Waste biomass from food processing industries can be seen as an abundant source for biorefineries that aim at the industrial production of biofuels and value-added chemicals [1
]. Dairy industries, for example, make an important contribution to the production of liquid effluents that are rich in organic substances characterised by significant contents of lactose (0.18–60 kg/m3
), protein (1.4–33.5 kg/m3
) and fats (0.08–10.58 kg/m3
]. The generation of a liquid effluent, i.e., cheese whey (CW), is estimated at 0.8–0.9 L per litre of treated milk, or 9 kg per kg of produced cheese [2
]. The total amount of produced CW worldwide is estimated to be around 180–190 million tons per year, and only half of this by-product is subsequently used for food or feed production [4
]. On the other hand, around 100 million tons per year are typically discarded as a waste by-product in the environment, representing a significant issue for traditional wastewater treatment plants. Moreover, different functional proteins with high nutritional and therapeutic properties can be obtained from CW purification using a wide variety of separation techniques [5
]. The chemical composition and the characteristics of the CW depend on the type of milk as well as on the adopted cheese production technique. However, generally, CW has a minerals content of about 0.46 ± 10%, and a concentration of total suspended solids that ranges between 0.1 and 22 g/L. Other typical characteristics are: pH in the range 3.3–9.0, a phosphorus content of 0.006–0.5 g/L, a total Kjeldahl nitrogen (TKN) of 0.01–1.7 g/L, chemical oxygen demand (COD) values in the interval 0.8–102 g/L and biological oxygen demand (BOD) values in the range 0.6–60 g/L [2
]. Therefore, while CW processing in conventional wastewater treatment plants can be quite challenging, this waste biomass could be conveniently used as valuable feedstocks for the production of biofuels and biochemicals.
A sustainable route for the production of biofuels and biochemicals is the development of biorefineries based on renewable biomass sources [6
]. This is why anaerobic-fermentation-based bioprocesses, such as dark fermentation (DF) and anaerobic digestion (AD), have been widely tested and applied for the production of bio-hydrogen (H2
), bio-methane (CH4
) and several biochemicals, including high-value organic acids (i.e., acetate, lactate, and butyrate) [7
]. Among others, lactic acid is probably one of the most interesting products of anaerobic fermentation, being widely used in food industries as a preservative compound and curing agent, and a flavoring agent in cosmetics and pharmaceuticals, such as skin care products. Moreover, lactic acid can be chemically treated for the production of biological plastics, a natural alternative to petrochemical plastics that represents, as is well-known, a significant environmental problem [7
]. Even though most of the lactic acid that is used today is derived from biological routes, its production cost could be much more competitive if the used feedstock, which accounts for more than 70% of the production costs [10
], is represented by organic wastes. For this, CW can serve as an excellent feedstock for the production of lactic acid. However, several previous studies have been carried out on pure cultures under axenic conditions [12
]. To the best of our knowledge, only a few studies used a mixed-culture inoculum and complex substrates as a fermentation feedstock [16
]. The operation and maintenance for axenic conditions and the use of a pH buffer may have significant cost implications for the lactic acid production economy [18
]. Therefore, mixed-culture and real waste feedstock were adopted in this study, as they represent crucial elements in the development and application of a sustainable waste-based biorefinery [19
The aim of this paper is to assess dark fermentative lactic acid production using CW mediated by mixed cultures. Among other biological conversion technologies, DF is considered to be particularly interesting, as it allows for the production of multiple biofuels and other platform chemicals from waste biomasses [10
]. The study also aims at maximising the lactate yields along with identifying the optimal process stability at the natural pH of the substrate. Different management techniques have been adopted using two lab-scale fermentative bioreactors for lactic acid production from CW. The effect of digestate addition on lactic acid production and fermentation performance has been highlighted. The presented results constitute an interesting starting point for scaled-up applications in the field of CW valorisation and lactic acid production by mixed-culture fermentation.
