Consider the Continuous Rolling Mill Depicted

Rolling Mill

A rolling mill in which rolls with a vertical axis roll the edges of the metal stock through the horizontal rolls between some of the passes.

From: Piping Materials Guide , 2005

Ferrous Metallurgical Process Industry

Santanu Chakraborty , in Treatise on Process Metallurgy: Industrial Processes, 2014

6.5 Rolling Mills

The rolling mill facilities were designed to meet the production requirement of Table 1. The rolling mills complex was proposed to include a light and medium merchant mill with breakdown group of stands for rolling blooms into billets and an intermediate in-line heat compensating furnace; a wire rod mill; a medium merchant and structural mill; and a universal beam mill. Specialized workshops, common to all mills, would be installed for repair and maintenance, roll turning and bearing inspection. All the rolling mills were scheduled to operate three 8   h shifts per day and 299 days per year.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B978008096988609982X

Metal Working: Cold Rolling

P. Montmitonnet , in Encyclopedia of Materials: Science and Technology, 2001

2.4 Automatic Control and Modeling

Modern rolling mills, and specially cold tandem mills with high productivity, are equipped with a number of transducers participating in automatic control systems: rolling force (on exception rolling torques), strip tensions between stands, average gauge and thickness profile at exit, strip tension profile (shape measuring systems) between certain stands, sometimes temperature, parameters pertaining to profile and flatness actuators (forces, oil pressures, linear or angular position), lubricant outputs and temperature, coil length and weight, etc. Data logging from and process control through these transducers require a large computing capacity.

Control systems (e.g., control of thickness by the force, or of the thickness profile via the roll bending forces) require models linking the input and output parameters: for instance how much more load δF is needed to change the exit thickness by δh; this of course depends on all other rolling conditions and parameters, the "working point." Simplified "on line" models are used to do this: they linearize the relationship (i.e., estimate ∂F/∂h) around a certain working point. Artificial intelligence techniques are more and more used for this purpose. They may also be derived from more complex "off-line" models solving the thermo-mechanical equations of the system. Full account must then be taken of the coupling between the elastic deformation of the tools, the settings of the profile and flatness actuators, and the plastic deformation of the strip. The intensity of the coupling is all the stronger as thin, hard strip and high friction are considered. Such knowledge models may be complemented by a thermal analysis of the system, by lubrication and friction models, and by microstructure evolution models, all of which may also be coupled to the main thermo-mechanical analysis.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B0080431526009608

Size effects in the production of micro strip by the flat rolling of wire

K. van Putten , ... G. Hirt , in 4M 2006 - Second International Conference on Multi-Material Micro Manufacture, 2006

3.2 Experimental flat rolling of thin round wire

Soft annealed OF-Cu bars cut out from wire with 4.0, 2.0, 1.0 and 0.5   mm in diameter were rolled flat with a reduction of 25, 50 and 75% respectively. For each combination of wire diameter and reduction five experiments were performed. The mean value of the resistance to forming out of these five experiments is used for comparison.

3.2.1 Experimental set-up

A precision rolling mill, built in 2003, with 140  mm roll diameter of the Prymetall company was used for the experiments. During the rolling of the OF-Cu bars the rolling force was measured. There were no forward and backward tensions applied on the copper bars and rolling speed was reduced to a minimum of about 0.75   m/min. The strip thickness h₁ and breadth b₁ were measured afterwards so that the contact area could be calculated.

3.2.2 Specimen preparation

Bars of 1.5   m length were cut from 4.0, 2.0, 1.0 and 0.5   mm diameter OF-Cu wire. The bars were annealed for 30   min. at 400   °C and straightened afterwards. The stresses induced during the straightening were reduced by a second heat treatment for 15   min. at 150   °C.

3.2.3 Experimental results

The resistance to forming in dependency of the wire diameter for each combination of wire diameter and reduction is plotted in Figure 5. The combinations with the same reduction 25, 50 and 75% are marked with open circles, squares and triangles respectively and connected by grey coloured curves.

