C. Andriotis
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
35 records found
1
Building Automation and Control Systems (BACS) enhance energy efficiency in office buildings but often leave occupants dissatisfied, especially in shared offices with diverse indoor environmental preferences. Personal Comfort Models (PCM) capture individual preferences, yet these models can be difficult to apply in shared environments. This study examines whether a generalized personalized modeling would be necessary, or instead aggregated thermal preferences adequately represent individual needs in shared spaces. By mining the ASHRAE Global Thermal Comfort Database II, we analyzed 11 naturally ventilated office buildings spanning four climate zones, each providing 187 to 568 data points of subjective thermal preferences and concurrent environmental measurements. To capture uncertainties in occupant feedback and monitoring, we developed a Bayesian multinomial logistic framework. We compare an aggregated model (all responses together) with a hierarchical (partial-pooling) model having occupant-specific intercepts, using the No‑U‑Turn Sampler for Hamiltonian Monte Carlo. The Widely Applicable Bayesian Information Criterion confirms that hierarchical approach outperforms the aggregated model in every case. In four of the eleven studies, aggregated air temperature ranges satisfied occupants’ demands within a 95% posterior credible interval. Moreover, in most buildings, these temperature ranges aligned well with the majority of occupant preferences, suggesting that prior knowledge of building conditions or occupant variability might reduce the need for highly customized thermal environment. Future research should explore these findings in conditioned buildings and across other indoor environmental quality domains, where greater variability could exist.
...
Building Automation and Control Systems (BACS) enhance energy efficiency in office buildings but often leave occupants dissatisfied, especially in shared offices with diverse indoor environmental preferences. Personal Comfort Models (PCM) capture individual preferences, yet these models can be difficult to apply in shared environments. This study examines whether a generalized personalized modeling would be necessary, or instead aggregated thermal preferences adequately represent individual needs in shared spaces. By mining the ASHRAE Global Thermal Comfort Database II, we analyzed 11 naturally ventilated office buildings spanning four climate zones, each providing 187 to 568 data points of subjective thermal preferences and concurrent environmental measurements. To capture uncertainties in occupant feedback and monitoring, we developed a Bayesian multinomial logistic framework. We compare an aggregated model (all responses together) with a hierarchical (partial-pooling) model having occupant-specific intercepts, using the No‑U‑Turn Sampler for Hamiltonian Monte Carlo. The Widely Applicable Bayesian Information Criterion confirms that hierarchical approach outperforms the aggregated model in every case. In four of the eleven studies, aggregated air temperature ranges satisfied occupants’ demands within a 95% posterior credible interval. Moreover, in most buildings, these temperature ranges aligned well with the majority of occupant preferences, suggesting that prior knowledge of building conditions or occupant variability might reduce the need for highly customized thermal environment. Future research should explore these findings in conditioned buildings and across other indoor environmental quality domains, where greater variability could exist.
Journal article
(2026)
-
Mohammad Saifullah, K. G. Papakonstantinou, A. Bhattacharya, S. M. Stoffels, C. P. Andriotis
Life-cycle management of large-scale transportation systems is a computationally intensive task. It requires determining a sequence of inspection and maintenance decisions to minimize long-term risks and costs while dealing with multiple uncertainties and constraints that lie in high-dimensional spaces. Traditional approaches, such as static age- or condition-based maintenance and risk-based or periodic inspection plans, have been widely applied but often suffer from limitations related to optimality, scalability, and the ability to properly handle uncertainty in the loop of optimization. Moreover, many existing methods rely on unconstrained formulations that overlook critical hard and soft operational constraints. We address these issues in this work by casting the optimization problem within the framework of constrained Partially Observable Markov Decision Processes (POMDPs), which provide a robust mathematical foundation for stochastic sequential decision-making under observation uncertainties, in the presence of risk and resource limitations. To tackle the high dimensionality of state and action spaces, we propose DDMAC-CTDE, a Deep Decentralized Multi-Agent Actor-Critic (DDMAC) reinforcement learning architecture with Centralized Training and Decentralized Execution (CTDE). To demonstrate the utility of the proposed framework, we also develop a new comprehensive benchmark environment representing an existing transportation network in Virginia, U.S., with heterogeneous pavement and bridge assets undergoing nonstationary degradation. This environment incorporates multiple practical constraints related to budget limits, performance guidelines from State and Federal agencies, traffic delays, and risk considerations. On this benchmark, DDMAC-CTDE consistently outperforms standard transportation management baselines, producing more reliable, cost- and risk-efficient policies that respect operational constraints. Together, the proposed framework and benchmark provide (i) a scalable, constraint-aware methodology for sequential decision-making under uncertainty, and (ii) a realistic, rigorous testbed for comprehensive evaluation of Deep Reinforcement Learning (DRL) solutions for transportation infrastructure management.
...
Life-cycle management of large-scale transportation systems is a computationally intensive task. It requires determining a sequence of inspection and maintenance decisions to minimize long-term risks and costs while dealing with multiple uncertainties and constraints that lie in high-dimensional spaces. Traditional approaches, such as static age- or condition-based maintenance and risk-based or periodic inspection plans, have been widely applied but often suffer from limitations related to optimality, scalability, and the ability to properly handle uncertainty in the loop of optimization. Moreover, many existing methods rely on unconstrained formulations that overlook critical hard and soft operational constraints. We address these issues in this work by casting the optimization problem within the framework of constrained Partially Observable Markov Decision Processes (POMDPs), which provide a robust mathematical foundation for stochastic sequential decision-making under observation uncertainties, in the presence of risk and resource limitations. To tackle the high dimensionality of state and action spaces, we propose DDMAC-CTDE, a Deep Decentralized Multi-Agent Actor-Critic (DDMAC) reinforcement learning architecture with Centralized Training and Decentralized Execution (CTDE). To demonstrate the utility of the proposed framework, we also develop a new comprehensive benchmark environment representing an existing transportation network in Virginia, U.S., with heterogeneous pavement and bridge assets undergoing nonstationary degradation. This environment incorporates multiple practical constraints related to budget limits, performance guidelines from State and Federal agencies, traffic delays, and risk considerations. On this benchmark, DDMAC-CTDE consistently outperforms standard transportation management baselines, producing more reliable, cost- and risk-efficient policies that respect operational constraints. Together, the proposed framework and benchmark provide (i) a scalable, constraint-aware methodology for sequential decision-making under uncertainty, and (ii) a realistic, rigorous testbed for comprehensive evaluation of Deep Reinforcement Learning (DRL) solutions for transportation infrastructure management.
