3.1. Phosphorus Removal Performance of the Upgraded Septic Tank and Drainfield
The P removal performance of the sidestream system is shown in Figure 2
for the septic tanks and Figure 3
for the drainfields. The control septic tank without a slag filter had an influent pH of 7.2 ± 0.2 which remained stable in the R3b compartment. The R3b compartment released o-PO4
), probably due to sludge hydrolysis. The slag filter resulted in a favorable pH above 11, and an effluent o-PO4
and TP below 0.1 and 0.3 mg P/L, respectively, during the 275 days of the tests. The water quality of the slag filter effluent was consistent with previous observations from the slag filters of the same media [17
]. The cumulative P retention of the slag filter increased slowly up to 105 mg TP/kg slag (Figure 4
), which is lower than the cumulative removal of 300–1700 mg TP/kg slag observed in six recent slag filter studies for domestic wastewater treatment [6
]. The low P retention observed in the filter might be at least due to the short duration of the experiment; the filter had not yet reached its saturation. The mean P removal percentage was similar in all tested recirculation ratios (87%, 85% and 86% at the recirculation ratios of 25%, 50% and 75%, respectively).
The septic tank with a sidestream slag filter improved the TP removal efficiency compared to the control septic tank at recirculation ratios of 50% and 75% (Figure 2
), reaching removal efficiencies of 11% and 32% at recirculation ratios of 50% and 75%, respectively. The o-PO4
removal was markedly improved, reaching 40% at a recirculation percentage of 75% (Table 5
, all data different from the control at t < 0.02, except when indicated). TP and o-PO4
removal at the 25% recirculation ratio was not significantly different from the control test. A larger pH resulted in lower TP and o-PO4
concentrations in the R3a compartment of the experimental septic tank, in which the pH increased to 8.9 at the 75% recirculation ratios. The pH increase was due to the slow dissolution of the slag filter, which added hydroxide and calcium into the septic tank supernatant. Note that the P removal efficiency of the slag filter and the septic tanks was not affected by the seasonal changes in the experiment at the 25% recirculation ratio (e.g., the experiment started in the Fall, but the recirculation ratio changed to 50% in Winter). This observation contrasts with other field slag filters experiments that showed seasonal fluctuations in removal efficiency [15
The mean TP concentration at the effluent of the septic tank with a sidestream slag filter was 3.1 mg P/L, which is not as low as other applications of alkaline filters fed with primary effluent in the main stream mode (e.g., all influent passing through the filter). Indeed, a TP concentration of 0.55 mg P/L at the effluent of a vertical flow oil shale ash filter was reported [26
However, TP concentrations in the range of 2 to 6 mg P/L were observed at the effluent of field scale slag filters fed with the effluent of constructed wetlands [15
]. In these experiments, the pH at the effluent of filters was between 8 and 9, due to a short hydraulic retention time and possibly because of the exposure to atmospheric CO2
. These results outline the need for a pH above 10 at the effluent of an alkaline slag filter to reach a low TP concentration.
The type of sand had an impact on the phosphorus removal in the drainfield (Figure 3
). The phosphorus removal by the non-calcareous drainfield was very large in both control and steel slag filter systems, reaching a mean TP (and o-PO4
) concentration below 0.1 mg P/L for the 275 days of the experiment. In the limestone drainfield, however, such a high o-PO4
removal was observed only for the first 100 d (o-PO4
below 0.1 mg P/L at the drainfield effluent). After 100 d, the o-PO4
concentration at the effluent from the limestone drainfield increased in both control and slag filter systems, possibly because of sorption saturation. Interestingly, the TP concentration in limestone drainfield effluents increased after only 25 d, even if the o-PO4
concentration was still low and stable. The good removal efficiency of the non-calcareous drainfield could be explained either by its sorption capacity or by its precipitation mechanisms. The sorption capacity potential of the non-calcareous soil might not have been reached yet, and a phosphorus breakthrough could be expected after a few months or years of operation. Robertson et al. proposed, however, that equilibrium with aluminum or ferric phosphate minerals can explain the low concentrations of phosphorus monitored on a long-term [20
After reaching breakthrough, the TP removal efficiency of the limestone drainfield was improved by the sidestream slag filter. In the limestone control system, the mean TP removal efficiency between day 100 and day 275 was 54%, resulting in a TP concentration between 1 mg P/L and 3 mg P/L in the drainfield effluent.
In the system with a slag filter, this efficiency increased to 76% TP removal in the drainfield between day 100 and day 275, which resulted in a mean TP concentration of 0.7 mg P/L in the drainfield effluent.
