Wineries, distilleries and other grape processing industries annually generate large volumes of wastewater. This mainly originates from various washing operations during the crushing and pressing of grapes, as well as rinsing of fermentation tanks, barrels and other equipment or surfaces. In general, regulatory bodies dictate that wastewater should have a pH of 5.5 to 7.5 and the chemical oxygen demand (COD) should not exceed 75 mg/L before discharge to the environment (South African Water Act no. 36, 1998). However, winery wastewater typically has a pH of 3 to 4, with a COD of 800 to 12 800 mg/L. Furthermore, the COD can increase to 25 000 mg/L depending on the harvest load and processing activities. In general, wineries in urban areas channel the wastewater to local sewage treatment facilities that result in heavy penalties due to the low acid pH and high COD. In contrast, rural wineries often have very little or no treatment operations for wastewater, which is often irrigated onto grass fields.
Table 1. Chemical composition of synthetic wastewater and the COD contribution of the respective constituents in pure solutions at similar concentrations.
Several criteria should be considered when deciding on a treatment system for winery wastewater. These include an eco-friendly process that is flexible enough to handle various concentration loads and characteristics; low capital and operating costs; the system should require minimal personal attention and not occupy too much land, and the desired degree of degradation should be achieved without a need for dilution with water. A number of biological systems have been evaluated for winery wastewaters, such as anaerobic digesters and activated sludge reactors that are efficient in COD removal, but require long retention times. Furthermore, the capital and running costs of these treatment systems usually put them out of reach of smaller wineries.
Fig 1. Schematic representation of the set-up for the evaluation of the RBC.
The objective of this research project was to develop an eco-friendly biological system that would be flexible enough to handle varying concentration loads, keep both the capital and operation costs low, require minimal personal attention and not occupy too much land. Most importantly, it should lower the COD and increase the pH to acceptable levels without the addition of chemicals or dilution with water. We therefore evaluated a Rotating Biological Contactor (RBC) for the biological treatment of winery wastewater, since it is relatively easy to operate, has a short start-up, requires little maintenance and is effectively oxygenated with little sloughing of biomass. The system is based on a microbial biofilm that develops on the surface of discs mounted onto a horizontal shaft with at least 40% of the discs submerged in the wastewater. Rotation of the shaft results in alternating contact of the discs with wastewater and air that allows for the aerobic growth of the micro-organisms on the surface of the discs. Various operating parameters can be controlled, such as disc rotation speed, recirculation and hydraulic retention time.
Table 2. Wastewater composition of the distillery effluent during the evaluation period.
Chemical analysis of winery wastewater
Chemical analyses of winery wastewater were done during the 1999 harvest season to reflect the composition during the peak season when large volumes of high acid, high COD wastewater is discarded. The results indicate a large variation in COD (320 mg/L to 5670 mg/L), pH (3.9 to 4.9) and chemical composition. The wastewater generated by the destemming and pressing operations contained higher concentrations of glucose, fructose and malic acid that originate from the grape berries themselves. The considerable variation in the chemical composition of the wastewater can be ascribed to different varieties of grapes, harvest load, operation procedures, etc.
The development of synthetic wastewater (Table 1) provided a defined substrate with known composition that could be used for comparative analysis of different treatment variables. The COD of pure solutions of the various synthetic effluent constituents with concentrations corresponding to that of the synthetic effluent, indicated that the fermentable sugars (glucose and fructose) contributed almost half of the COD (Table 1). Ethanol and acetic acid also contributed to the COD, but to a lesser extent than the fermentable sugars.
Fig. 2. Recovery of biofilm after different physical or chemical shocks were introduced.
Biodegradation of winery wastewater by natural occurring micro-organisms
If a biofilm-based treatment system is to be considered for the treatment of winery wastewater, it is necessary to determine whether the naturally occurring micro-organisms are able to produce biofilms, while simultaneously reducing the COD of the wastewater. Microscopic analysis of glass slides suspended in wastewater streams showed thick biofilms on the surface of the slides that contained a large number of yeast and filamentous cells. Preliminary identification of the most abundant isolates indicated eight bacterial isolates and seven yeast isolates. When evaluated in synthetic wastewater, the bacterial isolates were ineffective with only two isolates showing a decrease in COD after 48 hours. The yeast isolates were more effective in reducing the COD of the wastewater, e.g. MEA5 reduced the COD by 95% after 24 hours under aerated conditions.