Fig. 5. Resistance to forming k_w for the flat rolling of pure copper (OF-Cu) wires with various wire diameters d_wire .

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780080452630500623

Applications of Radioisotope Instruments in Industry

J.F. Cameron , C.G. Clayton , in Radioisotope Instruments, 1971

Cold-strip reduction mills

A tandem cold rolling mill usually consists of three to five stands of rollers, each of which reduces the strip thickness by about 20 to 40% so that the total reduction in one pass may be by a factor ten.

It has long been recognized that some, if not most, of the variations in thickness of strip produced by a cold-reduction mill have their origin in the hot rolling process. For example, it has been shown that in many coils of metal produced by a cold mill several heavy areas can be identified which result from water-cooled skids in the slab-heating furnace preceding the hot-strip mill. Another common characteristic on lines not using thickness gauges on the hot-strip mill is a gradual increase in thickness from beginning to end of the strip due to a slow decrease in temperature of the slab in the hot mill. It is the proper function of the cold-reduction mill to remove these variations within the coil and to keep the average thickness to the desired value.

The first step in the control of a tandem mill is to eliminate variations in thickness in the incoming strip. If the strip entering the second stand of the mill has a uniform thickness, then the second and all succeeding stands require little adjustment. The best way to achieve the required adjustment is to apply automatic screw control to the first stand. ⁽²³⁾ At this point the steel is still relatively soft and moves slowly so that the mill screws have time to act.

Many tandem mills have contactor-operated, fixed-speed, screw-down motors, but on the most modern plant the screw motors are actuated by variable-speed systems incorporating rotary or magnetic amplifiers. Such plants lend themselves to fast, accurate control with gauges.

Several variables are, in principle, available to the control engineer for applying corrective action; for instance the screw setting on each stand and the inter-stand tension. In practice all but one or two of these are preset and thickness gauges are included only on the closed loop to control the remaining one or two variables. In the U.S.A. a method has been developed which uses a radioisotope instrument after the first stand to control the screws of this stand and a second gauge after the last stand to control the inter-stand tension. The first successful installation using automatic control of the first stand was made in 1953 and dual control of first and last stands was accomplished in 1955. ^(23–26)

Figure 3.55(a) shows a portion of a chart recording the strip thickness after the first stand of a tandem mill producing steel plate for a tin-plate mill. At the time this recording was made, the screws were not being controlled automatically. The variations shown are typical. The gradual variation in thickness from beginning to end of the coil is apparent. Especially striking is the sudden change of thickness at the point where the head of one coil is welded to the tail of the preceding one. Figure 3.55(b) shows the chart from the same mill producing material to the same specification but with automatic screw control. It can be seen that the strip is being held almost entirely within the tolerance limits. When a sudden change occurs, the screw setting is changed as fast as the screw-down motors can be made to move.

FIG. 3.55. Recorder charts showing variations in steel strip thickness on the first stand of a tandem mill, (a) with manual control and (b) with automatic screw-down control.

A reversing, single-stand, cold reduction mill can be considered similar to the first stand of a tandem mill. Without instrumental control a great deal of skill is required to achieve uniform lateral and longitudinal thickness using a combination of screw-down on the reducing rollers and tension between the reducing rollers and the wind-on reel. By using a bremsstrahlung gauge to control automatically the screw-down pressure, the operator can concentrate on the other control. ^(27–32) Two gauges are used, one on each side of the mill and the control circuits are switched automatically from one to the other as the mill reverses. With bremsstrahlung gauges, particularly in reversing mills which produce a wider range of alloys than do most tandem mills, a separate calibration curve is required for each metal, e.g. carbon steel, stainless steel, brass, etc.

The measuring head of a typical gauge used in such a control system is shown in Fig. 3.56. ^(27–29) , ³¹ This photograph was taken on the "coiler-side" of the mill. The operator first presets the required thickness on a dial (top right) and, when the strip is under tension, the measuring head is moved into position and the mill is accelerated. Automatic control is introduced at a predetermined threshold speed. If the operator adjusts the coil tension during the pass so as to improve uniformity in thickness across the strip, the resulting thickness change is immediately corrected.

