N. Baptiste, R.P.

Chapuis / Engineering Geology 184 (2015) 111118 117

for constant-head tests with fully penetrating MWs, but slightly lower 6. Discussion
MW measuring capacities. The differences can be explained by two
reasons: The aquifer K values may be assessed using several methods, and at
different scales. The small scale is that of soil samples: their quality must
1. The slug test was performed using a partially penetrating MW be assessed and their grain size distribution curve (GSDC) analysed to
whereas the constant-head tests were performed using a fully pene- check for mixes of sub-layers, before using reliable methods to predict
trating MW. the K values. The middle scale is that of eld permeability tests in driven
2. The slug test has a smaller radius of inuence due to the short dura- ush-joint casings and monitoring wells (MWs), for which reliable
tion of initial slugging. methods must be used (Chapuis and Chenaf, 2002). The large scale is
that of pumping tests.
It means that for slug tests in MWs, the maximum Ktest value is about This paper has studied only permeability tests with MWs (middle
103 m/s for large slots and correct lter sand, and they may be as low scale evaluation). It has determined how the parasitic head losses in
as 5 106 m/s for ne slots and some eld clogging exceeding the the- the screen and lter pack inuence the Ktest value and by how much
oretical rule of 50% clogging that is used for fully developed pumping Ktest differs from the true K3 of the tested soil. These ndings are valid
well screens. only for permeability tests in MWs. If the same MWs are used to moni-
tor a pumping test, the screen and lter will have a different inuence
5.2.4. Cases of properly designed MWs on the extracted K (pumping test) value, which will be mostly due to
After having shown that usual MWs have a eld measuring capacity some increased time-lag in the drawdown curve. The study of this inu-
in the range of 105 to 103 m/s, let us look at perfectly designed and ence during pumping test is outside the scope of the paper.
installed MWs, following the rules of Table 2. Consider rst the case of Because pumping tests involve large volumes, which are more likely
uniform aquifer and lter materials (indexes am and fm): if the GSDCs to meet preferential ow paths, one may believe that they may yield
are parallel, then the D50 ratio is equal to the D10 ratio and thus the larger K values than tests at a smaller scale (Bradbury and Muldoon,
ratio K2/K3 is nearly equal to the square of the D10 ratio (Chapuis, 1990; Rovey and Niemann, 2001). Thus, the K value may increase
2004, 2012a). As a result, K2 is higher than about 25 K3. This means with the tested volume, which is viewed as a scale effect. However,
that the hydraulic head losses in the lter material can be neglected there is no consensus about this scale effect. Many studies tried to
for a eld permeability test. In this rst case, the MW eld measuring check or challenge academic opinions, examining testing techniques,
capacity is controlled exclusively by the screen. investigated scales, and geologic media. Alas, the quality of each K
Consider a second case of an aquifer material with the following value was not questioned. Regrettably, the quality control of groundwa-
characteristics: D10 (am) = 0.92 mm, D 60 (am) = 0.74 mm, ter parameters, which must be methodically completed for engineered
CU (am) = 8, and D30 (am) = 0.15 mm. According to the criteria facilities, is rarely done for environmental studies (Chapuis, 1995).
of Table 2, since CU is higher than 5, in the graph (Figure 7) and at Butler and Healey (1998) rejected the idea of scale effect, judging that
y = 30% we locate the points of abscissa 6D30(am) and 9D30(am). it results from artefacts linked to incomplete well development and
The lter material, with C U(fm) 2.5, must fall between these low-K skins around well screens, but they did not study what produce
points, thus lter sand 1 is selected for the data in Fig. 7. This selec- a positive or negative skin. The skin phenomena and their effects on
tion assumes a homogeneous aquifer material. However, the aquifer the Ktest value are important for safety issues, and thus are more studied
material GSDC has a double curvature, which may indicate that the in geotechnique than in geosciences (Chapuis and Chenaf, 2010). This
aquifer material is stratied and formed of layers of two uniform paper has documented by how much poorly installed MWs underesti-
materials. The decomposition method which gives the GSDCs of mate the highest range of eld K values and thus, decrease the middle
the ner and coarser layers is explained in Chapuis et al. (2014). scale K values, which may contribute to create an articial scale effect.
In such a case, the lter sand 0 must be selected to stop the migra-
tion of solid particles in the ner material layers, otherwise the MW 7. Conclusion
screen will be severely clogged and the sampled groundwater may
remain turbid. Head losses in the MW screen and lter pack may strongly inuence
the permeability test results. Most often, this leads to underestimate the
sought K3 value of the tested material. Any screen-lter couple has an
inuence on the Ktest value obtained by a permeability test. The Ktest
100
aquifer material value is always equal to or smaller than the true K3 value of the tested
percentage (%) smaller than D

90 sand 2 material. The underestimate depends up on the hydraulic conductivities

sand 1
80 sand 0
K1 and K2 of the screen and lter material. Predictive equations give the
sand 00 lter K2 value. This paper gives the equivalent K1 of the screen. Several
70
6* and 9*D30 common screens were tested in a water tank. The test results for K1
60 were shown to depend up on the time that water rested in the tank.
50 Freshly poured water has the highest content of dissolved gas, and
yields the lowest K1 values. If water had been at rest for a few days, its
40
dissolved gas constant had decreased to come closer to the equilibrium
30 gas content as given by Henry's law, and the K1 value was increased.
20 Field condition differs from laboratory condition in a water tank. To de-
crease the air content (hardly visible micro bubbles) in the lter pack,
10
the lter material must be deaired in a water bucket before being slowly
0 poured with water into the drilling casing using a tremie. This precau-
0.01 0.1 1 10
tion may increase the MW measuring capacity by a factor of 10 or
grain size D (mm) more. The impact of high gas content in the lter pack was examined
elsewhere (Chapuis, 2009b).
Fig. 7. Selection of the lter sand according to the criteria of Table 2. If the aquifer material
is homogeneous, the lter sand 1 is appropriate. However, if the aquifer material is strat-
For studies to assess the eld K3 values, the lter material must be
ied, bimodal, the lter sand 0 must be selected, otherwise the MW screen will be severely selected according to known design rules. However, for environmental
clogged and the sampled groundwater will be turbid. studies, it is common to select a too ne lter pack. This supposedly