Claims:We Claim

1. A system for on-line prediction of mechanical properties of hot rolled coils from hot strip rolling mill comprising:

a module for inputting data for the rolling parameters with chemistry from the steel making stage;

a plurality of field devices positioned at reheating furnace, roughing stands, finishing stands for measuring process parameters in the hot rolling mill;

a programmable logic controller for acquiring data of measured parameters from said field devices and transmitting said data parameters to a central processing unit;

a mathematical module for converting and normalizing the measured data into the different parameters using empirical equations.

2. A system for on-line prediction of mechanical properties of hot rolled strip rolling as claimed in claim 1, wherein the empirical
grain growth equation during reheating is given by,
. . ... (1)
Where

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second

3. A system for on-line prediction of mechanical properties of hot rolled c f rolling as claimed in claim 1, wherein the empirical e i k strain given by:

, ... (2)

where

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy

4. A system for on-line prediction of mec of hot rolled coils from hot strip rolling as claimed in e empirical

equations includeCritical strain given by: ... (3)

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy

5. A system for on-line prediction of mechanical properties of hot rolled c f rolling as claimed in claim 1, wherein the empirical e grain size given by

, ... (4)

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy

6. A system for on-line prediction of mechanical properties of hot rolled f p rolling as claimed in claim 1, wherein the empirical
ti DRX grain size given by:

)

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy

7. A system for on-line prediction of mechanical properties of hot rolled coils from hot strip rolling as claimed in claim 1, wherein the empirical tio me required for 50% of static recrystallization is given

. ... (6)

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy

8. A system for on-line prediction of mechanical properties of hot rolled coils from hot strip rolling as claimed in claim 1, wherein the empirical
tion fraction of SRX, is given by

.
1 . ... (7)

where ...(8)

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy

9. A system for on-line predicti properties of hot rolled coils from hot strip rolling as cl wherein the empirical

equations includeSRX grain size, )

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy

10. A system for on-line prediction of mechanical properties of hot rolled c f h as claimed in claim 1, wherein the empirical
i rowth given by
. ... (10)

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy

11. A system for on-line prediction of mechanical properties of hot rolled coils from hot strip rolling as claimed in claim 1, wherein the empirical
equations include Ferrite grain size after cooling is given by :

d? = (1

b + cCe

+ (d + eCe

) C-0.5 + f(1- egd?

) ] ...(11)

where, C]+[%Mn]/6

e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy

12. A system for on-line prediction of mechanical properties of hot rolled coils from hot strip rolling as claimed in claim 1, wherein the empirical ati ge of stress(YS) of material is
[ ] [ ] + [ ] [ ] . + . ... (12)
13. A system for on-line prediction of mechanical properties of hot rolled c s olli i wherein the empirical
ti mat S) n by,
[ ] [ ] + [ ] [ ] [ ] + . ... (13)
14. A system for on-line prediction of mechanical properties of hot rolled coils from hot strip rolling as claimed in claim 1, wherein the empirical
equations include percentage elongation(%El) is given by,

? ? a ? b[C] ? c[Mn] ? d[Si] ? e[P] ? f [S ] ? gd ?0.5

... (14)

15. A system for on-line prediction of mechanical properties of hot rolled coils from hot strip rolling as claimed in claim 1, wherein the acquired data includes gap setup, roll force speed reference, speed actual, temperature, strip thickness, looper tension primary data input (PDI) including grade, target width, target thickness, chemical composition of steel [C], [Mn], [P], [S], [Si], [Al], [Ti].

16. A method for on-line prediction of mechanical properties of hot rolled coils from hot strip rolling mill using the system as claimed in any of the preceding claims.

Dated: this 19th day of March, 2016.

To,
The Controller of Patents,
The Patent Office, Kolkata.

