This research was carried out on a full-scale pure oxygen thermophilic plant, operated and monitored throughout a period of 11 years. The plant treats 60,000 t y−1 (year 2013) of high-strength industrial wastewaters deriving mainly from pharmaceuticals and detergents production and landfill leachate. Three different plant configurations were consecutively adopted: (1) biological reactor + final clarifier and sludge recirculation (2002–2005); (2) biological reactor + ultrafiltration: membrane biological reactor (MBR) (2006); and (3) MBR + nanofiltration (since 2007). Progressive plant upgrading yielded a performance improvement chemical oxygen demand (COD) removal efficiency was enhanced by 17% and 12% after the first and second plant modification, respectively. Moreover, COD abatement efficiency exhibited a greater stability, notwithstanding high variability of the influent load. In addition, the following relevant outcomes appeared from the plant monitoring (present configuration): up to 96% removal of nitrate and nitrite, due to denitrification; low-specific biomass production (0.092 kgVSS kgCODremoved−1), and biological treatability of residual COD under mesophilic conditions (BOD5/COD ratio = 0.25–0.50), thus showing the complementarity of the two biological processes.
Thermophilic aerobic wastewater treatment is a biological process carried out at a temperature above 45 °C. This system has been proposed for treatment of various high-strength wastewaters, such as potato slops from potato distilleries, dairy wastes, olive oil mill wastewater, lipid-rich wool scouring wastewaters, paper machine white water (de Sousa et al. 2011; Abeynayaka & Visvanathan 2011a), paper mill deinking wastewater (Simstich et al. 2012) and pharmaceutical wastewaters (Collivignarelli et al. 2014).
The main advantages of aerobic thermophilic process consist of high biodegradation rate, inactivation of pathogens (Visvanathan et al. 2007) and low excess sludge production (Wujcik et al. 2000; Suvilampi & Rintala 2003). However, aerobic thermophilic treatments, compared to mesophilic processes, have been reported to suffer for poorer effluent quality (Lapara & Alleman 1999); this is typically ascribed to higher chemical oxygen demand (COD) and turbidity, caused by the presence of a considerable amount of dispersed particles, such as free bacteria and colloids (Suvilampi et al. 2005). In effect, biomass washout is favoured by the small size of sludge flocs; about 16% of the thermophilic sludge volume is characterized by particles having a diameter lower than 5 μm, while the typical value of mesophilic sludge is 4% (Abeynayaka & Visvanathan 2011b). This drawback is crucial in case a final clarifier is used for biomass/effluent separation. A feasible alternative solution consists of replacing the sedimentation tank with a filtration unit, thus implementing a membrane biological reactor (MBR) (Abeynayaka & Visvanathan 2011a; Simstich et al. 2012).
Owing to process specificity, in terms of biological and chemical mechanisms as well as physical conditions, proper operation strategies should be implemented, as the knowledge acquired from conventional-activated sludge systems is unsuitable (Abeynayaka & Visvanathan 2011b). Regarding ammonia, for instance, removal mechanisms are affected by different growing microbial consortia, physical–chemical conditions such as temperature and pH, and aeration requirement (Abeynayaka & Visvanathan 2011b). In particular, nitrification is inhibited by the extreme scarcity of nitrifying bacteria (Lapara & Alleman 1999), so that ammonia can be removed only via stripping, assimilation, and precipitation (Kurian et al. 2005; Abeynayaka & Visvanathan 2011a). On the contrary, nitrite and nitrate can be denitrified.
Owing to the lack of scientific reports on case studies, the discussion of full-scale experiences should be encouraged, in order to get reliable data for increasing the level of knowledge, thus promoting the diffusion of such a promising technology.
In this paper we report the results of an 11-years monitoring of a thermophilic aerobic plant (1,000 m3 volume) which operates in a waste treatment facility located in northern Italy. The plant was progressively improved. The strength of this research lies, first of all, in the duration of the monitoring period: data were acquired daily; moreover, plant features themselves constitute a novelty and originality factor, represented by the use of pure oxygen and the different configurations, one of which included a double membrane filtration stage. Several aspects have been investigated, such as COD and nitrogen removal efficiency and mechanisms, process stability, sludge production, and complementarity with a downstream mesophilic process (which plays a major role in case effluents from similar facilities are discharged into municipal sewers, as often observed). Eventually, we define process conditions enabling reaching of certain yields, by indicating values which can be adopted for any other plant characterized by the same configuration.
