifying and monitoring loosening in bolted joints, combining ease of installation, data acquisition, real-time monitoring, and decision-making, while maintaining a low implementation cost.

Nonlinear dynamical systems with structural damage present significant challenges for analysis and monitoring due to their complex responses, the presence of multiple interacting nonlinearities, and the sensitivity of system behavior to small changes in conditions. These complexities are further compounded when the system exhibits both inherent nonlinear effects, such as those induced by magnetic forces, and localized damage mechanisms like breathing cracks, which can introduce time-varying stiffness and intermittent contact phenomena. This study investigates a mechanical system exhibiting both magnetically induced nonlinear behavior and damage in the form of a breathing crack. By relying solely on system output measurements, the research eliminates the need for input excitation data, offering a practical approach for complex or inaccessible systems. Transmissibility and coherence functions are then employed to exploit nonlinear dynamics, accounting for potential damages and uncertainties. These functions facilitate the identification and characterization of harmonics and other nonlinear features, while also suppressing the influence of unknown input excitation effects. Experimental investigations involve measuring outputs from multiple locations along the structure to explore the distinct features identified by a combination of each pair of output signals used in computing the transmissibility and coherence functions. The results demonstrate the capability of these functions to reveal key signatures of nonlinearity and damage, supporting their use in structural health monitoring and fault diagnosis for complex systems in modern engineering applications.

Piezoresistive nanocomposites have emerged as a promising alternative for structural monitoring, combining low sensor cost, simple fabrication and analysis, and real-time measurement capability. Despite advances in developing new nanocomposites and their experimental applications, there remains a gap in studies addressing the mathematical modeling of their electromechanical behavior. This work presents the development of a constitutive model for a piezoresistive nanocomposite composed of a natural rubber matrix reinforced with carbon nanotubes (CNT) and carbon black (CB), which has previously been employed to detect tightening loss in bolted joints with promising results. The sensor has a washer-shaped geometry and is installed between the bolt washer and the bolted joint interface, so that the deformation generated in the joint alters the electrical response of the sensor. Samples with different nanofiller proportions were fabricated and characterized to extract their piezoresistive properties. Mechanical stress and electrical conductivity under cyclic deformation were approximated by linear relationships whose coefficients varied with nanofiller concentration. Subsequently, a nonlinear formulation was proposed to describe the combined influence of CNT and CB on these coefficients. The model was calibrated using Bayesian inference, enabling the estimation of both conductivity and mechanical stress under various deformation conditions. The proposed constitutive model can be integrated into multiphysics simulation software, enabling virtual testing of the sensor within complex structures and optimizing its characteristics for different applications. The agreement between simulated and experimental results confirms the model’s capability to simulate and tune the sensitivity of nanocomposite sensors by adjusting CNT and CB contents, facilitating the design of customized real-time structural monitoring sensors for specific bolted joint configurations.

Bolted joints are widely used in industrial structures, but they require regular inspections to prevent failures such as bolt loosening. Conventional inspections are costly and time-consuming, as they involve manual checks and downtime. Therefore, indirect detection of loosening is a major challenge, further complicated by variability and nonlinear effects. This paper proposes a damage detection approach that avoids feature extraction by using the raw frequency response signal. A Gaussian Process Regression (GPR) model is directly trained on transmissibility functions to build a baseline model in the frequency domain. A GPR-based damage index is then formulated to detect outliers relative to this baseline. To improve sensitivity, a Global Sensitivity Analysis (GSA) with Sobol’s indices is applied as a feature selection method, identifying frequency ranges most affected by loosening. The paper also extends the GPR-based model to a three-dimensional formulation that captures the evolution of nonlinear behavior as input levels increase. In this way, the GPR-based damage index can distinguish different motion regimes and separate changes caused by nonlinear regime transitions from those caused by loosening. In summary, with proper feature selection, the proposed probabilistic frameworks reliably detect bolt loosening while minimizing false diagnoses in a clear way.

