Árpád Rózsás
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Árpád Rózsás has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Arthur Slobbe
Netherlands Organization for Applied Scientific Research (TNO)
5 shared publications Arthur Slobbe
3 shared publications- Agnieszka Bigaj‐Van Vliet
Netherlands Organization for Applied Scientific Research (TNO)
1 shared publication - Alice Cicirello
Department of Engineering, University of Cambridge
1 shared publication - Ana Teixeira
Deltares
1 shared publication - Andrés Martínez
Reliable Structures
1 shared publication
Latest Publications
- 6 April 2025
- Published by Elsevier in Engineering Structures
- Vol. 334
Towards practical Bayesian system identification of engineering structures with spatially dense measurements
- 6 April 2025
- Published by Elsevier in Engineering Structures
- Vol. 334
Towards practical Bayesian system identification of engineering structures with spatially dense measurements
- Open Access
Bayesian system identification for structures considering spatial and temporal correlation
The decreasing cost and improved sensor and monitoring system technology (e.g. fiber optics and strain gauges) have led to more measurements in close proximity to each other. When using such spatially dense measurement data in Bayesian system identification strategies, the correlation in the model prediction error can become significant. The widely adopted assumption of uncorrelated Gaussian error may lead to inaccurate parameter estimation and overconfident predictions, which may lead to sub-optimal decisions. This paper addresses the challenges of performing Bayesian system identification for structures when large datasets are used, considering both spatial and temporal dependencies in the model uncertainty. We present an approach to efficiently evaluate the log-likelihood function, and we utilize nested sampling to compute the evidence for Bayesian model selection. The approach is first demonstrated on a synthetic case and then applied to a (measured) real-world steel bridge. The results show that the assumption of dependence in the model prediction uncertainties is decisively supported by the data. The proposed developments enable the use of large datasets and accounting for the dependency when performing Bayesian system identification, even when a relatively large number of uncertain parameters is inferred.
Open AccessBayesian system identification for structures considering spatial and temporal correlation
The decreasing cost and improved sensor and monitoring system technology (e.g. fiber optics and strain gauges) have led to more measurements in close proximity to each other. When using such spatially dense measurement data in Bayesian system identification strategies, the correlation in the model prediction error can become significant. The widely adopted assumption of uncorrelated Gaussian error may lead to inaccurate parameter estimation and overconfident predictions, which may lead to sub-optimal decisions. This paper addresses the challenges of performing Bayesian system identification for structures when large datasets are used, considering both spatial and temporal dependencies in the model uncertainty. We present an approach to efficiently evaluate the log-likelihood function, and we utilize nested sampling to compute the evidence for Bayesian model selection. The approach is first demonstrated on a synthetic case and then applied to a (measured) real-world steel bridge. The results show that the assumption of dependence in the model prediction uncertainties is decisively supported by the data. The proposed developments enable the use of large datasets and accounting for the dependency when performing Bayesian system identification, even when a relatively large number of uncertain parameters is inferred.
- 14 December 2022
- Published by Wiley in Structural Concrete
- Vol. 24 (2) , 2979-3001
Reliability‐based calibration of design code formulas: Application to shear resistance formulas for reinforced concrete members without shear reinforcement
This paper presents a reliability‐based calibration method for design code formulas. The method is demonstrated on the shear design formulas in Eurocode 2 and fib Model Code 2010 (MC2010). We found that the partial factor in the current Eurocode 2 is about 20% lower than the optimal value and, thus, provides an insufficient safety margin. The obtained optimal partial factor in the (modified) Eurocode 2 and MC2010 formulas is 1.53 and 1.36, respectively. The difference stems from higher accuracy and, hence, lower uncertainty of the MC2010 model in predicting experimental results. Hence, on average, the MC2010 formula leads to about 13% larger design resistances compared to Eurocode 2 given that the target reliability for both design formulas is the same. To stimulate and facilitate future structural code development and derivation of partial factors, we make the used computer code freely available.
