CC BY
113
Structural-matrix models for long product rolling processes: modeling production traceability and forming consumer properties...
Inheritance of technological and organizational measures of previously passed metallurgical conversions as well as follow-up ones takes part in ensuring the required level of consumer properties that will let us predict, monitor and improve the quality of long products.
Created adaptive complex comprises: an information flow scheme (a); new quality indexes of the rolling process (b); methodology to assess and improve the technological competence of personnel (c); methodology of quality control during design, implementation, long products production and improving.
The considered solution of the information model organization correlates well with object-oriented database creation process used in modern automatic process control systems as well as on all CAM systems on all automation levels up to the MES systems and enterprise management systems.
Immutability of the essence of applied mathematical apparatus and the basic principles of its construction to solve technical and technological problems and challenges of quality management makes it possible to adapt and use the developed approach in overlapping technical areas.
References
1. Tulupov O.N. Structure-matrix models for improving the efficiency of rolling grades: Monograph. Magnitogorsk: Nosov Magnitogorsk State Technical University, 2002. 224 p.
2. Tulupov O.N., Ruchinskaya N.A., Moller A.B., Limarev A.S., Lutsenko A.N.Quality management of long products by using rational preventive actions in mill setting Vestnik Magnitogorskogo gosudarstvennogo tehnicheskogo universiteta im. Gl Nosova. [Vestnik of Nosov Magnitogorsk State Technical University]. 2007, no 4, pp. 73-80.
3. Lewandowski S.A. Effectiveness enhancement of section mills by the quality management model improving: PhD Dissertation. Magnitogorsk, 2006.
4. Alekseev A.M., Loginov V.G., Zaitsev A.A., Tulupov O.N., Moller A.B., Rashnikov S.F., Morozov A.A. A method of rod rolling. Patent RUS no. 2148443, 1998.
5. Tulupov O.N., Moller A.B., Kinzin D.I., Lewandowskiy S.A., Limarev A.S., Zavyalov K.A., Richkov S.S., Loginova I.V., Unruh C. J., Nowitskiy R.V. Resistance increasing of the 370 mill roller by cooling system improving. Research report («Magnitogorsk Iron and Steel Works»).
6. Kinzin D.I., Kalugina O.B. Evaluation of values impact of the distortion structure on widening in rolled section Vestnik Magnitogorskogo gosudarstvennogo tehnicheskogo universiteta im. G.I. Nosova. [Vestnik of Nosov Magnitogorsk State Technical University]. 2011, no 4, pp. 21-23.
7. Nalivayko A.V., Steblov A.B., Tulupov O.N., Rychkov S.S. Issledovaniye urovnya mekhanicheskikh svoystv armatury klassa A500S s tsel'yu otsenki vliyaniya osobennostey tekhnologii na pokazateli kachestva. Vestnik Magnitogorskogo gosudarstvennogo tehnicheskogo universiteta im. G.I. Nosova. [Vestnik of Nosov Magnitogorsk State Technical University]. 2010, no. 2, pp. 69-73.
8. Moller A.B., Tulupov O.N., Kinzin D.I., Levandovskiy S.A., Limarev A.S., Zav'yalov K.A., Nazarov D.V., Novitskiy R.V., Rychkov S .S., Loginova I.V. Razrabotka i opytno-promyshlennoye oprobovaniye ekspluatatsii bandazhi-rovannykh valkov v predchistovoy gruppe kletey stana 170 STS OAO «MMK» [Designing and experiment industrial testing of tire rolls running at millstand 170 of «MMK»]. Research report («Magnitogorsk Iron and Steel Works»).
9. Ruchinskaya N.A., Zaytsev O.YU., Tulupov O.N., Lutsenko A.N. Printsipy sozdaniya informatsionnykh modeley upravleniya kachestvom sortovogo prokata. [The principles of creating long products quality management information models]. Proizvodstvo prokata. [Rolling processes]. 2007, no. 8, pp. 33-41.
