Reaction water, a high-strength COD (chemical oxygen demand) petrochemical effluent, is generated during the Fischer-Tropsch reaction in the SASOL Synthol process at SASOL SynFuels, Secunda, South Africa. Distillation of the reaction water to remove non- and oxygenated hydrocarbons yields approximately 25 - 30 ML/d of an organic (carboxylic) acid-enriched stream (average COD of 16 000 mg/L) containing primarily C2 – C5 organic acids, light oils, aldehydes, ketones, cresols and phenols. Together with the Oily sewer water (API) and Stripped Gas Liquor (SGL) process streams, this process effluent is currently treated in ten dedicated activated sludge basins. However, the successful operation of these activated sludge systems has proven to be difficult with low organic loading rates (3.5 kg COD/m3.d) low COD removal efficiencies (<80 %) and high specific air requirements (60 - 75 m3 air/kg CODrem). It is hypothesised that these operational difficulties can be attributed to organic shock loadings, variation in volumetric and hydraulic loadings, as well as variations in the composition of the various process streams being treated. Due to the fact that the Fischer-Tropsch (Synthol) reaction water constitutes 70 % of the COD load on the activated sludge systems, alternative processes to improve the treatment cost and efficiency of the Fischer-Tropsch acid stream are being investigated. Various studies evaluating the aerobic and anaerobic treatment of Fischer-Tropsch reaction water alone in suspended growth wastewater treatment systems have proven unsuccessful. High rate fixed-film processes or biofilm reactors, of which the fluidised-bed reactors are considered to he one of the most effective and promising processes for the treatment of high-strength industrial wastewaters, could he a suitable alternative. The primary aim of this study was to evaluate the suitability of biological fluidised-bed reactors (BFBRs) for the treatment of Fischer-Tropsch reaction water. During this study, the use of aerobic and anaerobic biological fluidised-bed reactors (BFBR), using sand and granular activated carbon (GAC) as support matrices, were evaluated for the treatment of a synthetic effluent analogous to the Fischer-Tropsch reaction water stream. After inoculation, the reactors were operated in batch mode for 10 days at a bed height expansion of 30% and a temperature of 30 ºC to facilitate biofilm formation on the various support matrices. This was followed by continuous operation of the reactors at hydraulic retention times (HRTs) of 2 days. While the COD of the influent and subsequent organic loading rate (OLR) was incrementally increased from 1 600 mg/L to a maximum of 20 000 mg/L and 18 000 mg/L for the aerobic and anaerobic reactors, respectively. Once the maximum influent COD concentration had been achieved the OLR was further increased by decreasing the HRTs of the aerobic and anaerobic reactors to 24h and 8h, and 36h, 24h and 19h, respectively. The dissolved O2 concentration in the main reactor columns of the aerobic reactors was constantly maintained at 0.50 mg/L. Chemical Oxygen Demand (COD) removal efficiencies in excess of 80 % at OLR of up to 30 kg COD/m3.d were achieved in the aerobic BFBRs using both sand and GAC as support matrices. Specific air requirements were calculated to be approximately 35 and 41 m3 air/kg CODrem for the BFBRs using sand and GAC as support matrices, respectively. The oxygen transfer efficiency was calculated to be approximately 5.4 %. At high OLR (> 15 kg COD/m3.d) significant problems were experienced with plugging and subsequent channelling in the BFBR using GAC as support matrix and the reactor had to be backwashed frequently in order to remove excess biomass. Despite these backwash procedures, COD removal efficiencies recovered to previous levels within 24 hours. In contrast, no significant problems were encountered with plug formation and channelling in the BFBR using sand as support matrix. In general the overall reactor performance and COD removal efficiency of the aerobic BFBR using sand as support matrix was more stable and consistent than the BFBR using GAC as support matrix. This BFBR was also more resilient to variations in operational conditions, such as the lowering of the hydraulic retention times and changes in the influent pH. Both aerobic reactors displayed high resilience and COD removal efficiencies in excess of 80 % were achieved during shock loadings. However, both reactors were highly sensitive to changes in pH and any decrease in pH below the pKa values of the volatile fatty acids in the influent (pKa of acetic acid = 4.76) resulted in significant reductions in COD removal efficiencies. Maintenance of reactor pH above 5.0 was thus an essential facet of reactor