Διερεύνηση της αποτελεσματικότητας ανοξικών δεξαμενών επιλογής για τον έλεγχο της νηματοειδούς διόγκωσης σε Εγκαταστάσεις Επξεργασίας Λυμάτων

Abstract

Στόχος της παρούσας μεταπτυχιακής εργασίας ήταν η διερεύνηση της αποτελεσματικότητας ανοξικών φρεατίων επιλογής στο πρόβλημα της νηματοειδούς διόγκωσης (ν.δ.) στο Κέντρο Επεξεργασία Λυμάτων της Ψυττάλειας (ΚΕΛΨ). Στην εν λόγω εγκατάσταση, το πρόβλημα της ν.δ. οφείλεται σε μεγάλο βαθμό στην παρουσία στη βιοκοινότητα της ενεργού ιλύος νηματοειδών βακτηρίων (ν.δ.) χαμηλής οργανικής φόρτισης (Organic Load –OL) με προτίμηση στο εύκολα βιοδιασπάσιμο COD (RBCOD). Αυτά τα ν.β. είναι τα N. limicola, Thiothrix spp. και Type 1851 . Για την εξέταση του εν λόγω φαινομένου λειτούργησαν δύο πιλοτικά συστήματα επί τόπου στο ΚΕΛΨ. Το ένα σύστημα (Control) αποτελούσε προσομοίωση της βιολογικής βαθμίδας σύμφωνα με το status quo (αναερόβια, ανοξική, αερόβια δεξαμενή και δεξαμενή τελικής καθίζησης -ΔΤΚ), ενώ το δεύτερο (Experimental) ήταν καθ’ όλα παρόμοιο, με μόνη διαφοροποίηση τη διαμερισματοποίηση σε τρία τμήματα της ανοξικής δεξαμενής ώστε (μαζί με την αναερόβια) να αποτελέσουν μία διάταξη ανοξικών φρεατίων επιλογής. Να σημειωθεί ότι, τόσο στο ΚΕΛΨ όσο και στα πιλοτικά συστήματα, η αναερόβια δεξαμενή στην ουσία λειτουργούσε ως ανοξική, λόγω της απουσίας δευτερεύουσας απονιτροποίησης στη γραμμή ανακυκλοφορίας της ιλύος. Η περίοδος των πειραμάτων μπορεί να χωριστεί σε δύο φάσεις λειτουργίας. Η Α’ Φάση αποτέλεσε μία δοκιμαστική περίοδο συνεχών αλλαγών. Η λειτουργία των πιλοτικών συστημάτων ήταν σχετικά ασταθής, αλλά προέκυψαν ορισμένες σημαντικές πρακτικές παρατηρήσεις. Συνοπτικά, θα πρέπει να δίδεται προσοχή στην ώρα δειγματοληψίας για τη λήψη αντιπροσωπευτικών δειγμάτων, να ελαχιστοποιείται το μήκος των σωληνώσεων ώστε να αποφεύγεται η ανάπτυξη βιοφίλμ εντός αυτών, να υπάρχουν διατάξεις εγκλωβισμού στερεών μετά την τελική εκροή, να είναι οι όγκοι των ΔΤΚ μεγαλύτεροι από τους υπολογιζόμενους μέσω των φορτίσεων ώστε να μπορούν να απορροφηθούν φαινόμενα συγκέντρωσης στερεών εντός αυτών, να συγκεντρώνεται το σύνολο της ημερήσιας εκροής από κάθε σύστημα για τον ορθό προσδιορισμό των στερεών στην τελική εκροή, ενώ θα πρέπει να ελέγχεται και η θερμοκρασία, ώστε να διατηρείται εντός ενός επιθυμητό εύρος στο