CONCLUSIONES

5.2.2.1 Biological Filtration: Activated Carbon

Two activated carbon filtration systems were set up directly next to rainwater tanks B and tank C (Figure 5.1). The filtering systems consisted of a smaller container (20 L) containing holes at the bottom, which fitted into a larger container (25 L) with a tap connected for effluent collection (Figure 5.2). The smaller container was filled with a bottom layer of 5 cm marine pebbles and then a top layer of approximately 17 cm of activated carbon (Aquasorb Udectrading Pty. Ltd) which represented the filtering material. A garden hose was then connected to the tap of a DRWH tank on the one end, and then to a shower head (20 cm diameter) on the other end. A shower head was fitted to the lid of the smaller container to allow the harvested rainwater to trickle through the shower head and over the activated carbon filtering medium. A biofilm was allowed to establish in the slow activated carbon filtration system by filling the system with rainwater and then allowing the system to remain closed for one week. The system was then drained and in a continuous flow arrangement, the rainwater was allowed to flow through the system for the duration of sampling and subsequent filtrate analyses.

Figure 5.2. (A) Schematic diagram of the smaller container containing marine pebbles and activated carbon which fitted into a larger container allowing for a slow activated carbon filtration system. (B) Photograph of the slow activated carbon system.

Activated carbon Marine pebbles

Shower head

139 To determine the degree of chemical and bacterial pollution, samples were collected for five consecutive days. On each sampling day a 1 L water sample was collected directly from rainwater tanks B and C (before sample), respectively, and a 1 L rainwater sample was collected from the filtrate of the activated carbon filtration systems (an after sample) connected to tanks B and C, respectively.

5.2.2.2 Biological Filtration: Slow Sand Filtration

Two slow sand filtration systems were set up directly next to rainwater tanks B and tank C (Figure 5.1). The same system utilised for the activated carbon filtration (section 5.2.1 and Figure 5.2) was constructed with the exception that the slow sand filtration system consisted of a 5 cm layer of marine pebbles, placed in the bottom of the smaller container and approximately 22 cm of 0.61 mm silica sand (Cape Silica Suppliers CC, Cape Town, South Africa) placed on top of the pebbles. A biofilm was also allowed to establish for two weeks in a closed system before sampling took place by allowing water from the respective DRWH tanks to flow through the system. After the biofilm had formed, samples were collected every second day, for a total of six sampling events over a two week period. The water was allowed to continuously flow through the system between samplings. On each sampling day a 1 L water sample was collected directly from rainwater tanks B and C (before sample), respectively and a 1 L rainwater sample was collected from the filtrate of the slow sand filtration systems (an after sample) connected to tanks B and C, respectively.

5.2.2.3 Polyvinyl (alcohol) (PVA) nanofibre membrane filtration system

Polyvinyl (alcohol) nanofibres were produced by a process of needleless electrospinning utilising a Nanospider 200 Lab (Elmarco, s.r.o., Czech Republic). The substrate material onto which the nanofibres were deposited was a Tyvek material (Marshall Hinds, Johannesburg, South Africa) which was wound onto a core. A PVA polymer solution was made up by dissolving a PVA powder (Nippongohsei, Japan) in distilled water at 80˚C. The PVA polymer solution was modified by adding a cross-linker, acid and CuCl2 (proprietary information). The PVA polymer solution was

then poured into a polypropylene tub containing a stainless steel spinning electrode which was then partially submerged in the polymer solution. In order to create an electric field, a high voltage was connected to the spinning electrode with the collecting wire electrode grounded to create a potential difference. The spinning conditions were as follows; spinning distance was 100 mm, rotation speed of electrode was 3.2 rpm, high voltage was 80 kV, relative humidity was below 40% and speed of fabric was 0.1 m/min. Once the nanofibres were spun onto the Tyvek material, the newly synthesised membrane was cross-linked at 140°C for 15 min. A section of the membrane was analysed using scanning electron microscopy (SEM) at the Central Analytical Facility (CAF),

140 Stellenbosch University. Microscopy was performed using a LEO 1450VP SEM (Zeiss, Germany). The final product, a PVA nanofibre membrane was then used in a column flow through system. The column system that was directly attached to tanks B and C is indicated in Figure 5.3. A schematic diagram of the PVA nanofibre membrane column is represented in Figure 5.3A, where unfiltered rainwater (red arrows) was allowed to flow through the PVA nanofibre membrane to the centre of the column and then filtered rainwater (blue arrows) was collected. The column systems were designed as follows, an inner cylinder containing holes (Figure 5.3B) was fitted inside a larger column (Figure 5.3C, D). A PVA nanofibre membrane was then wrapped around the inner cylinder twice which was then covered with a red netting (Figure 5.3E). This PVA nanofibre membrane system was assessed for bacterial removal efficiency only (Figure 5.3F).

Figure 5.3. A) A schematic diagram of the PVA nanofibre membrane column, where unfiltered rainwater (red arrows) was allowed to flow through the PVA nanofibre membrane to the centre of the column and then filtered rainwater (blue arrows) was collected. B – F) A column system containing a PVA nanofibre membrane. G) Activated carbon was then layered around the PVA nanofibre membrane.

B . C . D . E. F . G . A . Outer column Inner cylinder PVA nanofibre membrane

141 To determine the bacterial and chemical contamination removal efficiency, activated carbon (Aquasorb udectrading Pty. Ltd) was then layered around the PVA nanofibre membrane in order to exclude larger contaminants before passing through the PVA nanofibre membrane (Figure 5.3G). An initial 1 L rainwater sample was collected directly from tanks B and C. Five 1 L samples were then individually collected after the water passed through the PVA nanofibre membrane/activated carbon filtration system. The PVA nanofibre membrane and activated carbon was then replaced and another sampling was repeated with the system connected to each tank.

5.2.3 Chemical Analysis

Filtered samples collected from the slow sand and activated carbon biological filtration systems (before and after biofilm formation) and the PVA activated carbon column filtration system were analysed for the following chemical parameters. For the determination of the metal concentrations, Falcon™ 50 mL high-clarity polypropylene tubes containing polyethylene caps, were pre-treated with 1% nitric acid before sampling. The concentrations of metals such as aluminium (Al), vanadium (V), chromium (Cr), manganese (Mn), iron (Fe), copper (Cu) and zinc (Zn), amongst others, were then determined using inductively coupled plasma atomic emission spectrometry (ICP- AES) according to Saleh et al. (2000) and nitric acid digestion. All samples were analysed for the presence of metals at the Central Analytical Facility (CAF), Stellenbosch University. The filtered rainwater samples were also sent to the Centre for Scientific and Industrial Research (CSIR) Stellenbosch for anion analyses. The anions detected along with the corresponding detection method are summarised in Table 5.1.

Table 5.1. Methods used in the detection of anions performed by the CSIR, Stellenbosch

Anion Method

Nitrate and Nitrite SALM 7.0 Automated Colorimetry Soluble phosphate SALM 9.0 Automated Colorimetry Sulphate MALS 6.5 ICP OES Detection Chloride SALM 1.0 Automated Colorimetry Fluoride SALM 11 Potentiometric measurement