Engineered Data Science - Petra Data Science

Case Studies

Big data comminution

Algorithms – Mining’s crystal ball

This article was first published by Mining Magazine, 3 June 2017 and is reproduced here with . Read more...
Big data minerals processing

Newcrest-PETRA collaboration on big data delivers results

In his address to the Sydney Mining Club today, Newcrest CEO Sandeep Biswas talks . Read more...
PETRA Data Science named as finalist in Austmine 2017 Awards

The team at PETRA Data Science is delighted to announce that their engineered data . Read more...
Conditioning Big data

PRODFINDER® is a finalist in 2016 Mining Magazine Awards!

PRODFINDER® algorithms automatically link 3D geological, dispatch and blast data to processing plant historical data, and simultaneously take into account geology, blast design and plant operating factors to reveal optimal blast designs. Read more...
Blasting fragmentation

Machine learning AI enters underground mines

Our FRAGx<sup>®</sup> algorithms uses 3d mapping point cloud data (e.g. UGPSRapidMapper, 3D Laser Mapping or MVS) to automatically assess ore fragmentation in less than a minute. Read more...
Conditioning Big data

Big data blasts into mill efficiency and productivity studies

PETRA is excited to introduce our new PRODFINDER big data integration and statistical algorithms. These algorithms make it possible to conduct low cost big data "mine to mill" type projects without site trials! Read more...

Recent Projects

PRODFINDER^®

Developed virtual stockpiles for a pit to plant study to optimise blended stockpiles for plant performance.

Find Out More

FORESTALL^®

Applied dynamic time warping and machine learning to predict pump pressure spikes.

Find Out More

FORESTALL^®

Algorithm to predict SAG mill overloads resulting in downtime (real-time warnings using real-time plant PI data).

Find Out More

PRODFINDER^®

Predict plant performance using historical blended stockpile data.

Find Out More

PRODFINDER^®

Big data mine to mill study identified opportunities to optimise drill and blast to improve SAG mill performance.

Find Out More

PRODFINDER^®

Multifactor analysis of truck dispatch data identified characteristics of the best operators.

Find Out More

PRODFINDER^®

Optimise primary crusher feed rate and start-up speed for variable weather, fragmentation and ore type.

Find Out More

FRAGx^®

Automated drawpoint fragmentation assessment using machine learning AI.

Find Out More

Previous Clients

Making Data Productive

Our experience as mining professionals guides our practices, processes and procedures. Our services have been developed to help mining organisations use their data efficiently and effectively to improve their operations.

PETRA enables mining companies to move beyond tracking, reporting and incremental improvement, and into machine learning prediction, data visualisation and advanced analytics.
FORESTALL^®

FORESTALL® machine-learning predicts breakdowns using real-time data (e.g PI Tag data). With FORESTALL®, it's easy to create real-time alerts. The algorithm implemented with our FORESTALL®service estimates the likelihood that an event will occur and predicts this based on situational and subjective computational functions.

FORESTALL® allows decision makers to undertake early, quantitative risk assessment.
PRODFINDER^®

PRODFINDER® deep analytics and data integration algorithms makes your data assets productive. Our PRODFINDER® services handle complex interactions between factors and bias to estimate how important each factor is to productivity. PRODFINDER® is designed to break down data silos and merge geological data with historical mine and mill data.
PRODVIEW^®

Our PRODVIEW® services bring your data to life with 3D visualisation of your mining analytics data. PRODVIEW® data visualisation algorithms integrate 3D geological/blasting data with production data to uncover complex productivity drivers. These innovative algorithms link data silos to uncover new opportunities for improved productivity.
LABYRINTH^®

Our LABYRINTH® service is a novel statistical method to quantify intact rock damage and conditioning from x-ray microtomography data. This algorithm allows tens of thousands to millions of microfractures to be analysed simultaneously and increases the sample size similar to standard intact rock strength tests. LABYRINTH® is applicable to rock pieces, cores and fragments for testing.

PETRA Data Science

PETRA exists to help industry move beyond tracking, reporting and incremental improvement. Our experienced team of data scientists have worked to develop proprietary data analysis services which help our clients get the most out of their data.

All projects are managed by PhD-qualified engineering professionals. Because we understand resource industries, we are very productive, and our average project turnaround is only six weeks. We look for opportunities to leverage your existing systems and data infrastructure.
Mining Professionals & Data Scientists

Our team of dedicated data scientists all come from a background in mining. All of our projects are managed by our PhD-qualified mining professionals. Our experience means we know precisely what problems you face and how to add value to your existing data mining.

The value of data is becoming increasingly understood and valued but still remains under-utilised by mining companies. We work with you and your team to understand your situation and how we can help to reduce downtime and improve.
For Mining Professionals, By Mining Professionals

PETRA enables resource companies to capitalise on their big data assets. We assist companies to move beyond tracking, reporting and incremental improvement, and into machine learning, data visualisation and deep analytics.

We help companies find, predict and show productivity improvements in all types of mining data, whether it be; equipment IoT data, 3D data (mine/geological/blast), historical data (process/monitoring), or specialist data formats (lidar, seismic, images or video data).
Our Previous Projects
We’ve worked with a wide array of clients on a huge number of projects. Our recently completed projects include:
- Machine learning algorithm to provide plant operator an hour warning for SAG mill overload
- Applied dynamic time warping and machine learning to identify an error in flocculent addition in one of the thickeners that caused the GEHO pump pressure to spike
- Big data Mine-to-Mill study identified opportunities to optimise drill and blast to improve SAG mill performance
- Predict plant performance using historical blended stockpile data.