2.1.3. Biogas Characterization

The produced biogas was sampled in 1 L multilayer foil gas sampling bags (Restek S.r.l.). Biogas content was analyzed using a dual channel micro gas chromatograph (Agilent, 490MicroGC), implementing Molesieve 5A and PoraPLOT Q columns, both running with helium and a micromachined Thermal Conductivity Detector (TCD), while biogas volume was determined according to the water displacement method.

### *2.2. Environmental and Economic Analyses*

Environmental and economic analyses of biogas and biomethane production from olive processing by-products through an anaerobic codigestion process were carried out using, respectively, LCA and LCC methodologies. As previously mentioned, LCA follows ISO standards, [14] and [15], which define the principles, framework, and requirements of handling a LCA study. Therefore, the procedure followed in this study includes the following four methodological steps: goal and scope definition; life cycle inventory; life cycle impact assessment (LCIA); and interpretation. The LCC methodology applied in this work was based on the approaches described by Ciroth et al. [26] and Moreau and Weidema [27] and is congruent with and complementary to the LCA methodology. Therefore, the system boundary and the functional unit were similar to those of the LCA (Figure 3). The LCC was also implemented in conjunction with specific economic indicators to assess the economic profitability of biogas production.

**Figure 3.** Methodological implementation of Life Cycle Assessment (LCA) and Life Cycle Costing (LCC). Source: Our elaboration.

#### 2.2.1. Scenarios Description

The production of biogas from anaerobic codigestion was evaluated both from environmental and economic points of view, considering the mixtures previously tested at the laboratory scale. Hence, a scaling up from the laboratory level to the industrial level was performed. Specifically, an anaerobic reactor that produces biogas and generates an electrical power of 200 kW was considered for the analyses.

Environmental and economic impact assessments were then performed considering only the mixtures containing OMWW (the control thesis was excluded). Particularly, the following was evaluated: biogas production from the mix containing 20% (*v*/*v*) OMWW and 80% (*v*/*v*) digestate with a retention time of 16 days (Thesis 2); biogas production from the mix containing 30% (*v*/*v*) OMWW and 70% (*v*/*v*) digestate with a retention time of 29 days (Thesis 3).

The scaling up was based on the results obtained in the laboratory experiments, modelling the size of the plant into an industrial one considering matrix availability. It has been assumed that the plant is located in the vicinity of another anaerobic digestion plant with an electrical power of 998 kW that produces an annual quantity of digestate

equal to 37,000 t, enough to satisfy the feeding needs of the plant fed with OMWW, which, according to the different retention times, requires annual quantities of matrices equal to 45,625.00 (Thesis 2) and 25,172.41 t (Thesis 3).

Specific assumptions for environmental and economic assessments are discussed in the following paragraph.

2.2.2. Goal and Scope Definition, Functional Unit and System Boundaries

For both LCA and LCC, the same assumptions were used for life cycle modelling so that the life cycle of biogas production from OMWW could be assessed according to common criteria from both environmental and economic perspectives. In particular, the function to be analyzed is OMWW recovery to energy; therefore, "1 m3 of normalized biogas" has been defined as the Functional Unit (FU). Since wastewater is normally considered a waste product of the olive milling process, it has been chosen to limit the system boundaries "from digester gate to the biogas production", considering OMWW as a residual product with zero impact (Figure 4).

#### 2.2.3. Specific LCA Implementation

Data on quantities of matrices, transport, quantities of biogas and generated heat as well as produced digestate were taken from laboratory trials and scaled up to the 200 kW electrical power plant. According to current regulatory requirements, the plant electricity consumption has been set at 11% for self-consumption. Secondary data on fuel production for transport and digester construction were taken from the Ecoinvent 3.5 database. The methane and ammonia emissions from digestate storage were assessed according to Lovarelli et al. [28]. Fugitive methane losses from digesters and post-digesters and losses during biogas treatment and combustion were considered equal to 2% following Dressler et al. [29].

The inventory data (Table 3) were processed through Simapro 8.5 software using the ILCD 2011 midpoint impact assessment method [30], through which the following impact categories were assessed: climate change; ozone depletion; human toxicity, non cancer effects; human toxicity, cancer effects; particulate matter; ionizing radiation HH; ionizing radiation E (interim); photochemical ozone formation; acidification; terrestrial eutrophication; freshwater eutrophication; marine eutrophication; freshwater ecotoxicity; land use; water resource depletion; mineral, fossil and ren resource depletion.


