Co-Digestion and Mono-Digestion of Sewage Sludge and Steam-Pretreated Winter Wheat Straw in Continuous Stirred-Tank Reactors—Nutrient Composition and Process Performance

Abstract

Wheat straw (WS) constitutes a considerable biomass resource and can be used to produce the energy carrier methane through anaerobic digestion. Due to the low contents of several nutrient elements and water in harvested WS, the use of sewage sludge (SS), consisting of primary sludge and waste-activated sludge, as a nutrient source in co-digestion with steam-pretreated wheat straw (PWS) was investigated theoretically and practically. WS was steam-pretreated, with acetic acid as the catalyst, at 190 °C for 10 min, ending with a rapid reduction in pressure. Process stability and specific methane production were studied for the mono-digestion and co-digestion of PWS and SS in continuous stirred-tank reactors for 208 days. The HRT was 22 days and the OLR 2.1 g_VS L⁻¹ d⁻¹. In co-digestion, the OLR was increased to 2.8 g_VS L⁻¹ d⁻¹ for one week. Nutrient elements were added to PWS mono-digestion at two different concentration levels. Co-digestion was stable, with a total concentration of short-chain fatty acids (SCFAs) at a safe level below 0.35 g L⁻¹ at both OLRs. The higher OLR during co-digestion would require an increase in reactor volume of 14%, compared to the mono-digestion of SS, but would increase the annual production of methane by 26%. The specific methane production levels for PWS mono-digestion, SS mono-digestion, and co-digestion were 170, 320, and 260 mL g⁻¹_VS, respectively. Co-digestion did not result in a synergistic increase in the methane yield. SCFAs accumulated in the mono-digestion of PWS when using lower levels of nutrient supplements, and the concentrations fluctuated at higher nutrient levels. The main conclusion is that PWS and SS can be co-digested with long-term process stability, without the addition of chemicals other than water and acetic acid. The specific methane production for mono-digestion of PWS was relatively low. The effect of using higher concentrations of micronutrients in PWS mono-digestion should be evaluated in future studies.

1. Introduction

The production of methane-containing biogas from waste materials and residues is advocated by the EU [1] to aid the transition from fossil-based to renewable fuels. Motives for the use of wheat straw (WS) for biogas production through anaerobic digestion (AD) have been outlined by Byrne et al., Croce et al., and Bondesson and Galbe [2,3,4]. Bondesson and Galbe [3,5] and Byrne et al. [4] have motivated the pretreatment of WS by acetic-acid-catalysed steam pretreatment prior to AD. However, WS that is harvested when the kernels are mature does not contain high enough content of many elemental nutrients, in relation to carbon, for stable and efficient mono-digestion [6,7]. For instance, carbon/nitrogen (C/N) quotients of 50–120 have been reported [7,8,9], while a quotient in the range 16–25 is desirable [2]. Furthermore, the dry matter (DM) content of harvested WS is high, typically about 90–96% [6,7,8]. The maximum DM concentration for AD in continuous stirred-tank reactors (CSTRs) is usually ~10%, and for dry fermentation it is ~35% [10]. Efficient mono-digestion of finely milled WS supplemented with mineral nutrients has been demonstrated by Nges et al. [7]. Mumme et al. [11] demonstrated the digestion of WS in a continuous solid-state reactor. However, the addition of mineral nutrients and fresh water in the mono-digestion of WS has negative consequences on both economic feasibility and environmental sustainability [12,13,14]. Both nutrients and water can be recirculated to a certain degree, but some additions will be necessary [7].

The co-digestion of two or more substrates can be used to balance the nutrient and water contents of different substrates, thereby reducing or eliminating the need for the addition of mineral nutrients [15]. Many different substrates have been investigated regarding co-digestion with wheat straw; for example, manure, food waste, seaweed, under-sown lucerne, sewage sludge, and rapeseed meal [4,8,16,17,18,19]. Large quantities of co-substrates are needed due to the abundance of WS. Manure is abundant in large parts of the world. However, in some regions, the amount of manure is too small in relation to the available WS or too dry for feasible co-digestion with WS [6,16]. Sewage sludge (SS) from municipal wastewater treatment plants (MWWTPs) is a promising substrate for co-digestion with WS as it typically contains 90–98% water and its contents of many nutrient elements are considerably higher than in WS. The nitrogen content of SS in relation to carbon is higher than optimal for AD, with a C/N quotient of about 4–12 [6,20]. Several studies have been carried out on the co-digestion of SS and WS [6,9,18,19]; however, most of these were performed using batch tests. Unfortunately, batch tests, with the use of a standard inoculum, have limitations in the investigation of synergistic effects in co-digestion [21]. Only one study on the co-digestion of WS and SS in semi-continuously fed CSTRs could be found in the literature [6]. In that study, SS was successfully used as a nutrient-supplementing co-substrate with finely milled WS [6]. Milling has a lower investment cost than steam pretreatment, but to reach the same particle size it requires considerably more energy [22,23]. Alkaline pretreatment is another efficient pretreatment method. However, the amount of alkali chemicals needed is relatively high compared to the demand of acid in acid-catalysed steam pretreatment, and the alkali must be recovered or neutralised prior to AD [4,24]. In acetic-acid-catalysed steam pretreatment, the catalyst is completely degradable in AD [4]. Although the application of steam pretreatment prior to AD has been researched for many years, studies on process performance during the mono-digestion of steam-pretreated WS in continuously or semi-continuously fed reactors are still lacking in the scientific literature.

Many MWWTPs already use anaerobic digestion in the treatment of sludge to reduce the organic content and the number of pathogens in the sludge. The value of water and nutrients in MWWTPs and the benefits of co-digestion of SS with other organic wastes and residues is now being recognised [25]. The SS used for digestion usually consists of a combination of thickened primary sludge and waste activated sludge (WAS). The amount of biogas produced by the AD of SS varies considerably [26]. Primary sludge is generally more easily degraded, and thus gives higher specific methane yields (SMYs) [27]. Thus, the proportion of primary sludge in the SS affects the efficiency of the AD process. The age of the sludge in the activated sludge process affects the degradability and biogas yield of WAS [28]; the greater the age, the lower the biogas yield. The high water content of SS limits the organic loading rate (OLR) that can be applied in a CSTR. The addition of carbon-rich WS with a higher total solid (TS) content than in SS would enable a higher OLR and an improved C/N quotient, while having only a minor effect on the HRT.

The aims of the present study were as follows:

• To investigate the process performance of PWS mono-digestion;
• To investigate the theoretical potential of balancing the nutrient composition in the co-digestion of SS and PWS based on the nutrient element/carbon ratio;
• To investigate the process performance and potential synergistic effects, in addition to the nutrient complementation, of co-digestion in relation to the mono-digestion of PWS and SS.

2. Materials and Methods

2.1. Materials

2.1.1. Substrates

The substrates used in the co-digestion experiments were SS and pretreated wheat straw (PWS). The SS was composed of primary sludge and WAS at a ratio of 7:3 based on volume (both having a density of about 1 g mL⁻¹). SS was provided by the Sjölunda Wastewater Treatment Plant, which is a full-scale plant located in Malmö, southern Sweden. The SS was prepared in two batches. SS from batch 1 was used until operational day 160, and SS from batch 2 was used for operational days 161–208. The SS was analysed with regard to pH, alkalinity, TS, and volatile solids (VS) content, elemental composition, and fibre composition. For more details regarding the sludges, see Supplementary file: Part 1.

