ArcelorMittal Long Products Canada gains insight into production processes with the PI System
- ChallengeMigrate legacy Fortran VAX system into the PI System to increase data access and standardization.
- SolutionSlab tracking uses AF process data structure and AF SDK to generate batches in Event Frames.
- BenefitsNew insight into processes and ability to tweak system behavior to improve process quality and reliability.
When operators with expertise in the legacy Fortran VAX production systems at ArcelorMittal Long Products (AMLP) Canada began retiring, the company realized it was time for a change. The company had adopted the legacy VAX-VMS system in 1985, and though it was reliable, it became increasingly challenging to find young people who know how to program in Fortran. AMLP specializes in the manufacture of semifinished products such as billets, slabs, bars, and rods. In 2016, it was named one of Fortune’s World’s Most Admired Companies. To keep its business sustainable, AMLP decided to migrate its legacy all-in-one system to an open, multilayer architecture in the PI System. With the PI System, operators, engineers, and ad hoc specialists had open access to data as well as standardized calculations and models. “We wanted to empower our people with data,” said Jean-Yves St-Onge, director of automation at AMLP.
Automating the slab-casting process
Until 2014, AMLP used the PI System primarily as a data historian. “Two years ago, we attacked the real automation part of it,” said St-Onge. “That's where we started to integrate the modern PI System and take advantage of Asset Framework and Event Frames to track everything.” ALMP first addressed its slab-casting processes. The company wanted to build a reliable slab-tracking system to understand where each centimeter of steel was in relation to the process.
Slab casting itself is a relatively simple process, explained Alexandre Côté, a software engineer at AMLP. However, the slab-casting machines themselves are gigantic, which presented a logistical challenge. To see what was happening in real time, engineers connected the PI System with its MES and a front end, where they could enter production and quality event information. Engineers could also dig into analyses to understand events affecting quality at each step of the process.
One awesome data structureThen, engineers began building an Asset Framework (AF) data structure. The first step was to bring all data from every system possible into one data store and add structure to reflect the slab-casting process. AMLP involved production staff and even operators to help the architects understand what was going on under the surface of the machines. The company based its AF data structure on reusable templates so that it could easily roll out future installations.
Asset-based templates also made data access simple. Previously, with its legacy system, production data was not clean, nor did it give AMLP the information it needed. AF helped eliminate redundant data, convert units, and structure data into attributes to give engineers and operators direct access to real-time, intelligible information.
Capturing knowledge of the process
AMLP incorporated two levels of analysis into its process control. For example, the company embedded 150-200 relatively simple calculations into its AF data structure for slab casting. For advanced process control, AMLP used AF SDK as a programmatic way to access the PI System to do more complex analysis. In these cases, AF SDK allowed AMLP to create code when its simple calculations were used for more complex calculations or to generate Event Frames. With Event Frames, AMLP could see slabs transit through each section of the machine. To address quality, AMLP used Event Frames to link quality events, time, and spatial position so it could geo-localize quality events onto the product.
For more information about ArcelorMittal Long Products Canada and the PI System, watch the full presentation here.