Anaerobic Co-Digestion of Tannery and Slaughterhouse Wastewater for Solids Reduction and Resource Recovery: Effect of Sulfate Concentration and Inoculum to Substrate Ratio

Abstract

Anaerobic digestion is considered unsuitable for the bioremediation of tannery effluent due to process inhibition, mainly due to high concentrations of sulfur species, and the accumulation of ${\mathrm{H}}_2\mathrm{S}$ and/or ${\mathrm{NH}}_3$. This study using the standardized biochemical methane potential protocol showed that efficient processing is possible with slaughterhouse wastewater, provided sufficient functional biomass is present at the start of the process and the ${\mathrm{SO}}_4{}^{2-}$ concentration is below inhibition threshold. Methanogenic activity (K = 13.4–17.5 and µm = 0.15–0.27) and ${\mathrm{CH}}_4$ yields were high when reactors were operated ISR ≥ 3 and/or lower ${\mathrm{SO}}_4{}^{2-}$ ≤ 710 mg/L while high ${\mathrm{SO}}_4{}^{2-}$ ≥ 1960 mg/L and ISR < 3.0 caused almost complete inhibition regardless of corresponding ISR and ${\mathrm{SO}}_4{}^{2-}$. The theoretical optimum operating conditions (922 mg/L ${\mathrm{SO}}_4{}^{2-}$, ISR = 3.72) are expected to generate 361 mL biogas/gVS, 235 mL ${\mathrm{CH}}_4$/gVS with reduction efficiencies of 27.5% VS, 27.4% TS, 75.1% TOC, 75.6% ${\mathrm{SO}}_4{}^{2-}$, and 41.1% COD. This implies that tannery sludge will be reduced by about 27% (dry mass) and ${\mathrm{SO}}_4{}^{2-}$ by 76%, with a fraction of it recovered as ${\mathrm{S}}_0$. The models displayed a perfect fit to the cumulative ${\mathrm{CH}}_4$ yields with high precision in the order Logistic > Cone > modified Gompertz > first order.

Highlights

• Tannery effluent characteristics exhibited significant batch to batch variability
• Inhibition was seemingly caused by ${\mathrm{SO}}_4{}^{2-}$, ${\mathrm{NH}}_3$ and ${\mathrm{H}}_2$ S, respectively
• The inhibitory effect of ${\mathrm{SO}}_4{}^{2-}$ led to a significant decrease in A, µm, and K
• Process recovered ${\mathrm{CH}}_4$, ${\mathrm{S}}_0$, and reusable water e.g for irrigation and construction
• Models perfectly fitted in the order Logistic > Cone > modified Gompertz > first order

1. Introduction

The leather tanning and products industries play prominent roles in the world’s economy, particularly for developing countries. South Africa is a net exporter of hides/skins, and is a renowned producer of exotic ostrich leather. The industry sustainably recycles skins which are by-products of the meat industry and prevents their disposal on to the environment. However, tanneries solve one problem and create another as they produce large amounts of tannery wastewater (TWW) loaded with toxic metal salts, and in/organic substances. Conventionally, a tonne of raw hide/skin yields approximately 200 kg of leather, 500 kg wet sludge, and 50–15,000 m³ of liquid effluent containing residual processing chemicals [1]. Typically, TWW contains high loads of chromium (Cr), sodium (Na), chlorides (Cl), nitrogen (TN), sulfate (${\mathrm{SO}}_4{}^{2-}$), sulfide (${\mathrm{S}}^{2-}{/\mathrm{H}}_2{\mathrm{S}/\mathrm{HS}}^{-}$), and suspended solids (SS) [2]. In addition, the process generates in-plant solid wastes such as untanned raw trimmings, fleshings, tanned waste blue splits, trimmings, and shavings. Most of the solids emanate from wet-blue processing and 80% is generated by the beamhouse processes [3]. Tanneries are therefore regarded as one of the most polluting industries particularly in developing countries that dominate the industry and supply >60% of the world’s skins/hides [1].

The adequate management of TWW and tannery solid waste, particularly sludge is onerous and expensive and impacts on the profitability of the tanneries. Sludge management costs account for nearly 40% of the overall budget and 55% of the process and maintenance costs of tannery wastewater treatment plants (TWWTPs) [4], while the costs of energy may be up to 60% of the total costs incurred in TWWTPs [5]. The amalgamation of slaughterhouses and tanneries presents an opportunity to integrate the value chain, promote the processing of fresh skins and co-treatment of TWW with slaughterhouse wastewater (SWW) [6]. The authors demonstrated the synergistic effect of co-digesting (AcoD) tannery waste activated sludge (TWAS) with slaughterhouse sludge (SHS). There is a lack of studies focusing on the AcoD of TWW with slaughterhouse wastewater (SWW) while evaluating process kinetics and recovering valuable resources. This is innovative approach will integrate the value chain and promote cleaner production through the processing of fresh or chilled skin/hides. This will eliminate sodium chloride (NaCl) preservation, the soaking stage, and NaCl availability in TWW.

However, the successful application of anaerobic digestion (AD) in treating TWW and SWW is hindered by microbial inhibitors, notably inorganic sulfur (S) species, ammonia (${\mathrm{NH}}_3$), volatile organic acids (VOA), and heavy metals. The presence of ${\mathrm{SO}}_4^{2-}$ favours the growth of sulfate-reducing bacteria (SRB) which compete with methanogens for available carbon sources (acetate (${\mathrm{CH}}_3\mathrm{COOH})$, and carbon dioxide/hydrogen (${\mathrm{CO}}_2/{\mathrm{H}}_2$). The availability of microbial inhibitors in TWW had prompted investigations on AcoD and pre-treatment, particularly coagulation in order to improve process efficiency. Successes in pretreatment studies have led to the realization that these ‘toxicants’ can potentially be recovered as value-added products. In light of the global adoption of circular bioeconomy principles, TWWTPs are increasingly being regarded as potential biorefineries. A review by Mpofu et al. [7] reported on the feasibility of using AD to recover sulfur species (${\mathrm{H}}_2\mathrm{S}$ and ${\mathrm{S}}_0$), ${\mathrm{H}}_2,$ and VOA rich biogas, biofertilizer/compost, metals, activated carbon, and/or reusable water. Therefore, the aim of this study was to determine the AD process efficiency and kinetics during the AcoD of TWW and SWW at different ${\mathrm{SO}}_4{}^{2-}$ concentrations using an acclimated inoculum. The study also seeks to ascertain whether the adoption of AD can: (i) improve the quality of treated TWW to a standard that promotes reuse (ii) and/or recover elemental sulfur (${\mathrm{S}}_0$), (iii) and/or recover renewable energy as biogas, and/or (iv) reduce the volume of sludge from TWWTPs, thereby improving the overall environmental and economic performance of tanneries.

2. Materials and Methods

2.1. Sampling

Samples used in this study were collected from an ostrich tannery that is integrated with a slaughterhouse (IOT). The tannery processes mainly fresh ostrich skins via wet-blue tanning. The tannery blends TWW and SWW and treats the blended effluents (ostrich tannery effluent (OTE)) using the activated sludge process (ASP). Six 50 L composite ostrich tannery effluent (OTE) samples were obtained from the balancing tank every 2 weeks over the course of 5 months (March to September 2018), to allow for fluctuations in effluent quality.

2.2. Analytical Methods

The loss on ignition standard methods were used to determine the concentration of total solids (TS) at 105 °C in an oven and total volatile solids (VS) in furnace at 550 °C, respectively [8]. Soluble fractions of the reactors were sampled before mixing and were analyzed to determine process efficiency. A Merck Spectroquant Pharo^® Spectrophotometer (Darmstadt, Germany) together with Merck cell tests or kits were used to determine the concentration of substances (

Table 1. Characteristics of different batches of ostrich tannery effluent.

