3.1. Process Characteristics, Productivity, and Stability
The main composition of the swine manure and manure fibers used in this study is presented in
Table 1. In general, the substitution of 1/3 of the dry matter of the influent with manure fibers resulted in a slight increase of organic matter in both digesters as compared to the reference digester. The swine manure presented an average content of organic matter,
VS (68% TS), as compared to values found in literature (60–77% TS) [
25,
26,
27,
28,
29]. The organic fraction of the NP manure fibers was of the same magnitude as in swine manure; however they were significantly richer in carbohydrates (cellulose and hemicellulose). Even though the fibers did not originate from the same batch of manure, this was expected as during solid-liquid separation, the sugars that are bound to the biomass remain in the solids resulting thus in a greater fraction of the dry matter. The composition of the pretreated fibers showed some differences in comparison to the NP fibers especially in regards to the hemicellulose and protein content, affecting the overall influent composition in the NP and AAS digesters. During optimal conditions of AAS at ambient temperature, partial solubilization of hemicellulose takes place while the cellulose content remains the same [
11,
14]. The solubilized hemicellulose was partly converted to oligosaccharides and monomeric sugars and partly to degradation products not detected by the methods applied. An increase of the protein content was also observed in the AAS fibers, which could be attributed to the formation of N-complexes during the pretreatment, as a small fraction of the reagent N is bound to the biomass [
14]. AAS also resulted in a reduced lipid content of manure fibers, implying that saponification might have occurred. Overall, aside from these differences, the organic fraction of the manure fibers was similar in both NP and AAS fibers, corresponding to ca. 70% TS, resulting in similar
OLRs in the two digesters (
Table 2).
Swine manure often contains high
TAN concentrations, which in combination with the mineralization of organic N during AD, increases the risk of partial inhibition of the methanogenesis step [
3]. Therefore, mono-digestion of this feedstock is often discouraged [
30]. The
TAN concentration in the reference digester, which was fed only with swine manure, was 2.73 g/L. The addition of fibers, both in the NP and AAS digester, led to a reduction of the
TAN concentration (
Table 2). This was expected, as aside from the lower
TAN concentration of the fibers (
Table 1), the mixture-based digesters had a lower organic N content as indicated by their protein content (
Table 1) and by the
C/Norg ratios (11.08 and 10.24 for the NP and AAS digester, respectively, in comparison to 9.44 for the reference digester), (
Table 2). Generally, the threshold value reported in literature at which NH
3 inhibition begins varies significantly and is dependent on the
pH, the temperature, the
TAN concentration, the presence of other ions and the acclimation of the inoculum [
31,
32]. According to a recent review [
33], a
TAN concentration above 1.7–1.8 g/L is inhibitory under mesophilic conditions without acclimation of the inoculum. The NP digester was the only digester with such low concentration. However, as the initial inoculum originated from a digester operated for 3 years on swine manure with
TAN concentrations similar to the reference digester′s concentration used, the inoculum could be considered well acclimated [
34]. Research has shown that both NH
4+ and NH
3 may produce inhibition to methanogenic cultures. The inhibition mechanism for each type differs; high concentration of NH
4+ is considered to produce inhibition of enzymes producing CH
4, while the inhibition mechanism of NH
3 (
FAN) is based on its diffusion through the cell membrane producing proton imbalance and deficiency of K
+ [
35]. Among the two forms,
FAN is widely considered to be a stronger and more direct inhibitor [
32]. However, there is no universal threshold for
FAN inhibition reported in literature due to the complexity of the involved factors in different systems [
32]. Inhibition due to
FAN is usually reflected in accumulation of
VFAs as the main group affected by
FAN is the methanogenic archaea. However, the concentration of
VFAs was very low during the entire period of experiments in all processes in this study (
Table 2), which further supports the acclimation of the inoculum.
