Design optimization of a simple, single family, anaerobic sewage treatment system

Van Haandel & Lettinga (1994) showed that other anaerobic systems tend to perform less well than the UASB reactor. Almost all large anaerobic treatment systems constructed in Brazil since 1990 are UASB reactors or variants. However, in small-scale units, UASB reactors are less common, although Coelho et al. (2003) showed that a 360 L reactor (as in Figure 3) is superior to the ST/anaerobic filter combination. In the latter case the ST is used mainly to retain gross solids that would otherwise clog the upflow anaerobic filter, which itself provides efficient removal of organic matter from the pre-treated effluent from the ST. In this paper a UASB reactor configuration is proposed that can substitute the traditional ST and be used as single family alternative for decentralized sewage treatment.

The digestion and settling zone diameters must be chosen so that an adequate upward velocity is maintained within those zones. Since this ‘adequate velocity’ was not known at first, six reactors of different diameters were constructed and tested (see Table 1). The important digester parameters were calculated once the basic parameters (HRT, and the digestion, transition and settling zone diameters, as well as the separator's inclination) were defined. Table 2 shows the expressions used to define the various dimensions. The diameter of the conical separator was 0.05 m larger than that of the digestion zone, so that there was an overlap and biogas bubbles could not enter the settling zone.

Table 2

Expressions defining aspects of the UASB configurations used in the study

Line n° .	Derived parameters .	Symbol .	Formula .
10	Digestion area
11	Settling area
12	Velocity in digestion zone	vdi	=Qa/Adi
13	Velocity in settling. Zone	Vs	=Qa/Ade
14	Overlap	O	=(Ds–Ddi)/2
15	Transition height	Htr	=(Dde–Ddi)/2*tanα
16	Separator height	Hse	=Ds/2*tanα
17	Settler height	Hde	=Hse – Hb
18	Settler volume
19	Transition volume
20	Digestion volume	Vdi	=Vtot–Vde–Vtr
21	Volumetric proportion in digestion zone	fdi	=Vdi/Vtot
22	Volumetric proportion in transition zone	ftr	=Vtr/Vtot
23	Volumetric proportion in settling zone	fde	=Vde/Vtot
24	Digester height	Hdi	=Vdi/Adi
25	Total height	Htot	=Hdi + Htr + Hde + Hb
26	Minimum area in settling zone.
27	Maximum area in settling zone.
28	Maximum velocity in settling zone	vmax	=Qa/Amin
29	Minimum velocity in settling zone	vmin	=Qa/Amax

Line n° .	Derived parameters .	Symbol .	Formula .
10	Digestion area
11	Settling area
12	Velocity in digestion zone	vdi	=Qa/Adi
13	Velocity in settling. Zone	Vs	=Qa/Ade
14	Overlap	O	=(Ds–Ddi)/2
15	Transition height	Htr	=(Dde–Ddi)/2*tanα
16	Separator height	Hse	=Ds/2*tanα
17	Settler height	Hde	=Hse – Hb
18	Settler volume
19	Transition volume
20	Digestion volume	Vdi	=Vtot–Vde–Vtr
21	Volumetric proportion in digestion zone	fdi	=Vdi/Vtot
22	Volumetric proportion in transition zone	ftr	=Vtr/Vtot
23	Volumetric proportion in settling zone	fde	=Vde/Vtot
24	Digester height	Hdi	=Vdi/Adi
25	Total height	Htot	=Hdi + Htr + Hde + Hb
26	Minimum area in settling zone.
27	Maximum area in settling zone.
28	Maximum velocity in settling zone	vmax	=Qa/Amin
29	Minimum velocity in settling zone	vmin	=Qa/Amax

View Large

The upward velocities in the digestion and settling zones are the basic parameters, when designing the configuration shown in Figure 4, needed to ensure good contact in the digestion zone and good sludge retention in the settling zone. Figure 4 and Table 1 show that, for sewage flow, Q_a, the UASB-SF reactor is defined by (1) the reactor volume, V_u, or retention time HRT, (2) the transition zone and separator inclination (α), and (3) the digestion (D_d), settling zone (D_s), and separator (D_p) diameters. These define all other reactor dimensions and were operated at constant flow rates of 0.5 and 1 m³.d⁻¹, corresponding to the minimum and maximum anticipated flow rates for a single family unit. In a second phase, variable flow rates were applied as significant variations in sewage flow rate with respect to time tend to happen in single family units.

