Development and Implementation of Decision Support Systems for Blast Smelting Control in the Conditions of PrJSC “Kamet-Steel”

Abstract

This article presents a description of three decision support systems (DSS) in the mode of an adviser to the technological personnel of blast furnaces (BF), which were implemented by the Iron and Steel Institute of Z.I. Nekrasov (Dnipro, Ukraine) or underwent pilot testing as part of the automated control system of the BF shop of PrJSC “Kamet-steel” (Kamianske, Ukraine). The first DSS for managing the thermal state was implemented in 2021; it includes the entire list of information necessary for personnel in a convenient and compact form, generates recommendations in case of technology deviations, and, in the case of incorrect actions by the personnel, signals the need for correct actions. The main recommendations from the DSS are to correct the raceway adiabatic flame temperature, coke consumption when its characteristics are specified in (indicators of strength and abrasion, fractional composition, humidity, ash and sulfur), and ore load change. Using the system allows both reducing the specific coke consumption and preventing unplanned downtime. The second DSS for controlling the distribution of fuel additives over air tuyeres is based on information on thermal loads determined on water-cooled elements of tuyere tools. The main recommendations from the DSS are to adjust the amount of injected pulverized coal fuel on individual tuyeres in order to ensure a uniform distribution of the raceway adiabatic flame temperature around the circumference of the BF and, as a result, the energy efficiency of BF smelting. The third DSS for adjusting the parameters of the charging mode is based on information from the means of controlling the temperatures of the gas flow above the surface of the charge in the BF. The functioning of this DSS is based on determining the reference curves for the distribution of the gas flow along the BF radii, corresponding to the minimum consumption of coke and maximum productivity, and on the search for solutions by direct and iterative optimization methods, which allow one, by adjusting the charging parameters, to ensure a rational distribution of charge materials and gas flow in the BF.

1. Introduction

Globally, the practice of automation and information support of blast furnaces (BF) is based on the maximum acceptable level of obtaining operational and reliable information by technological personnel under the provision of recommendations generated by computer systems in accordance with knowledge bases based on technological experience, as well as using mathematical and statistical models and information by means of automated control. Such computer systems are commonly referred to as decision support systems (DSS). An analysis of the literature data reveals a variety of both the use of mathematical models, model systems, expert modules, and monitoring systems [1,2,3,4,5,6,7] for decision support, as well as complex DSS in BF production [7,8,9,10,11].

It should be noted that the automation and information support of BFs in Ukraine, until recently, were inferior to the world leaders in iron producers. Basically, information support consisted of visualization for technological personnel of changes in time of the main controlled parameters. These parameters include top gas temperature; temperature of peripheral gases; blast pressure, top gas pressure, and static pressure at the level of the middle of the shaft; temperature, flow rate, and humidity of the total blast supplied to the BF; oxygen content in the blast; consumption of natural gas and pulverized coal fuel by air tuyeres; chemical composition of the top gas (with often unsatisfactory operation of gas analyzers); BF charging parameters; parameter cooling systems of mine, air tuyeres, and hearth; and temperature of the lining of the mine, hearth, and bottom. For the last three items, Ukrainian scientists have already implemented the corresponding mathematical models [1,4,12,13].

DSSs, similar to those used at metallurgical plants of the world’s leading iron producers, are extremely difficult to adapt to Ukrainian conditions, which are characterized by significant variability of charge and fuel conditions. Attempts to quickly adapt the systems of world manufacturers did not bring success. Thus, the creation of new domestic DSSs required deep preliminary studies for each specific BF. This problem is the subject of the present article.

In addition to preliminary studies, it is necessary to equip BFs with additional means (systems) for monitoring process parameters. Such important and promising means of control for Ukraine include the following: means of temperature control of the gas flow over the burden surface [14,15,16], radar means for monitoring the configuration of the burden surface [17,18,19], means for controlling the distribution of the chemical composition of gases along the radius of the BF above the burden surface and (or) in the upper zone of the BF [20], control of blast distribution for each air tuyere, and multi-point control of static pressure along the height of the BF shaft [21].

