# Model-Based Construction of Wastewater Treatment Plant Influent Data for Simulation Studies

## Abstract

**:**

_{4}-N and NO

_{3}-N) are available. The proposed method is highly suitable for calculating an online estimate of the influent concentrations, which can be used as input information for digital twins, such as observer models and predictive controllers, based solely on the online measurement of the influent flow rate.

## 1. Introduction

#### 1.1. Motivation

_{4}-N, NO

_{3}-N), preferably online measurements, of the plant. This validation can be simplified for the quasi-stationary case with an estimation of the average inflow used as input data for the simulation and the comparison of average simulation results with the composite samples of the effluent of the real plant. This simple form of validation is not sufficient in the situation where not only an average cleaning performance is required for a plant, but also limit values are specified that must be complied with at all times, i.e., also during special events such as rain events. This applies, for example, to all plants in Germany. In this situation, the model validation must show that the dynamics of discharge peaks (NH

_{4}-N and NO

_{3}-N) during rain events can be reproduced with the model. This proof is only possible with a dynamic simulation with the specification of dynamic and high-resolution (<2 h sampling time) inflow data (concentrations and volume flow). An established option in this situation is to carry out an intensive measurement campaign in which, for example, 2 h composite samples are sampled and analysed in the inlet over a longer period (>14 d). With these data and the continuously measured water flow rate, the suitable inflow data are then available for model validation. However, this procedure is very cost extensive and therefore only makes sense for larger plants (cost-effective) (refer Borzooei et al. 2019 [4]). In addition, only a relatively short period of time can be analysed, which may not always be suitable for model validation (relevant rainfall events).

_{4}-N and NO

_{3}-N. This method has been used successfully in a large number of simulation studies. However, difficulties have also arisen in some cases:

- In certain constellations of the shape parameters, the time series of the “urine” component generates negative values.
- In certain constellations of the form parameters, the time series of the “urine” component produces a second maximum (evening peak) that is greater than the morning peak, and this does not correspond to human activity and is an artefact.
- In some systems, the increase in the water flow rate in the morning is significantly steeper than can be represented by a second-order Fourier series (see similar findings in Rodríguez et al., 2013 [8]).
- For stormwater runoff, it is assumed too simplistically that no pollution loads from rainwater runoff are introduced.
- The form factors (t
_{min}, f_{Q,min}, t_{max}, f_{Q,max}) are related to the total dry weather inflow; for a system size-based specification, a reference to the fraction of wastewater only makes more sense. - The proportion of infiltration water is often not constant over the period under consideration in a simulation study.

^{#}(5.0).

#### 1.2. State of the Art

#### 1.3. Typical Data Available Based on Routine Measurements

## 2. Method

#### 2.1. Modelling Dry Weather Inflow Pattern

#### 2.1.1. Model Approach for Dry Weather Inflow Pattern according to HSG

_{min}) and the relative size (f

_{Q,min}) of the nighttime minimum and the time (t

_{max}) and the relative size (f

_{Q,max}) of the daytime maximum and the parameters of the Fourier series.

#### 2.1.2. Model Approach for Dry Weather Inflow Patter, Version 2023

^{3}, $TK{N}_{inf}=0$ gN/m

^{3}, ${P}_{inf}=0$ gP/m

^{3}). With the simplifying and arbitrary assumption that the $COD$ concentration in urine and grey water are equal ($CO{D}_{U}=CO{D}_{g}$) and the known $COD$ load (${L}_{COD}$) and the $COD$ concentration in the grey water can be calculated as follows (Equation (10)).

#### 2.2. Implementation of the Method in the SIMBA^{#} Simulation System

^{#}simulation system (ifak 2023 [19]). Figure 3 shows the first part of the parameter dialogue, with which the parameters ${L}_{COD}$, ${L}_{TKN}$, ${L}_{P}$, ${Q}_{m}$ and ${Q}_{inf}$ are specified. The time delay of the concentration curves is implemented in a separate block (plug flow). This division makes it possible to map rain events which then cause load peaks in the inlet of the wastewater treatment plant.

#### 2.3. Calculation of a Continuous Inflow for Medium-Long Period

#### 2.4. Implementation of the Method for Long-Term Dynamic Data Generation in the SIMBA^{#} Simulation System

^{#}simulation tool. The internal structure of this block is shown in Figure 8.

