FIELD OF THE INVENTION

The present invention relates to the field of anaerobic digestion processes with separate phases, in particular with recirculation of digestate.

BACKGROUND

In general, a process of anaerobic digestion with separate phases consists of two phases: from the first phase (fermentation) the organic substrate feeding the process is converted into a gaseous mixture, consisting mainly of carbon dioxide and gaseous hydrogen, and an effluent liquid, known as fermentate, rich in organic carbon in a soluble form, such as Volatile Fatty Acids (VFA), alcohols, lactic acid and other longer chain compounds. The produced fermentate or part thereof (in the case of VFA recovery) is subsequently fed in the second phase (digestion) from which biogas and a liquid effluent are obtained (the digestate).

The use of anaerobic digestion processes with separate phases in the treatment of organic waste is widely described and consolidated. In this field, different methods for the production of biogas and precursors of bioplastic materials, such as VFA, are applied.

Hydrogen mixed with methane, also known as hydromethane, is a second- generation fuel for internal combustion engines, which allows greater performance than traditional CNG and lower emissions of pollutants into the atmosphere. VFAs are currently of great interest both in the purification sector, as carbon for supporting the biological processes of nitrogen removal and phosphorus removal and recovery from civil and industrial wastewater, and in the bioplastics production sector.

In order to avoid the inhibition of fermentation processes or a metabolic change in the fermentation process towards less desirable by-products, such as lactic acid and alcohols, and reduction of produced hydrogen gas, it is necessary to control and maintain the pH in the fermenter reaction medium in a range comprised between 5 and 6, with an optimal value of 5.5 (Micolucci et al. ,“Automatic process control for stable biohythane production in two phase thermophilic anaerobic digestion of food waste” International Journal of Hydrogen Energy 2014, 39, 17563 - 17572). This control can be obtained through the addition of chemical substances (bases) or through the dosage of recirculated digestate (Fig. 1 ). Among these methods, the use of digestate recirculation is the cheapest solution that, however, requires particular attention in order to avoid an excessive accumulation of ammonia in the system with consequent inhibition of both the fermentation process itself and the methanogenic one.

It has been demonstrated (e.g. Micolucci et al., 2014) that by maintaining the recirculation flow rate stable, i.e. by operating with a fixed recirculation ratio (RR), the process does not guarantee stability over the long operating period. In the same work (Micolucci et al., 2014), the authors have shown that the process can remain stable operating with a variable RR but in that case the RR was decided day by day by the operator on the basis of the pH value in the fermenter and the weekly trend of the ammonia content in the digester, determined analytically.

In Gottardo et al. (“Dark fermentation optimization by anaerobic digested sludge recirculation: effects on hydrogen production”. Chemical Engineering Transactions 2013, Vol 32, 997 - 1002) the effect of a fixed RR in the stability of an anaerobic digestion process with separate phases has been studied, highlighting also how the fermentation process is negatively affected by the inhibitory effect caused by the accumulation of ammonia generated by a static RR.

In a second study conducted by Gottardo et al., ("Pilot scale fermentation coupled with anaerobic digestion of food waste - Effect of dynamic digestate recirculation" Renewable energy 2017, 1 14, 455-463.), the test that gave the best result was characterized by a recirculation ratio oscillated automatically between two predetermined values (0.5 and 0.7) with a three-weekly time frequency; in this case, the ammonia content in the digester was high (1 .5 g N-NH ₄ ⁺ /L) and close to the level of inhibition of the methanogenic component.

CN104866913 describes a method for predicting the concentration of ammoniacal nitrogen in an anaerobic fermenter in a recirculation process. The method can easily predict the dynamic change in the ammoniacal nitrogen concentration in the fermenter with the increase of the recirculation time through a formula based on the total nitrogen and the volume of the incoming load. It is therefore evident the need to operate a process of anaerobic digestion with separate phases with a dynamic recirculation of digestate, the variability of which is deduced with the aim of simultaneously satisfying the two conditions mentioned above, i.e. of maintaining the pH in the fermenter at a value close to 5.5 and the ammonia content in the digester below the inhibition threshold for the methanogenic component.