The aim of the present work is to increase the level of automation of the BF production through the development and implementation of new decision support systems for managing BF smelting in variable technological, fuel, and raw material conditions. The presented research and development of DSSs were carried out for the operating conditions of three BFs of PrJSC “Kamet-steel” (Kamianske, Dnipropetrovsk region, Ukraine, until September 2021; name: Dneprovsky Iron and Steel Integrated Works): BF No. 1M (useful volume 1500 m³, 20 air tuyeres, BF No. 9 (useful volume 1386 m³, 18 air tuyeres) and BF No. 12 (useful volume 1386 m³, 16 air tuyeres).

The first DSS to stabilize the thermal state of the BF was developed and implemented by the Iron and Steel Institute of Z.I. Nekrasov (ISI) at PrJSC “Kamet-steel” in 2021. The target function of managing the thermal state of BF smelting is to maintain the stability of the thermal regime at a level that ensures the production of cast iron with minimal deviations in chemical composition. The content of silicon in cast iron [Si] is most often used as the main indicator (criterion) of the thermal state [22]. Maintaining this value within the specified limits, on the one hand, provides the most economical thermal regime without excess heat reserve (that is, without excessive fuel consumption and with maximum productivity), and on the other hand, guarantees compliance with the requirements for cast iron in terms of chemical composition. Informative indicators of the thermal state of BF smelting are the iron temperature, the sulfur content of the iron, and How much FeO and alkaline oxide is in the slag. Characteristic indicators of the low thermal state for the conditions of the BFs of the PrJSC “Kamet-steel” are: the sulfur content in the iron is more than 0.8%, FeO and K₂O in the slag is more than 1.0% and 0.9%, respectively.

Many years of experience in the theory and practice of controlling the thermal state of BF smelting, as a rule, were reduced to the development and implementation of forecasting methods [Si], but most of the known methods of forecasting in practice have not always had a high degree of accuracy due to the multifactorial influence of technological parameters on the thermal state of BF smelting and uncontrollability of some of them [11]. An analysis of literature sources reveals a variety of methods for predicting the silicon content of cast iron; there are methods based on support vector machines [23,24], neural networks [25,26,27], the partial least squares method [28], and other approaches to finding a solution for this problem [29,30,31,32]. In addition, the accuracy of forecasting methods is determined by two main conditions: the prompt receipt of the full amount of reliable information about the technology parameters, and the stability of the fuel and blast parameters. Compliance with the first condition is difficult to implement (especially under Ukrainian conditions), and if the second condition is observed, the stability of the thermal state is ensured, which is accompanied by a change in the parameters of the thermal state within acceptable ranges, which excludes the feasibility of forecasting as such. Therefore, the actual task of the stability of the process is as follows: a clear understanding of the degree of influence of certain factors, prompt response to changes, and compliance with the technological instructions.

To implement the second DSS to ensure a uniform raceway adiabatic flame temperature (RAFT) distribution, it is necessary to control the distribution of blast costs for each air tuyere of a BF, which has been an extremely difficult task until recently. The reasons for the uneven distribution of the blast over the tuyeres are related to such factors such as the location of the hot blast supply to the BF; different gas permeability of materials in the above-tuyere zones due to the uneven circumferential distribution of the charge, change in the section of the tuyeres when they are overgrown with slag, work with distortions of the burden levels; uneven formation of the skull and the height of the lining of the mine, incomplete mixing of hot and cold air during operation of the mixing valve; circumferential unevenness of the melt in the hearth; geometry of tuyere sleeves, and ingress of fragments of refractory materials into the diffuser of tuyere devices with partial destruction of the combustion chamber of air heaters and the hot blast air duct [33].