(N. K. Gupta) Patent Agent, Of NICHE,
For SAIL
, Description:FORM 2

THE PATENTS ACT, 1970

(39 of 1970)
COMPLETE SPECIFICATION (Section 10 and rule 13)

TITLE

ON-LINE SYSTEM FOR PREDICTING MECHANICAL PROPERTIES OF HOT ROLLED COILS FROM HOT STRIP ROLLING MILL

APPLICANT

STEEL AUTHORITY OF INDIA LIMITED, A GOVT. OF INDIA ENTERPRISE,
RESEARCH & DEVELOPMENT CENTRE FOR IRON & STEEL, DORANDA, RANCHI-834002, STATE OF JHARKHAND

The following specification particularly describes the nature of the invention and the manner in which it is to be performed

ON-LINE SYSTEM FOR PREDICTING MECHANICAL PROPERTIES OF HOT ROLLED COILS FROM HOT STRIP ROLLING MILL

FIELD OF INVENTION

The invention belongs to the field of hot rolling technology for hot strip mills. More particularly, the present invention relates to online system and method for the prediction of mechanical properties of hot rolled coils from hot strip rolling.

BACKGROUND OF THE SYSTEM

The technical field of this invention is that of nondestructive materials characterization, particularly quantitative, model-based characterization of surface and mechanical properties using sensors. For determining the mechanical properties of a hot rolled coil from the hot strip mill, the usual practice is to perform a series of tests of the specimen in a testing machine. The specimen used for tensile testing is prepared from a cut-out sample of the outer wrap of the coil produced in the mill. The cut-out sample is then machined to prepare the specimen for tensile testing. One drawback of this existing method is that there is only one sample per coil that can be tested since the coil cannot be cut from the module for taking the samples.

As there is no means to know the variation in property in the body of the coil, the sample is not representative of the entire coil because the sample from the outer wrap of the coil does not represent the entire length of the coil. Since the variability of properties along the length need to be within control from the point of view of application and further processing, it is important to know this variation during rolling of the hot rolled coil in the hot strip mill so that corrective and preventive action can be taken.

Model based prediction of mechanical properties of steel products during hot rolling of steel has been a challenging task for researchers. Attempts have also been made at different steel plants throughout the world to predict mechanical properties on-line using the micro structural evolution models. On-line prediction of mechanical properties can help steel plants in faster release of hot bands as well as in modification of process parameters and alloy design for achieving desired mechanical properties at minimum input cost. The aims of this technology are to increase productivity, reduce manufacturing cost and improve product quality through on-line prediction and control.

In the prior art, a Chinese published Patent specification no. CN 1664550

A discloses a method for online test of steel plate mechanic property during rolling process by providing a comprehensiveness on-line predicting method based on the physico-metallurgy model and combined with the database, information technology through setting up a corresponding model of microscopic structure, finished size, and art component to the mechanical. The method including the following steps: (a) selecting and confirming the parameter of the model; (b) setting up the real-time traffic to the processing machine, calling the art parameter and alloying component dynamic data from the processing database; (c) predicting the ferrite grain size and temperature (d) predicting mechanical of the finished plate. The invention is used specially for low-carbon steel and mini alloyed steel, the adaptive process is heavy and medium plate mill or big mill and finisher in rolling process.

In another prior art, a Chinese Patent No. CN 1589986 discloses a method for optimization and automatic control of technical parameters of a rolling plate, defining an automatic control system that records operational data on the production line for optimizing the forces of the rolling rolls, thermo mechanical or not disposing microrestructurales predictive models, not taking into consideration, like the above patents, the mechanical properties of the product finally obtained.

In another prior art a PCT published application No. WO2004085087 (A2) discloses a system for on-line property prediction for hot rolled coils in a hot strip mill of a steel plant. The system comprises a unit (5) for capturing the chemistry from the steel making stage and providing the data on rolling schedule. Field devices (FD1 ... FDn) are provided at the instrumentation level for measuring process parameters during hot rolling. A programmable logic controller (1) is used for acquiring data of measured parameters from the field devices and feeding the data to a processor (2). Means (3) is provided for conversion of the measured data from time domain to space domain using segment tracking. A computation module (4) processes the converted space domain data for predicting mechanical properties along the length and through the thickness of the strip being rolled. A display unit (6) displays the predicted properties. The data obtained can be stored in a data warehouse (8) for future use. A unit (7) provided in the system can collect the predicted properties and feed the same to the scheduling unit (5).