SUMMARY OF THE INVENTION

The object of the present invention, with reference to Fig. 2, is a process of anaerobic digestion with separate phases with a dynamic recirculation of digestate, said process comprising a feed flow (Qo) and characterized in that the more appropriate recirculated digestate flow (QR) is established dynamically during the process through an algorithm based on inputs from a probes system; said probes system comprising a pH meter (a1 ) in the fermentation reactor, a pH meter (a2) and a conductivity probe (b2) in the digestion reactor, said algorithm being the following:

- if the pH of the reaction medium in the fermenter (pH _ai ), indicated by the probe (a1 ), is less than 5.2, then QR=QO and therefore the Recirculation Ratio (RR), defined as the ratio QR/QO, is 1 ;

- if pH _ai is greater than 5.7, then QR=0 and therefore the RR is 0 (zero);

- if pHai is between 5.2 and 5.7, then QR, and consequently also RR, is established on the basis of equation (I)

QR = 20([NH ₃ ]set - [NHajpred) + Qo/2 (I) where, [NH ₃ ]set is the total ammonia concentration value of the reaction medium in the digester fixed (Set) below the inhibition threshold of the methanogenic component, [NH ₃ ]p _r ed is the total ammonia concentration value of the reaction medium in the digester predicted (Pred) from equation (II)

[NHsjPred = -682.244(pHa2) + 235.543(Cb2) + 3874.920 (II) where pH _a 2 is the pH indicated by the pH meter (a2) in the digester and Cb2 is the conductivity value indicated by the conductivity probe (b2) in the digester. The present invention has the following advantageous characteristics:

- The automatic management of the fermentation process for:

o starting up the process

o stabilization in case of shock or

o maintaining optimal process conditions.

- The parameter control - pH and electrical conductivity - carried out online and continuously, not with punctual or titration tests;

- The ease of use of the probes for online and continuous parameter measurements;

- The estimation of the ammonia concentration carried out with a mathematical model;

- The control of the recirculate portion based on the ammonia concentration estimate.

The application of the control system according to the present invention has made it possible to improve the production performance of the AD process with separate phases, with respect to those obtained in the past, while maintaining a high process resilience and a rapid and automatic restoration capacity following stress conditions due to external forces.

The control method according to the present invention, due to its simplicity, requires minimum investment and management costs and allows the anaerobic digestion process to be controlled without the aid of external interventions even under transient conditions (such as during the process start up or a feed change).

The object of the present invention is also an apparatus, with reference to fig. 4, for the anaerobic digestion with separate phases with a dynamic recirculation of digestate according to the process of the invention, said apparatus comprising: a fermenter (100), where to conduct an anaerobic fermentation reaction, said fermenter comprising a pH meter (a1 ), an inlet for a feed flow (Qo optionally mixed with QR), an outlet for the gases produced by fermentation and an outlet for the fermentate;

a digester (101 ), where to conduct an anaerobic digestion reaction, said digester comprising a pH meter (a2), a conductivity probe (b2), an inlet in fluid communication with the fermentate outlet, an outlet for the gases produced from digestion and an outlet for the digestate in fluid communication with the feed flow

(Qo);

command and control means (103) functionally connected to all components to be activated/deactivated, and to those intended to detect and provide information and/or data.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 - Diagram of a process of anaerobic digestion with separate phases with a recirculation of digestate according to the state of the art

FIG. 2 - Diagram of a process of anaerobic digestion with separate phases with a dynamic recirculation of digestate according to the present invention

FIG. 3 - Algorithm for the dynamic management of the recirculated digestate according to the present invention

FIG. 4 - Pilot Plant Diagram used in the embodiments of the present invention FIG. 5 - A: Process configuration adopted for Runs 1 and 2 of Example 2; B: Process configuration adopted for Run 3 of example 2

FIG. 6 - A: PCR model (dashed plane) in the space of the variables: B: Results of the PCR model

FIG. 7 - Trend of pH in the fermenter and of recirculation in the embodiments of the process of the invention

FIG. 8 - Trend of the concentration of total ammoniacal nitrogen in the digester and of recirculation in the embodiments of the process of the invention

FIG. 9 - Trends of the ammonia concentration observed and predicted by the model over the whole period of time of the trial according to the present invention

FIG. 10 - Boxplot related to the specific productions shown in Table 5

DETAILED DESCRIPTION OF THE INVENTION

In the process according to the invention, having to be constant the total flow (QTOT) of feeding to the fermenter (i.e. QTOT=2QO), for values of recirculation ratio less than 1 , the portion of non-recirculated digestate with respect to its maximum value is replaced by an addition of equal volume of a dilution flow (Q _a ) that can be represented by water or other low-solids organic waste, i.e. Q _a = Qo - QR.

