Legume as Vegetal Nitrogen Source with Olive Mill Wastewater for Methane Production Through Two-Stage Anaerobic Co-Digestion Process

Abstract

Energy security fosters the development of biogas, particularly in the context of the rapid energy transition. Substrates suitable for anaerobic digestion serve as feedstocks to produce biogas. Determining the optimal feedstock ratio is a key factor for achieving viable anaerobic digestion (AD) processes with high methane yields. This study evaluated different two-stage AD assays using biomass from a leguminous crop (Lupinus Albus, lupin) and olive mill wastewater (OMW). The highest methane yields were obtained in assays with higher proportions of lupin in the feed mixture (532 L kg VS⁻¹ and 522 L kg VS⁻¹) and at a higher Organic Load Rate (OLR) evaluated (510 L kg VS⁻¹). Moreover, the presence of OMW in the feedstock significantly increased the Volatile Fatty Acid (VFA) concentration, as observed in the assay.

1. Introduction

The current global energy system increasingly relies on renewable sources such as biogas and biomethane. Each year, round 40 billion cubic meters of natural gas equivalent is produced, mainly in Europe, the United States, and China, with most of it used for local heat and electricity generation. Although biomethane production is growing at an annual rate of about 20%, it still accounts for only 0.2% of the global natural gas demand [1].

It is estimated that sustainable biogas could replace up to one-quarter of the current global demand for natural gas [1]. However, only about 5% of this potential is currently being exploited, primarily in Europe (around 40%).

Biogas is produced through anaerobic digestion (AD), a biotechnological process in which complex organic matter is degraded into simpler compounds in the absence of oxygen. This process has significant environmental benefits, including renewable energy production, organic waste treatment, and greenhouse gas mitigation. The microbial communities involved differ between the acidogenic and methanogenic stages, which also require different optimal conditions. In single-stage AD systems, operational conditions are usually selected to favor the slower-growing methanogens over the faster-growing acidogens, thus limiting efficiency. Two-stage AD, where acidogenesis and methanogenesis are separated into different reactors, has been shown to improve process stability and increase methane yields compared to traditional single-stage systems [2,3] The division of the process into two digesters provides a more flexible loading operation that allows for higher biological stability, even when using complex substrates [4].

The competitiveness of biogas can also be enhanced by emphasizing its positive externalities, such as waste valorization, greenhouse gas reduction, and biofertilizer production. In regions like Extremadura (Spain), large areas of olive groves generate about 500 million tons of waste [5], primarily in the form of olive oil mill wastewater (OMW). OMW is difficult to degrade due to its high polyphenol content, making two-stage AD a promising alternative to improve treatment efficiency and methane yield. Furthermore, balancing the carbon-to-nitrogen (C/N) ratio of substrates is critical for stable AD performance. OMW typically has a C/N ratio of around 40 [6], higher than the optimal 20–30 range [7]. Therefore, a nitrogen-rich co-substrate is required to improve the balance.

Lupin (Lupinus albus) is a leguminous crop well adapted to the Iberian Peninsula. It can grow in acidic and sandy soils with low fertility, requires little input, and is relatively resistant to drought thanks to its deep taproot [8]. With its high nitrogen content, lupin represents an excellent co-substrate for balancing the C/N ratio in AD. Previous research using another legume, Crotalaria latifolia L. (CL), reported maximum methane yields of 494 L kg VS⁻¹ at an organic loading rate of 1.4 g VS L⁻¹ d⁻¹ [7]. However, CL requires irrigation in the climatic conditions of Extremadura, whereas lupin can complete its growth cycle under rainfed conditions.

The aim of this study is to optimize the performance of two-stage AD using OMW as the primary carbon source and lupin as a nitrogen source. By evaluating these reactors, the influence of lupin and OMW on the methane yield and the main parameters of the AD process are evaluated. This study can offer key factors to improve and set up anaerobic digestion trials using OMW.

