A New Technique Using Headspace Gas Monitoring to Determine Carbon Source Addition in a BNR Process

A new method was developed using the carbon dioxide detected in the headspace of a batch reactor to determine the carbon source addition in the biological nutrient removal (BNR) process. The activated sludge used in this study was taken from the University of British Columbia (UBC) Wastewater Treatment Pilot Plant aerobic tanks undergoing simplified UCT process. Experiments showed that detectable changes of carbon dioxide concentration in the headspace reflected the acetate utilization in the anaerobic condition, i.e. the P release reaction. In the sodium acetate addition cases, the elapsed time (E Time) of carbon dioxide evolution rate changes was proportional to the amount of acetate added in solution. Results were successfully used to estimate the amount of acetate added in tests. This new methodology was also capable of estimating the amount of VFAs recovered from the thermophilic aerobic digestion (TAD) supernatant. This on-line “E Time” monitoring approach has the potential to optimize the carbon source addition in BNR processe

A New Technique Using Headspace Gas Monitoring to Determine Carbon Source Addition in a BNR Process Jowitt Z. Li * , Donald S. Mavinic * , and Harlan G. Kelly ** * University of British Columbia, Department of Civil Engineering, Environmental Engineering Group, Vancouver, B. C. V6T 1Z4, Canada ** Dayton & Knight Ltd., West Vancouver, B. C. V7V 3N9, Canada ABSTRACT A new method was developed using the carbon dioxide detected in the headspace of a batch reactor to determine the carbon source addition in the biological nutrient removal (BNR) process. The activated sludge used in this study was taken from the University of British Columbia (UBC) Wastewater Treatment Pilot Plant aerobic tanks undergoing simplified UCT process. Experiments showed that detectable changes of carbon dioxide concentration in the headspace reflected the acetate utilization in the anaerobic condition, i.e. the P release reaction. In the sodium acetate addition cases, the elapsed time (E Time) of carbon dioxide evolution rate changes was proportional to the amount of acetate added in solution. Results were successfully used to estimate the amount of acetate added in tests. This new methodology was also capable of estimating the amount of VFAs recovered from the thermophilic aerobic digestion (TAD) supernatant. This on-line “E Time” monitoring approach has the potential to optimize the carbon source addition in BNR processes. KEYWORDS Biological nutrient removal (BNR), carbon dioxide evolution, carbon source determination, on-line headspace monitoring, thermophilic aerobic digestion (TAD), volatile fatty acids (VFAs). INTRODUCTION Available carbon is an essential component for biological nutrient removal (BNR) enhancement. Volatile fatty acids (VFAs, C 2 -C 4 ) recovered from the thermophilic aerobic digested (TAD) sludge supernatant, under a microaerated operation, is a potential carbon source for denitrification and enhanced biological phosphorus removal (Chu, et al, 1996). However, high soluble COD and ammonia-N recovered from the TAD supernatant could disrupt the system performance and deteriorate effluent quality. Meanwhile, the VFA concentration in TAD supernatant fluctuates due to varied sludge conditions, such as the sludge age and ORP. TAD supernatant addition must be optimized to maximize the benefit of VFA utilization and prevent system overloading (Li, Mavinic, and Kelly, 1999). A quick and reliable method to estimate the available carbon source on-line would be considerably beneficial in optimizing the BNR performance. A new method, using the headspace carbon dioxide (CO 2 ) monitoring of a batch reaction under the anaerobic condition, was proposed to estimate the amount of carbon source addition. The fundamental concept was derived from the observation of P release corresponding to the acetate utilization. Complexity and interference of the carbonate-bicarbonate species and pH in the system were discussed elsewhere (Spérandio and Paul, 1997). The observed CO 2 concentration in the headspace is the overall outcome of various mechanisms, such as endogenous respiration, carbonate-bicarbonate balance, CO 2 reserve, and BNR. The objective of this study was to access the correlation between the carbon dioxide profiles observed in the headspace and the amount of acetate added in the activated sludge. - 61 - METHODOLOGY Experiment Apparatus The experiment apparatus, equipped with carbon dioxide (Vaislal  GMD20), ORP (Broadley-James  , Ag/Ag-Cl), pH (Oakton  , Ag/Ag-Cl), and temperature probes/sensors, is illustrated in Figure 1. Nitrogen gas was sparged through the entire test course to provide sufficient mixing of solution, prevent oxygen interference, and carry the gas for detection. Internal circulation was designed to enhance the transfer mixing in the headspace. The carbon dioxide sensor readings were verified by different volumes of air samples using a Shimadzu  TOC-500 inorganic carbon analyzer (samples taken at 1 atm, 100% relative humidity), the