Low-cost sensors for greywater recycling

The growing importance of water reuse systems necessitates reliable and affordable monitoring technologies to ensure safe greywater recycling. This project, funded by ZHAW start-up support, investigates the feasibility of low-cost sensor solutions for real-time monitoring of greywater quality, aiming to support decentralized treatment systems. A literature review identified key parameters for safe reuse, including turbidity, pH, chemical oxygen demand (COD), and microbiological indicators. Conventional monitoring often relies on laboratory analyses, leading to delays and recurring costs, whereas emerging sensor technologies—such as spectrophotometry, fluorescence sensing, and machine learning-based approaches—offer real-time, multi-parameter capabilities. Experimental evaluation of a low-cost sensor array deployed at the KREIS-Haus facility showed limitations in accuracy, stability, and maintenance requirements. While parameters like dissolved oxygen showed moderate correlation with commercial systems, issues such as sensor drift, calibration challenges, and biofilm interference reduced reliability. Two novel sensing approaches were further assessed. Image-based analysis proved unsuitable for predicting turbidity under realistic conditions, largely due to sensitivity to environmental factors. In contrast, multi-wavelength spectrophotometry combined with regression models demonstrated strong performance, enabling accurate prediction of COD (R² up to 0.84) using relatively simple methods. Building on these findings, a prototype low-cost sensor was developed in collaboration with the ZHAW Institute of Signal Processing, integrating absorbance, fluorescence, and scattering measurements. Initial tests confirmed its potential for estimating water quality parameters. Overall, the study highlights that while low-cost conventional sensors face significant limitations, optical and data-driven approaches offer a promising pathway toward affordable, real-time greywater monitoring. Future work should focus on improving robustness, calibration, and field validation to enable broader implementation.