CW fermentation was performed in two identical reactors operating under the same conditions and averaged values of all the investigated parameters are reported. Standard deviations between the two measures, corresponding to the variation range in this special case, are indicated by error bars and reported for lactic acid alone for clarity of representation.
During the start-up phase, a high F/M ratio and the anaerobic conditions led to an increase in organic acids (OA) production (Figure 1
a) and a corresponding gradual decrease of pH from 6 to 4.5 (Figure 1
b). Ethanol was detected as a significant fermentation by-product, and its concentration reached a maximum value of 5.5 g/L. It showed an increasing trend with a low concentration until Day 7. Acetic acid showed an opposite trend, reaching a stable low concentration (around 2 g/L) at Day 10. Moreover, a normalised hydrogen volume of 103 mL of H2
was produced within 20 days of fermentation (data not shown).
During phase 2 (Day 22–53), the adoption of a four-day HRT resulted in an increase in OA generation (Figure 2
a). Increasing the lactic acid production led to lower pH values (Figure 2
b). pH dropped from 4.5 to 3.8 at the end of phase 2. As shown in Figure 2
a, acetic acid and ethanol were still present in the reactor. The amount of acetic acid increased until a constant concentration of around 4.5 g/L was reached, while, again, an opposite trend was exhibited by ethanol, whose concentration was very low during the final days of phase 2.
During phase 3 (Days 56–67), lactic acid and acetic acid were the main by-products, and their concentration remained unchanged. The HRT allowed for a fermentative process producing the maximum lactic acid concentration (11.6 g/L), which, nonetheless, was lower than the maximum value obtained during phase 2 (17.2 g/L). The acetic acid concentration remained constant (≈3 g/L) and no ethanol production was observed.
During phase 4 (Days 70–123), the digestate was no longer added, and the HRT was not varied compared to the previous phase (HRT = 2 days). In these conditions, the pH varied between 3.6 and 3.4, remaining almost stable for the whole phase’s length, while the production of acetic acid slightly increased (Figure 3
Finally, the minimum HRT (1 day) adopted in phase 5 (Days 126–136), corresponding to the maximum OLR of 32.43 g COD/L/day, had negligible effects on lactic acid accumulation within the effluent (Figure 3
a). The pH value varied between 3.6 and 3.4 as in the previous phase (Figure 3
b). A maximum lactic acid concentration of 17.6 g/L was reached, which was lower than those reached during phase 4.
For mass balance purposes, Figure 3
b reports the trend of consumed soluble COD during phases 4 and 5. The reported values were calculated as the net soluble COD variation between the influent and the effluent. As can be seen, the maximum variation (ranging around 11 g/L) was registered during phase 4 at day 106.
shows, instead, the daily specific yield of lactic acid (Y*
) during the same operation phases, evaluated as mols of accumulated COD in the form of lactic acid by mols of net consumed soluble COD. The latter was calculated by removing the lactic acid contribution to the influent and effluent soluble COD value. The results showed a maximum and minimum yield of 0.33 (phase 4) and 0.10 (phase 5).
Hydrogen production was also detected during the initial phases of fermentation. The cumulative hydrogen generated during phases 1 and 2 reached a total normalised volume of 299 mL until Day 40 (data not shown). The hydrogen yields that were achieved during these phases were very low compared to other studies [22
]. After that period, only CO2
was generated. This result is in agreement with the observed trend of pH values. After phase 2, the pH value was always below 4, which generally corresponds to inhibiting conditions for hydrogen generation studies [24
Finally, Figure 5
illustrates the average lactic acid yields obtained per unit of fed substrate in terms of soluble COD. Each bar corresponds to a single HRT. Different colours are used for different operation phases. The main results are further summarised in Table 3
, which reports the average (Y) and the maximum (Ymax
) value of the yield, and the average (P) and the maximum (Pmax
) lactic acid production rate (daily amount of lactic acid produced per litre). All shown values refer to the net converted soluble COD that was obtained by subtracting the influent lactic acid contribution already contained in the cheese whey. Standard deviations related to each phase (Table 3
) refer to the averaged values of all the specified parameters during the different HRTs constituting a single phase.