operation. It has been reported that the VFA/alkalinity ratio can be used to assess the stability of biological reactors. The VFA/alkalinity ratios of the aerobic BFBRs containing sand and GAC as support matrices were stable (VFNalkalinity ratios of < 0.3 - 0.4) until the OLR increased above 10 kg/m3.d. At OLRs higher than 10 kg/m3.d the VFA/alkalinity ratios in the BFBR using sand support matrix increased to 4, above the failure limit value of 0.3 - 0.4. In contrast the VFA/alkalinity ratios of the BFBR using GAC support matrix remained stable until an OLR of 15 kg/m3.d was obtained, where the VFA/alkalinity ratios then increased to > 3. Towards the end of the study when an OLR of approximately 25 kg/m3.d was obtained the VFA/alkalinity ratios of both the BFBRs using sand and GAC as support matrices increased to 9 and 6 respectively, indicating the decrease in reactor stability and acidification of the process. Total solid (TS) and volatile solid (VS) concentrations in the aerobic BFBRs were initially high and decreased over time. While the total suspended solids (TSS) and volatile suspended solids (VSS) concentrations were initially low and increased over time as the OLR was increased, this is thought to be as a result of decreased HRT leading to biomass washout. The anaerobic BFBR using sand as support matrix never stabilised and COD removal efficiency remained very low (< 30 %), possibly due to the high levels of shear forces. Further studies concerning the use of sand as support matrix were subsequently terminated. An average COD removal efficiency of approximately 60 % was achieved in the anaerobic BFBR using GAC as a support matrix at organic loading rates lower than 10 kg COD/m3.d. The removal efficiency gradually decreased to 50 % as organic loading rates were increased to 20 kg COD/m3.d. At OLRs of 20 kg COD/m3.d, the biogas and methane yields of the anaerobic BFBR using GAC as support matrix were determined to be approximately 0.38 m3 biogas/kg CODrem (0.3 m3 biogas/m3reactor vol.d), and 0.20 m3 CH4/kg CODrem (0.23 m3 CH4/m3reactor vol.d), respectively. This value is 57 % of the theoretical maximum methane yield attainable (3.5 m3 CH4/kg CODrem). The methane yield increased as the OLR increased, however, when the OLR reached 8 kg/m3.d the methane yield levelled off and remained constant at approximately 2 m3 CH4/m3reactor vol.d. Although the methane content of the biogas was initially very low (< 30 %), the methane content gradually increased to 60 % at OLRs of 20 kg COD/m3.d. The anaerobic BFBR using GAC as support matrix determined that as the OLR increased (>12 kg/m3.d), the VFA/alkalinity ratio increased to approximately 5, this is indicative of the decrease in stability and acidification of the process. The anaerobic BFBR using GAC as support matrix experienced no problems with plug formation and channelling. This is due to the lower biomass production by anaerobic microorganisms than in the aerobic reactors. The TS and VS concentrations were lower than the aerobic concentrations but followed the same trend of decreasing over time, while the TSS and VSS concentrations increased due to decreased HRTs. The anaerobic BFBR was sensitive to dramatic variations in organic loading rates, pH and COD removal efficiencies decreased significantly after any shock loadings. Compared to the activated sludge systems currently being used for the biological treatment of Fischer-Tropsch reaction water at SASOL SynFuels, Secunda, South Africa, a seven-fold increase in OLR and a 55 % reduction in the specific air requirement was achieved using the aerobic BFBRs. The methane produced could also be used as an alternative source of energy. It is, however, evident that the support matrix has a significant influence on reactor performance. Excellent results were achieved using sand and GAC as support matrices in the aerobic and anaerobic BFBRs, respectively. It is thus recommended that future research be conducted on the optimisation of the use of aerobic and anaerobic BFBRs using these support matrices. Based on the results obtained from this study, it can be concluded that both aerobic and anaerobic treatment of a synthetic effluent analogous to the Fischer-Tropsch reaction water as generated by SASOL in the Fischer-Tropsch Synthol process were successful and that the application of fluidised-bed reactors (attached growth systems) could serve as a feasible alternative technology when compared to the current activated sludge treatment systems (suspended growth) currently used. Keywords: aerobic treatment, anaerobic treatment, biological fluidised-bed reactors, petrochemical effluent, Fischer-Tropsch reaction water, industrial wastewater. / Thesis (M. Omgewingswetenskappe)--North-West University, Potchefstroom Campus, 2004.

