Effect of biochar particles applied in bedding course of the innovative permeable pavement on enhancing nitrogen removal

Urbanization involves artificial structures and impermeable pavements replacing soil and vegetation, meaning the area covered by impervious surfaces increases dramatically. Impervious surfaces prevent rainwater infiltration and evaporation, and this intensifies the urban heat island effect (Yamagata et al. 2008). Stormwater runoff from impervious surfaces pollutes natural water bodies and exacerbates urban non-point-source pollution (Li & Davis 2014). Permeable pavements (PPs) are effective at preserving the hydro-ecological balance and have been used widely for paved areas, such as sidewalks, parking lots, and urban squares (Drake and Bradford 2013; Bean et al. 2019).

Nitrogen management is important for controlling urban non-point-source pollution (Preisner et al. 2020). However, it has recently been found that PPs may control nitrogen concentrations in runoff to some degree (Abdollahian et al. 2018), but NO₃⁻-N is often leached from PPs (Bean et al. 2007; Collins et al. 2010). NO₃⁻-N is mainly removed from runoff through denitrification. Anoxic conditions and the presence of ample electron donors can promote denitrification. Collins et al. (2010) prepared a PP with a sand layer 10 cm thick and found that the sand layer provided a large surface area to which microbes attached and on which a biofilm could form to enhance total nitrogen (TN) removal. In a permeable interlocking concrete pavement (PICP), the sand layer is very thin (2–3 cm) and does not effectively retain moisture. The results in conditions unfavorable for microbial growth. Systems lacking anoxic microsites give inefficient denitrification (Bean et al. 2007).

Nevertheless, relevant results for enhancing PPs' nitrogen removal have rarely been reported. Ostrom & Davis (2019) used elevated drainage outlets and sodium acetate as a source of carbon in a pilot PP experiment and found that the system decreased the TN load by 33% and the NO₃⁻-N load by 34%. However, assessments of correlations have indicated that internal water storage zones do not always favor nitrogen removal and can cause unstable nitrogen leaching. Experiments showed that nitrogen removal of PPs is affected by the aggregate properties, hydraulic retention time, loading method, and temperature, and the removal efficiencies are far from satisfaction (Hunt et al. 2006; Davis 2007; Braswell et al. 2018).

Liu et al. (2018) developed an evaporation-enhancing PP to mitigate the urban heat island effect. This innovative pavement (IPP) has a liner at the bottom and an overflow port 15 cm above the bottom to form an internal water storage zone. Terracotta capillary columns lift water to the pavement surface through capillary action. The structure ensures a water supply to the pavement surface to allow evaporation to occur and prolongs the cooling effect. IPP can be applied in high water table areas to prevent groundwater contamination by infiltration. However, results of the pilot experiment showed that the removal rate of TN of the IPP was not obviously improved.

To enhance nitrogen removal of PPs, mainly for the type as PICP, biochar was added to the bedding course sand of PPs in this study to improve the water retention capacity of the bedding courses. It is expected microbes will attach to and grow on biochar, then nitrification and denitrification will be enhanced. The combined effects of capillary action and the biochar in keeping the bedding course moist were investigated. The novel modified bedding layer was expected to improve the denitrification efficiencies of PPs, and denitrification efficiencies of innovative PPs and of conventional PPs will be assessed.

Two conventional PP units and four innovative PP units were constructed. The PPs were described in a previous publication (Liu et al. 2018). Each PP was a bench-scale unit made of polyvinylchloride and was 210 mm long, 210 mm wide, and 400 mm high. From bottom to top, each unit consisted of 30 cm of graded aggregate, a geotextile, a bedding course 3 cm deep, and 6 cm of terracotta bricks. Capillary columns were installed within the aggregate in the innovative PP units to give a capillary column section/unit section area ratio of 1:8. The conventional PP units did not contain capillary columns. An overflow port was installed 150 mm above the bottom of each unit. Schematics of the units are shown in Figure 1.