Building automation and control systems (BACS) are central to energy performance and occupant comfort in non-residential buildings. Comfort is inherently multi-domain, including thermal, visual, acoustic, and air quality requirements. Multi-domain BACS involves frequent trade-offs across domains when conflicting control actions arise, such as providing glare control versus daylight availability. Yet existing occupant-centric control research treats building services in isolation, and prior multi-domain comfort reviews rarely examine how multi-domain demands are integrated into BACS decision logic across services. We conducted a systematic review of 43 studies to examine how multi-domain occupant demands are represented and operationalized in BACS. Across the evidence base, thermal comfort is universal, while visual and air quality are frequently included. Acoustics is rarely addressed due controllability constraints. Most studies remain unimodal in their demand representation, even when multiple domains are in scope. Integrated BACS implementations are therefore largely built on within-domain formulations. Multimodal demand models that encode cross-domain and combined effects are uncommon and are rarely implemented in integrated BACS. Rule-based strategies dominate multi-domain controllers. Optimization-based and learning-based controllers are also used, but they often rely on fixed weights or reward terms that make trade-offs difficult to interpret. In addition, actuator choice is rarely made explicit when multiple services can achieve the same target state. Future research should benchmark unimodal and multimodal demand formulations under comparable control contexts, extend bottom-up multimodal models beyond thermal and air quality into integrated BACS, especially for façade control, and develop transparent, preference-aware policy designs that make priorities and service actions understandable.
...
Building automation and control systems (BACS) are central to energy performance and occupant comfort in non-residential buildings. Comfort is inherently multi-domain, including thermal, visual, acoustic, and air quality requirements. Multi-domain BACS involves frequent trade-offs across domains when conflicting control actions arise, such as providing glare control versus daylight availability. Yet existing occupant-centric control research treats building services in isolation, and prior multi-domain comfort reviews rarely examine how multi-domain demands are integrated into BACS decision logic across services. We conducted a systematic review of 43 studies to examine how multi-domain occupant demands are represented and operationalized in BACS. Across the evidence base, thermal comfort is universal, while visual and air quality are frequently included. Acoustics is rarely addressed due controllability constraints. Most studies remain unimodal in their demand representation, even when multiple domains are in scope. Integrated BACS implementations are therefore largely built on within-domain formulations. Multimodal demand models that encode cross-domain and combined effects are uncommon and are rarely implemented in integrated BACS. Rule-based strategies dominate multi-domain controllers. Optimization-based and learning-based controllers are also used, but they often rely on fixed weights or reward terms that make trade-offs difficult to interpret. In addition, actuator choice is rarely made explicit when multiple services can achieve the same target state. Future research should benchmark unimodal and multimodal demand formulations under comparable control contexts, extend bottom-up multimodal models beyond thermal and air quality into integrated BACS, especially for façade control, and develop transparent, preference-aware policy designs that make priorities and service actions understandable.
Journal article
(2026)
-
Pooyan Kazemi, Michela Turrin, Charalampos Andriotis, Alireza Entezami, Stefano Mariani, Aldo Ghisi
Early-stage design optimisation of tall buildings has been studied for several decades through performance-based design, parametric modelling, evolutionary algorithms and, more recently, machine-learning surrogates. A remaining challenge, however, is the computationally efficient integration of seismic time-history simulation, architectural form generation, multi-response prediction and multi-objective design exploration within a single reproducible workflow. This issue is particularly relevant for diagrid tall buildings, where geometric transformations such as tapering, twisting and curvilinearity strongly affect structural response, material use and environmental indicators. This study proposes a surrogate-assisted framework for early-stage multi-objective seismic design optimisation of diagrid tall buildings. A database of 1000 parametric building configurations was generated and analysed under seven bidirectional ground-motion records, resulting in 7000 OpenSeesPy time-history simulations. The original 22 input variables were transformed into 45 preprocessed input columns through scaling, categorical encoding and conditional masking of geometry-dependent variables. A multi-input, multi-output feed-forward neural network was then trained to predict 91 responses, including 69 structural responses, 14 environmental and cost-related responses and 8 geometry-dependent quantities. The trained surrogate was coupled with NSGA-II to screen candidate designs using selected objectives related to acceleration, displacement, inter-storey drift, von Mises stress and total mass. Approximately 2000 surrogate-based optimisation evaluations were produced over 10 generations and reduced to 408 unique candidate geometries. The proposed framework does not replace final nonlinear finite-element verification; rather, it provides a computationally efficient decision-support tool for rapidly reducing a high-dimensional diagrid design space before detailed engineering assessment.
...
Early-stage design optimisation of tall buildings has been studied for several decades through performance-based design, parametric modelling, evolutionary algorithms and, more recently, machine-learning surrogates. A remaining challenge, however, is the computationally efficient integration of seismic time-history simulation, architectural form generation, multi-response prediction and multi-objective design exploration within a single reproducible workflow. This issue is particularly relevant for diagrid tall buildings, where geometric transformations such as tapering, twisting and curvilinearity strongly affect structural response, material use and environmental indicators. This study proposes a surrogate-assisted framework for early-stage multi-objective seismic design optimisation of diagrid tall buildings. A database of 1000 parametric building configurations was generated and analysed under seven bidirectional ground-motion records, resulting in 7000 OpenSeesPy time-history simulations. The original 22 input variables were transformed into 45 preprocessed input columns through scaling, categorical encoding and conditional masking of geometry-dependent variables. A multi-input, multi-output feed-forward neural network was then trained to predict 91 responses, including 69 structural responses, 14 environmental and cost-related responses and 8 geometry-dependent quantities. The trained surrogate was coupled with NSGA-II to screen candidate designs using selected objectives related to acceleration, displacement, inter-storey drift, von Mises stress and total mass. Approximately 2000 surrogate-based optimisation evaluations were produced over 10 generations and reduced to 408 unique candidate geometries. The proposed framework does not replace final nonlinear finite-element verification; rather, it provides a computationally efficient decision-support tool for rapidly reducing a high-dimensional diagrid design space before detailed engineering assessment.
