ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part A: Civil Engineering

Forum papers are thought-provoking opinion pieces or essays founded in fact, sometimes containing speculation, on a civil engineering topic of general interest and relevance to the readership of the journal. The views expressed in this Forum article do not necessarily reflect the views of ASCE or the Editorial Board of the journal.

Recent research has emphasized the importance of social vulnerability in assessing overall community vulnerability. This is supported by the observation that socially marginalized groups tend to experience the worst disaster outcomes because they lack the advantages that benefit more dominant groups within stratified social systems. The importance of social marginalization is also reflected in recent discussions of resilience. Indicators of social stratification are used to measure social vulnerability and resilience in the context of disasters. These are often aggregated into composite indexes, which are useful in providing a quantitative metric that can be used to identify community priorities and inform decision-making. However, a major shortcoming in the use of the indexes has been the lack of validation with observable disaster outcomes. The purpose of this paper is to provide a link between evacuation behavior and these indexes through a review and analysis of prominent indexes measuring social vulnerability and resilience. The analysis presents a comparison of prominent indexes of community resilience, social vulnerability, and social capital with recognized factors that influence evacuation behavior and the different steps of evacuation modeling. The analysis revealed that no single index provided a comprehensive match for the factors relevant to evacuation behavior, although SoVI was found to provide the most overlap. This review has identified the potential to incorporate more social, behavioral, and human-centric indicators into evacuation modeling to improve the prediction of trip generation, trip distribution, and trip assignment.

Geotechnical analysis and design are usually based on traditional methods and empirical relationships. However, various challenges, such as the inability to model all parameters involved and the uncertainty of model input, make these methods ineffective. Therefore, probabilistic methods recently began to be used to estimate the probability of slope failure. In this study, slope stability is presented as an optimization problem and the simulated annealing (SA) method is used to identify the slip surface with the minimum safety factor. The SA algorithm is then integrated with the Monte Carlo sampling method to calculate the probability of slope failure. Next, the groundwater level is added to the problem formulation as a random variable to model the uncertainty for cases where the groundwater table is not accurately determined. The results of the analysis show that for slopes with low safety factors, the uncertainty of the groundwater table worsens the condition and increases the failure probability. Moreover, the parametric study shows that the insufficient understanding of groundwater-level distribution and the assumption of uniform distribution further increases the failure probability. The results of this study are presented in the form of a probability curve against the groundwater table. This presentation has practical applications in real projects.

During the 2003 Bachu earthquake in China, the liquefaction phenomenon was widely observed. However, the standard penetration test–based method suggested in the national seismic design code in China significantly underestimates the liquefaction potential in Bachu. In this paper, a model bias factor is used to represent the effect of model uncertainty. The variability of the model bias factor associated with the national seismic design code method is investigated using 160 case histories collected from 7 regions in China. It is found that the statistics of the model bias factor vary from one region to another, indicating that the national seismic design code method is associated with different levels of accuracy when applied in different regions. Using the knowledge about the model bias factor learned from the seven regions as prior information, the distribution of the model bias factor is then updated with the region-specific data from Bachu based on Bayes’ theorem. The updated model bias factor is verified through liquefaction potential, and the expected number of liquefaction cases is consistent with the observed number of liquefaction cases. Base on the calibrated model bias factor in Bachu, a new liquefaction assessment criterion is then suggested for Bachu to achieve the target reliability level.

The reliability simulation of ultrasonic nondestructive testing (NDT) based on the COMSOL Multiphysics simulation software is used to research the influence factors of the angle between the sound beam axis and the crack orientation, and the crack burial depth. First, the COMSOL Multiphysics simulation models of ultrasonic NDT are established to obtain the ultrasonic detection amplitude of cracks with the same size at different angles or crack burial depths. According to different crack lengths, the simulation of ultrasonic detection amplitude under different angles or crack burial depths are carried out, and the probability of detection (POD) curve forms, such as Weibull and Lockheed function, refereed so as to determine the optimal POD curve form considering the influence of angle or crack burial depth. Finally, the minimum detectable crack size with 90% detection probability considering the influence of angle or crack burial depth can be inversed, thus a simulation method of ultrasonic NDT reliability based on COMSOL Multiphysics is formed. The simulation results show that the POD curves and minimum detectable crack size at different angles or crack burial depths in a specific structure are effectively obtained through the proposed method. It can provide some suggestion for the formulation of an ultrasonic testing scheme affected by angle and crack burial depth in engineering practice.

