Abstract:Abstract In place of Portland cement concrete, alkali-activated materials (AAMs) are becoming more popular because of their widespread use and low environmental effects. Unfortunately, reliable property predictions have been impeded by the restrictions of conventional materials science methods and the large compositional variability of AAMs. A support vector machine (SVM), a bagging regressor (BR), and a random forest regressor (RFR) were among the machine learning models developed in this study to assess the compressive strength (CS) of AAMs in an effort to gain an answer to this topic. Improving predictions in this crucial area was the goal of this study, which used a large dataset with 381 points and eight input factors. Also, the relevance of contributing components was assessed using a shapley additive explanations (SHAP) approach. In terms of predicting AAMs CS, RFR outperformed BR and SVM. Compared to the RFR model’s 0.96 R 2 , the SVM and BR models’ R 2 -values were 0.89 and 0.93, respectively. In addition, the RFR model’s greater accuracy was indicated by an average absolute error value of 4.08 MPa compared to the SVM’s 6.80 MPa and the BR’s 5.83 MPa, which provided further proof of their validity. According to the outcomes of the SHAP research, the two factors that contributed the most beneficially to the strength were aggregate volumetric ratio and reactivity. The factors that contributed the most negatively were specific surface area, silicate modulus, and sodium hydroxide concentration. Using the produced models to find the CS of AAMs for various input parameter values can help cut down on costly and time-consuming laboratory testing. In order to find the best amounts of raw materials for AAMs, academics and industries could find this SHAP study useful.

Expansion prediction of alkali aggregate reactivity-affected concrete structures using a hybrid soft computing method

Alkali–silica reaction expansion prediction in concrete using hybrid metaheuristic optimized machine learning algorithms

Prediction of alkali-silica reaction expansion of concrete using artificial neural networks

Development of prediction models for evaluation of alkali-silica reaction in concrete

Correlating alkali-silica reaction (ASR) induced expansion from short-term laboratory testings to long-term field performance: A semi-empirical model

New hybrid support vector regression methods for predicting fresh and hardened properties of self-compacting concrete

Promoting low carbon construction using alkali-activated materials: A modeling study for strength prediction and feature interaction

Evaluation of multiple machine learning models for ASR expansion of concrete

Predicting the Expansion of Concrete Exposed to Sulfate Attack with a Regression Model Based on a Performance Classification

Numerical Calculation of Expansion Induced by Alkali Silica Reaction

Alkali-silica reaction expansion model for confined concrete with stress-dependency and casting direction anisotropy

Numerical Prediction of Swelling in Concrete Arch Dams Affected by Alkali-Aggregate Reaction

Calculation of alkali silica reaction (ASR) induced expansion before cracking of concrete

A Novel Multiscale Modelling Approach for Evaluation of the ASR in Concrete Structures

Prediction of Service Life of Concrete Structures Affected by Alkali-Silica Reaction

Meso‐scale Particle Modeling of Concrete Deterioration Caused by Alkali‐aggregate Reaction

Concrete Expansion Induced by Alkali–Silica Reaction in a Small Arch Dam

Chemo-Damage Model for Concrete Anisotropic Expansion Caused by Alkali-Aggregate Reaction

Mesoscale Modeling Study on Mechanical Deterioration of Alkali–Aggregate Reaction-Affected Concrete

Fresh state and strength performance evaluation of slag-based alkali-activated concrete using soft-computing methods

Predicting the deformation behaviour of concrete face rockfill dams by combining support vector machine and AdaBoost ensemble algorithm