HR: 0800h
AN: H31B-0376 [Abstracts]
TI: Aspects of Numerical Simulation of DNAPLs in Fractured Bedrock
AU: * Wealthall, G P
EM: g.wealthall@bgs.ac.uk
AF: British Geological Survey, Keyworth, Nottingham, NG12 5GG
United Kingdom
AU: Lerner, D N
EM: d.n.lerner@sheffield.ac.uk
AF: University of Sheffield, Mappin Street, Sheffield, S1 3JD
United Kingdom
AB:
The distribution of dense non-aqueous phase liquids (DNAPLs) is notoriously difficult to predict in heterogeneous geologies.
This single factor has serious implications for predicting DNAPL source zone architecture, assessing the risk to groundwater
from the dissolution of the non-aqueous phase, and targeting the source zone with a specific remediation technology.
When DNAPL enters a fractured bedrock the fluid will follow a complex path based predominantly on the heterogeneous
distribution of fractures in the rock mass where fluid migration is controlled by both large- and small-scale processes. At
the large-scale, connectivity of the fracture network defines whether a potential pathway is present. Whereas at the
small-scale, fracture aperture variability and pore size distribution of the matrix define the distribution of the DNAPL.
Simulating DNAPL migration in heterogeneous geologies demands numerical models which reconstruct the complex fracture network
architecture and at the same time couple the transport of the fluid phase through interconnected fractures.
Discrete fracture network models (DFNMs) should incorporate the 3-D characteristics of natural systems. DFNMs may be
deterministic or stochastic. Stochastic models are applied where an exact description of fractured rock mass is
unachievable. However, few DFNMs are available that incorporate 3-D non-orthogonal fracture geometries and, furthermore,
field measurements are often not available to adequately validate the fracture simulations.
DNAPL models may be classified into dynamic fully compositional simulators which are based on a detailed description of
two-phase flow physics and mass transfer terms, or static models which are limited to simpler invasion percolation concepts.
Problems of discretisation and convergence limit the application of coupled multiphase compositional simulators, which are,
at present, constrained to realisations of orthogonal fracture geometries. Furthermore, the computational time to run fully
compositional simulators in dense fracture networks often limits application to a limited number of model realisations and
therefore precludes application to stochastic methods. Investigators of DNAPL release to fractured bedrock are, therefore,
restricted to resolving representative descriptions of either the geology or the non-aqueous phase, but not both.
A recent study is described where a coupled stochastic DFNM and invasion percolation model is applied to conceptualise the
fate of DNAPLs in heterogeneous geologies. This approach is used, at a fundamental level, to explore the effect of
variability in natural fracture networks in estimating the distribution of DNAPLs in fractured rock. The implications of
this study are that prediction of DNAPL migration is not adequately described by deterministic approaches, that investigators
of fractured rock sites should expect to work with a range of values, and that the range may be large.
Discussion includes an assessment of the uncertainties associated with applying numerical simulation at real DNAPL release
sites in heterogeneous geologies and recommendations are presented to address future research needs.
DE: 1829 Groundwater hydrology
DE: 1831 Groundwater quality
DE: 1832 Groundwater transport
SC: Hydrology [H]
MN: 2004 AGU Fall Meeting