Wugang Ezhou Pellet Plant Chain Grate Machine–Development and Application of an Expert System for Optimizing Oxidized Pellet Production in Rotary Kiln
Fan Xiaohui1, Shu Fanghua2, Yang Guiming1, Chen Xuling1, Chen Jianhua2, Huang Hanchang2, Zhen Cailing2, Luo Fuhui2, Li Xi1
( 1. School of Resource Processing and Bioengineering, Central South University, Hunan Changsha 410083; 2. Wugang Ezhou Pellet Plant, Hubei Wuhan 430000)
Abstract: The complexity of the process, frequent fluctuations in raw materials, and differences in operator experience pose significant challenges to the stability and optimized control of the chain grate machine – rotary kiln oxidized pellet production. This paper establishes process simulation models and thermal balance models for the chain grate machine, rotary kiln, and annular cooler based on the heat transfer and mass transfer theory of oxidized pellet thermal processes, achieving transparency in production status; utilizing data mining technology combined with expert knowledge, an expert system for optimizing chain grate machine – rotary kiln oxidized pellet production was developed. The system has been in operation at Wugang Ezhou Pellet Plant since October 2013, and results show that the system interface is user-friendly, practical, and the model computation accuracy exceeds 90%, with operational guidance reaching expert levels on-site.
Keywords: Chain Grate Machine – Rotary Kiln; Oxidized Pellets; Mathematical Model; Expert System
1 Introduction
The design capacity of the Wugang Mining Company Ezhou Pellet Plant chain grate machine – rotary kiln is 5 million tons of acidic pellet ore per year, and it commenced operation in 2006, with the single-line production capacity ranking first in the country. The process flow is shown in Figure 1: Raw materials are dried, ground with high-pressure roller mills, and mixed with bentonite for ball preparation; qualified green balls enter the chain grate machine for wind drying, suction drying, first-stage preheating, and second-stage preheating; preheated balls enter the rotary kiln for roasting, with coal powder and natural gas as fuels, using axial and swirl air for secondary combustion; roasted balls enter the annular cooler for three-stage cooling, with the hot air from the cooler returning to the chain grate machine and rotary kiln for waste heat utilization.
Although the basic detection level of the pellet plant is relatively high, achieving monitoring of major parameters and control of some parameters, the recent rise in raw material prices and frequent fluctuations in on-site materials have raised higher requirements for the rationality and reliability of process operations. Therefore, this paper introduces mathematical models and artificial intelligence technology into the optimization of the pellet production process, developing an expert system for optimizing chain grate machine – rotary kiln oxidized pellet production. When raw material conditions change, it can effectively and accurately adjust operational parameters accordingly, thereby stabilizing production and optimizing the quality indicators of pellet ore.
2 Main Structure of the System
The main structure of the system is shown in Figure 2. Given the characteristics of significant fluctuations and large time delays in actual production, this system employs a combination of mathematical models under stable conditions, intelligent control under fluctuating conditions, optimized operation, and abnormal condition diagnosis to improve operational levels and ensure efficient and stable production.
3 Functional Modules of the System
3. 1 Data Acquisition
On-site production data is divided into mixing – batching and thermal parts, with data transmitted to the central control room using different methods. The mixing – batching part communicates data with PLC through I/O Server, with the operator machine and server on the same computer. Since the central control room engineer’s station is configured with InSQL and SQLServer, the data acquisition subsystem directly accesses the Runtime database of SQLServer remotely, saving the access results to the local database. The thermal part communicates data with PLC through DA Server, with the operator machine and server on different computers. The data acquisition subsystem uses OPC technology to access DA Server directly and saves the access results to the local database. The methods of accessing on-site data and the connection method of the production optimization expert system are shown in Figure 3.
