Drawing on experience of BNR
James Barnard, Pedro Curto and Frank Rogalla of Black & Veatch explore the past 30 years of biological nutrient removal and the various theories put into practice at different works
In the European Union (EU), the Urban Wastewater Directive (UWD) set the effluent standards for sensitive receiving waters at 2mg/l for smaller plants and at 1mg/l for larger plants. Such stringent treatment objectives require combinations of biological and chemical treatment in both suspended and attached growth processes. Achieving phosphorus concentrations of less than 0.01mg/l, which only occurred in some isolated instances of sensitive receiving waters, like the Wahnbach Reservoir in Germany or the Florida Everglades in the US are now under wider discussion and nitrogen concentrations of less than 1mg/l are presently being evaluated. The high cost of removal to such low levels makes a better evaluation of the effect of different nitrogen and phosphorus species on the environment necessary. bpr theory
The theory of biological phosphorus removal (BPR) is well-established, requiring an anaerobic contact zone free of nitrates for contact of phosphate accumulating organisms (PAO) with volatile fatty acids (VFA).
Since nitrogen removal to low levels are not always required and some nitrates may be present in the return activated sludge (RAS), processes to reduce the nitrates in the RAS before returning it to the anaerobic zone were introduced, such as the University of Cape Town (UCT) and the Johannesburg (JHB) configurations, as well as the Westbank process. The latter process, shown in Figure 1, is a successful variation that combines all the features of the afore mentioned basin sequence to safeguard the anaerobic zone against nitrates and dissolved oxygen from the return sludge.
In this process RAS is passed through the pre-anoxic zone for further denitrification and 5-10% of the influent may be added to the RAS. The high-mixed liquor concentration encourages endogenous denitrification, which is augmented with some carbon from the influent to reduce the nitrates. The main influent stream is passed to the anaerobic zone to which fermenter supernatant is added.
Storm flow is diverted to the first anoxic zone to protect the anaerobic zone. Mixed liquor from the nitrification zone is recycled to the first anoxic zone for denitrification. The removal of nutrients from wastewaters requires low concentrations of effluent suspended solids (ESS) because these contain high concentrations of nitrogen (N) and phosphorus (P).
The empirical formula for biomass demonstrates each 100mg ESS contains 12mg N and 2.2mg P. Through the BPR mechanisms, the particulate phosphorus concentration discharged from activated sludge plants, chemical or biological, can be as high as 10mg P / 100mg ESS. Taking into consideration that some works are consented to remove total phosphorus (TP) to less than 1mg TP/l, it is fundamental that solids are removed to very low concentrations (typically to less than 10mg/l). In order to produce low TP in the effluent, particulate phosphorus must be removed from the effluent and for this reason filtration or flotation must be added.
Building on the theory originated by Fuhs & Chen1, the need for a source of volatile fatty acids (VFA) in the influent was recognised. With enough volatile acids in the influent, or with the help of fermenters, BPR can reduce ortho-phosphates in the effluent to around 0.1-0.2mg/l.
In warmer climates with flat sewer grades and a number of forced mains from pumping stations, VFA will be produced in the solids deposited in the sewers and on the slime layers in forced mains. However, it was clear that in colder countries and with combined sewer systems, some augmentation of VFA would be required. Acid fermentation of primary sludge was first used in Kelowna, British Columbia, Canada. Follow-up studies showed fermentation of primary sludge was essential for BPR. With combined sewer systems, fermentation is essential due to the low strength and low temperature during winter rainfall events.activated sludge
During pilot tests, Barnard2 observed good phosphorus removal in a four-stage biological nitrogen removal plant with no formal anaerobic zone. As shown in Figure 2, a second anoxic zone had an inadvertent connection to a dead zone, which was supposed to be isolated from the flow but in fact had two 25mm holes connecting the two compartments. Phosphorus was released from the sludge in the second anoxic zone to 32mg/l. When this happened, the effluent soluble phosphorus was less than 0.2mg/l from samples taken every 2h over a period of six weeks. The influent total phosphorous concentration was more than 8mg/l. Based on the theory of Fuhs & Chen, Barnard concluded when the biomass is retained under anaerobic conditions in the absence of a source of VFA and nitrates, a secondary release of phosphorus will take place that is unrelated to the uptake of VFA.
Over-sizing of unaerated zones can lead to secondary release. Without uptake of VFA there will not be any internal energy source to effect uptake under aeration. The fermentation of a portion of the phosphorus-containing biomass in the pilot plant must thus produce sufficient VFA for the uptake of phosphorus to occur in the aerated zone, while avoiding the phosphorus release from the entire biomass. Fermenting a portion of the RAS was also applied in the South Cary (North Carolina) water reclamation facility.