Identifying features in nonlinear structures, especially when they already exhibit inherent nonlinear behavior in the undamaged conditions, remains a significant challenge in structural health monitoring (SHM), particularly when the damage mechanism is also nonlinear. This complexity further increases in field applications, where controlling and measuring excitation is particularly challenging. Traditional methods, which rely on frequency response functions (FRFs), are unsuitable for nonlinear structures because they require knowledge of the excitation source and cannot effectively separate inherent nonlinearities from structural changes. This paper presents an output-only approach that combines transmissibility functions and kernel principal component analysis (KPCA) for damage detection in structures with nonlinear behavior caused by cubic stiffness and breathing crack effects. A numerical validation on a nonlinear multi-degree-of-freedom (MDOF) system with multiple resonances, followed by an experimental application on a nonlinear flexible beam, was performed to demonstrate the method’s generalization and practical relevance. The experimental beam exhibits cubic nonlinearity in the undamaged condition, while the simulated damage, a breathing crack, introduced additional quadratic stiffness effects. KPCA successfully separated cubic and quadratic nonlinear effects by analyzing transmissibility functions between accelerometer pairs, enabling accurate damage detection. This framework addresses the limitations of FRF-based methods, providing a straightforward and valuable tool for practical applications in SHM that does not rely on complex machine learning techniques.

Fouling is a prevalent challenge that requires precise monitoring and identification to maintain optimal heat exchanger performance. Typically, monitoring methods focus on controlled variables such as pressure, temperature, and the flow rates of hot and cold fluids. However, while some heat exchangers benefit from extensive historical data and comprehensive operational knowledge, others face limitations due to incomplete or insufficient information. This paper proposes a methodology that combines data-driven modeling with transfer learning techniques, specifically Joint Distribution Adaptation (JDA), to diagnose the state of the heat exchanger. Using temperature data from a well-documented source heat exchanger, we identify a reference auto-regressive (AR) model under clean conditions to serve as a digital shadow of the physical system (heat exchanger). As the operating conditions of the heat exchanger change, deviations (errors) between the observed and predicted behavior evolve, providing a basis for monitoring. These changes are analyzed by extracting features from the error signals. To track the impact of fouling on system performance, a support vector machine (SVM) classifier is used to monitor the evolution of the global heat transfer coefficient. The knowledge and expertise gained from the source heat exchanger are then transferred to a target heat exchanger using a JDA algorithm, which may differ in geometry or operating conditions but requires more data to identify a new model or train a classifier from scratch. The results demonstrate the specific contexts and scenarios in which the trained and classified model from the source heat exchanger can be effectively applied to the target exchanger, highlighting the potential of this approach to extend operational insights to data-limited systems.

Delamination is a typical form of damage in laminated composites that must be detected and quantified early to prevent severe consequences, such as total structural failure. However, experimental testing to investigate this kind of damage mechanism can be expensive and, in some cases, unfeasible. The main contribution of this paper is to detect and quantify delamination damage using a cost-effective regression technique called co-Kriging, which combines two datasets of distinct natures. By leveraging a second, simpler, and more affordable co-variable, co-Kriging improves the prediction accuracy of a target variable while reducing data acquisition costs. The method integrates Lamb wave datasets from low-cost numerical simulations with experimental laminate tests. Based on a simplified plate theory model, the numerical data includes a basic delamination model to capture its influence on signal features without requiring a complex or fully detailed model. This numerical data is then integrated with experimental results from a laminate with piezoceramics used for actuation and sensing. The combined datasets are utilized to develop a surrogate model that provides more accurate predictions of the size of the delamination area. The results showcased that co-Kriging, by fusing low-cost numerical data with experimental information, significantly enhances the ability to detect and quantify early damage in composite structures.