Winter wheat straw (WS1 in Supplementary file: Table S1), harvested in autumn 2020 (Johan Håkansson Lantbruksprodukter, Dalby, Sweden), was dried in the field and was protected from precipitation during storage. It was subjected to steam pretreatment, biochemical methane potential (BMP) tests, elemental analysis, and fibre compositional analysis. After steam pretreatment it was analysed with regard to pH, content of SCFAs, alcohols, hydroxymethylfurfural (HMF), furfural, TS and VS content, elemental composition, fibre composition, and sugar release during enzymatic hydrolysis, and used in BMP tests and CSTR experiments. Elemental analysis was performed on another 12 samples of winter WS, WS2–WS13 (Supplementary file: Table S1), from various locations in Sweden, and the fibre compositions of WS2, WS4, WS6, and WS11 (see Supplementary file: Table S1) were additionally analysed to provide a reference composition. The samples were provided by Dr Karin Hamnér, Department of Soil and Environment, Swedish University of Agricultural Sciences, Uppsala, Sweden, from her own experimental work and from the saved samples from other studies of long-term cultivation trials [29]. Additional information on the reference WS samples can be found in reports on other studies in which this WS has been used (Supplementary file: Table S1).

2.1.2. Inocula

The inocula used in the CSTR experiments and the BMP test were collected from the mesophilic anaerobic digester at Sjölunda MWWTP (VA SYD, Malmö, Sweden), which was operated at 37 °C, with an average HRT of 22 days and an OLR of 2.1 g_VSL⁻¹d⁻¹. The batch of inoculum used for the CSTR experiments had a TS content of 3.11 (SD 0.01)% and a VS content of 2.10 (SD 0.00)% on a wet weight (WW) basis. The batch used for the BMP test had a TS content of 2.78 (SD 0.00)% and VS content of 2.10 (SD 0.00)% and 1.70 (SD 0.01)% on a WW basis. The content of ammonia–N and nitrate–N in the inoculum used in the CSTR experiments was 1250 mg kg⁻¹ and 0.8 mg kg⁻¹, respectively.

2.1.3. Continuous Stirred-Tank Reactors

The experiments were conducted in six 3 L glass CSTRs, each with an active volume of 2.6 L. The contents of each reactor were heated to 37 °C and mixed at 80 ± 10 rpm. Reactors, stirrers, and heating are described elsewhere [30]. Feeding was performed manually and daily as described by [30]. For additional information, see Supplementary file: Part 1.

2.2. Methods

2.2.1. Substrate Treatment

The sludge was hygienized by incubation for one hour once it had reached a temperature of at least 70 °C and was thereafter stored at 6 °C. For more details, see Supplementary file: Part 1. The WS was milled, soaked, and steam-pretreated. A fast-moving coarse mill (Retsch GmbH, Haan, Germany) was used to mill the WS. The particle size was determined by sieving 40 g of milled WS: 8% < 0.85 mm; 69% 0.85–4 mm; 21% 4–10 mm; and 2% > 10 mm. The winter wheat straw was steam-pretreated as described by Bondesson and Galbe [3], with minor modifications. Briefly, 5.4 kg of milled WS was soaked in 1% acetic acid (in cold tap water) at a WS/acetic acid solution ratio of 1:10 by weight, in a bucket with a weight on top, overnight. Next day, the soaking solution was drained off, and the solids pressed in a 25 L press (Tinkturenpressen HP5 M, Fischer Maschinen-fabrik GmbH, Burgkunstadt, Germany). In total, 90% of the liquid was removed. The resulting cake (TS content of 51.5%) was steam-pretreated at 190 °C for 10 min in a 10 L batch reactor, followed by a rapid pressure release (generally referred to as steam explosion), as described previously [31]. The PWS was frozen in aliquots and thawed prior to feeding. The PWS was analysed to determine the pH, TS, and VS content, fibre composition, sugar release during enzymatic hydrolysis, elemental composition, SCFAs, and alcohols.

2.2.2. Fibre Composition Analysis and Enzymatic Hydrolysis

The WS was analysed to determine water-extractive and ethanol-extractive compounds, structural carbohydrates, lignin, and ash according to NREL protocols [32,33]. The PWS was separated into a solid and a liquid fraction. The solid fraction was analysed to determine the structural carbohydrates, lignin, and ash, while the contents of oligomeric and monomeric sugars, SCFAs, and sugar degradation products were determined in the liquid fraction according to NREL protocols [32,33]. PWS was enzymatically hydrolysed according to a NREL protocol [34], with the modification of using a higher initial water insoluble solids (WIS) content of 2%. Cellic^® Ctec2 (Novozymes A/S, Bagsvaerd, Denmark). An enzyme loading of 15 FPU g_WIS⁻¹ was used. The sugar yields were calculated as the share of sugar polymers and oligomers in PWS.

2.2.3. Estimation of the Microbial Nutrient Demands

The nutrient demand (chemical element demand) for microbial metabolism was estimated theoretically. The mass of nutrients was calculated for the formation of microbial biomass from 12% of the carbon in the substrate, based on the biomass formation described by McCarty [35], and the carbon/element ratios in the microbial granules of an AD process presented by Byrne et al. [4].

2.2.4. Biochemical Methane Potential Tests

BMP tests were performed on milled WS, PWS, SS batch 1, and SS batch 2 using an automated methane potential test system (AMPTS II, Bioprocess Control AB, Lund, Sweden), according to the manufacturer’s handbook and recommendations by Holliger et al. [36,37]. The incubation temperature was 37 °C. The inoculum/substrate ratio used was 2:1, apart from milled WS, where the ratio was 1:1.7. The total mass per test replicate was 400 g except in the case of milled WS, where it was 408 g. Cellulose powder (Avicel PH-101, Fluka Biochemika, Switzerland) with a particle size of ~50 µm was used as the control material. Triplicates were used for substrate materials and controls. The headspaces of the bottles were flushed with nitrogen gas prior to starting.

2.2.5. Start-Up and Operation of AD in CSTRs

After a start-up period (see Supplementary file: Part 1), AD was run in duplicate reactors: mono-digestion of SS (1A and 1B), co-digestion of PWS:SS (2A and 2B), and mono-digestion of PWS (3A and 3B), as described in

Table 1. HRT, OLR, and the amount of each feeding component (expressed as WW) during operational periods 1 and 2.

	SS	PWS:SS	PWS:SS	PWS
Reactors	1A, 1B	2A, 2B	2A, 2B	3A, 3B
OLR (gVS L−1 d−1)	2.1 ± 0.1	2.1 ± 0.1	2.8 ± 0.1	2.1 ± 0.1
SS	100%	65%	88.1%	-
PWS	-	8.78%	11.9%	25.1%
Macronutrient supplements	-	-	-	38.04%
Micronutrient supplements	-	-	-	0 (23.67%) 2
Water 1	-	26.21%	-	36.86% (13.19%) 2

¹ Deionized during operational period 1 and MilliQ-filtered during operational period 2, added to adjust the HRT. ² Additions on 2 out of 3 days are given before the parentheses and additions every third day are given in parentheses.

. The PWS:SS ratio in co-digestion was 35:65 based on VS from operational day 5. The experiments were run with feeding in two operation periods: operational period 1, spanning 122 days, and operational period 2, spanning 103 days (

Table 2. Overview of important periods and changes during the CSTR experiments.