The South Cary WRF has a flexible design that can be operated in a number of configurations, all of which fermented a portion of the RAS before returning it to the anaerobic basin. Consistently good phosphorus removal resulted. It may thus be possible to simplify fermentation and produce excellent phosphorus removal.KEEPING THE SOLIDS
Besides depending on the available amount of volatile fatty acids, many BNR works have their capacity limited by the performance of the final tanks. Adding new final tanks or any additional solids separation units can be very costly and is not always feasible.
In the UK, for example, a 25m diameter clarifier can cost up to £0.5M, not including management fees and other considerations such as the footprint taken up, ground conditions, as well as changes in pipework and RAS pumping station.
This enhances the importance of finding options to optimise the secondary clarifier without the need for major upgrades. The first priority is to achieve well settling sludge to reduce the concentration of solids lost in the final effluent.
There are certain characteristics of the biological process in BNR works that can have a major impact on the characteristics of the solids entering the clarifier. These problems can be solved with changes in the biological treatment process configuration or operation. Step-feeding of the flows into the biological reactor is an option to increase solids inventory while reducing clarifier solids loading. Step-feed can also improve the clarifier performance, preventing denitrification and secondary phosphorus release, as less sludge needs to be stored in the final tank. The use of simulation models have shown that under step-feed configuration a reduction of the solids loading between 30-40% can be obtained when compared to conventional BNR design.
The use of a selector zone, usually anaerobic, to mix the RAS and influent wastewater before the aeration tanks can improve the sludge characteristics. The objective is to provide a short-term, high-substrate condition that favours certain floc-formers over the growth of filaments sensitive to low-food or micro-organism (F/M) ratios. This can solve sludge bulking problems in low F/M works and it has been incorporated in the design of many BNR works in the UK. When operating at higher solids concentrations, it might occur that there is inadequate capacity in the sludge collection system for the higher solids loading entering the tank. If the removal rate of sludge from the clarifier is not improved, this can cause denitrification problems and high-sludge blankets. Thus, in order to increase the transport of sludge to the hopper, some modifications can be considered. For conventional scraper mechanisms these can include replacement by a hydraulic suction system or spiral scraper, increase mechanism speed, increase length and depth of scraper and/or add additional arms.
If the RAS cannot be returned at the required rate, pump capacity can be upgraded to a certain limit but if the return rate is too high this can also deteriorate effluent quality due to increase in short-circuiting. Control of the RAS pumps should be checked and changed if required. In general, it appears most activated sludge operations perform well and require less attention when the constant RAS flow rate approach is used. With longer process sludge ages to achieve nitrification, floating sludge and biological scum can accumulate and cause effluent deterioration. There are a number of skimmer designs to remove scum. Some mechanisms collect the scum and recycle it with the RAS. These should be avoided as they keep the foaming organisms in the system promoting their growth. Some modifications to remove scum include submerging the centre well by a few mm to prevent scum accumulation.
With selective wastage of the scum from the aeration basin, the scum boards may be removed. With scum boards, the scum must be removed faster than it forms or more scum will grow and deteriorate effluent quality.modifications
Modifications to the hydraulic internal features of the clarifiers are used to control short-circuiting of the feed that reaches the underflow or effluent prematurely. These can take many forms but the most important ones are the modifications to the clarifier inlet and the use of baffles. The clarifier inlet design can have a large influence on the turbulence induced by the flow entering the tank. The installation of an energy dissipation inlet (EDI) can be used for the purpose of dissipating inlet energy, reducing disturbances to the sludge blanket caused by density currents and enhancing flocculation. The most successful EDI is claimed to be the LA - EDI.
Mixed liquor is discharged from the central EDI through outlets that impact on another, destroying the energy and assisting in flocculation. This type of EDI reportedly maintained a compact sludge blanket and was actually able to draw down an initially high blanket to the normal low blanket levels. Low ESS averaged 13mg/l. The tank equipped with the EDI sustained loadings of 2.65m3/m2/h and peak loadings of 2.89m3/m2/h for periods of 3h were achieved.
Research studies have demonstrated strategically located baffles within the final tank could significantly improve performance at relatively low cost. The use of baffles can prevent the development of density currents but when used in shallow tanks this still requires the sludge blanket to be kept low. The most common location for baffles is below the feed well (McKinney baffle) or in the side wall below effluent weir (Stamford baffle) or as an extension of the internal weir (Black & Veatch baffle).
