An integrated batch annealing furnace simulator

Transcription

1 J. Phys. IV France 120 (2004) EDP Sciences, Les Ulis DOI: /jp4: An integrated batch annealing furnace simulator S.S. Sahay Tata Research Development & Design Centre, 54B, Hadapsar Industrial Estate, Pune, India Abstract. An integrated batch annealing furnace simulator with the capability of predicting spatial and temporal evolution of temperature, microstructure and mechanical properties of the coils during the batch annealing operation has been developed. The prediction capability of this integrated simulator has been extensively validated with data collected from several industrial batch annealing operations. In this article, the problems in controlling a batch annealing operation via conventional temperature based control strategy has been highlighted. These problems can be effectively resolved by using the integrated simulator. Furthermore, the utility of this simulator has been illustrated by a case study on optimization of coil dimensions for maximization of furnace productivity. 1. INTRODUCTION During the past decade, attempts have been made in developing automotive panels from aluminum alloys to reduce weight (density of aluminum is 2.7 g.cm -3 as compared to 7.8 g.cm -3 for steel) for achieving the reduction in fuel consumption. However, the high cost of aluminum alloys as well as low Young modulus (70 GPa for aluminum as compared to 210 GPa for steel, demanding additional stiffening ribs) make their usage in passenger cars prohibitively expensive. As a result, 95% of the automotive panels are still being made from cold rolled annealed steel sheets. These sheets are extensively used in the high performance applications, such as automotive panels and white goods, where in addition to stringent mechanical property requirements, dimensional tolerance and surface finish are extremely important. In addition, the stiff global competition in this niche market necessitates an optimum and low cost operation. In a cold rolling mill, hot rolled steel strips are rolled at room temperature to achieve improved surface quality and mechanical properties. However, extensive deformation given to the steel at room temperature during the cold rolling operation significantly reduces the ductility and formability of cold rolled sheets. This necessitates annealing, where the cold rolled sheets get stress relieved through the mechanisms of recovery, recrystallization, and grain growth. Annealing of cold rolled steel sheets are carried out either in continuous annealing furnace, where the sheets after cold rolling enters an integral tunnel type furnace for annealing, or in a batch annealing furnace, Figure 1. Schematic diagram of batch annealing furnace. Furnace Cover H 2 Gas Coil

2 810 JOURNAL DE PHYSIQUE IV where the cold rolled sheets are coiled and subsequently stacked for annealing in a cylindrical furnace. In the batch annealing process, 4-5 cylindrical steel coils (typically tonnes each) are stacked on a furnace base (Fig. 1) and annealed under hydrogen atmosphere. The air-gaps between the sheets in the coils result in very low radial thermal conductivity. As a result spatial variation in temperature is prevalent during the annealing process. The outer surfaces of the coil, referred to as hot-spot, are heated faster and achieve the annealing temperature in shorter time as compared to the inner core of the coil, commonly referred to as cold spot. Since recrystallization and grain growth are thermally activated processes, such thermal lag leads to spatial variation in microstructure, with an associated variation in the mechanical properties within a coil. In addition due to the axial temperature variation along the furnace, there is coil-to-coil variation in the microstructure and mechanical properties [1,2]. Although an increase in soaking time usually results in reduction in microstructural and mechanical property variations, it also reduces the furnace productivity. Therefore, selection of soaking time in an industrial batch annealing operation requires an optimization between productivity and quality [3]. In addition, appropriate selection of heating rate, which has a metallurgical implication on precipitation and recrystallization kinetics, and annealing temperature form the crux of batch annealing thermal cycle design [4]. Batch annealing operation has significant influence on all the important plant performance parameters, such as energy consumption, plant productivity, emissions, as well as quality parameters, strength, ductility, drawability and formability [3]. In view of its relevance on all these key parameters, it is essential to optimize the batch annealing operation for maximum productivity and minimum energy consumption, while achieving the specified product quality. Despite being a critical operation, generally the batch annealing cycles at the industrial scale are designed through plant trials and empirical methods, which in addition to being time consuming and expensive, may at best provide sub-optimal result. Instead, the process cycle can be effectively optimized through a process simulator, which can emulate the batch annealing operation, thus reducing the number of plant trials required for plant optimization. Although, mathematical models for batch annealing operation have been available [5] for over a decade, these models are primarily thermal models, where only the temperature evolution during the batch annealing process are predicted. Using these models, the temperature differential between the hot and cold spots is considered as a criterion for the completion of soaking, which is generally used in controlling the batch annealing operations. Although, these models are extremely useful both for off-line calculations as well as for on-line process control, they are limited to the temperature estimation and do not attempt to predict the consequent microstructure or mechanical property variation across the annealed coils. In order to bridge this gap, an integrated batch annealing furnace simulator (BAFSIM TM ), with prediction capabilities extended to microstructural and final mechanical property, has been developed. This model has the ability to predict spatial and temporal evolution of temperature, microstructure and final mechanical properties of the coils undergoing batch annealing. The major advantage of this integrated model lies in its capability to design the batch annealing cycles directly on the basis of requirements of microstructure and mechanical property specifications, rather than indirectly estimating them using temperature differentials between the hot and cold spots. The prediction capability of this simulator has been validated with data collected from several large-scale industrial batch annealing operations. The details of this integrated model and its validation with plant data have been presented earlier [1,2]. In this article, the emphasis is on illustrating the utility of such an integrated simulator in process optimization and efficient control of the furnace operation. In the section 2 of this article, the simulator architecture is briefly outlined. The problems in controlling a batch annealing operation via conventional temperature based control strategy have been highlighted in section 3. Finally, a case study on optimization of coil dimensions for maximization of furnace productivity has been presented in the section 4.

