New techniques in processing

Predicting the processing performance of an untried ore deposit is essential to determining its economic viability.

In cases where an ore just makes product grade or is difficult to prepare for smelting, there are serious risks. The complex mineralogy of some ores can make it hard to predict optimal plant grade and recovery rates using traditional metallurgical testing techniques.

Also, in most cases, not all ore types or areas of an orebody can be examined in sufficient detail to highlight potential problems.

Dr. John Clout of the minerals division of Commonwealth Scientific and Industrial Research Organization (CSIRO), a government-funded body based in Dickson, Australia, has developed an alternative technique. Using image analysis of microscopy samples, he is able to predict grade and recovery of metal from ore, and to rank new deposits in terms of how easy it is to upgrade them.

Different flowsheet scenarios can be set up to see how well they perform with certain ore types and blends.

“It works well,” says Clout. “We saved one company from making a wrong decision that might have cost half a billion dollars. Our mineral tracking models showed that even if the beneficiation [preparation for smelting] processes worked at 100 per-cent efficiency, one ore type was never going to make the grade.

“The result was initially surprising. So they went back and tested again, and the results confirmed that the new ore could not be upgraded as the company had expected. The company therefore revised its plans and saved itself the capital cost of a new full-blown beneficiation plant.”

Clout was especially satisfied that the model closely predicted the subsequent experimental results for maximum grade and recovery. “The mineral tracking model is an effective tool for guiding selected experiments, avoiding excessive experimental work that may or may not establish an optimum,” he says, adding: “The difficulty with traditional indirect techniques, such as chemical assays and assays of size fractions, is that they are often unable to detect changes in ore texture or tell you exactly why you have a problem.”

It takes extensive laboratory and pilot-scale testing to work out the best way to treat the ore, or to find out that the ore is untreatable. Clout says it’s easy to end up with good test results that can never be quite repeated by the plant once it is commissioned.

Although the technique is applicable to any automated image analysis technique, Clout uses an optical image analysis system to determine the mineralogy, porosity, texture and mineral liberation.

“It gets right to the heart of the matter, defining the physical and textural nature of a large number of individual particles and tracking them through mathematical simulations of unit operations, right through to the flowsheet.”

— The preceding is an excerpt from Process Magazine, a monthly publication of the Commonwealth Scientific and Industrial Research Organization.