_{0}..a

_{3}, b

_{1}..b

_{3}), and the proportion of infiltration water is determined. Size-dependent (person equivalents—PEs) standard values are calculated for the parameters $d{T}_{N}$ and $\beta $.

## 3. Results

#### 3.1. Verification of the Modelling Approach for Dry Weather Inflow Pattern

_{0}, a

_{1}, a

_{2}, b

_{1}and b

_{2}.

^{3}) in the urine. The time function of the urine flow rate is characterised by a constant proportion (${f}_{U,min}{Q}_{U,M}$) and a nitrogen peak in the morning $(1-{f}_{U,min}){Q}_{U,M}$. This nitrogen peak is determined by the two shape parameters ${T}_{N,max}$ (time) and $\beta $ (the shape parameter of the Gumbel function and the width of the peak).

^{3}for raw wastewater and approximately 600 gCOD/m

^{3}for pre-treated water (data points labelled in red) are confirmed.

#### 3.2. Experiences with the Method to Generate Long-Term Dynamic Simulation Influent Data

_{4}-N, online PO

_{4}-P analyzer, daily 24 h composite samples with $COD$, $TKN$ and PO

_{4}-P) were available. In this case, the influent data synthesized with the method could be compared with measured values. Figure 17 compares the ammonium loads calculated from the measured data with the data synthesized using the method.

## 4. Summary and Outlook

_{4}-N effluent peaks occurring at the real plants very well. This is indispensable for a capacity estimate, especially against the background of the usual monitoring practice in Germany (peak values).

_{4}-N, SAK, COD via spectrometer) in the influent of the wastewater treatment plant. In this case as well, advanced methods for inflow data generation are possible (Alex et al. 2009 [15]).

