One of the common quality parameters for drinking water is residual aluminium. High doses of residual aluminium in drinking water or water used in the food industry have been proved to be at least a minor health risk or even to increase the risk of more serious health effects, and cause economic losses to the water treatment plant. In this study, the trend index is developed from scaled measurement data to detect a warning of changes in residual aluminium level in drinking water. The scaling is based on monotonously increasing, non-linear functions, which are generated with generalized norms and moments. Triangular episodes are classified with the trend index and its derivative. The severity of the situations is evaluated by deviation indices. The trend episodes and the deviation indices provide good tools for detecting changes in water quality and for process control.