Parameter	Batch 1	Batch 2	Batch 3	Batch 4	Batch 5	Batch 6	Mean	SD
TOC (mg/L)	2467	3380	4530	9080	485	820	3460	3148
COD (mg/L)	7945	8143	7903	15690	4387	7235	8551	3768
BOD (mg/L)	3532	1472	1542	1515	1531	1552	1857	821
VOAt (mg/L AAE)	3070	2800	2440	2480	2120	1800	2452	456
TN (mg/L)	440	235	180	260	220	530	311	140
TAN (mg/L NH3-N)	18.8	13.5	16.2	41.0	13.2	9.60	18.7	11.3
NO3 (mg/L)	143.8	39.0	28.6	18.7	11.5	54.7	49.4	48.7
TP (mg/L PO42−-P)	6.65	5.10	5.05	17.8	5.00	4.90	7.41	5.11
${\mathrm{SO}}_4{}^{2-}$ (mg/L)	1114	626	352	424	173	1186	646	417
HS− (mg/L)	ND	2.38	5.70	2.20	0.00	0.12	2.08	2.31
Cl (mg/L)	2038	1547	1294	1022	911	2369	1530	576
TS (g/L)	7.85	8.07	8.38	19.4	5.53	4.69	8.98	5.30
TVS (g/L)	3.61	4.06	4.97	14.6	2.82	2.32	5.40	4.61
K (mg/L)	11.7	19.9	13.2	12.3	10.7	6.6	12.4	4.3
Na (mg/L)	1477	1315	1953	2789	964	754	1542	740
Fe (µg/L)	3272	3081	606	497	193	282	1322	1446
Ca (mg/L)	11.7	24.0	24.2	6.9	17.6	16.8	16.9	6.8
Mg (mg/L)	19.4	15.5	39.5	55.3	14.9	13.2	26.3	17.2
Zn (µg/L)	1568	674	439	401	229	198	585	511
Cu (µg/L)	304	136	16.6	65.9	12.7	12.1	91.2	115
Co (µg/L)	7.7	4.7	1.6	1.9	44.2	1.9	10.3	16.8
Cd (µg/L)	2.27	1.08	0.18	0.27	0.18	0.18	0.69	0.85
Ni (µg/L)	73.1	18.4	18.7	21.3	5.6	8.4	24.2	24.7
Cr (µg/L)	766	57	1094	350	584	136	498	395
Pb (µg/L)	8.4	2.3	4.8	6.8	5.2	4.1	5.3	2.1
Al (µg/L)	1798	2366	583	624	85	101	926	941
Alk (g/L CaCO3)	245	236	330	264	297	308	280	37.0
EC (mS/cm)	8.22	8.27	8.81	11.87	4.04	3.61	7.47	3.13
pH	6.49	6.73	7.33	7.09	6.92	6.93	ND	ND
TVS:TS	0.46	0.50	0.59	0.76	0.51	0.49	ND	ND
BOD:COD	0.44	0.18	0.20	0.10	0.35	0.21	ND	ND
C:N	5.61	14.38	25.17	34.92	2.20	1.55	ND	ND
VFA:Alk	12.5	11.9	7.39	9.39	7.14	5.84	ND	ND
$\mathrm{COD}: {\mathrm{SO}}_4{}^{2-}$	7.13	13.0	22.4	37.0	25.3	6.10	ND	ND
COD:TVS	2.20	2.00	1.59	1.07	1.56	3.12	ND	ND

ND = no data given, SD = standard deviation.

): chemical oxygen demand (COD) (cat no: 14555), 5-day biological oxygen demand (BOD₅) (cat no: 00687), total organic carbon (TOC) (cat no: 14879), total volatile organic acids (VOA_t) as acetic acid equivalents (AAE) (cat no: 01763), total sulfate (${\mathrm{SO}}_4{}^{2-}$) (cat no: 118389), total (${\mathrm{S}}^{2-})\mathrm{as} \left({\mathrm{HS}}^{-}\right)$ (cat no: 14779), total nitrogen (TN) (cat no: 14537), nitrate $({\mathrm{NO}}_3{}^{-}$) (cat no: 114776), nitrite (${\mathrm{NO}}_2{}^{-}$) (cat no: 114776), total phosphate (TP) as phosphorous (${\mathrm{PO}}_4{}^{2-}-\mathrm{P}$) (cat no: 14729), total ammonia nitrogen (TAN) (${\mathrm{NH}}_3-\mathrm{N}$) (cat no: 00683), and total alkalinity (Alk) as calcium carbonate (${\mathrm{CaCO}}_3)$ (cat no: 101758), following the prescribed procedures.

The concentrations of metals (Al, Cd, Cr, Co, Cu, Fe, Ni, Pb, and Zn) and other cations (Ca, Cl, K, Mg, and Na) were acidified with nitric acid (HNO₃) and were quantified by inductively coupled plasma (ICP) atomic emission spectroscopy (AES) using a Thermo ICap 6200 ICP-AES instrument, while ultra-trace analyses were performed by ICP-mass spectrometry (MS) using an Agilent (Santa Clara, USA) 7900 ICP-MS instrument.

2.3. Biomethane Potential Experiments

The biomethane potential (BMP) experimental protocol described by Holliger et al. [9] was followed in this study. The constant inoculum amount added in all the reactors was 176 mL while the blended OTE varied from 850 mL to 2120 mL. The reactors were 2.5 L (total volume) screw-capped borosilicate bottles with modified lids fitted with o-rings containing stainless steel inserts with gas-tight ports and tubing to allow sampling and biogas collection. All reactors were topped up to 2.3 L using deionized water. There was a high TVS ratio between inoculum and OTE (~25:1), and the volume occupied by the inoculum was relatively low, even at high inoculum to substrate ratios (ISRs). An acclimated inoculum was prepared by feeding OTE to digestate obtained from mesophilic batch reactors treating TWAS. The inoculum was kept at 37 °C and was fed with OTE until biogas production and quality stabilized [10].

The effect of 2 numeric factors were investigated: ${\mathrm{SO}}_4{}^{2-}$ (665–2000 mg/L) and ISR (2–5), which were assessed using 2 responses: anaerobic biodegradability ${\mathrm{B}}_{\mathrm{o}}$ (%COD, %TS, %TVS, and %TOC reduction)], and maximum ${\mathrm{CH}}_4$ yield (ML ${\mathrm{CH}}_4$/gVS), using response surface methodology (RSM). The RSM was based on a full factorial central composite experimental design (CCD) with 13 runs, 5 replicates (R4, R8–R11) and 5 levels for each factor. The experimental design matrix (

Table 2. Experimental design matrix showing the methane yield and biodegradability results.

