April 24, 2015
Miner problem: billions of dollars invested, lower resource prices, squeezed margins. What would you do? Miners the world over are looking to the future. Yes, they’re streamlining and cutting costs. They’re also centralising knowledge, driving efficiencies in an industry where fractional changes make huge differences across vast operations, and they’re looking for not just new prospects, but the best new prospects. The future for mining is more high-tech and groundbreaking than you’ve ever imagined, and easier on the environment! Optimisation, productivity, lowered risk—the industry’s KPIs are increasingly being met, and then some.
Enter the data mine and beware of falling touchscreens—the size of a long wall, these lodes of information could knock you out!
Anglo-Australian mining company Rio Tinto first peered into the big-data tunnel on a visit to GE’s massive Transport facility in Erie, Pennsylvania. Back in the 1990s, the operation of some 6,000 heavy-haul locomotives across the US rail system were being monitored from the Erie control room. John McGagh, former head of innovation at Rio Tinto, says, “That was a big aha moment for me. We were wiring our world together on the procurement side, but man, GE had wired its world together across this whole thing—a network of locomotives! That’s the story I think that gets us to where we are today.”
For Rio Tinto today that means a brand new Analytics Excellence Centre at Pune, India. It’s the latest of several high-tech facilities around the globe, each assessing massive volumes of data captured by the array of sensors attached to the company’s fixed and mobile equipment, along with thousands of research and productivity data sets, and each building on the knowledge of the network. “The Centre will help us predict the future through the use of advanced data-analytic techniques, to pinpoint with incredible accuracy the operating performance of our equipment,” said Greg Lilleyman, Rio’s chief executive of technology and innovation, when the Pune facility came online on March 3. Data scientists at the centre will use predictive mathematics to analyse volumes of data with the aim of predicting and preventing unplanned machine downtime, and improving the safety, and of course productivity, of its procedures.
Pushbutton control, Perth switches on the Pilbara
In Australia, Rio’s adventures with remote monitoring, remote control and what it calls the Mine of the Future™, began in the west at its now vast Pilbara operations.
The Pilbara, says McGagh, “it’s bigger than Wales. There are 15 mines in a network, each one with its own geology. Rio runs GE trains up and down that system. It’s among the heaviest haul-train axle load in the world. Fifteen mines, three ports, heavy-haul rail systems, 300 trucks. You have to get the ore to the port, through the drill, blast, load cycle and down the rail and it’s got to be the quality that you’re looking for. It’s one of the world’s greatest challenges, and [Rio] now controls it from a room 1,500 kilometres away.”
Opened in 2010, Rio’s Operations Centre in Perth has developed such innovations as the autonomous drill rig, which allows more efficient and reliable drilling than cab-driven rigs and can be operated from safe sites either at the mine or outside the Pilbara—from Perth, or anywhere in the world, really.
Rio Tinto’s Operations Centre, Perth. Copyright © 2015 Rio Tinto
The Pilbara iron-ore region took Rio Tinto 35 years to develop from starting the first mines to achieving the first 100 million tonnes of productive capacity. It then took only five years to develop the Pilbara operations from 100 million to 200 million tonnes. The company now aims to take the system to 360 million tonnes of productive capacity in an even tighter time frame.
Conducting a symphony of heavy machinery
Future productivity of mines will come not only from autonomous or remotely operated systems. Dr Ana Duarte is a solution consultant at GE Intelligent Platforms (Mining), working with an advanced analytical software product known as Mine Performance, which drills deep down into the data available from any mine, to extract information that leads to greater efficiencies. It’s a two-part solution that encompasses both asset and operations optimisation. “It provides our customers with actionable notifications in regards to equipment failure and/or process variations and inefficiencies,” explains Duarte whose team works to analyse data for clients in South Africa, Australia and South America. “One side of the solution looks at process health. Once your process is running well, you can optimise that process to increase your throughput, to achieve lower energy consumption, better grade in your product and so on.
“Mine Performance increased the availability of one fleet of haul trucks from 70% to 85%.”
“The other side of the solution looks at equipment health. How well your machines, such as pumps, mills, dryers, haul trucks are running.” The software sifts and compares historical data on machinery with current data, for anomalies that indicate minute changes in performance. “When the real data starts to deviate from what is predicted that machine at that condition should be doing, then we start digging, do the diagnosis, identify the root cause and tell our customers, for example, that they should check truck number 112 where there is an injector failure in the engine. It is very specific,” says Duarte.
In this way, the solution allows for fast, planned repairs, only when needed, rather than a mine constantly having to do emergency repairs, or routinely taking equipment offline for servicing when service may not be necessary. Mine Performance increased the availability of one customer’s fleet of haul trucks from 70% to 85%.
