M.C.M. van Loosdrecht¹, L.M.M. de Bruin², H.F.R. van der Roest², M.K. de Kreuk¹

¹ Kluyverlaboratory for Biotechnology, Delft University of technology, Julianalaan 67, 2628 BC Delft, The Netherlands e-mail: m.c.m.vanloosdrecht@tnw.tudelft.nl
² DHV Water B.V., P.O. Box 484, 3800 AL Amersfoort, The Netherlands Abstract

Growing aerobic granules on synthetic wastewater in a batch-wise operated airlift reactor proved to be possible on the condition of a short feeding time and high dissolved oxygen concentrations. It was also demonstrated that a longer anaerobic feeding time combined with low dissolved oxygen concentrations had a positive influence on granulation and conversion processes. Experimental results obtained with a lab-scale reactor were used for designing two variants of a full-scale sewage treatment plant based on granulated sequencing batch reactors (GSBRs). In order to evaluate the feasibility of this new technology, a comparison was made to a conventional treatment plant based on activated sludge technology.

Based on total annual costs both GSBR variants prove to be more attractive than the reference alternative (7-17%). From a sensitivity analysis appears that the GSBR technology is less sensitive to the land price respectively more sensitive to a higher RWF. This means that the GSBR technology becomes more attractive at lower RWF/DWF ratios and higher land prices. Because of the high permissible volumetric load the footprint of the GSBR variants is only 25% compared to the reference. However, the GSBR with only primary treatment can not meet the present effluent standards for municipal wastewater in the Netherlands, mainly because of a too high suspended solids concentration in the effluent.

A growing number of sewage treatment plants in the Netherlands is going to be faced with more stringent effluent standards. In general, activated sludge plants will have to be extended with a post treatment step (e.g. sand filtration) or transformed into a Membrane Bio Reactor. In this case a GSBR variant with primary treatment as well as post treatment can be an attractive alternative.

Keywords aerobic granular sludge technology, activated sludge technology, potential, nitrogen removal, phosphate removal, sewage treatment.

Introduction

Conventional sewage treatment plants (STPs) based on activated sludge technology require a large footprint. This is caused by the relatively poor settling characteristics of activated sludge, resulting in low permissible dry solids (DS) concentrations in aeration tanks and in a low maximum hydraulic load of secondary sedimentation tanks. In the nineties of last century, compact attached growth technologies in several configurations were developed (immobilised bed, fluidised bed and airlift reactors). Main feature of these, continuously operated technologies is a high volumetric load, occasionally without a separate sludge/water separation step. The process conditions in airlift reactors are simple and the area requirement is limited. Because of the large specific biofilm area volumetric conversion capacities can be high (Heijnen et al., 1993 ). Main disadvantage of these systems is the high investment costs. Recent research showed that it is possible to grow granular sludge in a batch wise operated system without a carrier at high dissolved oxygen (DO) concentrations (>80%) resulting in high biomass concentrations and high volumetric loads (7,5kgCOD/(m³.d)) (Beun et al., 2000 , Etterer and Wilderer, 2001 ). Because of the outstanding settling properties the use of a traditional or integrated settler is not necessary. Separation of sludge and effluent occurs within the reactor during a short settling phase. So no long idle times due to sludge settling are required in these Granulated Sequencing Batch Reactors (GSBRs).

Presently, research is carried out to the mechanism of granulation and, although the precise mechanism is not known (yet) the following process conditions are assumed to play an important role:

o A short settling time results in the selection of well settling biomass, because biomass particles with low settling velocities are washed out with the effluent. At laboratory scale, granules with settling velocities higher than 12 m/h were maintained in the reactor;

o Applying high substrate gradients, for example by means of a pulse feed or a plug flow feed through the settled bed, is essential for granulation and for simultaneous conversion processes within the granule. Substrate diffuses into the core of the granules and because of the surplus of substrate, a part is converted into a storage product as poly-b-hydroxybutyrate (PHB). As a result substrate is equally distributed in the granule and the actual growth rate of the organisms will be lower, because of the growth on storage polymers. This growth rate combined with a non-transport-limited system, results in the formation of smooth and dense granules (Picioreanu et al., 1998 ). Another aspect of substrate diffusion throughout the granules is that shortages and decay of bacteria are minimised, which reduces disintegration of granules and which enables simultaneous COD and N-removal;

o Formation of smooth and dense granules is, as is showed in biofilm research (Gjaltema et al., 1997 , Tay et al., 2001 , Kwok et al., 1998 ), stimulated when granules are exposed to large shear forces caused by intensive (non-mechanical) mixing in the reactor. Because of the requirement of high shear forces laboratory research up to now has been carried out in an airlift reactor.