FIG. 3.56. Bremsstrahlung gauge installed on a single stand reversing mill. The measuring head can be seen immediately above the steel strip before it is wound on the reel: the indicating unit and large-area deviation meter are mounted to the right of the measuring head. The operator is preparing to switch-in the automatic control system as the mill speed increases at the beginning of a new reel.

(By courtesy of Nuclear Enterprises Ltd.)

This mill is used to cold-roll brass and copper into thin sheets. It is not possible to use flying micrometers for thickness measurement of this material because of the high degree of surface finish which is required and the malleable nature of the material. The mill produces brass strip up to 40 cm width and 0.1 mm thickness at rolling speeds approaching 200 m/min. The thickness of brass on the ingoing side is 0.7 to 0.15 mm and the finished thickness is 0.12 mm. While the lower thickness limits are within the β-gauge range, β-gauges are limited to about 0.5 mm steel while with ⁹⁰Sr/⁹⁰Y/Al bremsstrahlung a range of about 0.1 to 10 mm can be covered.

The measuring head is located just before the reeler, as shown in Fig. 3.56, and the output is fed to the main electronic console which is remote from the mill. The minimum stable automatic control tolerance is ±0.04 mm and the normal working tolerance is ±0.004 mm.

On thin materials lower energy bremsstrahlung sources (e.g. ¹⁴⁷Pm/Al) or β-sources ^(33–40) would be used. Bremsstrahlung gauges have the advantage over β-gauges of covering a wider thickness range and of being less affected by the presence of oil used to roll the sheet.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780080158020500082

Thinning Films and Tribological Interfaces

A.K. Tieu , ... Y.J. Liu , in Tribology Series, 2000

2.1 Experimental set up

A Hille-100 rolling mill with rolls of 225  mm diameter and 254   mm length, driven by a variable speed DC motor of 75 horse power, was used. The maximum rolling force, torque and speed are 1500kN, 13kN-m and 70   rpm respectively. The sensor roll nitrided surface hardness is 65 to 70 HRC. The diameter of the top and bottom roll were 254.2   mm and the average roughness (R_a) of the roll surface was 0.15   μm. Aluminum alloy H5052-H34, carbon steel and lubricant Rolkleen 485A were used in the experiment.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/S0167892200801510

History of Hot Strip Mills1

John G. Lenard , in Primer on Flat Rolling (Second Edition), 2014

2.11.3 Computer Control

The automation of hot rolling mills has matured such that the control system has been organized in hierarchical levels. At the topmost is what is known as level 3, which has the production planning and control function. This level is responsible for setting the optimized production for production orders and feeds the raw material and target product information to the next lower level of control. Level 3 also collects product information resulting from the rolling process and saves it in databases for future retrieval.

The next lower level of control, level 2, is known as process automation, where each product is rolled and tracked individually in the plant. This function also contains mathematical models which calculate the optimum rolling set-ups for the mill (e.g. gap, speed, force, temperatures).

The next lower level of control, level 1, is the basic automation which controls the basic sequences and automation functions of the individual equipment in the hot mill. These may include dedicated technological functions, such as automatic gage control, width control, temperature control, in the laminar flow system and others.

The lowest level, level 0, includes the basic drive equipment regulators and controls.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780080994185000020

Identification by the Least-Squares Method

Yucai Zhu , in Multivariable System Identification For Process Control, 2001

4.3 Industrial Case Studies

Now we have learned the least-squares principle and its applications in linear dynamic process identification. How does it perform in real world applications? To answer the question, typically, a theorist will start making assumptions and then analyze the properties of the method; an engineer, on the other hand, may try this tool on his process and give his judgment based on test results. Obviously the two ways of life are complementary. We will first take the engineers approach. In this section, two industrial processes are used to illustrate the least-squares method. The first process is the rolling mill; and the second one the glass tube drawing process. The extension of the least-squares method to multi-input multi-output (MIMO) process will be discussed when treating the problems.

