Pilot plant functioning was based on the slow-release NH₃ stripping process conducted under batch conditions. In this process, the recovery of NH₃ from the slurry or digestates occurred in a reactor, where air was introduced into the headspace. The airflow served a dual purpose: to encourage the release of NH₃ from the surface of the slurry or digestates and to carry it away. The NH₃-laden airflow was then directed into a wet scrubber, where NH₃ reacted with a (H₂SO₄) solution, resulting in the formation of AS.

The pilot plant (Figure 1) had a 30-m³ steel tank to receive the input slurry and digestates to be treated, from which a centrifugal pump (7.5 kW) sent the input slurry/digestates to a screw-press solid–liquid separator (SEPCOM Biogas Vertical—WAM Italia S.p.A, 4 kW, screen 0.25 mm) to remove coarse solids. The liquid fraction was collected in three 7.5-m³ high-density polyethylene (HDPE) tanks equipped with a radar sensor for electronic level measurement (VEGAPULS 31, Vega). A progressive cavity pump (C.M.O. Pompe s.n.c., 1.8 kW) loaded and emptied the liquid fraction into four 7.5-m³ HDPE stripping reactors. Each reactor was equipped with the following: a heating system using hot water circulated in a pipe exchanger inside the tank, a mechanical mixing system using an electric motor with a vertical paddle mixer (100 rpm), a temperature and pH sensor (B&C electronics PH 4326; probe B&C electronics SZ165), a slurry level sensor (VEGAPULS 31, Vega), and an effluent sampling valve. The hot water, which is used to heat the reactors, was produced by an electric boiler (36 kW) that simulates the available heat from a biogas plant. The reactors were half-filled at a volume of 3.75 m³, and the airflow necessary for NH₃ recovery was ensured by a blower (Project Plast, Cremona, Italy, 2.2 kW, 2000 m³ h⁻¹) that operated in a closed circuit and connected the reactors and scrubbers. The scrubber had a tank containing an acid solution for cleaning NH₃-laden air from the reactors. As the acid was gradually neutralized to form AS, it became necessary to add more acid. This was done when the solution reached a pH of 1.5, as suggested by Sigurnjak et al. (2019). The pilot plant was monitored and controlled using a programmable logic controller (PLC), which could also be remotely accessed.

The experimental trials were conducted in the pilot plant under batch conditions (Table 1). Seven trials (i.e., Pig1, Pig2, Pig3, Dig1, Dig2, Cattle1, and Cattle2) were conducted by heating the stripping reactors at 40°C. Considering that an electric boiler was used for heating the reactors, an upscaling of the pilot plant is feasible only with a surplus heat to be valorized (i.e., from a CHP of a biogas plant). Otherwise, it is advisable to conduct the treatment at ambient temperature or with the addition of additives. For this reason, trials Pig4.at and Cattle3.at were conducted at ambient temperature, and trials Pig5.sh and Cattle4.sh were conducted at ambient temperature (at) with the addition of sodium hydroxide (i.e., sh; NaOH 30% w/w). NaOH 30% w/w was added to the reactor before starting the trial to increase the pH. It was necessary to add 30.4 kg and 59.4 kg of a previously prepared solution of NaOH 30% w/w per cubic meter of slurry to raise the pH to 10.7 for cattle slurry (Cattle4.sh) and pig slurry (Pig5.sh), respectively.

The trials were monitored using the following procedure:

During the trials, every 5 min, the PLC recorded the pH, temperature, and level values of the reactors and the pH of the scrubber.

Furthermore, during each sampling, the level of the AS solution inside the scrubber tank was manually measured and recorded. When the safety level for AS was exceeded, the AS solution was unloaded by a dosing pump (model BT-MA/AD, Etatron D.S., Rome – Italy), and if necessary, more acid (H₂SO₄ at 50% w/w) was added to the scrubber, by a dosing pump (model Tekna Evo AKL 803, SEKO, Rieti – Italy). Each acid/AS load/unload was measured and recorded by the operator.

At each sampling, electricity consumption was recorded from the plant’s electricity meter.