Reaction water, a high-strength COD (chemical oxygen demand) petrochemical effluent, is generated during the Fischer-Tropsch reaction in the SASOL Synthol process at SASOL SynFuels, Secunda, South Africa. Distillation of the reaction water to remove non- and oxygenated hydrocarbons yields approximately 25 - 30 ML/d of an organic (carboxylic) acid-enriched stream (average COD of 16 000 mg/L) containing primarily C2 – C5 organic acids, light oils, aldehydes, ketones, cresols and phenols. Together with the Oily sewer water (API) and Stripped Gas Liquor (SGL) process streams, this process effluent is currently treated in ten dedicated activated sludge basins. However, the successful operation of these activated sludge systems has proven to be difficult with low organic loading rates (3.5 kg COD/m3.d) low COD removal efficiencies (<80 %) and high specific air requirements (60 - 75 m3 air/kg CODrem). It is hypothesised that these operational difficulties can be attributed to organic shock loadings, variation in volumetric and hydraulic loadings, as well as variations in the composition of the various process streams being treated. Due to the fact that the Fischer-Tropsch (Synthol) reaction water constitutes 70 % of the COD load on the activated sludge systems, alternative processes to improve the treatment cost and efficiency of the Fischer-Tropsch acid stream are being investigated. Various studies evaluating the aerobic and anaerobic treatment of Fischer-Tropsch reaction water alone in suspended growth wastewater treatment systems have proven unsuccessful. High rate fixed-film processes or biofilm reactors, of which the fluidised-bed reactors are considered to he one of the most effective and promising processes for the treatment of high-strength industrial wastewaters, could he a suitable alternative. The primary aim of this study was to evaluate the suitability of biological fluidised-bed reactors (BFBRs) for the treatment of Fischer-Tropsch reaction water. During this study, the use of aerobic and anaerobic biological fluidised-bed reactors (BFBR), using sand and granular activated carbon (GAC) as support matrices, were evaluated for the treatment of a synthetic effluent analogous to the Fischer-Tropsch reaction water stream. After inoculation, the reactors were operated in batch mode for 10 days at a bed height expansion of 30% and a temperature of 30 ºC to facilitate biofilm formation on the various support matrices. This was followed by continuous operation of the reactors at hydraulic retention times (HRTs) of 2 days. While the COD of the influent and subsequent organic loading rate (OLR) was incrementally increased from 1 600 mg/L to a maximum of 20 000 mg/L and 18 000 mg/L for the aerobic and anaerobic reactors, respectively. Once the maximum influent COD concentration had been achieved the OLR was further increased by decreasing the HRTs of the aerobic and anaerobic reactors to 24h and 8h, and 36h, 24h and 19h, respectively. The dissolved O2 concentration in the main reactor columns of the aerobic reactors was constantly maintained at 0.50 mg/L. Chemical Oxygen Demand (COD) removal efficiencies in excess of 80 % at OLR of up to 30 kg COD/m3.d were achieved in the aerobic BFBRs using both sand and GAC as support matrices. Specific air requirements were calculated to be approximately 35 and 41 m3 air/kg CODrem for the BFBRs using sand and GAC as support matrices, respectively. The oxygen transfer efficiency was calculated to be approximately 5.4 %. At high OLR (> 15 kg COD/m3.d) significant problems were experienced with plugging and subsequent channelling in the BFBR using GAC as support matrix and the reactor had to be backwashed frequently in order to remove excess biomass. Despite these backwash procedures, COD removal efficiencies recovered to previous levels within 24 hours. In contrast, no significant problems were encountered with plug formation and channelling in the BFBR using sand as support matrix. In general the overall reactor performance and COD removal efficiency of the aerobic BFBR using sand as support matrix was more stable and consistent than the BFBR using GAC as support matrix. This BFBR was also more resilient to variations in operational conditions, such as the lowering of the hydraulic retention times and changes in the influent pH. Both aerobic reactors displayed high resilience and COD removal efficiencies in excess of 80 % were achieved during shock loadings. However, both reactors were highly sensitive to changes in pH and any decrease in pH below the pKa values of the volatile fatty acids in the influent (pKa of acetic acid = 4.76) resulted in significant reductions in COD removal efficiencies. Maintenance of reactor pH above 5.0 was thus an essential facet of reactor