Method and system for producing pearlitic rail steels
Patent Information
- Application Number
- PCT/EP2026/058124
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-03-28
- Filing Date
- 2026-03-23
- Publication Date
- 2026-10-01
Smart Images
Figure EP2026058124_01102026_PF_FP_ABST
Abstract
Description
[0001] International patent application
[0002] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0003] 240919WO
[0004] Method and system for the production of pearlitic rail steels
[0005] The invention relates to a method for producing pearlitic rail steels, a system for producing pearlitic rail steels and a track component, in particular a rail for railway vehicles.
[0006] Worldwide, 71.1% of steel (in 2023) is produced via the blast furnace route. Pig iron is obtained from iron ore by melting it in a blast furnace together with reducing agents such as coke and additives. The molten pig iron is then further processed into crude steel in a converter process (e.g., with oxygen) or in an electric arc furnace (using scrap as raw material).
[0007] Primary metallurgical units used in the production of crude steel include: BF (blast furnace), RE-Ent-S (hot metal desulfurization), BOF (basic oxygen furnace), EAF (electric arc furnace), SAF (submerged arc furnace), IF (induction furnace), smelter, MOE (metal oxide electrolysis), plasma reduction metallurgy, etc. The crude steel is then further processed in secondary metallurgy to improve its composition and purity.Here, the steel is treated in secondary metallurgical units such as rinse stations, purging stations, ladle furnaces, RH plants (Ruhrstahl-Heraeus process), VD plants (vacuum degassing), DETEM plants, VOD plants (vacuum oxygen degassing), ASEA-SKF (oxygen vacuum treatment with stirring), CAS plants (composition adjustment by sealed argon bubbling), RS-V plants (Rheinstahl-Siemens vacuum), CAS-OB plants (composition adjustment by sealed argon bubbling - oxygen blowing), calcium-silicon treatment (CaSi injection), magnesium treatment (magnesium injection), VLD processes (vacuum ladle degassing), etc., where additives (metallic and non-metallic feedstocks) such as alloying elements and slag formers are added to achieve the desired chemical composition. and to achieve specific properties of the steel.
[0008] Scrap metal is not normally used in blast furnaces, as the process is primarily based on the reduction of iron ore to produce liquid pig iron. However, scrap metal is used in other process steps in steel production, especially in electric arc furnaces or as a coolant in converters. International patent application
[0009] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0010] 240919WO
[0011] The pig iron produced in primary metallurgical units is then further processed in an oxygen converter. In these converters, such as the LD converter (Linz-Donawitz process), liquid pig iron is decarburized by injecting oxygen and converted into crude steel. Scrap metal is added as a coolant, along with other feedstocks. The addition of scrap metal in the converter process helps to control the temperature, as the exothermic nature of the oxygen reactions heats the molten metal considerably.
[0012] Another important production method in the steel industry is the electric arc furnace (EAF). Here, up to 100% scrap metal can be used as feedstock, which is melted using electrical energy to produce liquid steel. This process is particularly environmentally friendly, as the more energy-intensive reduction process is usually unnecessary and therefore causes fewer CO₂ emissions than the blast furnace process.
[0013] In the electric arc furnace process, steel is produced using electrical energy. An electric arc is generated between graphite electrodes and the scrap material in the furnace, creating high temperatures (up to 3,500 °C in the arc zone) that melt the scrap and process it into liquid crude steel.
[0014] Raw steel production utilizes raw materials such as scrap metal. A distinction is made between metallic and non-metallic raw materials. These include, for example, iron ore, reducing agents, slag formers, sponge iron in the form of directly reduced iron (DRI) or hot briquetted iron (HBI), deoxidizing agents, alloying agents, and scrap metal.
[0015] Using scrap metal as a raw material for steel offers several advantages. Melting scrap metal requires only the energy needed for heating or melting, and because the energy-intensive reduction process usually involves upstream processes, it uses less energy than the primary production of crude steel from iron ore. Furthermore, scrap recycling reduces CO₂ emissions and the need for new raw materials, thus lowering the environmental impact. Since steel can be recycled again and again, using scrap metal makes a significant contribution to the circular economy in the steel industry.
[0016] Although the use of scrap in the electric arc furnace (EAF) offers many advantages, there are also some disadvantages and challenges that need to be considered.
[0017] The first challenge when using scrap metal is the fluctuating scrap quality. The scrap used in the EAF is often inhomogeneous and can contain impurities. International patent application
[0018] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0019] 240919WO
[0020] The scrap may contain elements that negatively affect steel quality. For example, unwanted elements such as copper, molybdenum, tin, nickel, and chromium can be present in the scrap. These are difficult to remove from the raw steel and can have negative effects on downstream secondary metallurgy and the properties of the final product for certain applications. Impurities can make the EAF process more complex and expensive, as additional processing steps or sorting may be required.
[0021] The second major challenge concerns controlling the chemical alloy composition. Because scrap can have different alloys and metal compositions, it can be difficult to precisely control the chemical composition of the final product. Adding alloying elements in the EAF (Electron Alloy Forming) therefore becomes more complicated, and additional adjustments are often necessary to achieve the desired steel specification.
[0022] Due to potential impurities and the fluctuating quality of scrap metal, it is difficult to guarantee the particularly low levels (< 0.005 wt%) of elements such as copper, tin, zinc, and molybdenum required for certain specialized applications using the EAF process. In such cases, the use of primary materials (e.g., pig iron, liquid pig iron, hot metal, DRI, HBI) or very high-quality, sorted scrap is necessary, which increases costs.
[0023] In addition to the chemical composition, which is subject to fluctuations due to the input material in an electric arc furnace process and is therefore often kept within predetermined tolerance limits, the effects of individual elements on the properties of the final product must also be taken into account.
[0024] In particular, during subsequent heat treatment or forming processes, such as rolling, the material properties of the steel are further controlled.
[0025] Several solutions are already known that take into account the undesirable accompanying elements in the alloy composition or provide for additional adjustments to the alloying elements.
[0026] It is common practice to define specific target ranges for a steel grade, specifying the minimum and maximum values for each alloying element. For accompanying elements that are not intentionally alloyed but are present in the raw materials in unknown concentrations, international patent applications exist.
[0027] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0028] 240919WO
[0029] Maximum levels are defined to ensure that, even if these levels are not reached, the material properties remain within the desired range. However, the actual levels of these accompanying elements in the liquid phase are not taken into account.
[0030] A disadvantage of this is that low maximum levels of accompanying elements necessitate the use of raw materials with low levels of these elements. This is primarily achieved through a high proportion of raw materials produced from primary sources.
[0031] From EP 3 956 481 Bl, a computer-aided method for monitoring the steelmaking process in a converter is known. In this method, various materials with specific properties are introduced into the converter to produce liquid crude steel and slag. The method first defines the desired properties of the liquid crude steel and slag to be produced. Then, the required quantities of each material that must be introduced into the converter to achieve the defined target characteristics of the crude steel and slag are calculated. The calculated material quantities are transmitted to the operator or to automatic loading systems, and the converter is loaded accordingly.
[0032] This section describes the loading of the converter with various feedstocks. A disadvantage is that only the composition of the melt produced in the converter is controlled. Subsequent manipulation of the alloy composition is not specified.
