Operational performance in the mining industry currently relies heavily on the integration between the mine and mill. Despite this, the majority of mining facilities struggle with having separate systems across geological, drilling, and milling operations that do not allow full visibility and efficiency.
This piece discusses the reasons mining data silos continue to be an issue, the costs of these silos, and how employing an integrated mine-to-mill data framework enables operations to enhance throughput, lower energy consumption, and accelerate smarter operational decisions.
This kind of fragmentation leads to three major challenges in the mining process:
Operational Inefficiency. When data is not integrated, it becomes difficult to sync the drilling process with the milling process.
Decision Delays. When information is housed in multiple disparate systems, decisions take longer.
Escalation of Costs. The combination of multiple redundant systems and manual reconciliation of data drives costs upwards. Mining operations that do not integrate their data from mining and milling activities can lose between $60 million and $180 million annually, without discussing the value of integration. Cited by Databricks (2023), siloed data diminishes visibility and accessibility for teams, creating operational inefficiencies, and increasing the complexity of governance across the entire value chain.
Other researched cases ofmine-to-mill optimization have shown throughput increases of between 5% and 20% when the initial mining activities, particularly blasting and size reduction, are linked with the subsequent milling activities (Hatch; U.S. Department of Energy, 2023). According to the U.S. Department of Energy, comminution is often the single most important process to integrate because it alone accounts for 30-40% of a site's total energy consumption.
A wide gap in the mining industry is the collection of data across the different sectors of mining operations. All the information is trapped in deep isolation silos due to investing in specific mining technology. Geological data is kept separate from the control system of the mill. Mining operations data never make it to inform planning. Sensor data is kept separate from the metallurgical team. All of this disconnect in mining leads to the following silos:
Outdated Systems: Many concentration plants use control systems that are over 20 years old. There is little to no means to relay information unless it is in a .csv format which is time-consuming and lacks context.
Vendor Lock-In: A lot of software that manage drills and blasts use databases that do not allow access to the information stored in them. Similarly, systems that manage fleet and data historians use closed and proprietary systems.
Misaligned Key Performance Indicators: A majority of the time geologists manage reconciliation variance and not mill recoveries. For blast engineers, mill recovery information is not available. Because of this disjoint in metrics, the siloed systems are optimized. This leads to a decline in the functionality of the entire mining system.
Data systems used for mining plants and geology systems should assimilate data from different mining operations and create a single data silo that streamlines decision-making. This should allow automation and access to the information for each team in real-time. Systems such as Databricks are capable of providing the stakeholders with the truth without discrepancies (Databricks, 2023).
Different departments should establish consistent naming conventions, units, and timestamps. For instance, a truck-payload tag, a blast-hole assay, and a SAG mill power reading should all be linked back to the same ore block. Without this, the downstream analysis will be inordinately slow and prone to errors.
The essence of bidirectional mine-to-mill integration is that mill data (throughput, power draw, recovery rates, etc.) must come back to planners and blast engineers so that they can make decisions in the upstream based on the actual results from the downstream, as opposed to just the results that they assumed.
When KPI structures converge, the silos that exist in teams begin to collapse. If the targets of blast engineers are tied to the mill throughput, and the planners' KPIs are based on ore recovery, then incentive alignment is achieved across all steps, making the integration of a technical approach less difficult.
When data is integrated, predictive machine learning models can uncover associations that manual analysis would miss, for example, the impact of specific blast parameters on SAG mill power draw three shifts later. As a result, predictive capability turns the integration of mine-to-mill data from a reporting function to a source of operational edge (Kohezion, n.d.).
AtNTWIST, we built our platform to address these needs by integrating data from geology, haulage, lab assays, and the process plant into a single cohesive decision layer. Combining these data sets into a singular decision layer allows for the MineMax Suite to assign and manage tasks to control process parameters while integrating input from the processing plant, which is a gap that exists between mine planning and mill execution.
With theMineMax Suite, we can reliably forecast and limit disruptions to production from negative variance from the mine-to-mill balance in the parameters of grade, hardness, and volume as well as enable adjustments on the fly to the blend to optimize the outcomes for the downstream plant performance and processing.
The mining businesses we've partnered with have proven that the previously perceived value opportunity of closing the gap between the mine and the mill is not just a theoretical value opportunity, it's a proven and measurable benefit that can be realized.
Understand ourMine-to-mill Optimization Solution.
The most significant yet persistent inefficiencies in the mining sector come from the compartmentalization of data and functions between the mine and the mill. However, mine-to-mill integration is a practical operational strategy that has proven effective with real data to back it for copper, gold, and iron ore mining operations across the globe. When mining operations unify siloed data, close feedback loops, and align functional teams toward a common goal, they fully realize the productive potential and operational efficiency of their assets while eliminating variability from the mix.
Q1. What is mine-to-mill integration?
The mine-to-mill integration process combines the upstream and downstream functions and activities of the entire mining value chain so that the activities in the downstream segment influence and determine the scope of the activities in the upstream segment and conversely. This includes integrating geology, surveying, drilling and blasting activities with crushing, grinding, and processing activities.
Q2. What are data silos in mining?
Data silos in mining operations form when mining operations implement individual technologies (e.g. fleet management, blast design, SCADA, process historians) from different vendors at different times, without any companion data standards. Further silos are created by structural barriers and conflicting KPIs across mine and mill teams.
Q3. What's the expected mine-to-mill integration improvement?
Documented cases have shown improvement in mill throughput by 5% to 20%, from less than stellar integration and depending on the ore type. Optimal blast fragmentation, combined with feedback from the mill, has shown energy savings in comminution of 10% to 29%.
Q4. For mine-to-mill integration, what data sources have to be linked?
The primary data sources include geological block models, drill and blast, fragmentation measurements, load and haul to the mill, stockpile grades, crusher settings, and process plant measurements (power draw, throughput, and recovery of SAG/ball mills). Integrating these data sources creates the foundation for the first analytical layer.
Q5. What is the time frame for mine-to-mill integration and how do you see the benefits?
With little more than basic data connectivity and a common dashboard, you can see the benefits within weeks. In reality, the predictive capability (in which ML models are genuinely predicting mill performance based on what is happening upstream) requires about 3 to 6 months of model development and testing with actual operational data.
Databricks. (2023, May 2). Data Silos Explained: Problems They Cause and Solutions. Retrieved from https://www.databricks.com/blog/data-silos-explained-problems-they-cause-and-solutions
Rovjok. (2023). Mine-to-Mill Analytics and Mill Optimisation. Retrieved from https://rovjok.com/casestudy/mine-to-mill-analytics-and-mill-optimisation/
Cognizant. (n.d.). Cloud-Based AI Analytics Solution for Mining—Case Study. Retrieved from https://www.cognizant.com/us/en/case-studies/mining-cloud-based-ai-analytics
Kohezion. (n.d.). Siloed Data: What is Data Silos, Problems and Solution. Retrieved from https://www.kohezion.com/blog/siloed-data