The Australian mining industry is no stranger to innovation. The generation of new revenue streams that materially benefit companies is crucial for any business that hopes to succeed in the modern era. Similarly, updating old streams by instituting new practices and technology is an integral component of the mining industry.
But for innovation to be meaningful, it is critical to review and understand your metrics and determine how new practices directly connect to value generation. Put simply, you must establish why institutional changes matter and how they matter to your business. For the mining industry, innovation for innovation's sake just isn't enough to compete in today's market. This is where data science comes into the picture.
So, why is data science so important to innovation? Bringing new ideas into the world requires reallocation of resources, disruption of the status quo and periods of downtime during the implementation process. Trepidation often binds decision-makers' hands, preventing them from innovating without a clear view of the future. Of course, this is understandable — without data that demonstrates potential advantages and analysis that provides a complete understanding of the new model, the payoffs can fail to materialise.
It's important to note that we are in the midst of a technological revolution when it comes to mining, with the majority of mining data having been collected and curated within just the last few years. A primary example is the way modernised mining operations have been able to function remotely, while still acquiring real-time data and making operating decisions. These improvements have saved money, time and — most importantly — lives.
However, this incredible influx of data has a major pitfall: data without consultation and analysis has no meaning. Mining companies that focus only on primary KPIs of and fallout from negative events are not taking the time to adequately utilise the library of information available. So many managers allow huge backlogs of information to go unstudied because the tax on their time is too immense.
Again, this is no surprise. Data needs to be constantly and efficiently organised for it to be accessible — and thus, easily analysable. Further analyses of the prepared data must be effective. Simple calculations of the potential advancement of a mine as a percentage of production improvement or cost reduction are not simple. Instead, they are based on hundreds of nested assumptions about maintenance, labor and downtime — each of which must be validated in order for the study to have any meaning.
Staying Conceptually Aligned
The analysis of data also needs to be timely in order to positively affect the flow of business. Managers make hundreds of decisions every day, each of which will have a comparative cost. In aggregate, these decisions and their respective costs will make the difference between surging ahead and falling behind. The symptoms of effective or ineffective decision-making appear slowly. As a result, it is not until operations have been fully disrupted (by a failure to understand the necessary data and integrate a comprehensive set of standards) that managers will realise they missed their window of opportunity.
While conducting performance diagnostics, we often come across islands of excellent understanding amidst a sea of data science miscues and failures. In these cases, five-year plans and annual schedules may be robustly understood and the orebody analysis immaculate — but all too often, these pieces are not linked together. Unfortunately, these gaps in analysis often lead to failures in spatial compliance as well as a misunderstanding of real performance versus what is achievable.
In order to move forward, it is essential for mining companies to fully understand key problems at every level of decision-making. Data science can help ensure each of the essential pieces in a data model can talk to each other. The framing of any problem — and its potential solutions — must remain linked to high-level business performance metrics, a clear and understandable timeline and well-characterised potential performance gaps.
Data Collection and Organisation
To paraphrase the saying: what gets measured gets managed. This is true of production data, fleet management and many other metrics so essential to Australian mining. If your managers can clearly see your current practices and the effect they're having on productivity and revenue, they're more likely to take steps to improve. With the cost of conventional mining increasing alongside the distance between attractive ore bodies and established infrastructure, this is where a significant amount of mining data collection occurs.
With the deployment of fleet management systems (FMS) and integrated production reporting software, we are able to find a basis for the dispatch of our fleets and control of our production schedules. This alone dramatically improves the profitability of our enterprises. Many mines use a management operating system (MOS) or commercial reporting system that automatically collates fleet productivity and other site data and then commences reporting and filing. With the correct initial setup, a reporting system can show huge savings regarding the amount of time analytics takes to extract value from the collected data. However, even with an excellent off-the-shelf solution, digital innovation in this space is crucial. Without major leaps in terms of how the data is analysed, which comparisons are made and which essential questions are being asked, companies will struggle to outpace the competition.
The Importance of Investing in Data Validation
One of the most important things to establish early on in such a project is a meaningful validation system through which data integrity is guaranteed and sustained. Without a validation structure that checks production data, validates it with offline reports and corrects any errors, the system will be unable to deliver results. With that in mind, there are hundreds of pitfalls data systems can fall into when they are initially installed.
We have noticed that FMS is frequently set up with inadequate consideration to the collection pipeline, causing incorrect referencing. Pipelines often lead to databases that are not optimised for cross-referencing with other systems. We have also seen time usage models applied and interpreted incorrectly.
Even more troubling are businesses that fail to put sufficient capital investment into their analysis capabilities. Without analyses in place to interpret which production targets are at risk and what actions need to be taken, companies cannot create improvement programs to carry that work meaningfully forward. However, by making sure these responsibilities are filled and resourced appropriately, mining companies can avoid these potential pitfalls.
Improving Objective Selection
So, now that we have established the critical nature of data collection and curation, how do we determine where to start? One of our favourite tools is bottleneck analysis. Bottleneck analysis is a powerful visual tool that can provide a great deal of insight into your current capabilities. It shows exactly which portion of the process could potentially unlock the most improvement.
When changes are implemented but fail to improve the enterprise, it is frequently because they took place at a part of the process that was outside the performance bottleneck. By establishing your highest impact areas before beginning other analyses, you can ensure you aren't wasting your time.
Accessing Your Value Performance diagnostics creates substantial gains for mining companies by giving all the aspects of data collection, curation and interpretation the time and energy they deserve. This helps you to unlock the full potential of your existing enterprise. With data science, you can optimise future expansion activities in ways that would be impossible without careful analysis, careful thought and careful hands.