To improve bedding course characteristic and enhance nitrogen removal, coconut shell and bamboo biochar, mixed with sand of bedding course in a certain proportion of mass. These two types of biochar were chosen because they have higher density and strength so that they are less likely to be crushed. Basic characteristics of the two biochars were provided by Lvyuan Environmental Technology Co., Ltd (Henan, China), which the biochars were purchased from. Biochar is granular, with a particle size of about 3 mm and a density of about 1.1–1.3 g/cm³. Specific surface areas of coconut shell and bamboo biochar are 165 and 143 m²·g⁻¹, respectively. Pore volumes of coconut shell and bamboo biochar are 0.16 and 0.12 cm³ ·g⁻¹, respectively. The units were designed in three groups of tests, which were blank control pavement (BCP) tests, coconut shell biochar pavement (CBP) tests, and bamboo biochar pavement (BBP) tests. The compositions of the bedding courses used in the six types of units that were used are shown in Table 1.

Steady and uniform simulated rainfall was produced using a clear plastic sheet divided into six boxes, on the bottom of each were 5 mm × 5 mm holes. Each hole was plugged with a geotextile in a funnel shape to slow the water droplets released from the hole. The simulated rainfall intensity was adjusted by adjusting the number of holes. The actual rainfall intensity was determined using the volumetric method. The plastic sheet was attached to a structure on wheels to allow the system to be moved to above the units during rainfall and to be removed after the rainfall had ended. A photograph of the system is shown in Figure 2.

Artificial runoff containing the components shown in Table 2 was used as the influent. The artificial runoff was prepared to match the water quality of general runoff from a parking lot (Tota-Maharaj & Scholz 2010).

A 2 L aliquot of outflow water from the secondary sedimentation tank of a wastewater treatment plant in Shanghai was added to each unit to add appropriate microbes and quickly promote biofilm formation. Starting 3 d after the outflow water had been added, residual nitrogen was flushed from each unit using tap water on a 2 d cycle for two weeks before the experiment was started.

The simulated rainfall intensity was selected to match adverse conditions in Shanghai. The total rainfall capacity of each experiment was 40 mm, and tests were performed using rainfall intensities of 10, 20, and 40 mm/h. For each unit, a 250 mL aliquot of effluent was collected at the end of each feeding period and the NH₄⁺-N, NO₃⁻-N, and TN concentrations and chemical oxygen demand (COD) were determined using standard methods (APHA 2012) within 24 h of the sample being collected. Nitrogen removal data for the different units were compared, and organic matter leaching was assessed. Simulated rainfall events were performed at 1 week intervals, and three tests were performed at each rainfall intensity.

The experiment lasted from August to December 2020. The units were placed on the roof of a building in the Tongji University campus in Shanghai from August to November. The units were then moved to a shed with a transparent roof and wrapped in foam insulation when the weather turned cold (5–10 °C) in December. The simulated rainfall intensity in December was 10 mm/h, but the inflow cycle was unchanged. The results acquired in September (rainfall intensity 10 mm/h, ambient temperature 22–28 °C) were compared with the results acquired in December to investigate the influence of the temperature on denitrification.

The nitrogen mass balances for the units were calculated after analyzing samples of the media in the bedding courses of the units. Each sample was dried, ground, and passed through a 200-mesh sieve, then the nitrogen content was determined using a Vario EL III elemental analyzer (Elementar, Hanau, Germany) (Wang et al. 2021).

The TN masses concerning liquid phase were determined from influent and effluent quantities and concentrations of each feeding operation.

At the end of the experiment, DNA was extracted from samples of the media in the bedding courses of the units using soil DNA kits. The target genes nirK, nirS, and 16S rRNA were quantified by performing quantitative polymerase chain reaction assays. Each sample was analyzed in triplicate.

The data were analyzed using SPSS version 20 software (IBM, Armonk, NY, USA). The data were found to have normal distributions at the level α = 0.05. Differences between denitrifying gene copies were identified by performing least significant difference tests as part of one-way analyses of variance. Differences between nitrogen concentrations were identified by performing Gams–Howell tests because the variance was heteroscedastic.

The mean NH₄⁺-N concentrations in the effluents from the experimental and control units were not significantly different (Figure 3), indicating that the composition of the bedding course had little effect on the nitrification efficiency of the biofilm in the bedding. The mean NH₄⁺-N removal rates for unit CBP2 at rainfall intensities of 10, 20, and 40 mm/h were 83.8%, 82.8%, and 82.1%, respectively, indicating that increasing the rainfall intensity had little effect on NH₄⁺-N removal rate (p > 0.05). However, the NH₄⁺-N concentrations were higher in the effluents in December than in the other months. The NH₄⁺-N removal rates for the units in December, when the ambient temperatures were 5–10 °C, were 51.2%–53.6%. This would be attributed to the nitrifying bacteria activities decreasing as the temperature decreased (Zhang et al. 2014).

