We can significantly increase the efficiency of any plant at no capital cost to the company.” A rather startling claim to make, but Mr John Finch and Dr Marcus Sztrimbely, two engineers working out of a modest office in Willowdale, Ont., stick by it with confidence. What they do for their clients seems simple enough. All they do is propose changes to the operating procedures of a plant so that it runs as efficiently as possible. But to do this they are using some pretty sophisticated technology — something that’s become known as artificial intelligence. Finch and Sztrimbely, together with help from another engineer and three computer scientists whom they employ through two separate companies (Northern Dynamics and Advanced Dynamic Systems), are about to deploy three very special computer software packages. They will be put to use in two major nickel smelters and a major steel-making plant in Canada. The computer programs, called expert systems, were written so as to run on hardware already owned by the companies operating these facilities. “I think this is the way the industry is headed,” Finch says. “Instead of computer applications being determined by the hardware a company owns, software companies are starting to make their products flexible enough to be adaptable to just about any hardware.”
The expert systems developed by the two companies use rules of good operating practice and actual operating information provided by the smelter operators to mimic the reasoning process of the “expert” smelter operator.
The introduction of these “intelligent” computer programs marks what must be the first application of artificial intelligence technology in the mining industry. The first prototype was delivered in July and should be on-line in September or October. This program is the result of months of hard work involving relatively brief plant visits and exhaustive questioning by Dr Sztrimbely. “I’ve worked in smelters before, so I’m very comfortable getting around in one and can relate well to the operators.” Weeks of questioning enabled him and his colleagues to document all the operating practices and procedures of each plant. “It’s the people on the shop floor who give a plant its personality and its that which we try to capture in the computer,” Sztrimbely says. From this comprehensive catalogue of information, the company constructs a decision-making flow chart that shows exactly how decisions are made in the plant. (A hypothetical flow sheet is shown in the example on page 64.)
“There were little things in these flow sheets that often startled our clients,” Sztrimbely says. “There were things there they could not believe they were doing. These flow sheets alone often paid for the cost of the entire implementation program by pointing out obvious inefficiencies. Most companies desperately need an outside objective set of eyes to look critically at their operations.”
Once the flow sheets were approved, Advanced Dynamic Systems coded these rules into the expert system. When the programs are on stream and all the bugs worked out, smelter superintendents will be able to quickly draft production schedules for a given day’s work and the computer will inform him if the schedule he proposes is physically and logistically possible. The computer simulates the proposed schedule by going through each function step by step and comes up with suggestions which would make it even more efficient. It will even explain why the schedule is not possible by reminding the superintendent of particular operating rules or the maintenance status of a particular piece of equipment. But the superintendent is always left to make the final decision on what to do.
What caused Advanced and Northern to become interested in this application of expert systems technology was the tricky problem of dealing with sulphur dioxide emissions from base metals plants in Canada. Delays in any smelter, for whatever reason, often require a converter filled with molten nickel to sit and idle. This results in the escape of sulphur dioxide into the ambient atmosphere of the plant which is then vented to the outside. Here, it is picked up by governmental air pollution monitors, designed to ensure that companies maintain strict emission guidelines. Idling expensive converters also increases the unit cost of producing nickel. If an expert system could help smelter operators avoid delays, the company could produce nickel at a lower cost and would put less sulphur dioxide into the atmosphere. The sulphur dioxide would go, instead, to the smelter’s acid-making plant where it is contained and prevented from entering the natural environment. This caused the federal government to become interested in the project. Through its Department of Regional Industrial Expansion program, Northern and Advanced both received Industrial and Regional Development Program grants (of less than $200,000 each) to develop the necessary program “shells.” By custom-designing each shell for a particular company’s needs, Northern and Advanced are able to sell the programs under licence. The licencing agreement prevents clients from selling the program to other mining companies, thus protecting the computer company’s right to the technology developed.
“We think we are helping the National Research Council fulfill the irap mandate,” Finch says. “In order for Canadian companies to compete internationally, they have to become smarter and more efficient — and expert systems are one way to do that.”
Right now the two companies are under a lot of pressure to produce a product that does its job well within a short time period. “We’re going to do well, but we’re not going to make a lot of money at it,” Sztrimbely admits. “But down the road clients are going to look at these three projects as examples of what we can do. Even further down the road they’re going to start asking us to do things we can’t do at the present time.
Be the first to comment on "Computers SMARTER SMELTERS"