A mathematical model was developed to investigate hydraulic transport and bacterial selection during slow up-flow anaerobic feeding (0.9 m h-1) of wastewater across the settled bed of granular sludge biofilms used in column-type sequencing-batch reactors to remove nutrients. A plug-flow regime with dispersion was identified from residence time distribution data (RMSE < 0.010, R2 = 0.999). Metabolic formulations allowed assessing the effect of environmental conditions on the competition of polyphosphate- (PAO, Accumulibacter) and glycogen-accumulating organisms (GAO, Competibacter) for the uptake of acetate during anaerobic feeding. Since PAO and GAO metabolisms rely on distinct dynamics of intracellular storage polymers, the feeding phase length at nominal flowrate was shown to impact bacterial selection. In addition, acetate was preferentially consumed by PAO under alkaline conditions (pH 7.5-8.0) independently of temperature (10-30°C). GAO were only able to outcompete PAO under combined acidic (pH 6.0-6.5) and higher mesophilic (25-30°C) conditions. However, the difference in uptake rates was only half of the one obtained under conditions selecting for PAO. The model can support the assessment of spatial stratification of conversion processes across the bed and the design of operation and bed geometries towards optimal bacterial resource management in granular sludge.