3. 2 Process Simulation
This system establishes mathematical models of the thermal processes for the chain grate machine, rotary kiln, and annular cooler based on heat transfer – mass transfer theory, selecting suitable model parameters according to on-site equipment and production conditions, making the production process transparent. Taking the chain grate machine as an example, the green balls undergo physical and chemical phenomena such as temperature increase, moisture evaporation, and magnetite oxidation during thermal exchange with high-temperature gases［1］. The heat balance theory of gas-solid two phases can be used to calculate the temperature distribution of the material layer, where Equation (1) is the gas heat balance equation, and Equation (2) is the pellet heat balance equation.
Where: G is the gas flow rate at the surface of the pellet, kg/(m2·s); ρb is the weight of the pellet per unit volume; Cg is the specific heat capacity of the gas at constant pressure, J/(kg·K); Cp is the specific heat capacity of the pellet at constant pressure, J/(kg·K); Tg is the gas temperature, K; Tp is the pellet temperature, K; z is the height of the material layer, m; t is the time of pellet movement, s; h is the effective convective heat transfer coefficient, J/(m2·s·K); A is the external surface area of the pellet per unit volume, m2/m3; Rw(Rcd) is the moisture evaporation (condensation) rate, kg/s; ∆Hw is the enthalpy of moisture evaporation/condensation; Rm is the magnetite oxidation rate, kg/s; ∆Hm is the enthalpy of the magnetite oxidation reaction; λ is the heat distribution coefficient.
The rate of moisture evaporation can be divided into constant speed and decreasing speed stages based on the moisture content of the pellets, as shown in Equations (3) and (4); the rate of magnetite oxidation uses the unreacted core model, as shown in Equation (5).
Pellet moisture≥5%when
Where: α, β, γ are adjustment coefficients; kw is the rate constant for moisture diffusion, m/s; WeH2O is the water vapor content at the gas-solid interface equilibrium, kg/m3; WH2O is the actual water vapor content in the gas phase, kg/m3; DeH2O is the effective diffusion coefficient of moisture, m2/s; εP is the porosity of the pellet; r is the initial radius of the pellet; rw is the radius of the wetted core, m.
The solution of the above equations adopts the finite element difference method, that is, the chain grate machine is meshed, and the equations (1) to (5) are applied to each mesh for computation［2，3］. The calculation results of the previous mesh serve as the initial conditions for the subsequent mesh, and when the production fluctuations are small and the system is close to equilibrium, the computation is initiated, with the results displayed in intuitive forms such as two-dimensional colored images and one-dimensional curves.
3. 3 Material and Heat Balance
In-depth research on the material and heat balance of the process is of positive significance for assessing production status and determining reasonable thermal regimes. This paper conducts heat balance calculations for the three main thermal devices: chain grate machine, rotary kiln, and annular cooler, and then performs material and heat balance calculations for the entire process. The income and expenditure items of the material balance use pre-processed on-site detection data (such as feed amount, qualified green ball amount), fully considering the changes in production cycles and the time series characteristics of the data; the income and expenditure items of the heat balance use detection data (such as wind box temperature, fan flow) and process simulation calculation results (such as the temperature of the pellets in the PH section, the oxidation rate of the pellets in the PH section). The results of the material and heat balance are displayed in tabular form.
3. 4 Expert Control［4～6］
The expert system applies artificial intelligence and computer technology to perform reasoning and judgment based on the knowledge and experience provided by domain experts, possessing strong capabilities for solving complex practical problems. This system combines the model calculation results and the changes in detection parameters, adopting expert rules to control the chain grate machine’s PH section grate plate temperature and the rotary kiln’s head discharge temperature as the control core, assisting in the judgment of UDD wind box temperature, TPH hood temperature, Zone1 hood temperature, Zone2 hood temperature, and Zone3 hood temperature states, providing adjustment suggestions for major operational variables such as machine speed, fan opening, and coal injection amount. Taking the chain grate machine as an example, its control logic sequence is shown in Figure 4.
The expert rules are obtained from expert knowledge, on-site operational experience, parameter analysis of mathematical models, and historical data mining, stored in the local database. Among them, for continuous numerical production data, the mining technology mainly adopts clustering analysis and binning-association rules, using Clementine 12.0 as the mining platform, with the basic mining process shown in Figure 5.
