Investigation of a basic nitrification–denitrification biofiltration system for primary wastewater treatment

The scarcity of water often leads countries to turn to alternative water supplies. Treated wastewater reuse is one of them (Hassen et al. 1998). Wastewater reuse covers two concepts, firstly the treatment of wastewater and secondly its reuse. Similar to other European countries and others whose legislation allowed the exploitation of wastewater, Tunisia has always encouraged this initiative of exploitation of this non-conventional resource. For this reason, there is rigorous legislation in Tunisia to control the reuse of treated wastewater (INNORPI NT106.03 1989). This policy of reutilization has been recommended in areas of strong constraints in terms of water availability due to low rainfall and the strong aridity of the climate in Tunisia. This last policy allows the irrigation of crops and green spaces, contributing to the preservation of the environment, and avoiding the withdrawal of water from dams in periods of low-flow water. In addition, it limits the overexploitation of the water table leading to the drying up of certain wetlands and, above all, avoiding the discharge of wastewater directly into the natural environment which protects sensitive downstream water uses, such as bathing areas, recreational activities, and other water uses (Hassen et al. 1998).

Wastewater is essentially water discharged by communities, industries, and individuals. Wastewater is often loaded with pollutants and various organic and inorganic micro-pollutants with discharges of metals, such as zinc, boron, aluminum, cadmium, mercury, etc., but also microorganisms pathogenic or not such as bacteria, viruses, protozoa, fungi, and helminths. The use of wastewater in its raw state is strongly discouraged and not allowed because these waters are vectors of an important sanitary risk for man and its environment.

To limit these problems, wastewater is sent to treatment plants (STEP) which are a set of processes designed to treat various pollutants. A ‘classic’ wastewater treatment plant (WWTP) comprises pre-treatment equipment, a primary treatment, followed by a secondary treatment, often of the biological type. In this last process, microorganisms are the key operators of water purification. The rate of depollution achieved by this simple biological process can meet the standards of discharge into the natural environment. The efficiency of recent plants can reach up to 90% of physical–chemical and microbiological pollution abatement. However, from a perspective of reuse in agriculture, a tertiary treatment of flocculation–coagulation, dephosphorylation, denitrification, disinfection aiming at eliminating persistent and especially microbial pollution is necessary (Brahmi & Hassen 2016; Ibrahim et al. 2020). This treatment may include a disinfection step for microbial inactivation or even their total elimination to fulfill with standards. To improve the microbiological quality of wastewater leaving the WWTP, some new tertiary processes have emerged during these last years, such as the advanced oxidation processes (AOP) by photocatalysis ensuring the generation and formation of hydroxyl radicals (OH°), which provide effective disinfection (Zhou & Smith 2001).

In Tunisia, the percentage of the population connected to a municipal sewage system is 90%. Many villages are not yet equipped with WWTPs. The cost of the installation increases considerably when the density of the inhabitants decreases. Several biological treatment processes are available to remove or reduce nitrogen levels in the water. These include activated sludge and fixed media processes. The latter is well suited to the constraints of small communities: limited budget, space, and human resources. These processes are compact, easily applicable for low flows, and often require less maintenance than suspended media processes (Dichiara et al. 2015). Some treatment plants use fixed-bed processes as complementary tertiary treatment for activated sludge processes. The inclusion of bacteria to the media also concentrates the biomass, increasing its activity, and decreasing sludge production (Vinod & Reddy 2005; Vrtovšek & Roš 2006).

Conventional biological treatment of ammonia nitrogen in WWTPs is achieved by oxidation of ammonia nitrogen via aerobic nitrifying flora, followed by reduction of nitrates to nitrogen by denitrifying flora, under anoxic conditions. In these biological processes, applying wastewater intermittency allows a good renewal of the gaseous phase process, ensures better performance of the reactor operation, and prevents the use of an air compressor, thus reducing the operating cost of the process. However, the reduction of nitrogen in separate columns for nitrification and denitrification has the advantage of separating the biomass of each process and requiring different conditions. It is then necessary to add an external carbon source to provide post-denitrification (Dong et al. 2012). Many carbon sources can be added, such as methanol, acetate, ethanol, sugar, and molasses (Fact 2013). In addition, wastewater recycling can be a source of carbon change in wastewater (Cappai et al. 2004).

This study aimed to evaluate the performance of nitrification/denitrification reactors as a biofilter, regarding nitrogen removal or transformation and formation of a microbial biofilm, for eventual future usage in a rural area without a common sewage system. In parallel, the impact of sewage sludge inoculation, as a source of carbon and enhancement of the operational sewage microbes in the system, is evidenced by the improved water quality at the system's exit.

Finally, the verification of functional relationships between biofilm characteristics and nitrification–denitrification performance will ensure a better design and optimization of the operation of the nitrification/denitrification reactor.

