Materials and methods

Characterization of fermented waste sewage sludge

FSS was obtained through anaerobic fermentation of secondary sewage sludge from the WWTP of Frielas. Sewage sludge was pre-treated at 80 °C for 3 h at pH = 8 and then fed semi-continuously into a pilot-scale (200 L) continuous stirred-tank reactor (CSTR) with an organic loading rate (OLR) of 11.1 gCOD L^-1 d^-1, sludge retention time (SRT) of 3.3 d, temperature of 35.1 ± 0.5 °C and pH = 4.9 ± 0.2. After fermentation, the FSS was collected and frozen until needed for characterization (Table 4.1) and use in the subsequent tests.

Experimental setup

4.2.2.1 Mg-enhanced chemical precipitation of ammonium

A set of experiments were carried out to evaluate the feasibility of removing ammonium from FSS through Mg-enhanced chemical precipitation. The optimal conditions were opti-mized using design of experiments (DoE). The factors chosen for this study were: time (0.33, 3 and 8h), pH (4.4 and 9) and the added magnesium to nitrogen ratio (0:1, 0.8:1 and 1.6:1 Mg-mol:N-mol). Since the quantity of Mg²⁺ in the FSS was not determined, 0 Mg-mol N-mol^-1 was chosen to test the possibility of not supplementing any Mg²⁺, in case the FSS contained enough; 1.6 Mg-mol N-mol^-1 was chosen as it was the optimal ratio according to literature [18].

A list of the experiments is shown in Table B1 in Appendix B.

Identical glass flasks with a total volume of 250 mL were filled with 200 mL of fermented sewage sludge and conditions set in each flask according to Table B1 in Appendix B. Mg²⁺ was added to the FSS in the form of MgO2H2, and the value corresponds to the molar ratio with relation to the NH4+ in the FSS. Flasks were stirred at 120 rpm and kept at 20 ºC throughout the experiment. Flasks were sampled at the end of the experiment to analyze for NH4+, PO4

3-and FP.

4.2.2.2 Stripping of ammonia

A set of experiments were carried out to evaluate the feasibility of removing ammonia from FSS through ammonia stripping and determine its optimal conditions using DoE. The factors chosen for DoE were the duration of the experiment (0.33, 1 and 3 h), pH (4.4, 9.5 and 11) and the presence/absence of aeration.

Table 4.1. Initial characterization of the FSS used in this study.

TS = total solids; TSS = total suspended solids; VSS = volatile suspended solids; VS = volatile solids; CODSOL = soluble COD; CODTOT = total COD; TNSOL = soluble fraction of total nitrogen; TN = total nitrogen.

Identical glass flasks with a total volume of 250 mL were filled with 200 mL of fermented sewage sludge and conditions set in each flask according to Table B2 in Appendix B. In the aerated tests, an air compressor Hailea V-20 was used with an air flow of 3.3 L min^-1. pH = 4.4 was used as it was the pH of the fermented sludge, pH = 9.5 as it is slightly above the pKa for the conversion of NH4+ to NH3 and pH = 11 because it has been shown to be among the most

Parameter Value

TS (gTS L^-1) 16.7 ± 0.5

TSS (gTSS L^-1) 9.20 ± 0.16

VSS (gVSS L^-1) 7.69 ± 0.07

VS (gVS L^-1) 16.6 ± 0.5

pH 4.41

CODSOL (gCOD L^-1) 8.30 ± 0.11

CODTOT (gCOD L^-1) 28.3 ± 1.6

TNSOL (gN L^-1) 1.12 ± 0.15

TN (gN L^-1) 1.67 ± 0.17

NH4+ (mgN L^-1) 993 ± 14

PO43- (mgP L^-1) 447 ± 19

NO3- (mg L^-1) 0 ± 0

NO2- (mg L^-1) 0 ± 0

Soluble proteins (g L^-1) 1.01 ± 0.00 Total proteins (g L^-1) 4.23 ± 0.00 Soluble carbohydrates (g L^-1) 0.34 ± 0.01 Total carbohydrates (g L^-1) 1.28 ± 0.01 VS/TS ratio (gVS gTS-1) 0.99 ± 0.04 VSS/TSS ratio (gVSS gTSS-1) 0.83 ± 0.02 CODSOL/CODTOT ratio (gCOD gCOD^-1) 0.29 ± 0.02 TNSOL/TN ratio (gN gN-1) 0.67 ± 0.11 FP/CODSOL ratio (gCODFP gCOD^-1) 0.80 ± 0.01

C:N:P molar ratio 100:39:3.8

FP (C-mmol L^-1) 181.8 ± 0.71

FP (gCODFP L^-1) 6.63 ± 0.02

efficient values for the stripping of ammonia [17]. Flasks were stirred at 120 rpm and kept at 20 ºC. Flasks were sampled at the end of the experiment to quantify NH4+, PO43- and FP.