**Table 3.** Inventory data. Source: Our elaboration.

Cut-off criteria were set ignoring all inventory data that would have an impact of less than 1%, such as energy for plant control computers.

The main limitation of the study lies in the scaling operation, whereby productions are directly proportional to those obtained in laboratory trials. Since a decrease in plant efficiency at full-scale is possible, a sensitivity analysis was carried out, reducing the biogas production by 10% and 20%.

#### 2.2.4. Specific LCC Implementation and Profitability Analysis

The LCC analysis aimed at evaluating the overall cost of the two scenarios of biogas production under study (Thesis 2 and Thesis 3). Data collection was conducted in parallel with the inventory phase of LCA to estimate costs related to plant acquisition, operation, and disposal in accordance with Gonzalez et al. [31]. As pointed out by Herbes et al. [32], the site-specific conditions in which the process takes place should be considered. Therefore, in performing the economic analysis, site-specific cost drivers are taken into account.

The initial investment cost for the plant acquisition was EUR 900,000 according to the current market prices, corresponding to a specific cost of EUR 4500 per kW.

Operating costs were split into three categories: materials and services, labor, quota, and other attributions. In the first category, only transport costs for matrix handling were considered, assuming an average distance of 500 km per year. The diesel average price was taken as EUR 0.92 per liter, taking into account an average consumption of 0.05 L.t<sup>−</sup>1.km−1.

In this work, the purchase price of both OMWW and digestate was assumed to be EUR 0 per ton. In the first case, we assumed that the transport cost is covered by olive mills, which avoid the traditional disposal of the wastes on the soil. In the second case, the cost of digestate was considered for free.

Within the labor costs, human labor cost based on local current wage (EUR 8 per hour) and administrative overheads (EUR 10.3 per hour) were included.

In the quota and other attributions category, all those cost items not directly attributable to specific biogas production process stage, represented by quotas (i.e., depreciation, maintenance and insurance), interests in advance capital and capital goods, land rent and levies, were considered.

The expected revenues were estimated considering only the sale of electricity after internal consumption. The electricity produced was assumed to be fed into the national grid. A FiT tariff of EUR 0.233 per kWh was considered [33].

Estimation of end of life costs for the biogas plant disposal was obtained from the literature [31]. The plant disposal was estimated by subtracting from disposal cost the used equipment revenue.

The following assumptions were made to carry out the economic analysis of the two scenarios:


In order to evaluate the investment feasibility of the biogas production scenarios, specific economic indicators were identified—i.e., Discounted Gross Margin (DGM), Net Present Value (NPV), Internal Rate of Return (IRR) and Discounted Payback Period (DPP). These represent the most common indicators used to compare investment options, which are based on the cash flow model [32].

The DGM indicator provides information on project profitability, as advised by Mel et al. [37] and Stillitano et al. [38], defined in Equation (1):

$$\text{DGM} = \sum\_{\mathbf{t}=1}^{\mathbf{n}} \frac{\text{TR}\_{\mathbf{t}}}{\left(1+\mathbf{r}\right)^{\mathbf{t}}} - \frac{\text{VC}\_{\mathbf{t}}}{\left(1+\mathbf{r}\right)^{\mathbf{t}}} \tag{1}$$

where TRt is the total revenue in the t-th year; VCt is the variable cost in the t-th year; t is the time of the cash flow (year); n is the plant lifetime (15 years) and r is the discount rate (5%).

The NPV and IRR indicators were calculated according to Equations (2) and (3), respectively, as suggested by Moreno et al. [39]:

$$\text{NPV} = \sum\_{\mathbf{t}=1}^{n} \frac{\text{CF}\_{\mathbf{t}}}{(\mathbf{1} + \mathbf{r})^{\mathbf{t}}} - \text{I}\_{0} \tag{2}$$

where t is the time of the cash flow (year); n is the plant lifetime (15 years); CFt is the net cash flow in the t-th year; r is the discount rate (5%) and I0 is the initial investment, which equals the total facility investment.

$$\sum\_{t=1}^{n} \frac{\text{CF}\_{\text{t}}}{\left(1 + \text{IRR}\right)^{\text{t}}} - \text{I}\_{0} = 0$$

where IRR is the discount rate, which will make the NPV equal to zero.

