Top Navigation Bar

ASAE Conference Proceeding

This is not a peer-reviewed article.

Assessment of Enhanced Biological Phosphorus Removal for Dairy Manure Treatment

K. A. Yanosek, M. L. Wolfe, and N. G. Love

Pp. 212-220 in the Animal, Agricultural and Food Processing Wastes, Proceedings of the Ninth International Symposium, 11-14 October 2003 (Raleigh, North Carolina, USA), ed. Robert Burns. ,11 October 2003 . ASAE Pub #701P1203


The goal of this research was to evaluate the potential for enhanced biological phosphorus removal (EBPR) as a nutrient removal technique for dairy manure. A critical factor in EBPR is the availability of carbon (C) from volatile fatty acids (VFAs), particularly acetic and propionic acids. Volatile fatty acids are the preferred energy source for phosphorus accumulating organisms (PAOs), the organism responsible for EBPR. The specific objectives of this research were to determine the VFA production, or fermentation, potential of dairy manure and to identify critical growth parameters needed to model EBPR. Volatile fatty acid production was determined using laboratory-scale reactors and gas chromatography. Acetic, propionic, and butyric acids accounted for 57, 23, and 20% of the total VFA measured, respectively, while valeric acid concentrations were below the detection level of the gas chromatograph. Fifteen percent of the readily biodegradable COD was converted to VFA-COD. The wastewater treatment design software, BioWin, was used to identify the most critical growth parameters in modeling EBPR. The effluent phosphate (PO 4 ) concentration was sensitive only to PAO growth parameters, suggesting that well-established populations of other heterotrophs would not interfere with EBPR, provided that environmental conditions favorable to growth of sufficient PAOs are maintained, and the sludge retention time is short enough to prevent growth of nitrifiers. The most critical microbial growth parameters in BioWin with respect to PO 4 concentration included one kinetic growth parameter, PAO maximum specific growth rate, and five stoichiometric parameters for PAOs: growth yield, aerobic PO 4 uptake rate per unit Poly-3-hydroxy-butyrate (PHB) utilized for growth, PHB yield per unit VFA uptake, PO 4 release per unit VFA uptake, and fraction of PO 4 taken up that can be released. These results suggest that sufficient VFA can be created by fermenting dairy waste, and improved EBPR design for treatment of dairy waste can be achieved by quantifying the six critical parameters.

KEYWORDS. Nutrient Management, Dairy Manure Wastewater, Fermentation Potential, Enhanced Biological Phosphorus Removal, Modeling.


Over the last two decades, livestock operations have become highly concentrated due to growing trends towards larger, more confined facilities and a decrease in cropland on smaller farms. This has led to greater amounts of excess manure nutrients on livestock farms, increasing the potential for nutrient pollution of water bodies from runoff. Excess nutrients in water bodies lead to water quality impairment, particularly eutrophication. The risk of phosphorus (P) runoff increases when manure is land-applied based on nitrogen (N) requirements of a crop because the N:P ratio required by the crop is greater than the N:P ratio in manure. If manure treatment could increase the N:P ratio of the manure, overapplicaiton of P with respect to crop needs could be reduced.

One treatment method that could potentially achieve this goal is enhanced biological phosphorus removal (EBPR). Enhanced biological phosphorus removal is achieved using intermittent aeration to promote growth of P accumulating organisms (PAOs). These bacteria store P, removing it from the liquid fraction of the waste stream. Biomass is then separated from the wastewater resulting in a lower P concentration in the liquid fraction and higher P concentration in the solid (sludge) fraction. The P sequestering activity of PAOs is referred to as EBPR.

A critical factor in EBPR is the availability of carbon (C) from volatile fatty acids (VFAs), the preferred energy source for PAOs. Liquid dairy manure tends to have a high concentration of phosphate (PO 4 ) relative to soluble chemical oxygen demand (COD) (Whichard, 2001). The greater the amount of PO 4 present in wastewater, the greater amount of VFAs will be required to remove the PO 4 via EBPR. Fermentation of the liquid fraction could potentially yield enough VFAs to support EBPR. It might be possible to ferment only a fraction of the wastewater and reblend it with the unfermented wastewater prior to EBPR treatment. This would reduce the volume of the fermentation vessel.