The feed for the process according to the invention can preferably be Putrescible Organic Waste such as the Organic Fraction of the Urban Solid Waste (FORSU) and the sludge from wastewater treatments, as well as zootechnical and agro industrial waste.

More preferably, the feed is FORSU, which can be pre-treated by any process existing in the state of the art, such as grinding, screening and deferrization or by squeezing.

According to the present invention the ammonia concentration [NH3], be it Set or Pred, is expressed as total ammoniacal nitrogen.

Preferably, according to the invention [Nhbjset is 0.70-1.3 g/L.

In case the feed is FORSU [NFbJset is preferably 0.73-0.77 g/L, more preferably 0.74-0.76 g/L, still more preferably 0.75 g/L.

The algorithm according to the present invention is represented in a block diagram in Fig. 3.

As shown in the block diagram reported in Fig. 3, the algorithm provides a two-level control. The first level uses the information received from the pH probe in the fermenter (a1 - Fig. 2) and imposes on the recirculation management system:

• if the pH is lower than the value of 5.2, to recirculate the established maximum portion (RR equal to 1 );

• if the pH is greater than the value of 5.7, to suspend the recirculation (RR equal to 0).

For the choice of the pH limit values for level 1 of the control algorithm, account was taken of:

• ensuring that the reaction medium in the fermenter can quickly reach an ideal buffer capacity to allow a significant reduction in the variability of the pH, around the optimal value of 5.5, as the RR adopted changes,

• ensuring that the request for intervention of level 1 takes place with a minimum frequency, preferentially only on occasions of the first start-up or restoration following variations caused by external events.

Going more into detail, it has been observed, from experimental tests, that by forcing, in the start-up or restoration phases of the process, the pH in the fermenter to assume a value comprised between 5.2 and 5.7, it is possible, in the subsequent regime phase, to carry out again the management of the RR exclusively as a function of the ammonia content in the digester (level 2 of the control algorithm) significantly limiting the risk that the variability of the pH is such as to compromise the desired stability of the fermentation process and consequently the use of a further level 1 control action.

Also the choice to put a unitary RR as the maximum value is attributable to a logic that derives from experimental considerations: in fact, it has been observed as with a RR greater than 1 proliferation of methanogenic biomass is more likely, in the fermentation reactor, with the consequent bioconsumption of VFA and hydrogen.

In the third case, i.e. with a pH of the fermenter comprised in the interval between 5.2 and 5.7, the determination of the RR, level 2, is obtained based on the data provided by the pH probes (a2 - Fig. 2) and electrical conductivity (b2 - Fig. 2) present in the digester. Based on the data provided by these two probes, the ammonia content in the digester reaction medium is predicted. This prediction takes place through equation (II) which was obtained through a mathematical model obtained by means of the multivariate regression method known as Principal Component Regression (PCR).

Once the ammonia content in the digester has been estimated through the model, the algorithm level 2 sets the RR value necessary to keep the ammonia content in the digester close to a predetermined value ([NH3]set); such control is obtained through a proportional approach, through equation (I).

Going into the details of equation (I), the argument comprised between the round brackets corresponds to the interval between the fixed ammonia concentration value (Set) and that predicted by the model (Pred) (equation (II)). Therefore, the management of level 2 does not take into account the pH value of the fermenter and for this reason it acts only if the condition described above at level 1 is respected. As regards the aforementioned ammonia concentration value at the set point ([NH3]set), in the case wherein the feed is FORSU, this has been set very preferably at 0.75 g/L; the choice of this value is linked to the fact that several studies (Salerno et al,“Inhibition of biohydrogen production by ammonia.” Water Research 2006, 40, 1 167 - 1 172; Gottardo et al.,“Dark fermentation optimization by anaerobic digested sludge recirculation: effects on hydrogen production”. Chemical Engineering Transactions 2013, Vol 32, 997 - 1002) have shown that ammonia can inhibit, with high concentrations, not only the methanogenic activity in the digester but can also slow down or even inhibit the acidogenic fermenting activity, starting already from total relatively low ammoniacal nitrogen values in the fermenter (0.67 g/L). Experimentally it has been observed that by maintaining an ammonia content in the digester around the aforementioned value it is possible to prevent the ammonia in the fermenter from reaching potentially inhibitory values for the fermenting activity, with a consequent reduction both in the production of gaseous hydrogen and of VFA.