## Funding

## Data Availability Statement

## Conflicts of Interest

## Appendix A. Parameter Estimation of Fourier-Series Parameters

## Appendix B. Relationship between Shape Parameters and Fourier Coefficients (Second Order)

## Appendix C. 8.3 Model Fit for the Sample Diurnal Pattern (Complete)

## References

- Ruiz, L.M.; Pérez, J.I.; Gómez, M.A. Practical review of modelling and simulation applications at full-scale wastewater treatment plants. J. Water Process Eng.
**2023**, 56, 104477. [Google Scholar] [CrossRef] - Hvala, N.; Vrečko, D.; Levstek, M.; Bordon, C. The use of Dynamic Mathematical Models for improving the Designs of upgraded Wastewater Treatment Plants. J. Sustain. Dev. Energy Water Environ. Syst.
**2017**, 5, 15–31. [Google Scholar] [CrossRef] - Gernaey, K.; van Loosdrecht, M.C.M.; Henze, M.; Lind, M.; Jørgensen, S.B. Activated sludge wastewater treatment plant modelling and simulation: State of the art. Environ. Model. Softw.
**2004**, 19, 763–783. [Google Scholar] [CrossRef] - Borzooei, S.; Amerlinck, Y.; Abolfathi, S.; Panepinto, D.; Nopens, I.; Lorenzi, E.; Meucci, L.; Zanetti, M.C. Data scarcity in modelling and simulation of a large-scale WWTP: Stop sign or a challenge. J. Water Process Eng.
**2019**, 28, 10–20. [Google Scholar] [CrossRef] - Langergraber, G.; Alex, J.; Weissenbacher, N.; Woerner, D.; Ahnert, M.; Frehmann, T.; Halft, N.; Hobus, I.; Plattes, M.; Spering, V.; et al. Generation of diurnal variation for influent data for dynamic simulation. Water Sci. Technol.
**2008**, 57, 1483–1486. [Google Scholar] [CrossRef] [PubMed] - Ahnert, M.; Alex, J.; Dürrenmatt, D.J.; Langergraber, G.; Hobus, I.; Schmuck, S.; Spering, V. Dynamische Simulation als Bestandteil einer Kläranlagenbemessung nach DWA-A 131. KA-Korresp. Abwasser Abfall
**2015**, 62, 615–624. [Google Scholar] - Alex, J.; Dürrenmatt, D.J.; Langergraber, G.; Hobus, I.; Spering, V. Voraussetzungen für eine dynamische Simulation als Bestandteil einer Kläranlagenbemessung nach DWA-A 131. KA-Korresp. Abwasser Abfall
**2015**, 62, 436–446. [Google Scholar] - Juan Pablo Rodríguez, J.P.; McIntyre, N.; Díaz-Granados, M.; Achleitner, S.; Hochedlinger, M.; Maksimovi, C. Generating time-series of dry weather loads to sewers. Environ. Model. Softw.
**2013**, 43, 133–143. [Google Scholar] [CrossRef] - Flores-Alsina, X.; Ort, C.; Martin, C.; Benedetti, L.; Belia Snip, L.; Saagi, R.; Talebizadeh, M.; Vanrolleghem, P.A.; Jeppsson, U.; Gernaey, K.V. Generation of (Synthetic) Influent Data for Performing Wastewater Treatment Modelling Studies. In Proceedings of the 4th IWA/WEF Wastewater Treatment Modelling Seminar, Spa, Belgium, 29 March–2 April 2014; Available online: http://www.biomath.ugent.be/WWTmod2014/ (accessed on 1 May 2023).
- Martin, C.; Vanrolleghem, P.A. Analysing, completing, and generating influent data for WWTP modelling: A critical review. Environ. Model. Softw.
**2014**, 60, 188–201. [Google Scholar] [CrossRef] - Wu, X.; Zheng, Z.; Wang, L.; Li, X.; Yang, X.; He, J. Coupling process-based modeling with machine learning for long-term simulation of wastewater treatment plant operations. J. Environ. Manag.
**2023**, 341, 118116. [Google Scholar] [CrossRef] [PubMed] - Ahnert, M.; Oppermann, J.; Hurzlmeier, S.; Barth, M.; Gerard, I.; Abel, T.; Bernatzky, C.; Marx, C.; Kühn, V. Das Forschungsprojekt Zeiteffiziente Analyse von Kläranlagen (ZAK)—Von der Idee zum Produkt. In Abwasser Abfall; DWA: Hennef, Germany, 2014; pp. 124–130. [Google Scholar]
- Arbeitsblatt, A.D.A. 131—Bemessung von Einstufigen Belebungsanlagen; DWA: Hennef, Germany, 2016. [Google Scholar]
- Hauduc, H.; Gillot, S.; Rieger, L.; Ohtsuki, T.; Shaw, A.; Takacs, I.; Winkler, S. Activated sludge modelling in practice: An international survey. Water Sci. Technol.
**2009**, 60, 1943–1951. [Google Scholar] [CrossRef] [PubMed] - Alex, M.; Hetschel, M. Ogurek: Simulation Study with Minimized Additional Data Requirements to Analyse Control and Operation of WWTP Dorsten, Water Science & Technology; IWA Publishing: London, UK, 2009. [Google Scholar]
- Mannina, G.; Cosenza, A.; Vanrolleghem, P.A.; Viviani, G. A practical protocol for calibration of nutrient removal wastewater treatment models. J. Hydroinform.
**2011**, 13, 575–595. [Google Scholar] [CrossRef] - Almeida, M.C.; Butler, D.; Friedler, E. At-source domestic wastewater quality. Urban Water
**1999**, 1, 49–55. [Google Scholar] [CrossRef] - Flores-Alsina, X.; Saagi, R.; Lindblom, E.; Thirsing, C.; Thornberg, D.; Gernaey, K.V.; Jeppsson, U. Calibration and validation of a phenomenological influent pollutant disturbance scenario generator using full-scale data. Wat. Res.
**2013**, 51, 172–185. [Google Scholar] [CrossRef] [PubMed] - SIMBA
^{#}. SIMBA^{#}Water 5.0 User Manual; ifak e.V. Magdeburg: Magdeburg, Germany; Available online: https://www.ifak.eu/de/produkte/simba (accessed on 1 May 2023). - Gernaey, K.V.; Flores-Alsina, X.; Rosen, C.; Benedetti, L.; Jeppsson, U. Dynamic influent pollutant disturbance scenario generation using a phenomenological modelling approach. Environ. Model. Softw.
**2011**, 26, 1255–1267. [Google Scholar] [CrossRef] - Arbeitsblatt, A.D.A. 198—Vereinheitlichung und Herleitung von Bemessungswerten für Abwasseranlagen; DWA: Hennef, Germany, 2003. [Google Scholar]
- Rojas, K.; Galvis, A.; Schütze, M. Modelling of sediments in the drainage system of Cali/Colombia. In Proceedings of the 14th International Conference on Urban Drainage, Prague, Czech Republic, 10–15 September 2017. [Google Scholar]
- Ahnert, M.; Blumensaat, F.; Langergraber, G.; Alex, J.; Woerner, D.; Frehmann, T.; Halft, N.; Hobus, I.; Plattes, M.; Spering, V.; et al. Goodness-of-fit measures for numerical modelling in urban water management—A review to support practical applications. In Proceedings of the LWWTP07 Conference, Vienna, Austria, 9–13 September 2007; pp. 69–72. [Google Scholar]

