In this study simultaneously monitored drinking patterns from multiple pens in one weaner section and one finisher section herd are modeled in a multivariate dynamic linear model (DLM). Correlations between the monitored data are incorporated in the model allowing it to include interactions between pens in each section. A two-sided tabular CUSUM is applied as control chart, whereby alarms for outbreaks of unwanted events amongst the pigs can be generated for specific pens in the herd. Area-specific alarms enable the manager to include specific knowledge of the animals in the targeted area and choose the best suited intervention.