Operational Day	Comments
−12	Mono-digestion of SS in all six reactors.
0	Reactors 2A, 2B, 3A, and 3B were shifted from SS mono-digestion to co-digestion. Start of operational period 1.
25	Determination of SCFAs prior to feeding, hourly for 7 h after feeding, and at 22 and 24 h after feeding.
33	Reactors 3A and 3B were shifted from co-digestion to PWS mono-digestion.
70	The IA/PA threshold quotient was increased from 0.3 to 0.40.
92	Determination of SCFAs prior to feeding, hourly for 9 h after feeding, and at 24 h after feeding.
81–93	Reference period 1 for reactors 1A, 1B, 2A, and 2B. The gas production of reactor 2B was excluded from SMP calculations and the determination of residual methane determination due to gas leakage.
94–102	The OLR was temporarily increased from 2.1 to 2.8 gVS L−1 d−1 for reactors 2A and 2B.
98	The IA/PA threshold quotient was increased from 0.4 to 0.45.
104–108	Reduced feeding to reactor 3A (only fed day 107) due to high IA/PA quotient.
110–114	Reference period 1 for reactors 3A and 3B. The gas production for reactor 3A was excluded due to reduced feeding at the end of operational period 1.
122	End of operational period 1.
122–124	There was a break in feeding for 50 days between operational periods 1 and 2. Residual methane production was measured. Reactors 2B and 3B were excluded due to gas leakages. For reactor 3B, the leakage occurred 5 days after operational period 1.
124	Start of operational period 2.
123–136	Start-up period with reduced feeding to reactors 1A, 1B, 2A, and 2B.
123–139	Start-up period with reduced feeding to reactors 1A, 1B, 2A, and 2B.
123–160	Start-up period with reduced feeding to reactor 3A.
123–170	Start-up period with reduced feeding to reactor 3B.
199–200	Remediation of leakages by improved sealing.
201–208	Reference period 2 for reactors 1A, 1B, 2A, and 2B.
208	End of operational period 2 for reactors 1A, 1B, 2A, and 2B. Residual methane yields from these four reactors were excluded due to a higher TS content of the SS used on days 209–221 than the SS used on days 1–208.
219–227	Reference period 2 for reactors 3A and 3B.
224	Determination of SCFAs in reactors 3A and 3B prior to and hourly for 9 h after feeding, and at 20 and 24 h after feeding.
227	End of operational period 2 for reactors 3A and 3B. Residual methane production was measured for reactors 3A and 3B for 58 days.

). The concentrations of supplemented nutrient elements in PWS mono-digestion were higher in operational period 2 than in operational period 1; see Section 2.2.6. Each operational period was followed by a period without feeding, but heating, stirring, and gas recording continued to determine the residual methane yield. In each operational period, a reference period was defined after the HRT had passed 3 times for each substrate (Table 2). Compositional analyses were performed on digestate samples taken during the reference periods, and the specific methane production (SMP) from these periods were used for comparison.

The HRT was 22 days (when fed 100%) during both operational periods. The HRT was based on the average HRT at Sjölunda MWWTP (VA SYD, Malmö, Sweden) during spring 2020 [38]. The distributions between the feeding components for the six reactors are described in Table 2. The OLR was 2.1 ± 0.1 g_VS L⁻¹ d⁻¹ (when fed 100%) during the entire experiment, except for days 94–102. During these days, the OLR was 2.8 g_VS L⁻¹ d⁻¹ for reactors 2A and 2B. At the start-up of operational period 2, feeding was increased step-wise (see Supplementary file: Part 1). The VS degradation was calculated based on mass balances according to Switzenbaum et al. [39]. Residual methane produced after each operational period was measured from 24 h after the last feeding until n days had passed. The methane was divided by the mass of VS fed over 22 days and expressed as a fraction of the SMP from reference period 1 or 2.

Normally, pH, alkalinity, and SCFAs were measured every third day in samples collected before feeding, and more frequently when the quotient of intermediate alkalinity/partial alkalinity (IA/PA) was above a certain threshold value. This quotient can be used as an indicator of the risk of a pH drop in the reactor in the near future, and the value is substrate-dependent [40]. The threshold value was initially set to 0.3 [40], but was increased to 0.40 after day 70, and to 0.45 from day 98, since the concentrations of SCFAs at quotients of 0.3 and 0.4 were below the concentrations likely to be inhibiting at an acidic pH [41]. When the IA/PA quotient increased above the threshold value, feeding was suspended on that day in operational period 1 and reduced with 25% in operational period 2, with a few exceptions (see Supplementary file: Part 1). The IA/PA quotient was measured again the next day prior to feeding.

2.2.6. Nutrient Supplements

The intention was to provide the same concentrations of macronutrients and micronutrients to PWS mono-digestion (reactors 3A and 3B) as those present in co-digestion (reactors 2A and 2B). However, the addition of nutrients was lower than intended during both operational periods due to a misunderstanding. During operational period 1, the addition of all added nutrients, except N, were one-fifth to one-twentieth of that intended. During operational period 2, the addition of Mg was half that intended, and the micronutrients added were 25% of the intended. See Supplementary file: Part 1 for information on the total concentrations of elements in the substrate and the supplements in the mono-digestion of PWS during operational periods 1 and 2, expressed as the concentration of elements in co-digestion. Macronutrient supplements and water were fed together with the PWS to reactors 3A and 3B, according to

Table 3. The supplementary macro- and micronutrients fed in the mono-digestion of PWS during operational periods 1 and 2. Macronutrients were supplemented daily and micronutrients every third day. Values are given on a daily basis.

Compound	Operational Period 1		Operational Period 2
	Stock Solution Concentration (g L−1)	Content of Element in Feed (mg kg−1) 1	Stock Solution Concentration (g L−1)	Content of Element in Feed (mg kg−1) 1
Macronutrient solutions
(NH)2CO	47.880 2	1723 (N)	40.296 2	1874 (N)
Na2HPO4	2.458 2	40.8 (P)	41.982 2	705 (P)
(NH4)2SO4	1.51 2	27.8 (S)	27.53 2	508 (S)
CaCl2	1.25 2	34.4 (Ca)	19.91 2	547 (Ca)
MgCl2	0.428 2	8.31 (Mg)	3.578 3	34.7 (Mg)
KH2PO4		0	0.938 3	10.3 (K)
Micronutrient solutions 4
FeCl2·4H2O	4.64	103 (Fe)	14.10	312 (Fe)
CuCl2·2H2O	3.16 × 10−2	9.28 × 10−2 (Cu)	1.02 × 10−1	3.00 (Cu)
ZnCl2	2.39 × 10−4	9.06 × 10−3 (Zn)	7.92 × 10−2	3.00 (Zn)
MnCl2·4H2O	1.35 × 10−4	2.96 × 10−3 (Mn)	5.08 × 10−2	1.11 (Mn)
NiCl2·6H2O	1.09 × 10−5	2.10 × 10−4 (Ni)	3.69 × 10−3	7.19 × 10−2 (Ni)
NaSeO3·5H2O	6.76 × 10−8	1.61 × 10−6 (Se)	2.99 × 10−4	7.07 × 10−3 (Se)
(NH4)6Mo7O24·4H2O	2.90 × 10−8	1.15 × 10−6 (Mo)	2.89 × 10−4	1.24 × 10−2 (Mo)
CoCl2·6H2O	5.31 × 10−7	9.93 × 10−6 (Co)	1.57 × 10−3	3.07 × 10−2 (Co)
Na2O4W·2H2O	1.58 × 10−7	7.07 × 10−6 (W)	5.40 × 10−4	2.37 × 10−2 (W)
Na2HPO4 NaSeO3·5H2O Na2O4W·2H2O	See above.	60.6 (Na)	See above.	1046 (Na)

¹ Additions of nutrient solutions were calculated based on concentration and added based on weight, assuming a density of 1 g mL⁻¹. The actual densities varied from 1.00 to 1.04 g mL⁻¹. ² 8.99 g of each solution was added daily. ³ 4.50 g of each solution was added daily. ⁴ A total of 27.97 g of a combined stock solution, containing all the micronutrients, was added every third day.