4.3.1 Identification and Control of the Rolling Mill

Let us revisit the rolling mill introduced in Chapters 1 and 3. It was decided to use drawing force and roll gap as control inputs and keep the rolling speed constant; the input thickness cannot be manipulated and will be used for feedforward control. So, for the feedforward and feedback control of the process, a 3-input 1-output model needs to be identified; see Fig. 4.3.1.

Figure 4.3.1. A 3-input 1-output rolling mill model

Thus the process model has the following form

(4.3.1) $y(t)=G_1(q\text{)}{u}_1\left(t\text{)+}{G}_2\right(q\text{)}{u}_2\left(t\text{)+}{G}_3\right(q\text{)}{u}_3\left(t\text{)+}v\right(t\text{)}$

Based on a staircase experiment, we learned that the process is reasonably linear around working points and process step responses are very short. Hence an FIR model is used in identification:

(4.3.2) $y\text{(t)=}{\displaystyle \sum _{k=1}^{n_1}{g}_{1,k}^o{u}_1\text{(t−}k\text{)+}}{\displaystyle \sum _{k=1}^{n_2}{g}_{2,k}^o{u}_2\text{(t−}k\text{)+}}{\displaystyle \sum _{k=1}^{n_3}{g}_{3,k}^o{u}_3\text{(t−}k\text{)+}}v(t\text{)}$

Introduce the residuals

(4.3.3) $\epsilon (t)=y(t\text{)−}\left[\underset{k=1}{\overset{n_1}{{{\displaystyle \sum }}^{\text{}}}}{g}_{\text{1,}k}{u}_1\left(t\text{−}k\text{)+}\underset{k=1}{\overset{n_2}{{{\displaystyle \sum }}^{\text{}}}}{g}_{\text{2,}k}{u}_2\right(t\text{−}k\text{)+}\underset{k=1}{\overset{n_3}{{{\displaystyle \sum }}^{\text{}}}}{g}_{\text{3,}k}{u}_3(t\text{−}k\text{)}\right]$

Then we are ready to apply the least-squares method.

Remark. There is a principal difference between the output disturbance v(t) and the residual ε(t), although they appear at the same place in equations (4.3.2) and (4.3.3). The term v(t) accounts for the effect of all the unmeasured disturbances acting at the process output; the term ε(t) is used to account for model misfit which is a function of model parameters, ε(t) = ε(t, θ). Note that in equation (4.3.2) the parameters are fixed (unknown) true values; in equation (4.3.3) the parameters are the variables to be estimated.

For the model estimation input-output data are collected, where u ₁ (drawing force) and u ₂(roll gap) are driven by two PRBS test signals, and u ₃ (input thickness) is measured disturbance. Denote the data sequence as

$Z^N\text{:=}y\text{(1)}{u}_1\text{(1)}{u}_2\text{(1)}{u}_3\text{(1)}\cdots \cdots y\left(N\text{)}{u}_1\right(N\text{)}{u}_2\left(N\text{)}{u}_3\right(N\text{)}$

Then it is straightforward to extend the least-squares formula (4.3.2) for a 3-input 1-output FIR model as follows

(4.3.4) $\widehat{\theta }{\text{=[Φ}}^{\text{T}}{\text{Φ]}}^{-1}{\text{Φ}}^{\text{T}}\text{y}$

where

$\begin{array}{l}\begin{array}{cccccc}\text{y}& =& \left[\begin{array}{c}\text{y(}n\text{+1)}\\ y(n\text{+2)}\\ ⋮\\ y(N)\end{array}\right],& \theta & =& \left[\begin{array}{c}{g}_{\text{1,1}}\\ ⋮\\ g_1,n_1\\ g_{\text{2,1}}\\ ⋮\\ g_2,n_2\\ g_{3,1}\\ ⋮\\ g_3,n_3\end{array}\right]\end{array}\\ \begin{array}{ccc}\text{Φ}& \text{=}& \left[\begin{array}{lllllllll}{u}_{\text{l}}(n_1\text{)}\hfill & \cdots \hfill & u_1\text{(l)}\hfill & u_2{\text{(n}}_2\text{)}\hfill & \cdots \hfill & \cdots \hfill & u_3(n_3)\hfill & \cdots \hfill & u_3(1)\hfill \\ ⋮\hfill & \hfill & ⋮\hfill & ⋮\hfill & \hfill & \hfill & ⋮\hfill & \hfill & ⋮\hfill \\ u_1(N\text{−l)}\hfill & \cdots \hfill & u_1(N\text{−}{n}_1\text{)}\hfill & u_2(N\text{−l)}\hfill & \cdots \hfill & \cdots \hfill & u_3(N-1)\hfill & \cdots \hfill & u_3(N-n_3)\hfill \end{array}\right]\end{array}\end{array}$

