The progressive deterioration of the water resource available for different uses, including human consumption, has led the scientific community in recent years to investigate the occurrence, the fate and the hazards posed to both human health and the environment by human-made chemicals. These substances, created for industrial and pharmaceutical purposes constitute a serious threat to the environment also because the conventional technologies to treat wastewater had not been designed to remove this broad category of pollutants that occurs in water bodies at trace concentrations. This has led, in many countries, to the decision of adopting measures to tackle the problem of these substances' occurrence in waters, which are collectively labelled as “Contaminants of Emerging Concern” (CEC). On a technical level, this has led many wastewaters treatment plants to comply with the water quality safety standard via the implementation of advanced tertiary treatments resulting in a reclaimed wastewater resource having a high resulting quality level both for agricultural and human consumption standards and on the legislative level this has led many countries to identify a list of hazardous CECs and to set restrictive limits on the occurrence of these compounds in the reclaimed resource. The legislative apparatuses of many national and international entities are currently directed towards the definition of the acceptable levels of these substances and their almost complete removal or reduction to safely reuse the reclaimed resource both for agricultural and human consumption purposes. There are many advanced treatment technologies, which can be used to achieve this goal and, among them, a class of treatments, known as Advanced Oxidation Processes (AOPs) seem the most suitable for addressing CEC occurrence in wastewater treatment plants (WWTPs). The AOPs are treatments that use the synergic reactions of reagents and/or UV radiation to create highly reactive radical species that strongly oxidize many CEC, thus resulting in their mineralization or conversion into less harmful substances. The present work aims to accurately monitor the fate of CEC during AOPs using spectroscopic measures, particularly fluorescence. Fluorescence is a propriety exhibited by certain substances that have absorbed light or re-emitting a light with a longer wavelength. It can be useful for the determination of the good quality of the water matrix since it gives information about the content of fluorescing organic matter inside the water sample. WW samples exhibit the presence of characteristic peaks in the Excitation-Emission Matrix, five of which have been selected in the present work to be used to monitor CEC occurrence. The present work has investigated water samples originating in two WWTPs and one WW reuse facility (WWRF) that employs UV/hydrogen peroxide and ozone/hydrogen peroxide AOPs to achieve water reuse for agricultural purposes. Using techniques of statistical analysis, artificial neural networks and a real-time fluorescence sensor the present work has successfully predicted the fate of 8 benchmark CEC. The statistical approach has employed techniques such as PCA, OLS simple, stepwise, and multiple linear regressions, successfully predicting the fate of CEC. However, the results were affected by the water matrix effect and exhibited strong prediction accuracy only when analyzing waters with the same initial characteristics. To generalize the results on waters of different XII initial quality three linked artificial neural network (ANN) models were created. The first model successfully predicted CEC removal during AOPs using input data on a combination of five, three, and one fluorescence peaks. The most suitable indicator of CEC removal was identified in the I₄ peak (275/345). A second inverse model was then created that successfully predicted the fluorescence intensities of the five peaks associated with the benchmark CEC concentrations measured. A third model was created, which has successfully predicted the ozone, hydrogen peroxide and UV radiation to be dosed using as predictors as few as three peaks (I₃, I₄, I₅). This multiple ANN model can be reliably employed to assess CEC occurrence and monitor AOPs efficiencies, estimate the fluorescence intensities associated with the monitored CEC thresholds prescribed by the national regulations, and optimize AOPs by estimating the least required doses to achieve a certain level of fluorescence reduction. Finally, the present work has successfully tested a real-time I₄ peak fluorescence sensor, both in a lab environment and at a pilot scale during AOP treatments, and has highlighted its potential for continuous monitoring.
