Strength Prediction Model Research Articles

Development of a machine learning model for on-site evaluation of concrete compressive strength by SonReb

Coring or destructive testing is typically the default choice for the evaluation of concrete compressive strength in reinforced concrete (RC) structures. However, it can be impractical and/or not representative of all areas of interest in a structure. While various non-destructive test (NDT) methods can be correlated to concrete strength, the accuracy of any single NDT used for this purpose is generally low. The SonReb method, which combines ultrasonic pulse velocity readings and rebound number, has been shown to have improved accuracy over single test methods. However, empirical SonReb equations, calibrated to specific datasets using regression analysis, cannot necessarily be applied to concrete from other sources without introducing significant errors. This study presents a practical machine learning (ML) model for on-site concrete strength prediction. A large database was created from available literature along with new experimental test data. Three different ML models based on an adaptive neuro-fuzzy inference system (ANFIS) were developed along with a graphical user interface application to facilitate its use in the field. In addition to the ML models, linear and non-linear regression analyses were also conducted and compared with existing equations in the literature. The accuracy of each model was subsequently validated against core samples extracted from a reinforced concrete slab. The results show that the proposed ML model and non-linear regression provided the most reliable predictions of concrete strength of the validation specimen with a mean absolute error of less than 10 % compared with twelve core samples. The findings suggest that ML can be a potential tool to evaluate the in-place compressive strength of RC structures in combination with simple NDT methods.

Surface response regression and machine learning techniques to predict the characteristics of pervious concrete using non-destructive measurement: Ultrasonic pulse velocity and electrical resistivity

It is crucial to assess the characteristics of pervious concrete even post-construction. The quality monitoring of such a procedure is tricky in pervious concrete that it is typically avoided. As a potential means of enhancing the aforementioned quality control, the current study investigates the possibility of predicting characteristics of pervious concrete through response surface methodology and machine learning techniques using non-destructive test measurement (ultrasonic velocity and electrical resistivity). A total of 225 datasets from the experimental study were taken for this study. To recognize the best reliable model for predicting characteristics of pervious concrete, response surface methodology up to sixth order polynomial and five different machine learning techniques were used as statistical assessment tools. Using both ultrasonic pulse velocity and electrical resistivity as predictors for estimating porosity and compressive strength via response surface methodology, using a quadratic model for porosity prediction and a cubic model for compressive strength prediction are recommended. The machine learning models used in the research exhibited superior performance compared to the response surface methodology. Among the many machine learning models evaluated in this study, boosted decision tree regression model better predicted porosity (R2 = 0.92) and compressive strength (R2 = 0.92) of pervious concrete. Therefore, prediction models for the characteristics of pervious concrete are created using non-destructive measurement and machine learning techniques, which may ensure that the construction sector can utilize the offered models without any theoretical expertise.

Experimental study on macro–micro-structural properties and strength prediction model of wastewater concrete

This study is intended to investigate the mechanical, durability, and drying shrinkage properties of concrete mixtures incorporating wastewater from ready-mixed concrete plants. The microstructural attributes were investigated by X-ray diffraction (XRD), scanning electron microscope (SEM) imaging, energy dispersive spectrometry (EDS), and mercury intrusion porosimetry (MIP). Meanwhile, rough set theory (RST) was used to analyze the influence of mean pore radius, porosity, fractal dimensions, and pore size distribution (PSD) on the compressive strength of wastewater concrete. The results showed that the seven-day cube compressive strengths for C20, C40, and C60 concrete specimens increased when wastewater was utilized, with the highest rise in the strength properties being observed as 16.84% for C40 mixture. Nevertheless, the 28-day cube compressive strengths of C20, C40, and C60 concretes were reduced by 4.85%, 9.33%, and 18.63%, respectively. The durability results showed that wastewater mixing reduced the frost resistance of C40 and C60 concrete, but improved for C20 concrete. The microstructural investigation revealed that mixing concrete with wastewater densifies the early-age concrete microstructure and refines the pore size. Nevertheless, the compactness of concrete microstructure at the later stage was reduced, and the pores in concrete were enlarged due to the poor cohesiveness of wastewater particles. The porosity and PSD have the greatest influence on the compressive strength of wastewater concrete. A strength and pore structure model of wastewater concrete was also developed and validated by considering PSD and porosity as bivariate parameters. Highlights The addition of wastewater can improve the early compressive strength of concrete. The addition of wastewater reduces the carbonization resistance of concrete. A high-precision model of wastewater concrete strength was established.

