A novel sequential mechanical filter system was developed as an alternative primary treatment method for onsite wastewater treatment. The filter combines traditional screening with a novel type of counter-flow filter using wood-shavings as a biodegradable filter matrix. This study tested the system in a batch loading regime simulating high frequency toilet flushing using blackwater from a student dormitory. The filter removed 78–85% of suspended solids, 60–80% of chemical oxygen demand, and 42–57% of total-P in blackwater, giving a retentate with a dry matter content of 13–20%. Data analysis clearly indicated a tendency towards higher removal performance with high inlet concentrations, hence, the system seems to be most applicable to blackwater or other types of wastewater with a high content of suspended solids.
Filtration has gained increasing attention as an alternative primary treatment method. For municipal treatment plants, advanced mechanical filter systems were developed (Tchobanoglous et al. 2002; Rusten & Odegaard 2006) while for smaller onsite applications, simpler organic percolation filters are proposed, which are easier to operate (Lens et al. 1994; Taylor et al. 2003; Todt et al. 2014b). Filtration is an interesting option for blackwater treatment in source separating sanitary systems (Todt et al. 2014b) since blackwater contains 60% of total suspended solids (TSS) in a small volume fraction (Meinzinger & Oldenburg 2009). However, simple organic percolation filters were shown to be vulnerable to clogging, which limits practical implementation (Lens et al. 1994; Todt et al. 2014b). To overcome these problems, a novel mechanical filter device was developed as a primary treatment step for the blackwater discharge from mountain cabins with 20–40 accommodation places.
The filter system is combining traditional mechanical screening with a new type of counter-flow filter using an organic media as biodegradable filter matrix. For the coarse filtration in the first filtration step, a downscaled standard drum screen (Tchobanoglous et al. 2002) was chosen. The design of the counter-flow filter was based on the outcomes from a previous study on blackwater treatment with static filter columns (Todt et al. 2014b). The objectives of this study were to assess the system performance regarding treatment of blackwater directly connected to a vacuum toilet system and to gain knowledge on the filtration mechanisms.
Counter-flow matrix filter and initial test setup
The prototype of the sequential mechanical filter unit consists of a pre-screening unit that removes coarse particulate material followed by a counter-flow filter removing finer particles (Figure 1). The pre-screening unit was based on a drum screen with an opening size of 3 mm. The counter-flow filter uses an organic filter matrix that is transported upwards with the help of a conveyor. The sewage passes the filter matrix where particulate matter is retained and the liquid fraction leaves the system via a 3 mm screen at the bottom of the filter tube. The conveyor tubes had an inner diameter of 150 mm. For this experiment, wood-shavings with a grain size of 10–30 mm were used. The rotating intervals of the conveyors were controlled by the number of loading batches. At a speed of 0.5 RPM, the pre-screening conveyor was run for 3 s in each of the three loading batches and the counter-flow conveyor was run for 10 s in each 10 loading batches, giving a transportation distance per feeding interval of 6 and 20 mm, respectively.
The experiment used blackwater from student dormitories that are equipped with vacuum toilets with a 1.2 L flushing volume (Todt et al. 2014a). For the experiment, a dosing arrangement simulating discharges of a vacuum sewer system was established (Figure 1(b)). The filter system was loaded for 33 days, interrupted by weekends with no loading. Loading took place in one daily sequence of 1.1 L batches (simulated toilet flushes) at intervals of 90 s. The hydraulic load per day was determined by the blackwater availability from the student dormitories and ranged from 60 to 312 batches.
Sampling and statistical analysis
Daily composite liquid samples were collected from the stirred blackwater holding tank (S1) and counter-flow filter (S3). The liquid effluent from the pre-screening unit (S2) was sampled using grab samples. All liquid samples were analysed for TSS, which was determined with 1.2 μm glassfiber filters (GF-C, 47 mm, Whatman, Little Chalfont, UK). A selected number of samples were also analyzed for chemical oxygen demand (COD) and phosphorus using spectrophotometric test kits (Hach-Lange, Berlin, Germany). The raw blackwater was additionally analyzed for the content of large-sized (>1 mm) suspended solids by filtration over a calibrated woven textile filter with 1 mm mesh size (PETEXTM 07-1000/45, Sefar AG, Heiden, Switzerland). To minimize filter-cake formation and potential retention of smaller particles, the filtration volume was limited to 5 mL in the TSS determination and in the determination of solids >1,000 μm. These were the smallest volumes that resulted in a good repeatability of the filtration procedure of a well-mixed sample (<20% variance between parallels). Solid retentate samples were collected in the discharge chamber (S4) and analyzed for dry matter by drying at 105 °C and loss of ignition (LOI) at 550 °C.