The purpose of these is to deflect the gravity current of mixed liquor running up the perimeter wall. A successful case of the installation of Stamford baffles in one WwTW in the US has demonstrated an improvement on efficiency of the clarifier of around 14-22%, although higher rates have been reported in other works. Modifications to FST internal features can be hydraulically modelled using a high accuracy clarifier model (HACM).
The characteristics of this model allows simulations for different operating conditions such as flow rate, MLSS concentration, SVI and with or without baffles. Where space is available and in warmer climates, the activated sludge system is still the low-cost choice due to the potential use of internal carbon sources for denitrification and the saving in chemical costs. The five-stage Bardenpho (Barnard denitrification and phosphorus removal) process and effluent filtration are used extensively in warmer regions for achieving TN standards of less than 3mg/l and phosphorus to less than 1 mg/l. In colder regions the low rate of endogenous respiration in the second anoxic zone would require a large volume, alternatively a source of carbon must be added.
Since effluent filtration becomes necessary to attain low effluent TN and TP values, denitrification sand filters that combine attached growth denitrification with filtration could be used to reduce nitrates even from the simplest processes such as the Modified Lutzack Ettinger (MLE) pre-anoxic zone configuration without resorting to a second anoxic zone.
For lower space requirements, attached growth systems or a combination of attached growth nitrification and denitrification systems to the tail-end of high-rate activated sludge plants are more competitive in colder temperature climates. Chemically enhanced primary treatment (CEPT) followed by biological aerated filters (BAF) nitrification and denitrification is used successfully in the VEAS plant in Norway. However, a Westbank configuration for nitrogen and phosphorus removal is used successfully in Grimstad at mixed liquor temperature as low as 5°C. BAFs allow some filtering action as well and are also used for nitrification due to the greater simplicity. Denitrifying sand filters serve the double purpose of filtering and denitrification.
Chemical addition is required for these applications. The Broomfield plant in Colorado was retro-fitted as an integrated fixed film activated sludge (IFAS) system by installing screens in the aeration basin and filling the tank with 67% of floating Kaldnes plastic media. Mixed liquor passes through the plant and is settled in the final clarifiers.
The media with nitrifier growth are retained in the aeration basin by the screens and kept in suspension by the aeration device. RAS is returned to the anoxic zone, together with the feed to the plant. The addition of the media saved the expansion of the aeration basin to allow for nitrification, while still having the benefits of the MLE process for removal of TN.
The results of a pilot study with mobile media at the Broomfield plant in Colorado showed that at a temperature of 14°C it was possible to get nitrification with a four and a half day SRT, which is half of what was expected. Good simultaneous nitrification and denitrification was also observed leading to more than 85% removal of total nitrogen.
Pilot plant results are shown in Figure 3. For even lower residuals, chemicals can be added. A simple moving bed biological reactor (MBBR) process at Lillehamer with separate nitrification and denitrification, with the addition of methanol for nitrate reduction and Alum for phosphorus removal, achieved the results shown in Figure 4.
For the best period below the TN averaged 3.5mg/l and the TP averaged 0.15. Such low residuals are targeted as the future of BNR in the US by proposed legislation to set the total maximum daily load (TMDL). These limits fix the total mass that can be discharged in future to a river basin and allocates this to the dischargers. With future expansions of the plants a reduction of the effluent concentrations would be required with each expansion in capacity. In other sensitive locations proposed legislation will require the TN be reduced to lass than 1mg/l and the TP to less than 0.01mg/l.
Many plants in Florida and on the east coast of the US must already comply with a TN standard of 3mg/l and with TP less than 1mg/l. On inland water bodies the emphasis is on phosphorus and in some locations the effluent phosphorus discharge concentration is limited to less than 0.1mg TP/l.
At the Durham plant in Oregon, the limit never to exceed is 0.07 and the Pinery water plant in Colorado must average 0.03mg TP/l. There are a number of small plants in the state of New York that discharge into the reservoirs that supply water to New York, which must reduce phosphorus to less than 0.01mg/l.
The Durham plant consists of primary tanks, a three-stage Phoredox (A2O) and chemical post treatment with filtration. Initially the required effluent concentration was achieved by adding 170mg/l of Alum to the primary, secondary and tertiary stages.