A TP mass balance in the septic tank and drainfield with or without a sidestream slag filter is shown in Table 6
for the least favorable drainfield media, which was limestone sand. The calculations were made assuming a raw wastewater influent TP concentration of 6 mg P/L, a recirculation ratio of 75% in the steel slag filter and a mature limestone drainfield (e.g., after the initial high P removal period). The system with slag filter resulted in 8% of TP in the drainfield effluent, which is significantly less than the system without the slag filter, in which 33% of the TP is released in the seepage. This represents a significant reduction of TP load to the underlying groundwater in limestone soil application: for 10,000 septic tanks, the net P capture of a slag filter upgrade is 24.3 kg P/L. According to the results of this study, a septic tank improved with a sidestream steel slag filter at a 75% recirculation ratio reaches the target of 1 mg P/L at the seepage of the drainfield, which is comparable to common nutrient removal targets in advanced secondary or tertiary treatment processes [18
]. This target was not reached in the control system with the limestone drainfield, which is in agreement with previous groundwater monitoring below drainfields [20
]. In drainfields located in natural, non-calcareous sand, however, efficient long-term removal of phosphorus is possible [27
], and steel slag filters might not be needed in those applications.
The phosphorus recovery potential of the system was improved by the presence of the slag filter, assuming that the TP in the septic tank can be recovered in a subsequent centralized sludge treatment process [18
]. In the control system, 33% of TP was accumulated in the first and second compartments of the septic tank, compared to 43% in the presence of a steel slag filter. However, the 24% TP fraction captured in the slag filter was considered unrecoverable, due to technical challenges related to phosphorus extraction from exhausted media.
The COD and TSS removal efficiency of the drainfields with or without a slag filter was similar (Figure S4
, Supplementary Materials
), which indicates that the biological activity in the drainfield was not affected by the slag filter effluent. In the steel slag filter system, the R3a effluent pH did not exceed 9.0. Such pH rise is not expected to strongly inhibit BOD5
removal by heterotrophic bacteria, which tolerate a pH range of 6.0 to 9.0 [18
As the pH in the drainfield was buffered to about 8.0 by contact with atmospheric CO2
and biological activity (Figure 3
), biological polishing in the drainfield is not expected to be inhibited despite a septic tank effluent pH higher than 9.0. Having a slag filter had a minor impact on the effluent (seepage) pH of the drainfield (0.1 to 0.2 pH increase). The monitoring of calcium concentration, alkalinity and DIC concentration in drainfields is shown in Figure S5
3.3. Phosphorus Removal Mechanisms in the Septic Tank
-pH relationship in the septic tank effluent is shown in Figure 6
. In this figure, experimental results are compared to former data of the effluent of a lab-scale septic tank with sidestream slag filter fed by the effluent of the second compartment of the septic tank [13
]. Equilibrium curves of hydroxyapatite and vivianite are shown as possible P removal mechanisms by precipitation. Hydroxyapatite equilibrium curves were drawn at relevant fixed calcium concentrations: the effluent in the 2015 study had a stable calcium concentration of 30 mg/L, while most of the present study samples had a calcium concentration between 100 mg/L and 175 mg/L. Vivianite equilibrium curves were drawn at iron concentrations of 0.1 mg/L and 0.5 mg/L, which is the approximate range observed in ten different real septic tanks [20
The septic tank effluent o-PO4
and Ca concentrations from this 2015 project were equilibrated with finely-grained hydroxyapatite for pH between 8.3 and 9.0. Results suggest that in a septic tank improved with a sidestream slag filter, phosphorus is removed by hydroxyapatite precipitation as observed in steel slag filters [17
]. Note that equilibrium with finely-grained hydroxyapatite with a solubility product of 10−46
resulted in a realistic o-PO4
concentrations between 0.1 mg/L and 10 mg P/L for pH between 7.5 and 9.0. The bulk hydroxyapatite solubility product (e.g., 10−57
according to [3
]) is commonly considered as in a recent wastewater modeling study [32
], but instead results in equilibrated o-PO4
concentrations that are much lower, while supersaturation with bulk hydroxyapatite is observed in slag filters [23
] or biological reactors [33
Further mineralogical observations would be needed to confirm the presence and size of hydroxyapatite in biological reactors, and to determine an appropriate hydroxyapatite solubility product.