Removal efficiencies of different units under different operation conditions for NO₃⁻-N were shown in Figure 4. The units containing biochar performed better than the units not containing biochar. The NO₃⁻-N removal rate was better for unit CBP3 than unit BCP2, the mean removal efficiency for unit CBP3 being 41.4% at a rainfall intensity of 10 mm/h and at 22–28 °C. The removal rates for the innovative PP units containing biochar were 54.0% under the same conditions. This indicated that adding biochar to conventional PPs could improve the nitrogen removal efficiency, moreover, the improvement in nitrogen removal would be better for innovative PPs than conventional PPs. The biochar had good water retention properties and could maintain a high water content in the bedding course between rainfall events. The microorganism quantity and activity in the bedding course was therefore maintained well by the biochar particles, and local anoxic environments would have formed in pores of biochar to ensure that denitrification occurred (Qiu et al. 2020). However, the bedding courses of the conventional units quickly became dry between rainfall events because gravity drew the water down and capillary water evaporated, so the biochar particles would gradually have dried out. The bedding courses in the innovative PP units remained moist because of strong wicking by the capillary columns, and this would have helped to maintain the denitrification effect. During the experiment, the NO₃⁻-N concentration was lower in the CBP2 effluent than the BBP effluent and the mean NO₃⁻-N removal rates for units CBP2 and BBP were 44.6% and 41.6%, respectively. This indicated that NO₃⁻-N was removed more effectively by the units containing coconut shell biochar than the units containing bamboo biochar. More NO₃⁻-N leaching occurred as the rainfall intensity increased or the temperature decreased, and this could cause the concentration in the effluent to be higher than the concentration in the influent. Less denitrification occurred at low temperatures than at high temperatures. The results for December and September were compared. NO₃⁻-N leaching occurred in the control units in December (low temperatures) but not September (high temperatures), and the NO₃⁻-N removal rates for the experimental units were reduced to 29.3%–41.2% in December.

The experimental units effectively decreased the TN concentration (Figure 5). In September, the mean TN removal rate for unit BCP1 was 14.1% and the overall mean TN removal rate for unit BCP2 was 6.76%, while the TN removal rates for the innovative units CBP2 were 57.7% (Figure 6), and the TN removal rate for conventional unit CBP3 was 48.8%. Nitrogen in the effluent was mainly in the form NO₃⁻-N, and the TN removal rate varied in a similar way to the NO₃⁻-N removal rate. The TN concentration in effluent was significantly affected by the rainfall intensity (p < 0.01) and temperature (p < 0.01). The TN removal rates for the innovative PP units were 52.6%–57.7% when the rainfall intensity was 10 mm/h and at 22–28 °C. Ning (2006) found that 99.7% of the rainfall events in Shanghai in the last two decades had rainfall intensities ≤10 mm/h. This indicated that the PP system with biochar in bedding, especially for the innovative one could give good nitrogen removal rates during most rainfall events in Shanghai.

Removal efficiencies of different units under different operation conditions for COD were shown in Figure 7. The COD removal rates of the experimental units at rainfall intensities higher than common natural rainfall intensities were 83.7%–87.1%, which were higher than achieved in practice using PPs not containing biochar (Niu et al. 2016). The COD in effluent from the experimental units remained <20 mg/L throughout the experiment, indicating that adding biochar to the bedding course at a mass ratio of 6% would not lead to marked leaching of organic matter (AMHL et al. 2018).

There were low number of microorganisms in each unit at the start of the experiment, and the initial source of nitrogen was mainly from organic material such as the biochar in the bedding course. Negligible nitrogen was retained by the inorganic materials such as the terracotta bricks and aggregate at the end of the experiment. The bedding course with large inner and external surfaces was the main medium in which microorganisms occurred, so there was more biomass in the bedding course than elsewhere. This meant that most of the nitrogen in each unit was in the bedding course. The nitrogen balance for the bedding course could therefore be used to represent the overall nitrogen balance for the unit. The nitrogen contents of the two types of biochar were determined. N_in was calculated by multiplying the nitrogen content of the biochar by the mass of biochar. N_r was determined by analyzing samples of the bedding course media. The results are shown in Table 3. The bedding course of the control group contained no organic matter and very small numbers of microorganisms, and the initial and retained amounts of nitrogen in the media were below the detection limit, so each related value was given the value zero. The TN masses concerning liquid phase were determined from influent and effluent quantities and concentrations of each feeding operation (Table 4).