operation. It has been reported that the VFA/alkalinity ratio can be used to assess the stability of biological reactors. The VFA/alkalinity ratios of the aerobic BFBRs containing sand and GAC as support matrices were stable (VFNalkalinity ratios of < 0.3 - 0.4) until the OLR increased above 10 kg/m3.d. At OLRs higher than 10 kg/m3.d the VFA/alkalinity ratios in the BFBR using sand support matrix increased to 4, above the failure limit value of 0.3 - 0.4. In contrast the VFA/alkalinity ratios of the BFBR using GAC support matrix remained stable until an OLR of 15 kg/m3.d was obtained, where the VFA/alkalinity ratios then increased to > 3. Towards the end of the study when an OLR of approximately 25 kg/m3.d was obtained the VFA/alkalinity ratios of both the BFBRs using sand and GAC as support matrices increased to 9 and 6 respectively, indicating the decrease in reactor stability and acidification of the process. Total solid (TS) and volatile solid (VS) concentrations in the aerobic BFBRs were initially high and decreased over time. While the total suspended solids (TSS) and volatile suspended solids (VSS) concentrations were initially low and increased over time as the OLR was increased, this is thought to be as a result of decreased HRT leading to biomass washout. The anaerobic BFBR using sand as support matrix never stabilised and COD removal efficiency remained very low (< 30 %), possibly due to the high levels of shear forces. Further studies concerning the use of sand as support matrix were subsequently terminated. An average COD removal efficiency of approximately 60 % was achieved in the anaerobic BFBR using GAC as a support matrix at organic loading rates lower than 10 kg COD/m3.d. The removal efficiency gradually decreased to 50 % as organic loading rates were increased to 20 kg COD/m3.d. At OLRs of 20 kg COD/m3.d, the biogas and methane yields of the anaerobic BFBR using GAC as support matrix were determined to be approximately 0.38 m3 biogas/kg CODrem (0.3 m3 biogas/m3reactor vol.d), and 0.20 m3 CH4/kg CODrem (0.23 m3 CH4/m3reactor vol.d), respectively. This value is 57 % of the theoretical maximum methane yield attainable (3.5 m3 CH4/kg CODrem). The methane yield increased as the OLR increased, however, when the OLR reached 8 kg/m3.d the methane yield levelled off and remained constant at approximately 2 m3 CH4/m3reactor vol.d. Although the methane content of the biogas was initially very low (< 30 %), the methane content gradually increased to 60 % at OLRs of 20 kg COD/m3.d. The anaerobic BFBR using GAC as support matrix determined that as the OLR increased (>12 kg/m3.d), the VFA/alkalinity ratio increased to approximately 5, this is indicative of the decrease in stability and acidification of the process. The anaerobic BFBR using GAC as support matrix experienced no problems with plug formation and channelling. This is due to the lower biomass production by anaerobic microorganisms than in the aerobic reactors. The TS and VS concentrations were lower than the aerobic concentrations but followed the same trend of decreasing over time, while the TSS and VSS concentrations increased due to decreased HRTs. The anaerobic BFBR was sensitive to dramatic variations in organic loading rates, pH and COD removal efficiencies decreased significantly after any shock loadings. Compared to the activated sludge systems currently being used for the biological treatment of Fischer-Tropsch reaction water at SASOL SynFuels, Secunda, South Africa, a seven-fold increase in OLR and a 55 % reduction in the specific air requirement was achieved using the aerobic BFBRs. The methane produced could also be used as an alternative source of energy. It is, however, evident that the support matrix has a significant influence on reactor performance. Excellent results were achieved using sand and GAC as support matrices in the aerobic and anaerobic BFBRs, respectively. It is thus recommended that future research be conducted on the optimisation of the use of aerobic and anaerobic BFBRs using these support matrices. Based on the results obtained from this study, it can be concluded that both aerobic and anaerobic treatment of a synthetic effluent analogous to the Fischer-Tropsch reaction water as generated by SASOL in the Fischer-Tropsch Synthol process were successful and that the application of fluidised-bed reactors (attached growth systems) could serve as a feasible alternative technology when compared to the current activated sludge treatment systems (suspended growth) currently used. Keywords: aerobic treatment, anaerobic treatment, biological fluidised-bed reactors, petrochemical effluent, Fischer-Tropsch reaction water, industrial wastewater. / Thesis (M. Omgewingswetenskappe)--North-West University, Potchefstroom Campus, 2004.