[0033] German patent application DE 10 2021 211 320 A discloses a method for controlling and regulating a production plant for rolled products made of metallic alloys such as steel, iron, or aluminum. The aim of the method is to optimize the use of raw materials and reduce production costs. Initially, specific target values for the material, surface, and geometric properties of the final product, as well as permissible chemical compositions with defined tolerance ranges, are established for each order. By applying process models, predictive actual values of the product properties are calculated for each order and compared with the target values. Only compositions that lie within the specified tolerances are selected. For each selected chemical composition, cost parameters such as alloy, feedstock, energy, iron, additive, and CO2 costs are determined. These are quantified in the form of penalty points or a penalty function.The intersection of the international patent application is then determined.
[0034] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0035] 240919WO
[0036] Permissible chemical compositions for various combinations of production orders that are to be melted in a common batch were determined.
[0037] The disadvantage of this approach is that tolerance ranges are determined for each composition. Precise control of the composition is not possible with this model.
[0038] From EP 4 183 498 A1, an AI-based predictive model is known that can predict mechanical properties from previously measured input values, such as the composition of the melt. To influence the mechanical properties, an adjustment of the manufacturing parameters, such as pre-roll ratio, finish-roll ratio, rolling start temperature, cooling start time, cooling rate, or line speed, is proposed as output.
[0039] The model is optimized for specific production lines and processes. If production conditions or requirements need to change quickly, the system may have difficulty adapting.
[0040] US Patent 5,462,613 A discloses a process for producing wire rod with a predefined tensile strength. An empirical model is used to first predict the tensile strength based on a sample analysis of the steel melt. The same model is then used to calculate a "moving setpoint" for one or more control elements, such as carbon, required to achieve the target tensile strength. The melt is then "trimmed" by adding these elements before being processed into wire rod.
[0041] From CN 112 233 735 A, a method for adjusting the chemical composition of pearlitic rail steel to meet specific tensile strength requirements is known. The method combines the software models JMatPro and DEFORM-3D to improve the tensile strength prediction accuracy of JMatPro.
[0042] WO 2022 / 110812 Al discloses a high-strength pearlitic steel rail and a manufacturing process for it. The process combines a specific chemical composition, comprising C, Si, Mn, Cr, V and Nb, with controlled rolling and cooling.
[0043] Pearlitic rail steels are a special group of steels characterized by a pearlitic microstructure, which offers high strength and wear resistance. These steels are subject to an international patent application.
[0044] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0045] 240919WO
[0046] are particularly suitable for use in railway tracks and sleepers, as they can withstand the high loads caused by train traffic and high axle loads.
[0047] The production of these steels generally involves the following steps: steelmaking, shaping and cooling, and, if necessary, heat treatment.
[0048] Pearlitic rail steels are usually produced in an electric arc furnace or an oxygen converter. The main constituents besides iron are carbon (approx. 0.7–0.8%) and manganese (approx. 1%), with other elements such as silicon, chromium, and vanadium being added to achieve specific properties.
[0049] After the rails are cast and rolled, the cooling process is precisely controlled to create a fine-grained pearlitic microstructure. This pearlitic structure consists of alternating layers of ferrite (soft and ductile) and cementite (hard and brittle), resulting in an excellent combination of hardness and toughness.
[0050] In certain applications, heat treatment can be performed to further improve wear resistance. "Head hardening" is a process that specifically treats the rail head to increase hardness in this highly stressed area.
[0051] The pearlitic microstructure in these steels is created, among other things, by controlled cooling and is crucial for their material properties.
[0052] Furthermore, pearlitic rail steels are characterized by high wear resistance, which is particularly important since the rail surface is exposed to mechanical stresses and abrasion.
[0053] Furthermore, the combination of ferrite and cementite gives the steels the necessary toughness to withstand the high dynamic loads in rail transport.
[0054] Pearlitic rail steels are used in various areas of rail transport, such as high-speed lines, heavy goods transport or curved tracks.
[0055] Pearlitic rail steels are therefore specially developed materials that, through their composition and controlled heat treatment, offer an ideal microstructure for rail transport. The advantages of their pearlitic structure include high hardness, strength, and are subject to an international patent application.
[0056] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0057] 240919WO
[0058] Toughness makes them a durable and robust choice for rails used in demanding traffic environments.
[0059] Therefore, maintaining consistent product characteristics is of great importance.
[0060] The object of the invention is to provide a method for the production of pearlitic rail steels, with which constant product properties are ensured even with larger fluctuations in the accompanying elements.
[0061] The problem is solved by a method having the features of claim 1.
[0062] Advantageous further training options are indicated in the sub-requirements.
[0063] Furthermore, the invention aims to provide a system for the production of pearlitic rail steels, which makes it possible to guarantee constant product properties even with larger fluctuations in the accompanying elements.
[0064] The problem is solved with a system having the features of claim 16.
[0065] Advantageous further training options are indicated in the sub-requirements.
[0066] Furthermore, it is an object of the invention to provide a track component, in particular a rail, for rail vehicles.
[0067] This problem is solved with a track section having the features of claim 23.
[0068] This is a concept according to the invention for dynamic alloying. In this process, the element boundaries for alloying elements are adapted in the liquid phase.
[0069] An analysis of the melt determines the actual values of the element contents. These actual values are fed into a data-driven predictive model suitable for a specified rail steel, so that the influence of the respective elements on a material-related, especially mechanical, property is modeled. If necessary to bring the material-related, especially mechanical, property into a desired target range, a correction model identifies a response in the form of adapted alloying element limits, and these are fed to the alloy calculator as adapted target values. The alloy calculator uses the adapted alloying element limits to determine the masses of raw materials. International patent application
[0070] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0071] 240919WO
[0072] The inventors recognized that a distinction should be made between two defined types of elements when dynamically correcting element levels: influencing elements and compensation elements.
[0073] Various raw materials are used in the steelmaking process. These include iron ore, sponge iron in the form of directly reduced iron (DRI) or hot briquetted iron (HBI), and scrap metal. These raw materials naturally introduce not only iron but also so-called accompanying elements and trace elements.
[0074] Trace elements are usually unavoidable elements, such as Si, Mn, S, P, O, N, and H, and are removed during a subsequent metallurgical process, if possible. If removal is not possible, the respective trace element and its influence on metallurgical processes must be taken into account.
[0075] For each type of steel, which is defined by the chemical composition of the alloy and the manufacturing route, there exists a defined class of accompanying elements resulting from practical experience; that is, certain upper limits for the levels of accompanying elements must be observed.
[0076] Trace elements are explicitly disruptive accompanying elements (e.g., As and Sb) that are present in very small quantities. These are not taken into account during the (dynamic) alloying process carried out here.
[0077] Alloying elements are elements that are deliberately added to give steel desired material properties. These properties can include mechanical properties such as hardness, tensile strength, toughness, and others, as well as chemical properties such as resistance to corrosion or hydrogen embrittlement, and others.
[0078] When using raw materials, especially metallic and non-metallic raw materials, it must be taken into account that they contain accompanying elements from their raw material history and alloying elements from their metallurgical history.
