This study deals with the application of data reconciliation to wastewater treatment processes which are subject to dynamic conditions and therefore do not reach a steady-state behaviour sensu stricto. The SHARON partial nitritation process, which is operated cyclically with alternating aerated and anoxic periods, is studied as an example. The collected data long-term dynamic data set was split up into data subsets corresponding with different pseudo-steady-state operations, which allowed a better gross error detection. Mass balances were set up taking into account off-gas measurements besides liquid phase measurements and including kinetic relations between measurements based on the biological conversions in the reactor. As a result, a higher number of variables could be reconciled, more key variables could be identified, and gross error detection was facilitated. In order to draw conclusions on the process performance in a shorter period of operation, e.g., on the N₂O emission factor, the average value of the whole data set should be used with caution. The strong dependence of infiltrated air on the aeration regime and gross error in grab sampling (magnitude of 20%) had a substantial impact on calculating N₂O emission. It is recommended that the process performance indicators are derived and checked separately for steady state data subsets to guarantee reliable outcomes.

A stepwise experimental design procedure to obtain reliable data from wastewater treatment plants (WWTPs) was developed. The proposed procedure aims at determining sets of additional measurements (besides available ones) that guarantee the identifiability of key process variables, which means that their value can be calculated from other, measured variables, based on available constraints in the form of linear mass balances. Among all solutions, i.e. all possible sets of additional measurements allowing the identifiability of all key process variables, the optimal solutions were found taking into account two objectives, namely the accuracy of the identified key variables and the cost of additional measurements. The results of this multi-objective optimization problem were represented in a Pareto-optimal front. The presented procedure was applied to a full-scale WWTP. Detailed analysis of the relation between measurements allowed the determination of groups of overlapping mass balances. Adding measured variables could only serve in identifying key variables that appear in the same group of mass balances. Besides, the application of the experimental design procedure to these individual groups significantly reduced the computational effort in evaluating available measurements and planning additional monitoring campaigns. The proposed procedure is straightforward and can be applied to other WWTPs with or without prior data collection.