calibration curve is shown in Figure 2. Batch Tests The activated sludge used in this study was taken from the University of British Columbia (UBC) Wastewater Treatment Pilot Plant aerobic tank. Sludge was pre-sparged with nitrogen gas, at a 5 L/min rate for 10 minutes in a preparation tank, to remove the foam, if any, and carbon dioxide reserve in solution; then the sludge was transferred to the experimental reactor (3.6 L in total with 1.6 L headspace volume). The nitrogen flow rate was maintained at 1,200 mL/min in the experiment reactor. In each batch test, a certain volume of sodium acetate (1,000 mg/L as HAc) was added and the carbon dioxide profiles were monitored until the reaction completed. Unknown concentrations of sodium acetate were tested using the same procedure; their concentrations were estimated using the calibration curve generated from the known concentration tests. TAD supernatant samples were tested to derive CO 2 monitoring information, and this was compared with the known NaAc calibration curve, to estimate their VFA content. Due to the variance of sludge characteristics and activity, the comparison was made only on the basis of tests using the same batch of sludge sample. Acetate and VFAs concentrations were verified using a Hewlett-Packard ® 5880A gas chromatograph, equipped with a flame ionization detector (FID). Figure 1. Experiment setup of CO 2 headspace monitoring Spike and sampling ORP pH Temp. Data logger CO 2 sensor KOH Nitrogen gas V ent Sampling Gauge Internal circulation P Flowmeter P Fill pump Data logger - 62 - Figure 2. Carbon dioxide sensor calibration with air samples TAD operation A pilot-scale, single-stage TAD (75L) equipped with a Turborator ® aerator (Turborator ® Technologies Inc.) was operated to produce digested sludge for tests. TAD operations in this study were deliberately maintained at a microaerated condition (typically the ORP below –300 mV) to accumulate VFAs (Chu, et al., 1996). Temperature was maintained within the range of 50 to 60 ºC and volatile solid destruction averaged about 30% at 7 days of HRT. Digested sludge was centrifuged for 10 minutes at 3,000 rpm and supernatant was withdrawn for tests. VFAs concentration in supernatant averaged about 490 (100-1,230) mg/L as HAc. RESULTS AND DISCUSSIONS Figure 3 shows a typical CO 2 /pH profile of sodium acetate addition in the CO2 profile; the rising time between the left foot of the slope and right edge of the peak was defined as the “E Time” (elapsing time). Chemical tracing studies demonstrated that the CO 2 change was highly reflective of VFA utilization upon the spiking and depletion in solution. Results indicated that the CO 2 changes accompanied to the typical enhanced biological P release reaction. The pH profile showed a mirror image to the CO 2 and decreased (less than 0.01/minute in rate) after the addition; this implied that the pH change was due to the biological reaction. A 14 C tracing study suggested that the CO 2 monitoring was not practical in a biological phosphorus release reaction, because of insignificant CO 2 changes (Bordacs and Chiesa, 1989). However, in comparison with pH-buffered tests in this study, it was confirmed that the pH change in the reaction helped to result in distinguishing overall CO 2 change. The “E Times” of different acetate concentration additions are shown in Figure 4. There was a high linear correlation between the initial acetate concentrations and determined “E Times”. Results suggested that the unknown concentration could be estimated by interpolation. Unknown sodium acetate concentrations were estimated using this linear correlation and their verification (by gas chromatograph analysis) are shown in Table 1. The estimated error was less than 2 %, in four out of five tests. At lower concentrations, this method appeared to overestimate the concentration slightly. Results revealed that the CO 2 monitored using this approach and experimental design was able to estimate the initial acetate concentration in the reactor. CO2 verify-1 (n=16) y = 175.54x R 2 = 0.9998 0 200 400 600 800 1000 1200 1400 0.0 2.0 4.0 6.0 8.0 Reading (V) CO2 (ppm) - 63 - Figure 3. CO 2 and pH profiles of NaAc addition Table 1: Acetate concentration estimation and analytical verification Sample # E Times (minutes) Estimated conc. (mg/L as HAc) Analytical conc. (mg/L as HAc) Error in % * 1 8.3 10 9.9 1.0 2 16.5 20 19.8 1.0 3 14.1 17 16.8 1.2 4 4.6 5 4.8 4.2 5 30.2 36 35.8 0.6 *Error estimation in percentage: (estimated conc. – analytical conc.)