Conservative Predictions Research Articles

Inspection data‐based prediction on fatigue crack of orthotropic steel deck using interpretable machine learning method

AbstractThe prediction of fatigue cracks on orthotropic steel decks is of great significance to the maintenance of bridges. However, fatigue cracks are affected by various uncertainties in reality, which encourages a data‐driven study for the sake of reliability and accuracy of predictions. Based on the crack inspection data from orthotropic steel decks on actual bridges in China, the feature engineering is conducted considering fatigue crack behaviors, and the machine learning models are trained and tested for predicting cracks, including XGBoost, random forest, and multiple decision trees. According to the receiver operating characteristic curves of the three models, the XGBoost model has the best performance, whereas the average AUC is about 0.75, limited by the insufficient data volume of positive samples. With the SHAP values of all features, the interpretation of the machine learning model is presented, indicating that the global effects, that is, the longitudinal position, the loading condition, and the bridge age, are always influential factors for fatigue cracks. The local features concerning the interactions between cracks have an effect on crack behaviors to a certain extent, but less important. Accordingly, the interpretable machine learning model can provide conservative predictions in a rather transparent way on this issue, which can benefit decision‐making in bridge designs, maintenance, and management.

Assessment of design models for aluminium alloy CHS beam-columns with stocky cross-sections

There is little information on the prediction accuracy of the international design standards for aluminium alloy circular hollow section (CHS) beam-columns. To bridge this gap, this paper presents a comprehensive numerical investigation on the performance of aluminium alloy CHS beam-columns with stocky cross-sections. A non-linear finite element (FE) model was developed and validated against experimental results. The validated FE model was employed to conduct an extensive parametric study of aluminium CHS members subjected to combined axial compression and bending over a range of lengths, cross-sections and applied eccentricities. The effect of these parameters was investigated for two material grades, 6061-T6 and 6082-T6, resulting in a total of 168 FE models. The numerically obtained capacities were used to assess the accuracy of the strength predictions of Eurocode 9 (EC9), the Aluminium Design Manual (AA) and the Direct Strength Method (DSM). On the basis of the results, EC9, with the exponent of the interaction formula, ψc, calculated as a function of the buckling reduction factor χ, was found to provide the most accurate strength predictions with a mean value equal to 1.01 for both examined material grades. The most conservative predictions were given by the DSM, whilst the most scattered predictions by the AA, with a Coefficient of Variation (COV) equal to 0.17. Finally, based on the DSM buckling strength and the plastic cross-sectional moment resistance, a new interaction curve is proposed for CHS beam-columns with stocky cross-sections.

Post-buckling damage tolerance of welded omega-stiffened thermoplastic panels with initial damage

Welded omega-stiffened panels made of thermoplastic carbon composite with initial damage in the conduction welded joint are analysed and tested to investigate the damage tolerance in post-buckling. Finite element analyses are performed, using the virtual crack closure technique to investigate skin-stringer separation for both the pristine welded joint and joints with initial damage. A sensitivity study is executed for the initial damage size and location with different geometrical imperfections. Four omega-stiffened panels are tested, of which three have initial damage consisting of a foil at the welded skin-stringer interface. During the test, digital image correlation is used to measure the panels’ deformation to determine the evolution of the buckling shape and the interaction with the initial damage. A high-speed camera is placed on the stringer side of the panel to capture the final failure. The panels fail in post-buckling when skin-stringer separation occurs, starting from the initial damage. The finite element analysis is able to predict the overall structural behaviour well, with conservative failure load predictions for panels with initial damage in the middle stringer. Although the initial buckling shape is predicted well, the buckling shape evolution at higher loads is difficult to predict.