Tall buildings with diagrids
Design exploration and assessment of the influence of the architectural form on the structural behavior
Journal article
(2026)
-
Pooyan Kazemi, Alireza Entezami, Michela Turrin, Charalampos Andriotis, Stefano Mariani, Aldo Ghisi
Tall buildings with diagrids offer several advantages over alternative structural solutions, including the possibility of reducing the dimensions of interior cores and freeing up interior space. However, the interplay between structural systems and architectural form is not always adequately considered during the early stages of design to support quantitative decision-making. In this study, a database comprising 1000 tall buildings has been generated using parametric modeling. The buildings range in height from 140 to 300 meters. These buildings have then been subjected to seven ground motions selected from the NGA-West2 database, for a total of 7000 simulations. Data related to geometric, structural, and environmental variables will subsequently be processed to support investigations based on artificial intelligence-driven generative approaches aimed at producing alternative optimized solutions. In this paper, the data are described in terms of specific 22 input (10 architectural, 12 related to the ground motion) and 99 output (69 structural, 23 environmental, 7 geometry-dependent) features, demonstrating the extent to which the database has been expanded with respect to the original design space. The dataset is made available through a repository for the architecture, engineering, and construction community.
...
Tall buildings with diagrids offer several advantages over alternative structural solutions, including the possibility of reducing the dimensions of interior cores and freeing up interior space. However, the interplay between structural systems and architectural form is not always adequately considered during the early stages of design to support quantitative decision-making. In this study, a database comprising 1000 tall buildings has been generated using parametric modeling. The buildings range in height from 140 to 300 meters. These buildings have then been subjected to seven ground motions selected from the NGA-West2 database, for a total of 7000 simulations. Data related to geometric, structural, and environmental variables will subsequently be processed to support investigations based on artificial intelligence-driven generative approaches aimed at producing alternative optimized solutions. In this paper, the data are described in terms of specific 22 input (10 architectural, 12 related to the ground motion) and 99 output (69 structural, 23 environmental, 7 geometry-dependent) features, demonstrating the extent to which the database has been expanded with respect to the original design space. The dataset is made available through a repository for the architecture, engineering, and construction community.
Life-cycle fragility analysis of aging reinforced concrete bridges
A dynamic Bayesian network approach
Reinforced concrete bridges are predominant structural systems in transportation infrastructure. Their exposure to chronic and sudden stressors, such as corrosion and earthquakes, make them prone to risks with severe socioeconomic consequences. While time-dependent single-component seismic fragility formulations have advanced the frontier of life-cycle probabilistic risk assessment, state-dependent multi-component representations of damage and deterioration, paramount for structural integrity management, still lack a systematic probabilistic framework. This paper develops a novel dynamic Bayesian network to evaluate the life-cycle fragility functions of aging bridges, encapsulating the impacts of corrosion and seismic phenomena over time. The network establishes Markovian transitions among deterioration states for various bridge components integrating chloride diffusion and corrosion propagation models with non-stationary Gamma processes. A methodology for deriving and state-dependent fragility at the component and system levels depending on several deterioration scenarios is presented. Our framework is exemplified in an archetypical 4-span bridge, demonstrating the longitudinal effects of corrosion on the system's seismic fragility for splash and atmospheric conditions. Insights from the multi-component analysis highlight the capabilities in understanding the pathologies and evolving mechanical interactions among components. The adaptability in accommodating on-site observations and advanced decision-making algorithms is discussed, demonstrating the suitability of the framework for applications requiring flexible and updatable virtual environments.
...
Reinforced concrete bridges are predominant structural systems in transportation infrastructure. Their exposure to chronic and sudden stressors, such as corrosion and earthquakes, make them prone to risks with severe socioeconomic consequences. While time-dependent single-component seismic fragility formulations have advanced the frontier of life-cycle probabilistic risk assessment, state-dependent multi-component representations of damage and deterioration, paramount for structural integrity management, still lack a systematic probabilistic framework. This paper develops a novel dynamic Bayesian network to evaluate the life-cycle fragility functions of aging bridges, encapsulating the impacts of corrosion and seismic phenomena over time. The network establishes Markovian transitions among deterioration states for various bridge components integrating chloride diffusion and corrosion propagation models with non-stationary Gamma processes. A methodology for deriving and state-dependent fragility at the component and system levels depending on several deterioration scenarios is presented. Our framework is exemplified in an archetypical 4-span bridge, demonstrating the longitudinal effects of corrosion on the system's seismic fragility for splash and atmospheric conditions. Insights from the multi-component analysis highlight the capabilities in understanding the pathologies and evolving mechanical interactions among components. The adaptability in accommodating on-site observations and advanced decision-making algorithms is discussed, demonstrating the suitability of the framework for applications requiring flexible and updatable virtual environments.
Predicting building operational energy under material degradation and climate uncertainty
A sensitivity analysis
Conference paper
(2025)
-
A.M. Koniari, C. Andriotis, S. Bianchi, Pablo G. Morato, S. Khademi, M. Overend
Building energy prediction models expedite performance assessment and assist in decision making, from early-stage design to retrofit planning at single- or multi-building scales. However, the number of parameters involved in the energy performance evaluation often impede the prediction process requiring the assimilation of high-dimensional, uncertain input. This is compounded further at multi-building scale e.g. urban energy modelling, due to the increased complexity of evaluating diverse building geometries. While single-building sensitivity and uncertainty analysis is well-established for identifying the most influential input parameters and evaluate the uncertainty effects on energy demand, these are hard to generalize at multi-building scale which remains relatively unexplored. The present study advances existing research by applying a variance-based sensitivity analysis to assess the impact of varying (i) building façade layout, (ii) envelope thermal properties, (iii) envelope air tightness and (iv) building occupancy. The analysis is conducted for multiple buildings under two future climate variations, while also considering the degradation of material thermal properties. The latter is derived from known deterioration models for single-building uncertainty propagation, relying on experimental and simulated data. The approach is applied to a temperate oceanic climate with particular focus on the Dutch building stock, including a sample of buildings with diverse geometric characteristics in Rotterdam. First-order Sobol indices are computed to evaluate the impact with respect to the heating, cooling and total energy demand. Our findings indicate that infiltration is the most influential factor for heating energy demand, whereas cooling is mostly affected by the envelope thermal properties and, particularly, window solar heat gain coefficient. Common patterns regarding the impact of insulation across different envelope components can be identified among buildings with similar orientation and compactness ratio indicating the importance of considering these geometric properties in retrofit decision-making workflows.
...