Logunova O.S., Matsko I.I., Posochov I.A.
INTEGRATED SYSTEM STRUCTURE OF INTELLIGENT MANAGEMENT SUPPORT OF MULTISTAGE METALLURGICAL PROCESSES
Abstract. The necessity to invent a universal technology for creating the intelligent management support of multistage processes, which is able to interface process variables between local loops at each stage of production is determined. The integrated system structure of intelligent management support of multistage metallurgical processes and production stage models are suggested. The improved method of billet macrostructure assessment and block designing technology of intelligent management support of continuous casting billet production are described.
Keywords: control systems, intelligent support, multistage metallurgical processes, continuous cast billet, billet macrostructure assessment.
The actuality of the study
Modern Industries require new systems for multistage process management. These requirements are due to the new priority trends in accordance with Russian state policy. One of these trends is the development of information and telecommunication technologies, which are integrated in automated control systems (ACS) of large industrial enterprise production. The use of new ACS modules for multistage manufacturing processes facilitates unit performance increasing and provides reduced quantity of low-quality products.
From the management point of view, multistage technology of steel products is a complicated process.
Such technologies require the system allowing to monitor output product quality in on-line and providing intelligent decision support of process control.
In developing and implementing new modules, which enlarge already existing ACS, the necessity to use graphic information obtained during quality estimation procedure of finished products and semi-finished products emerges.
The effectiveness of using graphic information and decision-making in ACS production is illustrated by theory and practice.
Methods of image obtaining, processing and segmentation can be found in foreign and Russian scientific pub-
lications. Mathematical theory development in improving and graphical information segmentation was defined in scientific papers [1-3]. An overview on the topic of decision-making on basis of tree-type structure was presented by Quinlan J.R., Janikow C.Z., Hastie T., Tibshirani R., Friedman J., Berestneva O.G. and others. The practical application of fuzzy set theory and fuzzy logic refers to Zadeh L.A., Esposito F., Malebra D., Semeraro G., J.J. Dyulichevoy and others.
However, despite the studies carried out and large number of publications in the field of automated control systems (ACS) in steel production [4-7], the following issues are of current interest:
- the lack of automated systems permitting to ensure technological processes control, based on output product quality information;
- the lack of techniques of graphic information collection and processing about the quality of steel products using low-contrast images with irregular shaped elements;
- the lack of application packages for intelligent decision support in automated control systems (ACS) of multistage production, based on adaptable fuzzy trees with dynamic structure, taking into account the values of attributive quality characteristics of products.
In the current circumstances, there is a need to develop a universal technology of creating the intelligent support system of multistage process management, which is able to interface process variables between local loops at each stage of production.
The structure of research object
Fig. 1 illustrates the diagram of a manufacturing process with N stages. The following notations are displayed in Fig. 1: {Zi} is a task vector for technological parameters values of the i-th process; {Ri} is a decision vector of parameters values changing of the i-th process; {Mi} is a vector of estimation (indicators) of product or semiproduct quality, resulting in redistribution at the N selected stages, Z is a value of final parameters for product assigned position.
A typical example of a multistage manufacturing process is the production of continuous cast billet, with three main stages: steel melting in alternating-current (AC) electric arc furnace (EAF), steel processing in a ladle furnace unit (LFU) and continuous steel casting in billet continuous casting machines (BCCM).
An illustrative diagram of a multistage production for the chosen technology, where N = 3, is represented in Fig. 2. All input and output parameters, shown in Fig. 2, become of specific physical and technologic significance (see Table).
Values, which are transmitted between production stages and determine the operation mode choice of the next step unit, are highlighted in Table. The presence of binding parameters permits to organize the integrated management of all production stages.