A total of 196 samples were collected during the trials. The samples were stored at +4°C and analyzed within 24 h, to determine the content of total solids (TS), volatile solids (VS), ash, total Kjeldahl nitrogen (TKN), total ammoniacal nitrogen (TAN), pH, EC, phosphorus (P), and potassium (K). In particular, P and K were evaluated using the rapid tests NANOCOLOR 0–55 (REF 985055) and NANOCOLOR 0–45 (REF 985045), respectively. All parameters were analyzed using standard methods (APHA/AWWA/WEF, 2012). Otherwise, organic N was derived by subtracting the TAN content from the TKN content.

The concentrations of the major elements (Na, Mg, Ca, K, and P) and trace elements (Al, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Mo, Cd, and Pb) in the AS solution were analyzed by inductively coupled plasma mass spectrometry (EPA, 2007; APHA/AWWA/WEF, 2012) using an ICP-MS (Bruker Aurora-M90).

Using both the results of the analyses on the samples and the data recorded during monitoring, the performance of the pilot plant was evaluated in terms of the TAN removal rate, TKN removal rate, and slurry volume reduction during the stripping treatment.

To determine the percentage of removal efficiencies achieved by the pilot plant, the difference between the effluent and influent masses of the above-mentioned parameters was considered.

The technical performances of the process were assessed by interpolating the TAN removal efficiency achieved in the pilot plant trials with heated reactors, to define a regression model over time considering the results of all trials performed. For the trials conducted at ambient temperature or with NaOH addition, the TAN removal efficiency was assessed by averaging the values obtained from the trials. The electricity consumption, acid requirement, and NaOH addition (when used) were then evaluated to determine the operational costs of the process and its economic sustainability.

Furthermore, the AS solution was evaluated in terms of quantity by assessing its N concentration and the mass of AS produced.

The electricity consumption recorded in each trial included the pretreatment of solid–liquid separation; reactor loading and unloading; reactor mixing, heating, and aeration; scrubber functioning; and PLC. Considering that electricity was used in the pilot plant to heat the reactors, the electricity consumption and thermal energy consumption have been analyzed separately. The H₂SO₄ requirement was the amount introduced into the scrubber during each trial, whereas the NaOH requirement was the amount added to the slurry or digestates in the reactors. The operational costs related to electricity consumption were 0.2297 € kWh⁻¹ as determined from the annual report referred to in 2021 by the Italian Regulatory Authority for Energy, Networks, and Environment (ARERA). In the trials with reactors heated to 40°C, the costs do not include the electricity consumption resulting from heating, considering that an upscaling of the pilot plant is feasible only with a surplus heat to be valorized. The cost of pure H₂SO₄ and NaOH 30% w/w were, respectively, 0.21 and 0.21 € kg⁻¹ (Vingerhoets et al., 2023). Consumption and costs were expressed per unit of N removed in each trial.

In addition to the treatment costs, it is worth noting that the AS solution produced has an economic value as a mineral fertilizer. Because there is no market for this product, the value of AS was considered a cost avoided in the purchase of mineral fertilizers based on the price of 1 kg of N as urea (46% N), with reference to the average market price in Europe in spring 2023 (March, April, and May) equal to 495 €/t, i.e., 1.08 € kgN⁻¹ (WUR, 2024).

The analyses reported in Table 2 refer to the mean values of the samples collected from each reactor at the beginning of each trial. The TS content of the treated slurries in trials Pig5.sh and Cattle4.sh also included the addition of NaOH.

The cattle slurry used in trials Cattle1 and Cattle2 had a high dilution with TS and lower nutrient content than that found in samples taken in the same area (Finzi et al., 2015). This was due to the consistent use of cooling systems based on the use of water (showers) during the period in which the slurry was treated in the pilot plant. The cattle slurry used in the trial in Cattle3.at was collected in a barn that did not have cooling systems with showers and therefore had a higher TS and nutrient content.

The digestates had a lower TS and nutrient content than that found in the same area (Finzi et al., 2015) probably because they were obtained from biogas plants fed mainly with livestock slurries (supplemented with small amounts of energy crop silage). In the Dig1 trial, the digestate used was from a biogas plant fed with pig slurry, while in the Dig2 trial, the digestate used was from a biogas plant fed with cattle slurry. Conversely, the pig slurry had a TS and nutrient content in line with (Pig1, Pig2, and Pig3) or higher (Pig4.at) than the values recorded in the same area (Finzi et al., 2015).