The waste treatment facility and the thermophilic unit
The waste treatment facility spreads over an area of 14,000 m2 and is authorized for a total capacity of 60,000 t y−1 of non-hazardous and hazardous (apart from carcinogens and mutagens) industrial (solid and liquid) wastes. The plant consists of two lines, for the treatment of solid and liquid wastes, respectively (Figure 1).
The solid waste line consists of shredding, reconditioning, and/or drying of packages of expired products (such as food, beverages, and cosmetics) and other industrial wastes, which are treated before being sent to recovery or disposal.
The liquid waste line includes several chemical–physical pre-treatment steps (as specified in Figure 1) followed by a pure oxygen thermophilic biological stage and tertiary treatments.
The pure oxygen thermophilic unit, which is the main subject of this research, has been submitted to progressive improvements. Configuration #1: from 2002 to 2005 the bioreactor was coupled with a final clarifier (with sludge recycling). This set-up was firstly converted into an MBR system (configuration #2): two parallel ultrafiltration units were installed, each of them consisting of three vessels with 99 ultrafiltration (UF) ceramic membranes (300 kDa cut-off). This device, which is still running, is operated at a pressure ranging between 3 and 5 bar, thus ensuring a permeate flow of 8 m3 h−1; the concentrate is recirculated back to the bioreactor. A further improvement (early 2007: configuration #3), aimed at refining the UF permeate quality, was achieved by the adoption of a second filtration stage (NF = nanofiltration: polyamide; spiral wood; 300 Da cut-off): each of four parallel lines includes two modules, equipped with four membranes, respectively. Work pressure varies between 10 and 20 bar. The retentate (about 20 m3 d−1) is recirculated back to the chemical–physical process upstream the bioreactor.
Table 1 shows, for each plant configuration, the range of variation of flowrate and main quality parameters (pH, as well, varied remarkably and reached basic values up to 11; data not shown). A large value scattering appears, as expected in such a real case. Moreover, influent chloride concentration (data not shown) was as high as 1.39%; it has to be noted that saline wastewaters (>10,000 mg L−1) may on the contrary inhibit metabolic processes of mesophilic biomass, thus yielding negative effects on organic and nutrient removal (especially denitrification) (Hong et al. 2007).
|Parameter||M.U.||10th percentile||Average||90th percentile||10th percentile||Average||90th percentile||10th percentile||Average||90th percentile|
|Parameter||M.U.||10th percentile||Average||90th percentile||10th percentile||Average||90th percentile||10th percentile||Average||90th percentile|
Monitoring plan and analytical methods
Influent and effluent COD, nitrogen species, together with operating parameters (dissolved oxygen, pH, temperature, biomass concentration in the biological reactor and excess biomass extraction) were monitored daily for most of the observation period. Based on available data, organic loading rate (OLR) and sludge loading rate (SLR) were calculated. Chemical–physical analyses were performed in compliance with Standard Methods (APHA et al. 2012). Ammonia utilization rate (AUR) were performed at 49 °C using thermophilic biomass sampled into biological reactor (during plant configuration #3) in order to verify the nitrification ability of the biomass (ISO 9509 2006). Interfering substances on the AUR measurements were reduced by washing (centrifuging and resuspending with distilled water) the tested thermophilic biomass. In addition, oxygen uptake rate (OUR) tests and biological oxygen demand (BOD5) measurements were carried out occasionally, using thermophilic and mesophilic biomass. The aim of OUR tests was to evaluate the suitability of plant effluent to be submitted to a subsequent mesophilic polishing stage. Mesophilic OUR tests were carried out at 20 °C using mesophilic biomass withdrawn from a municipal wastewater treatment plant; thermophilic OUR tests were performed at 45 °C using thermophilic biomass withdrawn from the plant monitored (ISO 8192 2007). Both thermophilic and mesophilic tests were carried out with the same initial amount of biomass and substrate dosage. Moreover, BOD5 measurements were performed on plant permeate following two different procedures: in the first case (mesophilic BOD5) the standard BOD measurement was carried out, at a temperature of 20 °C and using a mesophilic biomass as inoculum; in the second case (thermophilic BOD5), the procedure was adapted in order to operate at 45 °C using thermophilic biomass as inoculum. The aim of these tests was, again, the investigation of the compatibility of thermophilic process permeate with a subsequent mesophilic treatment (Collivignarelli et al. 2014).