Wastewater treatment models have been developed to predict nitrification, denitrification, carbon (C) oxidation, and EBPR. Such a model could be a useful tool in evaluating the potential for using EBPR to treat dairy manure. The BioWin model (EnviroSim Associates Ltd., 2001) developed for municipal wastewater, might be appropriate for dairy manure applications because microbial processes that enable biological nutrient removal are the same in both municipal and dairy manure wastewater.

While there has been some application of EBPR to swine manure (Osada et al., 1991; Bicudo and Svoboda, 1995; Lee et al., 1997; Tilche et al., 1999), there has been very limited application of EBPR to dairy manure. To determine if EBPR could be a viable treatment alternative for dairy manure, it is first necessary to evaluate characteristics of dairy wastewater with respect to the EBPR process. The specific objectives addressed in this study were:

  • To determine the fermentation potential (volatile fatty acid (VFA) production) of dairy manure; and

  • To identify critical parameters of the EBPR component of the BioWin wastewater treatment model for application to dairy manure.


Fermentation Potential

Manure was collected from four dairy cows at the Virginia Tech Dairy. Two of the cows were fed a high P diet and two of the cows were fed a low P diet. Each of the four manure mixtures was passed separately through a mechanical solid separator with a 2-mm screen. The fraction passing through the screen was collected. Volatile fatty acid formation potential was determined using the following procedure as presented by Lie and Welander (1997). Fifty mL of the liquid fraction of solid separated manure was placed into 50-mL serum bottles. The bottles were sealed with butyl rubber stoppers. Triplicate bottles were prepared for each cow. Bottles were incubated at 23 ± 2 o C in the dark. A sample was obtained from each bottle daily by piercing the septum with a needle and withdrawing approximately 2 mL. Samples were analyzed for acetic, propionic, butyric, and valeric acid with a Hewlett Packard 5890 gas chromatograph equipped with a Carbopack (60/80) column (length 1.8 m, diameter 3.2 mm) and a flame ionization detector. The temperatures of the column, injector, and detector were 145??C, 225??C, and 250??C, respectively. Nitrogen gas saturated with phosphoric acid was used as a carrier gas (40 ml/min). Chemical oxygen demand was measured in duplicate for filtered, solid separated manure, as described in Standard Methods (APHA et al., 1998) by the Virginia Tech Dairy Nutrition Laboratory. Bottles were sampled over a period of 13 days at which point the VFA concentration was no longer increasing. The peak concentrations of VFA that occurred during this time were used to determine the total VFA potential.

Critical Model Parameters

A sensitivity analysis of the BioWin model was conducted to identify critical parameters in the EBPR component. A baseline scenario was selected based on desired PO4 removal. The baseline consisted of a sequencing batch reactor (SBR) configured to promote EBPR. BioWin default values were used for most microbial growth parameters. Experimentally determined values found in the literature were used for some parameters (Table 1). Dairy manure wastewater characteristics (Table 2) were estimated based on literature and fermentation potential results. Solid separated dairy manure characteristics determined by Whichard (2001) were used for model input except for colloidal biodegradable COD and soluble readily biodegradable non-VFA COD. These values were adjusted for prefermentation based on experimentally determined fermentation potential. Based on the laboratory study, it was assumed that 15% of the combined soluble and colloidal COD would be recovered as VFAs from prefermentation of dairy manure. Setting the fermentation rate in the model equal to zero resulted in the experimentally determined values being used in the model instead of simulated values.

Table 1: Baseline parameter values determined experimentally by Whichard (2001) that differed from BioWin default values.




Baseline Value

Non-PAO heterotrophic stoichiometric parameters

Yield (aerobic growth)

g COD/ g COD



Non-PAO kinetic stoichiometric parameters

µ MAX (aerobic/anoxic growth)

d -1




mg/L as COD



Decay rate

d -1



Fermentation rate

d -1



Autotroph kinetic parameters


g COD/ g COD



K s NH 4

mg NH 4 /L



Decay rate

d -1



Table 2: Influent wastewater characteristics.

State Variable



Before fermenter *

After fermenter

X sp


Slowly biodegradable particulate COD



X sc


Colloidal biodegradable COD



X i


Inert particulate COD



X on

mg N/L

Organic particulate nitrogen



S bsc


Soluble readily biodegradable non-VFA COD



S bsa


Soluble VFA COD



NH 3 -N

mg N/L




N os

mg N/L

Organic soluble nitrogen



PO 4 -P

mg P/L




S us


Soluble unbiodegradable COD





Inert suspended solids








*Values from Whichard (2001).