Reactor	$\mathbf{A}:\mathbf{S}{\mathbf{O}}_4^{2-}$	B:ISR	Biogas Yield	Methane Yield	$\mathbf{Average}\mathbf{C}{\mathbf{H}}_4$	Biodegradability Indicators (% Reduction)
						TOC	Sulfate	TS	VS	COD
	(mg/L)		(mL/gVSadded)	(mL/gVSadded)	%	(%)	(%)	(%)	(%)	(%)
R1	1960	2.5	129	12.0	9.30	87.6	59.7	11.7	15.2	16.6
R2	1335	2.0	4.97	1.71	34.4	93.1	63.4	11.5	19.2	13.8
R3	710	2.5	83.6	17.0	20.3	76.2	77.1	20.1	28.5	19.2
R4	1335	3.0	79.8	25.7	32.2	81.3	73.6	19.3	26.7	24.4
R5	710	4.0	361	93.3	25.9	82.5	68.0	29.1	40.9	52.0
R6	1335	5.0	180	41.5	23.1	88.3	81.1	26.0	55.0	10.8
R7	1960	4.0	110	19.5	17.7	77.1	80.4	28.3	33.3	14.9
R8	1335	3.0	337	146	43.4	82.7	80.4	26.1	30.5	43.6
R9	1335	3.0	260	100	38.5	69.0	85.3	49.2	51.7	24.6
R10	1335	3.0	265	102	38.5	61.3	79.6	37.0	40.8	43.0
R11	1335	3.0	280	103	36.9	75.3	75.0	35.8	37.7	24.3
R12	2000	3.0	9.13	1.02	11.2	23.2	49.3	19.0	27.5	39.3
R13	665	3.0	290	130	44.8	78.5	77.8	29.0	35.4	25.5

COD = chemical oxygen demand, Conc = concentration, ISR = inoculum to substrate ratio, R = reactor, TOC = total organic carbon, TS = total solids, VS = volatile solids.

) was generated using Design-Expert^® Software Version 11 (Stat-Ease, Inc., Minneapolis, MN, USA). Two sets of each reactor and negative controls (inoculum and substrates only) reactors were set up. However, no positive controls were setup. Magnesium sulfate (${\mathrm{MgSO}}_4$) and potassium sulfate (${\mathrm{K}}_2{\mathrm{SO}}_4$) were added to the OTE of selected reactors to mimic the upper range of ${\mathrm{SO}}_4{}^{2-}$ expected in the effluent stream. Reactors were manually mixed once a day for two minutes.

Biogas Sampling and Analysis

The gas ports from the 2 L BMP bioreactors were connected to individual gas sampling bags. When sufficient biogas was produced (>200 mL), samples were analyzed qualitatively and quantitatively. The ${\mathrm{CH}}_4$, ${\mathrm{CO}}_2$, $\mathrm{CO}$ and oxygen (${\mathrm{O}}_2$) content (%vol), as well as the ${\mathrm{H}}_2\mathrm{S}$ content (parts per million (ppm) of the gas were determined using a Geotech biogas 5000 analyzer (Warwickshire, UK) according to the manufacturers’ instructions. Biogas volume was determined using a gas syringe.

3. Results and Discussion

3.1. Characteristics of Ostrich Tannery Effluent

It has been shown that there are significant differences in the TWW generated from the processing of either ostrich skins or bovine hides [7]. One contributing factor is the differences in the tanning and TWW treatment processes. In this study, the variations in the slaughterhouse and tannery operations was the main contributing factor.

3.1.1. Chemical and Biological Oxygen Demand and Solids Concentrations in the Ostrich Tannery Effluent

As expected, there was a batch-to-batch variation in the parameters that were measured in the OTE (Table 1). The TS, TVS, and COD significantly varied (ANOVA, p < 0.05) and batch 4 (June-winter) was higher. Apart from routine differences in the daily industrial processes, it was hypothesized that (i) samples containing different proportions of SWW and/or (ii) ostrich skins containing a higher amount of fat during colder months may have played contributory roles [10]. Indeed, the TS, TVS, and COD concentrations exhibited an inverse relationship with the monthly average temperature profile of the area. The BOD:COD ratio (=0.10) in batch 4 was the lowest of all the batches, but the BOD itself was within the range of the other batches. These results suggest that, if organic solids, notably fats, were responsible for the high TS, TVS, and COD concentrations in batch 4, then these were mainly recalcitrant in nature.

The gCOD:gTVS ratios of batches 1–6 were 2.20, 2.00, 1.59, 1.07, 1.56, and 3.12, respectively. These results suggested that batches 1, 2, and 6 were more lipid-like in character (gCOD:gTVS = 2.9) due to the low TVS content, whilst the other batches were more protein-like (gCOD:gTVS = 1.42) [11]. It was postulated that the insignificant differences (ANOVA, p > 0.05) in TVS:TS ratios and gCOD:gTVS were mainly due to the elucidated factors that caused differences in the TVS content of the OTE. Assuming no inhibition, ${\mathrm{CH}}_4$ yields of approximately 1000 mL/gTVS for lipid-like batches and 415–496 mL/gTVS for protein-like batches with 50–71% ${\mathrm{CH}}_4$ are expected [11]. Despite the lack of TVS:TS ratios being reported in literature, it was envisaged that batch 4 samples would be the most suitable for AD as TVS:TS ratios >0.8 are required for efficient reactor performance [12].

3.1.2. Concentration of Nitrogen, Carbon and Volatile Organic Acids in the Ostrich Tannery Effluent

Total nitrogen concentrations (TN) in TWW are widely reported in literature, and are typically high [7]. The TN determined in this study for OTE (Table 1) were in keeping with literature values. In contrast to TN, there is a lack of studies that report TOC and C:N ratios of TWW. Except for batch 5 (485 mg/L, end July 2018), the TOC (820–9080 mg/L, Table 1) in this study were higher than 510 mg/L reported by Bhattacharya et al. [13]. Although, the optimal C:N range for AD is 20–30 [14], the optimal range for TWW has been reported as 6–9 [15]. Anaerobic reactors operating at lower than optimal C:N ratios are likely to suffer ${\mathrm{NH}}_3$ and VOA inhibition during AD. In this study, although the TN of batch 3 and 4 (May–June) was low, the TOC concentrations were high compared to other batches. Batch 3 exhibited an optimal C:N ratio (25.2 ± 0.9) for AD whilst batch 4 was above optimal (34.9 ± 1.6) and the rest were below optimal (1.55–14.4).

Animal skins have a layer of fat that can increase in colder months and result in increased lipid concentrations in TWW. Fat floats were observed in the OTE, particularly in batch 4. Generally, lipids have high ${\mathrm{CH}}_4$ yields and B_o, and require long retention times due to low degradation rates [16]. High lipid concentrations (long chain fatty acids) in OTE may theoretically inhibit AD [16]. In this study, the measured VOA_t concentrations (1.80–3.07 g/L) were mostly below the inhibitory thresholds (VOA_t = 5.80–6.90 g/L) [17]. The VOA:ALK ratios of OTE (0.53–0.62) were >0.4, indicated the possibility of AD process instability and failure [18]. However, the speciation of ${\mathrm{NH}}_3-{\mathrm{NH}}_4$ during the AD of nitrogenous wastes serves as a buffer and plays a vital role in maintaining a relatively constant pH [15,19].

3.1.3. Inorganic Characteristics of Ostrich Tannery Effluents

The TN and ${\mathrm{NO}}_3{}^{-}$ were not significantly different (p > 0.05) in batches 2–5, and highest in batches 1 and 6. The ${\mathrm{NH}}_3$ concentration in batch 4 was notably higher than in the other batches. The batch-to-batch variation in N species was assumed to be due to process variations, environmental factors, effluent constituents and microbial activity (hydrolysis). The ${\mathrm{NH}}_3$ concentrations were all below the inhibiting range of 53–1450 mg/L for AD [20]. As expected in OTE, high ${\mathrm{SO}}_4{}^{2-}$, Na, Cl, Cr, Fe, and Ca concentrations were found. The concentration trends of Na, Cl, Fe, Cr, TS, and COD were similar (r = 0.64–0.86) from batch to batch.