This fixed biomass reactor was designed and built using a PVC column of 315 mm diameter and 200 cm height (Figure 1). At the bottom of this column was placed a layer of gravel 10 cm thick, above which is placed a 5 cm diameter recovery drain made of PVC. This draining mass (coarse gravel) was surmounted by a filtering mass made of 168 cm of finer gravel (8–10 mm). The experimental device was installed in the open air without shelter.

The gravel used as a support for the first nitrification column varied in size between 6 and 8 mm.

The primary water was pumped from the primary clarifier to the nitrification column with a variable flow pump (IM. B-L dosing system, France). The pump feeds the column with a flow rate of 1 L/min through a coil, which ensures a relatively homogeneous distribution of the water over the infiltration range.

The nitrification column was supplied by a daily load of 40 L/day divided into two batches, which allowed a better renewal of the gas phase and good oxidation of the oxidizable matters. The quantity of water added per batch remained in the column for 30 min before being ‘evacuated’. Thus, the water to be treated remained in contact for 30 min with the granular support placed inside the column.

Table 1 summarizes the average physico-chemical and bacteriological characteristics of primary wastewater intended to be nitrogen-free.

Table 1

Average composition of the primary wastewater used during the experiment (n = 16)

Parameters .	Minimum .	Maximum .	Average .
Temperature (°C)	14	29	21
pH	6.8	8.7	7.5
Electrical conductivity (ms/cm)	1.7	2.5	2.2
TSS (mg/L)	49	256	95
COD (mg/L)	148	480	297
BOD5 (mg/L)	90	300	219
N- (mg/L)	32	156	70
N- (mg/L)	0	1	0.5
Fecal coliforms (MPN/100 ml)	1.5 × 103	1.1 × 108	7 × 106
Fecal streptococci (MPN/100 ml)	3 × 102	2.3 × 105	1.2 × 105

Parameters .	Minimum .	Maximum .	Average .
Temperature (°C)	14	29	21
pH	6.8	8.7	7.5
Electrical conductivity (ms/cm)	1.7	2.5	2.2
TSS (mg/L)	49	256	95
COD (mg/L)	148	480	297
BOD5 (mg/L)	90	300	219
N- (mg/L)	32	156	70
N- (mg/L)	0	1	0.5
Fecal coliforms (MPN/100 ml)	1.5 × 103	1.1 × 108	7 × 106
Fecal streptococci (MPN/100 ml)	3 × 102	2.3 × 105	1.2 × 105

View Large

The second denitrification column was also designed and made in PVC, with a diameter of 345 mm and a height of 200 cm. At the bottom of this column was placed a layer of gravel (8–10 mm) of 5 cm thickness, above which was placed a similar 5 cm diameter recovery drain made of PVC. This draining mass was surmounted by a filtering mass made up of 165 cm of sand with a well-defined granulometry. There are two sampling devices installed in the gravel pack. This experimental device was also installed in the open air without shelter in series with the first nitrification column.

The sand used as a support for the second denitrification column was brought from the beach of Salloum, Tunisia in the coastal fringe between the city of Bouficha and the city of Enfida-Ville. It precedes a dry sieving of sand on sieves type AFNOR whose meshes are in geometric progression from bottom to top. The sand used was coarse to very coarse with diameters between 0.5 and 1 mm.

The secondary water was pumped from the water collection basin at the outlet of the first nitrification column with a flow rate of 3 l/min. The column was fed with a daily load of 47 l/day added in a single tank where it was saturated with water for 24 h before being ‘flushed’. From the 8th month of operation, inoculation of the sand filter (C2) with fresh wastewater sludge was performed to provide organic matter to the microorganisms colonizing the biofilm of the filter bed and to optimize denitrification. The sludge was taken from the primary clarifier and mixed with 1 ml/l of pre-treated wastewater. The mixture was then passed onto the filter bed. Monthly five cyclic additions of wastewater sludge were made. The physico-chemical characteristics of the activated sludge used in this study were pH = 7.2; electrical conductivity (EC) = 2.8 ms/cm; TSS = 351.4 mg/L; chemical oxygen demand (COD) = 700 mg/L; and Mohlman's index = 91.7.

To guarantee the system's continued proper operation, regeneration episodes for the two columns are programmed at a specific frequency. This regeneration is done under the action of a vigorous circulation of water in a counter-current for about 30 min.

The microbiological and physico-chemical analysis of the wastewater was carried out on samples taken at different points, as shown in Figure 1. During 22 months of the system operation, wastewater sampling was performed weekly from both columns (C1 and C2) in 1-l polyethylene bottles, taken manually through purge valves installed at the base of the reactor (Figure 1) and stored at 4 °C until analysis.

