Water Utility Journal 14: 19-27, 2016. © 2016 E.W. Publications

Kinetics of COD in wastewater stabilization ponds in northern Greece

M. Gratziou* and M. Chalatsi 1 Department of Civil Engineering, Democritus University of Thrace, Xanthi, 76133, Greece * e-mail: mgratzi@civil.duth.gr

Abstract: Only few models have been developed to simulate pollutant removal processes in Wastewater Stabilization Ponds (WSPs), especially for the Greek climate conditions, where the use of WSPs is limited. The knowledge of pollutant removal reaction rate constants is very useful to design and size Wastewater Stabilization Ponds. The present work studies the treatment efficiencies of two full scale WSPs systems situated in Northern Greece, in a Mediterranean climate. The mean monthly air temperatures range between 4oC to 29oC. These systems treat municipal wastewater of rural settlements and they consist of one facultative pond, two maturation ponds and a rock filter used for algae filtration before the final discharge. The study estimates the reaction rate constants of COD removal mechanisms. Before this estimation, the outflow data were corrected by the mass balance method to eliminate errors from atmospheric precipitation and evapotranspiration. The used models for the K estimation are the Kickuth Equation model, the combined model of first order and CSTR, the global First order substrate removal model for complete mixed system, the Stover–Kincannon model and the Kadlec and Knight model. Assessment of the models and the accuracy and reliability of the results were made through comparison with existing real data. The study estimates these essential design parameters pertaining to local conditions to optimize the design considerations and sizing requirements. The constant K has a very strong relationship with hydraulic retention time.

Key words: wastewater; stabilization ponds; COD kinetic; biodegradation rate constant

1. INTRODUCTION

Wastewater Stabilization Ponds (WSPs) are simple, in engineering terms, structures, with significant advantages, such as simplicity, low construction and operating costs and high efficiency in organic matter and pathogen organism removals (Penha-Lopes et al., 2012; Kayombo et al., 2005; Al-Sa'ed, 2007; Ferrara and Avci, 1982; Senzia et al., 2002; Alcalde et al., 2002; Amahmid et al., 2002). Despite their simplicity, WSPs are extremely complex as ecological systems (Kayombo et al., 2005). Most empirical models or design tools developed have used on-site or regionally-specific WSP data. Different characteristics, in terms of climate and hydrology, can lead to problems, when models are transferred without appropriate adjustment to local conditions. Using “uniform” simple methods in all countries often results to malfunctions or reduced efficiency of WSP system (Tilley et al., 2014). Several WSP design tools and models have been adopted; however, information on the dynamics of pollutants in these systems are lacking. Lack of such information, and the fact that in Greece the use of these systems is limited and the relevant infor-mation is insufficient, were the impetus for this work.

In the wastewater treatment literature, many kinetic models for biomass growth processes have appeared (Henze et al., 1997; Beltrán et al., 2000; Chlot et al., 2011). Most of these models describe the number of microorganisms, without considering the consumption of substrate. Zwietering et al. (1990) derived a modified mathematical relationship based on the Gompertz model (Kayombo et al., 2005) for the increase in biomass over time, which relates the population size over time to the specific growth rate, lag time, and asymptotic level of organisms. Parameters such as TSS, COD, BOD were used as substrate indicators for evaluation under the assumption that the removal was exclusively due to aerobic biodegradation (Carta et al., 2004).

The aim of the present research is the determination of COD pollutant removal reaction and the biodegradation rate constants KT of two full-scale WSPs systems situated in Northern part of Greece, treating municipal wastewater, pertaining to local conditions. One of the objectives was to

20 M. Gratziou & M. Chalatsi

determine the suitability and usefulness of different models, available in the literature. The models were compared using statistical efficiency criteria, by which the lack of fit of the models was compared.

2. MATERIALS AND METHODS

2.1 Data

The two systems (Vamvakofito, Charopo) are situated in a lowland area in mainland of northern Greece (latitude φ: 41ο up to 41ο15’ N; longitude λ: 23ο21΄ up to 23ο23΄E) and altitude from 30 m to 52 m, in a Mediterranean climate area. The mean monthly air temperatures vary from 4 oC to 29 oC. They treat only domestic wastewater and consist of one facultative pond, two maturation ponds and a rock filter used for algae filtration before the final discharge. Wastewater is discharged through an open channel of 0.75 m2 vertical section. Every system has a different size, shape (Figure 1) and total hydraulic retention time (HRT). Vamvakofito’s HRT is 68.7 d and Charopo’s 72.4 d.