In yet another prior art a Russian patent application No. RU 2263552 discloses a method of calibrating a continuous rolling installation, in which certain operating parameters are recorded, such as the size of the laminate or the engine speeds of the rolling rolls so that depending on said recorded parameters, these are predicted for subsequent laminations in order to minimize power consumption of the facility, without having thermo mechanical, micro structural tension or predictive models, not having as an objective the optimization of the mechanical properties of rolled steel.

SUMMARY OF INVENTION

With respect to the prior art problems, the present invention provides a method of predicting the rolling process line mechanical properties of the steel sheet. The present invention provides a mechanistic model of physical metallurgy, combined with the database, a comprehensive online performance prediction method of information technology, the

establishment of correspondence between the model of the microstructure, the finished size, composition and mechanical properties of the process between the hot rolling process to achieve Online real-time detection of the mechanical properties of steel, for process optimization procedures and chemical composition, reducing the number of samples to detect, reduce production costs and improve performance and quality of steel to provide evidence.

The mathematical model is developed from the theoretical equations of microstructure evolution during reheating, deformation, recrystallization, grain growth, phase transformation and structure-property correlation. The initial coefficients and exponents of the semi-empirical equations were determined from the experimental data generated in Gleeble-3500. The mathematical model is combined with ANN model in an innovative way to predict mechanical properties more accurately. The uniqueness of this combination is that the model trains itself very fast with on-line predicted and measured data retaining the strength of fundamental equations used in the mathematical model. The model has been validated with measured property data and found to be accurate.

An Object Linking and Embedding (OLE) for Process Control (OPC) based communication module is developed for continuous data acquisition from Programmable Logic Control (PLC) based mill control systems. The software, installed on a Microsoft Windows based model server, acquires online data of reduction schedule from the Siemens S7-400 PLC based Automatic Gauge Control (AGC) system. The mill speed and temperature data are acquired from the ABB 800PEC PLC based mill drive control system. A File Transfer Protocol (FTP) based communication software module is also developed for obtaining chemical composition data from Entrepreneur Resource Planning (ERP) system of the plant. Similarly, another software module receives primary data input of coils from the plant production- planning system using FTP. All these online input parameters

are being stored in a specifically structured Microsoft-Access based database in the model server.

An ASP.Net based web-portal is developed to broadcast the model predicted mechanical properties in the intranet of SAIL. The proxy server gateway setting of RSP is modified to allow the web-portal to work in the SAIL intranet. Since the SAIL intranet is connected to all steel plants, R&D centre and marketing centres throughout India, the hybrid model predicted mechanical property information is obtained at all these locations immediately after completion of hot strip rolling of a coil.

BRIEF DESCRIPTION OF ACCOMPANYING DRAWINGS

Fig. 1: illustrates a screenshot of output screen of Finishing stand module in accordance with the present invention;

Fig. 2: illustrates a Conceptual diagram of the Hybrid Model for prediction of YS in accordance with the present invention;

Fig. 3: illustrates a Conceptual diagram of integration of model with plant automation in accordance with the present invention;

Fig. 4: illustrates a Data flow diagram of the system in accordance with the present invention;

Fig. 5: illustrates a SAIL Intranet location with the hybrid-model server at

Rourkela Steel Plant (RSP) in accordance with the present invention;

Fig. 6: illustrates a Screenshot of On-line Predictive Model at HSM, RSP in accordance with the present invention;

Fig. 7: illustrates a Performance of Model for YS prediction in accordance with the present invention;

Fig. 8: illustrates a Performance of Model for UTS Prediction in accordance with the present invention;

Fig. 9: illustrates a screenshot of Website output in accordance with the present invention.

