Título

EXPERIENCE REPORT OF SURFACE QUALITY MANAGEMENT IN PICKLE LINES: IMPROVEMENT POTENTIAL OF PRODUCTIVITY AND QUALITY

EXPERIENCE REPORT OF SURFACE QUALITY MANAGEMENT IN PICKLE LINES: IMPROVEMENT POTENTIAL OF PRODUCTIVITY AND QUALITY

Autoria

Karlowitsch, Martin; Peintinger, Hans-Gerhard; Simpson, Rod

DOI

10.5151/2594-5300-0056

Downloads

Baixar Artigo 54 Downloads

Resumo

In today's sellers market, one usually expects capacity to dominate quality, because nearly the complete produced steel can be sold. However, capacity and quality are not as conflicting as it might appear on the first glance: This paper shall examine how upstream quality measures at the pickling line can help increasing both quality and capacity for downstream processes. Describing a strip steel imperfection through both a quality and a capacity perspective requires parameters (features) to describe the defect characteristics such as size and shape. More features deliver better results. Therefore, Parsytec developed more than 800 features. This has caused a quantum leap in recognition accuracy and has finally made classification feasible. Software tools are needed to create classifiers effectively. The job is obvious: selecting the right features to distinguish the defects, which have been included into a process optimization-oriented defect catalog. The results of this new software approach will be discussed with field studies from leading US steel producers like USS POSCO: applying this classification technology at the pickling enabled USS POSCO to decrease the tandem breaks at the subsequent tandem line by more than 50% - leading to improved capacity and throughput by simultaneously higher quality.

 

In today's sellers market, one usually expects capacity to dominate quality, because nearly the complete produced steel can be sold. However, capacity and quality are not as conflicting as it might appear on the first glance: This paper shall examine how upstream quality measures at the pickling line can help increasing both quality and capacity for downstream processes. Describing a strip steel imperfection through both a quality and a capacity perspective requires parameters (features) to describe the defect characteristics such as size and shape. More features deliver better results. Therefore, Parsytec developed more than 800 features. This has caused a quantum leap in recognition accuracy and has finally made classification feasible. Software tools are needed to create classifiers effectively. The job is obvious: selecting the right features to distinguish the defects, which have been included into a process optimization-oriented defect catalog. The results of this new software approach will be discussed with field studies from leading US steel producers like USS POSCO: applying this classification technology at the pickling enabled USS POSCO to decrease the tandem breaks at the subsequent tandem line by more than 50% - leading to improved capacity and throughput by simultaneously higher quality.

Palavras-chave

Surface inspection, Surface quality

Surface inspection, Surface quality