Prediction model for sandstone strength based on Weibull distribution of micro-element strength and power-law distribution of crack length

It is of the great significance to construct the strength prediction model that considers the influence of the macro and micro defects in the rock in an overall manner to ensure the construction of smart mines and the safety of underground personnel and equipment. Based on the Weibull distribution of rock micro-element strength and the power-law distribution of crack length and the discrete element calculation model, the prediction model for strength of damaged rock specimens were established by numerical simulation and theoretical derivation respectively. Then the rationality of the model was verified by the calculated results. The results show that: 1) The weibull distribution of the micro-element strength and the power-law distribution of the crack length in the PFC2D calculation model are realized by programming, and the quantitative corresponding relationship between the macro, micro and micro damage in the rock and the corresponding distribution parameters is analyzed. 2) 400 sandstone specimens with macro, micro and micro damage are numerically established and loaded to achieve statistical analysis of the relationship between micro element strength, pre-crack information and rock strength. According to the simulation results, the four-dimensional space scatter points of uniaxial compressive strength of sandstone specimens are constructed, and the flow law of compressive strength of sandstone specimens under the influence of multiple damage parameters is obtained. 3) Combined with Mori Tanaka method and rock damage probability distribution theory, a 12 parameter rock strength prediction model is established. The model can both describe the influence of the micro-element strength and macroscopic crack information on the rock strength, and the model calculation results are highly consistent with the simulated results, with the correlation coefficient up to 0.991.

Soft Computing for Comprehensive Concrete Strength Prediction – A Comparative Study

The evolution of concrete strength prediction methodologies has transitioned from empirical formulas based on experimental data to contemporary soft computing approaches. Initially, the concrete mix design was reliant on simple relationships between concrete mix proportions and compressive strength; later, the early techniques evolved to include statistical models incorporating material properties, curing conditions, and environmental variables. The advent of computational tools and artificial intelligence marked a paradigm shift, with accurate concrete strength prediction crucial for influencing structural integrity, safety, and cost-effectiveness in construction. The article explores empirical and analytical concrete strength prediction models before reviewing the application of soft computing approaches such as fuzzy logic, genetic algorithms, and neural networks. The integration of these models and hybrid approaches is discussed in this research study by highlighting their effectiveness in handling complex relationships within concrete mix parameters. A comparative analysis of various soft computing methods applied to structural and non-structural elements is carried out in this study to demonstrate their diverse applications and advantages in optimizing concrete mix designs, enhancing structural performance, and contributing to cost and time efficiency in construction processes.

Data-driven prediction and optimization of axial compressive strength for FRP-reinforced CFST columns using synthetic data augmentation

Fiber-reinforced polymer (FRP) sheets can be used as additional confinement to improve the load-bearing capacity and durability of concrete-filled steel tubular (CFST) columns. This study presents a data-driven approach aimed at accurately predicting the strength and optimizing the design of FRP-CFST columns under axial compressive loading. To overcome the limited database size (only 287 samples), a synthetic data augmentation technique was developed with recourse to the tabular generative adversarial networks (TGAN). Four machine learning (ML) models were rigorously trained using two distinct schemes “Trained with synthetic - Tested with real” (TSTR) and “Trained with real - Tested with real” (TRTR) to seek out the best strength prediction model. Then the winner model was integrated with the metaheuristic optimization algorithm—nondominated sorting genetic algorithm II (NSGA-II), to optimize both the axial strength and material cost of FRP-CFST columns. This fusion enabled a dual-objective optimization via generating the Pareto front, providing designers with a spectrum of optimal design solutions. It turned out that the categorical gradient boosting model trained with the TSTR mode (CATB-Syn) showed superior performance than the other ML models. Its accuracy also surpassed that of existing design equations. The combination of CATB-Syn with NSGA-II provided a reliable Pareto front, and the interpretability of CATB-Syn was also verified. Thus, the proposed framework opens a new avenue for the efficient design of such double-jacketed columns.