Significant differences (ANOVA, p < 0.05) in ${\mathrm{SO}}_4{}^{2-}$ and ${\mathrm{HS}}^{-}$ were assumed to emanate mainly from the differences in tanning operations rather than differences in the TWWTP performance and environmental factors. The high COD: ${\mathrm{SO}}_4{}^{2-}$ ratios (13–37) of batch 2–5 were above the reported ranges for favoring methanogenesis over sulphidogenesis [21]. However, batch 1 and 6 were within the 1–7 range and capable of supporting either methanogenesis or sulphidogenesis. The macronutrient (C:N:P:S) ratios were in the range 2.1–64:1.1–3.8:0.01–0.3:1 (Table 1) and were not equivalent to the optimal ratio of 500–600:15:5:1 for AD [22]. Therefore, acclimatization of the inoculum was most vital in ensuring effective AD.

3.1.4. Metal Characteristics of Ostrich Tannery Wastewater

The concentration of most metals (Na-Mg; Zn-Cu-Ni-Cd-Fe-Al; Ni-Pb; and Cr-Al) in the OTE samples displayed a similar trend (r = 0.73–0.99), with the first 2 batches having the highest concentrations. The IC₅₀ values for methanogens and acetogens have been reported as 11 g/L Na, 28 g/L K, 4.8 g/L Ca, 4–8 mg/L Cd, 100–400 mg/L Ni, 17–58 mg/L Zn, 67 mg/L Pb, 8.3–3000 mg/L Cr, and 0.7–5.65 g/L Fe [23,24]. The metal concentrations in the OTE were below the reported IC_50. Some metals, such as Ni, Zn, Co, Cu, and Ca are also necessary as metabolic co-factors, and in this study, they were either within or below the optimal range for AD. However, inhibiting and optimal metal concentrations strongly depend on their availability as free ions, their physico-chemical properties, operating conditions, microbial species and their adaptation [25]. The presence of metal mixtures in TWW may exhibit antagonistic and synergistic effects to inhibition [26].

3.2. Biochemical Methane Potential Experiments

Methanogenesis is generally considered to be the slowest, most sensitive, and often rate-limiting reaction when processing tannery effluents as they are laden with soluble and/or unionised toxicants such as ${\mathrm{NH}}_3$/${\mathrm{NH}}_4{}^{+}$, ${\mathrm{SO}}_4{}^{2-}$, ${\mathrm{H}}_2\mathrm{S}$/${\mathrm{HS}}^{-}$, VOA, and metals [7].

3.2.1. Cumulative Methane Generation

In this study, negligible biogas that could not be quantified was generated in the inoculum and substrate controls. In reactors with ISR ≥ 3, and ${\mathrm{SO}}_4{}^{2-}$ ≤ 710 mg/L lag phases between 5 and 23 days were experienced before ${\mathrm{CH}}_4$ generation commenced (Figure 1), reflecting complete, but transient inhibition. In addition, in four of five replicates with ISR = 3, close to 50 days were required for ${\mathrm{CH}}_4$ generation to reach completion. The ${\mathrm{SO}}_4{}^{2-}$ concentrations in these reactors was in the upper range expected in the OTE from the tannery concerned (${\mathrm{SO}}_4{}^{2-}$ = 352–1186 mg/L, n = 6, Table 1). The average cumulative ${\mathrm{CH}}_4$ yield in these reactors ranged from 98 to 146 mL/gVS, higher than the ${\mathrm{CH}}_4$ yield reported by Saxena et al. [27], but lower than that reported by Achouri et al. [28] for AD of TWW without pre-treatment (7.6 mL/gVS and 753 mL/gVS, respectively, after 35 and 37 days, respectively). Studies by Mpofu et al. [6,19] also reported long lag phases of >60 days during mono-digestion of ostrich TWAS, and 20 days during co-digestion (AcoD) (50%/50% v/v) with ostrich SHS that led to retention times of 108 and 50 days, respectively.

In the reactors with ${\mathrm{SO}}_4{}^{2-}$ ≥ 1960 mg/L (i.e., above the range expected in the OTE from the study site), methanogenesis was severely inhibited at all ISRs (Figure 1B). However, at ISR ≥ 3 and ${\mathrm{SO}}_4{}^{2-}$ ≤ 710 mg/L (mid concentration range from study site), no lag phase was experienced, and more than 92% of the cumulative ${\mathrm{CH}}_4$ generation of 130–139 mL was obtained within 10 days of operation (Figure 1C). These results are very promising and strongly suggest that by optimizing the sludge recycle ratio and/or SRT, efficient AD can be achieved, provided the ${\mathrm{SO}}_4{}^{2-}$ concentration is kept below a particular (high) threshold. Further experiments need to be conducted to optimize these, and other factors (such as mixing), in continuous or semi-continuous systems.

3.2.2. Hydrolysis and pH Changes

The lack of ${\mathrm{CH}}_4$ generation in some of the reactors in the first 15 days reflected poor or absent methanogenic activity, but not necessarily a lack of other metabolic processes. Analysis of selected physicochemical parameters of the reactor contents established that hydrolysis, acidogenesis and acetogenesis took place.

It was assumed that the primary mechanism for ${\mathrm{NH}}_3$ release (76–89% increase after 20 days), was protein hydrolysis. The overall increase in VOA in some reactors, and 38–80% reduction in FOG clearly indicated that hydrolysis of lipids also occurred. The ${\mathrm{NH}}_3$ concentration in samples taken at day 0 and at day 20 in all reactors fell above the minimum inhibitory concentration (MIC) reported in literature for AD (53 mg/L; [29]), but fell well below this MIC at the end of the study (day 62). Temporal ${\mathrm{NH}}_3$ increases (25–147%) between day 0 and day 20 were observed in reactors operating at higher ISRs (>2.5) and/or lower ${\mathrm{SO}}_4{}^{2-}$ ≤ 1335 mg/L, while decreases (5–51%) were observed for reactors operating at lower ISRs ≤ 2.5 and/or higher ${\mathrm{SO}}_4{}^{2-}$ ≥ 1335 mg/L. The pH in the reactors remained largely within the optimal range for methanogens over the first 20 days. However, values measured in all reactors at the end of the study were slightly higher than the optimal range (6.5–8.0) [30]. Buffering capacity provided by high ${\mathrm{NH}}_3$/${\mathrm{NH}}_4{}^{+}$ concentration probably compensated to some extent for the initial high VFA:ALK of the reactor contents.

In terms of VFA:ALK, initial ratios (>0.4) suggested that potentially unstable operational conditions for methanogenesis existed during start-up, which stabilized after 20 days (<0.3–0.4) in all reactors with the exception of R3 and R12 which generated minimal ${\mathrm{CH}}_4$. There was a notable increase in alkalinity, suggesting microbial utilisation of ${\mathrm{H}}^{+}$ in the reactors, for example by oxidising homoacetogens, chemolithotrophic sulfur oxidising bacteria (SOB), and/or hydrogenotrophic methanogens (HMs).

3.2.3. Acidogenesis, Acetogenesis and Changes in Volatile Organic Acid Concentration

The initial and final VOA concentrations in R1 and R12 increased by 16% and 17%, respectively. In contrast, decreases ranging from 7% to 60% were noted in other reactors. Together with changes in the VOA, ${\mathrm{SO}}_4{}^{2-}$ was reduced to ${\mathrm{H}}_2\mathrm{S}$, suggesting that both acidogenesis and acetogenesis occurred during the lag phase. The accumulation of VOA in the two reactors (R1 and R12) operating at high ${\mathrm{SO}}_4{}^{2-}$ ≥ 1960 mg/L and ISR ≤ 3 suggested the involvement of SRB in the breakdown of complex substrates.

