Characterization of roll bite heat transfers in hot steel strip rolling and their influence on roll thermal fatigue degradation

A temperature sensor with a thermocouple placed at ~0.5 mm from roll surface is used in hot rolling conditions to evaluate by inverse calculation heat transfers in the roll bite. Simulation analysis under industrial hot rolling conditions with short contact lengths (e.g. short contact times) and high rolling speeds (7 m./s) show that the temperature sensor + inverse analysis with a high acquisition frequency (> 1000 Hz) is capable to predict accurately (5 to 10% error) the roll bite peak of temperature. However as heat flux is more sensitive to noise measurement, the peak of heat flux in the bite is under-estimated (20% error) by the inverse calculation and thus the average roll bite heat flux is also interesting information from the sensor (these simulation results will be verified with an industrial trial that is being prepared). Rolling tests on a pilot mill with low rolling speeds (from 0.3 to 1.5 m./s) and strip reductions varying from 10 to 40% have been performed with the temperature sensor. Analysis of the tests by inverse calculation show that at low speed (<0.5 m./sec.) and large contact lengths (reduction: 30 to 40%), the roll bite peak of heat flux reconstructed by inverse calculation is correct. At higher speeds (1.5 m./sec.) and smaller contact lengths (reduction : 10-20%), the reconstruction is incorrect: heat flux peak in the bite is under-estimated by the inverse calculation though its average value is correct. The analysis reveals also that the Heat Transfer Coefficient HTC _roll-bite (characterizing heat transfers between roll and strip in the bite) is not uniform along the roll bite but is proportional to the local rolling pressure. Finally, based on the above results, simulations with a roll thermal fatigue degradation model under industrial hot rolling conditions show that the non-uniform roll bite Heat Transfer Coefficient HTC_roll-bite may have under certain rolling conditions a stronger influence on roll thermal fatigue degradation than the equivalent (e.g. same average) HTC _roll-bite taken uniform along the bite. Consequently, to be realistic the roll thermal fatigue degradation model has to incorporate this nonuniform HTCroll-bite.