**Figure 1.**Modelling approach for dry weather inflow Langergraber et al. 2008 [5].

**Figure 2.**Fourier approximation of the dry weather inflow Langergraber et al. 2008 [5].

**Figure 7.**Inlet divided into additional infiltration water, infiltration water, urine, grey water, and rainwater runoff.

**Figure 11.**Reproduction of daily pattern, variant 2023 (selection). The legends for measured Q and measured COD include the id (1–21) of the analysed sample plant. See Appendix C for all analysed data sets.

**Figure 12.**Shape parameter wastewater flow rate as a function of plant size (PE). (A linear regression line is plotted as grey dotted line which appears as solid grey line in the plots).

**Figure 13.**Shape parameter COD pattern. (A linear regression line is plotted as grey dotted line which appears as solid grey line in the plots).

**Figure 14.**Resulting $COD$ concentrations in the wastewater. (A linear regression line is plotted as grey dotted line which appears as solid grey line in the plots).

**Figure 15.**Shape parameter for TKN pattern. (A linear regression line is plotted as grey dotted line which appears as solid grey line in the plots).

**Figure 16.**Shape parameter P pattern. (A linear regression line is plotted as grey dotted line which appears as solid grey line in the plots).

Location | Data |
---|---|

Influent | Water quantity in the inlet of the plant (or in the outlet of the plant) as continuous online measurement (e.g., 15 min values) |

Influent | 24 h mixed samples (preferably flow-proportional mixing), laboratory analysis of COD, TKN/TN, P concentrations, measured values on different days but not for every day, sampling in the raw inlet, alternatively in the pre-treatment outlet, random samples for the ratio COD particulate/COD total, ISS (inert suspended solids) |

Internal flows | Internal flows (recirculation, return sludge) as continuous online measurement (e.g., 15 min values), alternatively the estimated flow rates from pump control signals (on, off, frequency) |

Aeration state | Typically, oxygen concentrations in the aeration tanks as continuous online measurement (e.g., 15 min values) |

Sludge produced | Excess sludge flow rate as continuous online measurement (e.g., 15 min values), alternatively the estimated flow rates from pump control (on, off, frequency). Helpful: TSS concentration measurements in the return sludge or in the excess sludge, continuously. Random samples: TSS (total suspended solids), ISS |

Activated sludge tanks | Available online measurements for NO_{3}-N, NH_{4}-N, PO_{4}-P and TSS as continuous online measurements (e.g., 15 min values),Grab samples of NO _{3}-N, NH_{4}-N, PO_{4}-P, TSS and temperature |

Effluent | Available online measurements for NO_{3}-N, NH_{4}-N, PO_{4}-P and TS as continuous online measurements (e.g., 15 min values)24 h composite samples of COD, NO _{3}-N, NH_{4}-N, P and TSS |

Location | Data |
---|---|

Primary sludge | Primary sludge flow rate; grab samples: TSS and ISS |

Anaerobic digestion | Amount of digested sludge, TSS, VSS (volatile suspended solids), ISS, NH_{4}-N, biogas flow rate and CH_{4} content |

Partial Flow | COD g COD/m ^{3} | TKN g N/m ^{3} | P g P/m ^{3} | Flow Rate m ^{3}/d | Average Flow Rate |
---|---|---|---|---|---|