3 ICTPMCS SIMULATOR ARCHITECTURE The basic objective of the simulator is to emulate the batch annealing operation (Fig. 2), such that based on the process inputs, e.g. coil dimensions and temperature set-points, it can predict the spatial distribution and temporal evolution of temperature, microstructure and mechanical properties of the coils. BAFSIM TM essentially comprises three modules: (i) thermal, (ii) microstructural and (iii) mechanical property modules. The transient temperature profiles at different locations in the coil are calculated by the thermal module, which serve as inputs to the microstructural module, where grain size is calculated on the basis of recrystallization and grain growth kinetics. Finally, the mechanical properties are estimated from the established microstructure-property correlation. These modules are Inputs Coil Dimensions Grade T-t setpoint Batch Annealing Furnace Simulator Outputs T-t distributions Microstructure Final Properties Figure 2. Typical inputs and outputs of the batch annealing simulator. briefly outlined below. In the thermal module, interactions among different components of the furnaces are considered. All the three modes of heat transfer (conduction, convection and radiation) are used to determine the transient temperature variation in coils and different furnace components, such as flue gas, furnace wall, protective cover, cooling hood, and inert gas. For example, the thermal profiles of coils are obtained by solving the following energy equation in cylindrical coordinates [1,2,5]. T k z T z 1 rk r r T r C m m m m m z r (1) where T m is the temperature, m is the density, C m and k z are temperature dependent specific heat and conductivity of the coil. The radial conductivity (k r ) of the coil depends on the sheet thickness and air gap between sheets. Using the appropriate boundary conditions [1,2], the solution to above equation provides the complete transient temperature history during heating and cooling cycles, at different locations of the coil. These temperature profiles at various locations, serve as the input to the microstructural module, where using the recrystallization and grain growth kinetics, the grain size distribution is computed at various locations of the coils. In the microstructural module, the recrystallization kinetics has been modeled using the JMAK approach, whereas grain growth behavior has been modeled using a Beck type correlation [6,7]. For example, in the JMAK theory the fraction recrystallized is given by [6]: n X 1 exp( k t rex rex ) (2) where X is the volume fraction recrystallized after time (t), k rex is the temperature dependent constant and n rex is the Avrami exponent. Effect of precipitation on recrystallization and grain growth was incorporated for AlK grade steel, whereas for interstitial free steel it was ignored. The mechanical properties - such as strain hardening exponent (n), tensile strengths ( y, UTS ), hardness and percent elongation (%El) - in various coils are derived through appropriate microstructure-property relations. For example, the effect of grain size on yield strength of steel is given by the Hall-Petch relationship [8]: k hp YS o (3) D