ανάμικτο υγρό. Μετά την επίλυση των περισσοτέρων ανωτέρω προβλημάτων, έγινε επανεκκίνηση των πιλοτικών συστημάτων, τα οποία λειτούργησαν πλέον πολύ πιο ομαλά κατά τη Β’ Φάση. Σε επίπεδο γενικής απόδοσης των συστημάτων, η απομάκρυνση οργανικού άνθρακα και αζώτου ήταν ικανοποιητική, τα MLSS ήταν αντίστοιχα του ΚΕΛΨ, ενώ τα επίπεδα διαλυμένου οξυγόνου διατηρήθηκαν ανάμεσα σε 2 και 4 mg/L. Όσο αφορά τη λειτουργία του επιλογέα, στο Experimental σύστημα επετεύχθη κλιμάκωση της OL σε επίπεδα αντίστοιχα με τα βιβλιογραφικά αναφερόμενα ως επαρκή για την αντιμετώπιση των ν.β. χαμηλής OL. Αντίθετα, στο Control σύστημα, η OL ήταν μεν υψηλή στην αναερόβια δεξαμενή (κοινή διάταξη για τα δύο συστήματα), αλλά έφθινε απότομα στην ανοξική. Από το προφίλ του διαλυτού COD κατά μήκος των συστημάτων έγινε σαφές ότι (και στα δύο) η πλειονότητα του RBCOD καταναλωνόταν εντός της πρώτης αναερόβιας δεξαμενής. Το στοιχείο αυτό υποστηρίζεται και από τις μετρήσεις του ρυθμού απονιτροποίησης, ο οποίος και στα δύο συστήματα ήταν υψηλός στα αναερόβια διαμερίσματα, ενώ στη συνέχεια έφθινε (απότομα για το Control ενώ σταδιακά για το Experimental σύστημα). To SVI ήταν και στα δύο πιλοτικά συστήματα σε παρόμοια επίπεδα, αλλά σαφώς βελτιωμένο σε σχέση με το ΚΕΛΨ. Η βελτίωση αυτή υποδεικνύεται και από το δείκτη νηματοειδών, ο οποίος κυμαινόταν σε επίπεδα 4,5 για το ΚΕΛΨ ενώ αντίστοιχα 3,5 για τα πιλοτικά συστήματα. Τα ν.β. N. limicola και Type 1851 συνέχισαν να είναι κυρίαρχα στα πιλοτικά συστήματα αλλά με σαφώς μειωμένη παρουσία, ενώ το ν.β Thiothrix spp. πρακτικά δεν παρατηρήθηκε σε αυτά. Η παρόμοια συμπεριφορά των δύο συστημάτων οφείλεται πιθανότητα στο γεγονός ότι η αναερόβια δεξαμενή ήταν επαρκής για την επιβολή της απαραίτητης κινητικής πίεσης, ώστε να αρθεί το πλεονέκτημα των ν.β. έναντι των συσσωματούμενων. Η διαφοροποίηση με το ΚΕΛΨ φαίνεται να οφείλεται στο γεγονός ότι, ενώ στα πιλοτικά συστήματα η αναερόβια δεξαμενή ήταν μία απόλυτα ξεχωριστή κατασκευή (καμία πιθανότητα αξονικής μίξης), στο σύστημα πλήρους κλίμακας ο διαχωρισμός της δεν είναι εξίσου σαφής, με αποτέλεσμα να υπάρχει αξονική μίξη και να μην επιτυγχάνεται η απαραίτητη αρχικά υψηλή OL. Ακόμα και ο σαφής διαχωρισμός αυτού μόνο του σταδίου, πιθανόν να οδηγούσε σε βελτίωση της καθιζησιμότητας στο ΚΕΛΨ.