[0079] If, for example, scrap metal is used as a raw material in steel production, its components, apart from iron, are considered accompanying and trace elements for the product to be manufactured using the scrap. International patent application
[0080] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0081] 240919WO
[0082] Their presence must therefore be taken into account, since they (as former alloying elements) can only be removed from the product via metallurgical processes with great difficulty or not at all. However, their effect can be considered and compensated for according to the invention.
[0083] In the classic manufacturing process, the starting point is a product (steel) that possesses specific material properties according to market requirements. It is then chemically / metallurgically adjusted to meet these requirements. Such a product is therefore an alloy, where the alloying elements are defined by a specific content with upper or lower limits. Thus, the properties of a given alloy are known.
[0084] If, due to changed conditions, the use of raw materials such as scrap metal is significantly increased, the product will consequently be loaded with a higher quantity of accompanying elements. These accompanying elements (as "former" alloying elements) have an influence, for example, on the material properties of the product.
[0085] Therefore, this must be taken into account when adding alloys. The following scenarios can occur.
[0086] A) For example, the proportion of element A introduced by the feedstock is below the desired level. In this case, less element A may need to be added.
[0087] B) For element A, a content is introduced via the starting material that is already within the target range; nothing further needs to be added.
[0088] C) More element A than desired is introduced. The content of element A can only be reduced metallurgically with increased effort. Accordingly, the effect of element A must be compensated for.
[0089] The above example becomes more complex when several accompanying elements or alloying elements interact to produce specific properties.
[0090] The invention addresses this by selecting alloying elements to complement the accompanying elements that influence one or more desired material properties (hereinafter referred to as influencing elements), which reduce the influence of the International Patent Application
[0091] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0092] 240919WO
[0093] Influencing elements can compensate for a desired scope of influence (hereinafter referred to as compensation elements).
[0094] Compensation elements are elements that are at best already regular alloying elements of the target alloy, so that only quantitative adjustments are made within the framework of the desired qualitative alloy composition.
[0095] A simple example should illustrate this. If, in addition to iron, a desired steel alloy contains elements A, B, and C, and if element A is present in an amount exceeding the desired quantity, and if C is an element that reduces the influence of A on a defined target parameter, then C is added in an amount that, in addition to its own effect, also compensates for the effect of A.
[0096] Another example is when element A is present in the raw materials in an amount exceeding the desired quantity. If element C has a comparable influence on a desired property as A, but is present in the alloy in small amounts or only in traces, only enough of it is added so that the combined effect of A and C achieves the target value for the desired property. In this case, that means less C is added than would normally be added to compensate for the excessively high A content.
[0097] Compensation in the sense of the invention therefore means that an adjustment of the target alloy composition can take place upwards or downwards.
[0098] In this case, B is, for example, an element that has no influence on the defined target parameter, such as strength.
[0099] Overall, this inventive procedure allows for a coarser pre-selection of the input material.
[0100] Input parameters for the forecasting model are all elements that are contained in the liquid phase and have been analyzed.
[0101] The starting parameters are adapted element limits for elements that are alloyed in the liquid phase. These are the so-called compensation elements, such as C, Si, Mn, and Cr. Compensation elements are therefore elements that are added in the liquid phase to counteract the influence of the other elements, so that the desired properties of the final alloy are achieved. International patent application
[0102] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0103] 240919WO
[0104] Both the influencing and the compensating elements are elements that are introduced into the melt by adding raw materials.
[0105] These are trace elements that inevitably occur in iron or steel and may be present in small quantities. They can originate from the raw materials used, such as ore or scrap, or be introduced into the melt during processing. These trace elements influence the properties of the steel and, depending on their concentration, can have positive or negative effects on strength, corrosion resistance, toughness, and ductility.
[0106] In steel production, "scrap" refers to recycled, reusable metallic secondary material used as feedstock for the production of new steel. Scrap consists primarily of iron-based alloys and is classified into various grades and varieties depending on its purity, shape, size, and chemical composition.
[0107] The present process, which uses raw materials, relates to the production of pearlitic rail steels.
[0108] The concept is based on producing a final product of a specific steel grade in which material properties, particularly mechanical properties such as yield strength and tensile strength, as well as other parameters such as carbon equivalent (CAE), remain unchanged or change only minimally despite increased levels introduced by the metallic feedstocks in the primary metallurgical aggregate. This is in comparison to the strength properties and other parameters that result when smaller proportions of feedstocks are used in the melt production, thus introducing lower levels of undesirable impurities (e.g., in the LD process). This also ensures that subsequent post-processing steps following steel production do not need to be modified.
[0109] This is achieved by adapting one or more alloying elements and is subsequently referred to as dynamic alloying. The adaptation is carried out by applying a predictive model, which initially models the expected material properties, particularly mechanical properties, based on actual values of the alloy composition of a primary metallurgical aggregate. Should these properties lie outside the target range specified by the desired rail steel grade, corrected alloying element contents are calculated using a correction model, so that the specified requirements for the material properties, particularly mechanical properties, are met.
[0110] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0111] 240919WO
[0112] The properties can be fulfilled. Adaptation is therefore achieved by applying a predictive model for material properties combined with a correction model.
[0113] For railway steels, the material properties relevant are, in particular, hardness, tensile strength, elongation, and metallographic characteristics, especially microstructure. The predictive model calculates the effect of the elements as a hardness factor (HF). The hardness factor describes a characteristic value that correlates, via constraints, with material properties, especially hardness, tensile strength, elongation, and metallographic characteristics, especially microstructure.
[0114] The hardness factor can take values between 0% and 100%, where 0% corresponds to a hardness of 200 HB and 100% to a technically feasible hardness.
[0115] According to EN 13674, this hardness is currently 440 HB.
[0116] The technically feasible hardness can currently be assumed to be up to 480 HB.
[0117] The examples given refer to a maximum hardness of 400 HB; if higher hardness values are taken into account, the factors of the specified formalism will change accordingly.
[0118] For the hardness factor (HF), at least one compensation element is determined, which adjusts the hardness factor. The corrected target values or the target range of the at least one compensation element are calculated using the correction model, and the at least one compensation element is added in a quantity that brings the hardness factor into the target range.
[0119] After adding compensating elements, a further analysis can be performed. The actual value of the alloy composition is checked and, if necessary, dynamically corrected again using the prediction and correction model.
[0120] Various machine learning models can be used to generate the forecast and / or correction model, for example, linear regression models, polynomial regression models, decision tree-based regression models, random forest regression models, nearest neighbor models, neural network models, support vector machine models, ADA boost models, regression boost models, HIST boost models, XGBOOST models, generalized additive models, and symbolic regression models. International patent application
[0121] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0122] 240919WO
[0123] These models are data-driven. Database-driven queries, for example, can be used as the data basis for chemical and mechanical values. Process parameters, particularly heat treatment temperature and holding times, can also be derived from corresponding database-driven queries. Furthermore, datasets can be augmented with simulation data to represent the influence of higher pollutant loads. The data basis is divided into training and validation data; for example, 80% of the data is used for training purposes and 20% for validation purposes.