Adjusted concentrations based on experimentally determined fermentation potential.

Growth parameters are highly dependent on physiological state and wastewater characteristics (Grady et al., 1999). However, Whichard (2001) showed that some parameters are similar for dairy manure and municipal wastes. For each BioWin parameter, ranges of variability were identified (Yanosek, 2002). The sensitivity of model output to each BioWin microbial parameter was evaluated by running simulations for the range of values. Simulations were run for all microbial growth parameters for each of the three microorganism classifications in BioWin: PAO heterotrophs, non-PAO heterotrophs, and autotrophs. The model output concentration of PO 4 was used to identify critical parameters, i.e. parameters to which PO 4 was most sensitive. Relative sensitivity (Saltelli et al., 2000) was calculated for the effluent PO 4 concentration for each range of variables. A scale was designated to classify the varying degrees of relative sensitivity as “slight”, “moderate”, “high”, or “extreme” (Table 3).

Table 3: Relative sensitivity scale.

Relative Sensitivity * (RS)

Level of sensitivity

0 – 0.1


0.1< RS = 1


1< RS = 10


10< RS = 100




* Absolute value of relative sensitivity.

Simulations were run for two different baselines: complete PO 4 removal (BL 1 ) and incomplete PO 4 removal (BL 2 ). Parameters to which PO 4 -P was highly or extremely sensitive (RS > 10) were considered critical. In some simulations, EBPR was not maintained in the system, but the PO 4 concentration remained only slightly sensitive to the parameter. Parameters that did not maintain EBPR within the tested range of values were also considered critical.

Results and Discussion

Fermentation Potential

Peak VFA concentrations of acetic, butyric, and propionic acids were identified for each of the manure samples from cow 1 (low P diet) and cow 2 (high P diet) (Figure 1). Samples from cows 3 and 4 could not be analyzed due to equipment failure. Although valeric acid peaks were detected, they were below the reliable detection level of 10 mg/L of the gas chromatograph. The total average peak VFA concentration was 4479 ± 909 mg COD/L (Table 4). This peak VFA concentration corresponded to 15% of the total COD prior to fermentation. Acetic acid was produced in greatest abundance accounting for 57% of the total VFA measured as COD. Propionic and butyric acid accounted for 23 and 20% of the total VFA COD measured, respectively. The predominance of acetic acid during fermentation of mammalian wastes is in agreement with the literature (GonCalves, 1994; Skalsky and Daigger, 1995). Butyric acid comprised a more significant fraction of the VFA production compared to experimentally determined values for human wastes (Wentzel et al., 1989; Elefsiniotis and Oldham, 1991; GonCalves, 1994; Skalsky and Daigger, 1995).

Table 4: Peak acid concentrations of fermented solid separated dairy manure.



Peak acid concentration (mg VFA-COD /L)





Cow 1 Average

Low P





Cow 2 Average

High P





Overall Average





Acid concentrations given as COD, which was theoretically calculated from measured VFA concentrations and

theoretical COD equivalents.

Whichard (2001) determined the COD of solid separated dairy manure obtained from a full-scale dairy operation in Pennsylvania. The soluble plus colloidal (1.5 µ g filtered) COD was 3960 mg/L and the total COD was 5480 mg/L. Assuming conversion of 15% of influent COD to VFA, the VFA concentration would be between 594 and 822 mg VFA-COD, depending on the extent of hydrolysis of particulate COD. Phosphate removal by the addition of individual VFAs of acetic, propionic, and butyric acids were 18.8, 31.5, and 39 mg COD/mg PO 4 , respectively, as reported by Abu-ghararah and Randall (1991). Assuming these PO 4 removal rates and the average production of acetic, propionic, and butyric acids from this research (57, 23, and 20% of the total VFA-COD, respectively), it was estimated that 36 mg PO 4 could be removed. Other reported requirements for removal of 1 mg PO 4 /L range from 6 to 20 mg acetic acid (as COD)/mg PO 4 removed (Abu-ghararah and Randall, 1991; Barnard, 1994; Lie et al., 1997). The amount of PO 4 removed per unit VFA determines the PO 4 removal ability of a particular system, as demonstrated in Figure 2.