Analysis of pore structure characteristics and strength prediction model of coarse-grained soil based on fractal theory

Analysis of pore structure characteristics and strength prediction model of coarse-grained soil based on fractal theory

Experimental study of macroscopic and microscopic properties of long-age hydraulic concrete based on high-temperature accelerated curing

Determining the long-term mechanical properties of hydraulic concrete currently requires time-consuming and expensive tests, which has hampered investigations into the mechanical properties of long-age hydraulic concrete. In this study, the indoor macroscopic mechanical property tests of long-age hydraulic concrete with different curing ages (90, 180 days and 1, 2, 3 years) and different fly ash contents (0%, 15%, 35%) were designed and carried out on hydraulic concrete with 0.5 water-to-binder ratio by using a high-temperature accelerated-curing method based on equivalent age theory. Further tests were carried out using SEM, XRD, and MIP tests on long-age hydraulic concrete specimens with different fly ash contents to determine and analyze the long-term evolution of the macroscopic and microscopic properties of hydraulic concrete. The results showed that the fly ash content significantly affects the activation energy Ea of hydraulic concrete and that the curing time required to reach the designed hydration degree increases with increasing fly ash content. The long-age compressive strength and splitting tensile strength of hydraulic concrete with different fly ash contents increased logarithmically with curing age. After hydration for the curing age of 3 years, the overall microscopic structure of the mortar region of the long-age hydraulic concrete becomes dense, and the main XRD peak is due to quartz. The porosity decreases with increasing curing age, and the main pore types transition from multi-harmful pores (>200 nm) to harmless pores (<20 nm) and less harmful pores (20–50 nm). The multiple correlation coefficient of the long-term strength prediction model established by using the combined exponential prediction model ranges from 0.9152 to 0.9765, which indicates that the combined exponential prediction model based on the strength at the curing age of 28 days better replicates the long-term performance of hydraulic concrete.

Optimized artificial neural network model for accurate prediction of compressive strength of normal and high strength concrete

This study develops and presents an Artificial Neural Network (ANN) model employing the Levenberg-Marquardt Backpropagation (LMBP) training algorithm to predict the compressive strength of both normal and high strength concrete. The model's robustness was evaluated using an extensive dataset comprising 1637 samples. Eight input variables, including the cement content, blast furnace slag, fly ash, fine aggregate, coarse aggregate, water content, superplasticizer, and testing age, were considered. The optimal number of hidden layers and neurons in the layer were identified through analysis, and the effectiveness of the model was assessed through k-fold cross-validation and statistical measures, including correlation coefficient (R), coefficient of determination (R2), Root Mean Square Error (RMSE), and Mean Absolute Error (MEA). Comparison with other models was carried out, and the perturbation/super-position method was employed for parametric studies to investigate the effect of each input variable on the output variable. The k-fold cross-validation confirmed the generalizability of the model, and statistical measures showed good results, with unit cement content and superplasticizers having the highest impact on compressive strength. The findings demonstrate that the suggested ANN model is an extremely precise, economical, and practical predictive tool for concrete compressive strength.

Failure Mode Identification and Shear Strength Prediction of Rectangular Hollow RC Columns Using Novel Hybrid Machine Learning Models

Failure mode identification and shear strength prediction are critical issues in designing reinforced concrete (RC) structures. Nevertheless, specific guidelines for identifying the failure modes and for accurate predictions of the shear strength of rectangular hollow RC columns are not provided in design codes. This study develops hybrid machine learning (ML) models to accurately identify the failure modes and precisely predict the shear strength of rectangular hollow RC columns. For this purpose, 121 experimental results of such columns are collected from the literature. Eight widely used ML models are employed to identify the failure modes and predict the shear strength of the column. The moth-flame optimization (MFO) algorithm and five-fold cross-validation are utilized to fine-tune the hyperparameters of the ML models. Additionally, seven empirical formulas are adopted to evaluate the performance of regression ML models in predicting the shear strength. The results reveal that the hybrid MFO-extreme gradient boosting (XGB) model outperforms others in both classifying the failure modes (accuracy of 93%) and predicting the shear strength (R2 = 0.996) of hollow RC columns. Additionally, the results indicate that the MFO-XGB model is more accurate than the empirical models for shear strength prediction. Moreover, the effect of input parameters on the failure modes and shear strength is investigated using the Shapley Additive exPlanations method. Finally, an efficient web application is developed for users who want to use the results of this study or update a new dataset.

