An Integrated Approach for Analyzing and Evaluating Mix Design Proportions and Compressive Strength of Geopolymer Concrete Sections Lacking Technical Data by Integrating Digital Image Processing and Machine Learning

Keshavarzi, Mansour; Pourrostam, Towhid; Javaheri Tafti, Mohammad Reza

doi:10.22034/ijumes.2026.743389

An Integrated Approach for Analyzing and Evaluating Mix Design Proportions and Compressive Strength of Geopolymer Concrete Sections Lacking Technical Data by Integrating Digital Image Processing and Machine Learning

Document Type : Case Study

Authors

Mansour Keshavarzi ¹

Towhid Pourrostam ²

Mohammad Reza Javaheri Tafti ¹

¹ Department of Civil Engineering, Taft. C., Islamic Azad University, Taft, Iran

² Department of Civil Engineering, CT. C., Islamic Azad University, Tehran, Iran

10.22034/ijumes.2026.743389

Abstract

Given the complex behavior of geopolymer concretes and the simultaneous influence of mix design characteristics, aggregate properties, binder composition, and curing conditions on compressive strength, the development of rapid. This study presents an approach based on digital image processing and analysis combined with data-driven modeling to identify quantitative characteristics of geopolymer concrete constituents and predict compressive strength. To this end, images of specimen sections were preprocessed and then analyzed using RGB-to-HSV color space transformation, grayscale conversion, contrast enhancement, edge detection, thresholding, and masking. Subsequently, a dataset comprising 117 compressive strength records was used to develop the prediction model, with 75% of the data allocated for training and 25% for testing. The input variables included coarse aggregate percentage, fine aggregate percentage, specimen age, molarity, sodium silicate-to-sodium hydroxide solution ratio, and ground granulated blast furnace slag content. The prediction model was developed using gene expression programming machine learning to identify the nonlinear relationship between the variables governing the mix design and compressive strength. The results indicated that digital image processing can extract useful quantitative information regarding the distribution of constituent phases within concrete sections, and that this information, together with mix design parameters and testing conditions, provides adequate capacity for use in compressive strength modeling. By combining image-based analysis and data-driven modeling, the proposed approach offers a framework for reducing reliance on purely empirical assessments and facilitating the analysis of relationships among microstructural characteristics, mix design parameters, and the mechanical performance of geopolymer concretes.

Keywords

Aggregate properties

blast furnace slag

digital image processing

geopolymer concrete

image analysis