The results of the introduction of systems for automatic distribution of blast over tuyeres are known, where ceramic plugs; uncooled shortened measuring nozzles; and butterfly valves made of heat-resistant steel, throttling valves, etc., were used for control and regulation [33,34]. The low efficiency of such systems, due to the short service life of the measuring equipment, necessitated the introduction of a new generation system, taking into account the existing possibility of controlling the consumption of pulverized coal for each individual tuyere [35]. The problems of the efficiency of injection of pulverized coal as an alternative and economically justified substitute for coke are widely covered in the literature. Known works are devoted to peculiarities of preparation and transportation of pulverized coal before blowing into a BF; increasing the completeness of combustion of pulverized coal, including when using various designs for supplying pulverized coal to the blast channel of the tuyere tool; and requirements necessary to achieve high specific consumption of pulverized coal. A literature review on solving these problems is presented in [36]. At the same time, almost no attention has been paid to the problem of ensuring a uniform or directed (given) uneven distribution of the blast during the injection of pulverized coal. This fact led to the solution for this issue, as covered in the present article.

BF charging control is an important element in controlling the operation of a BF and achieving high technical and economic indicators. Literature analysis reveals a variety of mathematical models used to assist in the distribution of charge materials, using discrete element modeling [37,38,39,40,41], experiments on physical models [40,41,42], dynamic clustering method based on dynamic time deformation and adaptive resonance theory, charge distribution control based on the inverse dynamic model [43], other methods of calculation and forecasting [43], and known methods based on radar technology [17,18,19,44] to determine the surface of the charge. To implement the third DSS for adjusting the parameters of the charging mode, it is proposed to use information on the dynamics of changes in the temperatures of the gas flow above the charge surface. This is due to the use of stationary thermal probes in Ukrainian BFs and a series of long-term studies of the influence of BF charging parameters on the gas flow distribution along the BF radius [2,12,14,15,16].

2. Materials and Methods

Experimental research methods, methods of mathematical simulation, methods of object-oriented programming, statistical data processing methods of control facilities, and industrial testing of the developed technical solutions are used in this study.

The main principle of the implementation of the DSSs was the development of algorithms for making management decisions in real time using direct and iterative optimization methods.

The algorithms of the first DSS were based on the knowledge accumulated during the scientific and technological support of the plant’s BFs; the algorithms took into account the main technological errors that were made by the technological personnel when adjusting the thermal state of the smelt. In addition, technological, fuel, and raw material factors characteristic of the facility, which affect the smelting process, were taken into account.

The algorithms of the second DSS were based on information new to researchers and practitioners about the change in blast flow rates over time for each individual air tuyere of a BF. The partial blast flow rates were determined from the heat removals measured by the calorimetric method on the water-cooled elements of the tuyere instruments, in accordance with the principle that the distribution of heat removals around the circumference of the blast furnace is proportional to the change in blast costs. Based on this information, a method was developed to provide a given thermal state of the tuyere zone of a BF with a regulating factor—the consumption of pulverized coal supplied to each air tuyere.

The algorithms of the third DSS were based on information about the change in the temperature of the gas flow above the burden surface in the BF in conjunction with the applied charging modes and technical and economic indicators of BF smelting. This information is recorded continuously by four stationary fixed thermal probes at eight control points along each BF radius. The algorithms are based on the obtained statistical dependencies and known theoretical concepts of the influence of charging parameters on the nature of the temperature gas distribution, obtained for the conditions of a particular BF.

