Sampling for PPCPs in wastewater systems: comparison of different sampling modes and optimization strategies

The aim of this study was to assess uncertainties associated with different sampling modes when evaluating loads of pharmaceuticals and personal care products (PPCPs) in sewers and influents to sewage treatment plants (STPs). The study demonstrates that sampling uncertainty can range from "not significant" to "far greater than the uncertainty due to chemical analysis", which is site- and compound-specific and depends on the (in)accuracy of the analytical method. Conventional sampling devices operated in common time- or flow-proportional sampling modes, and applying traditional sampling intervals of 30 min or longer can result in the collection of nonrepresentative samples. At the influent of a STP, wastewater may appear as a continuous stream, but it is actually composed of a number of intermittently discharged, individual wastewater packets from household appliances, industries, or subcatchments in pressurized sewer systems. The resulting heterogeneity can cause significant short-term variations of pollutant loads. We present different experimental results and a modeling approach showing that the magnitude of sampling uncertainty depends mainly on the number of pollutant peaks and the sampling frequency; sampling intervals of 5 min or shorter may be required to properly account for temporal PPCP variations in influents of STPs. A representative sample is a prerequisite for providing meaningful analytical results and cannot be compensated with a large number of samples, accurate chemical analysis, or sophisticated statistical evaluation. This study highlights that generalizing from one case to another is difficult and hence a careful systems analysis of the catchment under investigation, or precautionary choice for a sophisticated sampling mode, is necessary to prove reproducibility.