A view from the APCOM plenary session. The event attracts 175 registrants.

High quality papers, plenty of networking and a strong focus on technology, E&MJ reviews APCOM’s 39th conference

By Carly Leonida, European Editor

E&MJ went back to school in early June, quite literally, for the 39th APCOM conference which was held at the Wrocław University of Science and Technology in Poland.

APCOM or, to use the organization’s full name, Applications for Computers and Operations Research in the Minerals Industries, was founded in 1961 by four American universities — the University of Stanford, the University of Arizona, the Pennsylvania State University and the Colorado School of Mines.

As the name suggests, the organization’s biannual conference aims to promote the application of computer-based technologies and research within the mining industry, and the event has become a firm fixture on the global events calendar since its inception in 1986.

The 2019 incarnation brought together 175 attendees from 21 countries. This year’s contingent was weighted on the side of academia, however, that was no bad thing.

E&MJ was invited to close the conference by summarizing some highlights from the program — I chose my five favorite papers — and what follows is an expanded version of that presentation.

Gamification Enters Mining

Day one kicked off with an excellent plenary session. Dr. Sean Dessureault of MST Global talked about IoT-enabled devices such as tablets and smart phones, and how the technologies within these can be harnessed to provide a low-cost alternative to traditional fleet management systems (FMS).

Most tablets and phones feature GPS technology for tracking, Bluetooth for wireless communication with other devices, and inertial sensors (the latter allows Google Maps to tell when you get out of a car and switched to walking during navigations).

Dessureault explained that FMS were mainly developed for use at open-pit porphyry copper mines and are therefore difficult to adapt for use underground. However, this limitation can be overcome using smart devices.

For example, the MST team deployed tablets in trucks at an underground mine and, over a month of monitoring day-to-day activity, they found that 30% of the time during uphill hauls, the trucks were sitting idle on the ramp due to traffic issues, which is extremely inefficient. Thanks to the tablet’s inbuilt technology, MST was able to capture this data and help solve the bottleneck, which might otherwise have gone unnoticed.

Mobile phones and tablets can also be used for personnel tracking and proximity detection. These devices are widely available and cheap to purchase, which helps to lower the barrier to entry for smaller mining companies.

Dessureault gave the example of Bluetooth low energy (BLE), which was a system originally developed for tracking people in the retail sector. He compared the cost of deploying a tracking system based on this to an RFID-based one, which is the norm in most underground operations, and the contrast was stark.

Each RFID beacon (and a reasonably size mine would need many) costs around $250-$350. RFID readers are also required with each costing approximately $3,000 and the associated infrastructure would run to $100,000-plus.

A BLE system uses low-powered Bluetooth chips that can be embedded on machinery and are detected by smart devices such as a phone or tablet. Multiple beacons would be needed to transfer the signal around a mine (approx. $40 each). Most employees already have a mobile phone in their pocket that could act as a detector, or tablets (approx. $400 each) could be distributed. No further infrastructure is required.

“I know mining companies today that are installing proximity detection systems for $15,000-$20,000 per machine,” he told the audience. “Your phone can do exactly the same for just a few dollars with centimeter-level accuracy. That information can also be captured and used to automate cycles.”

To finish, Dessureault talked about gamification and how tools and techniques employed by the gaming industry to keep gamers engaged can be applied in mining.

Drawing upon his experience as an underground machine operator, he explained how the job can at times be dull and repetitive. However, by introducing tablets and a scoreboard system, this can be counteracted.

For example, points can be allocated to operators based on number of loads moved, how well they follow instructions and for filling in safety feedback forms. This instigates competition and promotes interaction between colleagues. Rewards can then be given based on high scores.

Initiatives like this are important because they create a greater sense of job satisfaction for those already in the industry, and can also help make the sector more attractive to new recruits.

Developing a Virtual Mine

Dessureault’s presentation led nicely into one from Rudolph Suppes and Yannick Feldmann. The duo are researchers at Aachen University in Germany and have been working to develop a learning tool powered by virtual reality (VR) to train engineering students.

The pair explained how some of the teaching tools used in current mine engineering courses had become outdated, and they felt that a lack of practical experience was hampering students learning efforts.

“Most students haven’t even seen the inside of a mine. That’s where we got the idea to use VR to enhance the learning experience,” Feldmann told the audience.