When the conditions NPV > 0 and IRR > r occur, the investment is profitable; otherwise, it should be rejected [40].

The formula for calculating the DPP indicator is presented in Equation (4), as suggested by Tse et al. [41]:

$$\text{DPP} = \text{LNC} \frac{\text{ADC}}{\text{DCA}} \tag{4}$$

where LNC is the last period with a negative discount cumulative cash flow; ADC is the absolute value of discount cumulative cash flow at the end of the period LNC; DCA is the discount cash flow during the period after LNC.

As argued by Ong and Chun [42], the payback period, defined as the expected number of years required to recover the initial investment, is often used as an indicator of a project's riskiness. In any case, the payback period must be shorter than the time horizon considered.

Lastly, each indicator value has been defined for the FU of 1 m3 of normalized biogas.

As a final step, a sensitivity analysis was performed for the two scenarios to examine the influence of varying specific parameters over the economic indicators under study [43]. The variables independently evaluated were discount rate (r) set to be floated with ± 20% and biogas yields floated with −10% and −20%.

#### **3. Results and Discussion**

*3.1. Results of Biochemical Methane Potential (BMP) of Olive Mill Wastewater*

3.1.1. Matrix and Substrate Characterization

The results of the initial characterization of the matrices and the substrates are, respectively, reported in Table 4.

**Table 4.** Matrix and substrate preliminary characterization. Values are expressed as mean ± St. Dev of minimum three replicates for each parameter and each matrix/substrate. Source: Our elaboration.


According to the obtained data, the pH value of OMWW is, as expected, very low and similar to values reported by other authors [44], while values inherent to theses subjected to anaerobic digestion process are between 6.93 ± 0.03 and 7.97 ± 0.16, with optimal values for both mixes, meaning that the comatrix, i.e., the digestate, exerted a good buffering effect. Dry content (DC) in all cases does not exceed 10%, indicating that the process runs in wet conditions, which consists of the operating mode of most of large-scale reactors worldwide [45,46]. Volatile solid (VS) or organic substance content also represents an important parameter for the anaerobic digestion process as it refers to the susceptible content to be decomposed [47]. In addition, the chemical oxygen demand (COD), whose values are between 70.35 ± 4.47 and 80.82 ± 1.59 g.L−<sup>1</sup> for the substrates subjected to AcoD, measures the content of oxidizable compounds in the substrate [48] and theoretically enable predicting methane production as 1 g of converted COD corresponds to a maximum of 350 mL of methane [49]. In the three theses, the carbon/nitrogen ratio (C/N) was below the recommended value that should be comprised between 20 and 30, with an optimal value of 30. However, Guarino et al. [50] investigated the effect of a wider C/N interval ranging from 9 to 50 on anaerobic digestion of buffalo manure and obtained a high biomethane productivity (around 60–70%) even with lower values than those obtained in our experiments. C/N values decreased in favor of nitrogen as OMWW content in the reactor increased.

As the trials aim at assessing the BMP of the OMWW, it was important to quantify polyphenol (PP) contents since they represent inhibiting compounds of the bacterial pool, particularly methanogens. The analysis revealed an amount of 4.60 g.L<sup>−</sup>1.

#### 3.1.2. Biogas and Methane Yields

The biogas volume recorded in each sampling date as well as methane content were normalized to normal liters (dry gas, at temperature = 0 ◦C and pressure = 1013 hPa), according to the standard procedures described in the VDI 4630 [48], as carried out in [44]. Cumulative biogas production of the tested theses during the AcoD period is represented in Figure 5. Higher biogas production was registered in Thesis 2, which contains 20% *v/v* olive mill wastewater, until day 20. After that, this tendency changed in favor of Thesis 3, which registered a total amount of biogas equal to 5.80 ± 1.77 NL.L−<sup>1</sup> of substrate, corresponding to 80.22 ± 24.49 NL.kgSV<sup>−</sup>1.