After the pretreatment of the data, a FIR model is estimated. Then the model fit is checked using a data sequence from another PRBS experiment. Figure 4.3.2 shows part of the measured exit thickness and the simulated one. The power of the simulation error is about 5% of that of the output. For control purposes this is an accurate model.

Figure 4.3.2. Model fit of the rolling mill

Feedback and feedforward controllers have been designed based on the identified model, and on extensive simulations of the control system. Figure 4.3.3 shows the control scheme. There is a large measurement delay at the output due to the placement of the thickness meter. In such a case the feedback controller is only effective at low frequencies, which means that only slow disturbances can be compensated for by the feedback loop. Thus, if possible, a feedforward controller should to be used to compensate for fast disturbances.

Figure 4.3.3. The control scheme for the rolling mill, C _fb1 and C _fb2 are the two feedback controllers, C _ff1 and C _ff2 are the two feedforward controllers.

According to simulation of the controlled system, a considerable reduction of thickness variation can be achieved by the designed controllers. The industrial tests have confirmed this. Figure 4.3.4 shows the measured strip thickness during manual control and during computer control respectively. A 70% reduction of standard deviation has been realized! This is considered to be a big achievement.

Figure 4.3.4. Manual control versus computer control of the rolling mill

4.3.2 Identification of the Glass Tube Process

Let us recall the inputs and outputs selected for the glass drawing process: Inputs (MV's):

u ₁(t):

gas pressure

u ₂(t):

drawing speed

Outputs (CV's):

y ₁(t):

wall thickness

y ₂(t):

diameter.

A general linear relation between the inputs and outputs can be described by a transfer function matrix

(4.3.5) $\left[\begin{array}{l}{y}_1(t\text{)}\hfill \\ y_2(t\text{)}\hfill \end{array}\right]\text{=}\left[\begin{array}{ll}{G}_{11}^o\text{(q)}\hfill & G_{12}^o\text{(q)}\hfill \\ G_{21}^o\text{(q)}\hfill & G_{22}^o\text{(q)}\hfill \end{array}\right]\left[\begin{array}{l}{u}_1(t\text{)}\hfill \\ u_2(t\text{)}\hfill \end{array}\right]+\left[\begin{array}{l}{v}_1\text{(t)}\hfill \\ v_2\text{(t)}\hfill \end{array}\right]$

or

$\{\begin{array}{l}{y}_1(t)=G_{11}^o(q\text{)}{u}_1\left(t\text{)+}{G}_{12}^o\right(q\text{)}{u}_2(t)\text{+}{v}_1(t\text{)}\hfill \\ y_2(t)=G_{21}^o(q\text{)}{u}_1\left(t\text{)+}{G}_{22}^o\right(q\text{)}{u}_2\left(t\text{)+}{v}_2\right(t\text{)}\hfill \end{array}$

where G^o _ij (q) is a rational of polynomials in the delay operator q ⁻¹.

First we must parametrize the model. The generalization of the difference equation model leads to a so called matrix fraction description (MFD). For a 2-input 2-output process model a left MFD is denned as

(4.3.6) $G(q)={\left[\begin{array}{ll}{A}_{11}(q\text{)}\hfill & A_{12}(q\text{)}\hfill \\ A_{21}(q\text{)}\hfill & A_{22}(q\text{)}\hfill \end{array}\right]}^{-1}\left[\begin{array}{ll}{B}_{11}(q\text{)}\hfill & B_{12}(q\text{)}\hfill \\ B_{21}(q\text{)}\hfill & B_{22}(q\text{)}\hfill \end{array}\right]$

where A_ij (q), B_ij (q) are polynomials in the delay operator q ⁻¹.