The main physico-chemical parameters, COD, total suspended solids (TSS), ammonium (⁠⁠), and nitrates (⁠⁠) are analyzed according to standard methods recommended by AFNOR (1992). Bacterial loads in the water were assessed by enumeration of fecal pollution indicator bacteria (fecal coliforms: FC and fecal streptococci: FS) using the most probable number (MPN) method (Rodier 1987).

Biofilm samples were taken from the filter beds (gravel and sand) of both columns according to the feeding time (T0: 1 week after the start of feeding, T1: After seven months of feeding, T2: After 1 week of the first sludge inoculation at the second column (C2), T3: After one month of the first sludge inoculation at the C2, T4: After 1 week of the second sludge inoculation at the C2, and T5: After one month of the second sludge inoculation at the C2) and according to the sample collection level (5, 80, and 168 cm below the infiltration range of the C1 and at 5, 10, 83, and 165 cm below the infiltration range of the C2). Samples were collected in 100 ml sterile bottles and placed at −20 °C until analysis. All analyses were performed within 24 h of sampling.

The biofilm samples are to be examined under the SEM. SEM, as reported by El Abed et al. (2012), lets us explore the biofilm sample surface. The observations can be performed in ‘conventional SEM’ mode where the sample was subjected to a high vacuum, or in ‘low-vacuum’ mode, the latter of which is under nitrogen or air. The EDS (energy-dispersive spectroscopy) detector cooled with liquid nitrogen allowed non-destructive chemical analysis (point or distribution maps) of the samples during the observation.

A mass of 5 g of filtering materials is taken from coarse gravel in C1 and/or fine sand in C2, dissolved by mechanical agitation in 50 ml of distilled water, and centrifuged at 8,000 g at 4 °C for 30 min to remove insoluble material. The supernatant was used for exopolysaccharides (EPS) analysis. EPS was determined by the phenol/sulfuric acid with glucose method recommended by Dubois et al. (1956). The protein content (PN) of EPS was determined according to the method described by Bradford (1976) using bovine serum albumin as a reference protein. The Bradford (1976) technique was preferred to the Lowry protein assay because it was more sensitive and allowed the determination of small amounts of protein in the medium.

C_in and C_eff were the concentrations in the influent and effluent, respectively.

The average value obtained in triplicate from each analysis was subjected to an analysis of variance (ANOVA), and the results were separated by the Student–Newman–Keuls test at p < 0.05.

The operating principle of the fixed biomass process consists of provoking the development of autochthonous bacteria carried by the wastewater, which depending on the time and the flow, gathers in a more or less thick film by physical and chemical action. This film is known as biofilm referring to the mainly microbial activity developed following the retention and assimilation of the organic and mineral pollution carried by the wastewater (Ben Rajeb et al. 2015). It is usually accepted that the formation of a biofilm takes place in four successive stages: transferring bacteria to the support, the initial adhesion of the bacteria to the surfaces, the proliferation helping to form micro-colonies, and the maturation of the biofilm. Thus, the various microbes, mainly bacteria, move from the liquid phase to the solid support by diffusion or convection linked to the dynamic flow of the aqueous solution or active movements due to structures such as flagella of certain microbial species (Van Benthum et al. 1998). Thus, a complex mixture of microbes, proteins, glycoproteins, and organo-mineral nutrients will be implemented, conditioning biofilm formation on the surface of granular gravel or sand materials. Then, a fraction of the microorganisms is adsorbed to the surface. This adhesion to the substrate is still reversible at the beginning and strengthens and grows with the advancement of the water flow and the process operation. Thus, the microbes get closer to the support and create several strong anchoring points. The adhesion then becomes irreversible. The bacterial physiology is then oriented toward a phenotype ‘biofilm’ characterized, in particular, by the overproduction of exo-polymers that will ‘cement’ the bacterial consortium. This irreversible adhesion can lead to isolated cells or even aggregates on the material surface. The colonization stage includes a more or less intense and rapid accumulation by cell division of adhered bacteria, but also by continuous recruitment of bacteria from the liquid phase. This process corresponds to a progressive occupation of the free surface (horizontal colonization) and a development in thickness, helping to form micro-colonies. This biofilm is specified according to time, the flow of pollutants, and especially the chemical quality of these pollutants. Since the waters are commonly loaded with nitrogen, microbial flora specialized in the transformation and assimilation of nitrogen will be benefited. Other microbial communities specialize in the assimilation of phosphates, etc. The increase in biofilm thickness is correlated with the balance between the organic load applied and the intensity of the stresses caused by the prevailing unfriendly wastewater components. This thickness reaches a stationary level when the growth phenomena, conditioned by the access to the substrate, are counterbalanced by the biofilm detachment. The maturation of the biofilm depends on the surface load of the substrate applied, i.e., the quantity of substrates applied by a unit of surface and time. The biomass increases with the nutrients available until the diffusional resistance is too great. The increase in biofilm thickness then reaches a maximum value. At this point, detachment occurs, letting the microorganisms find a more favorable environment (O'Toole et al. 2000).