The basic parameters in Table 1 influence the values of the derived parameters and must be chosen such that the values of the latter are adequate. The most important variables are:

• (1) The liquid velocity in the digestion and settling zones. The velocity in the digestion zone must allow retention of a large sludge mass, but be high enough to avoid inorganic solids accumulation there. Velocities of less than 0.3 m.h⁻¹ lead to grit and sand accumulation and are thus too low. The lower the velocity at the top of the settling zone, the more efficient the sludge retention, but an evaluation is needed to determine whether the retention of hard-to-settle particles and their return to the digestion zone is really advantageous, as it might affect sludge settleability. The maximum upward fluid velocity at the bottom of the settling zone is dictated by the difference between the settling zone and phase separator diameters. If it is low, it is easier to return retained particles to the digestion zone.

• (2) The proportional volume of the different zones. The optimal division of volumes of the settling and digestion zones is not obvious: a large digestion zone allows a large sludge mass, but a large settling zone is required for efficient sludge retention and hence a high sludge concentration in the digestion zone, although it is the settling area, not its volume, that determines particle retention. Nevertheless, the settling volume can be important in promoting flocculation of small particles into better settling flocs. The division of the volume over the settling and the digestion zone was a key point of investigation.

• (3) Reactor height. A tall reactor has a small footprint, but its implementation may be complicated. In order to avoid pumping, the top of the reactor must be below the level at which the sewage arrives at the treatment unit, but digging deep may be difficult, especially in areas with rocky soil.

The biological, chemical and mechanical properties of the sludge that developed in the reactors are also important. The most important biological sludge properties are its activity or methane production potential, and its stability (the fraction of biodegradable organic material it contains). The most relevant chemical property is the fraction of inorganic sludge, while its settleability is the most significant mechanical property. All were evaluated for each reactor as a function of operational conditions.

All reactors had sludge sampling points at several heights. Table 3 shows their levels from bottom to top. The total sludge mass was calculated from the sum of the products of the volume of each reactor segment corresponding to a sampling point and the sludge concentration measured there.

Raw sewage from Campina Grande, Paraíba (Brazil), was used in the study. The reactors were initially operated at constant flow and load conditions for retention times of 12 and 6 hours. The flows were 500 and 1,000 L.d⁻¹, respectively, which were taken as representing the sewage from one and two, four-person, households (125 L.inh⁻¹.d⁻¹). It was expected that these flows could be treated easily, as Coelho et al. (2003) design had been shown to do so efficiently.

In Equation (1) the sludge age denotes the value calculated on the basis of the volatile solids. The sludge age on the basis of total sludge can also be calculated but the values of MX_v and X_ve must be substituted by MX_t and X_te, where the index ‘_t’ denotes the total solids content of the sludge.

Tests to characterize the performance of the UASBs were carried out over 18 months. Their objective was to determine the division of influent COD into its fractions in the effluent, the sludge and biogas. COD and suspended solids tests of influent and effluent were carried out following procedures described in Standard Methods for the Examination of Water and Wastewater (APHA et al. 2012). COD determinations were done on raw and settled effluent samples, and the COD fraction transformed into sludge was taken as the difference between the two values.

The specific methane production potential (SMPP) is the methane production per unit mass of sludge per unit time unit, under ideal anaerobic digestion conditions. It was determined in accordance with the procedure described by Van Haandel & Lettinga (1994), using 250 mL bottles (effective volume 200 mL) at a constant temperature of 35 °C. Solutions were added to ensure the presence of macro- and micro- nutrients, and that pH = 7 was maintained during the test. A substrate mass of 0.8 g COD was added as acetate solution, (4 g-COD.l⁻¹), capable of producing about 200 mL of methane at 0 °C and 1 bar (standard conditions). The anticipated duration of the test was 3 to 5 days. Cumulative methane production was determined from the volume of solution displaced in a Mariotti flask, and SMPP was determined from the maximum methane production rate and calculated as g-COD.gX_v⁻¹.d⁻¹. Substitution of a mixture of acetate and propionate for the acetate feed did not affect the methane production rate, whence it was concluded that biogas production in the treatment units was due to acetotrophic rather than hydrogenotrophic methanogens. SMPP tests were carried out with sludge samples withdrawn from the reactor and samples obtained by settling the raw effluent to obtain the sludge that was expelled from the reactor. Thus it was possible to determine if the SMPP of the sludge in the reactors was the same as that of the sludge expelled from the reactors.

