Special Issue on Advanced Technologies in Seismic Design, Assessment and Retrofitting

1. Introduction

The seismic design of new structures, as well as the assessment of seismic vulnerability and the retrofitting or rehabilitation of existing structures, are two important research fields in civil engineering. The main objective of this research is to increase safety against seismic risk and, at the same time, minimize the required costs. The need for more effective seismic protection and strengthening over time is made more apparent by the representative diagram of economic losses due to earthquakes worldwide over the last 100 years (Figure 1).

In general, both the design of new earthquake-resistant structures and the assessment and the rehabilitation or the retrofitting of existing structures are processes in which the objective is the optimal solution that leads to a required level of safety (defined by the seismic codes) without being too economically costly. The problem could therefore be described, in general terms, as an optimization problem with specific constraints imposed by safety and economic cost requirements. In order to achieve this optimization, methods and procedures are required that are part of many scientific fields, such as modelling and analysis of structures, seismic risk assessment, structural health monitoring, experimental studies and materials engineering.

The need to establish guidelines that lead to the reliable assessment and retrofitting or rehabilitation of existing structures has driven the development and continuous updating of seismic codes worldwide (e.g., [1,2,3,4,5]). In addition, there is an increasing amount of research (see e.g., [6,7,8,9,10,11,12,13,14]) that aims to develop new and more effective methods, as illustrated in the diagram in Figure 2.

To design the diagram in Figure 2, advanced searches were carried out in the Scopus database using the following criteria combinations.

• Criteria combination 1: Search documents concerning research on conventional methods of assessment and rehabilitation or retrofitting of existing structures: (“Seismic Assessment” OR “Seismic Retrofit” OR “Seismic Rehabilitation” OR “Seismic Evaluation”) AND (LIMIT-TO (SUBJAREA, “ENGI”)).
• Criteria combination 2: Search for documents concerning research on methods for the assessment and rehabilitation or retrofitting of existing structures containing machine learning procedures: (“Seismic Assessment” OR “Seismic Retrofit” OR “Seismic Rehabilitation” OR “Seismic Evaluation”) AND (“Artificial Intelligence” OR “AI” OR “Machine learning”) AND (LIMIT-TO (SUBJAREA, “ENGI”)).

It is evident from Figure 2 that machine learning methods have provided new potential in the seismic assessment and rehabilitation or retrofitting of existing structures, as methods are able to handle multiparametric problems using existing experience. In addition, it should be stressed that existing structures also include historic buildings (mainly of masonry) of great cultural value. For these structures, there is also an extensive number of journal or conference papers and books (see e.g., [7,8,14]) presenting methods of assessment and retrofitting or rehabilitation. In these structures, there is also the additional requirement of not altering their historical character.

Thus, following the current trend for improving methods of design and the assessment and retrofitting or rehabilitation of structures, the present Special Issue invited researchers to present their work in the scientific fields that contribute to the upgrading of relevant technology. Eight papers were submitted and accepted. These papers cover important areas of the scientific fields that contribute to the reliable assessment of the seismic vulnerability of structures and to effective retrofitting using new technologies. More specifically, the eight papers generally deal with new technologies in the following scientific fields:

• Advanced modelling and analysis methods for the seismic vulnerability assessment of structures [15,16].
• Advanced techniques to improve seismic response and seismic damage control using dampers [17,18].
• Advanced technologies that increase the effectiveness of seismic retrofitting or rehabilitation using new special materials and new construction techniques [19,20].
• Machine learning algorithms for the prediction of seismic response of structures [21,22].

However, it should be noted that, apart from the above-mentioned scientific fields, the published papers also touch on scientific fields such as structural monitoring (ambient vibration measurements) [15,21], the calibration of structural models using experimental results [19], as well as the seismic assessment of heritage masonry buildings [15].

2. A Brief Overview of the Published Papers

2.1. Advanced Modelling and Analysis Methods for the Seismic Vulnerability Assessment of Structures

The accurate and dependable modelling and analysis methods of existing structures are the cornerstone for the proper assessment of their seismic vulnerability and the appropriate selection of retrofitting or rehabilitation techniques. The use of advanced analysis methods assisted by modern available software enables a more accurate assessment of their actual seismic response.

Corrado Chisari et al. [15] assessed the seismic vulnerability of the bell tower of St. Lucia Church in Cellole (Italy). Following the required procedure for the assessment of an existing structure, they first collected the necessary data (structural geometry, material properties and dynamic characteristics). To capture the geometry, a visual inspection and application of a complete 3D laser scan survey was carried out. For the estimation of the material properties of the structure, a series of non-destructive or minimally destructive methods (e.g., sclerometer tests, penetrometers) were utilised. For the estimation of the dynamic characteristics of the structure, the appropriate instrumentation of the tower and the recording of the ambient vibrations, fast Fourier transform and frequency domain decomposition procedures were used and their evaluation by modal assurance criterion tables were carried out. Thus, the first eigenfrequencies and eigenmodes were identified. This was then followed by an assessment of the seismic behaviour with two levels of evaluation [EL1 (simplified mechanical) and EL2 (kinematic limit analysis)]. Both assessment procedures led to the conclusion that the retrofitting of the structure was required. The results of the assessment are also consistent with the picture of extensive cracking identified during the visual inspection. As a future extension of the research, the authors considered adding existing damage to the structural model, as well as determining the required procedures for rehabilitation.