Obviously in the apparatus according to the present invention the fermenter (100) and the digester (101 ) are provided with stirrer and means for controlling the temperature of the reaction medium.

Preferably, both the fermenter (100) and the digester (101 ) consist of a Continuous- flow Stirred-Tank Reactor (CSTR).

Preferably, the apparatus according to the invention therefore comprises a first conduit (104) for feeding the fermenter, said first conduit in fluid communication with the inlet of the fermenter; a second conduit (105) for feeding the digester, said second conduit in fluid communication with the outlet of the fermentate and with the inlet of the digester; a third conduit (106) of the digestate recirculation, said third conduit in fluid communication with the digestate outlet and with the first conduit. Preferably, the apparatus according to the invention also comprises a pump system; preferably comprising at least 3 pumps: a first pump (107) for managing the feed flow entering the fermenter, a second pump (108) for managing the flow of fermentate from the fermenter to the digester and a third pump (109) for the digestate flow management.

Preferably, the apparatus according to the invention also comprises at least one valve, preferably a solenoid valve (1 10) for the management of the recirculated digestate flow QR, said valve positioned downstream of the third pump (109), i.e. in the conduit (106) of the digestate recirculation.

Preferably, the apparatus according to the invention further comprises a mixing vessel (1 1 1 ), also provided with stirrer, comprising an inlet for the dilution flow (Q _a ) (e.g. water from a water network or other organic waste with a low-solids content), an inlet for the feed flow (Qo) (pre-treated or not), an inlet for the recirculated digestate flow (QR) and an outlet in fluid communication with the inlet of the fermenter. If the mixing vessel (1 1 1 ) is present, then the apparatus also has a fourth pump (1 14) for managing the feed flow (Qo) entering the mixing vessel (1 1 1 ). A fifth pump (1 15) is useful for managing the flow of the dilution medium Q _a . Said fifth pump (1 15) can be replaced by a solenoid valve according to whether the dilution medium Q _a is an organic waste with a low solid content or water from a water network.

Preferably, the apparatus according to the invention comprises an intermediate vessel (1 13) comprising an inlet in fluid communication with the outlet of the fermentate from the fermenter and an outlet in fluid communication with the second pump (108). Said intermediate vessel (1 13) is advantageously present for the recovery of fermentate to be sent to other lines (e.g. production of polyhydroxyalkanoates for bioplastics) for the use of the VFAs.

Obviously, the apparatus according to the invention comprises an electric power supply means adapted to supply electric currents and voltages necessary for the operation of the components that need it.

The present invention can be better understood in light of the following embodiment examples.

EXPERIMENTAL PART

EXAMPLE 1 - Diagram of the pilot plant

In order to verify the efficiency of the control system according to the present invention, a pilot plant has been realized consisting of 2 CSTR type reactors (the 200-litre fermenter and the 760-litre digester) to which the probes ( a1 , a2 and b2) have been applied, as in the description reported above, and a 150-litre mixing tank, used for mixing the feed with the recirculated digestate (figure 4).

From the mechanical point of view, 5 EP MIDEX pumps were used, with TΊ/4 attachments, 1400 Rpm and a maximum prevalence of 27 metres.

From the electrical point of view, both reactors have been equipped with an electrical panel provided with temperature control, switch for the stirrer, a main circuit breaker and a cut-out switch. An electrical panel was provided for the management of the pumps; this was equipped with 5 three-phase contactors, one for each pump. An additional electrical panel has been adopted to supply the control hardware; this was realized with 220V sockets plus a 24V transformer for the jack-reed connection of the gas metres and for the connection to analytical instruments.

The cRIO - compact Reconfigurable Input Output - was chosen as the hardware system. A CompactRIO (cRIO) system is a hardware device consisting of a chassis and a controller with a processor characterized by a programmable field array (FPGA), i.e. an integrated circuit designed to be configured by the user via software. In the chassis there are one or more interchangeable I/O modules (exchange of electrical impulses, i.e. in or out information), conditioned by the programming logic LabVIEW (National Instruments) or by external sources (probes, motors, electric impulses, sensors, etc.) which can be entered and managed by the processor. The modules that can be connected to the chassis of the cRIO system offer direct connectivity of the sensors used in the industrial system to be monitored and/or controlled, as well as functions that can be directly managed by the LabVIEW software system.