This detachment of the biofilm can be because of several phenomena such as erosion or damage of biofilm small sections; sloughing, a massive and rapid loss of biofilm related to a sudden change in the environment, and finally abrasion that occurs during collisions between microbial supports. At this stage of biofilm maturation, it is possible to observe a natural detachment of micro-colonies for no apparent reason (O'Toole et al. 2000). The biofilm thus protects the microbes, pathogenic or not, and lets them survive in hostile environmental conditions and benefits water purification.

The initial water coming from the outlet of the primary clarifier of the pilot plant is used to feed the first nitrification column with a hydraulic load of 50 cm/day. This C1 ensures a good circulation of air naturally conditioned with the gravel of rather coarse mesh and guarantees the maximum conditions of aerobiosis and oxidation, bringing the reduced forms of nitrogen in oxidized ones, mainly as N-NO₃. Thus, the C1 column is named the nitrification column, while the C2 is built by a dense mass of siliceous sand, not friable, and more or less fine and becoming more compact with the development of microbial biofilm, and the increase of dissolved oxygen (DO) consumption favoring the dominance of anaerobic denitrifying bacteria; hence the name of the C2 as denitrification column. The nitrification (C1)/denitrification (C2) reactor operates naturally with no oxygen, with rigorous and continuous monitoring of the oxygen levels in both columns. Regeneration episodes of the two columns are programmed at well-determined frequencies to ensure the durability of the good functioning of the system. The placement of sludge consolidated, reinforced, and especially stabilized the biological activity in the columns by a continuous and measured contribution of nutritive matter.

The inoculation of fresh wastewater sludge was carried out in column C2 to provide organic matter to the microorganisms colonizing the biofilm of the gravel pack and to optimize the denitrification process. The sludge was taken from the primary clarifier and mixed with 1 ml/l of wastewater to be treated. Figure 2 illustrates the evolution of the biofilm and its development in the sandy gravel pack to show its purification potential activity.

The dominance of calcium is related to the dominance of all calcareous forms of water in Tunisia represented by the calcium carbonate of the soil. The latter exerts a predominant action on the pH and the activity of ⁠. Silica is present in abundance in natural environments, in the dissolved, crystalline or amorphous state. By hydration, surface hydroxyl groups are formed in the sand and allow the fixation of metal cations by physical and chemical adsorption (Bourg 1988). Silica has a smaller specific surface area than other soil compounds such as clays or amorphous oxides of iron and manganese, which contributes to mask its role and function in the overall retention phenomena, and more precisely the retention of metal cations (Plassard 1999). With the development of the biofilm, we see an enrichment of the biofilm in calcium but a concomitant decrease in oxygen levels.

During the initial stage, the surface of the sand grain showed crevices and anfractuosities. These structures will be seats and places for the installation of microorganisms carried by the water to be treated (Figure 3(a)). Before adding the sludge (T1), the observation of the sand mass showed a ‘moon landscape’ generated by the biomass and the microorganisms, as reported by Nilsson (1990). They form a thin layer of biofilm that encompasses the surface and covers the crevices of the grains. Chemical element analysis revealed that the phosphorus showed an increase at the surface of the sand grains, which can strongly promote microbial growth (Figure 3(b)). But Nilsson (1990) showed that phosphorus content and biofilm formation are closely related. After adding the sludge (T2), the surface of the sand grains became more and more homogeneous, with the formation of a biofilm thick layer (Figure 3(c)). Wastewater containing phosphorus, even at low concentrations, can strongly promote microbial growth, knowing that microorganisms have a high affinity and need for phosphorus assimilation (Schmidt et al. 2011).

In the C1 and at 5 cm of depth, the PN showed a gradual increase to 0.9 ± 0.2 mg bovine serum albumin (BSA)/g media (Figure 4(a)). While this content is revealed more important at 80 and 168 cm depth of the filter bed, with respective values of 1.4 ± 0.25 and 1.45 ± 0.2 g BSA/g media. At the C2 and after seven months of operation (T1), the PN reached does not exceed 0.5 mg of BSA/g of biofilm (Figure 4(b)) for the three depths or levels 5, 10, 83, and 165 cm of the filter column. While after the second introduction of the sludge, the PN increased significantly to 2.25 ± 0.12 mg BSA/g media at 5 cm depth. Toward the first 5 cm of the filter bed, the PN was the highest, but it did not show a significant variation (p < 0.05) as compared to the one recorded at 10 cm depth of the filter media (2.15 ± 0.14 mg BSA/g media). Meanwhile, the amounts of PN recorded at 83 and 165 cm depth of the gravel pack were less significant, and they were in the order of 1.6 and 1.5 mg BSA/g media, respectively.