Infiltration water (constant) | $CO{D}_{inf}$ | $TK{N}_{inf}$ | ${P}_{inf}$ | ${Q}_{inf}$ | ${Q}_{inf}$ |

Infiltration water time variable | $CO{D}_{inf}$ | $TK{N}_{inf}$ | ${P}_{inf}$ | ${q}_{inf,extra}\left(t\right)$ | |

Rainfall runoff | $CO{D}_{rain}$ | $TK{N}_{rain}$ | ${P}_{rain}$ | ${q}_{rain}\left(t\right)$ | |

Grey water | $CO{D}_{g}$ | $TK{N}_{g}$ | ${P}_{g}$ | ${q}_{g}\left(t\right)$ | ${Q}_{g,m}$ |

Urine | $CO{D}_{u}$ | $TK{N}_{u}$ | ${P}_{u}$ | ${q}_{u}\left(t\right)$ | ${Q}_{u,m}$ |

Wastewater (grey water + urine) | $CO{D}_{w}$ | $TK{N}_{w}$ | ${q}_{w}\left(t\right)$ | ${Q}_{w,m}$ | |

Total dry weather inflow | $CO{D}_{dw}\left(t\right)$ | $TK{N}_{dw}\left(t\right)$ | ${P}_{dw}\left(t\right)$ | ${q}_{dw}\left(t\right)$ | ${Q}_{m}$ |

Total inflow | $COD\left(t\right)$ | $TKN\left(t\right)$ | $P\left(t\right)$ | $q\left(t\right)$ |

Variable | Description | Unit |
---|---|---|

a_{0}..a_{3}, b_{1}..b_{3} | Coefficients of Fourier series for total dry weather flow | m^{3}/d |

${f}_{u,min}$ | Constant fraction of urine flow | - |

${T}_{N,max}$ | Time of urine peak | d |

$\beta $ | Form parameter of Gumbel function, width of peak | - |

$d{T}_{N}$ | Time shift of maximum flow to maximum of total flow rate | d |

${L}_{COD}$ | Average COD load per day for dry weather | g COD/d |

${L}_{TKN}$ | Average TKN load per day for dry weather | g N/d |

${L}_{P}$ | Average P load per day for dry weather | g COD/d |

$\omega $ | Unit frequency | 1/d |

$T$ | Period length = 1 d | 1 d |

${f}_{TKN,COD,g}$ | TKN/COD ratio of grey water | gN/gCOD |

${f}_{P,COD,g}$ | P/COD ratio of grey water | gP/gCOD |

${f}_{Q,min}$ | Ratio of minimum to mean wastewater flow | - |

${Q}_{min}$ | Minimum dry weather flow rate | m^{3}/d |

${T}_{min}$ | Time of minimum dry weather flow rate | d |

${f}_{Q,max}$ | Ratio of maximum to mean wastewater flow | - |

${Q}_{max}$ | Maximum dry weather flow rate | m^{3}/d |

${T}_{max}$ | Time of maximum dry weather flow rate | d |

${t}_{plug}$ | Time constant of plug flow element | d |

${V}_{plug}$ | $\mathrm{Volume}\mathrm{of}\mathrm{plug}\mathrm{flow}\mathrm{element},{V}_{plug}={t}_{plug}{Q}_{m}$ | m^{3} |

${V}_{i}$ | Volume of one element in the cascade of CSTRs to model the plug-flow element | m^{3} |

${T}_{0}$ | Simulation time step for plug-flow model | d |

Abbreviations | ||

PE | Person equivalent, virtual number of persons connected to a WWTP representing the observed load | |

WWTP | Wastewater treatment plant | |

$COD$ | COD—Chemical Oxygen Demand | |

$TKN$ | Total Kjehldahl Nitrogen | |

$P$ | Phosphorus | |

HSG | The university simulation group HSG (hsgsom.org) |

Concentration | Raw Wastewater | Pre-Settled Wastewater | Unit |
---|---|---|---|

$CO{D}_{w}$ | $\frac{120\frac{g}{ped}}{{q}_{pe}}$ | $\frac{78\frac{g}{ped}}{{q}_{pe}}$ | g COD/m^{3} |

$TK{N}_{w}$ | $\frac{12\frac{g}{ped}}{{q}_{pe}}$ | $\frac{10.8\frac{g}{ped}}{{q}_{pe}}$ | g N/m^{3} |