. Micronutrient supplements were fed daily for the first 22 days of PWS mono-digestion. Micronutrients were combined and added as one solution. From the 23rd day of PWS mono-digestion (operational day 58), micronutrients were fed to reactors 3A and 3B every third day (according to Table 3), two hours prior to feeding with PWS and the macronutrients, based on recommendations by Takashima and Speece [42]. Extra micronutrients, 12 times the daily dose, were added to reactors 3A and 3B on operational day 135.

2.2.7. Determination of SCFA over 24 h

The concentrations of acetic acid and propionic acid prior to and hourly after feeding (at 0 h) were determined for all reactors on day 25 (prior to starting mono-digestion of PWS) and on day 92, and in reactors 3A and 3B on day 224. On day 25, hourly sampling was performed up to 7 h after feeding and once after 22 h. On day 92, the hourly measurements were extended to 9 h after feeding, and sampling was performed once after 24 h. On day 224, measurements were made up to 9 h after feeding, and at 20 h and at 24 h after feeding.

2.3. Analytical Methods

2.3.1. Gas Volume Determination and Leakage Detection

The volume of gas produced was measured with the AMPTS II after stripping from carbon dioxide using 3 M NaOH as described in the AMPTS protocol (Bioprocess control AB, Lund, Sweden). Larger bottles (1 L) than standard (100 mL), containing 800 mL NaOH solution, were used as carbon dioxide traps for gas from CSTRs. The gas volumes recorded by the AMPTS were reported as dry methane gas, normalised [36] to 0 °C and 101.325 kPa. The software function for the elimination of gas overestimation was applied for the BMP test. The detection of gas leakages in CSTR experiments is described in Supplementary file: Part 1. Gas leakages were remedied on days 199 to 200 by improved sealing.

2.3.2. pH, Alkalinity, and SCFAs

Partial and total alkalinity were determined according to Jenkins et al. [43] and intermediate alkalinity was calculated according to Ripley et al. [40]. The digestate samples were centrifuged at 5000 rpm for 10 min (Thermo Scientific™ Labofuge™ 200 Centrifuge, Waltham, MA, USA) prior to the determination of alkalinity. Measurements of TS and VS were performed according to standard methods [44]. For an analysis of volatile compounds in PWS, triplicate samples were prepared by steeping PWS in deionized water, based on Porter and Murray [45]. The TS and VS contents of PWS were corrected for the volatilization of SCFAs and alcohols, according to correction factors developed for grass silage [45] as suggested by Kreuger et al. [46]. For the analysis of SCFAs (formic, acetic, propionic, butyric, isobutyric, valeric, caproic, lactic, and succinic acids), glycerol, methanol and ethanol, and sugar, the refractive index spectra after high-performance liquid chromatography (HPLC)-separation was used, and for an analysis of hydroxymethylfurfural and furfural, the ultraviolet spectra after HPLC-separation were used. For further information, see Supplementary file: Part 1.

2.3.3. Nutrient Elements and Ions

Elements and ions were analysed with the same instruments and methods as described by Byrne et al. [4], with the difference that Zn was analysed with inductively coupled plasma optical emission spectroscopy (ICP-EOS) in the current study. In short, phosphate and chloride were analysed with ion chromatography, total C and total N with an CN element analyser, the elements B, Ca, Cu, Fe, K, Mg, Mn, Na, P, S, and Zn were analysed with ICP-OES, Co, Cr, Mo, Ni, Se, and W were analysed with inductively coupled plasma mass spectroscopy (ICP-MS) and total ammoniacal nitrogen (TAN) was analysed with a flow injection analyser. The concentration of free ammonia nitrogen was calculated with the assumption of an ideal equilibrium solution [47].

2.4. Statistical Analysis

Statistical analyses were performed with the statistical software Prism (Prism 5 for Mac OS X, version 5.0b; GraphPad Software Inc., La Jolla, CA, USA). An analysis of outliers was performed using Dixon’s test when the sample size was 3 to 7, and with Grubb’s test when the sample size was larger than 7 [48]. A nested analysis of variance (ANOVA) was used to analyse the influence of substrates and reactors on the methane yield, based on recommendations by Hofmann et al. [49]. The SMP for all substrates for reference period 2 and for SS only for reference period 1 were included. The daily SMP was considered as random statistical samples; n = 16, 16, and 15 per reactor for the SMP for SS, co-digestion, and PWS, respectively, for reference period 1. For reference period 2, n = 8 per reactor. The SMP for reactors 2B and 3A for reference period 1 were not included in the analyses due to gas leakage and to reduced feeding, respectively.

A one-way ANOVA and Tukey’s multiple comparison test were used for comparison of the SMY for BMP tests, and the SMP for CSTRs from reference periods 1 and 2. The accumulated SMYs from individual test flasks after 22 days in the BMP tests were considered as random statistical samples; n = 3 for BMP tests. The SMP for CSTR experiments was treated as described above. A significance level of 5% was used in all analyses.

3. Results and Discussion

3.1. Substrate Characteristics

The contents of TS, VS, total ammonia nitrogen (TAN), and nitrate–N of sludges are given in

Table 4. Content of TS, VS, total ammonia nitrogen, and nitrate–N of substrates, on wet weight basis. SD is given in parenthesis.

	SS Batch 1	SS Batch 2	Primary Sludge	WAS
TS, % of WW	5.44 (0.1)	4.75 (0.48)	6.25 (0.1)	4.44 (0.03)
VS, % of TS	81.6 (1.6)	79.3 (0.2)	82.8 (0.1)	79.6 (0.2)
n TS and VS measurements	33	12	4	4
TAN (mg kg−1)	274	ND	120	752
NO3-N (mg kg−1)	1.1	ND	0.7	0.8

WW—wet weight. TAN—total ammonia nitrogen. ND—not determined.

. The TS content of primary sludge was higher than the TS content of WAS. The pH and total alkalinity of SS batch 1 were pH 5.98 and 6900 mg CaCO₃ L⁻¹.

The contents of the nutrient elements in substrates and inoculum on a total solid basis are presented in

Table 5. Contents of nutrient elements in substrates and inoculum and the average content and SD of 13 WS samples (Supplementary file: Table S1), expressed as mg kg_TS⁻¹ for all compounds except C and N, which are expressed as g kg_TS⁻¹.

