Mei C., Zhou J., Peng X. Simulation and Optimization of Furnaces and Kilns for Nonferrous Metallurgical Engineering
Подождите немного. Документ загружается.
Introduction
$IJ.FJBOE)POHSPOH$IFO
The furnaces and kilns for nonferrous metallurgical engineering (FKNME) are a big
family. There were few reports on systematic research of FKNME, especially the
accurate and quantitative analysis. In this chapter ideas and analysis methods are
introduced for comprehensive and in-depth understanding of the FKNME system.
1.1 Classification of the Furnaces and Kilns for Nonferrous
Metallurgical Engineering (FKNME)
In most industrial applications and research activities, metals are generally categorized
into two groups: the ferrous metals and the nonferrous metals. Chromium, manganese
and iron are ferrous metals while all the rest are the nonferrous metals. The nonferrous
metals can be further classified by their applications as follows (Zhao, 1992):
a) Heavy metals: copper, lead, zinc, tin.
b) Light metals: aluminum, magnesium, titanium.
c) Precious metals: gold, silver, platinum group.
d) Metals in the iron and steel industry: nickel, chromium, cobalt, molybdenum,
tungsten, vanadium, niobium, tantalum.
e) Metals in the electronic industry: cadmium, gallium, germanium, mercury,
indium, rhenium, selenium, tellurium, manganese.
f) Metals in the nuclear industry: uranium, zirconium, hafnium, cesium,
rubidium, beryllium, rare earth.
g) Metals in the chemical industry: bismuth, antimony, lithium.
The nonferrous metals family is much larger and plays an important role in
industrial activities and people’s daily life. The furnaces and kilns built for
extracting and processing nonferrous metals are generally called FKNME. They
vary a lot in structure and function. Currently up to a hundred different FKNME
Chi Mei and Hongrong Chen
can be named. For the convenience of investigation, the FKNME, are classified as
shown in Table 1.1, by their intended functions, heating sources, heating methods,
working mechanisms, furnace structures and thermal performances.
Table 1.1 A classification of the FKNME (Mei, 2000)
Criterion Examples
Intended functions
Drying furnace (kiln), baking furnace, calciner, heating furnace,
chlorinator, smelter, melting furnace, converting furnace, refining
furnace, heat treatment furnace, reduction furnace, fuming
furnace, volatilizing furnace, distillation furnace, diffusion
furnace, anode furnace, cathode furnace
Heating sources Autothermic furnace, fuel furnace, electrical furnace
Heating methods
Flam furnace, downdraft furnace (kiln), muffle, resistance
furnace, inducting furnace, arc furnace, ore smelting electric arc
furnace, electron bombard furnace, ion furnace, thermal plasma
furnace
Working mechanisms
Fluidized furnace, cyclone furnace, blast furnace, flash furnace,
top blown converter, side blown converter, bottom blown
furnace, bath smelter, hot air recycling furnace, air cushion
furnace, suspension roasting furnace
Furnace structures
Rotary kiln, reverberatory furnace, multiple hearth furnace,
vertical well furnace, crucible furnace, chamber furnace, vertical
distillation kettle, tower redistillation furnace, carbon tube
furnace, tungstate rod furnace, molybdenum wire furnace, bell
furnace, walking beam furnace, carbon particle furnace, muffle
Thermal performances
Simple-structured stoves (such as combustion chamber, air
heating furnace)
Heater (heating furnace, heat treatment furnace, melting furnace,
downdraft kiln, tunnel kiln, etc.)
High temperature reactor (converting furnace, bath smelter,
molten salt electrolysis cell)
1.2 The Thermophysical Processes and Thermal Systems of the
FKNME
Despite the large variety, most of the thermophysical processes taking place in the
kilns and furnaces can be fundamentally disassembled into the following basic
processes:
a) Fuel combustion or thermo-electrical conversion.
b) Confined gas flow, gas-particle multi-phase flow or general fluidized bed flow.
c) Mechanical or electromagnetic movements of materials (loads).
d) Heat and mass transfer.
e) Physical-chemical reactions under high temperature.
1 Introduction
Note that these processes are: on one hand, the purpose of the furnaces (kilns)
are designed for; on the other hand, the joint effects of the other processes listed in
clusters a~d.
The thermal system of the furnaces and kilns generally consists of the main
body and the supporting parts as listed in Table 1.2.
