This project investigates different models used and possibly usable for
modelling of resistance changes from sorption in MBR systems. The goal
was to create a model for prediction of time needed to reach
predetermined resistance values. Data was taken from a constant pressure submerged MBR setup,
logging was done every 30 seconds over one year. Data filtering was
based on standard deviation from a moving average. All models had only a
subset of the full data available for fitting, whose size increased one
by one after each fit.

The widely used first order exponential model failed to produce any
result. A modified version, the stretched exponential model had a better
outcome, but it was still prone to produce errors. It was found that
these models presume the existance of only one steady-state resistance
value while in practice that may be not a valid assumption. A more
sophisticated model was developed to overcome this difficulty, but lack
of necessary data type prevented its use. Finally, an empirical model
based on rate of resistance increase was able to predict cycle endtimes
with greater accuracy.