It is evident that, after 16 cycles (during phase 4) conducted under different operational conditions, the reactors reached a stable performance until the end of the last phase (18 more HRT turnovers) even if an HRT change was adopted between phase 4 and phase 5. During phase 4, the maximum yield of 0.37 g COD(LA) g/COD(CW) was reached, while the maximum value of lactic acid productivity of 10.6 g LA/L/day was observed during the final phase at the HRT of 1 Day.
The high organic carbon content and the prevalent rapidly biodegradable COD fraction of CW allow for a promising conversion of this liquid effluent in dark fermentative bioreactors [27
]. During the start-up phase of continuous bioreactors, digestate from anaerobic digestion is normally used as an inoculum for DF experiments, as it contains fermentative acidogenic microbial consortia [31
]. Different pre-treatment strategies have been introduced during the last few years [22
] that lead to methanogenic activity inhibition by favoring acidogenic species [23
]. In this study, the significant amount of digestate used as an inoculum for CW conversion had an important buffering effect during the start-up phase, and limited the pH drop (Figure 1
b) due to acid generation [25
]. Hydrogen production was low, while ethanol was the main fermentation by-product. This result was related to the fact that alcohol production processes (i.e., solventogenesis) are usually associated with the presence of acetic acid, which is produced by the Wood–Ljungdahl pathway using hydrogen as an electron acceptor [26
]. However, the concomitant presence of ethanol and acetic acid could also have been due to the presence of heterofermentative bacteria. Heterofermentation, in fact, is one of the different pathways that lactic acid bacteria can follow, leading to lactic acid, carbon dioxide, ethanol, and acetic acid production [27
]. As a consequence of the mentioned conversion processes, low lactic acid concentrations characterised phase 1. The maximum measured lactic acid concentration value, in fact, was below 5 g/L.
The lactic acid concentration clearly increased in the second phase of operation. The adoption of a repeated-batch feeding strategy, which was different from the start-up phase, strongly affected the lactic acid concentration and yields. In industrial-scale applications, lactic acid is generally produced in batch mode, and by using pure cultures, whose growth is optimised by the addition of a significant amount of chemicals [13
]. Previous studies demonstrate that the adoption of a repeated-batch mode could give a higher yield than the adoption of a single-batch mode [10
]. During the same phase, high concentrations of acetic and lactic acid were reached, and the pH profile indicated that the buffering capacity due to digestate addition did not affect the pH evolution.
During the third phase, the more stable OA concentrations suggested that the microbial community was more acclimated to the CW conversion. Although the lactic acid production rate was more stable, the higher ORL led to a lower substrate conversion rate, reducing the bioreactor’s performance. These results were likely due to the reduced time for slower biological reactions contributing to lactic acid production during phase 2.
The acetic acid concentration, which varied between 2 and 4 g/L during the digestate addition phases (1, 2 and 3), drastically decreased to 1 g/L at the beginning of phase 4. Digestate is rich in acetogenic bacteria, which are able to convert organic compounds in acetic acid [31
]. The lower presence of acetogenic micro-organisms in the feeding cheese whey led to their progressive washing-out during phases 4 and 5. Conversely, the lactic acid production considerably increased, and a maximum concentration of 20.1 g/L was obtained. This result was probably due to the prevalence of autochthonous lactic bacteria, which led to a higher lactic acid percentage compared to the previous phases [33
]. Indeed, lactic acid bacteria are more acid-tolerant than other fermentative bacteria as they are able to regulate their intracellular pH [34
]. Moreover, they are able to grow at extremely low pH [14
]. In previous studies, acidic pre-treatments on fermentation inoculum were adopted to favor lactic acid bacteria growth and proliferation [14
]. This procedure limited the production of undesirable compounds in the fermentation broth, increasing the purity of the produced lactic acid.
It is worth noting that the lactic acid concentration measured during phases 4 and 5 (without inoculum addition) did not reach the peak characteristics of phase 2 (operated with inoculum addition). Nonetheless, the concentration fluctuations were restrained (Figure 3
a), suggesting that the process was more stable along the different feeding cycles.