Flexural behavior and serviceability of steel fiber-reinforced lightweight aggregate concrete beams reinforced with high-strength steel bars

Six beams reinforced with HRB600 bars and one reinforced with HRB400 bars fabricated using steel fiber-reinforced lightweight aggregate concrete (SFLWAC) were tested under a four-point bending load to investigate the load-carrying capacity and service performance, with different reinforcement ratios, clear span lengths, concrete strengths, and reinforcement strengths. The test results showed that cracks ran through lightweight aggregates and developed rapidly without an aggregate interlocking mechanism, and the specimens eventually failed in the mode that the compressive concrete was crushed following longitudinal bars yielded, with 3 ∼ 4 major cracks. The specimens satisfied the allowable deflection and crack width required in the codes, with a ductility coefficient exceeding 4.5, and 1/150 of the span length was recommended to be the maximum deflection at the serviceability limit state. Generally, the increasing reinforcement ratio significantly improved the flexural strength and serviceability but reduced the ductility, the linear stiffness decreased with the growing span length, and the SFLWAC strength mainly affected the cracking moment and ductility. Replacing HRB400 bars with equal strength of HRB600 bars led to a substantial drop in serviceability, while an equal area of HRB600 bars improved the ultimate moment by 38.7 %. The calculation results indicated that by neglecting the residual tensile strength, fracture characteristics, and bond properties of SFLWAC, the prediction equations in the Chinese, American, and European codes underestimated the ultimate moment and overestimated the crack width, and the Chinese code yielded the conservative deflection prediction. The analytical flexural models for HRB600-reinforced SFLWAC beams were derived by incorporating the effects of steel fibers, lightweight aggregate concrete, and bond stress, exhibiting accurate predictions.

Behavior of hybrid FRP-concrete-steel double-skin tubular member T-joints subjected to brace axial compression

Hybrid fiber reinforced polymer (FRP)-concrete-steel double-skin tubular members (DSTMs) are a novel form of structural members comprised of an outer FRP tube, an inner steel tube, and an annular concrete sandwiched between the two tubes. By leveraging the advantageous properties of each constituent material, DSTMs possess superior mechanical performance and exceptional durability, making them highly promising for applications in ocean structures. Despite extensive research highlighting the excellent performance of individual DSTMs such as beams and columns since their inception, investigations into the behavior of DSTM joints have been scarce. This critical knowledge gap presents a significant obstacle to the widespread practical application of DSTMs. To address this gap, the present study was proposed to investigate the behavior of circular DSTM T-joints subjected to brace axial compression. An experimental program was conducted to systematically examine the influence of key factors such as FRP/steel tube thickness, brace-to-chord diameter ratio, void ratio, and concrete strength on the performance of DSTM T-joints. The experimental results demonstrated that the DSTM T-joints exhibited a ductile behavior; FRP confinement was demonstrably effective in enhancing the joint bearing capacity; the brace-to-chord diameter ratio and void ratio were identified as two crucial factors influencing the joint bearing capacity. Finally, simple design equations capable of providing reasonably accurate and slightly conservative predictions for the bearing capacities of the test DSTM T-joint specimens were proposed.

Data-Driven Weather Forecasting and Climate Modeling from the Perspective of Development

Accurate and rapid weather forecasting and climate modeling are universal goals in human development. While Numerical Weather Prediction (NWP) remains the gold standard, it faces challenges like inherent atmospheric uncertainties and computational costs, especially in the post-Moore era. With the advent of deep learning, the field has been revolutionized through data-driven models. This paper reviews the key models and significant developments in data-driven weather forecasting and climate modeling. It provides an overview of these models, covering aspects such as dataset selection, model design, training process, computational acceleration, and prediction effectiveness. Data-driven models trained on reanalysis data can provide effective forecasts with an accuracy (ACC) greater than 0.6 for up to 15 days at a spatial resolution of 0.25°. These models outperform or match the most advanced NWP methods for 90% of variables, reducing forecast generation time from hours to seconds. Data-driven climate models can reliably simulate climate patterns for decades to 100 years, offering a magnitude of computational savings and competitive performance. Despite their advantages, data-driven methods have limitations, including poor interpretability, challenges in evaluating model uncertainty, and conservative predictions in extreme cases. Future research should focus on larger models, integrating more physical constraints, and enhancing evaluation methods.

