An improved framework for multi-objective optimization of cementitious composites using Taguchi-TOPSIS approach

The traditional Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) methodology is commonly used for the multi-objective optimization of cementitious composites, allowing the simultaneous optimization of various mechanical and physical properties. Due to the significant scale differences among these properties, such as target strength (ranging from tens to hundreds) and strain (typically 0-1%), normalization is essential for accurate comparison. However, current civil engineering practices often employ fixed normalization methods, which may not always lead to optimal performance. This study addresses this limitation by proposing a novel framework for evaluating normalization methods within the TOPSIS process. The framework integrates metrics such as the Ranking Consistency Index (RCI), Spearman Correlation (SC), Rank Variance (RV), plurality voting, and Pareto dominance sorting to identify and exclude unsuitable normalization techniques. It was validated using three experimental datasets: hybrid fibre engineered cementitious composites, recycled aggregate concrete, and geopolymer concrete. The results showed considerable variation in optimization outcomes depending on the normalization method. For the tested datasets, the framework identified the Linear max-min and Lai and Hwang methods as superior due to their higher RCI, SC and lower RV, and these methods also resulted in optimal properties, thereby confirming the effectiveness of the framework. Overall, the study highlights the critical role of selecting suitable normalization methods in multi-response optimization and demonstrates how the proposed framework improves optimization accuracy.