Automation and Thickness Control in Hot Strip Mill
Automation and Thickness Control in Hot Strip Mill
The economic efficiency of metal rolling processes is strongly correlated to the quality level of the end-rolled products. Rolling of flat steel products is a complex process where the quality of the product is influenced by a range of factors such as incoming material, mechanical and electrical equipment, operating parameters, lubrication, and automation and control strategies etc. The significant quality parameters are material thickness, material shape and surface, and the homogeneity of stress distribution. For optimized cost-efficiency and to maximize material usage, tight tolerances for the thickness are essential, to enable the strip to be rolled down as closely as possible to the minimum permissible thickness. Product quality can only be effectively optimized if the mechanical, electrical and instrumentation equipment as well as the control strategy solution combine together well.
The hot strip mills (HSM) process reduces by compressing the continuously cast steel slabs having rectangular cross section and thickness in the range of 250 mm to 350 mm into steel flat strip till a desired thickness is reached. Several HSMs have capabilities to produce strips having as small as 1 mm thickness. The process steps in a typical HSM are (i) reheating of slabs in a pusher or a walking beam type reheating furnace in order to reach the optimal temperature, (ii) a roughing mill (either a reversing mill or a continuous mill consisting of a number of stands) for achieving a preliminary thickness reduction, (iii) the finishing mill consisting of 5 to 7 consecutive rolling stands which reduces the thickness to the desired value, and (iv) coiling the long strip in a coiler.
In the finishing mill of HSM, an important task is performed by a hydraulic arm, called the looper, placed in the middle between two consecutive stands and whose purpose is to keep the strip tension at a constant value. This mechanical system is subject to particularly unstable dynamics which make the control issue tricky
The processing of slab to hot rolled strip in the HSM is achieve through the several process steps whose complexity involves mechanical and automation technologies. Hot rolling in a HSM needs not only mechanical solutions but also appropriate control technologies. The process of rolling in the HSM can be controlled though a standard software and automation architecture which includes four automation levels.
The automation system is not the sole determinant of performance of the hot strip mill. However, for any given configuration of mechanical and electrical equipment, the potential performance of the mill is only being achieved with high-performance control and automation. Attention is required to be focused on throughput and quality, where control is particularly important in achieving good performance. Normally the throughput and quality interact in both positive and negative ways and these interactions are to be considered in defining the control system.
Throughput – The ultimate throughput which can be achieved in a mill is limited by the capabilities of the mechanical and electrical hardware. To achieve throughputs consistently close to this limit needs high-quality control and automation. At high throughputs, three or more work pieces can be in the rolling mill at different stages of processing at the same time. To avoid catastrophic collisions in the mill, accurate tracking is essential. The tracking system uses signals from mill instrumentation and process information (for example, as a piece is rolled, so its length increases) to maintain a dynamic map of the mill. It is, of course, to be robust against the loss of individual mill instruments.
Throughput control looks ahead at the rolling schedule and determines which part of the mill installation, furnace, roughing mill, finishing mill, or coiler, can limit throughput. The limiting process is then controlled to achieve maximum throughput and other parts of the process are controlled to match this throughput. This results in an improvement in energy efficiency and a reduction in wear and tear on the equipment, thus reducing costs.
Throughput and quality also interact. As throughput increases, control becomes more difficult, and to maintain the required level of quality and yield needs careful design of the control system. Quality and throughput control also interact in positive ways. For example, for achieving a greater range and accuracy of temperature control out in the finishing mill, inter-stand cooling sprays are normally installed. These are to be controlled to maintain the strip temperature at the mill exit but, further, they can be used to increase the speed at which the work piece is rolled in the finishing mill, while maintaining the target exit temperature.
Quality – A primary aim of the automation system is to control the mill equipment so that the rolled coils meet the dimensions (gauge, width, profile, and flatness) and material properties as per the requirement of the specifications. There are two aspects to controlling the quality parameters namely (i) control of the head end of the work piece as it threads the mill, and (ii) control of the mill equipment to maintain the desired quality parameters through the rolling of the coil.