The fates of the initial and added nitrogen in each unit were assessed using three possibilities, retained in the bedding course, discharged in the effluent, and removed from the system.

The amount of nitrogen retained by the bedding course was different for each unit and was affected by the biochar dose and type. For the experimental units, the amount of nitrogen discharged in the effluent was <5% of the total amount of nitrogen added to each unit. Nitrification and denitrification removed 781.58 mg of nitrogen from unit CBP2. This was more than the 733.20 mg of nitrogen removed from the conventional unit CBP3 with the same biochar amendment, indicating that adding biochar improved nitrogen removal more in the innovative PP units than in the conventional PP units. Biochar in a conventional PP without capillary columns was easy to dry out, then, the temperature of the bedding course would have increased when the sun caused the temperature of the surface paving to increase, and this may have decreased the activities of microbes on the biochar. Nitrogen removal by the innovative PP units was not much better than that by conventional PP units according to Figure 8, which might be because during dry periods when the pavers were dried out and their surfaces were very hot under the noon sun, living conditions for microbes within biochar in the bedding course were not so difficult. The problems will be investigated in a future study.

Samples from the bedding course of each unit were collected at the end of the experiment. The amounts of denitrifying microorganisms in the units were assessed from the nirS and nirK gene copy numbers.

As shown in Figure 9, the biochar markedly increased the abundances of the nirS and nirK genes in the units. The nirS and nirK gene abundances were 3.04 and 1.29 times higher, respectively, in the CBP3 samples than the BCP2 samples. The nirS and nirK gene abundances were 5.50–8.46 and 3.74–5.29 times higher in the samples from the innovative PP units containing biochar than the BCP1 samples, respectively. The 16S rRNA copy number has been used to indicate the total number of microbes (Fierer et al. 2005). Biochar can increase the biomass by providing a large surface area for microbes to become attached to and by increasing the moisture content of the bedding course. The effluent analysis results indicated that the gene copy numbers for the units positively correlated with the degree of denitrification. The nirS and nirK gene copy numbers in the CBP2 samples were 1.46 × 10⁷ and 1.62 × 10⁷ copies/g, respectively, which gave the strongest denitrification effect. The nirS and nirK gene copy numbers were an order of magnitude lower for the blank, BCP1 and BCP2 samples than for the CBP2 samples, i.e., weaker denitrification occurred in the BCP1 and BCP2 units. The copy numbers for the CBP1 and CBP2 samples were significantly different for all of the genes that were quantified (p < 0.05), indicating that the increase in biomass was related to the amount of biochar added. The nirS gene abundance can reflect anoxic conditions (Henderson et al. 2010). The nirS gene copy numbers were markedly higher for the experimental units than the control units, indicating that the biochar changes contributed to the formation of local anoxic environments.

This study strengthens the nitrogen removal effect of the innovative permeable pavement by adding biochar to the bedding course, which is very promising for application. Subsequent field trials can be conducted to verify the practicability of IPP.

• (1)

  Adding biochar to a PP bedding course can increase the amount of nitrogen removed. The NO₃⁻-N removal rates for the innovative PP units containing biochar were 48.6%–54.0%, and the TN removal rates were 52.6%–57.7%, which will not cause problems of organic matter leaching.

• (2)

  Coconut shell biochar is a better choice than bamboo biochar for adding to PP units. The innovative PP units performed better if the bedding course contained 6% coconut shell biochar than if the bedding course contained 3% coconut shell biochar. The innovative PP units performed better than the conventional PP units with the same amounts of biochar added. This would have been because the combination of capillary columns and biochar caused the bedding course to maintain a high moisture content and therefore local anoxic environments were maintained in the biochar particles.

• (3)

  Denitrification in the units with biochar added was markedly weakened during heavy rain and at ambient temperatures <10 °C.

This research was supported by the National Water Pollution Control and Management Technology Major Project (No. 2017ZX07207001).

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