Due to the high natural variation in wastewater samples, as well as the limited number of data points produced by the experiments, a descriptive, quartile-based statistics method was chosen to present the data, and non-parametric Wilcoxon signed rank tests with a significance level of p < 0.01 for sample comparisons. The data are presented using box-plots showing 1st and 3rd quartiles, supplemented with error bars showing the minimum and maximum values. Variation ranges expressed in the text refer to 1st and 3rd quartile. To explore the dependency of TSS reduction versus the TSS concentration, simple linear regression plots were established with Microsoft-Excel.
RESULTS AND DISCUSSION
Treatment performance and filter media consumption
The filtration with the pre-screening unit resulted in a TSS reduction of 57–72%, which was further reduced to 49–64% by the counter-flow filter (Figure 2(b)), giving a final effluent TSS concentration ranging from 1,150 to 1,650 mg L−1 (Figure 2(a)). The whole system obtained a TSS reduction of 78–85% (Figure 2(b)), 60–80% for COD (Figures 2(d) and 2(e)), and 42–57% for total-P (Figures 2(f) and 2(g)). No significant (p = 0.13) reduction was found for dissolved orthophosphate (not shown), hence, the majority of the retained P was particle-bound. The obtained particle (TSS) reduction was close to reported values (90%) from rotating belt sieves with significant smaller mesh sizes (350 μm) (Rusten & Odegaard 2006) or static percolator filters (Lens et al. 1994; Todt et al. 2014b). However, the inlet TSS concentration range in the blackwater used by our study (Figure 2(a)) was substantially higher than the settled blackwater (0.4–1.5 g TSS L−1) used by Todt et al. (2014b) or municipal wastewater used by the other above-cited studies. Considering the dependency of filter efficiency on inlet concentration as indicated by Figure 3(a), a direct comparison of our results to the currently available literature data needs to be done with care.
The retained solid material (retentate) discharged from the filtration steps showed a dry matter content of 13–14% for the pre-screening retentate and 18–20% for the counter-flow retentate (Figure 2(c)). The higher dry matter content in the retentate of the counter-flow filter was most likely a result of the wood-shavings amendment. The obtained dry matter content was considerably higher than the dry matter content of 0.7–10%, which is typical for gravity sedimentation in a septic tank (Henze & Comeau 2008). The high dry matter content might be advantageous for a further processing of the retentate by composting or drying with solar heat. The pre-screening retentate showed a high LOI of 91–92% (Figure 2(c)) indicating that mainly organic matter is retained in the first filtration step. This is in accordance with the findings from other experiments with screening systems (Ruiken et al. 2013). The retentate of the counter-flow showed an LOI of 92–94% (Figure 2(c)), which is surprisingly low, considering its high content of wood-shavings having an LOI of 98%. This indicates that the retained solids in the second filtration step have an lower LOI than the solids retained in the pre-screening unit. Hence, the counter-flow filter has also retained a certain quantity of the inorganic compounds from the wastewater. This is supported by the phosphorus reduction (Figures 2(f) and 2(g)).
The wood-shavings consumption was in the range of 1–1.2 L per 100 simulated toilet flushes. The measurements of the wood-shavings load were notably lower than a theoretical filter-matrix consumption, which was calculated on 1.76 L per 100 simulated flushes based on the tube cross-section and conveyor speed. These can be explained if only 60–70% of the tube cross-sectional area was filled with wood-shavings, as was confirmed by visual observations. In the first preliminary experiments, the peat-sawdust mixture used in our previous study (Todt et al. 2014b) was used as a filter matrix. Its hydraulic capacity, as well as structural stability, was shown to be insufficient so that wood-shavings were selected for the experiment. Other filter media with high hydraulic capacity, for example bark, may be an alternative to wood-shavings. To avoid washing out filter media or clogging the effluent grid, as well as ensuring sufficient hydraulic capacity, the grain size of the filter media should be 10 mm or larger.
Dependency of suspended solids retention on inlet concentration and hydraulic load
The dependency of the filter performance on the inlet concentration for suspended solids is elucidated by the regression plots in Figure 3. The regression plots in Figures 3(a)–3(c) indicate a tendency towards greater relative TSS retention at higher TSS concentration loads for both filtration steps. This tendency was further pronounced for the last filtration step, as indicated by the inclination of the regression lines and R-squared values (Figures 3(b) and 3(c)), which points towards different filtration mechanisms in the two filtration steps.