Temporal and spatial variation and prediction of water yield and water conservation in the Bosten Lake Basin based on the PLUS-InVEST model

Temporal and spatial variation and prediction of water yield and water conservation in the Bosten Lake Basin based on the PLUS-InVEST model

Behavior of wire arc additively manufactured 316L austenitic stainless steel single shear bolted connections

This paper aims to investigate the behavior of single shear bolted connections made of wire arc additively manufactured 316L austenitic stainless steel. A set of 44 wire arc additive manufacturing (WAAM) 316L austenitic stainless steel single shear bolted connections were included with consideration of various bolt positions, surface conditions and loading orientations respective to the printing layer direction. The geometric dimensions of the WAAM austenitic stainless steel plates were measured with the assistance of non-contact 3D laser scanning prior to tensile testing. Monotonic tensile tests were carried out to investigate the load-deformation responses, failure patterns and resistances determined by both the deformation and strength criteria of the single shear bolted connections. The effects of geometric and printing parameters on the single shear bolted connections were analyzed. Due to the absence of codified design provisions for WAAM austenitic stainless steel bolted connections, the suitability of the existing design rules originally developed for traditionally manufactured carbon steel and stainless steel bolted connections was examined. Design resistances calculated by the Eurocode 3 (prEN 1993-1-8 and prEN 1993-1-4), the American Specification (ANSI/AISC 370-21) as well as the prevalent design recommendations proposed in existing literature were compared with the obtained experimental results. It is shown that the abovementioned design methods offer conservative predictions for the resistances of WAAM 316L austenitic stainless steel single shear bolted connections. Further study is needed to improve the accuracy of design resistance predictions.

Growth limits of psychrotrophic Bacillus cereus as a function of temperature, pH, water activity, and lactic or acetic acid

This work focuses on the effects of temperature, pH, water activity, and concentrations of acetic or lactic acid on the growth limits of psychrotrophic Bacillus cereus sensu lato (s.l.). A previously published growth boundary model, based on an ‘interaction term’, was extended by the integration of new environmental factors. Further development has been made by replacing, wherever possible, the single values for strain-dependent parameters by statistical distributions, making it possible to describe the intra-group variability in B. cereus s.l. behaviour. The parameters associated with organic acid (i.e., the Minimum Inhibitory Concentrations, MIC) were determined for one strain for lactic acid and three strains for acetic acid. The MICs estimated were close to previously published values for mesophilic reference group III strain F4810/72. The growth/ no growth interface for psychrotrophic B. cereus s.l. in absence of organic acid was defined by the lower growth limits obtained separately for “groups II and V” and “group VI”. The model predictions for the transition between the “no-growth only” and “possible growth” provide fail-safe predictions for ComBase and literature data (468 records). To investigate behaviour of psychrotrophic B. cereus s.l. under organic acid, growth/ no growth data were generated at 15 °C (simulating mild temperature abuse) for three B. cereus s.l. strains (one from group II and two from group VI) at different pH levels (between 4.8 and 6.2), water activities (between 0.974 and 0.997) and concentrations of acetic acid (up to 45 mM) or lactic acid (up to 100 mM). Each of the three strains was studied separately for a total of 312 experiments. The minimum pH levels required for growth increase in the presence of lactic or acetic acid, highlighting their inhibitory effects. These inhibitory effects are enhanced at the lowest water activity tested. Although, group VI strains were reported to be more affected by low aw, only small differences were observed between group II and group VI at aw 0.974. The developed model was found to provide conservative (i.e. fail-safe) predictions for the growth limits under acetic or lactic acid at 15 °C.

Critical distance-based fatigue evaluation of welded joints in steel bridge under bending loading

Welded joints are prone to fatigue cracking under bending loading in steel bridges. However, certain design curves for these joints primarily focus on tensile or bending-tension loads, potentially leading to inaccurate assessments. This study conducted bending fatigue tests on fillet welded joints (FWJs), evaluating the suitability of specification methods for assessing their fatigue performance and attempting to determine the strength curves required for fatigue assessment. Furthermore, it validated the feasibility of the traditional theory of critical distance (TCD) in predicting the fatigue life of FWJs under bending loading. On this basis, a TCD assessment method incorporating the welding residual stress (WRS) was proposed. The results indicate that the fatigue design curve of the specifications provides conservative predictions for the fatigue performance of FWJs under bending loading, and FAT112 from Eurocode3 may be suitable for fatigue assessment using hot-spot stress and nominal stress methods. Furthermore, predicting fatigue life for FWJs under bending loading with traditional TCD is feasible. Compared to traditional TCD, the proposed method significantly improves prediction accuracy, with a minimum scatter factor of just 1.5. The proposed TCD not only considers the WRS effect but also reduce the need for numerous fatigue tests to establish the design curve for a specific welded joint.

