ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part B: Mechanical Engineering

Inspection and Maintenance methods development have a pivotal role in preventing the uncertainty-induced risks in the oil and gas industry. A key aspect of inspection is evaluating the risk of equipment from the scheduled and monitored assessment in the dynamic system. This activity includes assessing the modification factor's probability of failure and calculating the equipment's remaining useful life (RUL). The traditional inspection model constitutes a partial solution to grouping the vast amount of real-data inspection and observations at equal intervals. This literature review aims to offer a comprehensive review concerning the benefit of machine learning in managing the risk while incorporating time-series forecasting studies and an overview of risk-based inspection methods (e.g., quantitative, semiquantitative, and qualitative). A literature review with a deductive approach is used to discuss the improvement of the clustering Gaussian mixture model to overcome the noncircular shape data that may show in the K-Means models. Machine learning classifiers such as Decision Trees, Logistic Regression, Support Vector Machines, K-nearest neighbors, and Random Forests were selected to provide a platform for risk assessment and give a promising prediction toward the actual condition and the severity level of equipment. This work approaches complementary tools and grows interest in embedded artificial intelligence in Risk Management systems and can be used as the basis of more robust guidance to organize complexity in handling inspection data, but further and future research is required. This article is available in the ASME Digital Collection at .

Wind turbines suffer from mass imbalance due to manufacturing, installation, and severe climatic conditions. Condition monitoring systems are essential to reduce costs in the wind energy sector. Many attempts were made to improve the detection of faults at an early stage to plan predictive maintenance strategies, but effective methods have not yet been developed. Artificial intelligence has a huge potential in the wind turbine industry. However, several shortcomings related to the datasets still need to be overcome. Thus, the research question developed for this paper was “Can data augmentation and fusion techniques enhance the mass imbalance diagnostics methods applied to wind turbines using deep learning algorithms?” The specific aims developed were: (i) to perform sensitivity analysis on classification based on how many samples/sample frequencies are required for stabilized results; (ii) to classify the imbalance levels using Gramian angular summation field and Gramian angular difference field and compare against data fusion; and (iii) to classify the imbalance levels using data fusion for augmented data. Convolutional neural network (CNN) techniques were employed to detect rotor mass imbalance for a multiclass problem using the estimated rotor speed as an input variable. A 1.5-MW turbine model was considered and a database was built using the software turbsim, fast, and simulink. The model was tested under different wind speeds and turbulence intensities. The data augmentation and fusion techniques used along with CNN techniques showed improvement in the classification and hence the diagnostics of wind turbines. This article is available in the ASME Digital Collection at .

The dynamic response of a Duffing system from self-induced resonance to system resonance is studied in this paper. From numerical simulation, it is found that the system response gradually transits from self-induced resonance to system resonance with the increase of the pulse amplitude of the signal. In order to describe this process, we define the quality factor of the system response. With the evolution from self-induced resonance to system resonance, the quality factor gradually increases from 0 to 1. Then, based on the evolution, a novel method is developed to evaluate the severity of rolling bearing early damage. The results show that the method can be used not only to describe the process of a rolling bearing from healthy to damaged, but also to evaluate the severity of the early damage of a rolling bearing. The quality factor is a key index to reflect the severity of a rolling bearing. In addition, the sensitivity of the quality factor is superior to other traditional indices formerly used in the early damage evaluation. The effective method gives a new way for rolling bearing early damage evaluation. This article is available in the ASME Digital Collection at .

This work focuses on in situ residual unbalance estimation of the dual rotor system with the implementation of active magnetic bearing (AMB) as a controller and exciter. The excessive vibration generated due to the presence of residual unbalances limits the operating speed of the system. The compact structure of the dual rotor system provides constraints to the conventional balancing procedure that requires manual addition of the trial unbalances for balancing. In order to overcome the difficulty in balancing of the dual rotor system, an identification algorithm based on the modified influence coefficient method (MICM) is developed for the simultaneous estimation of residual unbalances in both inner and outer rotors with the generation of virtual trial unbalances as magnetic excitation through AMB. The controlling action of AMB attenuates the vibrational response of the system within the required limit and allows the safe operation of the system in the presence of rotor faults and additional excitations. The vibrational responses of the system at the limited locations and the magnitude and phase of the virtual trial unbalances are only required in the MICM for the estimation of unbalances. To numerically illustrate the present methodology, the displacement response is obtained from the developed finite element model of the dual rotor system with discrete disk unbalances and randomly distributed shaft. The robustness of the algorithm in the estimation of residual unbalances is verified with the addition of a different percentage of measurement noises. After balancing, the dual rotor system is found to traverse its critical speed with a less vibrational response. This article is available in the ASME Digital Collection at .

