- About Us
- Contact Us
- Our History
- Find Software Agent
- Office Locations
- Training & Events
Application of Monitoring Data for Calibrating Numerical Models
James Jung, P.G.: Senior Geotechnical Engineer, Rio Tinto Kennecott Copper, Operational & Technical Support
Tatyana Katsaga, Ph.D.: Senior Rock Mechanics Engineer, Itasca Consulting Canada, Inc.
Loren Lorig, Ph.D., P.E.: Principal Engineer, Itasca Consulting Group, Inc.
Three-dimensional numerical models are playing increasingly important roles in stability analyses for open pit mining.
In order to increase confidence in the results, it is crucial that model predictions represent observations, measurements and realistic mechanisms. This paper lays out practical application and best practices in achieving reliable model calibration, which is crucial to having confidence in predictive results. Field data should be utilized at every stage of large modelling projects, as follows.
During model setup, geotechnical model data such as down-hole measurements, can inform how structures are treated (i.e., explicit model interface vs. discretized zone), the role of water and how strengths are distributed spatially. During this stage, target calibration criteria need to be established among stakeholders according to the problem being considered.
During model calibration, the model is compared to historical monitoring records and slope conditions. Success at this stage hinges upon data available, understanding data limitations, and good knowledge of the historical mechanism being represented.
Once the model has met pre-defined calibration criteria, model predictions should be reconciled against current monitoring data at ongoing intervals as mining progresses to reconcile model inputs and original predictions. During this stage, model predictions can also be used to guide new instrument layouts based on predicted mechanisms and deformation.
Throughout the modelling process, confidence in the model resulting from underlying data, model idealizations and the ability to represent realistic mechanisms, etc. should be discussed with management to provide necessary resources for model improvement and to select appropriate risk acceptance criteria.
An example of displacement on a structure vs shear predicted by a model.