Rensning af grundvand med aktivt kul for BAM og atrazin

Rensning af grundvand med aktivt kul for BAM og atrazin

Summary

The Danish drinking water resource is increasingly polluted by pesticides, especially BAM and atrazine. This problem may lead to a need for treatment technologies to remove pesticides from drinking water. Activated carbon filtration is an obvious technology that is used worldwide, but primarily for treatment of surface water, whereas Danish drinking water supply is based on groundwater.

The purpose of the project has therefore been to investigate the opportunities for using activated carbon (AC) filtration to remove the pesticides BAM (2,6-Dichlorobenzamide) and atrazine from Danish groundwater. The main objective of the study was to determine the AC capacity, which is an expression of the amount of pesticide which can be bound to a certain amount of AC and maintain the outlet concentration below a specified value. The calculation of this theoretical value is different for different physical systems, and therefore the capacity depends on the technical system applied.

The AC capacity was investigated for three different types of activated carbon (Chemviron Filtrasorb F400, Norit ROW 0.8, and Lurgi, Hydraffin CC 8 x 30); each based on a different raw material (bitumen, peat and coconut). The investigations were based on groundwater from a limestone aquifer (Hvidovre Waterworks) and from a glacial sand aquifer (Kisserup Waterworks), both with relatively low amounts of natural organic matter (NVOC = 1.1-1.7 mg/l). Furthermore, the AC capacity for these water types was compared to the AC capacity in organic free water (MilliQ-water).

The types of AC were characterized with respect to specific surface area, micropore area, micropore volume, and pore volume determined by the BET method, and the grain size distribution determined by sieving analysis.

Experimentally three different approaches were used: Column experiments at bench-scale (near-realistic scale (one meter scale)), small-scale column test (centimetre scale) and adsorption isotherm test (batch test with suspended AC). Experimental set-ups for small-scale column tests and adsorption isotherm tests are not clearly defined in literature, and considerable effort has therefore been made to identify how such tests should be carried out experimentally.

In order to simulate the AC capacity for BAM and atrazine at a realistic scale, a bench-scale experiment with a realistic hydraulic retention time and a realistic filtration velocity was set up at Hvidovre Waterworks. The bench-scale experiment showed a significant headloss in the upper 2 cm of the adsorber during operation, due to iron precipitation. This problem was easily solved by manually removing the upper layer of activated carbon. In practise it may be an advantage to design the adsorber with easy access to the carbon surface, allowing removal of the upper carbon layer. Manual removal of iron from the carbon surface can probably eliminate expenses associated with installation of backwashing facilities. Backwashing of activated carbon adsorbers can result in migration of saturated carbon particles to the fresh activated carbon in the bottom of the adsorber and thereby cause early breakthrough of pesticide in the effluent water.

With a column length of 64 cm and a maximum effluent concentration of 0.1 µg/L, the bench-scale experiment at Hvidovre Waterworks resulted in capacities of 48-57 µg BAM/g AC and 43-66 µg atrazine/g AC. The influent concentrations of BAM and atrazine were 0.27 µg/L and 0.21 µg/L respectively, and the influent concentration of NVOC was 1.1 mg/L. The maximum capacity of BAM was observed in the activated carbon made from coconuts (Lurgi, Hydraffin CC 8 x 30), whereas the maximum capacity for atrazine was observed in the activated carbon made from bitumen (Chemviron Filtrasorb F400). At a maximum effluent concentration of 0.01 µg/L the capacity was 20-34 µg BAM/g AC and 15-27 µg atrazine/g AC, and the maximum capacity of both compounds were observed in the carbon type Chemviron Filtrasorb F400. However, in general the calculated capacities of the three investigated carbon types were surprisingly similar, taking the different raw materials and the different pore structures into account.
Furthermore, the adsorption capacity for BAM was of the same order of magnitude as the adsorption capacity for atrazine, which is surprising due to the much more polar (hydrophilic) structure of BAM.

In the adsorption isotherm tests the capacities were 170-250 µg BAM/g AC and 110-180 µg atrazine/g AC with water from Hvidovre Waterworks, while water from Kisserup Waterworks resulted in capacities of 160-175 µg BAM/g AC and 270-560 µg atrazine/g AC. The activated carbon capacity was far higher with MilliQ-water; 2700-3600 µg BAM/g AC and 3900-6300 µg atrazine/g AC. A natural organic matter concentration of 1-2 mg NVOC/L thereby reduced the capacity for BAM and atrazine significantly due to adsorption site competition. It should be noticed that adsorption capacities reported by AC manufactures are often from adsorption isotherm tests with MilliQ-water.

In the adsorption isotherm experiments, equilibrium was not reached until 2-3 weeks. Usually equilibrium is expected after a few hours or days. This means, however, that diffusion of pesticides into the pores of activated carbon is a slow process.

The activated carbon capacities determined by adsorption isotherm tests were significantly higher than the capacities determined by realistic bench-scale experiments. Specifically BAM capacities were 3-4 times higher, while atrazine capacities were 2-3 times higher.

The small-scale column experiments were able to predict the activated carbon capacity of the first column (with a length of 32 cm) of the bench-scale experiments. However, predicting the AC capacity of two bench-scale columns (with a length of 64 cm) resulted in a capacity two times higher than actually observed in the realistic bench-scale experiment. This difference is probably caused by preloading of the bench-scale columns with natural organic matter, which due to the short operation time is not observed in the small-scale column experiments. Therefore, at present time, the small column test is not applicable for prediction of full-scale adsorber performance.
Furthermore, use of small column experiments are not possible without verifying data with a bench-scale experiments, because it is not yet possible to predict which scaling theory to use.

Modelling of break-through curves with the program AQUASIM of tracer experiments carried out in the bench-scale columns and of pesticide adsorption in bench-scale columns and small-scale columns has shown a potential for model-based scaling. It is possible to model sorption of BAM in both bench-scale columns and small-scale columns with isotherm data for BAM (K_d=1.15 m ³/g AC). By modelling of the tracer experiments from the bench-scale columns, it is possible to estimate mass transfer parameters necessary for modelling the breakthrough profiles of BAM.

In general the project shows that AC adsorption is a suitable method for removal of BAM from groundwater. Furthermore, contracy to expectations, the capacities for BAM are in the same order of magnitude as capacities for atrazine. It was expected that the adsorption of BAM onto activated carbon would be considerably lower compared to that of atrazine, because of the more polar structure of BAM (hydrophilic). The capacity for BAM and atrazine did not vary significantly with the different types of activated carbon or groundwater investigated in this study.