Table 1.2 The thermal system for the furnaces and kilns
Item Components Description
The furnace foundation Concrete
The hearth and refractory wall
Refractory brick and pounding
materials
The insulation Insulation materials
The holes and doors
The shell and external
reinforcement
Steel structure
Main body
Running machinery
Heat supply by fuel-firing or
electrical heating
Fuel preparing and combustion
facilities
Ventilation and exhausting
Voltage and frequency changing
element, air heating device,
vacuum system
Materials charging and discharging
Cooling system
Water-cooling or air-cooling
equipment
Heat recuperation
Heat exchanger, storage, waste
heat steam system
Supporting parts
Monitoring system
Measuring instrument and
control system
Even though separately described as the main body and the supporting parts,
the thermal system should be considered inseparable as for delivering the
performance. A malfunction of any element may result in failure, if not serious
damage, of the whole system. Therefore, the developer of a furnace or kiln
should make sure the main body and the supporting parts are integrately
designed. A researcher should also keep the whole system in mind when
investigating any individual part. This “integrate thinking and systematic
approach” is actually a key rule applicable to the practice of investigations and
design of the FKNME system.
Chi Mei and Hongrong Chen
1.3 A Review of the Methodologies for Designs and Investi-
gations of FKNME
In past years,methodologies to design and investigate the FKNME were greatly
progressed.With development of computer technology,mathematical models and
numerical computations have become more and more important to research of the
FKNME.
1.3.1 Methodologies for design and investigation of FKNME
Methodologies for design and investigation of FKNME can be put into following
categories.
1.3.1.1
Empiric analysis
ü
analog estimation
ü
approximate quantification
Even though this method is considered very “primitive” in collecting and developing
knowledge about the furnaces and kilns, it is still quite efficient in some specific
cases as well as for the SME’s (small-sized and middle-sized enterprises) in less
sizable furnaces or kilns, because it is easy to understand, reliable and easy to use.
Nevertheless, this methodology is quite limited in adapting modern research
techniques and tools for better effectiveness (Mei, 1996).
1.3.1.2
Simulating experiment
ü
similarity analysis
ü
empirical formulation
This method has been developed since the mid 19th century and is still being widely
used nowadays. The theoretical foundation of this method is the Similarity Theory (the
positive law, the reverse law and the ± law), based on which the similitude
experiments can be designed and carried out. The results of these experiments are usually
formulated into dimensionless numbers and empirical equations (Mei, 1996).
The simulating experiments can be designed in two ways. One is called the
similar model experiment, which is used when the model and the real application
share some identical properties, such as the wing profile and high-speed train
contour experiments in the wind tunnels. The other is called the analog model
experiment (or heterogeneous model experiment), which means using a situation
that is more easily observed or measured to analogize another one that is difficult
to observe or measure. An example of it is using electrical phenomenon or
hydraulic phenomenon to simulate heat transfer. This method can be very useful in
optimizations of the furnace (kiln) structure and internal geometry as well as
investigations of the fluid flow and heat (mass) transfer in the furnace (kiln). It
becomes inappropriate, however, when the geometry becomes very complicated or
when temperature, pressure or flow velocity goes too high or when toxic
substances are presented.
1 Introduction
1.3.1.3
Mathematic modeling
ü
numerical analysis
Starting from the end of the 1960s and beginning of the 1970s, mathematic models
and computer program packages have been developed in the western countries to
simulate mono-process representing the magnetic field, temperature field or cell
shell stresses field of the aluminium electrolytic cells. These packages have been
soon used in assisting the engineering design of the cells. Similar works started in
China in the 1980s with numerical studies and optimization of the resistance furnace
and the inner liner structure of the periodic working furnaces (kilns). In recent
decades, more and more FKNME mathematic models have been developed with
higher reliability and better applicability as a result of the remarkable development
of numerical techniques for fluid dynamics, heat transfer and combustion.
1.3.1.4
Mathematic model
ü
hologram simulation
ü
systematic optimization
Based on the accomplishment of the mono-process orientated mathematic modeling
researches, engineers and scientific researchers started to develop more sophisticated
mathematic models for optimizations of the FKNME structures and their operational
conditions. However, focus changed from modeling mono-process to modeling multiple
processes, and approaches also changed from ignoring or simplifying the coupling
effects of the concurrent processes to modeling these effects with extensive details and
high precision. This kind of simulations is called the “hologram simulations”, because
they enable highly truthful and comprehensive reflection of the processes in the
furnaces (kilns). The course of using hologram simulation techniques to find
automatically the operational conditions for the optimized or expected furnace (kiln)
performance is called the “hologram simulation experiment”. The collection of
numerical simulation techniques and tools based on hologram simulation experiments
targeting multi-objective system optimization is called the “numerical modelü
hologram simulationüsystematic optimization” method (MHSO).
