June/July 2013

Illuminating the cave

The growing number of block cave mines in development and advances in modelling software are giving miners a better picture of underground operations

By Eavan Moore

A crew works on the No. 2 shaft at Rio Tinto’s Oyu Tolgoi mine in 2012. This block cave operation scheduled to begin production this summer, and others in development promise valuable feedback that will help refine the modelling process for mass mining | Courtesy of Rio Tinto

When the Henderson block cave mine opened in Colorado in 1976, there was no commercial mine planning software that catered to the block cave method, which undermines a large ore body and collects the fragmented ore from an array of drawpoints as the rock mass collapses under its own weight. The operations team at the mine was compelled to write its own.

While some miners continue to develop their own software in-house (see: “The power to plan”), today’s caving operations have alternatives. Commercial software developers, researchers and consultants have collaborated to apply increasingly sophisticated computer modelling in the planning of block caving operations. This stage is particularly critical given the extensive underground development required before any ore can be extracted.

The first planning step is to take an existing resource model and work out the rough footprint of the block of ore to be mined, explains John Barber, technical director for underground mining at AMEC E&C. To do this, the mine planning consultants can apply basic parameters like cost and grade to in-house algorithms, coming up with the best “height of draw” (HOD) or column height.

For the more detailed work, planners can use a range of commercially available software packages. Many mines use Gemcom PCBC, a production scheduling system developed 25 years ago under the Geovia brand of Dassault Systèmes. For every one of a mine’s thousands of drawpoints, PCBC can extend a virtual draw column rising to intersect with the block model. While consultants can generate a basic footprint on their own, a “Best HOD” utility within PCBC exists, allowing users to refine the height of draw with column-by-column data.

Material mixing algorithms then estimate what will actually come out of the draw point. One of the more useful updates to PCBC, according to Chuck Brannon, manager of underground planning at Freeport McMoRan, is a 2006 template that can simulate a variety of mixing mechanisms, including vertical mixing, rilling, fines migration, and toppling from an adjacent open pit.

PCBC generates a production schedule that opens drawpoints in sequence and reports the tonnes and grade mined from each drawpoint. It makes use of inputs like the planned development sequence, tonnage required, desired cave shape, and numerous possible constraints such as tunnel capacity and material handling limitations during start-up. A typical schedule can be created in 20 minutes and exported in a variety of formats, including Excel spreadsheets.

Although AMEC’s Barber believes linking production schedules to dispatch is still a weak point, there are special features for block caving in the dispatch software sold by companies like Micromine, Caterpillar and Modular Mining. The vice-president of Geovia’s Caving Business Unit, Tony Diering, notes that Sandvik’s Automine is particularly well-integrated with PCBC, thanks to a 2004 request by the Finsch diamond mine.

Much of the modelling work in PCBC and other programs aims at drawing down the cave as uniformly as possible. “The name of the game in my book is cave management,” says Barber. “In an ideal world, which we don’t live in, every day you would have the same number of tonnes or buckets of muck drawn from every draw point in the cave. If your draw gets too far out of sequence, you can start introducing early dilution ingress. You can lose resources. You can cause geotechnical issues.”

Predictive power

For the mine to get a uniform draw, it is important to anticipate what will emerge from the draw point. How will waste material dilute the ore as both move through the cave, and what sort of secondary fragmentation will broken rock chunks experience as they travel? Too little fragmentation could result in blocked drawpoints; too much can produce over-fine material.

Predicting what happens inside the cave is very complex, explains Otto Richter, principal consultant at Snowden Group. He emphasizes that in caving operations, everything that has happened to the rock mass in the past – from geological formation to mining – continues to impact the cave’s performance throughout the mine’s life. “With open pit methods, history is history and you only work with remaining in situ material,” he says. “Not with block caving. Unless you know what happened in the past, you will never be able to accurately predict the future in a block cave over any reasonable period.”

But a mine in its design stage needs predictive tools, and that is a focus of the industry-sponsored Mass Mining Technology (MMT) project, based out of the University of Queensland, which seeks to minimize key caving risks through research. Participating developer Itasca International Inc. drew on physical models, theories originating from bin and silo flow, and an existing numerical code called PFC to develop a rapid gravity flow simulator called REBOP (rapid emulator based on PFC). In use at about 20 mines worldwide, it abstracts a draw column into layers that grow according to incremental rules based on common rock flow mechanisms such as dilation, collapse and rilling.

Improved fragmentation modelling

Matt Pierce, principal engineer at Itasca, explains that REBOP uses four key inputs: drawpoint locations; the shapes and sizes of the drawbells and drawpoints; a draw schedule; and a block model that specifies the expected primary fragmentation.

“Understanding the primary fragmentation is key to understanding how the material will move down,” says Pierce, adding that this is no easy task. He is part of a team developing a new fragmentation tool, Synthetic Rock Mass (SRM), which is a numerical code that tests the effects of compression on a virtual rock mass. The rock model employs a discrete fracture network – a rendering of geotechnical data such as the number, orientation and density of joints.

Davide Elmo, assistant professor of rock mechanics at the University of British Columbia, uses discrete fracture network and SRM modelling in his research and consulting work. To produce a good fragmentation model, he cautions, the mine needs to include detailed information on its jointing structure. In particular, joint length has emerged as a critical parameter underrepresented in existing rock mass classification systems; new mines are just starting to add exploration work to learn how far their jointing structures extend. “We made it more complicated,” explains Elmo. “Instead of just using rock mass specification systems, we said, ‘Those are okay, but they cannot really solve your problems.’”

Elmo has also worked on a second key research area: predicting the surface subsidence that will inevitably result from a large caving process. “Predict” is too strong a word, in Elmo’s view, but he believes fairly good results can be obtained from finite element modelling tools like Itasca’s FLAC3D, Rockfield Software Ltd.’s ELFEN, and Dassault Systemes’ Abaqus.

In these models, according to Elmo, a finite element mesh subjected to stress distorts to the point of caving in an iterative process. He says, “Typically you have triggers in this model: ‘If you have a displacement greater than x, then switch the properties from rock mass to cave.’ And from there, your analysis gets updated, so now we have lots of elements that become caved and everything gets weaker.”

The test of time

The finite element models are currently applied at some of the larger block cave operations such as the Henderson mine in Colorado, and the Finsch diamond mine in South Africa. Newer versions of the software are also being used on mines like Resolution in Arizona, Newcrest’s Cadia East and Rio Tinto’s Oyu Tolgoi, which have yet to start production. The predictive power of the software has improved greatly in the last few years and as these new mines come online they will continue to improve. The issue today is that the calibration of the models has not kept pace with development of the new tools. “I would say probably in the next five years, we’re going to start to see what extra work we need to make these methods better,” says Elmo.

Meanwhile, the MMT project will have more than enough work to keep researchers busy; caving is increasingly widespread, over a wider range of rock masses. “I’ve been seeing a systematic increase in interest in caving research over the last 15 years,” says Pierce. “You put a lot of money up front before you extract your first ore, which means you’d better get it right.”

Post a comment


PDF Version