Fractures, FracMan and Fragmentation

Fractures, FracMan® and Fragmentation Applications of DFN Models to Block & Panel Caving

P62-1461 1

Background Golder Associates are one of the pioneering groups in the use of the Discrete Fracture Network (DFN) approach. DFN models seek to describe the heterogeneous nature of fractured rock masses by explicitly representing key elements of the fracture system as discrete objects in space with appropriately defined geometries and properties. By building geologically realistic models that combine the larger observed deterministic structures with smaller stochastically inferred fractures, DFN models capture both the geometry and connectivity of the fracture network as well as the geometry of the associated intact rock blocks. They are therefore ideally suited to addressing a wide range of both geomechanical and hydraulic problems associated with fractured rock masses.

The first commercially available DFN code was FracMan, released in 1986 by Golder Associates Inc. This code has been developed continuously over the past 20 years, and is now a Windows XP based object oriented code capable of modeling over 20 million fractures per simulation. Whilst it was initially developed with mining and civil engineering applications in mind, to date it has been applied primarily to oil and gas and environmental projects, including radioactive waste management. FracMan is now being applied to a wide range of mining geomechanical problems including slope and tunnel stability, in situ fragmentation prediction and groundwater management.

In recognition of this interest in the use of its DFN capabilities, Golder is releasing a preliminary version of FracMan known as FracMan Geomechanics Edition (FRAG) which will contains tools aimed at solving a number of mining related fracture problems. In addition to the tools included in FRAG, Golder has developed a series of technologies applications and algorithms for discrete fracture analysis of Block and Panel Caving (BPC) problems that have not yet been incorporated to FracMan.

The purpose of this brief document is to lay out ideas about some of the issues and applications of DFN modeling and FracMan have to both in situ and primary fragmentation assessment in order to stimulate the ongoing efforts in this area.

Developing a FracMan DFN Model

Building a DFN model comprises more than simply deriving the statistical properties of the fracture system. A key element of DFN modeling is the derivation of the underlying conceptual fracture model and how the fractures spatially distribute themselves within the rock volume. As such a DFN model is not just a statistical description but an ordered geologically realistic representation of the distribution, interaction and geometry of rock mass fractures. Figure 1 shows a flow diagram for the development and application of a DFN model.

The process of building a DFN model involves 4 main stages: data analysis; conceptual model development; model building and validation. These activities are seen within Figure 1 as the “validation loop”. This is a refining process of testing the developed DFN model against various validation data sets such as borehole or mapping fracture data in order to ensure a good fit between the simulated and actual fracture data. The analysis of the data and conceptualization of the fracture system needs to be constantly updated until the validation is optimized and this can often be a long term process with an early conceptual model


P62-1461 2

being found to be inadequate as new data are acquired and new understanding developed. However once the level of validation achieved is satisfactory, then the DFN model is available for use.

One significant area that requires consideration is the development of methodologies and tools for integrating data from increasingly complex and varied data acquisition systems. With the increasing use of photogrammetric techniques, our ability to map the rock mass has leapt ahead of our ability to easily capture that data and convert it into a DFN model. As the accurate capturing of the rock mass structure is a cornerstone of the DFN approach, effort is needed to ensure that the full value is extracted from the exotic and often expensive data sources such as Lidar, photogrammetry, borehole cameras, televiewers, micro-seismics as well as other surface and borehole geophysics.

An important aspect of a FracMan model is that while a single realization may be used, the stochastic nature of the models means that multiple realizations can be generated and the likely range of outcomes assessed using Monte Carlo simulation techniques. This means that quantification of the rock mass structure, design of the cave and even equipment selection can feasibly be based upon a probabilistic assessment.

FracMan and In Situ Fragmentation

It is our belief that that unless you obtain the best possible assessment of in situ fragmentation, any attempts to predict primary or secondary fragmentation is unlikely to be successful. With this in mind, there are two main applications of a DFN model that are relevant to in situ fragmentation.

In Situ Fragmentation Assessment

Golder has developed several technologies for searching through a fracture network in order to define the natural pre caving (in situ) state of the rock mass fragmentation. Whilst previous research emphasis has been placed on defining blocks in 2D, FracMan offers the considerable advantage that its DFN fragmentation analyses are 3D. This can be performed in a number of ways either rapidly using a ray tracing algorithm for approximating block volume or through more complex 3D search algorithms (such as the convex hulls algorithm) that define the extent of any fully or partially defined rock block, including block volume, mass, size and shape (see Figure 2). One area that needs addressing further is the nature of the rock bridges that prevent full block formation. Current efforts have focused largely on defining fragmentation in terms of block formation but quantifying the geometry and properties of the intact rock is a key part of the overall in situ fragmentation assessment and critical to understanding primary fragmentation and the rock mass caveability.

Pre-Conditioning Evaluation

There is an increasing trend in the use of pre-conditioning as a means of improving the in situ fragmentation and therefore caveability of a rock mass. However much of this pre-conditioning is undertaken without quantitatively examining the likely impact that induced hydraulic fractures might have on the in situ fragmentation of a rock mass. FracMan provides a means to generate additional fractures in the DFN model to represent the induced fractures based upon the fracing design and in


P62-1461 3

situ stress field. The resultant model can then be re-evaluated in terms of improved fracture connectivity and fragmentation in order to assess the likely relative performance of the planned pre-conditioning as well as providing a means to optimize the design and layout of the pre-conditioning system. A simple example is shown in Figure 3.