The results of the trials conducted (Table 3) showed TKN removal rates between 42 and 72% and TAN removal rates between 35 and 79% in the trial with reactors heated to 40°C; TKN removal rates between 3 and 26% and TAN removal rates between −5 and 25% in the trial with reactors kept at ambient temperature; and TKN removal rates between 50 and 67% and TAN removal rates between 76 and 82% in the trial with pH reactors adjusted to 10.7. The values depended on the treatment duration, which differed among trials. The TKN removal rate was lower than the TAN removal rate because the stripping treatment affected the amount of ammoniacal N. The TKN removal rate was higher than the TAN removal rate only in trials Pig2 and Cattle3.at. In trial Pig2, this removal effect could be due to a lower TAN/TKN ratio of the treated pig slurry compared with the expected values for this type of slurry (Finzi et al., 2015). The low TAN/TKN ratio could be due to the high content of organic N. A fraction of organic N had undergone a mineralization process (equal to 27% of the initial TKN content), producing TAN and leading to higher removal of TKN compared with TAN, the content of which was increased by the amount of organic N mineralized. In fact, organic N transformation to ammonium is temperature-dependent, as it is related to the transformation of organic matter. Sommer et al. (2007) observed the highest rates during the storage of pig slurry at 20°C in the initial period of 10–15 days when mineralization of organic N of approximately 19% (relative to the initial TKN content). In trial Cattle3.at, the removal effect could be due to the very slow TAN removal because the mineralization rate of organic N was 6%, consistent with the mean value of 10% found in other trials.

During the treatment, together with NH₃, a part of the water in the effluent also evaporated. A slurry/digestates volume reduction from 3 to 23% was observed in the trials with reactors heated to 40°C. The reduction in the volume of the treated effluent was very variable and depends on the duration of the treatment. It may seem like an additional advantage to be able to reduce the volume of the effluent in addition to N, but this would be true if water exits the process. However, these results were only due to the absence of insulation of pipes and scrubbers in the pilot plant. The water vapor in the exhaust air from the reactors condensed in the pipes and scrubber, especially when the external temperature was significantly lower than that of the reactor. The volume reduction can be limited in an insulated installation, or when the external temperature is close to the temperature of the reactor. The volume reduction in trial Pig1 was 3% after 15 days of treatment, with an average difference of 13.6°C between the temperatures in the reactors and the external temperature. The trials carried out at ambient temperature, with a temperature difference of 3.7–4.3°C, had a volume reduction of 1–3% in a process of 15–21 days. A higher volume reduction (23%) occurred in trial Pig3 with a temperature difference of 27.2°C and a duration of 28 days.

The TAN removal efficiencies achieved in the trials at 40°C were used to define a simple regression model over time that provides an achievable performance, regardless of the slurry characteristics and treatment duration (Figure 2). Each point on the graph corresponds to 1 sampling day and represents the removal of the TAN content from the start of the trial; 1 week after the start of treatment, a TAN removal rate of approximately 25% can be expected, 45% after 2 weeks, and 55% after 3 weeks. These average removal rates fit within a confidence interval, presenting an overall margin of variability of approximately ±15%, linked to the different operating conditions. The advantage of this process is the possibility of modulating the intensity of the treatment. The treatment can be stopped at any time once the necessary N removal level has been reached, based, for example, on the N surplus that a livestock farm has to manage. By observing the levels of TAN removal as a function of treatment time, the process highlights higher removal rates in the early stages. This means that longer treatment times allow higher levels of overall removal, but in the advanced stage of the process, the TAN release rate is lower than that in the initial stage. Therefore, each unit of N removed in this stage will have a higher unit cost.

Considering the trend in the removal of TAN, as shown in Figure 2, and based on the increased knowledge gained during the trials conducted in the pilot plant, a treatment duration of 2 weeks resulted as a tradeoff between the achievable N removal rate and consumption/costs. In detail, after 2 weeks of treatment, the performances shown in Figure 3 were achieved. In trials with reactors heated to 40°C, the digestates showed an average TAN removal rate of 54%, compared with 35% for cattle slurry and 34% for pig slurry. Without considering the Pig2 trial, which was characterized by high N mineralization, the average removal rate of TAN from pig slurry was 47%. The trials conducted at ambient temperature showed significant variability, with the Pig4.at trial close to 25% of TKN removal and the Cattle3.at trial characterized by high N mineralization. The trials with NaOH addition showed higher values than the other trials, with an average TAN removal rate of 76%. It should be noted that the addition of NaOH allowed a TAN removal rate of 64% to be achieved after just 1 week of treatment.