According to the study of Lazarova & Manem (1995), the PNs in the biofilm are positively correlated with the rate of the substrate (N-⁠) removal. Thus, nitrification seemed greater at 80 and 168 cm depth compared to the one recorded at 5 cm. However, the polysaccharides (PS) > 2.6 mg glucose/g substrate seemed within the typical range usually reported in the literature, as by Ragusa et al. (2004), and with a value of about 5 mg/g. However, these results do not coincide with those found by Mahendran et al. (2012) who registered in a fixed biofilm process a PS varying between 44 and 71 mg/g of dry matter.

Besides, the PS obtained at the first 5 cm were high, but this value did not show a significant variation (p < 0.05) with the one recorded at 10 cm of the filter bed (4.2 ± 0.11 mg glucose/g of media). While at 83 and 165 cm depths of the filter bed, the PS was, respectively, around 3.7 and 3.4 mg of glucose/g of media.

Thus, at the level of the C2, the accumulation rates of EPS seemed higher at 5 cm than at 10 cm depth (Figure 3(d)), which showed active biomass at this level of the filter bed (Di Iaconi et al. 2004). Beyond 10 cm, low levels of EPS were seen. These results agreed with those recorded by Chabaud (2007) and showed that the accumulation rates decreased with the filter depth, and remained relevant in the first 10 cm; but at 30 and 50 cm, these EPS rates are insignificant. Polymeric materials were mainly present at the top of a sand column, as reported by Hilger et al. (2000). These last results showed a developed biofilm between 0 and 10 cm for the reactor supplied at 70 cm/day. According to Chabaud (2007), biofilm is more likely to develop in the 0–2.5 cm part of a reactor fed at 5 cm/day, and in a deeper 0–10 cm zone of a reactor fed at 12 cm/day. The infiltration zone, therefore, will develop deeper because of the importance of a load of wastewater supply at this level.

In the C1, the PN/EPS ratio increased exponentially with the operating time reaching a value of 0.25 ± 0.02. While the study by Mahendran et al. (2012) showed that the PN/PS ratio varied between 1 and 1.2 in a fixed biofilm system. But, following the addition of the sewage sludge in the C2, the PN/PS ratio became constant at a value of 0.5 ± 0.02. In a similar study conducted on developed biofilm in a wastewater irrigation system, Yan et al. (2012) have seen PN/PS mass ratios between 0.21 and 1.9 and showed that EPS was more abounding than protein in the mature biofilm.

The resulting contents of the parameters of COD, BOD5, and TSS during this experiment are summarized in Table 2. The TSS content at the inlet of the nitrification/denitrification reactor varied with the feeding time. At the exit of the reactor, the average TSS was around 16 ± 0.25 mg/L and agreed with the EHS (Environmental, Health, and Safety) guidelines. The removal of TSS by the nitrification/denitrification system is revealed significant; it was of the order of 83.5%. This result agreed with the one reported by Tyagi et al. (2009), who registered an average TSS removal of about 90%.

Table 2

The mean values of physico-chemical and bacteriological parameters at the inlet and outlet of the first and second column reactor

.	.	.	Physico-chemical parameters .						Fecal indicator bacteria .
.	.	.	Content (mg/L) .						MPN of bacteria (MPN/100 ml) .
.	.	Level of abatement .	COD .	BOD5 .	TSS .	N- .	N- .	N- .	FC .	FS .
Input			340 ± 2 d	229 ± 2 c	97 ± 1 d	63 ± 0.06 d	0.1 ± 0.01 a	0.22 ± 0.02 a	1.2 × 105 d	7 × 106 c
Output C1			114 ± 04 c	56.2 ± 2 b	35.3 ± 0.6 c	25 ± 0.06 c	7.8 ± 0.02 c	24.5 ± 0.05 d	6.43 × 103 c	8.8 × 104 b
Output C2	Without sludge		84 ± 2 b	28 ± 2 a	16 ± 0.25 a	7.5 ± 0.02 b	6.4 ± 0.2 c	17.5 ± 0.06 c	9 × 102 a	7.7 × 102 a
	With sludge		70 ± 3.5 a	26.4 ± 1.4 a	24 ± 0.5 b	3.7 ± 0.67 a	2 ± 0.3 b	8.1 ± 0.28 b	9.3 × 102 a	1.7 × 103 a
Removal efficiency (%)	Without sludge		75.3	88.4	83.5	88.1	ND	ND	3.2 U-log10	4.8 U-log10
	With sludge		79.4	88.5	75.3	94.1	ND	ND	3.3 U-log10	3.7 U-log10

