Simulation of water resource recovery facilities with an open source software

Abstract

Digitalization is in an uprising trend for more than a decade on many aspects of wastewater treatment processes and these days we are coming across with the term more than ever. Simulation softwares are virtual platforms, a projection of a particular configuration created by the users that can process the data provided with the help of consistent mathematical model implementations. By doing this, environmental engineers are able to control and optimize the operational parameters and use if for finding the most cost-efficient treatment configuration while upgrading an existing facility process scheme or even before constructing it. In other words, engineers can prevent excessive construction and operational costs along with excessive energy consumptions. The motivations of this thesis study is to emphasize the need for popularizing creating functionable softwares with user friendly interfaces, creating specific softwares for divergent configurations and usage of modelling in academy as it is so benefitial for the students to familiarize with the fundamentals of modeling during their undergraduate lectures in terms of the convenience it provides for operational and kinetic parameters. An open-source software able to perform simulations of water resource recovery facilities with Modified Ludzack-Ettinger configuration has been developed within the scope of this study. Python programming language has been chosen for the development of the software due to its easy to learn syntax and its open-source libraries that contain powerful packages such as NumPy, SciPy, PySide2, Matplotlib and Pandas. The data handling of inputs and outputs have been achieved with the help of useful built-in functions of NumPy and Pandas, whereas the graphical user interface of the software have been created with PySide2. SciPy.integrate's solve_ivp function has been used for performing computations of ordinary differential equations with the backward differentiation formula (BDF) method which is a multi-step variable-order implicit method used in solving stiff problems. Lastly for the development phase, figure canvas class of Matplotlib package has been integrated to the interface for visualizing the results of performed simulations. A biochemical process model, consisting of 10 processes and 2 operational parameters defined for 15 state variables, have been created for the specific configuration that includes hydrolization processes of rapidly hydrolyzable COD, slowly hydrolyzable COD, soluble organic nitrogen and particulate organic nitrogen along with the growth and decay processes of heterotrophic and autotrophic biomasses. Activated Sludge Model No. 1 (ASM1) has been taken as a base model for the creation of software model meanwhile endogenous respiration process definitions for two different heterotrophic organism species were adopted from the Activated Sludge Model No. 3 (ASM3). Modifications have been made to the hybrid process model as the ammonification of soluble organic nitrogen process from Activated Sludge Model No. 1 and the storage mechanism of Activated Sludge Model No. 3 were removed from the process model in this thesis study. Once the process model was created, mass balance equations of each state variable were implemented in the software. Configuration reactors were considered as Continuously Stirred Tank Reactors (CSTR) and therefore were assumed as ideal reactors. The reactant concentrations were considered to be distributed homogenously through the reactors meaning that the reactant concentrations within the reactor are assumed to be equal to the effluent concentrations of the reactors. Rate of accumulation in the reactors were computed for each state variable for defining the mass balance equations of the specific configuration. Cofefficients and stoichiometric parameters defined on process model matrix were multiplied by the process rates of each component for calculating the rate of accumulation in the reactors. Operational processes like constant feed of dissolved oxygen and sludge disposal process for the particulate matter that are going to be wasted were included in the matrix. Computation of sludge disposal was achieved by a sludge retention time input parameter and correction factors for the process rates of denitrifiers were also included to kinetic parameters alongside the coefficients of heterotrophic and autotrophic growth and decay processes. Lastly, hydrolysis rates and coefficients were appended to the model. Calibration and validation of the process model have been achieved by using the data set of an existing WRRF. First 220 days of the data set of 363 days were used for the calibration and last 143 days were used for the validation of the parameter coefficients. Root Mean Square Error (RMSE) and Janus Coefficient methods have been selected for evaluating the precision of model simulation outputs. The most precise predictions in the calibration were achieved for the NH4-N and the NO3-N parameters with Root Mean Square Error values of 1,73 and 2,01, respectively while in the validation phase, the most precise predictions were achieved for the NH4-N and the TKN parameters with Root Mean Square Error values of 0,65 and 0,78, respectively. The least precise predictions were computed for the COD and pCOD parameters on both of the calibration and validation processes with Root Mean Square Error values of 14,41 and 14,14, respectively for the calibration and 5,82 and 7,93, respectively for the validation processes. The verification of the developed software was achieved by implementing the Modified-Ludzack Ettinger model in AQUASIM, an acknowledged simulation software used in environmental science, and comparing the results obtained from AQUASIM and the developed software created in this thesis study. Several simulations were done using the same operational parameters, kinetic and stoichiometric coefficients in each software while changing the parameters and coefficients each time a simulation was performed. Similarly, simulation outputs of each software were compared with simulations having different step sizes like 10-1, 10-2 and 10-3. On all of the simulations mentioned, it was seen that the outputs of the developed software matched the outputs of AQUASIM software. In conclusion, a useful tool to predict the performances of nitrogen removal process schemes for different water quality and treatment requirements was created in this thesis study. Considering a decent automation integration is achieved to the software, the developed software will increase the control of facility operators over the operation of the systems. The need for specific case studies on the modeled configuration will reduce with the efficient use of the software and younger generations of environmental engineers will be provided a better mean of comprehension for the operational, kinetic and stoichiometric parameters and their impacts on the processes.