**Figure 5.** Cumulative biogas production for 30-day AcoD of olive mill wastewater. Values are the mean production values obtained from the three replicates of each thesis at different sampling time. Source: Our elaboration.

Considering biogas total amount, Thesis 2 (with 4.88 ± 2.03 NL.L−<sup>1</sup> of substrate corresponding to 64.06 ± 26.64 NL.kgSV−1) and Thesis 3, respectively, recorded 2- and 2.5-times higher productions than that of the blank equal to 2.37 ± 0.37 NL.L−<sup>1</sup> of substrate corresponding to 31.89 ± 4.98 NL.kgSV−<sup>1</sup> (Figure 6), meaning that reactor content in OMWW favored biogas production. Nevertheless, statistical analysis by performing oneway ANOVA did not show any significant difference (F = 4.082; df (2; 6), Pr (>F) = 0.076).

**Figure 6.** Mean values ± St. Dev. of total biogas specific production for 30 days AcoD of olive mill wastewater. Source: Our elaboration.

Regarding biogas composition, the highest methane percentage, 75.48%, was obtained by Thesis 3. Methane content had the same tendency as biogas production, with

higher amount in Thesis 2 at the beginning of AcoD process until the 16th day, after which a decline was observed in both Theses 1 and 2, whereas higher amounts were found in Thesis 3 until day 29, after which they decreased by 58.12% (Figure 7). Total amount of methane during the whole process was equal to 0.99 ± 0.06 NL.L−<sup>1</sup> of substrate, corresponding to 13.29 ± 0.79 NL.kgSV−1, 2.55 ± 1.15 NL.L−<sup>1</sup> of substrate, corresponding to 33.53 ± 15.15 NL.kgSV−1, and 3.45 ± 1.27 NL.L−<sup>1</sup> of substrate, corresponding to 47.68 ± 17.55 NL.kgSV−1, respectively, for Theses 1, 2 and 3. Additionally, for methane content, no significant difference was found (F = 4.997; df (2; 6), Pr (>F) = 0.0528). Figure 8 illustrates the biogas composition considering the overall process for the three tested theses.

**Figure 7.** Methane content in the biogas expressed as percentage. Source: Our elaboration.

**Figure 8.** Biogas composition considering the whole process of OMWW AcoD. Source: Our elaboration.

Biogas and biomethane results obtained in this study are lower than those reported by other authors, who implemented OMWW in an anaerobic codigestion process. Indeed, experimental trials performed by Zema et al. [44] provided 0.362 Nm3.kgTVS–1 of biogas (0.187 Nm3.kgTVS–1 of methane) and 0.176 Nm3 kgTVS–1 of biogas (0.067 Nm3.kgTVS–1 of methane) after 25 days using, respectively, 20% and 30% OMWW with a polyphenol concentration of 2.8 g.kg−<sup>1</sup> in a mesophilic AcoD process with digested liquid manure. In contrast with the results presented here, they yielded higher amounts of biogas and methane with lower quantities of OMWW. Bovina et al. [51] reported an increase in biogas and methane yield with the increasing of OMWW content instead of sewage sludge, obtaining the best performances with 25% OMWW (with 1.01 ± 40 g.L−<sup>1</sup> of polyphenols) i.e., 116 NL.kgVS–1 of methane. Calabrò et al. [52] used raw and concentrated OMWW with polyphenols values ranging between 1.1 ± 0.12 and 4.4 ± 0.03 g.L−<sup>1</sup> in order to obtain up to 2 g.L−<sup>1</sup> PPs in the blends they tested in batch under mesophilic conditions. They obtained 0.419 NL.gTVS–1 with a PP concentration of 0.5 g.L<sup>−</sup>1. The blend with 2 g.L−<sup>1</sup> PPs provided better results due to the adaptation of the inoculum to polyphenols (170 against 45 NL.gTVS–1 when using non acclimated inoculum).

Accordingly, it can be stated that the lower amounts obtained in this study are mainly due to the high polyphenol (PP) contents, equal to 4.60 g.L<sup>−</sup>1. Regarding this aspect, Borja et al. [53] and Fedorak et al. [54] suggest not exceeding a phenol concentration of 2 g.L−<sup>1</sup> to avoid an inhibiting effect on the methanation process.