Fig. 10: illustrates a flowchart showing method steps in accordance with the present invention.

DETAILED DESCRIPTION

The disclosed online system and method for the prediction of mechanical properties of hot rolled coils from hot strip rolling is intended to improve the quality and to achieve the stringent property requirements. Such on-line prediction helps the operator to take corrective actions so as to get nearly uniform mechanical properties along the length of the strip.

METHODOLOGY OF ON-LINE HYBRID MODEL DEVELOPEMNT

The on-line hybrid model has been developed in a series of steps which include selection of empirical equations, development of mathematical model, and development of Mathematical-ANN hybrid model and integration of the hybrid model with plant automation system to predict the mechanical properties on-line.The system captures the chemistry of the hot rolled coil from the steel making stage and the process parameters during the hot rolling stage.

Selection of Empirical Equations

As discussed earlier, a large number of empirical relationships are published in the literature. The equations are converted into generalized form and given below (eqn. 1 to 13). The nomenclatures used in the equations are given in Table-1.

Table-1: Nomenclature of Symbols

Symbol Description & Unit
[C] C in steel composition (%)
[Mn] Mn in steel composition (%)

[P] P in steel composition (%)
[S] S in steel composition (%)
[Si] Si in steel composition (%)
[Al] Al in steel composition (%)
[Nf] Free N in steel composition (%)
e Strain
?? Strain rate (sec-1)
? r Accumulated strain below recrystallization temperature
R Gas Constant (8.31451 J/K/mol)
T Absolute Temperature
t Time in second
Q Activation energy
Cooling rate (0C/sec),
austenite grain size prior to transformation into ferrite

(micron)
A, m, a, b, c, d,e,f,g, n Coefficients and Exponents of different equations. These material specific values are different for different equations

quation during reheating is given by,
. . ... (1)
During deformation in the rolling process static recrystallization (SRX), dynamic recrystallization (DRX), metadynamic recrystallization (MDRX) and grain gr Empirical equations for of these processes are given bel

Peak strain,

Critical strain ... (3) RX grain size, 4)
MDRX grain size, = )
Time required for 50% of static recrystallization is given by,

. ... (6)
me f

.
1 . ... (7)

where

... (9)

. ... (10)

Ferrite ze after cooling is given by

d? = (1

b + cCe

+ (d + eCe

) C-0.5 + f(1- egd?

) ] ... (11)

where, C] + [%Mn]/6

Experiments were conducted in Dynamic Thermo Mechanical Simulator, Gleeble 3500 to find stress, stress and ferrite grain size for at different strain rates and temperatures. Using the data generated in the Gleeble
3500, the coefficients and exponents of the above equations were determined by parameter estimation technique minimizing root mean square error using a multiple multivariable optimization technique. The details of the technique are described in an earlier publication [25]. The initial values of the coefficients required for the optimization were taken from literature [26],[27].

The structure-property correlation equations were also generalized as lo

of of al is given by
[ ] [ ] + [ ] [ ] . . ... (12)
l nsile e
[ ] [ ] + [ ] [ ] [ ] + . ... (13)
Percentage elongation (%El) is given by,

? ? a ? b[C] ? c[Mn] ? d[Si] ? e[P] ? f [S ] ? gd ?0.5

... (14)

The coefficients of the above mechanical properties equations proposed by different researchers are well documented in the books of Lenard et al[26] and Ginsburg [27]. Different researchers have proposed different equations for the three properties.

The approach adopted in this present work was not to evaluate the equations proposed by individual researchers. All the equations were taken as components of the hybrid model.

Development of Mathematical Models

The mathematical model is developed based on the empirical equations mentioned above and the modular design approach for hot strip mill. The hot strip mill line has been divided into 4 parts: reheating furnace, roughing stands, finishing stands and run out table. Individual modules have been developed for prediction of grain size after each part separately and then these parts have been integrated. Based on the above concepts, modules are developed in VB.Net programming language. A typical output of finishing stand module is shown in Fig. 1.