There are two control modes namely (i) mill set-up, and (ii) dynamic control. A fundamental difference in control strategy is imposed on the two modes by the availability of measurements. As the strip threads in the mill, there are no measurements of the final quality parameters, the strip simply has not reached the measuring instruments, and control is achieved by feed forward and model-based control. Once the mill is full, direct measurements of (some of) the final quality parameters are available, and dynamic feed-back control comes into operation. Accuracy in both modes of control is important, and good head-end quality parameters lead to a high yield. Width control in the HSM is also important. Coils are frequently marketed by length rather than by weight, and hence, any excess width represents a yield loss.
The shape of the strip is defined by two parameters which interact namely (i) profile, and (ii) flatness. Profile is the thickness variation across the width of the strip and, for downstream processing it is required to be controlled. There is a need for uniform thickness both along the strip length and across its width. Flatness is the ability of the strip to lie flat without applying any external forces and is also important for the downstream processing of the strip. Flatness defects are induced by poor control of proportional profile (profile divided by thickness) through the mill and, hence, there is an interaction between the profile and the flatness control.
The flatness-control problem also differs from that of profile since flatness is important, not just at the mill exit, but in the inter-stand gaps between the finishing mill stands. Bad flatness defects between stands can lead to instability of the rolling process in the finishing mill, resulting in a complete loss of control and the destruction of the coil resulting into a cobble. This represents a yield loss and also affects mill availability by stopping the process while the mill is cleared of cobble.
In addition to the dimensional parameters, there are other quality parameters which are important. A particularly important objective is the control of the mechanical properties of the finished strip. Mechanical properties are determined, to a large extent, by microstructure, and the microstructure itself is determined, to some extent, by the strain, and to a large extent by the temperature history of the rolled coil. In current control and automation systems, control of microstructure is achieved indirectly by controlling the temperature evolution as the strip is cooled on the run out table between the mill exit and coiler. The mill metallurgist defines the target cooling trajectory, and the control system adjusts the cooling sprays on the run out table and the finishing mill speed to match the desired temperature trajectory as accurately as possible.
Control-system structure – The control objectives for the HSM are expressed in terms of throughput and product quality parameters. However, the practical scope of control covers a very wide range of applications ranging from individual local high-speed position control loops with operational speeds at the milli seconds or sub-milli seconds level to the overall work piece scheduling task which operates on an hourly or longer time-scale. All these controls contribute to the overall performance needed from the automation system but the objectives are frequently expressed in terms of sub-goals more appropriate to the time-scale of the particular controller. For example, a position loop’s goal can be expressed in terms of the rate of change and overshoot in response to a demanded position change, while the overall scheduling of products through the mill can be expressed in terms of speed of satisfying the production plans.
The automation technology applied in the hot strip mill is normally divided in four levels referred to as levels 0, 1, 2, and 3, respectively. In all these automation levels which need to cooperate hierarchically in order to achieve the best performances and the highest productivity levels, a number of control technologies, mathematical models of physical phenomena and optimization algorithms are implemented. The hierarchical structure of a control automation system normally adopted for hot strip mill is shown in Fig 1.
This difference in time-scale and scope of the individual controls is reflected in the multi-level structure of the control systems now widely used on rolling mills. In Fig 1, the block diagram of such a multi-level system is shown. The separation of functions between the various levels is not sharp, and whether specific functions are implemented in, say, level 1 or level 2 can vary from installation to installation and in response to the development of better control methods and equipment.
Fig 1 Structure of a typical automation system for hot strip mill process
Level 0 – It is the lowest level of control and includes, for example, the control loops for hydraulic capsules used to position the rolls in the rolling mills and for the main electric motors powering the mill.
Level 1 – It is primarily concerned with in-piece control. At this level, the quality parameters such as strip thickness and temperature start to appear. However, the objectives for the level-1 loops are frequently sub-goals supplied by level 2. For example, level 1 control loops function to control the exit thickness out of intermediate stands in the finishing mill, and this exit thickness pattern through the mill is set by level 2 to achieve the required mill exit thickness within machine and process constraints.
