NASSCOM Community

Analytics Case Study Series: 1. Mining

Blog Post created by NASSCOM Community on Jun 9, 2016

The demand for energy is on the rise and resources being harder to locate, analytics will be imperative to meet demand.

Key Areas of application:

  • Reducing the time taken to locate natural resources
  • Reducing the costs associated with locating natural resources
  • Increasing the efficiency of both new and old deposit

The Case Details:

Organisation: Absolutdata

Client: A Mining major

Geography: US

Category: Predictive

The Situation:

  • Identifying failure risk by predicting engine failure
  • Changing engine oil at the ‘optimum time’ using predictive maintenance
  • Decreasing false alarm rates for operational efficiencies

The Solution:

  • Used terabytes of truck sensor’s data along with truck’s operations and alarms data
  • Merged data sources using Big Data, applied machine learning algorithms such as gradient boosted regression and cluster analysis
  • Used sentiment analysis and word cloud inputs to further refine the models

Results:

  • Alarms categorised into 5 categories, with critical alarms forming only 6% of the alarms as against 24%
  • 3-5 days of savings on every oil change leading to engine oil savings
  • Engine replacement costs down by 50%

Outcomes