BO-Stacking: A novel shear strength prediction model of RC beams with stirrups based on Bayesian Optimization and model stacking

Shear strength prediction for reinforced concrete (RC) beams is a complex non-linear problem impacted by many factors. Traditional Machine Learning (ML) models for shear strength prediction still need manual single-model selection and hyperparameter optimization, limiting the model deployment efficiency and prediction accuracy in practical engineering. This study proposes a novel shear strength prediction model based on Bayesian Optimization and multi-layer stacking (BO-Stacking). It trains and stacks several base ML models in a multi-layer structure to avoid manual model selection and boost shear strength prediction accuracy concurrently. Besides, Bayesian Optimization is developed to automatically find optimal hyperparameters of each base ML model, substantially reducing computational cost during training. With the developed techniques, the proposed model is able to automatically achieve high accuracy by simply specifying training time. Furthermore, SHapley Additive exPlanation is adopted to interpret the feature contributions in shear strength prediction. The proposed model is implemented on an experimental dataset consisting of 406 RC beams with stirrups and then compared with seven base ML models and three shear design codes. Comparative results show that BO-Stacking achieves the highest accuracy on the test set with R2, RMSE and MAE of 0.9551, 39.98 and 25.99 respectively. The reliability and robustness of BO-Stacking is further verified through the trend analysis and Monte Carlo simulation. The proposed BO-Stacking model automates the tedious tasks of implementing an ML model for shear strength prediction and achieves stable and accurate results, facilitating the efficiency and reliability of model deployment in practice.

Unboxing machine learning models for concrete strength prediction using XAI

Concrete is a cost-effective construction material widely used in various building infrastructure projects. High-performance concrete, characterized by strength and durability, is crucial for structures that must withstand heavy loads and extreme weather conditions. Accurate prediction of concrete strength under different mixtures and loading conditions is essential for optimizing performance, reducing costs, and enhancing safety. Recent advancements in machine learning offer solutions to challenges in structural engineering, including concrete strength prediction. This paper evaluated the performance of eight popular machine learning models, encompassing regression methods such as Linear, Ridge, and LASSO, as well as tree-based models like Decision Trees, Random Forests, XGBoost, SVM, and ANN. The assessment was conducted using a standard dataset comprising 1030 concrete samples. Our experimental results demonstrated that ensemble learning techniques, notably XGBoost, outperformed other algorithms with an R-Square (R2) of 0.91 and a Root Mean Squared Error (RMSE) of 4.37. Additionally, we employed the SHAP (SHapley Additive exPlanations) technique to analyze the XGBoost model, providing civil engineers with insights to make informed decisions regarding concrete mix design and construction practices.

Compressive strength of fly ash based geopolymer utilizing waste completely decomposed granite

An attempt to recycle construction waste soils in southern cities of China was presented in this study, where completely decomposed granite (CDG) was utilized to replace fly ash in the production of environment-friendly low-carbon geopolymers. Five crucial parameters were identified and incorporated into the experimental design using the Taguchi method, including fine and gravel particle content of CDG, fly ash content, water-binder ratio, and curing temperature. Test results demonstrate that the newly formulated geopolymer can achieve a compressive strength of up to 14 MPa with a relatively high usage of CDG. The strength initially increases and then stabilizes with an increase in fine particle content. However, no significant relationship is observed between the compressive strength and gravel particle content. Higher curing temperatures and greater fly ash content lead to increased strength, while the strength initially decreases and then slightly increases with changes in the water-binder ratio. To examine the significance of these factors, Analysis of variance (ANOVA) were conducted and a correlation heatmap was drawn, revealing the following order of significance: curing temperature > fly ash content > water-binder ratio > fine particle content > gravel particle content. Furthermore, a robust strength prediction model was modified through multiple regression analysis. Another two additional groups of tests were performed to validate the proposed model, demonstrating the model's strong performance, particularly for curing ages beyond 14 days.