The level 1 automation directly interacts with low-level devices (actuators and transducers). Real-time control loops and logic sequences are implemented here. Fast sampling (1 milli second) and high computing power are achieved, for example, through VME (Versa Module European) architecture technology. Conventional PLC, instead, guarantees a minimal sample time of 10 milli seconds. The Human–Machine Interface (HMI) offers to the operator a real-time look at the process.
Level 2 – It directly addresses the control of the quality and throughput parameters. Its domain of operation is much wider than level 1, and, in a well developed system, covers the integrated control of the reheating furnace, roughing mill, finishing mill, and run out table. It is very much concerned with the set-up control of the mill from work piece to work piece, but frequently includes part of the dynamic in-piece control also. Normally the dynamic control at level 2 is concerned with the overall coordinated control of, say, the finishing mill. Local control loops are more appropriate to level 1. Much of the control at level 2 is feed forward in nature and model based. Good control performance needs predictive models of the process (e.g. deformation and temperature models) and dynamic models of the mill machines and sensors. Most of the process models necessary for state of the art control of rolling mills reside here.
The level 2 automation provides higher-level control functions and utilities, like optimal plant setup calculation, generation of production reports and statistical analysis of product quality. In particular mathematical models of technological processes are used to generate proper plant setups. Reliability of physical models, at different and even time-varying working conditions, is achieved by self-adaptation that is, identification techniques based on plant feed-back that improve recursively the reliability of the model predictions. Technological information and historical archive of production are stored into the data base (DB), while the process work station (PWS) offers a graphic interface to the level 2 utilities.
Level 3 – The level 3 automation system is implemented in order to provide additional utilities for top-level production supervision (this is also known as ‘Manufacturing Execution System’ (MES) functions), storage yard management and coordination among levels 2 of the different processes belonging to the same plant. The level 3 automation system is in charge of coordinating the production scheduling between the production process of HSM and the production processes of the upstream and downstream units.
The level 3 functions mainly as a scheduler of the hot strip mill. It takes the order book for the mill and organizes it into rounds of, typically, 100 – 200 work pieces which comply with the scheduling rules developed for the mill. As well as the HSM itself, level 3 takes account of upstream and downstream processes and stock areas. The scheduling rules used are, essentially, a global model of the rolling mill which enables the scheduler to organize the round so that the required quality parameters can be achieved within the constraints imposed by the mill equipment. Traditionally, the most important factor is the evolution of the profiles of the rolls in the mills caused by their wear and thermal expansion. From this derives the `coffin’ schedule which means start narrow, quickly build up to wider material as the thermal crowns increase on the rolls, and gradually fade back to narrow as the rolls wear.
Control technologies applied in HSM
The use of advanced control and modeling solution for HSM has been subject to several developments in various directions in the past 40-50 years. These include (i) the use of multivariable control techniques which has been proposed for the finishing mill since the 1970s and now it is considered a consolidated tool for controlling the generic rolling stand together with the downstream looper or the downstream coiler, (ii) development of various models in order to predict the material characteristics according to the material temperature and the rolling process and application of controls for regulating the coiling temperature, (iii) development of advanced control techniques which are applied in order to compensate friction phenomena, (iv) development of models and controllers in order to improve the material flatness and profile, and (v) development of steering control techniques which have been recently introduced in order to increase the productivity levels by reducing the probability of cobble events. Fig 2 shows typical layout of hot strip mill with level-1 controls.
Fig 2 Typical layout of hot strip mill with level-1 controls
Control technologies for thickness regulation
Fig 3 shows an example of a thickness regulation as applied to the HSM normally with the provision of the following sensors. The instruments and sensors needed for the thickness regulations are given below.
Thickness and profile gauge – This gauge is based on x-ray technology and is aimed at measuring the thickness in the centre-line of the work piece. The gauge is rarely mounted on a moving carriage and can measure the whole thickness profile along the width of the coil. Normally one thickness / profile measurement system is installed at the end of the last stand in the mill.
Load cells – These are provided in order to have a measurement of the rolling force which represents a basic measurement signal for the HSM thickness regulation. In case the load cells are not provided then the measurement of the hydraulic force signal generated by the pressure transducers installed in the main cylinder can be used exploited as an alternative measure.