He pointed out that the application of VR tools in mining is not new; VR headsets and theatres have been used for many years in equipment safety training.

“It’s already used at some universities for educational purposes,” Feldmann said. “But I think we can do better.”

Benefits of this approach include unlimited access to equipment; time savings (no downtime or travel required); and simulations of hazardous situations in a safe environment.

As part of this initiative, Aachen University has teamed up with TTU and Wolfram Raw Materials to develop a virtual version of the Mittersill tungsten mine in Austria. Suppes and Feldmann explained that they wanted to focus on the underground environment first as they felt the health and safety benefits would be greatest at these operations. Photogrammetry was used to make the mine as realistic as possible.

“Details matter in mining,” Suppes reminded the audience.

Challenges included achieving a high degree of immersion and making the virtual environment as realistic as possible.

Suppes explained that each student wears an Occulus Rift headset and can walk freely around the mine. The idea is not to provide them with a lecture, rather to demonstrate a scenario or technique and assign them problem-solving tasks. Students are given freedom of choice to make decisions and, importantly, mistakes. They can then be awarded points afterward (gamification again) and can review their performance with a tutor to determine areas for improvement.

The team has added “360° video bubbles” in certain areas of the mine, which students can walk into to watch interviews with actual mine personnel on certain topics.

“There are various trials under way,” Suppes said. “Process-based learning will hopefully increase the uptake of training for engineers and benefit the job market in a safe and cost-efficient way.”

All going well, Aachen University will begin using the virtual mine for training this autumn.

From Aviation to Mining

An unusual but interesting addition to the conference program came from Finn Hovgaard on day two. Hovgaard is a senior airline captain with Scandinavian Airlines (SAS) and chief of flight operations for the Danish Air Force. After a chance meeting with Dr. Christoph Mueller of MT-Silesia who organized the APCOM 2019 conference, the pair realized there were multiple synergies between the work they were doing in both the mining and aviation industries.

Hovgaard now acts as a consultant for Mueller’s company, Minetronics, which specializes in electronics and communications systems, helping to develop optimized human-machine interfaces (HMIs) for its equipment. “I have huge experience [of automated systems] from a user point of view, and I know the pros

and cons,” he told the audience. “My presentation is designed to offer thoughts and reflections, instead of presenting solutions.”

Hovgaard shared his analysis of air crash investigations with the audience. The crux of the discussion was that following these types of incidents, there are always thorough investigations into the causes. These nearly always conclude that failures in automation, and the subsequent inability of the pilot to regain control of the aircraft is to blame.

“It is a big responsibility for the software designers who program these systems, because once the pilot has pushed the button, they are just a passenger,” Hovgaard told the audience.

He explained how important it is for pilots to be able to take back control if automated systems fail, and this is where the subject intersects with mining.

“The questions and issues are similar,” Hovgaard said. “Fully manual operation of machines with well-educated operators is safe. Fully automatic operation under well-defined and regular conditions is safe. But what happens when there is a variation from regular conditions?

“Who is ultimately in control? What is not safe is if a system restricts the operator’s ability to intervene in these situations.”

Hovgaard went on to discuss how HMIs can be designed to ensure that operators can easily understand all key parameters and, most importantly, regain manual control if required. He used the example of a dashboard from an underground train and talked about the color coding and menus used for the on-screen displays. The layout and appropriate selection of numbers and graphics can greatly affect an operator’s ability to grasp information quickly. There should be no room for confusion.

“Imagine if you were in your car and it was skidding on black ice,” he said. “And you had to disconnect all of the electronics before you were able to retake control. You would undoubtably crash.”

Hovgaard concluded that there were several lessons that companies in the mining space, particularly OEMs (there were several representatives in the room), could take away: to look carefully at the design and use of HMIs; rigorous training and testing for operators, particularly in safety situations; and selecting the right operators. Experience is vital when safety is at stake.

Bringing 5G Into Mines

Day three saw two stand-out presentations. The first was from Wesley Santos, whose background is in telecommunications. He now works with Epiroc’s automation R&D group on machine control systems. His presentation focused on LTE (long-term evolution) communication technologies, and specifically 5G. Epiroc has been studying the topic to enable the company to collaborate effectively on its automation projects.

