Masonry construction is a cornerstone of civil engineering, employing meticulously arranged units like bricks, stones, or concrete blocks to create long-lasting structures. Its advantages include high compressive strength for walls and foundations, excellent durability, inherent fire resistance, and aesthetic versatility (Sathiparan, 2015). Sandcrete blocks, also known as concrete blocks or cement-sand blocks, have become a popular choice in masonry construction for several reasons. Their affordability compared to fired bricks or natural stones makes them cost-effective. Production is fast and efficient, meeting high construction demands (Olaiya et al., 2023). Modern production methods ensure consistent block sizes, simplifying construction and reducing waste. Additionally, sandcrete blocks come in various shapes, sizes, and strengths, adaptable for building walls, partitions, and even some structural elements (Xu et al., 2024).
The major environmental concerns surrounding sandcrete block production stem from the usage of cement. Cement production is a significant contributor to greenhouse gases, particularly carbon dioxide (CO2), due to the high temperatures required during the manufacturing process (Belaïd, 2022). Estimates suggest cement production accounts for around 8% of global CO2 emissions. Researchers are actively searching for alternative materials to replace cement and reduce its environmental impact. However, choosing a suitable substitute isn't straightforward. Several factors need to be considered, including the material's overall quality, ease of availability, and the cost of transporting it to production facilities (Mahasenan et al., 2003).
Rice husks, a byproduct of rice milling, are a globally abundant form of municipal waste. Unfortunately, their potential uses haven't been fully explored. Despite the massive rice production of nearly 500 million metric tons in 2020 (Anburuvel et al., 2023), with each ton generating roughly 0.28 kg of rice husk waste (Kumar Das et al., 2022), most of this husk ends up being burned in the open or dumped in landfills. These disposal methods cause serious environmental problems. Only a small fraction of rice husk ash (RHA), the leftover material after burning, is used for purposes like fertilizer additives, animal bedding, cooking oil filtration, or even in landfills and paving applications (Pode, 2016). The incorporation of RHA as a cement substitute in cementitious materials has been demonstrated to not only enhance the mechanical performance of these materials but also promote a more sustainable construction industry. Its pozzolanic activity, a key factor in enhancing the performance of cement-based materials, is influenced by its amorphous silica content, specific surface area, and particle fineness (Zerbino et al., 2011). Controlled combustion and grinding processes can significantly improve these properties for optimal utilization in cement-based materials (Jamil et al., 2013; Subramaniam and Sathiparan, 2022). Jittin et al. (2020) explored the pozzolanic potential of RHA and its applications in various construction materials, including concrete, alkali-activated binders, pavements, and bricks. They identified several factors impacting RHA's pozzolanic reactivity: fineness, incineration conditions, available alkali media, mix design, and amorphous silica concentration. Siddika et al. (2021) and Amran et al. (2021) delved deeper into these aspects, investigating the mechanical properties, curing behavior, microstructure, and durability of RHA-blended concrete. Similarly, Christopher et al. (2017) and Thomas (2018) conducted comprehensive studies on concrete incorporating RHA as a partial replacement for cement. Their work emphasizes the importance of processing factors and material characteristics in this approach. The existing literature suggests that RHA can be effectively utilized as a cement substitute in cement-based materials, up to a 10–20% replacement level, without compromising performance due to its strong pozzolanic properties. Additionally, studies have shown reduced workability when incorporating RHA at replacement levels of 15% and 20% by weight (Antiohos et al., 2014).
Building on existing research, several studies investigated the potential of RHA as a partial replacement for cement in sandcrte blocks. Findings supported the notion that incorporating RHA as a cement substitute demonstrably affects the mechanical properties of sandcrete (Khan et al., 2021). The extent of this influence is contingent on several factors, including the amount of RHA replacing cement, the fine aggregate-to-binder ratio, the water-to-binder ratio, and curing time. Given these dependencies, it is crucial to analyze the impact of these factors on compressive strength and develop a methodology specifically for predicting compressive strength in RHA-blended sandcrete blocks. While existing literature explores compressive strength prediction in conventional RHA blended concrete (Bassi et al., 2023; Huang et al., 2023; Kashem et al., 2024; Nasir Amin et al., 2023; Paul et al., 2024) and pervious concrete (Malami et al., 2022), a gap remains regarding a dedicated prediction model for sandcrete blended RHA. This study aims to address this gap by proposing a methodology for such predictions.
The field of material science is increasingly embracing machine learning (ML) for predicting building material properties (Feng et al., 2020; Kaveh and Shabani Rad, 2023; Marani and Nehdi, 2020; Sathiparan and Jeyananthan, 2023a, b). Sandcrete blended with RHA exhibits complex strength behavior influenced by the replacement level and various other parameters. Machine learning methods are particularly well-suited for capturing these non-linear relationships (Kaveh et al., 2021). This research aims to leverage ML to predict the compressive strength of RHA blended sandcrete. This approach is proposed to reduce reliance on laboratory testing, which can generate construction waste, and empower engineers with accurate prediction tools (Sathiparan et al., 2023; Subramaniam et al., 2023). Traditional mathematical models may struggle to capture the intricacies of these relationships, whereas ML can offer effective solutions for both linear and non-linear scenarios (Gao et al., 2019; Sathiparan et al., 2023; Wijekoon et al., 2023).
In the present study, a data-driven approach was employed, utilizing published data and evaluating five distinct ML algorithms to identify the most effective and reliable method for predicting compressive strength in RHA blended sandcrete. The developed models offer a promising avenue for improving the accuracy of compressive strength prediction in this sustainable construction material.