4 812 JOURNAL DE PHYSIQUE IV where YS is the yield stress, o is the frictional stress required to move dislocations, k hp is the Hall- Petch slope, and D is the grain size. Although the basic framework of the simulator is based on fundamental equations, for engineering realization of mathematical models several parameters - such as heat transfer coefficients, gas emissivity, and material properties, which are specific to the furnace and operating conditions - must be used. In reality, these values are plant-specific and therefore may vary from plant to plant. These parameters are evolved from the data obtained from controlled plant experiments, for tuning the model. The thermal models of the simulator were tuned to actual industrial furnaces through the inplant experiments, where, by embedding several thermocouples in the stack, temperatures were monitored across the coils during the annealing process. The microstructural evolution coefficients for the relevant grades were determined by conducting kinetic experiments in the relevant temperature time regimes. Finally, mechanical properties were correlated to the grain size by carrying out tensile and hardness tests on samples with wide range of grain sizes. The results from these experiments have been presented earlier [1,2]. It must be noted that the thermal profiles obtained at different locations in the coil are non-isothermal, whereas the recrystallization and grain growth kinetics described by JMAK and Beck type formulations are for isothermal conditions. These isothermal kinetics equations were utilized for monitoring recrystallization and grain growth over the nonisothermal temperature profiles by segmenting the thermal cycles into small isotherms [9] and integrating the recrystallization and grain growth kinetics over time. As this quasi-thermal approach does not capture special nonisothermal effects [10], special heating rate controlled experiments [11] were also carried out to improve the prediction capability. In addition, efforts are underway to develop better fundamental models where heating rate effects can also be incorporated in the model. At present, in the integrated model, the thermal profiles at various locations are obtained, these temperature profiles are segmented and subsequently the progress of recrystallization at various locations is monitored. After the recrystallization is complete, grain growth is initiated at that location. Using the final grain size distribution across the coils, mechanical property parameters such as yield strength, tensile strength and elongation are computed using Hall-Petch type relations. 3. TEMPERATURE VS. MICROSTRUCTURE BASED FURNACE CONTROL Most of the modern batch annealing operations are controlled on the basis of temperature differential ( T) between hot and cold spots at the end of soaking cycle. Depending on the grade and permissible process variability, the temperature differential varies from 20 to 40 o C. It is generally assumed that by controlling the temperature differential, the variability of the quality parameters, namely grain size, tensile strength, ductility and hardness can be controlled. A series of simulations were conducted using the batch annealing simulator to verify this common presumption. In these simulations, a stack of four interstitial free grade steel coils was taken. The inner and outer coil diameters were varied in the ranges of 400 to 725 mm and 1500 to 1950 mm, respectively, for which Heating Soaking Cooling around 150 simulations were carried out. For each stack, the outer and inner diameters 400 were kept constant, the sheet thickness was Control TC Hot Spot 1mm and the coil width was 1200 mm. The heating profile was kept constant for all 200 Cold Spot these simulations, whereas soaking was carried out until a T of 40 o 0 C was achieved. The cooling was carried out until a core temperature of 150 o C was achieved. A Time, hrs Figure 3. A typical temperature typical temperature profile of the control profile. Temperature, o C

5 ICTPMCS 813 thermocouple, hot spot and cold spot is shown in the Fig. 3. This temperature profile is similar to an industrial batch annealing operation. As the T was kept constant at 40 o C for all the simulations, it is expected that the grain size will not vary with the coil dimensions. It must be noted that in an industrial operation, the inner coil diameter is generally kept constant while outer coil diameter vary from batch to batch, depending on the customer requirements. As can be seen from the results presented in Fig. 4, both minimum and maximum grain sizes, on which the mechanical properties are also strongly dependent, vary significantly with the coil dimensions. These results suggest that the conventional control of batch annealing furnaces on the basis of T will result in variation in final product quality. With increasing thrust on reducing the process variability and increasing the Sigma Level of the process, such variation in product quality is not desirable. On the other hand, if the operation is controlled by the integrated model, irrespective of the coil dimensions, the soaking time can be optimally controlled to achieve the desired microstructure (grain size) as well as the mechanical properties, and thereby the process variability can be tightly controlled. 28 (a) Maximum Grain Size 23 (b) Minimum Grain Size Grain Size m Grain Size m Figure 4. Variation in (a) maximum and (b) minimum grain sizes as a function of outer coil diameter for constant temperature differential ( T) of 40 o C at the end of soaking cycle. 4. OPTIMIZATION OF COIL DIMENSIONS The coil dimensions, e.g. inner and outer diameters, are expected to have significant influence on the furnace productivity. However, the changes in coil dimensions do not have a linear relationship with the plant performance parameters, such as productivity, energy consumption or quality. For example, when the inner diameter is decreased or the outer coil diameters is increased, the stack weight increases, which positively affects the furnace productivity. On the other hand, with increase in outer coil diameter and decrease in inner coil diameter, the cycle time - corresponding to a constant product quality - also increases. The increase in cycle time is due to increase in required soaking time - for achieving equivalent grain size/mechanical properties - as well as increase in cooling time, to attain equivalent core temperature at the cover removal. The overall BAF productivity will increase with increase in outer coil diameter or decrease in inner coil diameter, only if the positive contribution from increased stack weight dominates over the increase in cycle time. In this section, these conflicting effects of coil outer and inner diameters on the furnace productivity have been examined by carrying out simulations with inner diameter in the range of 400 to 700 mm and outer diameter in the range of 1500 to 1950 mm. The sheet thickness was kept constant at 1 mm and the coil width was taken to be 1200 mm. With four identical dimensional coils in each stack, around 150 simulations were carried out for this study. It must be noted that in these simulations, the objective was to determine the furnace productivity as a function of inner and outer coil diameters, while achieving identical quality parameters, namely, grain size, tensile strength, hardness and core temperature at the end of the cycle. In order to calculate the annual furnace productivity, the stack weight and number of cycles per year is required. The stack weight was directly computed from the coil dimensions, whereas for the number

