Mine Pit Stability Analysis Using Remote Mapping - A Multidisciplinary Matlab Application George Poropat 20 August 2014 ENERGY FLAGSHIP
Objective Remote mapping of the orientation and position of exposed structures in the wall of a mine to improve mine productivity and safety 2 George Poropat CSIRO
Rationale Rock mass structure controls the stability of mine pits and thus the safety of mining operations. Structure is one of the major factors affecting operations such as pit excavation and resource extraction and thus the profitability of a mining operation. 3 George Poropat CSIRO
Rationale Mine pits can suffer catastrophic failures that destroy equipment, injure and kill workers and may sterilise the resource. 4 George Poropat CSIRO
Objective Remote mapping of the orientation and position of exposed structures in the wall of a mine The Problem The slopes of mine pits are so large that measurement methods that require physical access (contact) with the rock are extremely dangerous. Remote mapping of the orientation and position of surfaces is required and it must be easy, cheap, fast and accurate. 5 George Poropat CSIRO
Objective Remote mapping of the orientation and position of exposed structures in the wall of a mine The Approach Create a point cloud model of the surface to be mapped Initial: Laser Scanning to Create 3D Point Cloud Models The advantage of laser scanning is that is easily understood by users. The problems with laser scanning start with cost and get worse with weight and resolution requirements. Final: Create Point Clouds Using Photogrammetry Digital imaging is cheap although when the project started it was not so cheap and resolution was limited (so take a chance on technology evolving). The problem is computing power requirements and lack of user understanding of how it works. 6 George Poropat CSIRO
Objective Remote mapping of the orientation and position of exposed structures in the wall of a mine The User Requirements Initial: Estimate the direction of the normal to an exposed plane surface in a rock mass. The Real Requirements: Estimate the orientation of visible fractures i.e. the exposed edges of discontinuities in the rock mass. Makes using laser systems even more difficult. Estimate the spacing between the planes. Estimate the size and position of the planes / discontinuities. Analyse the geometrical relationships between the planes. Do all that quickly. 7 George Poropat CSIRO
3D Imaging Solution A Digression: What is A 3D Image? Shape and colour integrated to create representations of objects that can be viewed and analysed by humans or computers. 3D images are accurate 3D objects visualised and manipulated using computer graphics. 3D images can be used to extract spatial information that cannot be obtained easily or quickly by other methods. 8 George Poropat CSIRO
3D Imaging Computer generated 3D image Computer created 3D visualisation of data 9 George Poropat CSIRO
3D Imaging Computer rendered 3D image A 3D image of a real world object 10 George Poropat CSIRO
Implications of User Requirements Computational time had to be reasonable Graphics interface had to be flexible Development time had to be short Fast code development Maximum use of standard libraries e.g. matrix inversion Good debugging to support development and bug fixes Easy compilation. 11 George Poropat CSIRO
The Development Requirements Image Processing Ability to handle image files Large computational load By the standards of 15 years ago very large loads Visualisation Detailed graphics display Textured overlay User Interaction Real time interaction in 3D Make it happen fast and at low cost!!!!! 12 George Poropat CSIRO
The Result: 3D Images of Mining Operations 13 George Poropat CSIRO
The Result: Large Scale Aerial 3D Imaging 14 George Poropat CSIRO
The Development Initial test code built using Matlab 5 (1999) First release build used Matlab 6 Matlab 6 handled 8 bit and single precision data types. Compiler created C code Required editing to eliminate functions that would not compile Updated to Matlab 7 compiler model Performance Improvements Hand coded high performance correlation functions 15 George Poropat CSIRO
The Development Commercial Release Completely re-engineered front end based on C#.NET Combination of client and customer requirements Retained Matlab 7 compiler model Performance improvements in custom libraries OpenMP multicore implementation of correlation functions GPU implementation of correlation functions 16 George Poropat CSIRO
Advantages Code development is fast. Development of mathematically based algorithms is assisted by the various debugging tools although other tools are getting more flexible. Debugging is fast. A programmer who is familiar with the application typically finds bugs easily and fixes are usually turned around within 24 hours 17 George Poropat CSIRO
Downside Shipping and installing the MCR which changes with each release of Matlab is a pain although the system tracks the versions. Clients may find the size of the installation pack an issue Code security must be managed 18 George Poropat CSIRO
Summary The application combined: Image file handling Image processing Intensive mathematic processing Large scale parameter estimation using matrix algebra Probabilistic modelling etc. Computational geometry Visualisation The Matlab development environment and in-built functional libraries contributed significantly to the development. 19 George Poropat CSIRO
The Future Investigating the use of the Matlab Coder A return to the original compiler architecture? The Matlab Coder (the original compiler model) but may improve performance (current tests on some functions ~4 times speed up) and make delivery and maintenance easier. 20 George Poropat CSIRO
Thank you Energy Flagship George Poropat Research Scientist Telephone +61 7 3327 4425 email George.poropat@csiro.au www.csiro.au ENERGY FLAGSHIP