System of Vamvakofito System of Charopo

Figure 1. Ground Plan of WSP systems From each system, instantaneous samples were taken from the inflow of the facultative pond and

the outflow of the last maturation pond, during the years 2006, 2007 and 2012, twice a month or more frequently (Chalatsi, 2014). Samples were collected approximately at the same morning period, but different day of the week, according to the suggestions of Mara and Pearson (1987), while meteorological data were recorded. The samples were placed into 1,000 mL polyethylene bottles, and were transferred immediately to the wastewater laboratory of Serres City (Chalatsi, 2014). To enhance the range and accuracy of data, each of the samples was analyzed separately twice, with methods proposed by Simplified Laboratory Procedures for Wastewater Examination

1st Maturation

H=1.50 m

E=1,783m2

2nd Maturation

H=1.50 m

E=1,783m2

Facultative

H=2.40 m E=2,449m2

1st Maturation H=1.50 m E=3,665m2

2nd Maturation

H=1.50 m

E=1,875m2

Facultative

H=2.40 m E=1,875m2

Water Utility Journal 14 (2016) 21

(Eaton et al., 2005) and the averages were considered. The inflow and the outflow rates were measured with a handheld electromagnetic flow meter, under the assumption that the wastewater supply was constant during the day. The climate is classified as dry to semi-humid with excess of water in winter. The average annual air temperature is 15.2 °C and the average annual rainfall is 448.5 mm. The winds in the area are very weak and do not normally exceed 6 km/h.

The outflow data were corrected by the mass balance method to eliminate errors from atmospheric precipitation and evapotranspiration, as many researchers believe that the mass balance is the most authoritative method to approach mechanisms and parameters that determine the performance of natural systems and the changes occurring in these (Heliotis and DeWitt, 1983; Breen, 1990; Korkusuz et al., 2005).

The water balance estimation, described by Eq. (1), uses the principles of conservation of mass in a closed system:

Qout = Qin + I – PET (1)

where Qout is the wastewater outflow rate [m3/d], Qin is the wastewater inflow rate [m3/d], I is the water quantity which enters the system via precipitation [m3/d] and PET is the water quantity lost from the system via evapotranspiration [m3/d]. The precipitation depth Hrain was obtained from the Hellenic Meteo Service, Bureau of Serres, and the evapotranspiration depth HPET was calculated by applying the Thornthwaite method (Korkusuz et al., 2005; Chalatsi, 2014), due to the small number of required data for its implementation, compared to the Perman-Monteith equation, which is considered as more reliable (Ward and Robinson, 2000). The Thornthwaite method gives a very good estimation of the water balance for the purposes of this research. Having estimated by Thornthwaite method the evapotranspiration depth HPET, and knowing the amount of Hrain, the change of pond water level ΔH [cm] could be calculated. Multiplying ΔH by the surface of each system, changes in volume ΔV [m3] were estimated. Dividing ΔV by the number of the days of each month, the term ΔV/d [m3/d] could be obtained, i.e., the daily change of the pond volume for each month. The term ΔV/d is subtracted from the initial daily flow, Q, thus resulting in a new term Q΄ (Eq. 2):

Q΄ = Q – (ΔV/d) (2)

Multiplying the new daily flow Q΄ [m3/d] by the mean of the concentrations [mg/L] and by the number of days elapsed between sampling (d),

!!"#!!!!"#!!!!

, the output mass Massout [kg] was estimated - with appropriate conversion of units (Eq. 3). Similarly the input mass in the system, Massin [kg] was also estimated (Eq. 4).

Massout = Q΄ Δt !!"#!!!!"#!!!

! 10-3 (3)

Massin = Q Δt !!"!!!!"!!!

