SCIENTIA IRANICA
Year:2020 | Volume:27 | Issue:6 (Transactions A: Civil Engineering)|Papers Count:14

This article presents a review of selected articles about structural engineering applications of Machine Learning (ML) in the past few years. It is divided into the following areas: structural system identification, structural health monitoring, structural vibration control, structural design, and prediction applications. Deep neural network algorithms have been the subject of a large number of articles in civil and structural engineering. There are, however, other ML algorithms with great potential in civil and structural engineering that are worth exploring. Four novel supervised ML algorithms developed recently by the senior author and his associates with potential applications in civil/structural engineering are reviewed in this paper. They are the Enhanced Probabilistic Neural Network (EPNN), the Neural Dynamic Classification (NDC) algorithm, the Finite Element Machine (FEMa), and the Dynamic Ensemble Learning (DEL) algorithm.

Evaluating the reliability of trenches against sliding failure is complicated since most alluvial deposits are heterogeneous and spatially variable. This means that rather than perfectly circular or linear failure surfaces, trench failure tends to be more complex, following the weakest path or zones through the material; thereby, a new method called Continuous Slip-Surface (CSS) was adopted to calculate the critical excavation depth. CSS runs an algorithm to seek a continuous slip surface. The Finite Difference Method (FDM) coupled with random  field theory and CSS method is well suited for calculating slope stability since it allows the failure surface to seek out the weakest path through the soil. For an unsupported vertical cut, it was shown that the critical excavation depth acquired from CSS method was indeed an upper bound solution. In terms of numerical and analytical terms, an idealized variation model was used to illustrate that increasing the un-drained shear strength density attenuated the effect of shear strength variability. Correlation structure of the input variable was shown to in fluence the results, although its behavior varied on low and high scales of fluctuation.

Space headway calculation and analysis play an important role in identifying surrounding obstacles and understanding traffic scenes. However, the performance of existing methods is limited by the complexity of computer processing. In addition, it is quite difficult to obtain space headway at turn movement trajectories, mainly owing to the limitation of rectilinear propagation. Therefore, a hybrid model based on spline curve and numerical integration was proposed to estimate the distance of the front vehicle and vehicle trajectory in this study. The space headway at turn movement trajectories was analogous to the track of a vehicle, which could be  fitted by a quadratic spline curve. Newton-Cotes numerical integration was employed to calculate distance due to its meshing flexibility and ease of implementation. Data collected from Lankershim Boulevard in the city of Los Angeles, California (USA) were used to evaluate performance of the hybrid model. Compared with another algorithm based on computer vision and trilinear method, the results showed that the proposed model worked successfully and outperformed the competing method in terms of accuracy and reliability. Finally, the proposed method was applied to investigate the effects of vehicle speed, relative speed of vehicles, and time period on the spacing of vehicles during car-following.

Damaged structures are not usually reliable to tolerate designed loads and therefore, it is required to retro t structural parts. The main purpose of this paper is to utilize High-Performance Fibre-Reinforced Cement-based Composite (HPFRCC) as a high-performance material to recover the damaged beams and improve their ductility and moment capacity with experimental approaches. In addition, a retrofitting method was presented using HPFRCC. The experimental study was performed on three simply supported beams with the same dimension, materials, and reinforcement configuration. The first beam, known as the reference beam, is subjected to the pure bending condition until its failure and the others are prone to a certain amount of load according to the  final capacity of the first beam. Thereafter, two damaged beams are retrofitted using HPFRCC in the created grooves on the tensile surface of the beam; finally, these retrofitted beams are loaded to determine the bending behavior. Experimental results demonstrate that retrofitting can improve the first crack strength, load in yield condition, and maximum load capacity. Also, the proposed method increases the ductility and energy absorption of retrofitted beams.

Freight transport policy analysts attempt to shift truck freight movements to rail so as to diminish transportation externalities including environmental costs and safety issues. Therefore, policy-makers need to be aware of the consequences of their decisions beforehand. This study is mainly focused on two policies targeting fuel price and access to rail transportation. A nation-wide freight mode choice model was developed for Iran, and shippers' tendency to choose rail or truck freight transportation was analyzed by considering the shipping time and cost, commodity weight, commodity type, and rail accessibility. Total fuel consumption and air pollution costs were compared in various scenarios. Based on the results, environmental transportation costs are significantly reduced as a result of the modal shift from truck to rail freight transportation if the government reallocates gasoline subsidy to the construction of prioritized railroads.