The list of major task vectors for continuous cast billet production stages
Vector set Vector coordinates Physical or technological significance of coordinates
{*} d charge materials ratio (crowbar / pig iron), %
{Ci} percentage of chemical elements in steel at EAF exit,%
ti arc time under the current, min
I1 current strength, MA
Ui current voltage, MV
{li} elements proportion of non-metallic charge, %
V} flow-rate of solid oxidizer additives kg
g} flow-rate of gas oxygen, m3 / h
f} discharge intensity of ferroalloys, kg / t
T1 metal temperature at EAF exit, °C
{Z2} T1 metal temperature at EAF exit, °C
C percentage of chemical elements in steel at EAF exit, %
t2 arc time under the current, min
I2 current strength, MA
U2 current voltage, MV
g} argon gas flow rate m3 / h
f} discharge intensity of ferroalloys, kg / t
T2 metal temperature in steel teeming ladle at LFU exit, °C
{Z3} T2 metal temperature in steel teeming ladle at LFU exit, °C
T3 metal temperature in tundish ladle at BCCM exit, °C
{Qe} percentage of chemical elements in steel teeming ladle at EAF exit, %
v billet drawing rate, m / min
F} water flow rate for secondary cooling zones BCCM, m3 / h
D air flow rate for secondary cooling zones BCCM, m3 / h
w crystallizer oscillation frequency
F water flow rate on crystallizer m3 / h
{M3} task aimed to billet macrostructure evaluation, grade
{Z} {MB} task aimed to billet macrostructure evaluation, grade
{c4 percentage of chemical elements in tundish ladle BCCM, (last test), %
{I} Information about template selection for quality evaluation
{M} {0} actual results of quality evaluation, grade
{h} Information about template selection for quality evaluation
Fig. 1. Diagram of multistage production with integrated system of intelligent management support
Fig. 2. Diagram of multi-stage continuous cast billet production
The structural model of the i-th production stage
The block of each stage (Fig. 1) has a complex structure, which comprises: mathematical process models, visualization module of expected result, current process monitoring module, decision-making block, systems of local loops of process variable control (technologic unit or its field ). The structural diagram of each stage can be unitized according to the diagram shown in Fig. 3.
The following notations are introduced in Fig. 3: {RiM} is a decision vector leading to the i-th process parameter points changing, as a result of modelling; { AZi } is a correction vector for the task point of process control; { Zient } is a task vector for the system of local loops of control object; { Ziex } is a vector of technologic parameter values transmitting to the control object; { Zi+1} is a vector of technologic parameters, obtained after the i-th stage production ending and transmitting to the stage i +1 as a task.
Decision-making blocks, which can be implemented using different technologies, both in automated and automatic modes are represented in Fig. 1 and Fig. 3. Currently, decision-making technologies based on the classification with the help of neural networks, fuzzy logic and treelike structures are the most widespread [4, 8, 9]. A combination of many tree-like structures and fuzzy membership functions about the received quality indicator level is one of the options, taking into account the information fuzzi-ness and taxonomic criterion of product quality. Taking
into account plenty quality indicators and production process dynamism, an adaptive forest of dynamic structure fuzzy trees is formed. Each tree adaptability permits to retrain the decision tree in changing the technologic process running and response to the situations, unspecified in the decision tree. Dynamism allows to change the relevance level of process control variables after a set of tree adaptations towards providing shorter decision branches, reducing the parameters, requiring correction.
View module (Fig. 3) is designed to exhibit results of modelling, and its construction is a separate scientific problem in the field of information processing. The module is optional and it is not included in many process control systems because of implementation complexity in mapping results in real time.
The block «System of local loops» also has a complex structure and can combine M subsystems, which are responsible for process variable selection and stabilization. The peculiarity of the suggested solution is the integrated result examination in changing values of many process parameters not only at the selected i-th stage but also between stages. Therefore, vector representation of input and output parameters is shown between the diagram blocks, shown in Fig. 1 and Fig. 3.
Fig. 4 illustrates the example of structural scheme of continuous steel casting stage applicable for arc-furnace steelmaking plant of OJSC «Magnitogorsk Iron and Steel Works».