Comparing the effluent TSS concentration obtained via the pre-screening unit (Figure 2(a)) with the concentration determined for suspended solids >1 mm (Figure 3(d)) indicates that a majority of large-sized particles were retained in the pre-screening unit. The observed weak tendency towards greater relative particle removal with higher TSS concentration loads (Figure 3(b)) is in accordance with observations in other drum-screen systems with mechanical retentate removal (Rusten & Odegaard 2006; Ali 2013). In our experiment, this tendency is probably influenced by the high-fraction large-sized particles in the blackwater (Figure 3(d)) and filter-cake dynamics. Without the rotating conveyor, the mechanism behind the filter cake dynamics in the filter drum may be comparable to static textile filters. Experiments with textile filters showed that larger-sized solids accumulate into a filter cake, which enhanced the filtration effect by retaining a greater number of particles smaller than the apparent opening size of the textile, but at the same time substantially decreased the hydraulic capacity of the filter (Faure et al. 2006). The main function of the rotating conveyor inside our filter drum is to sustain hydraulic capacity and prevent clogging by antagonizing filter-cake formation. However, a qualitative optical investigation of the filter drum showed that slight filter-cake formation still occurred, especially on days with a high TSS concentration load, which might increase the retention of small-sized particles. The magnitude of such an additional filtration effect due to filter-cake formation within the screen drum is most likely determined by the TSS load and the movement of the conveyor. Hence, it might be possible to improve the TSS removal capacity by adjusting the rotation speed and intervals of the conveyor. To determine this relationship, further experiments are needed.
The mechanisms in the wood-shaving filter matrix in the counter-flow filter are comparable to smaller-sized vertical-flow percolators or compost filter systems. In such systems, the tortuous filter matrix was shown to facilitate both filter-cake formation and biofilm growth (Zhao et al. 2009; Hua et al. 2013), which again enhances the retention of smaller-sized particles. These processes may be behind the observed greater dependency of the TSS reduction on the TSS concentration load in the counter-flow filter (Figure 3(c)) compared to the first filtration step (Figure 3(b)). As shown with our previous experiment (Todt et al. 2014b) and others (Lens et al. 1994; Taylor et al. 2004), in a static vertical-flow percolator, a majority of the removed particles tend to be retained in a distinct layer on the top of the filter matrix. In an earlier experiment, we obtained a 72% TSS reduction in vertical-flow percolator columns filled with a wood-shavings mixture and could not find a significant difference between 150 and 300 mm filter height (Todt et al. 2014b). The transverse filtration across the 150 mm filter tube, which is applied to the second filtration step in this study (Figure 1(a)), should theoretically have reached a comparable efficiency. However, the TSS reduction obtained in the second filtration step was, at 49–64% (Figure 2(b)), notably lower, which points to a significant impact of the mechanical conveyor system on the retention process. The rotation of the filter matrix (wood-shavings) likely reduce the performance of the filter system by breaking up developed filter-cakes as well as mixing accumulated particles into deeper layers from where a re-suspension and outwash into the effluent can take place. As indicated by the measured wood-shavings consumption (‘Treatment performance and filter media consumption’ section), the feeding mechanism of the wood-shavings was not perfect, which resulted in an incomplete filling of the tube. The retention time of the wood-shavings matrix of 5–6 days in cross-flow section of the filter tube may also be too short to facilitate a greater biofilm development.
The above findings indicate that filter efficiency of the second filtration step can be increased by improving the wood-shavings feeding and transport mechanism. The present conveying length was selected to obtain a complete renewal of the infiltration surface by turning the wood-shavings matrix 120 degrees at each conveying interval. The conveying interval was determined from trial runs aimed to keep ponding in the infiltration funnel below 10 cm. Periodical ponding of 3–5 cm was observed, which corresponds to a volume of 1.2–1.9 L. After a conveying event, this ponded volume percolated into the wood-shavings matrix within 30–60 s, which means three to four times higher infiltration velocity than the average at the applied load of 1.1 L every 90 s. A higher infiltration rate may reduce filter performance. Hence, the conveying regime can influence both capacity and treatment. Additional research is needed to optimize rotation versus these key parameters.