Il progressivo deterioramento delle risorse idriche disponibili per diversi usi, tra cui il consumo umano, ha portato negli ultimi anni la comunità scientifica a studiare la presenza, il destino e i pericoli posti alla salute umana e all'ambiente dalle sostanze chimiche prodotte dall'uomo. Queste sostanze, create per scopi industriali e farmaceutici, costituiscono una seria minaccia per l'ambiente anche perché le tecnologie convenzionali per il trattamento delle acque reflue non sono state progettate per rimuovere questa ampia categoria di inquinanti che si trovano nei corpi idrici a concentrazioni in tracce. Ciò ha portato, in molti Paesi, alla decisione di adottare misure per affrontare il problema della presenza di queste sostanze nelle acque, che sono collettivamente denominate "contaminanti di interesse emergente" (CEC). A livello tecnico, ciò ha portato molti impianti di trattamento delle acque reflue a conformarsi agli standard di sicurezza della qualità dell'acqua attraverso l'implementazione di trattamenti terziari avanzati, ottenendo una risorsa di acque reflue bonificate con un elevato livello di qualità sia per gli standard agricoli che per il consumo umano; a livello legislativo, ciò ha portato molti Paesi a identificare un elenco di CEC pericolosi e a fissare limiti restrittivi sulla presenza di questi composti nella risorsa bonificata. Gli apparati legislativi di molti enti nazionali e internazionali sono attualmente orientati alla definizione dei livelli accettabili di queste sostanze e alla loro quasi completa rimozione o riduzione per riutilizzare in sicurezza la risorsa bonificata sia per scopi agricoli che per il consumo umano. Esistono molte tecnologie di trattamento avanzate che possono essere utilizzate per raggiungere questo obiettivo e, tra queste, una classe di trattamenti, nota come Processi di Ossidazione Avanzata (AOP), sembra la più adatta per affrontare la presenza di CEC negli impianti di trattamento delle acque reflue (WWTP). Gli AOP sono trattamenti che utilizzano reazioni sinergiche di reagenti e/o radiazioni UV per creare specie radicaliche altamente reattive che ossidano fortemente molti CEC, determinandone la mineralizzazione o la conversione in sostanze meno dannose. Il presente lavoro mira a monitorare accuratamente il destino dei CEC durante le AOP utilizzando misure spettroscopiche, in particolare la fluorescenza. La fluorescenza è una proprietà esibita da alcune sostanze che hanno assorbito la luce o che riemettono una luce con una lunghezza d'onda maggiore. Può essere utile per determinare la buona qualità della matrice acqua, poiché fornisce informazioni sul contenuto di materia organica fluorescente all'interno del campione d'acqua. Può essere utile per determinare la buona qualità della matrice idrica, poiché fornisce informazioni sul contenuto di materia organica fluorescente all'interno del campione d'acqua. I campioni WW mostrano la presenza di picchi caratteristici nella matrice di eccitazione-emissione, cinque dei quali sono stati selezionati nel presente lavoro per essere utilizzati per monitorare la presenza di CEC. Il presente lavoro ha analizzato i campioni di acqua provenienti da due impianti di depurazione e da un impianto di riutilizzo di acque reflue (WWRF) che impiega le AOP UV/perossido di idrogeno e ozono/perossido di idrogeno per ottenere il riutilizzo dell'acqua per scopi agricoli. Utilizzando tecniche di analisi statistica, reti neurali artificiali e un sensore di fluorescenza in tempo reale, il presente lavoro ha previsto con successo il destino di 8 CEC di riferimento. L'approccio statistico ha impiegato tecniche come PCA, OLS semplice, stepwise e regressioni lineari multiple, prevedendo con successo il destino dei CEC. Tuttavia, i risultati sono stati influenzati dall'effetto matrice dell'acqua e hanno mostrato una forte accuratezza di previsione solo quando sono state analizzate acque con le stesse caratteristiche iniziali. Per generalizzare i risultati su acque di diversa qualità iniziale, sono stati creati tre modelli di rete neurale artificiale (RNA) collegati tra loro. Il primo modello ha previsto con successo la rimozione del CEC durante le AOP utilizzando dati di input su una combinazione di cinque, tre e un picco di fluorescenza. L'indicatore più adatto della rimozione di CEC è stato identificato nel picco I₄ (275/345). È stato quindi creato un secondo modello inverso che ha previsto con successo le intensità di fluorescenza dei cinque picchi associati alle concentrazioni di CEC di riferimento misurate. È stato creato un terzo modello che ha previsto con successo l'ozono, il perossido di idrogeno e le radiazioni UV da dosare utilizzando come predittori solo tre picchi (I₃, I₄, I₅). Questo modello ANN multiplo può essere impiegato in modo affidabile per valutare la presenza di CEC e monitorare l'efficienza delle AOP, stimare le intensità di fluorescenza associate alle soglie di CEC monitorate prescritte dalle normative nazionali e ottimizzare le AOP stimando le dosi minime necessarie per ottenere un certo livello di riduzione della fluorescenza. Infine, il presente lavoro ha testato con successo un sensore di fluorescenza di picco I₄ in tempo reale, sia in ambiente di laboratorio che su scala pilota durante i trattamenti AOP, evidenziandone il potenziale per il monitoraggio continuo.