This paper applies transfer learning in the context of structural health monitoring (SHM) to two almost identical bridges located side-by-side, whose construction dates are separated by almost three decades. The uniqueness of this study is enhanced by the fact that the newer bridge has been reported as damaged for almost one decade, with no monitoring data available from its undamaged condition. To overcome data scarcity and uncertainty in the training of machine learning algorithms, this paper proposes a multidisciplinary framework to reuse monitoring data in the undamaged condition from the older bridge to address damage detection in the new one. A numerical model solved by the finite element method is developed to simulate the undamaged condition of the new bridge. The model is calibrated to account for sources of epistemic uncertainty using Bayesian inference through Markov-Chain Monte Carlo simulations with the Metropolis-Hastings algorithm. During the Bayesian updating process, a global sensitivity analysis using Sobol indices is proposed to identify the main parameters influencing the model's outputs. The results show that the numerical model is capable of simulating the dynamics of the new bridge in its undamaged condition, and transfer learning through domain adaptation is capable of adapting the data from the old bridge so that it can be reused to train a machine learning algorithm to classify observations from the new bridge, taking into account random uncertainty. This framework provides substantial benefits in addressing data scarcity and uncertainty, model updating, and machine learning challenges in the context of SHM but also reveals some limitations of unsupervised transfer learning.

Asymmetries in bistable energy harvesters — arising from manufacturing tolerances, assembly misalignments, or operational conditions — are traditionally viewed as detrimental to performance. This study challenges that as- sumption through a comprehensive experimental investigation, demonstrating that controlled asymmetries can, under specific conditions, enhance energy harvesting. A prototype system was tested with asymmetry introduced via magnet rotation and shaker base tilting. We explored a wide range of configurations, excitation levels, and initial conditions. Results show that rotating the magnet weakens the magnetic attraction, shifting the resonance frequency downward. While base tilting utilizes gravitational effects to modify the magnetic interaction, further reducing its strength. Although pronounced asymmetries reduce power output at higher frequencies, they can significantly improve performance at lower frequencies, depending on system dynamics and excitation. Monostable and chaotic responses were also characterized numerically and experimentally. Overall, the findings reveal that asymmetry, when properly controlled, is not a flaw but a tunable design parameter to optimize bistable energy harvester performance.

Bolted joints are widely used in various sectors due to the ease of connecting two or more parts, but the loss of torque in the bolts of this connection can cause catastrophic failures. This article presents a new piezoresistive sensor composed of a nanocomposite material with rubber, carbon black, and carbon nanotubes designed for monitoring structural health in bolted joints using a new approach for detecting tightening variation. The sensor combines rubber-like flexibility with enhanced conductivity thanks to incorporating carbon nanostructures. The properties of the rubber are thoroughly examined, and the sensor is tested in various scenarios. Its behavior and resistive response are first evaluated under cyclic loading to assess durability and functional reliability. The sensor is also analyzed for its response to environmental effects such as temperature and humidity, and their influence on measurements is evaluated. After that, the sensor is studied as an automatic and static indicator for detecting torque variation when coupled with bolted joints in a structure. This application leverages the sensor’s ability to detect variations in electrical resistance caused by changes in tightening torque, providing a method for identifying such issues. The study indicates the new sensor reliably responds to cyclic loading with resistance variation. Environmental conditions can introduce variations in influence depending on the material’s conductivity level. The sensor was applied to a bolted joint, and the variation in bolt torque can be related to the variation in electrical resistance. In addition, the percentage of nanomaterials makes the sensor more or less sensitive to the torque level, making it possible to produce a sensor for different torque ranges. For the most sensitive sensor, there was a variation of 60% in the measurements after varying by 2 Nm. At the same time, the second, with more nanomaterials, changed the resistance by another 40% for a range of 6 Nm. Furthermore, its application in identifying tightening variations has proven sensitive and effective, successfully detecting torque variations in bolts.