In general, parametrization of MIMO MFD models for identification is not an easy task which will not be discussed in this book. The diagonal form MFD is, however, simple and physically appealing. For the 2-input 2-output process, the diagonal form MFD is given as

(4.3.7) $G(q)={\left[\begin{array}{ll}{A}_1(q\text{)}\hfill & 0\hfill \\ 0\hfill & A_2(q\text{)}\hfill \end{array}\right]}^{-1}\left[\begin{array}{ll}{B}_{11}(q\text{)}\hfill & B_{12}(q\text{)}\hfill \\ B_{21}(q\text{)}\hfill & B_{22}(q\text{)}\hfill \end{array}\right]$

Applying this description to the general model (4.3.6) yields

(4.3.8) $\{\begin{array}{c}{A}_1^o\left(q\text{)}{y}_1\text{(t)=}{B}_{\text{I1}}^o\right(q\text{)}{u}_1\left(t\text{)+}{B}_{12}^o\right(q\text{)}{u}_2\left(t\text{)+}{A}_1^o\right(q\text{]}{v}_1(t\text{)}\\ A_2^o\left(q{\text{)y}}_2\right(t)=B_{21}^o\left(q{\text{)u}}_1\right(t\text{)+}{B}_{22}^o\left(q\text{)}{u}_2\right(t\text{)+}{A}_2^o\left(q\text{)}{v}_2\right(t\text{)}\end{array}$

We note that in a diagonal form the model is decoupled into two two-input single-output sub-models; for each sub-model there is a common denominator polynomial. Thus the two sub-models can be estimated separately.

Let the degrees of the all the polynomials in a sub-model be equal and call this degree the order of the sub-model. Denote the orders of the two sub-models n ₁ and n ₂ respectively; then [n ₁, n ₂] defines the model structure. Denote the input-output data sequence as

$Z^N\text{:=}{y}_1\text{(1)}{y}_2\text{(1)}{u}_1\text{(1)}{u}_2\text{(1)}\cdots \cdots y_1\left(N\text{)}{y}_2\right(N\text{)}{u}_1\left(N\text{)}{u}_2\right(N\text{)}$

where n =max{n ₁, n ₂}. Then the least-squares estimate of the parameters of the first sub-model which minimizes the loss function

$V_1\text{=}{\displaystyle \sum _{t=n+1}^N{\left(A_1\right(q\text{)}{y}_1\left(t\text{)-[}{B}_{11}\right(q\text{)}{u}_1\left(t\text{)+}{B}_{12}\right(q\text{)}{u}_2(t\text{)])}}^2}$

is

(4.3.9) $\widehat{\theta }{\text{=[Φ}}_1^T{\text{Φ}}_1{\text{]}}^{-1}{\text{Φ}}_1^T{y}_1$

where

$\begin{array}{ccc}y& =& \left[\begin{array}{c}{y}_1\left(n_1+1\right)\\ y_1\left(n_1+2\right)\\ ⋮\\ y_1\left(N\right)\end{array}\right],{\theta }_1=\left[\begin{array}{c}{a}_{1,1}\\ ⋮\\ a_{1,n_1}\\ b_{11,1}\\ ⋮\\ b_{11,n_2}\\ b_{12,1}\\ ⋮\\ b_{12,n_3}\end{array}\right]\hfill \\ {\Phi }_1& =& \left[\begin{array}{cccccccc}{y}_1\left(n_1\right)& \cdots & y_1\left(1\right)& u_1\left(n_1\right)& ⋮& u_1\left(1\right)& ⋮& u_2\left(1\right)\\ ⋮& & ⋮& ⋮& & ⋮& & ⋮\\ y_1\left(N-1\right)& ⋮& y_1\left(N-n_1\right)& u_1\left(N-1\right)& ⋮& u_1\left(N-n_1\right)& ⋮& u_2\left(N-n_1\right)\end{array}\right]& \end{array}$

The same can be done for the second sub-model.