Load cells in some cases are mounted on the loopers in order to get a direct measurement of the inter-stand strip tensions. Also in this case an alternative measure is represented by the force signal generated by the pressure transducers mounted in the hydraulic cylinder acting on the looper.
Fig 3 Thickness control in hot strip mill
There are certain features which distinguish between basic controllers and external controllers. The basic controllers are those controllers which are in charge of implementing references for physical actuators. On the other hand, the external controllers are those controllers which produce references for basic controllers in order to reach the desired target. The thickness control is achieved by means of the following basic controllers.
Hydraulic gap control (HGC) – HGC is done by a controller which receives a gap reference and measures the gap coming from position encoders placed in the hydraulic cylinder and produces the servo-valve command which indeed controls the oil mass flow generating the movement of the cylinder. Obviously, the measured gap can be significantly different from the physical gap of the stand because of the stand elastic stretch.
Torque controller (TC) – The torque controller controls the torque generated by the two reels. These controllers receive a torque reference which is produced by the ‘tension control by torque (TCT) controller which aims at keeping constant the strip coiling / uncoiling tensions.
Speed controller (SC) – The speed controller is in charge of regulating the stand speed. Obviously, in order to achieve the rolling stability, the speed reference is to be coordinated with the other operations of the rolling mill.
Hydraulic torque controller (HTC) – The hydraulic torque controller is in charge of controlling the torque generated by the looper.
Automatic gauge control
The automatic gauge control (AGC) system is provided in HSM for the purpose of achieving the desired thickness of the hot strip. It is the system for the regulation of the thickness. For its applications in HSM, the AGC requires strictly the acquisition of the stretch for each stand. The acquisition of the stand stretch is very important in the case of HSM while in the case of cold rolling mill it is much less important.
The stand stretch represents the elastic behaviour of the mechanical structure of the stand when a compressing force is generated by the main hydraulic cylinder (i.e., the HGC cylinder). This characteristic is to be known in advance for implementing the AGC in HSM and for this reason a suitable control sequence is implemented and executed off-line, that is, before rolling, known as the ‘stretch acquisition sequence (SAS).
The SAS is obtained by putting the work rolls in contact and linearly modifying the position reference for the HGC from a minimum value to a maximum value. For each position reference, the force measured by the load cells (or by the HGC hydraulic force measurement) is recorded in order to build a stretch characteristic similar to the one depicted in Fig 4. The records are, in general, performed twice. The first records are made with increasing HGC position references (up readings) and the second records are made with decreasing HGC position references (down readings).
The differences between the up readings and the down readings are connected with a non negligible hysteresis in the elastic behaviour of the stand. Finally, a best-fit polynomial curve satisfying the following equation is stored in order to perform the AGC task. F in the equation is the measured force.
Fig 4 Stretch characteristics of the hot strip mill
It needs to be pointed out that the acquisition of the stretch characteristic ‘stretch’ (F) can be made full use of during rolling to derive an indirect measurement of the work piece exit thickness h because of the equation h’ = S + F (stretch) where h is the strip exit thickness for the considered stand, h’ is its estimate derived from the previous equation, S is the measured gap for the considered stand derived from the encoders mounted in the hydraulic cylinder, and F is the measured rolling force (from load cells or from the HGC pressures). This equation is normally being referred as ‘gauge meter equation’ and is frequently simplified by introducing the so-called ‘Mill Modulus’, Mm of the stand, that is, the elastic constant of the stand. The simplified equation is h’ = S + F/Mm. In general, the real implementation of conventional AGC is based on the first equation, whereas, the advanced controller synthesis based on models can exploit the linear version represented by the second equation.
The AGC in HSM has the purpose of keeping constant the thickness of the strip by acting on the position references for all the HGC by compensating several phenomena, for example the hysteresis of the stand stretch, the variation of the material hardness caused by possible fluctuations of the material temperature, and so on. To do this, it is necessary to take into account that the presentence of a looper between one stand and the following one implies that the regulation performed by one stand does not influence the regulation performed by the adjacent stands provided an effective inter-stand tension control is ensured by the looper. This fact represents the main reason why the control architecture of AGC for the hot strip mill and the cold rolling mill are significantly different.
