Stockpiles offer helpful alternatives for mine planning to ensure the ore quality and amount of material required by the processing plant. In order to satisfy industrial requirements of grades and tonnes, reducing stockpile fluctuations represent an important tool particularly for short-term mine planning.

The latest Release of our Stockpile Application, allows mine sites to implement more efficient stockpile management and processing of the material at the right time it needs to. The Stockpile Application feeds information to the Material Transport Model which supports other optimization applications for processing plant circuits.

On the other hand, it also enables customers to see the quality of the data that feeds the stockpiles 3D block models and provides a solution to fix it if required to improve the accuracy of the models.

Challenge 1

We don’t know what is sent to the plant, given that our operation carries out blending by the actions of bulldozers. Our WAM model doesn’t take into account dozing.


Development of a stockpile model that incorporates dozer pushes.


  • Delivery of a known, consistent feed into the processing plant.

  • Grade knowledge and ability to reconcile to crushers grades.

Challenge 2

We don’t know exactly from which stockpile we are feeding the crushers.


Develop a Reclaim Destination feature to allow material flows to be organized according to reclaim destination. Material properties and tonnes reclaimed from different stockpiles and sent to Crushers can be tracked in time series.

Ensure the model is providing accurate estimations of the feed characteristics when reclaiming.

Challenge 3

Unknown quality of the FMS data sent as input for the stockpile model.


Generate Data Quality Metrics/dashboards to show customers the quality of the input data, therefore how robust the stockpile model is.


  • Enable users to quickly detect and correct reporting or operational errors/mistakes.

  • In the case where Stockpile model outputs are questioned, it can assist with showing where data (rather than the model) has been at fault.

Challenge 4

To export the content of a "reporting dashboard" into a CSV file for internal purposes, so it is possible to copy and paste at once those values into an Excel sheet and avoid making mistakes when typing.


To enable text boxes to be exported into CSV files, besides adding on the button of the file, average, and totals for each variable in consideration.


Increase the use of the application onsite.

Challenge 5

Unknown quality of the scan sent for base/initialization/conformation of the stockpile model.

Scan Data quality analysis takes too long.


Create a functionality that automates the scan data quality process and generates a PDF report attached to the due asset in the app.

The scan checker converter carries out four types of checks:

  • I/O checks: whether the raw scan can be read

  • Topology checks: whether the scan is aligned with the associated base scan and other scans of the stockpile in question.

  • Physical checks: whether the scan displays physically plausible characteristics.


Customers can easily and quickly validate the quality of their scans through the in-app scan report and take the actions for improvements if due.

Challenge 6

A large amount of invalid/missing GPS coordinates from the FMS decrease the accuracy of the 3D stockpile Block model.


Implementation of GPS Filling functionality to identify complex spatial pattern issues within stockpile events and fill them properly into the 3D Stockpile Block model.

The Neural network fill approach

The GPS filling problem can be approached as a supervised learning problem. A Neural Network can be trained on historical data without missing GPS values. Consider a moving window of 10 events prior to each event. The moving window of previous (x,y) coordinates can be used as training input, with the next (x,y) coordinates used as the training label. This approach is expected to perform best and could identify more complex spatial patterns within stockpile events.

GPS Filling metrics

The base metrics give additional info on GPS filling frequency, predictions, and uncertainty. Example of GPS filling metrics are:

  • A number of dumps/loads coordinates filled with GPS filling.

  • Mass of dumps/loads coordinates filled with GPS filling.

  • Filled dumps/loads x coordinate prediction uncertainty from GPS filling

  • Filled dumps/loads y coordinate predicted by GPS filling.


Accurate models enable better mine planning decisions, therefore optimizing the value and performance of the plant.

Contact Us so we can show you how the latest Stockpile Application can help you improve your mine planning processes.

Did this answer your question?