The widely used First Order Reliability Method (FORM) is efficient for uncertainty quantification and safety assessment in Uncertainty-based Multidisciplinary Design Optimization (UBMDO). However, the Rosenblatt transformation is necessary for FORM. This transformation process can significantly increase the degree of nonlinearity of the computing process, especially in a multidisciplinary coupled computing environment. To deal with this case, the First Order Saddlepoint Approximation (FOSPA) strategy was utilized in this study. The method of UBMDO using FOSPA was proposed to enhance the accuracy of reliability evaluation in the calculation process. Moreover, the decoupling strategy of UBMDO was introduced here for higher computing efficiency. Two examples were utilized to illustrate the application of the proposed method.

This study aims to present a probabilistic framework for inspection, maintenance, and repair planning of jacket platforms based on the Bayesian probabilistic networks. To this end, considering damage and deterioration mechanisms in such a platform, some practical models are proposed to predict the extent of the damage in each element. Next, models are presented for prediction of the loads on the platform. Consequently, the condition of the structure is evaluated under the effect of damages as well as the applied loads, and the failure probabilities of structural elements are computed given their internal forces and the residual capacity of the members at the time of planning. By applying this framework, which is formulated by using Bayesian networks, it would be possible to update failure probability based on the latest information received from the inspection results or structural monitoring. Finally, the application of the proposed probabilistic Bayesian network is presented to provide an inspection plan for a sample platform and demonstrates how to incorporate the information obtained from the inspections or monitoring.

The geotechnical properties of soil deposit and the variability associated with their probable distributions have a profound impact on the seismic response of a site. In the present work, the influences of soil profile characterizations corresponding to the shear wave velocity ($V_s$), density, and material degradation using various probabilistic distributions are investigated. A stochastic process is introduced for solving the spatial variability in soil deposit via Monte Carlo simulations. The results are validated with those obtained from the reference solution using the Strata program version 0.5.5. Additionally, sensitivity analysis is conducted to investigate the effect of the random input variables in the soil profile. The analysis concludes that the consideration of probabilistic distributions of the geotechnical parameters plays a significant role in evaluating the reliability of a site. The variability in material degradation has a greater impact than the unit weight on site response. Furthermore, comparatively the variability in $V_s$ for both the Toro model and log-normal distribution is identical for periods greater than 1.0 s, while in the range of lower periods, the former is lower than the latter with maximum reductions of 11.14% and 20.86% in surface response spectra and amplification factor, respectively.

This paper explores the problem of data collaboration in health care, which is the one of the critical infrastructure sectors designated by the Department of Home Security. Limitations to data sharing in health care obstruct the development of a new generation of medical technology powered by artificial intelligence (AI). Collaborative machine learning helps to overcome these limitations through training models on distributed data sets without data sharing. Among other approaches to collaborative machine learning, federated learning in recent years has demonstrated multiple advantages. However, it had been developed and tested in a highly distributed data environment, which is different from the typical cases of health care data collaboration. The objective of this paper is to validate the known advantages of federated learning and to assess possible risks in a small multiparty setting. The experiments show that federated learning can be successfully applied in a multiparty collaboration setting. However, with a small number of parties, it becomes easier to overfit to each local data so that the averaging steps have to occur more frequently. In addition, for the first time, the risks of a membership inference attack were assessed for different methods of collaborative machine learning.

Corrosion and hurricanes pose substantial risk to coastal bridges. This risk is compounded by future climatic and socioeconomic changes due to the increasing temperature and humidity, rising sea level, changing frequency and intensity of hurricanes, and amplifying exposure and consequences to adverse events. These future conditions, however, involve deep uncertainty, making life-cycle risk assessment and management extremely difficult. To address this challenge, a novel approach is proposed to identify, among various factors related to climatic and socioeconomic changes, the vulnerability of coastal bridges that could lead to unacceptable life-cycle risk. The considered factors include temperature increase, sea level rise, hurricane frequency and intensity, and socioeconomic growth. The acceptable risk is derived from the threshold reliability index of existing structures and the associated failure consequences. The entire space of relevant factors is divided into acceptable and unacceptable categories using Latin hypercube sampling and life-cycle risk assessment. Advanced analysis tools, including feature scoring, scenario discovery, and dimensional stacking, are then used to pinpoint the driving factors of bridge vulnerability and their thresholds that are prone to create unacceptable risk. The proposed approach is applied to a representative bridge in a coastal setting. The implication of the approach on robust decision-making about climate adaptation is also discussed.

Hydropower asset managers use whole-life cost models to better manage risk costs and asset replacement strategies for their hydroelectric powertrain and balance of plant assets. Whole-life cost models can be simplified to four main input variables: Weibull scale ($\alpha$) and shape ($\beta$) parameters that inform the probability of failure; a planning horizon $\delta$, indicating the frequency of the decision-making opportunity; and a cost ratio (CR) that compares unplanned outage costs with planned outage costs. Both the probability of failure models and the cost estimates suffer from high uncertainty and a lack of information regarding confidence bounds. The sensitivity of model results to the uncertainty inherent in the CRs is highly important for informing facility and fleet-wide decision making. Weibull shape and scale parameters were modeled at $\pm 10\%$ errors for a wide range of CRs. This sensitivity study reveals that (1) a $\pm 10\%$ error in the shape parameter imparts a 0.04% to 5.89% change in the optimized annualized cost (OC) over the range of CRs; (2) at CRs above 1.4, the 4-year ($\pm 10\%$) error in the scale parameter causes a change in optimal point (OP) of replacement of 2 to 4 years; and (3) as CRs increase, the impact of errors on OP first measured as 7 years decreases to less than 1 year by the time CRs reach 2.4.