2. Materials and Methods

2.1. Inoculum and Substrates

The olive mill wastewater (OMW) and lLupin (L) used for the reactors were collected from a factory dedicated to the management of olive mill waste (Troil Vegas Altas S.C., Badajoz, Spain) and from the experimental fields of the Centre for Scientific and Technological Research of Extremadura (CICYTEX, Badajoz, Spain), respectively. These wastes were mixed with Prickly Pear (PP), Rapeseed Oil (RO), and Crotalaria latifolia L. (CL), also harvested at the same center. The vegetable raw materials, lupin and CL, were crushed before the ensiling process, and after, the silage vegetable materials were dried at 105 °C and milled. PP was ground and stored at −4 °C prior to use. OMW required no pretreatment and was stored under the same conditions. The inoculum was obtained from digestate used in previous studies by Parralejo et al. [7].

2.2. Experimental Design and Reactor Setup

The reactors were stainless steel laboratory digesters with a capacity of 5 L, operated in a semi-continuous two-stage AD configuration (hydrolytic and methanogenic stages). The working volume was 2 L in the first digester and 4 L in the second. The system was maintained at mesophilic temperature (38 °C), with partial recirculating of digestate from the second stage to the first. Temperature control was ensured by hot water circulating through an external jacket. Feedstock was added daily via a hopper, and the organic loading rate (OLR) was set at 3.2 g VS L⁻¹ d⁻¹ for all assays, except assay 7, which was performed at 3.6 g VS L⁻¹ d⁻¹. The values of OLR were selected according to the literature that supports high methane yields at similar OLRs [9,10]. The digester volume was maintained by removing an equivalent amount of digestate through a ball valve. The design and feedstock composition are summarized in

Table 1. Reactor design and composition of substrates evaluated.

Parameter	PP	RO	CL	Lupin	OMW
C, %	1.64 ± 0.10	74.00 ±0.50	44.80 ± 0.28	45.20 ± 0.57	11.8 ± 1.56
N, %	0.11 ± 0.01	0	1.97 ± 0.06	3.20 ± 0.03	0.33 ± 0.03
C/N	14.91	-	22.74	14.13	35.76
VST, %	6.65 ± 0.10	-	80.64 ± 0.12	93.79 ± 0.12	11.47 ± 0.42
Assay	Composition in the feedstock, % over VS
1	50	17	22	11	-
2	53	18	12	17	-
3	50	18	-	32	-
4	24	-	-	40	36
5	8	-	-	40	52
6	-	-	-	40	60
7 *	-	-	-	40	60

PP: Prickly Pear, RO: Rapeseed Oil, CL: Crotalaria legume, Lupin: Lupinus legume; OMW: olive mill wastewater, VS_T: Total Volatile Solids. * Assay 7: The OLR evaluated was 3.6 kg VS L⁻¹ d⁻¹.

and Figure 1. The feed composition specified in Table 1 was selected in order to bring in small modifications in the raw material amounts per each assay to include lupin and OMW in the feedstock and maintaining a balanced C/N ratio. The recirculation of a portion of digestate (175 g) was carried out to help reduce the high TS concentration of the feedstock.

Similar values for both legume plants were obtained in terms of their C content; however, the N content of lupin is higher than the N content of CL. This difference in values in the N composition for the two legumes entails a C/N ratio close to optimum values for the two legumes. However, the OMW has an increased C/N ratio, so lupin is better able maintain the suitable values of the C/N ratio for the mixture of the feedstock.