In this study, results were below the 100 mg Ca/L and 170 mg Ca/L finely-grained hydroxyapatite equilibrium curves, which suggests that other removal mechanisms took place. One possible mechanism is the sorption or coprecipitation of o-PO4
on freshly precipitated calcium carbonate. Such a removal mechanism has been proposed by Barca et al. based on the observation of crystals of different shapes and composition at the scanning electron microscope [15
]. Phosphate sorption on calcium carbonate can have a significant effect in high-alkalinity wastewater in which significant DIC reduction is observed, which was not the case in [13
] where the influent alkalinity was less than 250 mg CaCO3
/L, compared to 400 to 500 mg CaCO3
/L in this study. Calcium-carbonate-based media are known to have a phosphorus sorption capacity, such as 0.3 to 0.6 mg P/g for limestone and 3.5 mg P/g for shell sand and 0.8 mg P/g for oyster shells [34
]. Assuming a recirculation ratio of 75% in the septic tank, a reduction of 20 mg/L of DIC is expected, which corresponds to 166 mg/L of calcium carbonate precipitates. Assuming a sorption capacity of 3 mg P/g, 0.5 mg P/L of phosphate is expected to sorb on calcium carbonate precipitates, which explains a part of the phosphorus removal efficiency. A second possible removal mechanism is the precipitation of iron phosphate as vivianite, which is thermodynamically possible under anaerobic conditions prevailing in a septic tank [20
3.4. Effect of Influent Alkalinity on Steel-Slag-Filter Upgraded Septic Tank Operation and Costs
The recirculation ratio needed to reach a pH of 9 at the effluent of the septic tank is shown in Figure 7
, based on simulations of slag filter upgraded septic tank (influent calcium concentration fixed at 175 mg/L). Two slag filter hypotheses were tested: fresh slag which is assumed at the beginning of the filter lifetime, and long-term behavior of slag according to a slow slag exhaustion. Note that simulations agree with experimental data in this study (influent of 400 to 500 mg CaCO3
/L alkalinity, slag filter effluent of 11.1, recirculation ratio of 75% and pH at the effluent of septic tank of 8.7 to 9.0, experimental points shown in Figure 7
The simulated needed recirculation ratio depended strongly upon the influent alkalinity and the slag freshness. With fresh slag, the needed recirculation ratio was below 70% for alkalinity up to 425 mg CaCO3/L, but it increased following the slag exhaustion (needed ratio over 100% if alkalinity is above 225 mg CaCO3/L). In the present study, the slag filter effluent pH decreased from approximately 11.4 to 11.1 in the 215 days of operation, and is expected to decrease progressively until slag exhaustion at a pH of approximately 10.5. Therefore, the septic tank effluent pH will decrease as well, and the phosphorus removal capacity of the septic tank will be affected. The slag filter longevity was not reached in this study, but it can be estimated using slag filter modeling.
Claveau-Mallet et al. [9
] estimated the longevity of a steel slag filter operated under similar conditions (a series of two barrels of 5–10 mm slag followed by three barrels of 3–5 mm slag with a total empty bed contact time of 30 h) and fed with an influent alkalinity of 210 mg CaCO3
/L. The longevity was estimated at two years using simulations with the P-Hydroslag model [9
]. With a higher influent alkalinity of 400 to 500 mg CaCO3
/L, the longevity is expected to be less than two years because of increased precipitation and clogging by calcium carbonate. The use of a sidestream slag filter instead of a flow-through slag filter, however, increases the expected longevity to approximately 18 months as a significant part of calcium carbonate precipitation takes place in the septic tank instead of the slag filter. In this study at 75% recirculation ratio, 30% of influent DIC was removed in the septic tank, which means that the septic tank influent alkalinity was reduced by 30% compared to the septic tank influent. Such calcium carbonate control has an important impact on steel slag filter applications in onsite and decentralized treatment, where high-alkalinity influents are expected from some drinking water from groundwater supplies.
A simplified cost analysis of the implementation of septic systems for the treatment of domestic wastewater from a 3-bedrooms dwelling was conducted. Three scenarios were considered: (1) a conventional septic system without the slag filter upgrade; (2) conventional septic system upgraded with a sidestream slag filter and fed with a low-alkalinity influent (50 mg CaCO3
/L); and (3) conventional septic system upgraded with a sidestream slag filter and fed with a high-alkalinity influent (200 mg CaCO3
/L). The capital costs of the conventional septic system were spread over 20 years, which is a realistic lifetime for a septic system. The capital costs of a conventional septic system including installation were estimated to $
7500 (CAD) based on local price estimations. The slag filter upgrade was scaled up to a full-scale 3-bedroom dwelling, assuming an underground concrete reactor containing ten barrels in series [9
]. The analysis considered maintenance operations for the conventional septic system (sludge removal in the septic tank once every two years) and the slag filter upgrade (replacement of all barrels at fixed frequency). The barrel replacement frequency was estimated for each scenario based on the influent alkalinity and the hydraulic retention time of voids. The needed recirculation ratio was 40% in scenario 2 and 90% in scenario 3 (from Figure 7
), resulting in the hydraulic retention time of voids of 40 h and 18 h, respectively. The slag filter longevity was estimated at 10 years in scenario 2, and 2 years in scenario 3, based on previous longevity predictions [9
]. The resulting yearly costs and costs per removed unit of phosphorus mass are shown in Table 9
and Table 10
The yearly cost of a conventional septic system was estimated at $500. The implementation of a slag filter upgrade increased the yearly cost to $741 and $1217 in low-alkalinity and high-alkalinity influents, respectively. Such a cost increase for a single dwelling is significant, but remains realistic in comparison to the costs range of tertiary-level wastewater treatment processes in decentralized applications. The cost of the slag filter upgrade was highly influenced by the influent alkalinity, which determined the barrel replacement frequency. The cost per removed phosphorus mass was also influenced by the influent alkalinity. In scenario 2, the removal cost was $228/kg TP, which was similar than that observed in the conventional septic system ($211/kg TP). In scenario 3, however, the removal cost was much higher, reaching $374/kg TP. Note that this economic analysis did not consider costs inherent to commercialization, such as personnel time, R&D development costs, certification process costs, etc. Therefore, the real cost of hypothetical commercialized slag filter upgrades would be higher.