Over the centuries, structural design has evolved to meet the more stringent fundamental and architectural challenges and demands. With the advent of modern computational facilities, the horizon of analysis and design has been pushed continuously. However, even after these advancements, structures, once adequately designed, suffer deterioration when exposed to their intended operational environment. These changes are influenced by many factors, e.g., weathering, changes in working/boundary conditions, climate, and loading patterns/intensity, among many others. All these factors affect the material and geometric properties of a structure. Consequently, these changes affect the performance in terms of structural capacity and/or functionality. For example, a bridge operating over a significant time may experience weathering, the difference in axle loading, and vehicle count per day. These are reflected in the underperformance of the structure, i.e., unwanted displacement/vibration, cracks, and damages. In addition to the factors mentioned above, natural calamities like earthquakes, cyclones, floods, and landslides also damage structures. Thus, the sustainability of any structure needs periodic health monitoring and condition assessment.
In this process, the structural response is systematically analyzed to ascertain its in-situ conditions. Depending on the type of inputs used in this inverse analysis, structural health monitoring can be classified into different groups viz., acoustic emission-based approach, vibration-based approach, thermal signature-based approach. Among these options, vibration-based structural health monitoring has gained popularity among researchers and engineers. In this approach, the response of an existing structure due to ambient or particular loading is analyzed to verify its capacity and other critical parameters. Depending upon the type/nature of signal processing involved in this process, vibration-based structural health monitoring has three different paradigms, i.e., time domain, frequency domain, and time-frequency domain. Each of these techniques has its advantages and disadvantages. Given these options, the main activities of this group in structural health monitoring are focused on the following issues –
- Time-frequency based signal processing has proved its potential in different applications in science and engineering. Short-Time Fourier Transform, Hilber-Huang Transform, Wavelet Transform are used in structural health monitoring. Among them, wavelet transform is a powerful tool for frequency tracking and multi-resolution analysis. It opens up many avenues for further development and improvisation of this tool (e.g., multi-channel or synchrosquzeeing) for system identification and damage detection. This group aims to develop different time-frequency algorithms for system identification of civil infrastructures.
- Gaussian filtering is an effective strategy for inverse problems. Different versions of these filters exist along with different modifications. The Kalman filter is the most popular for tracking and identification that has witnessed a wide range of applications in different fields. This group aims to develop these techniques further for non-linear system identification at the phenomenological level (i.e., global load-deformation characteristics through models like Bouc-Wen) or material level (i.e., local stress-strain behaviour at the section level). Besides system identification, these filtering algorithms can also estimate inputs or loads applied to excite the parent system, which is a significant advantage. The efficiencies of these algorithms are demonstrated for simultaneous state and parameter estimation of hysteretic systems (i.e., damage quantification).
- Bayesian inference has played a vital role in the different decision-making processes in science and engineering. It is extensively used in system identification, FE model updating, and damage identification. However, this predictor-corrector algorithm needs to simulate samples from the conditional probability density functions. For this purpose, the Metropolis-Hastings algorithm or its advanced versions are primarily used. This random walk-based candidate generation needs a large number of samples due to its poor acceptance rate. In this context, hybrid simulation using the analogy of Hamiltonian dynamic (aka Hamiltonian Monte-Carlo simulation) offers an efficient alternative. The present activities of this group are to adopt/modify the leap-frog algorithm or use a closed-form sample generation scheme for candidate generation. Also, the group investigates the efficiency of these algorithms for the FE model updating of existing buildings/bridges and their accuracy for the solution of the inverse problem.
Besides health monitoring, vibration control plays a significant role in the design and operation of modern structures. It not only helps to reduce unwanted vibration for better serviceability but also improves long-term performance. For example, a vibration isolator reduces structural response, hence the stress level and its reversal, which mainly affects the fatigue life of any structure. Vibration control can be carried out in three different ways – (i) changing/adding stiffness to the original structure (e.g., bracing system), (ii) separating vibration at its source (e.g., base isolators), and (iii) by different active/semi-active/passive devices (e.g., tuned mass damper, magneto-rheological dampers). These controllers are tuned depending upon the architecture and physics behind the operational principle. For example, a passive tuned mass damper primarily uses resonance to absorb energy from the primary structure. The tuning strategy or the control law behind any device is tricky to extract maximum performance. It is difficult in the presence of uncertainty. With these in view, the activities of this group are mainly focused on –
- Tuning of passive controllers affects the performance of the overall system. The deterministic design uses different criteria to cast the objective function for the optimal solution. However, structures are often subjected to random input and parametric uncertainties. Optimal tuning in the presence of these uncertainties is the focus of the activities of this group. It includes robustness of the controller and reliability-based design optimization for maximum output.
- Base isolation has remained a popular tool for vibration control in civil infrastructure. The design of a base isolator primarily relies on recorded ground motions or their spectrum compatible versions for a particular site. However, a recorded ground motion is an outcome of a random process. Moreover, a specific site experiences ground motion due to a given scenario. Therefore, the base isolator design must accommodate process and scenario-specific ground motions for optimal tuning of base isolator parameters. This group is trying to explore different base isolator design strategies, keeping different seismic cases (process or scenario-specific) at the center.
- Smart controllers are another area that has received a lot of attention in the recent past due to their flexibility to adopt different operation conditions. In this context, the smartness of any controller is introduced by its architecture and/or by the use of advanced material. For example, hybrid base isolation made of a traditional isolator and an inerter can offer optimal results under multi-objective performance functions in different conditions. In this context, recent advancement in material technology helps to develop magneto-rheological elastomers. This smart isolator can be tuned in real-time to generate different hysteresis, thereby changing the performance level of the isolator as per requirement. This group aims to investigate the performance of these smart controllers under different operational difficulties and their optimal tuning algorithms.
