Soft computing techniques for predicting the properties of raw rice husk concrete bricks using regression-based...

Area of Science:

• Materials Science
• Civil Engineering
• Sustainable Construction

Background:

• Traditional concrete production has significant environmental impacts.
• Recycled materials like raw rice husk, fly ash, and hydrated lime offer potential for sustainable construction.
• Developing eco-friendly concrete alternatives is crucial for reducing carbon footprint.

Purpose of the Study:

• To evaluate the use of raw rice husk, fly ash, and hydrated lime as replacements for fine aggregate and cement in concrete bricks.
• To optimize the compressive strength, water absorption, and dry density of these novel concrete bricks.
• To validate the predictive accuracy of optimization models using machine learning techniques.

Main Methods:

• Response Surface Methodology (RSM) was employed to optimize concrete brick properties.
• Artificial Neural Network (ANN) and Multiple Linear Regression (MLR) were used for model validation.
• Sensitivity analysis was conducted to assess the influence of input parameters on outcomes.

Main Results:

• The ANN and RSM models demonstrated superior accuracy in predicting concrete brick properties compared to MLR.
• High R-squared values (R² > 0.9997 for ANN, R² > 0.9155 for MLR) confirmed model reliability.
• Optimized concrete bricks showed promising mechanical properties, with accurate predictions for 28-day compressive strength, water absorption, and density.

Conclusions:

• Machine learning, particularly ANN combined with RSM, provides an efficient and accurate method for predicting concrete mechanical properties.
• This approach conserves time, labor, and resources in civil engineering research and development.
• Raw rice husk-based concrete bricks represent a viable sustainable construction material.