Prediction Model of Residual Soil Shear Strength under Dry–Wet Cycles and Its Uncertainty

Granite residual soil is widely distributed in Southeast Fujian. Large-scale engineering construction leads to the exposure of residual soil slopes to the natural environment. Affected by seasonal climate factors, the soil of slopes experiences a dry–wet cycle for a long time. The repeated changes in water content seriously affect the shear strength of soil, and then affect the stability of the slope. In order to explore the influence of the dry–wet cycle on the shear strength of granite residual soil in Fujian, an indoor dry–wet cycle simulation test was carried out for shallow granite residual soil on a slope in Fuzhou, and the relationship between water content, dry–wet cycle times, and the shear strength index, including the cohesion and internal friction angle of the granite residual soil, was discussed. The results show that when the number of dry–wet cycles is constant, the cohesion and internal friction angle of the granite residual soil decrease with an increase in water content. The relationship between the cohesion, internal friction angle, and water content can be described using a power function. Meanwhile, the fitting parameters of the power function are also a function of the number of wet and dry cycles. The prediction formulas of the cohesion and internal friction angle considering the number of dry–wet cycles and water content are established, and then the prediction formula of shear strength is obtained. The ratio of the predicted value of shear strength to the test value shall be within ±15%. An error transfer analysis based on the point estimation method shows that the overall uncertainty of the predicted value of shear strength caused by the combined uncertainty of the predicted value of cohesion and the internal friction angle and the single-variable uncertainty of the predicted value of shear strength caused only by the uncertainty of the predicted value of either the cohesion or internal friction angle increases first and then decreases with an increase in the number of dry–wet cycles. All increase with an increasing water content. The maximum standard deviation of the proposed shear strength prediction model of granite residual soil is less than 9%.

Machine Learning Based Compressive Strength Prediction Model for CFRP-confined Columns

The study described in the paper focuses on the compressive strength of concrete columns that are reinforced with carbon fiber reinforced polymer (CFRP). The researchers aimed to identify the main parameters that affect the compressive strength of these confined columns. They used models to analyze and study the relationship between these parameters and the compressive strength. The study found that certain parameters had an inverse relationship with compressive strength. These parameters include the column diameter, CFRP fracture strain, and modulus of elasticity. On the other hand, the CFRP thickness and concrete strength exhibited a positive relationship with the compressive strength. The study also determined that the influence of column diameter and CFRP thickness was greater compared to the influence of CFRP fracture strain and elastic modulus on the compressive strength. To predict the compressive strength, the researchers developed a machine learning algorithm-based compressive strength prediction model. They found that a backpropagation (BP) neural network model showed high prediction accuracy and robustness in predicting the strength. Additionally, the researchers' model was analyzed, and it was found that while the calculated values from their model aligned well with the experimental results, there were some issues with overestimating or conservatively estimating the compressive strength in certain cases.

Adhesion and Sliding Constitutive Relationship between Basalt–Polypropylene Hybrid Fiber-Reinforced Concrete and Steel Bars

The extreme marine environment of the South China Sea Islands, which features high temperatures, high humidity levels and high salt levels, seriously affects the safety of building structures. The durability of concrete can be significantly improved by adding a basalt–polypropylene hybrid fiber, but its bonding mechanism with deformed bars is complicated. Therefore, the bonding performance of hybrid basalt–polypropylene fiber-reinforced concrete and deformed bars was studied by combining experiments and a theoretical analysis. We designed 38 groups of different concrete strengths, different thicknesses of concrete covers, different anchor lengths and different diameters of rebars. The bond strengths, bond–sliding curves and failure forms of each pull specimen were compared and analyzed. The results showed that the failure forms and bond–slip curves of the basalt–polypropylene hybrid fiber-reinforced concrete specimens and the ordinary concrete specimens were essentially the same. Based on the results of the axial tensile tests, an ultimate bond strength prediction model was developed, and a bond–sliding constitutive model for hybrid fiber-reinforced concrete and steel bars was also established.