The AAS digester generally presented an improved biogas production compared to the rest of the digesters. The average biogas productivity of the AAS digester was 0.48 L/L/d, corresponding to a 17% increase as compared to the productivity of the NP digester (0.41 L/L/d), and a 12% increase in comparison to the reference digester (0.43 L/L/d). The inhomogeneity of the feed had as a result a variability in the daily observed production and yield data; however, as shown in
Figure 1, the AAS digester performed consistently better than the NP digester. Interestingly, the NP digester presented lower biogas productivity than the reference digester, indicating that reducing the
TAN (and
FAN) concentration was not sufficient for improving the conversion efficiency of the process. This implies that the degradability of manure was a more important factor limiting the biogas production than the
TAN (and
FAN) concentration. The inoculum used for the digesters startup originated from a long-term AD process of liquid manure and thus it was better acclimated for the reference digester process than for the mixture-based digesters. However, given that liquid manure already contains manure fibers (in a lower percentage), a difference in biogas production due to the lack of inoculum acclimation would not be expected to be significant. The CH
4 productivity of the digesters presented similar trends to the biogas productivity, with the AAS digester performing better than the rest of the digesters. Nevertheless, the addition of fibers (both NP and AAS-treated) resulted in a reduction of the content of CH
4 in biogas (
Table 2). This was probably a result of the higher fraction of carbohydrates in the mixture-based digesters, that stoichiometrically produce a lower CH
4/CO
2 ratio in comparison to lipids and proteins in which manure was richer (
Table 1). Still, the highest CH
4 yield per g TS
fed was observed in the AAS digester.
The evolution of the CH
4 yield of the mixture-based digesters during the entire course of experiments is shown in
Figure 1. The theoretical maximum CH
4 yields of the mixture-based digesters were calculated to be 377 mL/g TS
fed (538 mL/g VS
fed) and 365 mL/g TS
fed (533 mL/g VS
fed) for the NP and AAS digester, respectively, while the yields obtained experimentally were 156 mL CH
4/g TS
fed for the NP digester and 215 mL CH
4/g TS
fed for the AAS digester. These correspond to 41% and 59% of the theoretical yields, respectively, and to a 38% increase of the CH
4 yield from the AAS digester compared to the NP digester. Assuming that the hydrolysis is the limiting step in the AD of manure fibers, a 1st order model was fitted to the CH
4 production of NP and AAS manure fibers of previously run batch experiments [
14] (data presented in
Supplementary material). This permitted calculating the hydrolysis constants
kh (0.1187 d
−1 for NP fibers and 0.1329 d
−1 for the AAS fibers), based on which, and by taking into consideration the
SRT of each digester, the expected yields of NP and AAS fibers in the continuous experiments (without manure) were calculated to correspond to 55 mL/g TS
fed and 161 mL/g TS
fed, respectively by using Equation (5). This corresponds to an increase of the CH
4 yield of the fibers of 193%. Given the feed ratio of the mixture-based digesters (2 g TS manure/ 1 g TS fibers) and assuming that the CH
4 yield of manure corresponded to the same as in the reference digester (204 mL/g TS
fed), the expected CH
4 yields of the mixture-based processes (manure and fibers) were calculated to be 151 mL/g TS
fed and 189 mL/g TS
fed for the NP and AAS digester, respectively. The theoretical predictions were similar to the experimental measurements for the NP digester, while the AAS digester exhibited a slightly higher efficiency experimentally (12%) in regards to the CH
4 yield. Therefore, the 1st order model was safely used for calculating the yields during the process simulation in HYSYS (
Section 3.3).
In the present study, the processes were operated at low
OLRs because of restrictions of the lab-scale pumping system used, making it not possible to increase the share of fibers in the feed due to clogging of the feeding tubing. While it could be expected that certain instability would be produced by increasing the
OLR and the share of fibers [
29], this would be largely associated to the availability of the added C and N. In this sense, the addition of fibers with an increased N content could be balanced if the C of the fibers is available for microbial uptake. Thus, if no further pretreatment of fibers with high N concentration is followed, their use for enriching liquid manure might result to be prohibitive. In a previous study, where raw manure fibers were added to pig manure, a significant increase in the
TAN concentration was observed, producing inhibition of the process when increasing the substitution of manure with fibers up to 60% [
28]. This difference was mainly attributed to the higher total N concentrations of manure fibers in comparison to raw manure [
28]. The efficiency of separating the organic N content of liquid manure depends on the technology used and, generally, it is increased with an increased efficiency of solids separation [
36,
37]. On the other hand, the
TAN concentration remains in the liquid fraction regardless of the separator used [
36]. The separation and AAS pretreatment of fibers for the enriching of raw manure can provide certain flexibility on the final total N content of the influent, given that an NH
3 removal step is necessarily applied after AAS and before AD. The source of manure fibers and the efficiency of solids separation could determine whether the organic N in the fibers is lower than in manure, while the NH
3 removal step following an AAS pretreatment would allow for a better control of the initial
TAN concentration of the feed.