The amount and N concentration of the AS solution showed variability among the trials conducted, due to treatment duration and the ambient temperature during each trial.

In the trials with high slurry/digestates volume reduction, the water evaporated from the reactors condensed in the scrubber diluting the AS solution. The level of the scrubber tank also increased the necessity of emptying the AS solution before it became saturated with NH₃ to prevent spilling liquid from the scrubber.

In trial Pig1 conducted with the heated reactors and low slurry volume reduction, the N concentration of the AS solution reached 69.7 g kg⁻¹ (AS solution concentration of 33.2% w/w). Therefore, in the trials with high slurry volume reduction, the concentration of N in AS was low (5.7–44.0 g kg⁻¹ of N).

Due to the high N removal rate of the Pig5.sh trial, conducted by adding NaOH and at ambient temperatures during the summer period (approximately 30°C of slurry temperature), the N concentration of the AS solution reached the maximum value of 85.9 g kg⁻¹ (AS solution concentration of 40.9% w/w).

In addition to the N content, the AS solution was also characterized for total solids and major and trace element contents, which are reported in Table 4.

The results of the trials conducted show TKN removal rates between 40 and 70% and TAN removal rates between 50 and 80%. These removal rates were similar to those found in other studies at the laboratory scale based on the slow-release stripping process. Provolo et al. (2017) found that the highest TAN removal efficiency (87%) was obtained at 40°C with pH adjusted to 9 using sodium hydroxide. However, at 40°C without pH adjustment, removal efficiencies of 69% for TAN and 47% for TKN were obtained in 10 days. Heidarzadeh Vazifehkhoran et al. (2022) reached TAN removal rates of 92, 83, and 67% with 20 days of hydraulic retention time (HRT) and 83, 60, and 41% with 12 days of HRT from the digestates, pig slurry, and dairy cattle slurry, respectively.

Regarding the stripping process at the pilot scale, there are studies in the literature (Baldi et al., 2018; Brienza et al., 2023; Vingerhoets et al., 2023), where the treatment was performed by treating the slurry in stripping towers. Vingerhoets et al. (2023) treated a liquid fraction of a digestate for 2-3 h and obtained significantly improved N removal efficiency, from 27% to 40%, by raising the pH from 8 to 10 at a temperature of 30°C and an airflow rate of 1,440 m³ h⁻¹. At ambient temperature, Brienza et al. (2023) found a TKN removal rate of 21% (32% TAN removal) from a liquid fraction of anaerobically digested pig slurry with a pH of 8, treated during three periods between spring and autumn, in a pilot plant with the capacity to process approximately 1 ton per hour. In the present study, treating pig slurry with a pH of 7.9, a comparable TAN removal rate of 25% was obtained at ambient temperature of 30°C (Pig4.at), while a higher TAN removal rate of 82% (Pig5.sh) was obtained by increasing pH from 7.9 to 10.7 (Figure 3). In both cases, 2 weeks of treatment were necessary, much longer than the few hours reported by Brienza et al. (2023) and Vingerhoets et al. (2023).

Using our pilot plant, the addition of NaOH allowed TAN removal rates 1.9 times higher than those achieved when heating the reactors to 40°C, consistent with the findings of Baldi et al. (2018), who reported TAN removal rates of 1.6 times higher in trials with pH adjusted to 10.5 using NaOH addition, compared with trials without pH modification. Baldi et al. (2018) obtained on a full-scale plant a removal efficiency of 59–66% on digestate with pH increased to 10.5 by adding NaOH and treated for 6 h at 56°C–58°C in a continuous process, and they achieved a higher removal efficiency of 79% in a semi-batch process on digestate with pH increased to 10.1, treated for 6 h at 52°C. In the trials without NaOH addition, conducted at a temperature of 45–55°C and a pH of 9.2–9.5, they obtained a lower N removal rate (23–39%). A similar result was found by Bolzonella et al. (2018), who reached a removal efficiency of 22% when performing a stripping trial on digestates at a temperature of 60–70°C. In our study, the trials conducted at 40°C achieved TAN removal rates, ranging from 10% (Pig2) to 62% (Dig2), but after 2 weeks of treatment.

