Computer simulation of activated sludge processes is a critical tool for design, operation, and troubleshooting, but simulating enhanced biological phosphorus removal (EBPR) systems has proven to be particularly challenging. This may be due in part to uncertainties in biokinetic models, but new research suggests it may also be due to deficiencies in conventional “lumped state” approaches to simulation, which model bulk concentrations of the polyphosphate accumulating organisms (PAOs) responsible for EBPR, as well as their microbial storage product (polyphosphate, glycogen, and polyhydroxyalkanoates) contents. A recently developed, alternative method for modeling biological treatment systems is the “distributed state” approach, which models individual bacteria as they move through a biological reactor system, rather than bulk concentrations. This approach predicts that PAO states (their microbial storage product contents) tend to diverge when reactors are completely mixed, and this can produce very different outcomes than those predicted by the conventional lumped approach. A MATLABbased distributed state program (DisSimulator 2.0) was applied to an A2O system and it was determined that distributed states tend to become more important with (1) shorter internal recycle ratios and (2) longer reactor hydraulic residence times. Consequently errors resulting from relying on lumped simulations are greatest in these conditions. This work illustrates that there appear to be interesting process phenomena related to changing state distributions that apparently cannot be accounted for by lumped simulations. These insights suggest that the continued advancement of the distributed simulator approach has the potential to improve design and operation of biological nutrient removal systems.
Computer simulation of activated sludge processes is a critical tool for design, operation, and troubleshooting, but simulating enhanced biological phosphorus removal (EBPR) systems has proven to be particularly challenging. This may be due in part to uncertainties in biokinetic models, but new research suggests it may also be due to deficiencies in conventional approaches to simulation, which infer unrealistic hydraulic characteristics. EBPR systems cycle bacteria through anaerobic and aerobic phases, and this selects for polyphosphate accumulating organisms (PAOs) containing several microbial storage products (polyphosphate, glycogen, and polyhydroxyalkanoates). It is critical that PAOs are maintained in each phase for sufficient time for the following processes to occur:
Volatile fatty acids (particularly acetic acid) are taken up and stored as PHAs (particularly poly-3-hydroxybutyrate, or PHB).
ATP for acetate uptake and PHB synthesis is supplied by degradation of intracellularly stored polyphosphate, leading to soluble P release from the cell.
Reducing equivalents for PHB synthesis are produced by degradation of stored glycogen and/or TCA cycle operation.
Stored PHAs are oxidized, producing energy and carbon substrates for metabolic reactions, including glycogen synthesis in some metabolic models.
Soluble P is taken up by EBPR organisms and stored intracellularly as polyphosphate.
Growth on stored PHAs occurs.
The conventional, “lumped” approach to simulating activated sludge systems assumes bulk biomass and microbial storage product concentrations as inputs to biokinetic models to calculate process rates (Gujer, 2002). This approach is used in the many currently available commercial software packages for wastewater treatment modeling.
However, completely mixed hydraulics lead to variable residence times in anaerobic and aerobic reactors, and under these conditions individuals within a PAO population are predicted to develop a distribution of microbial storage product contents (or “states”) (Schuler, 2005). These individuals exhibit a range of process rates that, when summed, can yield very different results than those obtained when assuming bulk values. Ignoring state distributions tends to result in overestimates of EBPR performance, and so could lead to undersized EBPR systems. It is therefore of interest for researchers and practioners to know under which conditions distributed states are most likely to be important.
The objectives of this work were to determine factors contributing to distributed state formation, the degree which these affect process performance, and to draw lessons for improved EBPR design. Two process characteristics were evaluated: hydraulic residence time (HRT), and internal recycle ratio. These were hypothesized to affect distributed predictions because they affect reactor hydraulic characteristics. This is the first presentation of distributed state analyses
to systems with internal recycle or nitrification/denitrification.