Digital

Today, progressive miners are restructuring their operations with digital technologies carrying the burden of their day-to-day tasks.

This enables engineers to focus on strategic projects and on enhancing productivity. Technologies from other industries, as diverse as natural language processing, bioelectric sensors and commercial drones are being used in a range of applications from health and safety to asset management, all with great impact.

By Gareth Halstead, subject matter expert: mineral processing, DataProphet

Understanding the challenges in driving digital innovation in mining

Like any paradigm shift, there is an initial resistance to innovation and the change it entails.

This article first appeared in Mining Review Africa Issue 1, 2020

Read the full digimag here or subscribe to receive a print copy here

Operational challenges such as equipment breakdowns, design problems, and even political instability are common distractions to the complex task of digital transformation.

Read: Digitisation – revitalising operations and local communities

Resources that would be allocated to the deployment and implementation of new technologies are re-allocated to address these issues. Whilst larger mining houses may benefit from dedicated teams to evaluate and deploy new technologies; mid-tier and junior miners should look to partner with technology vendors for support. Either way, mining houses need to find a way to leverage these disruptive technologies to improve their competitive advantage. 

Making a significant impact

Despite the challenges, there are a number of machine learning and AI applications making a significant impact in this industry today:

  • Predicting the location of mineral deposits: Machine learning models ingest satellite and drone imagery to better predict the location of mineral deposits. Given the cost of exploration, any improvement in finding a suitable ore body provides a profound cost saving.
  • Replacing hard sensors: Control mechanisms often require real-time measurements of systems which cannot be measured directly or are in an environment which is not conducive to physical sensors. In these cases, soft (virtual) sensors can be used to infer the state of the system without significant investment or compromising on process knowledge.   
  • Optimise flotation circuits: A well-known used case for soft sensors is the prediction of grade and recovery based on features extracted from images of the bubbles produced in froth flotation. Modern approaches to this problem rely heavily on branches of machine learning such as neural networks to interpret these features. 
  • Improved health and safety: Society has adopted a stance of zero-tolerance for injuries and fatalities on mines. The complex and multi-faceted environment in which these incidents occur is rich in unstructured data from multiple sources. A novel approach to this problem is the use of natural language processing to analyse large corpuses of text such as reports to identify risks related to health and safety. 
  • Optimising production: Intelligent control is the next step towards full automation of beneficiation plants. Brownfield sites are often data rich and machine learning models can easily be trained upon this data. The outcome is an optimised control plan which is cognisant of the interdependencies between parameters within the plant. Whilst continuously ingesting data from across the entire plant, the system provides operators with only the changes for their section to avoid information overload.

For a solution to be described as AI in its true sense, it needs to be integrated into the data environment to continuously learn and update its understanding of the environment.

Watch: How company culture should adopt digital technology

Many existing AI solutions sit on a continuum from being fully integrated with continuous learning, to simple models built off a static dump of data. Unless prescriptive, these AI technologies are effectively advanced analytics. 


Can digital technology help mines in Africa to overcome common challenges? 

Remote mines in developing countries face additional challenges to those in the developed world. These include the availability of spare parts and the ability to attract highly skilled engineers.

Developments in automation and intelligent control enables engineers to monitor and prescribe changes to process parameters remotely—making the necessary changes from anywhere in the world. At the same time, it is important to ensure local talent is upskilled through e-learning programmes and simulators.

How does the future look?

Having passed the peak of the last super cycle, mining companies are looking for efficiencies in their operations. Digital technologies, such as AI, present a unique opportunity for miners to improve their operations and achieve these efficiencies. 

Mines and beneficiation plants will be fully autonomous in the future. Engineers will be free from the burden of day-to-day operations and will monitor operations remotely, allowing them to spend more time on strategic improvements.

The ability to prescribe optimum control parameters, using an expert execution system (EES) from remote locations, will also lead to safer plants and enhance production outputs. Health and safety will be improved further by automating and using machine learning to better understand the inherent risks in the operation.

The future is about intelligent, autonomous and highly optimised mining operations—with an eye on extending operations outside of earth’s orbit.