ce/papers最新文献

The analysis and design process in structural engineering relies on the results obtained of the structural model from the black-box finite element analysis which causes implicit limit state function (i-LSF) in the structural reliability analysis (SRA). The current surrogate modeling techniques are based on evaluating the i-LSF to construct surrogates. However, even though their computational efficiencies and accuracies, the developed surrogates are mainly still implicit or yield highly complex i-LSFs. In this work, the Kolmogorov-Arnold Network (KAN) is used to discover an equivalent explicit LSF (ee-LSF) by generating a symbolic function for a given dataset. The discovered ee-LSF can be used in SRA since the expensive FEA is now able to be replaced by a simple explicit function. This paradigm allows us to unveil the implicitness of LSFs by discovering equivalent formulations through KANs which is novel to this work. Two examples are covered in this paper to present the ee-LSF approach. The ee-LSF approach demonstrates high accuracy, though its computational efficiency is currently lower compared to other surrogate modeling techniques. This limitation presents an opportunity for enhancement in future studies, particularly through integration with advanced sampling techniques.

The assessment of the load bearing capacity of existing structures is a task with growing importance, especially within the context of the aging infrastructure in Germany. This is usually done with information based on old documents from the construction phase. One of the main load cases of a concrete bridge is its own weight (dead loads) which depends on the geometry and material weights. As there are always some deviations between the real structure and the target state based on documents, a partial safety factor is applied to the loads. In contrast to the assessment of a new structure that is yet to be built these deviations are not theoretical for existing structures but can instead be measured. However, the acquisition and usage of the actual weight distribution of a bridge can be very work intensive and expensive. Therefore, it can be more practical to instead modify the safety elements for the dead loads as a function of the measured deviations based on a certain number of samples. This approach was applied on three existing concrete valley bridges resulting in lower partial safety factors and therefore more efficient assessments of the load bearing capacities.

Bridges demand detailed analysis and adherence to evolving engineering standards. Many existing bridges are experiencing challenges such as aging-related deterioration, increased traffic loads, and discrepancies with updated technical requirements. This paper aims to present the key elements of a newly developed recommendation for action about inspection-based reliability analysis of existing bridges in Germany.

The project seeks to integrate the benefits of semi-probabilistic and probabilistic assessment concepts, enabling the incorporation of measured data into the reliability analysis of existing structures through partial factor modification. The primary focus is on broader use of non-destructive testing methods for verifications in both the ultimate limit state and the serviceability limit state. The project, which started in 2022, is carried out by several research institutes and engineering companies. The project and the developed recommendation are outlined, covering the process from defining targeted inspection strategies to assessing the quality of measured data and performing structural, partial factor-based evaluations utilizing the quality-assessed on-site inspection results. The application is demonstrated through a case study involving an existing concrete frame bridge, temperature monitoring and inspection of the reinforcement.

This study investigates the potential of selected machine learning algorithms to predict the subsurface tensile strength of eco-friendly cementitious floor composites containing fly ash, ground granulated blast furnace slag, and granite processing waste. The experimental database, built on 23 mixtures with up to 30% SCM replacement, was completed and analysed statistically. Destructive testing was used to obtain reference values of subsurface tensile strength. Three machine learning algorithms k-Nearest Neighbors (kNN), Adaptive Boosting (AdaBoost), and Support Vector Machines (SVM) were trained using 5-fold cross-validation. The kNN model with Manhattan distance and distance-based weighting achieved the highest accuracy (R = 0.862, MAPE = 6.81%), outperforming the other models. The findings demonstrate that appropriately calibrated machine learning models can serve as reliable tools for non-destructive prediction of tensile strength in sustainable cement composites, reducing time and material losses in quality control.

Managing bridges during and after catastrophic events is a challenging task, requiring a balance between the safety of users and minimizing functional disruptions. Scour around bridge foundations is the leading cause of collapses in watercourse-spanning bridges. Currently, the most common method for scour monitoring is visual inspection but it is labor-intensive, inefficient, and unreliable. In response to that, Structural Health Monitoring (SHM) systems have gained interest in recent years. However, for large infrastructure networks, equipping all bridges with sensors is economically unfeasible, limiting sensor use to critical structures. This study implements a probabilistic framework for scour assessment in a bridge network using Bayesian Networks (BNs). This framework employs data from installed scour monitoring systems at key bridge locations to infer scour depths at unmonitored piers. It then enhances decision-making by integrating BN-derived scour data with analyses of both direct and indirect costs linked to various management plans. Finally, the proposed risk-based framework is applied to a case study involving a network of bridges spanning a same river.

Managing mega construction projects involves extensive processes that are often unstructured. The DigiPeC research project tackles this issue by developing a software solution.