We propose and test a novel sensor for structural health monitoring (SHM) based on a nanocomposite film consisting of indium tin oxide (ITO) nanowires embedded in a polyvinyl butyral (PVB) matrix to detect cracks. The sensor operates by measuring the electrical resistance between terminals when it is adhered to the surface of a structure. When damage mechanisms are present, the ITO nanowires either break or rearrange, causing a change in the sensor’s internal resistance. However, certain limitations must be addressed for broader applications. Among these, the effects of loading, humidity, and temperature fluctuations in the host structure pose significant challenges. Experimental tests were conducted to evaluate the environmental influences on measurements, revealing minimal data variation, indicating that these factors are not critical concerns. Additionally, sensitivity tests were performed to assess the sensor’s response to structural changes emulating damage mechanisms, and the sensor demonstrated accurate detection. In crack propagation experiments, the sensor detected cracks during their nucleation, propagation, and ultimate failure stages. The results show this novel sensor has strong potential for crack detection and monitoring applications. Another key advantage is the simplicity of signal collection and analysis, as measurements can be obtained through static readings without excitation signals in the structure. Furthermore, the sensor’s resistance correlates well with the extent of damage.

Bistable harvesters are attractive for capturing broadband vibration energy, yet their strong nonlinearity makes them sensitive to fabrication errors. This work evaluates the robustness of an asymmetric harvester that is corrected in situ by tilting its support, allowing gravity to cancel the magnetic mismatch. All mechanical, electrical, and geometric parameters are modeled as independent uniform variables, including the tilt angle, with coefficients of variation between 5 and 40%. A polynomial-chaos surrogate, coupled with Sobol global sensitivity metrics, identifies the factors that truly govern the spread of mean electrical power, while probability maps visualize the likelihood of performance gains or losses. Across every uncertainty level, excitation amplitude and frequency dominate the variance budget; electromechanical coupling follows at a distance, and damping terms are marginal. Once the support is set to the optimum tilt, a mounting or machining error of up to 3° changes the mean power by less than 10% and never becomes a leading driver of variability. Conversely, a 30% increase in either drive parameter retains a better than 60% chance of doubling the harvested power. These findings establish a clear, quantitative robustness margin for slope-angle compensation. Designers can, therefore, relax geometric tolerances and focus on adaptive strategies that track the operating vibration environment, leading to lighter, cheaper, and more reliable bistable energy harvesters.

Transfer learning has promised to generalize structural health monitoring (SHM) of bridges, as it permits one to reuse long-term monitoring data across similar structures. Studies have been published using numerical and monitoring data from bridges sharing global similarities. This paper presents the first application of unsupervised transfer learning between twin concrete bridges. The two bridges are located sideby-side, but their construction time is separated by almost three decades. This paper proposes a framework to reuse monitoring data in the undamaged condition from the old bridge to address data scarcity and uncertainty in the training of machine learning algorithms for SHM of the new bridge. To deal with the scarcity of data, a numerical model is developed to simulate the undamaged condition of the new bridge. The model is calibrated using Bayesian inference through Markov-Chain Monte Carlo simulations with the Metropolis-Hastings algorithm. To deal with sources of uncertainty, a transfer learning is used to perform domain adaptation of data sets from both bridges. The results show the numerical model is capable of simulating the dynamics of the new bridge and transfer learning is capable of adapting the distribution domain of the data from the old bridge in such a way that it can be reused to train machine learning algorithms to classify observations from the new bridge.

Images are widely employed to analyze vibration signals and extract information for damage detection in noninvasive structural health monitoring (SHM) techniques. However, video quality must often be high in resolution to accurately extract patterns associated with damage in classification processes. In this context, this article introduces a robust approach that enables classifying a structure’s health state using noisy, low-resolution images, which are much cheaper and easier to obtain. The approach involves decomposing these videos using optDMD—dynamic mode decomposition to extract patterns and assess changes over time, proposing a metric for SHM based on video data that detects and localizes damage while providing qualitative information on its extent. To illustrate the formulation, numerical tests on an Euler–Bernoulli beam and experimental validation on a beam with damage caused by varying crack sizes are conducted. Videos of different quality and resolution are used to demonstrate the extraction of both modal characteristics and the contributions of the identified modes to image reconstruction and the detection of the beam’s structural states.