Laboratory research showed good results concerning simultaneous COD-, N- and P-removal, when an anaerobic feeding phase was combined with a low DO in the aerated period. In a 3-litre laboratory scale reactor synthetic wastewater was fed as a plug-flow through the settled bed of granules, during an anaerobic period of one hour (total cycle time three hours). The composition of the synthetic wastewater based on COD (with acetate as carbon source), Nkj and Ptotal corresponds with sewage. The aforementioned feeding pattern proved to generate more stable granules than with a pulse feed of three minutes. This was due to the selection of relative slow growing phosphate accumulating organisms (PAOs). These organisms store acetate as PHB during the anaerobic feeding period by releasing phosphate, while during the aerated period PAOs use the stored PHB as carbon source and take up the released phosphate again. During aeration, ammonia is converted to nitrate, which can serve as an electron acceptor for PAOs in the core of granules where oxygen is depleted. To enlarge the anoxic zone volume within granules, a low DO was applied (20%). At these conditions high COD-, N- and P-removal efficiencies were reached and amounted respectively 100%, 98% and 99%.

The effect of different hydrodynamic conditions (shear stress) was tested by comparing granulation in an airlift reactor as well as in a bubble column. Presumed disadvantage of a bubble column were lower shear forces compared with an airlift reactor, which may influence granulation negatively. Besides a longer start-up phase, granulation occurred in the bubble column with a similar performance compared to the airlift reactor.

Also an experiment with presettled sewage showed good results towards granulation. Granular sludge with a diluted sludge volume index (SVI) of 36 ml/g and an average diameter of 1.1 mm was formed in the laboratory.

The complete laboratory research, which results are summarised above, will be further outlined in other publications. In this paper the feasibility of a full-scale STP based on aerobic granular sludge technology is evaluated by means of a comparison with a conventional STP based on activated sludge technology.

General assumptions

A standard wastewater composition is used, as defined by the Dutch foundation for applied research for water management (STOWA): 600 mg/l COD (216 mg/l dissolved; 384 mg/l suspended); 55 mg/l Nkj (45.7 mg/l dissolved; 9.3 mg/l suspended), 9 mg/l Ptotal and 250 mg/l suspended solids. For the average wastewater flow 160 l/(pe.d) is taken with a peak (RWF) of 34l/(pe.h). The effluent requirements are based on the discharge regulations for municipal wastewater in the Netherlands, meaning extensive total nitrogen en phosphate removal (Ntotal = 10 mg/l and Ptotal = 1mg/l). Primary treatment consists of conventional sedimentation and the removal of suspended solids depends on the dosing of flocculants. Removal efficiencies for suspended solids of 50% (no flocculants) and 80% (dosing of flocculants) are taken. Metal salts are assumed not to be dosed in the GSBR variants because of the need for the selection of PAOs.

For the calculation of the capital costs depreciation periods for civil parts and mechanical/electrical parts of respectively 30 and 15 years are used. Capital costs are calculated based on annuities with an interest rate of 6%. Operational costs are based on main cost factors such as sludge disposal costs, power use aerators, chemical use and maintenance costs. The power use is corrected for power production generated from biogas. In Table 1 the starting points for the calculation of the operating costs are given.

TABLE 1 Starting points costs calculations Cost factor Unit Value Electricity ?/kWh 0.054 Sludge disposal ?/tonDrySolids 320 Iron chloride (41%) ?/ton 115 Land price ?/m² 22.7 Poly electrolyte (liquid, 50%) ?/kg active 3.6 Maintenance civil parts % of investments 0.5 Maintenance mechanical parts % of investments 2 Maintenance electro technical parts % of investments 2

Treatment alternatives and global technological starting points

Two alternatives of the aerobic granular sludge technology were compared to an alternative of a conventional STP based on activated sludge technology. The following GSBR variants were taken into account:

· GSBR with primary treatment including chemical dosing with extensive removal of suspended solids. Post treatment is assumed not to be required;

· GSBR with only post treatment. The post treatment consists of removal of suspended solids from the effluent from the GSBR (see figure 1).