As to the effect of the HRT variation, it could be observed that the adoption of lower values did not strongly affect the lactic acid concentration. As reported in a previous study [38
], when low HRTs were adopted (12 h and 8 h), it was possible to reach a high extracted lactic acid mass in the presence of low lactic acid concentrations. HRT variations affected both the yield and the lactic acid productivity. The yield was higher when a higher HRT was adopted, while the production increased for lower HRT values. Therefore, it seems more convenient to use lower HRTs in order to extract a higher lactic acid amount in terms of daily mass.
Different studies demonstrate that the optimum pH value for lactic acid production is between 5.5 and 5.9 [13
]. The adoption of lower values led to low lactic acid concentrations (below 5 g/L) [14
]. In contrast, the results achieved in this study showed that it was possible to produce a high lactic acid amount (20.1 g/L) at low pH, under the adopted uncontrolled pH conditions. This result is highly relevant, as uncontrolled pH conditions are usually unfavorable for lactic acid production. Perez et al. [13
], for instance, studied CW fermentation by Lactobacillus helveticus
under uncontrolled pH conditions in batch mode. The maximum observed concentration value was about 15 g/L, while the same species was able to accumulate around 60 g/L at a fixed pH of 5.9, and 80 g/L with the supplementary addition of yeast extract. Liang et al. performed another example of uncontrolled pH fermentation in 2014 [16
]. The study used potato peel waste as a substrate and mixed culture in batch mode. The maximum concentration of 14.7 g/L of lactic acid was observed. Wu et al. [41
] studied acidogenic fermentation of fruit and vegetables wastes. In order to improve the lactic acid production, they varied the pH value of a Continuous-flow Stirred Tank Reactor (CSTR) from 4 to 5 by external addition of a NaOH solution. The maximum concentration of about 15 g/L was reached, which was again lower than the maximum concentration achieved in the present study. Regarding the lactic acid yields (Y and YMAX
in Table 3
), expressed as grams of produced lactic acid (in terms of COD) per gram of fed COD to the reactors, the higher YMAX
value was reached by Choi et al., 2016 [38
] using CW and performing the fermentation at a fixed pH of 5.5. When the same reactor was operated at pH 3, the average yields were below the value of 0.1. In this study, the values of Y ranged between 0.2 and 0.37, similarly to the case presented by Whu et al. in 2015 with different wastes [41
Other authors have evaluated the yield in terms of mol of produced lactic acid per mol of consumed lactose, carbohydrates or soluble sugars consumed during the fermentation process [42
]. This parameter adds qualitative information about the biological conversion and refers to the percentage of the feeding organic compound effectively converted into lactate. Ghaly et al. [42
] studied the batch fermentation of cheese whey using pure cultures and nutrient supplementation. They reached yields (in terms of grams of lactic acid per gram of lactose) between 0.56 and 0.72. Different strains were tested by Joudeikiene et al. [43
], who performed pure-culture cheese whey batch fermentation at 37 °C under stationary optimised conditions. The results showed significant lactic acid yields ranging between 0.33 and 0.065. In this study, the grams of lactic acid (in terms of COD) per gram of converted COD were evaluated under repeated-batch conditions. The maximum yield of 0.37 was achieved during phase 4, which was lower than the values reported under batch conditions using pure cultures.
No important pH variation was detected from the second phase to the end of the process, as the pH remained around the lactic acid pKa due to the prevalence of this compound in the culture medium. It was more convenient to feed the reactor with the cheese whey alone as it was possible to produce higher amounts of lactic acid, which had higher purity compared to the previous phases where other fermentation by-products were detected. The presented results demonstrate that mixed culture CW fermentation can be a promising alternative to the pure culture fermentation processes that are usually adopted for biological lactic acid production. This study represents a preliminary base for subsequent higher-scale applications and for mathematical modeling of mixed-culture fermentation processes [32