At the IPD Innovation Hub, established with DigiPeC's support, expertise in project management is consolidated. Based on this, the software DigiCon is developed to structure and streamline complex processes digitally.

DigiCon enables interactive, structured process mapping, enhancing risk management and innovative models like IPA or Allianz. The successfully implemented Project Risk Twin (PRT) has been integrated as a case study, applied in major projects such as Hamburg's U5, Munich's 2nd S-Bahn line, and the Brenner Base Tunnel. The PRT supports structured cost, risk, and schedule analysis throughout planning, construction, and operations. Its benefits include improved risk identification, proactive delay prevention, and opportunity utilisation. Digital integration enhances its usability.

This paper highlights DigiCon's ability to optimise processes, increase efficiency, and reduce project risks, leading to improved scheduling and lower uncertainty.

Concrete, a composite material made of aggregates embedded in a cement matrix, exhibits complex mechanical behavior at the mesoscale, where the distribution and interaction of components significantly influence macroscopic properties, including failure patterns. The traditional Lattice Discrete Particle Model (LDPM) has been effective for simulating concrete at the coarse aggregate level, but it oversimplifies the heterogeneity between aggregates and the cement matrix. In the standard LDPM, tetrahedral subdomains combine aggregates and matrix as uniform material, neglecting essential heterogeneity. This oversimplification fails to accurately capture the variability and scatter observed in experiments. To address this, a two-phase LDPM is proposed, explicitly considering aggregates and the matrix as separate phases within each subdomain. Hence, the local material property is linked to the heterogenous mesostructure of the material. The random fields of the spatially variable concrete stiffness are characterized by the determination of the correlation model, considering different particle sizes, positions, and specimen dimensions. This new approach improves the realism of LDPM simulations without increasing the computational cost.

The fatigue behaviour of plain (PC) and steel fibre reinforced concrete (SFRC) is usually assessed on the basis of experimentally derived SN curves. Test data exhibit large inherent scatter in numbers of cycles to failure leading to significant uncertainties in calculated lifetime predictions. The paper deals with the stochastic evaluation of these uncertainties introducing mechanically based SN curves. The curves take into account effects of the concrete's fracture energy and of the steel fibre's post cracking tensile strength as a function of fibre type, dosage, orientation and bond behaviour by a dimensionless ductility index. Response surface methodology serves to fit the approach to experimental data taken from the literature and to assess its prediction accuracy by specifying confidence intervals. Besides, variance-based sensitivity analyses using Latin Hypercube Sampling to generate random variables are numerically executed to theoretically quantify the effect of scattering input parameters on scattering numbers of cycles to failure. Two examples demonstrate the procedure and underline the experimental observation that the concrete's tensile strength governs fatigue performance at high stress levels while fibre dosage and orientation become meaningful at low ones.

A railway bridge in Switzerland has been analyzed for wind-induced structural loads using long-term meteorological data. The Swiss standards and also the Eurocodes assume direction-independent wind pressures, leading to conservative design values. This study employs a directional wind load assessment, considering both longitudinal and transverse wind components. As a linear structure, the bridge is primarily affected by wind from a single dominant direction, offering substantial optimization potential. Using a Gumbel distribution for extreme value modeling, characteristic wind pressures were derived from a 43-year dataset. The results indicate a significant reduction in the transverse wind pressure (q_p0,char = 0.79 kN/m²) compared to the normative approach (q_p0,char = 1.3 kN/m²). Furthermore, the long-term wind data do not show an increasing trend in wind speeds over time. This refined methodology offers potential for optimizing structural design and enhancing sustainability in bridge engineering. The findings support a probabilistic approach to wind load estimation, promoting cost-effective design adaptations for existing bridges.

This paper reports a data-driven approach using Artificial Neural Network (ANN) machine learning tools to predict crack widths in reinforced concrete due to irreversible serviceability limit state (SLS) load-induced cracking. SLS target reliability levels in design standards such as Eurocode and those of South Africa were assigned using ultimate limit state values. Where SLS cracking is the dominant criterion, these levels are insufficient, needing a full probabilistic analysis. With SLS cracking the limiting criterion in the design of reinforced concrete water retaining structures and bridges, these types of structures would benefit from improvements to both crack prediction and suitable reliability levels. The semi-analytical SLS load-induced crack formulations in design standards have a model uncertainty CoV in the order of 0,35 to 0,38, significant in probabilistic analysis and reliability (where general structural uncertainty CoV is 0,1. Model uncertainty as a random variable is highly dependent on the crack formulation considered, making target reliability assessment challenging. The ANN model aims to improve crack model uncertainty. A dataset compiled from experimental research on load-induced cracking is used to train the ANN model.