Element	WS1	Average of WS1–WS13 1	SD and CV, WS1–WS13 1	PWS	Primary Sludge 2	WAS 2	SS 2	AD Inoculum
Total C	435	430	20.5 (5%)	472	437	405	430	33
Total N 3	6.3	4.5	1.8 (41%)	6.1	49.6	63.7	53.8	41.6
Ca	2892	2185	617 (28%)	1591	15,464	15,798	15,972	22,368
Cu	3.11	2.47	0.65 (26%)	3.42	335	301	328	438
Cl	750	467	270 (57%)	148	307	335	321	780
Fe	28.3	31.0	22.5 (76%)	39.8	30,693	39,141	33,151	52,293
K	10,661	8121	2210 (27%)	2286	1618	4082	2291	3948
Mg	846	699	152 (22%)	244	1865	2650	2094	2979
Mn	25.4	15.4	11.2 (73%)	10.0	131	121	128	178
Na	127	80.0	36.1 (43%)	49.7	2581	3865	3019	4777
P	490	412	110 (7%)	235	16,818	25,726	19,085	27,725
S	1494	832	290 (27%)	607	16,665	9571	14,659	18,356
Si	722	627	117 (19%)	ND	ND	ND	ND	ND
B	8.31	6.54	1.55 (26%)	18.02	BDL	BDL	BDL	BDL
Cr	6.46	0.548	0.144 (23%)	0.532	4.52	6.22	8.14	10.47
Co	0.077	0.020	0.019 (43%)	0.038	2.19	3.49	4.14	4.99
Ni	3.90	0.560	0.226 (40%)	0.272	5.12	6.95	10.3	11.5
Zn	8.17	3.31	2.07 (46%)	6.01	278	268	289	472
Se	0.071	0.058	0.014 (24%)	BDL	1.24	2.52	2.65	2.61
Mo	0.383	0.642	0.318 (50%)	0.243	1.99	3.72	3.96	5.66
W	BDL	BDL	ND	BDL	0.957	2.70	2.36	5.14

BDL—below detection limit, CV—coefficient of variance, ND—not determined, SD—standard deviation. ¹ CV is given in parenthesis. n = 13 except for Cr and Ni where n = 12 due to exclusion of outliers (sample WS1) and n = 3 for Se due to 10 samples with concentrations below the detection limit. ² Batch 1. For SS, measurements were made on two samples, taken at different time points. ³ Determination of total nitrogen includes a drying step where part of the TAN is lost.

. The average composition and SD of 13 WS samples (including the one used in this study) from various locations in the southern half of Sweden are included for comparison. Contents of nutrient elements in individual WS samples are presented in Supplementary file: Table S2. The contents of Cr and Ni in WS1 were identified as outliers, and were excluded from the average of all samples. In PWS, the contents of Cr and Ni were lower than in WS1. The Zn content of WS1 was also significantly higher than the content of most other WS samples. However, the content of Zn in WS3 was also higher than most other WS samples and therefore WS1 does not fulfil the criteria of an outlier (Table 5). Arthur et al. [50] report 2–6-times higher content of Mn, Mo, and Zn, 18 times higher content of Fe, and 59 times higher content of Co in one batch of wheat straw cultivated in Germany and used for long-term AD.

The fibre compositions of WS1 and four other samples of WS were not significantly different; see

Table 6. Content of extractives and composition of fibres of WS1, and the averages of WS2, WS4, WS6, and WS11 (Supplementary file: Table S1) and of sludges as percentage of total solids. BC—below calibrated concentration.

	WS1	Average of 4 WS-Samples	SD	SS	Primary Sludge	WAS
Glucan	38.6	37.4	2.3	13.3	14.7	BC
Xylan	21.2	20.9	1.6	BC	BC	BC
Galactan	BC	BC	BC	BC	BC	BC
Arabinan	3.4	BC	BC	BC	BC	BC
Mannan	BC	BC	BC	BC	BC	BC
AIL	16.0	14.5	0.8	13.5	9.9	8.0
ASL	5.0	5.9	0.8	3.0	3.0	3.4
Lignin ash	0.2	0.6	1.0	2.8	2.7	1.9
Water extractives	13.0	13.5	1.8	23.0	23.9	30.1
Ethanol extractives	2.1	2.4	0.3	17.0	18.1	17.4
Sum	99.31	95.19	8.56	72.71	72.23	60.66

AIL—acid-insoluble lignin, ASL—acid-soluble lignin, SD—standard deviation.

. The TS content of WS1 was 96.4%. The pH of PWS was pH 3.64. The fibre composition and the contents of sugars, SCFAs, glycerol, ammonia–N, HMF, and furfural in PWS are shown in

Table 7. Composition of PWS on wet-weight and TS basis.

	Fibre (% of WW)	Monomers (% of WW)	Oligomers (% of WW)	Sum (% of WW)	Sum (% of TS)
Glucan/glucose	9.16	0.11	0.31	9.58	46.1
Xylan/xylose	0.99	0.68	3.09	4.76	22.9
Galactan/galactose		0.04	0.07	0.11	0.5
Arabinan/arabinose		0.25	0.23	0.48	2.3
Mannan/mannose
Formic acid		0.14		0.14	0.6
Acetic acid		0.62		0.62	3.0
Lactic acid		0.39		0.39	1.9
Glycerol		0.02		0.02	0.1
HMF		0.02		0.02	0.1
Furfural		0.20		0.20	0.9
Ammonia-N				0.016	0.003
AIL	3.96			3.96	19.0
ASL	0.40			0.40	1.9
Lignin ash	0.45			0.45	2.2
Sum				21.12	101.6
WIS				14.35 (0.16) 1
TS un-corrected				20.07 (0.65) 1
TS corrected				20.79 (0.68) 1
VS corrected				19.89 (0.55) 1	95.69 (0.10) 1
Total ash				0.90 (0.03) 1

AIL—acid-insoluble lignin, ASL—acid-soluble lignin, TS–total solids, WIS—water-insoluble solids, WW–wet weight. ¹ SD is given in parentheses; n = 2 for WIS content; n = 21 for TS, VS, and ash contents.

. The content of nitrate–N was 1% of ammonia–N in PWS. The sugar yields from enzymatic hydrolysis of sugar polymers and oligomers in PWS were 82.5% (SD 1.4%) of glucan and glucose oligomers and 65.2% (SD 1.0%) of xylan and xylose oligomers. The sugar degradation products HMF and furfural were found in concentrations of 0.2 g kg⁻¹ and 2.0 g kg⁻¹ in the wet PWS (Table 7). Both compounds were found to be inhibitory at concentrations of 0.8 g L⁻¹ and to completely inhibit methanogenesis at 2 g L⁻¹ [51]. However, the concentrations in the AD reactor will be lower than in the substrate in a continuous or semi-continuous process due to degradation during AD; see Section 3.3.1.

3.2. Nutrient Composition and Microbial Nutrient Demand

The elements present in SS were above the estimated microbial demand (Figure 1). The full-scale digestion of SS at the Sjölunda Wastewater Treatment Plant from which the SS and inoculum originated demonstrates sufficient nutrient availability for stable digestion, providing a SMP of 310–360 mL g_VS⁻¹ at the same HRT and OLR as applied in the current study [38].

Theoretical analysis of the elemental composition and estimated microbial nutrient demand showed that PWS had insufficient contents of the elements N, P, S, Na, Fe, Cu, Co, Mo, Ni, Se, Zn, and W in relation to C for mono-digestion, and that co-digestion with 65% SS (by VS) would provide the necessary amounts. These included the Co, P, and W, which were classified as critical raw materials by the EU in 2023 [14].

The PWS was found to have a lower content of all analysed nutrient elements except Ca, K, Mg, and Mn than needed for the estimated microbial demand (Figure 1). The content of K and several other elements (based on TS) were higher in raw wheat straw (WS1) than in PWS; see Table 5. In the present experiments, a significant share of these elements was solubilised and lost during the soaking step prior to steam pretreatment, as quantified and presented by Tosi [52]. It would be possible to recycle the soaking liquid in a full-scale plant so the elements could be utilised in the microbial process. The nutrient contents of the WS used in the digestion experiments (WS1 and PWS) are compared with the average content of 13 WS samples (including WS1) in Table 5. The content of Cr and Ni was significantly higher in WS1 than in the other WS samples. This could be a result of contamination from the grinding equipment. The other WS samples were ground in another place, another year, and likely with different equipment.