The figure 1 shows the screenshot of output screen of Finishing stand module with the model calculated parameters. It calculates strain, critical strain and conditions for dynamic recrystallization. When there is a dynamic recrystallization, the model calculates dynamic recrystallization fraction. Based on recrystallization kinetics, the model predicts grain size after the pass. The grain size becomes entry grain size to phase transformation module of Run-out-table.

After calculation of austenite grain size after finishing stand, the Run-out- table module calculates phase transformation kinetics from austenite to ferrite and pearlite. This calculation is made by incorporating cooling rate and composition to phase transformation equations. After the grain size of each phase and their fraction is calculated, the model calculates final mechanical properties: YS, UTS and % elongation.

Development of mathematical-ANN (artificial neural network) hybrid model

The mechanical properties predicted by the empirical models are not highly accurate as the empirical equations have been formulated with some simplified assumptions which are not suitable for practical industrial application. Therefore, an ANN program has been used along with the mathematical model as shown in Fig. 2.

The Conceptual diagram of the Hybrid Model for prediction of YS as shown in Fig. 2 includes 4 blocks: Block-1 represents the predicted values of Yield Stress (YS) from different mathematical model equations given in the books of Lenard et al and Ginzburg. Block-2 represents chemical composition of steel which plays an important role affecting the yield stress of material. Block-3 represents values of intermediate parameters calculated from mathematical models. Block-4 represents the ANN model having feed forward network with back propagation algorithm for training of the model. The ANN model gathers their knowledge by detecting the patterns and relationships in data and learns (or is trained) through experience and not from programming. The ANN model herein disclosed is formed from plurality of single field units, processing elements (PE), connected with coefficients (weights), which constitute the neural structure and are organised in layers.

The normalized values of mathematical model are treated with inverse of

ANN transfer function before passed into ANN model.

In a traditional back propagation algorithm initial weights and biases are taken as random numbers between 0 and 1. In the hybrid model, this case the initial values have been chosen selectively. The initial values of weights ith input layer to ith hidden layer is taken as 1. The initial values of weights from each hidden layer to output layer have been taken as 1/n. All other initial values have been taken as 0. The idea behind this innovation is that when the first set of data of first epoch is presented to ANN model

the output becomes arithmetic average of all mathematical model predicted YS values.

As the training goes on, the mathematical model for prediction of YS having more accuracy from the rest gets more weightage. The other factors of Block-2 and Block-3 affecting the YS also get some weightage. The network trains very fast whenever measured property data is incorporated.

Integration of hybrid model with plant automation system

The hybrid model is integrated with plant automation system for accessing on-line data automatically and predicts the mechanical properties after rolling of each individual coil. The integration diagram is given in Figure-3.

A level-2 system is installed in the mill for getting the data from mill automation system and controlling the cooling system. The figure 3 shows that the primary data like chemical composition, planned thickness etc are collected on line from the VAX and Steel Melting Shop systems. The online process data are collected from mill PLC systems and stored in the model database. A VB.Net based program is developed for transferring chemical composition data of any heat required by the model. The data flow of the disclosed system is shown in Fig. 4.

Table1: Details of Data Described in Fig. 4

DATA ID DATA DESCRIPTION
1 GAP SETUP, ROLL FORCE
2 SPEED REFERENCE (6), SPEED ACTUAL (6), TEMPERATURE (4), STRIP THICKNESS, LOOPER TENSION (6)
3 PRIMARY DATA INPUT (PDI) INCLUDING GRADE, TARGET WIDTH, TARGET THICKNESS