The AGC in HSM is achieved by some external controllers cooperating during rolling. In particular two regulators are in charge of controlling the looper. These are described below.
The Looper Control by Torque (LCT) – The LCT achieves the regulation of the inter-stand tension by acting on the torque reference used by the HTC. Normally, the LCT is fed by the tension error generated by a load cell mounted on the looper or, alternatively, by the estimation of the inter-stand tension derived by the looper hydraulic force.
The Looper Control by Speed (LCS) – The LCS aims at regulating the looper angular position by acting on the speed reference of the upstream stand (i.e., by acting on the reference for the SC acting on the upstream stand). This regulator is also referred to as the mass flow regulator.
The proper thickness regulation is achieved in a different way for the intermediate stands and for the final stand respectively. In reality, for the intermediate stands a direct thickness measurement is not available and thus an indirect measurement of the thickness is achieved from the gauge meter principle as being given in the above two equations. Hence, the AGC represented in Fig 4 is composed of the following two regulators.
The Absolute Gauge control, feed-Back via Gap (AGBG) – The AGBG is applied to all the intermediate stands which are not provided with a direct thickness measurement device and is based on the gauge meter principle and generates a trim for the gap reference of the corresponding HGC. This controller is also in charge of making some feed forward compensations connected to the variation of the oil film for the backup roll bearings, the thermal expansion of the work roll due to the contact with the strip and the variation of the roll diameters due to wear.
The Monitor Gauge control, feed-Back via Gap (MGBG) – The MGBG aims at keeping the strip thickness of the strip leaving the last stand of the finishing mill according to the proper target value by using the feedback of thickness coming from the x-ray located at the mill exit. The deviation signal is used to correct the gap references for the HGC of all the stands. Indeed, a dedicated algorithm defines how to distribute the corrections among all the finishing stands. However, the main problem in implementing the MGBG is that it is strictly necessary to take into account the transport delays between the x-ray and the stand which implements the required correction.
Finally, as shown in Fig 3, the LCT can receive a trim from the AGBG regulator in order to reduce the interactions between the LCT and the AGBG.
The speed of the stands and of the coilers is required to be coordinated in order to ensure the stability of the mill. This is done by feed -forward controller is known as ‘Speed Master’. In order to prevent instability problems for the hot rolling process, one stand is elected as ‘pivot stand’ and the speed variations of the pivot stand are compensated in feed-forward through suitable speed variations for the other stands. In order to do this, it is fundamental to know, as precisely as possible, the ‘forward slip’ (FS) for all the stands, that is, the following coefficient representing the relation between the stand motor ‘angular speed’ (As) and the exit strip speed (Vout). This is represented by the equation FS = Vout / RAs, where R is the radius of the work roll. Normally, the FS coefficients are estimated through suitable mathematical models installed in the level 2 automation system together with its sensitivities with respect to the tension set points and the strip speed.
The multivariable control applied for the finishing mill
During the recent years, advanced control technologies are implemented and are now considered well established in the control of the thickness in the HSM finishing mill. The main purpose is to provide a multivariable framework in order to integrate the main controllers acting in the HSM process (more precisely, the AGBG, LCT, and LCS) in only one controller which reduces possible interferences between the various tasks and allows not only to increase the performances but also to decrease the probability of cobbles during the realization of ultra-thin gauges.
The multivariable control is consequently applied in the intermediate stands in order to perform together the AGBG achieved by a certain (n-th) stand together with the LCT/LCS applied to the downstream looper (Fig 5). One more reason for using advanced control is represented by the necessity of introducing a priori robustness about a possible uncertainty associated to the knowledge of the stand stretch: in fact, it is possible to prove that the presence of a strong uncertainty in the knowledge of the mill modulus can cause the AGBG instability. On the other hand, the measurement of the stretch is performed off-line and it is subject to time-variability together with the stand wear.
Fig 5 Application of a multivariable control in hot strip mill