! 10-3 (4)

The difference Massin - Massout determines the variation in mass throughout the study period. The mass balance is described by the general expression (Eq. 5) (Kuo, 2014).

mass accumulation = Massin - Massout + mass generation or mass consumption (5)

2.2 Kinetic models and their evaluation

The most commonly used models described WSP and mainly Constructed Wetllands kinetics are the Zwietering model (Zwietering et al., 1990), the combined model of first-order and plug flow, initially proposed by Kickuth (Sun and Tanveer, 2009; Khosravi et al., 2013a), the combined model

22 M. Gratziou & M. Chalatsi

of first-order and CSTR (Continuous flow Stirred Tank Reactors) regime (Sun and Tanveer, 2009; Khosravi et al., 2013a) and the global first-order substrate removal model for a complete mixed system (Vavlin et al., 2001; Carta et al., 2004; Khosravi et al., 2013b). The Stover–Kincannon (1982) model expresses the substrate utilization rate as a function of the organic loading rate by monomolecular kinetics for biofilm reactors, such as biological filters and moving bed biofilm reactors (MBBR) (Borghei et al., 2008). The model was suggested by Khosravi et al. (2013b) having a good adaptation in the case of stabilization ponds. Similarly, the Kadlec and Knight (1996) model, developed for wetlands, was also considered in this research. As the examined stabilization pond systems have characteristics similar to wetlands - no sludge removal throughout the years of operation and a significant growth of self-sown reeds at the banks of the ponds (Figure 2) - this model was assumed that it could be used.

Facultative Pond of Vamvakofito WSP system Facultative Pond of Charopo WSP system

Maturation Pond of Vamvakofito WSP system Maturation Pond of Charopo WSP system

Figure 2. General view of WSP systems where is apparent the plants growth

Combining the first-order reaction kinetics and the plug flow regime, Eq. (6) can be reached, firstly proposed by Kickuth (1977). It has been widely used for the sizing of HSF CWs for domestic sewage treatment:

K1 = !!ln (𝐶!" − 𝐶!"#), [m d-1] (6)

where K1 is the first-order kinetic constant for organic pollutants removal, Q is the flow rate [m3d-1], A is the pond area [m2], Cin is the input pollutant concentration [mg L-1], Cout is the output pollutant concentration [mg L-1].

The combined model of first order and CSTR gives the following equation (Khosravi et al., 2013a):

K2 = !(!!"!!!"#)!(!!"# )

, [m d-1] (7)

Water Utility Journal 14 (2016) 23

According to the first-order substrate removal model, the first-order kinetic constant in a complete mixed system can be expressed as follows (Khosravi et al., 2013a):

K3 = !!"!!!"#!!×!!"#

, [d-1] (8)

where θΗ is the hydraulic retention time (HRT) [day].

By the model of Stover–Kincannon (1982), the saturation value constant KB was estimated as follows (Sun and Tanveer 2009):

KB = !!

!!"!!!"#! !!!"#

×!!"×!!"#

!! , [g L-1d-1] (9)

where Umax is the maximum substrate removal rate [mg COD L-1d-1]. The above models describe the kinetics, without considering the consumption of substrate.

Zwietering et al. (1990) derived a modified mathematical relationship based on the Gompertz model for the increase in biomass over time, which relates the population size over time to the specific growth rate, lag time, and asymptotic level of organisms. Kadlec and Knight (1996) developed a model, for wetlands, that is a combination of the basic equation of the plug flow model and the aqueous mass balance. This model is known as K-C* model. It differs from the original Kickuth equation in two ways. It is a reversible first-order reaction equation and includes a non-zero substrate concentration. It describes better the removal of pollutants, as they cannot be reduced to zero in wetlands or in ponds, due to the subsequent release of pollutants from the ponds into the treated water. The non-zero background concentration represents in a more realistic way the pollutants resulting from transformation processes within the sediments and from the interactions between the sediments and the wastewater. The main reason of these processes is the production of organics from the decomposition of organic materials and the endogenous autotrophic processes (IWA, 2000; Ronnie and Frazer, 2010). The substrate utilization rate was directly related to the specific growth rate of heterotrophic bacteria in the stabilization ponds, as it was also shown by Panikov (2000) and Kayombo et al. (2003). The K-C* model is written as in Eq. (10) (Kadlec and Knight, 1996):

𝐾! =!!ln !!"!!∗

!!"#!!∗ , [m d-1] (10)

where K5 is the first-order kinetic constant and C* is non-zero background concentration [mg/L]. The value of C* for COD, according to Kadlec and Knight (1996), is equal to 3.5 + 0.053 Cin.

All the above-mentioned equations have been used in this research. The model assessment, the accuracy and reliability of the results were evaluated by comparing model results with existing real data. For each stabilization pond system, the value of “KCOD” was obtained through the calibration of linear regression and as the median of the predicted values.