[0124] To improve the accuracy of the models, additional important process parameters besides the chemical composition, such as the temperature profile of the rolling and cooling processes, could be considered. Furthermore, it is useful to identify the most relevant parameters and, where applicable, negligible parameters with regard to a specific material-related, particularly mechanical, property.
[0125] Purely data-driven models are well-suited for prediction within the data space defined by the training data. However, prediction accuracy may deteriorate if there is a significant deviation from this known data space, i.e., if a significantly different alloy composition is incorporated into the calculation of the material properties, particularly mechanical properties, using the prediction and / or correction model. Furthermore, small training datasets may suffer from high variance, which can impair prediction accuracy. This effect can be minimized by using larger training datasets.
[0126] To identify the model with the highest accuracy for modeling the hardness factor, the actual values determined by melt analysis are fed into potential models (see list above). The prediction model and / or correction model with the highest accuracy, characterized by the smallest deviation / value of the root mean square error (RMSE) of the hardness factor, is then used to model the hardness factor. When considering multiple material properties, particularly mechanical properties, the mean value and / or the sum of the square roots of the mean square errors of the material properties, particularly mechanical properties, is used. International patent application
[0127] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0128] 240919WO
[0129] The influence of the individual elements on the material properties of the alloy is as follows.
[0130] Copper has a work-hardening and hardening effect on rails, while simultaneously reducing hot formability and elongation at break. Selective copper enrichment can also be observed on the surface of the rail steel. Copper can increase the strength of the steel through precipitation at concentrations >0.8% Cu. Below this concentration, the strength-enhancing effect of copper is attributable to solid solution strengthening. Low-melting phases at the grain boundaries can thus significantly impair hot formability. This subsequently leads to cracks in the finished product, which in the worst case can cause rail fractures during track operation. Therefore, it is advantageous to select a copper content of 0.001–0.4 wt.%, preferably 0.001–0.38 wt.%, particularly preferably 0.001–0.35 wt.%, and even more preferably 0.001–0.3 wt.%.
[0131] Nickel increases the toughness of steel, enhancing its hardness and strength. Furthermore, nickel increases the solubility of copper in rail steel. Therefore, a nickel content of 0.001–0.4 wt.%, preferably 0.001–0.38 wt.%, more preferably 0.001–0.35 wt.%, and particularly preferably 0.001–0.3 wt.% is advantageous.
[0132] Even small amounts of molybdenum significantly increase the hardenability of steel. However, excessive fluctuations in analysis lead to increasing instability of the microstructure. Besides the formation of fracture-inducing martensite phases in the segregation zone, this can result in the development of undesirable foreign microstructure components, which can lead to accelerated surface fatigue. Furthermore, the weldability of the steel decreases with increasing molybdenum content. The influence of molybdenum on the transformation behavior also increases the process uncertainty of the heat treatment plant downstream of the rolling process when significant fluctuations in analysis occur, as the accelerated cooling can lead to the formation of undesirable mixed phases on the rail head circumference. Corresponding operational experience has clearly shown that exceeding critical values increasingly results in the presence of mixed phases.Molybdenum promotes subcooling and thus has a positive effect on the flake fineness of pearlitic rail steels. Therefore, it is advantageous to select a molybdenum content of 0.001–0.06 wt.%, preferably 0.001–0.058 wt.%, and particularly preferably 0.001–0.055 wt.%.
[0133] Tin has a work-hardening and hardening effect, while simultaneously reducing elongation at break and tending towards greater segregation, thus affecting structural stability. Tin also... International patent application
[0134] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0135] 240919WO
[0136] Furthermore, it is surface-active and impairs hot formability. Therefore, it is advantageous to select a tin content of 0.001–0.03 wt.%, preferably 0.001–0.028 wt.%, preferably 0.001–0.026 wt.%, and particularly preferably 0.001–0.024 wt.%.
[0137] Carbon forms carbides with a wide variety of elements; therefore, hardness is directly proportional to the carbon content. With a hypereutectoid carbon content, the heat treatment must be adjusted to prevent the formation of secondary cementite at the grain boundaries. In railway steels, carbon acts as a wear-resistant element via the carbides and has high strength-enhancing properties, while simultaneously reducing elongation at break and weldability. Therefore, it is advantageous to select a carbon content of 0.38–1.4 wt.%, preferably 0.38–1.39 wt.%, more preferably 0.38–1.35 wt.%, particularly preferably 0.038–1.3 wt.%, and even more preferably 0.38–1.25 wt.%.
[0138] Silicon acts as a solid solution hardener in the ferrite, thus increasing the hardness of the pearlite, but is not relevant for wear behavior, since in the deformed zone at the wheel-rail contact point, the cementite in the pearlite is the determining parameter. Silicon significantly lowers the transformation temperature, thereby promoting a fully pearlitic microstructure with hypereutectoid carbon contents during natural cooling as well as in heat-treated rails. Furthermore, it inhibits cementite formation and thus prevents the precipitation of grain boundary cementite in hypereutectoid steels. Silicon also has a significant influence on electrical conductivity. Therefore, it is advantageous to select a silicon content of 0.1–1.2 wt.%, preferably 0.1–1.15 wt.%, more preferably 0.1–1.10 wt.%, particularly preferably 0.1–1.05 wt.%, and even more preferably 0.1–1.0 wt.%.
[0139] Manganese binds sulfur to form manganese sulfides. Manganese increases hardenability and therefore lowers the critical cooling rate. This allows higher hardness to be achieved at lower cooling rates. Manganese is a strong austenite stabilizer. However, in combination with higher carbon and chromium contents, there is a risk of bainite formation. Manganese also tends to segregate more strongly. Therefore, it is advantageous to choose a manganese content of 0.2–1.8 wt.%, preferably 0.2–1.78 wt.%, more preferably 0.2–1.75 wt.%, particularly preferably 0.2–1.7 wt.%, and even more preferably 0.2–1.65 wt.%. International patent application
[0140] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0141] 240919WO
[0142] Chromium increases hardenability and lowers the critical cooling rate. As with manganese, the risk of bainite formation arises with chromium in combination with higher carbon and manganese contents. The hardening effect of chromium is significantly stronger than that of manganese. Furthermore, chromium has a negative impact on weldability. The formation of chromium carbides is also possible. Therefore, it is advantageous to select a chromium content of 0.001–1.5 wt.%, preferably 0.001–1.45 wt.%, more preferably 0.001–1.4 wt.%, particularly preferably 0.001–1.35 wt.%, and even more preferably 0.001–1.3 wt.%.
[0143] Vanadium generally has a grain-refining and strength-enhancing effect. Limitations in this regard are primarily economic in nature, based on the volatility of the ferrovanadium used. Therefore, it is advantageous to select a vanadium content of 0.001–0.3 wt.%, preferably 0.001–0.30 wt.%, more preferably 0.001–0.28 wt.%, particularly preferably 0.001–0.26 wt.%, and even more preferably 0.001–0.24 wt.%.
[0144] The elements C, Si, Mn, Cr and V are considered as compensation elements.
[0145] It should be noted that the detrimental effects of influencing elements are particularly critical when combined, and their origin lies in corresponding impurities in the feedstock (during the feedstock mix used to create the melt). When using melting technologies with increased feedstock quantities, the inevitably higher concentration of influencing elements can lead to greater fluctuations in the performance of relevant material properties. This can result in altered properties with wider variability in RCF (rolling contact fatigue) resistance, wear, and weldability during track operation.