Building energy prediction models expedite performance assessment and assist in decision making, from early-stage design to retrofit planning at single- or multi-building scales. However, the number of parameters involved in the energy performance evaluation often impede the prediction process requiring the assimilation of high-dimensional, uncertain input. This is compounded further at multi-building scale e.g. urban energy modelling, due to the increased complexity of evaluating diverse building geometries. While single-building sensitivity and uncertainty analysis is well-established for identifying the most influential input parameters and evaluate the uncertainty effects on energy demand, these are hard to generalize at multi-building scale which remains relatively unexplored. The present study advances existing research by applying a variance-based sensitivity analysis to assess the impact of varying (i) building façade layout, (ii) envelope thermal properties, (iii) envelope air tightness and (iv) building occupancy. The analysis is conducted for multiple buildings under two future climate variations, while also considering the degradation of material thermal properties. The latter is derived from known deterioration models for single-building uncertainty propagation, relying on experimental and simulated data. The approach is applied to a temperate oceanic climate with particular focus on the Dutch building stock, including a sample of buildings with diverse geometric characteristics in Rotterdam. First-order Sobol indices are computed to evaluate the impact with respect to the heating, cooling and total energy demand. Our findings indicate that infiltration is the most influential factor for heating energy demand, whereas cooling is mostly affected by the envelope thermal properties and, particularly, window solar heat gain coefficient. Common patterns regarding the impact of insulation across different envelope components can be identified among buildings with similar orientation and compactness ratio indicating the importance of considering these geometric properties in retrofit decision-making workflows.
More with less
Topology optimization strategies for structural glass design
Journal article
(2025)
-
Jackson L. Jewett, Anna Maria Koniari, Charalampos P. Andriotis, Faidra Oikonomopoulou, Telesilla Bristogianni, Josephine V. Carstensen
Advances in structural glass have enabled a new paradigm in expressive and transparent architecture. Cast glass can further extend the possibilities of structural glass by allowing for more complex and sophisticated shapes than the current planar geometries of structural float glass. However, the use of cast glass is currently limited because of the lengthy annealing process, making massive component sizes impractical to fabricate. Topology optimization (TO) has been proposed as a solution to this problem, as it is known to generate structurally efficient designs with a low volume of material. If tailored appropriately, TO can reduce component sizes and thereby diminish the total annealing time needed, while intelligently placing material in the areas where it will be utilized most effectively. For TO of glass to be successful, algorithms must properly capture glass’s specific material behavior. This research proposes a suite of TO algorithmic frameworks that design specifically for structural glass. These algorithms are demonstrated in a 2D design space, and the resulting geometries are fabricated using cut float glass and tested for experimental comparison on a 4-point bending load case. The results of these experiments provide valuable insights into the development of TO for structural glass, and help inform future research in TO of large-scale cast glass structures.
...
Advances in structural glass have enabled a new paradigm in expressive and transparent architecture. Cast glass can further extend the possibilities of structural glass by allowing for more complex and sophisticated shapes than the current planar geometries of structural float glass. However, the use of cast glass is currently limited because of the lengthy annealing process, making massive component sizes impractical to fabricate. Topology optimization (TO) has been proposed as a solution to this problem, as it is known to generate structurally efficient designs with a low volume of material. If tailored appropriately, TO can reduce component sizes and thereby diminish the total annealing time needed, while intelligently placing material in the areas where it will be utilized most effectively. For TO of glass to be successful, algorithms must properly capture glass’s specific material behavior. This research proposes a suite of TO algorithmic frameworks that design specifically for structural glass. These algorithms are demonstrated in a 2D design space, and the resulting geometries are fabricated using cut float glass and tested for experimental comparison on a 4-point bending load case. The results of these experiments provide valuable insights into the development of TO for structural glass, and help inform future research in TO of large-scale cast glass structures.
This paper presents ongoing research that aims to develop a closed-loop and real-time error
detection and correction system in 3D clay printing (3DCP) using computer vision and machine
learning. [...] ...
detection and correction system in 3D clay printing (3DCP) using computer vision and machine
learning. [...] ...
This paper presents ongoing research that aims to develop a closed-loop and real-time error
detection and correction system in 3D clay printing (3DCP) using computer vision and machine
learning. [...]
detection and correction system in 3D clay printing (3DCP) using computer vision and machine
learning. [...]
Journal article
(2025)
-
Jesse van Remmerden, Maurice Kenter, Diederik M. Roijers, Charalampos Andriotis, Yingqian Zhang, Zaharah Bukhsh
In this paper, we introduce multi-objective deep centralized multi-agent actor-critic (MO-DCMAC), a multi-objective reinforcement learning method for infrastructural maintenance optimization, an area traditionally dominated by single-objective reinforcement learning (RL) approaches. Previous single-objective RL methods combine multiple objectives, such as probability of collapse and cost, into a singular reward signal through reward-shaping. In contrast, MO-DCMAC can optimize a policy for multiple objectives directly, even when the utility function is nonlinear. We evaluated MO-DCMAC using two utility functions, which use probability of collapse and cost as input. The first utility function is the threshold utility, in which MO-DCMAC should minimize cost so that the probability of collapse is never above the threshold. The second is based on the failure mode, effects, and criticality analysis methodology used by asset managers to assess maintenance plans. We evaluated MO-DCMAC, with both utility functions, in multiple maintenance environments, including ones based on a case study of the historical quay walls of Amsterdam. The performance of MO-DCMAC was compared against multiple rule-based policies based on heuristics currently used for constructing maintenance plans. Our results demonstrate that MO-DCMAC outperforms traditional rule-based policies across various environments and utility functions.
...
In this paper, we introduce multi-objective deep centralized multi-agent actor-critic (MO-DCMAC), a multi-objective reinforcement learning method for infrastructural maintenance optimization, an area traditionally dominated by single-objective reinforcement learning (RL) approaches. Previous single-objective RL methods combine multiple objectives, such as probability of collapse and cost, into a singular reward signal through reward-shaping. In contrast, MO-DCMAC can optimize a policy for multiple objectives directly, even when the utility function is nonlinear. We evaluated MO-DCMAC using two utility functions, which use probability of collapse and cost as input. The first utility function is the threshold utility, in which MO-DCMAC should minimize cost so that the probability of collapse is never above the threshold. The second is based on the failure mode, effects, and criticality analysis methodology used by asset managers to assess maintenance plans. We evaluated MO-DCMAC, with both utility functions, in multiple maintenance environments, including ones based on a case study of the historical quay walls of Amsterdam. The performance of MO-DCMAC was compared against multiple rule-based policies based on heuristics currently used for constructing maintenance plans. Our results demonstrate that MO-DCMAC outperforms traditional rule-based policies across various environments and utility functions.