3.2. Reduction Efficiency of Major Organic Components and Digestate Quality
The composition of the effluents of the NP and AAS digesters was analyzed in order to better estimate the effects produced by the optimized AAS pretreatment. The efficiency of VS reduction of the NP and AAS digesters along with the reduction efficiencies of the major organic components are shown in
Table 3. The AAS digester presented a higher efficiency on reducing the organic matter of the feed (50.7% reduction of VS) than the NP digester (45.8%). It is important to mention here that the VS increase due to the growth of microbial biomass has not been taken into account for the calculations. However, as the aim of this study was to compare the reduction efficiencies of the different organic components in the two digesters (NP and AAS), this was considered not to affect the evaluation significantly. Among the two processes, the AAS digester presented higher reduction efficiencies in all major organic components of the feed. The highest difference was observed in the carbohydrate fraction where the digestion of manure enriched with pretreated fibers resulted in a 60.3% and 65.3% reduction of the cellulose and hemicellulose fractions, respectively, in comparison to 42.6% and 54.1% from the digestion of manure with NP fibers. Therefore, a 42% and 21% increase of reduction efficiency for cellulose and hemicellulose was achieved, respectively, due to the AAS pretreatment. This was expected as AAS affects mostly the lignocellulosic fraction of the biomass, by increasing the efficiency of polysaccharide hydrolysis. As reported in earlier studies, the main mechanism of AAS on swine manure fibers appears to be a swelling effect [
38] together with a significant solubilization of the hemicellulose fraction [
14] resulting thus in both the cellulose and hemicellulose fractions being more accessible for microbial degradation.
Aside from the improved carbohydrate removal, AAS appears to have facilitated the reduction of lipids and proteins as well. A slight solubilization of organic N (associated to proteins) and decrease of ethanol extractives (associated to lipids) were observed after AAS [
14], probably facilitating their further degradation. The reduction of lignin was also significantly affected by the AAS treatment, as it reached 48.2% in the AAS digester in comparison to 38.5% in the NP digester. It is known that lignin is not significantly removed from manure fibers after optimal AAS at ambient temperature [
14], thus this could be a result of the swelling of the fibers that facilitated microbial access during AD resulting in degradation products. Generally, lignin is considered to be recalcitrant to bioconversion and negatively correlated to CH
4 production [
39,
40]. Nevertheless, upon degradation, certain byproducts have been reported to be converted into CH
4 under anaerobic conditions [
41].
Energy production is undoubtedly a very important asset of AD that has led to an increased focus on this technology lately. Nevertheless, another aspect of AD that deserves attention is the digestate (digester effluent) valorization. During AD the organic matter content of the substrate is reduced to a certain degree, resulting in a digestate that contains non-degraded fractions together with valuable nutrients that can be useful for further applications. For instance, organic N is converted to NH
4+ − N increasing thus the availability of this essential nutrient for bacteria, algae, and plants. Similarly, the solid fraction presents the recalcitrant fractions that still contain C that can be converted to energy through thermochemical conversion, e.g., pyrolysis, combustion, gasification, etc. or serve as a slow-release C source in agricultural soils. While research on alternative applications of digestate is expected to advance in the coming years [
42], currently it is mostly used as a soil amendment and crop fertilizer.