Un approccio innovativo basato sulle reti neurali artificiali (ANN) e sulla spettroscopia di fluorescenza per monitorare il destino dei contaminanti emergenti (CEC) durante gli AOP nelle acque reflue / Todesco, Ruggero. - (2024 Jul 08).
Un approccio innovativo basato sulle reti neurali artificiali (ANN) e sulla spettroscopia di fluorescenza per monitorare il destino dei contaminanti emergenti (CEC) durante gli AOP nelle acque reflue
TODESCO, RUGGERO
2024-07-08
Abstract
The progressive deterioration of the water resource available for different uses, including human consumption, has led the scientific community in recent years to investigate the occurrence, the fate and the hazards posed to both human health and the environment by human-made chemicals. These substances, created for industrial and pharmaceutical purposes constitute a serious threat to the environment also because the conventional technologies to treat wastewater had not been designed to remove this broad category of pollutants that occurs in water bodies at trace concentrations. This has led, in many countries, to the decision of adopting measures to tackle the problem of these substances' occurrence in waters, which are collectively labelled as “Contaminants of Emerging Concern” (CEC). On a technical level, this has led many wastewaters treatment plants to comply with the water quality safety standard via the implementation of advanced tertiary treatments resulting in a reclaimed wastewater resource having a high resulting quality level both for agricultural and human consumption standards and on the legislative level this has led many countries to identify a list of hazardous CECs and to set restrictive limits on the occurrence of these compounds in the reclaimed resource. The legislative apparatuses of many national and international entities are currently directed towards the definition of the acceptable levels of these substances and their almost complete removal or reduction to safely reuse the reclaimed resource both for agricultural and human consumption purposes. There are many advanced treatment technologies, which can be used to achieve this goal and, among them, a class of treatments, known as Advanced Oxidation Processes (AOPs) seem the most suitable for addressing CEC occurrence in wastewater treatment plants (WWTPs). The AOPs are treatments that use the synergic reactions of reagents and/or UV radiation to create highly reactive radical species that strongly oxidize many CEC, thus resulting in their mineralization or conversion into less harmful substances. The present work aims to accurately monitor the fate of CEC during AOPs using spectroscopic measures, particularly fluorescence. Fluorescence is a propriety exhibited by certain substances that have absorbed light or re-emitting a light with a longer wavelength. It can be useful for the determination of the good quality of the water matrix since it gives information about the content of fluorescing organic matter inside the water sample. WW samples exhibit the presence of characteristic peaks in the Excitation-Emission Matrix, five of which have been selected in the present work to be used to monitor CEC occurrence. The present work has investigated water samples originating in two WWTPs and one WW reuse facility (WWRF) that employs UV/hydrogen peroxide and ozone/hydrogen peroxide AOPs to achieve water reuse for agricultural purposes. Using techniques of statistical analysis, artificial neural networks and a real-time fluorescence sensor the present work has successfully predicted the fate of 8 benchmark CEC. The statistical approach has employed techniques such as PCA, OLS simple, stepwise, and multiple linear regressions, successfully predicting the fate of CEC. However, the results were affected by the water matrix effect and exhibited strong prediction accuracy only when analyzing waters with the same initial characteristics. To generalize the results on waters of different XII initial quality three linked artificial neural network (ANN) models were created. The first model successfully predicted CEC removal during AOPs using input data on a combination of five, three, and one fluorescence peaks. The most suitable indicator of CEC removal was identified in the I₄ peak (275/345). A second inverse model was then created that successfully predicted the fluorescence intensities of the five peaks associated with the benchmark CEC concentrations measured. A third model was created, which has successfully predicted the ozone, hydrogen peroxide and UV radiation to be dosed using as predictors as few as three peaks (I₃, I₄, I₅). This multiple ANN model can be reliably employed to assess CEC occurrence and monitor AOPs efficiencies, estimate the fluorescence intensities associated with the monitored CEC thresholds prescribed by the national regulations, and optimize AOPs by estimating the least required doses to achieve a certain level of fluorescence reduction. Finally, the present work has successfully tested a real-time I₄ peak fluorescence sensor, both in a lab environment and at a pilot scale during AOP treatments, and has highlighted its potential for continuous monitoring.File | Dimensione | Formato | |
---|---|---|---|
PhD_Thesis_Todesco_finale.pdf
accesso aperto
Tipologia:
Tesi di dottorato
Licenza:
PUBBLICO - Pubblico con Copyright
Dimensione
4.64 MB
Formato
Adobe PDF
|
4.64 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.