After determining the best conditions, nitrogen recovery was tested by passing the ex-haust gas through a sulfuric acid solution (0.5 M). That solution was sampled frequently to quantify its NH4+ concentration.

4.2.2.3 Data analysis and model building

The experimental results of the chemical precipitation and stripping tests were processed in RStudio to assess the influence of the individual factors and their interactions with the ni-trogen removal (response). Analysis of Variance (ANOVA) was performed on each model and the significance of the model was determined based on the F-test. Significant factors and sig-nificant interactions between factors were used to create a second-order polynomial model, with the criteria for statistical significance of P-value < 0.05. The model was created according to the following general equation:

𝑌 = 𝑎₀+ ∑ 𝑎_𝑖𝑋_𝑖

3

𝑖=1

+ ∑ 𝑎_𝑖𝑖𝑋_𝑖²

3

𝑖=1

+ ∑ 𝑎_𝑖𝑗𝑋_𝑖𝑋_𝑗

3

𝑖,𝑗,=1

(1)

In this equation, Y is the predicted response, X represents the significant factors and a represents the coefficients determined by the models. The coefficient of determination (R²) was calculated to determine the quality of the fit. The fitted polynomial equations were graphically represented as surface three-dimensional plots to visualize the relationship between response and the significant factors.

4.2.2.4 Sequential batch reactor

A 2 L sequential batch reactor was inoculated with activated sludge from the WWTP of Mutela in Almada, Portugal and operated with a working volume of 1.8 L and a feast and famine regime. The reactor was aerated using an air compressor (Hailea V-20) with air flow of 2 L min

-1 and stirred with a mechanical impeller. The reactor was fed with FSS and operated with an OLR of 75 FP C-mmol L^-1 d^-1, a sludge retention time of 4 d and hydraulic retention time of 1 d. The reactor was operated with a 12-h cycle consisting of periods of 11 h of aeration and 1 h of settling. In Phase 1, the cycle started with a 10 min feeding of FSS and biomass was purged every cycle at the end of the aeration period. After settling, reactor media was withdrawn. In Phase 2, these conditions remained the same, except that treated FSS was fed at the start of

the cycle and a solution containing ammonium recovered from the stripped stream was sup-plemented at the end of the feast phase, in a proportion to impose an overall 100:7 C-mol:N-mol in the SBR cycle.

4.2.2.5 Accumulation batch assays

A 1 L reactor was assembled to perform the PHA accumulation tests (BioFlofi/CelliGenfi Fermentor and Bioreactor, Eppendorf). The reactor was inoculated with biomass produced in the selection step, completely aerated and operated at constant temperature (20 ºC). The re-actor was fed using a dissolved oxygen (DO) based pulse-wise feeding strategy, and the vol-ume of each pulse was such that the food to microorganism ratio was constant for each pulse along the accumulation test. Alike to operation in SBR, in Phase 1, the biomass was fed with FSS, whereas in Phase 2, it was fed with the FSS submitted to ammonia stripping.

Analytical procedures

COD was assessed using LCK Hach Lange kits (Hach) and Total Kjeldahl Nitrogen (TKN) was determined through a cuvette kit test LCV 909 (Hach); FP (acetate, ethanol, propionate, lactate, butyrate, iso-butyrate, valerate, iso-valerate and caproate, heptanoate and octanoate) were quantified by high performance liquid chromatography as described by [17]; Gas chro-matography coupled with flame ionization detector was used to quantify PHA concentration as described by [18]; suspended solids were determined according to Standard Methods [19];

concentration of ammonia, nitrite, nitrate and phosphate were determined using a colorimetric method by a segmented flow analyzer (Skalar San++ System); Sugars were quantified accord-ing to Dubois method [20] and proteins accordaccord-ing to the Lowry method [21].

Calculations

Relevant parameters characterizing the performance of the reactors were calculated after a steady state was achieved, which was recognized after 3 SRTs after any change in process operation.

In the SBR, rates (rPHA, rFP, rNFEAST, rNFAMINE) were calculated by the slope of PHA, FP or N, respectively, as function of the corresponding time period. Yields (YPHA/FP, YXA/FPFEAST and YXA/PHAFAMINE) were calculated by the ratio of production or consumption rates of the respective compounds. PHA content was calculated as gPHA/gTS and XA by subtracting PHA from VSS.

In the accumulation reactor, parameters were calculated similarly. Volumetric PHA productivity (P ) was calculated by dividing the ΔPHA by the time necessary to produce that

amount of PHA. The points used for the calculation of the ΔPHA were PHA amount at the beginning of the test and the instant at which 90% of the total PHA was produced.