The control system provides for the possibility of monitoring the pH parameters, temperature and electrical conductivity by using probes directly connected to the cRIO through analog modules with input reading (analog inputs, Al). The use of digital modules (digital outputs, DO) allows the control of electrical impulses for the activation of the pumps and control valves. Analog type modules with output signal (analog outputs, AO) allow the control of the working frequency of the pumps or other types of instrumentation that provide an analog signal as a source of manoeuvre.

With regards to the pH meters used, the probes supplied by Mettler Toledo were chosen, suitable for being immersed in aggressive and fatty matrices. These are in fact equipped with a probe holder which avoids direct contact between the electrode and the substrate present in the reactors. For each probe, an analyser with 4 - 20mA output has been provided which allows an analog signal to be transmitted, then sent to a data collection system. With regards to the conductivity measurement, given the aggressive nature of the matrix to be analysed, it was preferred to opt for an induction conductivity probe supplied by Mettler Toledo. The sensor is made up of two metal ribbon coils: the guide coil is used to produce a current in the medium, the second one is a detection coil that is used to measure the current in the process solution. The magnitude of this current is proportional to the conductivity of the solution itself.

The conductivity probe was immersed inside the reactor and positioned at least 5 cm away from walls and stirrer, in order to avoid interferences and obtain a more accurate and precise measurement.

EXAMPLE 2 - OPERATING CONDITIONS

To evaluate the control method according to the invention, this was applied to two different configurations: the first one (Run 1 and 2) consists of the classical system of the AD process with separate phases, i.e. in which the whole produced fermentate from the fermenter (100) is fed into the digester (101 ) (Figure 5A). In this case the hydromethane mixture is the product.

The second configuration (Run 3) was designed for the simultaneous recovery of VFA and hydromethane; this is possible by treating a part of the fermentate using a solid-liquid separation system (centrifugation). At the outlet of the separation unit, a liquid fraction rich in VFA is obtained to be sent to any other process (such as PFIA production) as well as a fraction richer in solids ("Cake") to be sent along with the portion of untreated fermentate to anaerobic digestion (Figure 5B).

In the first configuration, the distinction of Run 1 and Run 2 is attributable to the change in the type of pre-treatment of the substrate used, i.e. the organic fraction of urban solid waste (FORSU): in Run 1 , the FORSU was pre-treated using an approach of the Wet Refine (WR) type while in Run 2 the pre-treatment took place by Pressing.

The first pre-treatment approach consists substantially in a series of mechanical operations aimed at removing plastics (screen), metal (iron remover) and reduction in size (grinding). The second approach, instead, consists in "squeezing" of the FORSU, producing a "liquid" phase called "squeezed" to be sent directly to the anaerobic treatment. Given the two different pre-treatment natures, the two substrates obtained showed different characteristics (Table 1 and 2). Table 1 : Chemical - physical characteristics of FORSU pre-treated with Wet Refine method

Table 2: FORSU pre-treated by Pressing

From the comparison between the two tables it is possible to highlight how: • the squeezed FORSU showed, unlike the one pre-treated by the WR method, an acid pH and a significant content of volatile fatty acids.

• the biodegradable carbon fraction (expressed as TVS or COD) contained in the dry waste was higher in the squeezed FORSU.

These aspects are important as they can justify how the change in the substrate can lead to a variation in the system operation; this change was therefore adopted in order to verify the response of the control to a variation in an operating condition. The operating conditions are shown in table 3.

Table 3: Main operational parameters adopted in the trial (HRT = Hydraulic residence time; OLR = Volumetric Organic Load; T = Operating Temperature).

EXAMPLE 3 - EXECUTION OF THE PROCESS ACCORDING TO THE

INVENTION Equation (II) was obtained through a mathematical model obtained through the multivariate regression method known as Principal Component Regression (PCR). This method is based on the idea of representing the space of the predictors with a reduced number of main components, recovering the maximum useful information and eliminating a marginal part of information, i.e. the one related to the directions of minimum variance, considered not important. The use in regression of main components as independent variables has the advantage of balancing the importance of the original independent variables, since the contribution of many "similar" variables is unified into a single component. A further advantage of this approach to regression is that of overcoming the problems related to the possible condition of multicollinearity between the explanatory variables: in fact, in the case they are highly correlated with each other, the calculated regression coefficients can often be unstable and the statistics for the evaluation of the model incorrect.