The VOA concentration at any point in time depends on the balance between the breakdown of the products of hydrolysis into VOA by acidogens, and utilization of the VOA by acetogens and/or aceticlastic methanogens. Increased VOA concentrations could therefore be attributed to inhibition of the latter two metabolic groups. In contrast, decreased VOA concentrations could be attributed to either inhibition of acidogens (decreased formation) and/or efficient acetogenic/methanogenic activity (utilization). High ${\mathrm{H}}^{+}$ partial pressures greater than 10⁻⁴ atmospheres are also known to inhibit propionate (HPr), butyrate (HBu), and ethanol degrading acetogens [31].

Lipid inhibition may have occurred in R1, R2 and R12, which may have led to decreased acidogenesis and/or acetogenesis and subsequent accumulation of VOA.

3.2.4. Sulfidogenesis, Sulfite Oxidation and Methanogenesis

Notable reductions in ${\mathrm{SO}}_4{}^{2-}$ occurred in all the reactors over the study period, with concomitant increases in ${\mathrm{H}}_2\mathrm{S}$ over the first 20 days of operation. It was assumed that sulfidogenesis was largely responsible for the simultaneous increase in ${\mathrm{H}}_2\mathrm{S}$ and reduction in ${\mathrm{SO}}_4{}^{2-}$, and that SRB dominated not only ${\mathrm{HS}}^{-}$ generation, but also contributed to organic substrate utilization. Furthermore, at pH > 8, ${\mathrm{H}}_2\mathrm{S}$ is solubilized to ${\mathrm{HS}}^{-}$, such that as the pH increased in the reactors, some of the ${\mathrm{HS}}^{-}$ may have precipitated with metals. This can reduce direct ${\mathrm{HS}}^{-}$ toxicity on functional microbial species, but it can also reduce the bioavailability of essential methanogenic micronutrients. The ${\mathrm{HS}}^{-}$ concentrations measured at day 20 fell within the inhibitory range (IC₅₀ 43–125 mg/L at pH 7–8) for methanogenesis [32].

Nonetheless, it was apparent that about 43–96% of the formed ${\mathrm{S}}^{2-}$ from sulfidogenesis was oxidized into elemental sulfur (${\mathrm{S}}_0$) as a white layer was formed at the interface of the bulk liquid and head space in the reactors. This phenomenon has been described by Sabumon [33,34], whom observed the formation of ${\mathrm{S}}_0$ as the main intermediary product of ${\mathrm{HS}}^{-}$ and ${\mathrm{H}}_2\mathrm{S}$ oxidation during treatment of TWW. Moraes [35] reported that ${\mathrm{SO}}_4{}^{2-}$ may be re-formed by oxidation of thiosulfate (${\mathrm{S}}_2{\mathrm{O}}_4{}^{2-}$) and elemental ${\mathrm{S}}_0$. This may have been the case with R1 and R2 where a 91% reduction in ${\mathrm{SO}}_4{}^{2-}$ by day 20 was followed by 99% increase by day 62. Chemolithotrophic SOB can simultaneously reduce oxidized N compounds (${\mathrm{NO}}_2{}^{-}$ and ${\mathrm{NO}}_3{}^{-}$) and oxidize ${\mathrm{S}}^{2-}$ under anoxic conditions (Equation (1) to Equation (4)). Other inorganic reduced S compounds such as ${\mathrm{S}}_2{\mathrm{O}}_4{}^{2-}$ and ${\mathrm{S}}_0$ can also be used as electron donors by SOB. In this study, the notable decrease in the ${\mathrm{NH}}_3$ between day 20 and day 62 in the reactors supports denitrification having occurred, which would have made ${\mathrm{NO}}_2{}^{-}$ and ${\mathrm{NO}}_3{}^{-}$ available as electron donors for SOB. Furthermore, consumption of ${\mathrm{H}}^{+}$ by SOB could explain the anomalous increase in alkalinity in all reactors, except R5 and R8. However, due to the complexity of the physicochemical and biological processes, detailed mass balances for S were not determined.

$$5\mathrm{H}{\mathrm{S}}^{-}+\mathrm{N}{\mathrm{O}}_3{}^{-}+ 3{\mathrm{H}}^{+}\to 5\mathrm{S}{\mathrm{O}}_4{}^{2-}+4{\mathrm{N}}_2+4{\mathrm{H}}_2\mathrm{O}\Delta \mathrm{G}-3848\mathrm{kJ}/\mathrm{mole}$$

(1)

$$3\mathrm{H}{\mathrm{S}}^{-}+8\mathrm{N}{\mathrm{O}}_2{}^{-}+ 5{\mathrm{H}}^{+}\to 3\mathrm{S}{\mathrm{O}}_4{}^{2-}+4{\mathrm{N}}_2+4{\mathrm{H}}_2\mathrm{O}\Delta \mathrm{G}-2944\mathrm{kJ}/\mathrm{mole}$$

(2)

$$5\mathrm{H}{\mathrm{S}}^{-}+2\mathrm{N}{\mathrm{O}}_3{}^{-}+ 7{\mathrm{H}}^{+}\to 5{\mathrm{S}}_0+{\mathrm{N}}_2+6{\mathrm{H}}_2\mathrm{O}\Delta \mathrm{G}-253\mathrm{kJ}/\mathrm{mole}$$

(3)

$$3\mathrm{H}{\mathrm{S}}^{-}+2\mathrm{N}{\mathrm{O}}_2{}^{-}+ 5{\mathrm{H}}^{+}\to 3{\mathrm{S}}_0+{\mathrm{N}}_2+4{\mathrm{H}}_2\mathrm{O}\mathrm{G}-306\mathrm{kJ}/\mathrm{mole}$$

(4)

The COD:${\mathrm{SO}}_4{}^{2-}$ ratios were consistently <10, suggesting that sulfidogenesis would be favored over methanogenesis. It was therefore hypothesized that (i) methanogenesis was favored in reactors R5 and R13 where no lag phase for ${\mathrm{CH}}_4$ generation was experienced (ISR ≥ 3 and ${\mathrm{SO}}_4{}^{2-}$ ≤ 710 mg/L), (ii) sulfidogenesis initially dominated, followed by methanogenesis in the reactors that generated ${\mathrm{CH}}_4$ after lag phases, and (iii) sulfidogenesis was favored, and methanogenesis was severely and continually inhibited in reactors with high ${\mathrm{SO}}_4{}^{2-}$ of ≥ 1960 mg/L (R1, R7, R12). This was supported by qualitative analysis of the biogas from R1 (230 mL biogas: 78 ppm ${\mathrm{H}}_2\mathrm{S}$, 0.1% ${\mathrm{O}}_2$, 0% ${\mathrm{CH}}_4$). It was hypothesised that the physicochemical and biological milieu in R1 and R12 inhibited aceticlastic methanogens (AMs) and/or acetogens, leading to an accumulation of VOA between day 20 and day 62, which exacerbated methanogenic inhibition.

The results strongly suggested that SRB played a key role in the syntrophic degradation of hydrolysis metabolites and VOAs in most of the reactors. This could in turn have led to the dominance of HMs after the lag phase, which was suggested by the ${\mathrm{CH}}_4:{\mathrm{CO}}_2$ (>1) for the reversible inhibited reactors [36]. It was conclusively established that ${\mathrm{SO}}_4{}^{2-}$ ≥ 1960 mg/L caused almost complete methanogenic inhibition, while no inhibition occurred when reactors were operated at ${\mathrm{SO}}_4{}^{2-}$ ≤ 710 mg/L and ISR ≥ 3. Furthermore, both reactors that were operated at the lowest ISR (=2.5) were almost completely inhibited, suggesting that the initial methanogenic population density was too low to become established in that particular environment.