Under aerobic conditions, the mixed biofilm communities isolated from the microscope slides also reduced the COD of the synthetic wastewater (up to 62% within 72 hours), whereas very little effect was observed under anaerobic conditions. These results suggest that the naturally occurring micro-organisms were able to form a stable biofilm and also reduce the COD, i.e. utilise organic compounds that are susceptible to oxidation, within an RBC for the treatment of winery wastewater. Furthermore, the yeast isolates could play an important role in the degradation of organic compounds under aerobic conditions, such as those associated with an RBC.
Fig. 3. Influence of incubation temperature on removal of % COD by planktonic microbial community.
On-site scale evaluation of the RBC
A small-scale RBC was designed using a stainless steel trough of 14 260 cm3 (Fig. 1), with 16 polystyrene discs (20 cm in diameter, total surface of 1206 cm2) rotating at approximately 6 rpm. The RBC was evaluated on-site at a winery during the 2001 harvest: winery wastewater was pumped into the RBC after the excess grape skins and seeds had been removed and the system operated at hydraulic retention times varying between 0.35 and 1.4 hours.
Evaluation of the small-scale RBC on-site at a local winery in 2001 indicated an average decrease of 43% in the COD of the outflow (3475 ± 1715 mg/L), relative to the inflow (6090 ± 3382 mg/L) with a retention time of approx. 1 hour. Furthermore, the pH was increased by an average of 0.70 pH units (from 4.34 ± 0.29 to 5.04 ± 0.78). Sloughing of the biofilm occurred when the biofilm became too thick and weak degradation was observed when the retention time was too short (e.g. 0.6 hours), demonstrating the need to assess the operational conditions required for optimum microbial activity.
Since biofilm communities are dynamic and thus constantly changing to adapt to their environment, it is important to determine if there were dominant species in the biofilm. The most dominant yeast isolates in the microbial biofilms were Saccharomyces cerevisiae, Candida intermedia, Hanseniaspora uvarum and Pichia membranifaciens. All these species are naturally associated with grapes and/or water, and with the exception of H. uvarum, they are able to form either simple or elaborate pseudohyphae. Three time intervals were used to determine a shift in the population within the biofilm and notable changes were observed between the different time intervals for both bacteria and yeast. These results demonstrated the responsiveness of microbial communities to fluctuations in their environment.
Evalution of scaled-up RBC at winery and bottling plants
The evaluation of the scaled-up model with a capacity of 250 L containing 52 polyurethane discs rotating at 6 rpm was done on-site at Distell, Stellenbosch in 2002. The wastewater from the cellar that flows into an open holding tank had the stems and some of the skins removed. The wastewater was pumped through a filter to a second open holding tank (feed reservoir in Fig. 1). The wastewater in this tank had most of the solids removed with the exception of a small amount of fibres. For the first five days, an average reduction in COD of 43% was obtained, comparable to that found with the lab-scale model. Despite technical problems with a disfunctional water pump, an average decrease of 21% and pH increase of 0.95 units were obtained over a period of 3 months (average retention time of 1 hr). The system was subsequently evaluated at a bottling plant and resulted in an average reduction in COD of 34% and a pH increase of 0.83 units. The system was again plagued by variable flow rates that produced retention times of 1 - 4 hours.
A number of shocks typical of practices that may occur at a winery were introduced to determine their effect on the biofilm as well as the
ability of the biofilm to recover from the shock. Treatment with bleach (0.35% Sodium Hypochloride) had little effect (Fig. 2), but all of the other treatments showed a decrease in biofilm density. The exposure to high pH (pH 12), high temperature (60°C) and 2.5% SO2 had the most dramatic affects on the biofilm. However, for all the treatments, the biofilm was able to recover within 11 days.
Evalution of scaled-up RBC at distillery plant
In 2003, the prototype RBC was further evaluated for its efficacy on distillery wastewater, especially with regard to the physical parameters. Sampling of the RBC and monitoring of its parameters were conducted regularly (every 1-2 weeks) at the Distell distillery at Goudini and analysed for pH, conductivity, concentrations of COD, TD solids, nitrate, ammonium, orthophosphates, potassium, sodium, chlorides, magnesium and calcium (Table 2).
The volume of wastewater from the distillery varied from a maximum of 8785 m3 in March to a minimum of 2000 m3 in June. The temperature of the wastewater in the RBC was dependent on the effluent stream from the distillery, the season, wind speed and the hydraulic retention time. In summer daytimes, wastewater at the RBC inlet cooled down from ~30°C to ~20°C at the outlet and from ~15°C to ~10°C in winter daytime. The retention time was set at 3.2 hours, but only 2% COD removal was achieved; increasing the retention time to 5 hours did not improve the COD removal either. No effect on COD removal could be observed when the disc rotational speed was varied. This could be due to the already filamentous and thick grown yeast biofilm with a smooth and slimy surface not being sloughed off by the decreasing shear force. The biofilm thickness on some discs reached 0.8 cm after 3 months of operation, resulting in bridging the space between discs by biofilm and accumulation of sludge between discs.