Therefore, the first step of the trial was to determine the coefficients of the model; for this purpose, laboratory-scale tests have been carried out which have allowed correlating the pH and conductivity measurements with the increase in ammonia concentration in the digestor reaction medium. From the acquired data it was possible to realize the data matrix from which the model (equation (II)) was elaborated.

[NH ₃ ]pred = -682.244(pHa2) + 235.543(Cb2) + 3874.920 (II)

The model was elaborated using the open source statistical software R (version 3.1 .3). Figure 6A graphically shows the model (the dashed plane) in the three- dimensional space constituted by the variables considered.

To verify the significance of the model, various analyses were performed; first, the value of R ² and of the adjusted R ² was determined; both parameters were equal to 0.988.

In order to verify the significance in the regression for each of the explanatory variables adopted, an ANOVA test was conducted which showed that this was significant, i.e. that the regression coefficient was significantly different from 0, for both explanatory variables (figure 6B). Furthermore, an F test was conducted which verified the acceptability of the assumption of the regression significance identified by the model as a whole (p-value <2.2e-16).

The trial had a total duration of 302 days: the first 100 days were used for Run 1 , the subsequent 102 days for Run 2 and the remaining 100 days for Run 3.

Figure 7 shows the pH trend in the fermenter reaction medium as a function of the adopted RR.

Figure 7 clearly shows how the control level 1 was called into question, as expected, exclusively in the first days of operation of Run 1 , i.e. in the start-up phase, and in the first days of Run 2, i.e. following the feeding change. In other words, it can be observed that, under conditions of stability, the pH of the fermenter settled around the optimal value of 5.5 managing the RR exclusively based on the ammonia content present in the digester (level 2). This was possible only thanks to the intervention of the control level 1 in the start-up and restoration phases of the process which ensured the rapid achievement, in the reaction medium of the fermenter, of an ideal buffer capacity to significantly reduce the variability of the pH around the optimal value, as the RR varies.

Figure 8 shows the trend of total ammoniacal nitrogen in the digester as a function of the adopted RR.

As can be seen from Figure 8, the level 2 control has effectively managed the ammonia content in the digester, as shown by the average observed value, i.e. analytically determined as total ammoniacal nitrogen, which settled at 756 mg/L (it should be noted that the assigned set point value was 750 mg/L). Clearly, the merit of this accuracy is largely due to the model that allowed optimally predicting ammonia concentrations in the digester; as a proof thereof, figure 9 shows the trends of ammonia observed and predicted by the model over the whole trial period. Comparing the values of observed ammonia concentration with those of ammonia concentration predicted by the model, it was possible to calculate the SDEP (Standard Deviation Error in Prediction) and the percentage of variance explained by the predicting model (Q ² ); these parameters were then compared with those obtained by the three models studied in Micolucci et al. (2014). Table 4: Comparison between the performances predicted by the model produced in this study with the models present in Micolucci et al. (2014)

Table 4 shows how the model deduced in the present study presents a decidedly better prediction capacity than the models obtained in the previous study.

• CONCLUSIONS

The trials described in the previous pages made it possible to verify the feasibility of the control system proposed in adequately managing the DA process with separate phases, both in its classic configuration, applied for Run 1 and 2, where the whole production of VFA is used for the production of methane, and with an alternative configuration, applied in Run 3, designed for the simultaneous recovery of VFA and biohydromethane. Furthermore, the control system was also placed in the position of managing a change in the type of substrate, managing to bring the system back to the operating condition without needing any external intervention.

Finally, in this second part of the discussion, the results obtained from the present process will be compared with the results highlighted in two previous studies (Micolucci et al., 2014 and Gottardo et al., 2017). In order to be able to compare the three trials, among the three tests conducted in the present study, exclusively Run 1 will be considered, since it is the one operating under the same operating conditions and characteristics of substrate as those of the two previously mentioned studies. In the study Micolucci et al. (2014), the recirculation ratio ¹ was decided day by day by the operator based on the pH value in the fermenter and the weekly trend of the ammonia content in the digester, determined analytically; in the study by Gottardo et al. (2017), on the other hand, the recirculation ratio was managed automatically, making it vary between two predetermined values with a given time frequency.