Figure 2 gives a global process flow diagram (PFD) of the reference variant. The calculations of the reference alternative are based on a process temperature of 10 ^oC, a design sludge load of the aeration tanks of 0.14 kgCOD/(kgDS.d) and a SVI of 150 ml/g for the design of secondary sedimentation tanks.

The design of the GSBR-alternatives is based on biological phosphate removal which is assumed to be possible in a full-scale GSBR applying alternating anaerobic feeding periods and aerobic reaction periods. Preliminary calculations showed that chemical phosphate removal is too costly because of a higher sludge production resulting in increased sludge disposal costs and also because

FIGURE 1 Global PFD GSBR with post treatment

FIGURE 2 Global PFD reference alternative

of a higher chemical use. In order to be assured of plug flow conditions while feeding from the bottom of the GSBR through the settled granules bed, the maximum hydraulic surface load during the feeding period is chosen at 7.5 m/h. This leads to a construction height of the GSBRs, which are assumed to be built as bubble columns, of 5-6 m. The total cycle time amounts to 60minutes, which is formed by 20 minutes anaerobic feeding period, 27 minutes aeration or reaction phase, 5 minutes sedimentation followed by 7.5 minutes decantation. The COD design load of the GSBR is chosen at 0.3kgCOD/(kgDS.d), which corresponds with a pilot research for the application of an airlift reactor treating municipal wastewater (STOWA, 1997 ). In this research the net sludge production was almost zero, meaning an equal sludge content in influent and effluent.

The number of parallel treatment lines is determined by the length of feeding time compared to the total cycle time. Assumed is that at the most one GSBR can be fed with wastewater which leads to three treatment lines if the filling time is a third part of total cycle time. In case of post treatment, the overflow from the GSBR is buffered. Because effluent from the GSBR is discharged by gravity, the initial flow is high. In order to reduce the dimensions of the post treatment step, the overflow of the GSBR has to be buffered.

Sludge treatment for the reference and GSBR alternatives consists of gravitational thickening of primary sludge, mechanical of thickening surplus sludge, digestion and dewatering.

Results

Table 2 shows the effluent qualities for both GSBR-alternatives. Based on the influent characteristics and technological starting points the GSBR with post treatment can meet the effluent requirements. The GSBR-alternative with primary treatment does not meet the required effluent quality with respect to suspended solids. This is caused by an insufficient removal of suspended solids in the primary treatment and the assumption that suspended solids in a GSBR are not removed. A high suspended solids effluent concentration results also in increased Nkj- and Ptotal-concentrations. TABLE 2 Effluent quality GSBR

Parameter GSBR with

primary treatment GSBR with

post treatment COD (mg/l) 80 40 Suspended solids (mg/l) 50 < 10 Nkj (mgN/l) 4.1 2.0 NO₃^- (mgN/l) 5.9 8.0 Ptotal (mg/l) 1.5 1.0

The effluent requirements for the alternative with only primary treatment can only be met if the suspended solids concentration in the wastewater fed to the GSBR is less than 10-30 mg/l. However, this does not mean that a GSBR with primary treatment is not an attractive concept. If a more stringent effluent quality is required (e.g. Ntotal = 2.2 mg/l, Ptotal = 0.15 mg/l), which is in the Netherlands actual for a growing number of STPs discharging on sensitive surface waters, conventional activated sludge systems have to be extended with a post treatment step (e.g. with sand filtration) or can be transformed in Membrane BioReactors. In this case a GSBR with primary treatment as well as post treatment can be an attractive alternative.

The footprint of the total treatment plant is calculated by the sum of the net surfaces of all process units and buildings, multiplied by a factor 1,3. The calculations show that the footprints of the GSBR-alternatives are only around 25% of the footprint of the reference alternative. It can be concluded that the GSBR-technology is very compact, which is an important advantage in relation to activated sludge technology, especially in densely populated areas.

Figures 3 and 4 show respectively the investment costs and the total specific annual costs (sum of capital and operational costs). Despite the fact that a GSBR with primary treatment can not comply with the effluent standards, the costs for this variant are also depicted in figures 1 and 2. As can be seen from figure 1 the investments for the GSBR-alternatives are lower compared to the reference alternative (on the average 5%). As can be expected, the investment costs for the GSBR-alternative with primary treatment appear to be lower than the GSBR alternative with post treatment. This is mainly caused by the lower hydraulic design load for secondary treatment in comparison with primary treatment.