The software’s process-health solution also relies on machine data. There can be many processes in any one mine—from concentrators of ore or of slag, to cleansing of gas byproducts of smelting—and poor performance of machines, varying grades of raw material, chemical imbalances and so on, may hamper optimal flow through a particular process. At Lonmin platinum mine in South Africa, for example, the process of drying the concentrate wasn’t working fast enough to meet demand from the furnaces. Mine Performance increased throughput in the drying section by more than 10%. It also enabled 1.5% greater recovery of platinum from recycling of slag from the plant, saving millions of dollars in metal that would otherwise have been lost.
Rio Tinto unveils its latest Mine of the Future™ innovation—the Processing Excellence Centre (PEC) in Brisbane, Australia, March 2014. Copyright © 2015 Rio Tinto
At Rio Tinto, a company with interests all over the globe, such golden nuggets of information can be shared from a central facility, to improve not just one mine’s productivity, but the productivity of several mines. In Brisbane at Rio’s Processing Excellence Centre, opened in March 2014, an expert mineral processing team works at a giant interactive screen to overlay different data sets from seven Rio Tinto mines extracting copper or coal, in Mongolia, the US and Australia. The result, as they analyse real-time data—for example, from some 90,000 sensors monitoring performance on a single copper concentrator—and make adjustments to processes in one mine, is that should those adjustments prove advantageous, they can be implemented within hours at another mine and another. Tweaking and experimentation across the network can lead to compounding leaps in company profitability and return on investment.
Gambling on the greenfields—a surer bet
Developing new mines is an enormously risky business. “You take an ore body that you’ve delineated. You’ve done some work on it, but you’ve probably only delineated 0.00001 per cent of what’s down there. Then you’re raising the capital.” says McGagh. He makes this point to illustrate that, in the mining world of today, it would be mad to compound the risk profile of a new mine by financially loading it with expensive, cutting-edge technology. “Today,” he says, “is a good time to be looking at productivity enhancements within the mining business, because we’re coming off the capital build. The laser-like focus of these organisations is moving from massive capital programs to productivity.”
In the future, productivity-promoting upfront investment may well be justified.
The mines of tomorrow will be better delineated in the first place. Data sets and processing power are creating a new frontier where currently hidden mother lodes and bursting seams will be more accurately pinpointed.
Geoscientists like Dr Klaus Gessner, 3D geoscience manager at the Geological Survey of Western Australia, which is part of the Department of Mines and Petroleum (DMP), are creating richer 3D models of what lies beneath the earth’s surface. Geophysical, geological and geochemical data are being stitched into a 3D framework to build a clearer picture of how mineral deposits have been formed and what patterns characterise their distribution. The main benefit of his team’s work, says Gessner “is being able to tell the industry with more confidence where it makes sense to look and where it doesn’t make sense to look.”
In this snapshot view of a 3D model, modelled surfaces – shown in transparent colours – extrapolate between geological structures that were imaged on seismic reflection profiles up to a depth of 40 km. The purpose of this 3D model is to better characterise the 2.8 billion year old Windimurra Intrusion that is being explored for vanadium in Western Australia. Image courtesy Geological Survey of Western Australia, Department of Mines and Petroleum.
In order to effectively explore for mineral resources, Gessner says, we have to understand the geology of the entirety of the earth’s crust to a depth of at least 40 km. In recent years computing power has made it possible to “not only draw geological maps as we have done for the past 200 years, but now we are also able to make predictions of the rocks that exist at depth”. Let’s be clear, these models are educated predictions of what lies beneath. So to increasingly accurately predict patterns and distribution of minerals at a mine-able depth, we need to gather data of what formed the earth’s crust over a far greater realm.
“Data sets and processing power are creating a new frontier where currently hidden mother lodes and bursting seams will be more accurately pinpointed.”
Gessner and his colleagues are agglomerating data as small scale as what can be seen by performing a CT scan on a rock, all the way up to the scale of our planet as imaged through earthquakes on neighbouring land masses. Gessner explains: “If there’s an earthquake in, say, China, shockwaves travel deep through the earth, and the signal is influenced by the rock structure it goes through. You can bury a lot of seismometers in the earth and measure these waves. Then, under a number of assumptions and a lot of maths, you can build a model of the distribution of rocks in the area you have investigated.” It’s called seismic tomography, and the DMP is working with Australian National University, Macquarie University and the University of Western Australia to gather data from two passive seismic arrays in WA to further plumb the geographical rock masses of the State.
Data wranglers at twenty paces
For some of the measurements being taken across the geological sciences, Gessner says, “we are still lacking a framework to understand what we are seeing. People are turning to approaches like big data, where they try to lump a lot of data together and try to see a pattern that we are not capable of grasping as humans.”
McGagh envisages that the race to successfully mine data for new clues to prospecting and productivity will result in “a war of the algorithms—can my algorithm tell me more about my world than your algorithm?” But, he adds, “Those algorithms can’t be developed unless you have the data. The data has become the asset. New business models will be built around a lot of this stuff. It’s where we all fit in that new business enterprise … It’s accelerating away. It’s fantastic.”