Some basic characteristics of the digestates (shown in
Table 4) can indicate their stability as well as whether further processing should take place prior to their integration in agricultural systems. The main elements of organic matter are C and N, and consequently a mature organic amendment should have these elements as stabilized as possible. Different criteria have been proposed for qualifying organic amendments such as compost and digestates. The ratio of cellulose/lignin has been proposed as an indicator of humic acids formation in soil and a ratio lower than 0.5 has been proposed to characterize a mature amendment [
43,
44]. Among the digestates of the enriched manure, the effluent of the AAS digester presented a lower cellulose/lignin ratio than the effluent of the NP digester (
Table 4) as expected due to more extended degradation of carbohydrates. The values of both digestates stand above the threshold mentioned, probably due to the degradation of lignin resulting in a simultaneous reduction of both carbohydrates and lignin. This indicator, however, does not take into account the total amount of remaining matter that can still be oxidized. As indicated by the
COD/TS ratio of the digestates (0.1 and 0.2 for the NP and AAS digestates, respectively) and based on the criterion of stability of 0.7 g COD/g TS [
43], the total content of the remaining fractions that could undergo oxidation is satisfactory. From a
C/N point of view, both digestates corresponded to values below 15, thus no alteration of microbial equilibrium of well-balanced soils should occur according to Bernal et al. [
45].
The values of the NH
4+ − N fraction of the total N of the digestates indicate that probably their use as fertilizers would be more appropriate than as soil amendments [
44]. This characteristic of the digestates is both due to the degradation of proteins during AD and the high NH
4+ − N concentration of the original feedstock, swine manure. Regarding the fertilizing value of the digestates, the soluble N, P, and K nutrients are considered to be readily available to plants [
43]. As expected due to the origin of the feedstock, high values of N and K and a low concentration of P were found in the soluble fraction (
Table 4). The N and K concentrations are similar to values reported elsewhere on liquid manure-based digestates [
46]. The soluble P is relatively low, which could be explained by the fact that usually the majority of P remains in the solid fraction of digested manure [
47], however it is also similar to values reported elsewhere [
44]. Finally, the pH values of the digestates are close to neutral, being thus compatible with most of plants.
In conclusion, the digestates of the processes presented in this study can be valuable for agricultural valorization under certain conditions. The differences among the characteristics of the digestates were very small (
Table 4), indicating that AAS does not affect the quality of the digestate negatively. The high fraction of NH
4+ − N indicates that the digestates should be treated as fertilizers rather than as soil amendments. Nevertheless, it is common to separate anaerobic digestates into a liquid and solid fraction in order to facilitate and make their management independent. In such cases, the solid fraction can be used as an amendment while the liquid fraction as a fertilizer. Even under these conditions, the high NH
4+ − N concentration might cause phytotoxicity according to Teglia et al. [
43]. This could be avoided by optimizing the NH
3 removal and recovery of AAS-treated fibers prior to AD, reducing thus the NH
4+ − N concentration of the digestate and improving the performance of the AD process. Additionally, the surplus NH
3 of the effluent could be recovered and used for the chemical needs of the pretreatment [
13], as also discussed in
Section 3.3.
3.3. Preliminary Techno-Economic Analysis of AD Coupled to AAS and NH3 Recovery
Based on the experimental results obtained so far, the AD process coupled to the AAS pretreatment of manure fibers and an NH
3 recovery process was simulated in HYSYS. The purpose of this analysis was to estimate the increase of biogas production due to AAS along with some key output values by setting target values for some important characteristics of the process (
Table 5), ultimately for presenting the potential and some important limitations of the proposed process.
Figure 2 shows the process flow diagram based on the HYSYS simulation. The manure fibers are delivered to the biogas plant where they undergo pretreatment targeting the optimal AAS conditions [
14]. After AAS, the pretreated fibers are subjected to a mechanical separation step in order to recover the majority of NH
3 in the liquid fraction and reuse it for subsequent pretreatments. This was a necessity as the energy demand for extracting NH
3 from the whole pretreated batch would be excessive due to heating requirements of the high liquid volume. The pretreated fibers still presenting a high NH
3 concentration for AD after separation are sent to an NH
3 extraction step, after which the concentration of NH
3 is targeted to be less than 0.5%. The NH
3 extraction takes place at 130 mbar and is facilitated by the injection of steam at 250 °C. In order to ensure the NH
3 needed for the pretreatment implementation and for adjusting the solids concentration of the NH
3 extraction step of the fibers, part of the liquid fraction of the effluent of the digester (which presents an increased NH
3 concentration) is introduced in the extraction step. The extracted NH
3 is condensed and sent back to the AAS pretreatment tank for covering the chemical needs of the pretreatment. The fibers after NH
3 extraction are inserted to an anaerobic digester along with swine manure sent from the farms, which is operated at continuous mode. The dry matter content of the digester is thus increased to ca. 9% compared to 6% if solely manure would be digested. After AD, the digestate is sent back to the farmers for crop fertilization. During NH
3 extraction, CO
2 is also extracted from the system, which is then sent to the effluent stream for avoiding CO
2 build up in the extraction process. A very small part of the NH
3 recovered (0.1%) from the extraction step is used for the catalytic reduction of NO
x emissions from gas engines as described previously [
13], providing thus a more environmental friendly electricity production.