The constants k and v₀ are linked to the mechanical properties of sludge: v₀ reflects the sludge settling velocity of particles in a sludge so dilute that the particles settle individually (unhindered settling). The constant k describes sludge compressibility and the higher its value, the more difficult it is to obtain a high concentration by settling. Vesilind's equation was originally proposed for the determination of activated sludge settling, but tests carried out in this study showed that it can also be used for anaerobic sludge such as generated in the UASB reactors. The Vesilind constants k and v_o were determined for all reactors, again, not only with samples taken from the reactors but also with sludge that was expelled from the reactors.

The performance of all reactors was evaluated in terms of influent organic matter removal efficiency as well as digestion efficiency. In this context, digestion efficiency is given by Equation (2c). The removal efficiency is the fraction of the influent COD that is found in the settled effluent and is expressed by Equation (2a). The digestion efficiency is the fraction of influent COD that is converted to methane and is expressed by Equation (2c). Removal efficiency is given by Equation (2a) plus the fraction transformed into settleable solids (Equation (2b)). Table 4 shows the organic matter removal efficiency and sludge retention data for the reactors operating at 12 and 6 hour HRTs.

Table 4 reveals some important aspects:

• (1) The COD removal efficiency (COD transformed into methane or sludge) was influenced only marginally by the configuration in the range used, apart from design T₁ whose performance was inferior to that of the others at both 6 and 12 h HRTs. The influence of hydraulic and organic loads on removal efficiency was very limited in the investigated range: reducing HRT from 12 to 6 h led to an insignificant decrease. Similarly the anaerobic digestion efficiency (COD fraction converted to methane) was also affected little by configuration or HRT.

• (2) The COD fraction converted into settleable sludge was about half of the non-settleable total. Thus it is very important to state whether raw or settled effluent was used to determine effluent COD.

• (3) The non-biodegradable content of soluble material in the influent is atypically high for municipal sewage, which must be attributed largely to the fact that the sewerage outfall was partially blocked and conveyed only a fraction of the generated sewage. Because of this, extensive anaerobic digestion was taking place in the sewerage network, reflected in a low influent COD concentration, with high non-biodegradable soluble and low non-biodegradable particulate fractions, due to settling in the sewer.

• (4) The soluble effluent COD consisted largely of non-biodegradable organic matter. This was apparent from the biochemical oxygen demand/COD ratio, which was less than 0.25 in all tests and had an average value of 0.17.

• (5) The volatile fatty acids (VFA) concentration was consistently low at less than 0.5 mol.l⁻¹.

• (6) As reactor configuration was shown to have little influence on treatment efficiency, other factors will determine the preferred design. These include the total reactor height (difficulty of installation, especially in rocky terrain) and instability caused by shock loads. Reactors with low upward fluid velocities, which tend to be shallow, are affected by shock loads less than those with relatively higher velocities.

• (7) All reactors had much smaller footprints than equivalent STs, which is a clear advantage in housing schemes with small plots. The Brazilian norm for the minimum area of ST is 1.5 m², which is 3 to 5 times larger than the reactors that were operated in this investigation.

In Table 5 the total and volatile sludge masses in each reactor are calculated from the concentration profiles. Also included are the averages of daily expelled sludge mass (total and volatile), and the sludge ages calculated from the ratio of the retained and expelled masses, for the different reactors for both total and volatile sludge, by using Equation (1).

The results in Table 5 show that there was a considerable reduction in sludge mass in the system when the influent flow increased from 0.5 to 1 m³.d⁻¹. Reactors T₁, T₂ and T₃ (digestion zone diameter 0.3 m) lost about 70, 40 and 10%, respectively, of the initial mass. This can be attributed to (1) the inoculum used in these reactors, which was taken from an anaerobic pond with high concentrations of inert and inorganic matter, and (2) reactors T₁ and T₂ had relatively small settlers, so that the sludge could not be retained when the flow rate was increased. Reactors S₁, S₂ and S₃, like UASB-Y, were inoculated with sludge from T₁, T₂ and T₃, and this conventional sludge mass was much less affected by flow increases, because the inoculum was better and they had larger settling areas.