3. Results and Discussion

For the most accurate assessment of the stability of the thermal state of BF smelting, it is customary to use the root-mean-square deviation (RMSD) of the silicon content of cast iron σ[Si], which is determined either within one day of BF operation (by the number of outlets, smelting products per day) or on a temporary interval equal to 10 issues. Estimating σ[Si] over a longer time period makes it difficult to analyze the identification of factors that influenced the RMSD, since the comparison is relative to the average value of [Si], which is influenced by various factors on a daily basis. When analyzing the dynamics of change in σ[Si] at the BF of PrJSC “Kamet-steel” in 2020–2021, the factors most influencing the decrease in this indicator were identified. This is the stability of the change in the RAFT (with an average value of over 2000 °C) and the parameters that determine it; the promptness of taking corrective actions at the first signs of a violation of the thermal state; change in the ore load in accordance with the trend of changes in the thermal state and with an increase step with an increase in the intensity of smelting, not exceeding 0.3 units; compliance with the release schedule; stability of coke quality indicators and chemical composition of agglomerate; satisfactory (without cluttering) condition of the hearths when coke screenings are used as part of the iron ore part of the charge; and the presence of heat reserves before the stops of the BF for repairs. In addition, it is necessary to take into account negative factors that exist at PrJSC “Kamet-steel”, such as as the technical condition of the BFs (mainly the condition of the hearths), lack of automatic coke moisture control, no screening of agglomerate, instability of oxygen supplied for blast enrichment, instability of changes in the consumption of pulverized coal, high RMSD indicators of coke quality and chemical composition of the sinter, and delay in obtaining the results of the chemical analysis of melt products due to the peculiarities of the operation of the pneumomail [11].

Therefore, taking into account the above, the target value of σ[Si] under the predicted operating conditions of the BF PrJSC “Kamet-steel” may be less than 0.100 units, given the achieved value of 0.081–0.091% in May–July 2021 after a major overhaul at BF No. 1M with shotcreting of the shaft and flushing of the hearth with high-oxide sinter and briquettes at BF No. 9 and BF No. 12. To achieve this value of σ[Si], it is necessary to comply with the requirements of the dynamics of changes in the RAFT and ore load, comply with the release schedule, carry out targeted washings of the hearth with hard-to-recover materials when it is cluttered [45,46,47,48,49], carry out preventive soft washings from flux residues and coke debris in the intermediate zone of the BF with a mixture of iron ore materials with a basicity of 0.65–0.85 [48,49], ensure a uniform distribution of the thermal state of the tuyere zones along the circumference of the BF by directional regulation of the pulverized coal consumption in the air tuyeres [35], use manganese-containing materials in the BF charge to maintain the manganese content of cast iron in the range of 0.25–0.35% [49], and comply with the recommendations of the developed DSS for managing the thermal state of BF smelting.

In accordance with the knowledge base obtained during the research in 2018–2021 at PrJSC “Kamet-steel” [11,35,48,49], and as a result of the work performed in 2020–2021 (detailed analysis of violations of the thermal regime of BF smelting), a DSS was developed in the mode of an adviser for managing the thermal state of BFs of PrJSC “Kamet-steel”. Screenshots of the DSS software are shown in Figure 1. DSS includes the entire list of information necessary for technological personnel in a convenient and compact form: a table of chemical analyses of cast iron and slag; a table of chemical analyses of the agglomerate; indicators of quality of coke and pulverized coal; graphs of the dynamics of changes in gas-blast parameters; graphs of changes in the ore load, silicon content, cast iron temperature, temperatures of the peripheral gas flow and top gas; the BF charging system; and the feed composition of the charge.

In addition, the DSS generates recommendations for certain technology deviations, namely, when changing the quality of coke and pulverized coal, in the case of deviations from the basic range of changes in the RAFT and the parameters that determine it, when the ore load changes incorrectly, etc. In the case of incorrect actions by the personnel, the system signals the need for corrective actions.

The second DSS to be implemented in 2021–2022 at PrJSC “Kamet-steel”, in combination with an automated control system for the distribution of blast over air tuyeres, will allow corrective actions to be taken to ensure uniform distribution of the thermal state of the tuyere zone along the circumference of the BF in terms of the RAFT.

The uneven distribution of the blast over the air tuyeres of the BF is typical for conditions of variable quality of iron ore raw materials and coke, unsatisfactory condition of the hot blast air duct, variable consumption of fuel additives, overgrowth of tuyeres with slag, and stopping injection of fuel additives into one or more air tuyeres [35]. These factors lead to an uneven distribution of the RAFT along the circumference of the combustion chamber, a decrease in the replacement coefficient of the pulverized coal, and a low degree of use of the thermal energy of the gas along the height of the combustion chamber. To prevent the negative impact of these factors in order to reduce the consumption of coke and increase the productivity of the BF, it is advisable to install systems for automated control of the distribution of blast over the air tuyeres of the BF [35].