3.2.5. Metal Inhibition or Stimulation and Macronutrient Limitation

Depending on the speciation and concentration, bioavailable metals and other ions may either promote or inhibit methanogenesis. In this study, the concentrations of all the essential metals decreased, except for Ni in R1, R5, R7, R11, R12, and R13. More holistically, the concentrations of most soluble metals decreased in the reactors, except for R7 and/or R12. In contrast Ca increased concurrently with a decrease in VOA, ${\mathrm{NH}}_3$, and ${\mathrm{H}}_2\mathrm{S}$ except in R2, R7, and R12 that operated at high ${\mathrm{SO}}_4{}^{2-}$ concentrations and/or lower ISR. Interestingly, the same reactors exhibited a low ${\mathrm{CH}}_4$ yield and experienced longer lag phases. This supports the observed flocs in the different reactors which may have formed in an effort to adapt and prevent metal toxicity or deficiency through excretion of extracellular polymeric substances and soluble microbial products [25].

It is plausible that inhibition was initially caused by ${\mathrm{SO}}_4{}^{2-}$, followed by ${\mathrm{NH}}_3$, ${\mathrm{H}}_2\mathrm{S}$, and/or nutrient limitation. The C:N range was below optimal in all reactors at the beginning and end of the study. The initial C:N:P ratios ranged from 280:37:1 to 111:26:1, indicating both N and P were limiting macronutrients. However, although there was a notable reduction in concentration, bioavailable (soluble) P was still present at day 62. Furthermore, the ${\mathrm{HS}}^{-}$ that was generated may have reduced the bioavailability of inhibitory and/or stimulatory micronutrients via precipitation [16]. The precipitates may have further reversibly inhibited functional microorganisms by blocking their access to substrates [37].

3.3. Optimisation of Cumulative Methane Yield and Anaerobic Biodegradability

The experimental gas yields (${\mathrm{CH}}_4$ and biogas) and ${\mathrm{B}}_{\mathrm{o}}$ (% reduction of TOC, TS, VS, COD) (Table 2) were modelled using linear equations and quadratic polynomials with up to second degree interaction terms (Figure 2).

The models (Equations (5)–(11)) were significant (F test, p < 0.05) and there was only 0.12–3.20% probability that this may have been caused by natural system variation (

Table 3. Summary of the statistical results of the fitted models.

Models	Std Dev	Overall F Test p-Value	F Test (LOF) p Value	R2	Adj R2	Adeq Prec	AIC
Biogas-Quadratic	2.41	0.0012	13.8	0.87	0.81	15.4	71.9
CH4-Quadratic	0.005	0.032	4.87	0.77	0.61	6.4	−83.6
TOC-Mean	ND	ND	ND	ND	ND	ND	112
Sulfate-Quadratic	6.87	0.027	4.72	0.71	0.55	7.70	99.3
TS-Mean	ND	ND	ND	ND	ND	ND	ND
VS-Mean	ND	ND	ND	ND	ND	ND	ND
COD-Quadratic	0.01	0.003	10.7	0.84	0.76	10.9	−68.7

Adeq Prec = adequate precision, Adj = adjusted, AIC = Akaike’s information criterion, LOF = lack of fit, ND = no data, Pred = predicted, R² = coefficient of determination, Std dev = standard deviation, TOC = total organic carbon, TS = total solids and VS = volatile solids.

). The F test showed that ISR and ${\mathrm{SO}}_4{}^{2-}$ and their interaction ISR², (${\mathrm{SO}}_4{}^{2-}$)² and ISR(${\mathrm{SO}}_4{}^{2-}$) were all significant (p < 0.05) model terms for biogas yield, while ISR and its interaction with ${\mathrm{SO}}_4{}^{2-}$ (ISR(${\mathrm{SO}}_4{}^{2-}$)) were the only non-significant terms (F test, p > 0.05) on ${\mathrm{CH}}_4$ yield. The interaction of both factors ISR (${\mathrm{SO}}_4{}^{2-}$) was the only significant factor (F test, p < 0.05) on sulfate reduction whilst (${\mathrm{SO}}_4{}^{2-}$) and its interaction (${\mathrm{SO}}_4{}^{2-}$)² were the only significant factors (F test, p < 0.05) affecting COD reduction (Table 3). Nonetheless, both factors were significant in achieving the maximization of gas yields and B_o. The correlation coefficients (R²) of the models (Table 3) indicated that only 13.3%, 23.2%, 29.1%, and 16.1% of the variability in biogas yield, ${\mathrm{CH}}_4$ yield, %${\mathrm{SO}}_4{}^{2-},$ and %COD reduction was not explained by the models, respectively. The models’ adj. R² values were 0.81, 0.61, 0.55, and 0.76, respectively, suggesting moderate to good predictability of the gas yields and B_o.

In contrast, the general quadratic polynomial and linear equations did not fit the %TOC, %VS, and %TS reduction data very well (adj R² ≤ 0.28) and returned negative predicted R². This indicated that the overall mean better predicted B_o (Stat-Ease, Inc., Minneapolis, MN, USA). This is in agreement with the observation that regardless of inhibition of ${\mathrm{CH}}_4$ generation, other processes occurred. Therefore, Equations (5)–(11) were used to simulate and optimize the B_o and cumulative gas yields as plotted in Figure 2.

$$\mathrm{Biogas} \mathrm{yield} =40.4{\mathrm{ISR}}^2+0.07\left[{\mathrm{SO}}_4\right]\mathrm{ISR}-422\mathrm{ISR}-0.2\left[{\mathrm{SO}}_4\right]-834$$

(5)

$${\mathrm{CH}}_4 \mathrm{yield} =0.12+1.6\times {10}^{-8}{\left[{\mathrm{SO}}_4\right]}^2+6.0{\mathrm{ISR}}^2-0.05\mathrm{ISR}-4.2\times {10}^{-5}\left[{\mathrm{SO}}_4\right]+2.6\times {10}^{-6}\left[{\mathrm{SO}}_4\right]\mathrm{ISR}$$

(6)

$$\% {\mathrm{TOC}}_{\mathrm{reduction}}=75.1$$

(7)

$$\% {\mathrm{VS}}_{\mathrm{reduction}}=27.5$$

(8)

$$\% {\mathrm{TS}}_{\mathrm{reduction}}=27.4$$

(9)

$$\% {\mathrm{COD}}_{\mathrm{reduction}} =\frac{1}{0.39-7\times {10}^{-5}\left[{\mathrm{SO}}_4\right]-0.19\left(\mathrm{ISR}\right)}$$

(10)

$$\% {\mathrm{SO}}_4{}_{\mathrm{reduction}} =0.02\left[{\mathrm{SO}}_4\right]\mathrm{ISR}-0.011\left[{\mathrm{SO}}_4\right]-19.9\mathrm{ISR}+115$$

(11)

The cumulative ${\mathrm{CH}}_4$ yields and the average %${\mathrm{CH}}_4$ varied from 0 to 146 mL/gVS and 9.3% to 44.8%, respectively. The corresponding reduction efficiencies were ${\mathrm{SO}}_4{}^{2-}$ (49.3–85.3%), TOC (23.2–93.1%), VS (15.2–55.0%), TS (11.5–49.2%), and COD (10.8–52.0%) (Table 2). The ${\mathrm{CH}}_4$ yields were compared to the controls (inoculum only) and corrected ${\mathrm{CH}}_4$ yield reported. Based on the interest to maximize ${\mathrm{CH}}_4$ yield and B_o, the theoretical optimum operating conditions were found to be at ${\mathrm{SO}}_4{}^{2-}$ = 922 mg/L and ISR = 3.72 with a desirability of 0.65. These optimum conditions are expected to generate 361 mL biogas/gVS, 235 ML ${\mathrm{CH}}_4$/gVS and reduction efficiencies of 27.5% VS, 27.4% TS, 75.1% TOC, 75.6% ${\mathrm{SO}}_4{}^{2-}$, and 41.1% COD. This implies that the mass of sludge will be reduced by about 27% (dry mass) and ${\mathrm{SO}}_4{}^{2-}$ by 76%, with a fraction of it recovered as ${\mathrm{S}}_0$.

