Abstract
In the subsurface, there are two reasons why contaminant transport
behavior may not be properly captured by a discretized
advection-dispersion equation (ADE) model, and instead require a
non-Fickian treatment. These are: (i) velocity fluctuations smaller than
the model support scale, and (ii) mobile-immobile mass transfer (MIMT)
processes such as kinetic sorption and diffusion into secondary
porosity. Non-Fickian phenomena have been much studied in the
literature, are reasonably well understood, and can be modeled using a
variety of techniques, such as the continuous-time random walk (CTRW),
fractional ADE, and subordination approaches. When these more
sophisticated approaches to contaminant transport are used in the
academic literature, it is commonly with spatially uniform, quasi-1D
velocity fields. In professional practice, the situation is essentially
the converse: elaborate, large-scale numerical models of flow are the
norm, but transport modeling is generally limited to use of the ADE.
Naturally, it would be beneficial to both camps to combine sophisticated
flow models with sophisticated transport models. To do so, it is natural
to use particle tracking, as this allows for spatially non-uniform
advection. However, the primary question is how to define the stochastic
behavior of the particles in a way that directly corresponds to
underlying physics, is advection-independent, and can be used for
prediction rather than for post-hoc fitting, which is essentially the
state of the art. We will answer this question, presenting a new
conceptual approach that may be thought of as CTRW-on-a-streamline, in
which the travel-time probability distribution is constructed by
successive subordination using two distributions that (i) have direct
physical interpretations, (ii) can be predicted a priori, and (iii) are
independent of the underlying velocity field.
Open-source software using the new approach to model the two key sources
of non-Fickian contaminant transport on any arbitrary spatially
heterogeneous, transient velocity field that can be generated with
MODFLOW is in late-stage development. The software will be briefly
described and an illustrative simulation of a push-pull tracer test,
subject to kinetic sorption and drift in a heterogeneous aquifer, will
be presented.
Original language | English GB |
---|---|
Journal | Geophysical Research Abstracts |
Volume | 22 |
State | Published - 1 Dec 2018 |
Externally published | Yes |
Keywords
- 1829 Groundwater hydrology
- HYDROLOGYDE: 1831 Groundwater quality
- HYDROLOGYDE: 1832 Groundwater transport
- HYDROLOGYDE: 1847 Modeling
- HYDROLOGY