A PWS:SS ratio of 35:65 based on VS was chosen for co-digestion. The elements present in this combination were above the estimated microbial demand. The content of nutrient elements in SS would support a higher share of WS based on the stoichiometric demand for microbial biomass formation. However, higher quantities could be needed due to the limited availability of the elements to microorganisms [53]. According to a review by Demirel and Scherer [53] on the requirements of micronutrients in AD, the concentrations in co-digestion would be sufficient for AD according to some studies, while other studies show that higher concentrations of Co, Mo, Ni, and Se are stimulatory. The nutrient supplements applied in the study by Nges et al. [7] resulted in stable mono-digestion of milled WS and high conversion to methane, 300 mL g_VS⁻¹, at 30 days HRT and an OLR of 3 g_VS L⁻¹ d⁻¹. The macronutrient/carbon ratios in co-digestion in the current study were equal to or above that applied by Nges et al. [7] for all macronutrients (Mg was not included in their analysis). In the study by Nges et al. [7] the concentrations of Ni, Co, Mo, and W were considerably higher in the digestate than in the feed, possibly due to the use of reactors made of stainless steel. Therefore, a comparison with concentrations in the digestate, rather than with the feed in their study, is relevant. In the present study, the feed concentrations of Ni, Co, Mo, and W in co-digestion were 49%, 18%, 1%, and 4%, respectively, of those in the digestate in one treatment without recycling and an OLR of 3 g_VS L⁻¹ d⁻¹ in the study by Nges et al. [7]. According to a study by Arthur et al. [50], the addition of 0.15 mg L⁻¹ of Ni and W (and ethylenediaminetetraacetic acid) resulted in a 64% increase (to 242 mL g_VS⁻¹) in the SMP of milled WS, compared to supplementation with macronutrients only, in long-term AD in CSTRs. The WS used by Arthur et al. [50] had a higher content of several micronutrients than the WS samples analysed in the current study; see Section 3.1.

Most nutrient elements in the feed for the mono-digestion of PWS during operational period 1 were lower than the estimated microbial demand. Most nutrient elements in the feed for the mono-digestion of PWS during operational period 2 were above the estimated microbial demand, with the exception of Mo, Ni, Se, and Zn, which were 30%, 10%, 50%, and 10% lower than the estimated microbial demand. The content of Se might be underestimated since the concentration in PWS, but not in WS1, was below the detection limit and thereby counted as 0 mg L⁻¹ (Table 5). The macronutrient/carbon ratios in the feed for the mono-digestion of PWS during operational period 2 was above that used in the study by Nges et al. [7] for all macronutrients except K. Nges et al. did not add Se, but nevertheless, the process continued for 2.4 times the HRT at an OLR of 2 and at an OLR of 3. The concentration of SCFAs was about 0.4 g L⁻¹ at an OLR of 3 in the study by Nges et al. [7].

At the end of operational period 1, the nutrient concentrations of most elements were lower than the estimated microbial demand in reactors 3A and 3B, and were thus also lower at the beginning of operational period 2. During operational period 2, all elements except Mg and micronutrients were added at the same concentrations in the feed as in the co-digestion experiments. Extra micronutrients were added to reactors 3A and 3B 12 days into operational period 2 (day 135) to increase the nutrient concentrations in the reactors more quickly than with daily feeding.

3.3. Process Variables in the Liquid of CSTR Experiments

3.3.1. Short-Chain Fatty Acids and Inhibitors in SS Mono-Digestion and Co-Digestion

It was experimentally confirmed that the nutrients in the theoretically calculated combination of 35% WS and 65% SS could support the stable co-digestion of PWS and SS at an OLR of 2.1 g_VS L⁻¹d⁻¹ and a HRT of 22 days for 208 days, without any supplemented nutrients. The concentrations of short-chain fatty acids were low (Figure 2 and Figure 3), as described in more detail below. Anti-foam agents were only added once. Peng et al. [6] added anti-foam daily for preventive purposes to the co-digestion of WS and SS, and it has now been demonstrated to be superfluous. Stable AD with low concentrations of SCFAs was also demonstrated at a higher OLR of 2.8 g_VS L⁻¹ d⁻¹ with 22 days of HRT during a 7-day test (Figure 2 and Figure 3). It is highly probable that long-term stability could be achieved under these conditions since the ratio of PWS to SS was the same as that applied with the lower OLR for 208 days, and there was no indication of inhibition in the form of increased concentrations of SCFAs at the higher OLR. However, this remains to be demonstrated. Co-digestion at this OLR would require only a 14% larger reactor volume than the mono-digestion of SS, while increasing the annual methane production by 26%.

During an initial period of 12 days, all reactors were fed with SS. When the feed to four of the reactors (2A, 2B, 3A, and 3B) was changed to PWS:SS (defined as operational day 0), a notable increase in acetic acid, and occasionally propionic acid, was seen in three of the four reactors, during the first two to three weeks (Figure 2). A brief overflow occurred in reactor 3B day 15, and anti-foam agent was added to this reactor on this single occasion. From day 22 until the end of operational period 1, the level of SCFAs in the mono-digestion of SS and co-digestion was below 0.35 mg L⁻¹ and dominated by acetic acid. No foaming was noted. This stable operation included one week with a higher OLR, i.e., 2.8 g_VS L⁻¹ d⁻¹, in the co-digestion reactors (days 94–102).

The SCFAs were more closely monitored over the course of 24 h after feeding operational days 25 and 92 (Figure 3a,b). Interestingly, the acid concentration profiles over 24 h after feeding were very similar in the two reactors with mono-digestion of SS and for the four reactors with co-digestion on operational day 25 (Figure 3a). Additionally, the acid concentration profiles were very similar for the mono-digestion of SS and co-digestion when the OLR was increased from 2.1 to 2.8 g_VS L⁻¹ d⁻¹ on operational day 92 (Figure 3b). The concentration of SCFAs was almost as low 8 h after feeding as 24 h after feeding in the mono-digestion of SS and co-digestion with increased OLR.

After a break in feeding of 49 days, the experiments were continued in operational period 2 with an OLR of 50% of the target OLR, which was gradually increased to 75% and 100% of the target OLR (Figure 2). The concentrations of SCFAs were below 50 mg/L in all reactors at start of operational period 2. SCFA concentrations were considerably higher in SS mono-digestion during the first two weeks than during the start-up in operational period 1. The highest total concentration was 2.3 g L⁻¹, including formic, acetic, propionic, butyric, iso-butyric, valeric, and iso-valeric acid. Interestingly, the concentration of total SCFAs during the start-up of operational period 2 was considerably lower in the co-digestion (less than 0.3 mg L⁻¹) than in the mono-digestion of SS. However, the concentration of propionic acid was higher in reactors 2A and 2B than in the same reactors during start-up in operational period 1 (Figure 2). The concentration then decreased in reactors 1A and 1B to below 0.3 mg L⁻¹ 15 days into operational period 2, and remained so until the end of the experiment. The concentrations of SCFAs in reactors 2A and 2B remained low throughout operational period 2. Both the mono-digestion of SS and co-digestion were stable under long-term digestion for 208 days.