Boyke and Nagao [16] investigated the advanced modelling of pile-supported wharfs using finite element models utilising both inertial and kinematic forces. Considering that a reliable assessment of an existing structure requires feasible modelling, the authors compared frame analysis, which is an analysis method with low computational resource requirements but limited applicability to pile-supported wharfs, with finite element analysis. The aim of the investigation was to develop an extended 2D frame analysis method that is capable of considering a combination of inertial and kinematic forces induced during an earthquake. Models with or without ground slopes were used. The results of the analyses and comparisons of the two methods also showed that the addition of the kinematic force to the analysis is an important factor leading to the extraction of realistic stress magnitudes (bending moments). Finally, through their research, the authors proposed an equation to estimate the damping factor for pile-supported wharfs based on its width and natural periods.

2.2. Advanced Techniques to Improve Seismic Response and Seismic Damage Control Using Dampers

The effect of damping on the seismic response of structures is proven to be favourable in terms of damage reduction (displacement and stress control). For this reason, in addition to the inherent damping of a structure, it is often desired to artificially increase it by means of special devices, the dampers. Dampers are generally used in special engineering structures, such as bridges. Thus, damper technology is an advanced technique for improving the seismic response of structures.

Zhao et al. [17] studied the seismic performance of bridges into which strengthened accelerated oscillator dampers had been built, enabling the control of the vibrations. Starting with the accelerated oscillator damper (AOD), which has been proven to be highly effective, they focused on the need for two upgrades. They proposed the innovative strengthened AOD with a linear spring (SAOD-LS), which can be installed and operated in traditional bridges. The proposed damper has a design that allows the installation of a secondary system and improved technical characteristics. In addition, the authors proposed the SAOD-NSD, an AOD additionally enhanced with a non-linear spring device (NSD) that provides protection against resonance during vibrations either due to earthquakes or wind. With these two upgrade proposals (SAOD-LS and SAOD-NSD), the drawbacks of the original AOD configuration are overcome. Finally, the proposed innovative upgrades/modifications of the AOD were also verified through finite element analyses.

D’Aniello et al. [18] studied a solution to the problem of reducing seismic damage in steel structures through an innovative design based on the introduction of additional seismic energy dissipation mechanisms by friction dampers. More specifically, they dealt with the addition of friction devices (“Free for Damage” devices or FREEDAM devices) to steel beam-to-column joints. FREEDAM devices are designed to ensure the elastic behaviour of joints even in strong seismic events, thus ensuring the structural integrity of structures. Within the framework of the study, the calibration of advanced models of joints composed of solid finite elements using results obtained from experimental procedures was carried out. The analyses performed resulted in a number of useful conclusions for the correct configuration of joints with FREEDAM dampers and for the description of their seismic behaviour through the formulation of appropriate moment–rotation diagrams (discrimination of four phases during their seismic responses).

2.3. Advanced Technologies That Increase the Effectiveness of Seismic Retrofitting or Rehabilitation Using New Special Materials and New Construction Techniques

Research on the use of appropriate materials and structural configurations used for the retrofitting or the rehabilitation of existing structures is an important subject in earthquake engineering. The use of appropriate materials and devices can contribute substantially to increasing safety against strong seismic excitations while reducing the required construction cost.

Sun et al. [19] studied the problem of the seismic performance of underwater concrete piers strengthened using the precast concrete segment assembly method (PCSAM). Considering the fact that the PCSAM has some disadvantages, such as the considerable strength loss of filled concrete, poor accuracy and the poor connection performance of the segment sleeves, they developed and proposed the application of an innovative and improved strengthening method, IPCSAM. This method is efficient, economical and does not disturb the shipping traffic. To test and verify the proposed method, they investigated the seismic behaviour of reinforced piers through experiments. For this purpose, nine specimens were initially designed at a 1/5 scale with different characteristics and were tested under low reversed cyclic loading. In addition, an extensive parametric analysis was carried out using numerical models adapted to the experimental data. By combining the results from the experiments and numerical analyses, a dynamic behaviour model for reinforced piers was developed using the proposed method. The research process proved that (a) IPCSAM significantly increased the bearing capacity, ductility and initial stiffness, as well as the energy dissipation capacity, and (b) the numerical dynamic behaviour model developed is highly accurate and can be used as the basis for the practical application of designing rehabilitation schemes for piers.