1.3.2
The characteristics of the MHSO method
The MHSO method has following characteristics (Mei, 1996; Mei et al., 1999):
a) The procedure of implementing the MHSO methodology includes maily
three steps. Firstly, the studied objects should be understood and measured at
macroscopic level, based on which the mathematic modeling and numerical
simulation can be done to obtain the vector (scalar) fields information at the
microscopic level. Secondly, the determining relationship of the microscopic
variables to the macroscopic performances must be revealed and the influences of
the building structure and operational conditions to the microscopic variables must
be understood. After all these relationships have been well determined, the
Chi Mei and Hongrong Chen
optimization of the macro performance can then be carried out by adjusting the
microscopic variables by modifying the operational conditions and the structure.
b) The research strategy of the MHSO method is first to build up qualitatively
the physical models, then to make an approximate quantitative decision by solving
equation set or by artificial intellectual techniques. This preliminary quantitative
decision is then refined for higher reliability through a series of validating and
correcting processes using information obtained from practice. This strategy
combines the advantages of both qualitative and quantitative approaches.
c) The effectiveness of the MHSO method is obvious. Application of the method
will directly improve the performance of furnaces and kilns which is automatically
verifiable in practice. Therefore, the MHSO method combines the theory-based
investigations with facts and figures obtained in experiments and applications, which
guarantees the practicability of this method on the one hand and allows continuous and
systematic optimization of the furnaces and kilns on the other hand.
Compared to other methods mentioned above, the MHSO method also enables
obtaining information that is more comprehensive, reliable and easily understood.
However, this method is more demanding regarding mathematical models or
artificial intellectual models and requires support of efficient algorithm, advanced
computer capacity and reliable online measuring technologies. Table 1.3 compares
the classic methodologies and modern methodologies used in scientific and
engineering investigations.
Table 1.3 Comparison of modern and classic methodologies used in
scientific and engineering investigations
Comparison
point
Traditional methodologies Modern methodologies
Research
perspectives
The dependency of system
outputs on system inputs
The determination of the “macro
structure” to the “macro
performance”
The dependency of the macro
performance on micro mechanisms
Currently more focus on acquiring
information at microscopic level
Basis
Empiric observation
Statistics
Similarity model experiment
Approximate analysis
Mathematic models based on theories
of computational fluid dynamics,
heat (mass) transfer, combustion,
metallurgical reaction kinetics, and
artificial intelligence, validated by
experimental and practical data
Model style Input-output “black box”
Hybrid “gray model” combining
theoretical and empirical knowledge
Artificial intelligence model
Information
collection
Space-averaged and time-averaged
info
Momentary info at sampling
points
Vector (scalar) fields info (such as fluid
flow field,temperature field,concentration
field and so on)
Dynamic and quasi-dynamic information
Tools
Classic mathematics methods
Computer-based simulation packages
Artificial neural network (auxiliary
analysis integrated system)
1 Introduction
1.4
Models and Modeling for the FKNME
Models and modeling are important to research of the FKNME.Though forms of
FKNME models are greatly different,the modeling consideration and progress
share something in common.
1.4.1
Models for the modern FKNME
From the viewpoint of practical exercises, any equation set, formulation, graph or
data table etc. that reflects the interactions and dependencies among the
functioning variables of a process can be regarded as the mathematic model of this
process. The forms of mathematic models are enormously different due to the
large variety of the ways to establish these models and the different characteristics
of the mathematical physical equations. There is so far no classification of the
existing models. The same model can be named differently depending on the
preference of the users or developers. For example, a country, a region or an
enterprise can use their own names to label their energy consumption model,
pollution model, equipment capacity model, valuable elements recovery model,
equipment service life model etc. Scientific researchers can name models by their
mathematic patterns as linear models, parabolic models, normal distribution
models, sinusoid models, periodic pulse models etc. Models can also be named by
their theoretical foundation, such as the statistical models, mechanism analysis (or
logic) model, system identification model, mechanism analysis-system identification
model, etc. According to how much the processes have been quantitatively
understood, they can also be named as “white box” models, “grey box” models and
“black box” models. For the engineers and researches in the FKNME field, we
propose a more generally applicable classification of the models as shown in Table
1.4 (Mei, 1996; Zuoteng, 1985; Smith, 1982).
In practice, the models in different classes can be jointly used for better
accuracy and effectiveness.