Asymmetry in bistable energy harvesters, often arising from manufacturing imperfections, assembly misalignments, or operational factors, has traditionally been viewed as detrimental to energy conversion efficiency. This study challenges that perspective through a comprehensive experimental investigation demonstrating that controlled asymmetry can enhance energy harvesting performance. A custom-designed prototype was tested under multiple configurations, introducing asymmetry through magnet rotation and base tilting. Frequency-sweep and single-frequency tests showed that magnet rotation weakens magnetic attraction, while base tilt introduces a gravitational bias, together reshaping the potential landscape, lowering resonance frequencies, and even mitigating asymmetry effects. This tunability proved particularly effective at low excitation levels, where conventional bistable systems are less efficient. Moreover, certain asymmetric configurations exhibited higher power output and broader bandwidths than the symmetric case. State-space analyses, including experimentally constructed Poincaré maps, confirmed that asymmetry significantly influences dynamic behavior, stability, and energy distribution, providing additional tuning flexibility. Although excessive asymmetry may reduce performance at high frequencies, the results demonstrate that it is not a defect to be minimized but a strategic design parameter for achieving adaptive and efficient energy harvesting in practical applications.

One significant challenge in machine learning for Structural Health Monitoring (SHM) is reusing previously trained classifiers. A classifier might be suitable for one situation but not for another. Transfer learning techniques try to overcome this difficulty. In SHM, it is common to use the modal parameters as features; however, they are highly influenced by boundary conditions, geometry, and the level of structural damage. This work proposes an innovative approach that performs a similarity analysis to select features before applying transfer learning, aiming at improving classification and damage detection. The reasoning is that a higher similarity leads to a more efficient transfer of learning and, consequently, a better classification. Transfer learning is conducted via the domain adaptation technique known as Transfer Component Analysis (TCA), and cases with low similarity are compared to those with high similarity. Two datasets are analyzed. The first consists of a beam under different boundary conditions, and data are generated through numerical simulations. The second derives from an experimental setup of bolted joints with loosening damage. The proposed strategy, which uses a cosine-type similarity, is shown to improve the transfer learning classification.

Decarbonizing the transportation sector remains a crucial challenge towards a sustainable energy framework. As economies work to substitute fossil fuels, hydrogen-powered fuel cell vehicles emerge as a viable solution. These vehicles utilize composite pressure vessels to store hydrogen under elevated pressure. Several sources of uncertainty can be captured in composite pressure vessel design. Statistical variability in material properties is, however, inevitable. In this context, this study aims to investigate the effects of material properties uncertainties on burst pressure of composite pressure vessels using the development of a probabilistic assessment methodology. Various material systems and helical winding angles were simulated. Furthermore, different failure criteria were used: Maximum stress, Tsai-Wu, and Tsai-Hill. A two-dimensional analytical modeling approach reflected the assumptions usually made in preliminary design stages, based on Classical Laminate Theory. Simulations employed Monte Carlo and Polynomial Chaos Expansion methods. Uncertainty and sensitivity analyses were performed, statistically describing output variation and ranking which input parameters contributed the most to burst pressure determination. Results showed how the incorporation of material uncertainty affects strength failure envelopes and the acting stress field. A comparison between failure criteria demonstrated that the Tsai-Wu criterion provided higher burst pressure values. The selection of material systems also played a role in output variability. Finally, burst pressure proved to be very sensible to the stacking sequence. Vessels with low helical winding angles had burst dominated by transverse tensile strength, transverse modulus, and longitudinal modulus, whereas vessels with high helical angles had burst pressure significantly dominated by transverse tensile strength.