Improvement, validation, and implementation of predictive models on full scale anaerobic digestion assets
File(s)
Author(s)
Oxtoby, Steven John
Type
Thesis
Abstract
Anaerobic digestion (AD) is the most extensively used method of sewage sludge treatment in the UK, but there is a lack of practical tools available to aid operators in sewage sludge AD optimisation. This could be achieved through a modelling approach to optimise reactor loading rates, and conditions. However, current models of the AD process, such as Anaerobic Digestion Model 1 (ADM1) approach this problem by representing the fundamental biochemical and microbiology processes intrinsic to the process, which is too complex for practical application within an industrial context. A more practicable modelling system is required for industrial sewage sludge AD optimisation. Giacalone (2018) developed a sewage sludge AD model based on bench scale experimentation, to quantify the interactive effects of different major operational variables, and fundamental relationships representing first order degradation in a continuously stirred tank reactor with input variables including digester temperature, feed dry solids (DS), primary fraction, biomethane potential (BMP) and C, H and N fractions. By extensive sampling, and data cross checking, data sets were generated from 13 sewage treatment works (6 conventional and 7 advanced digestion sites), covering a wide range of digestion conditions. Key model outputs including biogas yield (BGY) and digestate ammonium-N (NH¬4-N) predictions were validated, and predictions of digestate DS were indicative of full scale performance. The ADFUN model can be adapted to represent other sludge type as predictions are based on fundamental biological and chemical sludge properties, therefore, the model was applied to estimate the BGY from AD of novel sludge types (granular and integrated fixed film sludges) in bench scale experiments with relative success. The model was also used to predict the biogas and financial benefits of digester optimisation varying scenarios. The model was effective at predicting the full-scale AD process, and applying optimised process conditions predicted by the model yields a benefit of up to 20 Nm3/tDS.
Version
Open Access
Date Issued
2022-09
Date Awarded
2023-03
Copyright Statement
Creative Commons Attribution NonCommercial Licence
License URL
Advisor
Smith, Stephen
Sponsor
Stream IDC
Thames Water (Firm)
Anglian Water
Severn Trent Plc
Publisher Department
Civil and Environmental Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Engineering Doctorate (EngD)