FBXO11 variants are associated with intellectual disability and variable clinical manifestation in Chinese affected individuals.

F-box protein 11 (FBXO11) is a member of F-Box protein family, which has recently been proved to be associated with intellectual developmental disorder with dysmorphic facies and behavioral abnormalities (IDDFBA, OMIM: 618089). In this study, 12 intellectual disability individuals from 5 Chinese ID families were collected, and whole exome sequencing (WES), sanger sequencing, and RNA sequencing (RNA-seq) were conducted. Almost all the affected individuals presented with mild to severe intellectual disability (12/12), global developmental delay (10/12), speech and language development delay (8/12) associated with a range of alternate features including increased body weight (7/12), short stature (6/12), seizures (3/12), reduced visual acuity (4/12), hypotonia (1/12), and auditory hallucinations and hallucinations (1/12). Distinguishingly, malformation was not observed in all the affected individuals. WES analysis showed 5 novel FBXO11 variants, which include an inframe deletion variant, a missense variant, two frameshift variants, and a partial deletion of FBXO11 (exon 22-23). RNA-seq indicated that exon 22-23 deletion of FBXO11 results in a new mRNA structure. Conservation and protein structure prediction demonstrated deleterious effect of these variants. The DEGs analysis revealed 148 differentially expressed genes shared among 6 affected individuals, which were mainly associated with genes of muscle and immune system. Our research is the first report of FBXO11-associated IDDFBA in Chinese individuals, which expands the genetic and clinical spectrum of this newly identified NDD/ID syndrome.

Effect of BFRP ties on the axial performance of RC columns reinforced with BFRP and GFRP rebars

The adoption of fiber-reinforced polymer (FRP) bars in lieu of traditional steel reinforcement for concrete structures has gained prominence. Nevertheless, prevailing international standards have not incorporated provisions to assess the significant contribution of FRP materials to the axial capacity and deformability of compression members. This study investigates the impact of Basalt (BFRP) ties on the axial performance of full-scale concrete columns reinforced with BFRP and Glass (GFRP) rebars, alongside traditional steel longitudinal rebars. Ten columns were subjected to concentric loading, with parameters including the type of longitudinal rebars (BFRP, GFRP, and steel), spacing of BFRP ties (180 mm, 120 mm, and 60 mm), and diameter of steel rebars (16 mm and 20 mm). Varied steel rebar diameters aimed to discern the influence of axial stiffness on column behavior. Results indicate an 18 % to 26 % higher axial load capacity in steel-reinforced concrete (RC) columns compared to FRP-RC columns. Both types of FRP rebars contributed approximately 12 % to the axial capacity. Reducing BFRP tie spacing from 180 mm to 60 mm significantly increased deformability by up to 250 % in FRP-RC columns and ductility by up to 14 % in steel-RC columns. The study highlights conservative predictions in existing design equations (ACI 440.11–22 and CSA S806-12) for the axial load capacity of BFRP and GFRP RC columns, underestimating by 14 % to 20 %, as they neglect the contribution of FRP rebars. Utilizing a concrete crushing strain of 3,000 µε and 3,500 µε as the ultimate FRP compressive strain produced predictions closest to the experimental axial load capacity.

Strain-Based damage model for ductile tearing simulation under combined tensile and shear modes

This paper proposes a strain-based damage model to simulate ductile tearing under combined tensile and shear modes. For tensile mode, the multiaxial fracture strain model depending only on the stress triaxiality is used, determined by analysing smooth bar tensile test and fracture toughness test data. For shear mode, a simple constant fracture strain model is used, determined by analysing smooth bar torsion test and torsion fracture test using a deep surface cracked cylinder specimen. The proposed damage model is applied to simulate circumferential surface cracked pipe made of SUS 304 under combined bending and torsion. Comparison with full-scale surface cracked pipe test data suggests that the use of the combined tensile and shear damage model with the linear interaction rule gives good and conservative prediction results.