1.4.2
The modeling process
The FKNME modeling process is schematically illustrated in Fig. 1.1.
As indicated in Fig. 1.1, the modeling of the FKNME starts from
comprehensive and in-depth understanding of the system by observing and
measuring its structure, working mechanisms, operation conditions and
performance. Based on these understandings, necessary and reasonable
Chi Mei and Hongrong Chen
assumptions can be made so that the complex reality can be simplified and
modeled by the most representative processes and variables. The reasonably
simplified system is then described using mathematical language, so that the
governing equation set can be established with conservation equations closed
with other equations describing the properties, transfer rates and other
constraining relations. The equation set is eventually resolved through a series of
numerical processes. For those systems that cannot be easily formulated by
mathematical methods, or those involving stochastic process or multiple
processes with strong coupling and (or) lag effects, the artificial intellectual
methods (such as expert system, fuzzy analysis, artificial neural network etc.)
have to be used, together with qualitative and quantitative dynamic models and
judge regulation with “symbol” add “searching” system.(Mei and Zhou,
1991a,b,c; Mei et al., 1994a,b; Peng and Mei, 1996)
The validity of models must be checked before practical use. This is normally
conducted using measurements specifically prepared for model validation or data
collected in productions. The reliability of a mathematical model can only be
proved after repeated checking-and-correcting processes or by building in
self-learning and correcting functions in the model.
Table 1.4 Mathematic models classification for FKNME
Class Description Examples Applicable circumstances
Empirical and
statistical
models
Black box model
Regression equation
Curve fitting
expressions
For easy and brief calculation,
valid specifically within the
range where the data are
collected
Analytic models
with governing
equation(s)
White box model
with clearly under-
stood mechanism
Heat conduction in
solid medium
Velocities of
laminar flow in a
tube
Infrequently used in engineer-
ing practice
Hybrid models
(theoretial-
empirical)
Grey box model
Usual turbulence
models
Heat and mass
transfer models in
fluids
Combustion models
Widely used in engineering
practice
Artificial
intelligence
models
Dynamic
experience model
Expert systems
Fuzzy analysis
model
Artificial neural
network model
Mostly applicable in cases
involving unsteady state
multiple processes with strong
non-linear and (or) strong
coupling effects
1 Introduction
Fig. 1.1 General modeling of nonferrous metallugical furnaces and kilns
References
Mei Chi (1996) Simulation and optimization of the nonferrous metallurgical furnace (in
Chinese). Journal of Chinese Nonferrous Metals, 6(4): 19~28
Mei Chi (2000) Handbook of The Nonferrous Metallurgical Furnace Design (in Chinese).
Metallurgical Industry Press, Beijing
Mei Chi, Peng Xiaoqi, Zhou Jieming (1994a) An intelligence decision support system
(IDSS) on the process of nickel matte smelter. Journal of Central South University of
Technology, 1(1): 14~18
Chi Mei and Hongrong Chen
Mei Chi, Peng Xiaoqi, Zhou Jieming (1994b) Fuzzy and adaptive control model for process
in nickel matte smelting furnace. Tran. of Nonferrous Metals Society of China, 4(3):
9~11
Mei Chi, Ying Zhiyun, Zhou Ping (1999) “Hologram” simulation of the modern furnace
(kiln) (in Chinese). Journal of Central South University of Technology, 30(6): 592~596
Peng Xiaoqi, Mei Chi (1996) An intelligent decision support system (IDSS) in the
operation process of electric furnace for clearing slag. Journal of Central South
University of Technology, 3(2):170~180
Smith J M (1982) Mathematical Model and Numerical Simulation for Engineers and
Researchers. Wang Xingyong etc. translation. Nuclear Energy Press, Beijing
Zhao Tiancong (1992) Handbook of Nonferrous Metals Extraction (in Chinese).
Metallurgical Industry Press, Beijing
Zhou Jieming, Mei Chi (1991) A mathematic model of electric smelting furnace used for
improving design and operation. Extraction and Processing Division Congress’91, TMS:
331~352
Zhou Jieming, Mei Chi (1991) Computer simulation of electro-thermal field in nickel ore
smelting electric arc furnace (in Chinese). Journal of Central South Mining &
Metallurgical Institute, 22(1): 46~53
Zhou Jieming, Mei Chi (1991) Mathematic model and computer simulation of Soderberg
electrodes in electric smelting furnace. Electro Warm International (B), 10(1): 210~215
Zuoteng Chilong (1985) Mathematic Model (in Japanese). Gong Rongzhang etc. translation.
Mechanical industry Press, Beijing