System and method for determination of air entrapment in ladles

Patent metadata

Field	Value
Publication number	US-11475180-B2
Application number	US-201916269052-A
Country	US
Kind code	B2
Filing date	Feb 6, 2019
Priority date	Mar 9, 2018
Publication date	Oct 18, 2022
Grant date	Oct 18, 2022

Abstract

This disclosure relates generally to SYSTEM AND METHOD FOR DETERMINATION OF AIR ENTRAPMENT IN LADLES performing various refining operations at a secondary steel making stage of a ladle is challenging as it requires huge time and resources. The present subject matter discloses a technique for determination of air entrapment in a ladle. In an embodiment, method uses transient computational fluid dynamics modeling for simulating tapping process of liquid steel and tracking the interface with respect to time. By carrying out the parametric study for different geometrical and operational parameters, training and validation data is generated which is then used for training an artificial neural network model. The new ladle geometrical and operational input parameters for which output parameters are required are then used to predict the air entrapment.

First claim

What is claimed is:

1. A processor-implemented method comprising:
obtaining geometrical parameters associated with a ladle, wherein the geometrical parameters comprise ladle top and bottom diameters, ladle height, a number of plugs, position of the plugs, and diameters of the plugs;
capturing ladle process parameters for tapping, wherein the ladle process parameters comprise a range of tapping rates, a tapping inlet area, an initial slag height, a percentage of carry over slag in incoming steel and a duration of tapping;
creating input data using the ladle process parameters and the geometrical parameters;
performing multi-phase computational fluid dynamics (CFD) using the created input data to obtain simulation results;
calculating output data which is an air-steel interface area and an amount of air entrapment from the simulated results;
creating training data from the input data and corresponding output data;
training an artificial neural network (ANN) model using the obtained training data; and
determining an amount of air entrapment and an air-steel interface area in the ladle using the trained ANN model.

2. The method of claim 1 , wherein creating input data using the ladle process parameters and the geometrical parameters, comprises:
dividing the geometrical parameters and the ladle process parameters into a number of sets which captures whole range of processing conditions and geometrical details;
performing multi-phase computational fluid dynamics simulation for the parameter values with ranges in each sets;
identifying key parameters in each of the sets by calculating an influence index from CFD simulation results for each of the ranges of parameters and selecting one or more of the parameters with influence index greater than a predefined value; and
creating the input data for the key parameters.

3. The method of claim 2 , wherein performing the multi-phase computational fluid dynamics, comprises:
creating a virtual geometry of the ladle using the geometrical parameters;
generating a computational grid using the virtual geometry of the ladle;
generating boundary conditions required for modeling of the computational fluid dynamics using the ladle process parameters; and
performing the multi-phase computational fluid dynamics simulations using the computational grid, boundary conditions and predefined solver conditions.

4. The method of claim 1 , wherein calculating output data which is the air-steel interface area and the amount of air entrapment from the simulated results, comprises:
calculating the amount of air entrapment within the steel in the ladle by estimating fluid area where liquid steel is less than a predetermined value; and
calculating the air-steel interfacial area by estimating an liquid steel-air interface where the percentage of air is more than a predetermined value.

5. A system comprising:
one or more memories; and
one or more hardware processors, the one or more memories coupled to the one or more hardware processors, wherein the one or more hardware processors are configured to execute programmed instructions stored in the one or more memories to:
obtain geometrical parameters associated with a ladle, wherein the geometrical parameters comprise ladle top and bottom diameters, ladle height, a number of plugs, position of the plugs, and diameters of the plugs;
capture ladle process parameters for tapping, wherein the ladle process parameters comprise a range of tapping rates, a tapping inlet area, an initial slag height and a duration of tapping;
create input data using the ladle process parameters and the geometrical parameters;
perform multi-phase computational fluid dynamics using the created input data to obtain simulation results;
calculate output data which is an air-steel interface area and an amount of air entrapment from the simulated results;
create training data from the input data and corresponding output data;
train an artificial neural network (ANN) model using the obtained training data; and
determine an amount of air entrapment and an air-steel interface area in the ladle using the trained ANN model.

6. The system of claim 5 , wherein the one or more hardware processors are further capable of executing programmed instructions to:
divide the geometrical parameters and the ladle process parameters into a number of sets which captures whole range of processing conditions and geometrical details;
perform multi-phase computational fluid dynamics simulation for the parameter values with extreme ranges in each sets;
identify key parameters in each of the sets by calculating an influence index from CFD simulation results for each of the extreme ranges of parameters and selecting the parameters with influence index greater than the predefined value;
create the input data for the key parameters.

7. The system of claim 6 , wherein the one or more hardware processors are further capable of executing programmed instructions to:
create a virtual geometry of the ladle using the geometrical parameters;
generate a computational grid using the virtual geometry of the ladle;
generate boundary conditions required for modeling of the computational fluid dynamics using the ladle process parameters; and
perform the multi-phase computational fluid dynamics simulations using the computational grid, boundary conditions and predefined solver conditions.

8. The system of claim 6 , wherein the one or more hardware processors are further capable of executing programmed instructions to:
calculate the amount of air entrapment within the steel in the ladle by estimating fluid area where liquid steel is less than a predetermined value; and
calculate the air-steel interfacial area by estimating an liquid steel-air interface where a percentage of air is more than a predetermined value.

9. One or more non-transitory machine readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors causes the one or more hardware processor to perform a method for determination of air entrapment in ladles, said method comprising:
obtaining geometrical parameters associated with a ladle, wherein the geometrical parameters comprise ladle top and bottom diameters, ladle height, a number of plugs, position of the plugs, and diameters of the plugs;
capturing ladle process parameters for tapping, wherein the ladle process parameters comprise a range of tapping rates, a tapping inlet area, an initial slag height, a percentage of carry over slag in incoming steel and a duration of tapping;
creating input data using the ladle process parameters and the geometrical parameters;
performing multi-phase computational fluid dynamics (CFD) using the created input data to obtain simulation results;
calculating output data which is an air-steel interface area and an amount of air entrapment from the simulated results;
creating training data from the input data and corresponding output data;
training an artificial neural network (ANN) model using the obtained training data; and
determining an amount of air entrapment and an air-steel interface area in the ladle using the trained ANN model.

10. The one or more non-transitory machine readable information storage mediums of claim 9 , further comprising:
dividing the geometrical parameters and the ladle process parameters into a number of sets which captures whole range of processing conditions and geometrical details;
performing multi-phase computational fluid dynamics simulation for the parameter values with ranges in each sets;
identifying key parameters in each of the sets by calculating an influence index from CFD simulation results for each of the ranges

Assignees

• Tata Consultancy Services Ltd

Inventors

• Singh Umesh
• Singh Amarendra Kumar

Classifications

• G06N3/048

  Activation functions · CPC title

• G06F30/20 Primary

  Design optimisation, verification or simulation (optimisation, verification or simulation of circuit designs G06F30/30) · CPC title

• G06N3/10

  Interfaces, programming languages or software development kits, e.g. for simulating neural networks · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N3/084 Primary

  Backpropagation, e.g. using gradient descent · CPC title