In our previous edition, we explored the potential of Autonomous Decision Ecosystems for Peruvian mining. Today, we delve deeper into how these systems, powered by artificial intelligence, can drastically optimize one of the most critical and costly processes: blasting, starting from a detailed understanding of the rock mass.

The Challenge: Terrain Variability vs. Generic Blasting Plans

Every meter of rock presents unique geomechanical properties. However, blasting plans are often based on averages for large zones, which can lead to inefficient use of explosives like ANFO. The consequences are well-known:

  • Excessive ANFO Consumption: Directly impacting operational costs.

  • Suboptimal Fragmentation: Causing problems in loading, hauling, and subsequent plant processes (crushing, grinding).

  • Undesired Damage: Affecting the stability of the remaining rock mass.

The Solution: Decision Agents for Intelligent ANFO Loading per Blasthole

Imagine a "Digital Blasting Assistant." This AI agent, fed with precise estimates of rock mass properties (such as the Blastability Index, RQD, UCS) for each section of the pattern, can:

  1. Integrate Key Information: Combine the drill pattern design with the estimated geomechanical characteristics for each blasthole or small group of them.

  2. Analyze with Precision: Utilize advanced models to determine the optimal blasting energy and, therefore, the amount of ANFO needed blasthole by blasthole. This analysis considers how different rock properties (hardness, fracturing) will influence fragmentation.

  3. Generate Detailed Recommendations: Produce a highly granular ANFO loading plan, optimized for fragmentation targets and minimizing waste, which the blasting team can implement with greater confidence.

Potential Value

  • Reduction in ANFO Consumption: Blasthole-by-blasthole optimizations can lead to significant savings.

  • Improved Downstream Productivity: Consistent and adequate fragmentation improves efficiency throughout the value chain, from loading to the plant (potential impact on overall throughput).

  • Safer and More Sustainable Operations: Better-designed blasts mean greater control, reducing risks and environmental impact.

Looking Ahead: Can We Predict the Rock Before Drilling the Next Pattern?

The Constant Challenge: Planning with Geological Uncertainty

While characterizing the rock mass where we have already drilled is fundamental, a major operational and planning challenge is anticipating conditions in adjacent areas that have not yet been explored in detail.

AI as a Bridge: Inferring Geomechanical Properties Between Patterns

Here, artificial intelligence offers a fascinating prospective capability. An AI system could:

  1. Learn from Known Zones: Be trained with historical data from drill holes, geological interpretations, and detailed estimates of rock mass properties from already characterized areas.

  2. Identify Complex Relationships: Discover spatial patterns and correlations between geomechanical properties and other available data (e.g., lithology, alterations, geological structures, geophysical data if available).

  3. Project Towards the Unknown: Generate initial and probabilistic estimates of geomechanical properties (BI, RQD, UCS) for adjacent patterns where direct information is scarce or non-existent.

Benefits for Planning and Operation

  • Informed Drilling Planning: Guide future drilling campaigns towards areas of greatest interest or uncertainty.

  • Preliminary Blasting Design Estimation: Have an initial idea of the challenges and blasting requirements in new areas.

  • Proactive Geotechnical Risk Management: Anticipate potential changes in rock mass stability or behavior.

Important: This inference does not replace the need for confirmatory drilling, but it does make it more efficient and targeted.

Practical Implementation: A Phased Approach

In line with the philosophy of Autonomous Decision Ecosystems, the adoption of these AI technologies should be progressive:

  • Phase 1 (Quick Value): Implement a recommendation agent for ANFO optimization in a specific sector, using the best available geomechanical estimates. Validate savings and improvements.

  • Phase 2 (Expansion): Extend the ANFO agent to more areas and begin developing/testing geomechanical inference models for adjacent patterns, feeding them with collected data.

  • Phase 3 (Integration): Seek integration of these agents with planning and execution systems for more holistic optimization.

FLUENTDATA: Transforming Data into Valuable Decisions

At FluentData, we specialize in converting complex data into intelligent and actionable decision-making tools. We help Peruvian mining companies to:

  • Maximize Their Existing Data: Even with limited databases, we can identify opportunities to apply AI and generate impact.

  • Implement Focused AI Solutions: Such as ANFO optimization agents or predictive models for geomechanical properties. To accelerate value demonstration, we can use our geomechanical simulation and estimation platform, GeoPredict, to quickly generate a baseline rock mass model and test how an AI agent would interact with it in a Proof of Concept (PoC).

  • Build Capabilities Progressively: Accompanying them every step of the way towards a more optimized and autonomous operation.

"True innovation is not about implementing the latest without context, but about intelligently adapting technology to our reality to generate real value."