Confinement-based design and behaviour of concrete-filled stiffened steel tubular square slender columns

This paper presents the details, analysis and results of an investigation into the compressive behaviour of concrete-filled stiffened steel tubular (CFSST) slender columns. These are members comprising an outer steel section with longitudinal stiffeners welded to the interior surface. A finite element (FE) model was developed and validated. An improvement in ultimate strength compared to conventional concrete-filled steel tubular column has been found due to the presence of stiffeners that reduce the width-to-thickness ratio of the steel tube, which in turn increases the effective cross-sectional area of the steel tube. Moreover, it was found that stiffeners increase the lateral confining pressure on the concrete. FE modelling was then employed to conduct a thorough parametric study to investigate the influence of a wide range of parameters. These included the slenderness ratio, concrete compressive strength, as well as the geometry and details of the stiffeners. Both medium-length and long CFSST columns, classified on the basis of slenderness ratio, were carefully studied and discussed, as well as their inherent differences. The strengths predicted from the FE simulations were compared with the design strengths obtained from various international design codes, and it was shown that the design expressions provide very conservative capacity predictions. Accordingly, a new three-stage confinement-based design model was proposed, which takes into account different confining mechanisms of columns based on their slenderness ratio. It is illustrated that this model provides a very accurate prediction of the square CFSST slender columns, and captures the key behavioural characteristics.

Variant in EZR leads to defects in lens development

ABSTRACT Background Congenital cataract is a common cause of blindness. Genetic factors always play important role. Material and Methods This study identified a novel missense variant (c.1412C>T (p.P471L)) in the EZR gene in a four-generation Chinese family with nuclear cataract by linkage analysis and whole-exome sequencing. A knockout study in zebrafish using transcription activator-like effector nucleases was carried out to gain insight into candidate gene function. Results Conservative and functional prediction suggests that the P-to-L substitution may impair the function of the human ezrin protein. Histology showed developmental delays in the ezrin-mutated zebrafish, manifesting as multilayered lens epithelial cells. Immunohistochemistry revealed abnormal proliferation patterns in mutant fish. Conclusions The study suggests that ezrin may be involved in the enucleation and differentiation of lens epithelial cells.

Sensitivity Analysis of Offshore Platform Structures Under Varying Scour Depths

Several offshore platforms operating in the Java Sea have reported experiencing scour at varying depths, raising concerns about the safety and integrity of these structures. Scouring, an erosion phenomenon that occurs around these offshore platform structures due to their presence, is one of the most common issues encountered. The presence of scour can have a significant impact on the safety of these structures. To comprehend the implications of scour on structural safety, sensitivity analysis proves to be an invaluable tool. Sensitivity analysis establishes a relationship between changes in the safety parameters of the structure, obtained through linear analysis, and the depth of scour. By investigating this connection, sensitivity curves can be generated, enabling a conservative prediction of alterations in the strength parameters of the structure due to scour. In this study, a four-legged jacket platform structure underwent linear analysis under storm and seismic conditions using the SACS software. The scour phenomenon was simulated by adjusting the mudline's elevation beneath the structure, modifying the pile coordinates, reducing the length of piles beneath the mudline, and accounting for soil characteristics at each scour depth in the model. The sensitivity analysis revealed that the safety factors of the upper structural components, connections, and piles decrease at varying rates corresponding to each component type as the scour depth of the platform increases. By implementing these sensitivity curves, engineers and operators can make informed decisions regarding the maintenance and retrofitting of offshore platform structures to ensure their ongoing safety and structural integrity in the face of scour-related challenges. This research provides valuable insights into the critical relationship between scour depth and structural safety, enhancing our ability to protect offshore operations in the Java Sea and similar environments.

Impact of Recycled Concrete and Brick Aggregates on the Flexural and Bond Performance of Reinforced Concrete