Lateral spreading is one of the most significant destructive and catastrophic phenomena associated with liquefaction caused by earthquake and it can cause very serious damage to structures and engineering facilities. The aim of this study is to evaluate liquefaction-induced lateral spreading and  nd new relations using Gene Expression Programming (GEP), which is a new and developed generation of genetic algorithms approaches. Since there are complicated, nonlinear, and higher-order relationships among many factors affecting the lateral spreading, GEP was assumed to be capable of finding complex and accurate relationships among the involved factors. This study includes three main stages: (i) compiling available database (484 data); (ii) dividing data into training and testing categories; and (iii) building new models and proposing new relationships to predict ground displacement in free face, gentle slope, and general ground conditions. The results of modeling each of these different ground conditions were presented in the form of mathematical equations. At the end, the final GEP models for 3 different cases of ground conditions were compared with Multiple Linear Regression (MLR) and other published models. The statistical parameters indicated the higher accuracy of GEP models over other relations.

This research aims to identify the basic properties of flood mud and the efficiency of biomass silica (SH85) as a stabilizer to improve the strength of this mud. Unconfined Compressive Strength (UCS) testing was carried out on untreated soil and soil treated with 2%, 4%, and 9% SH85 contents within three and seven curing days. The microstructure of SH85 treated flood mud was investigated via Field-Emission Scanning Electron Microscopy (FESEM) and Energy-Dispersive X-ray (EDX) spectrometry. It was found that the strength of treated soil increased two to seven times that of the untreated soil where the highest strength was recorded at 949 kPa following seven-day soil treatment by 9% SH85 content. A polynomial trend was observed with an R2 value greater than 95% relationship between SH85 content and UCS in different curing periods. The seven-day UCS of SH85 treated flood mud met the strength requirement of 0. 8 MPa for Malaysian subgrade material of low traffic volume roads and the compressibility was significantly reduced when SH85 content was greater than 4%. According to the FESEM and EDX results, cementitious products that promoted soil strength filled the voids of the treated soil.

Over the last few decades, a considerable number of theoretical and experimental investigations have been conducted on the mechanical strength of composite bonded joints. Nevertheless, many issues regarding the debonding behavior of such joints still remain uncertain. The high near free-edge stress fields in most of these joints are the cause of their debonding failure. In this study, the performance of an externally bonded Fiber-Reinforced Polymer (FRP) composite to a concrete substrate prism joint subjected to mechanical and thermomechanical loadings was evaluated using the principles of lamination theory. An inclusive Matlab code was generated to perform computations. The bond strength was estimated to take place in a region-also termed as the boundary layer-where the peak interfacial shearing and transverse peeling stresses occurred, whereas the preceding stress field was observed to be the main failure mode of the joint. The proposed features were validated by using the existing experimental data points as well as the commercial Finite Element (FE) modeling software, Abaqus. A comparison between the calculated and experimental results demonstrated favorable accord, producing quite a high average ratio. The current approach facilitates failure modeling analysis, optimal design of bonded joints, and scaling analysis among others.

The aim of this paper is to analyze damage to dams by using two recent methodologies. The first method was used to de ne the performance and response curves of concrete gravity dams. The second method defined the seismic input that was obtained from power spectral density function consistent with the response spectrum. Both methods are efficient, practical, and useful to develop quite complicated analysis as the construction of the stochastic process to de ne the synthetic earthquake and the estimation of cracks in the dam's body. These methodologies were explained and modified to improve their use. The fluid behavior of arch dams was compared with that of storage tanks by studying the sloshing phenomenon, which is usually neglected for dams. For the mathematical modeling, interactive programming language was used.

Sometimes, natural frequency of beams is not within the allowable range despite the appropriate shear and bending design. Every so often, structural designers face cases in which it is not possible to increase the beam height. Steel cable is utilized in this research to control the natural frequency of beam due to its important advantages such as low weight, small cross-sectional area, and high tensile strength. For the first time, theoretical relations were developed to calculate the rate of increase in pre-tensioning force of steel cables under external loading based on the method of least work. Moreover, the natural frequency of steel beams with different support conditions without cable and with di erent patterns of cable was calculated based on Rayleigh's method. To verify the theoretical relations involved, the steel beam was modeled using ABAQUS software. The obtained results showed that the theoretical relations could appropriately predict the natural frequency of beams with different support conditions and cable patterns. In addition, simply supported and fixed supported beams were prestressed with V-shaped and modified V-shaped patterns of the cable. According to the obtained results, the modified V-shaped pattern of the cable was more efficient than the V-shaped pattern.

