Remote locations, harsh environment: High variability and unplanned shutdowns are daily occurrences in the operation of asset intensive industries such as Mining, where yields strongly rely on the performance of critical equipment within the value chain. Mobile equipment like shovels or haul trucks have to be coordinated with fixed equipment like conveyor belts, crushers or SAG mills and other critical assets. The aim is producing a product with a desired and consistent quality. Sounds easy? Certainly not. But the orchestration of a mine is not just about operating expensive assets, but also managing the challenges of volatile prices, increasing costs and restrictions. This has led to technology and specialized services becoming the key pillars in defining new business models to maintain competitiveness in this challenging industry.
Operation & Information Technologies (OT/IT) have enabled a substantial production increase in the industry through the integration of critical assets from mine to port. Several systems such as DCS, CMMS, ERP, LIMS connected by fast, reliable communication networks, allow the big picture of the business to be understood through real time control, accountability, remote connections and visualization. This scenario has opened a door for the creation of new services going beyond typical equipment maintenance and systems configuration. The machines are not just a mechanical piece of equipment anymore, but have a lot of sensors and systems providing a huge amount of data (big data), M2M connections (internet of things), remote monitoring, several databases and the most important: People working to get the most out of this technology. Often this means they must remove themselves from their core business so as to develop solutions, through customized analytical applications that can bring value to their operations. Here the service industry has a lot of opportunities to develop and support services and solutions that can lead to a proactive operation. This is not a simple objective and has several challenges:
- Big data is not necessarily good data: We have to deal with different formats of files or poor quality of records when analysing events or developing models. In addition, where are the big sensors? There is missing data and for some key process variables the sensors are not so reliable.
- M2M is not just about remote desktop; Advanced/predictive analytics is needed for proactive asset management and process optimization in real time.
- Integration is not easy without standardization: We need suitable architectures for integrating all different systems (just like ISA-95) and agnostic solutions, so as to enable the cross-processing of information from different sources.
- Skilled workforce sometimes is a reluctant workforce: People need training to incorporate this technology into their daily operation. This technology can deliver very useful insights and foresights but at the end of the day, if no action is taken, no value is captured! So the customer needs a real service and not just fancy high-tech.
Mining has become much more connected, reliable and safe but at same time is bringing a lot of challenges that require creative and sometimes complex solutions. Despite the fact we that we can find most of the features of the Industrial Internet of Things already incorporated into this industry, there is an urgent need to create value through new applications and services. Years ago a good engineer told me that remote desktop connection to control systems could result in a dawn of a new era of services and I thought he was exaggerating. Now I realize my lack of vision at that time.
Daniel is a Solutions Consultant with GE in Chile. He focuses on big data analytics and IoT for mining applications.
You also need to consider hetrogeneous data, for example data associated with reporting failures or populating work orders and test results. there is lots of free text that has value in fusing with sensor derived data to enable us to understand the context for some of the events we observe. We can also exploit statistics to deal with many of the quality issues.
I would agree that big data is “lower quality” in the sense that you have more errors than in smaller sample type data. Nevertheless the quantity of the data more than makes up for that. I would not however say that the quality is low. If it were the results would be problematic
I think your point about big data not necessarily being good data is actually the rule, not as implied the exception. I believe part of the challenge of successfully applying big data technology is to yield insights and value despite, heterogeneous, unstructured and low quality data. As for missing sensors, surely it is the promise of IIot that sensors (using MEMs technology), ultra low power, scavenging their own power, wireless mesh network enabled means instrumentation (for monitoring) is orders of magnitude lower cost. It is this that will deliver huge volumes of data, that will effectively lift the fog, that currently obscures much of what goes on in operations and maintenance.