No significant correlation was found between the daily hydraulic load and the TSS reduction for the overall system (Figure 3(e)), the two filtration steps (Figures 3(f) and 3(g)), and the residual TSS concentration in the filtrate (Figure 3(h)). The filtration in the pre-screening unit is mainly a physical process, while biological processes are unlikely to take place. Hence, the filter performance of the first filtration step is mainly dependent on the physical parameters of the screen and the flow rate of 0.73 L min−1, which was determined by the flushing interval. With the 90 s flushing interval, it should therefore be theoretically possible to increase the daily hydraulic load up to 1,056 L d−1 without significantly impacting the filter performance. In contrast to the pre-screening unit, the counter-flow filter facilitates biological processes that may impact the filter performance. These biological processes can be more easily influenced by the applied diurnal pattern of loading and non-loading periods. Hence, an extrapolation of the regression line in Figure 3(g), and subsequently also in Figures 3(e) and 3(h), to hydraulic loading rates exceeding the maximum load of 343 L d−1 would need to be verified.
An increase in peak loading (number of flushes per unit time) may have a greater impact on the filter performance than the daily hydraulic load at constant flushing intervals. This could cause washing out of retained particles in both filtration steps; however, this has not been investigated yet. In static filter columns, an inverse correlation of filter performance to the percolation velocity was shown (Lens et al. 1994). Considering the infiltration area of 3.1 dm2 provided by the infiltration funnel, an average percolation velocity of 430 cm h−1 was calculated for the applied loading rate. This is three orders of magnitude higher than the highest percolation velocity of 0.42 cm h−1 applied by Lens et al. (1994), but as pointed out above, the treatment performance of our filter is comparable.
Onsite test at Britannia Hut
In addition to the laboratory experiments, a pilot-scale test of the mechanical filter system was performed at Britannia Hut, a mountain in Switzerland, over two seasons. The system treated the blackwater discharge from two vacuum toilets with a 0.7–0.8 L flushing volume. These toilets were used daily by 10–60 guests, The TSS in the raw blackwater was 7–11 g L−1 (not shown) and slightly higher than the inlet in the laboratory experiment (Figure 2(a)). The hydraulic load ranged from 50 to 430 flushes, which translates into 40–340 L blackwater per day. The system obtained a TSS reduction of 70–90% and an effluent TSS in the range of 1.2–2 g L−1 (not shown). The flushing data logged during the onsite test at Britannia Hut showed diurnal peaks ranging from 30 to 70 toilet flushes per hour, translating to a peak flow rate of 0.38–0.88 L min−1. These figures indicate that the peak flow rate of 73 L min−1 applied in the laboratory experiment is reasonable for a system treating blackwater from a mountain hut with up to 50 guests. The daily energy consumption per 100 toilet flushes was 0.003–0.004 kWh for the mechanical filtration system only, and 0.4–0.5 kWh including the vacuum sewer system (vacuum-on-demand VODTM, Jets Vacuum AS, Hareid, Norway). For solar-powered operation of the system at Britannia Hut, approximately 2 m2 photovoltaic cells would be required to treat a daily load of 100 toilet flushes.
Design evaluation and practical application
The design of the tested filter is an outcome of our previous study where we concluded that organic percolation filters can be significantly improved by implementing mechanical filter media renewal (Todt et al. 2014b). The presented two-step filtration system was developed with the help of preliminary trials based on the idea to integrate an organic percolation filter into a simple drum-scree as typically used in municipal treatment plants (Tchobanoglous et al. 2002). In its present configuration, and with the applied conveying velocity, the system worked satisfactorily hydraulically both during the laboratory experiment and in the onsite test at Britannia Hut. An issue that needs to be improved for practical application is the filter media transport into the filter tube. Based on the results presented in the study, the maximum capacity of the system is estimated to be 20–40 PE, considering a peak loading rate that does not exceed the tested 40-toilet flushes per hour. On higher peak loading rates, a preceding equalizing tank may be needed. However, such a tank will raise the energy consumption due to the need for an additional feed pump.
A compact novel filter system was developed as a primary treatment system for the blackwater fraction of source-separated sanitary systems, or other types of wastewater with a high TSS concentration. The filtration unit showed a removal capacity of 78–85% for TSS, 60–80% for COD and 42–57% for phosphorus when treating blackwater from a vacuum toilet system. This reduction is comparable to organic filter systems that have a much lower hydraulic capacity per filter area. With a dry matter content of 13–20% the retentate is suitable for further processing by composting or drying with solar heat. The results show a positive correlation between filter performance and inlet concentration. The data on the impact of hydraulic load indicate that the system is feasible for applications up to 40 PE.
This study was greatly supported by the European Commission (FP7), and the companies: Jets Vacuum AS, Norway; Seecon GmbH Switzerland; Gysi-Berglas AG, Switzerland; Bioazul, Spai; SiA Norplast, Latvia; the Swiss Alpine Club, and special thanks go to the staff at Britannia Lodge. Grateful thanks also to Risch Tratschin (Seecon GmbH) for helping with the data collection at Britannia Hut.