The sludge ages obtained by applying Equation (1) show that the reactors with bigger sedimentation zones had better biomass retention, expelling less solids in the effluent while maintaining a longer sludge age. Other research has shown that there is no advantage in operating a UASB reactor at a sludge age exceeding 100 days, because removal of biodegradable matter was essentially complete (Santos et al. 2016). From the data in Table 5 it is noted that the sludge age determined on the basis of total sludge mass and expulsion rate is 10 to 30% longer than that based on the volatile sludge. This indicates that there was selective retention of inorganic solids in the reactors, especially when the sludge age was long. For shorter sludge ages the difference between the total and volatile sludge ages was much smaller.

Figure 6(a)–6(c) show the concentration profiles in the reactors. Considerable concentration stratification is evident. This is attributed to changes in sludge settleability with depth, since the upflow velocity of the liquid phase was constant and the mixing energy of the rising bubbles – from biogas generation – (<1 W.m⁻³) was insufficient to induce significant turbulence.

The experimental data underlying Figure 6(a)–6(c) also indicate qualitative stratification, with a greater proportion of inorganic matter in the lower regions of the reactors. The volatile fraction also decreased with the cross-sectional area of the digestion zone, especially when the upflow rate was low. It is possible that the higher inorganic fraction improved sludge settleability. This is not unexpected as selective expulsion of organic sludge could mean that inorganic sludge, which tends to have better settling characteristics, was retained more efficiently and accumulated in the reactor's lower regions. This may have been enhanced by slow sintering, with the inorganic sludge gradually acquiring greater mechanical rigidity. This was observed but not quantified during the study. Selective inorganic sludge retention is also reflected in the calculated sludge age (the ratio of sludge mass retained to daily mass expelled), as shown in Table 5, the total sludge age calculated is systematically higher than that calculated on the bases of volatile solids.

Figure 6(a)–6(c) also show that the sludge layer in most reactors did not extend upward through the digestion zone. In other words, it was not possible to build the sludge up over the full height of this zone, even if hydrodynamic conditions were adequate for sludge settling and accumulation. It is believed that sludge retention by gravity settling was counter-balanced by break-up of the sludge flocs, with parts of the flocs with poor settling characteristics being swept out of the reactor. The sludge particles leaving the digestion zone may be discharged with the effluent, but it is also possible that these latter will flocculate in the settling zone and form bigger particles with better settling characteristics that can then return to the digestion zone.

The biological and mechanical sludge properties were also investigated. The key biological and mechanical properties, respectively, were the SMPP, and the settleability as defined by the Vesilind constants. These are important because they indicate the treatment capacity limits of the reactors. The settleability constants enable estimation of the maximum theoretical sludge mass that could be retained. On the other hand, using the sludge mass in the reactors and the SMPP, the maximum theoretical extent of methanogenesis can be calculated. These theoretical values can then be compared to the actual experimental observations.

Table 6 presents the SMPPs for the 6 and 12 hour HRTs evaluated. Tests were carried out for both HRTs using samples withdrawn from the reactors as well as expelled sludge samples. The data indicate that the SMPP was approximately the same in all reactors, with no significant differences between the collected and expelled sludges. Reactor load seems to affect SMPP value: longer retention times and lower loads tend to have larger SMPPs. This is as expected, the higher the load, the greater the volatile fraction – the latter comprising flocculated particulate influent organic matter, reducing the fraction of methanogens in the volatile sludge (Van Haandel and Lettinga 1994).

The settleability tests confirmed that Vesilind's formula (Equation (3)) describes settling perfectly in all cases. Table 7 shows the test results for sludge within and expelled from the reactors. The sludge characteristics indicated that the sludge in all reactors had excellent settleability, all with very low compressibility constant, k. There were no great differences between the constants in different reactors, but there was a clear difference between the sludge in the reactors and that expelled from them. In particular, k, for the expelled sludge is significantly higher on average, indicating that expelled sludge settleability was inferior to that of the sludge in the reactors.

