Asset health technologies have transformed the reliability of mining equipment over the past generation. By tapping into the equipment’s onboard sensors, maintenance teams can observe and record hundreds of parameters that indicate equipment health. Understanding this data and its effects has empowered mines to expand mean time before failure (MTBF), uptime, and other maintenance KPIs more than any tools in recent memory.
Yet, these technologies have their limitations. When installed exclusively on premises, asset health systems miss the advantages available with the power of cloud computing. In 2021, many innovations in predictive maintenance demand a cloud infrastructure and its unique capabilities to deliver optimal value. Remote data storage and aggregation, access to machine learning algorithms, and IIoT automation all rely on cloud technologies that are increasingly necessary elements in a forward-thinking mine maintenance program.
Fortunately, advances in data processing and communications technologies are making cloud solutions more viable for the mining industry. While traditionally resistant to cloud implementations, mines are now leveraging the capabilities of cloud computing, and their maintenance departments are seeing the benefits. New solutions are empowering maintenance teams to do their jobs better in ways that were impossible a few years ago – predicting component fatigue from early warning signs at the edge, observing changes in equipment performance on a continuous basis, and even collaborating with OEMs on proactive asset management that leverages integrated digital platforms.
Real-time analytics, now at the edge
Edge devices installed on mobile and plant equipment are the point of entry for much of the data in any asset health infrastructure. Traditionally, these low-powered hardware units provided simple data processing near the source of operation, streaming that information to a cloud server for aggregation with other datasets and cross-platform analysis.
While this configuration can work well, the wealth of sensors and data now available to mines and their maintenance teams often proves too voluminous and costly to manage in this way. Bandwidth restrictions and communication costs mean that traditional cloud infrastructures struggle to handle the requirements of emerging IIoT systems. Instead, new solutions see more and more calculations happening at the edge itself.
Long-established vendors like Emerson, as well as startups like FogHorn, are bringing advanced capabilities like analytics and AI to lightweight devices near the source of a data stream. Today’s edge devices are able to take raw sensor data – temperature, pressure, vibration, events, and more – and perform complex computations independent of a powerful cloud server. Data ingestion, processing, and reporting can now happen near the source, providing real-time, cost-effective insights to maintenance personnel. After that time-sensitive information has been communicated, the systems can publish compressed data to their cloud counterparts for richer analysis and long-term storage.
“It’s a two-way street,” says Vien Dang, asset health specialist for Wenco International Mining Systems. “Edge and cloud solutions work together. You train edge devices using a cloud-hosted model of what a healthy equipment unit looks like, then set it loose to respond to real-world applications.
“Reliability teams get clean, accurate reporting quickly so they can respond quickly. Then, that data feeds up to the cloud, improving the model they started with. Over time, the whole process gets faster, more accurate, and more responsive – with very little latency or bandwidth issues.”
Digital twins deliver precise, specific asset health modelling
Today’s inexpensive sensors and edge devices can easily produce vast streams of data, but making sense of it is another challenge. Often, maintenance teams have access to volumes of data, but lack useful information to diagnose emerging problems and intervene to prevent failures.
Rithmik Solutions is changing that. The Montreal-based company’s Asset Health Analyzer (AHA) uses machine learning and a rapid analytics infrastructure to create accurate, site-specific equipment health baselines that enable early detection and diagnosis of maintenance issues.
Other asset health technology may claim to enable early issue detection, but AHA analytics go beyond manual error thresholds and standard AI models. In effect, AHA uses a multi-tiered AI approach with digital twins, which act as virtual companions for the entire equipment fleet. This approach fundamentally transforms a mine’s preventive maintenance program, letting technicians follow component health on an ongoing basis and examine the exact condition of monitored parts before pulling it down for maintenance.
“There are a lot of advantages to embedding digital twins within a multi-layered AI approach,” says Amanda Truscott, co-founder and CEO of Rithmik Solutions. “Earlier alarms without any threshold setting, insight about what’s going wrong, what’s about to go wrong, and what went wrong in the past, the ability to prioritize maintenance based on actual equipment health.”
AHA uses machine learning to quickly build a contextualized baseline for the best-performing equipment at the mine. It then monitors equipment for any difference from that tuned-in “normal” state, providing deep and early insights into equipment issues so mines can prevent small problems from escalating. By maintaining models of standard equipment in this way, AHA also allows for cross-asset comparison, highlighting how like assets are similar – and how they vary.
Trials of AHA have already shown strong results, providing alarms hours – or even days – ahead of OEM alerts. In one case, rod-bearing failures on Cat 793Ds were costing a site in Canada $4 million year due to a late OEM warning – coming only a few minutes before the failure occurred. AHA was able to find indicators of those failures 10 hours earlier – a relative lifetime for maintenance to intervene.
“In another recent trial in collaboration with our partner Wenco’s digital platform, our Asset Health Analyzer rapidly uncovered a customer’s fleet-wide inefficiency that had gone undetected for multiple years by both the equipment dealer and the mine maintenance team,” said Kevin Urbanski, co-founder and CTO of Rithmik Solutions.
“What had happened was that temperature regulators failed on 76% of the mine’s haul truck fleet. Fixing the issue is going to both extend the life of the engines and result in significant fuel savings.”
Urbanski says AHA also pulled out previously unknown failure mode indicators on two separate chronic machine issues, which Rithmik and its customer are now using to generate earlier alerts of the failure modes. These insights are also providing a deeper understanding of the total impact of these failure modes on the machine themselves.
Cloud platforms create an ecosystem of partners in mine asset health
Cloud-based platforms are another emerging development in asset health. While digital portals are already common in medicine, entertainment, and enterprise business systems, they are new for mine maintenance.
The concept mirrors existing asset health systems: Sensor data streams to a server, which processes and reports real-time or historical information that maintenance technicians use to understand equipment condition. However, transferring this data to a secure cloud platform instead of an on-premises server opens up many opportunities for mining companies, including access to IIoT and AI-based analysis and stronger collaboration with OEM dealers.
Wenco and Hitachi Construction Machinery (HCM) are currently developing such a cloud-based solution, known as ConSite Mine. Operating on a digital IIoT platform, ConSite Mine remotely aggregates and processes the large volume of data associated with asset health for every installed unit at a mine site, displaying it on a customized dashboard for each customer.