In North America and Europe, there is a significant number of centenary metallic bridges that require maintenance and strengthening operations. These structures need to be adapted to increasing traffic intensities, and their structural integrity should be constantly evaluated. Fatigue damages were not considered in the original design of old metallic bridges, and riveted connections are one of the most frequent sources of fatigue damages on these structures. This paper intends to be a contribution for the reliable assessment of the fatigue behavior of riveted connections by proposing S-N curves for these structural details. Experimental fatigue data is analyzed from multiple bridges across Europe, and different statistical methods were implemented. A normal distribution function was implemented following the ISO 12107 standard, and the results were compared with the implementation of the two-parameter Weibull distribution function. Different estimation methodologies were implemented to determine the parameters of the Weibull distribution. S-N curves obtained by the statistical analysis were then compared with design recommendations from North American and European standards.

This paper introduces the assessment method for ergonomic interaction design reliability of the engineering software operation flow in the field of civil engineering. On basic structure of human reliability assessment index system concerning engineering software operation flow, the approaches of multilevel attribute data analysis were expounded by improved entropy, fault tree analysis (FTA), and Delphi to improve user experience and increase engineering software operation efficiency. This study analyzed the key technologies used to evaluate and improve human reliability of ergonomic interaction design, which is significant for enhancing the application quality of engineering software. Guided by functional requirements, every detail complies with the cognition and feelings of the operator. By considering conscious planning, and design with clear goals, the characteristics of not respecting habits, unfriendly interface, no significant themes, user cognitive load, and the operation error rate can be reduced. User interviews, observation, questionnaire surveys, and other methods were adopted to conduct user research on engineering software interfaces in actual operation. Different interface information architectures were studied to explore user behavior habits and preferences when using engineering software interfaces. The principle of user behavior feedback in interface design should be arranged by multichannel input information so that users can locate and detect the current operation behavior more effectively, improving the user experience and enhancing ergonomic interaction design reliability. Existing problems in the strategic layer of the system were raised through user experience, and the interface design of engineering software was assessed based on nine elements of ergonomic interaction design reliability. Eye-tracker experiments were used to verify the validity and rationality of human reliability assessment based on entropy–FTA–Delphi. The human reliability assessment results of ergonomic interaction design reliability can guide the interface design of engineering software.

The infrastructure construction industry has always been plagued with significant uncertainty. It is therefore crucial for decision makers (DMs) to ascertain suitable investment timing, and there is still a need for a paradigm to assist them. Real options analysis (ROA) is a project evaluation method that allows DMs to incorporate volatility into their analyses. This study uses ROA to create a decision framework to determine the optimal infrastructure investment decision timing. Two types of ROA, the traditional option and the rainbow option, are used in the analysis to examine the impacts of these methods on timing. The paper’s case study involves construction of a gas station in the Bangkok metropolitan region and takes the price of and demand for gas as the sources of uncertainty. The paper also carries out a loss value analysis. The results suggest that the DM should postpone the decision further when volatility increases due to a “wait-and-see” strategy. We introduce loss value to compensate for depreciation in the value of the project due to deferment. The results of the evaluation and the sensitivity analysis confirm that compared to nonconsideration of investment value, consideration can lead to an earlier optimal time for the investment.

The legal concept of foreseeability is an important element of proof in flood loss litigation against the US Army Corps of Engineers. Although hydrologic forecasting and engineering uncertainty address foreseeability, the civil engineering literature lacks discussion of how flood risk managers should consider the legal concept during the life cycle of projects. The 1928 Flood Control Act holds federal agencies immune from liability due to flood projects, so the litigation takes the alternate approach of takings per the Fifth Amendment of the US Constitution. Although foreseeability is a long-standing legal concept in determination of personal injury causation, flood risk scenarios involve many more players and complexities. If judges use concepts from simpler situations to judge complex flood cases, it might bias outcomes. A conceptual example synthesizes the main actions in five riverine flood cases against the Corps of Engineers that were alleged to be the proximate causes of flood damages. Foreseeability issues are discussed for hydrologic uncertainty, channel capacity, the condition of reservoirs and levees, changed project purposes, and land development. Based on these, current practices of the Corps of Engineers to address uncertainty and foreseeability in flood projects are reviewed.