The biofilm on the discs of the RBC consisted of a mixed population of fungi and bacteria. The dominant fungal group was found to be yeast; budding cells could be observed and pseudo-hyphal to hyphal growth was remarkable. The biofilm bottom layer consisted mainly of filamentous organisms with only few yeast cell conglomerates. In contrast, the biofilm top layer consisted mainly of hypha, single yeast cells, pseudohyphal and hyphal growing yeast. Only few bacteria could be seen with cells in the top layer being embedded in slime. No protozoa could be observed, which may be an indication of the inhibiting nature of the distillery effluent as protozoa are typically associated with biofilm reactors and may play an important role in carbon flow.
A number of physical parameters were evaluated for their effect on the efficacy of the RBC. Flasks with distillery effluent containing a planktonic microbial community were incubated at room temperature (approx. 22°C), 30°C and 37°C. The total COD decreased within 7 days from 25 000-27 000 mg/L to 9 000-19 000 mg/L (Fig. 3), representing a reduction of between 30% and 60%. Raising the temperature to 30°C or 37°C resulted in little improvement in COD reduction.
When small cubes of porous support media (sponge mass ca. 0.65 g) were added to support biofilm growth, the total COD of the flasks incubated at room temperature, 30°C and 37°C decreased within 7 days from 14 000-17 000 mg/L to 2 000-5 000 mg/L, representing a decomposition of total COD of between 60 and 80% (data not shown). The communities incubated at room temperature decreased the COD slightly faster in the first three days. The COD decrease was accompanied by a pH increase, showing a pH value higher than 8.0 at the end of every incubation period for room temperature, 30 and 37°C. However, the biodegradability did not improve when the wastewater pH was adjusted to 6.0 and 7.0. This could be due to an acid adapted wastewater microbial community.
The distillery wastewater had a concentration of nitrate and ammonium ranging between 35-132 mg/L and 68-378 mg/L, respectively. This resulted in a COD/N ratio of 40-122 when only nitrate and ammonia nitrogen is considered. This ratio is much higher than the ratio of 10-20 recommended in literature. Total nitrogen and phosphorus concentrations in the wastewater were adjusted accordingly to yield COD/N/P = 100/10/1. The degrading performance with supplementation of nitrogen and phosphorus was compared with microbial performance on wastewater without nutrients supplementation, but little improvement was found.
Chemical analyses of winery wastewater indicated that the high concentration of sugars contributed largely to COD, whereas the organic acids play a more prominent role in the acidity of the wastewater. The RBC was an effective biological system for lowering the COD of winery wastewaters, providing sufficient aeration for the biofilm, requiring low maintenance and being cost-effective. The efficiency of the RBC is perhaps not comparable to that obtained with anaerobic digesters or activated sludge reactors, but in view of the short retention time of 1-4 hours, the RBC could provide an effective system to address the peaks of high COD and acidity experienced during the harvest season. Given the seasonal fluctuations in wastewater discarded by wineries, the RBC could therefore be an effective primary treatment system to lower the COD to more acceptable levels for secondary treatment by constructed wetlands or other biological, chemical or physiochemical processes.
Noteworthy is the large number of yeast isolates within the biofilms that formed under the harsh conditions associated with winery wastewaters. Although bacterial biofilms are well known, very little information is available on the role and dynamics of yeasts within biofilms. This study indicated changes in the biofilm composition over time that confirmed that the yeast population in the biofilm was dynamic and able to adapt to their changing environment. Furthermore, we have shown that the biofilm is very stable and able to recover within 11 days after harsh shocks were introduced.
The prototype RBC proved to be effective for the treatment of fresh winery effluent as well as effluent from a bottling plant. Although thick biofilms developed on the RBC discs, the system seems to be less effective for distillery effluent. A number of parameters were evaluated, including temperature, pH, addition of growth supplements, with none providing conclusive answers for the apparent lack in efficiency of the system for distillery effluent. We therefore suspect that, although they are able to survive, the metabolic activity of the microbial organisms are inhibited by the high concentration of organic acids, polyphenols, etc., in distillery effluent. This was confirmed by the laboratory experiments where the mixed population obtained from the RBC was effective in lowering the COD of synthetic wastewater that does not contain the potential hazardous compounds in similar concentrations. Evaluation of the RBC is being continued at the Goudini distillery in search for a solution.
This work was supported by research grants from WINETECH, THRIP and the University of Stellenbosch. We thank the management and personnel of Distell and Rostberg for their assistance.