2.3. Analytical Methods

During the performance of the evaluated processes, carried out according to APHA methods [11], the pH and redox potential were measured using a pH meter in the digestion medium [9]. Total and Volatile Solids (TS and VS) were determined by drying the samples at 105 °C for 24 h and at 550 °C for 5 h, respectively. Alkalinity was measured by titration. In addition, the concentrations of Chemical Oxygen Demand (COD) and ammonia nitrogen (N-NH₄) were determined following the methods of Rani et al. [12]. Total Volatile Fatty Acids (VFAs) were analyzed according to Buchauer’s volumetric method [13], and the initial C/N ratio of the substrates was analyzed by employing a True-Spec CHN Leco 4084 elementary analyzer (Leco, Benton Harbor, MI, USA) using the Dumas method according to the UNE-EN 16948 biomass analysis standard for C, N, and H [14]. Biogas volume and composition were measured daily according to the procedure reported in the research carried out by Parralejo A.I. [15]. Biogas volume produced in the reactor was collected in Tedlar bags, and it was previously determined through a gas meter (Ritter model MGC-1 V3.2 PMMA, Awite Bioenergie GmbH, Langenbach, Germany). Subsequently, an Awite System of Analysis Process series 9 analyzer (Bioenergie GmbH, Awite Bioenergie GmbH, Langenbach, Germany) was employed to determine the biogas composition.

2.4. Energy Potential and Statistical Analysis

The methane energy potential (E) was calculated following the method of Usmanbaha et al. [16]:

E (kJ kg VS⁻¹) = 0.716 g CH₄ L⁻¹ × 55.5 kJ gCH₄⁻¹ × Methane yield (L kg VS⁻¹)

(1)

Statistical analyses of pH, ammonia nitrogen, alkalinity, and VFA values from the second digester were performed using ANOVA with a significance level (p) of 0.05 using the SPSS Statistics 20 software (IBM, Armonk, NY, USA).

3. Results

3.1. Influence of Substrate Composition and OLR on Methane Production

Increasing the proportion of lupin (reactors 1–3) or OMW (reactors 4–6) affected the methane yields, as shown in Figure 2. The lupin proportion (reactors 1, 2, and 3) and the OMW proportion (reactors 4, 5, and 6) were increased in the feedstock to determine the total methane yield from the two digesters, as shown in Figure 2. The methane production level presented in Figure 2 belongs to the two stages of the developed AD.

In the study of the influence of the lupin proportion on the methane yield, an increase was reached, but the difference achieved was not significant. However, when the OMW was included in the feedstock, the yield rose. Figure 2 shows the most elevated values in reactor 5 (467 L kg VS⁻¹), with a feedstock composed of PP, lupin, and OMW.

Lupin was increased in the composition of the developed reactor feedstock because it is cheaper to produce than CL. As can be observed in Figure 2 and

Table 2. Parameters and results of two-stage AD process in reactors 1 to 7.

Parameter	Reactors 1	Reactors 2	Reactors 3	Reactors 4	Reactors 5	Reactors 6	Reactors 7
HRT, days	43	43	37	52	70	89	63
OLR, g VS L−1 d−1	3.2	3.2	3.2	3.2	3.2	3.2	3.6
Total methane yield, L kg VS−1	487	532	522	342	467	408	510
COD removal, %	87	85	88	74	80	70	58
Methane energy yield, kJ kg VS−1	19	21	21	14	19	16	20
Methane energy potential, kJ d−1	235	242	282	182	249	218	241
Methane concentration stage 1, %	42	40	42	48	55	62	59
Methane concentration stage 2, %	59	67	63	68	69	65	60
Total methane concentration, %	51	53	53	58	62	64	59

Reactor 1: 50% PP + 17% RO + 22%CL + 11% lupin; reactor 2: 53% PP + 18% RO + 12% CL + 17% lupin; reactor 3: 50% PP + 18% RO + 32% lupin; reactor 4: 24% PP + 40% lupin + 36% OMW; reactor 5: 8% PP + 40% lupin + 52% OMW; reactor 6: 40% lupin + 60% OMW; reactor 7: 40% lupin + 60% OMW.

, higher methane yields were obtained when lupin was used in a higher proportion in the feedstock (reactors 2 and 3). Also, OMW contributed to increasing the methane yield when the OLR was raised.

