Visual Quality Control of the Vacuum Degassing - Introduction

J-K Kamarainen <>

1. Project introduction

At the steel plants a high-purity steel production is always preferred. Inpurities like oxygen, carbon, phosphorus, sulphur, hydrogen and nitrogen must be reduced from the steel to provide customers the demanded quality. One popular and flexible method for reducing the inpurities is the tank degassing, also known as vacuum degassing. This project is established in the co-operation with the Imatra Steel steel plant. Imatra Steel is interested to apply a machine vision system to the vacuum degassing quality control, which is currently done by the vacuum degassing operators. The main goal is to classify the degassing process in the terms of hydrogen content of the result steel (excellent, good, weak etc.)

2. Reducing hydrogen content

In the vacuum degassing treatment a tank full of melt steel is first pumped to low pressure and then argon gass is passed to the tank. Now the melt should start ``boiling'' and argon gas bubbles should be on its surface. Hydrogen is transfered from the melt to the bubbles and this dehydrogenisation reduces the hydrogen content of the steel. Simple we can say the more bubbles in the melt surface the less hydrogen in the melt. There is two variables for the visual estimation of the dehydrogenisation treatment, time and the level of the bubbling. These two variables should be used by the computional methods to estimate the quality of the vacuum degassing treatment.

3. Image processing methods

The vacuum degassing tank contains a digital camera and pictures from that camera are grabbed by a video card. In figures 1 and 2 are shown two different kind of levels of melt bubbling in the vacuum degassing treatment. These kind of images can be grabbed from the real time stream during the degassing process (approx 30 minutes).

Figure 1: Bubbling in the melt is weak.

Figure 2: Bubbling in the melt is good.

3.1 Known problems

View hole for the camera is saturated by the chunk of inpurieties until the hole (and the whole deck) is cleaned. Thus, the cleaning is done only every month or so and this affects a dynamic error to the grabbed images. Right after the cleaning the view is unblocked, but after more treatments the view is more and more blocked. Sometimes some chunks drop by themselves and the view is less blocked again. This problem should be solved to always guarantee a right classification.

About this document ...

Visual Quality Control of the Vacuum Degassing - Introduction

This document was generated using the LaTeX2HTML translator Version 98.1p1 release (March 2nd, 1998)

Copyright © 1993, 1994, 1995, 1996, 1997, Nikos Drakos, Computer Based Learning Unit, University of Leeds.

The command line arguments were:
latex2html -no_navigation -show_section_numbers -split 0 projintro.tex.

The translation was initiated by Joni Kämäräinen on 1999-07-19

Joni Kämäräinen