A Novel Image Analysis Methodology for Data‐Driven Assessment of Polymer Combustion Dynamics in the Vertical Flame Test

ABSTRACT The Underwriters Laboratories (UL 94) vertical burning test remains a standard and indispensable method for assessing polymer flammability, acknowledging the significant role of gravity in fire behavior. However, the reliance on qualitative classification (e.g., V‐0, V‐2) fails to capture the non‐linear, transient combustion dynamics, thereby restricting the scientific provision of time‐resolved quantitative data critical for understanding flame retardancy (FR) mechanisms. To fundamentally transform this assessment into a quantitative, data‐driven process, this study introduces a novel methodology based on image analysis applied to videos captured during the vertical flame test. This approach facilitates the automatic and high‐precision extraction of time‐series features—such as fluctuations in flame height and area, and flame positional changes—providing a direct, quantifiable reflection of polymer combustion dynamics. The extracted features were statistically validated using Canonical Discriminant Analysis, which successfully classified eight common resin types with high accuracy. This confirms that the numerical data effectively capture the complex, resin‐specific combustion behavior. Incorporating this novel, data‐driven technique alongside existing, established evaluation methods will lead to a more profound understanding of combustion states during the vertical flame test. This work provides genuine Data‐Driven Insight into flammability, marking a crucial advancement toward the accelerated development of next‐generation FR composite materials.