.	.	.	Physico-chemical parameters .						Fecal indicator bacteria .
.	.	.	Content (mg/L) .						MPN of bacteria (MPN/100 ml) .
.	.	Level of abatement .	COD .	BOD5 .	TSS .	N- .	N- .	N- .	FC .	FS .
Input			340 ± 2 d	229 ± 2 c	97 ± 1 d	63 ± 0.06 d	0.1 ± 0.01 a	0.22 ± 0.02 a	1.2 × 105 d	7 × 106 c
Output C1			114 ± 04 c	56.2 ± 2 b	35.3 ± 0.6 c	25 ± 0.06 c	7.8 ± 0.02 c	24.5 ± 0.05 d	6.43 × 103 c	8.8 × 104 b
Output C2	Without sludge		84 ± 2 b	28 ± 2 a	16 ± 0.25 a	7.5 ± 0.02 b	6.4 ± 0.2 c	17.5 ± 0.06 c	9 × 102 a	7.7 × 102 a
	With sludge		70 ± 3.5 a	26.4 ± 1.4 a	24 ± 0.5 b	3.7 ± 0.67 a	2 ± 0.3 b	8.1 ± 0.28 b	9.3 × 102 a	1.7 × 103 a
Removal efficiency (%)	Without sludge		75.3	88.4	83.5	88.1	ND	ND	3.2 U-log10	4.8 U-log10
	With sludge		79.4	88.5	75.3	94.1	ND	ND	3.3 U-log10	3.7 U-log10

Means (n = 3) followed by the same lowercase letter are not significantly different according to the Student–Newman–Keuls test (p ≤ 0.05); COD, chemical oxygen demand; BOD5, biological oxygen demand after 5 days; N-⁠, ammonia nitrogen; N-⁠, nitric nitrogen; N-⁠, nitrate nitrogen; FC, fecal coliforms; FS, fecal streptococci; TSS, total suspended solids; C1, first column of the reactor; C2, second column of the reactor.

View Large

After adding the fresh sewage sludge, the TSS contents recorded at the exit of the nitrification/denitrification process showed a net increase, from 16 ± 0.25 to 24 ± 0.5 mg/L before and after the sewage sludge addition. These last results were confirmed statistically since a significant variation between the two mean values was registered according to the Student–Newman–Keuls test at p < 0.05. Adding the sewage sludge within the C2 of the sand filter process allowed a high growth and development of the biofilm biomass, which is conditioned by the availability of various easily metabolizable substrates within the filtering material. Christensen & Harremoës (1978) have mentioned that when a biofilm reached a determined thickness, biomass detachment occurred. As explained by Karnchanawong & Polprasert (1990), the biofilm detachment was caused and conditioned by the natural death of the deep layers of the biofilm and hydraulic shear. This may explain the elevated TSS observed at the exit of the nitrification/denitrification system after adding fresh sewage sludge.

For the COD content at the exit of the filtering process, the COD varied between 70 and 84 mg O2/l. This result corresponded to an abatement of 75.3 and 79.4% for the process supplied at a hydraulic load of about 50 cm/day, and without or with sewage sludge supplementation, respectively. Two phenomena can be to blame for the reduction of the organic compound load. The first one is related to the effect of physical filtration; the second one is mainly because of the degradation processes and/or biological consumption. The cyclical addition of the fresh sewage sludge in the system allowed an average increase in COD removal, from 75.3 to 79.4% under a hydraulic load of about 50 cm/day. A similar study by Potts et al. (2004) showed a COD removal higher than the one achieved in the present work, with an abatement of around 96% for a sand system supplied with wastewater at 4 and 12 cm/day. Thus, hydraulic loading can effectively influence the performance of a filtering system.

But, Rahimia et al. (2011) showed that COD removal by a sequential biological nitrification/denitrification filtering reactor varied between 85 and 95%. These results showed a COD increase to 80 mg O2/l after 1 week of inoculation with fresh wastewater sludge. The study by Vrtovšek & Roš (2006) showed that the effectiveness of denitrification increased noticeably as COD increased. COD decreased after 4 weeks of operating. Hence, inoculating the C2 of the reactor cyclically during each month must prevent the drop in COD wastewater contents. Wang et al. (2008) showed that the COD removal efficiency decreased continuously and stabilized at around 25% after 2 weeks of processing because of the lower organic load of the used secondary wastewater.

Considering the performance of the nitrification/denitrification reactors in terms of BOD5 reduction, an abatement of about 88% is obtained both before and after adding the fresh sewage sludge. Although the BOD5 removal efficiency is revealed high following the addition or not of fresh waste sludge, with around 88%; the statistical results of the Student–Newman–Keuls test (p < 0.05) revealed there is no significant variation between the BOD5 at the exit of the nitrification/denitrification reactor before and after adding fresh sewage sludge. The BOD5 and COD levels followed EHS guidelines, recommending values between 30 and 125 mg/L for BOD5 and COD, respectively. The BOD5 removal results agreed with those reported by Tyagi et al. (2009) who presented BOD5 removal greater than 80%. In addition, the decrease in chemical load expressed by COD was less significant than the one recorded for biological oxygen demand (BOD). This result may be because of the fraction of COD that escapes degradation by the purifying biomass and that is slowly and hardly biodegradable. The COD/BOD ratio 5 showed the biodegradability character of the wastewater. The higher the ratio, the less wastewater was biodegradable (Metcalf & Eddy 2003; Hassen et al. 2022).