The ${\mathrm{CH}}_4$ yields obtained in this study (Table 2) were comparable to those obtained by Vazifehkhoran et al. [38] and higher than those reported by Agustini et al. [39] and Saxena et al. [27]. However, Achouri et al. [28] reported higher gas yields and no lag phase, presumably due to the higher dilutions (34% and 53%) using tap water, uncorrected gas yields, addition of micronutrients and use of a blend of tanyard and beamhouse effluent. Interestingly, Mpofu et al. [6] operated 0.5 L batch reactors at ${\mathrm{SO}}_4{}^{2-}$ = 494–562 mg/L, 37 ± 2 °C, pH = 7.0 ± 0.5 and ISR (=4) while codigesting TWAS and SHS (50% v/v). The authors reported biogas yields of 333–431 mL/gVS, ${\mathrm{CH}}_4$ yield of 170–215 mL ${\mathrm{CH}}_4$/gVS, 50–53% ${\mathrm{CH}}_4$ (average), and B_o of 54.4–68.5% VS, 45.5–50.7% TS and 43.2–48.2% COD. The %COD reduction achieved in this study were lower compared to other studies using ASBR and other continuous reactors (UASB, UAFFB, UAFBR, SAnMBR, and UACF) [7]. They were however in the same range with 45%, 56%, 45%, and 43%, accomplished by Achour et al. [28], Berhe et al. [15], Daryapurkar et al. [40], and Saxena et al. [27] respectively, using batch reactors.

3.3.1. Water Reuse

The resulting treated OTE met the stipulated wastewater limit values applicable for the irrigation of land with up to 50 and 500 m³/day. However, treated OTE did not meet the limit values for irrigating with 2000 m³/day in terms of SS, COD, ${\mathrm{NH}}_3$, and Cl [41]. This can be mitigated by mixing treated OTE and tap water at an optimised volumetric ratio that meets the irrigation standard. Maqboo et al. [42] concluded that irrigating with 50:50 (v/v) tap water and TWW might be a sustainable alternative for increasing vegetable growth.

3.3.2. Correlative Analysis of Variables on Methane Yield and Anaerobic Biodegradability

There was a weak linear insignificant relationship (r ≤ −0.29, (F test, p > 0.05)) between ${\mathrm{SO}}_4{}^{2-}$ and %solids reduction. Similarly, ISR did not linearly correlate with B_o (TS, COD, and TOC), average %${\mathrm{CH}}_4$ and ${\mathrm{CH}}_4$ yield (−0.16 ≤ r ≤ 0.12, (F test, p > 0.05)). There was a strong significant positive linear correlation (r = 0.71, (F test, p < 0.05)) between ISR and %VS reduction, while weak insignificant negative linear relationships (−0.29 < r ≤ −0.21, (F test, p > 0.05)) existed between ${\mathrm{SO}}_4{}^{2-}$ with %COD and %TOC reduction and ISR with biogas yield and %${\mathrm{SO}}_4{}^{2-}$ reduction (0.31 ≤ r ≤ 0.34, (F test, p > 0.05)). Mpofu et al. [10] also reported a lack of correlation (r < 0.19, (F test, p > 0.05)) between ISR and %solids reduction, strong positive correlation (r = 0.84, (F test, p < 0.05)) with gas yields and a moderate positive correlation (r = 0.46, (F test, p > 0.05)) with %COD reduction. A moderate negative linear relationship (−0.39 < r ≤ −0.53, (F test, p > 0.05)) existed between ${\mathrm{SO}}_4{}^{2-}$ with %${\mathrm{SO}}_4{}^{2-}$ reduction, average %${\mathrm{CH}}_4$, biogas and ${\mathrm{CH}}_4$ yield. These results confirm that (i) high ${\mathrm{SO}}_4{}^{2-}$ promoted sulfidogenesis over methanogenesis, which negatively affected gas yields, (ii) gas yields and ${\mathrm{SO}}_4{}^{2-}$ removal could be improved by decreasing the influent ${\mathrm{SO}}_4{}^{2-}$ to around 922 mg/L and increasing the ISR to 3.7, and (iii) the strong significant linear relationship between %${\mathrm{SO}}_4{}^{2-}$ reduction with %TS (r = 066, (F test, p < 0.05)) and %VS (r = 0.64, (F test, p < 0.05)) reduction indicated the importance of SRB in degrading hydrolysis products. This may have created a better environment for the hydrolytic bacteria.

3.4. Kinetic Study of Cumulative Methane Production

There is a lack of studies that report on the AD kinetics while treating TWW and worse for OTE. In order to evaluate the performance and the kinetics of the BMP experiment, the modified Gompertz, logistic, first order, and cone models were fitted onto the cumulative ${\mathrm{CH}}_4$ yield data (Table 3) using non-linear regression. The models displayed a perfect fit to the cumulative ${\mathrm{CH}}_4$ yields with high precision in the order Logistic > Cone > modified Gompertz > first order (Figure 3,

Table 4. Kinetic parameters and goodness of fit obtained from evaluated models.

Reactor	Model	Kinetic Parameters					Adj R2	p Value Prob > F	AIC	RMSE
$({\mathbf{SO}}_4{}^{2-}/\mathbf{ISR})$		A (mLCH4/gVS)	µm (mLCH4/gVSd)	λ (d)	K	n
R1 (1960/2.5)	Cone	11.5	ND	ND	0.043	4.92	0.965	0.44	90.5	0.46
	Logistic	11.2	0.61	14.4	ND	ND	0.956	0.41	105	0.51
	First order	13.0	0.03	ND	ND	ND	0.827	0.08	195	1.02
	Gompertz	3.55	0.69	13.5	ND	ND	0.676	0.23	235	1.38
R3 (710/2.5)	Logistic	16.7	3.63	20.9	ND	ND	0.999	0.50	99.0	0.11
	Cone	16.7	ND	ND	0.043	21.9	0.999	0.50	99.2	0.11
	Gompertz	6.14	1.76	22.7	ND	ND	0.999	0.50	99.2	0.11
	First order	28.8	0.02	ND	ND	ND	0.831	0.18	251	1.57
R4 (1335/3.0)	Logistic	26.5	1.06	12.9	ND	ND	0.955	0.45	209	1.13
	Gompertz	9.99	0.36	20.6	ND	ND	0.951	0.45	213	1.18
	Cone	28.2	ND	ND	0.04	3.25	0.949	0.46	217	1.20
	First order	47.3	0.015	ND	ND	ND	0.883	0.24	272	1.84
R5 (710/4.0)	Gompertz	51.2	4.20	2.75	ND	ND	0.996	0.50	183	0.93
	First order	139	0.172	ND	ND	ND	0.984	0.43	270	1.82
	Logistic	138	13.40	0	ND	ND	0.979	0.38	286	2.06
	Cone	146	ND	ND	0.27	1.22	0.966	0.47	319	2.65
R6 (1335/5.0)	Gompertz	15.3	1.71	7.80	ND	ND	0.968	0.45	222	1.25
	Cone	41.9	ND	ND	0.112	3.78	0.966	0.44	225	1.29
	Logistic	41.6	4.32	4.59	ND	ND	0.961	0.42	234	1.37
	First order	43.3	0.083	ND	ND	ND	0.883	0.23	306	2.40
R7 (1960/4.0)	Logistic	20.3	0.88	24.5	ND	ND	0.962	0.46	165	0.81
	Cone	21.1	ND	ND	0.03	5.60	0.960	0.47	168	0.83
	Gompertz	7.72	0.31	32.2	ND	ND	0.960	0.47	169	0.83
	First order	25.0	0.02	ND	ND	ND	0.763	0.04	285	2.04
R8; R9; R10 & R11 (1335/3.0)	Logistic	117	4.82	26.7	ND	ND	0.987	0.50	314	2.55
	Cone	124	ND	ND	0.025	5.68	0.982	0.48	337	3.04
	Gompertz	46.6	1.55	35.3	ND	ND	0.979	0.49	347	3.29
	First order	38.0	0.021	ND	ND	ND	0.437	0.03	648	4.81
R13 (665/3.0)	Logistic	129	17.5	3.12	ND	ND	0.991	0.47	265	1.75
	Gompertz	47.5	6.43	5.55	ND	ND	0.991	0.49	267	1.77
	Cone	130	ND	ND	0.15	3.44	0.990	0.50	271	1.83
	First order	132	0.125	ND	ND	ND	0.939	0.29	389	4.54