Conference paper
(2025)
-
P. Martinez-Alcaraz, P. de la Barra, C. P. Andriotis, Y. Wang, A. Luna-Navarro
It is a challenge for traditional building control systems to meet occupants’ needs in shared spaces due to the lack of understanding of individual occupant thermal preferences. This is a barrier to balancing energy efficiency and indoor environmental quality (IEQ). Advanced statistical learning methods offer new solutions towards more energy-efficient and user-centric control logics. In this work, a control logic is proposed to optimise the heating, ventilation and air conditioning (HVAC) operation based on thermal comfort archetype preferences, leveraging the ASHRAE Global Thermal Comfort Database II in conjunction with energy simulations. First, we apply the k-means clustering algorithm to categorize occupants into different archetypes regarding their common feedback on the thermal environment. Then, we fit a Bayesian logistic regression model to predict the thermal comfort preferences of different archetypes based on IEQ data. Finally, we identify two occupant-centric control logics to optimize HVAC operation to meet occupants’ requirements: (i) considering a unified response of thermal comfort in the space, and (ii) ensuring the dynamic optimal setpoint when conflicting occupant archetypes are present. Having compared this control logic with a common rule-based logic, our results demonstrate the potential of occupant-centric controls and the importance of multi-objective metrics in accounting for energy efficiency.
...
It is a challenge for traditional building control systems to meet occupants’ needs in shared spaces due to the lack of understanding of individual occupant thermal preferences. This is a barrier to balancing energy efficiency and indoor environmental quality (IEQ). Advanced statistical learning methods offer new solutions towards more energy-efficient and user-centric control logics. In this work, a control logic is proposed to optimise the heating, ventilation and air conditioning (HVAC) operation based on thermal comfort archetype preferences, leveraging the ASHRAE Global Thermal Comfort Database II in conjunction with energy simulations. First, we apply the k-means clustering algorithm to categorize occupants into different archetypes regarding their common feedback on the thermal environment. Then, we fit a Bayesian logistic regression model to predict the thermal comfort preferences of different archetypes based on IEQ data. Finally, we identify two occupant-centric control logics to optimize HVAC operation to meet occupants’ requirements: (i) considering a unified response of thermal comfort in the space, and (ii) ensuring the dynamic optimal setpoint when conflicting occupant archetypes are present. Having compared this control logic with a common rule-based logic, our results demonstrate the potential of occupant-centric controls and the importance of multi-objective metrics in accounting for energy efficiency.
Data-driven prediction of infrastructure aging is challenging due to the complex stochastic nature of degradation effects and the ill-documented historical records. Degradation modeling is, however, crucial for predictive maintenance that is key for infrastructure integrity. This study presents a multi-attribute, data-driven approach for modelling stochastic degradation and maintenance effects of roads, mining an extensive database of geo-located historical inspection and maintenance records from the municipality of Amsterdam. Inspection data track pavement conditions at irregular intervals across ten discrete states per road segment, following the Dutch CROW 146 protocol. Damage severity and extent for eight damage modes is captured, i.e., for transverse unevenness, irregularities, ravelling, edge damage, crack formation, joint filling, joint width, and settling. The maintenance dataset includes >25k minor interventions across 17k road segments, indicating repair requirements, and 200+ major maintenance projects, covering 21k segments where interventions are planned, all without verifying completion. This complicates accurate modelling of natural degradation as it is confounded by maintenance effects. To address the issue of irregular inspections, degradation is first modelled as a continuous-time Markov chain. Thereby, transition rates are estimated, which are then converted to discrete-time Markov chain transition probability matrices to eventually support regular maintenance planning. We further learn the effects of minor and major maintenance activities, as defined and recorded in the database. Based on the estimated degradation transitions, pre-maintenance and post-maintenance state distributions are estimated. Instantaneous maintenance transition matrices are computed by minimizing the cross-entropy between the pre-maintenance state after the intervention and the post-maintenance state. The model allows for a multi-attribute approach, segmenting roads based on construction material (e.g., asphalt, tiled pavement) and traffic loads (e.g., residential, commercial/pedestrian). The approach is exemplified for tiled pavements for a section of the road network of Amsterdam, where the effects of minor and major maintenance are ablated for long-term predictions. Although applied to Amsterdam, this method is relevant to any infrastructure system with discrete state datasets, providing a foundation for data-driven decision-making in infrastructure management.
...
Data-driven prediction of infrastructure aging is challenging due to the complex stochastic nature of degradation effects and the ill-documented historical records. Degradation modeling is, however, crucial for predictive maintenance that is key for infrastructure integrity. This study presents a multi-attribute, data-driven approach for modelling stochastic degradation and maintenance effects of roads, mining an extensive database of geo-located historical inspection and maintenance records from the municipality of Amsterdam. Inspection data track pavement conditions at irregular intervals across ten discrete states per road segment, following the Dutch CROW 146 protocol. Damage severity and extent for eight damage modes is captured, i.e., for transverse unevenness, irregularities, ravelling, edge damage, crack formation, joint filling, joint width, and settling. The maintenance dataset includes >25k minor interventions across 17k road segments, indicating repair requirements, and 200+ major maintenance projects, covering 21k segments where interventions are planned, all without verifying completion. This complicates accurate modelling of natural degradation as it is confounded by maintenance effects. To address the issue of irregular inspections, degradation is first modelled as a continuous-time Markov chain. Thereby, transition rates are estimated, which are then converted to discrete-time Markov chain transition probability matrices to eventually support regular maintenance planning. We further learn the effects of minor and major maintenance activities, as defined and recorded in the database. Based on the estimated degradation transitions, pre-maintenance and post-maintenance state distributions are estimated. Instantaneous maintenance transition matrices are computed by minimizing the cross-entropy between the pre-maintenance state after the intervention and the post-maintenance state. The model allows for a multi-attribute approach, segmenting roads based on construction material (e.g., asphalt, tiled pavement) and traffic loads (e.g., residential, commercial/pedestrian). The approach is exemplified for tiled pavements for a section of the road network of Amsterdam, where the effects of minor and major maintenance are ablated for long-term predictions. Although applied to Amsterdam, this method is relevant to any infrastructure system with discrete state datasets, providing a foundation for data-driven decision-making in infrastructure management.