The essence of the blast distribution control offered by the ISI is to determine the blast flow rate along the circumference of the BF in proportion to the thermal loads measured on the water-cooled elements of the tuyere tools. As a result of previous studies performed at the ISI when using natural gas in a blast, where a water-cooled nozzle acted as a calorimeter, a close relationship was established between the thermal loads on water-cooled nozzles and the non-uniformity of blast distribution measured by other methods; i.e., it was proven that the amount of heat that is lost on the water-cooled elements of tuyere devices unambiguously depends on the amount of physical heat and the gas-dynamic mode of blast movement per air tuyere [50,51,52]. In recent years, it has been proven that water-cooled flanges of suspended (movable) knees can act as a calorimeter (Figure 2), since nozzle cooling has only been used on large-volume BFs [35].

The prerequisite for the implementation of an automated control system for the distribution of blast over air tuyeres, including the corresponding DSS, was, first of all, the possibility of installing the injection of pulverized coal at PrJSC “Kamet-steel” in percentage terms to regulate the consumption of pulverized coal on individual air tuyeres and thereby ensure circumferential uniformity blowing mode parameters. It should be noted that this option of installing the injection of pulverized coal in Ukraine is available only at PrJSC “Kamet-steel”. Secondly, the ISI performed jointly with the BF shop in 2018–2021 measurements of thermal loads on the flanges of the suspended (movable) knees of tuyere tools made it possible to estimate the circumferential distribution of the thermal state of the tuyere zone in terms of the RAFT and, in order to equalize it, to realize the directed non-uniformity of pulverized coal flow rates [35]. As a result of determining the non-uniformity of the blast flow at the studied BFs, the determining influence of the configuration of the hot blast path supply at the point of insertion into the annular pipeline was established, which leads to the formation of a constant zone with a reduced blast flow on individual air tuyeres.

Figure 3 shows an illustration of tuyere hearths with an ideal blast distribution (No. 1); characteristic (working) for the BF PrJSC “Kamet-steel” (No. 2); and with unevenness for the BF PrJSC “Kamet-steel”, requiring corrective actions (No. 3). Efficiency of long-term pilot-industrial use of directional irregularity of pulverized coal consumption around the circumference of BFs in order to create a uniform circumferential distribution of the RAFT was as follows: at BF No. 12, 2.2% saving of the reduced coke consumption, with a 12% decrease in the RMSD of the silicon content of the cast iron at the outlet, and at BF No. 1M, 2.9% saving of the reduced coke consumption, with a decrease in the temperature of the peripheral gas flow by 100–120 °C. At the same time, the efficiency of using the uneven distribution of pulverized coal at BF No. 12 was observed when working with the amount of pulverized coal in the blast up to 120 kg/t of cast iron [35], which, among other things, contributed to the further implementation of an increase in the diameter of air tuyeres from 140 to 150 mm, as recommended by the ISI.

It should be noted that the measures to equalize the thermal state of the tuyere zones along the BF circumference without using an automated system require at least daily control measurements of thermal loads on the water-cooled flanges of the suspended (movable) elbows of the tuyere tools. However, when such measurements are carried out, the reliability of the results is significantly reduced, namely, when using the metrological tools available in the BF shop, when carrying out measurements in non-stationary periods of work (violation of the descent of the charge, a shift in the schedule for the release of smelting products, etc.), under conditions of impossibility of measurements in gas hazardous areas and in case of corrosive wear of cooling tubes, and when measuring by different teams of plumbers and BF shop masters (human factor).