[0146] The invention thus relates to a method for producing pearlitic rail steels with a hardness factor range specified for a rail grade, which is ensured by a target alloy, wherein at least one primary metallurgical unit leads to a first melt, wherein at least one analysis of the alloy element contents is carried out from the first melt, wherein actual values of the melt analysis are transmitted to a control system, wherein a distinction is made between influencing and compensating elements, wherein influencing elements are elements that have an influence on the hardness factor of an end product, and compensating elements are elements that have an influence on the hardness factor and compensate for the influence of influencing elements, wherein the magnitude of the influence of the actual values of all influencing and compensating elements on the International Patent Application
[0147] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0148] 240919WO
[0149] The hardness factor of the final product is known, wherein at least one actual value is recorded with respect to those influencing and compensating elements that have an influence on the hardness factor of the final product, wherein at least one compensating element is determined for the hardness factor of the final product, which adjusts the hardness factor, wherein one or more compensating elements are added to the melt in an amount that brings the hardness factor into a target range and compensates for the influence of one or more influencing elements on the hardness factor, and wherein two components are used for this purpose in the control system: a forecasting model and a correction model, wherein the forecasting model calculates the influence of the influencing and compensating elements on the hardness factor of the final product and the correction model calculates corrected target values of the compensating elements in order to bring the hardness factor into a target range.
[0150] The quantity of added compensation elements can also be 0 in accordance with the invention if this is deemed necessary to compensate for the effect or influence of the influencing elements.
[0151] Further training stipulates that the alloy is adjusted so that the final product complies with the microstructure of the standard EN 13674 or EN 14811.
[0152] Further training stipulates that the subsequent post-treatment steps following steel production are carried out unchanged until the final product is obtained.
[0153] Further training stipulates that the analysis must be carried out at least in the first melt of the last primary metallurgical aggregate of a production route, in particular from a tapping of crude steel, before or after a first addition of alloying elements and at least after a second addition of alloying elements.
[0154] Further training stipulates that the analysis is carried out by means of sampling according to EN ISO 14284.
[0155] A further development provides that if a first element is introduced by a starting material in a proportion that is below a desired target range defined by the target alloy, the first element, provided it is a desired alloying element, is added to the extent necessary to achieve the target range; or, if the first element is introduced by a starting material in a proportion that already corresponds to the target range, it is no longer added; or, if the first element is introduced by a starting material in an international patent application
[0156] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0157] 240919WO
[0158] Any excess of the alloy that exceeds the target range is compensated for by a reduced alloying element of a second similarly acting element or an increased alloying element of a counteracting third element.
[0159] Further training stipulates that the influencing elements whose influence on the hardness factor of the final product is to be compensated include one, several or all from the group of Cu, Mo, Ni, Sn.
[0160] Further training stipulates that the influencing elements whose influence on the hardness factor of the final product is to be compensated include Cu, Mo, Ni and / or Sn.
[0161] Further training stipulates that one, several or all of the elements from the group of C, Si, Mn, Cr, V are chosen as compensatory elements.
[0162] Further training stipulates that C, Si, Mn, Cr and / or V are chosen as compensating elements.
[0163] Further training stipulates that the forecasting and / or correction model is generated using one, several, or all of the following models: linear regression models, polynomial regression models, decision tree-based regression models, random forest regression models, nearest neighbor models, neural network models, support vector machine models, ADA boost models, regression boost models, HIST boost models, XGBOOST models, generalized additive models, and symbolic regression models.
[0164] Further training stipulates that the forecasting and / or correction model is generated using linear regression models, polynomial regression models, decision tree-based regression models, random forest regression models, nearest neighbor models, neural network models, support vector machine models, ADA boost models, regression boost models, HIST boost models, XGBOOST models, generalized additive models and / or symbolic regression models.
[0165] Further training provides that, when considering the hardness factor, at least one recorded actual value of the analysis is assigned to at least one forecasting model and / or correction model using the square root of the mean squared deviation of the hardness factor, whereby the square root of the mean squared deviation of the hardness factor is used for all forecasting models and / or International Patent Application
[0166] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0167] 240919WO
[0168] Correction models are determined and, based on the results, the forecasting model and / or correction model with the highest accuracy is selected.
[0169] Further training stipulates that, when several material properties are considered, at least one recorded actual value of the analysis is assigned to at least one prediction model and / or correction model using the mean and / or the sum of the square roots of the mean square deviations of the material properties, whereby the mean and / or the sum of the square roots of the mean square deviations of the material properties is determined for all prediction models and / or correction models for the at least one actual value, and the prediction model and / or correction model with the highest accuracy is selected based on the results.
[0170] A further training course stipulates that the following formula be used in the forecasting model:
[0171] H
[0172]
[0173] F = [K + {S” =1 x (A) 7 )}] xy-z,
[0174] where
[0175] HF is a hardness factor in %.
[0176] K is a basic factor
[0177] a« an nxm matrix with factors which are determined empirically and / or data-driven for each element A, is,
[0178] A is an n x 1 matrix of the elements in wt.% in the order C, Mn, Cr, Si, Mo, Cu, Ni, Sn and V,
[0179] n is the number of elements,
[0180] m maps the power of the function, in particular 2, and
[0181] y, z normalization factors are used to scale the hardness factor between a fixed lower and a fixed upper limit between 0 and 100%.
[0182] Further training stipulates that the following factors are used as starting points: International patent application
[0183] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0184] 240919WO
[0185] K = 42.5;
[0186] / \
[0187] 326.29 -113.04
[0188] 36.75 3.03
[0189] 138.2 -100.02
[0190] 197.87 -245.21 i '
[0191] 111.08 3728.02
[0192] 69.92 -176.82
[0193] 84.36 -444.17
[0194] 346.39 -13112.26 /
[0195]
[0196] 83.06 -146.5 /
[0197] y = 0.6623;
[0198] z = 152.23.
[0199] Further training stipulates that for a specific type of steel, the hardness factor HF lies between a lower limit HFu and an upper limit HFo, where HFu = 0% corresponds to a hardness value of 200 HB and HF0 = 100% corresponds to a hardness value of 480 HB, in particular 440 HB, in particular 400 HB.
[0200] Further training stipulates that the correction model used establishes a ranking of the different compensation elements according to resolution- and output-specific as well as economic considerations.
[0201] The term "dissolution" refers to the dissolution of the raw materials in the melt. In this context, this means that, in case of doubt, an element is unlikely to be considered as a compensating element if its dissolution in the melt takes too long and would, in particular, delay the entire process.
[0202] The term "output" refers to yield.
[0203] Further training stipulates that the dynamically corrected contents of the compensation elements are determined in step sizes, with the step sizes corresponding to 5-10% of the interval in which the corresponding alloying element may fluctuate due to manufacturing.
[0204] This ensures that over-regulation is avoided and that procedural costs are kept as low as possible.