Long-term inspection and maintenance (I&M) planning, a multi-stage stochastic optimization problem, can be efficiently formulated as a partially observable Markov decision process (POMDP). However, within this context, single-agent approaches do not scale well for large multi-component systems since the joint state, action and observation spaces grow exponentially with the number of components. To alleviate this curse of dimensionality, cooperative decentralized approaches, known as decentralized POMDPs, are often adopted and solved using multi-agent deep reinforcement learning (MADRL) algorithms. This paper examines the centralization vs. decentralization performance of MADRL formulations in I&M planning of multi-component systems. Towards this, we set up a comprehensive computational experimental program focused on k-out-of-n system configurations, a common and broadly applicable archetype of deteriorating engineering systems, to highlight the manifestations of MADRL strengths and pathologies when optimizing global returns under varying decentralization relaxations.
...
Long-term inspection and maintenance (I&M) planning, a multi-stage stochastic optimization problem, can be efficiently formulated as a partially observable Markov decision process (POMDP). However, within this context, single-agent approaches do not scale well for large multi-component systems since the joint state, action and observation spaces grow exponentially with the number of components. To alleviate this curse of dimensionality, cooperative decentralized approaches, known as decentralized POMDPs, are often adopted and solved using multi-agent deep reinforcement learning (MADRL) algorithms. This paper examines the centralization vs. decentralization performance of MADRL formulations in I&M planning of multi-component systems. Towards this, we set up a comprehensive computational experimental program focused on k-out-of-n system configurations, a common and broadly applicable archetype of deteriorating engineering systems, to highlight the manifestations of MADRL strengths and pathologies when optimizing global returns under varying decentralization relaxations.
Integrating Dynamic Seismic Simulation in Architectural Design
A Parametric Approach for Tall Buildings with Outer Diagrids
Book chapter
(2025)
-
Pooyan Kazemi, Michela Turrin, Charalampos Andriotis, Alireza Entezami, Stefano Mariani, Aldo Ghisi
This paper presents a comprehensive computational workflow to generate, evaluate, and compare different outer diagrid designs for tall buildings, focusing on their structural efficiency under various load conditions, including seismic actions. Such diagrid systems not only enhance both structural resilience and architectural flexibility but also provide primary resistance to lateral loads, allowing for smaller internal core structures. Despite their advantages, widely used architectural design software, such as Grasshopper®, offer limited support for the seismic simulations essential for such complex structures. To address this limitation, a novel integration of the Alpaca4d plug-in, based on OpenSees software, is developed to facilitate seismic simulations for architects and engineers. The methodology generates a diverse dataset of tall buildings by varying critical design parameters, including the geometries of the top and bottom floors, floor heights, and vertical transformations such as tapering or twisting. From over 61,200 theoretical combinations, the Latin hypercube sampling technique is adopted to select 1000 models to ensure a representative diversity. Each model is subject to cross-section optimisation based on Eurocode load combinations, incorporating both permanent and live loads according to code specifications. Dynamic analyses are then performed under seven ground motions selected from the PEER NGA-West2 database, to capture structural responses under realistic conditions. The computational workflow generates key outputs, including inter-storey drifts, base reaction forces and moments, and acceleration distributions, recorded at critical nodes and elements to optimise data collection and computational efficiency. These outputs provide a detailed understanding of the seismic performance of each model and serve as the basis for further structural/materials optimization. By leveraging these results, the workflow will support in the future early decision-making in structural and architectural design, also in view of the progresses in the artificial intelligence field. Overall, this approach offers a streamlined and adaptable method for systematically evaluating and optimising the seismic design of tall buildings for architectural design.
...
This paper presents a comprehensive computational workflow to generate, evaluate, and compare different outer diagrid designs for tall buildings, focusing on their structural efficiency under various load conditions, including seismic actions. Such diagrid systems not only enhance both structural resilience and architectural flexibility but also provide primary resistance to lateral loads, allowing for smaller internal core structures. Despite their advantages, widely used architectural design software, such as Grasshopper®, offer limited support for the seismic simulations essential for such complex structures. To address this limitation, a novel integration of the Alpaca4d plug-in, based on OpenSees software, is developed to facilitate seismic simulations for architects and engineers. The methodology generates a diverse dataset of tall buildings by varying critical design parameters, including the geometries of the top and bottom floors, floor heights, and vertical transformations such as tapering or twisting. From over 61,200 theoretical combinations, the Latin hypercube sampling technique is adopted to select 1000 models to ensure a representative diversity. Each model is subject to cross-section optimisation based on Eurocode load combinations, incorporating both permanent and live loads according to code specifications. Dynamic analyses are then performed under seven ground motions selected from the PEER NGA-West2 database, to capture structural responses under realistic conditions. The computational workflow generates key outputs, including inter-storey drifts, base reaction forces and moments, and acceleration distributions, recorded at critical nodes and elements to optimise data collection and computational efficiency. These outputs provide a detailed understanding of the seismic performance of each model and serve as the basis for further structural/materials optimization. By leveraging these results, the workflow will support in the future early decision-making in structural and architectural design, also in view of the progresses in the artificial intelligence field. Overall, this approach offers a streamlined and adaptable method for systematically evaluating and optimising the seismic design of tall buildings for architectural design.
Structural systems must satisfy multiple performance and functionality requirements during their life cycle, withstanding safety-reducing degradation mechanisms and hazards. Intervention strategies must be planned accordingly to maintain structural integrity and minimize total life-cycle costs and risks, posing a complex optimization problem. Recent advances in multi-agent deep reinforcement learning (DRL) in conjunction with partially observable Markov Decision Processes (POMDPs) have shown great potential for determining optimal structural integrity management policies for systems with large state and action spaces compared to traditional decision practices. This paper tackles the maintenance optimization problem of aging bridges in seismic-prone areas, creating an updatable environment that embeds chloride-induced corrosion and state-dependent seismic fragility throughout the bridge life-cycle. The evolution of the environment is captured by a dynamic Bayesian network, and it is further integrated with decentralized multi-agent DRL algorithms to identify near-optimal lifecycle decisions under risk constraints. Results on a multi-component bridge system show the suitability of the developed framework for minimizing expected life-cycle costs, and for providing detailed and adaptive policies that significantly outperform traditional condition- and time-based maintenance plans.
...
