with durations varying from 10 days in the column to a year in the tank. In contrast,
Heiderscheidt et al. (2008) reported a poor efficiency of NAPL layers oxidation by
KMnO 4 that was justified by the presence of the NAPL creating flow bypass,
without scale comparison. Scale comparison was done for biodegradation in heterogeneous systems (Davis et al. 2003) and clearly evidenced a decrease of biodegradation rates each time the scale of the experiment was increased. The role of
sediment heterogeneity on the efficiency of surfactant dissolution has been clearly
demonstrated by Page et al. (2007) in a 2D vertical tank experiment, indicating even
an increased concentration downgradient after treatment. Modeling of these types of
experiments also enhances the much lower dissolution rates in 2D experiments
compared to column results (Schaerlaekens and Feyen 2004).
For thermal treatments, although at a higher temperature, Lagadec et al. (2000)
also showed the same efficiency at small (8 g of soil) and intermediate (8 kg scale),
while results also show lower efficiency in presence of high heterogeneity (Tzovolou
et al. 2011).
4.7 Conclusions and Perspectives for Field Applications
The first important result of this research is that only low-temperature treatment is a
predictable and scalable technique, which can reach a 100% pollutant removal.
Indeed, heat was well spread across the deeper part of the porous medium at the
pilot scale. However, an additional heat source would have been needed near the soil
surface. It is quite interesting to outline that in column experiments, the removal
yield was lower than 100% in presence of clay leading to a hydraulic conductivity of
10
À5 m s
À1 , while it reached 100% in the lens of the 3D tank with the same property
but prepared with fine sand. Therefore, the main limitation of thermal remediation is
the presence of clays that prevents the vertical movement of gas bubbles. Except for
these conditions, thermal treatment only requires a correct dimensioning to reach the
target temperature, often 80
C, in the whole contaminated area.
The other techniques require a more careful analysis for a successful application
in the field. Concerning oxidation, batch and column experiments showed that
persulfate was a good candidate for BTX oxidation because it was remnant in the
field. However, it requires high concentrations (several g L
À1 ) due to its low electron
acceptor content, and its higher efficiency at these concentrations. These high
concentrations lead to density-driven vertical flow. The success of the technique is
then conditioned by a strict control of the dense solution behavior. This may be done
by localized injection on top of the contaminated lenses, which requires a very dense
network of contaminant characterization.
Surfactant injection looked promising in columns although it is less efficient on
soluble hydrocarbons like toluene than it has been observed on chlorinated solvents
(Pennell et al. 1994). This may explain why the columns’ results showed a removal
lower than 100%. This slight limited access to contaminant in column experiments
was more evidenced at the tank scale where the low permeability lenses were much
234
F. Jousse et al.
Heiderscheidt et al. (2008) reported a poor efficiency of NAPL layers oxidation by
KMnO 4 that was justified by the presence of the NAPL creating flow bypass,
without scale comparison. Scale comparison was done for biodegradation in heterogeneous systems (Davis et al. 2003) and clearly evidenced a decrease of biodegradation rates each time the scale of the experiment was increased. The role of
sediment heterogeneity on the efficiency of surfactant dissolution has been clearly
demonstrated by Page et al. (2007) in a 2D vertical tank experiment, indicating even
an increased concentration downgradient after treatment. Modeling of these types of
experiments also enhances the much lower dissolution rates in 2D experiments
compared to column results (Schaerlaekens and Feyen 2004).
For thermal treatments, although at a higher temperature, Lagadec et al. (2000)
also showed the same efficiency at small (8 g of soil) and intermediate (8 kg scale),
while results also show lower efficiency in presence of high heterogeneity (Tzovolou
et al. 2011).
4.7 Conclusions and Perspectives for Field Applications
The first important result of this research is that only low-temperature treatment is a
predictable and scalable technique, which can reach a 100% pollutant removal.
Indeed, heat was well spread across the deeper part of the porous medium at the
pilot scale. However, an additional heat source would have been needed near the soil
surface. It is quite interesting to outline that in column experiments, the removal
yield was lower than 100% in presence of clay leading to a hydraulic conductivity of
10
À5 m s
À1 , while it reached 100% in the lens of the 3D tank with the same property
but prepared with fine sand. Therefore, the main limitation of thermal remediation is
the presence of clays that prevents the vertical movement of gas bubbles. Except for
these conditions, thermal treatment only requires a correct dimensioning to reach the
target temperature, often 80
C, in the whole contaminated area.
The other techniques require a more careful analysis for a successful application
in the field. Concerning oxidation, batch and column experiments showed that
persulfate was a good candidate for BTX oxidation because it was remnant in the
field. However, it requires high concentrations (several g L
À1 ) due to its low electron
acceptor content, and its higher efficiency at these concentrations. These high
concentrations lead to density-driven vertical flow. The success of the technique is
then conditioned by a strict control of the dense solution behavior. This may be done
by localized injection on top of the contaminated lenses, which requires a very dense
network of contaminant characterization.
Surfactant injection looked promising in columns although it is less efficient on
soluble hydrocarbons like toluene than it has been observed on chlorinated solvents
(Pennell et al. 1994). This may explain why the columns’ results showed a removal
lower than 100%. This slight limited access to contaminant in column experiments
was more evidenced at the tank scale where the low permeability lenses were much
234
F. Jousse et al.