The models denominated by Equations (6) to (10) were evaluated by comparing the real obser-ved values, Cout of collected data from stabilization ponds, with the predicted by these equations F(Cout) values. To evaluate model performance, efficiency criteria were defined, as mathematical measures explaining how well the model simulation fits the available observations (Krause et al., 2005; Moriasi et al. 2007). The used efficiency criteria, in this study, were: (i) the coefficient of determination r2, defined as the squared value of the coefficient of correlation according to Bravais-Pearson. It provides a measure of how well observed outcomes are replicated by the model, as the proportion of total variation of outcomes explained by the model. The range of r2 lies between 0 (no correlation) and 1.0 (the dispersion of the prediction is equal to that of the observation). Τhe fact that only dispersion is quantified is one of the major drawbacks of r2. When considered alone, it is advisable to take into account additional information that can cope with that problem. (ii) The Nash-

24 M. Gratziou & M. Chalatsi

Sutcliffe efficiency E. The range of E lies between 1.0 (perfect fit) and -∞. An efficiency of 0 (E=0) indicates that the model predictions are as accurate as the mean of the observed data, whereas an efficiency less than zero (E<0) occurs when the observed mean is a better predictor than the model. Essentially, the closer the model efficiency is to 1, the more accurate the model is. (iii) The unitized risk or coefficient of variation CV that is defined as the ratio of the standard deviation σ to the mean µ. It shows the extent of variability in relation to the mean of the population. CV measures are often used as quality controls for quantitative laboratory assays. The closer to zero is the CV value, the better the fit. The combination of the above criteria gives more information about the efficiency of used equations.

3. RESULTS AND DISCUSION

After the mathematical and statistical processing of the collected data, the estimated values of KCOD are presented in Tables 1 and 2.

Table 1. Stabilization Ponds of Vamvakofito – COD biodegradation rate constant (the obtained removal was 69%)

Equation

6 Kickuth model

7 1st order and

CSTR

8 1st order in

complete mixed

9 Stover–Kincannon

10 K-C* model

m d-1 m d-1 d-1 g L-1 d-1 m d-1 KCOD 0.0993 0.0349 0.1817 22.4078 0.0229

r2 0.8553 0.8389 0.2357 0.4090 0.8389 E 0.9767 0.8530 0.2749 0.4894 0.8314

CV 0.0531 0.1572 0.9000 0.4265 0.1068 Median 0.0991 0.0351 1.0832 33.7538 0.0231

STD 0.0052 0.0055 1.6872 13.5981 0.0024 MIN 0.0861 0.0220 0.1704 10.1372 0.0168 MAX 0.1062 0.0503 6.9358 59.5981 0.0296

Table 2. Stabilization Ponds of Charopo - COD biodegradation rate constant (the obtained removal was 46%)

Equation

6 Kickuth model

7 1st order and

CSTR

8 1st order in

complete mixed

9 Stover–Kincannon

10 K-C* model

m d-1 m d-1 d-1 g L-1 d-1 m d-1

KCOD 0.0808 0.0161 0.0774 27.9845 0.0121 r2 0.9270 0.9212 0.6250 0.8534 0.9212 E 0.9363 0.9262 0.7528 0.9570 0.8828

CV 0.0512 0.2108 0.8938 0.3874 0.1706 Median 0.0818 0.0160 0.5820 25.9313 0.0128

STD 0.0041 0.0034 0.9522 9.5863 0.0022 MIN 0.0741 0.0096 0.1368 10.2163 0.0085 MAX 0.0880 0.0224 3.9478 47.2700 0.0169

The values used in Eq. (9) were Umax the maximum substrate removal rate for Vamvakofito WSP

system, which according to measurements was 34.68, while for Charopo, a value of 19.74 [mg COD L-1d-1] was measured.

Taking into account all the equations, the efficiency statistic criteria (r2, E, CV) show that for the two WSP systems the simple Kickuth Equation gives the best results for “KCOD”. However, each WSP system, as an alternative ecological system, operates with different rates. So, for Vamvakofito WSP system, a KCOD value of 0.0993 [m d-1] was estimated, while for Charopo WSP system, the

Water Utility Journal 14 (2016) 25

KCOD proposed value is equal to 0.0808 [m d-1]. The saturation value constant for Charopo KB is 27.9845 [g COD L-1d-1]. The respective saturation value constant for Vamvakofito system KB is equal to 22.4078 [g COD L-1d-1], values to be used with caution, due to the unsatisfactory statistical correlation. Shahryari et al. (2013), in a stabilization pond system in Birjand, East Iran, derived the respective values for COD removal rate as 0.06 [m d-1] and 23.81 [g COD L-1d-1].