Structural systems must satisfy multiple performance and functionality requirements during their life cycle, withstanding safety-reducing degradation mechanisms and hazards. Intervention strategies must be planned accordingly to maintain structural integrity and minimize total life-cycle costs and risks, posing a complex optimization problem. Recent advances in multi-agent deep reinforcement learning (DRL) in conjunction with partially observable Markov Decision Processes (POMDPs) have shown great potential for determining optimal structural integrity management policies for systems with large state and action spaces compared to traditional decision practices. This paper tackles the maintenance optimization problem of aging bridges in seismic-prone areas, creating an updatable environment that embeds chloride-induced corrosion and state-dependent seismic fragility throughout the bridge life-cycle. The evolution of the environment is captured by a dynamic Bayesian network, and it is further integrated with decentralized multi-agent DRL algorithms to identify near-optimal lifecycle decisions under risk constraints. Results on a multi-component bridge system show the suitability of the developed framework for minimizing expected life-cycle costs, and for providing detailed and adaptive policies that significantly outperform traditional condition- and time-based maintenance plans.
Multi-criteria design methods in façade engineering
State-of-the-art and future trends
Façade engineering is facing an era of extraordinary challenge to meet the surge in demand for buildings that are environmentally sustainable and enhance occupant wellbeing. Facades, also known as building envelopes, play a major role in the resource-efficiency of buildings and the quality of its indoor environment. Consequently, the development of effective design approaches is crucial for generating appropriate façade solutions. Façade design is complex and multi-disciplinary involving several and oftentimes conflicting performance criteria. Systematic and holistic design procedures are, therefore, required to achieve optimal trade-offs. Over the last decades, researchers in this field have used computational tools and power to address this challenging problem within the context of multi-criteria design approaches. This paper reviews the existing research in this field, and presents the state-of-the-art review from simple to advanced decision-making procedures currently used at the early design stages, where decisions have a disproportionally large impact on the façade performance. The paper provides a complete description of the design variables and objectives typically involved. Alternative multi-criteria design methodologies regarding discrete decisions and automated optimization are reviewed, each with salient pros/cons, and overall conclusions are drawn. Finally, the paper discusses ongoing trends and research needs, namely, the development of uncertainty-based procedures to enable more informed decision-making; the inclusion of structural/seismic safety considerations in the design process to achieve higher socio-economic benefits; the integration of smart building information modeling and processing technologies to facilitate smarter design decisions; and the adoption of integrated design approaches to promote climate-adaptive solutions that enhance resilience.
...
Façade engineering is facing an era of extraordinary challenge to meet the surge in demand for buildings that are environmentally sustainable and enhance occupant wellbeing. Facades, also known as building envelopes, play a major role in the resource-efficiency of buildings and the quality of its indoor environment. Consequently, the development of effective design approaches is crucial for generating appropriate façade solutions. Façade design is complex and multi-disciplinary involving several and oftentimes conflicting performance criteria. Systematic and holistic design procedures are, therefore, required to achieve optimal trade-offs. Over the last decades, researchers in this field have used computational tools and power to address this challenging problem within the context of multi-criteria design approaches. This paper reviews the existing research in this field, and presents the state-of-the-art review from simple to advanced decision-making procedures currently used at the early design stages, where decisions have a disproportionally large impact on the façade performance. The paper provides a complete description of the design variables and objectives typically involved. Alternative multi-criteria design methodologies regarding discrete decisions and automated optimization are reviewed, each with salient pros/cons, and overall conclusions are drawn. Finally, the paper discusses ongoing trends and research needs, namely, the development of uncertainty-based procedures to enable more informed decision-making; the inclusion of structural/seismic safety considerations in the design process to achieve higher socio-economic benefits; the integration of smart building information modeling and processing technologies to facilitate smarter design decisions; and the adoption of integrated design approaches to promote climate-adaptive solutions that enhance resilience.
It is now within reach to use generative artificial intelligence (AI) to autonomously generate full building geometries. However, existing literature utilizing 3D data has focused to a limited degree on architecture and engineering disciplines. A critical first step to expanding the use of generative deep learning models in generative design research is making training data available. This study investigates 3D building model data characteristics that make it suitable for generative AI applications. Key data set attributes are identified through a systematic review of the object-containing datasets currently used to train state-of-the-art 3D GANs. These requirements are then compared to attributes of existing available building datasets. This comparison shows that publicly available data sets of 3D building models lack essential characteristics for generative deep learning. Features that make these building models inadequate for the task include but are not limited to, their mesh formats, low resolution and levels of detail, and inclusion of irrelevant geometry. To achieve the desired properties in this work, necessary transformations of the data are incorporated into a tailored preprocessing pipeline. The pipeline is applied to an existing dataset that contains 3D models of single-family homes. The transformed dataset is tested within state-of-the-art GAN models to assess training performance and document future data requirements for applying deep generative design to buildings. Our experiments show promise for the impact that architectural datasets can make on deep learning applications within the discipline. It also highlights the need for additional 3D building model data to increase the diversity and robustness of new designs.
...
It is now within reach to use generative artificial intelligence (AI) to autonomously generate full building geometries. However, existing literature utilizing 3D data has focused to a limited degree on architecture and engineering disciplines. A critical first step to expanding the use of generative deep learning models in generative design research is making training data available. This study investigates 3D building model data characteristics that make it suitable for generative AI applications. Key data set attributes are identified through a systematic review of the object-containing datasets currently used to train state-of-the-art 3D GANs. These requirements are then compared to attributes of existing available building datasets. This comparison shows that publicly available data sets of 3D building models lack essential characteristics for generative deep learning. Features that make these building models inadequate for the task include but are not limited to, their mesh formats, low resolution and levels of detail, and inclusion of irrelevant geometry. To achieve the desired properties in this work, necessary transformations of the data are incorporated into a tailored preprocessing pipeline. The pipeline is applied to an existing dataset that contains 3D models of single-family homes. The transformed dataset is tested within state-of-the-art GAN models to assess training performance and document future data requirements for applying deep generative design to buildings. Our experiments show promise for the impact that architectural datasets can make on deep learning applications within the discipline. It also highlights the need for additional 3D building model data to increase the diversity and robustness of new designs.
Effective transportation network management systems should consider safety and sustainability objectives. Existing research on large-scale transportation network management often employs the assumption that bridges can be considered individually under these objectives. However, this simplification misses accurate system-level representations, induced by multiple components, network topology, and global maintenance actions. To address these limitations, this paper presents a deep reinforcement learning (DRL) framework that draws inspiration from biological learning behaviors to determine optimal life-cycle management policies. It incorporates synergetic branches and hierarchical rewards, factorizing the action space and, thereby, diminishing system complexity from exponential to linear with respect to the number of bridges. Extensive experiments based on a realistic case study demonstrate that the proposed method outperforms expert maintenance strategies and state-of-the-art decision-making methods. Overall, the proposed DRL framework can assist engineers by offering adaptive solutions to maintenance planning. It also provides solutions that address large action spaces within complex systems.