Fig. 3. Structural diagram of the i-th production stage
{DZ3}
System of local loops {Z3ex} BCCM
Fig. 4. Structural scheme of continuous steel casting stage
As shown in Fig. 4, a signal of suggested billet quality {R3} is received from the intelligent support block. Binding components of vector {Z3} enter the modeling block of continuous casting of steel. A system of sub-models is represented in this block, including the model of identification of heat transfer coefficient from the billet surface, the model of the billet thermal state, the model of billet damage forming, the model of billet quality forecasting, etc. The simulation data {^3m} enter consistently the rendering module, implemented on the SCADA platform, and the decision-making block. The use of adaptive fuzzy trees with dynamic structure is one of the ways of decision-making block building aimed to adjust the values of process variables. Melting mass; metal temperature after EAF discharging; metal temperature in pony ladle; oxidation of metal; steel casting rate; the content of basic chemical elements in steel are defined as the linguistic variables for the decision making tree of the possible damage origin.
A method of billet macrostructure evaluation
Speaking about steel product quality, firstly macro-and microstructures values are noticed, and these values are foundational for such mechanical properties as strength, formability, wearability. Discontinuity flaw geometric dimensions, obstacles quantity, density from metal distribution in given volume are the characteristics of metal macro- and microstructures. The determination of the degree of discontinuities flaw extension is generally defined by the appropriate field and state standards such as OST 4.14.73 or OST 14-1-235-91. The process of expert knowledge base formation, in order to assess metal-lurgic product quality, is the most time-consuming. Hitherto, information acquisition and processing are carried out by traditional methods, based on product visual in-
spection results. With modern computing facilities it becomes possible to develop automated system of metal macro- and microstructure assessment. However, the development of a decisive rule base for decision-making is possible only on the basis of «practical» expert knowledge. Such knowledge processing leads to the input of membership function of fuzzy sets and rules of their use. Intelligent support system of billet continuous cast production at arc-furnace plant of the OJSC Magnitogorsk Iron and Steel Works (MMK) was built based on technology, developed by the authors [9]. Points of the internal macro-damage extension have been selected as product quality indicators. Macro-damage rating is carried out on crosscut templates and sulphuric marks. The following types of damage such as axial looseness, dotty obstacle and cracks, perpendicular to the billet edges were examined. Length, width, strength and ratio of total damage area to template size are the characteristics of damages. According to the OST 14-1-235-91 membership functions for evaluating each type of damage were classified in tabulating.
Algorithm S4.5 and algorithm of decision-making fault reducing in fuzzy tree using were selected as the basis of classification algorithm. In order to construct a fuzzy tree forest, which allows to take into account predicable quality classification criteria from the tabulation, linguistic variables to describe damage parameters such as width, length, relative length, relative width, area, maximum area, relative area, length-width ratio, width-length ratio, strength and section attachment have been introduced. Tree type structure building results in order to assess the axial looseness of continuous casting billet are shown in Fig. 5. Full forest of fuzzy trees contains from 800 to 2,000 leaves. Branches for damage, which has an average size, are allocated in Fig. 5 and its part is located in
section 3. If 10% of damage is in section 3, then template attachment to target class (qualitative template) is 0.334 that corresponds to 2.5 points according to the OST.
Fig. 5. A segment of the classification tree of internal macro damages of continuous casting billet
Building block method of intelligent management support of continuous cast billet production
In creating the integrated system of intelligent management support of multi-stage processes it is necessary to establish cause-and-effect relationships between product quality indexes and process variable values, which need to be built at the stage of expert knowledge formalization.
One of the ways to formalize expert knowledge is adaptive fuzzy tree with dynamic structure (Fig. 6).
Melting mass; metal temperature after EAF discharging; metal temperature after secondary treatment; metal temperature in pony ladle; oxidation of metal; steel casting rate; the content of the basic chemical elements in steel are defined as the linguistic variables for the decision making tree of the possible damage origin.