If the expelled sludge's Vesilind constants are used in Equation (3a) it can be shown that the theoretical sludge concentration that could be retained greatly exceeds, in all cases, the concentrations achieved (see Figure 6). In other words, sludge particles with poor settling characteristics are expelled from the digestion zone because they cannot be retained. They tend to flocculate in the settling zone, however, and acquire better settleability, so that they can settle again and become part of the sludge in the digestion zone. It is possible, therefore, that, if the settled solids from a settler were returned to the UASB reactor, much of the material would be retained. If this were so, a larger sludge mass could be maintained and the reactor operated at longer sludge age, and thus at higher efficiency and with lower sludge production.

Table 8 shows the theoretical sludge concentrations that could be retained, calculated using Equation (3a), for the constant values for the different reactors and the average sludge concentration as determined experimentally. The theoretical sludge mass values tend to be larger than the experimental equivalents for most reactors. As can be seen in Figure 6, however, the actual sludge concentration in the bottoms of the reactors is much higher than the theoretical maximum calculated for expelled sludge. This is because some sludge had better settleability than the average value used to calculate the theoretical concentration. Equally, in the digestion zone upper sections, the sludge concentration is much lower than the theoretical maximum, confirming that mechanisms other than settling are relevant to sludge retention in the digestion zone, otherwise, the digestion zone would fill entirely with sludge at a concentration equal to the theoretical maximum. Thus, the low sludge concentrations in the upper sections of the reactors is attributed to floc break-up, with small particles carried upwards by the liquid flow. Some of these can flocculate in the transition or settling zones, and return to the sludge bed. Others will be discharged with the effluent.

It is also possible to compare the theoretical and actual methane production capacities. The theoretical production level is the product of the volatile sludge mass and SMPP for each reactor. The true production can be calculated as the product of the digested COD fraction and the applied load. Both theoretical and actual are shown in Table 8. In all cases the theoretical treatment capacity or methane production is much greater than that achieved, i.e., methane production was not the limiting process for anaerobic treatment. This is confirmed because the VFA concentration (including acetate, the main methanogenic substrate) was always low.

• (1) A simple, 250 L, UASB-based reactor can treat sewage from a single family efficiently. Its organic matter removal efficiency is much greater than can be achieved in a traditional ST, or combination ST and anaerobic filter.

• (2) A set of 8 × 250 L reactors with differing geometries could all digest raw sewage efficiently at 12 and 6 h HRTs. The experiments showed clearly that, within the size range investigated, geometry had no significant influence on performance. Thus, configuration preferences can be based on considerations of ease of manipulation and installation, and a shallow unit (depth 1.5 m) with a relatively large footprint (diameter 0.7 m) can be chosen.

• (3) Sludge retention in the settling zone, above the digestion zone, contributes to sludge mass maintenance, but settling is not the sludge retention determining mechanism in UASBs.

• (4) Sludge age was confirmed as the fundamental operating parameter determining UASB performance and, particularly, division of the influent COD into three fractions: (i) discharged in the effluent, (ii) converted to sludge, and (iii) converted to methane. If the sludge age is the same, UASBs treating the same sewage at different HRTs will tend to have the same effluent quality and sludge production rate, and therefore the same digestion efficiency. The lower the sludge age, the higher the proportions of influent COD going to the effluent or the excess sludge.

• (5) The SMPP of all sludge investigated in the study was within the range commonly found for sludges from treating raw sewage, i.e. 0.10 to 0.15 g-COD.g⁻¹X_v.d⁻¹. The SMPP of the sludge in the reactors did not differ significantly from that of the expelled sludge. In all cases, it was much greater than the actual specific methane production, showing that methanogenesis was not the efficiency limiting process for anaerobic digestion. In fact the incomplete removal of biodegradable organic material arises mainly because the preceding hydrolytic, acidogenic and acetogenic processes could not develop fully.

• (6) The settleability of the sludge within the reactors was always superior to that of the sludge expelled from them, although sludge retention was not the sludge mass retention determining mechanism. In most reactors the digestion zone was not completely filled with sludge, showing that a break-up mechanism exists, producing small particles that settle poorly and tend to escape from the reactor. Some of these tend to flocculate in the settling zone and may return to the digestion zone as recovered sludge, others are discharged with the effluent.

The experimental investigation was carried out with financial support from agencies of the Brazilian Government (CNPq, ANA and FINEP).