- 14 December 2022
- Published by Wiley in Structural Concrete
- Vol. 24 (2) , 2979-3001
Reliability‐based calibration of design code formulas: Application to shear resistance formulas for reinforced concrete members without shear reinforcement
This paper presents a reliability‐based calibration method for design code formulas. The method is demonstrated on the shear design formulas in Eurocode 2 and fib Model Code 2010 (MC2010). We found that the partial factor in the current Eurocode 2 is about 20% lower than the optimal value and, thus, provides an insufficient safety margin. The obtained optimal partial factor in the (modified) Eurocode 2 and MC2010 formulas is 1.53 and 1.36, respectively. The difference stems from higher accuracy and, hence, lower uncertainty of the MC2010 model in predicting experimental results. Hence, on average, the MC2010 formula leads to about 13% larger design resistances compared to Eurocode 2 given that the target reliability for both design formulas is the same. To stimulate and facilitate future structural code development and derivation of partial factors, we make the used computer code freely available.
- Open Access
- 13 October 2022
- Published by Taylor & Francis in International Journal of Architectural Heritage
- Vol. 17 (1) , 147-169
Siamese Convolutional Neural Networks to Quantify Crack Pattern Similarity in Masonry Facades
This paper proposes an automated approach to predict crack pattern similarities that correlate well with assessment by structural engineers. We use Siamese convolutional neural networks (SCNN) that take two crack pattern images as inputs and output scalar similarity measures. We focus on 2D masonry facades with and without openings. The image pairs are generated using a statistics-based approach and labelled by 28 structural engineering experts. When the data is randomly split into fit and test data, the SCNNs can achieve good performance on the test data ( ). When the SCNNs are tested on ”unseen” archetypes, their test values are on average 1% lower than the case where all archetypes are ”seen” during the training. These very good results indicate that SCNNs can generalise to unseen cases without compromising their performance. Although the analyses are restricted to the considered synthetic images, the results are promising and the approach is general.- 13 October 2022
- Published by Taylor & Francis in International Journal of Architectural Heritage
- Vol. 17 (1) , 147-169
Open AccessSiamese Convolutional Neural Networks to Quantify Crack Pattern Similarity in Masonry Facades
This paper proposes an automated approach to predict crack pattern similarities that correlate well with assessment by structural engineers. We use Siamese convolutional neural networks (SCNN) that take two crack pattern images as inputs and output scalar similarity measures. We focus on 2D masonry facades with and without openings. The image pairs are generated using a statistics-based approach and labelled by 28 structural engineering experts. When the data is randomly split into fit and test data, the SCNNs can achieve good performance on the test data ( ). When the SCNNs are tested on ”unseen” archetypes, their test values are on average 1% lower than the case where all archetypes are ”seen” during the training. These very good results indicate that SCNNs can generalise to unseen cases without compromising their performance. Although the analyses are restricted to the considered synthetic images, the results are promising and the approach is general. - 12 October 2022
- Published by Taylor & Francis in Structure and Infrastructure Engineering
- Vol. 20 (5) , 741-759
Efficient Bayesian model selection and calibration using field data for a reinforced concrete slab bridge
In this work, a reinforced concrete slab bridge (instrumented and tested in 2018) is investigated. Based on field data, a finite element model of the bridge is calibrated. Model selection is performed both based on log evidence and posterior predictive capabilities. It is investigated if the models selected based on the log evidence also induce the most accurate posterior predictions. The influence of different assumptions on modelling the spatial distribution of the stiffness and different possible suggestions on how to include prediction errors and model bias are investigated. Comparing the conclusions based on log evidence and posterior predictions, only using the log evidence for model selection could be debated. Models performing best when considering the log evidence led to the least accurate posterior predictions, and models rejected based on the log evidence could still have good predictive capabilities. Considering the different model classes, introducing spatial variation of the stiffness leads to a posterior prediction closer to the measurements. Introducing a global model bias leads to a better match between predictions and measurements compared to not including this model bias. Even better posterior predictions are achieved if this model bias is quantified locally for the different considered datapoints.