To facilitate the simulation process, some necessary assumptions were made. Based on the simulation presented in
Figure 2, the temperature of the pretreatment would correspond to 14 °C. According to preliminary experiments of AAS of manure fibers at 4 °C and 25 °C, there was no significant difference in the resulting CH
4 yield (data shown in
Supplementary material), thus this setting was considered to be acceptable. The efficiency of AAS with recirculated NH
3 is expected to be similar compared to the efficiency when using NH
3 diluted with tap water. To this end, preliminary experiments showed that when centrifuged effluent was added to the NH
3 evaporation process instead of tap water (see
Section 2.2), no significant difference was observed in the CH
4 yield as evidenced from BMP tests (data shown in
Supplementary material). However, other assumptions as of the stability of the process at increased
OLR and the effect of the removal of part of the soluble organic matter (that is recirculated to the AAS pretreatment after solid-liquid separation) prior to AD still remain to be tested at a larger scale. The solubilized fraction of AAS fibers has been measured to correspond to 14.62%
COD of the total COD [
14]. However, as degradation products are formed in the soluble fraction, their fate during AD is not known and thus their share in the final CH
4 yield cannot be directly calculated. Finally, the energy calculations were based mainly on the heat from condensation and steam requirements while the heating losses have not been taken into account.
The ultimate goal of the simulation was to have an estimation of the energy gain from the integration of AAS and NH
3 recovery to the process. The simulation showed that the biogas output from treating 10 kg manure with 3 kg manure fibers pretreated with AAS (scenario 3) would correspond to 10.70 MJ. The steam preparation and the compressor would require 2958 kJ and 485 kJ, respectively, while 2687 kJ can be recovered from the NH
3 condensation unit through heat transfer. Thus, a 756 kJ input is required for the implementation of the process, which corresponds to 7.0% of the total energy output (10.7 MJ) from the AD of manure and AAS-treated fibers. By using the same input data for the CH
4 yields of manure and NP fibers and the same
HRT as for the simulation (20 days), it was calculated that the energy produced from the AD of manure and NP manure fibers (scenario 2) would correspond to 7.04 MJ, while that of solely manure (scenario 1) would correspond to 5.17 MJ (calculations are presented in
Supplementary material). Given that the price of biogas in Denmark corresponds to 0.1765 DKK/MJ (including subsidies) [
48,
49], a 72% increase of revenue would be achieved by integrating AAS and NH
3 recovery compared to the AD of manure and NP fibers, and a 135% increase compared to the revenue from digesting only manure (
Table 6). In this simulation, a manure to fibers TS ratio of 2:3 was applied, resulting in approximately 9% TS in the anaerobic digester. However, CSTR-type digesters may permit a solid loading of up to 12% without the need to replace the pumping and stirring infrastructure [
50,
51]. Thus, a higher share of fibers could be permitted in existing full-scale liquid AD installations. As discussed in
Section 3.1, the quality and composition of manure and manure fibers are of great importance on the success of the proposed process. Nevertheless, given the above-mentioned considerations, and the flexibility the NH
3 recovery process offers to the final
C/N ratio of the AD process, a higher share of AAS-treated manure fibers is expected to result in a significantly higher improvement of biogas production. While further research should investigate different ratios of manure fibers to manure, as well as to address the potential deviations derived from the assumptions made, this preliminary estimation indicates that the proposed process can be viable and result in an impressive increase of revenue.