4 Development and Application of the System
Considering the process characteristics and on-site automation levels, the main modules developed include the data acquisition module, chain grate machine module, rotary kiln module, and annular cooler module, divided into data acquisition subsystem and production process optimization subsystem.
4. 1 Chain Grate Machine Module
The functions of this module include: online display of important production information, online calculation of temperature distribution in the chain grate machine material layer, online calculation of moisture distribution in the drying section, online calculation of pellet oxidation rate, online calculation of material layer permeability, heat balance calculation for the chain grate machine, reminders of operational status and adjustment suggestions, etc. (see Figure 6). Important production information mainly includes: wind box temperature and pressure, hood temperature and pressure, fan flow, chain grate machine speed, material layer height. After preprocessing, the production data serves as input for mathematical models such as heat transfer and mass transfer, with calculation results displayed in two-dimensional colored images, one-dimensional curves, and two-dimensional tables.
The triggering or activation of expert rules mainly depends on the judgment of production status, which changes with production plans and raw material properties (e.g., under low processing volume, a lower wind box temperature is considered normal; under high processing volume, a higher wind box temperature is needed to maintain normal status). Therefore, the module has added state parameter setting buttons (as shown in Figure 7), allowing users to input and modify operations, enabling operators to adjust the state judgment standards based on processing volume and raw material properties in a timely manner, enhancing the applicability and expandability of the optimization system.
4. 2 Rotary Kiln and Annular Cooler Modules
This module is similar to the chain grate machine module, featuring online display of production data, online calculation of temperature fields in the rotary kiln, online calculation of temperature distribution in the annular cooler, heat balance calculations for the rotary kiln, heat balance calculations for the annular cooler, reminders of operational status and adjustment suggestions, etc. The model calculation results make the production process transparent, and the system’s rich graphical interface helps operators accurately grasp the current production status, while expert operation suggestions provide important decision support and theoretical guidance for technicians in production adjustments.
4. 3 System Application
The production optimization expert system was debugged and applied at Wugang Mining Company’s Ezhou Pellet Plant in October 2013, showing that the developed system has a user-friendly interface, good practicality, high accuracy of mathematical model calculation results during stable production, and high consistency of operational suggestions provided by the system with the actual operational direction of technicians in the central control room. The system automatically provides guidance based on the characteristics of iron-containing raw materials, production plans, and production parameters, optimizing production operations, ensuring efficient and stable production of iron ore oxidized pellets. Since its operation, it has greatly facilitated the management and operation of on-site technical personnel, reduced the labor intensity of operators, and achieved good usage effects and economic benefits.
References
［1］Barati M. Dynamic simulation of pellet induration process in straight-grate system［J］. International Journal of Minerals Processing，2008，89( 1 － 4) : 30 － 39.
［2］Thurlby J. A. , Batterham R. J. , Turner R. E. Development and validation of a mathematical model for the moving grate induration of iron ore pellets［J］. International Journal of Minerals Processing，1979，6( 1) : 43 － 63.
［3］Wang Yi, Fan Xiao-hui, Chen Xu-ling. Mathematical models and expert system for grate-kiln process of iron ore oxide pellet production ( Part I ) : Mathematical models of grate process［J］．Journal of Central South University，2012，19( 4) : 1092 －1097.
［4］Wang Yi. Research on the simulation model and control guidance expert system of iron ore oxidized pellet chain grate machine – rotary kiln［D］. Changsha: Central South University, 2012.
［5］Fan Xiao-hui, Wang Yi, Chen Xu-ling. Mathematical models and expert system for grate-kiln process of iron ore oxide pellet production ( Part II) : Rotary kiln process control［J］．Journal of Central South University，2012，19( 6) : 1724 － 1727.
［6］Xiaohui Fan, Yi Wang, Xuling Chen. Expert system for grate-kiln pellet production based on mathematical models of temperature field［C］．Proceedings of TMS［A］．Orlando，2012: 165 － 173.