A = ultimate ${\mathrm{CH}}_4$ yield, µm = maximum CH₄ production rate (specific microbial growth rate), λ = lag phase, K = specific rate constant, n = shape factor constant, ND = no data, AIC = Akaike’s information criterion, RMSE = root mean square error, R = reactor, ${\mathrm{R}}^2$ = correlation coefficient, RT = retention time.

).

According to the statistical parameters (Table 4), the first order model was the worst predictor (Adj R² = 0.437–0.763) for most reactors that experienced a lag phase except for R5 (Adj R² = 0.984), R6 (Adj R² = 0.883) and R13 (Adj R² = 0.939) that experienced shorter lag phases of 3 to 6 days. The exponential–plateau curve displayed by the first order model accurately fits data without or with short lag phases. The first order model predicted the lowest kinetic values for maximum microbial specific growth rate (µm). In contrast, the rest of the models best fitted (0.827 ≤ Adj R² ≤ 0.999) reactors with long lag phases of 14 to 35 days as they display a sigmoidal shape with lag, exponential and stationary phases. However, the modified Gompertz model yielded the highest lag phases (λ) and significantly under predicted the ultimate ${\mathrm{CH}}_4$ yields (A) by >63%. The cone (Adj R² = 0.960–0.999) and logistic (Adj R² = 0.956–0.985) models gave better approximations of the kinetic constants.

The range of the kinetic constants: A, µm, K and λ obtained in this study were 11.2–139 mL${\mathrm{CH}}_4$/gVS, 0.171–17.5 mL${\mathrm{CH}}_4$ /gVSd, 0.025–0.27 day⁻¹ and 0–35 days, respectively. The highest K and µm were found in reactors R5 and R13 operating at lower ${\mathrm{SO}}_4{}^{2-}$ of 710 and 665 mg/L and ISR of 4 and 3, respectively. This confirms that from the range of parameters tested, higher ISR and low ${\mathrm{SO}}_4{}^{2-}$ provided the most ideal environment for proliferation of methanogens. The ${\mathrm{CH}}_4$ production μm = 0.015–17.5 mL ${\mathrm{CH}}_4$/gVSd obtained in this study were comparable to μm = 2.04–5.48 mL ${\mathrm{CH}}_4$/gVSd reported by Sri Bala Kameswari et al. [14], and μm = 0.08–5.49 reported by Mpofu et al. [19] for AD of tannery sludge. Furthermore, they were similar to μm = 6.0–18.1 mL ${\mathrm{CH}}_4$/gVSd reported by Mpofu et al. [6] while co-digesting TWAS and SHS. The K values in this study were higher than K = 0.0185–0.0239 d⁻¹ reported by Thangamani [43,44] during the AcoD of tannery sludge and solid wastes and K = 0.008–0.14 day⁻¹ reported by Mpofu et al. [6,19] during the mono and AcoD of TWAS and SHS, respectively. Generally, low K values indicate efficient AD similar to natural systems that operate under slow but steady reaction rates [45]. This study proved the advantage of liquid (wet/low solids) over solid (dry/high solids) AD and the synergistic effect of AcoD of OTE with SWW. Nonetheless, there is a need for more studies on the kinetics of AD of TWW.

In order to understand the influence of ISR and ${\mathrm{SO}}_4{}^{2-}$ on the process kinetics, reactors operating at the same ISR and same ${\mathrm{SO}}_4{}^{2-}$ were analyzed. For reactors operating at constant ISR, an increase in ${\mathrm{SO}}_4{}^{2-}$ led to a significant (ANOVA, p < 0.05) decrease in A, µm and K, and a significant (ANOVA, p < 0.05) increase λ due to the inhibitory effect of S species on methanogenesis. In contrast, an increase in ${\mathrm{SO}}_4{}^{2-}$ led to a 34% decrease in λ and a constant K while operating at lower ISR (=2.5). Moderate to strong negative linear correlation existed between ${\mathrm{SO}}_4{}^{2-}$ with µm (r = −0.74), K (r = −0.61) and A (r = −0.60). In reactors operating at near optimal ISR (=3–4), increases in ${\mathrm{SO}}_4{}^{2-}$ from 665 to 710, 710 to 1335 and 1335 to 1960 mg/L led to an increase in A and K with a decrease in µm and λ; decrease in A, K and µm with an increase in λ; and an increase in K with a decrease in A, µm, and λ respectively (Figure 4). The results demonstrated a non-monotonic relationship between process parameters and kinetics. Therefore, the optimal ${\mathrm{SO}}_4{}^{2-}$ that promoted a higher A, K, and µm and lower λ was determined to be 922 mg/L at ISR = 3.7 (Section 3.3).

Generally, AD process instability is caused by a metabolic imbalance between acidogenesis and methanogenesis and/or sulfidogenesis. In order to understand the hydrolytic-methanogenic balance, it was important to study the relationship between K and μm by converting the units of μm to d⁻¹. The K/μm ratio was >1.0 in all reactors, showing that the rate of hydrolysis rate was faster than methanogenesis rate, and the imbalance increased (r = 0.65, (F test, p < 0.05)) with ${\mathrm{SO}}_4{}^{2-}$. This consolidates the hypothesis that hydrolysis and acidogenesis proceeded faster than methanogenesis (Section 3.2.4). Thus, methanogenesis was the rate limiting step in this study.

3.5. Conclusions

Efficient AD of OTE is capable of reducing solids by up to 49% (dry mass) while recovering irrigation water, elementary sulfur, and up to 146 mL ${\mathrm{CH}}_4$/gVS. Methanogenic activity was highest (K = 13.4–17.5 and µm = 0.15–0.27) when reactors were operated at ISR ≥ 3 and/or lower ${\mathrm{SO}}_4{}^{2-}$ ≤ 710 mg/L while high ${\mathrm{SO}}_4{}^{2-}$ ≥ 1960 mg/L and ISR < 3.0 caused almost complete inhibition regardless of corresponding ISR and ${\mathrm{SO}}_4{}^{2-}$. It is acknowledged that while ${\mathrm{SO}}_4{}^{2-}$ will vary by tannery, and on a temporal basis at each tannery, concentrations measured in this study were ideal for AD and resource recovery. This presents integrated tanneries and slaughterhouses as potential biorefineries that can promote the sustainable economic development of developing countries.