[0205] Further training stipulates that the correction steps for Mn are 0.01 wt.% and for Cr 0.005 wt.%. International patent application
[0206] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0207] 240919WO
[0208] Further training stipulates that the target alloy must comprise the following target alloy composition in wt.%:
[0209] C 0.38-1.4
[0210] Si 0.1-1.2
[0211] Mn 0.2-1.8
[0212] in addition, iron and unavoidable impurities,
[0213] where optionally one, several or all of the following elements are included in wt.%:
[0214] Cr 0.001-1.5
[0215] Cu 0.001-0.4
[0216] Mon 0.001-0.06
[0217] Ni 0.001-0.4
[0218] V 0.001-0.3
[0219] Sn 0.001-0.03
[0220] Further training stipulates that the target alloy must comprise the following target alloy composition in wt.%:
[0221] C 0.38-1.4
[0222] Si 0.1-1.2
[0223] Mn 0.2-1.8
[0224] in addition, iron and unavoidable impurities,
[0225] the following elements are optionally included in wt.%:
[0226] Cr 0.001-1.5
[0227] Cu 0.001-0.4
[0228] Mon 0.001-0.06
[0229] Ni 0.001-0.4
[0230] V 0.001-0.3 and / or
[0231] Sn 0.001-0.03.
[0232] Furthermore, the invention relates to a control system for the production of pearlitic rail steels of a hardness factor range specified for a rail quality, which is described in an international patent application.
[0233] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0234] 240919WO
[0235] a target alloy is ensured, and for carrying out the method according to the invention, wherein the control system comprises two components: a prediction model and a correction model, wherein the prediction model is designed to calculate the influence of influencing elements and compensation elements on the hardness factor of the final product and the correction model is designed to calculate corrected target values of compensation elements in order to bring the hardness factor into a target range.
[0236] Further training stipulates that the forecasting and / or correction model is generated using one, several, or all of the following models: linear regression models, polynomial regression models, decision tree-based regression models, random forest regression models, nearest neighbor models, neural network models, support vector machine models, ADA boost models, regression boost models, HIST boost models, XGBOOST models, generalized additive models, and symbolic regression models.
[0237] Further training stipulates that the forecasting and / or correction model is generated using linear regression models, polynomial regression models, decision tree-based regression models, random forest regression models, nearest neighbor models, neural network models, support vector machine models, ADA boost models, regression boost models, HIST boost models, XGBOOST models, generalized additive models and / or symbolic regression models.
[0238] A further training course stipulates that the following formula be used in the forecasting model:
[0239] H
[0240]
[0241] F = [K + {S” =1 x (A) 7 )}] xy-z,
[0242] where
[0243] HF is a hardness factor in %.
[0244] K is a basic factor
[0245] a« an nxm matrix with factors which are determined empirically and / or data-driven for each element A, is,
[0246] A is an n x 1 matrix of the elements in wt.% in the order C, Mn, Cr, Si, Mo, Cu, Ni, Sn and V,
[0247] n is the number of elements, International patent application
[0248] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0249] 240919WO
[0250] m maps the power of the function, in particular 2, and
[0251] y, z normalization factors are used to scale the hardness factor between a fixed lower and a fixed upper limit between 0 and 100%.
[0252] Further training stipulates that the following factors should be used as starting points:
[0253] K = 42.5;
[0254] / \
[0255] 326.29 -113.04
[0256] 36.75 3.03
[0257] 138.2 -100.02
[0258] 197.87 -245.21
[0259] 111.08 3728.02
[0260] 69.92 -176.82
[0261] 84.36 -444.17
[0262] 346.39 -13112.26 /
[0263]
[0264] \ 83.06 -146.5 '
[0265] y = 0.6623;
[0266] z = 152.23.
[0267] Further training stipulates that the correction model used ranks the target values to be calculated for different compensation elements according to resolution- and output-specific as well as economic considerations.
[0268] Further training stipulates that the dynamically corrected contents of the compensation elements are determined in step sizes, with the step sizes corresponding to 5-10% of the interval in which the corresponding alloying element may fluctuate due to manufacturing.
[0269] Further training stipulates that the target alloy must comprise the following target alloy composition in wt.%:
[0270] C 0.38-1.4
[0271] Si 0.1-1.2
[0272] Mn 0.2-1.8
[0273] in addition, iron and unavoidable impurities,
[0274] where only optionally one, several or all of the following elements are included in wt.%: International patent application
[0275] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0276] 240919WO
[0277] Cr 0.001-1.5
[0278] Cu 0.001-0.4
[0279] Mon 0.001-0.06
[0280] Ni 0.001-0.4
[0281] V 0.001-0.3
[0282] Sn 0.001-0.03
[0283] Further training stipulates that the target alloy must comprise the following target alloy composition in wt.%:
[0284] 0.38-1.4
[0285] 0.1-1.2
[0286] 0.2-1.8
[0287] in addition, iron and unavoidable impurities,
[0288] the following elements are optionally included in wt.%:
[0289] Cr 0.001-1.5
[0290] Cu 0.001-0.4
[0291] Mon 0.001-0.06
[0292] Ni 0.001-0.4
[0293] V 0.001-0.3 and / or
[0294] Sn 0.001-0.03.
[0295] The invention further relates to a track component, in particular a rail for rail vehicles, made of rail steel, manufactured according to the inventive method.
[0296] The invention is illustrated by way of example with a drawing. The drawing shows:
[0297] Figure 1 shows the process flow of dynamic alloying;
[0298] Figure 2 shows an evaluation of the model prediction accuracy;
[0299] Figure 3 shows a summary of the chemical elements and material properties of the first embodiment;
[0300] Figure 4 shows a summary of the chemical elements and material properties of the second embodiment; International patent application
[0301] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0302] 240919WO
[0303] Figure 5 shows a summary of the chemical elements and material properties of the third embodiment;
[0304] Figure 6 shows a possible microstructure when a critical molybdenum content is exceeded;
[0305] Figure 7 shows a microstructure of a standard rail with levels of influencing elements exceeding the standard limits.
[0306] The inventors have recognized that it is particularly advantageous to distinguish between influencing and compensating elements.
[0307] The effect or influence of the influencing elements is compensated by adding compensating elements.
[0308] The effect or influence of the elements already present in the melt on the material properties of the final product is determined using the prediction model. Any necessary correction is determined using a correction model.
[0309] Figure 1 illustrates the inventive method in a simplified manner. Based on a melt analysis, a decision is made as to whether the analysis lies within the specified limits for the influencing element content or deviates from them. If the influencing element content lies within the limits, the standard process is carried out, i.e., no dynamic alloying occurs. However, if the influencing element content lies outside the specified limits, the influence on at least one specified material property, in particular a mechanical property, is determined using the predictive model in order to ascertain whether the predicted property lies within or outside the target range defined by the desired rail steel grade. If the property lies within the specified target range, the standard process is carried out, i.e., no dynamic alloying occurs.However, if the predicted property lies outside the specified target range, adapted alloy element contents are calculated using the correction model. By adding appropriate compensating elements, the effect of the influencing elements can be offset to bring the property in question into the specified target range. The standard process is then resumed. After dynamic alloying, actual analysis values can also be fed back into the prediction model, if necessary. If the predicted property lies within the specified target range, an international patent application is filed.