The COD/BOD5 ratio at the exit of the nitrification/denitrification filtering process averaged about 3.5 ± 0.9 and 1.9 ± 0.36, before and after adding the wastewater sludge, respectively. The change in the COD/BOD5 ratio by adding fresh sewage sludge improved the biodegradability of the organic compounds in the wastewater. In addition, adding the sludge led to an increase and development of the purifying biofilm within the filtering process. This increase raised the passage time of the wastewater, through the filtering materials, and the contact time and reaction of the organic matter with the purifying biomass.

The N- contents recorded at the exit of the C1 were, on average, about 24 mg/L (Figure 5(a)). A better N- transformation yield is obtained with an average value of around 96%. The N- removal efficiency obtained by the nitrification/denitrification reactor was higher, around 96% than the one reported by Rodríguez et al. (2011). These authors showed that the nitrogen removal by nitrification/denitrification process in a sequential biological reactor was only 70%. Although they worked on an aerated sequencing batch reactor (SBR) system equipped with a submersible aerator (AQUA 200). This ammonia oxidation efficiency can be attributed to the good oxygenation of the filter bed, of the order of 7 mg O2/l value close to saturation with the content of 9 mg O2/l at 20 °C (Tromans 2000). The oxygenation of the filtering pack is ensured by the fractional/sequential water supply of the nitrification/denitrification reactor to allow oxygen transfers in order not to alter the degradation of organic matter and to develop a clogging too important of the filtering system (Leverenz et al. 2009).

The nitrate nitrogen contents of 24 mg/L N-NO₃ recorded during the experimentation in the cold season with an average ambient temperature of 15 °C at the exit of the C1 of the reactor, ensuring the transformation of ammonia nitrogen NH4-N, were lower than the one of 34 mg/L N-NO₃ obtained during the hot season with an average ambient temperature of 35 °C) (Figure 5(b)).

The nitrification of is a biological process related to the oxidative action of nitrating bacteria such as Nitrobacter, Nitrococcus, Nitrospira, which oxidize nitrite to nitrate and are conditioned by the oxygen availability in the biofilter. Low water temperatures at the inlet of the nitrification/denitrification reactor slow down the kinetics of nitrogen oxidation and severely limit the nitrification process (Grey 1989). At a temperature of 20 °C, the accumulation of nitrite was low at the exit of the C1 of the reactor, which shows an effective oxidation and transformation into nitrate. Likewise, the variation of the O2 content inside the filtering material could also be involved. A decrease in DO at the exit of the C1 of 4 mg O2/l can be noted.

The N- content obtained at the exit of the nitrification/denitrification system was about 80 mg/L, and this value exceeded the value assigned by the WHO guidelines (50 mg/L of N-⁠). However, after sludge inoculation, a decrease in N- content at the exit of the nitrification/denitrification reactor was noted, with a value of around 28 mg/L (Figure 5(b)). This value meets the WHO standards. After one month of inoculation, the N- levels recorded at the exit of the reactor increased again. Further inoculations with waste sludge were applied, and these nitrate levels decreased again to an average value of about 27 mg/L of N- at the reactor exit.

The study conducted by Ouyang et al. (2000) with a combined pre-denitrification/nitrification submerged biofilter showed that N- recycling can be removed in an anoxic biofilter fed with synthetic wastewater. With a large residence time of 12 h, denitrification can be achieved inside an aerobic biofilter. Thus, at a long residence time, the denitrification performance can be improved, although the low denitrification performance caused by the limitation and lack of bioavailability of bio-assimilable carbon in wastewater imposed and required the addition of an exogenous carbon source to improve wastewater treatment (Marchetto et al. 2003). Vocks et al. (2005) showed that even without adding an external carbon source, denitrification rates well above endogenous rates could be achieved by a post-denitrification system and that the anaerobic reactor can have a positive impact on denitrification rates. Endogenous denitrification was significant when it was ranging between 0.2 and 0.6 mg N-/h g VDM (Kujawa & Klapwijk 1999). These data were obtained from conventional activated sludge processes. However, Innocenti (2004) measured an endogenous denitrification rate in a membrane biological reactor (MBR) of less than 0.002 mg N-/h g VDM after 24 h of aeration. Consistent with the literature, the denitrification rate recorded at the C2 of the reactor operating under anaerobic/anoxic conditions (DO about 1 mg O2/L) was between 0.3 and 0.8 mg N-/h g VDM after 24 h of aeration.

