Saldanha Steel boosts its yield - Monday, February 11, 2008

At Saldanha Steel, one of the problems was cobbling.

The Rolling Plant receives 20 ton steel slabs from the Thin Slab Caster. These go through four stages of rolling:

• Firstly, thickness is reduced from 82 mm to 20 mm by two roughing mill stands.
• The sheets are then coiled in a coil box.
• After uncoiling, the material goes through a five-stand finishing mill to reach a final thickness of between 1 mm and 4mm.
• In the final stage the sheets are coiled in a down coiler.

Cobbling can occur at any stage of the Hot Strip Mill, when the steel jams rather than passing smoothly between the rolls – similar to a paper jam in a printer. The cost of a cobble is significant, as both the affected coils and the upstream slabs at the caster (the kick-off slabs) are usually scrapped. In some cases the cobbled coils can be used after being reworked.

At Saldanha Steel, 0.4 percent of all coils produced were cobbled. The elimination of cobbling would therefore increase the yield of the plant, as well as its productivity and revenue.

By using CSense's Batch Troubleshooter it was possible, within 30 minutes, to identify the major causes of cobbling at the down coiler:

• A principle component analysis (PCA) fingerprint was created by using the process data relating to cobble-free coils. Three clear operating clusters were observed representing, respectively, the start, middle and end of a rolling instance.
• Next, the process data relating to a specific cobble instance was loaded. With a click of a button the major contributor to cobbling was identified: the pressure transducer on the side guide.
• By replacing the pressure transducer, the root cause for the cobble at the down coiler was eliminated.

Andre Theart, one of Saldanha Steel's Six Sigma black belts in training, says that with CSense, the plant's project cycle times were significantly reduced, process setpoints intelligently established and implemented, and knowledge of the overall process greatly enhanced. The result was a significant improvement in business value.

"Now that we've proved that CSense can effectively and rapidly find the cause of cobbles, our next step should include its application to finding the root causes of cobbles in the Finishing Mill, the Roughing Mill and the Coil Box."

Launching a new product with confidence

CSense not only enabled Saldanha Steel to improve the production of its existing products, but also helped it launch a new product.

The company had identified an opportunity to supply the market with high quality steel for cut-sheet applications having a flatness measurement of less than 56 I-units. Andre Theart employed the CSense Batch Troubleshooter to establish the process conditions under which this specialist product could be rolled. He extracted this set of rules from historical production data making use of, among others, CSense's intelligent decision tree. This methodology, facilitated by CSense, gave the manufacturer the confidence to implement the setpoints recommended by the software prior to the first rolling campaign.

"Since commencement of production at the beginning of 2007, we've made this product with consistently high quality and corresponding customer satisfaction," says Theart. "Over this short period we've succeeded in rolling more than 20 000 tons of steel in line with the strict flatness specifications and supplied it to our various customers – a first for the plant."

As a result of its contribution to process and productivity improvements, Theart plans to continue using CSense for Six Sigma projects at Saldanha Steel – "and we hope to get even more astonishing results in the future!"