This paper investigates the behavior of low-, medium-, and high-rise protected steel moment-resisting frames under post-earthquake  re through two different methods. In the first method, the pushover analysis was utilized to simulate the response of the sample structures for various target displacements. Then, the thermal-mechanical analysis was implemented to evaluate the behavior of the damaged frames under fire, assuming that the fi reproofing was delaminated at the end regions of the beams. In the second method, the seismic response of the frames under two sets of the MCE-scaled near-and far-fault ground motion records was determined employing the time history analysis. In this method, the damage of  fireproo fing was characterized by the maximum inter-story drift ratio. The results of the study revealed that the first method produced similar results to that of the second method for most cases. It was also found that for sufficiently large drift demands, the collapse of the frames under post-earthquake  re occurred in the side-way mode, while for lower seismic responses, the local failure of beams dominated other failure modes. Moreover, it was found that the reduction of fire resistance time due to the effects of Maximum Considered Earthquake (MCE) seismic loads ranged from 4% to 27% for the considered structures.

This paper describes the behavior of walls under in-plane cyclic shear compression of a new reinforced masonry system composed of horizontal and vertical reinforcements based on Iran's national building regulation codes in two groups. In the first group, grid-type steel bars were mounted on the cement core between solid clay bricks (double-wythe); in the second group, common grid-type steel bars were mounted on perforated bricks and trusses as horizontal reinforcements using advanced numerical simulation (LS-DYNA). A nonlinear fi nite element discrete modeling according to stress-strain models was applied to represent the previously modeled masonry walls. Masonry units included perforated bricks and solid clay bricks, and the mortar and bonding interfaces were shown as continuum elements. In order to validate the micro-modeling strategy, the input data were based on a reinforced masonry wall previously tested in the laboratory with clear identification and justification. Accordingly, the main objective of this paper is to (a) examine results of specimens in terms of maximum strength, ductility, energy absorption, and failure modes, (b) investigate the effect of aspect ratio and reinforcement type, and (c) compare the modeled walls with other reinforced systems.

Soil in cold regions experiences repetitive freeze-thaw cycles that are considered as one of the most significant phenomena in cold region engineering. Approximately 30% of soil all around the world and a large portion of fertile lands are exposed to daily or seasonal freeze-thaw cycles. These cycles cause considerable changes in water content, solute movement, permeability, strength parameters, erosion rate, and other physical or chemical characteristics of soil. Nowadays, one of the approaches to improving the physical and mechanical characteristics of the soil is to incorporate geosynthetic material as a layer between the embankment and the ground surface. This paper presents the results of California bearing ratio that tests clayey sandy soil. Moreover, the effect of freezethaw cycles on the compressive strength of geotextile-reinforced soil was investigated. The geotextile layer was placed in five positions at different depths of 1. 3, 2. 6, 3. 9, 5. 85, and 7. 8 cm beneath the surface of the mold and then, the sample was exposed to freeze-thaw cycles. It was found that the optimum depth of the geotextile layer was 3. 9 cm. In addition, it could be observed that reinforcing the soil could decrease the weakening effects of freeze thaw cycles by up to 41. 7%.

Improving management of complex and congested facilities in hospital buildings is a potential point for both reducing money spent and enhancing quality level of the medical services provided in public hospitals of Iran. Although Building Information Modeling (BIM) is identified as an effective tool for improving Facility Management (FM), use of advantages it offers to the FM processes of hospitals has been neglected thus far in the country. To address this issue, this research aims to investigate the BIM capabilities and the supporting organizational structure of public hospitals in Iran which can help improve their FM processes. A comprehensive literature review of applicable capabilities of BIM to the FM processes was conducted. Hierarchical FM structure of public hospitals in the country was recognized through a review of the related regulations. A public hospital case was chosen for in-depth recognition of FM process operations and validation of the proposed BIM-based improvements. It was argued that the use of BIM capabilities could substantially improve the FM processes of public hospitals. Reduced duration of FM activities, improved facility layouts, enhanced communication and coordination, facilitated training, and improved emergency management are some of the expected outcomes.