Variable | Description | Unit |
---|---|---|

${q}_{extra}\left(t\right)$ | Sum of additional infiltration water and rain runoff water | m^{3}/d |

${q}_{rain}\left(t\right)$ | Rain runoff water | m^{3}/d |

${q}_{inf,extra}\left(t\right)$ | Additional infiltration water compared to analysed dry weather period, seasonal effect or rain event caused | m^{3}/d |

${\widehat{q}}_{inf,extra}\left(t\right)$ | $\mathrm{Estimated}\mathrm{value}\mathrm{for}{q}_{inf,extra}\left(t\right)$ | m^{3}/d |

${T}_{inf,up}$ | $\mathrm{Time}\mathrm{constant}\mathrm{estimation}\mathrm{filter}\mathrm{for}{q}_{extra}\left(t\right){\widehat{q}}_{inf,extra}\left(t\right)$ | d |

${T}_{inf,down}$ | $\mathrm{Time}\mathrm{constant}\mathrm{estimation}\mathrm{filter}\mathrm{for}{q}_{extra}\left(t\right){\widehat{q}}_{inf,extra}\left(t\right)$ | d |

${Q}_{inf,COD}$ | Estimated infiltration flow rate based on COD concentration for wastewater | m^{3}/d |

${Q}_{inf,TKN}$ | Estimated infiltration flow rate based on TKN concentration for wastewater | m^{3}/d |

${Q}_{inf,Qmin}$ | Estimated infiltration flow rate based on night minimum | m^{3}/d |

${f}_{Qinf,Qmin}$ | Factor to specify infiltration flow rate based on night minimum | - |

Variable | Description | Unit |
---|---|---|

${t}_{plug}$ | $\mathrm{Case}\mathrm{specific}\mathrm{setting},\mathrm{default}\mathrm{value}{t}_{plug}=0.1$ d, Figure 6 can serve as orientation, larger values for pre-settled influent data | d |

$d{T}_{N}$ | $d{T}_{N}=0.1939-0.0219{\mathrm{log}}_{10}\left(PE\right)$, see also result section, Figure 15 | d |

${T}_{N,max}$ | $\mathrm{Calculated}\mathrm{from}\mathrm{the}\mathrm{time}\mathrm{of}\mathrm{flow}\mathrm{maximum}\mathrm{and}\mathrm{estimated}\mathrm{time}\mathrm{shift},d{T}_{N}$$:{T}_{N,max}={T}_{max}-d{T}_{N}$ | d |

${f}_{u,min}$ | $\mathrm{default}\mathrm{value},{f}_{u,min}=0.55$, approximately mean value from Figure 8 | - |

$\beta $ | $\beta =-0.0292+0.0273{\mathrm{log}}_{10}\left(PE\right)$, see also result section Figure 15 | - |

${f}_{TKN,COD,g}$ | ${f}_{TKN,COD,g}=0.06$ gTKN/gCOD, a larger industrial wastewater fraction could require adaptation | gTKN/gCOD |

${f}_{P,COD,g}$ | ${f}_{P,COD,g}=0.008$ gP/gCOD, see also result section Figure 16 | gP/gCOD |

${\mathrm{f}}_{Qu,Qw}$ | ${\mathrm{f}}_{Qu,Qw}=\raisebox{1ex}{${Q}_{u,m}$}\!\left/ \!\raisebox{-1ex}{${Q}_{w,m}$}\right.=0.1$ | - |

$TK{N}_{u}$ | $TK{N}_{u}=400$ gN/m^{3} | gN/m^{3} |

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |

© 2024 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Alex, J.
Model-Based Construction of Wastewater Treatment Plant Influent Data for Simulation Studies. *Water* **2024**, *16*, 564.
https://doi.org/10.3390/w16040564

**AMA Style**

Alex J.
Model-Based Construction of Wastewater Treatment Plant Influent Data for Simulation Studies. *Water*. 2024; 16(4):564.
https://doi.org/10.3390/w16040564

**Chicago/Turabian Style**

Alex, Jens.
2024. "Model-Based Construction of Wastewater Treatment Plant Influent Data for Simulation Studies" *Water* 16, no. 4: 564.
https://doi.org/10.3390/w16040564