[0310] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0311] 240919WO
[0312] The standard process is then followed. However, if the predicted property is still not within the specified target range, dynamic alloying can be performed again.
[0313] Figure 2 shows an evaluation of the model prediction accuracy. It is evident that the measured values of the hardness factor HF correlate strongly with the calculated values.
[0314] Figure 3 shows the results of an initial compensation test. This test includes a standard eutectoid rail grade for mixed traffic areas with and without influencing element loads, as well as the casting of a corresponding compensation alloy based on model predictions. It is evident that increasing the influencing element loads results in a significantly higher hardness factor (HF) than with the standard base alloy. Furthermore, the microstructure analysis reveals the presence of undesirable foreign phase components (bainite), which inevitably correlates with the increased HF value in this rail grade.
[0315] A corresponding counter-alloy proposal based on the forecast model prediction was cast as a compensation target alloy, whereby the target contents of manganese and silicon were reduced by the compensation model to maintain the original HF value corresponding to the base alloy.
[0316] Other minor fluctuations in the chemical analysis are due to metallurgical and metrological factors and are considered irrelevant. As a result, the negative effects of the influencing elements were eliminated, the HF value was reduced back to the desired standard value of the base alloy, and material and microstructure properties suitable for track operation were obtained, meaning the compensation test can be considered entirely successful.
[0317] Figure 4 shows the results of a second compensation test. This test includes a eutectoid standard rail grade with increased chromium content for applications in tight curves and heavy-duty applications, with and without influencing element load, as well as the casting of a corresponding compensation alloy based on the model prediction. It is evident that increasing the influencing element load results in a significantly higher hardness factor (HF) than with the standard base alloy. Furthermore, the microstructure analysis reveals the presence of undesirable foreign phase components (bainite), which inevitably correlates with the increased HF value in this rail grade. International patent application
[0318] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0319] 240919WO
[0320] A corresponding counter-alloy proposal based on the forecast model prediction was cast as a compensation target alloy, whereby the target contents of manganese and chromium were reduced by the compensation algorithm to maintain the original HF value corresponding to the base alloy.
[0321] Other minor fluctuations in the chemical analysis are due to metallurgical and metrological factors and are considered irrelevant. As a result, the negative effects of the influencing elements were eliminated, the HF value was reduced back to the desired standard value of the base alloy, and material and microstructure properties suitable for track operation were obtained, meaning the compensation test can be considered entirely successful.
[0322] Figure 5 shows the results of a third compensation test. This test involved a hypereutectoid standard rail grade with increased carbon content for applications under the highest stresses in heavy-duty applications, both with and without influencing element loads, as well as the casting of a corresponding compensation alloy based on the model prediction. It is evident that increasing the influencing element loads resulted in a significantly higher hardness factor (HF) than with the standard base alloy. Furthermore, the microstructure analysis revealed the presence of undesirable foreign phase components (bainite), which inevitably correlates with the increased HF value in this rail grade.
[0323] A corresponding counter-alloy proposal based on the model prediction was cast as a compensation target alloy, whereby the target manganese content was reduced by the compensation algorithm to maintain the original HF value corresponding to the base alloy.
[0324] Other minor fluctuations in the chemical analysis are due to metallurgical and metrological factors and are considered irrelevant. As a result, the negative effects of the influencing elements were eliminated, the HF value was reduced back to the desired standard value of the base alloy, and material and microstructure properties suitable for track operation were obtained, meaning the compensation test can be considered entirely successful.
[0325] If the proportion of undesirable influencing elements is too high, cracks will occur in the final product, which in the worst case, for example in railway operations, can lead to rail fractures. International patent application
[0326] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0327] 240919WO
[0328] Excessive molybdenum content can lead to undesirable microstructures (Figure 6). If a critical molybdenum content is exceeded, a mixed microstructure with pearlitic and bainitic components can form.
[0329] The potential impact of the combination of influencing elements can be investigated using end-quench tests and large-scale rolling tests on finished rails.
[0330] The end-quench tests show a significant increase in hardness of approximately 30 HB, accompanied by a substantial shift in the transformation behavior to shorter times when the influencing element load is increased to the permissible limits according to EN 13674-1. This is critical for the potential process area of large-scale production, as it promotes the formation of undesirable foreign microstructures. A further problem arises during batch transitions, as the chemical analysis, and thus the uniformity of the material properties along the rail length, is impaired.
[0331] Large-scale rolling trials of a standard rail alloyed with Z Cu + Mo + Ni = 0.28% largely confirmed the expectations from the end-quench trials and demonstrated the possibility of unacceptable microstructure formation when the valid influencing element standard limits are exceeded in combination with the standard large-scale heat treatment process. In addition to the expected bainite content, the exceedance of the standard limits resulted in the presence of additional, unusual martensite phases, which can be attributed to segregation (Figure 7). Due to their fracture-promoting nature, these martensite components certainly represent a critical case with regard to microstructure formation and must be avoided at all costs from a safety perspective.
[0332] The present method thus enables dynamic and flexible alloying, in which the influence or effect of the influencing elements is compensated by a targeted addition of the compensating elements.
[0333] The compensation elements can be selected taking into account, among other things, resolution- and yield-specific as well as economic aspects, thus significantly increasing the cost-efficiency of the overall process. International patent application
[0334] Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH
[0335] 240919WO
[0336] Furthermore, the present method makes it possible to reliably maintain the desired material properties despite the use of raw materials with an increased load of influencing elements.
Claims
International patent application Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH 240919WO Claims 1. A method for producing pearlitic rail steels of a hardness factor range specified for a rail grade, which is ensured by a target alloy, wherein at least one primary metallurgical unit leads to a first melt, wherein at least one analysis of the alloy element contents is carried out from the first melt, and wherein actual values of the melt analysis are transmitted to a control system. characterized by the fact that A distinction is made between influencing and compensating elements, whereby Influencing elements are elements that have an influence on the hardness factor of a final product, and compensating elements are elements that have an influence on the hardness factor and compensate for the influence of influencing elements, whereby the influencing elements whose influence on the hardness factor of the final product is to be compensated comprise one, several or all from the group of Cu, Mo, Ni, Sn, wherein as compensation elements one, several or all from the group of C, Si, Mn, Cr, V are chosen, wherein the magnitude of the influence of the actual values of all influencing and compensating elements on the hardness factor of the final product is known, whereby which at least one actual value is recorded with regard to those influencing and compensating elements that have an impact on the hardness factor of the final product, whereby at least one compensation element is determined for the hardness factor of the final product, which adjusts the hardness factor, whereby one or more compensating elements are added to the melt in an amount that brings the hardness factor into a target range and compensates for the influence of one or more influencing elements on the hardness factor, and wherein International Patent Application Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH 240919WO For this purpose, two components are used in the control system: a forecasting model and a correction model, whereby the forecasting model calculates the influence of the influencing and compensating elements on the hardness factor of the final product and the correction model calculates corrected target values of the compensating elements in order to bring the hardness factor into a target range.
2. Method according to claim 1, characterized in that the alloy is adjusted so that the final product conforms to the microstructure of standard EN 13674 or EN 14811.
3. Method according to claim 1 or 2, characterized in that subsequent post-treatment steps following steel production are carried out unchanged until the final product is obtained.