Figure 1: Volatile fatty acid concentrations in three laboratory-scale fermenters containing manure from cows 1 (low P diet) and 2 (high P diet). Sample number designation refers to cow; letter designation refers to fermented triplicate.


Figure 2: Phosphate removal for given removal ratios for different amounts of VFA-COD derived from fermentation. The lower line indicates fermentation of soluble COD fractions and the higher line indicates fermentation of colloidal and soluble COD fractions.


Critical Model Parameters

Sensitivity analysis of the Biowin model showed, as expected, that PO 4 removal was not affected by changes in autotrophic growth parameters, indicating that inadequate growth conditions existed in the reactor for propagation of these organisms. Relative sensitivity to all autotrophic growth parameters was zero. Phosphate was only slightly sensitive to changes in some non-PAO heterotrophic parameters. Fermentation by non-PAO heterotrophs was not necessary in the SBR since sufficient VFA-COD was available for PAO heterotrophic growth in the pre-fermented influent. Therefore, in simulations where changes in growth parameters eliminated non-PAO heterotrophs from the system, PAO-heterotrophs continued to grow.

Six critical PAO heterotrophic parameters were identified with respect to PO 4 concentrations in BioWin simulations (Figure 3). These included one kinetic growth parameter, maximum specific growth rate ( µ MAX,ZBP ), and five stoichiometric growth parameters: growth yield (Y ZBP ), aerobic PO 4 uptake rate per unit Poly-3-hydroxy-butyrate (PHB) utilized for growth (f P/PHB,AER ), PHB yield per unit VFA uptake (Y PHB ), PO 4 release per unit VFA uptake (f P/AC ), and fraction of PO 4 taken up that can be released (Y PP-LO ).

For BL 1 , PO 4 concentration was highly sensitive to Y ZBP and extremely sensitive to µ MAX,ZBP , f P/AC , Y PHB , Y PP-LO , and f P/PHB,AER . For BL 2 , PO 4 concentration was only slightly sensitive to Y ZBP and moderately sensitive to µ MAX,ZBP , f P/AC , Y PHB , Y PP-LO , and f P/PHB,AER . After mixed liquor concentration of PAO heterotrophs dropped to zero, no EBPR took place. The reduction in PO 4 after washout of PAOs was due to nutrient requirements for cell growth and precipitation. The PO 4 effluent concentrations at the point of washout were 52.4 and 52.9 mg PO 4 -P/L for BL 1 and BL 2 , respectively.

212-220isaafpw_files/image3.gif 212-220isaafpw_files/image4.gif

212-220isaafpw_files/image5.gif 212-220isaafpw_files/image6.gif

Figure 3: Concentrations of PO 4 effluent (from the SBR) and corresponding mixed liquor (in the SBR) concentrations for model outputs at distinct parameter values for simulations completed using BL 1 and BL 2 . Solid symbols indicate parameter values enabling EBPR and open symbols indicate washout of PAOs.

212-220isaafpw_files/image7.gif 212-220isaafpw_files/image8.gif

For each of the six parameters to which PO 4 was most sensitive, there was a range of values within which the growth and activity of PAOs was supported according to the BioWin simulations. The ranges of parameter values within which EBPR occurred were similar for both baselines (Table 5).

Table 5: Parameter ranges enabling EBPR in BioWin for complete PO 4 removal (BL 1 ) and incomplete PO 4 removal (BL 2 ).



Range Tested

EBPR Range

BL 1

BL 2


d -1




f P/AC
















g P/g COD









Ranges summarized in Table 5 include only the values at which EBPR was achieved. The exact parameter values at which PAO biomass growth could be sustained were not determined.


Addition of VFA is required to enable EBPR in dairy manure. Prefermentation can provide a significant amount of VFA to enable EBPR. In this study it was found that 15% of the total COD was converted to VFA. Whether the VFA production is sufficient to meet PO 4 removal requirements depends on the initial PO 4 and COD concentrations as well as the PO 4 removal ratio of the VFAs produced from fermentation. Further experimental investigation is required to characterize EBPR activity in dairy manure wastewater.