Figure 3 Investment costs

On the basis of the total specific annual costs the picture does not change (Figure 4). Again the GSBR-alternatives prove to be the most attractive. The total annual costs of the GSBR-alternatives with primary respectively post treatment are on the average 17% and 7% lower compared to the reference alternative. The capital costs of the GSBR-alternatives are relatively high because of the high share of the mechanical/electrical works on the investments (40-45%). In general the part of the mechanical/electrical works for conventional activated sludge systems amount to 25-30%.

Figure 4 Total specific annual costs

Sensitivity analysis

The influence of the RWF/DWF ratio and the land price on the total annual costs was calculated for a GSBR with post treatment as well for the reference, both for a capacity of 120.000 p.e. The results of the calculations are given in figure 5. The GSBR concept appears to be less sensitive to the land price than the reference, which is logical because of the compactness of the technology. On the other hand the GSBR technology is more sensitive to an increasing RWF compared to activated sludge technology. The reason for this higher sensitivity is because of the large impact of the maximum batch volume on the design of the GSBR. At higher RWF the maximum batch to be treated increases and as a result of this also the volume of the GSBR increases.

Figure 5 Sensitivity analysis

Conclusions

The aerobic granular sludge technology is a very promising technology and should be further developed.

The granulation process with acetate as carbon source is demonstrated to be stable in a batch wise operated airlift or bubble column reactor, with a pulse feed or anaerobic feeding period. A longer anaerobic period results in the selection of PAOs. This selection of PAOs in combination with a low DO makes a stable granulation process and high removal efficiencies for Ntotal and Ptotal possible.

Based on total annual costs both GSBR variants prove to be more attractive than the reference alternative (7-17%). Additional calculations showed a low sensitivity to the land price and a high sensitivity to the RWF.

Because of the high permissible volumetric load the footprint of the GSBR variants is only 25% compared to the reference. However, the GSBR with only primary treatment can not meet the present effluent standards for municipal wastewater in the Netherlands, mainly because of a too high suspended solids concentration in the effluent.

A growing number of sewage treatment plants in the Netherlands is going to be faced with more stringent effluent standards. In general, activated sludge plants will have to be extended with a post treatment step (e.g. sand filtration) or will have to be transformed into a Membrane Bio Reactor. In this case a GSBR variant with primary treatment as well as post treatment can be an attractive alternative.

Acknowledgements

This research was funded by the Dutch Foundation for Water Research (STOWA) within the framework of the “aerobic granule reactors” project (TNW99.262) as well as by the Dutch technology Foundation (STW) under project number DPC.5577.

References

Beun, J.J., M.C.M. van Loosdrecht and J.J. Heijnen (2000). Aerobic Granulation. Water Science and Technolology 41 (4-5): 41-48.

Etterer, T. and P.A. Wilderer (2001). Generation and properties of aerobic granular sludge. Water Science & Technology 43 (3): 19-26.

Gjaltema, A., J.L. Vinke, M.C.M. Van Loosdrecht and J.J. Heijnen (1997). Abrasion of suspended biofilm pellets in airlift reactors: Importance of shape, structure and particle concentrations. Biotechnology and Bioengineering 53 : 88 - 99.

Heijnen, J.J., M.C.M. Loosdrecht, R. Mulder, R. Weltevrede and A. Mulder (1993). Development and scale-up of an aerobic biofilm air-lift suspension reactor. Water Science and Technology 27 (5-6): 253-261.

Kwok, W.K., C. Picioreanu, S.L. Ong, M.C.M. van Loosdrecht, W.J. Ng and J.J. Heijnen (1998). Influence of biomass production and detachment forces on biofilm structures in a biofilm airlift suspension reactor. Biotechnology and Bioengineering 58 (4): 400- 407.

Picioreanu, C., M.C.M. Van Loosdrecht and J.J. Heijnen (1998). Mathematical modeling of biofilm structure with a hybrid differential-discrete cellular automaton approach. Biotechnology and Bioengineering 58 (1): 101-116.

STOWA (1997). Zuivering van stedelijk afvalwater in een airlift-biofilmreactor - uitontwikkeling en demonstratie. Utrecht, STOWA: 96.

STOWA (2001). Handboek biologische fosfaatverwijdering. Utrecht, Stichting Toegepast Onderzoek Waterbeheer: 234.

Tay, J.H., Q.S. Liu and Y. Liu (2001). The effects of shear force on the formation, structure and metabolism of aerobic granules. Applied Microbiology and Biotechnology 57 : 227 - 233.