6 814 JOURNAL DE PHYSIQUE IV of cycle per year for a constant product quality, cycle time as a function of inner and outer diameter was determined from the simulator Optimization of outer coil diameter The effect of outer coil diameter on stack weight, cycle time and furnace productivity is presented in Fig. 5. As is evident from Fig. 5a, the outer diameter of the coil has significant effect on the stack weight. For example, when the outer coil diameter is increased from 1500 to 1950 mm, for a constant inner coil diameter of 625 mm, the stack weight increases by around 83%. The heating time corresponding to the end of the soaking segment, to achieve a minimum grain size of 19 m in the stack, was determined from the simulator. As can be seen from Fig. 5b, the heating time increases for all the inner coil diameters, when the coil outer diameter is increased. For example, for a constant inner diameter of 625 mm, the heating time increases by around 15%, when the outer coil diameter is increased from 1500 mm to 1950 mm. Similarly, the cooling time to achieve a maximum core temperature of 150 o C was determined from the simulator. Fig. 5c show that the cooling time also increases with increase in outer coil diameter. In the case of inner coil diameter of 625 mm, when the outer coil diameter is increased from 1500 to 1950 mm, the cooling time increases by around 55%. It must be noted that the heating and cooling time have inverse relationship with the furnace productivity, i.e. when the heating and cooling time is increased, the furnace productivity decreases. From these results, the furnace annual productivity was determined from the stack weight, annual available time (24 hrs x 300 working days) and cycle time (sum of ramp-up time, soaking time and cooling time). The results are presented in Fig. 5d, where the annual furnace productivity is plotted as a function of outer coil diameter. This figure shows that the furnace productivity increases with an increase in the outer coil diameter, which suggest that the large increase in stack weight more than compensates the negative effects due to increase in cycle time. Continuing with the case of stacks with 625 mm inner diameter, there is around 44% increase in furnace productivity, when the outer coil diameter is increased from 1500 to 1950 mm. This effect is mainly due to the fact that the rate of increase of the coil weight per unit coil thickness is higher than the rate of increase in the cycle time per unit coil thickness. This study indicates that the outer coil diameter should be as large as possible to obtain maximum productivity and offers an avenue to explore when higher productivity is needed. However, the maximum coil size will be limited by the hot coil/cold coil design, furnace, handling equipment capabilities, etc., which need to be considered. Fig. 5d, also show that the furnace productivity, especially at the high outer coil diameters, is less sensitive to the coil inner diameter, which has been investigated in details in the subsequent subsection Optimization of inner coil diameter The variation in stack weight, cycle time and furnace productivity as a function of inner coil diameter is presented in Fig. 6. Similar to the simulations described in section 4.3, these simulations were carried out with 4 coils stack configuration, having coil width of 1200 mm and sheet thickness of 1 mm. With decrease in inner coil diameter the coil thickness decreases, and therefore the stack weight as well as the cycle time is expected to increase. However, as can be seen from Fig. 6a, decrease in coil inner diameter has marginal influence on stack weight (Fig. 6a) and although the heating and cooling time increases (Fig. 6b,c), the productivity show negligible influence (Fig. 6d). For example, for an outer coil diameter of 1750 mm, when the inner coil diameter decreases from 400 to 725 mm, the stack weight marginally increases by 8%, the heating and cooling time show a mere increase of 1.4% and 5%, respectively, which results in nearly insignificant (2.6%) increase in productivity. The results indicate that as the coil inner diameter is decreased, the effect of increase in stack weight is more or less compensated by the increase in cycle time, resulting in nearly constant furnace productivity. This is in contrast to the effect of outer coil diameter (section 4.1), which dominated and exhibited a significant influence on productivity. When the variation in productivity

