열역학 계산과 생성적 적대 신경망을 이요한 미세합금강의 비정상 결정립 성장 거동 분석 = Thermodynamic Calculations and Generative Adversarial Network-Based Analysis of Abnormal Grain Growth Behavior in Microalloyed Steels

다국어 초록 (Multilingual Abstract)

Carburizing heat treatment is generally carried out using specially designed microalloyed steels. Recently, high-temperature carburizing has attracted increasing attention as an effective means to shorten the carburizing time and thereby reduce both processing cost and CO2 emissions, in line with global carbon-neutrality initiatives. However, such high-temperature carburizing often causes severe austenite grain coarsening, leading to deterioration of mechanical properties. In this study, Nb and Ti were added to SCR420H steel to promote grain refinement during high-temperature carburizing. The resulting microstructures and precipitates in a steel without microalloying elements were compared with those in steels containing Nb or Ti. Thermodynamic analysis revealed that Al steel formed AlN and a small amount of (Ti,Nb)(C,N) precipitates; AlNb steel predominantly contained (Nb,Ti)(C,N) precipitates; and AlTi steel exhibited both (Ti,Nb)(C,N) and Ti₄C₂S₂ precipitates. The tendency for abnormal grain growth (AGG) was evaluated as a function of alloy chemistry and heat-treatment parameters. The addition of Nb and Ti increased the volume fraction of (Ti, Nb)(C, N) precipitates. Ti promoted the formation of Ti₄C₂S₂ particles, which remained stable at high temperatures. Furthermore, the grain coarsening temperature (GCT) increased further with increasing reheating temperature due to the high high temperature stability of Ti based precipitates above 1040°C. Unlike previous studies that primarily relied on visual observation or simple ratio based approaches using the Novikov equation, this study used MatCalc thermodynamic software to qualitatively and quantitatively evaluate the AGG behavior. Finally, we utilized artificial intelligence to learn grain characteristics in Al, AlNb, and AlTi alloys and heat treatment conditions, and predicted and verified the grain coarsening temperature using images generated by a generative adversarial network(GAN).

목차 (Table of Contents)

• 1. 서 론 1
• 2. 이론적 배경 4
• 3. 실험방법 12
• 4. 결과 및 고찰 22
• 5. 결 론 46
• 참 고 문 헌 47