【作者】 范国锋；

【导师】 王华； 卿山；

【作者基本信息】 昆明理工大学 ， 钢铁冶金， 2013， 博士

【摘要】 云南省高磷铁矿资源十分丰富,其中惠民铁矿、罗茨铁矿等高磷铁矿总储量超过20亿吨,但因其磷含量高,难以大量应用于炼铁工业。而惠民矿属于高磷褐铁矿的种,富含结晶水、比重小、磁性弱、可浮性低,故难以通过物理选矿法提高其品位。由于传统高炉炼铁生产发展受到资源、环保等方面的制约,我国不仅要继续完善和改进高炉炼铁工艺,而且还应适度发展直接还原和熔融还原等技术,特别是直接利用粉矿、粉煤为原料的非高炉炼铁新技术。为了揭示高磷铁矿粉的还原过程动力学行为特征及传热传质规律,本文采用数学建模与实验研究相结合的方法对惠民铁矿粉在高温下的普通煤粉还原过程动力学行为开展研究,通过热分析和时间序列分析方法对矿粉的还原过程非平衡态进行阶段建模、特征提取和还原过程中传热传质行为的考察,并使用分析测试技术进行实验验证,为昆明钢铁集团早日应用Hismelt熔融还原技术处理矿粉提供理论依据。本文在实验的基础上运用失重法研究了40℃～1400℃条件下,普通煤粉还原惠民高磷铁矿粉的动力学过程。首先利用热重分析仪对惠民铁矿进行还原实验来获取熔融还原过程的TG(质量变化)-DSC(热量变化)曲线,分别根据其曲线的升降变化和峰值特征得出反应的热力学数据及动力学数据,将整个过程分为：(1)结晶水脱除阶段：(2)赤铁矿预还原阶段；(3)熔化态阶段；(4)Fe2O3完全还原阶段：(5)Fe3O4还原为FeO阶段；(6)FeO还原为Fe阶段。其次运用热分析动力学中的积分法结合48个经验机理模型构建出备反应阶段的表观动力学模型,包括动力学参数,如表观活化能、频率因子和最概然机理函数(E,A,f(α))的计算,发现随着温度的升高,关键阶段所需的表观活化能呈现了递增的趋势,说明主元反应依次序进行,而通过纵向的补偿效应分析后发现E和1gA之间相互依赖且协同变化明显,说明还原过程中各类反应进行的难易程度与碰撞概率之间相互影响显著。然后通过公式推导和数据拟合建立了表观活化能随温度、升温速率的分布函数表达式,其函数曲线呈现出类高斯分布,并表现为后拖尾现象,表明表观活化能在高温段的分布稀疏且变化较快,体现了与熔融还原的能量补给需要一致；经过拟合得出了还原过程的频率因子表达式,其曲线拐点揭示出还原过程在熔化状态下活化分子之间的碰撞次数明显增加,有利于还原的进行。经过综合考虑提炼出能反映还原系统运行全貌及其细节的物理量—表观反应速率(DTG时间序列数据),该物理量不仅能够反映冶炼系统进行过程中的动力学行为,也能体现过程中的传热传质行为。具体内容如下：(1)运用动力系统理论中的相重构技术对实验所获取的DTG序列数据进行重构分析,并从表观反应速率的演化中恢复熔融还原过程中的动力学行为,计算结果表明：该序列的一阶时间延迟和二维嵌入能够准确地反映系统的动力学行为；该确定性复杂冶炼系统求属随机离散动力系统范畴,具有双“∞”的混沌吸引子结构,我们将其命名为铁还原过程吸引子：通过文献比较推断双“∞”曲线中内“∞”上的点表征着冶炼系统进入界面化学反应控制阶段,外“∞”L的点则表征着系统进入扩散控制阶段,而散布在该曲线周围的点表征着混合控制的结果及其偏离程度。通过对DTG序列的递归图分析、庞加莱截面和关联维数的计算以及最大李亚普诺指数的计算过程共同表明了惠民铁矿还原过程呈现出了弱混沌现象,且该行为是铁的还原引起的,因此确定该混沌行为为有用过程,加强该行为有助于实现冶炼的强化。(2)铁还原过程吸引子不同于Lorenz系统、Rossler系统和logistic I映射系统,而混沌性态完全不同于它们,则方程也无法参照上述的类型进行构建。普适方程的计算结果表明DTG的影响因素繁多,难以应用通常的微分方程进行求解,基于此运用时间序列分析中的预测建模方法对DTG进行表达,局部加权线性模型的良好预测性能体现出反应系统内部存在着短程相关性(局部)以及主次差别性(加权),进一步建立的EMD-wAR模型能够准确的表达出能量传输过程的频率分布特征(EMD),体现了反应系统内部存在着短程相关性(AR)以及主次差别性(黄金加权),EMD-wAR较局部加权线性法误差下降了一个数量级的,而EMD-wAR-CPSO-SVR模型的建立使得误差又较EMD-wAR模型下降了一个数量级。EMD-wAR-CPSO-SVR模型更有利于建立还原过程吸引子方程。(3)通过DTG信号的功率谱分析以及比较发现纯的氧化铁还原过程其能量传输集中在低频段,而矿石还原过程集中在中高频,因此在电炉冶炼中适当加强高频段的加热方式或调整工艺参数使得冶炼过程中的中高频段能量传输加强,这些都有助于实现冶炼过程的强化。最后在几个关键温度段对惠民铁矿进行竖式电炉还原实验,利用各阶段渣相的表观形貌变化推测反应过程的动力学行为特征。更多还原