4 CHEMICAL COMPOSITION OF STEEL [C], [Mn], [P], [S], [Si], [Al], [Ti]
5 MODEL CALIBRATION FACTORS (FINE TUNNED)
6 MODEL OUTPUT (YS, UTS, ELONGATION) TO USERS IN INTRANET FOR VIEW THROUGH WEBSITE
7 MODEL OUTPUT (YS, UTS, ELONGATION) TO OPERATION ENGINEERS FOR FUTURE DECISIONS
8 MODEL OUTPUT (YS, UTS, ELONGATION) TO MODEL DEVELOPERS FOR MODEL TRAINING
9 MODEL OUTPUT (YS, UTS, ELONGATION) TO MILL OPERATORS FOR ON SITE CORRECTIVE ACTION

Fig. 4 describes the data flow diagram (DFD) of the system. There are five input systems from where the data comes to the model and there are 4 output systems to which data output flows. Details of data flow are described in Table-1.

Advantages of Intranet Portal Based Website

Fig. 5 shows the location of Website server with respect to the intranet of SAIL developed by the SAIL corporate office. The model server is hooked to the SAIL network so that the users of the system can view the model output to all steel plants, R&D centre and marketing centresof SAIL throughout India.

RESULTS AND DISCUSSIONS

On-line predictive model as shown in Fig. 6 is configured to predict the aforesaid discussed mechanical properties (YS, UTS and %elongation) along coil length after rolling of each coil.

The performance of the hybrid model is shown in Histograms in the Fig. 7 and Fig. 8. Percentage of error has been shown against percentage of coils in these histograms. This figures shows that error of YS prediction is -
6% to +4% (Fig. 7) while that of UTS prediction (Fig. 8) is +/-4%. This low error for properties prediction indicates that the model is highly accurate.

An ASP Net based web-portal software has been developed to broadcast the model predicted mechanical properties in the intranet of SAIL. Fig. 9 shows the screenshot of the output screen of the web-portal. This typical output is a replica of actual test certificate issued by the plant to the customers. This typical sample output shows the actual measured values of chemical composition along with the predicted properties (YS, UTS and
% Elongation) of a particular coil.

The proxy server gateway setting of RSP is modified to allow the web- portal to work in the SAIL intranet. Since the SAIL intranet is connected to all steel plants, R&D centre and marketing centers throughout India, the hybrid model predicted mechanical property information is obtained at all these locations immediately after completion of hot strip rolling of a coil.

The system described in this patent has been developed in the following steps:

• Development of microstructure evolution model for hot rolled coils using

Mathematical-ANN hybrid modeling technique

• Installation of PLC-Server based Data Communication system at Hot strip Mill, RSP

• Experimental validation of mechanical properties

• Development of on-line intranet website for display of predicted mechanical properties data after rolling of each coil

• Any user in the SAIL network throughout the country can view the hybrid-model predicted mechanical properties of hot rolled coils

The uniqueness of the system is described below:

• The intranet website allows the display of predicted mechanical properties of hot rolled coils throughout the SAIL network all over India.

• Development of a new innovative model using Mathematical-ANN

hybrid technique for prediction of mechanical properties

• Data transfer among heterogeneous system (PLC, VAX System, and

Web Server)

• On-line display of Mechanical properties along coil length after rolling of each coil

• The model has an auto-calibration module which works on minimization of error using Generic Algorithm – ANN

Although the foregoing description of the present invention has been shown and described with reference to particular embodiments and applications thereof, it has been presented for purposes of illustration and description and is not intended to be exhaustive or to limit the invention to the particular embodiments and applications disclosed. It will be apparent to those having ordinary skill in the art that a number of changes, modifications, variations, or alterations to the invention as described herein may be made, none of which depart from the spirit or scope of the present invention. The particular embodiments and applications were chosen and described to provide the best illustration of the principles of the invention and its practical application to thereby enable one of ordinary skill in the art to utilize the invention in various embodiments and with various modifications as are suited to the particular use contemplated. All such changes, modifications, variations, and alterations should therefore be seen as being within the scope of the present invention as determined by the appended claims when interpreted in accordance with the breadth to which they are fairly, legally, and equitably entitled.