Research On Prediction And Optimization Of Microstructure And Mechanical Properties Of X70Pipeline Steel In Continuous Casting And Rolling For Medium-Thin Slab

As one of three technologies on thin slab casting and rolling, continuous casting and rolling for medium-thin slab (ASP) plays an important role in the production of hot strip rolling. There are nine ASP production lines in China at present. Some iron and steel enterprises have produced X70pipeline steel on ASP line successfully. But there are some problems existing in the production of X70pipeline steel, such as bigger fluctuation on mechanical properties, instability about evolution of micro structure. So this article systematically studied up on the microstructure evolution and mechanical properties prediction of X70pipeline steel based on ASP1700line of Jinan Iron and Steel Group. The temperature field model which is suitable for ASP line was established. Subsequently, this article also had an in-depth study on recrystallization, phase transformation, prediction of mechanical properties with artificial intelligence and optimization of chemical composition. So the main contents and innovative points about this paper are as follows:(1) According to the characters of continuous casting and rolling for medium-thin slab, the temperature field model which is suitable for X70pipeline steel was established by finite difference method. The temperature field was simulated during X70pipeline steel rolling. And the result showed one-dimensional temperature field along rolling direction (temperature varies with time or distance) and two-dimensional temperature field of rolled piece’s cross section. The average temperature difference was20℃by comparison of simulated and measured temperatures for which X70pipeline steel of6.47mm thick was rolled at the position of thermometric instrument. In order to eliminate such temperature difference, the heat transfer coefficients was modified by Powell algorithm. The research showed that the simulated and measured temperatures for Q345B test steel of11.5mm thick agree with each other very well subsequently. It was proved that the modified heat transfer coefficients were suitable for rolling of ASP1700line. The simulated result also showed that the temperature uniformity of strip’s cross section was increased before coiling by edge-masking laminar cooling water. (2) The activation energy of dynamic and static recrystallization of X70pipeline steel was obtained by thermal simulation experiment technology. Based on the previous research, austenitic microstmcture evolution model of X70pipeline steel produced by ASP line was established. The research object was X70pipeline steel with thickness of11.36mm which was rolled by medium-thin slab of135mm thick. The recrystallization of austenite and changes of mean flow stress were simulated under actual rolling schedule. In this basis, the process means of how to refine austenitic grain and increase accumulated strain were investigated, too. It was showed that the actual austenitic recrystallization behavior was described accurately by the established recrystallization model. The two methods of how to refine austenitic grain and increase accumulated strain (increasing press quantity of F1-F3and decreasing finishing temperature by adding intercooling), both could refine austenitic grain in finish rolling exit to15μm. But the residual strain of latter was bigger, which made the means flow strain of F6increased by60MPa.(3) The mathematical relationship between chemical compositions and critical carbon concentration when transformation from austenite to ferrite was obtained by thermal simulation experiment. Ferrite fraction was calculated by UBC models. The relationship between parameters of ferrite transformation and chemical composition was obtained by regression. The start temperature of ferrite transformation, fraction and grain size of ferrite were simulated under actual cooling schedule. It was indicated that the start temperature of ferrite transformation under continuous cooling was proportional to the one under equilibrium state. The relationship of between Ar3and product of cooling rate and initial austenitic grain size satisfied power function. The calculated ferrite fraction with regressed parameters agreed with measured fraction very well. After transformation of the X70pipeline steel with thickness of11.36mm done, the deviation of ferrite grain size was2μm between surface and center of cross section. The start temperatures of ferrite transformation were different under three cooling schedule. But the end temperature of ferrite transformation and ferrite fraction were equivalent. The routes of ferrite transformation were different within each temperature section.(4) Based on a large number of field production data, the mechanical properties of X70pipeline steel were predicted by artificial intelligence model. The artificial intelligence model was modified by eliminating the effect of input samples sequence. Rolling process parameters satisfied the normal distribution which verified by SPSS commercial software. So the predicted samples could be screened with3a rule. The result showed that the predicted precision increased by three methods, such as eliminating the effect of input samples sequence, screening predicted samples with3o rule and increasing number of predicted samples. Predicted precision of yield strength and tensile strength respectively increased33%and9%within±6%error by eliminating the effect of input samples sequence. The standard deviations of predicted error were within±3%by predicted samples preprocessing. The effect tendency of chemical compositions and finishing temperature to mechanical properties was obtained by this method.(5) The main factors which affect mechanical properties were studied by BP network and genetic algorithm. The method of optimization of chemical composition and temperature schedule were established.In this case, chemical composition and temperature schedule were optimized with presupposed mechanical properties. It was showed that BP network and genetic algorithm could be used to optimize the chemical composition and temperature schedule under considering performance price ratio. The optimized result also showed that decreased final data temperature had the same effect by adding microalloy elements.