In this study, with decreasing N- levels, the COD/N ratio showed an increase from 4 to 11. Liu et al. (2010) showed that with decreasing nitrates, the COD/N ratio also showed an increase from 3 to 5, while Carrera et al. (2004) showed that to ensure complete denitrification, the COD/N ratio must increase from 4 to 7.1. The greatest consumption of organic matter inside the filtering system also occurred after inoculation of this organic matter, with values ranging from 5.6 to 11 mg COD consumed/mg N- removed, depending on the COD/N ratio used. The organic matter consumed during this period would correspond to the readily biodegradable fraction in the wastewater (Peng et al. 2007). In our study, the denitrification process became incomplete with low COD levels, resulting in increased nitrate levels in the water at the exit of the nitrification/denitrification system. Nitrite accumulation was seen, with a maximum of 5.5 mg/L, while this is not the case with high COD levels.

The bacteriological results essentially showed that the number of fecal bacteria is effectively reduced in the nitrification/denitrification filtering system. For example, the fecal coliform removal showed an average of 1.6 log10 units at the exit of the C1 throughout the operational experimentation (Table 2). The fecal coliform abatement registered at the exit of the nitrification/denitrification system ranged from 1.7 to 5.4 log10 units. This range of values confirmed the good bacterial abatement of the filtering system.

After inoculation of the system with fresh sewage sludge, the number of fecal bacteria grew little, with average numbers of 9.3 × 10² and 1.7 × 10³ MPN/100 ml, respectively, for FC and FS. These numbers of fecal bacteria before inoculation were for CF = 9 × 10² MPN/100 ml and FS = 7.7 × 10² MPN/100 ml, respectively. The number of FC was within the standards (2 × 10³ MPN/100 ml) recommended by the European Bathing Water Quality Directive. While the number of FS is slightly higher than the standard with 10³ MPN/100 ml recommended by the European Bathing Water Quality Directive. The number of FC and FS in the wastewater treated by the nitrification/denitrification filtering system showed a significant variation according to the Student–Newman–Keuls test at p < 0.05, before and after the inoculation with wastewater sludge (Table 2).

Despite the good performance of the nitrification/denitrification filtering systems or biofilter regarding the removal of fecal bacteria, the quality of the treated water does not follow the standards of the European directive regarding the number of FS. Torrens et al. (2009) reported that the removal rates of the fecal bacteria by the biofilter would depend mainly on the retention time of the wastewater in the filtering system, the depth of the filter, the hydraulic load, and the volume of wastewater brought per batch with an intermittent filling of the system. That means bacterial removal was directly related to the applied load, with around 50 cm/day in our case. The higher the load, the lower the microbial removal, as reported by Stevik et al. (1999). Also, the wastewater load in the biofilter largely influenced the adhesion and trapping material processes as described by Pundsack et al. (2001). Van Cuyk et al. (2001) confirmed that increased hydraulic retention time speeded up and intensified bacterial inactivation.

Besides, the temperature appeared and remained the major important factor of microbial abatement in water treatment. It is well known that temperature is a vital and determining parameter that formally affects the survival of all living and their development. In the present study, the number of fecal bacteria registered during the cold season of winter with a usual temperature below 15 °C was lower than the one obtained during the hot season of summer with a usual temperature above 20 °C. These findings follow those of Pundsack et al. (2001), who found a clear seasonal impact on the overall performance of the treatment system.

After inoculation with sewage sludge, the number of fecal bacteria became more important. This variation in bacterial removal rates can be related to the variation in environmental conditions, before and after adding fresh sewage sludge, like predation by protozoa, and variations in the supply of organic matter.

The nitrification/denitrification biofilter system investigated in this study could reduce almost completely the total nitrogen load, and let the treated wastewater average contents under World Health Organization (WHO) and EHS guidelines, except for N- levels, which exceeded the standards of 50 mg/L of N-⁠. Cyclic inoculation or intermittent supply of fresh sewage sludge to the C2 of the biofilter system allowed for a reduction in N- levels, allowing for compliance with WHO and EHS standards and guidelines. The sewage sludge inoculation of the C2 affected slightly but not significantly the bacteriological qualities of the treated water at the exit of the reactor by increasing the load of fecal bacteria, and it is still recommended and advisable to associate a safe and non-bulky disinfection system, such as UV-C254 irradiation at the exit of the C2 of the nitrification/denitrification bio-filtering system. This typical bio-filtering system has been tested in at least two rural pilot areas in Tunisia, confirming its usefulness for reuse in the agricultural sector or for treating rural effluents often discharged in natural watercourses, causing harmful effects on the chemical and biological quality of streams and altering the natural environmental landscape.

This work was carried out within the framework of a Contract Program project entitled ‘Monitoring and treatment of nitrates in polluted water’ and subsidized by the Tunisian Ministry of Research and High Education. The authors want to thank the staff of CERTE, in particular Nessrine Khelifi, Hechemi, and Mondher Zairi, for their valuable help during this work.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.