RESULTS AND DISCUSSION
The biomass concentration in the bioreactor significantly increased after the replacement of the final clarifier with the UF unit and the addition of the NF stage (Figure 2(a)) the total suspended solids (TSS) concentration moved from an average value of about 50 kgTSS m−3 up to 150 kgTSS m−3 (plant configuration #3), to recently (at the end of 2013), 190 kgTSS m−3. The SLR decreased, accordingly, from an average value of 0.14 kgCOD kgTSS−1 d−1 down to less than 0.034 kgCOD kgTSS−1 d−1 (Figure 2(b)). Over time, a reduction in the volatile suspended solids (VSS)/TSS ratio was also observed (from 0.30 to 0.24), probably as a consequence of the increase of sludge retention time (SRT), achieved after the conversion to the MBR system. In fact, SRT increased from 78 d (average value for plant configuration #1) to 125 d (average value for plant configurations #2 and #3). Temperature variations (data not shown) recorded during the operation of plant configuration #1 (average value 45 °C) were reduced by means of the introduction (in 2008) of a heat exchanger on the recirculation line and a control system the average temperature was 49 °C for plant configuration #3 (with a maximum variation range of ±3 °C during the last 6 years). The dissolved oxygen concentration measured in the bioreactor has decreased from 2.3 to 1.4 mg L−1 in the last period (data not shown), likely as a consequence of higher OLR and VSS concentration.
pH (data not shown) decreased from 7.6 (average) down to 6.7 in the third period, despite the basic pH of the fed wastes (see section ‘Wastewater characteristics’). This was caused by the greater CO2 production deriving from aerobic biodegradation of organic substances as reported below, COD removal was enhanced, once the MBR configuration was installed.
Figure 3 shows the trend of the influent and effluent OLR. The influent COD load has been progressively raised (up to 4.3 kg COD m−3 d−1), thanks to the good and stable performance of the system. The residual effluent load, in effect, was reduced to 0.4 kg COD m−3 d−1 in July 2008. Currently, influent COD concentrations over 60,000 mg L−1 are recorded.
Figure 4 clearly displays the improvement achieved after upgrading measures: average monthly COD removal efficiency rose from 69% up to 90%. Efficiency is not negatively affected by increasing the coefficient of variation (CV) of influent load in the first phase, even though it is basically moderate (69%). All along the second phase, yield grows, without a negative effect of CV for values up to 20–25%. Phase 3 shows the highest removal rates and CV effect is further reduced: 33% induces only a slight worsening (from 90 to 87%).
Nitrogen mass balance
Figure 5 shows the nitrogen mass balance calculated based on data from July 2004 to December 2005 (configuration #1) and from January 2007 to October 2013 (configuration #3), by considering ammonia, nitrite, nitrate and organic compounds as measured in the liquid streams; nitrogen in the excess sludge was assumed equal to 8–10% of VSS (Bertanza et al. 2006). Indeed, chemical precipitation as struvite (NH4MgPO4·6H2O) is also reported as possible mechanism in the literature (de-Bashan & Bashan 2004; Juteau et al. 2004; Juteau 2006), but some preliminary tests (still in progress) by means of powder X-ray diffraction showed a negligible presence of this compound. As concerns nitrification, several AUR tests were performed at a temperature of 49 °C with thermophilic biomass (properly washed) taken from the biological reactor. Results showed a very low nitrification rate (<0.01 mgN-NO3− gVSS−1 h−1; data not shown), so that biological ammonia oxidation can be neglected with respect to assimilation and especially stripping (Kurian et al. 2005; Abeynayaka & Visvanathan 2011a). Hence, the stripping contribution can be calculated by subtracting the effluent load of ammonia + organic nitrogen (both suspended and dissolved) to the corresponding influent load. Finally, denitrification (occurring in the anoxic zone of the reactor; see section ‘The waste treatment facility and the thermophilic unit’) extent is given by the difference between N-NOx− influent and effluent loads.
