I'm arrived on saturday here in Pittsburgh and it's a real nice place.
Here is the link to the conference http://www.aist.org/08_aistech/2008aistech.htm
Today, I've "played" 2 sessions:
Data Warehouse and Mining Tools for Steel Production Control
Data-warehousing and mining technologies have been recently adopted by ILVA in Taranto plant and they are used to collect data from the different production lines (i.e. hot strip mills and plate mills) and consolidate them into an huge data hub named Quality Data Warehouse.
The Quality Data Warehouse provides Historical Data:
- To optimize the processes planning;
- To build the engineering plans;
- To build the reports to evaluate the quality results of produced coils and plates
All these information are easily and quickly accessed through a WEB application over the Intranet based on Microsoft Sharepoint.
The most innovative feature of the system are the mining tools which are used with excellent results by researchers to find patterns and trends or to evaluate the hidden relationships during the stages in the manufacturing processes for quality prediction purposes.
Plant Vehicles Tracking With GPS/DGPS/GPRS Technologies
The ILVA Taranto plant is an huge area with about 150 miles of railways with more than 50 trains and over 1700 freight cars which connect the different production lines (blast furnaces and hot strip mills). The operational management of the logistics and the maintenance of this infrastructure was a huge challenge.
An integrated system has been developed with the following features:
- Train Tracking with GPS/DGPS/GPRS using Windows-based remote units
- Telemetry Application to get trains data and alarms in real time to prevent damaging
- Maintenance Application to report the trains usages for planning purposes
- Management Application integrated with the ERP system to support and manage the daily operations of railways infrastructure
The integrated system has been released in 2005 with success and it was extended, leveraging the existing infrastructure, also for the tracking of industrial trucks.
A case success paper has been published by Microsoft in Italy:
Instead I've attended to the following sessions:
Fully Configurable Level 2 Infrastructure for Metal Rolling and Processing Control Systems
From John McMillen (TMEIC-GE)
I've already seen the GE architecture and I was very impressed because it has been conceived and design 15 years ago using today's best practice...
Configuration and not development, message bus, scripting to manage message processing, sinks, instrumentation and so on are all features included in GE framework.
From my perspective is an excellent design, my only concern is that some custom features in the platform today are offered off-the-shelves by integration with moder operating systems (ie instrumentation with WMI or message bus WS-Messaging or SQL Server Brokers)
PC-based System for "High-Resolution" Weld Data Acquisition, Archival and Analysis
From T. Roach, (U. S. Steel)
Excellent job regarding the best practices about development of high-perfomance software on Windows Environment (C++ unmanaged obviously).
The application written by Dr. Roach is a signal software recorded with good analysis capabilities.
The Arvedi Endless Strip Production Line (ESP) — From Liquid Steel to Hot Rolled Coil in Seven Minutes
From J. Siegl, Siemens VAI Metals Technologies GmbH & Co.
The abstract is:
Acciaieria Arvedi SpA, and Italian steel producer, installed the world’s first Arvedi ESP Endless Strip Production plant in Cremona, Italy. Start-up is planned for 2008. The unique design and plant configuration of this facility will allow fully continuous casting and endless rolling of high-quality strip in a wide range of steel grades. Due to the highly compact layout (total length 190 m), lower investment costs are incurred compared with conventional thin slab casting and direct rolling plants. Productivity of the ESP plant, based on a single casting line, will be about 2 million tonnes/year, and production costs will be noticeably lower than those of other thin slab plants and also of the ISP plant already operating in Cremona. With the line's ability to produce ultrathin hot rolled strip gauges of 0.8 mm and lower, a subsequent cold rolling process will no longer be necessary for many strip applications. Due to endless rolling operations, the production of strip with uniform and repeatable mechanical properties will be possible along the entire strip width and length. This fully integrated production facility, incorporating advanced technological packages, is the basis for overall plant reliability, superior products and a high line output.
Nothing more to say. I'm very impressed by the overall engineering and I little bit proud because the plant is in Italy!
CQE - Controlling Mechanical Properties of Hot Rolled Coil
A. Polo, Danieli Automation
The abstract is:
Danieli's Coil Quality Estimator (CQE) is an on-line system for prediction of mechanical properties of hot rolled coil, such as the yield strength (YS), tensile strength (TS), elongation (EL) and hardness (HV), at different locations over the length of the strip. The estimation is based on the final microstructure, the evolution of which is traced during entire thermomechanical processing using physically based models for simulation of thermal, deformational, microstructural, precipitational, and finally the structure-property correlation. The uncertainties in actual processing are taken into consideration with the help of the Artificial Neural Network (ANN), which acts in additional to the mathematical models. The system is available for different steel grades such as low-, medium- and high-carbon steels, and high-strength low-alloy steels. The accuracy of prediction is ±20 MPa for the YS, ±20 MPa for UTS, and ±5% MPa for EL. The system is useful for process control and optimization.
The project is very similar to our paper on "DWH and Mining Tools for steel production controls" but we were focused on IT technologies and methodogies, while Danieli has explained their mathematical models and excellend achieved results.
It has been very interesting, because now I've also understood what there is inside the mixed neuronal/genetic algorithm DLL (coming from ILVA) that we've integrated with SQL analysis services...