N2O emissions and aeration efficiency in wastewater treatment: improved monitoring, mechanistic modelling and data mining

Due to the growing awareness of climate change, there is a need to quantify GHGs from different sources. The water industry, which provides the water supply, wastewater collection, and treatment and discharge, contributes significantly to total energy consumption and consequently to GHG emissions in developed countries. In WRRFs, large amounts of organic and inorganic matter are transformed and transferred from the water phase to the atmosphere, lithosphere, and/or biosphere through emissions from process tanks, and treated effluents and biosolids that are disposed in the environment. All three main GHGs (i.e. CO2, CH4, and N2O) are emitted from WRRFs. N2O is currently the GHG of major concern with regards to direct emissions from WRRFs. N2O has a GWP 265–298 times that of CO2 for a 100-year timescale which makes this the single most important ozone depleting compound of our century. Anthropogenic activity is responsible for about 40% of the global N2O production and a 15% concentration increase has been observed since 1750. To date 3% of the anthropogenic N2O production is recognized to be generated by wastewater treatment. WRRFs designed for nutrients removal have been observed to emit up to 7% of the influent nitrogen load as gaseous N2O. In a WRRFs the bioreactor is currently recognized as the most emitting treatment step in these terms. In addition to this, the aerated compartment of bioreactors is generally recognized to cause between 45 to 75 % of the plant’s energy expenditure. Considering both the contribution of N2O emissions and energy expenditure, the biological step of a WRRF represents the large majority of the CFP of a WRRF. Measuring and accounting for N2O emissions and aeration efficiency requires standard methods allowing to obtain comparable measurements among different WRRFs and reproducible within the same facility in order to derive solid classifications. However, especially for N2O, there is the need of a unified protocol with general standardized guidelines for a sound assessment at different WRRFs. Both N2O and aeration efficiency measurements protocols present major lacks and assumptions. This thesis puts in evidence major weaknesses of protocols for N2O emission and aeration efficiency measurements proposing possible improvements in terms of sampling strategy, calculation methods and equipment. As measurements of N2O emissions and aeration efficiency are used to understand process dynamics and design new CFP minimization scenarios, also modelling WRRFs is very important in this view, given the system complexity. Modelling tools allow to design new plant operation and control strategies (aimed at minimizing these emissions) and evaluate their long term effect on the WRRF limiting trials (and risks). Current N2O kinetic models are highly developed in describing the biochemical processes, however, as they are developed in lab-controlled conditions, they are yet troublesome when it comes to full-scale applications. This is most probably due to a poor representation of local concentrations by the plant’s model layout and often to an over-parametrization of the biokinetic model. The modelled description of the plant’s layout is nowadays often erroneously underestimated, but its design should be one of the most important steps in the definition of a plant’s model as it has important effects on the calculation effort, the calibration of the kinetic model, and nonetheless, on its predictive power. This thesis considers one of the most advanced kinetic models available in the literature and shows how, using a better representation of hydrodynamics, this can improve its performances. As effective applications, and applicability, of kinetic models for N2O prediction in full scale are still limited, possible modelling alternatives are evaluated in this work. The application of a qualitative, knowledge-based risk assessment model (N2O risk model) to a full-scale datasets is provided to prove the concept of its use. The N2O risk model shows to be effective in helping to unravel the dynamics behind N2O production and to be able to give valuable insight in the mechanisms of N2O production. In addition to this, seen the crescent quantity of data that current WRRFs have available, and the fact that the amount of information is too often unused wasting part of the value of sensors and SCADA systems. A data mining approach is also presented. In this regard, this thesis gives a practical application of a data mining technique to derive potential relations with respect to N2O emission among variables that are routinely measured at WRRFs. The testing of different clustering algorithms and their critical evaluation is shown in view of an online application. This is furnishing a possible new root to the use of SCADA data for understanding and mitigating N2O emissions by translating hidden information into clear operational instructions. In summary, this thesis raises the main concerns about N2O and aeration efficiency assessment analyzing major weaknesses and suggesting possible solutions for developing more robust standardized methods. It further provides an overview of different N2O modelling approaches proposing possible developments to enhance capabilities to recognize sources of emission and provide clues for developing CFP reduction strategies.