Taking into account that the built tree leaves are presented in Fig. 6, the decision uses only two branches marked with a dotted line. The solution pertains to the upper branch per 0.14 and to the lower branch per 0.86. It can be defined for this solution that the resulting billet has 0,357 grade of membership to a class of usual quality billets, and that corresponds to 2.5 points according to the OST.
The billet is qualitative if the point is less than 2. Consequently, it is necessary to modify existing solutions. According to the developed algorithm, the control parameter change is performed. It means to reduce fuel cooler flow from 444 to 438
l / min, in this case billet quality assessment is reduced to 2 points, and that is sufficient for its ordinary quality classification.
To 2,3 m/min yes: 104,89; no: 23,38
Water flow on SCZ
More 2,3 to 4,m/min : yes: 97,86; no: 57 j
Water flow on SCZ
From 7 to 215 l/min yes: 36,76; no: 5,321
More215 to 431 l/min yes: 62,59; no: 5,43
From 431 to 654 l/min yes: 5,54; no: 12,62
From 7 to 215 l/min yes: 6,84; no: 6,23
More 215 to 431 l/min:
j yes: 77,81; no: 5,05 j
s s ^ From 431to 654l/min) I yes: 13,2; no: 45,71 "
From 7 to 215 l/min yes: 8,5; no: 16,69
More 215 to 431 l/min yes: 11,13; no: 2,15
From 431 to 654 l/min yes: 22,9; no: 2,47
Fig. 6. Moving path fraction in decision-making based of fuzzy tree structure in order to forecast continuous casting billet quality
Conclusion
Thus, granting modern decisionmaking methods it is possible to develop existing production control systems for their intellectualization. The use of an integrated system of intelligent management support of multistage metallurgical processes permits to carry out multiple factor analysis. Herewith, both well-known deterministic and statistical modelling process methods and unitized way of decision making, based on «practical» expert knowledge, are used in this analysis.
References
1. Shapiro. L., Stockman G. Computer Vision. Moscow: BI-NOM. Knowledge laborator, 2006, 752 p.
2. Gonzales R.C. and Woods R.E. Tsyfrovaya obrabotka izobrazheniy [Digital Image Processing], Tekhnosfera, Moscow, Russia, 2005, 1072 p.
3. Pratt W.K. Digital Image Processing. New York: Wiley, 1978. Translated under the title Tsifrovaya obrabotka izobrazhenii, Moscow: Mir, 1982, 312 p.
4. Ryabchikov M.Y., Parsunkin B.N., Andreev S.M., Logunova O.S., Ryabchikova E.S., Golov-ko N.A., Polko P.G. Achieving maximum perfor-
mance optimized ore grinding process in utilizing fuzzy extreme control principles. Vestnik Magnitogorskogo gosudarstvennogo tehnicheskogo universiteta im. G.I. Nosova. [Vestnik of Nosov Magnitogorsk State Technical University]. 2011. no. 2, pp. 5-9.
5. Tutarova V.D., Logunova O.S. Analysis of the surface temperature of continuosly cast ingot beyond the zones of air cooling. Steel in Translation. 1998. no. 8. pp. 21-23.
6. Logunova O.S., Matsko I.I., Safonov D.S. Modelling of the thermal state of the infinitely extended body with the dynamically changing boundary conditions of the third kind. Bulletin of the South Ural State University. Series:
Mathematical modeling and programming. 2012, no. 27, pp. 74-85.
7. Logunova O.S. Internal-defect formation and the thermal state of continuous-cast billet. Steel in Translation. 2008, vol. 38, no. 10, pp. 849-852.
8. Golovko N.A., Logunova O.S., Parsunkin B.N., Andreev S.M. Adaptive system of automatic control of stochastic nonlinear processes. Scientific Review, 2013, no. 1, pp. 166-170.
9. Matsko 1.1., Logunova O.S., Pavlov V.V., Mai^o O.C. Adaptive fuzzy decision tree with dynamic structure for automatic process control system of continuous-cast billet production. IOSR Journal of Engineering. 2012, vol. 2, no. 8, pp. 53-55.