/analytical conc. ×100 Slight foaming was also observed but in this study, its accumulation was not severe. Foaming had to be eliminated during the testing, to prevent the gas been trapped and affecting the CO 2 profiles. The carrier gas sparging rate also played a key role in deliver a distinguishable CO 2 profile, and it had to be carefully selected for each specific test condition. Occasionally, a short plateau was observed in the CO 2 profile; this was probably due to the dynamic equilibrium established in the system. A time lag after the acetate spike was consistently observed in every case. Transfer mechanisms inside the reactor and the carrier gas flow rate were probably the key factors resulting in this delay. However, according to ORP and pH profiles, this time lag seemed to be consistent and it did not have a significant effect on the determination of the “E Time” defined in this study. Preliminary experiments showed that the TAD supernatant addition resulted in comparable BNR enhancement, as the sodium acetate addition cases. A higher P/VFAs molar ratio, observed in TAD addition case, suggested there were other potential carbon substrate available for P release (Li, Mavinic and Kelly, 1999). TAD supernatant samples were tested using this CO 2 monitoring approach, to estimate its VFA concentrations. Figure 5 shows a typical comparison of CO 2 profile of TAD supernatant and NaAc addition. Observed CO 2 evolution rates were similar in both cases, and the TAD supernatant VFA estimations are shown in Table 2. Estimations showed a substantial overestimation, compared with the analytical results using the GC method. However, this is explained by the fact that the TAD supernatant contains other potential carbon sources, which could be converted to VFAs or utilized directly in the reactions. J1206-0-CO2 110 120 130 140 150 160 170 180 190 200 210 100 120 140 160 180 200 Time elapsed (min.) CO2 ppm E Time J1206-0-pH 7.30 7.35 7.40 7.45 7.50 7.55 7.60 7.65 7.70 100 120 140 160 180 200 Time elapsed (min.) pH - 64 - Figure 4: NaAc conc. vs. the “E Time” Figure 5: Comparison of CO 2 profile of TAD supernatant and NaAc additions. Table 2: TAD supernatant VFAs estimation based on “E Time” approach Estimated VFAs Analytical VFAs Analytical HAc Comparison in VFAs Comparison in HAc J0413-2-1 185 mg/L 169 mg/L 166 mg/L +9.5 % +11.4 % J0413-2-2 199 mg/L 169 mg/L 166 mg/L +17.8 % +19.9 % J0418-3 533 mg/L 507 mg/L 421 mg/L +5.1 % +26.6 % J0515-1 367 mg/L 343 mg/L 343 mg/L +7.0 % +7.0 % J0516-1 448 mg/L 399 mg/L 397 mg/L +12.2 % +12.8 % Average (S.D.) +10.3% (5.0%) +15.5% (7.7%) SUMMARY Headspace CO 2 information derived in this study was found to correspond highly to the VFA utilization and P release in anaerobic conditions. This new CO 2 monitoring technique showed promise in being able to determine the initial VFA concentration using the “E Time” approach. The CO 2 change in headspace was detectable with the help of the pH change during the P release reaction. This method was also able to estimate the TAD supernatant VFA concentration, or more specifically, the available carbon sources in the supernatant from a thermophilic aerobic digestion process. With the sludge directly withdrawn from the system and reacting with the carbon supplements in this batch-based test, results revealed the real state of utilizable carbon substrates involved in the particular BNR reaction. It also showed the potential, using the observed CO 2 profiles and “E Time” correlations with the carbon source concentrations, to determine on-line the carbon substrates in TAD samples, as well the BNR process performance of the P release reaction under anaerobic conditions. In a sequencing batch reactor, CO 2 J0515-1 and -2 0 50 100 150 200 250 300 0102030 Time elapsed (min.) CO2 (ppm) TAD NaAc R 2 = 0.9977 n=7 0 5 10 15 20 25 0 102030 NaAc Conc. mg/L as HAc E Time (min.) - 65 - monitoring information could be used to regulate the sequencing time, especially the stage of anaerobic conditions. In a continuous process, this semi-on-line CO 2 measurement could be applied to optimize the external carbon source dosing rates. ACKNOWLEDGEMENTS This research was supported by grants form the National Science and Engineering Research Council (NSERC) of Canada and the Dayton and Knight Ltd., West Vancouver, BC, Canada. The excellent technical support provided by the Environmental Engineering Laboratory and Wastewater Treatment Pilot Plant at the University of British Columbia is acknowledged and appreciated. REFERENCES Bordacs, K. and S. C. Chiesa (1989), Carbon flow patterns in enhanced biological phosphorus accumulating activated sludge cultures, Water Science and Technology, v21, n1, p387-396. Chu, A, D. S. Mavinic, W. D. Ramey, and H. G Kelly (1996), A Biological Model Describing Volatile fatty Acid Metabolism in thermophilic Aerobic Digestion of Wastewater Sludge, Water Research, v30, n8, p1759. Li, J. Z., D. S. Mavinic and H. G. Kelly (1999), Using Thermophilic Aerobic Digested Sludge Supernatant as a Potential Carbon Source in Biological Nutrient Removal System, The 7th IAWQ Asia-Pacific Regional Conference Proceeding Volume I, Taipei, p 289-p295. Spérandio, M. and E. Paul (1997), Determination of Carbon Dioxide Evolution Rate Using On- Line Gas Analysis During Dynamic Biodegradation Experiments, Biotechnology & Bioengineering, v53, n3, Feb. p243-p252. - 66 - . in the reactions. J1206-0-CO2 11 0 12 0 13 0 14 0 15 0 16 0 17 0 18 0 19 0 200 210 10 0 12 0 14 0 16 0 18 0 200 Time elapsed (min.) CO2 ppm E Time J1206-0-pH 7.30 7.35. HAc J0 413 -2 -1 185 mg/L 16 9 mg/L 16 6 mg/L +9.5 % +11 .4 % J0 413 -2-2 19 9 mg/L 16 9 mg/L 16 6 mg/L +17 .8 % +19 .9 % J0 418 -3 533 mg/L 507 mg/L 4 21 mg/L +5 .1 % +26.6