【Abstract】 The resources of high-phosphorus iron ore are very abundant in Yunnan Province, the reserves such as the Huimin, Luoci iron ore, etc. have over2billion tons. Because of high phosphorus content, it is difficultly to be employed with large scale by the iron making industry. The Huimin iron ore is one kind of high phosphorus and Limonite, due to its some embedded features, such as rich crystallization water, the small proportion, and the weak magnetic flotability, the iron grade of ore is difficult to be improved by physical beneficiation approaches. The development of the traditional blast furnace iron production is recently constrained by resources and environmental protection consideration, in the meanwhile, our country not only has to continue to refine and improve iron making process of blast furnace, but also should moderately develop the technology of direct reduction and smelting reduction, particularly for the technology of non-blast furnace iron making focusing on taking iron ore fines or pink-coal as raw material directly. In order to reveal the kinetic behaviors and heat and mass transfer law for reduction of high-phosphorus iron ore fines, the combination methods of mathematical modeling and experimental are employed to study the kinetic behaviors in reduction process of the Huimin iron ore with the ordinary coal at high temperature. The thermal analysis and time series analysis approaches are applied to stage modeling, feature extraction slag and investigation of the behavior of heat and mass transfer process for reduction process non-equilibrium. From experimental verification of the analysis and testing technology, it can provide the theoretical application of Hismelt smelting reduction for Kunming Iron and Steel company.In this paper, the weight loss method is used to the kinetic process for coal reduction of the Huimin high phosphorus iron ore fines with the conditions from40℃to1400℃. Firstly, the TG-DSC curve of smelting reduction process is obtain by the iron ore experiment of Huimin reduction in TGA, and the change or peak of the curve characteristics is then used to explore the chemical reaction rule of various stages in the process of smelting reduction. The whole process is divided into six stages or segments:(1) crystalline water removal stage;(2) hematite pre-reduction stage;(3) melting transition phase;(4) Fe2O3complete conversion segment;(5) Fe3O4transformed to FeO phase;(6) FeO transformed to Fe phase. Secondly, the integration method of the Thermal analysis is employed to build the apparent kinetic model of the main stage from experience mechanism model, including the calculation of the kinetic parameters such as the apparent activation energy, frequency factor and infer the most probable mechanism function (E, A,f(α)). It is found the fact that as the temperature rises, the apparent activation energy of the key stage shows the increasing trend, in the other words, the reaction has the descending order. The longitudinal compensation effect analysis shows that the interdependence and synergy between E and lgA change significantly, the degree of difficulty and the probability of collision with each other significantly affect for all kinds of reaction during the reduction process. The apparent activation energy is got by the derivation and numerical fitting with the distribution function of the temperature and heating rate as follow:as the fitting function curve appears right tailling, the activation energy presents the sparse distribution and fast changes in the high-temperature section, the need is consistent with for energy supply in smelting reduction. After fitting process, the frequency factor expression of the reduction process can be obtained, the inflection point of the curve reveals the number of collisions between the activation of molecular significantly increase during the process in the molten state, it is helpful reduction.The apparent reaction rate (DTG data) of physical quantities in the system is extracted, and it can reflect the whole reduction process and details for comprehensive consideration. The physical quantity can not only reflect the dynamic behavior of the smelting system, but also can reflect the behavior for process of heat and mass transfer. The specific contents are as follows.(1) the phase reconstruction technique of dynamical systems theory is employed to reconstruction analysis, the evolution of the characteristics of the apparent reaction rate reveals the dynamic behavior in the smelting reduction process, the results show that the dynamic behavior of the sequence of a first-order time delay and two-dimensional embedding was accurate for system. it is found that the complex uncertainty smelting system which belongs to the random discrete dynamical system has the chaotic attractor structure, and the structure shows the graphic feature of double ’∞’ it is so-called iron reduction attractor. The points inside ’∞’ within the curve of double ’∞’ characterize that the smelting reduction system comes into the stage of interfacial chemical reaction control; it can be inferred that points outside ’∞’ characterize the system comes into the stage of diffusion-controlled while the points are scattered around the curve characterized the stage of the mixed control. The recurrence plot of DTG sequence, Poincare section, the calculation of the correlation dimension and the largest Lyapunov index calculation process all show that reduction of Huimin iron ore is provided with weak chaos, and the behavior is caused by the reduction of iron, so we determine the chaotic behavior as useful process, strengthen contribute to the smelting.(2) The iron reduction attractor is different from the Lorenz system, Rossler system and logistic map system, and the chaotic behavior is completely different from them, then the equation can not refer to the type. The universal equation calculation results show that there are many factors influencing the DTG, it is difficult to apply the usual differential equations to solve. Thus the predictive modeling methods in the analysis of time series is employed, the locally weighted linear prediction perfonnance reflects the reaction within the system appear the short-range correlation (local) as well as the differences of primary and secondary (weighted), In further the EMD-wAR model can accurately express the frequency distribution characteristics of the energy transfer process (EMD), also reflects that there is the short-range correlation (AR) as well as primary and secondary difference in nature (golded weighted) within the reaction system, the EMD-wAR declined an order of magnitude error than the locally weighted, the establishment of the EMD-wAR-CPSO-SVR model made errors again declined an order of magnitude, the expression of DTG is more accurate, so that it is helpful the establishment of the attractor equation in the reduction process.(3)The power spectral analysis was employed to the DTG signal and it is found the frequency concentrated in the low during energy transfermamtion of the reduction process for pure iron oxide, the ore reduction process is concentrated in the middle and high frequency. It can be appropriate to strengthen heating method in the process of smelting furnace, it will also help in the strengthening of the smelting process.Finally, the reduction experiment with key temperature ranges of high phosphorus iron ore by vertical electric stove were carried out, the reaction process and kinetic behavior characteristics were speculated through the phase morphology changes of various stages in slag, so as to verify the characteristics of the redcution process.更多还原