Compared to the pilot-scale slow-release stripping process tested in this study, the cited studies that treated slurry in stripping towers achieved comparable N removal rates, although with shorter treatment times. The main difference in the process was due to the lower surface of liquid exposed to air in the slow-stripping process, as the slurry was just mixed in the reactor without creating drops or enhancing the surface with packed beds. In fact, similar results were obtained but using a higher overall quantity of air. Considering the G/L ratio, defined by Palakodeti et al. (2021) as the ratio of the total volume of used gas to the volume of the stripped liquid, in our case, the value was approximately 44,000 compared to values well below 10,000 for packed beds (Palakodeti et al., 2021). The disadvantage of stripping towers is the treatment of slurry in columns with packed beds, which have a high risk of clogging and foaming, making this process more difficult to manage than the slow-release stripping process. Additionally, stripping towers require more intensive slurry solid/liquid separation to avoid clogging, increasing overall energy consumption and treatment costs.

The results obtained highlight the possibility of obtaining AS with 69.7 g kg⁻¹ of TAN in the trial Pig1, with the heated reactors and 85.9 g kg⁻¹ of TAN in the trial Pig5.sh, with the addition of NaOH. In the AS obtained in these trials, the sulfur trioxide (SO₃) contents (stoichiometrically determined) were 199.2 and 245.6 g kg⁻¹ for heated and NaOH-added trials, respectively. These AS solutions reached the compositional requirements of Fertilizing Product Regulation (EU) 1009/2019, which establishes rules for placing EU fertilizing products on the market (Regulation (EU) 1009, 2019). This regulation defines a liquid inorganic macronutrient fertilizer [PFC 1(C)(I)(b)(ii)] as a substance that contains only one declared main macroelement [N, phosphorus pentoxide (P₂O₅), or potassium oxide (K₂O)] for at least 15 g kg⁻¹ by mass and one or more declared secondary macroelements [calcium oxide (CaO), magnesium oxide (MgO), sodium oxide (Na₂O), or sulfur trioxide (SO₃)] for at least 7.5 g kg⁻¹ by mass. The sum of all declared contents of major and minor macronutrients must be at least 70 g kg⁻¹ by mass.

Overall, the AS solutions obtained had a quality in line with Sigurnjak et al. (2019), who reported trials on two full-scale plants, where the NH₃ was removed from biogas or digestates in the treatment plant or at the end-of-pipe pathway, in which AS with 53–74 g kg⁻¹ of TAN was obtained. Similar results were reported by Ledda et al. (2013), who obtained AS with 51–61 g kg⁻¹ of TAN in a digestate treatment plant characterized by sequential integration of solid–liquid separations, ultrafiltration, reverse osmosis, and cold NH₃ stripping. Pigoli et al. (2021) obtained an AS with 71.7 g kg⁻¹ of TAN in a full-scale plant where organic wastes were transformed by high-solid thermophilic anaerobic digestion coupled with N stripping technology.

The results obtained in this study are in line with the literature when considering the trials of limited water addition to the AS. As explained in the results, this condition was not achieved in some cases due mainly to the set-up of the pilot plant but could be easily improved in full-scale installations.

The average TS content (18.0% ± 7.9%) of the AS solution obtained in this study was consistent with literature findings of 20% ± 10% reported by Pigoli et al. (2021).

Contents of the major and trace elements in the AS solutions of all the trials were consistent with the data, as shown by Pigoli et al. (2021) and Brienza et al. (2021). In addition, the Cu and Zn contents of the AS solution were lower than the threshold established in Regulation (EU) 1009/2019 (Cu ≤600 mg kg⁻¹ DW and Zn ≤1,500 mg kg⁻¹ DW).

Regarding TOC content in the AS solutions of all the trials, its value was calculated as the difference between TS content attributable to AS and TS contents attributable to residual acid and was always under the threshold of Regulation (EU) 1009/2019 (≤10 g kg⁻¹ FW).