For both systems, the suggested biodegradation rate constant values K, after calibration, from all data and linear regression are: KCOD = 0.095 [md-1] and KB-COD = 25.20 [g COD L-1d-1], with r2 > 0.79 for 108 mg L-1 < CODin < 301 mg L-1.

Figure 3. Relationship between the outflow corrected value of COD concentration (real data) Cout and the COD concentration outflow value Ĉout generated by Kickuth model (KCOD = 0.095)

The rate constant (K) has a strong positive relationship (R² = 0.985) with HRT expressed as follow:

KCOD = 0.3069 – 0.0031HRT (11)

where HRT is in days.

Figure 4. Relationship between KCOD and hydraulic detention time (HRT)

4. CONCLUSIONS

Even when the terms of climate and hydrology are the same, the WSP systems have different characteristics and behavior, as their operation is multiparametric and their nature represents a complex ecological system. The Kickuth Equation, which is typically used in the design of

26 M. Gratziou & M. Chalatsi

stabilization ponds, showed a good mathematical relationship between theoretically predicted results and real data. For conditions in Greece and in similar WSP systems, a COD biodegradation rate constant (K) equal to 0.095 [m d-1] and a saturation value constant KB-COD of 25.20 [g COD L-1d-

1] can be proposed, with r2 > 0.79 for 108 mg L-1 < CODin < 301 mg L-1. This constant K has a very strong relationship with the hydraulic retention time (HRT). The estimated parameters can effecti-vely be applied in WSP sizing in typical Mediterranean climatic conditions.

ACKNOWLEDGEMENTS

An initial version of this paper has been presented in the WASTEnet Program Conference, “Sustainable Solutions to Wastewater Management: Maximizing the Impact of Territorial Co-operation”, Kavala, Greece, June 19-21, 2015.

REFERENCES

Al-Sa'ed R. (2007) Sustainability of natural and mechanized aerated ponds for domestic and municipal wastewater in Palestine. Water International, 32(2): 310-324.

Alcalde L., G. Oron G., L. Gillerman, M. Salgot and Y. Manor (2002) Removal of fecal coliforms, somatic coliphages and F-specific bacteriophages in a stabilization pond and reservoir system in arid regions. Water Science and Technology, 3(4): 177-184

Amahmid O., S. Asmama and K. Bouhoum (2002) Urban wastewater treatment in stabilization ponds: occurrence and removal of pathogens. Urban Water, 4(3): 255-262

Beltrán F.J., J.F. García-Araya and P.M. Alvarez (2000) Estimation of biological kinetic parameters from a continuous integrated ozonation activated sludge system treating domestic wastewater. Biotechnology Progress, 16(6): 1018-24.

Carta F., J. Pereda, P. Álvarez, F.Romero and M.M Durán (2004) Aerobic purification of dairy wastewater in continuous regime part II: kinetic study of the organic matter removal in two reactor configurations. Process Biochemistry, 22(2): 117–124.

Chalatsi M. (2014) WSP’s qualitative and hydrodynamics characteristic in N. Greece. PhD Thesis, Department of Civil Engineering, Democritus University of Thrace.

Chlot S., A. Widerlund, D. Siergieiev, F. Ecke, E. Husson and B. Öhlander (2011) Modelling nitrogen transformation in waters receiving mine effluents. Science of the Total Environment, 409(21): 4585-4595

Eaton A.D., American Public Health Association, American water Works Association and Water Environment Federation (2005) Standard Methods for the Examination of Water and Wastewater. 21st ed., Washington, D.C.: APHA-AWWA-WEF

Ferrara R.A. and C.M. Avci (1982) Nitrogen Dynamics in Waste Stabilization Ponds. J. Water Pollution Control Federation, 54(4): 361-369.

Heliotis F.D. and DeWitt C.B. (1983) A conceptual model of nutrient cycling in wetlands used for wastewater treatment: A literature analysis. Wetlands, 3, No 1, p.p.134-152.

Henze M., P. Harremoes, J.I.C. Jansen and E. Arvin (1997) Wastewater Treatment Biological and Chemical Processes. 2nd ed. Springer, Germany.

IWA (2000) Constructed Wetlands for Pollution Control: Processes, Performance, Design and Operation. Scientific and Technical Report No.8, IWA Publishing, London, UK.