7 ICTPMCS 815 with the inner coil diameter was closely examined, it was noted that the productivity goes through a maximum value before decreasing. As shown in Fig. 7, when the inner coil diameter is increased from 400 to 725 mm, the productivity go through a maximum for the inner coil diameter of 475 mm, before it starts to decrease. It suggest that in the initial phase ( mm), when the coil inner diameter is increased the resultant decrease in cycle time more than compensates the decrease in stack weight, thereby increases the furnace productivity. However, as noted earlier, the over all productivity change due to increase in inner coil diameter is insignificant. The results presented in this section illustrate that the outer coil diameter has more pronounced effect on the furnace productivity, and therefore should be focused in a furnace optimization exercise. Stack W eight, tonnes Cooling Time, hrs (a) Stack Weight (c) Cooling Time for Constant Core Temperature Heating Time, hrs Productivty, tonnes/year ,000 16,000 12,000 (b) Heating Time for Constant Grain Size (d) Furnace Productivity for Constant Quality Figure 5. Variation in (a) stack weight, (b) heating and soaking time to achieve a constant minimum grain size, (c) cooling time to achieve a constant core temperature and (d) furnace productivity for constant quality, as a function of outer coil diameter. 5. SUMMARY Batch annealing is an important process in the cold rolling mill, having influence on all the key plant performance parameters. An integrated process model, which can emulate the batch annealing process, has been developed. This simulator has the capability to predict the evolution of temperature, microstructure and mechanical properties at various locations in the coils during the batch annealing operation. In this article, it has been shown that the detailed predictions from the integrated simulator (e.g. temperature, microstructure, mechanical property) can be also related to the furnace performance parameters, such as product variability and furnace productivity. In the first case study, it has been shown that controlling a batch annealing operation via conventional temperature based control

8 816 JOURNAL DE PHYSIQUE IV strategy can result in high variability in the product quality, which can be effectively reduced by using the integrated simulator. Subsequently, it has been shown that the simulator can be used to optimize the coil dimensions for maximum furnace productivity. The outer coil diameter was found to have a significant effect on the furnace productivity, whereas inner coil diameter marginally affected the furnace productivity. It was noted that an increase in outer coil diameter monotonically increases the furnace productivity, whereas decrease in furnace inner diameter increases the furnace productivity up to a maximum value, beyond which it showed a decrease. Therefore, for maximum BAF productivity, the outer coil diameter should be kept at the maximum possible value, while the inner diameter should be kept at the optimum diameter. Stack Weight, tonnes Cooling Time, hrs (a) Stack Weight (c) Cooling Time for Constant Core Temperature Heating Time, hrs Productivty, tonnes/year (b) Heating Time for Constant Grain Size ,000 (d) Furnace Productivity 16,000 12, Figure 6. Variation in (a) stack weight, (b) heating and soaking time to achieve a constant minimum grain size, (c) cooling time to achieve a constant core temperature and (d) furnace productivity for constant quality, as a function of inner coil diameter. Productivity, tonnes/year 16,400 16,200 Outer Coil Diameter 1750 mm 16, Figure 7. Variation in furnace productivity as a function of inner coil diameter for a constant outer coil diameter of 1750 mm.

9 ICTPMCS 817 Acknowledgements The author is grateful to Professor M. Joseph, Executive Director, TRDDC, for approving and supporting this work. References [1] Sahay S.S, Kumar A.M., Singh S.B., Bhagat A.N., and Sharma M.S.S., Tata Search (2001) [2] Sahay S.S., Kumar A.M. and Chatterjee A., Ironmaking and Steelmaking (2003) in press. [3] Sahay S.S. and Kumar A.M., Mater Manuf Process, 17 (2002) [4] Hutchinson, Int. Met. Rev., 29 (1984) [5] Jaluria Y., Int. J. Num. Meth. Eng. 25 (1988) [6] Humphreys F.J., Hatherly, M. Recrystallization and Related Annealing Phenomenon, (Pergamon, Elsevier Science Ltd., Oxford, UK, 1996) pp [7] Kozeschnik E., Pletenev V., Zolotorevsky N., Buchmayr B., Metall. Mater. Trans. 30A (1999) [8] Meyers M.A., and Chawla K.K., Mechanical Behavior of Materials, (Prentice-Hall Int., London, 1999) pp [9] Jiao S., Penning J., Leysen F., Houbaert Y., and Aernoudt E., ISIJ Int., 40 (2000), [10] Sahay S.S., Malhotra C.P., and Kolkhede A.M., Acta Mater. 51 (2003) [11] Sahay S.S., Kumar B.V.H., and Krishnan S.J., Microstructural Evolution during Batch Annealing, Proc. Intl. Conf. on Advances in Materials and Materials Processing, Kharagpur, India, Feb. 2002, pp