A comparison between the methane yield and methane content evolution in all the reactors (Figure 3 and Figure 4) was carried out to evaluate the influence on the methane yield and methane content using different feedstock composition and the OLR employed in the feed, maintaining a proportion of 40:60% (on a VS basis) of lupin to OMW, respectively. Figure 2 and Figure 3 indicate that the methane yield obtained was higher when the OLR was raised. In terms of the methane yield evolution (Figure 3), different behavior was observed amongst the reactors, with a certain stability remaining in reactors as the amount of substrates in the feedstock decreased. Regarding the OLR employed, the results shown in Figure 3 suggest that an increase in the amount fed seemed to help microorganisms degrade the organic matter in the two-stage AD process. Ribeiro et al. [10] developed a two-stage process in fluidized-bed reactors, employing sugarcane vinasse and molasse, and they reported methane yield values of 270 L kg COD⁻¹ and 307 L kg COD¹ at 6 kg COD m⁻³ d⁻¹ and 9 kg COD m⁻³ d⁻¹. When the OLR values were increased, the methane yields obtained did not follow the same trend. However, De La Lama et al. [9], in their study about thermal pretreatments of olive pomace before a two-stage process, noticed that the maximum methane production rate was achieved when the OLR used was 4.5 g VS L⁻¹ d⁻¹, with the OLR evaluated ranging from 2–7 g VS L⁻¹ d⁻¹. Figure 4 indicates that the methane concentration in the biogas generated remained constant amongst all the reactors evaluated, which means that the fluctuations shown in Figure 3 for reactors 1 to 3 were due to the biogas volume produced.

A steady state was observed, as shown in Figure 3, for reactors 4 to 7 practically at the beginning of the process, and the values remained stable, achieving a rise in reactors 7 after about 45 days, which may be correlated positively with OLR and the proportion of protein-rich feed (higher lupin content), while reactor 6 presented a slight decrease after nearly 60 days of the process. However, reactors 1 to 3 reflect the peak in the methane yield along the entire process (Figure 3), probably due to the heterogeneity in the feedstock composition.

A summary of performance in the two-stage AD process is detailed in Table 2. It can be observed that the highest methane yields belonged to reactors 1, 2, and 3 when the OLR studied was 3.2 g VS L⁻¹ d⁻¹ without OMW in the feedstock composition, and the trend of COD removal was calculated according to the trend that was observed with the methane yield. Therefore, the higher values of COD removal for the reactors without OMW (reactors 1, 2, and 3) seem to indicate that the organic matter degradation was more elevated than the degradation in the reactors with OMW (assays 4 to 6). When the OLR value was increased, the methane yield increased, but the COD removal indicated poor organic matter degradation, with OLR values above 3.2 g VS L⁻¹ d⁻¹ maybe hindering the degradation. De la Lama et al. [9], in their research evaluating a substrate like OMW, obtained the maximum methane yield values at 4.5 g VS L⁻¹ d⁻¹, but a pretreatment was carried out.

The methane concentration showed a sharp rise when OMW was introduced into the feed mixture (reactors 4 to 7), mainly in stage 1. This suggests that the adaptation of the methanogenic microorganisms in the first stage increased as the AD process progressed and the OMW amount was higher. This could be due to the recirculation of the substrate between the two phases. Lukitawesa et al. [17] studied the effect of recirculation on biogas production from citrus waste in two-stage AD, obtaining a higher methane yield in the reactor evaluated with recirculation than in the reactor without recirculation used as control.

Usmanbaha et al. [16] reported values ranging from 65–70% for the methane concentration in the second stage and an HRT of 52 days for the co-digestion of palm oil mill effluent and Ceratophyllum demersum, values close to several of the assays studied in this research. In addition, the methane energy yield (or methane energy potential), determined to be between 14 and 21 kJ g VS⁻¹ (or 182–282 kJ d⁻¹), was very similar to other authors. Devens et al. [18] studied a comparative analysis of the performance between two-stage and single-stage AD for cassava processing wastewater and glycerol, and the values obtained for the methane energy potential ranged 111–152 kJ d⁻¹ in the best assays developed for the two-stage processes, lower than values obtained in this research. Other authors obtained values very similar to the values presented in this work, such as Ribeiro et al. [10], who reported values ranging from 115–235 kJ d⁻¹ in the two-stage AD of sugarcane waste with a varying OLR.