The present M.Sc thesis was conducted in the context of the Inter-Departmental Graduate Studies Program: “Water Resources Science and Technology” (Environmental and Water Quality Engineering specification), coordinated by the School of Civil and Environmental Engineering at the National Technical University of Athens. It aims to investigate the efficiency of the implementation of anoxic selector tanks in controlling filamentous bulking, a problem frequently occurring at several Waste Water Treatment Plants (WWTP). Specifically, the impact of introducing a combined anaerobic- anoxic selector at the WWTP of Psittalia was investigated through pilot scale experiments. The pilot systems were located on the island of Psittalia, where the majority of the experiments was also held. The research included the design and set up of two separate systems: • A Control system, which was a model of the secondary treatment unit of Psittalia according to the status quo (anaerobic, anoxic and aerobic reactors along with a secondary clarifier) • An Experimental system, almost identical to the afore described one, with the sole differentiation being the compartmentalization of the anoxic tank into three sectors

The existing implementation of anaerobic, anoxic and aerobic reactors indicates that metabolic selection already takes place at the WWTP at hand, applying a selection pressure that leads to the predominance of bacteria with specific metabolic characteristics in the activated sludge consortium. The compartmentalization of the presently unipartite anoxic tank aims to its transformation into a series of separate anoxic selector tanks, also introducing kinetic selection additionally to the metabolic one. The WWTP of Psittalia is no exception to the numerous plants facing the issue of activated sludge bulking. Due to the high Solids’ Retention Times applied (8-12 d), the problem at hand can be attributed to the overgrowth of low organic load filamentous bacteria. Specifically, the observed filaments at the WWTP of Psittalia can be divided into two main categories. • Filaments pertaining to filamentous bulking and foaming that are not expected to be affected by the implementation of anoxic selector tanks. Those are M. parvicella and Nocardia (Cordona). Those filaments seem to be able to grow on slowly biodegradable COD (though the latter is believed to be able to use RBCOD as well). Consequently, the use of kinetic selection is not expected to influence them. • Filaments pertaining to filamentous bulking that grow on RBCOD. N. Limicola, Types 1851 & 021N και Thiothrix spp. Belong in this category. Those are the bacteria that are targeted through the use of kinetic selection. It is worth mentioning at this point that Type 0092 is also a filamentous bacterium predominant in the activated sludge consortium of the WWTP of Psittalia. It is believed to grow on slowly biodegradable COD, but its presence is not of great importance to filamentous, since it grow mainly inside the floc’s body. One of the most commonly used measures to deal with the overgrowth of low organic load filaments is the adoption of selector tanks. A selector is defined as the entering part of a bioreactor and it is characterized by a low dispersion rate and a sufficient substrate gradient. It may be a small separate contact zone which constitutes the entrance of the incoming wastewater and the recirculated sludge. It must present a high RBCOD uptake rate, resulting almost in the complete removal of this component from the bulk liquid. The above logic was also the one followed during the design of the research at hand. The anaerobic reactor of the full scale treatment plant constituted the first part of the selector in the model. It must be noted at this point that, due to the absence of a secondary denitrification tank in the recirculated activated sludge line, the anaerobic tank actually functions as an anoxic one. The anoxic tank of the Experimental system was compartmentalized so that its first sector had the same volume as the anaerobic tank, the second one was two times that volume and finally, the third part had the remaining anoxic volume, so that the two systems (control and experimental) had the same overall anoxic volume. This way, the achievement of an originally high and gradually descending organic load was attempted, so that the floc formers could gain an advantage over the filamentous bacteria. The basic properties were common for the two pilot systems and they were set based on the operating principles of the full scale treatment plant. Specifically: • The flowrate entering each system (QIN) was set around 100 L/d • The activated sludge recirculation was originally set to 1 × QIN, but it was at times increased (up to 2 × QIN) in an attempt to solve the emerging problem of solids accumulation in the secondary clarifiers and hence the loss of a part of them with the final effluent • The internal recirculation was kept at all times around 2 × QIN (as in the full scale system) • The SRT was aimed to follow the SRTs of the full scale system (8-12 d)

To start with, both systems were inoculated with mixed liquor from the full scale plant aerobic reactors. Monitor and maintenance of the pilot systems included total and volatile suspended solids, temperature, dissolved oxygen, total and soluble COD, ammonia nitrogen and nitrate nitrogen measurements, along with microscopic evaluation. The operation of the systems can be divided into two separate periods: • Days 1-59: This period was actually a “trial” one. All the experimental analyses were conducted according to the existing program, but throughout the whole time there were alterations/ optimizations taking place in the systems’ configurations, according to the emerging needs and problems. The systems’ stability presented great variability during this period, resulting in the decision to “reboot” both of them (new biomass). • Days 60-108: After the implementation of all the corrective measures deriving from the first period’s experience, a new operational phase for the two systems began. The efficiency of both of the pilot systems and their overall operational stability were undoubtedly improved, rendering the occurring results much more reliable.

Despite the fact that the first period analyses’ results cannot be confidently used to derive conclusions considering filamentous bulking, some observations of special interest need to be noted. The large scale and the location of the pilot plants on site at the WWTP of Psittalia, place the research at hand among the quite original ones in the filed in Greece until today. Therefore, the derived practical experience may prove to be useful to future similar magnitude attempts. • Specific care should be given to sampling time (or composite samples should be used) to avoid non-representative of the overall daily mean measurements due to variability during the day • Tubing length leading to the systems’ entrance should be minimized, in order to avoid the formation of a biofilm in it. This can lead to the consumption of a non-negligible part of the COD in its way to the system • Temperature should be controlled in the area where the systems are located, to maintain mixed liquor temperature in a specific range • Solids’ captivation configurations should be adopted after the secondary clarifiers. This way, in case of excessive loss of solids with the effluent, part of it can be returned to the systems • It is advised to gather on a daily basis the entity of the final effluent from each system. This way, the TSS measurements will be accurate and representative of the daily mean and thus the SRT control will be much more efficient • The secondary clarifier’s volume is advisable to be much bigger than the one indicated by the solids’ load or the area load calculations, in order to avoid problems of excessive solids’ loss with the final effluent