Shatokhin I.M., Bigeev V.A., Shaymardanov K.R., Manashev I.R.
INVESTIGATION OF COMBUSTION IN TITANIUM-FERROSILICON SYSTEM
Abstract. Results of self-sustaining combustion process in the titanium-ferrosilicon system investigations are presented. These data were used for experimental-industrial technology developing of production ferro silico titanium with high titanium content for steel alloying.
Keywords: titanium, titanium containing steel, ferrotitanium, ferrosilicotitanium, titanium ferrosilicide, self-propagating high-temperature synthesis, combustion rate, combustion temperature.
At present, titanium is actively used for alloying of wide assortment of steel thanks to its specific properties. Titanium is used for modern HSLA-steels, pipe-steels, stainless steels and others. Mainly, the mechanism of its influence on steel quality is associated with formation of titanium carbides, nitrides and carbonitrides in steel.
Ferrotitanium is usually used for steel alloying. Depending on the method of production, ferrotitanium can be with high (~70% Ti) and low (<40% Ti) titanium content. High-grade ferrotitanium is usually obtained by melting titanium-containing waste in induction furnaces. Low-grade ferrotitanium is produced by recovery from ilmenite in special melting aggregates. Ilmenite concentrate, iron ore, aluminum powder, ferrosilicon and lime are used as raw materials. Sometimes, when high-purity steel grades are produced, vacuum-melted ferrotitanium is used. Standard ferrotitanium contains considerable amount of impurities (such as non-ferrous metals, nitrogen, oxygen, hydrogen, carbon, sulfur, phosphorus), which come to alloy from raw materials and atmosphere during production. Moreover, assimilation degree of titanium from ferrotitanium remains low.
That is why, full or partial replacement of ferrotitani-um by alternative titanium containing alloys is actual. This problem can be solved by creating of complex master-alloys, which contain titanium as basic element and high-level elements, such as Si, Al, Ca, etc. It is supposed, that these elements will protect titanium against oxidation, because they are strong deoxidizers. In this way, titanium assimilation will be higher.
Thus, to be most effective, new alloy should meet the following requirements:
- high titanium content;
- low impurities content;
- the presence of elements with high affinity to oxygen.
This alloy will give high and stable titanium assimilation, will allow to produce steels with narrow titanium
limits and will reduce impurities content in metal. The most economical alternative to ferrotitanium can be ferro silico titanium.
Usually ferro silico titanium is obtained by melting titanium, metallic silicon and low-carbon steel in induction furnace or by recovery from ilmenite ore.
But it is impossible to produce ferro silico titanium with high titanium content (more than 30% Ti) by furnace methods, because of high melting point of titanium sili-cides and strong liquation during crystallization. Moreover, alloy, obtained by furnace methods, has higher concentration of nitrogen, oxygen and hydrogen, penetrating into the melt from the atmosphere. So, it is important to use fundamentally different method of titanium ferrosili-cide obtaining that enables high product recovery with low impurities content with the lowest electricity costs. Such method is self-propagating high-temperature synthesis (SHS). When SHS is applied, no traditional furnaces are used. The process is carried out in special SHS-reactors at the atmosphere of inert gas or vacuum. At the combustion synthesis, as well as at the traditional metallo-thermic process, energy source is the heat of chemical reactions. But unlike metallothermy, SHS is slagless.
The Ti-Si system has 5 silicides: TiSi, TiSi2, Ti5Si3, Ti3Si4 h Ti3Si. Ti5Si3 has the highest melting point. Physical and chemical properties of titanium silicides are presented in Table 1. The formation of titanium sili-cides occurs with great heat release, so that adiabatic combustion temperature is high. Regularities of interaction of titanium and silicon are studied in detail in works [3, 4]. It was shown, that combustion in powder compound of titanium and silicon can be implemented in a wide range of parameters changing such as the components ratio, powders dispersion, etc. It can be expected, that heat release during chemical interaction in ternary system Ti-Si-Fe will be enough to carry out the process in self-propagating mode.
