Odour concentrations prediction based on odorants concentrations from biosolid emissions

Abstract

Biosolids storage areas are a significant contributor to wastewater treatment plant (WWTPs) odour emissions which can cause sensorial impact to surrounding communities. Most odour impact regulations are based on odour concentration (COD) measurements determined by dynamic olfactometry. Understanding the relationship between odorants concentrations and COD in the biosolids emission is important to identify how the measurement and monitoring can be conducted using analytical rather than sensorial techniques. Some of the odorants are unknown in biosolid emissions, increasing the uncertainty in predicting COD. In this study, emissions from 56 biosolid samples collected from two WWTPs located in Sydney, Australia, were analysed by analytical and sensorial methods, including olfactory detection port (ODP) and dynamic olfactometry. Concentrations of 25 odorants and two ordinal variables represented odour events detected by ODP assessors were linked to COD values. Bayesian Model Averaging and Variable Selection with Bayesian Adaptive Sampling were applied to model the relation between COD and odorants concentrations. Results indicate the usability of the probabilistic methods and nonlinear transformations in modelling the odour concentrations based on odorants concentrations from biosolids emission and the accuracy of a small dataset.