Under these assumptions, the following outcomes can be drawn:
In the first phase, 37% of influent nitrogen was gassified, the contribution of stripping and denitrification being equal to 6% and 31%, respectively. Ammonia stripping was favoured by pH and temperature conditions in the bioreactor in particular, the pH occasionally exceeded 8 (average = 7.6).
In the third phase, 60% of influent nitrogen was gassified. According to our estimation, stripping was negligible and this can be confirmed by the reduction of pH (average 6.7; see section ‘Operating parameters’), with respect to the previous phase.
Specific biomass production
In configuration #1, sludge was not directly extracted from the system, because of the effluent washout. Therefore, the amount of excess sludge corresponded to the effluent VSS load (the concentration being about 1 kgVSS m−3). In contrast, in case of configurations #2 and #3, the actual extraction could be taken into account, since cells are retained by membranes. The amount of extracted sludge was then divided by the removed COD load, so as to obtain the specific sludge yield. Estimated sludge production resulted in 0.101 and 0.092 kgVSSproduced kgCODremoved−1, for configuration #1 and #2–3, respectively, thus exhibiting a 9% reduction. These values are close to the lower limit of the typical range reported in the literature for thermophilic aerobic processes (0.08–0.19 kgVSSproduced kgCODremoved−1, according to Kurian et al. (2005)).
Complementarity between thermophilic and mesophilic processes
Residual biodegradability of UF and NF units permeate was evaluated by carrying out BOD analyses under mesophilic conditions and calculating the BOD5/COD ratio. These measurements displayed evidence that thermophilic process enables not only a direct biodegradation of organic substances, but also favours a subsequent further removal by mesophilic biomass. Figure 6(a) highlights the significant BOD5/COD values (in case of UF and NF, respectively, configuration #3), ranging from 25 to 50%.
OUR tests were then performed under mesophilic and thermophilic conditions, respectively. Based on our findings (mesophilic OURs are almost always greater than thermophilic OURs Figure 6(b)), it can be postulated that mesophilic and thermophilic processes can be seen as complementary for organic substance biodegradation. The mesophilic and thermophilic BOD5 confirm this result (data not shown).
The 11-years intensive monitoring of the full-scale thermophilic plant led to the following reliable operational outcomes for a prompt applicability in design and management of such a system.
Plants having similar configurations enable the operation under load conditions equal to 3.5, 4.0 and 4.3 kg COD d−1 m−3 (TSS concentration in the reactor ranging from 50 up to 190 kg m−3) in the presence of settling tank, UF and UF together with NF, respectively. Relative SRT values range from 70–80 to 120–130 d (without and with a membrane filtration stage, respectively).
Corresponding average expected efficiency of COD removal is 70, 80 and 90% for each aforementioned configuration.
The biological treatability of residual COD under mesophilic conditions (shown by BOD5/COD and OUR measurements) demonstrates the complementarity with a downstream-activated sludge process; this can be crucial in case effluents from similar facilities are discharged into municipal sewers, as often observed.
Wide variations of influent loads are well withstood. Furthermore, they should be maintained below a value of ±20%, in order to guarantee the yields reported above under the specified operation conditions.
Mixtures of wastewater characterized by highly variable compositions can be fed to the plant without affecting process efficiency; a wide range of liquid wastes was fed to the bioreactor, thus leading to a large scattering of concentrations of main parameters of the influent mixture, including COD (12,000–32,500 mg L−1), pH (up to 11) and chloride (up to 1.39%).
Specific sludge production resulted in 0.092 kgVSSproduced kgCODremoved−1, for configurations including MBR and 0.101 without membranes, close to the lower limit of the typical range reported in the literature for thermophilic aerobic processes.
As far as nitrogen removal is concerned, nitrification, and precipitation as struvite results were almost negligible; the main removal mechanisms consisted of denitrification (31–60%), assimilation in the excess sludge (5–11%), ammonia stripping (0–6%), the last occurring only in case of average pH >7.5; the appreciable extent of ammonification of organic nitrogen occurred with MBR configurations might be considered as a positive consequence, since ammonia can be recovered from the effluent by means of stripping and subsequent acidic absorption.
The authors thank Idroclean Spa (Casirate d'Adda – BG – Italy) for giving financial and technical support to the research.