Kadlec R.H. and Knight R.L. (1996) Treatment Wetlands. Lewis Publishers, CRC Press, New York. Kayombo S., T.S.A. Mbwette, J.H.Y Katima N. Ladegaard and S.E. Jørgensen (2005) Waste Stabilization Ponds and Constructed

Wetlands Design Manual. IWA Publishing. Kayombo S., T.S.A. Mbwette, J.H.Y Katima and S.E. Jørgensen (2003) Effects of substrate concentrations on the growth of

heterotrophic bacteria and algae in secondary facultative ponds. Water Research, 37(12): 2937–2943 Korkusuz E.A, M. Beklioglu and G.N. Demirer (2005), Comparison of the treatment performances of blast furnace slag-based and

gravel-based vertical flow wetlands operated identically for domestic wastewater treatment in Turkey. Ecological Engineering, 24(3):185–198

Khosravi R., M. Khodadadi, A. Gholizadeh, E.A. Mehrizi, T. Shahriary and A. Shahnia (2013a) BOD5 removal kinetics and wastewater flow pattern of stabilization pond system in Birjand. European Journal of Experimental Biology, 3(2): 430-436

Khosravi R., T. Shahryari, A. Halvani, M. Khodadadi, F. Ahrari and E. A. Mehrizi (2013b) Kinetic analysis of organic matter removal in stabilization pond in the wastewater treatment plant of Birjand. Advances in Environmental Biology, 7(6): 1182-1187

Krause P., D.P. Boyle, and F.Bäse (2005), Comparison of different efficiency criteria for hydrologic model assessment, Advances in Geosciences, 5: 89-97

Kuo J. (2014) Practical Design Calculations for Groundwater and Soil Remediation. Chapter 4: Mass-Balance Concept and Reactor Design, 2nd ed., Taylor & Francis Group.

Mara, D.D. and Pearson, H.W. (1987) Waste Stabilization Ponds: Design Manual for Mediterranean Europe. Copenhagen, Denmark: WHO Regional Office for Europe.

Moriasi, D. N., Arnold, J. G., Van Liew, M. W., Bingner,R. L., Harmel, R. D., Veith, T. L. (2007), Model Evaluation Guidelines for Systematic Quantification of Accuracy in Watershed Simulations. Transactions of the ASABE, 50(3): 885–900.

Water Utility Journal 14 (2016) 27

Penha-Lopes G., M.R. Flindt, B.Ommen, E. Kristensen, P.Garret and J.Paula (2012) Organic carbon dynamics in a constructed mangrove wastewater wetland populated with benthic fauna: A modelling approach. Ecological Modelling, 232: 97–108

Panikov N.S. (2000) Microbial Growth Kinetics. 2nd Ed., Chapman &Hall, London. Ronnie A. D. Frazer-Williams (2010) A review of the influence of design parameters on the performance of constructed wetlands.

Journal of Chemical Engineering, 25(1): 29-42 Senzia M.A., A.W. Mayo, T.S.A. Mbwette, J.H.K. Katima and S.E. Jørgensen (2002) Modelling nitrogen transformation and

removal in primary facultative ponds. Ecological Modelling, 154(3): 207-215 Shahryari T., A. Halvani, M. Khodadadi, F. Ahrari and E. A. Mehrizi (2013) Kinetic analysis of organic matter removal in

stabilization pond in the wastewater treatment plant of Birjand. Advances in Environmental Biology, 7(6): 1182-1187 Sun G. and S.Tanveer (2009) Kinetic modeling of organic matter removal in 80 horizontal flow reed beds for domestic sewage

treatment, Process Biochemistry, 44(7): 717-722 Thornthwaite C.W. (1948) An approach toward a rational classification of climate. Geographical Review, 38(1): 55–94. Tilley E., L. Urlich, C. Leuethi, P. Reymond and C. Zurbruegg (2014) Waste stabilisation ponds. Swiss Federal Institute of Aquatic

Science and Technology, SSWM (http://www.sswm.info/category/implementation-tools/wastewater-treatment/hardware/semi-centralised-wastewater-treatments/w)

Vavilin V., S.V. Rytov, L.V. Lokshira and G. Lyberatos (2001) Simplified hydrolysis models for the optimal design of two-stage anaerobic digestion. Water Research, 35(17): 4247–4251

Zwietering M.H., I. Jongenburger, F.M. Rombouts and K. Van't Riet (1990) Modeling of the bacterial growth curve. Applied and Environmental Microbiology, 56(6): 1875-1881