FracMan and Primary Fragmentation

FracMan clearly has application to the definition and assessment of the in situ fragmentation as that is a natural fracture problem that FracMan is designed (although not necessarily optimised) to evaluate. However we believe that FracMan can also add real value to the primary fragmentation assessment and below are outlined three differing routes by which we believe this may be achieved (summarised in Figure 4). It is our belief that the ELFEN code developed by RockField Software Ltd is one of the best codes available for simulating fragmentation through fractured rock masses and we have already made some progress in linking FracMan and Elfen together. However we also recognise that the ability to take a FracMan model in some form into a wide range of advanced numerical codes offers the most flexible route to the wider community. Whilst the sections below refer primarily to Elfen there may be other current or future codes offering similar or improved capabilities.

Direct FracMan-Elfen Output

Whilst the direct simulation of a FracMan DFN model within Elfen is a laudable aim, in the near term the input DFN model needs to be simplified significantly to make the problem tractable and the run times sensible. Whether exporting 2D or 3D models, a series of model processing tools are needed to simplify the DFN model so that it can be used to generate a viable Elfen grid. See Figure 4a.

Partially Upscaled FracMan-Elfen Output

Golder has considerable experience in generating upscaled equivalent continuum permeability values from an underlying DFN model. In a similar way, it is believed that parts of a DFN model could be converted to a continuum equivalent of realistic rock mass properties complete with directional anisotropy, whilst preserving the more important fractures as discrete elements (see Figure4a). Conventional rock mass classification systems poorly handle the underlying fracture system and therefore using a DFN approach to deriving a continuum equivalent is a logical route to follow. Again this mixed Discrete-Continuum approach would make the problem far more tractable and reduce run times (Figure 4b).

“Reversing the problem” – the Elfen-FracMan solution

An alternative approach to the numerical simulation of a DFN model is to turn the problem around. By undertaking numerical experiments in Elfen in order to understand how the fragmentation process progresses (i.e. which fractures are most likely to extend, when do new ones form, which blocks break etc), the additional fractures caused during primary fragmentation could be stochastically generated using FracMan. In this way multiple realisations could be generated using a Monte Carlo approach and therefore a possibly better constrained result generated than is possible from a single realisation of a more complex and time consuming numerical simulation (see Figure 4a).


P62-1461 4

Summary

The FracMan suite of codes provides access to a wide range of validated applications and algorithms developed over a number of years, but these are not yet integrated into a single modeling system that is optimized for addressing in situ fragmentation problems. FracMan represents a versatile tool for building geologically realistic and statistically valid descriptions of the rock mass fracture system and as such is a logical starting point for addressing both in situ and primary fragmentation issues. Despite the work that remains to be done, we believe the FracMan is the best available set of tools from which to build a powerful caving-oriented fracture modeling system. Whatever the approach that is adopted to defining both in situ and primary fragmentation, we strongly believe that it needs to be grounded in a realistic description of the fracture network and associated rock mass, defined in a tractable manner allowing practical modeling run times and with the ability to address and quantify uncertainty and variability.

3D Sampling

Geological Model

Data Analysis & DFN Inputs

ModelValidation DFN Model

Conceptual Fracture Model

2D Applications 3D Applications

1. Block Identification2. Block Volume3. Block Kinematics4. Rock Bridge Quantification5. Pre-Conditioning evaluation6. Fracture Connectivity7. Flow simulation8. Equivalent property calculation

Simulation of borehole dataSimulation of tunnel trace mapsSimulation of well tests, KH

Borehole data

Well Testing

Geophysics

Mapping/Imaging

Main Data Sources

Validation Loop

1. Step path analysis2. Rock Bridge Quantification3. Equivalent property calculation

Export 2D/3D fracture model to numerical codes

Data

Analysis

Conceptualisation

Validation

Application

Figure 1: Generalised work flow for the development and application of a DFN model


P62-1461 5

Figure 2: An example where a DFN model (left) was searched to define the in situ fragmentation around a circular tunnel (right). See inset of an example block that is defined by multiple fracture planes as an exotic polyhedron. This is the basis for how we would define 3D in situ fragmentation σhmaxa) c)

σhmaxb)

σhmin σ2

σ2σhmin

Figure 3: Using a DFN model to evaluate the impact of rock mass pre conditioning through hydraulic fracture creation. a) consider a simple poorly connected fracture system in a rock mass and for this illustration, σ2 is vertical. (b) If σhmax is aligned with the fracture system hydro-fracs that are created (red fractures) will not alter the fracture connectivity or block forming potential. However in c) hydro-fracs created in alternative direction (purple) significantly increase the connectivity and block forming potential of the rockmass


P62-1461 6

Figure 4: a) Three alternative approaches to improving the tractability of fragmentation modelling using FracMan and Elfen. b) Illustration of the “upscaling” of part of a DFN model to equivalent cellular properties whilst preserving the key discrete structures

FracManDFN Model

Elfen

FracManDFN Model

Elfen

FracManDFN Model

FracManDFN Model

In SituFragmentationAssessment

PrimaryFragmentationAssessment

FracMan –ElfenInterface

Model Simplification

Model Simplification

SelectiveUpscaling

Fragmentation Rules

1 2 3a)

b)

ElfenExperiments

Fractures, FracMan and Fragmentation

Documents