Long term operation of a thermophilic anaerobic reactor : 2 process stability and efficiency at decreasing sludge retention time 3 4

The aim of this study was to evaluate the performance of thermophilic sludge digestion at decreasing sludge retention time (SRT) and increasing organic loading rate (OLR), in terms of methane production, effluent stabilisation, hygienisation and dewaterability. Focus was put on determining indicators to help prevent process failure. To this end, a lab-scale reactor was operated for nearly 2 years at 55 degrees Celsius. Methane production rate was increased (from 0.2 to 0.4-0.6 m(3)((ch)(4)) m(-3)(reactor) d(-1)) by decreasing the SRT from 30 to 15-10 days, while increasing the OLR from 0.5 to 2.5-3.5 kg VS m(-3)(reactor) d(-1). Sludge dewaterability was worsened at SRT below 15 days; while pathogen destruction was always successful. The following concentrations might be used to prevent process failure: VFA C2-C5 (3.7 g CODL(-1)), acetate (0.6 g L(-1)), acetate/propionate (0.5), intermediate alkalinity (1.8 g CaCO(3) L(-1)), intermediate/partial alkalinity (0.9), intermediate/total alkalinity (0.5), CH(4) in biogas (55%).


Introduction
In anaerobic digesters, biogas production depends on the amount of organic 52 matter biodegraded by anaerobic microorganisms. Thus, it depends on the composition 53 of the substrate, and presence and equilibrium between anaerobic consortia in the 54 reactor. Design and operation parameters of the process include sludge retention time changing from low-solids to high-solids sludge. Each subsequent SRT decrease was 148 made once the digester had reached stable operation (i.e. fairly constant performance in 149 terms of biogas production, VFA concentration and pH in the reactor) as proposed by 150 other authors (Angelidaki and Ahring, 1994;El-Mashad et al., 2004). This digester was 151 operated for 21 months, under the conditions summarised in Table 1. 152 153 154 155 The solids content of sludge was determined according to the Standard Methods 156 procedure 2540G (APHA, 1999). TS and VS were determined directly from sludge 157 samples, whereas total dissolved solids (TDS) and volatile dissolved solids (VDS) were 158 determined from the supernatant of samples centrifuged at 7000 rpm. Supernatants 159 underwent filtration through 1.2 µm nominal pore size glass fibber filters (Albet 160 FVC047, Spain). The particulate fractions, total suspended solids (TSS) and volatile 161 suspended solids (VSS) were subsequently deduced. pH, alkalinity and VFA (acetic, 162 propionic, iso-butyric, n-butyric, iso-valeric and n-valeric acids) were also analysed 163 from the filtrate supernatant. Samples for VFA analysis were further filtered through a 164 0.45 µm nylon syringe filter. VFA and biogas composition were determined by gas 165 chromatography (Perkin-Elmer AutoSystem XL Gas Chromatograph), as described in 166 Ferrer et al. (2008). 167 Total, partial and intermediate alkalinities were determined as proposed by 168 Ripley et al. (1986). The method consists of a two step titration: a first one down to pH 169 5.75, which is due to HCO 3 species and is known as partial alkalinity (PA); and a 170 second one down to pH 4.3, which corresponds to the total alkalinity (TA). The 171 intermediate alkalinity (IA), which is related to VFA concentration, is then estimated as 172 the difference between TA and PA. It can be used as an indirect measurement of VFA 173 P r e -p r i n t concentration. The alkalinity ratio (AR), defined as the ratio between intermediate and 174 total alkalinity (IA/TA), or between intermediate and partial alkalinity (IA/PA); may 175 also be a useful indicator of the concentration of VFA in the sample. 176 Sludge dewaterability was determined using the Capillary Suction Time (CST) 177 test, according to the Standard Methods procedure 2710G (APHA, 1999 et al., 2002;Gavala et al., 2003). 207 The best results were obtained at the lowest SRTs (15 and 10 days), with OLR 208 of 1-1.6 and 1.5-2 kg VS m -3 reactor d -1 , respectively. In particular, the highest biogas and 209 methane production rates (up to 0.56 and 0.36 m 3 m -3 d -1 , respectively) correspond to 10 210 days SRT (Table 2, period VI).

211
After switching from low-solids to high-solids sludge (40-60 g TS L -1 ; 30-35 g 212 VS L -1 ), OLR as high as 3-4 kg VS m -3 reactor d -1 were maintained ( Figure 1). Biogas 213 production rate was almost doubled from 0.5 to 1 m 3 biogas m -3 reactor d -1 at 10 days SRT 214 feeding low-and high-solids sludge, respectively (Table 2, periods VI and VII). 215 However, higher effluent VFA (> 4 g COD L -1 ) were detected (Figure 3(a)). 216 The SRT was gradually decreased to 6 days with OLR ranging from 4.5 to 6.5 217 kg VS m -3 reactor d -1 (Figure 1), which are amongst the highest OLR and lowest SRT 218 reported for single stage sludge digestion (Buhr and Andrews, 1977;Speece, 1988;De 219 la Rubia et al., 2006a;De la Rubia et al., 2006b). Initially, biogas production reached its 220 highest rates (~ 1.5 m 3 biogas m -3 reactor d -1 ), with 58-69 % CH 4 in biogas. However, these COD L -1 ), as shown in Figure 3(a). Methane content in biogas drop below 50 % ( Figure   223 2) and VS removal to 13 %. To avoid digester failure, the SRT was set back to 10 days. concentration were less pronounced, the trend followed by the acetate to propionate 250 ratio (A/P ratio) was similar to that of acetate, as can be seen from Figure 3(b).