The construction industry strongly relies on concrete and clay bricks for various applications. The escalating demand for these materials, driven by rapid population growth, has led to resource depletion and increased construction and demolition waste (CDW). Recycling CDW into construction materials, particularly in the form of recycled concrete aggregates (RCAs) and recycled brick aggregates (RBAs), has emerged as a promising solution. This study deals with the structural performance of concrete incorporating RCAs and RBAs. The experimental program encompasses material characterization, concrete mix design, and several tests to assess density, compressive strength, bond behavior, and flexural properties. The results indicate that the replacement of fine natural aggregate (NA) with fine RCAs or RBAs has a negligible impact on density, while the partial replacement of coarse NAs with RAs yields modest reductions in compressive strength. Notably, the bond strength between steel rebar and concrete is influenced by the type and content of RA, with specimens containing RCAs exhibiting a higher bond strength than those with RBAs. Empirical models used to predict bond strength generally align with experimental results, with conservative predictions by some models, such as ACI 318, and overestimation by others, such as models proposed by AS-3600 and CEB-FIB. The flexural tests of beams highlight the variation in stiffness and load-bearing capacity with the proportion of NAs replaced by RAs. While beams with 50% NA replacement demonstrate comparable performance to control beams, those with 100% RA replacement exhibit lower cracking and yielding stiffness. Cracking patterns in beams with RAs differ from control beams, with RA-containing beams showing more cracks and an altered crack distribution. The findings underscore the feasibility of using recycled aggregates in construction, with partial NA replacement offering a balance between sustainable material usage and desired structural properties.

Robust Uncertainty Quantification Using Conformalised Monte Carlo Prediction

Deploying deep learning models in safety-critical applications remains a very challenging task, mandating the provision of assurances for the dependable operation of these models. Uncertainty quantification (UQ) methods estimate the model’s confidence per prediction, informing decision-making by considering the effect of randomness and model misspecification. Despite the advances of state-of-the-art UQ methods, they are computationally expensive or produce conservative prediction sets/intervals. We introduce MC-CP, a novel hybrid UQ method that combines a new adaptive Monte Carlo (MC) dropout method with conformal prediction (CP). MC-CP adaptively modulates the traditional MC dropout at runtime to save memory and computation resources, enabling predictions to be consumed by CP, yielding robust prediction sets/intervals. Throughout comprehensive experiments, we show that MC-CP delivers significant improvements over comparable UQ methods, like MC dropout, RAPS and CQR, both in classification and regression benchmarks. MC-CP can be easily added to existing models, making its deployment simple. The MC-CP code and replication package is available at https://github.com/team-daniel/MC-CP.

Conformal Prediction Regions for Time Series Using Linear Complementarity Programming

Conformal prediction is a statistical tool for producing prediction regions of machine learning models that are valid with high probability. However, applying conformal prediction to time series data leads to conservative prediction regions. In fact, to obtain prediction regions over T time steps with confidence 1--delta, previous works require that each individual prediction region is valid with confidence 1--delta/T. We propose an optimization-based method for reducing this conservatism to enable long horizon planning and verification when using learning-enabled time series predictors. Instead of considering prediction errors individually at each time step, we consider a parameterized prediction error over multiple time steps. By optimizing the parameters over an additional dataset, we find prediction regions that are not conservative. We show that this problem can be cast as a mixed integer linear complementarity program (MILCP), which we then relax into a linear complementarity program (LCP). Additionally, we prove that the relaxed LP has the same optimal cost as the original MILCP. Finally, we demonstrate the efficacy of our method on case studies using pedestrian trajectory predictors and F16 fighter jet altitude predictors.

Experimental and numerical investigations of austenitic stainless steel semi-oval hollow sections under combined compression and bending

Semi-oval hollow section is an innovative cross-section profile, including one semi-circular flange, one flat flange and two flat webs. While the semi-circular flange (exposed to fluid or wind) offers a low level of hydrodynamic or aerodynamic drag, the flat elements facilitate connections with other members. This paper presents experimental and numerical investigations into the local buckling behaviour and resistances of austenitic stainless steel semi-oval hollow sections under combined compression and bending. A testing programme was firstly conducted and included initial local geometric imperfection measurements and ten eccentric compression tests. In conjunction with the testing programme, a numerical modelling programme was performed, where finite element models were developed to validate against the test results and conduct parametric studies for expanding the test data pool over a wider range of cross-section dimensions and loading combinations. The obtained test and numerical data were used to evaluate the applicability of the relevant design interaction curves for austenitic stainless steel rectangular hollow sections, as prescribed in the European code and American specification, to austenitic stainless steel semi-oval hollow sections. On the basis of the evaluation results, the codified design interaction curves were found to provide conservative resistance predictions. Finally, an improved design interaction curve, with more accurate end points and suitable shape, was developed and led to more accurate and consistent resistance predictions for austenitic stainless steel semi-oval hollow sections under combined compression and bending.