The work reports the systematic approach to the study of artificial intelligence (AI) in addressing the complexity of inline inspection (ILI) data management to forecast the risk in natural gas pipelines. A recent conventional standard may not be sufficient to address the variation data of corrosion defects and inherent human subjectivity. Such methodology undermines the accuracy assessment confidence and is ineffective in reducing inspection costs. In this work, a combination of unsupervised and supervised machine learning and deep learning has profoundly accelerated the probability of failure (PoF) assessment and analysis. K-means clustering and Gaussian mixture models show direct relevance between the corrosion depth and corrosion rate, while the overlapping PoF value is scattered in three clusters. Logistic regression, support vector machine, k-nearest neighbors, and ensemble classifiers of AdaBoost, random forest, and gradient boosting are constructed using particular features, labels, and hyperparameters. The algorithm correctly predicted the score of PoF from 4790 instances and confirmed the 25% metal loss at a location of 13.399 m. The artificial neural network (ANN) is designed with various layers (input, hidden, and output) architecture. It is optimized using an activation function to predict that 74% of the pipeline's anomalies that classified at low-medium and medium-high risk. Furthermore, it provides a quick and precise prediction about the external defects at 13.1 m and requires the personnel to conduct wrapping composite. This work can be used as a standard guideline for risk assessment based on ILI and applies to industry and academia. This article is available in the ASME Digital Collection at .

Experimental toxicology studies for the purposes of setting occupational exposure limits for aerosols have drawbacks including excessive time and cost which could be overcome or limited by the development of computational approaches. A quantitative, analytical relationship between the characteristics of emerging nanomaterials and related in vivo toxicity can be utilized to better assist in the subsequent mitigation of exposure toxicity by design. Predictive toxicity models can be used to categorize and define exposure limitations for emerging nanomaterials. Model-based no-observed-adverse-effect-level (NOAEL) predictions were derived for toxicologically distinct nanomaterial clusters, referred to as model-predicted no observed adverse effect levels (MP-NOAELs). The lowest range of MP-NOAELs for the polymorphonuclear neutrophil (PMN) response observed by carbon nanotubes (CNTs) was found to be 21–35 μg/kg (cluster “A”), indicating that the CNT belonging to cluster A showed the earliest signs of adverse effects. Only 25% of the MP-NOAEL values for the CNTs can be quantitatively defined at present. The lowest observed MP-NOAEL range for the metal oxide nanoparticles was Cobalt oxide nanoparticles (cluster III) for the macrophage (MAC) response at 54–189 μg/kg. Nearly 50% of the derived MP-NOAEL values for the metal oxide nanoparticles can be quantitatively defined based on current data. A sensitivity analysis of the MP-NOAEL derivation highlighted the dependency of the process on the shape and type of the fitted dose-response model, its parameters, dose selection and spacing, and the sample size analyzed. This article is available in the ASME Digital Collection at .

Engineering decisions that have the greatest effect on worker and public safety occur early in the design process. During these decisions, engineers rely on their experience and intuition to estimate the severity and likelihood of undesired future events like failures, equipment damage, injuries, or environmental harm. These initial estimates can then form the basis of investment of limited project resources in mitigating those risks. Behavioral economics suggests that most people make significant and predictable errors when considering high consequence, low probability events. Yet, these biases have not previously been studied quantitatively in the context of engineering decisions. This paper describes results from a set of computer-based engineering assessment and decision experiments with undergraduate engineering students estimating, prioritizing, and making design decisions related to risk. The subjects included in this experiment overestimated the probability of failure, deviated significantly from anticipated risk management preferences, and displayed worsening biases with increasing system complexity. These preliminary results suggest that considerably more effort is needed to understand the characteristics and qualities of these biases in risk estimation and understand what kinds of interventions might best ameliorate these biases and enable engineers to more effectively identify and manage the risks of technology. This article is available in the ASME Digital Collection at .