While some VFA was present in the feed it was not sufficient for sustained biological phosphorus removal. Two existing tanks were then converted to fermenters to enhance the VFA supply. The effluent ortho-phosphates from the biological plant now range for the most part between 0.1 and 0.2mg P/l. Only about 25mg/l of Alum is still used to reduce the phosphorus to less than 0.07mg TP/l in a tertiary stage. At the Pinery water plant in Colorado, as shown in Figure 5, the average effluent phosphorus concentration of less than 0.03mg TP/l is achieved in a plant consisting of a five-stage Bardenpho plant followed by a chemical contact process with alum addition and sand filtration.
The biological plant averages effluent phosphorus of between 0.1-0.2mg TP/l. About 75mg/l of Alum is added to reduce TP to less than 0.03mg/l. Effluent phosphorus concentrations averaging 0.12 mg/l is achieved in Kalispell, Montana without chemical addition but with sand filtration. The future requirements for both low nitrogen and phosphorus in the effluent lead to conflicts in a number of the methods used.
For example, the five-stage Bardenpho plant with sand filters in Orange County, Florida achieves effluent soluble phosphorus of 0.1-0.15mg TP/l only when the nitrates are higher than 3mg/l. To achieve low effluent TN, nitrates in the mixed liquor have to be reduced to around 1 mg/l, which then results in the soluble phosphorus rising to 0.3-0.5mg TP/l.
Similar observations were made in the Bonnybrook plant in Calgary, Alberta. When the nitrate concentration in the effluent is low, some secondary release of phosphorus takes place in the final clarifiers, resulting in higher effluent phosphorus. Conflicts also develop in denitrifying sand filters for achieving low total nitrogen and phosphorus. Bacteria growing on the sand media use the carbon source for denitrification of all the residual nitrates in the effluent. The growth of heterotrophic denitrifying organisms requires phosphorus. If the phosphorus concentration in the filter feed is too low, it could impair the functioning of the denitrification mechanism. In the Truckee Meadows plant in Nevada the high-rate Phoredox effluent soluble phosphorus averages 0.2mg TP/l. Some phosphorus is required for the nitrifying trickling filters and for the denitrifying sand filters. It was observed phosphorus could be limiting in the latter, requiring the addition of phosphorus for high-rate nitrogen removal. This may again impact the level to which the phosphorus could be removed.
The plant configuration is shown in Figure 6. Methanol is added to the trickling filter effluent being discharged to the filters. Good clarification is required to ensure low effluent suspended solids.
Experience in other plants with some solids in the effluent was that the particulate matter removed in the filter contained accumulated phosphorus. The phosphorus may be released to the liquid phase under the anoxic conditions required to remove nitrates.
It may therefore be necessary to separate the attached growth denitrification stage from the final stage of phosphorus removal and filtration, as shown in Figure 7. Most membrane biological reactors (MBR) operate in the nitrification or denitrification mode with chemical addition for phosphorus removal.
The plants upstate on the watershed of New York city achieve TN of less than 3mg/l and phosphorus of less than 0.01mg/l by adding a carbon source for denitrification and alum for phosphorus removal. The price of the membranes used in the MBR processes is coming down but it can still not compete with some of the above processes except under conditions where a high-quality effluent must be produced. The membrane system replaces final clarifiers, sand filters and disinfection and produces an effluent that could be re-used.
The methodology for achieving effluent TP values to as low as 0.03mg/l and TN to as low as 2mg/l have been demonstrated. The study of methods for getting low levels of both nitrogen and phosphorus in the wastewater is ongoing under the leadership of Doctor Krishna Pagilla and the related report by the Water Environment Research Foundation (WERF) will be published about a year from now.
Phosphorus can be removed by chemicals but as the desired concentration goes down the consumption of chemicals goes up due to unwanted side reactions. Pretreatment by biological means can reduce the ortho-phosphorus to less than 0.1mg/l, with the addition of fatty acids or generation of these acids on site. A lot less chemicals would then be required to reduce the effluent phosphorus to less than 0.03mg/l, as has been demonstrated in various plants using Bardenpho, three-stage Phoredox or similar configurations, including pretreatment and post-chemical addition and effluent filtration. It has been demonstrated in small full-scale plants in the state of New York that effluent phosphorus well below 0.01mg/l was possible with the use of membranes. Configurations for getting both very low nitrogen and phosphorus are being investigated l
1 Fuhs GW and Chen M, Microbiological basis of phosphate removal in the activated sludge process for the treatment of wastewater, (1975), Microbiol. Ecol. 2(2), 119-138.2 Barnard JL, Cut P and N without chemicals, (1974), Water and Wastes Engineering, Part 1, 11(7), 33-36; Part 2, 11(8), 41-43.