This article is 3 years old. Images might not display.
Orica's Design for Outcome (DfO) applies machine learning algorithms to automatically characterise the rock for blasting, setting a new benchmark for designing blasts in a digital age. It unlocks the power of using previously underutilised data sources both upstream and downstream of blasting to design the best blast every time.
By automatically processing data to better understand rock mass variability and tailoring blast designs accordingly, more consistent and targeted blast results are realised. This improves productivity and reduces cost down through the mining value chain.
MACHINE LEARNING ENABLED BLASTING - THE OPPORTUNITY
Blasting today commonly produces variable outcomes. This adversely impacts downstream productivity and overall costs through poorer fill factors and shovel productivity, oversize, reduced crusher throughput, _and reduced comminution productivity.
A key contributor is mismatched explosives energy to variable rock properties. Despite an increase in available sensor data at mines, conventional blast design techniques render it practically impossible for an engineer to combine and process all the data sources necessary to tailor blast designs to rock conditions at a high resolution. Most importantly, to do it in an everyday mining operations workflow. Design for Outcome applies machine learning and automation to the problem, creating a solution that allows drill and blast engineers to implement data-enabled tailored blast designs practically.
FUNCTIONALITY
Design for Outcome utilises machine learning algorithms and cloud-hosted processing to match rock characteristics to blast energy at high resolution to achieve targeted and consistent outcomes. It also automates the process and presents the user with a simple web-based interface that makes generating high resolution, bespoke blast designs easy. Design for Outcome is provided in various configurations: • Post-drill classification module This module automatically ingests Measure While Drill (MWD) data and produces blastability domains based on hardness.
Archived article: image not displayed.
The output generated is used in blast design packages with loading rules capability, such as Orica's SHOTPlus™ Premier blast design application.
• Pre-drill classification module
The pre-drill classification module utilises block model data augmented by existing high resolution MWD data. Machine learning algorithms generate blastability domains that inform drilling patterns for subsequent blasts. Blastability domaining is_then further refined post-drilling Below: Automated blastability domaining using Design for Outcome post drill classification. Raw MWD is cleansed and normalised to produce domains for automated loading rules using the post-drill classification module.
BENEFITS
Designed in collaboration with Orica's customers, domain experts and data scientists, the technology can be used to efficiently allocate drilling and explosives energy for consistent results, resulting in the following benefits:
• Reducing drill and blast costs while maintaining productivity by eliminating excessive use of explosives energy
• Improving fragmentation and downstream productivity by allocating appropriate energy to harder rock domains
• Utilising data from sensors implemented at mines in operational workflows rather than retrospectively
• Efficiently generating tailored blast_designs in production workflow timescales despite using high volumes of data
• Ease of use through a Web interface for drill and blast engineers ROY HILL CASE STUDY Through an innovation partnership, Design for Outcome was deployed at the Roy Hill iron ore mine in Western Australia to improve mining production and increase mining profit by reducing the variability in blast performance.
In collaboration with Roy Hill, Orica integrated drill design data through blast excavation in a cloud-hosted platform, providing data analytics to generate valuable insights on geological hardness, energy deployment, and mining productivity.
The Design for Outcome machine learning algorithms use MWD data from the autonomous drill fleet to domain the geology for the drill holes and match explosives energy to each domain, generating automated loading rules for blast charging and deploying to the field through BlastIQ™ enabled smart explosives delivery units.
The closed loop system is accompanied by automated excavation productivity reporting. Using Design for Outcome, Roy Hill has been able to target higher energy only where required, allowing drill and blast patterns and costs to be significantly reduced across the mine while maintaining benchmark excavation productivity.
WHAT'S NEXT
Design for Outcome will continue to expand value creation through better targeting drilling and explosives energy for cost and productivity benefits.
This includes:
• utilising new and emerging sources of orebody knowledge data to better inform high resolution domaining
• integrating blast outcome data sources into the machine learning algorithms
• advancing end-to-end integrated capability from orebody knowledge through to process optimisation, through integration with technologies such as the Integrated Extraction Simulator (IES) to enable sustainable Mine to Mill solutions.