For the estimation of the model of this glass tube process, a PRBS experiment was performed. After the pretreatment of the data, 1269 samples are available. We shall use the first 600 samples for model estimation and the remaining 669 samples for model validation. The model structure [4, 4] is used. Figure 4.3.5 shows the result of model validation; the relative simulation errors are 39.8% and 41.7%. We find that this model is rather inaccurate. With this poor quality we feel insecure about the applicability of the model.

Figure 4.3.5. Model fit for the glass tube process "solid" := measured output; "dashed" := model output

Why does least-squares function well for the rolling mill process but perform poorly for the glass tube process? In the next section we are going to analyze the least-squares method and answer this question. Also, based on the analysis, ways to improve the least-squares method are highlighted.

At this point, we will congratulate you if the least-squares method does solve your problem in process modeling. You may stop reading and spend your time on other more important business. However, those readers who cannot obtain satisfactory results with the least-squares method, we encourage you to read on; this also holds for readers who would like to know more about recent progress in process identification.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780080439853500065

Practice of Micro Flexible Rolling

Zhengyi Jiang , ... Haibo Xie , in Microforming Technology, 2017

15.1 Equipment and Tools for Micro Deep Drawing

Fig. 15.1 shows the micro flexible rolling (MFR) mill systems, including rolling set system, driving system, and operation panel. The mill stand consists of a mill house containing two cylindrical work rolls, which perform the reduction. The rolling force can be measured by a load cell.

Figure 15.1. Micro flexible rolling systems.

Fig. 15.2 shows the driving device. This universal joint has increased joint twist angle, improved drive dynamic performance, increased drive service life, noiseless operation, and high performance in transferring the required torque.

Figure 15.2. Driving device.

Fig. 15.3 illustrates roll screw down system. Roll position is controlled by a closed loop position regulator that automatically adjusts the flow according to a hydraulic servo valve. There are regulators on both sides of operator and drive, respectively. Both sides are coupled together by software control. The control maintains the level of two sides when they move together. The hydraulic cylinders set the pass line and change gap opening while rolling to maintain strip thickness. The position regulator calculates the error between the reference and feedback position. The error is used by a speed balance control to keep either the operator or drive side from lagging behind during long movements of both sides. The control acts by reducing the servo valve flow command for the side with the smaller error and increasing it for the side with the larger error. Hydraulic cylinder control also includes a total force regulator with a differential position loop to maintain the stand level.

Figure 15.3. Roll screw down system.

Fig. 15.4 shows control panel in the flexible rolling systems. This is a user-friendly color operation panel. Operator can input or revise rolling parameters via the touch screen panel.

Figure 15.4. Control panel.

Table 15.1 lists the basic parameters of MFR mill, where the rolling speed is 1–2   m/min and the rolling force limit is 15   KN.

Table 15.1. Parameters of Micro Flexible Rolling Mill

Material Width (mm)	Material Thickness (mm)	Rolling Force Limit (KN)	Rolling Speed (m/min)	Roll Diameter (mm)
10–20	0.1–0.5	15	1–2	25

Generally speaking, flexible rolling can be performed by changing the roll gap during rolling [1]. The difficulty is the precise control of size and shape of longitudinal section in each variable thickness region, and the control is usually accomplished using absolute automatic gage control. In order to obtain the products with various thicknesses, rolling mill and its control system need to have the following characteristics and capabilities: rolling mill with high stiffness; stand and roll system with good manufacturing accuracy; main motor with sufficient power margin, good speed response characteristics, and regulating performance; hydraulic system with fast response, especially the servo valve and cylinder with high frequency response characteristics; the absolute Automatic Gauge Control (AGC) system and high-precision mathematical models; good detectors for rolling parameters, such as workpiece and rolling speed, displacement, force and energy, and data acquisition and processing system. It is better to have the self-learning function in the process control system.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780128112120000157

Managing Maintenance Resources

Anthony Kelly , in Plant Maintenance Management Set, 2006

9.3.5 Maintenance systems

The operation of the work planning system (see Figure 9.10), was based on a multiterminal maintenance documentation system. The system had a manual loop, i.e. work request to the unit engineer, he vetted it and entered it into the backlog. The downshift program was established at a Wednesday meeting and was in the hands of the core team planner on Thursday before the downday (the Monday).