251
As well as individual and total VFA, some authors have proposed acetate 252 concentration and A/P ratio as valuable indicators to predict process failure (Marchaim 253 and Krause, 1993;Pind et al., 2002). For manure, an acetic acid concentration of 0.8 g 254 L -1 and an A/P ratio of 1.4 have been proposed as limit values (Hill et al., 1987;cited in 255 Marchaim and Krause, 1993). To our knowledge, such limit values for thermophilic 256 sewage sludge digestion have not yet been proposed. In the present study, acetate 257 concentration was usually below 0.6 g L -1 (Table 2, all periods) and only in cases of 258 organic overloading or temperature fluctuations (due to operating problems) did this 259 value rise above 0.6 g L -1 and up to 2 g L -1 . Furthermore, concentrations above 1 g L -1 260 were only reached when the SRT was reduced to 6 days, with OLR greater than 5 kg 261 VS m -3 reactor d -1 , as shown in Figure 3. Therefore, a limit concentration of 0.6 g L -1 of 262 acetic acid would seem more appropriate to predict digester failure in the case of 263 thermophilic sludge digestion. Similarly, during stability periods the A/P ratio was 264 below 0.5 (

Alkalinity
According to Ripley et al. (1986), the total alkalinity (TA) of a sample is a result 274 of HCO 3 species, which is known as partial alkalinity (PA); and VFA, which is known  all other indicator parameters were above the limit values proposed, the pH was still 8. 311 The reason for this is that the alkalinity of the system was also the highest; hence the 312 buffer capacity of the system prevented from pH drop resulting from VFA 313 accumulation. In sewage sludge digesters, sufficient alkalinity is generally found (3-5 g 314 CaCO 3 L -1 ) to prevent the pH from falling below the limit for methanogenesis inhibition 315 (Metcalf and Eddy, 2003). Studies with high-solids sludge (4-10 % TS) have shown that 316 the optimum pH range for high rate digestion is 6.6-7.8, while the acceptable pH range 317 is 6.1-8.3; meaning that below 6.1 the process may fail due to an excessively low 318 methanogenesis rate compared to acidogenesis rate, while above 8.3 the process might 319 be inhibited by free ammonia (Lay et al., 1997). Ammonia inhibition is favoured by high process temperature (Angelidaki and Ahring, 1994) and is pointed out as a major 321 cause for low biogas production treating pig slurries (Bonmatí and Flotats, 2003 For the purposes of this study, the SRT was gradually reduced from 30 to 6 days. 335 However, because the feeding sludge was collected weekly from the WWTP, seasonal 336 variations and operational changes affected its composition and organic content. 337 Furthermore, low-solids and high-solids sludge were used. Whilst operating under a 338 fixed SRT, the OLR was affected by the sludge organic content; thus it was also 339 necessary to assess the effect of OLR on the thermophilic sludge digestion. 340 Figure 5 shows methane production rate, effluent VFA and effluent VS as a 341 function of the OLR. In general, high correlations were obtained for methane production 342 rate and VFA (R 2 =0.96). This means that daily methane production, hence 343 methanogenic activity, was very much dependant on the OLR, regardless of the SRT. 344 Similarly, acidogenesis increased with the OLR (Figure 5), but short SRT were not enough to convert all VFA to methane, which means that a portion of hydrolysed 346 organic compounds did not end up yielding methane. for a thermophilic process would be lower. 359 In the present study, the minimum SRT assayed was 6 days, but the minimum 360 SRT ensuring a stable performance was also 10 days. Methane production under 361 thermophilic conditions was improved by decreasing the SRT from 30 to 10 days. It 362 was further enhanced at 6 days SRT with an OLR higher than 5 kg VS m -3 d -1 , feeding 363 high-solids sludge. However, when the OLR eventually increased (> 6 kg VS m -3 d -1 ) as 364 a result of fluctuations in the solids content of the feed sludge, methanogenic activity 365 was severely affected; as indicated by decreased biogas production, with methane 366 content below 50 %, and a sudden accumulation of VFA, with a total concentration 367 higher than 6 g L -1 . Furthermore, the quality of the effluent in terms of VS content was 368 worsened.
On the other hand, working at SRT of 10 days still with high OLR (3-4 kg VS 370 m -3 reactor d -1 ), the process was more stable. Biogas and methane production rates ( this study, it seems that digested sludge dewaterability was deteriorated with TS higher 391 than 26 g L -1 and VS higher than 17 g L -1 ; which corresponded to OLR above 3 kg VS 392 m -3 reactor d -1 and SRT below 10 days. P r e -p r i n t    P r e -p r i n t P r e -p r i n t   Note: The start-up period has not been included P r e -p r i n t