Colloid clogging of saturated porous media under varying ionic strength and roughness during managed aquifer recharge

Column experiments were conducted to examine the clogging effects of colloids under controlled conditions of solution ionic strength (IS) and porous media roughness. The results showed that colloids in recharge water play an important role in the clogging process of saturated porous media, such that even a small amount of colloid may cause a large reduction in the permeability of the porous medium. Clogging at the pore throat was inferred to be the main reason for the severe permeability reduction of porous media. The characteristics of colloid clogging were clearly in ﬂ uenced by both IS and medium roughness. Recharge water with a higher IS facilitated greater attachment of colloids to the surface of the saturated porous medium, which lead to super ﬁ cial clogging, while collectors with a rough surface resulted in greater clogging than collectors with a smooth surface.


INTRODUCTION
An important issue for any managed aquifer recharge (MAR) system is the decline of permeability, typically called clogging (Dillon et al. ). Clogging is an inevitable problem and recognized as perhaps the most significant challenge in MAR operations; it is attributed to physical, chemical and biological processes. According to survey data, physical clogging is the most common, affecting 70% of MAR cases (Dillon et al. ).
Suspension clogging, gas clogging and compaction clogging are all types of physical clogging, with clogging caused by suspended particles being the most common (Xiao & Reddi ). with clogging that arises from a colloidal suspension in the MAR process having been much less studied. Mays & Hunt () concluded that solution pH, ionic strength (IS) and exchangeable ions together determine colloid stability, and hence the morphology of the deposited colloids and the resulting permeability of the formation. However, there is little experimental data on the effects of colloid effects on clogging in the MAR process. One exception was Roth et al. (), who showed that a reduction in permeability was strongly associated with the fractal dimension of the deposited colloid's morphology. However, the transport, release and deposition of colloids in porous media are very complicated, as the colloids are simultaneously under the influence of various factors, including the glass beads ranged from 224 to 250 μm in diameter.

Column experiments
The experimental set-up included a column, a peristaltic pump (BT100-1F, Longer Company, China), a fraction collector (CBS-A, Huxi Company, China), two pressure transmitters (A-10, WIKA Company, Germany), and a data acquisition system (Figure 1). The column was made of plexiglass and measured 16 cm in length with a 2-cm internal diameter. Gauze mesh placed inside the end caps was used to support the porous media and also helped spread the input solution laterally throughout the column. The column was wet packed with the river sand or glass beads.
The peristaltic pump was used to move solutions from the supply bottle into the column at a constant Darcy velocity (v) of 5 m/d. Flow direction was vertically upward. Note that the potential effect of gravity on the kinetics of colloid transport and deposition was not considered in the present study. Colloid transport was assessed by injecting of a pulse of the colloid suspension into the column for 460 min (i.e., phase 1), followed by the injection of a colloid-free solution of the same chemistry for 100 min (i.e., phase 2). A fully-automated sample collector continuously gathered the effluent samples at specified time intervals (approximately every 11.5 min) and then measured them with an ultraviolet spectrophotometer (PerkinElmer, USA) at a wavelength of 486 nm. A linear correlation was established between the absorbance reading  Table 1.
At the end of the experiment, the sand or beads were excavated at 1 cm intervals from the column and placed in a centrifuge tube that contained the same electrolyte solution as used in phase 1. The tube was shaken vigorously with a vortex mixer for 3 min to separate the retained colloids from the collector surface. Next, the colloid concentration in the electrolyte solution was measured by an ultraviolet spectrophotometer.

Hydraulic conductivity
The hydraulic conductivity (K ) of the porous medium in each column was calculated by Darcy's law (Mbonimpa et al. ).       (Table 2).
Based on the data from Table 2, the retained colloid only occupied a very small proportion of the pore space, so a change of porosity can be neglected. Obviously, the relationship between hydraulic conductivity and porosity could not be interpreted by using the Kozeny-Carman equation (Equation (2)), which is the most common approach for estimating the permeability of the porous media (Roth et al. ). So, the decrease of hydraulic conductivity was not caused by clogging of the whole pore space, but only at the pore throat ( Figure 8). ( 2) where K is hydraulic conductivity (m/s), n (À) is the porosity of porous media and d 50 is median grain size (mm). V Re /V P (À) 2.31 × 10 À5 1.54 × 10 À4 1.55 × 10 À4 8.79 × 10 À6 K/K 0 (À) 65% 55% 30% 96%