Contaminant Fate and Reactive Transport in Groundwater - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 867

• sequential (iterative or non-iterative) approach (operator splitting), which solves

separately the transport and reaction terms. Widely used multispecies and multi-
component reactive transport models such as RT3D (Clement 1997 ) and PHT3D
(Prommer et al. 2003 ) are based on this method.

19.3.3 Model Application

The application of a groundwater flow and transport model, incorporating the princi-
pal hydrogeological and hydrogeochemical conditions of a particular contaminated
site, can be described as a succession of several steps (e.g. Bear et al. 1992 ; Zheng
and Bennet 2002 ):

(1) Formulation of objectives: the identification of the goals and purposes of the
modeling study, which could include an improvement of the understanding of
the transport regime at a particular site, or setting the priority for remediation
depending on contaminant transport characteristics.
(2) Review and interpretation of the available data: compilation of all the sources
of information from the site characterization.
(3) Development (or improvement) of the conceptual model based on the compiled
information including geological, hydrogeological, geochemical data, together
with information on the site history and presumable location of contaminant
sources.
(4) Field data collection: at this stage the need to collect new data in addition to
those available from the site characterization should be evaluated.
(5) Selection of an appropriate computer code: the choice should depend on differ-
ent factors including not only the capability to represent the principal processes
determining the contaminant transport at the site, but also on the goals of the
modeling study and the expertise of the modeler.
(6) Construction of the contaminant transport model: by preparing the input data,
the conceptual model is translated in the site-specific mathematical model.
(7) Model calibration and sensitivity analysis: after the model has been run with
some initial estimates of model parameters, these input parameters must be
adjusted, either manually (trial-and-error) or assisted by a computer optimiza-
tion code, in order to match the field observations to a reasonable degree.
Sensitivity analysis is performed before and after calibration to test the sen-
sitivity of the computed model results to some input parameters (e.g. Hill and
Tiedeman 2007 ).
(8) Predictive simulations: after the contaminant transport model has been cali-
brated it can be run to predict future contamination scenarios and/or the impact
of remediation activities.
(9) Uncertainty analysis: many uncertainty sources are associated with a field scale
modeling study such as incomplete understanding of physical and biogeochem-
ical processes, spatial and temporal variability of model parameters and/or