Stochastic modeling to identify requirements for centralized monitoring of distributed wastewater treatment
Distributed (decentralized) wastewater treatment can, in many situations, be a valuable alternative to a centralized sewer network and wastewater treatment plant. However, it is critical for its acceptance whether the same overall treatment performance can be achieved without on-site staff, and whether its performance can be measured. In this paper we argue and illustrate that the system performance depends not only on the design performance and reliability of the individual treatment units, but also significantly on the monitoring scheme, i.e. on the reliability of the process information. For this purpose, we present a simple model of a fleet of identical treatment units. Thereby, their performance depends on four stochastic variables: the reliability of the treatment unit, the respond time for the repair of failed units, the reliability of on-line sensors, and the frequency of routine inspections. The simulated scenarios show a significant difference between the true performance and the observations by the sensors and inspections. The results also illustrate the trade-off between investing in reactor and sensor technology and in human interventions in order to achieve a certain target performance. Modeling can quantify such effects and thereby support the identification of requirements for the centralized monitoring of distributed treatment units. The model approach is generic and can be extended and applied to various distributed wastewater treatment technologies and contexts.