A Comparison of Wastewater Process Modelling Tools: Case Study of Potsdam WWTW

Govender, Neeren

A Comparison of Wastewater Process Modelling Tools: Case Study of Potsdam WWTW

Master Thesis

2022

Abstract

A wastewater process model can either be based on a steady-state or dynamic mathematical modelling approach. The characterisation of the influent wastewater and specific wastewater parameters are critical input parameters to the mathematical model, whether for design, optimisation or to measure the expected performance of a wastewater treatment works (WWTW). A wastewater treatment process is inherently dynamic because of the changes in wastewater characteristics and flow rates, impacting the plant's process capacity and performance. Mathematical models have the capability to model such changes with ease and predict the expected performance of the plant. This dissertation involves a theoretical modelling approach to compare steady-state and dynamic-state modelling tools on an existing full-scale WWTW, namely the Potsdam WWTW located in Cape Town (South Africa). The models adopted for this research included: i) A plant-wide steady-state model developed by the Author as part of this dissertation (Author's PWM); ii) Plant-Wide Steady-State Design (PWSSD) developed by Wu (2014); iii) BioWin 5.0 (developed by Envirosim Associates Ltd; Barker and Dold, 1997; Envirosim 2007); and iv) UCTPHO dynamic model (Wentzel et al., 1992). The goal was to compare the output and results of the various models adopted for this research under steady-state and dynamic-state conditions where the models have been calibrated on historical field data and to examine the impact that specific wastewater characteristics have on the process capacity and the overall performance of the WWTW. To accomplish this goal, the following objectives were achieved: 1. The characterisation of the influent wastewater is critical in modelling. For a new plant (green field site), the wastewater characterisation will be based on typical norms as published in the relevant research literature. For an existing plant, the wastewater characterisation is typically based on historical sampling data. There can be challenges with historical sampling data, as data is often not measured, missing or not credible. In this case, designers are required to make assumptions to fill the gaps in the wastewater characterisation, adopting typical norms as published in the research literature and using sound process engineering judgement which was the scenario for the case study WWTW. 2. The documented design capacity of the Potsdam WWTW is 47 Mℓ/day (47,000 kgCOD/day) whereas the total theoretical process capacity estimated as part of this dissertation is 60.75 Mℓ/day (64,660 kgCOD/day), 37.5% higher than the documented process capacity. 3. The various models correlated well in terms of output results for both steady-state and dynamic-state conditions when compared across all process units for both the liquid and sludge streams, using the same input process parameters (flow rates, load patterns, wastewater characterisation, fractions and design assumptions). The steady-state models (Author's PWM and PWSSD) were almost identical in output results except for the aerobic digester. The output results of the dynamic-state models (BioWin and UCTPHO) were similar to the steady-state models but did differ for a few variables, attributed to the fact that the dynamic models use dynamic kinetic equations under constant or dynamic flow and load conditions, and this will produce different results than the less complex steady-state models which are based on constant flow and load conditions. 4. An evaluation of the impact of selected influent characteristic parameters on the system performance variables for biological nutrient (nitrogen and phosphorus) removal was performed for each of the steady-state and dynamicstate models. The parameters selected included the influent TKN (TKN/COD ratio), the maximum specific growth rate of nitrifiers (µMax) and readily biodegradable COD (RBCOD), and during the analysis, all other input parameters to the models were kept constant. The influent TKN and RBCOD are specific wastewater characteristics that can vary during the lifetime of a WWTW having a strong impact on its N and P removal performance. µMax is also a critical wastewater parameter and at sludge ages close to the minimum sludge age for nitrification, this parameter impacts severely on the performance of biological nitrogen removal systems. It was concluded that substantial changes in the influent TKN, µMax and RBCOD will significantly impact a WWTW, specifically concerning N and P removal, therefore, impacting effluent and sludge quality. The various models followed similar trends; however, the following discrepancies were noted: ▪ When the TKN was increased, only the BioWin model considered the impact that the pH outside the range of 7.2 – 8.0 would have on the µMax and nitrification capacity of the bioreactor. This is a shortfall in the other models ▪ When the µMax was varied, the different models followed the same trend, but nitrification problems occurred at different µMax values. The BioWin model showed partial nitrification at a higher µMax threshold value than the other models, which occurred under minimum temperature conditions only ▪ When the RBCOD was varied, the only difference in the models is the RBCOD fraction at which complete biological phosphate removal took place. The steady-state models had the same RBCOD fraction, and the dynamic-state model (i.e., BioWin) had a higher RBCOD fraction at which complete biological phosphorous removal took place due to the fact that BioWin adopts two different maximum specific growth rates for the poly accumulating organisms (PAOs), namely a higher growth rate constant under phosphorous rich conditions which results in a higher uptake of phosphorous, and a lower growth rate constant under phosphorous limited conditions, where phosphorous uptake is limited. With a lower PAO growth in the BioWin model resulting in a lower PAO population, there is less potential for aerobic polyphosphate uptake, resulting in higher effluent Ortho-P concentrations. A WWTW is complex with many interactions of different processes (biological, chemical and physical) and products taking the form of various phases (aqueous, gas and solid), complicated further by variations in influent characteristics, concentrations and flows. To manage these complexities, wastewater process models have been developed over the last three decades from stand-alone models for individual process units to plant-wide computational steady-state and dynamic models, which cater for a broad spectrum of wastewater engineering objectives. Steady-state models are powerful as they comprise simple and explicit algebraic equations that easily allow the estimation of design requirements and operation requirements for a WWTW with much less input information than dynamic models. They are often pre-processers to the dynamic models. In contrast, dynamic models require detailed input formation and sophisticated mathematical solvers but are more accurate in predicting effluent quality, system responses to dynamic conditions and the inhibitory effects of pH, temperature and metabolic products. The decision of selecting a steady-state or dynamic-state model is influenced by several factors, such as available information and influent data, user competency and modelling experience, size and complexity of the plant, as well as the detail and accuracy required by the designer. Wastewater process models therefore serve as a valuable tool for design, optimisation and operational and control strategies.

Keywords

Water Quality Engineering

Reference:

Collections

Masters

Full item page