Herrera et al. [20] focused their research on improving the modelling of connectivity and constraint conditions between existing masonry-infilled reinforced concrete moment frames and mortar walls reinforced with steel wire mesh for seismic rehabilitation. This rehabilitation method created shear walls that provided a building with the required specified stiffness and helped to protect against undesirable collapse mechanisms under seismic loading. A condition for the formation of shear walls when using the proposed retrofitting method is the sufficient connection between the existing elements (moment frame) and the new ones (mortar walls). Therefore, a key feature of this retrofitting method is the construction of structural elements with cross-sections consisting of different materials. For the modelling of elements of this type, fibre models were used, from which the capacity curves for the retrofitted frames were derived. These curves are necessary in the context of non-linear methods of seismic assessment. The derived curves were calibrated by means of corresponding curves obtained using the Bernoulli–Euler beam theory, moment–curvature analyses and the well-known plastic hinge model. The main conclusion that emerged from the research procedure is that reliable capacity curves for frames retrofitted with mortar walls reinforced with steel wire mesh can be obtained using fibre models and by including intermediate connectivity nodes between the top and bottom of the frame, where rigid link constraints connect the existing frame with the new wall.

2.4. Machine Learning Algorithms for the Prediction of the Seismic Response of Structures

Machine learning algorithms are increasingly used to investigate methods to assess the seismic vulnerability of existing structures (Figure 2). The inherent ability of these algorithms to deal with multiparametric problems and to be continuously “trained” with new data resulting from real seismic excitations makes them powerful computational tools.

Damikoukas and Lagaros [21] developed the MLPER (machine learning-based prediction of structures’ earthquake responses) model. The MLPER model was applied to predict the acceleration time history of the top floor translational degree of freedom of an n-story building without performing finite element analysis. The architecture of MLPER is based on the operation of a trained convolutional neural network (CNN) and is realized in three basic stages (encoding, latent space and decoding). The operation of CNNs requires the input of data in image form. In the case of MLPER, the input “parameters” are 2D images of the time history of the ambient responses of buildings in combination with the target seismic event. It is important to note that the input time histories are transformed from the time domain to the frequency domain in order to address a number of problems, such as the problems arising from the exceptionally large difference in the values of ambient response, those of earthquake excitation and those of the responses under earthquake. Thus, the input format of the CNNs is, in this case, two pairs of images, i.e., amplitude and phase spectrograms, corresponding to ambient responses and earthquake data, respectively. Similarly, the output is a pair of images consisting of amplitude and phase spectrograms. It should be noted, however, that the output images concern earthquake-induced responses. The time domain evaluation of the model (performed with various metrics, such as the mean absolute percentage error (MAPE) and the mean deviation angle (MDA)), showed very satisfactory results. The development of MLPER sets the basis for addressing the problem of the accurate and rapid prediction of the seismic response of buildings of various types considering different significant factors (i.e., the site effects, the second-order effects and the behaviour that differs from the bilinear stiffness).

Morfidis et al. [22] developed RASDA (rapid seismic damage assessment) software to support the rapid visual inspection (RVI) of existing reinforced concrete buildings. In this software, they incorporated trained artificial neural networks (ANNs) and, more specifically, multilayered feedforward perceptron neural networks (MFPNN). The seismic vulnerability assessment of existing buildings is generally divided into three levels, with increasing complexities in each of them. The first level is based on the RVI procedure carried out using specific templates without the need for complex calculations. The procedure followed is therefore simplified in order to allow the inspection of a large number of existing buildings in the shortest possible time. The proposed software has a built-in, user-friendly graphic user interface (GUI), with which it is possible to quickly input and process the required data collected during a RVI either pre-seismically or post-seismically. Moreover, with the built-in trained MFPNN, it has the capability—based on the data entered by the user and by solving a pattern recognition problem—to extract a prediction for the level of seismic damage of the studied buildings by classifying them in one of three damage categories/classes (null–slight, moderate and severe–collapse), as is typically carried out in the context of RVI after strong earthquakes. The evaluation of the proposed software was documented using an extended, numerically generated dataset of reinforced concrete buildings subjected to previous strong earthquakes. For this evaluation, confusion matrices (which are the main tools utilised to check the reliability of pattern recognition problems) were used.

3. Conclusions

The present Special Issue comprises a series of research papers covering a wide range of scientific fields that contribute to the proper assessment of the seismic vulnerability of existing structures and their effective retrofitting or rehabilitation. In these papers, research is presented on the application of advanced methods that are feasible today due to the availability of modern computational tools (e.g., software for the implementation of detailed modelling and non-linear analysis methods, as well as software for the implementation of machine learning methods) but also modern special equipment (advanced dampers, devices for structural monitoring and experimental procedures). A common conclusion in all the published papers is that there is the potential for a significant reduction in the expected losses due to strong seismic excitations through a more precise estimation of the expected seismic response of existing structures, the vast majority of which were designed either with the application of old seismic codes that do not reflect modern scientific knowledge or without the application of any seismic code.