Skip to main content
Publicly Available Case
How research teams improved blast design through rock-mass classification from MWD
Supervised ML turns pressure/torque/RPM into geotech classes. Better inputs cut overbreak/underbreak and downstream losses.
Context
Geotechnical characterization relied on limited sampling and manual interpretation, making it difficult to evaluate rock properties at scale. This created uncertainty in blast design and contributed to overbreak or underbreak.
Solution
ML models were trained on MWD data to predict rock strength and geological units along drill holes. The results provided continuous profiles used to design more accurate blasts.
Results
The enhanced visibility enabled better blast design and reduced overbreak and underbreak, improving downstream fragmentation quality.
Data Inputs
MWD pressure, torque, RPM, penetration rate