FEATURE ADVANCED MATERIALS & PROCESSES | APRIL 2025 53 ed dilatometry curve with the fast-quenching rate (HTC of 2000) matches the experimental data well for the phenomena in the circled regions A and B. The root causes for the difference between the fast quenching and the uniform cooling dilatometry curves are investigated. BEGINNING OF QUENCHING STEP At the beginning of quenching, the sample is cooled from both the outer diameter (OD) and the end surfaces. This leads to a lower temperature and more shrinkage in the corner region, as shown in Fig. 3, with snapshots of temperature and sample shape at the 0.2 second during quenching. The part shape change is magnified by 10 times, and it shows clearly that the end face has a crown shape instead of being flat. The dilatometry displacement is measured from the largest distance between the end faces of the cylinder, and the temperature is taken from the middle of the OD surface, as shown by the red dot in Fig. 3. A series of modeling investigations show that the inaccuracy in the dilatometry data in Region A can be controlled by using a slower quenching rate, using a cylinder sample with a smaller diameter, and using a cylinder sample with a crown end face. The third recommendation will make the axial displacement measurement using a strain gauge from two points instead of two surfaces, which improves the data consistency. of the axial displacement takes the corner points, and the temperature reading is from the middle point of the OD surface. The temperature nonuniformity in the sample during martensitic phase transformation is the root cause of the data inaccuracy. For steels with a higher harden- ability, the quenching rate can be decreased to improve the temperature uniformity while suppressing the diffusive transformations. For steels with a low hardenability, the sample size (smaller diameter) and sample geometry (crown end face) can be customized to improve the data accuracy. By reducing the quenching rate from the HTC of 2000 W/(m2K) to 200 W/(m2K), the calculated dilatometry curve is close to the ideal cooling condition, as shown in Fig. 5. The simulated Ms difference between the slow cooling and the ideal cooling is about 5°C. FURTHER MODELING ANALYSIS The cylinder size affects the temperature uniformity during quenching. Figure 6 compares the temperature distribution profiles at the beginning of martensitic Fig. 3 — Demonstration of sample shape change and its effect on the accuracy of collected displacement data at the beginning of quenching. MARTENSITIC TRANSFORMATION In this article, only the martensitic transformation is considered during quenching to simplify the dilatometry strain curve. The quenching rate used in the model is fast enough to miss the diffusive phase transformations. The martensite start temperature (Ms) for AISI 4140 steel is about 325°C. At the 2.2 second point during quenching, the core temperature of the cylinder sample is about 390°C (above Ms), while the corner temperature is about 290°C (below Ms). The martensitic transformation occurs in the corner region at this time snapshot, which causes a volume expansion, as shown in Fig. 4. The measurement Fig. 4 — Demonstration of sample shape change and its effect on the accuracy of collected displacement data when martensite transformation starts. Fig. 5 — Effect of quenching rate on the accuracy of the collected displacement data. 9
RkJQdWJsaXNoZXIy MTYyMzk3NQ==