Cost control maintenance documentation and spares management were briefly examined. They seemed to be generally satisfactory. The reporting of the top ten low reliability and high maintenance cost areas good. The quality of history recording compared well with that of the top quartile of maintenance departments I have audited.

Because of the Hot Mill problem, I concentrated my efforts on plant reliability control system (PRC). The overriding purpose of a PRC system is the identification and eradication of 'reliability hotspots'. Figure 9.11 shows the three levels of organizational effort necessary to carry this out. (PRC is discussed in depth in Maintenance Systems and Documentation.)

Figure 9.11. PRC system for an alumina refinery

My comments on PRC at the rolling mill were as follows:

•

Within the team procedures, a level 1 system was in operation. In terms of concept and philosophy it was a good system and worked well for all the teams, with the exception of the Hot Mill team. This was in part due to the reactive nature of maintenance which was preventing the unit engineer/team from concentrating on designing out unreliability.

•

The level 2 and 3 systems were not operating in a satisfactory way, in particular in the Hot Mill area. This was caused by:

–

lack of definition of the PRC system and of the roles within the system;

–

lack of enthusiasm on the part of the project engineers for helping with maintenance problems, they felt they should concentrate on new projects;

–

too few professional maintenance engineers in the centralized maintenance group.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780750669955500266

Theory, Performance and Constructional Features of Induction Motors

K.C. Agrawal , in Industrial Power Engineering Handbook, 2001

1.11 Types of induction motors

These are of two types:

1

Squirrel cage rotor

1

Slip-ring or wound rotor

In a squirrel cage motor the rotor winding is made of solid metallic rods short-circuited at both ends of the rotor. Short-circuiting of rotor bars leads to fixed rotor parameters. In slip-ring motors the rotor is also wound like the stator and the six winding terminals are connected to a slip-ring assembly. This gives an opportunity to vary the rotor circuit impedance by adding external resistance and thus vary the rotor circuit parameters to achieve the required performance.

Although it may seem easy to alter the speed–torque and speed–current characteristics of such a motor through its rotor circuit, the use of such motors is recommended only for specific applications where the use of a squirrel cage motor may not be suitable. The reason is its slip-rings and the brushes which are a source of constant maintenance due to arcing between the rings and the brushes, besides a much higher initial cost and equally expensive control gears.

Specific applications for such motors are rolling mills, rice mills, paper mills and cranes etc. for one or more of the following reasons:

1

To contain the start-up inrush current, as a result of low start-up impedance and to control the same as needed through external resistance in the rotor circuit.

2

To provide a smoother start.

3

To meet the requirement of a high start-up torque and yet contain the start-up inrush current.

4

To meet the load demand for more frequent starts and reversals, as for cranes and other hoisting applications.

5

To achieve the required speed variation through variation in rotor circuit impedance.

Note The latest trend, however, is to select only squirrel cage motors as far as practicable and yet fulfil most of the above load requirements. Fluid couplings and static (IGBT or thyristor) drives can meet the above requirements by starting at no load or light load and controlling the speed as desired, besides undertaking energy conservation. See also Chapters 6 and 8.

1.11.1 Choice of voltage

Because of heavy start-up inrush currents, the use of LT motors should be preferred up to a medium sized ratings, say, up to 160 kW, in squirrel cage motors and up to 750 kW in slip-ring motors. For still higher ratings, HT motors should be used.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780750673518500799

meeksbeerbeen.blogspot.com

Source: https://www.sciencedirect.com/topics/engineering/rolling-mill

Share this post