Santos explained exactly what 5G is, how it works, and the benefits and challenges the mining industry is facing in its implementation. “LTE is a global wireless broadband standard that was developed by a company called 3GPP in 2004,” Santos explained. “It was commercially realized in 2008 and is better known as 4G.”

4G was the first data-centric, packet-switched, IP-based mobile broadband technology, and 5G is the next step or evolution of the LTE standard. 5G was first established in 2016 and is due to be introduced commercially later this year or in early 2020. It offers improved connectivity, higher capacity and lower latency.

Santos explained that Wi-Fi, which is commonly used in mines today, is insufficient for connecting large numbers of devices — a key requirement in automation. Wi-Fi is relatively low-powered and has a short range. It runs on a limited, unlicensed spectrum, which creates noise and interference, and is also costly to install and maintain.

“I’ve seen a lot of frustration over the past 12 years because mines have coverage but not connectivity,” he said. “Autonomous systems have different, time-sensitive requirements for connectivity.”

5G will deliver massive device access, which is what autonomous systems require to operate effectively.

Santos explained how today, we tend to take connectivity for granted, but physical systems — architecture — are required to make this possible. At present, consumers including mining companies, are reliant on telecoms companies to provide this. What makes 5G a game changer is that users can set up and run networks themselves. It will effectively remove miner’s dependency on telecoms providers and give them full autonomy.

“With 5G, mine sites can have tailored, private EPC [evolved packet core] deployments,” he said.

What this means is that mines can buy the required hardware and applications and run them through a standard computer to create their own network. This approach offers better traffic control, higher bit rates and lower latency, all of which support automation applications and their time-critical communication.

5G-based networks are also IoT ready and offer improved security because they handle encryption in a different way to previous iterations. As with any new technology, there are limitations. The business model for 5G has yet to be defined, and spectrum and licensing are still under development. The technology is also not fully compatible with industrial communications standards yet, and some of the big telecoms providers, including Nokia, Huawei and Ericsson are working to establish new standards.

“There is a lot of hype at the moment,” Santos said. “But it’s important that miners and vendors understand the limitations when considering applications.”

Companies have been undeterred so far, and several have already begun to buy up spectrum in anticipation of becoming their own network providers. BHP, which runs remote operations centers in Western Australia for its network of mines in the Pilbara, has purportedly been very keen to do so.

Harnessing AI

Another presentation that grabbed E&MJ’s attention on day three was from Newtrax Technologies’ Louis-Pierre Campeau. The company’s expertise in the handling and analysis of data has not gone unnoticed in the industry — APCOM 2019’s silver sponsor, Sandvik, completed its acquisition of Newtrax shortly after the conference concluded — and I was keen to find out more.

Newtrax provides IoT-based solutions for real-time data collection at underground hard-rock mines. It supplies the hardware required to collect and transmit the data as well as the software solutions for safety and productivity applications.

Campeau discussed in detail the artificial intelligence (AI)-based tools, specifically machine learning algorithms, that the company has developed to valorize data sets related to the predictive maintenance of machines.

He described the three main types of maintenance: corrective, for example, changing a flat tire on a vehicle; preventative, such as a regular oil change; and predictive. “If you can detect when a failure might happen, then you can do the [predictive] maintenance right on time,” Campeau told the audience.

He discussed the merits of both supervised and unsupervised algorithms in these applications. With a supervised algorithm, labels are provided on data sets to help the algorithm identify anomalies. Newtrax uses supervised learning in the detection of specific events that have already been observed in historical data such as engine or transmission failures. The data labels are taken from maintenance logs supplied by clients.

With unsupervised learning, algorithms are trained to identify groups of data that are generated under “normal” operating conditions and identify variations. These can then be used to detect outliers, which might indicate machine faults ahead of time.

These algorithms are currently used with data generated from individual sensors to detect faulty behavior. Newtrax has found that sensors which produce more outliers than average are almost always faulty, and this can be useful in detecting faults on transmissions, with engine power or battery problems.

Campeau explained that the advantage of unsupervised algorithms is that no labeling needs to be done, and that they are more likely to detect never-before-seen failures.

He added the Newtrax is currently researching the use of deep neural network algorithms to detect outliers and potential failures, and also the potential application of AI techniques in planning optimization, automatic planning, and optimization of resource allocation for underground mines.