A sensitivity analysis of BioWin demonstrated the complexity of EBPR activity and the range of output values that could result from various numerical inputs for microbial growth parameters. BioWin PO 4 output was most sensitive to maximum specific growth rate, growth yield, aerobic PO 4 uptake rate per unit PHB utilized for growth, PHB yield per unit VFA uptake, PO 4 release per unit VFA uptake, and fraction of phosphate taken up that can be released. Due to the sensitivity of PO 4 to the critical parameter values, BioWin should not be used for treatment system design without calibration and/or parameter determination. Experimental determination of the most critical microbial growth parameters would facilitate modeling and design of EBPR systems.


Funding was provided by the Virginia Tech ASPIRES program and the Departments of Biological Systems Engineering and Civil and Environmental Engineering. We acknowledge the assistance of Dr. Katharine Knowlton and Julie McKinney from the Dairy Nutrition Laboratory, and Ms. Jody Smiley and Ms. Julie Petruska of the Environmental Engineering Analytical Laboratory at Virginia Tech.


Abu-ghararah, Z.H. and C.W. Randall. 1991. The effect of organic compounds on biological phosphorus removal. Water Science and Technology 23(4-6): 585-594.

APHA. 1998. Standard Methods for the Evaluation of Water and Wastewater 20 th Edition. American Public Health Association, Washington D.C. pp. 2:57-59, 5:15-16.

Barnard, J.L. 1994. Alternative prefermentation systems. In Proceedings of the 67 th Annual Water Environment Federation Conference, Chicago . Alexandria, VA: Water Environment Federation.

Bicudo, J.R. and I.F. Svoboda. 1995. Effects of intermittent-cycle extended-aeration treatment on the fate of nutrients, metals and bacterial indicators in pig slurry. Bioresource Technology 54: 63-72.

Elefsiniotis, P. and W.K. Oldham. 1991. The effect of operational parameters on the acid-phase anaerobic ferementation in the biological phosphorus removal process. In Proceedings of the ASCE Environmental Engineering Division, National Conference on Environmental Engineering, Reno, NV, 325-330 . New York: American Society of Civil Engineers.

EnviroSim Associates Ltd. 2001. User Manual for BioWin32. Flamborough, Ontario, Canada.

GonCalves, R.F., A.C. Charlier, and F. Sammut. 1994. Primary fermentation of soluble and particulate organic matter for wastewater treatment. Water Science and Technology 30(6): 53-62.

Grady, C.P., G.T. Daigger, and H.C. Lim. 1999. Biological Wastewater Treatment . New York: Marcel Dekker, Inc.

Lee, S., J. Park, K. Ko, and B. Koopman. 1997. Effect of fermented swine wastes on biological nutrient removal in sequencing batch reactors. Water Science and Technology 31(7): 1807-1812.

Lie, E. and T. Welander. 1997. A method for determination of the readily fermentable organic fraction in municipal wastewater. Water Research 31(6): 1269-1274.

Lie, E., M. Christensson, K. Jonsson, P. Johansson, and T. Welander. 1997. Carbon and phosphorus transformations in a full-scale enhanced biological phosphorus removal process. Water Research 31(11):2693-2698.

Osada, T., K. Haga, and Y. Harada. 1991. Removal of nitrogen and phosphorus from swine wastewater by activated sludge units with the intermittent aeration process. Water Resources 25(11): 1377-1388.

Saltelli, A., K. Chan, and E.M. Scott. 2000. Sensitivity Analysis, 10. New York: John Wiley and Sons Ltd.

Skalsky, D.S. and G.T. Daigger. 1995. Wastewater solids fermentation for volatile acid production and enhanced biological phosphorus removal. Water Environment Research 67(2): 230-237.

Tilche, A., E. Bacilieri, G. Bortone, F. Malaspina, S. Piccinini, and L. Stante. 1999. Biological phosphorus and nitrogen removal in a full scale sequencing batch reactor treating piggery wastewater. Water Science and Technology 40(1): 199-206.

Wentzel, M.C., P.L. Dold, G.A. Ekama, and G.v.R. Marais. 1989. Enhanced polyphosphate organism cultures in activated sludge systems. Part III: Kinetic model. Water South Africa 15(2): 89-102.

Whichard, D.P. 2001. Nitrogen removal from dairy manure wastewater using sequencing batch reactors. M.S. Thesis. Blacksburg: Virginia Tech.

Yanosek, K.A. 2002. Enhanced biological phosphorus removal from dairy manure to meet nitrogen:phosphorus crop nutrient requirements. M.S. Thesis. Blacksburg, VA: Virginia Tech.