3.3.2. Short-Chain Fatty Acids and Inhibitors in PWS Mono-Digestion

To the best of our knowledge, this is the first published study in which steam-pretreated WS has been mono-digested semi-continuously in CSTRs. When the mono-digestion of PWS was started in reactors 3A and 3B on day 35, the concentration of acetic acid was low (<0.2 g L⁻¹) and the concentration of propionic acid was very low during the following 40–50 days (Figure 2). Thereafter, the concentrations of acetic and propionic acid increased gradually. When the SCFAs and the inhibitors HMF and furfural were more closely monitored over the course of 24 h after feeding on operational day 92 (Figure 3b), the concentration of SCFAs, especially acetic acid, in PWS mono-digestion, was higher than in the other reactors. Furthermore, the acetic acid concentration did not decrease as much in 24 h as in the other reactors. The concentrations of HMF and furfural were below the calibrated concentration of 0.2 g L⁻¹ for all substrates, and below concentrations shown to be inhibitory to hydrolysis and methanogenesis [51] (Figure 2 and Figure 3). The IA/PA quotient increased above 0.45 in reactor 3A on day 104 and in reactor 3B on day 116. The nutrients added during operational period 1 were thus not sufficient to support a stable process. Our results confirm the results presented by Peng et al. [6], who reported stable co-digestion of WS and SS and acidification in the mono-digestion of WS without the supplementation of nutrient elements [6].

SCFA concentrations were considerably higher in PWS mono-digestion during the first two weeks of operational period 2 than during the start-up in operational period 1. The SCFA concentrations in PWS mono-digestion at the beginning of operational period 2 did not decrease as quickly as in SS mono-digestion (Figure 2). Despite a higher concentration of nutrient supplements in the feed during this period than during operational period 1, it takes time for the concentrations to change in the reactor. Since the SCFAs decreased considerably more slowly in PWS mono-digestion than in SS mono-digestion, when the OLR was reduced, a deficiency in micronutrients was suspected [54] (Figure 2). A dose of 12 times the daily micronutrient dose was therefore added on day 135 to PWS mono-digestion to bring about a more rapid increase in the micronutrient concentrations in the reactors than through daily feeding. Around day 141, the concentration of SCFAs had decreased to a low level in PWS mono-digestion, and the OLR was gradually increased again. Reactors with PWS mono-digestion were fed full OLR from days 161 and 172, respectively, and the IA/PA quotient remained below 0.45 in both reactors until the end of the experiment on day 228. The concentrations of SCFAs varied in these reactors. A decreasing trend was observed at the end of the experiment, and not an increasing and deteriorating trend. The average IA/PA quotient during this period of full OLR was 0.35 and 0.34 in reactors 3A and 3B, respectively. The maximum IA/PA quotient during the same period was 0.43 and 0.40 for reactors 3A and 3B, respectively. These values indicate that the process was working but had an elevated risk for acidification. The stability of the long-term operation was not studied.

The fluctuating concentrations of SCFAs in the mono-digestion of PWS in operational period 2 could be a result of too-low concentrations of micronutrients. The long-term stability was not studied. The concentrations of Mo of 0.03 mg L⁻¹ in feed was lower than the estimated demand for microbial biomass formation (0.036 mg L⁻¹) and critically low according to Lebuhn et al. [54]. However, the concentration in the effluent (0.08 mg L⁻¹) was higher than the estimated demand for microbial biomass formation. The concentration of Co (0.03 mg L⁻¹ in and out) was lower than the critical concentration 0.07 mg L⁻¹ reported by Munk and Lebuhn [55] and additions in higher concentrations than applied in the current study may increase the SMP and methane production rate and reduce SCFA concentrations. The increased addition of 0.10–0.15 mg L⁻¹ each of Ni and W instead of 0.072 mg L⁻¹ of Ni and 0.024 mg L⁻¹ of W could likely lead to increased SMP and methane production rate, based on the results of Arthur et al. [50]. Co was not added during the experiments performed by Arthur et al. [50] since the WS contained sufficient content, in contrast to the WS materials analysed and used in the current study. Little is known about the influence of the concentration of Se on the SMP of WS. In the study by Arthur et al. [50], the Se concentration in the WS was below the detection limit.

3.3.3. Ammonia and pH

The average pH in reactors 1A and 1B decreased from 8.1 in weeks 2 and 3 to 7.8 during the last two weeks of operational period 1 (days 0–120). Over the same period, the average pH in reactors 2A and 2B decreased from 8.1 to 7.5. The pH and IA/PA quotient are presented in Supplementary file: Table S3. The higher pH in reactors 1A and 1B initially, compared to later, could be due to a lack of feeding of the inoculum for 8 days prior to start of the experiments. The additional decrease in the pH seen in reactors 2A and 2B, in relation to that in reactors 1A and 1B, could be related to the higher C/N quotient in reactors 2A and 2B (12) than in reactors 1A and 1B (8).

On day 88, the TAN concentration in reactors 2A and 2B was 67% of that in reactors 1A and 1B (660 mg L⁻¹, SD 40 mg L⁻¹ vs. 980 mg L⁻¹, SD 70 mg L⁻¹), while the free ammonia nitrogen concentration was 29% of that in reactors 1A and 1B (26 mg L⁻¹, SD 0.1 mg L⁻¹ vs. 90 mg L⁻¹, SD 17 mg L⁻¹). The concentrations of TAN and free ammonia nitrogen in the mono-digestion of PWS during the last week of operational period 2 were 580 (SD 110) mg L⁻¹ and 28 (SD 1) mg L⁻¹, respectively.

3.4. Methane Production in CSTR Experiments and Methane Yield in BMP Tests

3.4.1. Methane Production and VS Conversion in CSTR Experiments

A significant difference was found between the SMPs for the substrates digested in the CSTR experiments according to a nested ANOVA (p = 0.0002). SS mono-digestion resulted in the highest SMP (320 mL g⁻¹ VS⁻¹) and PWS mono-digestion in the lowest SMP (167 mL g⁻¹ VS⁻¹) (Figure 4). No synergistic increase in SMP was observed in co-digestion (SMP 261 mL g⁻¹ VS⁻¹) in this study, apart from that caused by reduced or eliminated nutrient deficiency, compared to the mono-digestion of PWS without supplements (which leads to acidification and ultimately no gas production). The confidence intervals (95%) for the average SMP during reference period 2 were, for SS mono-digestion, co-digestion, and PWS mono-digestion, 311–327, 251–271, and 163–172 mL g⁻¹ VS⁻¹, respectively. The SMPs for all substrates for reference period 2, and for SS digestion only for reference period 1, were included in the analysis. No significant influence of reactors was found. Tukey’s multiple comparison test showed a significant difference between all substrate pairs. No significant difference was found between the SMPs for SS in reference periods 1 and 2 (Figure 4 and Figure S1). The degradation of VS in SS mono-digestion, co-digestion, and PWS mono-digestion was 61.7% (SD 1.7%), 60.9% (SD 0.7%), and 47% (SD 0.9%), respectively.

SS could be mono-digested as well as co-digested. No antagonistic effects were observed in co-digestion. However, the mistakes made when adding the nutrients limit the value of the control for the determination of antagonistic effects of co-digestion. After co-digestion had been run for more than three times the HRT during operational period 1, the OLR was increased to 2.8 g_VS L⁻¹d⁻¹ for one week (days 96–103), while the HRT was maintained. The SMP for reactor 2A during the week of increased OLR was not significantly different from that during reference periods 1 and 2 (Figure S1).