All of the above matters had to be addressed during the first period. Their resolution rendered the systems’ operation from that point onwards much more stable. During the second operational period, the mean SRT was 10 d for the Control system and 11 d for the Experimental one, while the average MLSS concentrations were 3.500 mg/L and 4000 mg/L respectively. Regarding the Control system, the mean soluble COD at the final effluent was 39 mg/L, the NH4-N removal efficiency was 85% and the mean NO3-N concentration at the final effluent was 4,2 mg/L. The corresponding values for the Experimental system were 35 mg/L, 93% and 7,3 mg/L. The dissolved oxygen levels in the aeration basins were maintained for both systems between 2 and 4 mg O2/L, values high enough to avoid the overgrowth of low dissolved oxygen filaments. It is obvious that both of the systems efficiency was quite satisfactory. Regarding the comparison between the two pilot plants in terms of anoxic selector tanks operation, the experimental system presented an organic load gradient (6,32,81,30,8 kg COD/ kg MLSS/ d beginning from the anaerobic tank and ending at the third anoxic compartment). The respective values for the control system were 6,5 kg COD/ kg MLSS/ d in the anaerobic tank and 1,0 kg COD/ kg MLSS/ d in the anoxic one. It is obvious that the anoxic reactor compartmentalization imposed the desirable organic load gradient conditions, in levels corresponding to the ones mentioned in the existing literature as adequate (631,5 kg COD/ kg MLSS/ d for a three stage anoxic selector). Moreover, the mean soluble COD at the outlet of the second anoxic stage (where the selector effect theoretically wears out) was 54 mg/L. This value is lower than the threshold of 60 mg/L mentioned in the literature as the higher permitted soluble COD level at the exit point of a selector, in order to effectively combat the overgrowth of low organic load filaments. In both of the systems, the higher soluble COD levels were measured in the anaerobic tank (73 mg/L and 67 mg/L for the control and experimental systems respectively), while the soluble COD concentration along the rest of the systems did not reduce significantly (from 39 to 57 mg/L and from 35 to 56 mg/L for the control and Experimental systems respectively). This indicates that the readily biodegradable COD is almost entirely removed at the anaerobic basin at both systems. This observation is also bolstered by the calculated denitrification rates at both systems. Specifically, the denitrification rate for the Control system was calculated at 6,3 mg NO3-N/ g MLVSS/ h in the anaerobic and 2,3 mg NO3-N/ g MLVSS/ h in the anoxic tank. The corresponding calculations for the Experimental system were 7,34,62,41,3 mg NO3-N/ g MLVSS/ h, starting at the anaerobic reactor and ending at the third anoxic compartment. Theoretically, denitrification rates above 4 mg NO3-N/ g MLVSS/ h indicate the consumption of readily biodegradable COD, a range of 2-4 mg NO3-N/ g MLVSS/ h characterizes denitrification in the presence of slowly biodegradable COD while values below 2 mg NO3-N/ g MLVSS/ h correspond to endogenous denitrification. Taking into account this classification, it is concluded that the readily biodegradable COD is almost entirely consumed in the anaerobic tank at both systems. As far as the sludge volume index is concerned, its mean value was found to be 171 ml/ g SS for both systems! Though this measurement does not indicate a perfect settleability, it is undisputably far improved in comparison to the SVI measured at the full scale system during the same period (243 ml/ g SS). The two systems’ similar behavior leads to the conclusion that the first anaerobic tanks (common feature of both systems), suffices to impose the necessary kinetic pressure, in order to retract the filamentous bacteria low organic load advantage. The difference with the full scale system can probably be attributed to insufficient separation between the anaerobic and the anoxic tanks at the WWTP of Psittalia. In both the pilot systems, the anaerobic tank was a completely separate construction from the anoxic one, meaning that there was no possibility for axial mix to occur. On the contrary, at the full scale plant, the anaerobic and the anoxic tanks constitute parts of the same configuration. It seems that the compartmentalization of those two stages is not adequately clear, resulting to axial mix and therefore inability to maintain a high enough organic load at the entering part of the bioreactor. This leads to the distribution of the available substrate among many microorganisms (and not just the ones at the anaerobic tank, thus low F/M ratio) therefore vanishing the advantage floc formers have over filamentous bacteria at high organic loads. The microscopic observation results were in agreement with the SVI measurements. The average Filament Index was 3,5 for both the pilot systems, a value clearly improved compared to the one of the full scale treatment plant (4,5). From the Specific Filament Index measurements it is concluded that the filaments targeted through the use of anoxic selector tanks (N. limicola, Type 1851, Thiothrix spp.) were indeed reduced in comprison to the WWTP of Psittalia. The similar reduction in both of the pilot plants, however, again leads to the conclusion that a clearly separated first anoxic zone suffices to impose the necessary kinetic pressure.