...
Effective transportation network management systems should consider safety and sustainability objectives. Existing research on large-scale transportation network management often employs the assumption that bridges can be considered individually under these objectives. However, this simplification misses accurate system-level representations, induced by multiple components, network topology, and global maintenance actions. To address these limitations, this paper presents a deep reinforcement learning (DRL) framework that draws inspiration from biological learning behaviors to determine optimal life-cycle management policies. It incorporates synergetic branches and hierarchical rewards, factorizing the action space and, thereby, diminishing system complexity from exponential to linear with respect to the number of bridges. Extensive experiments based on a realistic case study demonstrate that the proposed method outperforms expert maintenance strategies and state-of-the-art decision-making methods. Overall, the proposed DRL framework can assist engineers by offering adaptive solutions to maintenance planning. It also provides solutions that address large action spaces within complex systems.
Generative Artificial Intelligence (AI) promises to make a vast impact across disciplines, including transforming the architectural design process by autonomously generating full building geometries. One form of generative deep learning that has been used to create 2D and 3D representations of objects is Generative Adversarial Networks (GANs). Existing literature, however, has limited applications that utilize 3D data for building geometry generation, with previous studies focused on low-scale 3D geometries suitable for objects such as chairs or cars. This paper develops a new GAN architecture to produce high-resolution feasible building geometry. The training dataset used is a selection of 3D models of single-family homes from an existing database, pre-processed for the specific application. State-of-the-art GAN models are initially tested to establish baseline performance and applicability potential. Then, a systematic study is performed to identify the structure and hyperparameters necessary to successfully fit a GAN to this design task. The successful architecture, named 3DBuildingGAN, uses a combination of Wasserstein loss with gradient penalty, leaky rectified linear units for neuron activation in the generator and the critic, and the root mean squared propagation optimizer with a fixed learning rate. The proposed model generates outputs similar in size, shape, and proportion to the training data with minimal noise in the output. Evaluation of memorization properties indicates open research directions, such as incorporating memorization rejection and training on larger data sets. Finally, the study reflects on how AI algorithms can reshape creativity through data-driven design solutions.
...
Generative Artificial Intelligence (AI) promises to make a vast impact across disciplines, including transforming the architectural design process by autonomously generating full building geometries. One form of generative deep learning that has been used to create 2D and 3D representations of objects is Generative Adversarial Networks (GANs). Existing literature, however, has limited applications that utilize 3D data for building geometry generation, with previous studies focused on low-scale 3D geometries suitable for objects such as chairs or cars. This paper develops a new GAN architecture to produce high-resolution feasible building geometry. The training dataset used is a selection of 3D models of single-family homes from an existing database, pre-processed for the specific application. State-of-the-art GAN models are initially tested to establish baseline performance and applicability potential. Then, a systematic study is performed to identify the structure and hyperparameters necessary to successfully fit a GAN to this design task. The successful architecture, named 3DBuildingGAN, uses a combination of Wasserstein loss with gradient penalty, leaky rectified linear units for neuron activation in the generator and the critic, and the root mean squared propagation optimizer with a fixed learning rate. The proposed model generates outputs similar in size, shape, and proportion to the training data with minimal noise in the output. Evaluation of memorization properties indicates open research directions, such as incorporating memorization rejection and training on larger data sets. Finally, the study reflects on how AI algorithms can reshape creativity through data-driven design solutions.
Assessing life-cycle seismic safety of aging reinforced concrete bridges is a challenging engineering task. Deterioration phenomena reduce structural capacity, exacerbating poor design choices that are typical of old bridges, while also being characterized by major uncertainties. Management of engineering systems in highly uncertain environments can be efficiently addressed through Markov decision processes, which rely on dynamic Bayesian networks to model the deteriorating system’s life-cycle. However, there is still a gap in developing virtual environments that can seamlessly fit in such advanced algorithmic decision-making frameworks, especially under life-cycle seismic behavior considerations. In this study, we develop a dynamic Bayesian network capable of incorporating disparate uncertainties related to chloride-induced corrosion and seismic action, aiming at providing fragility curves over the bridge service life. The framework is applied to a prototype bridge encapsulating key risk-prone features. Using a multi-component approach, the developed network provides valuable insights into the fragility evaluation of both the system and individual components. Markovian transitions among component deterioration states are computed by combining corrosion initiation and propagation models with non-stationary Gamma processes. Subsequently, state-dependent fragilities are obtained through probabilistic seismic assessment based on non-linear dynamic analyses and multinomial logistic regression. Results show that the approach sheds light on the risk interplay mechanisms between components and the system, and on how different corrosion scenarios affect the system fragility. Discussion is finally provided on how these risk considerations can be interpreted for decision-making, allowing for better repair and retrofit strategies.
...
Assessing life-cycle seismic safety of aging reinforced concrete bridges is a challenging engineering task. Deterioration phenomena reduce structural capacity, exacerbating poor design choices that are typical of old bridges, while also being characterized by major uncertainties. Management of engineering systems in highly uncertain environments can be efficiently addressed through Markov decision processes, which rely on dynamic Bayesian networks to model the deteriorating system’s life-cycle. However, there is still a gap in developing virtual environments that can seamlessly fit in such advanced algorithmic decision-making frameworks, especially under life-cycle seismic behavior considerations. In this study, we develop a dynamic Bayesian network capable of incorporating disparate uncertainties related to chloride-induced corrosion and seismic action, aiming at providing fragility curves over the bridge service life. The framework is applied to a prototype bridge encapsulating key risk-prone features. Using a multi-component approach, the developed network provides valuable insights into the fragility evaluation of both the system and individual components. Markovian transitions among component deterioration states are computed by combining corrosion initiation and propagation models with non-stationary Gamma processes. Subsequently, state-dependent fragilities are obtained through probabilistic seismic assessment based on non-linear dynamic analyses and multinomial logistic regression. Results show that the approach sheds light on the risk interplay mechanisms between components and the system, and on how different corrosion scenarios affect the system fragility. Discussion is finally provided on how these risk considerations can be interpreted for decision-making, allowing for better repair and retrofit strategies.