The SMP for the mono-digestion of PWS was stable towards the end of the experiment (after remediating leakages), despite the low supplementation of micronutrients. The daily SMP for mono-digestion of PWS for each individual reactor showed a low SD of 5.1–5.5% over the last 36 days of the experiment. The decreasing SMP after switching from the co-digestion to mono-digestion of PWS on experimental day 34 (Supplementary file: Figure S1) correlates with the decreasing content of SS, which showed a higher SMP than PWS.

We did not analyse the nutrient availability nor investigate whether higher concentrations of any nutrient element of those applied would improve the methane production rate or the SMP under the conditions used.

3.4.2. Methane Yields in Biochemical Methane Potential Tests

The SMYs of the substrates and of milled WS are shown in Figure 4. In line with previous studies, PWS was converted to methane more quickly than WS milled to particle sizes of 1–10 mm, leading to a higher yield of methane with the former after 22 days’ digestion in a BMP test at 37 °C. After 45 days of digestion, the SMY from the PWS had not increased compared to that at 22 days. It was, however, still significantly higher (242 mL g_VS⁻¹) than that from milled WS (137 mL g_VS⁻¹); see Figure 4. No lag phase, indicating inhibition, was observed. The SMYs of the two batches of SS used in the reactor experiments were not significantly different: neither between batches nor between 22 days and 45 days of digestion (Figure 4). The SMY obtained from the cellulose control in the BMP test was within the reference yield and showed a low variation [37].

The SMP of PWS obtained by digestion in CSTRs with 22 days of HRT was significantly lower (31%) than the SMY obtained from BMP tests after 22 days of digestion, while no significant difference between CSTR and BMP was observed for SS (Figure 4). The SMYs from the BMP tests of SS and PWS did not increase between 22 and 45 days of incubation, while the residual SMPs for CSTR digestates indicate that an increased HRT would increase the SMP (Figure 5). Furthermore, the SMP for the mono-digestion of SS was not significantly different from that at the full-scale wastewater treatment plant from which the inoculum and SS were obtained; 310–360 mL g_VS⁻¹, according to a personal communication [38]. The difference between the SMP for the mono-digestion of PWS in CSTR and the SMY for PWS in the BMP test is larger than the resulting difference from using the extrapolation coefficient of 0.8 to 0.9 suggested by Holliger et al. [56]. The reason for the lower SMP of PWS for CSTR experiments than for the BMP test has not been elucidated, but it could be due to nutrient limitations [57] in the mono-digestion of PWS, as discussed in previous sections. Another explanation could be some kind of inhibition [57]. However, no signs of inhibition (such as a lag phase) were seen when PWS was digested in the BMP tests and the sugar degradation products formed during pretreatment (HMF and furfural) were below the concentrations in reactor liquids reported to be inhibiting [51] (Figure 2 and Figure 3).

The SMY of PWS in the BMP test (240 mL g_VS⁻¹) is in the lower range of values reported in previous studies (240–330 mL g_VS⁻¹) [8,17,58,59,60,61]. The steam pretreatment of WS was not optimised in the present study since much research has already been performed on this for both ethanol production and AD. Although an evaluation of the pretreatment per se was not one of the aims of the present study, the degree of conversion to methane influences other aspects of the study, such as the nutrient demand. In addition, it influences the economic feasibility of scaling up the processes. The choice of pretreatment conditions in the current study was based on Bondesson and Galbe [3,5]. The temperature and time they used prior to ethanol production and subsequent AD did not diverge from those in studies on optimisation for AD. Bauer et al. [58] reported an SMY above 300 mL per g_VS⁻¹ following steam pretreatment of WS between 160 °C and 200 °C for 10 min, compared to 240 mL per g_VS⁻¹ after treatment at 190 °C for 10 min in the present study. Theuretzbacher et al. [60] reported SMYs from 261 to 275 mL g_VS⁻¹ after steam pretreatment of WS at 140 °C to 178 °C for 30 min. An acid catalyst was used in the present study, in contrast to the studies by Bauer et al. [58] and Theuretzbacher et al. [60]. The severity of the pretreatment is known to increase conversely with decreasing pH below neutral. Ambye-Jensen et al. [62] reported lower xylose recovery after pretreatment at 190 °C for 10 min during pretreatment, including SCFAs produced during ensiling, compared to pretreatment without previous ensiling. The use of an acid catalyst improved the pretreatment efficiency at 170 °C and 180 °C, leading to similar total sugar yields to that following treatment at 190 °C without an acid catalyst. The degradation of xylose and other hemicellulose sugars could be one reason for the relatively low SMY in the present study, and the use of a lower pretreatment temperature should be evaluated in future studies. Ambye-Jensen et al. [62] also pointed out, with reference to Monavari et al. [63], that the longer storage time of WS with acids and water during 4 weeks of ensiling could improve the pretreatment efficiency compared to a shorter time, such as that used in the present study with overnight soaking.

3.4.3. Residual Methane Yield after 22 Days’ HRT

In the present study, water was added to the mono-digestion of PWS and co-digestion to achieve the same HRT and OLR as for SS. If water was not added, the HRT for co-digestion would be 30 days. If neither water nor supplements were added to the mono-digestion of PWS, the HRT would be 87 days. Nutrients must be added to the mono-digestion of PWS but could be added in more concentrated forms than in the present study. The residual methane production (Figure 5) shows that increasing the HRT increases the SMP of all substrates. With an HRT of 30 days, the SMP from the mono-digestion of SS, co-digestion, and mono-digestion of PWS would likely increase by about 5%, 8%, and 12%, respectively, based on the residual methane yield after 8 days of incubation (Figure 5). A further increase in the HRT to above 70 days would likely lead to increases in the specific methane yield of about 9%, 15%, and 19–33% for the mono-digestion of SS, co-digestion, and mono-digestion of PWS, based on the residual methane yield after 49 days of incubation (Figure 5). A higher SMP would therefore be expected for PWS mono-digestion and co-digestion without water added. However, for the mono-digestion of SS, the SS would have to be concentrated further to increase the HRT above 22 days, without reducing the OLR.

4. Conclusions

The co-digestion of PWS and SS at a VS ratio of 35:65 is stable in semi-continuous anaerobic digestion at an OLR of 2.1 g_VS L⁻¹ d⁻¹ and HRT of 22 days, without the addition of chemicals other than water and small amounts of acetic acid (as a catalyst in steam pretreatment). In the co-digestion of PWS and SS at this ratio, SS complements PWS with 12 nutrient elements quantitatively, including Co, P, and W. The co-digestion of PWS and SS reduces or eliminates the need for nutrient supplementation, in comparison to the mono-digestion of PWS, and therefore reduces or eliminates the need for the addition of mined nutrient elements, or other external nutrient supplements. Co-digestion did not result in a synergistic increase in the SMP, compared to the mono-digestion of PWS and SS. The HRT can be increased to 30 days in co-digestion, with a maintained OLR at 2.1 g_VS L⁻¹ d⁻¹, which is predicted to result in a slightly higher methane yield. Introducing PWS as a co-substrate in the digestion of SS simplifies the start-up of the process after breaks in operation. The findings of this study indicate that PWS can be mono-digested. However, long-term stability was not studied, and further studies are needed to determine the reason for the relatively low SMP. The effect of applying higher concentrations of micronutrients in PWS mono-digestion should be evaluated in future studies.