3.2. Interactions of Parameters Measured in the Process

Figure 5 illustrates the interaction of pH, alkalinity, ammonia nitrogen, and VFAs for the different reactors, as analyzed from the second digester. An increase was observed in the pH values when the lupin substrate was introduced in larger quantities in the mixture fed to the digestor. Likewise, the OMW caused the same effect on the pH parameter because it increased for reactors 4, 5, and 6. A similar trend was observed for the alkalinity values, where they progressively rose as the lupin and OMW substrates were introduced into the feedstock. This was a consequence of the pH increase, because the alkalinity measures the buffer power of the digestion medium through the chemical equilibrium of carbon dioxide–bicarbonate [19]. Regarding the ammonia nitrogen and VFAs, there were clear differences between the reactors carried out without OMW (lower values of ammonia nitrogen) and with OMW (higher values of ammonia nitrogen). Low values of ammonia nitrogen (1913–4790 mg L⁻¹) and VFA (1275–2747 mg L⁻¹) concentrations were found in the research carried out by Parralejo et al. [20] when the presence of vegetable substrates in the employed feedstock was predominant. In any case, the values represented in Figure 5 were far from the threshold values that can inhibit the AD process (above 8000 mg L⁻¹) for the VFA concentration [21] and the value of 5000 mg L⁻¹ for the ammonia nitrogen content, according to Drogs B [22].

The VFA parameter was significantly different for reactors 1, 2, and 3 (p < 0.05) compared to reactor 5 and reactors 4, 6, and 7 (

Table 3. ANOVA analysis of evaluated parameters (alkalinity, pH, VFAs, and ammonia nitrogen) in reactors operated with Lupin and OMW.

Assay	pH	VFA, mg L−1	N-NH4, mg L−1	Alkalinity, mg CaCO3 L−1
1	7.92 ± 0.12	818 ± 178 a	1015 ± 47	8533 ± 464
2	7.98 ± 0.18	816 ± 129 a	1028 ± 10	9007 ± 205
3	8.08 ± 0.02	722 ± 71 a	1640 ± 136	13,630 ± 1682
4	8.05 ± 0.13	2828 ± 99 c	2695 ± 118	17,775 ± 525
5	8.09 ± 0.03	1996 ± 142 b	2380 ± 52	15,796 ± 159
6	8.19 ± 0.04	2740 ± 291 c	2368 ± 188	17,275 ± 1183
7	8.19 ± 0.03	2766 ± 300 c	2805 ± 109	19,472 ± 447

Different letters indicate significant differences between values of the same parameter for the assays carried out. No significant differences were found in the parameters without any letter. For calculating the standard deviation, four samples were taken over the process.

). This seems indicate that the presence of OMW in the feedstock increases the VFA concentration of the digestion medium; therefore, two-stage AD is a suitable process to treat this kind of OMW. pH, ammonia nitrogen, and alkalinity did not present a homogeneity of variances according to Leven statistics, so the ANOVA statistical analysis was not realized.

4. Conclusions

This research demonstrates a positive influence of an increase in lupin in the feedstock and an increase in the OLR on the methane yield, with the highest values achieved in reactors 2 (53% PP, 18% RO, 12% CL, 17% lupin), 3 (50% PP, 18% RO, 32% lupin), and 7 (40% lupin, 60% OMW), with values of 532 L kg VS⁻¹, 522 L kg VS⁻¹, and 510 L kg VS⁻¹, respectively. These methane yields reflect methane energy potentials ranging between 182–282 kJ d⁻¹, close to values reported by other authors. Although the methane yields from the assays with OMW produced slightly lower methane yields, the two-stage system effectively prevented inhibition and maintained process stability. Recirculation of the digestate supported the operation of reactors with a high OMW cotent. Significant differences in VFAs confirmed the suitability of two-stage AD for OMW treatment. Further research should evaluate higher OLRs and pilot-scale applications to optimize energy potential.