4. Method according to one of the preceding claims, characterized in that the analysis is carried out at least in the first melt of the last primary metallurgical aggregate of a production route, in particular from a tapping of crude steel, before or after a first addition of alloying elements and at least after a second addition of alloying elements.
5. Method according to one of the preceding claims, characterized in that the analysis is carried out by means of sampling according to EN ISO 14284.
6. A method according to one of the preceding claims, characterized in that, if a first element is introduced by a feedstock in a proportion that lies below a desired target range defined by the target alloy, the first element, if it is a desired alloying element, is added to the extent necessary to achieve the target range, or, if the first element is introduced by a feedstock in a proportion that already corresponds to the target range, is no longer added, or, if the first element is introduced by a feedstock in a proportion that lies above the target range, is compensated by a reduced alloying proportion of a second similarly acting element or an increased alloying proportion of an opposing third element.
7. Method according to one of the preceding claims, characterized in that the forecasting and / or correction model is carried out using one, several or all of the International Patent Application Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH 240919WO The following models are generated: linear regression models, polynomial regression models, decision tree-based regression models, random forest regression models, nearest neighbor models, neural network models, support vector machine models, ADA boost models, regression boost models, HIST boost models, XGBOOST models, generalized additive models, symbolic regression models.
8. Method according to one of the preceding claims, characterized in that, when considering the hardness factor, the at least one recorded actual value of the analysis is assigned to the at least one prediction model and / or correction model using the square root of the mean square deviation of the hardness factor, wherein the square root of the mean square deviation of the hardness factor is determined for all prediction models and / or correction models for the at least one actual value, and the prediction model and / or correction model with the highest accuracy is selected based on the results.
9. Method according to one of the preceding claims, characterized in that, when several material properties are considered, the at least one recorded actual value of the analysis is assigned to the at least one prediction model and / or correction model using the mean value and / or the sum of the square roots of the mean square deviations of the material properties, wherein the mean value and / or the sum of the square roots of the mean square deviations of the material properties is determined for the at least one actual value for all prediction models and / or correction models, and the prediction model and / or correction model with the highest accuracy is selected based on the results.
10. Method according to one of the preceding claims, characterized in that the following formula is used in the forecasting model: H F = [K + x (A) 7 )}] xy-z, where HF is a hardness factor in %. K is a basic factor, International patent application Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH 240919WO a« an nxm matrix with factors which are determined empirically and / or data-driven for each element A, is, A is an n x 1 matrix of the elements in wt.% in the order C, Mn, Cr, Si, Mo, Cu, Ni, Sn and V, n is the number of elements, m maps the power of the function, in particular 2, and y, z normalization factors are used to scale the hardness factor between a fixed lower and a fixed upper limit between 0 and 100%.
11. Method according to claim 10, characterized in that the following factors are used as starting points: K = 42.5; / \ 326.29 -113.04 36.75 3.03 138.2 -100.02 197.87 -245.21 i ' 111.08 3728.02 69.92 -176.82 84.36 -444.17 346.39 -13112.26 / 83.06 -146.5 / y = 0.6623; z = 152.
23.
12. Method according to one of the preceding claims, characterized in that for a specific steel grade the hardness factor HF is between a lower limit HF U and an upper limit HFo, where HFu = 0% corresponds to a hardness value of 200 HB and HFo = 100% corresponds to a hardness value of 480 HB, in particular 440 HB, in particular 400 HB.
13. A method according to one of the preceding claims, characterized in that the correction model used defines a sequence of the different compensation elements according to resolution- and output-specific as well as economic considerations. International patent application Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH 240919WO 14. Method according to one of the preceding claims, characterized in that the dynamically corrected contents of the compensation elements are determined in step sizes, wherein the step sizes correspond to 5-10% of the interval in which the corresponding alloying element may vary due to manufacturing.
15. Method according to one of the preceding claims, characterized in that the target alloy comprises the following target alloy composition in wt.%: C 0.38-1.4 Si 0.1-1.2 Mn 0.2-1.8 in addition, iron and unavoidable impurities, where optionally one, several or all of the following elements are included in wt.%: Cr 0.001-1.5 Cu 0.001-0.4 Mon 0.001-0.06 Ni 0.001-0.4 V 0.001-0.3 Sn 0.001-0.
03.
16. Control system for the production of pearlitic rail steels of a hardness factor range specified for a rail grade, which is ensured by a target alloy, and for carrying out the method according to one of the preceding claims, characterized in that the control system comprises two components: a prediction model and a correction model, wherein the prediction model is designed to calculate the influence of influencing elements and compensation elements on the hardness factor of the final product and the correction model is designed to calculate corrected target values of compensation elements in order to bring the hardness factor into a target range by adding at least one compensation element.
17. Tax system according to claim 16, characterized in that the forecasting and / or correction model uses one, several or all of the following International patent application Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH 240919WO Models that are generated include: linear regression models, polynomial regression models, decision tree-based regression models, random forest regression models, nearest neighbor models, neural network models, support vector machine models, ADA boost models, regression boost models, HIST boost models, XGBOOST models, generalized additive models, and symbolic regression models.
18. Control system according to claim 16 or 17, characterized in that the following formula is used in the forecasting model: HF = [K + {S” =1 x (A) 7 )}] xy-z, where HF is a hardness factor in %. K is a basic factor a« an nxm matrix with factors which are determined empirically and / or data-driven for each element A, is, Ai an n x 1 matrix of the elements in wt.% in the order C, Mn, Cr, Si, Mo, Cu, Ni, Sn and V is, n is the number of elements, m maps the power of the function, in particular 2, and y, z normalization factors are used to scale the hardness factor between a fixed lower and a fixed upper limit between 0 and 100%.
19. Control system according to claim 18, characterized in that the following factors are used as starting points: K = 42.5; / \ 326.29 -113.04 36.75 3.03 138.2 -100.02 197.87 -245.21 111.08 3728.02 69.92 -176.82 84.36 -444.17 346.39 -13112.26 / 83.06 -146.5 'International patent application Voestalpine Rail Technology GmbH; Voestalpine Stahl Donawitz GmbH 240919WO y = 0.6623; z = 152.
23.
20. Control system according to one of the preceding claims 16 to 19, characterized in that the correction model used ranks the target contents to be calculated on different compensation elements according to resolution- and output-specific as well as economic considerations.
21. Control system according to one of the preceding claims 16 to 20, characterized in that the dynamically corrected contents of the compensation elements are determined in step sizes, wherein the step sizes correspond to 5-10% of that interval in which the corresponding alloying element may vary due to manufacturing.
22. Control system according to one of the preceding claims 16 to 21, characterized in that the target alloy comprises the following target alloy composition in wt.%: C 0.38-1.4 Si 0.1-1.2 Mn 0.2-1.8 in addition, iron and unavoidable impurities, where optionally one, several or all of the following elements are included in wt.%: Cr 0.001-1.5 Cu 0.001-0.4 Mon 0.001-0.06 Ni 0.001-0.4 V 0.001-0.3 Sn 0.001-0.
03. Track component, in particular rail for rail vehicles, made of rail steel, manufactured according to the method of one of claims 1 to 15.