In figures 7 and 8 it is possible to highlight how the control system of the present invention has maintained, on average, except for the start-up phase, a recirculation ratio equal to 0.5 with a maximum and minimum value of 0.59 and 0.41 , respectively. Also in Micolucci et al. (2014), the recirculation ratio was on average 0.5, but in this case, however, there are large oscillations of RR as shown by the minimum and maximum values, also prolonged over time, of 0 and 0.7 respectively. In Micolucci et al. (2014), the total ammoniacal nitrogen content in the digester was higher than that recorded in the present trial (1.2 g/L vs 0.745 g/L) but still below the limit value for the inhibition of the methanogenic component (approximately 1.8 g/L). However, what can be highlighted is that in Micolucci et al. (2014) the pH of the fermenter was on average equal to 5.2, i.e. coinciding with the lower limit of the control level 1 of the present invention, and more distant from the optimal value (5.5) with respect to that found in the present trial (5.48). In Micolucci et al. (2014), the high ammonia concentration in the digester and the low pH value in the fermenter have in fact made the control of the system particularly problematic, which has forced the operator to have to continually adjust one of the two parameters, thus negatively affecting the other one. In the study conducted by Gottardo et al. (2017), the only operating condition that allowed the process to be maintained stable was that in which the RR was made to vary between the value of 0.5 and the value of 0.7 with three weekly frequency. Given the higher recirculation ratio adopted, the total ammoniacal nitrogen content in the digester in Gottardo et al (2017) was high (1.5 g/L) and close to the level of inhibition of the methanogenic component. In spite of this, the pH of the fermenter in Gottardo et al. (2017) was (5.3) almost similar to that observed in Micolucci et al. (2014). By observing the specific yields for the three

¹ In Micolucci et al (2014) only, the recirculation ratio was determined as the ratio of the recirculation portion added to the feeding flow to the total feeding flow rate products of the process (table 5 and Figure 10), it can be seen how the current trial had the best specific productions, both in terms of produced gas and in terms of VFA (intended as whole mass flow produced by the fermenter). Table 5: Specific average productions and results of the Duncan - Waller post - hoc test. For the means having different apices the null hypothesis Ho with a = 0.05 is rejected.

In fact, Table 5 and Figure 10 show that the specific productions associated with the fermentation process were significantly higher (ANOVA, Duncan - Waller post-hoc test, a = 0.05) in the present study than those found both in Micolucci et al. (2014) and in Gottardo et al. (2017).

This may be due both to the fact that for the present study, the pH value of the fermenter was closer to the optimal value (5.5) than that observed in the two previous studies, and to the fact that the concentration of total ammoniacal nitrogen in the fermenter was in both previous studies, higher (0.705 g/Land 0.687 g/L respectively for Micolucci et al. (2014) and Gottardo et al. (2017) vs 0.555 g/L found in the present study). In this regard it is recalled that Gottardo et al. (2013) had highlighted a reduction in the fermenter performances starting from a total ammoniacal nitrogen content in the fermenter of 0.67 g/L. Table 5 also shows a clear improvement in digester performances in the present trial with respect to the two previous studies; in fact in the current trial a significantly higher average specific production of methane was observed (ANOVA, Duncan - Waller post-hoc test, a = 0.05) with respect to those recorded in Micolucci et al. (2014) and Gottardo et al. (2017). This is in line with the results obtained by Gottardo et al. (2017) which showed a connection between the management of the fermenter, and in particular the pH, and the efficiency of conversion of the substrate into biogas in the digester.

Given the greater specific productions related to gaseous products, even for the overall mixture, the hydromethane, the specific production was greater (ANOVA, Duncan - Waller post - hoc test, a = 0.05) in the present case with respect to the two previous studies (on average 0.69 m ³ /KgTVSAUM and 0.79 m ³ /KgTVSAUM, respectively for Micolucci et al. (2014) and Gottardo et al. (2017) vs 0.88 m ³ /KgTVSAUM found in the current study).

In conclusion, the application of the control system has made it possible to improve the production performances of the DA process with separate phases, with respect to those obtained in the past, while maintaining a high process resilience and a rapid and automatic restoration capacity following stress conditions due to external forces.