- 12 October 2022
- Published by Taylor & Francis in Structure and Infrastructure Engineering
- Vol. 20 (5) , 741-759
Efficient Bayesian model selection and calibration using field data for a reinforced concrete slab bridge
In this work, a reinforced concrete slab bridge (instrumented and tested in 2018) is investigated. Based on field data, a finite element model of the bridge is calibrated. Model selection is performed both based on log evidence and posterior predictive capabilities. It is investigated if the models selected based on the log evidence also induce the most accurate posterior predictions. The influence of different assumptions on modelling the spatial distribution of the stiffness and different possible suggestions on how to include prediction errors and model bias are investigated. Comparing the conclusions based on log evidence and posterior predictions, only using the log evidence for model selection could be debated. Models performing best when considering the log evidence led to the least accurate posterior predictions, and models rejected based on the log evidence could still have good predictive capabilities. Considering the different model classes, introducing spatial variation of the stiffness leads to a posterior prediction closer to the measurements. Introducing a global model bias leads to a better match between predictions and measurements compared to not including this model bias. Even better posterior predictions are achieved if this model bias is quantified locally for the different considered datapoints.
- Open Access
- 11 April 2022
- Published by Elsevier in Engineering Fracture Mechanics
- Vol. 268
Uncertainty quantification of the failure assessment diagram for flawed steel components in BS 7910:2019
- 11 April 2022
- Published by Elsevier in Engineering Fracture Mechanics
- Vol. 268
Open AccessUncertainty quantification of the failure assessment diagram for flawed steel components in BS 7910:2019
- 3 April 2022
- Published by Wiley in Structural Concrete
- Vol. 23 (6) , 3569-3600
Structural and load parameter estimation of a real‐world reinforced concrete slab bridge using measurements and Bayesian statistics
This paper describes a static diagnostic load testing and measurement campaign of a reinforced concrete road bridge in Amsterdam. We consider 29 vertical translation sensors and 37 strain sensors. Multiple Bayesian parameter estimations are performed to estimate two structural and two load parameters of a three‐dimensional finite element (FE) model. The structural parameters are the concrete elastic modulus of the deck and the rotational spring stiffness at the piers. The load (truck) parameters are the load mass and position, which are estimated from separate measurements. We found that estimating (updating) the two selected structural parameters considerably improves the vertical translation model prediction accuracy over the FE model that was built before the measurement campaign, that is, the R2 score increases from 0.90 to 0.98. Moreover, the load (truck) mass and position parameters can be estimated with high accuracy and high precision. When all measurements are used, the load mass and position are estimated with an error, respectively, less than 1.5 tonnes and 0.5 m with respect to the ground truth. On the negative side: when only certain strain measurements are used, the 90% posterior credible interval is about 3 m and 10 tonnes off the ground truth. Although the results are promising, they are restricted to the considered case and, more importantly, to operational loading conditions where the loading is well controlled. All data, models, and code are freely available upon request.
- 3 April 2022
- Published by Wiley in Structural Concrete
- Vol. 23 (6) , 3569-3600
Structural and load parameter estimation of a real‐world reinforced concrete slab bridge using measurements and Bayesian statistics
This paper describes a static diagnostic load testing and measurement campaign of a reinforced concrete road bridge in Amsterdam. We consider 29 vertical translation sensors and 37 strain sensors. Multiple Bayesian parameter estimations are performed to estimate two structural and two load parameters of a three‐dimensional finite element (FE) model. The structural parameters are the concrete elastic modulus of the deck and the rotational spring stiffness at the piers. The load (truck) parameters are the load mass and position, which are estimated from separate measurements. We found that estimating (updating) the two selected structural parameters considerably improves the vertical translation model prediction accuracy over the FE model that was built before the measurement campaign, that is, the R2 score increases from 0.90 to 0.98. Moreover, the load (truck) mass and position parameters can be estimated with high accuracy and high precision. When all measurements are used, the load mass and position are estimated with an error, respectively, less than 1.5 tonnes and 0.5 m with respect to the ground truth. On the negative side: when only certain strain measurements are used, the 90% posterior credible interval is about 3 m and 10 tonnes off the ground truth. Although the results are promising, they are restricted to the considered case and, more importantly, to operational loading conditions where the loading is well controlled. All data, models, and code are freely available upon request.
