Rationalised Steel Grade Dataset Using the K-Means Reduction Process (KMRP)

This dataset provides a rationalised collection of carbon and stainless steel grades compiled from multiple internationally recognised designation systems. It integrates mechanical properties, chemical composition ranges, processing indicators, and alloy identifiers from major global standards to enable comparative analysis and data-driven investigation of steel grade behaviour and redundancy. Steel grades were initially consolidated from the following international standards and designation systems: AISI / SAE (American Iron and Steel Institute / Society of Automotive Engineers);UNS (Unified Numbering System for Metals and Alloys);ASTM (American Society for Testing and Materials);EN / EN 10027 (European standards for steel classification);GB / GB/T (Chinese national standards for ferrous alloys). Mechanical, chemical, and processing information was collected from three publicly accessible materials databases: MatWeb (matweb.com), MakeItFrom (makeitfrom.com), and Steel-Grades (steel-grades.com). Data from these sources were cross-referenced and harmonised to ensure consistency across property definitions, compositional ranges, and processing descriptors. The dataset includes the following information for each steel grade: Mechanical properties, including ultimate tensile strength (UTS), yield strength, elongation, and hardness (where reported);Chemical composition ranges, expressed as minimum and maximum values for principal alloying elements as well as tramp and impurity elements;Processing-related attributes, represented through binary indicators for common thermomechanical and heat treatment conditions (e.g. annealed, quenched, hardened);Standard identifiers and nomenclature, including AISI, SAE, UNS, EN, GB/T, and ASTM grade names and numbers. This dataset represents the final output of the K-Means Reduction Process (KMRP), a custom data-driven rationalisation framework developed to systematically reduce the number of existing steel grades while preserving full mechanical property coverage. The KMRP algorithm clusters steel grades based on similarity in mechanical performance and subsequently eliminates redundant grades whose property ranges are fully covered by other representatives within the same cluster. Each steel grade is annotated with a categorical Status label indicating whether it was retained (Remaining) or removed (Eliminated) during the maximum-elimination strategy. The retained grades collectively span the full mechanical performance space of the original dataset, thereby demonstrating that a substantially smaller subset of grades can represent the same application-driven performance envelope.