Prediction model of compressive strength of foamed concrete considering pore size distribution

To more accurately evaluate the relationship between pores and compressive strength of foamed concrete, the effects of density, water-cement ratio, and foaming agent type on porosity, pore size distribution, and compressive strength were studied. It was found that as the density decreased and the water-cement ratio increased, the number of larger-sized pores increased and the pore size distribution became wider and more non-uniform. Further, the pore structure of lower-density foamed concrete was more susceptible to the influence of density and water-cement ratio more likely to produce larger size pores. Particularly, foaming agent type had a significant effect on the pore size distribution. The pore size variation of foamed concrete was primarily attributed to the initial size of fresh foam and the complex merging and escape phenomena occurring in the foam in the slurry, the process of preparation and forming. It was also found that smaller porosity and more concentrated and uniform pore size distribution were more beneficial for strength. Additionally, to quantify the combined effect of porosity and pore size distribution on compressive strength, this paper innovatively proposes the parameters of the “pore synthesis impact factor”. Finally, a new empirical model relating the porosity, size distribution, and compressive strength is presented.

Experimental research on the thermal and mechanical response of flowing pipelines under the horizontal jet flame

The safe operation of parallel pipelines has become a focus of attention. The jet flame generated by the initial pipe accident may cause adjacent pipe section failure. Therefore, it is necessary to study the thermal impact of jet flame on adjacent pipelines. In this paper, we design and establish a thermal effect test platform to determine the change in pipe wall temperature and mechanical properties under the horizontal jet flame. Firstly, the flame combustion characteristics, such as flame form, temperature, and heat flux, are tested and analyzed. Then, the quantitative variation law of the pipe wall and oil temperature under different flow rates, wall thicknesses, and heating lengths is obtained. Combined with the mechanical test results of experimental pipelines, a pipe-bearing capacity calculation model is established via the GA-BP machine learning algorithm. Finally, the safety evaluation method and classification of the stress margin are established. The results indicate that the temperature and heat flux of the fire-facing surface is significantly higher than other clock orientations, where the values are 1287.38 ℃ and 12.37 kW/m2, respectively. As the flow rate decreases and the wall thickness and heating length increase, the pipe wall temperature rises linearly, and the tensile strength and impact toughness reduce logarithmically. The average relative error of the pipe strength prediction model is less than 2.44 %. The established safety evaluation method can be applied to determine the thermal risk level of steel pipelines under the jet flame.

Development of prediction models for interlayer shear strength in asphalt pavement using machine learning and SHAP techniques

The interlayer bonding condition in asphalt pavement significantly affects pavement performance. This study employed machine learning techniques to predict interlayer shear strength (ISS). Feed-forward artificial neural networks (ANN) and random forest (RF) models were developed and compared with traditional multiple linear regression (MLR). Utilizing 156 datasets, divided into 70% training and 30% testing, model performance was assessed using R-squared, mean squared error (MSE), root mean squared error (RMSE), and mean absolute error (MAE). SHapley Additive exPlanations (SHAP) was utilized for model interpretation. The results indicated that the ANN and RF models outperformed MLR, explaining over 95% of experimental data. RF exhibited superior performance with lowest MSE, RMSE, and MAE (0.0029, 0.0538, and 0.0376 MPa). SHAP analysis highlighted the significance of temperature, normal stress, shear deformation rate, and curing time as influential variables in ISS prediction. Elevated temperature adversely influenced ISS, while normal stress, shear deformation rate, and curing time positively contributed to ISS.

An investigation of fatigue behavior and residual strength model of steel-GFRP composite bar

Steel-fiber reinforced polymer (FRP) composite bar (SFCB) combines the advantages of corrosion resistance of FRP and good ductility of steel, and is a promising reinforcement material to improve the service life of marine structures. However, the fatigue behavior of SFCB is complex and has not been explored. Therefore, in this study, fatigue tests were conducted at different stress levels (0.32, 0.36, and 0.41 of quasi-static tensile strength with a fixed stress ratio of 0.1) and tensile tests were performed after fatigue pre-damage for a certain number of cycles (25%, 50%, and 75% of the fatigue life) to investigate the fatigue behavior of SFCB. The results show that fatigue failure is caused by steel core fracture, and the FRP layer can improve the fatigue life. Compared with the change of residual strength, the stiffness degradation of SFCB during fatigue loading was not significant. Finally, a residual strength prediction model containing known parameters of the S-N curve is also proposed, which can well predict the residual strength decay of SFCB after fatigue damage. In conclusion, these findings and the proposed residual strength prediction model contribute to the understanding and design of SFCBs.