- 1 February 2022
- Published by Elsevier in Computers and Geotechnics
- Vol. 142
Probabilistic parameter estimation and reliability assessment of a simulated sheet pile wall system
- 1 February 2022
- Published by Elsevier in Computers and Geotechnics
- Vol. 142
Probabilistic parameter estimation and reliability assessment of a simulated sheet pile wall system
- 25 June 2019
- Published by Wiley in Structural Concrete
- Vol. 21 (1) , 32-47
On the value of a reliability‐based nonlinear finite element analysis approach in the assessment of concrete structures
The objective of this paper is to explore the value of reliability‐based nonlinear finite element analysis (NLFEA) over the currently available, standardized assessment methods. To our knowledge, no studies are available on this subject, and this paper provides a first insight into the value and reliability level of these assessment methods. The exploration is illustrated through three reinforced concrete structural members: a continuous girder, a continuous deep beam, and a high‐strength deep beam. The analysis is performed gradually: step by step advancing the approximation level of the mechanical and the probabilistic models. The added value of the reliability‐based NLFEA over the semi‐probabilistic Eurocode (EC) method is found to be on average 0.60. In other words: even if according to the semi‐probabilistic EC method the design action (Ed) is 60% higher than the design resistance (Rd), the compliance with the target reliability criterion can be demonstrated by a reliability‐based NLFEA. Furthermore, it is observed that the gain and its cause (i.e., more advanced mechanical or probabilistic models) are different for the three cases. Though this outcome is restricted to the analyzed cases and should be interpreted as an upper limit added value, it indicates that a more detailed physical representation of the problem and an explicit treatment of uncertainties may uncover substantial reserves compared with the currently available, standardized assessment methods. Hence, the reliability‐based NLFEA method offers a promising alternative in the assessment of existing structures, enabling to avoid expensive measures that might be needed based on simplified methods.
- 25 June 2019
- Published by Wiley in Structural Concrete
- Vol. 21 (1) , 32-47
On the value of a reliability‐based nonlinear finite element analysis approach in the assessment of concrete structures
The objective of this paper is to explore the value of reliability‐based nonlinear finite element analysis (NLFEA) over the currently available, standardized assessment methods. To our knowledge, no studies are available on this subject, and this paper provides a first insight into the value and reliability level of these assessment methods. The exploration is illustrated through three reinforced concrete structural members: a continuous girder, a continuous deep beam, and a high‐strength deep beam. The analysis is performed gradually: step by step advancing the approximation level of the mechanical and the probabilistic models. The added value of the reliability‐based NLFEA over the semi‐probabilistic Eurocode (EC) method is found to be on average 0.60. In other words: even if according to the semi‐probabilistic EC method the design action (Ed) is 60% higher than the design resistance (Rd), the compliance with the target reliability criterion can be demonstrated by a reliability‐based NLFEA. Furthermore, it is observed that the gain and its cause (i.e., more advanced mechanical or probabilistic models) are different for the three cases. Though this outcome is restricted to the analyzed cases and should be interpreted as an upper limit added value, it indicates that a more detailed physical representation of the problem and an explicit treatment of uncertainties may uncover substantial reserves compared with the currently available, standardized assessment methods. Hence, the reliability‐based NLFEA method offers a promising alternative in the assessment of existing structures, enabling to avoid expensive measures that might be needed based on simplified methods.