In order to effectively control the distribution of blast over the air tuyeres of the combustion chamber and further change the distribution of the pulverized coal along the circumference of the BF in order to equalize or create a given unevenness of the RAFT, an automated control system is currently being implemented, including the installation of a water flow meter for each tuyere device supplied to the cooled elements, a suspended (movable) elbow of a tuyere tool; a thermocouple in the section of the drain pipeline after cooling; and a DSS (which includes visualization of primary information on the distribution of pulverized coal consumption and heat loads along the circumference of the BF; calculated information about the distribution of the actual RAFT; flow rate, and blast velocity; the geometry of tuyere hearths; and a block for generating recommendations for adjusting the consumption of fuel additives on individual tuyeres to equalize the distribution of RAFT, as well as signaling the presence of the so-called “slagging” of tuyeres; and if necessary, recommendations for changing the diameters of air tuyeres, etc).

The third DSS (name adopted at PrJSC “Kamet-steel”: “Expert Module”), implemented in 2021–2022 in the automatic control system of the BF shop in combination with the installation of means for measuring the temperatures of the gas flow above the surface of the charge materials (thermoprobes), will make it possible to reasonably and in a timely manner correct the parameters of the charging mode [2,11,12,14,15,16,53].

In recent years, to assess the distribution of the intensity of the gas flow, thermal probes are increasingly used, which are installed above the surface of the charge in the BF (Figure 4). An important advantage of thermal probes is continuous measurement of the temperature distribution of gases, where the information obtained helps with evaluating and correcting the BF charging mode [2,14,15], operational reliability, durability, and low cost of construction. The disadvantages of thermal probes include the change in readings when adjusting the level of filling of materials in the BF due to mixing and deflection of the gas flow, underestimation of temperatures under the influence of the refrigerant, and the formation of charge depressions under the thermal probes. These shortcomings are not significant, since the analysis of the dynamics of temperature changes over time and the assessment of changes in relative values, rather than absolute ones, quite objectively makes it possible to identify the influence of certain parameters of the charging mode on the change in the configuration of the temperature field and to determine the permissible limiting temperatures corresponding to the high technical and economic indicators of the BF smelting.

The functioning of the DSS for controlling the charging mode using the readings of thermal probes has the following algorithmic structure. At the first stage, we have the formation of the “reference” database, including technological parameters, smelting indicators, and BF charging modes (characteristics and charging systems) with the corresponding temperature distributions over the surface of the charge. That is, for the entire list of charging modes used at the BF, the distribution curves of the average daily temperatures over the surface of the charge along the top radii corresponding to the minimum consumption of coke and the maximum productivity of the BF are “bound”. Next, the formation of criteria for the influence of the parameters of the BF charging is carried out: the charging system, the level of the charge, the volumes of coke and iron-containing materials, and the operating modes of the charge distributor around the circumference on the shape of the temperature distribution curves above the surface of the charge (that is, on the temperature change at each point along the BF radii).

At the next stage, the actual data of the BF operation are compared with the reference database, and the deviations of the gas temperature distribution over the charge surface are determined. The search for solutions to approximate the deviations to the “reference” database is carried out using the methods of direct and iterative optimization. The direct optimization method consists of searching the database for the best BF charging parameters corresponding to the reference temperature distribution. The iterative optimization method consists of generating further corrective changes in one or more parameters of the BF charging until the maximum approximation of the temperature distribution to the reference one is reached. Examples of the functioning of the DSS (expert module) are shown in Figure 5.

In addition, optimization methods include the following: continuous training of the DSS analytical unit with statistical information about the operation of the BF, including technical and economic indicators and information from all measuring equipment available at the BF; accumulated experience in studying the relationship between the distribution of charge materials and gases on the top; and well-known provisions of the theory and practice of the BF charging on the effect of charging parameters on the efficiency of melting and furnace productivity.

4. Conclusions

The decision support systems developed by the Iron and Steel Institute of Z.I. Nekrasov were implemented and passed pilot testing as part of the automated control system of the BF shop of the PJSC “Kamet-steel”. Furthermore, they can significantly increase the level of production automation. In addition, timely information support and decision-making recommendations allow you to interactively adjust the process and, as a result, ensure the energy efficiency of smelting by minimizing the number of transition periods and breakdowns of the BF.